bump version

Contents/Code/__init__.py

@@ -24,12 +24,11 @@ import support
 import interface
 sys.modules["interface"] = interface
 
-from subliminal.cli import MutexLock
 from subzero.constants import OS_PLEX_USERAGENT, PERSONAL_MEDIA_IDENTIFIER
 from interface.menu import *
 from support.plex_media import media_to_videos, get_media_item_ids, scan_videos
 from support.subtitlehelpers import get_subtitles_from_metadata
-from support.storage import whack_missing_parts, save_subtitles, get_subtitle_storage
+from support.storage import whack_missing_parts, save_subtitles
 from support.items import is_ignored
 from support.config import config
 from support.lib import get_intent
@@ -43,13 +42,7 @@ def Start():
     HTTP.CacheTime = 0
     HTTP.Headers['User-agent'] = OS_PLEX_USERAGENT
 
-    try:
-        subliminal.region.configure('dogpile.cache.dbm', expiration_time=datetime.timedelta(days=30),
-                                    arguments={'filename': os.path.join(config.data_items_path, 'subzero.dbm'),
-                                               'lock_factory': MutexLock})
-    except:
-        Log.Warn("Not using file based cache!")
-        subliminal.region.configure('dogpile.cache.memory')
+    config.init_cache()
 
     # clear expired intents
     intent = get_intent()

Contents/Code/interface/__init__.py

@@ -18,3 +18,6 @@ sys.modules["interface.refresh_item"] = refresh_item
 
 import item_details
 sys.modules["interface.item_details"] = item_details
+
+import sub_mod
+sys.modules["interface.modification"] = sub_mod

Contents/Code/interface/advanced.py

@@ -148,24 +148,34 @@ def TriggerStorageMaintenance(randomize=None):
 
 @route(PREFIX + '/get_logs_link')
 def GetLogsLink():
+    if not config.plex_token:
+        oc = ObjectContainer(title2="Download Logs", no_cache=True, no_history=True,
+                             header="Sorry, feature unavailable",
+                             message="Universal Plex token not available")
+        return oc
+
     # try getting the link base via the request in context, first, otherwise use the public ip
     req_headers = Core.sandbox.context.request.headers
+    get_external_ip = True
+    link_base = ""
 
     if "Origin" in req_headers:
         link_base = req_headers["Origin"]
         Log.Debug("Using origin-based link_base")
+        get_external_ip = False
 
     elif "Referer" in req_headers:
         parsed = urlparse.urlparse(req_headers["Referer"])
         link_base = "%s://%s:%s" % (parsed.scheme, parsed.hostname, parsed.port)
         Log.Debug("Using referer-based link_base")
+        get_external_ip = False
 
-    else:
+    if get_external_ip or "plex.tv" in link_base:
         ip = Core.networking.http_request("http://www.plexapp.com/ip.php", cacheTime=7200).content.strip()
         link_base = "https://%s:32400" % ip
         Log.Debug("Using ip-based fallback link_base")
 
-    logs_link = "%s%s?X-Plex-Token=%s" % (link_base, PREFIX + '/logs', config.universal_plex_token)
+    logs_link = "%s%s?X-Plex-Token=%s" % (link_base, PREFIX + '/logs', config.plex_token)
     oc = ObjectContainer(title2="Download Logs", no_cache=True, no_history=True,
                          header="Copy this link and open this in your browser, please",
                          message=logs_link)

Contents/Code/interface/item_details.py

@@ -1,23 +1,19 @@
 # coding=utf-8
 import os
-import traceback
 
-from babelfish import Language
-
-from subzero.constants import PREFIX
+from sub_mod import SubtitleModificationsMenu
 from menu_helpers import debounce, SubFolderObjectContainer, default_thumb, add_ignore_options, get_item_task_data, \
     set_refresh_menu_state
+
 from refresh_item import RefreshItem
+from subzero.constants import PREFIX
+from support.config import config
 from support.helpers import timestamp, cast_bool, df, get_language
 from support.items import get_item_kind_from_rating_key, get_item, get_current_sub
-from support.plex_media import get_plex_metadata, scan_videos
 from support.lib import Plex
-from support.storage import get_subtitle_storage, save_subtitles
-from support.config import config
+from support.plex_media import get_plex_metadata, scan_videos
 from support.scheduler import scheduler
-
-from subliminal_patch import PatchedSubtitle as Subtitle
-from subzero.modification import registry as mod_registry
+from support.storage import get_subtitle_storage
 
 
 @route(PREFIX + '/item/{rating_key}/actions')
@@ -141,69 +137,6 @@ def SubtitleOptionsMenu(**kwargs):
     return oc
 
 
-@route(PREFIX + '/item/sub_mods/{rating_key}/{part_id}', force=bool)
-@debounce
-def SubtitleModificationsMenu(**kwargs):
-    rating_key = kwargs["rating_key"]
-    part_id = kwargs["part_id"]
-    language = kwargs["language"]
-    current_sub, stored_subs, storage = get_current_sub(rating_key, part_id, language)
-    kwargs.pop("randomize")
-
-    oc = SubFolderObjectContainer(title2=kwargs["title"], replace_parent=True)
-    for identifier, mod in mod_registry.mods.iteritems():
-        oc.add(DirectoryObject(
-            key=Callback(SubtitleApplyMod, mod_identifier=identifier, randomize=timestamp(), **kwargs),
-            title=mod.description
-        ))
-
-    oc.add(DirectoryObject(
-        key=Callback(SubtitleApplyMod, mod_identifier=None, randomize=timestamp(), **kwargs),
-        title="Restore original version",
-        summary=u"Currently applied mods: %s" % (", ".join(current_sub.mods) if current_sub.mods else "none")
-    ))
-
-    return oc
-
-
-@route(PREFIX + '/item/sub_add_mod/{rating_key}/{part_id}/{mod_identifier}', force=bool)
-@debounce
-def SubtitleApplyMod(mod_identifier=None, **kwargs):
-    if mod_identifier is not None and mod_identifier not in mod_registry.mods:
-        raise NotImplementedError
-
-    rating_key = kwargs["rating_key"]
-    part_id = kwargs["part_id"]
-    lang_a2 = kwargs["language"]
-    item_type = kwargs["item_type"]
-
-    language = Language.fromietf(lang_a2)
-
-    current_sub, stored_subs, storage = get_current_sub(rating_key, part_id, language)
-    current_sub.add_mod(mod_identifier)
-
-    storage.save(stored_subs)
-    metadata = get_plex_metadata(rating_key, part_id, item_type)
-    scanned_parts = scan_videos([metadata], kind="series" if item_type == "episode" else "movie", ignore_all=True)
-    video, plex_part = scanned_parts.items()[0]
-
-    subtitle = Subtitle(language, mods=current_sub.mods)
-    subtitle.content = current_sub.content
-    subtitle.plex_media_fps = plex_part.fps
-    subtitle.page_link = "modify subtitles with: %s" % (", ".join(current_sub.mods) if current_sub.mods else "none")
-    subtitle.language = language
-
-    try:
-        save_subtitles(scanned_parts, {video: [subtitle]}, mode="m", bare_save=True)
-        Log.Debug("Modified %s subtitle for: %s:%s with: %s", language.name, rating_key, part_id,
-                  ", ".join(current_sub.mods) if current_sub.mods else "none")
-    except:
-        Log.Error("Something went wrong when modifying subtitle: %s", traceback.format_exc())
-
-    kwargs.pop("randomize")
-    return SubtitleModificationsMenu(randomize=timestamp(), **kwargs)
-
-
 @route(PREFIX + '/item/search/{rating_key}/{part_id}', force=bool)
 @debounce
 def ListAvailableSubsForItemMenu(rating_key=None, part_id=None, title=None, item_title=None, filename=None,

Contents/Code/interface/menu.py

@@ -1,5 +1,7 @@
 # coding=utf-8
 import logging
+import os
+
 import logger
 
 from item_details import ItemDetailsMenu
@@ -11,7 +13,7 @@ from advanced import DispatchRestart
 from subzero.constants import ART, PREFIX, DEPENDENCY_MODULE_NAMES
 from support.scheduler import scheduler
 from support.config import config
-from support.helpers import timestamp,  df
+from support.helpers import timestamp, df
 from support.ignore import ignore_list
 from support.items import get_all_items, get_items_info, \
     get_item_kind_from_rating_key
@@ -147,7 +149,6 @@ def RefreshMissing(randomize=None):
 @route(PREFIX + '/ValidatePrefs', enforce_route=True)
 def ValidatePrefs():
     Core.log.setLevel(logging.DEBUG)
-    Log.Debug("Validate Prefs called.")
 
     # cache the channel state
     update_dict = False
@@ -182,9 +183,44 @@ def ValidatePrefs():
         Core.log.removeHandler(logger.console_handler)
         Log.Debug("Stop logging to console")
 
+    Log.Debug("Validate Prefs called.")
+
+    # SZ config debug
+    Log.Debug("--- SZ Config-Debug ---")
+    for attr in [
+            "app_support_path", "data_path", "data_items_path", "enable_agent",
+            "enable_channel", "permissions_ok", "missing_permissions", "fs_encoding"]:
+        Log.Debug("config.%s: %s", attr, getattr(config, attr))
+
+    for attr in ["plugin_log_path", "server_log_path"]:
+        value = getattr(config, attr)
+        access = os.access(value, os.R_OK)
+        if Core.runtime.os == "Windows":
+            try:
+                f = open(value, "r")
+                f.read(1)
+                f.close()
+            except:
+                access = False
+
+        Log.Debug("config.%s: %s (accessible: %s)", attr, value, access)
+
+    # fixme: check existance of and os access of logs
+    Log.Debug("Platform: %s", Core.runtime.platform)
+    Log.Debug("OS: %s", Core.runtime.os)
+    Log.Debug("----- Environment -----")
+    for key, value in os.environ.iteritems():
+        if key.startswith("PLEX"):
+            if "TOKEN" in key:
+                outval = "xxxxxxxxxxxxxxxxxxx"
+
+            else:
+                outval = value
+            Log.Debug("%s: %s", key, outval)
+    Log.Debug("-----------------------")
+
     Log.Debug("Setting log-level to %s", Prefs["log_level"])
     logger.register_logging_handler(DEPENDENCY_MODULE_NAMES, level=Prefs["log_level"])
     Core.log.setLevel(logging.getLevelName(Prefs["log_level"]))
 
     return
-

Contents/Code/interface/sub_mod.py

@@ -0,0 +1,231 @@
+# coding=utf-8
+
+import traceback
+import types
+
+from babelfish import Language
+
+from menu_helpers import debounce, SubFolderObjectContainer
+from subliminal_patch import PatchedSubtitle as Subtitle
+from subzero.modification import registry as mod_registry, SubtitleModifications
+from subzero.constants import PREFIX
+from support.plex_media import get_plex_metadata, scan_videos
+from support.storage import save_subtitles
+from support.helpers import timestamp, pad_title
+from support.items import get_current_sub
+
+
+@route(PREFIX + '/item/sub_mods/{rating_key}/{part_id}', force=bool)
+@debounce
+def SubtitleModificationsMenu(**kwargs):
+    rating_key = kwargs["rating_key"]
+    part_id = kwargs["part_id"]
+    language = kwargs["language"]
+    current_sub, stored_subs, storage = get_current_sub(rating_key, part_id, language)
+    kwargs.pop("randomize")
+
+    oc = SubFolderObjectContainer(title2=kwargs["title"], replace_parent=True)
+    for identifier, mod in mod_registry.mods.iteritems():
+        if mod.advanced:
+            continue
+
+        if mod.exclusive and identifier in current_sub.mods:
+            continue
+
+        oc.add(DirectoryObject(
+            key=Callback(SubtitleSetMods, mods=identifier, mode="add", randomize=timestamp(), **kwargs),
+            title=pad_title(mod.description), summary=mod.long_description or ""
+        ))
+
+    fps_mod = SubtitleModifications.get_mod_class("change_FPS")
+    oc.add(DirectoryObject(
+        key=Callback(SubtitleFPSModMenu, randomize=timestamp(), **kwargs),
+        title=pad_title(fps_mod.description), summary=fps_mod.long_description or ""
+    ))
+
+    shift_mod = SubtitleModifications.get_mod_class("shift_offset")
+    oc.add(DirectoryObject(
+        key=Callback(SubtitleShiftModUnitMenu, randomize=timestamp(), **kwargs),
+        title=pad_title(shift_mod.description), summary=shift_mod.long_description or ""
+    ))
+
+    if current_sub.mods:
+        oc.add(DirectoryObject(
+            key=Callback(SubtitleSetMods, mods=None, mode="remove_last", randomize=timestamp(), **kwargs),
+            title=pad_title("Remove last applied mod (%s)" % current_sub.mods[-1]),
+            summary=u"Currently applied mods: %s" % (", ".join(current_sub.mods) if current_sub.mods else "none")
+        ))
+        oc.add(DirectoryObject(
+            key=Callback(SubtitleListMods, randomize=timestamp(), **kwargs),
+            title=pad_title("Manage applied mods"),
+            summary=u"Currently applied mods: %s" % (", ".join(current_sub.mods))
+        ))
+
+    oc.add(DirectoryObject(
+        key=Callback(SubtitleSetMods, mods=None, mode="clear", randomize=timestamp(), **kwargs),
+        title=pad_title("Restore original version"),
+        summary=u"Currently applied mods: %s" % (", ".join(current_sub.mods) if current_sub.mods else "none")
+    ))
+
+    return oc
+
+
+@route(PREFIX + '/item/sub_mod_fps/{rating_key}/{part_id}', force=bool)
+@debounce
+def SubtitleFPSModMenu(**kwargs):
+    rating_key = kwargs["rating_key"]
+    part_id = kwargs["part_id"]
+    item_type = kwargs["item_type"]
+
+    oc = SubFolderObjectContainer(title2=kwargs["title"], replace_parent=True)
+
+    metadata = get_plex_metadata(rating_key, part_id, item_type)
+    scanned_parts = scan_videos([metadata], kind="series" if item_type == "episode" else "movie", ignore_all=True)
+    video, plex_part = scanned_parts.items()[0]
+
+    target_fps = plex_part.fps
+
+    kwargs.pop("randomize")
+
+    for fps in ["23.976", "24.000", "25.000", "29.970", "30.000", "50.000", "59.940", "60.000"]:
+        if float(fps) == float(target_fps):
+            continue
+
+        if float(fps) > float(target_fps):
+            indicator = "subs constantly getting faster"
+        else:
+            indicator = "subs constantly getting slower"
+
+        mod_ident = SubtitleModifications.get_mod_signature("change_FPS", **{"from": fps, "to": target_fps})
+        oc.add(DirectoryObject(
+            key=Callback(SubtitleSetMods, mods=mod_ident, mode="add", randomize=timestamp(), **kwargs),
+            title="%s fps -> %s fps (%s)" % (fps, target_fps, indicator)
+        ))
+
+    return oc
+
+
+POSSIBLE_UNITS = (("ms", "milliseconds"), ("s", "seconds"), ("m", "minutes"), ("h", "hours"))
+POSSIBLE_UNITS_D = dict(POSSIBLE_UNITS)
+
+
+@route(PREFIX + '/item/sub_mod_shift_unit/{rating_key}/{part_id}', force=bool)
+@debounce
+def SubtitleShiftModUnitMenu(**kwargs):
+    oc = SubFolderObjectContainer(title2=kwargs["title"], replace_parent=True)
+
+    kwargs.pop("randomize")
+
+    for unit, title in POSSIBLE_UNITS:
+        oc.add(DirectoryObject(
+            key=Callback(SubtitleShiftModMenu, unit=unit, randomize=timestamp(), **kwargs),
+            title="Adjust by %s" % title
+        ))
+
+    return oc
+
+
+@route(PREFIX + '/item/sub_mod_shift/{rating_key}/{part_id}/{unit}', force=bool)
+@debounce
+def SubtitleShiftModMenu(unit=None, **kwargs):
+    if unit not in POSSIBLE_UNITS_D:
+        raise NotImplementedError
+
+    oc = SubFolderObjectContainer(title2=kwargs["title"], replace_parent=True)
+
+    kwargs.pop("randomize")
+
+    rng = []
+    if unit == "h":
+        rng = range(-10, 11)
+    elif unit in ("m", "s"):
+        rng = range(-59, 60)
+    elif unit == "ms":
+        rng = range(-900, 1000, 100)
+
+    for i in rng:
+        if i == 0:
+            continue
+
+        mod_ident = SubtitleModifications.get_mod_signature("shift_offset", **{unit: i})
+        oc.add(DirectoryObject(
+            key=Callback(SubtitleSetMods, mods=mod_ident, mode="add", randomize=timestamp(), **kwargs),
+            title="%s %s" % (("%s" if i < 0 else "+%s") % i, unit)
+        ))
+
+    return oc
+
+
+@route(PREFIX + '/item/sub_set_mods/{rating_key}/{part_id}/{mods}/{mode}', force=bool)
+@debounce
+def SubtitleSetMods(mods=None, mode=None, **kwargs):
+    if not isinstance(mods, types.ListType) and mods:
+        mods = [mods]
+
+    rating_key = kwargs["rating_key"]
+    part_id = kwargs["part_id"]
+    lang_a2 = kwargs["language"]
+    item_type = kwargs["item_type"]
+
+    language = Language.fromietf(lang_a2)
+
+    current_sub, stored_subs, storage = get_current_sub(rating_key, part_id, language)
+    if mode == "add":
+        for mod in mods:
+            identifier, args = SubtitleModifications.parse_identifier(mod)
+            if identifier not in mod_registry.mods_available:
+                raise NotImplementedError("Mod unknown or not registered")
+
+            current_sub.add_mod(mod)
+    elif mode == "clear":
+        current_sub.add_mod(None)
+    elif mode == "remove":
+        for mod in mods:
+            current_sub.mods.remove(mod)
+
+    elif mode == "remove_last":
+        if current_sub.mods:
+            current_sub.mods.pop()
+    else:
+        raise NotImplementedError("Wrong mode given")
+    storage.save(stored_subs)
+
+    metadata = get_plex_metadata(rating_key, part_id, item_type)
+    scanned_parts = scan_videos([metadata], kind="series" if item_type == "episode" else "movie", ignore_all=True)
+    video, plex_part = scanned_parts.items()[0]
+
+    subtitle = Subtitle(language, mods=current_sub.mods)
+    subtitle.content = current_sub.content
+    subtitle.plex_media_fps = plex_part.fps
+    subtitle.page_link = "modify subtitles with: %s" % (", ".join(current_sub.mods) if current_sub.mods else "none")
+    subtitle.language = language
+
+    try:
+        save_subtitles(scanned_parts, {video: [subtitle]}, mode="m", bare_save=True)
+        Log.Debug("Modified %s subtitle for: %s:%s with: %s", language.name, rating_key, part_id,
+                  ", ".join(current_sub.mods) if current_sub.mods else "none")
+    except:
+        Log.Error("Something went wrong when modifying subtitle: %s", traceback.format_exc())
+
+    kwargs.pop("randomize")
+    return SubtitleModificationsMenu(randomize=timestamp(), **kwargs)
+
+
+@route(PREFIX + '/item/sub_list_mods/{rating_key}/{part_id}', force=bool)
+@debounce
+def SubtitleListMods(**kwargs):
+    rating_key = kwargs["rating_key"]
+    part_id = kwargs["part_id"]
+    language = kwargs["language"]
+    current_sub, stored_subs, storage = get_current_sub(rating_key, part_id, language)
+
+    kwargs.pop("randomize")
+
+    oc = SubFolderObjectContainer(title2=kwargs["title"], replace_parent=True)
+    for identifier in current_sub.mods:
+        oc.add(DirectoryObject(
+            key=Callback(SubtitleSetMods, mods=identifier, mode="remove", randomize=timestamp(), **kwargs),
+            title="Remove: %s" % identifier
+        ))
+
+    return oc

Contents/Code/support/activities.py

@@ -11,9 +11,9 @@ class PlexActivityManager(object):
     def start(self):
         activity_sources_enabled = None
 
-        if config.universal_plex_token:
+        if config.plex_token:
             from plex import Plex
-            Plex.configuration.defaults.authentication(config.universal_plex_token)
+            Plex.configuration.defaults.authentication(config.plex_token)
             activity_sources_enabled = ["websocket"]
             Activity.on('websocket.playing', self.on_playing)
 

Contents/Code/support/config.py

@@ -9,6 +9,7 @@ import datetime
 import subliminal
 import subliminal_patch
 from babelfish import Language
+from subliminal.cli import MutexLock
 from subzero.lib.io import FileIO, get_viable_encoding
 from subzero.constants import PLUGIN_NAME, PLUGIN_IDENTIFIER, MOVIE, SHOW
 from lib import Plex
@@ -45,6 +46,7 @@ class Config(object):
     data_path = None
     data_items_path = None
     universal_plex_token = None
+    plex_token = None
     is_development = False
 
     enable_channel = True
@@ -68,6 +70,7 @@ class Config(object):
     sections = None
     enabled_sections = None
     remove_hi = False
+    fix_ocr = False
     enforce_encoding = False
     chmod = None
     forced_only = False
@@ -91,6 +94,7 @@ class Config(object):
         self.data_path = getattr(Data, "_core").storage.data_path
         self.data_items_path = os.path.join(self.data_path, "DataItems")
         self.universal_plex_token = self.get_universal_plex_token()
+        self.plex_token = os.environ.get("PLEXTOKEN", self.universal_plex_token)
 
         self.set_plugin_mode()
         self.set_plugin_lock()
@@ -109,6 +113,7 @@ class Config(object):
         self.permissions_ok = self.check_permissions()
         self.notify_executable = self.check_notify_executable()
         self.remove_hi = cast_bool(Prefs['subtitles.remove_hi'])
+        self.fix_ocr = cast_bool(Prefs['subtitles.fix_ocr'])
         self.enforce_encoding = cast_bool(Prefs['subtitles.enforce_encoding'])
         self.chmod = self.check_chmod()
         self.forced_only = cast_bool(Prefs["subtitles.only_foreign"])
@@ -118,6 +123,21 @@ class Config(object):
         self.default_mods = self.get_default_mods()
         self.initialized = True
 
+    def init_cache(self):
+        use_fallback_cache = True
+        if Core.runtime.os != "Windows":
+            try:
+                subliminal.region.configure('dogpile.cache.dbm', expiration_time=datetime.timedelta(days=30),
+                                            arguments={'filename': os.path.join(config.data_items_path, 'subzero.dbm'),
+                                                       'lock_factory': MutexLock})
+                use_fallback_cache = False
+            except:
+                pass
+
+        if use_fallback_cache:
+            Log.Warn("Not using file based cache!")
+            subliminal.region.configure('dogpile.cache.memory')
+
     def set_log_paths(self):
         # find log handler
         for handler in Core.log.handlers:
@@ -144,6 +164,8 @@ class Config(object):
         else:
             Log("Did NOT find Preferences file - please check logfile and hierarchy. Aborting!")
 
+        # fixme: windows
+
     def set_plugin_mode(self):
         if Prefs["plugin_mode"] == "only agent":
             self.enable_channel = False
@@ -412,6 +434,8 @@ class Config(object):
         mods = []
         if self.remove_hi:
             mods.append("remove_HI")
+        if self.fix_ocr:
+            mods.append("OCR_fixes")
 
         return mods
 

Contents/Code/support/helpers.py

@@ -110,9 +110,9 @@ def str_pad(s, length, align='left', pad_char=' ', trim=False):
         raise ValueError("Unknown align type, expected either 'left' or 'right'")
 
 
-def pad_title(value):
+def pad_title(value, width=49):
     """Pad a title to 30 characters to force the 'details' view."""
-    return str_pad(value, 49, pad_char=' ')
+    return str_pad(value, width, pad_char=' ')
 
 
 def get_plex_item_display_title(item, kind, parent=None, parent_title=None, section_title=None,

Contents/Code/support/tasks.py

@@ -80,6 +80,7 @@ class Task(object):
         return
 
     def run(self):
+        Log.Info(u"Task: running: %s", self.name)
         self.time_start = datetime.datetime.now()
 
     def post_run(self, data_holder):
@@ -88,6 +89,7 @@ class Task(object):
         if self.time_start:
             self.last_run_time = self.last_run - self.time_start
         self.time_start = None
+        Log.Info(u"Task: ran: %s", self.name)
 
 
 class SearchAllRecentlyAddedMissing(Task):
@@ -248,7 +250,7 @@ class DownloadSubtitleMixin(object):
         if subtitle.content:
             try:
                 whack_missing_parts(scanned_parts)
-                save_subtitles(scanned_parts, {video: [subtitle]}, mode=mode)
+                save_subtitles(scanned_parts, {video: [subtitle]}, mode=mode, mods=config.default_mods)
                 Log.Debug("Manually downloaded subtitle for: %s", rating_key)
                 download_successful = True
                 refresh_item(rating_key)
@@ -362,13 +364,21 @@ class FindBetterSubtitles(DownloadSubtitleMixin, SubtitleListingMixin, Task):
                 return
 
         now = datetime.datetime.now()
+        min_score_series = int(Prefs["subtitles.search.minimumTVScore2"].strip())
+        min_score_movies = int(Prefs["subtitles.search.minimumMovieScore2"].strip())
 
         subtitle_storage = get_subtitle_storage()
         recent_subs = subtitle_storage.load_recent_files(age_days=max_search_days)
 
         for fn, stored_subs in recent_subs.iteritems():
             video_id = stored_subs.video_id
-            cutoff = self.series_cutoff if stored_subs.item_type == "episode" else self.movies_cutoff
+
+            if stored_subs.item_type == "episode":
+                cutoff = self.series_cutoff
+                min_score = min_score_series
+            else:
+                cutoff = self.movies_cutoff
+                min_score = min_score_movies
 
             # don't search for better subtitles until at least 30 minutes have passed
             if stored_subs.added_at + datetime.timedelta(minutes=30) > now:
@@ -420,7 +430,7 @@ class FindBetterSubtitles(DownloadSubtitleMixin, SubtitleListingMixin, Task):
                         better_downloaded = False
                         better_tried_download = 0
                         for sub in subs:
-                            if sub.score > current_score:
+                            if sub.score > current_score and sub.score > min_score:
                                 Log.Debug("Better subtitle found for %s, downloading", video_id)
                                 better_tried_download += 1
                                 ret = self.download_subtitle(sub, video_id, mode="b")

Contents/DefaultPrefs.json

@@ -399,6 +399,12 @@
     "id": "subtitles.remove_hi",
     "label": "Remove Hearing Impaired tags from downloaded subtitles",
     "type": "bool",
+    "default": "false"
+  },
+  {
+    "id": "subtitles.fix_ocr",
+    "label": "Fix common OCR errors in downloaded subtitles",
+    "type": "bool",
     "default": "true"
   },
   {

Contents/Info.plist

@@ -13,7 +13,7 @@
         <key>CFBundleSignature</key>
         <string>????</string>
         <key>CFBundleVersion</key>
-        <string>2.0.0.10</string>
+        <string>2.0.0.13</string>
         <key>PlexFrameworkVersion</key>
         <string>2</string>
         <key>PlexPluginClass</key>
@@ -32,7 +32,7 @@
 
 &lt;h1&gt;Sub-Zero for Plex&lt;/h1&gt;&lt;i&gt;Subtitles done right&lt;/i&gt;
 
-Version 2.0.0.10 DEV
+Version 2.0.0.13 DEV
 
 Originally based on @bramwalet's awesome &lt;a href=&quot;https://github.com/bramwalet/Subliminal.bundle&quot;&gt;Subliminal.bundle&lt;/a&gt;
 

Contents/Libraries/Shared/enzyme/mkv.py

@@ -369,7 +369,8 @@ class Chapter(object):
         if chapterdisplays:
             string = chapterdisplays[0].get('ChapString')
             language = chapterdisplays[0].get('ChapLanguage')
-        return cls(start, hidden, enabled, end, string, language)
+            return cls(start, hidden, enabled, end, string, language)
+        return cls(start, hidden, enabled, end)
 
     def __repr__(self):
         return '<%s [%s, enabled=%s]>' % (self.__class__.__name__, self.start, self.enabled)

Contents/Libraries/Shared/subliminal_patch/subtitle.py

@@ -7,8 +7,8 @@ import chardet
 import pysrt
 import pysubs2
 from bs4 import UnicodeDammit
-from subliminal import Subtitle
 from subzero.modification import SubtitleModifications
+from subliminal import Subtitle
 
 logger = logging.getLogger(__name__)
 
@@ -138,8 +138,6 @@ class PatchedSubtitle(Subtitle):
 
     def get_modified_content(self):
         """
-        :param language: 
-        :param fps: 
         :return: string 
         """
         if not self.mods:
@@ -148,14 +146,12 @@ class PatchedSubtitle(Subtitle):
         encoding = self.guess_encoding()
 
         submods = SubtitleModifications()
-        submods.load(content=self.text, fps=self.plex_media_fps)
+        submods.load(content=self.text, language=self.language)
         submods.modify(*self.mods)
         return submods.to_string("srt", encoding=encoding).encode(encoding=encoding)
 
     def get_modified_text(self):
         """
-        :param language: 
-        :param fps: 
         :return: unicode 
         """
         content = self.get_modified_content()

Contents/Libraries/Shared/submod_test.py

@@ -1,7 +1,9 @@
 # coding=utf-8
 
-import sys
 import logging
+import sys
+
+from babelfish import Language
 
 logger = logging.getLogger(__name__)
 
@@ -14,7 +16,9 @@ if debug:
     logging.basicConfig(level=logging.DEBUG)
 
 submod = SubMod(debug=debug)
-submod.load(fn)
-submod.modify("remove_HI")
+submod.load(fn, language=Language.fromietf("eng"))
+#submod.modify("remove_HI", "OCR_fixes")
+#submod.modify("OCR_fixes")
+submod.modify("change_FPS(from=24,to=25)")
 
 #print submod.f.to_string("srt")

Contents/Libraries/Shared/subzero/modification.py

@@ -1,246 +0,0 @@
-# coding=utf-8
-
-import re
-import traceback
-from collections import OrderedDict
-
-import pysubs2
-import logging
-
-logger = logging.getLogger(__name__)
-
-
-class SubtitleModifications(object):
-    debug = False
-
-    def __init__(self, debug=False):
-        self.debug = debug
-
-    def load(self, fn=None, content=None, fps=None):
-        """
-        
-        :param fn:  filename
-        :param content: unicode 
-        :param fps: 
-        :return: 
-        """
-        try:
-            if fn:
-                self.f = pysubs2.load(fn, fps=fps)
-            elif content:
-                self.f = pysubs2.SSAFile.from_string(content, fps=fps)
-        except (IOError,
-                UnicodeDecodeError,
-                pysubs2.exceptions.UnknownFPSError,
-                pysubs2.exceptions.UnknownFormatIdentifierError,
-                pysubs2.exceptions.FormatAutodetectionError):
-            if fn:
-                logger.exception("Couldn't load subtitle: %s: %s", fn, traceback.format_exc())
-            elif content:
-                logger.exception("Couldn't load subtitle: %s", traceback.format_exc())
-
-    def modify(self, *mods):
-        new_f = []
-        for line in self.f:
-            applied_mods = []
-            for identifier in mods:
-                if identifier in registry.mods:
-                    mod = registry.mods[identifier]
-
-                    # don't bother reapplying exclusive mods multiple times
-                    if mod.exclusive and identifier in applied_mods:
-                        continue
-
-                    new_content = mod.modify(line.text, debug=self.debug)
-                    if not new_content:
-                        if self.debug:
-                            logger.debug("%s: deleting %s", identifier, line)
-                        continue
-
-                    line.text = new_content
-                    new_f.append(line)
-                    applied_mods.append(identifier)
-
-        self.f.events = new_f
-
-    def to_string(self, format="srt", encoding="utf-8"):
-        return self.f.to_string(format, encoding=encoding)
-
-    def save(self, fn):
-        self.f.save(fn)
-
-
-SubMod = SubtitleModifications
-
-
-class SubtitleModRegistry(object):
-    mods = None
-    mods_available = None
-
-    def __init__(self):
-        self.mods = OrderedDict()
-        self.mods_available = []
-
-    def register(self, mod):
-        self.mods[mod.identifier] = mod
-        self.mods_available.append(mod.identifier)
-
-registry = SubtitleModRegistry()
-
-
-class Processor(object):
-    """
-    Processor base class
-    """
-    name = None
-
-    def __init__(self, name=None):
-        self.name = name
-
-    @property
-    def info(self):
-        return self.name
-
-    def process(self, content):
-        return content
-
-    def __repr__(self):
-        return "Processor <%s %s>" % (self.__class__.__name__, self.info)
-
-    def __str__(self):
-        return repr(self)
-
-    def __unicode__(self):
-        return unicode(repr(self))
-
-
-class StringProcessor(Processor):
-    """
-    String replacement processor base
-    """
-
-    def __init__(self, search, replace, name=None):
-        super(StringProcessor, self).__init__(name=name)
-        self.search = search
-        self.replace = replace
-
-    def process(self, content):
-        return content.replace(self.search, self.replace)
-
-
-class ReProcessor(Processor):
-    """
-    Regex processor
-    """
-    pattern = None
-    replace_with = None
-
-    def __init__(self, pattern, replace_with, name=None):
-        super(ReProcessor, self).__init__(name=name)
-        self.pattern = pattern
-        self.replace_with = replace_with
-
-    def process(self, content, debug=False):
-        return self.pattern.sub(self.replace_with, content)
-
-
-class NReProcessor(ReProcessor):
-    """
-    Single line regex processor
-    """
-
-    def process(self, content, debug=False):
-        lines = []
-        for line in content.split(r"\N"):
-            a = super(NReProcessor, self).process(line, debug=debug)
-            if not a:
-                continue
-            lines.append(a)
-        return r"\N".join(lines)
-
-
-class SubtitleModification(object):
-    identifier = None
-    description = None
-    exclusive = False
-    pre_processors = []
-    processors = []
-    post_processors = []
-
-    @classmethod
-    def _process(cls, content, processors, debug=False):
-        if not content:
-            return
-
-        new_content = content
-        for processor in processors:
-            old_content = new_content
-            new_content = processor.process(new_content, debug=debug)
-            if not new_content:
-                if debug:
-                    logger.debug("Processor returned empty line: %s", processor)
-                break
-            if debug:
-                if old_content == new_content:
-                    continue
-                logger.debug("%s: %s -> %s", processor, old_content, new_content)
-        return new_content
-
-    @classmethod
-    def pre_process(cls, content, debug=False):
-        return cls._process(content, cls.pre_processors, debug=debug)
-
-    @classmethod
-    def process(cls, content, debug=False):
-        return cls._process(content, cls.processors, debug=debug)
-
-    @classmethod
-    def post_process(cls, content, debug=False):
-        return cls._process(content, cls.post_processors, debug=debug)
-
-    @classmethod
-    def modify(cls, content, debug=False):
-        new_content = content
-        for method in ("pre_process", "process", "post_process"):
-            new_content = getattr(cls, method)(new_content, debug=debug)
-
-        return new_content
-
-
-class SubtitleTextModification(SubtitleModification):
-    post_processors = [
-        # empty tag
-        ReProcessor(re.compile(r'({\\\w+1})[\s.,-_!?]+({\\\w+0})'), "", name="empty_tag"),
-
-        # empty line (needed?)
-        NReProcessor(re.compile(r'^\s+$'), "", name="empty_line"),
-
-        # empty dash line (needed?)
-        NReProcessor(re.compile(r'(^[\s]*[\-]+[\s]*)$'), "", name="empty_dash_line"),
-
-        # clean whitespace at start and end
-        ReProcessor(re.compile(r'^\s*([^\s]+)\s*$'), r"\1", name="surrounding_whitespace"),
-    ]
-
-
-class HearingImpaired(SubtitleTextModification):
-    identifier = "remove_HI"
-    description = "Remove Hearing Impaired tags"
-    exclusive = True
-
-    processors = [
-        # brackets
-        NReProcessor(re.compile(r'(?sux)[([].+[)\]]'), "", name="HI_brackets"),
-
-        # text before colon (and possible dash in front)
-        NReProcessor(re.compile(r'(?u)(^[A-z\-]+[\w\s]*:[^0-9{2}][\s]*)'), "", name="HI_before_colon"),
-
-        # all caps line (at least 3 chars)
-        NReProcessor(re.compile(r'(?u)(^[A-Z]{3,}$)'), "", name="HI_all_caps"),
-
-        # dash in front
-        NReProcessor(re.compile(r'(?u)^\s*-\s*'), "", name="HI_starting_dash"),
-    ]
-
-
-registry.register(HearingImpaired)

Contents/Libraries/Shared/subzero/modification/__init__.py

@@ -0,0 +1,5 @@
+# coding=utf-8
+
+from registry import registry
+from mods import hearing_impaired, ocr_fixes, fps, offset
+from main import SubtitleModifications, SubMod

Contents/Libraries/Shared/subzero/modification/dictionaries/__init__.py

@@ -0,0 +1,3 @@
+# coding=utf-8
+
+from data import data

Contents/Libraries/Shared/subzero/modification/dictionaries/data.py

@@ -0,0 +1,174 @@
+import re
+from collections import OrderedDict
+data = {'dan': {'BeginLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'EndLines': {'data': OrderedDict(),
+                      'pattern': None},
+         'PartialLines': {'data': OrderedDict(),
+                          'pattern': None},
+         'PartialWordsAlways': {'data': OrderedDict([(u'\xa4', u'o'), (u'IVI', u'M'), (u'lVI', u'M'), (u'IVl', u'M'), (u'lVl', u'M')]),
+                                'pattern': None},
+         'WholeLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'WholeWords': {'data': OrderedDict([(u'Haner', u'Han er'), (u'JaveL', u'Javel'), (u'Pa//e', u'Palle'), (u'bffte', u'bitte'), (u'Utro//gt', u'Utroligt'), (u'Kommerdu', u'Kommer du'), (u'smi/er', u'smiler'), (u'/eg', u'leg'), (u'harvinger', u'har vinger'), (u'/et', u'let'), (u'erjeres', u'er jeres'), (u'hardet', u'har det'), (u't\xe6nktjer', u't\xe6nkt jer'), (u'erjo', u'er jo'), (u'sti/', u'stil'), (u'Iappe', u'lappe'), (u'Beklagel\xe7', u'Beklager,'), (u'vardet', u'var det'), (u'afden', u'af den'), (u'snupperjeg', u'snupper jeg'), (u'ikkejeg', u'ikke jeg'), (u'bliverjeg', u'bliver jeg'), (u'hartravit', u'har travlt'), (u'pandekagef/ag', u'pandekageflag'), (u'Stormvarsell', u'Stormvarsel!'), (u'stormvejn', u'stormvejr.'), (u'morgenkomp/et', u'morgenkomplet'), (u'/yv', u'lyv'), (u'varjo', u'var jo'), (u'/eger', u'leger'), (u'harjeg', u'har jeg'), (u'havdejeg', u'havde jeg'), (u'hvorjeg', u'hvor jeg'), (u'n\xe5rjeg', u'n\xe5r jeg'), (u'g\xe5rvi', u'g\xe5r vi'), (u'atjeg', u'at jeg'), (u'isine', u'i sine'), (u'f\xe5rjeg', u'f\xe5r jeg'), (u'k\xe6rtighed', u'k\xe6rlighed'), (u'skullejeg', u'skulle jeg'), (u'laest', u'l\xe6st'), (u'laese', u'l\xe6se'), (u'g\xf8rjeg', u'g\xf8r jeg'), (u'g\xf8rvi', u'g\xf8r vi'), (u'angrerjo', u'angrer jo'), (u'Hvergang', u'Hver gang'), (u'erder', u'er der'), (u'villetilgive', u'ville tilgive'), (u'\ufb01eme', u'fjeme'), (u'genopst\xe5ri', u'genopst\xe5r i'), (u'svigtejer', u'svigte jer'), (u'kommernu', u'kommer nu'), (u'n\xe5rman', u'n\xe5r man'), (u'erfire', u'er fire'), (u'Hvorfor\ufb01nderdu', u'Hvorfor \ufb01nder du'), (u'undertigt', u'underligt'), (u'itroen', u'i troen'), (u'erl\xf8gnt', u'er l\xf8gn!'), (u'g\xf8rden', u'g\xf8r den'), (u'forhelvede', u'for helvede'), (u'hjpe', u'hj\xe6lpe'), (u'togeti', u'toget i'), (u'M\xe5jeg', u'M\xe5 jeg'), (u'savnerjer', u'savner jer'), (u'erjeg', u'er jeg'), (u'vaere', u'v\xe6re'), (u'geme', u'gerne'), (u'trorp\xe5', u'tror p\xe5'), (u'forham', u'for ham'), (u'afham', u'af ham'), (u'harjo', u'har jo'), (u'ovema\ufb01et', u'overnattet'), (u'begae\ufb01ighed', u'beg\xe6rlighed'), (u'sy\u2019g', u'syg'), (u'Imensjeg', u'Imens jeg'), (u'bliverdu', u'bliver du'), (u'\ufb01ser', u'fiser'), (u'manipuierer', u'manipulerer'), (u'forjeg', u'for jeg'), (u'iivgivendefor', u'livgivende for'), (u'formig', u'for mig'), (u'Hardu', u'Har du'), (u'fornold', u'forhold'), (u'defrelste', u'de frelste'), (u'S\xe5jeg', u'S\xe5 jeg'), (u'varjeg', u'var jeg'), (u'g\xf8rved', u'g\xf8r ved'), (u'kalderjeg', u'kalder jeg'), (u'\ufb02ytte', u'flytte'), (u'handlerdet', u'handler det'), (u'trorjeg', u'tror jeg'), (u'\ufb02ytter', u'flytter'), (u'soverjeg', u'sover jeg'), (u'\ufb01nderud', u'\ufb01nder ud'), (u'naboerp\xe5', u'naboer p\xe5'), (u'ervildt', u'er vildt'), (u'v\xe6reher', u'v\xe6re her'), (u'hyggerjer', u'hygger jer'), (u'borjo', u'bor jo'), (u'kommerikke', u'kommer ikke'), (u'folkynde', u'forkynde'), (u'farglad', u'far glad'), (u'misterjeg', u'mister jeg'), (u'\ufb01nt', u'fint'), (u'Harl', u'Har I'), (u'bedejer', u'bede jer'), (u'synesjeg', u'synes jeg'), (u'vartil', u'var til'), (u'eren', u'er en'), (u'\\Al', u'Vil'), (u'\\A', u'Vi'), (u'fjeme', u'fjerne'), (u'Iigefyldt', u'lige fyldt'), (u'ertil', u'er til'), (u'fa\ufb01igt', u'farligt'), (u'\ufb01nder', u'finder'), (u'\ufb01ndes', u'findes'), (u'irettesae\ufb01else', u'irettes\xe6ttelse'), (u'ermed', u'er med'), (u'\xe8n', u'\xe9n'), (u'gikjoi', u'gik jo i'), (u'Hvisjeg', u'Hvis jeg'), (u'ovema\ufb01er', u'overnatter'), (u'hoident', u'holdent'), (u'\\Adne', u'Vidne'), (u'fori', u'for i'), (u'vei', u'vel'), (u'savnerjerjo', u'savner jer jo'), (u'elskerjer', u'elsker jer'), (u'harl\xf8jet', u'har l\xf8jet'), (u'eri', u'er i'), (u'\ufb01ende', u'fjende'), (u'derjo', u'der jo'), (u'sigerjo', u'siger jo'), (u'menerjeg', u'mener jeg'), (u'Harjeg', u'Har jeg'), (u'sigerjeg', u'siger jeg'), (u'splitterjeg', u'splitter jeg'), (u'erjournalist', u'er journalist'), (u'erjoumalist', u'er journalist'), (u'Forjeg', u'For jeg'), (u'g\xe2rjeg', u'g\xe5r jeg'), (u'N\xe2rjeg', u'N\xe5r jeg'), (u'a\ufb02lom', u'afkom'), (u'farerjo', u'farer jo'), (u'tagerjeg', u'tager jeg'), (u'Virkerjeg', u'Virker jeg'), (u'morerjer', u'morer jer'), (u'kommerjo', u'kommer jo'), (u'istand', u'i stand'), (u'b\xf8m', u'b\xf8rn'), (u'frygterjeg', u'frygter jeg'), (u'kommerjeg', u'kommer jeg'), (u'eriournalistelev', u'er journalistelev'), (u'harfat', u'har fat'), (u'f\xe5r\ufb01ngre', u'f\xe5r \ufb01ngre'), (u'sl\xe2rjeg', u'sl\xe5r jeg'), (u'bam', u'barn'), (u'erjournalistelev', u'er journalistelev'), (u'politietjo', u'politiet jo'), (u'elskerjo', u'elsker jo'), (u'vari', u'var i'), (u'fornemmerjeres', u'fornemmer jeres'), (u'udkl\xe6kketl', u'udkl\xe6kket!'), (u'\xed', u'i'), (u'nyi', u'ny i'), (u'Iumijelse', u'forn\xf8jelse'), (u'vures', u'vores'), (u'I/Vash\xedngtan', u'Washington'), (u'opleverjeg', u'oplever jeg'), (u'PANTEL\xc3NER', u'PANTEL\xc5NER'), (u'Gudmurgen', u'Godmorgen'), (u'SKYDEV\xc3BEN', u'SKYDEV\xc5BEN'), (u'P\xLine truncated
+                        'pattern': u"(um)\\bHaner\\b|\\bJaveL\\b|\\bPa\\/\\/e\\b|\\bbffte\\b|\\bUtro\\/\\/gt\\b|\\bKommerdu\\b|\\bsmi\\/er\\b|\\b\\/eg\\b|\\bharvinger\\b|\\b\\/et\\b|\\berjeres\\b|\\bhardet\\b|\\bt\\\xe6nktjer\\b|\\berjo\\b|\\bsti\\/\\b|\\bIappe\\b|\\bBeklagel\\\xe7\\b|\\bvardet\\b|\\bafden\\b|\\bsnupperjeg\\b|\\bikkejeg\\b|\\bbliverjeg\\b|\\bhartravit\\b|\\bpandekagef\\/ag\\b|\\bStormvarsell\\b|\\bstormvejn\\b|\\bmorgenkomp\\/et\\b|\\b\\/yv\\b|\\bvarjo\\b|\\b\\/eger\\b|\\bharjeg\\b|\\bhavdejeg\\b|\\bhvorjeg\\b|\\bn\\\xe5rjeg\\b|\\bg\\\xe5rvi\\b|\\batjeg\\b|\\bisine\\b|\\bf\\\xe5rjeg\\b|\\bk\\\xe6rtighed\\b|\\bskullejeg\\b|\\blaest\\b|\\blaese\\b|\\bg\\\xf8rjeg\\b|\\bg\\\xf8rvi\\b|\\bangrerjo\\b|\\bHvergang\\b|\\berder\\b|\\bvilletilgive\\b|\\b\\\ufb01eme\\b|\\bgenopst\\\xe5ri\\b|\\bsvigtejer\\b|\\bkommernu\\b|\\bn\\\xe5rman\\b|\\berfire\\b|\\bHvorfor\\\ufb01nderdu\\b|\\bundertigt\\b|\\bitroen\\b|\\berl\\\xf8gnt\\b|\\bg\\\xf8rden\\b|\\bforhelvede\\b|\\bhjpe\\b|\\btogeti\\b|\\bM\\\xe5jeg\\b|\\bsavnerjer\\b|\\berjeg\\b|\\bvaere\\b|\\bgeme\\b|\\btrorp\\\xe5\\b|\\bforham\\b|\\bafham\\b|\\bharjo\\b|\\bovema\\\ufb01et\\b|\\bbegae\\\ufb01ighed\\b|\\bsy\\\u2019g\\b|\\bImensjeg\\b|\\bbliverdu\\b|\\b\\\ufb01ser\\b|\\bmanipuierer\\b|\\bforjeg\\b|\\biivgivendefor\\b|\\bformig\\b|\\bHardu\\b|\\bfornold\\b|\\bdefrelste\\b|\\bS\\\xe5jeg\\b|\\bvarjeg\\b|\\bg\\\xf8rved\\b|\\bkalderjeg\\b|\\b\\\ufb02ytte\\b|\\bhandlerdet\\b|\\btrorjeg\\b|\\b\\\ufb02ytter\\b|\\bsoverjeg\\b|\\b\\\ufb01nderud\\b|\\bnaboerp\\\xe5\\b|\\bervildt\\b|\\bv\\\xe6reher\\b|\\bhyggerjer\\b|\\bborjo\\b|\\bkommerikke\\b|\\bfolkynde\\b|\\bfarglad\\b|\\bmisterjeg\\b|\\b\\\ufb01nt\\b|\\bHarl\\b|\\bbedejer\\b|\\bsynesjeg\\b|\\bvartil\\b|\\beren\\b|\\b\\\\Al\\b|\\b\\\\A\\b|\\bfjeme\\b|\\bIigefyldt\\b|\\bertil\\b|\\bfa\\\ufb01igt\\b|\\b\\\ufb01nder\\b|\\b\\\ufb01ndes\\b|\\birettesae\\\ufb01else\\b|\\bermed\\b|\\b\\\xe8n\\b|\\bgikjoi\\b|\\bHvisjeg\\b|\\bovema\\\ufb01er\\b|\\bhoident\\b|\\b\\\\Adne\\b|\\bfori\\b|\\bvei\\b|\\bsavnerjerjo\\b|\\belskerjer\\b|\\bharl\\\xf8jet\\b|\\beri\\b|\\b\\\ufb01ende\\b|\\bderjo\\b|\\bsigerjo\\b|\\bmenerjeg\\b|\\bHarjeg\\b|\\bsigerjeg\\b|\\bsplitterjeg\\b|\\berjournalist\\b|\\berjoumalist\\b|\\bForjeg\\b|\\bg\\\xe2rjeg\\b|\\bN\\\xe2rjeg\\b|\\ba\\\ufb02lom\\b|\\bfarerjo\\b|\\btagerjeg\\b|\\bVirkerjeg\\b|\\bmorerjer\\b|\\bkommerjo\\b|\\bistand\\b|\\bb\\\xf8m\\b|\\bfrygterjeg\\b|\\bkommerjeg\\b|\\beriournalistelev\\b|\\bharfat\\b|\\bf\\\xe5r\\\ufb01ngre\\b|\\bsl\\\xe2rjeg\\b|\\bbam\\b|\\berjournalistelev\\b|\\bpolitietjo\\b|\\belskerjo\\b|\\bvari\\b|\\bfornemmerjeres\\b|\\budkl\\\xe6kketl\\b|\\b\\\xed\\b|\\bnyi\\b|\\bIumijelse\\b|\\bvures\\b|\\bI\\/Vash\\\xedngtan\\b|\\bopleverjeg\\b|\\bPANTEL\\\xc3NER\\b|\\bGudmurgen\\b|\\bSKYDEV\\\xc3BEN\\b|\\bP\\\xc3LIDELIG\\b|\\bavertalte\\b|\\bOms\\\xedder\\b|\\blurteb\\\xe5d\\b|\\bTelrslning\\b|\\bmiU\\\xf8\\b|\\bg\\\xe5ri\\b|\\bFan\\/el\\b|\\babe\\\ufb01\\\xe6s\\b|\\bhartalt\\b|\\b\\\\\\\xc5rkelig\\b|\\bbeklagerjeg\\b|\\bN\\\xe5rjeg\\b|\\brnaend\\b|\\bvaskebjorn\\b|\\bIvil\\b|\\bbesog\\b|\\bVaer\\b|\\bUndersogte\\b|\\bmodte\\b|\\btoj\\b|\\bfodt\\b|\\bgore\\b|\\bprovede\\b|\\bforste\\b|\\bigang\\b|\\bligenu\\b|\\bclet\\b|\\bStrombell\\b|\\btmvlt\\b|\\bstudererjournalistik\\b|\\binforrnererjeg\\b|\\bomk\\\ufb01ng\\b|\\btilAsg\\\xe5rd\\b|\\bKederjeg\\b|\\bjaettetamp\\b|\\berjer\\b|\\batjulehygge\\b|\\bUeneste\\b|\\bfoltsaetter\\b|\\bA\\/ice\\b|\\btvivlerjeg\\b|\\bhenterjer\\b|\\bforst\\\xe5rjeg\\b|\\bhvisjeg\\b|\\b\\/\\\xe6rt\\b|\\bvfgtrgt\\b|\\bhurtigtjeg\\b|\\bkenderjo\\b|\\bseiv\\b|\\b\\/\\\xe6gehuset\\b|\\bherjo\\b|\\bstolerjeg\\b|\\bdigi\\b|\\btaberi\\b|\\bsl\\\xe5rjeres\\b|\\blaere\\b|\\btr\\\xe6nerwushu\\b|\\befterjeg\\b|\\be\\\ufb01er\\b|\\bdui\\b|\\ba\\\ufb01en\\b|\\bbliveri\\b|\\bacceptererjer\\b|\\bdrikkerjo\\b|\\b\\\ufb01anjin\\b|\\berl\\\xe6nge\\b|\\berikke\\b|\\bmedjer\\b|\\bTmykke\\b|\\b\\'\\\ufb01anjins\\b|\\bMesteri\\b|\\bsagdetil\\b|\\bindei\\b|\\bo\\\ufb01e\\b|\\b\\'\\\ufb01lgiv\\b|\\bLf\\\xe5r\\b|\\bviserjer\\b|\\bRejsjerblot\\b|\\b\\'\\\ufb01llad\\b|\\biiiie\\\ufb01nger\\b|\\bVILOMFATTE\\b|\\bmo\\\ufb01o\\b|\\bg\\\xf8rjer\\b|\\bgifi\\b|\\bhardu\\b|\\bgi\\\ufb01\\b|\\bIaeggerjeg\\b|\\biet\\b|\\bsv\\/yte\\b|\\bti\\/\\b|\\bWdal\\b|\\b\\\ufb01\\\xe5et\\b|\\bHvo\\/for\\b|\\bhellerikke\\b|\\bWlle\\b|\\bdr\\/ver\\b|\\bV\\\\\\\ufb02lliam\\b|\\bV\\\\\\\ufb02lliams\\b|\\bVk\\\ufb01lliam\\b|\\bv\\\xe5dejakke\\b|\\bk\\\xe6\\\ufb02l\\b|\\bsagdejeg\\b|\\boven\\/ejet\\b|\\bkarameisauce\\b|\\bLf\\\xf8lgej\\\xf8disk\\b|\\bblevjo\\b|\\basiateri\\b|\\berV\\\\\\\ufb02lliam\\b|\\blidt\\\ufb02ov\\b|\\bsagdejo\\b|\\berlige\\b|\\bVt\\\ufb01lliam\\b|\\bW\\\ufb01II\\b|\\ba\\\ufb02darede\\b|\\bhj\\\xe6iperjeg\\b|\\bladerjeg\\b|\\bH\\\xe2ndledsbeskyttere\\b|\\bLsabels\\b|\\bG\\\xf8rjeg\\b|\\bm\\\xe2jeg\\b|\\bogjeg\\b|\\bgjordejeg\\b|\\bvillejeg\\b|\\bVl\\\ufb02lliams\\b|\\bDajeg\\b|\\biorden\\b|\\bfandtjeg\\b|\\bTilykke\\b|\\bk\\\xf8rerjer\\b|\\bg\\\xf8fjeg\\b|\\bSelvflgelig\\b|\\bfdderLine truncated
+ 'deu': {'BeginLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'EndLines': {'data': OrderedDict(),
+                      'pattern': None},
+         'PartialLines': {'data': OrderedDict(),
+                          'pattern': None},
+         'PartialWordsAlways': {'data': OrderedDict([(u'IVI', u'M'), (u'IVl', u'M'), (u'I\\/I', u'M'), (u'I\\/l', u'M'), (u'lVI', u'M'), (u'lVl', u'M'), (u'l\\/I', u'M'), (u'l\\/l', u'M'), (u'\xa4', u'o')]),
+                                'pattern': None},
+         'WholeLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'WholeWords': {'data': OrderedDict([(u'/a', u'Ja'), (u'/ch', u'Ich'), (u'/d/of', u'Idiot'), (u'/ebte', u'lebte'), (u'/eid', u'leid'), (u'/hn', u'ihn'), (u'/hnen', u'Ihnen'), (u'/hr', u'Ihr'), (u'/hre', u'Ihre'), (u'/hren', u'Ihren'), (u'/m', u'im'), (u'/mmer', u'immer'), (u'/n', u'In'), (u'/ndividuen', u'Individuen'), (u'/nn', u'Inn'), (u'/oe', u'Joe'), (u'/sf', u'ist'), (u'/sf/0/1n', u'Ist John'), (u'/ungs', u'Jungs'), (u'/Vfinuten', u'Minuten'), (u'/\xe9nger', u'l\xe4nger'), (u'/\xe9uft', u'l\xe4uft'), (u'0/1', u'Oh'), (u'0/me', u'ohne'), (u'0/vne', u'ohne'), (u'00om', u'000 m'), (u'100m', u'100 m'), (u'120m', u'120 m'), (u'13Oj\xa7hrie', u'130-j\xe4hrige'), (u'145m', u'145 m'), (u'150m', u'150 m'), (u'160m', u'160 m'), (u'165m', u'165 m'), (u'19m', u'19 m'), (u'20m', u'20 m'), (u'27m', u'27 m'), (u'30m', u'30 m'), (u'37m', u'37 m'), (u'38m', u'38 m'), (u'3km', u'3 km'), (u'5/ch', u'sich'), (u'5/cher', u'sicher'), (u'5/cherer', u'sicherer'), (u'5/e', u'Sie'), (u'5/nd', u'Sind'), (u'500m', u'500 m'), (u'5ulSere', u'\xe4u\xdfere'), (u'60m', u'60 m'), (u'6de', u'\xf6de'), (u'6dere', u'\xf6dere'), (u'6ffne', u'\xf6ffne'), (u'6ffnen', u'\xf6ffnen'), (u'6ffnet', u'\xd6ffnet'), (u'6fter', u'\xf6fter'), (u'750m', u'750 m'), (u'85m', u'85 m'), (u'90m', u'90 m'), (u'a//em', u'allem'), (u'A//es', u'Alles'), (u'abbeif$en', u'abbei\xdfen'), (u'abdrficken', u'abdr\xfccken'), (u'aBen', u'a\xdfen'), (u'abergl\xe9iubischen', u'abergl\xe4ubischen'), (u'aberja', u'aber ja'), (u'aberjemand', u'aber jemand'), (u'Aberjetzt', u'Aber jetzt'), (u'abf\xe9hrt', u'abf\xe4hrt'), (u'abf\xe9illt', u'abf\xe4llt'), (u'abgef\xe9rbt', u'abgef\xe4rbt'), (u'abgeh\xe9ngt', u'abgeh\xe4ngt'), (u'abgeh\xe9rt', u'abgeh\xf6rt'), (u'abgelost', u'abgel\xf6st'), (u'abgesprengli', u'abgesprengt!'), (u'abgestfirztl', u'abgest\xfcrzt'), (u'abgestilrzt', u'abgest\xfcrzt'), (u'abgestofien', u'abgesto\xdfen'), (u'abgew\xe9hlt', u'abgew\xe4hlt'), (u'abgew\xe9hnen', u'abgew\xf6hnen'), (u'abgew\xe9hnt', u'abgew\xf6hnt'), (u'abge\xe9nderten', u'abge\xe4nderten'), (u'abh\xe9ingt', u'abh\xe4ngt'), (u'abh\xe9ngen', u'abh\xe4ngen'), (u'abh\xe9ngig', u'abh\xe4ngig'), (u'abh\xe9ngiges', u'abh\xe4ngiges'), (u'Abh\xe9rstationen', u'Abh\xf6rstationen'), (u'Abjetzt', u'Ab jetzt'), (u'abkfihlen', u'abk\xfchlen'), (u'Abkfirzung', u'Abk\xfcrzung'), (u'abkommlich', u'abk\xf6mmlich'), (u'Ablegenl', u'Ablegen!'), (u'ablfisen', u'abl\xf6sen'), (u'ablosen', u'abl\xf6sen'), (u'Ablosung', u'Abl\xf6sung'), (u'abreif$en', u'abrei\xdfen'), (u'Abrijcken', u'Abr\xfccken'), (u'abr\xe9umen', u'abr\xe4umen'), (u'Absch/ed', u'Abschied'), (u'abschiefien', u'abschie\xdfen'), (u'abschlief$en', u'abschlie\xdfen'), (u'abschliefien', u'abschlie\xdfen'), (u'abschwiiren', u'abschw\xf6ren'), (u'abstoflsend', u'absto\xdfend'), (u'Abtrijnnige', u'Abtr\xfcnnige'), (u'abwiirgt', u'abw\xfcrgt'), (u'abw\xe9gen', u'abw\xe4gen'), (u'abzuh\xe9ren', u'abzuh\xf6ren'), (u'abzuschwiiren', u'abzuschw\xf6ren'), (u'abzusfofien', u'abzusto\xdfen'), (u'ACh', u'Ach'), (u'Achtungl', u'Achtung!'), (u'Achzen', u'\xc4chzen'), (u'ACHZT', u'\xc4CHZT'), (u'Acic', u'Acid'), (u'ADDRESSDATEI', u'ADRESSDATEI'), (u'Adi\xf6s', u'Adi\xf3s'), (u'Admiralitat', u'Admiralit\xe4t'), (u'Admiralit\xe9it', u'Admiralit\xe4t'), (u'Admiralit\xe9t', u'Admiralit\xe4t'), (u'Aff\xe9re', u'Aff\xe4re'), (u'Aff\xe9ren', u'Aff\xe4ren'), (u'AFl', u'AFI'), (u'aggresivem', u'aggressivem'), (u'Agypten', u'\xc4gypten'), (u'aher', u'aber'), (u'AI/en/vichtigste', u'Allerwichtigste'), (u'Ain/vays', u'Airways'), (u'AIs', u'Als'), (u'Aktivit\xe9iten', u'Aktivit\xe4ten'), (u'Aktivit\xe9ten', u'Aktivit\xe4ten'), (u'AKTMERT', u'AKTIVIERT'), (u'Alarmsfufe', u'Alarmstufe'), (u'albem', u'albern'), (u'Albtriiume', u'Albtr\xe4ume'), (u'ale', u'als'), (u'alleinl', u'allein!'), (u'allejubeln', u'alle jubeln'), (u'allern\xe9chsten', u'allern\xe4chsten'), (u'Allmiichtigerl', u'Allm\xe4chtiger!'), (u'Allm\xe9chtige', u'Allm\xe4chtige'), (u'Allm\xe9chtiger', u'Allm\xe4chtiger'), (u'allm\xe9hlich', u'allm\xe4hlich'), (u'Allm\xe9ichtiger', u'Allm\xe4chtiger'), (u'Allsparkl', u'Allspark!'), (u'allt\xe9iglichen', u'allt\xe4glichen'), (u'ALTESTE', u'\xc4LTESTE'), (u'Altester', u'\xc4ltester'), (u'Alzte', u'\xc4rzte'), (u"Amerx'kaner", u'Amerikaner'), (u'amfisierst', u'am\xfcsierst'), (u'Amiilsierst', u'Am\xfcsierst'), (u'amiisieren', u'am\xfcsieren'), (u'amiisierenl', u'am\xfcsieren!'), (u'amiisierte', u'am\xfcsierte'), (u'Amijsant', u'Am\xfcsant'), (u'amlllsant', u'am\xfcsant'), (u'amlllsiert', u'am\xfcsiert'), (u'amtlsant', u'am\xfcsant'), (u'Amusanf', u'Am\xfcsant'), (u'amusant', u'am\xfcsant'), (u'Amusiert', u'Am\xfcsiert'), (u'Anderst', u'\xc4nderst'), (u'Anderung', u'\xc4nderung'), (u'Anderungen', u'\xc4nderungen'), (u"anfa'ngt", u'anf\xe4ngt'), (u'Anffihrer', u'Anf\xfchrer'), (u'Anffingerl', u'Anf\xe4nger!'), (u'Anfiihrer', u'Anf\xfchrer'), (u'anfijhlt', u'anf\xfchlt'), (u'Anfingerl', u'Anf\xe4nger!'), (u'Anfuhrer', u'Anf\xfchrer'), (u'Anfuhrern', u'Anf\xfchreLine truncated
+                        'pattern': u"(um)\\b\\/a\\b|\\b\\/ch\\b|\\b\\/d\\/of\\b|\\b\\/ebte\\b|\\b\\/eid\\b|\\b\\/hn\\b|\\b\\/hnen\\b|\\b\\/hr\\b|\\b\\/hre\\b|\\b\\/hren\\b|\\b\\/m\\b|\\b\\/mmer\\b|\\b\\/n\\b|\\b\\/ndividuen\\b|\\b\\/nn\\b|\\b\\/oe\\b|\\b\\/sf\\b|\\b\\/sf\\/0\\/1n\\b|\\b\\/ungs\\b|\\b\\/Vfinuten\\b|\\b\\/\\\xe9nger\\b|\\b\\/\\\xe9uft\\b|\\b0\\/1\\b|\\b0\\/me\\b|\\b0\\/vne\\b|\\b00om\\b|\\b100m\\b|\\b120m\\b|\\b13Oj\\\xa7hrie\\b|\\b145m\\b|\\b150m\\b|\\b160m\\b|\\b165m\\b|\\b19m\\b|\\b20m\\b|\\b27m\\b|\\b30m\\b|\\b37m\\b|\\b38m\\b|\\b3km\\b|\\b5\\/ch\\b|\\b5\\/cher\\b|\\b5\\/cherer\\b|\\b5\\/e\\b|\\b5\\/nd\\b|\\b500m\\b|\\b5ulSere\\b|\\b60m\\b|\\b6de\\b|\\b6dere\\b|\\b6ffne\\b|\\b6ffnen\\b|\\b6ffnet\\b|\\b6fter\\b|\\b750m\\b|\\b85m\\b|\\b90m\\b|\\ba\\/\\/em\\b|\\bA\\/\\/es\\b|\\babbeif\\$en\\b|\\babdrficken\\b|\\baBen\\b|\\babergl\\\xe9iubischen\\b|\\baberja\\b|\\baberjemand\\b|\\bAberjetzt\\b|\\babf\\\xe9hrt\\b|\\babf\\\xe9illt\\b|\\babgef\\\xe9rbt\\b|\\babgeh\\\xe9ngt\\b|\\babgeh\\\xe9rt\\b|\\babgelost\\b|\\babgesprengli\\b|\\babgestfirztl\\b|\\babgestilrzt\\b|\\babgestofien\\b|\\babgew\\\xe9hlt\\b|\\babgew\\\xe9hnen\\b|\\babgew\\\xe9hnt\\b|\\babge\\\xe9nderten\\b|\\babh\\\xe9ingt\\b|\\babh\\\xe9ngen\\b|\\babh\\\xe9ngig\\b|\\babh\\\xe9ngiges\\b|\\bAbh\\\xe9rstationen\\b|\\bAbjetzt\\b|\\babkfihlen\\b|\\bAbkfirzung\\b|\\babkommlich\\b|\\bAblegenl\\b|\\bablfisen\\b|\\bablosen\\b|\\bAblosung\\b|\\babreif\\$en\\b|\\bAbrijcken\\b|\\babr\\\xe9umen\\b|\\bAbsch\\/ed\\b|\\babschiefien\\b|\\babschlief\\$en\\b|\\babschliefien\\b|\\babschwiiren\\b|\\babstoflsend\\b|\\bAbtrijnnige\\b|\\babwiirgt\\b|\\babw\\\xe9gen\\b|\\babzuh\\\xe9ren\\b|\\babzuschwiiren\\b|\\babzusfofien\\b|\\bACh\\b|\\bAchtungl\\b|\\bAchzen\\b|\\bACHZT\\b|\\bAcic\\b|\\bADDRESSDATEI\\b|\\bAdi\\\xf6s\\b|\\bAdmiralitat\\b|\\bAdmiralit\\\xe9it\\b|\\bAdmiralit\\\xe9t\\b|\\bAff\\\xe9re\\b|\\bAff\\\xe9ren\\b|\\bAFl\\b|\\baggresivem\\b|\\bAgypten\\b|\\baher\\b|\\bAI\\/en\\/vichtigste\\b|\\bAin\\/vays\\b|\\bAIs\\b|\\bAktivit\\\xe9iten\\b|\\bAktivit\\\xe9ten\\b|\\bAKTMERT\\b|\\bAlarmsfufe\\b|\\balbem\\b|\\bAlbtriiume\\b|\\bale\\b|\\balleinl\\b|\\ballejubeln\\b|\\ballern\\\xe9chsten\\b|\\bAllmiichtigerl\\b|\\bAllm\\\xe9chtige\\b|\\bAllm\\\xe9chtiger\\b|\\ballm\\\xe9hlich\\b|\\bAllm\\\xe9ichtiger\\b|\\bAllsparkl\\b|\\ballt\\\xe9iglichen\\b|\\bALTESTE\\b|\\bAltester\\b|\\bAlzte\\b|\\bAmerx\\'kaner\\b|\\bamfisierst\\b|\\bAmiilsierst\\b|\\bamiisieren\\b|\\bamiisierenl\\b|\\bamiisierte\\b|\\bAmijsant\\b|\\bamlllsant\\b|\\bamlllsiert\\b|\\bamtlsant\\b|\\bAmusanf\\b|\\bamusant\\b|\\bAmusiert\\b|\\bAnderst\\b|\\bAnderung\\b|\\bAnderungen\\b|\\banfa\\'ngt\\b|\\bAnffihrer\\b|\\bAnffingerl\\b|\\bAnfiihrer\\b|\\banfijhlt\\b|\\bAnfingerl\\b|\\bAnfuhrer\\b|\\bAnfuhrern\\b|\\bAnf\\\xe9inger\\b|\\bAnf\\\xe9ingergliick\\b|\\bAnf\\\xe9nge\\b|\\banf\\\xe9ngst\\b|\\banf\\\xe9ngt\\b|\\bangebrfillt\\b|\\bangebrullt\\b|\\bangefiihrt\\b|\\bANGEHCHDRIGE\\b|\\bangehfirt\\b|\\bAngehtirigen\\b|\\bangeh\\\xe9ren\\b|\\bangeh\\\xe9rt\\b|\\bangel\\\xe9chelt\\b|\\bangerilhrt\\b|\\banger\\\ufb02hrt\\b|\\bangeschweifit\\b|\\bangespruht\\b|\\bangetiltert\\b|\\bAngriffspl\\\xe9nen\\b|\\bAngstschweili\\b|\\banhiiren\\b|\\bAnh\\\xe9inger\\b|\\banh\\\xe9lt\\b|\\banh\\\xe9ngen\\b|\\banh\\\xe9ren\\b|\\bankijndigen\\b|\\banliigen\\b|\\banlugen\\b|\\banmal3ende\\b|\\bann\\\xe9hern\\b|\\banriihrst\\b|\\banrijuhren\\b|\\banst\\\xe9indig\\b|\\banst\\\xe9indiger\\b|\\banst\\\xe9indiges\\b|\\banst\\\xe9ndig\\b|\\banst\\\xe9ndige\\b|\\banst\\\xe9ndigen\\b|\\bAnst\\\xe9ndiges\\b|\\bAntik\\\xe9rper\\b|\\bAntiquit\\\xe9t\\b|\\bAntistrahlenger\\\xe9t\\b|\\bantwortenl\\b|\\bAnwe\\/sung\\b|\\bAnwe\\/sungen\\b|\\bAnw\\\xe9iltin\\b|\\bAnw\\\xe9lte\\b|\\bAnw\\\xe9ltin\\b|\\bAnzilge\\b|\\bAnztinden\\b|\\bAnzuge\\b|\\bAnzugen\\b|\\banzuhiiren\\b|\\banzuhoren\\b|\\banzundenl\\b|\\banzupiibeln\\b|\\bAn\\\xe9sthesie\\b|\\bAn\\\xe9sthesieprofessor\\b|\\bAn\\\xe9sthesieteam\\b|\\bAn\\\xe9sthesist\\b|\\bAn\\\xe9sthesisten\\b|\\bAn\\\xe9sthetikum\\b|\\bARBEITERI\\b|\\bArbeitsfl\\\ufb02gel\\b|\\bArmeefunkger\\\xe9t\\b|\\bArmel\\b|\\bArschkichern\\b|\\bArschliicher\\b|\\bArschliichern\\b|\\bArschl\\\xe9cher\\b|\\bArschl\\\xe9chern\\b|\\bArzte\\b|\\bArzten\\b|\\bArztin\\b|\\bAtemger\\\xe9usche\\b|\\bAtlantikkiiste\\b|\\bAtlantikkuste\\b|\\bATMOSPHARE\\b|\\bAtmosph\\\xe9re\\b|\\bAtmosph\\\xe9renbereich\\b|\\bAtmosph\\\xe9reneinflugsequenz\\b|\\bAtmosph\\\xe9reneintritt\\b|\\bAttenfaiter\\b|\\bAttent\\\xe9iter\\b|\\bAttent\\\xe9ter\\b|\\bAttent\\\xe9ters\\b|\\bAttraktivit\\\xe9t\\b|\\bauBen\\b|\\bAubenblick\\b|\\bAuBenbord\\b|\\bAuBenwelt\\b|\\bauBer\\b|\\bAuBerdem\\b|\\bauBerhalb\\b|\\bauc\\/1\\b|\\bauchl\\b|\\bAuf\\$erdem\\b|\\bauf3er\\b|\\baufAugenh6he\\b|\\baufblilhende\\b|\\bauff\\'a\\'ngt\\b|\\bAuff\\\xe9lliges\\b|\\baufgebltiht\\b|\\baufgeftlhrt\\b|\\baufgeh\\\xe9ingt\\b|\\baufgeh\\\xe9rt\\b|\\baufgekl\\\xe9rt\\b|\\baufger\\\xe9umt\\b|\\baufgespiefit\\b|\\baufgewiihlter\\b|\\baufgez\\\xe9hlt\\b|\\bAufh6ren\\b|Line truncated
+ 'eng': {'BeginLines': {'data': OrderedDict([(u'Don,t ', u"Don't "), (u'Can,t ', u"Can't "), (u'Let,s ', u"Let's "), (u'He,s ', u"He's "), (u'I,m ', u"I'm "), (u'I,ve ', u"I've "), (u'I,ll ', u"I'll "), (u'You,ll ', u"You'll "), (u'T hey ', u'They '), (u'T here ', u'There '), (u'W here ', u'Where '), (u'M aybe ', u'Maybe '), (u'S hould ', u'Should '), (u'|was', u'I was'), (u'|am ', u'I am'), (u'No w, ', u'Now, '), (u'l... I ', u'I... I '), (u'L... I ', u'I... I '), (u'lrn gonna', u"I'm gonna"), (u"-l don't", u"-I don't"), (u"l don't", u"I don't"), (u'L ', u'I '), (u'-L ', u'-I '), (u'- L ', u'- I '), (u'-l ', u'-I '), (u'- l ', u'- I '), (u'G0', u'Go'), (u'GO get', u'Go get'), (u'GO to', u'Go to'), (u'GO for', u'Go for'), (u'Ifl', u'If I'), (u"lt'll", u"It'll"), (u"Lt'll", u"It'll"), (u'IVIa', u'Ma'), (u'IVIu', u'Mu'), (u'0ne', u'One'), (u'0nly', u'Only'), (u'0n ', u'On '), (u'lt ', u'It '), (u'Lt ', u'It '), (u"lt's ", u"It's "), (u"Lt's ", u"It's "), (u'ln ', u'In '), (u'Ln ', u'In '), (u'l-in ', u'I-in '), (u'l-impossible', u'I-impossible'), (u'l- impossible', u'I-impossible'), (u'L- impossible', u'I-impossible'), (u"l-isn't ", u"I-isn't "), (u"L-isn't ", u"I-isn't "), (u"l- isn't ", u"I-isn't "), (u"L- isn't ", u"I-isn't "), (u'l- l ', u'I-I '), (u'L- l ', u'I-I '), (u'l- it ', u'I-it '), (u'L- it ', u'I-it '), (u'Ls it ', u'Is it '), (u'Ls there ', u'Is there '), (u'Ls he ', u'Is he '), (u'Ls she ', u'Is she '), (u'L can', u'I can'), (u'l can', u'I can'), (u"L'm ", u"I'm "), (u"L' m ", u"I'm "), (u"Lt' s ", u"It's "), (u"I']I ", u"I'll "), (u'...ls ', u'...Is '), (u'- ls ', u'- Is '), (u'...l ', u'...I '), (u'Ill ', u"I'll "), (u'L hope ', u'I hope '), (u"|\u2019E'$ ", u"It's "), (u"The y're ", u"They're "), (u'<i>/ ', u'<i>I '), (u'/ ', u'I '), (u'<i>/n ', u'<i>In '), (u'/n ', u'In '), (u'<i>Ana\u2019 ', u'<i>And '), (u"<i>Ana' ", u'<i>And '), (u'~ ', u'- '), (u"<i>A'|'|'EN BOROUGH:", u'<i>ATTENBOROUGH:'), (u"A'|'|'EN BOROUGH:", u'ATTENBOROUGH:'), (u"DAVID A'|'|'EN BOROUGH:", u'DAVID ATTENBOROUGH:'), (u"AT|'EN BOROUGH:", u'ATTENBOROUGH:'), (u"DAVID AT|'EN BOROUGH:", u'DAVID ATTENBOROUGH:'), (u"<i>AT|'EN BOROUGH:", u'<i>ATTENBOROUGH:'), (u'TH REE ', u'THREE '), (u'NARRA TOR', u'NARRATOR'), (u'<i>NARRA TOR', u'<i>NARRATOR'), (u"lt'syour", u"It's your"), (u"It'syour", u"It's your"), (u'Ls ', u'Is ')]),
+                        'pattern': u"(um)^\\bDon\\,t\\ \\b|\\bCan\\,t\\ \\b|\\bLet\\,s\\ \\b|\\bHe\\,s\\ \\b|\\bI\\,m\\ \\b|\\bI\\,ve\\ \\b|\\bI\\,ll\\ \\b|\\bYou\\,ll\\ \\b|\\bT\\ hey\\ \\b|\\bT\\ here\\ \\b|\\bW\\ here\\ \\b|\\bM\\ aybe\\ \\b|\\bS\\ hould\\ \\b|\\b\\|was\\b|\\b\\|am\\ \\b|\\bNo\\ w\\,\\ \\b|\\bl\\.\\.\\.\\ I\\ \\b|\\bL\\.\\.\\.\\ I\\ \\b|\\blrn\\ gonna\\b|\\b\\-l\\ don\\'t\\b|\\bl\\ don\\'t\\b|\\bL\\ \\b|\\b\\-L\\ \\b|\\b\\-\\ L\\ \\b|\\b\\-l\\ \\b|\\b\\-\\ l\\ \\b|\\bG0\\b|\\bGO\\ get\\b|\\bGO\\ to\\b|\\bGO\\ for\\b|\\bIfl\\b|\\blt\\'ll\\b|\\bLt\\'ll\\b|\\bIVIa\\b|\\bIVIu\\b|\\b0ne\\b|\\b0nly\\b|\\b0n\\ \\b|\\blt\\ \\b|\\bLt\\ \\b|\\blt\\'s\\ \\b|\\bLt\\'s\\ \\b|\\bln\\ \\b|\\bLn\\ \\b|\\bl\\-in\\ \\b|\\bl\\-impossible\\b|\\bl\\-\\ impossible\\b|\\bL\\-\\ impossible\\b|\\bl\\-isn\\'t\\ \\b|\\bL\\-isn\\'t\\ \\b|\\bl\\-\\ isn\\'t\\ \\b|\\bL\\-\\ isn\\'t\\ \\b|\\bl\\-\\ l\\ \\b|\\bL\\-\\ l\\ \\b|\\bl\\-\\ it\\ \\b|\\bL\\-\\ it\\ \\b|\\bLs\\ it\\ \\b|\\bLs\\ there\\ \\b|\\bLs\\ he\\ \\b|\\bLs\\ she\\ \\b|\\bL\\ can\\b|\\bl\\ can\\b|\\bL\\'m\\ \\b|\\bL\\'\\ m\\ \\b|\\bLt\\'\\ s\\ \\b|\\bI\\'\\]I\\ \\b|\\b\\.\\.\\.ls\\ \\b|\\b\\-\\ ls\\ \\b|\\b\\.\\.\\.l\\ \\b|\\bIll\\ \\b|\\bL\\ hope\\ \\b|\\b\\|\\\u2019E\\'\\$\\ \\b|\\bThe\\ y\\'re\\ \\b|\\b\\<i\\>\\/\\ \\b|\\b\\/\\ \\b|\\b\\<i\\>\\/n\\ \\b|\\b\\/n\\ \\b|\\b\\<i\\>Ana\\\u2019\\ \\b|\\b\\<i\\>Ana\\'\\ \\b|\\b\\~\\ \\b|\\b\\<i\\>A\\'\\|\\'\\|\\'EN\\ BOROUGH\\:\\b|\\bA\\'\\|\\'\\|\\'EN\\ BOROUGH\\:\\b|\\bDAVID\\ A\\'\\|\\'\\|\\'EN\\ BOROUGH\\:\\b|\\bAT\\|\\'EN\\ BOROUGH\\:\\b|\\bDAVID\\ AT\\|\\'EN\\ BOROUGH\\:\\b|\\b\\<i\\>AT\\|\\'EN\\ BOROUGH\\:\\b|\\bTH\\ REE\\ \\b|\\bNARRA\\ TOR\\b|\\b\\<i\\>NARRA\\ TOR\\b|\\blt\\'syour\\b|\\bIt\\'syour\\b|\\bLs\\ \\b"},
+         'EndLines': {'data': OrderedDict([(u', sin', u', sir.'), (u' mothen', u' mother.'), (u" can't_", u" can't."), (u' openiL', u' open it.'), (u' of\ufb02', u' off!'), (u'pshycol', u'psycho!')]),
+                      'pattern': u"(um)\\b\\,\\ sin\\b|\\b\\ mothen\\b|\\b\\ can\\'t\\_\\b|\\b\\ openiL\\b|\\b\\ of\\\ufb02\\b|\\bpshycol\\b$"},
+         'PartialLines': {'data': OrderedDict([(u' /be ', u' I be '), (u" aren '1'", u" aren't"), (u" aren'tyou", u" aren't you"), (u" doesn '1'", u" doesn't"), (u" fr/eno'", u' friend'), (u" fr/eno'.", u' friend.'), (u" haven 'z' ", u" haven't "), (u" haven 'z'.", u" haven't."), (u' I ha ve ', u' I have '), (u" I']I ", u" I'll "), (u' L am', u' I am'), (u' L can', u' I can'), (u" L don't ", u" I don't "), (u' L hate ', u' I hate '), (u' L have ', u' I have '), (u' L like ', u' I like '), (u' L will', u' I will'), (u' L would', u' I would'), (u" L'll ", u" I'll "), (u" L've ", u" I've "), (u' m y family', u' my family'), (u" 's ", u"'s "), (u" shou/dn '1 ", u" shouldn't "), (u" won 'z' ", u" won't "), (u" won 'z'.", u" won't."), (u" wou/c/n 'z' ", u" wouldn't "), (u" wou/c/n 'z'.", u" wouldn't."), (u" wou/dn 'z' ", u" wouldn't "), (u" wou/dn 'z'.", u" wouldn't."), (u'/ did', u'I did'), (u'/ have ', u'I have '), (u'/ just ', u'I just '), (u'/ loved ', u'I loved '), (u'/ need', u'I need'), (u'|was11', u'I was 11'), (u'at Hrst', u'at first'), (u"B ullshiz'", u'Bullshit'), (u'big lunk', u'love you'), (u"can 't", u"can't"), (u"can' t ", u"can't "), (u"can 't ", u"can't "), (u'CHA TTERING', u'CHATTERING'), (u'come 0n', u'come on'), (u'Come 0n', u'Come on'), (u"couldn 't", u"couldn't"), (u"couldn' t ", u"couldn't "), (u"couldn 't ", u"couldn't "), (u"Destin y's", u"Destiny's"), (u"didn 't", u"didn't"), (u"didn' t ", u"didn't "), (u"didn 't ", u"didn't "), (u"Doesn '1'", u"Doesn't"), (u"doesn '1' ", u"doesn't "), (u"doesn '1\u2018 ", u"doesn't "), (u"doesn 't", u"doesn't"), (u"doesn'1' ", u"doesn't "), (u"doesn'1\u2018 ", u"doesn't "), (u"don '1' ", u"don't "), (u"don '1\u2018 ", u"don't "), (u"don '2' ", u"don't "), (u" aren '2'", u" aren't"), (u"aren '2' ", u"aren't "), (u"don '2\u2018 ", u"don't "), (u"don 't", u"don't"), (u"Don' t ", u"Don't "), (u"Don 't ", u"Don't "), (u"don'1' ", u"don't "), (u"don'1\u2018 ", u"don't "), (u"there '5 ", u"there's "), (u'E very', u'Every'), (u'get 0n', u'get on'), (u'go 0n', u'go on'), (u'Go 0n', u'Go on'), (u"H3993' birthday", u'Happy birthday'), (u"hadn 't", u"hadn't"), (u"he 's", u"he's"), (u"He 's", u"He's"), (u'He y', u'Hey'), (u'he)/', u'hey'), (u'He)/', u'Hey'), (u'HEA VY', u'HEAVY'), (u'Hold 0n', u'Hold on'), (u'I am. ls', u'I am. Is'), (u'I d0', u'I do'), (u"I 'm", u"I'm"), (u"I 'rn ", u"I'm "), (u"I 've", u"I've"), (u'I0 ve her', u'love her'), (u'I0 ve you', u'love you'), (u"I02'", u'lot'), (u"I'm sony", u"I'm sorry"), (u"isn' t ", u"isn't "), (u"isn 't ", u"isn't "), (u'K)/le', u'Kyle'), (u'L ook', u'Look'), (u'let me 90', u'let me go'), (u'Let me 90', u'Let me go'), (u"let's 90", u"let's go"), (u"Let's 90", u"Let's go"), (u'lfl had', u'If I had'), (u'lova you', u'love you'), (u'Lova you', u'love you'), (u'lovo you', u'love you'), (u'Lovo you', u'love you'), (u'ls anyone', u'Is anyone'), (u'ls he', u'Is he'), (u'-ls he', u'- Is he'), (u'ls it', u'Is it'), (u'-ls it', u'- Is it'), (u'ls she', u'Is she'), (u'-ls she', u'- Is she'), (u'ls that', u'Is that'), (u'-ls that', u'- Is that'), (u'ls this', u'Is this'), (u'-ls this', u'- Is this'), (u'Maze] tov', u'Mazel tov'), (u"N02' ", u'Not '), (u' of 0ur ', u' of our '), (u' ot mine ', u' of mine '), (u'PLA YING', u'PLAYING'), (u'REPEA TING ', u'REPEATING '), (u'Sa y', u'Say'), (u"she 's", u"she's"), (u"She 's", u"She's"), (u"shouldn 't", u"shouldn't"), (u'sta y', u'stay'), (u'Sta y', u'Stay'), (u'SWO rd', u'Sword'), (u'taka care', u'take care'), (u'Taka care', u'Take care'), (u'the Hrst', u'the first'), (u'toc late', u'too late'), (u'uf me', u'of me'), (u'uf our', u'of our'), (u'wa y', u'way'), (u'Wal-I\\/Iart', u'Wal-Mart'), (u"wasn '1' ", u"wasn't "), (u"Wasn '1' ", u"Wasn't "), (u"wasn '1\u2018 ", u"wasn't "), (u"Wasn '1\u2018 ", u"Wasn't "), (u"wasn 't", u"wasn't"), (u"Wasn 't", u"Wasn't"), (u"we 've", u"we've"), (u"We 've", u"We've"), (u"wem' off", u'went off'), (u"weren 't", u"weren't"), (u"who 's", u"who's"), (u"won 't", u"won't"), (u'would ha ve', u'would have '), (u"wouldn 't", u"wouldn't"), (u"Wouldn 't", u"Wouldn't"), (u'y()u', u'you'), (u'you QUYS', u'you guys'), (u"you' re ", u"you're "), (u"you 're ", u"you're "), (u"you 've", u"you've"), (u"You 've", u"You've"), (u"you' ve ", u"you've "), (u"you 've ", u"you've "), (u'aftera while', u'after a while'), (u'Aftera while', u'After a while'), (u'THUN DERCLAPS', u'THUNDERCLAPS'), (u'(BUZZI N G)', u'(BUZZING)'), (u'[BUZZI N G]', u'[BUZZING]'), (u'(G RU NTING', u'(GRUNTING'), (u'[G RU NTING', u'[GRUNTING'), (u'(G ROWLING', u'(GROWLING'), (u'[G ROWLING', u'[GROWLING'), (u' WAI LS)', u'WAILS)'), (u' WAI LS]', u'WAILS]'), (u'(scu RRYING)', u'(SCURRYING)'), (u'[scu RRYING]', u'[SCURRYING]'), (u'(GRUNT5)', u'(GRUNTS)'), (u'[GRUNT5]', u'[GRUNTS]'), (u'NARRA TOR:', u'NARRATOR:'), (u'(GROAN ING', u'(GROANING'), (u'[GROAN ING', u'[GROANING'), (u'GROAN ING)', u'GROANING)'), (u'GROAN ING]', u'GROANING]'), (u'(LAUGH ING', u'(LAUGHING'), (u'[LAUGH ING', u'[LAUGHING'), (u'LAUGLine truncated
+                          'pattern': None},
+         'PartialWordsAlways': {'data': OrderedDict([(u'\xa4', u'o'), (u'lVI', u'M'), (u'IVl', u'M'), (u'lVl', u'M'), (u'I\\/I', u'M'), (u'l\\/I', u'M'), (u'I\\/l', u'M'), (u'l\\/l', u'M'), (u'IVIa', u'Ma'), (u'IVIe', u'Me'), (u'IVIi', u'Mi'), (u'IVIo', u'Mo'), (u'IVIu', u'Mu'), (u'IVIy', u'My')]),
+                                'pattern': None},
+         'WholeLines': {'data': OrderedDict([(u'H ey.', u'Hey.'), (u'He)\u2019-', u'Hey.'), (u'N0.', u'No.'), (u'-N0.', u'-No.'), (u'Noll', u'No!!'), (u'(G ROANS)', u'(GROANS)'), (u'[G ROANS]', u'[GROANS]'), (u'(M EOWS)', u'(MEOWS)'), (u'[M EOWS]', u'[MEOWS]'), (u'Uaughs]', u'[laughs]'), (u'[chitte rs]', u'[chitters]'), (u'Hil\u2018 it!', u'Hit it!'), (u'<i>Hil\u2018 it!</i>', u'<i>Hit it!</i>'), (u'ISIGHS]', u'[SIGHS]')]),
+                        'pattern': None},
+         'WholeWords': {'data': OrderedDict([(u'$COff$', u'scoffs'), (u'$ergei', u'Sergei'), (u"$'llOp", u'Stop'), (u"/'//", u"I'll"), (u"/'/I", u"I'll"), (u'/ennifer', u'Jennifer'), (u'/got', u'I got'), (u'/have', u'I have'), (u'/hope', u'I hope'), (u'/just', u'I just'), (u'/love', u'I love'), (u"/'m", u"I'm"), (u'/mmerse', u'immerse'), (u'/nsu/ts', u'Insults'), (u'/ong', u'long'), (u'/ook', u'look'), (u"/t's", u"It's"), (u"/'ve", u"I've"), (u"\\/\\/e'd", u"We'd"), (u"\\/\\/e're", u"We're"), (u"\\/\\/e've", u"We've"), (u'\\/\\/hat', u'What'), (u"\\/\\/here'd", u"Where'd"), (u'\\/\\/hoo', u'Whoo'), (u'\\/\\/hy', u'Why'), (u"\\/\\/hy'd", u"Why'd"), (u"\\/\\le're", u"We're"), (u'\\/Ve', u'We'), (u"\\Ne're", u"We're"), (u"\\Nhat's", u"What's"), (u"\\Nhere's", u"Where's"), (u"|'mjust", u"I'm just"), (u'\xa4ff', u'off'), (u'\xa4Id', u'old'), (u'\xa4Ids', u'olds'), (u'\xa4n', u'on'), (u'\xa4ne', u'one'), (u'\xa4nly', u'only'), (u'\xa4pen', u'open'), (u'\xa4r', u'or'), (u'\xa4rder', u'order'), (u'\xa4ther', u'other'), (u'\xa4ur', u'our'), (u'\xa4ut', u'out'), (u'\xa4ver', u'over'), (u'\xa4wn', u'own'), (u"\u20acV\u20acI'y", u'every'), (u"0'clock", u"o'clock"), (u'0f', u'of'), (u'0fEngland', u'of England'), (u'0fft0', u'off to'), (u'0l/er', u'over'), (u'0n', u'on'), (u'0ne', u'one'), (u"0ne's", u"one's"), (u'0r', u'or'), (u'0rders', u'orders'), (u"0thers'", u"others'"), (u'0ut', u'out'), (u"0utlaw's", u"outlaw's"), (u"0utlaws'", u"outlaws'"), (u'0ver', u'over'), (u'13oos', u'1300s'), (u'18oos', u'1800s'), (u'195os', u'1950s'), (u"1et's", u"let's"), (u'1o', u'10'), (u'1oo', u'100'), (u'1ooth', u'100th'), (u'1oth', u'10th'), (u'2E_', u'2E.'), (u"2'IST", u'21ST'), (u"2'Ist_", u"2'1st."), (u'2o', u'20'), (u'2oth', u'20th'), (u'3o', u'30'), (u'3oth', u'30th'), (u'4o', u'40'), (u'4os', u'40s'), (u'4oth', u'40th'), (u'50rry', u'sorry'), (u'5o', u'50'), (u'5oth', u'50th'), (u'6o', u'60'), (u'6os', u'60s'), (u"'6os", u"'60s"), (u'6oth', u'60th'), (u'7o', u'70'), (u"'7os", u"'70s"), (u'7oth', u'70th'), (u'8o', u'80'), (u"'8os", u"'80s"), (u'8oth', u'80th'), (u'9/aim', u'alarm'), (u'9o', u'90'), (u'9oth', u'90th'), (u'9UnShQt', u'gunshot'), (u'a//', u'all'), (u'a/bum', u'album'), (u'a/so', u'also'), (u'A/ways', u'Always'), (u'abcut', u'about'), (u'aboutjoining', u'about joining'), (u'aboutposh', u'about posh'), (u'aboutus', u'about us'), (u'aboutyou', u'about you'), (u'accldent', u'accident'), (u'Acool', u'A cool'), (u'afier', u'after'), (u'affraid', u'afraid'), (u'Afriend', u'A friend'), (u'afterall', u'after all'), (u'afterthe', u'after the'), (u'afulcrum', u'a fulcrum'), (u'Afunny', u'A funny'), (u'aga/nst', u'against'), (u'ahsolutes', u'absolutes'), (u'AI_I_', u'ALL'), (u'AIien', u'Alien'), (u'AIex', u'Alex'), (u'AII', u'All'), (u'AIIan', u'Allan'), (u'AIIow', u'Allow'), (u'AIive', u'Alive'), (u"ain'tgotno", u"ain't got no"), (u"Ain'tgotno", u"Ain't got no"), (u'airstrike', u'air strike'), (u'AIVIBULANCE', u'AMBULANCE'), (u'ajob', u'a job'), (u'ajockey_', u'a jockey.'), (u'ajoke', u'a joke'), (u'Ajoke', u'A joke'), (u'ajoking', u'a joking'), (u'al/', u'all'), (u'al/en', u'alien'), (u'alittle', u'a little'), (u'allgasp', u'all gasp'), (u'alljustforshow', u'all just for show'), (u"aln't", u"ain't"), (u'alot', u'a lot'), (u'Alot', u'A lot'), (u'An5wer', u'Answer'), (u'Andit', u'And it'), (u"Andit's", u"And it's"), (u'andl', u'and I'), (u'andlaughs', u'and laughs'), (u'andleave', u'and leave'), (u'andthe', u'and the'), (u'andyou', u'and you'), (u'Andyou', u'And you'), (u'ANNOUNC/NG', u'ANNOUNCING'), (u'anotherstep', u'another step'), (u'ANSWER/NG', u'ANSWERING'), (u'answerwhat', u'answer what'), (u'antiquejoke', u'antique joke'), (u"anyhcdy's", u"anybody's"), (u'Anyidea', u'Any idea'), (u"anyone's_", u"anyone's."), (u'apejust', u'ape just'), (u'ARABlc', u'ARABIC'), (u"aren't_", u"aren't."), (u"arerl't", u"aren't"), (u"Arnsteln's", u"Arnstein's"), (u'atleast', u'at least'), (u'Atough', u'A tough'), (u'Awhole', u'A whole'), (u'awoman', u'a woman'), (u'Awoman', u'A woman'), (u'barelytalked', u'barely talked'), (u'bcat', u'boat'), (u'Bcllvla', u'Bolivia'), (u'bcmb', u'bomb'), (u'bcmbs', u'bombs'), (u'be//y', u'belly'), (u'becuase', u'because'), (u"Beep/'ng", u'Beeping'), (u'bejumpy', u'be jumpy'), (u'besldes', u'besides'), (u'bestfriend', u'best friend'), (u'bestguy', u'best guy'), (u'bestjob', u'best job'), (u'BIack', u'Black'), (u'BIess', u'Bless'), (u'bigos___', u'bigos...'), (u'BIame', u'Blame'), (u'BIind', u'Blind'), (u'BIood', u'Blood'), (u'BIue', u'Blue'), (u'BLOVVS', u'BLOWS'), (u'blowholel', u'blowhole!'), (u'blt', u'bit'), (u'Bo99', u'Bogg'), (u'bodiedyoung', u'bodied young'), (u'breakf\xe9wtl', u'breakfast!'), (u'bulldozlng', u'bulldozing'), (u'butjust', u'but just'), (u'butl', u'but I'), (u'Butl', u'But I'), (u'butljust', u'but I just'), (u"Butljustcan'tgetenough", u"But I just can't get enough"), (u"Butyou're", u"But you're"), (u'buythem', u'buy them'), (u'buyyou', u'buy you'), (u'byjust', u'by just'), (u'bythe', u'Line truncated
+                        'pattern': u"(um)\\b\\$COff\\$\\b|\\b\\$ergei\\b|\\b\\$\\'llOp\\b|\\b\\/\\'\\/\\/\\b|\\b\\/\\'\\/I\\b|\\b\\/ennifer\\b|\\b\\/got\\b|\\b\\/have\\b|\\b\\/hope\\b|\\b\\/just\\b|\\b\\/love\\b|\\b\\/\\'m\\b|\\b\\/mmerse\\b|\\b\\/nsu\\/ts\\b|\\b\\/ong\\b|\\b\\/ook\\b|\\b\\/t\\'s\\b|\\b\\/\\'ve\\b|\\b\\\\\\/\\\\\\/e\\'d\\b|\\b\\\\\\/\\\\\\/e\\'re\\b|\\b\\\\\\/\\\\\\/e\\'ve\\b|\\b\\\\\\/\\\\\\/hat\\b|\\b\\\\\\/\\\\\\/here\\'d\\b|\\b\\\\\\/\\\\\\/hoo\\b|\\b\\\\\\/\\\\\\/hy\\b|\\b\\\\\\/\\\\\\/hy\\'d\\b|\\b\\\\\\/\\\\le\\'re\\b|\\b\\\\\\/Ve\\b|\\b\\\\Ne\\'re\\b|\\b\\\\Nhat\\'s\\b|\\b\\\\Nhere\\'s\\b|\\b\\|\\'mjust\\b|\\b\\\xa4ff\\b|\\b\\\xa4Id\\b|\\b\\\xa4Ids\\b|\\b\\\xa4n\\b|\\b\\\xa4ne\\b|\\b\\\xa4nly\\b|\\b\\\xa4pen\\b|\\b\\\xa4r\\b|\\b\\\xa4rder\\b|\\b\\\xa4ther\\b|\\b\\\xa4ur\\b|\\b\\\xa4ut\\b|\\b\\\xa4ver\\b|\\b\\\xa4wn\\b|\\b\\\u20acV\\\u20acI\\'y\\b|\\b0\\'clock\\b|\\b0f\\b|\\b0fEngland\\b|\\b0fft0\\b|\\b0l\\/er\\b|\\b0n\\b|\\b0ne\\b|\\b0ne\\'s\\b|\\b0r\\b|\\b0rders\\b|\\b0thers\\'\\b|\\b0ut\\b|\\b0utlaw\\'s\\b|\\b0utlaws\\'\\b|\\b0ver\\b|\\b13oos\\b|\\b18oos\\b|\\b195os\\b|\\b1et\\'s\\b|\\b1o\\b|\\b1oo\\b|\\b1ooth\\b|\\b1oth\\b|\\b2E\\_\\b|\\b2\\'IST\\b|\\b2\\'Ist\\_\\b|\\b2o\\b|\\b2oth\\b|\\b3o\\b|\\b3oth\\b|\\b4o\\b|\\b4os\\b|\\b4oth\\b|\\b50rry\\b|\\b5o\\b|\\b5oth\\b|\\b6o\\b|\\b6os\\b|\\b\\'6os\\b|\\b6oth\\b|\\b7o\\b|\\b\\'7os\\b|\\b7oth\\b|\\b8o\\b|\\b\\'8os\\b|\\b8oth\\b|\\b9\\/aim\\b|\\b9o\\b|\\b9oth\\b|\\b9UnShQt\\b|\\ba\\/\\/\\b|\\ba\\/bum\\b|\\ba\\/so\\b|\\bA\\/ways\\b|\\babcut\\b|\\baboutjoining\\b|\\baboutposh\\b|\\baboutus\\b|\\baboutyou\\b|\\baccldent\\b|\\bAcool\\b|\\bafier\\b|\\baffraid\\b|\\bAfriend\\b|\\bafterall\\b|\\bafterthe\\b|\\bafulcrum\\b|\\bAfunny\\b|\\baga\\/nst\\b|\\bahsolutes\\b|\\bAI\\_I\\_\\b|\\bAIien\\b|\\bAIex\\b|\\bAII\\b|\\bAIIan\\b|\\bAIIow\\b|\\bAIive\\b|\\bain\\'tgotno\\b|\\bAin\\'tgotno\\b|\\bairstrike\\b|\\bAIVIBULANCE\\b|\\bajob\\b|\\bajockey\\_\\b|\\bajoke\\b|\\bAjoke\\b|\\bajoking\\b|\\bal\\/\\b|\\bal\\/en\\b|\\balittle\\b|\\ballgasp\\b|\\balljustforshow\\b|\\baln\\'t\\b|\\balot\\b|\\bAlot\\b|\\bAn5wer\\b|\\bAndit\\b|\\bAndit\\'s\\b|\\bandl\\b|\\bandlaughs\\b|\\bandleave\\b|\\bandthe\\b|\\bandyou\\b|\\bAndyou\\b|\\bANNOUNC\\/NG\\b|\\banotherstep\\b|\\bANSWER\\/NG\\b|\\banswerwhat\\b|\\bantiquejoke\\b|\\banyhcdy\\'s\\b|\\bAnyidea\\b|\\banyone\\'s\\_\\b|\\bapejust\\b|\\bARABlc\\b|\\baren\\'t\\_\\b|\\barerl\\'t\\b|\\bArnsteln\\'s\\b|\\batleast\\b|\\bAtough\\b|\\bAwhole\\b|\\bawoman\\b|\\bAwoman\\b|\\bbarelytalked\\b|\\bbcat\\b|\\bBcllvla\\b|\\bbcmb\\b|\\bbcmbs\\b|\\bbe\\/\\/y\\b|\\bbecuase\\b|\\bBeep\\/\\'ng\\b|\\bbejumpy\\b|\\bbesldes\\b|\\bbestfriend\\b|\\bbestguy\\b|\\bbestjob\\b|\\bBIack\\b|\\bBIess\\b|\\bbigos\\_\\_\\_\\b|\\bBIame\\b|\\bBIind\\b|\\bBIood\\b|\\bBIue\\b|\\bBLOVVS\\b|\\bblowholel\\b|\\bblt\\b|\\bBo99\\b|\\bbodiedyoung\\b|\\bbreakf\\\xe9wtl\\b|\\bbulldozlng\\b|\\bbutjust\\b|\\bbutl\\b|\\bButl\\b|\\bbutljust\\b|\\bButljustcan\\'tgetenough\\b|\\bButyou\\'re\\b|\\bbuythem\\b|\\bbuyyou\\b|\\bbyjust\\b|\\bbythe\\b|\\bC\\/latter\\/\\'\\/7g\\b|\\bca\\/\\/\\/ng\\b|\\bca\\/I\\b|\\bcall\\/ng\\b|\\bcallyou\\b|\\bcan\\*t\\b|\\bcan\\'i\\b|\\bcan\\'I\\b|\\bcanlgetyou\\b|\\bcannotchange\\b|\\bcannut\\b|\\bcan\\'T\\b|\\bcan\\'t\\_\\b|\\bcan\\'tjust\\b|\\bcan\\'tletgo\\b|\\bCar0l\\b|\\bCarhorn\\b|\\bcarrled\\b|\\bCcug\\b|\\bCcugs\\b|\\bCcver\\b|\\bcellularchange\\b|\\bcff\\b|\\bcfycu\\b|\\bcfycur\\b|\\bCh\\/rping\\b|\\bchaletgirl\\b|\\bchangejobs\\b|\\bCharliejust\\b|\\bChatter\\/\\'rtg\\b|\\bCHATTERWG\\b|\\bChequeredlove\\b|\\bcHIRPINcs\\b|\\bchjldishness\\b|\\bchlldrcn\\b|\\bchlldren\\b|\\bchocolatte\\b|\\bCho\\/r\\b|\\bcHucKl\\_Es\\b|\\bCIark\\b|\\bCIear\\b|\\bcircumcised\\_\\b|\\bckay\\b|\\bcl\\_osEs\\b|\\bCLATTERWG\\b|\\bcn\\b|\\bcne\\b|\\bcnes\\b|\\bCoincidenta\\/\\/y\\b|\\bCOm\\\u20ac\\b|\\bcomp\\/etey\\b|\\bcomplainingabout\\b|\\bcoms\\b|\\bcond\\/lion\\b|\\bconfdence\\b|\\bconhrmed\\b|\\bconnrm\\b|\\bConsecutivelyl\\b|\\bCOUGH5\\b|\\bCouGHING\\b|\\bcouIdn\\'t\\b|\\bcouldjust\\b|\\bcouldn\\'T\\b|\\bcouldn\\'tjust\\b|\\bCouldyou\\b|\\bcrappyjob\\b|\\bCRAsHING\\b|\\bcrder\\b|\\bCrowdcheers\\b|\\bCruoially\\b|\\bcther\\b|\\bcuuld\\b|\\bcver\\b|\\bd\\/\\'dn\\'t\\b|\\bd\\/squietude\\b|\\bD\\\xa4esn\\'t\\b|\\bd\\\xa4n\\'i\\b|\\bd\\\xa4n\\'t\\b|\\bd\\\xb09\\b|\\bd0\\b|\\bD0\\b|\\bD0asn\\'t\\b|\\bDad\\'\\/\\/\\b|\\bdairyjust\\b|\\bDar\\/\\/ng\\b|\\bdc\\b|\\bDcbby\\b|\\bdccsn\\'t\\b|\\bdcctcr\\b|\\bDces\\b|\\bdcgs\\b|\\bdcing\\b|\\bdcn\\'I\\b|\\bdcn\\'t\\b|\\bDcn\\'t\\b|\\bdcughnut\\b|\\bdeclslons\\b|\\bdeedhas\\b|\\bDehnitely\\b|\\bdesewes\\b|\\bdesperate\\/\\\xbby\\b|\\bD\\\ufb02nk\\b|\\bDIAl\\_lNcs\\b|\\bdidn\\'\\!\\b|\\bdidnt\\b|\\bdidn\\'T\\b|\\bdIdn\\'t\\b|\\bdidn\\'t\\_\\b|\\bdidrft\\b|\\bdidrl\\'t\\b|\\bdidyou\\b|\\bdivorcing\\_\\b|\\bdld\\b|\\bdldn\\'t\\b|\\bdlfflcull\\b|\\bdlg\\b|\\bdlsobeyed\\b|\\bdoasn\\'t\\b|\\bDoasn\\'t\\b|\\bdoctoh\\b|\\bDoctor\\_\\_\\_tell\\b|\\bdoesnt\\b|\\bdoesn\\'T\\b|\\bdoesri\\'t\\b|\\bdoesrt\\b|\\bDoesrt\\Line truncated
+ 'fin': {'BeginLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'EndLines': {'data': OrderedDict(),
+                      'pattern': None},
+         'PartialLines': {'data': OrderedDict(),
+                          'pattern': None},
+         'PartialWordsAlways': {'data': OrderedDict(),
+                                'pattern': None},
+         'WholeLines': {'data': OrderedDict([(u'Katsokaa pa.', u'Katsokaapa.'), (u'Mik!\r\n""e\u201c9ir\xe4\u0131', u'Mik!\r\n-Hengit\xe4!'), (u'Tu lta!', u'Tulta!'), (u'...0I1...', u'...on...'), (u'Ken g\xe4n nauh oja?', u'Keng\xe4nnauhoja?'), (u'k\xe3mmott\xe3V\xe3ll\xe3 mUiStoll\xe3.', u'kammottavana muistona.'), (u'H\xe4n n\xe4ki naisen menev\xe4n\r\nkellarikerroksen ksen asun to o ns\xe4.', u'H\xe4n n\xe4ki naisen menev\xe4n\r\nkellarikerroksen asuntoonsa.'), (u'Min\xe4 etsin k\xe4siini miehen, joka\r\non aurtanutiestradea ja minua:', u'Min\xe4 etsin k\xe4siini miehen, joka\r\non auttanut Lestradea ja minua:'), (u'Huomaa erityisesti\r\npunaisella merkitty "kitoris.', u'Huomaa erityisesti\r\npunaisella merkitty "klitoris".'), (u'Tulkaa, meill\xe4 on\r\nHa va\u0131ji-bileet suihkussa', u'Tulkaa, meill\xe4 on\r\nHavaiji-bileet suihkussa'), (u'Ta rkoitatko ett\xe4\r\nh\xe4net myrkytettiin?', u'Tarkoitatko ett\xe4\r\nh\xe4net myrkytettiin?'), (u'Odotta kaa soittoani\r\nIev\xe4hdyspaikalla.', u'Odottakaa soittoani\r\nlev\xe4hdyspaikalla.'), (u'Nyt kuuntelet, perska rva.', u'Nyt kuuntelet, perskarva.'), (u'Ta patko h\xe4net sitten?', u'Tapatko h\xe4net sitten?'), (u'Seuraa vissa va/oissa.', u'Seuraavissa valoissa.'), (u'A\u0131o\u0131t rapattaa minut.\r\n- En.', u'Aioit tapattaa minut.\r\n- En.'), (u'Todella vaku uttavaa.', u'Todella vakuuttavaa.'), (u'I-le ovat tuolla alhaalla', u'He ovat tuolla alhaalla'), (u'Nainen kuuluu minulle.\r\n- I-I\xe0ivy siit\xe4, ylikasvuinen hyttynen!', u'Nainen kuuluu minulle.\r\n- H\xe4ivy siit\xe4, ylikasvuinen hyttynen!'), (u'I-Ialuatte k\xe4ytt\xe4\xe4 pyh\xe0\xe0 kive\xe4\r\ndynastian aarteen l\xf6yt\xe4miseksi.', u'Haluatte k\xe4ytt\xe4\xe4 pyh\xe0\xe0 kive\xe4\r\ndynastian aarteen l\xf6yt\xe4miseksi.'), (u'Mit\xe4f?', u'Mit\xe4.. ?'), (u'Kuuluuko Hiru ko-klaanista mit\xe4\xe4n?\r\n- Ninjasoturit ovat edell\xe4mme.', u'Kuuluuko Hiruko-klaanista mit\xe4\xe4n?\r\n- Ninjasoturit ovat edell\xe4mme.'), (u'Anteeks\u0131} painoin kai... -', u'Anteeksi, painoin kai... -'), (u'ja Rea! Ho usew\xedves.', u'ja Real Housewives.'), (u'Et halu n n ut Julkkistansseihinkaan.', u'Et halunnut Julkkistansseihinkaan.'), (u'Laard i k\xe4si?', u'Laardik\xe4si?'), (u'Varo kaa!', u'Varokaa!'), (u'N\xe4ytt\xe4v\xe4t k\xf6 tyt\xf6t v\xe4h \xe4n\r\nh uorahtavi m m i lta?', u'N\xe4ytt\xe4v\xe4tk\xf6 tyt\xf6t v\xe4h\xe4n\r\nhuorahtavimmilta?'), (u'Stif... Ier.', u'Stif... ler.'), (u'J u mantsu kka! M it\xe4?', u'Jumantsukka! Mit\xe4?'), (u'Varasin Ch u m bawam ban,', u'Varasin Chumbawamban,'), (u'J u malavita!', u'Jumalavita!'), (u'S', u'Isi...'), (u'Haluan kertoa jotai n', u'Haluan kertoa jotain'), (u'I-Ialuatte', u'Haluatte')]),
+                        'pattern': None},
+         'WholeWords': {'data': OrderedDict([(u'kellojo', u'kello jo'), (u'onjo', u'on jo'), (u'senj\xe4lkeen', u'sen j\xe4lkeen'), (u'Ts\xe4u', u'Tsau'), (u'hydraulinenjousitus', u'hydraulinen jousitus'), (u'Kevyetja', u'Kevyet ja'), (u'Oleijo', u'Olet jo'), (u'viimeinenjuna', u'viimeinen juna'), (u'n\xe4ytt\xe4\xe4jotenkin', u'n\xe4ytt\xe4\xe4 jotenkin'), (u'onjoku', u'on joku'), (u'Iapsuuteen', u'Lapsuuteen'), (u'Ieikitjunill\xe4', u'leikit junilla'), (u'Sfiy', u'Stig'), (u'Iukulasejani', u'Lukulasejani'), (u'Ieikkiminen', u'Leikkiminen'), (u'K\xe4\xe4lik\xe4\xe4rylepizz\xe4', u'Kaalik\xe4\xe4rylepizza'), (u'sukl\xe4\xe4k\xe4stikkeen', u'suklaakastikkeen'), (u'p\xe4istetun', u'paistetun'), (u'b\xe4n\xe4\xe4nin', u'banaanin'), (u'pifi', u'piti'), (u'a\u0131keasfaan', u'oikeastaan'), (u'Stg', u'Stig'), (u'sikain\ufb02uenssatja', u'sikainfluenssat ja'), (u'Hienojuna', u'Hieno juna'), (u'kiinnostuajunista', u'kiinnostua junista'), (u'h\xe4y\u0131yvetu\u0131\xedn', u'h\xf6yryveturin'), (u'pienaismalli', u'pienoismalli'), (u'ena', u'eno'), (u'racfios\xe3hk\xe4ff\xe1j\xe3n\xe3', u'radios\xe4hk\xf6tt\xe4j\xe4n\xe4'), (u'Amenkan', u'Amerikan'), (u'laiva/la', u'laivalla'), (u'0/in', u'Olin'), (u'Junaha\u0131rasfus', u'Junaharrastus'), (u'enasfa', u'enosta'), (u'/V\u0131\xedn', u'Niin'), (u'Oliivarmasii', u'Olit varmasti'), (u'Internetist\xe4juuri', u'Internetist\xe4 juuri'), (u'hu\u0131janj\xe3nniff\xe5v\xe3\xe0l', u'hurjan j\xe4nnitt\xe4v\xe4\xe4.'), (u'Vaikkajunat', u'Vaikka junat'), (u'k\xe4\xe4ntyija', u'k\xe4\xe4ntyi ja'), (u'osaajotain', u'osaa jotain'), (u'Ieikill\xe4mme', u'leikill\xe4mme'), (u'Bellinjuna', u'Beltin juna'), (u't\xe4st\xe4johtolangasta', u't\xe4st\xe4 johtolangasta'), (u'tuntisijonkun', u'tuntisi jonkun'), (u'Olgajotain', u'Olga jotain'), (u'tulokseivuodesia', u'tulokset vuodesta'), (u'tuloksetvuodesta', u'tulokset vuodesta'), (u'Vainjalkapallo', u'Vain jalkapallo'), (u'lloveyou', u'I love you'), (u'po\ufb02\u2039aisi', u'potkaisi'), (u'minufulos', u'minut ulos'), (u'Ieiv\xe4npaahdinl', u'leiv\xe4npaahdin!'), (u'Ieiv\xe4npaahdinta', u'leiv\xe4npaahdinta'), (u'asuajossain', u'asua jossain'), (u'Koiratjuoksentelivat', u'Koirat juoksentelivat'), (u'tiet\xe4\xe4jotain', u'tiet\xe4\xe4 jotain'), (u'osaatjuna', u'osaat juna'), (u'va/as', u'valas'), (u'va/asperhe', u'valasperhe'), (u'koskeejotain', u'koskee jotain'), (u'I\xe4\xe4kkeens\xe4', u'l\xe4\xe4kkeens\xe4'), (u'Iaulavan', u'laulavan'), (u'Kuulitj\xe4\xe4n', u'Kuulit j\xe4\xe4n'), (u'Ievolle', u'levolle'), (u'Va/aa//e', u'Valaalle'), (u'va/aa//a', u'valaalla'), (u'I\xe4hdenkin', u'l\xe4hdenkin'), (u'Iakkiin', u'lakkiin'), (u'Iohik\xe4\xe4rmeell\xe4', u'lohik\xe4\xe4rmeell\xe4'), (u'Mik\u0131', u'Mik!'), (u'Iukossa', u'lukossa'), (u'Va/aan', u'Valaan'), (u's\xe4ve/m\xe4', u's\xe4velm\xe4'), (u'j\xe4//een', u'j\xe4lleen'), (u'P\xf6\xf6\u0131', u'P\xf6\xf6!'), (u'Hnjaa', u'Hiljaa'), (u'Iiftasin', u'liftasin'), (u'Jute//aan', u'Jutellaan'), (u'Iohduttamaan', u'lohduttamaan'), (u"L'lYP\xe4f\xe4\xe4n", u'Hyp\xe4t\xe4\xe4n'), (u'\xc4\u0131r\u0131\u0131', u'\xc4iti!'), (u'F\u0131\u0131\u0131p\u0131', u'Filip!'), (u'Yr\u203ar\u203a\xe4\xe4\u0131', u'Hypp\xe4\xe4!'), (u'Miki', u'Mik!'), (u'Heng\u0131r\xe4\u0131', u'Hengit\xe4!'), (u'Va/as', u'Valas'), (u'Ko///is', u'Koillis'), (u'Audieja', u'Audie ja'), (u'I/man', u'Ilman'), (u'/entotukia/ukse/ta', u'lentotukialukselta'), (u'/aivastosta', u'laivastosta'), (u'/entotukikohtaan', u'lentotukikohtaan'), (u'tu/ee', u'tulee'), (u'akse/iva//at', u'akselivallat'), (u'a\u0131\u0131ek\u0131rjo\u0131rerr\u0131\u0131n', u'allekirjoitettiin'), (u'Iitkua', u'litkua'), (u'0li', u'Oli'), (u'Iaskuvarjojoukkoihin', u'laskuvarjojoukkoihin'), (u'Iaskuvarjojoukoissa', u'laskuvarjojoukoissa'), (u'Tarvitsetjotain', u'Tarvitset jotain'), (u'Ieukaremmi', u'leukaremmi'), (u'Iainaatko', u'lainaatko'), (u'p\xe4\xe4llik\xf6lleja', u'p\xe4\xe4llik\xf6lle ja'), (u'I\xe4hell\xe4ni', u'l\xe4hell\xe4ni'), (u'Soti/aan', u'Sotilaan'), (u'va//attava', u'vallattava'), (u'viho//ise/ta', u'viholliselta'), (u'/iittoutuneet', u'liittoutuneet'), (u'voitjohtaa', u'voit johtaa'), (u'pohditkojotain', u'pohditko jotain'), (u'viho//ista', u'vihollista'), (u'sej\xe4isi', u'se j\xe4isi'), (u'Ient\xe4jill\xe4', u'lent\xe4jill\xe4'), (u'Soltutl', u'Soltut!'), (u'Iaulustamme', u'laulustamme'), (u'Iakkiani', u'lakkiani'), (u'Iiemeen', u'liemeen'), (u'kanadala\u0131\u0161ia', u'kanadalaisia'), (u'si//anp\xe4\xe4asema', u'sillanp\xe4\xe4asema'), (u'vahv\u0131\u0161tettiin', u'vahv\u0131stettiin'), (u'tu/ituki', u'tulituki'), (u'eliitt\u0131joukkoa', u'eliittijoukkoa'), (u'pa\u0131ka\u0131\u0131as\u0131\u0131', u'paikallasi!'), (u'Ieikkinyt', u'leikkinyt'), (u'Iujitettu', u'lujitettu'), (u'Ete/\xe4', u'Etel\xe4'), (u'viho//isranna//e', u'vihollisrannalle'), (u'sivusta/ta', u'sivustalta'), (u'/oppuisi', u'loppuisi'), (u'Iaskuvarjojoukkojen', u'laskuvarjojoukkojen'), (u'p\xe4\xe4tt\xe4nytj\xe4\xe4d\xe4', u'p\xe4\xe4tt\xe4nyt j\xe4\xe4d\xe4'), (u'IuutnanteillLine truncated
+                        'pattern': u"(um)\\bkellojo\\b|\\bonjo\\b|\\bsenj\\\xe4lkeen\\b|\\bTs\\\xe4u\\b|\\bhydraulinenjousitus\\b|\\bKevyetja\\b|\\bOleijo\\b|\\bviimeinenjuna\\b|\\bn\\\xe4ytt\\\xe4\\\xe4jotenkin\\b|\\bonjoku\\b|\\bIapsuuteen\\b|\\bIeikitjunill\\\xe4\\b|\\bSfiy\\b|\\bIukulasejani\\b|\\bIeikkiminen\\b|\\bK\\\xe4\\\xe4lik\\\xe4\\\xe4rylepizz\\\xe4\\b|\\bsukl\\\xe4\\\xe4k\\\xe4stikkeen\\b|\\bp\\\xe4istetun\\b|\\bb\\\xe4n\\\xe4\\\xe4nin\\b|\\bpifi\\b|\\ba\\\u0131keasfaan\\b|\\bStg\\b|\\bsikain\\\ufb02uenssatja\\b|\\bHienojuna\\b|\\bkiinnostuajunista\\b|\\bh\\\xe4y\\\u0131yvetu\\\u0131\\\xedn\\b|\\bpienaismalli\\b|\\bena\\b|\\bracfios\\\xe3hk\\\xe4ff\\\xe1j\\\xe3n\\\xe3\\b|\\bAmenkan\\b|\\blaiva\\/la\\b|\\b0\\/in\\b|\\bJunaha\\\u0131rasfus\\b|\\benasfa\\b|\\b\\/V\\\u0131\\\xedn\\b|\\bOliivarmasii\\b|\\bInternetist\\\xe4juuri\\b|\\bhu\\\u0131janj\\\xe3nniff\\\xe5v\\\xe3\\\xe0l\\b|\\bVaikkajunat\\b|\\bk\\\xe4\\\xe4ntyija\\b|\\bosaajotain\\b|\\bIeikill\\\xe4mme\\b|\\bBellinjuna\\b|\\bt\\\xe4st\\\xe4johtolangasta\\b|\\btuntisijonkun\\b|\\bOlgajotain\\b|\\btulokseivuodesia\\b|\\btuloksetvuodesta\\b|\\bVainjalkapallo\\b|\\blloveyou\\b|\\bpo\\\ufb02\\\u2039aisi\\b|\\bminufulos\\b|\\bIeiv\\\xe4npaahdinl\\b|\\bIeiv\\\xe4npaahdinta\\b|\\basuajossain\\b|\\bKoiratjuoksentelivat\\b|\\btiet\\\xe4\\\xe4jotain\\b|\\bosaatjuna\\b|\\bva\\/as\\b|\\bva\\/asperhe\\b|\\bkoskeejotain\\b|\\bI\\\xe4\\\xe4kkeens\\\xe4\\b|\\bIaulavan\\b|\\bKuulitj\\\xe4\\\xe4n\\b|\\bIevolle\\b|\\bVa\\/aa\\/\\/e\\b|\\bva\\/aa\\/\\/a\\b|\\bI\\\xe4hdenkin\\b|\\bIakkiin\\b|\\bIohik\\\xe4\\\xe4rmeell\\\xe4\\b|\\bMik\\\u0131\\b|\\bIukossa\\b|\\bVa\\/aan\\b|\\bs\\\xe4ve\\/m\\\xe4\\b|\\bj\\\xe4\\/\\/een\\b|\\bP\\\xf6\\\xf6\\\u0131\\b|\\bHnjaa\\b|\\bIiftasin\\b|\\bJute\\/\\/aan\\b|\\bIohduttamaan\\b|\\bL\\'lYP\\\xe4f\\\xe4\\\xe4n\\b|\\b\\\xc4\\\u0131r\\\u0131\\\u0131\\b|\\bF\\\u0131\\\u0131\\\u0131p\\\u0131\\b|\\bYr\\\u203ar\\\u203a\\\xe4\\\xe4\\\u0131\\b|\\bMiki\\b|\\bHeng\\\u0131r\\\xe4\\\u0131\\b|\\bVa\\/as\\b|\\bKo\\/\\/\\/is\\b|\\bAudieja\\b|\\bI\\/man\\b|\\b\\/entotukia\\/ukse\\/ta\\b|\\b\\/aivastosta\\b|\\b\\/entotukikohtaan\\b|\\btu\\/ee\\b|\\bakse\\/iva\\/\\/at\\b|\\ba\\\u0131\\\u0131ek\\\u0131rjo\\\u0131rerr\\\u0131\\\u0131n\\b|\\bIitkua\\b|\\b0li\\b|\\bIaskuvarjojoukkoihin\\b|\\bIaskuvarjojoukoissa\\b|\\bTarvitsetjotain\\b|\\bIeukaremmi\\b|\\bIainaatko\\b|\\bp\\\xe4\\\xe4llik\\\xf6lleja\\b|\\bI\\\xe4hell\\\xe4ni\\b|\\bSoti\\/aan\\b|\\bva\\/\\/attava\\b|\\bviho\\/\\/ise\\/ta\\b|\\b\\/iittoutuneet\\b|\\bvoitjohtaa\\b|\\bpohditkojotain\\b|\\bviho\\/\\/ista\\b|\\bsej\\\xe4isi\\b|\\bIent\\\xe4jill\\\xe4\\b|\\bSoltutl\\b|\\bIaulustamme\\b|\\bIakkiani\\b|\\bIiemeen\\b|\\bkanadala\\\u0131\\\u0161ia\\b|\\bsi\\/\\/anp\\\xe4\\\xe4asema\\b|\\bvahv\\\u0131\\\u0161tettiin\\b|\\btu\\/ituki\\b|\\beliitt\\\u0131joukkoa\\b|\\bpa\\\u0131ka\\\u0131\\\u0131as\\\u0131\\\u0131\\b|\\bIeikkinyt\\b|\\bIujitettu\\b|\\bEte\\/\\\xe4\\b|\\bviho\\/\\/isranna\\/\\/e\\b|\\bsivusta\\/ta\\b|\\b\\/oppuisi\\b|\\bIaskuvarjojoukkojen\\b|\\bp\\\xe4\\\xe4tt\\\xe4nytj\\\xe4\\\xe4d\\\xe4\\b|\\bIuutnanteille\\b|\\bI\\\xe4ht\\\xf6valmiiksi\\b|\\bIokoisat\\b|\\bKorjaus\\/ukemat\\b|\\btanke\\/\\/\\/e\\b|\\bky\\\u0131\\\u0131\\\xe4\\\u0131\\b|\\bIisti\\\xe4\\b|\\bKunniamita\\/in\\b|\\bVemon\\b|\\bfied\\\xe3n\\b|\\bEi\\\u0131is\\b|\\bInlem\\\xe4tion\\\xe4\\b|\\bholelfss\\\xe4\\b|\\bHan\\\u0131ey\\b|\\bpeikk\\\xe4\\b|\\bEhisl\\\xe3\\b|\\bY\\\ufb02\\\xe3ll\\\xe3v\\\xe3\\b|\\bmahdolisuus\\b|\\bloisteiaan\\b|\\baku\\b|\\bGaronilie\\b|\\bIikainen\\b|\\bvaiehtelijoita\\b|\\bpuutari\\\u0131aasi\\b|\\bGarunin\\b|\\boravaietti\\\xe4\\b|\\bpaijon\\b|\\bsutja\\b|\\bgospeikvartetissa\\b|\\bIauiajaksi\\b|\\bviihdylt\\\xe4j\\\xe4\\b|\\bhannaala\\b|\\bIaulajalta\\b|\\bsaatjotain\\b|\\bL\\\xf6yd\\\xe4njonkun\\b|\\bpy\\\xfcr\\\xedyi\\b|\\bCn\\\u0131dupia\\b|\\bIiikaa\\b|\\bke\\\u0131taa\\b|\\bThafs\\b|\\bKappaietta\\b|\\bsoijoka\\b|\\bLevytl\\\xe4\\b|\\bpaiaamista\\b|\\bkuonnunn\\b|\\bTytt\\\xfcyst\\\xe4v\\\xe4t\\b|\\bk\\\xe3p\\\xe3l\\\xf6ifiin\\b|\\bOiko\\b|\\btuipalo\\b|\\bhebofiavaa\\b|\\bongeina\\b|\\bongeimia\\b|\\bAjateikaa\\b|\\bku\\\u0131ki\\b|\\blhailen\\b|\\bL\\\xe4ht\\\xfcsi\\b|\\byhti\\\xf6ile\\b|\\beitoimi\\b|\\bvoij\\\xe4\\\xe4d\\\xe4\\b|\\bPomkat\\b|\\bkon\\\u0131ata\\b|\\bkonsentien\\b|\\bSep\\\xe3\\b|\\bhebolus\\b|\\bk\\\xfcyryss\\\xe4\\b|\\brepiat\\b|\\bkuikee\\b|\\bpon\\\u0131koita\\b|\\bnaisetja\\b|\\bten\\\u0131etuileeksi\\b|\\byl\\\xe4puolelia\\b|\\bkuullut180\\b|\\bIapsellista\\b|\\besiinnytja\\b|\\bIaulat\\b|\\bHenanjestas\\b|\\bteilie\\b|\\bIIe\\b|\\bt\\\xe4ilaisia\\b|\\bVaru\\b|\\bnimeli\\\xe4\\\xe4n\\b|\\banneijaan\\b|\\bpen\\\u0131staa\\b|\\bminkkitu\\\u0131kkia\\b|\\bminkkituiiiilla\\b|\\bTen\\\u0131etuioa\\b|\\bkuitalevy\\\xe4\\b|\\bei\\\xe4m\\\xe4ss\\\xe4\\b|\\bjutkaisua\\b|\\bFo\\\u0131t\\b|\\boien\\b|\\basioilie\\b|\\bEitarvitsekaan\\b|\\bp\\\xf6kenyin\\b|\\byll\\\xe4iiyy\\b|\\bkomeaita\\b|\\bChadie\\b|\\blaulaisilteko\\b|\\bIaittaa\\b|\\biytt\\\xe4renne\\b|\\bpikkutytLine truncated
+ 'fra': {'BeginLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'EndLines': {'data': OrderedDict(),
+                      'pattern': None},
+         'PartialLines': {'data': OrderedDict([(u" I'", u" l'"), (u" |'", u" l'")]),
+                          'pattern': None},
+         'PartialWordsAlways': {'data': OrderedDict(),
+                                'pattern': None},
+         'WholeLines': {'data': OrderedDict([(u'"D\'ac:c:ord."', u'"D\'accord."'), (u'\u201ci QU\xce gagne, qui perd,', u'ni qui gagne, qui perd,'), (u"L'ac:c:ent est mis \r\n \r\n sur son trajet jusqu'en Suisse.", u"L'accent est mis \r\n \r\n sur son trajet jusqu'en Suisse."), (u"C'est la plus gentille chose \r\n \r\n qu'Hitchc:oc:k m'ait jamais dite.", u"C'est la plus gentille chose \r\n \r\n qu'Hitchcock m'ait jamais dite."), (u"Tout le monde, en revanche, qualifie \r\n \r\n Goldfinger d'aventu re structur\xe9e,", u"Tout le monde, en revanche, qualifie \r\n \r\n Goldfinger d'aventure structur\xe9e,"), (u'et le film Shadow of a man \r\n \r\n a lanc\xe9 sa carri\xe8re au cin\xe9ma.', u'et le film <i>Shadow of a man</i> \r\n \r\n a lanc\xe9 sa carri\xe8re au cin\xe9ma.'), (u'En 1948, Young est pass\xe9 \xe0 la r\xe9alisation \r\n \r\n avec One night with you.', u'En 1948, Young est pass\xe9 \xe0 la r\xe9alisation \r\n \r\n avec <i>One night with you</i>.'), (u'Il a construit tous ces v\xe9hicules \r\n \r\n \xe0 C)c:ala, en Floride.', u'Il a construit tous ces v\xe9hicules \r\n \r\n \xe0 Ocala, en Floride.'), (u'Tokyo Pop et A Taxing Woman? Return.', u"Tokyo Pop et A Taxing Woman's Return."), (u'Peter H u nt.', u'Peter Hunt.'), (u'"C\'est bien mieux dans Peau. \r\n \r\n On peut s\ufb02\xe9clabousser, faire du bruit."', u'"C\'est bien mieux dans l\'eau. \r\n \r\n On peut s\'\xe9clabousser, faire du bruit."')]),
+                        'pattern': None},
+         'WholeWords': {'data': OrderedDict([(u'@immatriculation', u"d'immatriculation"), (u'acquer', u'acqu\xe9r'), (u'acteurjoue', u'acteur joue'), (u'aerien', u'a\xe9rien'), (u'agreable', u'agr\xe9able'), (u'aientjamais', u'aient jamais'), (u'AII', u'All'), (u'aitjamais', u'ait jamais'), (u'aitjus', u'ait jus'), (u'alle', u'all\xe9'), (u'alles', u'all\xe9s'), (u'appele', u'appel\xe9'), (u'apres', u'apr\xe8s'), (u'aujourdhui', u"aujourd'hui"), (u'aupres', u'aupr\xe8s'), (u'beaute', u'beaut\xe9'), (u'cabossee', u'caboss\xe9e'), (u"carj'", u"car j'"), (u"Carj'", u"Car j'"), (u'carla', u'car la'), (u'CEdipe', u'\u0152dipe'), (u'Cest', u"C'est"), (u"c'etaient", u"c'\xe9taient"), (u'C\xe9taient', u"C'\xe9taient"), (u"c'etait", u"c'\xe9tait"), (u"C'etait", u"C'\xe9tait"), (u'C\xe9tait', u"C'\xe9tait"), (u'choregraphiee', u'chor\xe9graphi\xe9e'), (u'cinema', u'cin\xe9ma'), (u"cl'AIcatraz", u"d'Alcatraz"), (u'cles', u'cl\xe9s'), (u'c\u0153urjoie', u'c\u0153ur-joie'), (u'completer', u'compl\xe9ter'), (u'costumiere', u'costumi\xe8re'), (u'cree', u'cr\xe9\xe9'), (u'daccord', u"d'accord"), (u"d'AIbert", u"d'Albert"), (u"d'AIdous", u"d'Aldous"), (u"d'AIec", u"d'Alec"), (u'danniversaire', u"d'anniversaire"), (u"d'Arra'bida", u"d'Arrabida"), (u"d'autod\xe9rision", u"d'auto-d\xe9rision"), (u'dautres', u"d'autres"), (u'debattait', u'd\xe9battait'), (u'decor', u'd\xe9cor'), (u'decorateurs', u'd\xe9corateurs'), (u'decors', u'd\xe9cors'), (u'defi', u'd\xe9fi'), (u'dej\xe0', u'd\xe9j\xe0'), (u'd\xe9j\xe0m', u'd\xe9j\xe0...'), (u'dejeunait', u'd\xe9jeunait'), (u'dengager', u"d'engager"), (u'd\xe9quipement', u"d'\xe9quipement"), (u'd\xe9rni\xe8r\xe9', u'derni\xe8re'), (u'Desole', u'D\xe9sol\xe9'), (u'dessayage', u"d'essayage"), (u'dessence', u"d'essence"), (u'd\xe9taient', u"c'\xe9taient"), (u'detail', u'd\xe9tail'), (u'dexcellents', u"d'excellents"), (u'dexp\xe9rience', u"d'exp\xe9rience"), (u'dexp\xe9riences', u"d'exp\xe9riences"), (u"d'h\xe9ro'l'ne", u"d'h\xe9ro\xefne"), (u"d'idees", u"d'id\xe9es"), (u"d'intensite", u"d'intensit\xe9"), (u'dontj', u'dont j'), (u'doublaitAlfo', u'doublait Alfo'), (u'DrNo', u'Dr No'), (u"e'", u'\xe9'), (u'ecrit', u'\xe9crit'), (u'elegant', u'\xe9l\xe9gant'), (u'Ell\xe9', u'Elle'), (u'\xe9n', u'en'), (u'equipe', u'\xe9quipe'), (u'erjus', u'er jus'), (u'estjamais', u'est jamais'), (u'\xe9t', u'et'), (u'etaient', u'\xe9taient'), (u'etait', u'\xe9tait'), (u'ete', u'\xe9t\xe9'), (u'etiez', u'\xe9tiez'), (u"etj'", u"et j'"), (u"Etj'", u"Et j'"), (u'etje', u'et je'), (u'Etje', u'Et je'), (u'EtsouvenL', u'Et souvent'), (u'eviter', u'\xe9viter'), (u'Fabsence', u"l'absence"), (u'fadapter', u"t'adapter"), (u'fadore', u"j'adore"), (u'F\xe2ge', u"l'\xe2ge"), (u'Fagent', u"l'agent"), (u'faiessay\xe9', u"j'ai essay\xe9"), (u'Failure', u"l'alllure"), (u'Fambiance', u"l'ambiance"), (u'Famener', u"l'amener"), (u'Fanniversaire', u"l'anniversaire"), (u'Fapparence', u"l'apparence"), (u'Fapres', u"l'apres"), (u'Fapr\xe8s', u"l'apr\xe8s"), (u'Farm\xe9e', u"l'arm\xe9e"), (u'Farri\xe8re', u"l'arri\xe8re"), (u'Farriv\xe9e', u"l'arriv\xe9e"), (u'Fascenseur', u"l'ascenseur"), (u'Fascension', u"l'ascension"), (u'Fassaut', u"l'assaut"), (u'Fassomme', u"l'assomme"), (u'Fatmosph\xe8re', u"l'atmosph\xe8re"), (u'Fattention', u"l'attention"), (u'Favalanche', u"l'avalanche"), (u'F\xe9clairage', u"l'\xe9clairage"), (u'F\xe9cran', u"l'\xe9cran"), (u'F\xe9motion', u"l'\xe9motion"), (u'Femplacement', u"l'emplacement"), (u'Fendroit', u"l'endroit"), (u'Fenseigne', u"l'enseigne"), (u'Fensemble', u"l'ensemble"), (u'Fentouraient', u"l'entouraient"), (u'Fentr\xe9e', u"l'entr\xe9e"), (u'F\xe9paisseur', u"l'\xe9paisseur"), (u'F\xe9poque', u"l'\xe9poque"), (u'F\xe9quipe', u'\xc9quipe'), (u'Fespace', u"l'espace"), (u'fesp\xe9rais', u"j'esp\xe9rais"), (u'Fesp\xe8re', u"l'esp\xe8re"), (u'Festh\xe9tique', u"l'esth\xe9tique"), (u'Fetranger', u"l'etranger"), (u'F\xe9vasion', u"l'\xe9vasion"), (u'F\xe9voque', u"l'\xe9voque"), (u'Fexp\xe9rience', u"l'exp\xe9rience"), (u'Fexplique', u"l'explique"), (u'Fexplosion', u"l'explosion"), (u'Fext\xe9rieur', u"l'ext\xe9rieur"), (u'Fhabituelle', u"l'habituelle"), (u'Fh\xe9licopt\xe8re', u"l'h\xe9licopt\xe8re"), (u'Fh\xe9liport', u"l'h\xe9liport"), (u'Fh\xe9listation', u"l'h\xe9listation"), (u'Fhonneur', u"l'honneur"), (u'Fhorloge', u"l'horloge"), (u'Fid\xe9e', u"l'id\xe9e"), (u'Fimage', u"l'image"), (u'Fimportance', u"l'importance"), (u'Fimpression', u"l'impression"), (u'Finfluence', u"l'influence"), (u'Finscription', u"l'inscription"), (u'Fint\xe9rieur', u"l'int\xe9rieur"), (u'Fintrigue', u"l'intrigue"), (u'Fobjectif', u"l'objectif"), (u'Foccasion', u"l'occasion"), (u'Fordre', u"l'ordre"), (u'Forigine', u"l'origine"), (u'fr\xeare', u'fr\xe8re'), (u'gaylns', u'gaijins'), (u'general', u'g\xe9n\xe9ral'), (u'hawa\xefennel', u'hawa\xefenne'), (u"hawa'l'en", u'hawa\xefen'), (u'Ia', u'la'), (u'I\xe0', u'l\xe0'), (u'Iaryngotomie', u'laryngotomie'), (u'idee', u'id\xe9e'), (u'idees', u'id\xe9es'), (u'Ie', u'le'), (u'IeLine truncated
+                        'pattern': u"(um)\\b\\@immatriculation\\b|\\bacquer\\b|\\bacteurjoue\\b|\\baerien\\b|\\bagreable\\b|\\baientjamais\\b|\\bAII\\b|\\baitjamais\\b|\\baitjus\\b|\\balle\\b|\\balles\\b|\\bappele\\b|\\bapres\\b|\\baujourdhui\\b|\\baupres\\b|\\bbeaute\\b|\\bcabossee\\b|\\bcarj\\'\\b|\\bCarj\\'\\b|\\bcarla\\b|\\bCEdipe\\b|\\bCest\\b|\\bc\\'etaient\\b|\\bC\\\xe9taient\\b|\\bc\\'etait\\b|\\bC\\'etait\\b|\\bC\\\xe9tait\\b|\\bchoregraphiee\\b|\\bcinema\\b|\\bcl\\'AIcatraz\\b|\\bcles\\b|\\bc\\\u0153urjoie\\b|\\bcompleter\\b|\\bcostumiere\\b|\\bcree\\b|\\bdaccord\\b|\\bd\\'AIbert\\b|\\bd\\'AIdous\\b|\\bd\\'AIec\\b|\\bdanniversaire\\b|\\bd\\'Arra\\'bida\\b|\\bd\\'autod\\\xe9rision\\b|\\bdautres\\b|\\bdebattait\\b|\\bdecor\\b|\\bdecorateurs\\b|\\bdecors\\b|\\bdefi\\b|\\bdej\\\xe0\\b|\\bd\\\xe9j\\\xe0m\\b|\\bdejeunait\\b|\\bdengager\\b|\\bd\\\xe9quipement\\b|\\bd\\\xe9rni\\\xe8r\\\xe9\\b|\\bDesole\\b|\\bdessayage\\b|\\bdessence\\b|\\bd\\\xe9taient\\b|\\bdetail\\b|\\bdexcellents\\b|\\bdexp\\\xe9rience\\b|\\bdexp\\\xe9riences\\b|\\bd\\'h\\\xe9ro\\'l\\'ne\\b|\\bd\\'idees\\b|\\bd\\'intensite\\b|\\bdontj\\b|\\bdoublaitAlfo\\b|\\bDrNo\\b|\\be\\'\\b|\\becrit\\b|\\belegant\\b|\\bEll\\\xe9\\b|\\b\\\xe9n\\b|\\bequipe\\b|\\berjus\\b|\\bestjamais\\b|\\b\\\xe9t\\b|\\betaient\\b|\\betait\\b|\\bete\\b|\\betiez\\b|\\betj\\'\\b|\\bEtj\\'\\b|\\betje\\b|\\bEtje\\b|\\bEtsouvenL\\b|\\beviter\\b|\\bFabsence\\b|\\bfadapter\\b|\\bfadore\\b|\\bF\\\xe2ge\\b|\\bFagent\\b|\\bfaiessay\\\xe9\\b|\\bFailure\\b|\\bFambiance\\b|\\bFamener\\b|\\bFanniversaire\\b|\\bFapparence\\b|\\bFapres\\b|\\bFapr\\\xe8s\\b|\\bFarm\\\xe9e\\b|\\bFarri\\\xe8re\\b|\\bFarriv\\\xe9e\\b|\\bFascenseur\\b|\\bFascension\\b|\\bFassaut\\b|\\bFassomme\\b|\\bFatmosph\\\xe8re\\b|\\bFattention\\b|\\bFavalanche\\b|\\bF\\\xe9clairage\\b|\\bF\\\xe9cran\\b|\\bF\\\xe9motion\\b|\\bFemplacement\\b|\\bFendroit\\b|\\bFenseigne\\b|\\bFensemble\\b|\\bFentouraient\\b|\\bFentr\\\xe9e\\b|\\bF\\\xe9paisseur\\b|\\bF\\\xe9poque\\b|\\bF\\\xe9quipe\\b|\\bFespace\\b|\\bfesp\\\xe9rais\\b|\\bFesp\\\xe8re\\b|\\bFesth\\\xe9tique\\b|\\bFetranger\\b|\\bF\\\xe9vasion\\b|\\bF\\\xe9voque\\b|\\bFexp\\\xe9rience\\b|\\bFexplique\\b|\\bFexplosion\\b|\\bFext\\\xe9rieur\\b|\\bFhabituelle\\b|\\bFh\\\xe9licopt\\\xe8re\\b|\\bFh\\\xe9liport\\b|\\bFh\\\xe9listation\\b|\\bFhonneur\\b|\\bFhorloge\\b|\\bFid\\\xe9e\\b|\\bFimage\\b|\\bFimportance\\b|\\bFimpression\\b|\\bFinfluence\\b|\\bFinscription\\b|\\bFint\\\xe9rieur\\b|\\bFintrigue\\b|\\bFobjectif\\b|\\bFoccasion\\b|\\bFordre\\b|\\bForigine\\b|\\bfr\\\xeare\\b|\\bgaylns\\b|\\bgeneral\\b|\\bhawa\\\xefennel\\b|\\bhawa\\'l\\'en\\b|\\bIa\\b|\\bI\\\xe0\\b|\\bIaryngotomie\\b|\\bidee\\b|\\bidees\\b|\\bIe\\b|\\bIes\\b|\\bIester\\b|\\bII\\b|\\bIimit\\b|\\bIIs\\b|\\bimmediatement\\b|\\binsufflee\\b|\\bintegrer\\b|\\binteressante\\b|\\bIogions\\b|\\bIorsqu\\b|\\bisee\\b|\\bIumiere\\b|\\bIynchage\\b|\\bJ\\'espere\\b|\\bJessaie\\b|\\bj\\'etais\\b|\\bJ\\'etais\\b|\\blat\\\xe9ral\\\xe9m\\\xe9nt\\b|\\blci\\b|\\bLci\\b|\\bl\\\xe9\\-\\b|\\blepidopteres\\b|\\blitteraire\\b|\\bll\\b|\\bLl\\b|\\blls\\b|\\bLls\\b|\\bmaintenanu\\b|\\bmaniere\\b|\\bmariee\\b|\\bMayer\\/ing\\b|\\bmeilleurjour\\b|\\bmelange\\b|\\bn\\'avai\\\xe9nt\\b|\\bn\\'etait\\b|\\boitjamais\\b|\\boitjus\\b|\\bontete\\b|\\boperateur\\b|\\bouv\\\xe9rt\\\xe9\\b|\\bP\\\xe9preuve\\b|\\bpere\\b|\\bplateforme\\b|\\bpourjouer\\b|\\bprecipice\\b|\\bpreferes\\b|\\bpremierjour\\b|\\bpresenter\\b|\\bprevu\\b|\\bprevue\\b|\\bpropriete\\b|\\bprot\\\xe8geraient\\b|\\bqu\\\xe9\\b|\\bqwangoiss\\\xe9\\b|\\brealisateur\\b|\\breception\\b|\\bre\\\xe9valu\\b|\\brepute\\b|\\breussi\\b|\\bs\\'arr\\\xe9tait\\b|\\bs\\'ave\\'rer\\b|\\bscenario\\b|\\bscene\\b|\\bscenes\\b|\\bseances\\b|\\bsequence\\b|\\bs\\\ufb02\\\xe9crasa\\b|\\bspeciale\\b|\\bSupen\\b|\\btorturee\\b|\\bUadmirable\\b|\\bUensemblier\\b|\\bUexplosion\\b|\\bUouvre\\b|\\bVaise\\b|\\bvecu\\b|\\bvehicules\\b|\\b\\\u0178appr\\\xe9ciais\\b|\\b\\\u0178esp\\\xe8re\\b|\\b\\\xff\\\xe9trangle\\b"}},
+ 'hrv': {'BeginLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'EndLines': {'data': OrderedDict(),
+                      'pattern': None},
+         'PartialLines': {'data': OrderedDict([(u'Ako ej', u'Ako je'), (u'ako ej', u'ako je'), (u'bez svesti', u'bez svijesti'), (u'Bi\u0107u uz', u'Bit \u0107u uz'), (u'bi ja', u'bih ja'), (u'bi la', u'bila'), (u'biti uredu', u'biti u redu'), (u'bi da bude', u'bi biti'), (u'Bi ste', u'Biste'), (u'bilo ko', u'bilo tko'), (u'\u0107e da do\u0111e', u'\u0107e do\u0107i'), (u'da li \u0107u', u'ho\u0107u li'), (u'dali \u0107u', u'ho\u0107u li'), (u'Da li je', u'Je li'), (u'da li je', u'je li'), (u'Da li ste', u'Jeste li'), (u'Da li si', u'Jesi li'), (u'dali si', u'jesi li'), (u'da li \u0107e', u'ho\u0107e li'), (u'dali \u0107e', u'ho\u0107e li'), (u'do srede', u'do srijede'), (u'Dobro ve\u010de', u'Dobra ve\u010der'), (u'Dobro ve\u010der', u'Dobra ve\u010der'), (u'Dobar ve\u010der', u'Dobra ve\u010der'), (u'gdje ide\u0161', u'kamo ide\u0161'), (u'Gdje ide\u0161', u'Kamo ide\u0161'), (u'Gdje sada', u'Kamo sada'), (u'gle ko', u'gle tko'), (u'ho\u0107u da budem', u'\u017eelim biti'), (u'Ho\u0107u da budem', u'\u017delim biti'), (u'ho\u0107u da ka\u017eem', u'\u017eelim re\u0107i'), (u'ho\u0107e\u0161 da ka\u017ee\u0161', u'\u017eeli\u0161 re\u0107i'), (u'ho\u0107e da ka\u017ee', u'\u017eeli re\u0107i'), (u'Izvinite me', u'Ispri\u010dajte me'), (u'Izvinite nas', u'Ispri\u010dajte nas'), (u'izvinite nas', u'ispri\u010dajte nas'), (u'ja bi', u'ja bih'), (u'Ja bi', u'Ja bih'), (u'Jel sam ti', u'Jesam li ti'), (u'Jeli se', u'Je li se'), (u'Jeli sve', u'Je li sve'), (u'Jeli ti', u'Je li ti'), (u'ko zna', u'tko zna'), (u'moglo da bude', u'moglo biti'), (u'moje sau\u010de\u0161\u0107e', u'moja su\u0107ut'), (u'mora da bude', u'mora biti'), (u'Moram da idem', u'Moram i\u0107i'), (u'Moramo da idemo', u'Moramo i\u0107i'), (u'mora\u0161 da ide\u0161', u'mora\u0161 i\u0107i'), (u'na ve\u010der', u'nave\u010der'), (u'Na ve\u010der', u'Nave\u010der'), (u'neko ko', u'netko tko'), (u'nedjelju dana', u'tjedan dana'), (u'Ne mogu da verujem', u'Ne mogu vjerovati'), (u'od kako', u'otkako'), (u'pravo u o\u010di', u'ravno u o\u010di'), (u'Si dobro', u'Jesi li dobro'), (u'Svako ko', u'Svatko tko'), (u'Svo vreme', u'Sve vrijeme'), (u'Svo vrijeme', u'Sve vrijeme'), (u'smeo da', u'smio'), (u'smeli da', u'smjeli'), (u'\u0160to ej', u'\u0160to je'), (u'\u0161to ej', u'\u0161to je'), (u'to ej', u'to je'), (u'To ej', u'To je'), (u'tamo natrag', u'tamo iza'), (u'tamo je natrag', u'tamo je iza'), (u'Tamo je natrag', u'Tamo je iza'), (u'zato sto', u'zato \u0161to'), (u'znati ko', u'znati tko'), (u'\u017eelim da budem', u'\u017eelim biti'), (u'\u017delim da budem', u'\u017delim biti'), (u'\u017eeli\u0161 da bude\u0161', u'\u017eeli\u0161 biti'), (u'\u017eelim da odem', u'\u017eelim oti\u0107i'), (u'\u017eeli\u0161 da ode\u0161', u'\u017eeli\u0161 oti\u0107i'), (u'\u017eelim da znam', u'\u017eelim znati'), (u'\u017eeli\u0161 da zna\u0161', u'\u017eeli\u0161 znati')]),
+                          'pattern': None},
+         'PartialWordsAlways': {'data': OrderedDict(),
+                                'pattern': None},
+         'WholeLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'WholeWords': {'data': OrderedDict([(u'advokati', u'odvjetnici'), (u'Advokati', u'Odvjetnici'), (u'advokatima', u'odvjetnicima'), (u'Advokatima', u'Odvjetnicima'), (u'amin', u'amen'), (u'Amin', u'Amen'), (u'ba\u0161ta', u'vrt'), (u'Ba\u0161ta', u'Vrt'), (u'ba\u0161te', u'vrtovi'), (u'Ba\u0161te', u'Vrtovi'), (u'ba\u0161ti', u'vrtu'), (u'ba\u0161tu', u'vrt'), (u'Ba\u0161tu', u'Vrt'), (u'ba\u0161tom', u'vrtom'), (u'bedan', u'bijedan'), (u'bede', u'bijede'), (u'bedi', u'bijedi'), (u'bejah', u'bijah'), (u'bele\u0161ci', u'bilje\u0161ci'), (u'bimso', u'bismo'), (u'bioskop', u'kino'), (u'bioskopi', u'kina'), (u'bitci', u'bitki'), (u'blije\u0111i', u'blje\u0111i'), (u'beg', u'bijeg'), (u'begu', u'bijegu'), (u'beja\u0161e', u'bija\u0161e'), (u'bekstvo', u'bijeg'), (u'bekstvu', u'bijegu'), (u'begstvu', u'bijegu'), (u'bes', u'bijes'), (u'besa', u'bijesa'), (u'besan', u'bijesan'), (u'besom', u'bijesom'), (u'besu', u'bijesu'), (u'be\u0161e', u'bje\u0161e'), (u'bioje', u'bio je'), (u'bi smo', u'bismo'), (u'bi ste', u'biste'), (u'bled', u'blijed'), (u'blede', u'blijede'), (u'boksuje\u0161', u'boksa\u0161'), (u'bolesan', u'bolestan'), (u'braon', u'sme\u0111a'), (u'bregu', u'brijegu'), (u'bti', u'biti'), (u'ceo', u'cijeli'), (u'Ceo', u'Cijeli'), (u'cepa', u'cijepa'), (u'cev', u'cijev'), (u'cevi', u'cijevi'), (u'cev\u010dica', u'cjev\u010dica'), (u'cev\u010dicu', u'cjev\u010dicu'), (u'cvetu', u'cvijetu'), (u'cvjetu', u'cvijetu'), (u'cvjetom', u'cvijetom'), (u'\u010dar\u0161av', u'plahta'), (u'\u010dar\u0161ave', u'plahte'), (u'\u010dar\u0161avima', u'plahtama'), (u'\u010das', u'sat'), (u'\u010detvoro', u'\u010detvero'), (u'\u010cita\u0107u', u'\u010citat \u0107u'), (u'\u010derku', u'k\u0107er'), (u'\u010cerku', u'K\u0107er'), (u'\u0107erku', u'k\u0107er'), (u'\u0106erku', u'K\u0107er'), (u'\u0107erkama', u'k\u0107erima'), (u'\u010derkama', u'k\u0107erima'), (u'\u0107erkom', u'k\u0107eri'), (u'\u010derkom', u'k\u0107eri'), (u'k\u0107erkom', u'k\u0107eri'), (u'k\u010derkom', u'k\u0107eri'), (u'\u0107ebe', u'deku'), (u'\u0107ebetu', u'deki'), (u'\u010dk', u'\u010dak'), (u'\u0107\u0161', u'\u0107e\u0161'), (u'\u0107ale', u'tata'), (u'\u0107aletom', u'ocem'), (u'\u0107aleta', u'oca'), (u'\u0107aletu', u'ocu'), (u'\u0107orsokak', u'slijepa ulica'), (u'\u0107orsokaku', u'slijepoj ulici'), (u'\u0107o\u0161ak', u'ugao'), (u'\u0107o\u0161ku', u'uglu'), (u'\u0107erka', u'k\u0107i'), (u'\u0106erka', u'K\u0107i'), (u'\u0107mo', u'\u0107emo'), (u'\u0107te', u'\u0107ete'), (u'\u010du\u0107e', u'\u010dut \u0107e'), (u'\u010du\u0107emo', u'\u010dut \u0107emo'), (u'\u010cu\u0107ete', u'\u010cut \u0107ete'), (u'\u010cu\u0107e', u'\u010cut \u0107e'), (u'\u0107utao', u'\u0161utio'), (u'\u0106utao', u'\u0160utio'), (u'daga', u'da ga'), (u'damas', u'danas'), (u'deda', u'djed'), (u'Deda', u'Djed'), (u'dede', u'djeda'), (u'dedi', u'djedu'), (u'dedom', u'djedom'), (u'defini\u0161e', u'definira'), (u'dele\u0107i', u'dijele\u0107i'), (u'deo', u'dio'), (u'Deo', u'Dio'), (u'de\u0161ava', u'doga\u0111a'), (u'dete', u'dijete'), (u'djete', u'dijete'), (u'Dete', u'Dijete'), (u'Djete', u'Dijete'), (u'detektuje', u'detektira'), (u'dodju', u'do\u0111u'), (u'dole', u'dolje'), (u'Dole', u'Dolje'), (u'Donije\u0107u', u'Donijet \u0107u'), (u'do ovde', u'dovde'), (u'dospeo', u'dospio'), (u'dospeju', u'dospiju'), (u'do\u0111avola', u'dovraga'), (u'Do\u0111avola', u'Dovraga'), (u'drug', u'prijatelj'), (u'drugde', u'drugdje'), (u'duuga', u'd\xfaga'), (u'dve', u'dvije'), (u'Dve', u'Dvije'), (u'\u0111avo', u'vrag'), (u'\u0111avola', u'vraga'), (u'\u0111emper', u'd\u017eemper'), (u'd\u017eanki', u'ovisnik'), (u'\u0111ubre', u'sme\u0107e'), (u'eksperti', u'stru\u010dnjaci'), (u'Eksperti', u'Stru\u010dnjaci'), (u'ekspertima', u'stru\u010dnjacima'), (u'en', u'ne'), (u'emitovao', u'emitirao'), (u'emituje', u'emitira'), (u'emitujem', u'emitiram'), (u'emituju', u'emitiraju'), (u'familijama', u'obiteljima'), (u'familiji', u'obitelji'), (u'flertuje', u'o\u010dijuka'), (u'foka', u'tuljan'), (u'foku', u'tuljana'), (u'foke', u'tuljani'), (u'fokama', u'tuljanima'), (u'gde', u'gdje'), (u'Gde', u'Gdje'), (u'Gre\u010den', u'Gretchen'), (u'gre\u0161e', u'grije\u0161e'), (u'gre\u0161ki', u'gre\u0161ci'), (u'Historija', u'Povijest'), (u'Historiju', u'Povijest'), (u'Historije', u'Povijesti'), (u'Historiji', u'Povijesti'), (u'Istorija', u'Povijest'), (u'Istoriju', u'Povijest'), (u'Istorije', u'Povijesti'), (u'Istoriji', u'Povijesti'), (u'historije', u'povijesti'), (u'historiji', u'povijesti'), (u'istorije', u'povijesti'), (u'istoriji', u'povijesti'), (u'ho\u010du', u'ho\u0107u'), (u'Ho\u010du', u'Ho\u0107u'), (u'hte\u0107e', u'htjet \u0107e'), (u'htedoh', u'htjedoh'), (u'i\u010di', u'i\u0107i'), (u'I\u010di', u'I\u0107i'), (u'iko', u'itko'), (u'ignori\u0161i', u'ignoriraj'), (u'Ignori\u0161i', u'Ignoriraj'), (u'ignori\u0161ite', u'ignorirajte'), (u'Ignori\u0161ite', u'Ignorirajte'), (u'ignori\u0161u', u'ignoriraju'), (u'Interesuje', u'Zanima'), (u'Interesuju', u'Zanimaju'), (u'intLine truncated
+                        'pattern': u'(um)\\badvokati\\b|\\bAdvokati\\b|\\badvokatima\\b|\\bAdvokatima\\b|\\bamin\\b|\\bAmin\\b|\\bba\\\u0161ta\\b|\\bBa\\\u0161ta\\b|\\bba\\\u0161te\\b|\\bBa\\\u0161te\\b|\\bba\\\u0161ti\\b|\\bba\\\u0161tu\\b|\\bBa\\\u0161tu\\b|\\bba\\\u0161tom\\b|\\bbedan\\b|\\bbede\\b|\\bbedi\\b|\\bbejah\\b|\\bbele\\\u0161ci\\b|\\bbimso\\b|\\bbioskop\\b|\\bbioskopi\\b|\\bbitci\\b|\\bblije\\\u0111i\\b|\\bbeg\\b|\\bbegu\\b|\\bbeja\\\u0161e\\b|\\bbekstvo\\b|\\bbekstvu\\b|\\bbegstvu\\b|\\bbes\\b|\\bbesa\\b|\\bbesan\\b|\\bbesom\\b|\\bbesu\\b|\\bbe\\\u0161e\\b|\\bbioje\\b|\\bbi\\ smo\\b|\\bbi\\ ste\\b|\\bbled\\b|\\bblede\\b|\\bboksuje\\\u0161\\b|\\bbolesan\\b|\\bbraon\\b|\\bbregu\\b|\\bbti\\b|\\bceo\\b|\\bCeo\\b|\\bcepa\\b|\\bcev\\b|\\bcevi\\b|\\bcev\\\u010dica\\b|\\bcev\\\u010dicu\\b|\\bcvetu\\b|\\bcvjetu\\b|\\bcvjetom\\b|\\b\\\u010dar\\\u0161av\\b|\\b\\\u010dar\\\u0161ave\\b|\\b\\\u010dar\\\u0161avima\\b|\\b\\\u010das\\b|\\b\\\u010detvoro\\b|\\b\\\u010cita\\\u0107u\\b|\\b\\\u010derku\\b|\\b\\\u010cerku\\b|\\b\\\u0107erku\\b|\\b\\\u0106erku\\b|\\b\\\u0107erkama\\b|\\b\\\u010derkama\\b|\\b\\\u0107erkom\\b|\\b\\\u010derkom\\b|\\bk\\\u0107erkom\\b|\\bk\\\u010derkom\\b|\\b\\\u0107ebe\\b|\\b\\\u0107ebetu\\b|\\b\\\u010dk\\b|\\b\\\u0107\\\u0161\\b|\\b\\\u0107ale\\b|\\b\\\u0107aletom\\b|\\b\\\u0107aleta\\b|\\b\\\u0107aletu\\b|\\b\\\u0107orsokak\\b|\\b\\\u0107orsokaku\\b|\\b\\\u0107o\\\u0161ak\\b|\\b\\\u0107o\\\u0161ku\\b|\\b\\\u0107erka\\b|\\b\\\u0106erka\\b|\\b\\\u0107mo\\b|\\b\\\u0107te\\b|\\b\\\u010du\\\u0107e\\b|\\b\\\u010du\\\u0107emo\\b|\\b\\\u010cu\\\u0107ete\\b|\\b\\\u010cu\\\u0107e\\b|\\b\\\u0107utao\\b|\\b\\\u0106utao\\b|\\bdaga\\b|\\bdamas\\b|\\bdeda\\b|\\bDeda\\b|\\bdede\\b|\\bdedi\\b|\\bdedom\\b|\\bdefini\\\u0161e\\b|\\bdele\\\u0107i\\b|\\bdeo\\b|\\bDeo\\b|\\bde\\\u0161ava\\b|\\bdete\\b|\\bdjete\\b|\\bDete\\b|\\bDjete\\b|\\bdetektuje\\b|\\bdodju\\b|\\bdole\\b|\\bDole\\b|\\bDonije\\\u0107u\\b|\\bdo\\ ovde\\b|\\bdospeo\\b|\\bdospeju\\b|\\bdo\\\u0111avola\\b|\\bDo\\\u0111avola\\b|\\bdrug\\b|\\bdrugde\\b|\\bduuga\\b|\\bdve\\b|\\bDve\\b|\\b\\\u0111avo\\b|\\b\\\u0111avola\\b|\\b\\\u0111emper\\b|\\bd\\\u017eanki\\b|\\b\\\u0111ubre\\b|\\beksperti\\b|\\bEksperti\\b|\\bekspertima\\b|\\ben\\b|\\bemitovao\\b|\\bemituje\\b|\\bemitujem\\b|\\bemituju\\b|\\bfamilijama\\b|\\bfamiliji\\b|\\bflertuje\\b|\\bfoka\\b|\\bfoku\\b|\\bfoke\\b|\\bfokama\\b|\\bgde\\b|\\bGde\\b|\\bGre\\\u010den\\b|\\bgre\\\u0161e\\b|\\bgre\\\u0161ki\\b|\\bHistorija\\b|\\bHistoriju\\b|\\bHistorije\\b|\\bHistoriji\\b|\\bIstorija\\b|\\bIstoriju\\b|\\bIstorije\\b|\\bIstoriji\\b|\\bhistorije\\b|\\bhistoriji\\b|\\bistorije\\b|\\bistoriji\\b|\\bho\\\u010du\\b|\\bHo\\\u010du\\b|\\bhte\\\u0107e\\b|\\bhtedoh\\b|\\bi\\\u010di\\b|\\bI\\\u010di\\b|\\biko\\b|\\bignori\\\u0161i\\b|\\bIgnori\\\u0161i\\b|\\bignori\\\u0161ite\\b|\\bIgnori\\\u0161ite\\b|\\bignori\\\u0161u\\b|\\bInteresuje\\b|\\bInteresuju\\b|\\binteresuju\\b|\\bisekao\\b|\\bisterao\\b|\\bistera\\\u0161\\b|\\bIstera\\\u0161\\b|\\bisuvi\\\u0161e\\b|\\bivica\\b|\\bivice\\b|\\bivici\\b|\\bivicu\\b|\\bIzolujete\\b|\\bIzolujte\\b|\\bizgladneo\\b|\\bizmeriti\\b|\\bizmerio\\b|\\bizneti\\b|\\bizvinim\\b|\\bizvini\\\u0161\\b|\\bizviniti\\b|\\bjedamput\\b|\\bjelda\\b|\\bju\\\u010de\\b|\\bJu\\\u010de\\b|\\bkakava\\b|\\bkampuje\\b|\\bkampujemo\\b|\\bkampuju\\b|\\bkancelariji\\b|\\bkancelarijom\\b|\\bkancera\\b|\\bkarmin\\b|\\bkarminom\\b|\\bk\\\u0107erci\\b|\\bk\\\u0107erka\\b|\\bk\\\u010derka\\b|\\bK\\\u0107erka\\b|\\bK\\\u010derka\\b|\\bk\\\u0107erkama\\b|\\bker\\b|\\bKer\\b|\\bkerova\\b|\\bkidnapovan\\b|\\bKidnapovan\\b|\\bkiji\\b|\\bkijim\\b|\\bklasifikuju\\b|\\bKo\\b|\\bkoa\\b|\\bkoaj\\b|\\bkola\\b|\\bkolima\\b|\\bkomandni\\b|\\bkombinuju\\b|\\bkompanija\\b|\\bkom\\\u0161ija\\b|\\bkom\\\u0161iji\\b|\\bkom\\\u0161iluk\\b|\\bkom\\\u0161iluku\\b|\\bkom\\\u0161ije\\b|\\bkom\\\u0161ijama\\b|\\bkontroli\\\u0161u\\b|\\bKontroli\\\u0161u\\b|\\bKontroli\\\u0161i\\b|\\bkontroli\\\u0161i\\b|\\bKontroli\\\u0161ite\\b|\\bkontroli\\\u0161ite\\b|\\bkorpu\\b|\\bkritikuju\\b|\\bkrsta\\b|\\bkrstu\\b|\\bkrsti\\\u0107a\\b|\\bkrsti\\\u0107u\\b|\\bKrsta\\b|\\bKrstu\\b|\\bKrsti\\\u0107a\\b|\\bKrsti\\\u0107u\\b|\\bkrstom\\b|\\bkrstove\\b|\\bKrstove\\b|\\bkrstovima\\b|\\bKrstovima\\b|\\bkri\\\u017eom\\b|\\bkupatilo\\b|\\bkupatilu\\b|\\bkvalifikuju\\b|\\bla\\\u017eeju\\b|\\bla\\\u017eov\\b|\\bla\\\u017eovi\\b|\\bla\\\u017eovu\\b|\\bla\\\u017eovima\\b|\\blebdeo\\b|\\ble\\\u010di\\b|\\blje\\\u010di\\b|\\bLe\\\u010di\\b|\\bLje\\\u010di\\b|\\bLejn\\b|\\bLenja\\b|\\blenji\\b|\\blenja\\b|\\ble\\\u0161nik\\b|\\bleto\\b|\\bletiti\\b|\\bleve\\b|\\blo\\\u017ei\\b|\\bmajca\\b|\\bmajcu\\b|\\bmajcom\\b|\\bMalopre\\b|\\bmalopre\\b|\\bmaloprije\\b|\\bmanifestuje\\b|\\bmehur\\b|\\bmenom\\b|\\bmera\\\u010d\\b|\\bmeri\\b|\\bmere\\b|\\bmerdevine\\b|\\bmerdevinama\\b|\\bme\\\u0161ane\\b|\\bme\\\u0161avina\\b|\\bme\\\u0161avine\\b|\\bme\\\u0161avinu\\b|\\bminut\\b|\\bmo\\\u010d\\b|\\bmuzejem\\b|\\bmuzici\\b|\\bna\\\u010di\\b|\\bnaduvan\\b|\Line truncated
+ 'hun': {'BeginLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'EndLines': {'data': OrderedDict(),
+                      'pattern': None},
+         'PartialLines': {'data': OrderedDict(),
+                          'pattern': None},
+         'PartialWordsAlways': {'data': OrderedDict(),
+                                'pattern': None},
+         'WholeLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'WholeWords': {'data': OrderedDict(),
+                        'pattern': None}},
+ 'nld': {'BeginLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'EndLines': {'data': OrderedDict(),
+                      'pattern': None},
+         'PartialLines': {'data': OrderedDict(),
+                          'pattern': None},
+         'PartialWordsAlways': {'data': OrderedDict(),
+                                'pattern': None},
+         'WholeLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'WholeWords': {'data': OrderedDict([(u'ls', u'Is'), (u'ln', u'In'), (u'lk', u'Ik'), (u'ledereen', u'Iedereen'), (u'ledere', u'Iedere'), (u'lemand', u'Iemand')]),
+                        'pattern': u'(um)\\bls\\b|\\bln\\b|\\blk\\b|\\bledereen\\b|\\bledere\\b|\\blemand\\b'}},
+ 'nor': {'BeginLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'EndLines': {'data': OrderedDict(),
+                      'pattern': None},
+         'PartialLines': {'data': OrderedDict(),
+                          'pattern': None},
+         'PartialWordsAlways': {'data': OrderedDict(),
+                                'pattern': None},
+         'WholeLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'WholeWords': {'data': OrderedDict(),
+                        'pattern': None}},
+ 'por': {'BeginLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'EndLines': {'data': OrderedDict(),
+                      'pattern': None},
+         'PartialLines': {'data': OrderedDict([(u'IN 6-E', u'N 6 E'), (u'in tegrar-se', u'integrar-se'), (u'in teresse', u'interesse'), (u'in testinos', u'intestinos'), (u'indica \xe7\xe3o', u'indica\xe7\xe3o'), (u'inte tino', u'intestino'), (u'intes tinos', u'intestinos'), (u'L da', u'Lda'), (u'mal estar', u'mal-estar'), (u'mastiga \xe7\xe1o', u'mastiga\xe7\xe3o'), (u'm\xe9di cas', u'm\xe9dicas'), (u'mineo rais', u'minerais'), (u'mola res', u'molares'), (u'movi mentos', u'movimentos'), (u'movimen to', u'movimento'), (u'N 5-Estendido', u'N\xba 5 Estendido'), (u'oxig\xe9 nio', u'oxig\xe9nio'), (u'pod mos', u'podemos'), (u'poder-se ia', u'poder-se-ia'), (u'pos sibilidade', u'possibilidade'), (u'possibi lidades', u'possibilidades'), (u'pro duto', u'produto'), (u'procu rar', u'procurar'), (u'Q u e', u'Que'), (u'qualifi cam', u'qualificam'), (u'R egi\xe3o', u'Regi\xe3o'), (u'unsuficien temente', u'insuficientemente')]),
+                          'pattern': None},
+         'PartialWordsAlways': {'data': OrderedDict(),
+                                'pattern': None},
+         'WholeLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'WholeWords': {'data': OrderedDict([(u'abitual', u'habitual'), (u'\xe0cerca', u'acerca'), (u'acessor', u'assessor'), (u'ac\xf3lico', u'ac\xf3lito'), (u'a\xe7oreano', u'a\xe7oriano'), (u'actuacao', u'actua\xe7\xe3o'), (u'acucar', u'a\xe7\xfacar'), (u'a\xe7ucar', u'a\xe7\xfacar'), (u'advinhar', u'adivinhar'), (u'africa', u'\xc1frica'), (u'ajuisar', u'ajuizar'), (u'album', u'\xe1lbum'), (u'alcool\xe9mia', u'alcoolemia'), (u'aldi\xe3o', u'alde\xe3o'), (u'algerino', u'argelino'), (u'ameixeal', u'ameixial'), (u'amia\xe7a', u'amea\xe7a'), (u'analizar', u'analisar'), (u'and\xe1ste', u'andaste'), (u'anemona', u'an\xe9mona'), (u'antartico', u'ant\xe1rctico'), (u'ant\xe1rtico', u'ant\xe1rctico'), (u'antep\xf4r', u'antepor'), (u'ap\xe1rte', u'aparte'), (u'apiadeiro', u'apeadeiro'), (u'apiar', u'apear'), (u'apreciacao', u'aprecia\xe7\xe3o'), (u'arctico', u'\xe1rctico'), (u'arrazar', u'arrasar'), (u'\xe1rtico', u'\xe1rctico'), (u'artifice', u'art\xedfice'), (u'artif\xedcial', u'artificial'), (u'ascen\xe7\xe3o', u'ascens\xe3o'), (u'ass\xfacar', u'a\xe7\xfacar'), (u'aste', u'haste'), (u'aster\xedstico', u'asterisco'), (u'aver\xe7\xe3o', u'avers\xe3o'), (u'avizar', u'avisar'), (u'avulsso', u'avulso'), (u'ba\xednha', u'bainha'), (u'banca-rota', u'bancarrota'), (u'bandeija', u'bandeja'), (u'b\xe9b\xe9', u'beb\xe9'), (u'beige', u'bege'), (u'ben\xe7\xe3o', u'b\xean\xe7\xe3o'), (u'benefici\xeancia', u'benefic\xeancia'), (u'beneficiente', u'beneficente'), (u'benvinda', u'bem-vinda'), (u'benvindo', u'bem-vindo'), (u'boasvindas', u'boas-vindas'), (u'borborinho', u'burburinho'), (u'Brazil', u'Brasil'), (u'bussula', u'b\xfassola'), (u'cabo-verdeano', u'cabo-verdiano'), (u'caimbras', u'c\xe3ibras'), (u'calc\xe1reo', u'calc\xe1rio'), (u'calsado', u'cal\xe7ado'), (u'calv\xedce', u'calv\xedcie'), (u'camoneano', u'camoniano'), (u'campi\xe3o', u'campe\xe3o'), (u'can\xe7acos', u'cansa\xe7os'), (u'caracter', u'car\xe1cter'), (u'caract\xe9res', u'caracteres'), (u'catequeze', u'catequese'), (u'catequisador', u'catequizador'), (u'catequisar', u'catequizar'), (u'ch\xedcara', u'x\xedcara'), (u'ciclano', u'sicrano'), (u'cicrano', u'sicrano'), (u'cidad\xe3es', u'cidad\xe3os'), (u'cidad\xf5es', u'cidad\xe3os'), (u'cincoenta', u'cinquenta'), (u'cinseiro', u'cinzeiro'), (u'cinsero', u'sincero'), (u'citacoes', u'cita\xe7\xf5es'), (u'coaliz\xe3o', u'colis\xe3o'), (u'c\xf4dia', u'c\xf4dea'), (u'comb\xf3io', u'comboio'), (u'comp\xf4r', u'compor'), (u'concerteza', u'com certeza'), (u'constituia', u'constitu\xeda'), (u'constitu\xedu', u'constituiu'), (u'contato', u'contacto'), (u'contens\xe3o', u'conten\xe7\xe3o'), (u'contribuicoes', u'contribui\xe7\xf5es'), (u'c\xf4r', u'cor'), (u'corass\xe3o', u'cora\xe7\xe3o'), (u'cor\xe7ario', u'cors\xe1rio'), (u'cor\xe7\xe1rio', u'cors\xe1rio'), (u'cornprimidosinbo', u'comprimidozinho'), (u'cr\xe2neo', u'cr\xe2nio'), (u'dE', u'de'), (u'defeni\xe7\xe3o', u'defini\xe7\xe3o'), (u'defenido', u'definido'), (u'defenir', u'definir'), (u'deficite', u'd\xe9fice'), (u'degladiar', u'digladiar'), (u'deiche', u'deixe'), (u'desinteria', u'disenteria'), (u'despendio', u'disp\xeandio'), (u'desp\xeandio', u'disp\xeandio'), (u'desplic\xeancia', u'displic\xeancia'), (u'dificulidade', u'dificuldade'), (u'dispender', u'despender'), (u'dispendio', u'disp\xeandio'), (u'distribuido', u'distribu\xeddo'), (u'dru\xedda', u'druida'), (u'\xe9cr\xe3', u'ecr\xe3'), (u'ecran', u'ecr\xe3'), (u'\xe9cran', u'ecr\xe3'), (u'\xeale', u'ele'), (u'elice', u'h\xe9lice'), (u'\xe9lice', u'h\xe9lice'), (u'emiratos', u'emirados'), (u'engolis-te', u'engoliste'), (u'engulir', u'engolir'), (u'enguliste', u'engoliste'), (u'entertido', u'entretido'), (u'entitular', u'intitular'), (u'entreterimento', u'entretenimento'), (u'entreti-me', u'entretive-me'), (u'env\xf3lucro', u'inv\xf3lucro'), (u'er\xf3i', u'her\xf3i'), (u'escluir', u'excluir'), (u'esclus\xe3o', u'exclus\xe3o'), (u'escriv\xf5es', u'escriv\xe3es'), (u'esqueiro', u'isqueiro'), (u'esquesito', u'esquisito'), (u'estacoes', u'esta\xe7\xf5es'), (u'esteje', u'esteja'), (u'excava\xe7\xe3o', u'escava\xe7\xe3o'), (u'excavar', u'escavar'), (u'exdr\xfaxula', u'esdr\xfaxula'), (u'exdr\xfaxulas', u'esdr\xfaxulas'), (u'exitar', u'hesitar'), (u'explicacoes', u'explica\xe7\xf5es'), (u'exquisito', u'esquisito'), (u'extende', u'estende'), (u'extender', u'estender'), (u'f\xe0cilmenfe', u'facilmente'), (u'f\xe0cilmente', u'facilmente'), (u'fariam-lhe', u'far-lhe-iam'), (u'FARM\xc1ClAS', u'FARM\xc1CIAS'), (u'farmec\xeautico', u'farmac\xeautico'), (u'fassa', u'fa\xe7a'), (u'f\xe9bre', u'febre'), (u'fecula', u'f\xe9cula'), (u'f\xe9mea', u'f\xeamea'), (u'femenino', u'feminino'), (u'femininismo', u'feminismo'), (u'f\xedsiologista', u'fisiologista'), (u'fiz\xe9mos', u'fizemos'), (u'fizes-te', u'fizeste'), (u'fl\xf4r', u'flor'), (u'for\xe3o', u'foram'), (u'formalisar', u'formalizar'), (u'f\xf4ro', u'foro'), (u'fos-te', u'foste'), (u'frag\xe2ncia', u'fragr\xe2ncia'), (u'fran\xe7\xeas', u'franc\xeas'), (u'frasqutnho', u'frasquiLine truncated
+                        'pattern': u'(um)\\babitual\\b|\\b\\\xe0cerca\\b|\\bacessor\\b|\\bac\\\xf3lico\\b|\\ba\\\xe7oreano\\b|\\bactuacao\\b|\\bacucar\\b|\\ba\\\xe7ucar\\b|\\badvinhar\\b|\\bafrica\\b|\\bajuisar\\b|\\balbum\\b|\\balcool\\\xe9mia\\b|\\baldi\\\xe3o\\b|\\balgerino\\b|\\bameixeal\\b|\\bamia\\\xe7a\\b|\\banalizar\\b|\\band\\\xe1ste\\b|\\banemona\\b|\\bantartico\\b|\\bant\\\xe1rtico\\b|\\bantep\\\xf4r\\b|\\bap\\\xe1rte\\b|\\bapiadeiro\\b|\\bapiar\\b|\\bapreciacao\\b|\\barctico\\b|\\barrazar\\b|\\b\\\xe1rtico\\b|\\bartifice\\b|\\bartif\\\xedcial\\b|\\bascen\\\xe7\\\xe3o\\b|\\bass\\\xfacar\\b|\\baste\\b|\\baster\\\xedstico\\b|\\baver\\\xe7\\\xe3o\\b|\\bavizar\\b|\\bavulsso\\b|\\bba\\\xednha\\b|\\bbanca\\-rota\\b|\\bbandeija\\b|\\bb\\\xe9b\\\xe9\\b|\\bbeige\\b|\\bben\\\xe7\\\xe3o\\b|\\bbenefici\\\xeancia\\b|\\bbeneficiente\\b|\\bbenvinda\\b|\\bbenvindo\\b|\\bboasvindas\\b|\\bborborinho\\b|\\bBrazil\\b|\\bbussula\\b|\\bcabo\\-verdeano\\b|\\bcaimbras\\b|\\bcalc\\\xe1reo\\b|\\bcalsado\\b|\\bcalv\\\xedce\\b|\\bcamoneano\\b|\\bcampi\\\xe3o\\b|\\bcan\\\xe7acos\\b|\\bcaracter\\b|\\bcaract\\\xe9res\\b|\\bcatequeze\\b|\\bcatequisador\\b|\\bcatequisar\\b|\\bch\\\xedcara\\b|\\bciclano\\b|\\bcicrano\\b|\\bcidad\\\xe3es\\b|\\bcidad\\\xf5es\\b|\\bcincoenta\\b|\\bcinseiro\\b|\\bcinsero\\b|\\bcitacoes\\b|\\bcoaliz\\\xe3o\\b|\\bc\\\xf4dia\\b|\\bcomb\\\xf3io\\b|\\bcomp\\\xf4r\\b|\\bconcerteza\\b|\\bconstituia\\b|\\bconstitu\\\xedu\\b|\\bcontato\\b|\\bcontens\\\xe3o\\b|\\bcontribuicoes\\b|\\bc\\\xf4r\\b|\\bcorass\\\xe3o\\b|\\bcor\\\xe7ario\\b|\\bcor\\\xe7\\\xe1rio\\b|\\bcornprimidosinbo\\b|\\bcr\\\xe2neo\\b|\\bdE\\b|\\bdefeni\\\xe7\\\xe3o\\b|\\bdefenido\\b|\\bdefenir\\b|\\bdeficite\\b|\\bdegladiar\\b|\\bdeiche\\b|\\bdesinteria\\b|\\bdespendio\\b|\\bdesp\\\xeandio\\b|\\bdesplic\\\xeancia\\b|\\bdificulidade\\b|\\bdispender\\b|\\bdispendio\\b|\\bdistribuido\\b|\\bdru\\\xedda\\b|\\b\\\xe9cr\\\xe3\\b|\\becran\\b|\\b\\\xe9cran\\b|\\b\\\xeale\\b|\\belice\\b|\\b\\\xe9lice\\b|\\bemiratos\\b|\\bengolis\\-te\\b|\\bengulir\\b|\\benguliste\\b|\\bentertido\\b|\\bentitular\\b|\\bentreterimento\\b|\\bentreti\\-me\\b|\\benv\\\xf3lucro\\b|\\ber\\\xf3i\\b|\\bescluir\\b|\\besclus\\\xe3o\\b|\\bescriv\\\xf5es\\b|\\besqueiro\\b|\\besquesito\\b|\\bestacoes\\b|\\besteje\\b|\\bexcava\\\xe7\\\xe3o\\b|\\bexcavar\\b|\\bexdr\\\xfaxula\\b|\\bexdr\\\xfaxulas\\b|\\bexitar\\b|\\bexplicacoes\\b|\\bexquisito\\b|\\bextende\\b|\\bextender\\b|\\bf\\\xe0cilmenfe\\b|\\bf\\\xe0cilmente\\b|\\bfariam\\-lhe\\b|\\bFARM\\\xc1ClAS\\b|\\bfarmec\\\xeautico\\b|\\bfassa\\b|\\bf\\\xe9bre\\b|\\bfecula\\b|\\bf\\\xe9mea\\b|\\bfemenino\\b|\\bfemininismo\\b|\\bf\\\xedsiologista\\b|\\bfiz\\\xe9mos\\b|\\bfizes\\-te\\b|\\bfl\\\xf4r\\b|\\bfor\\\xe3o\\b|\\bformalisar\\b|\\bf\\\xf4ro\\b|\\bfos\\-te\\b|\\bfrag\\\xe2ncia\\b|\\bfran\\\xe7\\\xeas\\b|\\bfrasqutnho\\b|\\bfrustado\\b|\\bfur\\\xe1\\b|\\bgaz\\b|\\bg\\\xe1z\\b|\\bgeito\\b|\\bgeneceu\\b|\\bgeropiga\\b|\\bglic\\\xe9mia\\b|\\bgorgeta\\b|\\bgrangear\\b|\\bguizar\\b|\\bhectar\\b|\\bherm\\\xe9ticamente\\b|\\bhernia\\b|\\bhigi\\\xe9ne\\b|\\bhilariedade\\b|\\bhiperac\\\xeddez\\b|\\bhontem\\b|\\bigiene\\b|\\bigienico\\b|\\bigi\\\xe9nico\\b|\\bigreija\\b|\\biguasu\\b|\\bilac\\\xe7\\\xe3o\\b|\\bimbigo\\b|\\bimpecilho\\b|\\b\\\xedncas\\b|\\binc\\\xeasto\\b|\\binclusiv\\\xe9\\b|\\binc\\\xf4modos\\b|\\bincontest\\\xe1velmente\\b|\\bincontest\\\xe0velmente\\b|\\bindespens\\\xe1veis\\b|\\bindespens\\\xe1vel\\b|\\bIndia\\b|\\bindiguina\\\xe7\\\xe3o\\b|\\bindiguinado\\b|\\bindiguinar\\b|\\binflac\\\xe7\\\xe3o\\b|\\bingreja\\b|\\bINSCRICOES\\b|\\bintens\\\xe3o\\b|\\bintertido\\b|\\bintoxica\\b|\\bintrega\\b|\\binveros\\\xedmel\\b|\\biorgute\\b|\\bipop\\\xf3tamo\\b|\\bipsilon\\b|\\bipslon\\b|\\bisquesito\\b|\\bju\\\xedz\\b|\\bjuiza\\b|\\bj\\\xfaniores\\b|\\bjustanzente\\b|\\bjuz\\b|\\bkilo\\b|\\blaborat\\\xf3rio\\-porque\\b|\\bladravaz\\b|\\blament\\\xe0velmente\\b|\\blampe\\\xe3o\\b|\\blargartixa\\b|\\blargarto\\b|\\bl\\\xeam\\b|\\bleuc\\\xe9mia\\b|\\blicensa\\b|\\blingu\\\xedsta\\b|\\blisongear\\b|\\blogista\\b|\\bma\\\xe7ajar\\b|\\bMacfadden\\-o\\b|\\bmae\\b|\\bmagestade\\b|\\bm\\\xe3gua\\b|\\bmangerico\\b|\\bmangerona\\b|\\bmanteem\\-se\\b|\\bmantega\\b|\\bmantem\\-se\\b|\\bmassi\\\xe7o\\b|\\bmassisso\\b|\\bm\\\xe9dica\\-Rio\\b|\\bmenistro\\b|\\bmerciaria\\b|\\bmetrelhadora\\b|\\bmiscegena\\\xe7\\\xe3o\\b|\\bmisogenia\\b|\\bmisogeno\\b|\\bmis\\\xf3geno\\b|\\bm\\\xba\\b|\\bm\\\xf4lho\\b|\\bmonument\\\xe2nea\\b|\\bmortandela\\b|\\bmorteIa\\b|\\bmuinto\\b|\\bnasaias\\b|\\bn\\\xeale\\b|\\bnest\\b|\\bNivea\\b|\\bnonagessimo\\b|\\bnonag\\\xe9ssimo\\b|\\bnornal\\b|\\bnot\\\xe0velmente\\b|\\bobcess\\\xe3o\\b|\\bobesidae\\b|\\b\\\xf3bviamente\\b|\\b\\\xf2bviamente\\b|\\bofecina\\b|\\boje\\b|\\bomem\\b|\\bopcoes\\b|\\bop\\\xf3brio\\b|\\bopr\\\xf3bio\\b|\\borf\\\xe3o\\b|\\borganigrama\\b|\\borganisar\\b|\\borg\\\xe3o\\b|\\borta\\b|\\b\\\xf3tima\\b|\\b\\\xf3timos\\b|\\bparaliza\\\xe7\\\xe3o\\b|\\bparalizado\\b|\\bparalizar\\b|\\bpar\\\xe1ste\\b|\\bP\\Line truncated
+ 'rus': {'BeginLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'EndLines': {'data': OrderedDict(),
+                      'pattern': None},
+         'PartialLines': {'data': OrderedDict(),
+                          'pattern': None},
+         'PartialWordsAlways': {'data': OrderedDict(),
+                                'pattern': None},
+         'WholeLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'WholeWords': {'data': OrderedDict([(u'\u041d\u0404\u0419', u'\u041d\u0415\u0419'), (u'\u041e\u0420\u0413\u0417\u041d\u0418\u0417\u041c\u041e\u0411', u'\u041e\u0420\u0413\u0410\u041d\u0418\u0417\u041c\u0410'), (u'\u0427\u04570', u'\u0427\u0422\u041e'), (u'\u041d\u042d', u'\u041d\u0410'), (u'\u0421\u041e\u0421\u0404\u0414\u041d\u042e\u042e', u'\u0421\u041e\u0421\u0415\u0414\u041d\u042e\u042e'), (u'\u041f\u041b\u0417\u041d\u0404\u0422\u0423', u'\u041f\u041b\u0410\u041d\u0415\u0422\u0423'), (u'\u0417\u0417\u0413\u042d\u0414\u041e\u041a', u'\u0417\u0410\u0413\u0410\u0414\u041e\u041a'), (u'\u0421\u041e\u0422\u0412\u041e\u0420\u0404\u041d\u0418\u042f', u'\u0421\u041e\u0422\u0412\u041e\u0420\u0415\u041d\u0418\u042f'), (u'\u041c\u0418\u0420\u042d', u'\u041c\u0418\u0420\u0410'), (u'\u041f\u041e\u042f\u0411\u041b\u0404\u041d\u0418\u042f', u'\u041f\u041e\u042f\u0412\u041b\u0415\u041d\u0418\u042f'), (u'\u0417\u0404\u041c\u041b\u0404', u'\u0417\u0415\u041c\u041b\u0415'), (u'\u0404\u0429\u0404', u'\u0415\u0429\u0401'), (u'\u0422\u0404\u041c\u041d\u042c\u0406\u0425', u'\u0422\u0415\u041c\u041d\u042b\u0425'), (u'\u0421\u0404\u0420\u042c\u0404\u0417\u041d\u042c\u0406\u041c', u'\u0421\u0415\u0420\u042c\u0415\u0417\u041d\u042b\u041c'), (u'\u041f\u041e\u0428\u0406\u04060', u'\u041f\u041e\u0428\u041b\u041e'), (u'\u041f\u04400\u0418\u04170\u0428\u0404\u041b', u'\u041f\u0420\u041e\u0418\u0417\u041e\u0428\u0415\u041b'), (u'\u0421\u0404\u041a\u0420\u0404\u0422\u042d\u041c\u0418', u'\u0421\u0415\u041a\u0420\u0415\u0422\u0410\u041c\u0418'), (u'\u041c\u042d\u0422\u0404\u0420\u0418\u0417\u041b\u042c\u0406', u'\u041c\u0410\u0422\u0415\u0420\u0418\u0410\u041b\u042b'), (u'\u041f\u042f\u0422\u0404\u041d', u'\u041f\u042f\u0422\u0415\u041d'), (u'\u041f\u041b\u0430\u041d\u0404\u0457\u0404', u'\u041f\u041b\u0410\u041d\u0415\u0422\u0415'), (u'\u041a\u0417\u0422\u042d\u041a\u041b\u0418\u0417\u041c', u'\u041a\u0410\u0422\u0410\u041a\u041b\u0418\u0417\u041c'), (u'\u041e\u041a\u0417\u0417\u0417\u041b\u0421\u042f', u'\u041e\u041a\u0410\u0417\u0410\u041b\u0421\u042f'), (u'\u0414\u042d\u041b\u042c\u0428\u0415', u'\u0414\u0410\u041b\u042c\u0428\u0415'), (u'\u0422\u0412\u041a', u'\u0422\u0410\u041a'), (u'\u041f\u041b\u0417\u041d\u0404\u0422\u0417', u'\u041f\u041b\u0410\u041d\u0415\u0422\u0410'), (u'\u0427\u0404\u0413\u041e', u'\u0427\u0415\u0413\u041e'), (u'\u0423\u0417\u041d\u042d\u0422\u042c', u'\u0423\u0417\u041d\u0410\u0422\u042c'), (u'\u041f\u041b\u042d\u041d\u0404\u0422\u0404', u'\u041f\u041b\u0410\u041d\u0415\u0422\u0415'), (u'\u041d\u0404\u041c', u'\u041d\u0415\u041c'), (u'\u0411\u041e\u0417\u041c\u041e\u0416\u041d\u0417', u'\u0412\u041e\u0417\u041c\u041e\u0416\u041d\u0410'), (u'\u0421\u041e\u0411\u0404\u0420\u0428\u0404\u041d\u041d\u041e', u'\u0421\u041e\u0412\u0415\u0420\u0428\u0415\u041d\u041d\u041e'), (u'\u0418\u041d\u042d\u0427\u0404', u'\u0418\u041d\u0410\u0427\u0415'), (u'\u0411\u0421\u0404', u'\u0412\u0421\u0415'), (u'\u041d\u0415\u0414\u041e\u0421\u0422\u0417\u0422\u041a\u0418', u'\u041d\u0415\u0414\u041e\u0421\u0422\u0410\u0422\u041a\u0418'), (u'\u041d\u041e\u0412\u042c\u0406\u0404', u'\u041d\u041e\u0412\u042b\u0415'), (u'\u0412\u0404\u041b\u0418\u041a\u041e\u041b\u0404\u041f\u041d\u042d\u042f', u'\u0412\u0415\u041b\u0418\u041a\u041e\u041b\u0415\u041f\u041d\u0410\u042f'), (u'\u041e\u0421\u0422\u042d\u0406\u0406\u041e\u0421\u042c', u'\u041e\u0421\u0422\u0410\u041b\u041e\u0421\u042c'), (u'\u041d\u0417\u041b\u0418\u0427\u0418\u0404', u'\u041d\u0410\u041b\u0418\u0427\u0418\u0415'), (u'\u0431\u042b', u'\u0431\u044b'), (u'\u041f\u0420\u041e\u0426\u0412\u0415\u0422\u0412\u0422\u042c', u'\u041f\u0420\u041e\u0426\u0412\u0415\u0422\u0410\u0422\u042c'), (u'\u041a\u0417\u041a', u'\u041a\u0410\u041a'), (u'\u0412\u041e\u0414\u0417', u'\u0412\u041e\u0414\u0410'), (u'\u041d\u0417\u0428\u0415\u041b', u'\u041d\u0410\u0428\u0415\u041b'), (u'\u041d\u0404', u'\u041d\u0415'), (u'\u0422\u041e\u0416\u0404', u'\u0422\u041e\u0416\u0415'), (u'\u0412\u0423\u041b\u041a\u042d\u041d\u0418\u0427\u0404\u0421\u041a\u041e\u0419', u'\u0412\u0423\u041b\u041a\u0410\u041d\u0418\u0427\u0415\u0421\u041a\u041e\u0419'), (u'\u042d\u041a\u0422\u0418\u0411\u041d\u041e\u0421\u0422\u0418', u'\u0410\u041a\u0422\u0418\u0412\u041d\u041e\u0421\u0422\u0418'), (u'\u041f\u041e\u042f\u0412\u0418\u041b\u0417\u0421\u042c', u'\u041f\u041e\u042f\u0412\u0418\u041b\u0410\u0421\u042c'), (u'\u041d\u041e\u0412\u0417\u042f', u'\u041d\u041e\u0412\u0410\u042f'), (u'\u0421\u0422\u0420\u042d\u0422\u0404\u0413\u0418\u042f', u'\u0421\u0422\u0420\u0410\u0422\u0415\u0413\u0418\u042f'), (u'\u0423\u0421\u041f\u0404\u0428\u041d0', u'\u0423\u0421\u041f\u0415\u0428\u041d\u041e'), (u'\u041f\u041e\u0421\u0417\u0414\u041a\u0423', u'\u041f\u041e\u0421\u0410\u0414\u041a\u0423'), (u'\u0413\u041e\u0422\u041e\u0411\u042b', u'\u0413\u041e\u0422\u041e\u0412\u042b'), (u'\u041d\u0417\u0427\u0417\u0422\u042c', u'\u041d\u0410\u0427\u0410\u0422\u042c'), (u'\u041e\u0425\u041e\u0422\u042d', u'\u041e\u0425\u041e\u0422\u0Line truncated
+                        'pattern': u'(um)\\b\\\u041d\\\u0404\\\u0419\\b|\\b\\\u041e\\\u0420\\\u0413\\\u0417\\\u041d\\\u0418\\\u0417\\\u041c\\\u041e\\\u0411\\b|\\b\\\u0427\\\u04570\\b|\\b\\\u041d\\\u042d\\b|\\b\\\u0421\\\u041e\\\u0421\\\u0404\\\u0414\\\u041d\\\u042e\\\u042e\\b|\\b\\\u041f\\\u041b\\\u0417\\\u041d\\\u0404\\\u0422\\\u0423\\b|\\b\\\u0417\\\u0417\\\u0413\\\u042d\\\u0414\\\u041e\\\u041a\\b|\\b\\\u0421\\\u041e\\\u0422\\\u0412\\\u041e\\\u0420\\\u0404\\\u041d\\\u0418\\\u042f\\b|\\b\\\u041c\\\u0418\\\u0420\\\u042d\\b|\\b\\\u041f\\\u041e\\\u042f\\\u0411\\\u041b\\\u0404\\\u041d\\\u0418\\\u042f\\b|\\b\\\u0417\\\u0404\\\u041c\\\u041b\\\u0404\\b|\\b\\\u0404\\\u0429\\\u0404\\b|\\b\\\u0422\\\u0404\\\u041c\\\u041d\\\u042c\\\u0406\\\u0425\\b|\\b\\\u0421\\\u0404\\\u0420\\\u042c\\\u0404\\\u0417\\\u041d\\\u042c\\\u0406\\\u041c\\b|\\b\\\u041f\\\u041e\\\u0428\\\u0406\\\u04060\\b|\\b\\\u041f\\\u04400\\\u0418\\\u04170\\\u0428\\\u0404\\\u041b\\b|\\b\\\u0421\\\u0404\\\u041a\\\u0420\\\u0404\\\u0422\\\u042d\\\u041c\\\u0418\\b|\\b\\\u041c\\\u042d\\\u0422\\\u0404\\\u0420\\\u0418\\\u0417\\\u041b\\\u042c\\\u0406\\b|\\b\\\u041f\\\u042f\\\u0422\\\u0404\\\u041d\\b|\\b\\\u041f\\\u041b\\\u0430\\\u041d\\\u0404\\\u0457\\\u0404\\b|\\b\\\u041a\\\u0417\\\u0422\\\u042d\\\u041a\\\u041b\\\u0418\\\u0417\\\u041c\\b|\\b\\\u041e\\\u041a\\\u0417\\\u0417\\\u0417\\\u041b\\\u0421\\\u042f\\b|\\b\\\u0414\\\u042d\\\u041b\\\u042c\\\u0428\\\u0415\\b|\\b\\\u0422\\\u0412\\\u041a\\b|\\b\\\u041f\\\u041b\\\u0417\\\u041d\\\u0404\\\u0422\\\u0417\\b|\\b\\\u0427\\\u0404\\\u0413\\\u041e\\b|\\b\\\u0423\\\u0417\\\u041d\\\u042d\\\u0422\\\u042c\\b|\\b\\\u041f\\\u041b\\\u042d\\\u041d\\\u0404\\\u0422\\\u0404\\b|\\b\\\u041d\\\u0404\\\u041c\\b|\\b\\\u0411\\\u041e\\\u0417\\\u041c\\\u041e\\\u0416\\\u041d\\\u0417\\b|\\b\\\u0421\\\u041e\\\u0411\\\u0404\\\u0420\\\u0428\\\u0404\\\u041d\\\u041d\\\u041e\\b|\\b\\\u0418\\\u041d\\\u042d\\\u0427\\\u0404\\b|\\b\\\u0411\\\u0421\\\u0404\\b|\\b\\\u041d\\\u0415\\\u0414\\\u041e\\\u0421\\\u0422\\\u0417\\\u0422\\\u041a\\\u0418\\b|\\b\\\u041d\\\u041e\\\u0412\\\u042c\\\u0406\\\u0404\\b|\\b\\\u0412\\\u0404\\\u041b\\\u0418\\\u041a\\\u041e\\\u041b\\\u0404\\\u041f\\\u041d\\\u042d\\\u042f\\b|\\b\\\u041e\\\u0421\\\u0422\\\u042d\\\u0406\\\u0406\\\u041e\\\u0421\\\u042c\\b|\\b\\\u041d\\\u0417\\\u041b\\\u0418\\\u0427\\\u0418\\\u0404\\b|\\b\\\u0431\\\u042b\\b|\\b\\\u041f\\\u0420\\\u041e\\\u0426\\\u0412\\\u0415\\\u0422\\\u0412\\\u0422\\\u042c\\b|\\b\\\u041a\\\u0417\\\u041a\\b|\\b\\\u0412\\\u041e\\\u0414\\\u0417\\b|\\b\\\u041d\\\u0417\\\u0428\\\u0415\\\u041b\\b|\\b\\\u041d\\\u0404\\b|\\b\\\u0422\\\u041e\\\u0416\\\u0404\\b|\\b\\\u0412\\\u0423\\\u041b\\\u041a\\\u042d\\\u041d\\\u0418\\\u0427\\\u0404\\\u0421\\\u041a\\\u041e\\\u0419\\b|\\b\\\u042d\\\u041a\\\u0422\\\u0418\\\u0411\\\u041d\\\u041e\\\u0421\\\u0422\\\u0418\\b|\\b\\\u041f\\\u041e\\\u042f\\\u0412\\\u0418\\\u041b\\\u0417\\\u0421\\\u042c\\b|\\b\\\u041d\\\u041e\\\u0412\\\u0417\\\u042f\\b|\\b\\\u0421\\\u0422\\\u0420\\\u042d\\\u0422\\\u0404\\\u0413\\\u0418\\\u042f\\b|\\b\\\u0423\\\u0421\\\u041f\\\u0404\\\u0428\\\u041d0\\b|\\b\\\u041f\\\u041e\\\u0421\\\u0417\\\u0414\\\u041a\\\u0423\\b|\\b\\\u0413\\\u041e\\\u0422\\\u041e\\\u0411\\\u042b\\b|\\b\\\u041d\\\u0417\\\u0427\\\u0417\\\u0422\\\u042c\\b|\\b\\\u041e\\\u0425\\\u041e\\\u0422\\\u042d\\b|\\b\\\u041f\\\u0420\\\u0418\\\u0417\\\u041d\\\u0417\\\u041a\\\u0417\\\u041c\\\u0418\\b|\\b\\\u041f\\\u04400\\\u0428\\\u041b\\\u041e\\\u041c\\b|\\b\\\u041d\\\u042d\\\u0421\\\u0422\\\u041e\\\u042f\\\u0429\\\u0404\\\u041c\\b|\\b\\\u041f\\\u0423\\\u0421\\\u0422\\\u041e\\\u0422\\\u0417\\\u0425\\b|\\b\\\u0411\\\u041b\\\u0417\\\u0416\\\u041d\\\u041e\\\u0419\\b|\\b\\\u041f\\\u041e\\\u0427\\\u0411\\\u0404\\b|\\b\\\u041c\\\u042c\\\u0406\\b|\\b\\\u0421\\\u0404\\\u0419\\\u0427\\\u0417\\\u0421\\b|\\b\\\u0404\\\u0421\\\u041b\\\u0418\\b|\\b\\\u0417\\\u0417\\\u0422\\\u0420\\\u041e\\\u041d\\\u0415\\\u041c\\b|\\b\\\u041e\\\u041f\\\u0417\\\u0421\\\u0417\\\u0404\\\u041c\\\u0421\\\u042f\\b|\\b\\\u0421\\\u0418\\\u041b\\\u042c\\\u041d0\\b|\\b\\\u041e\\\u0422\\\u041b\\\u0418\\\u0427\\\u0417\\\u0404\\\u0422\\\u0421\\\u042f\\b|\\b\\\u0420\\\u042d\\\u041d\\\u042c\\\u0428\\\u0404\\b|\\b\\\u041d\\\u0417\\\u0417\\\u042c\\\u0406\\\u0412\\\u0417\\\u042e\\\u0422\\b|\\b\\\u0422\\\u0404\\\u041a\\\u041b3\\b|\\b\\\u041e\\\u0421\\\u0417\\\u0414\\\u041e\\\u0427\\\u041d\\\u042b\\\u041c\\\u0418\\b|\\b\\\u041f\\\u041e\\\u0421\\\u0422\\\u0404\\\u041f\\\u0404\\\u041d\\\u041d0\\b|\\b\\\u0418\\\u0421\\\u041f\\\u042d\\\u0420\\\u042f\\\u041b\\\u0417\\\u0421\\\u042c\\b|\\b\\\u0404\\\u041e\\\u041b\\\u042c\\\u0428\\\u041e\\\u0404\\b|\\b\\\u041a\\\u041e\\\u041b\\\u0418\\\u0427\\\u0404\\\u0421\\\u0422\\\u0411\\\u041e\\b|\\b\\\u0413\\\u0404\\\u041c\\\u0417\\\u0422\\\u0418\\\u0422\\\u0415\\b|\\b\\\u041f\\\u041e\\\u041b\\\u0423\\\u0427\\\u042d\\\u0404\\\u0422\\b|\\b\\\u041d\\\u0404\\\u0414\\\u041e\\\u0421\\\u0422\\\u0417\\\u0427\\\u041d0\\b|\\b\\\u041f\\\u0418\\\u0422\\\u042d\\\u041d\\\u0418\\\u042f\\b|\\b\\\u041f\\\u041e\Line truncated
+ 'spa': {'BeginLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'EndLines': {'data': OrderedDict([(u'.\xbb.', u'\xbb.')]),
+                      'pattern': u'(um)\\b\\.\\\xbb\\.\\b$'},
+         'PartialLines': {'data': OrderedDict([(u'de gratis', u'gratis'), (u'si quiera', u'siquiera'), (u'Cada una de los', u'Cada uno de los'), (u'Cada uno de las', u'Cada una de las'), (u'haber que', u'a ver qu\xe9'), (u'haber qu\xe9', u'a ver qu\xe9'), (u'Haber si', u'A ver si'), (u' que hora', u' qu\xe9 hora'), (u'yo que se', u'yo qu\xe9 s\xe9'), (u'Yo que se', u'Yo qu\xe9 s\xe9'), (u' tu!', u' t\xfa!'), (u' si!', u' s\xed!'), (u' mi!', u' m\xed!'), (u' el!', u' \xe9l!'), (u' tu?', u' t\xfa?'), (u' si?', u' s\xed?'), (u' mi?', u' m\xed?'), (u' el?', u' \xe9l?'), (u' aun?', u' a\xfan?'), (u' mas?', u' m\xe1s?'), (u' que?', u' qu\xe9?'), (u' paso?', u' pas\xf3?'), (u' cuando?', u' cu\xe1ndo?'), (u' cuanto?', u' cu\xe1nto?'), (u' cuanta?', u' cu\xe1nta?'), (u' cuantas?', u' cu\xe1ntas?'), (u' cuantos?', u' cu\xe1ntos?'), (u' donde?', u' d\xf3nde?'), (u' quien?', u' qui\xe9n?'), (u' como?', u' c\xf3mo?'), (u' adonde?', u' ad\xf3nde?'), (u' cual?', u' cu\xe1l?'), (u'\xbfSi?', u'\xbfS\xed?'), (u'\xbfesta bien?', u'\xbfest\xe1 bien?'), (u'\xbfPero qu\xe9 haces?', u'\xa1\xbfPero qu\xe9 haces?!'), (u'\xbfpero qu\xe9 haces?', u'\xa1\xbfpero qu\xe9 haces?!'), (u'\xbfEs que no me has escuchado?', u'\xa1\xbfEs que no me has escuchado?!'), (u'\xa1\xbfes que no me has escuchado?!', u'\xa1\xbfes que no me has escuchado?!'), (u'\xbfaun', u'\xbfa\xfan'), (u'\xbftu ', u'\xbft\xfa '), (u'\xbfque ', u'\xbfqu\xe9 '), (u'\xbfsabes que', u'\xbfsabes qu\xe9'), (u'\xbfsabes adonde', u'\xbfsabes ad\xf3nde'), (u'\xbfsabes cual', u'\xbfsabes cu\xe1l'), (u'\xbfsabes quien', u'\xbfsabes qui\xe9n'), (u'\xbfsabes como', u'\xbfsabes c\xf3mo'), (u'\xbfsabes cuan', u'\xbfsabes cu\xe1n'), (u'\xbfsabes cuanto', u'\xbfsabes cu\xe1nto'), (u'\xbfsabes cuanta', u'\xbfsabes cu\xe1nta'), (u'\xbfsabes cuantos', u'\xbfsabes cu\xe1ntos'), (u'\xbfsabes cuantas', u'\xbfsabes cu\xe1ntas'), (u'\xbfsabes cuando', u'\xbfsabes cu\xe1ndo'), (u'\xbfsabes donde', u'\xbfsabes d\xf3nde'), (u'\xbfsabe que', u'\xbfsabe qu\xe9'), (u'\xbfsabe adonde', u'\xbfsabe ad\xf3nde'), (u'\xbfsabe cual', u'\xbfsabe cu\xe1l'), (u'\xbfsabe quien', u'\xbfsabe qui\xe9n'), (u'\xbfsabe como', u'\xbfsabe c\xf3mo'), (u'\xbfsabe cuan', u'\xbfsabe cu\xe1n'), (u'\xbfsabe cuanto', u'\xbfsabe cu\xe1nto'), (u'\xbfsabe cuanta', u'\xbfsabe cu\xe1nta'), (u'\xbfsabe cuantos', u'\xbfsabe cu\xe1ntos'), (u'\xbfsabe cuantas', u'\xbfsabe cu\xe1ntas'), (u'\xbfsabe cuando', u'\xbfsabe cu\xe1ndo'), (u'\xbfsabe donde', u'\xbfsabe d\xf3nde'), (u'\xbfsaben que', u'\xbfsaben qu\xe9'), (u'\xbfsaben adonde', u'\xbfsaben ad\xf3nde'), (u'\xbfsaben cual', u'\xbfsaben cu\xe1l'), (u'\xbfsaben quien', u'\xbfsaben qui\xe9n'), (u'\xbfsaben como', u'\xbfsaben c\xf3mo'), (u'\xbfsaben cuan', u'\xbfsaben cu\xe1n'), (u'\xbfsaben cuanto', u'\xbfsaben cu\xe1nto'), (u'\xbfsaben cuanta', u'\xbfsaben cu\xe1nta'), (u'\xbfsaben cuantos', u'\xbfsaben cu\xe1ntos'), (u'\xbfsaben cuantas', u'\xbfsaben cu\xe1ntas'), (u'\xbfsaben cuando', u'\xbfsaben cu\xe1ndo'), (u'\xbfsaben donde', u'\xbfsaben d\xf3nde'), (u'\xbfde que', u'\xbfde qu\xe9'), (u'\xbfde donde', u'\xbfde d\xf3nde'), (u'\xbfde cual', u'\xbfde cu\xe1l'), (u'\xbfde quien', u'\xbfde qui\xe9n'), (u'\xbfde cuanto', u'\xbfde cu\xe1nto'), (u'\xbfde cuanta', u'\xbfde cu\xe1nta'), (u'\xbfde cuantos', u'\xbfde cu\xe1ntos'), (u'\xbfde cuantas', u'\xbfde cu\xe1ntas'), (u'\xbfde cuando', u'\xbfde cu\xe1ndo'), (u'\xbfsobre que', u'\xbfsobre qu\xe9'), (u'\xbfcomo ', u'\xbfc\xf3mo '), (u'\xbfcual ', u'\xbfcu\xe1l '), (u'\xbfen cual', u'\xbfen cu\xe1l'), (u'\xbfcuando', u'\xbfcu\xe1ndo'), (u'\xbfhasta cual', u'\xbfhasta cu\xe1l'), (u'\xbfhasta quien', u'\xbfhasta qui\xe9n'), (u'\xbfhasta cuanto', u'\xbfhasta cu\xe1nto'), (u'\xbfhasta cuantas', u'\xbfhasta cu\xe1ntas'), (u'\xbfhasta cuantos', u'\xbfhasta cu\xe1ntos'), (u'\xbfhasta cuando', u'\xbfhasta cu\xe1ndo'), (u'\xbfhasta donde', u'\xbfhasta d\xf3nde'), (u'\xbfhasta que', u'\xbfhasta qu\xe9'), (u'\xbfhasta adonde', u'\xbfhasta ad\xf3nde'), (u'\xbfdesde que', u'\xbfdesde qu\xe9'), (u'\xbfdesde cuando', u'\xbfdesde cu\xe1ndo'), (u'\xbfdesde quien', u'\xbfdesde qui\xe9n'), (u'\xbfdesde donde', u'\xbfdesde d\xf3nde'), (u'\xbfcuanto', u'\xbfcu\xe1nto'), (u'\xbfcuantos', u'\xbfcu\xe1ntos'), (u'\xbfdonde', u'\xbfd\xf3nde'), (u'\xbfadonde', u'\xbfad\xf3nde'), (u'\xbfcon que', u'\xbfcon qu\xe9'), (u'\xbfcon cual', u'\xbfcon cu\xe1l'), (u'\xbfcon quien', u'\xbfcon qui\xe9n'), (u'\xbfcon cuantos', u'\xbfcon cu\xe1ntos'), (u'\xbfcon cuantas', u'\xbfcon cu\xe1ntas'), (u'\xbfcon cuanta', u'\xbfcon cu\xe1nta'), (u'\xbfcon cuanto', u'\xbfcon cu\xe1nto'), (u'\xbfpara donde', u'\xbfpara d\xf3nde'), (u'\xbfpara adonde', u'\xbfpara ad\xf3nde'), (u'\xbfpara cuando', u'\xbfpara cu\xe1ndo'), (u'\xbfpara que', u'\xbfpara qu\xe9'), (u'\xbfpara quien', u'\xbfpara qui\xe9n'), (u'\xbfpara cuanto', u'\xbfpara cu\xe1nto'), (u'\xbfpara cuanta', u'\xbfpara cu\xe1nta'), (u'\xbfpara cuantos', u'\xbfpara cu\xe1ntos'), (u'\xbfpara cuantas', u'\xbfpara cu\xe1ntas'), (Line truncated
+                          'pattern': None},
+         'PartialWordsAlways': {'data': OrderedDict(),
+                                'pattern': None},
+         'WholeLines': {'data': OrderedDict([(u'No', u'No.')]),
+                        'pattern': None},
+         'WholeWords': {'data': OrderedDict([(u'KBs', u'kB'), (u'Vd', u'Ud'), (u'N\xb0', u'N.\xb0'), (u'n\xb0', u'n.\xb0'), (u'nro.', u'n.\xb0'), (u'Nro.', u'N.\xb0'), (u'aca', u'ac\xe1'), (u'actuas', u'act\xfaas'), (u'actues', u'act\xfaes'), (u'adios', u'adi\xf3s'), (u'agarrenla', u'ag\xe1rrenla'), (u'agarrenlo', u'ag\xe1rrenlo'), (u'agarrandose', u'agarr\xe1ndose'), (u'algun', u'alg\xfan'), (u'alli', u'all\xed'), (u'alla', u'all\xe1'), (u'alejate', u'al\xe9jate'), (u'ahi', u'ah\xed'), (u'angel', u'\xe1ngel'), (u'angeles', u'\xe1ngeles'), (u'apagala', u'ap\xe1gala'), (u'aqui', u'aqu\xed'), (u'asi', u'as\xed'), (u'bahia', u'bah\xeda'), (u'busqueda', u'b\xfasqueda'), (u'busquedas', u'b\xfasquedas'), (u'callate', u'c\xe1llate'), (u'carcel', u'c\xe1rcel'), (u'camara', u'c\xe1mara'), (u'caido', u'ca\xeddo'), (u'cabron', u'cabr\xf3n'), (u'camion', u'cami\xf3n'), (u'codigo', u'c\xf3digo'), (u'codigos', u'c\xf3digos'), (u'comence', u'comenc\xe9'), (u'comprate', u'c\xf3mprate'), (u'consegui', u'consegu\xed'), (u'confias', u'conf\xedas'), (u'convertira', u'convertir\xe1'), (u'corazon', u'coraz\xf3n'), (u'crei', u'cre\xed'), (u'creia', u'cre\xeda'), (u'creido', u'cre\xeddo'), (u'creiste', u'cre\xedste'), (u'cubrenos', u'c\xfabrenos'), (u'comio', u'comi\xf3'), (u'dara', u'dar\xe1'), (u'dia', u'd\xeda'), (u'dias', u'd\xedas'), (u'debio', u'debi\xf3'), (u'demelo', u'd\xe9melo'), (u'dimelo', u'd\xedmelo'), (u'denoslo', u'd\xe9noslo'), (u'deselo', u'd\xe9selo'), (u'decia', u'dec\xeda'), (u'decian', u'dec\xedan'), (u'detras', u'detr\xe1s'), (u'deberia', u'deber\xeda'), (u'deberas', u'deber\xe1s'), (u'deberias', u'deber\xedas'), (u'deberian', u'deber\xedan'), (u'deberiamos', u'deber\xedamos'), (u'dejame', u'd\xe9jame'), (u'dejate', u'd\xe9jate'), (u'dejalo', u'd\xe9jalo'), (u'dejarian', u'dejar\xedan'), (u'damela', u'd\xe1mela'), (u'despues', u'despu\xe9s'), (u'diciendome', u'dici\xe9ndome'), (u'dificil', u'dif\xedcil'), (u'dificiles', u'dif\xedciles'), (u'disculpate', u'disc\xfalpate'), (u'dolares', u'd\xf3lares'), (u'hechar', u'echar'), (u'examenes', u'ex\xe1menes'), (u'empezo', u'empez\xf3'), (u'empujon', u'empuj\xf3n'), (u'empujalo', u'emp\xfajalo'), (u'escondanme', u'esc\xf3ndanme'), (u'esperame', u'esp\xe9rame'), (u'estara', u'estar\xe1'), (u'estare', u'estar\xe9'), (u'estaria', u'estar\xeda'), (u'estan', u'est\xe1n'), (u'estaran', u'estar\xe1n'), (u'estabamos', u'est\xe1bamos'), (u'estuvieramos', u'estuvi\xe9ramos'), (u'exito', u'\xe9xito'), (u'facil', u'f\xe1cil'), (u'fiscalia', u'fiscal\xeda'), (u'fragil', u'fr\xe1gil'), (u'fragiles', u'fr\xe1giles'), (u'frances', u'franc\xe9s'), (u'gustaria', u'gustar\xeda'), (u'habia', u'hab\xeda'), (u'habias', u'hab\xedas'), (u'habian', u'hab\xedan'), (u'habrian', u'habr\xedan'), (u'habrias', u'habr\xedas'), (u'hagalo', u'h\xe1galo'), (u'haria', u'har\xeda'), (u'increible', u'incre\xedble'), (u'incredulo', u'incr\xe9dulo'), (u'intentalo', u'int\xe9ntalo'), (u'ire', u'ir\xe9'), (u'jovenes', u'j\xf3venes'), (u'ladron', u'ladr\xf3n'), (u'linea', u'l\xednea'), (u'llamame', u'll\xe1mame'), (u'llevalo', u'll\xe9valo'), (u'mama', u'mam\xe1'), (u'maricon', u'maric\xf3n'), (u'mayoria', u'mayor\xeda'), (u'metodo', u'm\xe9todo'), (u'metodos', u'm\xe9todos'), (u'mio', u'm\xedo'), (u'mostro', u'mostr\xf3'), (u'morira', u'morir\xe1'), (u'muevete', u'mu\xe9vete'), (u'murio', u'muri\xf3'), (u'numero', u'n\xfamero'), (u'numeros', u'n\xfameros'), (u'ningun', u'ning\xfan'), (u'oido', u'o\xeddo'), (u'oidos', u'o\xeddos'), (u'oimos', u'o\xedmos'), (u'oiste', u'o\xedste'), (u'pasale', u'p\xe1sale'), (u'pasame', u'p\xe1same'), (u'paraiso', u'para\xedso'), (u'parate', u'p\xe1rate'), (u'pense', u'pens\xe9'), (u'peluqueria', u'peluquer\xeda'), (u'platano', u'pl\xe1tano'), (u'plastico', u'pl\xe1stico'), (u'plasticos', u'pl\xe1sticos'), (u'policia', u'polic\xeda'), (u'policias', u'polic\xedas'), (u'poster', u'p\xf3ster'), (u'podia', u'pod\xeda'), (u'podias', u'pod\xedas'), (u'podria', u'podr\xeda'), (u'podrian', u'podr\xedan'), (u'podrias', u'podr\xedas'), (u'podriamos', u'podr\xedamos'), (u'prometio', u'prometi\xf3'), (u'proposito', u'prop\xf3sito'), (u'pideselo', u'p\xeddeselo'), (u'ponganse', u'p\xf3nganse'), (u'prometeme', u'prom\xe9teme'), (u'publico', u'p\xfablico'), (u'publicos', u'p\xfablicos'), (u'publicamente', u'p\xfablicamente'), (u'quedate', u'qu\xe9date'), (u'queria', u'quer\xeda'), (u'querrias', u'querr\xedas'), (u'querian', u'quer\xedan'), (u'rapido', u'r\xe1pido'), (u'rapidamente', u'r\xe1pidamente'), (u'razon', u'raz\xf3n'), (u'rehusen', u'reh\xfasen'), (u'rie', u'r\xede'), (u'rias', u'r\xedas'), (u'rindete', u'r\xedndete'), (u'sacame', u's\xe1came'), (u'sentian', u'sent\xedan'), (u'sientate', u'si\xe9ntate'), (u'sera', u'ser\xe1'), (u'soplon', u'sopl\xf3n'), (u'sueltalo', u'su\xe9ltalo'), (u'tambien', u'tambi\xe9n'), (u'teoria', u'teor\xeda'), (u'tendra', u'tendr\xe1'), (u'telefono', u'tel\xe9fono'), (u'tipica', u't\xedpica'), (u'todavia', u'todav\xeda'), (u'tomalo', u't\xf3malo'), (u'tLine truncated
+                        'pattern': u'(um)\\bKBs\\b|\\bVd\\b|\\bN\\\xb0\\b|\\bn\\\xb0\\b|\\bnro\\.\\b|\\bNro\\.\\b|\\baca\\b|\\bactuas\\b|\\bactues\\b|\\badios\\b|\\bagarrenla\\b|\\bagarrenlo\\b|\\bagarrandose\\b|\\balgun\\b|\\balli\\b|\\balla\\b|\\balejate\\b|\\bahi\\b|\\bangel\\b|\\bangeles\\b|\\bapagala\\b|\\baqui\\b|\\basi\\b|\\bbahia\\b|\\bbusqueda\\b|\\bbusquedas\\b|\\bcallate\\b|\\bcarcel\\b|\\bcamara\\b|\\bcaido\\b|\\bcabron\\b|\\bcamion\\b|\\bcodigo\\b|\\bcodigos\\b|\\bcomence\\b|\\bcomprate\\b|\\bconsegui\\b|\\bconfias\\b|\\bconvertira\\b|\\bcorazon\\b|\\bcrei\\b|\\bcreia\\b|\\bcreido\\b|\\bcreiste\\b|\\bcubrenos\\b|\\bcomio\\b|\\bdara\\b|\\bdia\\b|\\bdias\\b|\\bdebio\\b|\\bdemelo\\b|\\bdimelo\\b|\\bdenoslo\\b|\\bdeselo\\b|\\bdecia\\b|\\bdecian\\b|\\bdetras\\b|\\bdeberia\\b|\\bdeberas\\b|\\bdeberias\\b|\\bdeberian\\b|\\bdeberiamos\\b|\\bdejame\\b|\\bdejate\\b|\\bdejalo\\b|\\bdejarian\\b|\\bdamela\\b|\\bdespues\\b|\\bdiciendome\\b|\\bdificil\\b|\\bdificiles\\b|\\bdisculpate\\b|\\bdolares\\b|\\bhechar\\b|\\bexamenes\\b|\\bempezo\\b|\\bempujon\\b|\\bempujalo\\b|\\bescondanme\\b|\\besperame\\b|\\bestara\\b|\\bestare\\b|\\bestaria\\b|\\bestan\\b|\\bestaran\\b|\\bestabamos\\b|\\bestuvieramos\\b|\\bexito\\b|\\bfacil\\b|\\bfiscalia\\b|\\bfragil\\b|\\bfragiles\\b|\\bfrances\\b|\\bgustaria\\b|\\bhabia\\b|\\bhabias\\b|\\bhabian\\b|\\bhabrian\\b|\\bhabrias\\b|\\bhagalo\\b|\\bharia\\b|\\bincreible\\b|\\bincredulo\\b|\\bintentalo\\b|\\bire\\b|\\bjovenes\\b|\\bladron\\b|\\blinea\\b|\\bllamame\\b|\\bllevalo\\b|\\bmama\\b|\\bmaricon\\b|\\bmayoria\\b|\\bmetodo\\b|\\bmetodos\\b|\\bmio\\b|\\bmostro\\b|\\bmorira\\b|\\bmuevete\\b|\\bmurio\\b|\\bnumero\\b|\\bnumeros\\b|\\bningun\\b|\\boido\\b|\\boidos\\b|\\boimos\\b|\\boiste\\b|\\bpasale\\b|\\bpasame\\b|\\bparaiso\\b|\\bparate\\b|\\bpense\\b|\\bpeluqueria\\b|\\bplatano\\b|\\bplastico\\b|\\bplasticos\\b|\\bpolicia\\b|\\bpolicias\\b|\\bposter\\b|\\bpodia\\b|\\bpodias\\b|\\bpodria\\b|\\bpodrian\\b|\\bpodrias\\b|\\bpodriamos\\b|\\bprometio\\b|\\bproposito\\b|\\bpideselo\\b|\\bponganse\\b|\\bprometeme\\b|\\bpublico\\b|\\bpublicos\\b|\\bpublicamente\\b|\\bquedate\\b|\\bqueria\\b|\\bquerrias\\b|\\bquerian\\b|\\brapido\\b|\\brapidamente\\b|\\brazon\\b|\\brehusen\\b|\\brie\\b|\\brias\\b|\\brindete\\b|\\bsacame\\b|\\bsentian\\b|\\bsientate\\b|\\bsera\\b|\\bsoplon\\b|\\bsueltalo\\b|\\btambien\\b|\\bteoria\\b|\\btendra\\b|\\btelefono\\b|\\btipica\\b|\\btodavia\\b|\\btomalo\\b|\\btonterias\\b|\\btorci\\b|\\btraelos\\b|\\btraiganlo\\b|\\btraiganlos\\b|\\btrio\\b|\\btuvieramos\\b|\\bunion\\b|\\bultimo\\b|\\bultima\\b|\\bultimos\\b|\\bultimas\\b|\\bunica\\b|\\bunico\\b|\\bvamonos\\b|\\bvayanse\\b|\\bvictima\\b|\\bvivira\\b|\\bvolvio\\b|\\bvolvia\\b|\\bvolvian\\b|\\breir\\b|\\bfreir\\b|\\bsonreir\\b|\\bhazmerreir\\b|\\boir\\b|\\boirlo\\b|\\boirte\\b|\\boirse\\b|\\boirme\\b|\\boirle\\b|\\boirla\\b|\\boirles\\b|\\boirnos\\b|\\boirlas\\b|\\bbi\\\xe9n\\b|\\bcr\\\xedmen\\b|\\bfu\\\xe9\\b|\\bfu\\\xed\\b|\\bqui\\\xe9res\\b|\\bt\\\xed\\b|\\bd\\\xed\\b|\\bv\\\xe1\\b|\\bv\\\xe9\\b|\\bv\\\xed\\b|\\bvi\\\xf3\\b|\\b\\\xf3\\b|\\bcl\\\xf3n\\b|\\bdi\\\xf3\\b|\\bgui\\\xf3n\\b|\\bd\\\xf3n\\b|\\bf\\\xe9\\b|\\b\\\xe1quel\\b|\\b\\\xe9ste\\b|\\b\\\xe9sta\\b|\\b\\\xe9stos\\b|\\b\\\xe9stas\\b|\\b\\\xe9se\\b|\\b\\\xe9sa\\b|\\b\\\xe9sos\\b|\\b\\\xe9sas\\b|\\bs\\\xf3lo\\b|\\bcoktel\\b|\\bcocktel\\b|\\bconciente\\b|\\bcomenz\\\xe9\\b|\\bdesilucionarte\\b|\\bdijieron\\b|\\bempez\\\xe9\\b|\\bhize\\b|\\bilucionarte\\b|\\binconciente\\b|\\bquize\\b|\\bquizo\\b|\\bverguenza\\b|\\bNu\\\xf1ez\\b|\\bIvan\\b|\\bJapon\\b|\\bMonica\\b|\\bMaria\\b|\\bJose\\b|\\bRamon\\b|\\bGarcia\\b|\\bGonzalez\\b|\\bJesus\\b|\\bAlvarez\\b|\\bDamian\\b|\\bRene\\b|\\bNicolas\\b|\\bJonas\\b|\\bLopez\\b|\\bHernandez\\b|\\bBermudez\\b|\\bFernandez\\b|\\bSuarez\\b|\\bSofia\\b|\\bSeneca\\b|\\bTokyo\\b|\\bCanada\\b|\\bParis\\b|\\bTurquia\\b|\\bMexico\\b|\\bMejico\\b|\\bMatias\\b|\\bValentin\\b|\\bmejicano\\b|\\bmejicanos\\b|\\bmejicana\\b|\\bmejicanas\\b|\\bio\\b|\\bia\\b|\\bie\\b|\\bIo\\b|\\bIa\\b|\\bAI\\b|\\bIe\\b|\\bEI\\b|\\bsuba\\\ufb02uente\\b|\\ba\\\ufb02\\\xf3jalo\\b|\\bA\\\ufb02\\\xf3jalo\\b|\\bperdi\\b|\\bPodria\\b|\\bconfia\\b|\\bpasaria\\b|\\bPodias\\b|\\bresponsabke\\b|\\bTodavia\\b|\\benvien\\b|\\bQueria\\b|\\btio\\b|\\btraido\\b|\\bAsi\\b|\\belegi\\b|\\bhabria\\b|\\bencantaria\\b|\\bleido\\b|\\bconocias\\b|\\bharias\\b|\\bAqui\\b|\\bdecidi\\b|\\bmia\\b|\\bCrei\\b|\\bpodiamos\\b|\\bavisame\\b|\\bdebia\\b|\\bpensarias\\b|\\breuniamos\\b|\\bPO\\\xcf\\b|\\bvendria\\b|\\bcaida\\b|\\bvenian\\b|\\bcompa\\\xf1ias\\b|\\bleiste\\b|\\bLeiste\\b|\\bfiaria\\b|\\bHungria\\b|\\bfotografia\\b|\\bcafeteria\\b|\\bDigame\\b|\\bdebias\\b|\\btendria\\b|\\bC\\\xcfGO\\b|\\banteg\\b|\\bS\\\xf3Io\\b|\\bIlam\\\xe1ndola\\b|\\bC\\\xe1\\\ufb02at\\\xe9\\b|\\bIlamaste\\b|\\bdaria\\b|\\bIargaba\\b|\\bYati\\b|\\bquerias\\b|\\bIimpiarlo\\b|\\bIargado\\b|\\bgaleria\\b|\\bBartomeu\\b|\\bIocalizarlo\\b|\\bIl\\\xe1mame\\b'}},
+ 'srp': {'BeginLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'EndLines': {'data': OrderedDict(),
+                      'pattern': None},
+         'PartialLines': {'data': OrderedDict([(u'bi smo', u'bismo'), (u'dali je', u'da li je'), (u'dali si', u'da li si'), (u'Dali si', u'Da li si'), (u'Jel sam ti', u'Jesam li ti'), (u'Jel si', u'Jesi li'), (u"Jel' si", u'Jesi li'), (u"Je I'", u'Jesi li'), (u'Jel si to', u'Jesi li to'), (u"Jel' si to", u'Da li si to'), (u'jel si to', u'da li si to'), (u"jel' si to", u'jesi li to'), (u'Jel si ti', u'Da li si ti'), (u"Jel' si ti", u'Da li si ti'), (u'jel si ti', u'da li si ti'), (u"jel' si ti", u'da li si ti'), (u'jel ste ', u'jeste li '), (u'Jel ste', u'Jeste li'), (u"jel' ste ", u'jeste li '), (u"Jel' ste ", u'Jeste li '), (u'Jel su ', u'Jesu li '), (u'Jel da ', u'Zar ne'), (u'jel da ', u'zar ne'), (u"jel'da ", u'zar ne'), (u'Jeli sve ', u'Je li sve'), (u'Jeli on ', u'Je li on'), (u'Jeli ti ', u'Je li ti'), (u'jeli ti ', u'je li ti'), (u'Jeli to ', u'Je li to'), (u'Nebrini', u'Ne brini'), (u'nedaj', u'ne daj'), (u'ne \u0107u', u'ne\u0107u'), (u'Nemogu', u'Ne mogu'), (u'ne mogu', u'ne mogu'), (u'Nemora\u0161', u'Ne mora\u0161'), (u'od kako', u'otkako'), (u'Si dobro', u'Jesi li dobro'), (u'Svo vreme', u'Sve vrijeme'), (u'Svo vrijeme', u'Sve vrijeme'), (u'Cijelo vrijeme', u'Sve vrijeme')]),
+                          'pattern': None},
+         'PartialWordsAlways': {'data': OrderedDict(),
+                                'pattern': None},
+         'WholeLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'WholeWords': {'data': OrderedDict([(u'\u010du', u'\u0107u'), (u'\u010de\u0161', u'\u0107e\u0161'), (u'\u010de', u'\u0107e'), (u'\u0107\u0161', u'\u0107e\u0161'), (u'\u0107mo', u'\u0107emo'), (u'\u0107te', u'\u0107ete'), (u'\u010demo', u'\u0107emo'), (u'\u010dete', u'\u010dete'), (u'djete', u'dijete'), (u'Hey', u'Hej'), (u'hey', u'hej'), (u'htjeo', u'htio'), (u'Ho\u010de\u0161', u'Ho\u0107e\u0161'), (u'ho\u010de\u0161', u'ho\u0107e\u0161'), (u'i\u010di', u'i\u0107i'), (u'jel', u"je l'"), (u'Jel', u"Je l'"), (u'nedaj', u'ne daj'), (u'Rje\u0161it', u'Rije\u0161it'), (u'smjeo', u'smio'), (u'uop\u010de', u'uop\u0107e'), (u'valda', u'valjda'), (u'\u017eelila', u'\u017eeljela')]),
+                        'pattern': u'(um)\\b\\\u010du\\b|\\b\\\u010de\\\u0161\\b|\\b\\\u010de\\b|\\b\\\u0107\\\u0161\\b|\\b\\\u0107mo\\b|\\b\\\u0107te\\b|\\b\\\u010demo\\b|\\b\\\u010dete\\b|\\bdjete\\b|\\bHey\\b|\\bhey\\b|\\bhtjeo\\b|\\bHo\\\u010de\\\u0161\\b|\\bho\\\u010de\\\u0161\\b|\\bi\\\u010di\\b|\\bjel\\b|\\bJel\\b|\\bnedaj\\b|\\bRje\\\u0161it\\b|\\bsmjeo\\b|\\buop\\\u010de\\b|\\bvalda\\b|\\b\\\u017eelila\\b'}},
+ 'swe': {'BeginLines': {'data': OrderedDict([(u'Ln ', u'In '), (u'U ppfattat', u'Uppfattat')]),
+                        'pattern': u'(um)^\\bLn\\ \\b|\\bU\\ ppfattat\\b'},
+         'EndLines': {'data': OrderedDict(),
+                      'pattern': None},
+         'PartialLines': {'data': OrderedDict(),
+                          'pattern': None},
+         'PartialWordsAlways': {'data': OrderedDict(),
+                                'pattern': None},
+         'WholeLines': {'data': OrderedDict(),
+                        'pattern': None},
+         'WholeWords': {'data': OrderedDict([(u'l\xe2rt', u'l\xe4rt'), (u'hederv\xe5rda', u'hederv\xe4rda'), (u'storm\xe2stare', u'storm\xe4stare'), (u'Avf\xe2rd', u'Avf\xe4rd'), (u't\xe2lten', u't\xe4lten'), (u'\xe2rjag', u'\xe4r jag'), (u'\xe4rjag', u'\xe4r jag'), (u'j\xe2mlikar', u'j\xe4mlikar'), (u'Riskako\ufb02', u'Riskakor'), (u'Karamellen/', u'Karamellen'), (u'Lngen\xfcng', u'Ingenting'), (u'\xe4rju', u'\xe4r ju'), (u'S\xe1', u'S\xe5'), (u'n\xe4rjag', u'n\xe4r jag'), (u'alltjag', u'allt jag'), (u'g\xf6rjag', u'g\xf6r jag'), (u'trorjag', u'tror jag'), (u'varju', u'var ju'), (u'g\xf6rju', u'g\xf6r ju'), (u'kanju', u'kan ju'), (u'blirjag', u'blir jag'), (u's\xe4gerjag', u's\xe4ger jag'), (u'beh\xe5llerjag', u'beh\xe5ller jag'), (u'pr\xf8blem', u'problem'), (u'r\xe4ddadeju', u'r\xe4ddade ju'), (u'hon\xf8m', u'honom'), (u'Ln', u'In'), (u'sv\xe5r\ufb02\xf6rtad', u'sv\xe5rfl\xf6rtad'), (u'\xf8ch', u'och'), (u'\ufb02\xf6rtar', u'fl\xf6rtar'), (u'k\xe4nnerjag', u'k\xe4nner jag'), (u'\ufb02ickan', u'flickan'), (u'sn\xf8', u'sn\xf6'), (u'gerju', u'ger ju'), (u'k\xf8ntakter', u'kontakter'), (u'\xf8lycka', u'olycka'), (u'n\xf8lla', u'nolla'), (u'sinnenajublar', u'sinnena jublar'), (u'ijobbet', u'i jobbet'), (u'F\xe5rjag', u'F\xe5r jag'), (u'Ar', u'\xc4r'), (u'liggerju', u'ligger ju'), (u'um', u'om'), (u'lbland', u'Ibland'), (u'skjuterjag', u'skjuter jag'), (u'Vadd\xe5', u'Vad d\xe5'), (u'pratarj\xe4mt', u'pratar j\xe4mt'), (u'harju', u'har ju'), (u'sitterjag', u'sitter jag'), (u'h\xe4\ufb02a', u'h\xe4rja'), (u's\ufb01\xe4l', u'stj\xe4l'), (u'F\xd6U', u'F\xf6lj'), (u'varf\xf6rjag', u'varf\xf6r jag'), (u's\ufb01\xe4rna', u'stj\xe4rna'), (u'b\xf6\ufb02ar', u'b\xf6rjar'), (u'b\xf6\ufb02an', u'b\xf6rjan'), (u'st\xe4ri', u'st\xe5r'), (u'p\xe4', u'p\xe5'), (u'harjag', u'har jag'), (u'attjag', u'att jag'), (u'Verkarjag', u'Verkar jag'), (u'K\xe4nnerjag', u'K\xe4nner jag'), (u'd\xe4rjag', u'd\xe4r jag'), (u'tu\ufb01', u'tuff'), (u'lurarjag', u'lurar jag'), (u'varj\xe4ttebra', u'var j\xe4ttebra'), (u'allvan', u'allvar'), (u'deth\xe4r', u'det h\xe4r'), (u'va\ufb02e', u'varje'), (u'F\xf6Uer', u'F\xf6ljer'), (u'personalm\xf6tetl', u'personalm\xf6tet!'), (u'harjust', u'har just'), (u'\xe4rj\xe4tteduktig', u'\xe4r j\xe4tteduktig'), (u'd\xe4rja', u'd\xe4r ja'), (u'lngen\xfcng', u'lngenting'), (u'iluften', u'i luften'), (u'\xf6sen', u'\xf6ser'), (u'tv\xe2', u'tv\xe5'), (u'Uejerna', u'Tjejerna'), (u'h\xe5n*', u'h\xe5rt'), (u'\xc4rjag', u'\xc4r jag'), (u'keL', u'Okej'), (u'F\xf6rjag', u'F\xf6r jag'), (u'varj\xe4ttekul', u'var j\xe4ttekul'), (u'k\xe4mpan', u'k\xe4mpar'), (u'mycketjobb', u'mycket jobb'), (u'Uus', u'ljus'), (u'serjag', u'ser jag'), (u'vetjag', u'vet jag'), (u'f\xe5rjag', u'f\xe5r jag'), (u'hurjag', u'hur jag'), (u'f\xf6rs\xf6kerjag', u'f\xf6rs\xf6ker jag'), (u't\xe1nagel', u't\xe5nagel'), (u'va\xfce', u'varje'), (u'Uudet', u'ljudet'), (u'amhopa', u'allihopa'), (u'V\xe4\xfc', u'V\xe4lj'), (u'g\xe4ri', u'g\xe5r'), (u'r\xf6d\xfcus', u'r\xf6dljus'), (u'Uuset', u'ljuset'), (u'Rid\xe0n', u'Rid\xe5n'), (u'vi\xfca', u'vilja'), (u'g\xe5ri', u'g\xe5r i'), (u'Hurd\xe5', u'Hur d\xe5'), (u'inter\\/juar', u'intervjuar'), (u'menarjag', u'menar jag'), (u'spyrjag', u'spyr jag'), (u'bri\xfcera', u'briljera'), (u'N\xe4rjag', u'N\xe4r jag'), (u'ner\\/\xf6s', u'nerv\xf6s'), (u'ilivets', u'i livets'), (u'n\xe4got', u'n\xe5got'), (u'p\xe0', u'p\xe5'), (u'Lnnan', u'Innan'), (u'Uf', u'Ut'), (u'lnnan', u'Innan'), (u'D\xe0ren', u'D\xe5ren'), (u'F\xe0rjag', u'F\xe5r jag'), (u'Vad\xe4rdetd\xe4L', u'Vad \xe4r det d\xe4r'), (u'sm\xe0tjuv', u'sm\xe5tjuv'), (u't\xe0gr\xe5nare', u't\xe5gr\xe5nare'), (u'dit\xe0t', u'dit\xe5t'), (u's\xe4', u's\xe5'), (u'v\xe0rdsl\xf6sa', u'v\xe5rdsl\xf6sa'), (u'n\xe0n', u'n\xe5n'), (u'kommerjag', u'kommer jag'), (u'\xe4rj\xe4ttebra', u'\xe4r j\xe4ttebra'), (u'\xe4rj\xe4vligt', u'\xe4r j\xe4vligt'), (u'\xe0kerjag', u'\xe5ker jag'), (u'ellerjapaner', u'eller japaner'), (u'attjaga', u'att jaga'), (u'eften', u'efter'), (u'h\xe4stan', u'h\xe4star'), (u'Lntensivare', u'Intensivare'), (u'fr\xe0garjag', u'fr\xe5gar jag'), (u'pen/ers', u'pervers'), (u'r\xe0barkade', u'r\xe5barkade'), (u'styrkon', u'styrkor'), (u'Dif\xe5f', u'Dit\xe5t'), (u'h\xe4nden', u'h\xe4nder'), (u'f\xf6\ufb01a', u'f\xf6lja'), (u'Idioten/', u'Idioter!'), (u'Varf\xf6rjagade', u'Varf\xf6r jagade'), (u'd\xe4rf\xf6rjag', u'd\xe4rf\xf6r jag'), (u'forjag', u'for jag'), (u'Iivsgladje', u'livsgl\xe4dje'), (u'narjag', u'n\xe4r jag'), (u'sajag', u'sa jag'), (u'genastja', u'genast ja'), (u'rockument\xe0ren', u'rockument\xe4ren'), (u'turne', u'turn\xe9'), (u'fickjag', u'fick jag'), (u'sager', u's\xe4ger'), (u'Ijush\xe5rig', u'ljush\xe5rig'), (u'tradg\xe5rdsolycka', u'tr\xe4dg\xe5rdsolycka'), (u'kvavdes', u'kv\xe4vdes'), (u'd\xe0rja', u'd\xe4r ja'), (u'hedersgaster', u'hedersg\xe4ster'), (u'Nar', u'N\xe4r'), (u'smaki\xf6sa', u'smakl\xf6sa'), (u'lan', u'Ian'), (u'Lan', u'Ian'), (u'eri', u'er i'), (u'universitetsamne', u'universitets\xe4mne'), (u'garna', u'g\xe4rna'), (u'Line truncated
+                        'pattern': u"(um)\\bl\\\xe2rt\\b|\\bhederv\\\xe5rda\\b|\\bstorm\\\xe2stare\\b|\\bAvf\\\xe2rd\\b|\\bt\\\xe2lten\\b|\\b\\\xe2rjag\\b|\\b\\\xe4rjag\\b|\\bj\\\xe2mlikar\\b|\\bRiskako\\\ufb02\\b|\\bKaramellen\\/\\b|\\bLngen\\\xfcng\\b|\\b\\\xe4rju\\b|\\bS\\\xe1\\b|\\bn\\\xe4rjag\\b|\\balltjag\\b|\\bg\\\xf6rjag\\b|\\btrorjag\\b|\\bvarju\\b|\\bg\\\xf6rju\\b|\\bkanju\\b|\\bblirjag\\b|\\bs\\\xe4gerjag\\b|\\bbeh\\\xe5llerjag\\b|\\bpr\\\xf8blem\\b|\\br\\\xe4ddadeju\\b|\\bhon\\\xf8m\\b|\\bLn\\b|\\bsv\\\xe5r\\\ufb02\\\xf6rtad\\b|\\b\\\xf8ch\\b|\\b\\\ufb02\\\xf6rtar\\b|\\bk\\\xe4nnerjag\\b|\\b\\\ufb02ickan\\b|\\bsn\\\xf8\\b|\\bgerju\\b|\\bk\\\xf8ntakter\\b|\\b\\\xf8lycka\\b|\\bn\\\xf8lla\\b|\\bsinnenajublar\\b|\\bijobbet\\b|\\bF\\\xe5rjag\\b|\\bAr\\b|\\bliggerju\\b|\\bum\\b|\\blbland\\b|\\bskjuterjag\\b|\\bVadd\\\xe5\\b|\\bpratarj\\\xe4mt\\b|\\bharju\\b|\\bsitterjag\\b|\\bh\\\xe4\\\ufb02a\\b|\\bs\\\ufb01\\\xe4l\\b|\\bF\\\xd6U\\b|\\bvarf\\\xf6rjag\\b|\\bs\\\ufb01\\\xe4rna\\b|\\bb\\\xf6\\\ufb02ar\\b|\\bb\\\xf6\\\ufb02an\\b|\\bst\\\xe4ri\\b|\\bp\\\xe4\\b|\\bharjag\\b|\\battjag\\b|\\bVerkarjag\\b|\\bK\\\xe4nnerjag\\b|\\bd\\\xe4rjag\\b|\\btu\\\ufb01\\b|\\blurarjag\\b|\\bvarj\\\xe4ttebra\\b|\\ballvan\\b|\\bdeth\\\xe4r\\b|\\bva\\\ufb02e\\b|\\bF\\\xf6Uer\\b|\\bpersonalm\\\xf6tetl\\b|\\bharjust\\b|\\b\\\xe4rj\\\xe4tteduktig\\b|\\bd\\\xe4rja\\b|\\blngen\\\xfcng\\b|\\biluften\\b|\\b\\\xf6sen\\b|\\btv\\\xe2\\b|\\bUejerna\\b|\\bh\\\xe5n\\*\\b|\\b\\\xc4rjag\\b|\\bkeL\\b|\\bF\\\xf6rjag\\b|\\bvarj\\\xe4ttekul\\b|\\bk\\\xe4mpan\\b|\\bmycketjobb\\b|\\bUus\\b|\\bserjag\\b|\\bvetjag\\b|\\bf\\\xe5rjag\\b|\\bhurjag\\b|\\bf\\\xf6rs\\\xf6kerjag\\b|\\bt\\\xe1nagel\\b|\\bva\\\xfce\\b|\\bUudet\\b|\\bamhopa\\b|\\bV\\\xe4\\\xfc\\b|\\bg\\\xe4ri\\b|\\br\\\xf6d\\\xfcus\\b|\\bUuset\\b|\\bRid\\\xe0n\\b|\\bvi\\\xfca\\b|\\bg\\\xe5ri\\b|\\bHurd\\\xe5\\b|\\binter\\\\\\/juar\\b|\\bmenarjag\\b|\\bspyrjag\\b|\\bbri\\\xfcera\\b|\\bN\\\xe4rjag\\b|\\bner\\\\\\/\\\xf6s\\b|\\bilivets\\b|\\bn\\\xe4got\\b|\\bp\\\xe0\\b|\\bLnnan\\b|\\bUf\\b|\\blnnan\\b|\\bD\\\xe0ren\\b|\\bF\\\xe0rjag\\b|\\bVad\\\xe4rdetd\\\xe4L\\b|\\bsm\\\xe0tjuv\\b|\\bt\\\xe0gr\\\xe5nare\\b|\\bdit\\\xe0t\\b|\\bs\\\xe4\\b|\\bv\\\xe0rdsl\\\xf6sa\\b|\\bn\\\xe0n\\b|\\bkommerjag\\b|\\b\\\xe4rj\\\xe4ttebra\\b|\\b\\\xe4rj\\\xe4vligt\\b|\\b\\\xe0kerjag\\b|\\bellerjapaner\\b|\\battjaga\\b|\\beften\\b|\\bh\\\xe4stan\\b|\\bLntensivare\\b|\\bfr\\\xe0garjag\\b|\\bpen\\/ers\\b|\\br\\\xe0barkade\\b|\\bstyrkon\\b|\\bDif\\\xe5f\\b|\\bh\\\xe4nden\\b|\\bf\\\xf6\\\ufb01a\\b|\\bIdioten\\/\\b|\\bVarf\\\xf6rjagade\\b|\\bd\\\xe4rf\\\xf6rjag\\b|\\bforjag\\b|\\bIivsgladje\\b|\\bnarjag\\b|\\bsajag\\b|\\bgenastja\\b|\\brockument\\\xe0ren\\b|\\bturne\\b|\\bfickjag\\b|\\bsager\\b|\\bIjush\\\xe5rig\\b|\\btradg\\\xe5rdsolycka\\b|\\bkvavdes\\b|\\bd\\\xe0rja\\b|\\bhedersgaster\\b|\\bNar\\b|\\bsmaki\\\xf6sa\\b|\\blan\\b|\\bLan\\b|\\beri\\b|\\buniversitetsamne\\b|\\bgarna\\b|\\bar\\b|\\bbaltdjur\\b|\\bvarjag\\b|\\b\\\xe0r\\b|\\bf\\\xf6rf\\\xf6rst\\\xe0rkare\\b|\\barjattespeciell\\b|\\bh\\\xe0rg\\\xe5r\\b|\\bIa\\b|\\bIimousinen\\b|\\bkrickettra\\b|\\bh\\\xe5rdrockv\\\xe0rlden\\b|\\btr\\\xe0bit\\b|\\bMellanvastern\\b|\\barju\\b|\\bturnen\\b|\\bkanns\\b|\\bbattre\\b|\\bv\\\xe0rldsturne\\b|\\bdar\\b|\\bsj\\\xe0lvant\\\xe0nder\\b|\\bjattelange\\b|\\bberattade\\b|\\bS\\\xe4\\b|\\bvandpunkten\\b|\\bN\\\xe0rjag\\b|\\blasa\\b|\\bskitl\\\xe0skigt\\b|\\bsambandsv\\\xe0g\\b|\\bvaldigt\\b|\\bStamga\\\ufb01el\\b|\\b\\\xe0rjag\\b|\\btajming\\b|\\butg\\\xe4ng\\b|\\bH\\\xe0r\\\xe5t\\b|\\bh\\\xe0r\\\xe5t\\b|\\banvander\\b|\\bharjobbat\\b|\\bimageide\\b|\\bkla\\\ufb01en\\b|\\bsjalv\\b|\\bdvarg\\b|\\bdetjag\\b|\\bdvargarna\\b|\\bfantasiv\\\xe0rld\\b|\\b\\\ufb01olliga\\b|\\bmandoiinstr\\\xe0ngar\\b|\\bmittjobb\\b|\\bSkajag\\b|\\blandari\\b|\\bgang\\b|\\bDetjag\\b|\\bNarmre\\b|\\bI\\\xe5tjavelni\\b|\\bH\\\xe5llerjag\\b|\\bvisionarer\\b|\\bT\\\xfclvad\\b|\\bmilit\\\xe0rbas\\b|\\bjattegiada\\b|\\bFastjag\\b|\\bs\\\xe5jag\\b|\\brockvarlden\\b|\\bsaknarjag\\b|\\ballafall\\b|\\b\\\ufb01anta\\b|\\bKr\\\xe0ma\\b|\\bstammer\\b|\\bbudb\\\xe0rare\\b|\\bIivsfiiosofi\\b|\\bf\\\xf6rj\\\xe4mnan\\b|\\bgillarjag\\b|\\bIarvat\\b|\\bklararjag\\b|\\bhatta\\\ufb01\\'\\\xe0r\\b|\\bD\\\xe0\\b|\\bupp\\\ufb01nna\\b|\\bR\\\xe0ttf\\\xe5glar\\b|\\bSv\\\xe4\\\xfcboda\\b|\\bP\\\xe5b\\\xf6\\\ufb02ar\\b|\\bslutarju\\b|\\bni\\\ufb01skebu\\\xfcken\\b|\\bh\\\xe4rj\\\xe4keln\\b|\\bH\\\xdfppa\\b|\\bf\\\xf6rst\\\xf6rds\\b|\\bvarj\\\xe4ttegoda\\b|\\bKor\\\\\\/\\b|\\bbr\\\xfcl\\\xe9el\\b|\\bHei\\b|\\b\\\xe4lskarjordgubbsglass\\b|\\bSn\\\xf6bom\\b|\\bSn\\\xf6boH\\b|\\bSn\\\xf6bol\\b|\\bsn\\\xf6boH\\b|\\bL\\\xe4ggerp\\\xe5\\b|\\blnge\\\ufb02\\b|\\bS\\\xe4gerj\\\xe4ttesmarta\\b|\\bdopplen\\/\\\xe4derradar\\b|\\bs\\\xe4kertj\\\xe4ttefin\\b|\\b\\\xe4rj\\\xe4ttefin\\b|\\bverkarju\\b|\\bblirju\\b|\\bkor\\\\\\/\\b|\\bnaturkatastro\\\ufb01\\b|\\bstickerjag\\b|\\bj\\\xe4ttebu\\\ufb01\\\xe9\\b|\\bbe\\\ufb01nner\\b|\\bSp\\\ufb02ng\\b|\\btrec\\\ufb01e\\b|\\bryckerjag\\b|\\bskullejag\\Line truncated
+for lang, grps in data.iteritems():
+    for grp in grps.iterkeys():
+        if data[lang][grp]["pattern"]:
+            data[lang][grp]["pattern"] = re.compile(data[lang][grp]["pattern"])

Contents/Libraries/Shared/subzero/modification/dictionaries/make_data.py

@@ -0,0 +1,59 @@
+# coding=utf-8
+
+import re
+import os
+import pprint
+from collections import OrderedDict
+
+from bs4 import BeautifulSoup
+
+TEMPLATE = """\
+import re
+from collections import OrderedDict
+data = """
+
+TEMPLATE_END = """\
+
+for lang, grps in data.iteritems():
+    for grp in grps.iterkeys():
+        if data[lang][grp]["pattern"]:
+            data[lang][grp]["pattern"] = re.compile(data[lang][grp]["pattern"])
+"""
+
+if __name__ == "__main__":
+    cur_dir = os.path.dirname(os.path.realpath(__file__))
+    xml_dir = os.path.join(cur_dir, "xml")
+    file_list = os.listdir(xml_dir)
+
+    data = {}
+
+    for fn in file_list:
+        if fn.endswith("_OCRFixReplaceList.xml"):
+            lang = fn.split("_")[0]
+            soup = BeautifulSoup(open(os.path.join(xml_dir, fn)), "xml")
+
+            fetch_data = (
+                    # group, item_name, pattern
+                    ("WholeLines", "Line", None),
+                    ("WholeWords", "Word", lambda d: (ur"(um)" + u"|".join([ur'\b' + re.escape(k) + ur'\b' for k in d.keys()])) if d else None),
+                    ("PartialWordsAlways", "WordPart", None),
+                    ("PartialLines", "LinePart", None),
+                    ("BeginLines", "Beginning", lambda d: (ur"(um)^"+u"|".join([ur'\b' + re.escape(k) + ur'\b' for k in d.keys()])) if d else None),
+                    ("EndLines", "Ending", lambda d: (ur"(um)" + u"|".join([ur'\b' + re.escape(k) + ur'\b' for k in d.keys()]) + ur"$") if d else None,),
+            )
+
+            data[lang] = dict((grp, {"data": OrderedDict(), "pattern": None}) for grp, item_name, pattern in fetch_data)
+
+            for grp, item_name, pattern in fetch_data:
+                for grp_data in soup.find_all(grp):
+                    for line in grp_data.find_all(item_name):
+                        data[lang][grp]["data"][line["from"]] = line["to"]
+
+                if pattern:
+                    data[lang][grp]["pattern"] = pattern(data[lang][grp]["data"])
+
+    f = open(os.path.join(cur_dir, "data.py"), "w+")
+    f.write(TEMPLATE)
+    f.write(pprint.pformat(data, width=1))
+    f.write(TEMPLATE_END)
+    f.close()

Contents/Libraries/Shared/subzero/modification/dictionaries/test_data.py

@@ -0,0 +1,10 @@
+# coding=utf-8
+
+from babelfish import Language
+from data import data
+
+#for lang, data in data.iteritems():
+#    print Language.fromietf(lang).alpha2
+
+for find, rep in data["dan"].iteritems():
+    print find, rep

Contents/Libraries/Shared/subzero/modification/dictionaries/xml/dan_OCRFixReplaceList.xml

@@ -0,0 +1,638 @@
+<OCRFixReplaceList>
+  <WholeWords>
+    <Word from="Haner" to="Han er" />
+    <Word from="JaveL" to="Javel" />
+    <Word from="Pa//e" to="Palle" />
+    <Word from="bffte" to="bitte" />
+    <Word from="Utro//gt" to="Utroligt" />
+    <Word from="Kommerdu" to="Kommer du" />
+    <Word from="smi/er" to="smiler" />
+    <Word from="/eg" to="leg" />
+    <Word from="harvinger" to="har vinger" />
+    <Word from="/et" to="let" />
+    <Word from="erjeres" to="er jeres" />
+    <Word from="hardet" to="har det" />
+    <Word from="tænktjer" to="tænkt jer" />
+    <Word from="erjo" to="er jo" />
+    <Word from="sti/" to="stil" />
+    <Word from="Iappe" to="lappe" />
+    <Word from="Beklagelç" to="Beklager," />
+    <Word from="vardet" to="var det" />
+    <Word from="afden" to="af den" />
+    <Word from="snupperjeg" to="snupper jeg" />
+    <Word from="ikkejeg" to="ikke jeg" />
+    <Word from="bliverjeg" to="bliver jeg" />
+    <Word from="hartravit" to="har travlt" />
+    <Word from="pandekagef/ag" to="pandekageflag" />
+    <Word from="Stormvarsell" to="Stormvarsel!" />
+    <Word from="stormvejn" to="stormvejr." />
+    <Word from="morgenkomp/et" to="morgenkomplet" />
+    <Word from="/yv" to="lyv" />
+    <Word from="varjo" to="var jo" />
+    <Word from="/eger" to="leger" />
+    <Word from="harjeg" to="har jeg" />
+    <Word from="havdejeg" to="havde jeg" />
+    <Word from="hvorjeg" to="hvor jeg" />
+    <Word from="nårjeg" to="når jeg" />
+    <Word from="gårvi" to="går vi" />
+    <Word from="atjeg" to="at jeg" />
+    <Word from="isine" to="i sine" />
+    <Word from="fårjeg" to="får jeg" />
+    <Word from="kærtighed" to="kærlighed" />
+    <Word from="skullejeg" to="skulle jeg" />
+    <Word from="laest" to="læst" />
+    <Word from="laese" to="læse" />
+    <Word from="gørjeg" to="gør jeg" />
+    <Word from="gørvi" to="gør vi" />
+    <Word from="angrerjo" to="angrer jo" />
+    <Word from="Hvergang" to="Hver gang" />
+    <Word from="erder" to="er der" />
+    <Word from="villetilgive" to="ville tilgive" />
+    <Word from="fieme" to="fjeme" />
+    <Word from="genopståri" to="genopstår i" />
+    <Word from="svigtejer" to="svigte jer" />
+    <Word from="kommernu" to="kommer nu" />
+    <Word from="nårman" to="når man" />
+    <Word from="erfire" to="er fire" />
+    <Word from="Hvorforfinderdu" to="Hvorfor finder du" />
+    <Word from="undertigt" to="underligt" />
+    <Word from="itroen" to="i troen" />
+    <Word from="erløgnt" to="er løgn!" />
+    <Word from="gørden" to="gør den" />
+    <Word from="forhelvede" to="for helvede" />
+    <Word from="hjpe" to="hjælpe" />
+    <Word from="togeti" to="toget i" />
+    <Word from="Måjeg" to="Må jeg" />
+    <Word from="savnerjer" to="savner jer" />
+    <Word from="erjeg" to="er jeg" />
+    <Word from="vaere" to="være" />
+    <Word from="geme" to="gerne" />
+    <Word from="trorpå" to="tror på" />
+    <Word from="forham" to="for ham" />
+    <Word from="afham" to="af ham" />
+    <Word from="harjo" to="har jo" />
+    <Word from="ovemafiet" to="overnattet" />
+    <Word from="begaefiighed" to="begærlighed" />
+    <Word from="sy’g" to="syg" />
+    <Word from="Imensjeg" to="Imens jeg" />
+    <Word from="bliverdu" to="bliver du" />
+    <Word from="fiser" to="fiser" />
+    <Word from="manipuierer" to="manipulerer" />
+    <Word from="forjeg" to="for jeg" />
+    <Word from="iivgivendefor" to="livgivende for" />
+    <Word from="formig" to="for mig" />
+    <Word from="Hardu" to="Har du" />
+    <Word from="fornold" to="forhold" />
+    <Word from="defrelste" to="de frelste" />
+    <Word from="Såjeg" to="Så jeg" />
+    <Word from="varjeg" to="var jeg" />
+    <Word from="gørved" to="gør ved" />
+    <Word from="kalderjeg" to="kalder jeg" />
+    <Word from="flytte" to="flytte" />
+    <Word from="handlerdet" to="handler det" />
+    <Word from="trorjeg" to="tror jeg" />
+    <Word from="flytter" to="flytter" />
+    <Word from="soverjeg" to="sover jeg" />
+    <Word from="finderud" to="finder ud" />
+    <Word from="naboerpå" to="naboer på" />
+    <Word from="ervildt" to="er vildt" />
+    <Word from="væreher" to="være her" />
+    <Word from="hyggerjer" to="hygger jer" />
+    <Word from="borjo" to="bor jo" />
+    <Word from="kommerikke" to="kommer ikke" />
+    <Word from="folkynde" to="forkynde" />
+    <Word from="farglad" to="far glad" />
+    <Word from="misterjeg" to="mister jeg" />
+    <Word from="fint" to="fint" />
+    <Word from="Harl" to="Har I" />
+    <Word from="bedejer" to="bede jer" />
+    <Word from="synesjeg" to="synes jeg" />
+    <Word from="vartil" to="var til" />
+    <Word from="eren" to="er en" />
+    <Word from="\Al" to="Vil" />
+    <Word from="\A" to="Vi" />
+    <Word from="fjeme" to="fjerne" />
+    <Word from="Iigefyldt" to="lige fyldt" />
+    <Word from="ertil" to="er til" />
+    <Word from="fafiigt" to="farligt" />
+    <Word from="finder" to="finder" />
+    <Word from="findes" to="findes" />
+    <Word from="irettesaefielse" to="irettesættelse" />
+    <Word from="ermed" to="er med" />
+    <Word from="èn" to="én" />
+    <Word from="gikjoi" to="gik jo i" />
+    <Word from="Hvisjeg" to="Hvis jeg" />
+    <Word from="ovemafier" to="overnatter" />
+    <Word from="hoident" to="holdent" />
+    <Word from="\Adne" to="Vidne" />
+    <Word from="fori" to="for i" />
+    <Word from="vei" to="vel" />
+    <Word from="savnerjerjo" to="savner jer jo" />
+    <Word from="elskerjer" to="elsker jer" />
+    <Word from="harløjet" to="har løjet" />
+    <Word from="eri" to="er i" />
+    <Word from="fiende" to="fjende" />
+    <Word from="derjo" to="der jo" />
+    <Word from="sigerjo" to="siger jo" />
+    <Word from="menerjeg" to="mener jeg" />
+    <Word from="Harjeg" to="Har jeg" />
+    <Word from="sigerjeg" to="siger jeg" />
+    <Word from="splitterjeg" to="splitter jeg" />
+    <Word from="erjournalist" to="er journalist" />
+    <Word from="erjoumalist" to="er journalist" />
+    <Word from="Forjeg" to="For jeg" />
+    <Word from="gârjeg" to="går jeg" />
+    <Word from="Nârjeg" to="Når jeg" />
+    <Word from="afllom" to="afkom" />
+    <Word from="farerjo" to="farer jo" />
+    <Word from="tagerjeg" to="tager jeg" />
+    <Word from="Virkerjeg" to="Virker jeg" />
+    <Word from="morerjer" to="morer jer" />
+    <Word from="kommerjo" to="kommer jo" />
+    <Word from="istand" to="i stand" />
+    <Word from="bøm" to="børn" />
+    <Word from="frygterjeg" to="frygter jeg" />
+    <Word from="kommerjeg" to="kommer jeg" />
+    <Word from="eriournalistelev" to="er journalistelev" />
+    <Word from="harfat" to="har fat" />
+    <Word from="fårfingre" to="får fingre" />
+    <Word from="slârjeg" to="slår jeg" />
+    <Word from="bam" to="barn" />
+    <Word from="erjournalistelev" to="er journalistelev" />
+    <Word from="politietjo" to="politiet jo" />
+    <Word from="elskerjo" to="elsker jo" />
+    <Word from="vari" to="var i" />
+    <Word from="fornemmerjeres" to="fornemmer jeres" />
+    <Word from="udklækketl" to="udklækket!" />
+    <Word from="í" to="i" />
+    <Word from="nyi" to="ny i" />
+    <Word from="Iumijelse" to="fornøjelse" />
+    <Word from="vures" to="vores" />
+    <Word from="I/Vashíngtan" to="Washington" />
+    <Word from="opleverjeg" to="oplever jeg" />
+    <Word from="PANTELÃNER" to="PANTELÅNER" />
+    <Word from="Gudmurgen" to="Godmorgen" />
+    <Word from="SKYDEVÃBEN" to="SKYDEVÅBEN" />
+    <Word from="PÃLIDELIG" to="PÅLIDELIG" />
+    <Word from="avertalte" to="overtalte" />
+    <Word from="Omsíder" to="Omsider" />
+    <Word from="lurtebåd" to="lortebåd" />
+    <Word from="Telrslning" to="Tekstning" />
+    <Word from="miUø" to="miljø" />
+    <Word from="gåri" to="går i" />
+    <Word from="Fan/el" to="Farvel" />
+    <Word from="abefiæs" to="abefjæs" />
+    <Word from="hartalt" to="har talt" />
+    <Word from="\Årkelig" to="Virkelig" />
+    <Word from="beklagerjeg" to="beklager jeg" />
+    <Word from="Nårjeg" to="Når jeg" />
+    <Word from="rnaend" to="mænd" />
+    <Word from="vaskebjorn" to="vaskebjørn" />
+    <Word from="Ivil" to="I vil" />
+    <Word from="besog" to="besøg" />
+    <Word from="Vaer" to="Vær" />
+    <Word from="Undersogte" to="Undersøgte" />
+    <Word from="modte" to="mødte" />
+    <Word from="toj" to="tøj" />
+    <Word from="fodt" to="født" />
+    <Word from="gore" to="gøre" />
+    <Word from="provede" to="prøvede" />
+    <Word from="forste" to="første" />
+    <Word from="igang" to="i gang" />
+    <Word from="ligenu" to="lige nu" />
+    <Word from="clet" to="det" />
+    <Word from="Strombell" to="Strombel!" />
+    <Word from="tmvlt" to="travlt" />
+    <Word from="studererjournalistik" to="studerer journalistik" />
+    <Word from="inforrnererjeg" to="informerer jeg" />
+    <Word from="omkfing" to="omkring" />
+    <Word from="tilAsgård" to="til Asgård" />
+    <Word from="Kederjeg" to="Keder jeg" />
+    <Word from="jaettetamp" to="jættetamp" />
+    <Word from="erjer" to="er jer" />
+    <Word from="atjulehygge" to="at julehygge" />
+    <Word from="Ueneste" to="tjeneste" />
+    <Word from="foltsaetter" to="fortsætter" />
+    <Word from="A/ice" to="Alice" />
+    <Word from="tvivlerjeg" to="tvivler jeg" />
+    <Word from="henterjer" to="henter jer" />
+    <Word from="forstårjeg" to="forstår jeg" />
+    <Word from="hvisjeg" to="hvis jeg" />
+    <Word from="/ært" to="lært" />
+    <Word from="vfgtrgt" to="vigtigt" />
+    <Word from="hurtigtjeg" to="hurtigt jeg" />
+    <Word from="kenderjo" to="kender jo" />
+    <Word from="seiv" to="selv" />
+    <Word from="/ægehuset" to="lægehuset" />
+    <Word from="herjo" to="her jo" />
+    <Word from="stolerjeg" to="stoler jeg" />
+    <Word from="digi" to="dig i" />
+    <Word from="taberi" to="taber i" />
+    <Word from="slårjeres" to="slår jeres" />
+    <Word from="laere" to="lære" />
+    <Word from="trænerwushu" to="træner wushu" />
+    <Word from="efterjeg" to="efter jeg" />
+    <Word from="efier" to="efter" />
+    <Word from="dui" to="du i" />
+    <Word from="afien" to="aften" />
+    <Word from="bliveri" to="bliver i" />
+    <Word from="acceptererjer" to="accepterer jer" />
+    <Word from="drikkerjo" to="drikker jo" />
+    <Word from="fianjin" to="Tianjin" />
+    <Word from="erlænge" to="er længe" />
+    <Word from="erikke" to="er ikke" />
+    <Word from="medjer" to="med jer" />
+    <Word from="Tmykke" to="Tillykke" />
+    <Word from="'fianjins" to="Tianjins" />
+    <Word from="Mesteri" to="Mester i" />
+    <Word from="sagdetil" to="sagde til" />
+    <Word from="indei" to="inde i" />
+    <Word from="ofie" to="ofte" />
+    <Word from="'filgiv" to="Tilgiv" />
+    <Word from="Lfår" to="I får" />
+    <Word from="viserjer" to="viser jer" />
+    <Word from="Rejsjerblot" to="Rejs jer blot" />
+    <Word from="'fillad" to="Tillad" />
+    <Word from="iiiiefinger" to="lillefinger" />
+    <Word from="VILOMFATTE" to="VIL OMFATTE" />
+    <Word from="mofio" to="motto" />
+    <Word from="gørjer" to="gør jer" />
+    <Word from="gifi" to="gift" />
+    <Word from="hardu" to="har du" />
+    <Word from="gifi" to="gift" />
+    <Word from="Iaeggerjeg" to="lægger jeg" />
+    <Word from="iet" to="i et" />
+    <Word from="sv/yte" to="svigte" />
+    <Word from="ti/" to="til" />
+    <Word from="Wdal" to="Vidal" />
+    <Word from="fiået" to="fået" />
+    <Word from="Hvo/for" to="Hvorfor" />
+    <Word from="hellerikke" to="heller ikke" />
+    <Word from="Wlle" to="Ville" />
+    <Word from="dr/ver" to="driver" />
+    <Word from="V\fllliam" to="William" />
+    <Word from="V\fllliams" to="Williams" />
+    <Word from="Vkfilliam" to="William" />
+    <Word from="vådejakke" to="våde jakke" />
+    <Word from="kæfll" to="kæft!" />
+    <Word from="sagdejeg" to="sagde jeg" />
+    <Word from="oven/ejet" to="overvejet" />
+    <Word from="karameisauce" to="karamelsauce" />
+    <Word from="Lfølgejødisk" to="Ifølge jødisk" />
+    <Word from="blevjo" to="blev jo" />
+    <Word from="asiateri" to="asiater i" />
+    <Word from="erV\fllliam" to="er William" />
+    <Word from="lidtflov" to="lidt flov" />
+    <Word from="sagdejo" to="sagde jo" />
+    <Word from="erlige" to="er lige" />
+    <Word from="Vtfilliam" to="William" />
+    <Word from="WfiII" to="Will" />
+    <Word from="afldarede" to="afklarede" />
+    <Word from="hjæiperjeg" to="hjælper jeg" />
+    <Word from="laderjeg" to="lader jeg" />
+    <Word from="Hândledsbeskyttere" to="Håndledsbeskyttere" />
+    <Word from="Lsabels" to="Isabels" />
+    <Word from="Gørjeg" to="Gør jeg" />
+    <Word from="mâjeg" to="må jeg" />
+    <Word from="ogjeg" to="og jeg" />
+    <Word from="gjordejeg" to="gjorde jeg" />
+    <Word from="villejeg" to="ville jeg" />
+    <Word from="Vlfllliams" to="Williams" />
+    <Word from="Dajeg" to="Da jeg" />
+    <Word from="iorden" to="i orden" />
+    <Word from="fandtjeg" to="fandt jeg" />
+    <Word from="Tilykke" to="Tillykke" />
+    <Word from="kørerjer" to="kører jer" />
+    <Word from="gøfjeg" to="gør jeg" />
+    <Word from="Selvflgelig" to="Selvfølgelig" />
+    <Word from="fdder" to="fadder" />
+    <Word from="bnfaldt" to="bønfaldt" />
+    <Word from="t\/ehovedede" to="tvehovedede" />
+    <Word from="EIler" to="Eller" />
+    <Word from="ringerjeg" to="ringer jeg" />
+    <Word from="blevvæk" to="blev væk" />
+    <Word from="stárjeg" to="står jeg" />
+    <Word from="varforbi" to="var forbi" />
+    <Word from="harfortalt" to="har fortalt" />
+    <Word from="iflere" to="i flere" />
+    <Word from="tørjeg" to="tør jeg" />
+    <Word from="kunnejeg" to="kunne jeg" />
+    <Word from="má" to="må" />
+    <Word from="hartænkt" to="har tænkt" />
+    <Word from="Fárjeg" to="Får jeg" />
+    <Word from="afdelingervar" to="afdelinger var" />
+    <Word from="0rd" to="ord" />
+    <Word from="pástá" to="påstå" />
+    <Word from="gráharet" to="gråharet" />
+    <Word from="varforbløffende" to="var forbløffende" />
+    <Word from="holdtjeg" to="holdt jeg" />
+    <Word from="hængerjo" to="hænger jo" />
+    <Word from="fikjeg" to="fik jeg" />
+    <Word from="fár" to="får" />
+    <Word from="Hvorforfølerjeg" to="Hvorfor føler jeg" />
+    <Word from="harfeber" to="har feber" />
+    <Word from="ándssvagt" to="åndssvagt" />
+    <Word from="0g" to="Og" />
+    <Word from="vartre" to="var tre" />
+    <Word from="abner" to="åbner" />
+    <Word from="garjeg" to="går jeg" />
+    <Word from="sertil" to="ser til" />
+    <Word from="hvorfin" to="hvor fin" />
+    <Word from="harfri" to="har fri" />
+    <Word from="forstarjeg" to="forstår jeg" />
+    <Word from="Sä" to="Så" />
+    <Word from="hvorfint" to="hvor fint" />
+    <Word from="mærkerjeg" to="mærker jeg" />
+    <Word from="ogsa" to="også" />
+    <Word from="nárjeg" to="når jeg" />
+    <Word from="Jasá" to="Jaså" />
+    <Word from="bándoptager" to="båndoptager" />
+    <Word from="bedárende" to="bedårende" />
+    <Word from="sá" to="så" />
+    <Word from="nár" to="når" />
+    <Word from="kunnejo" to="kunne jo" />
+    <Word from="Brammertil" to="Brammer til" />
+    <Word from="serjeg" to="ser jeg" />
+    <Word from="gikjeg" to="gik jeg" />
+    <Word from="udholderjeg" to="udholder jeg" />
+    <Word from="máneder" to="måneder" />
+    <Word from="vartræt" to="var træt" />
+    <Word from="dárligt" to="dårligt" />
+    <Word from="klaretjer" to="klaret jer" />
+    <Word from="pavirkelig" to="påvirkelig" />
+    <Word from="spekulererjeg" to="spekulerer jeg" />
+    <Word from="forsøgerjeg" to="forsøger jeg" />
+    <Word from="huskerjeg" to="husker jeg" />
+    <Word from="ifavnen" to="i favnen" />
+    <Word from="skullejo" to="skulle jo" />
+    <Word from="vartung" to="var tung" />
+    <Word from="varfuldstændig" to="var fuldstændig" />
+    <Word from="Paskedag" to="Påskedag" />
+    <Word from="turi" to="tur i" />
+    <Word from="spillerschumanns" to="spiller Schumanns" />
+    <Word from="forstárjeg" to="forstår jeg" />
+    <Word from="istedet" to="i stedet" />
+    <Word from="nárfrem" to="når frem" />
+    <Word from="habertrods" to="håber trods" />
+    <Word from="forførste" to="for første" />
+    <Word from="varto" to="var to" />
+    <Word from="overtil" to="over til" />
+    <Word from="forfem" to="for fem" />
+    <Word from="holdtjo" to="holdt jo" />
+    <Word from="passerjo" to="passer jo" />
+    <Word from="ellerto" to="eller to" />
+    <Word from="hartrods" to="har trods" />
+    <Word from="harfuldstændig" to="har fuldstændig" />
+    <Word from="gårjeg" to="går jeg" />
+    <Word from="giderjeg" to="gider jeg" />
+    <Word from="forjer" to="for jer" />
+    <Word from="erindrerjeg" to="erindrer jeg" />
+    <Word from="tænkerjeg" to="tænker jeg" />
+    <Word from="GAEt" to="GÅET" />
+    <Word from="hørerjo" to="hører jo" />
+    <Word from="forladerjeg" to="forlader jeg" />
+    <Word from="kosterjo" to="koster jo" />
+    <Word from="fortællerjeg" to="fortæller jeg" />
+    <Word from="Forstyrrerjeg" to="Forstyrrer jeg" />
+    <Word from="tjekkerjeg" to="tjekker jeg" />
+    <Word from="erjurist" to="er jurist" />
+    <Word from="tlLBUD" to="TILBUD" />
+    <Word from="serjo" to="se rjo" />
+    <Word from="bederjeg" to="beder jeg" />
+    <Word from="bilderjeg" to="bilder jeg" />
+    <Word from="ULVEtlME" to="ULVETlME" />
+    <Word from="skærerjo" to="skærer jo" />
+    <Word from="afjer" to="af jer" />
+    <Word from="ordnerjeg" to="ordner jeg" />
+    <Word from="giverjeg" to="giver jeg" />
+    <Word from="rejservi" to="rejser vi" />
+    <Word from="fangerjeg" to="fanger jeg" />
+    <Word from="erjaloux" to="er jaloux" />
+    <Word from="glemmerjeg" to="glemmer jeg" />
+    <Word from="Behøverjeg" to="Behøver jeg" />
+    <Word from="harvi" to="har vi" />
+    <Word from="ertyndere" to="er tyndere" />
+    <Word from="fårtordenvejr" to="får tordenvejr" />
+    <Word from="varfærdig" to="var færdig" />
+    <Word from="hørerfor" to="hører for" />
+    <Word from="varvel" to="var vel" />
+    <Word from="erforbi" to="er forbi" />
+    <Word from="AIle" to="Alle" />
+    <Word from="læserjo" to="læser jo" />
+    <Word from="Edgarer" to="Edgar er" />
+    <Word from="hartaget" to="har taget" />
+    <Word from="derer" to="der er" />
+    <Word from="stikkerfrem" to="stikker frem" />
+    <Word from="haraldrig" to="har aldrig" />
+    <Word from="ellerfar" to="eller far" />
+    <Word from="erat" to="er at" />
+    <Word from="turtil" to="tur til" />
+    <Word from="erfærdig" to="er færdig" />
+    <Word from="følerjeg" to="føler jeg" />
+    <Word from="jerfra" to="jer fra" />
+    <Word from="eralt" to="er alt" />
+    <Word from="harfaktisk" to="har faktisk" />
+    <Word from="harfundet" to="har fundet" />
+    <Word from="harvendt" to="har vendt" />
+    <Word from="Kunstneraf" to="Kunstner af" />
+    <Word from="ervel" to="er vel" />
+    <Word from="ståransigt" to="står ansigt" />
+    <Word from="Erjeg" to="Er jeg" />
+    <Word from="venterjeg" to="venter jeg" />
+    <Word from="Hvorvar" to="Hvor var" />
+    <Word from="varfint" to="var fint" />
+    <Word from="ervarmt" to="er varmt" />
+    <Word from="gårfint" to="går fint" />
+    <Word from="flyverforbi" to="flyver forbi" />
+    <Word from="Dervar" to="Der var" />
+    <Word from="dervar" to="der var" />
+    <Word from="meneråndeligt" to="mener åndeligt" />
+    <Word from="forat" to="for at" />
+    <Word from="herovertil" to="herover til" />
+    <Word from="soverfor" to="sover for" />
+    <Word from="begyndtejeg" to="begyndte jeg" />
+    <Word from="vendertilbage" to="vender tilbage" />
+    <Word from="erforfærdelig" to="er forfærdelig" />
+    <Word from="gøraltid" to="gør altid" />
+    <Word from="ertilbage" to="er tilbage" />
+    <Word from="harværet" to="har været" />
+    <Word from="bagoverellertil" to="bagover eller til" />
+    <Word from="hertaler" to="her taler" />
+    <Word from="vågnerjeg" to="vågner jeg" />
+    <Word from="vartomt" to="var tomt" />
+    <Word from="gårfrem" to="går frem" />
+    <Word from="talertil" to="taler til" />
+    <Word from="ertryg" to="er tryg" />
+    <Word from="ansigtervendes" to="ansigter vendes" />
+    <Word from="hervirkeligt" to="her virkeligt" />
+    <Word from="herer" to="her er" />
+    <Word from="drømmerjo" to="drømmer jo" />
+    <Word from="erfuldkommen" to="er fuldkommen" />
+    <Word from="hveren" to="hver en" />
+    <Word from="erfej" to="er fej" />
+    <Word from="datterforgæves" to="datter forgæves" />
+    <Word from="forsøgerjo" to="forsøger jo" />
+    <Word from="ertom" to="er tom" />
+    <Word from="vareftermiddag" to="var eftermiddag" />
+    <Word from="vartom" to="var tom" />
+    <Word from="angerellerforventninger" to="anger eller forventninger" />
+    <Word from="kørtejeg" to="kørte jeg" />
+    <Word from="Hvorforfortæller" to="Hvorfor fortæller" />
+    <Word from="gårtil" to="går til" />
+    <Word from="ringerefter" to="ringer efter" />
+    <Word from="søgertilflugt" to="søger tilflugt" />
+    <Word from="ertvunget" to="er tvunget" />
+    <Word from="megetjeg" to="meget jeg" />
+    <Word from="varikke" to="var ikke" />
+    <Word from="Derermange" to="Der e rmange" />
+    <Word from="dervilhindre" to="der vil hindre" />
+    <Word from="erså" to="er så" />
+    <Word from="DetforstårLeggodt" to="Det forstår jeg godt" />
+    <Word from="ergodt" to="er godt" />
+    <Word from="vorventen" to="vor venten" />
+    <Word from="tagerfejl" to="tager fejl" />
+    <Word from="ellerer" to="eller er" />
+    <Word from="laverjeg" to="laver jeg" />
+    <Word from="0mgang" to="omgang" />
+    <Word from="afstár" to="afstår" />
+    <Word from="pá" to="på" />
+    <Word from="rejserjeg" to="rejser jeg" />
+    <Word from="ellertage" to="eller tage" />
+    <Word from="takkerjeg" to="takker jeg" />
+    <Word from="ertilfældigvis" to="er tilfældigvis" />
+    <Word from="fremstar" to="fremstår" />
+    <Word from="ertæt" to="er tæt" />
+    <Word from="ijeres" to="i jeres" />
+    <Word from="Sagdejeg" to="Sagde jeg" />
+    <Word from="overi" to="over i" />
+    <Word from="plukkerjordbær" to="plukker jordbær" />
+    <Word from="klarerjeg" to="klarer jeg" />
+    <Word from="jerfire" to="jer fire" />
+    <Word from="tábeligste" to="tåbeligste" />
+    <Word from="sigertvillingerne" to="siger tvillingerne" />
+    <Word from="erfaktisk" to="er faktisk" />
+    <Word from="gár" to="går" />
+    <Word from="harvasket" to="har vasket" />
+    <Word from="harplukketjordbærtil" to="har plukket jordbær til" />
+    <Word from="plukketjordbær" to="plukket jordbær" />
+    <Word from="klaverfirehændigt" to="klaver firehændigt" />
+    <Word from="erjævnaldrende" to="er jævnaldrende" />
+    <Word from="tierjeg" to="tier jeg" />
+    <Word from="Hvorerden" to="Hvor er den" />
+    <Word from="0veraltjeg" to="overalt jeg" />
+    <Word from="gårpå" to="går på" />
+    <Word from="finderjeg" to="finder jeg" />
+    <Word from="serhans" to="ser hans" />
+    <Word from="tiderbliver" to="tider bliver" />
+    <Word from="ellertrist" to="eller trist" />
+    <Word from="forstårjeres" to="forstår jeres" />
+    <Word from="Hvorsjælen" to="Hvor sjælen" />
+    <Word from="finderro" to="finder ro" />
+    <Word from="sidderjeg" to="sidder jeg" />
+    <Word from="tagerjo" to="tager jo" />
+    <Word from="efterjeres" to="efter jeres" />
+    <Word from="10O" to="100" />
+    <Word from="besluttedejeg" to="besluttede jeg" />
+    <Word from="varsket" to="var sket" />
+    <Word from="uadskillige" to="uadskillelige" />
+    <Word from="harjetlag" to="har jetlag" />
+    <Word from="lkke" to="Ikke" />
+    <Word from="lntet" to="Intet" />
+    <Word from="afslørerjeg" to="afslører jeg" />
+    <Word from="måjeg" to="må jeg" />
+    <Word from="Vl" to="VI" />
+    <Word from="atbygge" to="at bygge" />
+    <Word from="detmakabre" to="det makabre" />
+    <Word from="vilikke" to="vil ikke" />
+    <Word from="talsmandbekræfter" to="talsmand bekræfter" />
+    <Word from="vedatrenovere" to="ved at renovere" />
+    <Word from="forsøgeratforstå" to="forsøger at forstå" />
+    <Word from="ersket" to="er sket" />
+    <Word from="morderpå" to="morder på" />
+    <Word from="frifodiRosewood" to="fri fod i Rosewood" />
+    <Word from="holdtpressemøde" to="holdt pressemøde" />
+    <Word from="lngen" to="Ingen" />
+    <Word from="lND" to="IND" />
+    <Word from="henterjeg" to="henter jeg" />
+    <Word from="lsabel" to="Isabel" />
+    <Word from="lsabels" to="Isabels" />
+    <Word from="vinderjo" to="vinder jo" />
+    <Word from="rødmerjo" to="rødmer jo" />
+    <Word from="etjakkesæt" to="et jakkesæt" />
+    <Word from="glæderjeg" to="glæder jeg" />
+    <Word from="lgen" to="Igen" />
+    <Word from="lsær" to="Især" />
+    <Word from="iparken" to="i parken" />
+    <Word from="nårl" to="når I" />
+    <Word from="tilA1" to="til A1" />
+    <Word from="FBl" to="FBI" />
+    <Word from="viljo" to="vil jo" />
+    <Word from="detpå" to="det på" />
+    <Word from="KIar" to="Klar" />
+    <Word from="PIan" to="Plan" />
+    <Word from="EIIer" to="Eller" />
+    <Word from="FIot" to="Flot" />
+    <Word from="AIIe" to="Alle" />
+    <Word from="AIt" to="Alt" />
+    <Word from="KIap" to="Klap" />
+    <Word from="PIaza" to="Plaza" />
+    <Word from="SIap" to="Slap" />
+    <Word from="Iå" to="lå" />
+    <Word from="BIing" to="Bling" />
+    <Word from="GIade" to="Glade" />
+    <Word from="Iejrbålssange" to="lejrbålssange" />
+    <Word from="bedtjer" to="bedt jer" />
+    <Word from="hørerjeg" to="hører jeg" />
+    <Word from="Fårjeg" to="Får jeg" />
+    <Word from="fikJames" to="fik James" />
+    <Word from="atsnakke" to="at snakke" />
+    <Word from="varkun" to="var kun" />
+    <Word from="retterjeg" to="retter jeg" />
+    <Word from="ernormale" to="er normale" />
+    <Word from="viljeg" to="vil jeg" />
+    <Word from="Sætjer" to="Sæt jer" />
+    <Word from="udsatham" to="udsat ham" />
+  </WholeWords>
+  <PartialWordsAlways>
+    <WordPart from="¤" to="o" />
+    <WordPart from="IVI" to="M" />
+    <WordPart from="lVI" to="M" />
+    <WordPart from="IVl" to="M" />
+    <WordPart from="lVl" to="M" />
+  </PartialWordsAlways>
+  <PartialWords>
+    <!-- Will be used to check words not in dictionary -->
+    <!-- If new word(s) exists in spelling dictionary, it(they) is accepted -->
+    <WordPart from="fi" to="fi" />
+    <WordPart from="fl" to="fl" />
+    <WordPart from="/" to="l" />
+    <WordPart from="vv" to="w" />
+    <WordPart from="m" to="rn" />
+    <WordPart from="l" to="i" />
+    <WordPart from="€" to="e" />
+    <WordPart from="I" to="l" />
+    <WordPart from="c" to="o" />
+    <WordPart from="i" to="t" />
+    <WordPart from="cc" to="oo" />
+    <WordPart from="ii" to="tt" />
+    <WordPart from="n/" to="ry" />
+    <WordPart from="ae" to="æ" />
+    <!-- "f " will be two words -->
+    <WordPart from="f" to="f " />
+    <WordPart from="c" to="e" />
+    <WordPart from="o" to="e" />
+    <WordPart from="I" to="t" />
+    <WordPart from="n" to="o" />
+    <WordPart from="s" to="e" />
+    <WordPart from="\A" to="Vi" />
+    <WordPart from="n/" to="rv" />
+    <WordPart from="Ã" to="Å" />
+    <WordPart from="í" to="i" />
+  </PartialWords>
+  <PartialLines />
+  <PartialLinesAlways />
+  <BeginLines />
+  <EndLines />
+  <WholeLines />
+  <RegularExpressions />
+</OCRFixReplaceList>

Contents/Libraries/Shared/subzero/modification/dictionaries/xml/fra_OCRFixReplaceList.xml

@@ -0,0 +1,270 @@
+<OCRFixReplaceList>
+  <WholeWords>
+    <Word from="@immatriculation" to="d'immatriculation" />
+    <Word from="acquer" to="acquér" />
+    <Word from="acteurjoue" to="acteur joue" />
+    <Word from="aerien" to="aérien" />
+    <Word from="agreable" to="agréable" />
+    <Word from="aientjamais" to="aient jamais" />
+    <Word from="AII" to="All" />
+    <Word from="aitjamais" to="ait jamais" />
+    <Word from="aitjus" to="ait jus" />
+    <Word from="alle" to="allé" />
+    <Word from="alles" to="allés" />
+    <Word from="appele" to="appelé" />
+    <Word from="apres" to="après" />
+    <Word from="aujourdhui" to="aujourd'hui" />
+    <Word from="aupres" to="auprès" />
+    <Word from="beaute" to="beauté" />
+    <Word from="cabossee" to="cabossée" />
+    <Word from="carj'" to="car j'" />
+    <Word from="Carj'" to="Car j'" />
+    <Word from="carla" to="car la" />
+    <Word from="CEdipe" to="Œdipe" />
+    <Word from="Cest" to="C'est" />
+    <Word from="c'etaient" to="c'étaient" />
+    <Word from="Cétaient" to="C'étaient" />
+    <Word from="c'etait" to="c'était" />
+    <Word from="C'etait" to="C'était" />
+    <Word from="Cétait" to="C'était" />
+    <Word from="choregraphiee" to="chorégraphiée" />
+    <Word from="cinema" to="cinéma" />
+    <Word from="cl'AIcatraz" to="d'Alcatraz" />
+    <Word from="cles" to="clés" />
+    <Word from="cœurjoie" to="cœur-joie" />
+    <Word from="completer" to="compléter" />
+    <Word from="costumiere" to="costumière" />
+    <Word from="cree" to="créé" />
+    <Word from="daccord" to="d'accord" />
+    <Word from="d'AIbert" to="d'Albert" />
+    <Word from="d'AIdous" to="d'Aldous" />
+    <Word from="d'AIec" to="d'Alec" />
+    <Word from="danniversaire" to="d'anniversaire" />
+    <Word from="d'Arra'bida" to="d'Arrabida" />
+    <Word from="d'autodérision" to="d'auto-dérision" />
+    <Word from="dautres" to="d'autres" />
+    <Word from="debattait" to="débattait" />
+    <Word from="decor" to="décor" />
+    <Word from="decorateurs" to="décorateurs" />
+    <Word from="decors" to="décors" />
+    <Word from="defi" to="défi" />
+    <Word from="dejà" to="déjà" />
+    <Word from="déjàm" to="déjà..." />
+    <Word from="dejeunait" to="déjeunait" />
+    <Word from="dengager" to="d'engager" />
+    <Word from="déquipement" to="d'équipement" />
+    <Word from="dérnièré" to="dernière" />
+    <Word from="Desole" to="Désolé" />
+    <Word from="dessayage" to="d'essayage" />
+    <Word from="dessence" to="d'essence" />
+    <Word from="détaient" to="c'étaient" />
+    <Word from="detail" to="détail" />
+    <Word from="dexcellents" to="d'excellents" />
+    <Word from="dexpérience" to="d'expérience" />
+    <Word from="dexpériences" to="d'expériences" />
+    <Word from="d'héro'l'ne" to="d'héroïne" />
+    <Word from="d'idees" to="d'idées" />
+    <Word from="d'intensite" to="d'intensité" />
+    <Word from="dontj" to="dont j" />
+    <Word from="doublaitAlfo" to="doublait Alfo" />
+    <Word from="DrNo" to="Dr No" />
+    <Word from="e'" to="é" />
+    <Word from="ecrit" to="écrit" />
+    <Word from="elegant" to="élégant" />
+    <Word from="Ellé" to="Elle" />
+    <Word from="én" to="en" />
+    <Word from="equipe" to="équipe" />
+    <Word from="erjus" to="er jus" />
+    <Word from="estjamais" to="est jamais" />
+    <Word from="ét" to="et" />
+    <Word from="etaient" to="étaient" />
+    <Word from="etait" to="était" />
+    <Word from="ete" to="été" />
+    <Word from="etiez" to="étiez" />
+    <Word from="etj'" to="et j'" />
+    <Word from="Etj'" to="Et j'" />
+    <Word from="etje" to="et je" />
+    <Word from="Etje" to="Et je" />
+    <Word from="EtsouvenL" to="Et souvent" />
+    <Word from="eviter" to="éviter" />
+    <Word from="Fabsence" to="l'absence" />
+    <Word from="fadapter" to="t'adapter" />
+    <Word from="fadore" to="j'adore" />
+    <Word from="Fâge" to="l'âge" />
+    <Word from="Fagent" to="l'agent" />
+    <Word from="faiessayé" to="j'ai essayé" />
+    <Word from="Failure" to="l'alllure" />
+    <Word from="Fambiance" to="l'ambiance" />
+    <Word from="Famener" to="l'amener" />
+    <Word from="Fanniversaire" to="l'anniversaire" />
+    <Word from="Fapparence" to="l'apparence" />
+    <Word from="Fapres" to="l'apres" />
+    <Word from="Faprès" to="l'après" />
+    <Word from="Farmée" to="l'armée" />
+    <Word from="Farrière" to="l'arrière" />
+    <Word from="Farrivée" to="l'arrivée" />
+    <Word from="Fascenseur" to="l'ascenseur" />
+    <Word from="Fascension" to="l'ascension" />
+    <Word from="Fassaut" to="l'assaut" />
+    <Word from="Fassomme" to="l'assomme" />
+    <Word from="Fatmosphère" to="l'atmosphère" />
+    <Word from="Fattention" to="l'attention" />
+    <Word from="Favalanche" to="l'avalanche" />
+    <Word from="Féclairage" to="l'éclairage" />
+    <Word from="Fécran" to="l'écran" />
+    <Word from="Fémotion" to="l'émotion" />
+    <Word from="Femplacement" to="l'emplacement" />
+    <Word from="Fendroit" to="l'endroit" />
+    <Word from="Fenseigne" to="l'enseigne" />
+    <Word from="Fensemble" to="l'ensemble" />
+    <Word from="Fentouraient" to="l'entouraient" />
+    <Word from="Fentrée" to="l'entrée" />
+    <Word from="Fépaisseur" to="l'épaisseur" />
+    <Word from="Fépoque" to="l'époque" />
+    <Word from="Féquipe" to="Équipe" />
+    <Word from="Fespace" to="l'espace" />
+    <Word from="fespérais" to="j'espérais" />
+    <Word from="Fespère" to="l'espère" />
+    <Word from="Festhétique" to="l'esthétique" />
+    <Word from="Fetranger" to="l'etranger" />
+    <Word from="Févasion" to="l'évasion" />
+    <Word from="Févoque" to="l'évoque" />
+    <Word from="Fexpérience" to="l'expérience" />
+    <Word from="Fexplique" to="l'explique" />
+    <Word from="Fexplosion" to="l'explosion" />
+    <Word from="Fextérieur" to="l'extérieur" />
+    <Word from="Fhabituelle" to="l'habituelle" />
+    <Word from="Fhélicoptère" to="l'hélicoptère" />
+    <Word from="Fhéliport" to="l'héliport" />
+    <Word from="Fhélistation" to="l'hélistation" />
+    <Word from="Fhonneur" to="l'honneur" />
+    <Word from="Fhorloge" to="l'horloge" />
+    <Word from="Fidée" to="l'idée" />
+    <Word from="Fimage" to="l'image" />
+    <Word from="Fimportance" to="l'importance" />
+    <Word from="Fimpression" to="l'impression" />
+    <Word from="Finfluence" to="l'influence" />
+    <Word from="Finscription" to="l'inscription" />
+    <Word from="Fintérieur" to="l'intérieur" />
+    <Word from="Fintrigue" to="l'intrigue" />
+    <Word from="Fobjectif" to="l'objectif" />
+    <Word from="Foccasion" to="l'occasion" />
+    <Word from="Fordre" to="l'ordre" />
+    <Word from="Forigine" to="l'origine" />
+    <Word from="frêre" to="frère" />
+    <Word from="gaylns" to="gaijins" />
+    <Word from="general" to="général" />
+    <Word from="hawaïennel" to="hawaïenne" />
+    <Word from="hawa'l'en" to="hawaïen" />
+    <Word from="Ia" to="la" />
+    <Word from="Ià" to="là" />
+    <Word from="Iaryngotomie" to="laryngotomie" />
+    <Word from="idee" to="idée" />
+    <Word from="idees" to="idées" />
+    <Word from="Ie" to="le" />
+    <Word from="Ies" to="les" />
+    <Word from="Iester" to="Lester" />
+    <Word from="II" to="Il" />
+    <Word from="Iimit" to="limit" />
+    <Word from="IIs" to="Ils" />
+    <Word from="immediatement" to="immédiatement" />
+    <Word from="insufflee" to="insufflée" />
+    <Word from="integrer" to="intégrer" />
+    <Word from="interessante" to="intéressante" />
+    <Word from="Iogions" to="logions" />
+    <Word from="Iorsqu" to="lorsqu" />
+    <Word from="isee" to="isée" />
+    <Word from="Iumiere" to="lumiere" />
+    <Word from="Iynchage" to="lynchage" />
+    <Word from="J'espere" to="J'espère" />
+    <Word from="Jessaie" to="J'essaie" />
+    <Word from="j'etais" to="j'étais" />
+    <Word from="J'etais" to="J'étais" />
+    <Word from="latéralémént" to="latéralement" />
+    <Word from="lci" to="Ici" />
+    <Word from="Lci" to="Ici" />
+    <Word from="lé-" to="là-" />
+    <Word from="lepidopteres" to="lépidoptères" />
+    <Word from="litteraire" to="littéraire" />
+    <Word from="ll" to="il" />
+    <Word from="Ll" to="Il" />
+    <Word from="lls" to="ils" />
+    <Word from="Lls" to="Ils" />
+    <Word from="maintenanu" to="maintenant" />
+    <Word from="maniere" to="manière" />
+    <Word from="mariee" to="mariée" />
+    <Word from="Mayer/ing" to="Mayerling" />
+    <Word from="meilleurjour" to="meilleur jour" />
+    <Word from="melange" to="mélange" />
+    <Word from="n'avaiént" to="n'avaient" />
+    <Word from="n'etait" to="n'était" />
+    <Word from="oitjamais" to="oit jamais" />
+    <Word from="oitjus" to="oit jus" />
+    <Word from="ontete" to="ont été" />
+    <Word from="operateur" to="opérateur" />
+    <Word from="ouvérté" to="ouverte" />
+    <Word from="Pépreuve" to="l'épreuve" />
+    <Word from="pere" to="père" />
+    <Word from="plateforme" to="plate-forme" />
+    <Word from="pourjouer" to="pour jouer" />
+    <Word from="precipice" to="précipice" />
+    <Word from="preferes" to="préférés" />
+    <Word from="premierjour" to="premier jour" />
+    <Word from="presenter" to="présenter" />
+    <Word from="prevu" to="prévu" />
+    <Word from="prevue" to="prévue" />
+    <Word from="propriete" to="propriété" />
+    <Word from="protègeraient" to="protégeraient" />
+    <Word from="qué" to="que" />
+    <Word from="qwangoissé" to="qu'angoissé" />
+    <Word from="realisateur" to="réalisateur" />
+    <Word from="reception" to="réception" />
+    <Word from="reévalu" to="réévalu" />
+    <Word from="repute" to="réputé" />
+    <Word from="reussi" to="réussi" />
+    <Word from="s'arrétait" to="s'arrêtait" />
+    <Word from="s'ave'rer" to="s'avérer" />
+    <Word from="scenario" to="scénario" />
+    <Word from="scene" to="scène" />
+    <Word from="scenes" to="scènes" />
+    <Word from="seances" to="séances" />
+    <Word from="sequence" to="séquence" />
+    <Word from="sflécrasa" to="s'écrasa" />
+    <Word from="speciale" to="spéciale" />
+    <Word from="Supen" to="Super" />
+    <Word from="torturee" to="torturée" />
+    <Word from="Uadmirable" to="L'admirable" />
+    <Word from="Uensemblier" to="L'ensemblier" />
+    <Word from="Uexplosion" to="L'explosion" />
+    <Word from="Uouvre" to="L'ouvre" />
+    <Word from="Vaise" to="l'aise" />
+    <Word from="vecu" to="vécu" />
+    <Word from="vehicules" to="véhicules" />
+    <Word from="Ÿappréciais" to="J'appréciais" />
+    <Word from="Ÿespère" to="J'espère" />
+    <Word from="ÿétrangle" to="s'étrangle" />
+  </WholeWords>
+  <PartialWordsAlways />
+  <PartialWords />
+  <PartialLines>
+    <LinePart from=" I'" to=" l'" />
+    <LinePart from=" |'" to=" l'" />
+  </PartialLines>
+  <PartialLinesAlways />
+  <BeginLines />
+  <EndLines />
+  <WholeLines>
+    <Line from="&quot;D'ac:c:ord.&quot;" to="&quot;D'accord.&quot;" />
+    <Line from="“i QUÎ gagne, qui perd," to="ni qui gagne, qui perd," />
+    <Line from="L'ac:c:ent est mis &#xD;&#xA; &#xD;&#xA; sur son trajet jusqu'en Suisse." to="L'accent est mis &#xD;&#xA; &#xD;&#xA; sur son trajet jusqu'en Suisse." />
+    <Line from="C'est la plus gentille chose &#xD;&#xA; &#xD;&#xA; qu'Hitchc:oc:k m'ait jamais dite." to="C'est la plus gentille chose &#xD;&#xA; &#xD;&#xA; qu'Hitchcock m'ait jamais dite." />
+    <Line from="Tout le monde, en revanche, qualifie &#xD;&#xA; &#xD;&#xA; Goldfinger d'aventu re structurée," to="Tout le monde, en revanche, qualifie &#xD;&#xA; &#xD;&#xA; Goldfinger d'aventure structurée," />
+    <Line from="et le film Shadow of a man &#xD;&#xA; &#xD;&#xA; a lancé sa carrière au cinéma." to="et le film &lt;i&gt;Shadow of a man&lt;/i&gt; &#xD;&#xA; &#xD;&#xA; a lancé sa carrière au cinéma." />
+    <Line from="En 1948, Young est passé à la réalisation &#xD;&#xA; &#xD;&#xA; avec One night with you." to="En 1948, Young est passé à la réalisation &#xD;&#xA; &#xD;&#xA; avec &lt;i&gt;One night with you&lt;/i&gt;." />
+    <Line from="Il a construit tous ces véhicules &#xD;&#xA; &#xD;&#xA; à C)c:ala, en Floride." to="Il a construit tous ces véhicules &#xD;&#xA; &#xD;&#xA; à Ocala, en Floride." />
+    <Line from="Tokyo Pop et A Taxing Woman? Return." to="Tokyo Pop et A Taxing Woman's Return." />
+    <Line from="Peter H u nt." to="Peter Hunt." />
+    <Line from="&quot;C'est bien mieux dans Peau. &#xD;&#xA; &#xD;&#xA; On peut sfléclabousser, faire du bruit.&quot;" to="&quot;C'est bien mieux dans l'eau. &#xD;&#xA; &#xD;&#xA; On peut s'éclabousser, faire du bruit.&quot;" />
+  </WholeLines>
+  <RegularExpressions />
+</OCRFixReplaceList>

Contents/Libraries/Shared/subzero/modification/dictionaries/xml/hun_OCRFixReplaceList.xml

@@ -0,0 +1,25 @@
+<OCRFixReplaceList>
+  <WholeWords />
+  <PartialWordsAlways />
+  <PartialWords />
+  <PartialLines />
+  <PartialLinesAlways />
+  <BeginLines />
+  <EndLines />
+  <WholeLines />
+  <RegularExpressions>
+    <!-- nagy I-l javítások -->
+    <RegEx find="([\x41-\x5a\x61-\x7a\xc1-\xfc])II" replaceWith="$1ll" />
+    <RegEx find="II([\x61-\x7a\xe1-\xfc])" replaceWith="ll$1" />
+    <RegEx find="([\x61-\x7a\xe1-\xfc])I" replaceWith="$1l" />
+    <RegEx find="([\x20])I([^aeou\x41-\x5a\xc1-\xdc])" replaceWith="$1l$2" />
+    <RegEx find="\bl([bcdfghjklmnpqrstvwxz])" replaceWith="I$1" />
+    <RegEx find="([\x41-\x5a\xc1-\xdc])I([\x61-\x7a\xe1-\xfc])" replaceWith="$1l$2" />
+    <RegEx find="([\x61-\x7a\xe1-\xfc][\-])I([\x61-\x7a\xe1-\xfc])" replaceWith="$1l$2" />
+    <RegEx find="([\x41-\x5a\xc1-\xdc])I([\-][\x41-\x5a\xc1-\xdc][\x61-\x7a\xe1-\xfc])" replaceWith="$1l$2" />
+    <RegEx find="\b([AEÜÓ])I([^\x41-\x5a\xc1-\xdc])" replaceWith="$1l$2" />
+    <RegEx find="\bI([aáeéiíoóöuúüy\xf5\xfb])" replaceWith="l$1" />
+    <RegEx find="\b(?:II|ll)" replaceWith="Il" />
+    <RegEx find="([\xf5\xfb])I" replaceWith="$1l" />
+  </RegularExpressions>
+</OCRFixReplaceList>

Contents/Libraries/Shared/subzero/modification/dictionaries/xml/nld_OCRFixReplaceList.xml

@@ -0,0 +1,24 @@
+<OCRFixReplaceList>
+  <WholeWords>
+    <Word from="ls" to="Is" />
+    <Word from="ln" to="In" />
+    <Word from="lk" to="Ik" />
+    <Word from="ledereen" to="Iedereen" />
+    <Word from="ledere" to="Iedere" />
+    <Word from="lemand" to="Iemand" />
+  </WholeWords>
+  <PartialWordsAlways />
+  <PartialWords />
+  <PartialLines />
+  <PartialLinesAlways />
+  <BeginLines />
+  <EndLines />
+  <WholeLines />
+  <RegularExpressions>
+    <RegEx find="\blk(?=\p{Ll}{2})" replaceWith="Ik" />
+    <RegEx find="\bln(?=\p{Ll}{2})" replaceWith="In" />
+    <RegEx find="\bls(?=\p{Ll}{2})" replaceWith="Is" />
+    <RegEx find="\beIk" replaceWith="elk" />
+    <RegEx find="\bler(land|se|s|)\b" replaceWith="Ier$1" />
+  </RegularExpressions>
+</OCRFixReplaceList>

Contents/Libraries/Shared/subzero/modification/dictionaries/xml/nor_OCRFixReplaceList.xml

@@ -0,0 +1,43 @@
+<OCRFixReplaceList>
+  <WholeWords />
+  <PartialWordsAlways />
+  <PartialWords>
+    <!-- Will be used to check words not in dictionary -->
+    <!-- If new word(s) exists in spelling dictionary, it(they) is accepted -->
+    <WordPart from="¤" to="o" />
+    <WordPart from="fi" to="fi" />
+    <WordPart from="fl" to="fl" />
+    <WordPart from="/" to="l" />
+    <WordPart from="vv" to="w" />
+    <WordPart from="IVI" to="M" />
+    <WordPart from="lVI" to="M" />
+    <WordPart from="IVl" to="M" />
+    <WordPart from="lVl" to="M" />
+    <WordPart from="m" to="rn" />
+    <WordPart from="l" to="i" />
+    <WordPart from="€" to="e" />
+    <WordPart from="I" to="l" />
+    <WordPart from="c" to="o" />
+    <WordPart from="i" to="t" />
+    <WordPart from="cc" to="oo" />
+    <WordPart from="ii" to="tt" />
+    <WordPart from="n/" to="ry" />
+    <WordPart from="ae" to="æ" />
+    <!-- "f " will be two words -->
+    <WordPart from="f" to="f " />
+    <WordPart from="c" to="e" />
+    <WordPart from="I" to="t" />
+    <WordPart from="n" to="o" />
+    <WordPart from="s" to="e" />
+    <WordPart from="\A" to="Vi" />
+    <WordPart from="n/" to="rv" />
+    <WordPart from="Ã" to="Å" />
+    <WordPart from="í" to="i" />
+  </PartialWords>
+  <PartialLines />
+  <PartialLinesAlways />
+  <BeginLines />
+  <EndLines />
+  <WholeLines />
+  <RegularExpressions />
+</OCRFixReplaceList>

Contents/Libraries/Shared/subzero/modification/dictionaries/xml/por_OCRFixReplaceList.xml

@@ -0,0 +1,508 @@
+<OCRFixReplaceList>
+  <WholeWords>
+    <Word from="abitual" to="habitual" />
+    <Word from="àcerca" to="acerca" />
+    <Word from="acessor" to="assessor" />
+    <Word from="acólico" to="acólito" />
+    <Word from="açoreano" to="açoriano" />
+    <Word from="actuacao" to="actuação" />
+    <Word from="acucar" to="açúcar" />
+    <Word from="açucar" to="açúcar" />
+    <Word from="advinhar" to="adivinhar" />
+    <Word from="africa" to="África" />
+    <Word from="ajuisar" to="ajuizar" />
+    <Word from="album" to="álbum" />
+    <Word from="alcoolémia" to="alcoolemia" />
+    <Word from="aldião" to="aldeão" />
+    <Word from="algerino" to="argelino" />
+    <Word from="ameixeal" to="ameixial" />
+    <Word from="amiaça" to="ameaça" />
+    <Word from="analizar" to="analisar" />
+    <Word from="andáste" to="andaste" />
+    <Word from="anemona" to="anémona" />
+    <Word from="antartico" to="antárctico" />
+    <Word from="antártico" to="antárctico" />
+    <Word from="antepôr" to="antepor" />
+    <Word from="apárte" to="aparte" />
+    <Word from="apiadeiro" to="apeadeiro" />
+    <Word from="apiar" to="apear" />
+    <Word from="apreciacao" to="apreciação" />
+    <Word from="arctico" to="árctico" />
+    <Word from="arrazar" to="arrasar" />
+    <Word from="ártico" to="árctico" />
+    <Word from="artifice" to="artífice" />
+    <Word from="artifícial" to="artificial" />
+    <Word from="ascenção" to="ascensão" />
+    <!-- <Word from="assucar" to="açúcar" /> assucar é uma palavra existente no dicionário -->
+    <Word from="assúcar" to="açúcar" />
+    <Word from="aste" to="haste" />
+    <Word from="asterístico" to="asterisco" />
+    <Word from="averção" to="aversão" />
+    <Word from="avizar" to="avisar" />
+    <Word from="avulsso" to="avulso" />
+    <Word from="baínha" to="bainha" />
+    <Word from="banca-rota" to="bancarrota" />
+    <Word from="bandeija" to="bandeja" />
+    <Word from="bébé" to="bebé" />
+    <Word from="beige" to="bege" />
+    <Word from="benção" to="bênção" />
+    <Word from="beneficiência" to="beneficência" />
+    <Word from="beneficiente" to="beneficente" />
+    <Word from="benvinda" to="bem-vinda" />
+    <Word from="benvindo" to="bem-vindo" />
+    <Word from="boasvindas" to="boas-vindas" />
+    <Word from="borborinho" to="burburinho" />
+    <Word from="Brazil" to="Brasil" />
+    <Word from="bussula" to="bússola" />
+    <Word from="cabo-verdeano" to="cabo-verdiano" />
+    <Word from="caimbras" to="cãibras" />
+    <Word from="calcáreo" to="calcário" />
+    <Word from="calsado" to="calçado" />
+    <Word from="calvíce" to="calvície" />
+    <Word from="camoneano" to="camoniano" />
+    <Word from="campião" to="campeão" />
+    <Word from="cançacos" to="cansaços" />
+    <Word from="caracter" to="carácter" />
+    <Word from="caractéres" to="caracteres" />
+    <Word from="catequeze" to="catequese" />
+    <Word from="catequisador" to="catequizador" />
+    <Word from="catequisar" to="catequizar" />
+    <Word from="chícara" to="xícara" />
+    <Word from="ciclano" to="sicrano" />
+    <Word from="cicrano" to="sicrano" />
+    <Word from="cidadães" to="cidadãos" />
+    <Word from="cidadões" to="cidadãos" />
+    <Word from="cincoenta" to="cinquenta" />
+    <Word from="cinseiro" to="cinzeiro" />
+    <Word from="cinsero" to="sincero" />
+    <Word from="citacoes" to="citações" />
+    <Word from="coalizão" to="colisão" />
+    <Word from="côdia" to="côdea" />
+    <Word from="combóio" to="comboio" />
+    <Word from="compôr" to="compor" />
+    <Word from="concerteza" to="com certeza" />
+    <Word from="constituia" to="constituía" />
+    <Word from="constituíu" to="constituiu" />
+    <Word from="contato" to="contacto" />
+    <Word from="contensão" to="contenção" />
+    <Word from="contribuicoes" to="contribuições" />
+    <Word from="côr" to="cor" />
+    <Word from="corassão" to="coração" />
+    <Word from="corçario" to="corsário" />
+    <Word from="corçário" to="corsário" />
+    <Word from="cornprimidosinbo" to="comprimidozinho" />
+    <!-- <Word from="cota-parte" to="quota-parte" /> é uma palavra existente no dicionário -->
+    <Word from="crâneo" to="crânio" />
+    <Word from="dE" to="de" />
+    <Word from="defenição" to="definição" />
+    <Word from="defenido" to="definido" />
+    <Word from="defenir" to="definir" />
+    <Word from="deficite" to="défice" />
+    <Word from="degladiar" to="digladiar" />
+    <Word from="deiche" to="deixe" />
+    <Word from="desinteria" to="disenteria" />
+    <Word from="despendio" to="dispêndio" />
+    <Word from="despêndio" to="dispêndio" />
+    <Word from="desplicência" to="displicência" />
+    <Word from="dificulidade" to="dificuldade" />
+    <Word from="dispender" to="despender" />
+    <Word from="dispendio" to="dispêndio" />
+    <Word from="distribuido" to="distribuído" />
+    <Word from="druída" to="druida" />
+    <Word from="écrã" to="ecrã" />
+    <Word from="ecran" to="ecrã" />
+    <Word from="écran" to="ecrã" />
+    <Word from="êle" to="ele" />
+    <Word from="elice" to="hélice" />
+    <Word from="élice" to="hélice" />
+    <Word from="emiratos" to="emirados" />
+    <Word from="engolis-te" to="engoliste" />
+    <Word from="engulir" to="engolir" />
+    <Word from="enguliste" to="engoliste" />
+    <Word from="entertido" to="entretido" />
+    <Word from="entitular" to="intitular" />
+    <Word from="entreterimento" to="entretenimento" />
+    <Word from="entreti-me" to="entretive-me" />
+    <Word from="envólucro" to="invólucro" />
+    <Word from="erói" to="herói" />
+    <Word from="escluir" to="excluir" />
+    <Word from="esclusão" to="exclusão" />
+    <Word from="escrivões" to="escrivães" />
+    <Word from="esqueiro" to="isqueiro" />
+    <Word from="esquesito" to="esquisito" />
+    <Word from="estacoes" to="estações" />
+    <Word from="esteje" to="esteja" />
+    <Word from="excavação" to="escavação" />
+    <Word from="excavar" to="escavar" />
+    <Word from="exdrúxula" to="esdrúxula" />
+    <Word from="exdrúxulas" to="esdrúxulas" />
+    <Word from="exitar" to="hesitar" />
+    <Word from="explicacoes" to="explicações" />
+    <Word from="exquisito" to="esquisito" />
+    <Word from="extende" to="estende" />
+    <Word from="extender" to="estender" />
+    <Word from="fàcilmenfe" to="facilmente" />
+    <Word from="fàcilmente" to="facilmente" />
+    <Word from="fariam-lhe" to="far-lhe-iam" />
+    <Word from="FARMÁClAS" to="FARMÁCIAS" />
+    <Word from="farmecêutico" to="farmacêutico" />
+    <Word from="fassa" to="faça" />
+    <Word from="fébre" to="febre" />
+    <Word from="fecula" to="fécula" />
+    <Word from="fémea" to="fêmea" />
+    <Word from="femenino" to="feminino" />
+    <Word from="femininismo" to="feminismo" />
+    <Word from="físiologista" to="fisiologista" />
+    <Word from="fizémos" to="fizemos" />
+    <Word from="fizes-te" to="fizeste" />
+    <Word from="flôr" to="flor" />
+    <Word from="forão" to="foram" />
+    <Word from="formalisar" to="formalizar" />
+    <Word from="fôro" to="foro" />
+    <Word from="fos-te" to="foste" />
+    <Word from="fragância" to="fragrância" />
+    <Word from="françês" to="francês" />
+    <Word from="frasqutnho" to="frasquinho" />
+    <Word from="frustado" to="frustrado" />
+    <Word from="furá" to="furar" />
+    <Word from="gaz" to="gás" />
+    <Word from="gáz" to="gás" />
+    <Word from="geito" to="jeito" />
+    <Word from="geneceu" to="gineceu" />
+    <Word from="geropiga" to="jeropiga" />
+    <Word from="glicémia" to="glicemia" />
+    <Word from="gorgeta" to="gorjeta" />
+    <Word from="grangear" to="granjear" />
+    <Word from="guizar" to="guisar" />
+    <Word from="hectar" to="hectare" />
+    <Word from="herméticamente" to="hermeticamente" />
+    <Word from="hernia" to="hérnia" />
+    <Word from="higiéne" to="higiene" />
+    <Word from="hilariedade" to="hilaridade" />
+    <Word from="hiperacídez" to="hiperacidez" />
+    <Word from="hontem" to="ontem" />
+    <Word from="igiene" to="higiene" />
+    <Word from="igienico" to="higiénico" />
+    <Word from="igiénico" to="higiénico" />
+    <Word from="igreija" to="igreja" />
+    <Word from="iguasu" to="iguaçu" />
+    <Word from="ilacção" to="ilação" />
+    <Word from="imbigo" to="umbigo" />
+    <Word from="impecilho" to="empecilho" />
+    <Word from="íncas" to="incas" />
+    <Word from="incêsto" to="incesto" />
+    <Word from="inclusivé" to="inclusive" />
+    <Word from="incômodos" to="incómodos" />
+    <Word from="incontestávelmente" to="incontestavelmente" />
+    <Word from="incontestàvelmente" to="incontestavelmente" />
+    <Word from="indespensáveis" to="indispensáveis" />
+    <Word from="indespensável" to="indispensável" />
+    <Word from="India" to="Índia" />
+    <Word from="indiguinação" to="indignação" />
+    <Word from="indiguinado" to="indignado" />
+    <Word from="indiguinar" to="indignar" />
+    <Word from="inflacção" to="inflação" />
+    <Word from="ingreja" to="igreja" />
+    <Word from="INSCRICOES" to="INSCRIÇÕES" />
+    <Word from="intensão" to="intenção" />
+    <Word from="intertido" to="entretido" />
+    <Word from="intoxica" to="Intoxica" />
+    <Word from="intrega" to="entrega" />
+    <Word from="inverosímel" to="inverosímil" />
+    <Word from="iorgute" to="iogurte" />
+    <Word from="ipopótamo" to="hipopótamo" />
+    <Word from="ipsilon" to="ípsilon" />
+    <Word from="ipslon" to="ípsilon" />
+    <Word from="isquesito" to="esquisito" />
+    <Word from="juíz" to="juiz" />
+    <Word from="juiza" to="juíza" />
+    <Word from="júniores" to="juniores" />
+    <Word from="justanzente" to="justamente" />
+    <Word from="juz" to="jus" />
+    <Word from="kilo" to="quilo" />
+    <Word from="laboratório-porque" to="laboratório porque" />
+    <Word from="ladravaz" to="ladrava" />
+    <Word from="lamentàvelmente" to="lamentavelmente" />
+    <Word from="lampeão" to="lampião" />
+    <Word from="largartixa" to="lagartixa" />
+    <Word from="largarto" to="lagarto" />
+    <Word from="lêm" to="lêem" />
+    <Word from="leucémia" to="leucemia" />
+    <Word from="licensa" to="licença" />
+    <Word from="linguísta" to="linguista" />
+    <Word from="lisongear" to="lisonjear" />
+    <Word from="logista" to="lojista" />
+    <Word from="maçajar" to="massajar" />
+    <Word from="Macfadden-o" to="Macfadden o" />
+    <Word from="mae" to="mãe" />
+    <Word from="magestade" to="majestade" />
+    <Word from="mãgua" to="mágoa" />
+    <Word from="mangerico" to="manjerico" />
+    <Word from="mangerona" to="manjerona" />
+    <Word from="manteem-se" to="mantêm-se" />
+    <Word from="mantega" to="manteiga" />
+    <Word from="mantem-se" to="mantém-se" />
+    <Word from="massiço" to="maciço" />
+    <Word from="massisso" to="maciço" />
+    <Word from="médica-Rio" to="médica Rio" />
+    <Word from="menistro" to="ministro" />
+    <Word from="merciaria" to="mercearia" />
+    <Word from="metrelhadora" to="metralhadora" />
+    <Word from="miscegenação" to="miscigenação" />
+    <Word from="misogenia" to="misoginia" />
+    <Word from="misogeno" to="misógino" />
+    <Word from="misógeno" to="misógino" />
+    <Word from="mº" to="º" />
+    <Word from="môlho" to="molho" />
+    <Word from="monumentânea" to="momentânea" />
+    <Word from="mortandela" to="mortadela" />
+    <Word from="morteIa" to="mortela" />
+    <Word from="muinto" to="muito" />
+    <Word from="nasaias" to="nasais" />
+    <Word from="nêle" to="nele" />
+    <Word from="nest" to="neste" />
+    <Word from="Nivea" to="Nívea" />
+    <Word from="nonagessimo" to="nonagésimo" />
+    <Word from="nonagéssimo" to="nonagésimo" />
+    <Word from="nornal" to="normal" />
+    <Word from="notàvelmente" to="notavelmente" />
+    <Word from="obcessão" to="obsessão" />
+    <Word from="obesidae" to="obesidade" />
+    <Word from="óbviamente" to="obviamente" />
+    <Word from="òbviamente" to="obviamente" />
+    <Word from="ofecina" to="oficina" />
+    <Word from="oje" to="hoje" />
+    <Word from="omem" to="homem" />
+    <Word from="opcoes" to="opções" />
+    <Word from="opóbrio" to="opróbrio" />
+    <Word from="opróbio" to="opróbrio" />
+    <Word from="orfão" to="órfão" />
+    <Word from="organigrama" to="organograma" />
+    <Word from="organisar" to="organizar" />
+    <Word from="orgão" to="órgão" />
+    <Word from="orta" to="horta" />
+    <Word from="ótima" to="óptima" />
+    <Word from="ótimos" to="óptimos" />
+    <Word from="paralização" to="paralisação" />
+    <Word from="paralizado" to="paralisado" />
+    <Word from="paralizar" to="paralisar" />
+    <Word from="paráste" to="paraste" />
+    <Word from="Pátria" to="pátria" />
+    <Word from="paúl" to="Paul" />
+    <Word from="pecalço" to="percalço" />
+    <Word from="pêga" to="pega" />
+    <Word from="periodo" to="período" />
+    <Word from="pertubar" to="perturbar" />
+    <Word from="perú" to="peru" />
+    <Word from="piqueno" to="pequeno" />
+    <Word from="pirinéus" to="Pirenéus" />
+    <Word from="poblema" to="problema" />
+    <Word from="pobrema" to="problema" />
+    <Word from="poden" to="podem" />
+    <Word from="poder-mos" to="pudermos" />
+    <Word from="ponteagudo" to="pontiagudo" />
+    <Word from="pontuacoes" to="pontuações" />
+    <Word from="prazeiroso" to="prazeroso" />
+    <Word from="precaridade" to="precariedade" />
+    <Word from="precizar" to="precisar" />
+    <Word from="preserverança" to="perseverança" />
+    <Word from="previlégio" to="privilégio" />
+    <Word from="primária-que" to="primária que" />
+    <Word from="priúdo" to="período" />
+    <Word from="probalidade" to="probabilidade" />
+    <Word from="progreso" to="progresso" />
+    <Word from="proibído" to="proibido" />
+    <Word from="proíbido" to="proibido" />
+    <Word from="própia" to="própria" />
+    <Word from="propiedade" to="propriedade" />
+    <Word from="propio" to="próprio" />
+    <Word from="própio" to="próprio" />
+    <Word from="provocacoes" to="provocações" />
+    <Word from="prsença" to="presença" />
+    <Word from="prustituta" to="prostituta" />
+    <Word from="pudérmos" to="pudermos" />
+    <Word from="púlico" to="público" />
+    <Word from="pús" to="pus" />
+    <Word from="pusémos" to="pusemos" />
+    <Word from="quadricomia" to="quadricromia" />
+    <Word from="quadriplicado" to="quadruplicado" />
+    <Word from="quaisqueres" to="quaisquer" />
+    <Word from="quer-a" to="quere-a" />
+    <Word from="quere-se" to="quer-se" />
+    <Word from="quer-o" to="quere-o" />
+    <Word from="químco" to="químico" />
+    <Word from="quises-te" to="quiseste" />
+    <Word from="quizer" to="quiser" />
+    <Word from="quizeram" to="quiseram" />
+    <Word from="quizesse" to="quisesse" />
+    <Word from="quizessem" to="quisessem" />
+    <Word from="raínha" to="rainha" />
+    <Word from="raíz" to="raiz" />
+    <Word from="raizes" to="raízes" />
+    <Word from="ratato" to="retrato" />
+    <Word from="raúl" to="raul" />
+    <Word from="razar" to="rasar" />
+    <Word from="rectaguarda" to="retaguarda" />
+    <Word from="rédia" to="rédea" />
+    <Word from="reestabelecer" to="restabelecer" />
+    <Word from="refeicoes" to="refeições" />
+    <Word from="refêrencia" to="referência" />
+    <Word from="regeitar" to="rejeitar" />
+    <Word from="regurjitar" to="regurgitar" />
+    <Word from="reinvidicação" to="reivindicação" />
+    <Word from="reinvidicar" to="reivindicar" />
+    <Word from="requer-a" to="requere-a" />
+    <Word from="requere-se" to="requer-se" />
+    <Word from="requer-o" to="requere-o" />
+    <Word from="requesito" to="requisito" />
+    <Word from="requisicoes" to="requisições" />
+    <Word from="RESIDENCIA" to="RESIDÊNCIA" />
+    <Word from="respiraçáo" to="respiração" />
+    <Word from="restablecer" to="restabelecer" />
+    <Word from="réstea" to="réstia" />
+    <Word from="ruborisar" to="ruborizar" />
+    <Word from="rúbrica" to="rubrica" />
+    <Word from="sàdia" to="sadia" />
+    <Word from="saiem" to="saem" />
+    <Word from="salchicha" to="salsicha" />
+    <Word from="salchichas" to="salsichas" />
+    <Word from="saloice" to="saloiice" />
+    <Word from="salvé" to="salve" />
+    <Word from="salve-raínha" to="salve-rainha" />
+    <Word from="salvé-rainha" to="salve-rainha" />
+    <Word from="salvé-raínha" to="salve-rainha" />
+    <Word from="sao" to="são" />
+    <Word from="sargeta" to="sarjeta" />
+    <Word from="seções" to="secções" />
+    <Word from="seija" to="seja" />
+    <Word from="seissentos" to="seiscentos" />
+    <Word from="seje" to="seja" />
+    <Word from="semiar" to="semear" />
+    <Word from="séniores" to="seniores" />
+    <Word from="sensibilidadc" to="sensibilidade" />
+    <Word from="sensívelmente" to="sensivelmente" />
+    <Word from="setessentos" to="setecentos" />
+    <Word from="siclano" to="sicrano" />
+    <Word from="Sifilis" to="Sífilis" />
+    <Word from="sifílis" to="sífilis" />
+    <Word from="sinão" to="senão" />
+    <Word from="sinmtoma" to="sintoma" />
+    <Word from="sintéticamente" to="sinteticamente" />
+    <Word from="sintetisa" to="sintetiza" />
+    <Word from="SÓ" to="só" />
+    <Word from="sôfra" to="sofra" />
+    <Word from="sôfregamente" to="sofregamente" />
+    <Word from="somáste" to="somaste" />
+    <Word from="sombracelha" to="sobrancelha" />
+    <Word from="sombrancelha" to="sobrancelha" />
+    <Word from="sombrancelhas" to="sobrancelhas" />
+    <Word from="suavisar" to="suavizar" />
+    <Word from="substituido" to="substituído" />
+    <Word from="suburbio" to="subúrbio" />
+    <!-- <Word from="sues" to="seus" /> sues existe "Cuidado, não sues muito." -->
+    <Word from="suI" to="sul" />
+    <Word from="Suiça" to="Suíça" />
+    <Word from="suiças" to="suíças" />
+    <Word from="suiço" to="suíço" />
+    <Word from="suiços" to="suíços" />
+    <Word from="supôr" to="supor" />
+    <Word from="tabeliões" to="tabeliães" />
+    <Word from="taínha" to="tainha" />
+    <Word from="tava" to="estava" />
+    <Word from="têem" to="têm" />
+    <Word from="telemovel" to="telemóvel" />
+    <Word from="telémovel" to="telemóvel" />
+    <Word from="terminacoes" to="terminações" />
+    <Word from="toráxico" to="torácico" />
+    <Word from="tou" to="estou" />
+    <Word from="transpôr" to="transpor" />
+    <Word from="trasnporte" to="transporte" />
+    <Word from="tumors" to="tumores" />
+    <Word from="úmida" to="húmida" />
+    <Word from="umidade" to="unidade" />
+    <Word from="vai-vem" to="vaivém" />
+    <Word from="vegilância" to="vigilância" />
+    <Word from="vegilante" to="vigilante" />
+    <Word from="ventoínha" to="ventoinha" />
+    <Word from="verosímel" to="verosímil" />
+    <Word from="video" to="vídeo" />
+    <Word from="virus" to="vírus" />
+    <Word from="visiense" to="viseense" />
+    <Word from="voçe" to="você" />
+    <Word from="voçê" to="você" />
+    <Word from="vôo" to="voo" />
+    <Word from="xadrês" to="xadrez" />
+    <Word from="xafariz" to="chafariz" />
+    <Word from="xéxé" to="xexé" />
+    <Word from="xilindró" to="chilindró" />
+    <Word from="zaíre" to="Zaire" />
+    <Word from="zepelin" to="zepelim" />
+    <Word from="zig-zag" to="ziguezague" />
+    <Word from="zoô" to="zoo" />
+    <Word from="zôo" to="zoo" />
+    <Word from="zuar" to="zoar" />
+    <Word from="zum-zum" to="zunzum" />
+  </WholeWords>
+  <PartialWordsAlways />
+  <PartialWords />
+  <PartialLines>
+    <LinePart from="IN 6-E" to="N 6 E" />
+    <LinePart from="in tegrar-se" to="integrar-se" />
+    <LinePart from="in teresse" to="interesse" />
+    <LinePart from="in testinos" to="intestinos" />
+    <LinePart from="indica ção" to="indicação" />
+    <LinePart from="inte tino" to="intestino" />
+    <LinePart from="intes tinos" to="intestinos" />
+    <LinePart from="L da" to="Lda" />
+    <LinePart from="mal estar" to="mal-estar" />
+    <LinePart from="mastiga çáo" to="mastigação" />
+    <LinePart from="médi cas" to="médicas" />
+    <LinePart from="mineo rais" to="minerais" />
+    <LinePart from="mola res" to="molares" />
+    <LinePart from="movi mentos" to="movimentos" />
+    <LinePart from="movimen to" to="movimento" />
+    <LinePart from="N 5-Estendido" to="Nº 5 Estendido" />
+    <LinePart from="oxigé nio" to="oxigénio" />
+    <LinePart from="pod mos" to="podemos" />
+    <LinePart from="poder-se ia" to="poder-se-ia" />
+    <LinePart from="pos sibilidade" to="possibilidade" />
+    <LinePart from="possibi lidades" to="possibilidades" />
+    <LinePart from="pro duto" to="produto" />
+    <LinePart from="procu rar" to="procurar" />
+    <LinePart from="Q u e" to="Que" />
+    <LinePart from="qualifi cam" to="qualificam" />
+    <LinePart from="R egião" to="Região" />
+    <LinePart from="unsuficien temente" to="insuficientemente" />
+  </PartialLines>
+  <PartialLinesAlways />
+  <BeginLines />
+  <EndLines />
+  <WholeLines />
+  <RegularExpressions>
+    <!-- <RegEx find="\bi\b" replaceWith="I" /> just an example - do not use this regex -->
+    <RegEx find="([0-9]) +º" replaceWith="$1º" />
+    <RegEx find="\Bcao\b" replaceWith="ção" />
+    <RegEx find="\Bcoes\b" replaceWith="ções" />
+    <!-- <RegEx find="\Bccao\b" replaceWith="cção" /> não faz sentido ter este e ter a linha de cima -->
+    <!-- <RegEx find="\Bccoes\b" replaceWith="cções" /> não faz sentido ter este e ter a linha de cima -->
+    <RegEx find="\b(m|M)ae\b" replaceWith="$1ãe" />
+    <RegEx find="\Bdmnis\B" replaceWith="dminis" />
+    <RegEx find="\Blcól\B" replaceWith="lcoól" />
+    <RegEx find="\b(t|T)a[nm]b[eé]m\b" replaceWith="$1ambém" />
+    <RegEx find="\bzeppeli[mn]\b" replaceWith="zepelim" />
+    <RegEx find="\b(s|S)ufe?ciente\b" replaceWith="$1uficiente" />
+    <RegEx find="\b(n|N)ao\b" replaceWith="$1ão" />
+    <RegEx find="\b(B|b)elem\b" replaceWith="$1elém" />
+    <RegEx find="\b(s|S)u[íi]sso(s)?\b" replaceWith="$1uíço$2" />
+    <RegEx find="\b(s|S)u[íi]ssa(s)?\b" replaceWith="$1uíça$2" />
+    <RegEx find="\b(p|P)rivelig[ie]\p{Ll}d" replaceWith="$1rivelegiad" />
+    <RegEx find="\bpud(?:és|e-)se\b" replaceWith="pudesse" />
+    <RegEx find="\biquilíbr(?:e|i)o\b" replaceWith="equilíbrio" />
+    <RegEx find="\b(c|C)orregi\B" replaceWith="$1orrigid" />
+    <RegEx find="(?&lt;=A|a)ssociacao" replaceWith="ssociação" />
+    <RegEx find="(?&lt;=N|n)inguem" replaceWith="inguém" />
+    <RegEx find="(?&lt;=g|G)rat(?:uí|úi)to" replaceWith="ratuito" />
+    <RegEx find="(?&lt;=d|D)esiquilíbr[ei]o" replaceWith="esequilíbrio" />
+    <RegEx find="\b[k|K]il(ogramas?|ómetros?)" replaceWith="qui$1" />
+  </RegularExpressions>
+</OCRFixReplaceList>

Contents/Libraries/Shared/subzero/modification/dictionaries/xml/rus_OCRFixReplaceList.xml

@@ -0,0 +1,257 @@
+<OCRFixReplaceList>
+  <WholeWords>
+    <Word from="НЄЙ" to="НЕЙ" />
+    <Word from="ОРГЗНИЗМОБ" to="ОРГАНИЗМА" />
+    <Word from="Чї0" to="ЧТО" />
+    <Word from="НЭ" to="НА" />
+    <Word from="СОСЄДНЮЮ" to="СОСЕДНЮЮ" />
+    <Word from="ПЛЗНЄТУ" to="ПЛАНЕТУ" />
+    <Word from="ЗЗГЭДОК" to="ЗАГАДОК" />
+    <Word from="СОТВОРЄНИЯ" to="СОТВОРЕНИЯ" />
+    <Word from="МИРЭ" to="МИРА" />
+    <Word from="ПОЯБЛЄНИЯ" to="ПОЯВЛЕНИЯ" />
+    <Word from="ЗЄМЛЄ" to="ЗЕМЛЕ" />
+    <Word from="ЄЩЄ" to="ЕЩЁ" />
+    <Word from="ТЄМНЬІХ" to="ТЕМНЫХ" />
+    <Word from="СЄРЬЄЗНЬІМ" to="СЕРЬЕЗНЫМ" />
+    <Word from="ПОШІІ0" to="ПОШЛО" />
+    <Word from="Пр0ИЗ0ШЄЛ" to="ПРОИЗОШЕЛ" />
+    <Word from="СЄКРЄТЭМИ" to="СЕКРЕТАМИ" />
+    <Word from="МЭТЄРИЗЛЬІ" to="МАТЕРИАЛЫ" />
+    <Word from="ПЯТЄН" to="ПЯТЕН" />
+    <Word from="ПЛаНЄїЄ" to="ПЛАНЕТЕ" />
+    <Word from="КЗТЭКЛИЗМ" to="КАТАКЛИЗМ" />
+    <Word from="ОКЗЗЗЛСЯ" to="ОКАЗАЛСЯ" />
+    <Word from="ДЭЛЬШЕ" to="ДАЛЬШЕ" />
+    <Word from="ТВК" to="ТАК" />
+    <Word from="ПЛЗНЄТЗ" to="ПЛАНЕТА" />
+    <Word from="ЧЄГО" to="ЧЕГО" />
+    <Word from="УЗНЭТЬ" to="УЗНАТЬ" />
+    <Word from="ПЛЭНЄТЄ" to="ПЛАНЕТЕ" />
+    <Word from="НЄМ" to="НЕМ" />
+    <Word from="БОЗМОЖНЗ" to="ВОЗМОЖНА" />
+    <Word from="СОБЄРШЄННО" to="СОВЕРШЕННО" />
+    <Word from="ИНЭЧЄ" to="ИНАЧЕ" />
+    <Word from="БСЄ" to="ВСЕ" />
+    <Word from="НЕДОСТЗТКИ" to="НЕДОСТАТКИ" />
+    <Word from="НОВЬІЄ" to="НОВЫЕ" />
+    <Word from="ВЄЛИКОЛЄПНЭЯ" to="ВЕЛИКОЛЕПНАЯ" />
+    <Word from="ОСТЭІІОСЬ" to="ОСТАЛОСЬ" />
+    <Word from="НЗЛИЧИЄ" to="НАЛИЧИЕ" />
+    <Word from="бЫ" to="бы" />
+    <Word from="ПРОЦВЕТВТЬ" to="ПРОЦВЕТАТЬ" />
+    <Word from="КЗК" to="КАК" />
+    <Word from="ВОДЗ" to="ВОДА" />
+    <Word from="НЗШЕЛ" to="НАШЕЛ" />
+    <Word from="НЄ" to="НЕ" />
+    <Word from="ТОЖЄ" to="ТОЖЕ" />
+    <Word from="ВУЛКЭНИЧЄСКОЙ" to="ВУЛКАНИЧЕСКОЙ" />
+    <Word from="ЭКТИБНОСТИ" to="АКТИВНОСТИ" />
+    <Word from="ПОЯВИЛЗСЬ" to="ПОЯВИЛАСЬ" />
+    <Word from="НОВЗЯ" to="НОВАЯ" />
+    <Word from="СТРЭТЄГИЯ" to="СТРАТЕГИЯ" />
+    <Word from="УСПЄШН0" to="УСПЕШНО" />
+    <Word from="ПОСЗДКУ" to="ПОСАДКУ" />
+    <Word from="ГОТОБЫ" to="ГОТОВЫ" />
+    <Word from="НЗЧЗТЬ" to="НАЧАТЬ" />
+    <Word from="ОХОТЭ" to="ОХОТА" />
+    <Word from="ПРИЗНЗКЗМИ" to="ПРИЗНАКАМИ" />
+    <Word from="Пр0ШЛОМ" to="ПРОШЛОМ" />
+    <Word from="НЭСТОЯЩЄМ" to="НАСТОЯЩЕМ" />
+    <Word from="ПУСТОТЗХ" to="ПУСТОТАХ" />
+    <Word from="БЛЗЖНОЙ" to="ВЛАЖНОЙ" />
+    <Word from="ПОЧБЄ" to="ПОЧВЕ" />
+    <Word from="МЬІ" to="МЫ" />
+    <Word from="СЄЙЧЗС" to="СЕЙЧАС" />
+    <Word from="ЄСЛИ" to="ЕСЛИ" />
+    <Word from="ЗЗТРОНЕМ" to="ЗАТРОНЕМ" />
+    <Word from="ОПЗСЗЄМСЯ" to="ОПАСАЕМСЯ" />
+    <Word from="СИЛЬН0" to="СИЛЬНО" />
+    <Word from="ОТЛИЧЗЄТСЯ" to="ОТЛИЧАЕТСЯ" />
+    <Word from="РЭНЬШЄ" to="РАНЬШЕ" />
+    <Word from="НЗЗЬІВЗЮТ" to="НАЗЫВАЮТ" />
+    <Word from="ТЄКЛ3" to="ТЕКЛА" />
+    <Word from="ОСЗДОЧНЫМИ" to="ОСАДОЧНЫМИ" />
+    <Word from="ПОСТЄПЄНН0" to="ПОСТЕПЕННО" />
+    <Word from="ИСПЭРЯЛЗСЬ" to="ИСПАРЯЛАСЬ" />
+    <Word from="ЄОЛЬШОЄ" to="БОЛЬШОЕ" />
+    <Word from="КОЛИЧЄСТБО" to="КОЛИЧЕСТВО" />
+    <Word from="ГЄМЗТИТЕ" to="ГЕМАТИТА" />
+    <Word from="ПОЛУЧЭЄТ" to="ПОЛУЧАЕТ" />
+    <Word from="НЄДОСТЗЧН0" to="НЕДОСТАТОЧНО" />
+    <Word from="ПИТЭНИЯ" to="ПИТАНИЯ" />
+    <Word from="ПОКЗ" to="ПОКА" />
+    <Word from="БЬІХОДИЛИ" to="ВЫХОДИЛИ" />
+    <Word from="ЗЄМІІЄ" to="ЗЕМЛЕ" />
+    <Word from="ВЄСЬІИЗ" to="ВЕСЬМА" />
+    <Word from="ЗЄМЛИ" to="ЗЕМЛИ" />
+    <Word from="бЬІЛО" to="БЫЛО" />
+    <Word from="КИЗНИ" to="ЖИЗНИ" />
+    <Word from="СТЗНОВИЛЗСЬ" to="СТАНОВИЛАСЬ" />
+    <Word from="СОЛЄНЄЄ" to="СОЛЁНЕЕ" />
+    <Word from="МЭГНИТНЫМ" to="МАГНИТНЫМ" />
+    <Word from="ЧТОбЬІ" to="ЧТОБЫ" />
+    <Word from="СОЗДЕТЬ" to="СОЗДАТЬ" />
+    <Word from="МЗГНИТНОЄ" to="МАГНИТНОЕ" />
+    <Word from="КЭЖУТСЯ" to="КАЖУТСЯ" />
+    <Word from="ОЗНЗЧЗЄТ" to="ОЗНАЧАЕТ" />
+    <Word from="МОГЛЗ" to="МОГЛА" />
+    <Word from="ИМЄТЬ" to="ИМЕТЬ" />
+    <Word from="КОСМОСЭ" to="КОСМОСА" />
+    <Word from="СОЛНЄЧНЗЯ" to="СОЛНЕЧНАЯ" />
+    <Word from="СИСТЄМЗ" to="СИСТЕМА" />
+    <Word from="ПОСІІУЖИЛО" to="ПОСЛУЖИЛО" />
+    <Word from="МЗГНИТНОГО" to="МАГНИТНОГО" />
+    <Word from="ПЛВНЄТЫ" to="ПЛАНЕТЫ" />
+    <Word from="ЛОКЗЛЬНЬІХ" to="ЛОКАЛЬНЫХ" />
+    <Word from="ПОЛЄЙ" to="ПОЛЕЙ" />
+    <Word from="КЗЖУТСЯ" to="КАЖУТСЯ" />
+    <Word from="КЗКОГО" to="КАКОГО" />
+    <Word from="СТРЗШНОГО" to="СТРАШНОГО" />
+    <Word from="СТОЛКНОЕЄНИЯ" to="СТОЛКНОВЕНИЯ" />
+    <Word from="МЕСТЗМИ" to="МЕСТАМИ" />
+    <Word from="СДЄЛЗТЬ" to="СДЕЛАТЬ" />
+    <Word from="СТЗЛО" to="СТАЛО" />
+    <Word from="МЭГНИТНОГО" to="МАГНИТНОГО" />
+    <Word from="ЗЗКЛЮЧЗВШЄЙСЯ" to="ЗАКЛЮЧАВШЕЙСЯ" />
+    <Word from="ЄГО" to="ЕГО" />
+    <Word from="ЯДРЄ" to="ЯДРЕ" />
+    <Word from="НЗ" to="НА" />
+    <Word from="ИСЧЄЗЛ3" to="ИСЧЕЗЛА" />
+    <Word from="СЧИТЗЮ" to="СЧИТАЮ" />
+    <Word from="ШЭНСЫ" to="ШАНСЫ" />
+    <Word from="ИНЗЧЄ" to="ИНАЧЕ" />
+    <Word from="СТЗЛ" to="СТАЛ" />
+    <Word from="ТРЗТИТЬ" to="ТРАТИТЬ" />
+    <Word from="НЗПРЗВЛЯЄТСЯ" to="НАПРАВЛЯЕТСЯ" />
+    <Word from="ОБЛЭСТИ" to="ОБЛАСТИ" />
+    <Word from="ЯВЛЯІОТСЯ" to="ЯВЛЯЮТСЯ" />
+    <Word from="ГЛЭВНОЙ" to="ГЛАВНОЙ" />
+    <Word from="ДОКЗЗЗТЄЛЬСТВ" to="ДОКАЗАТЕЛЬСТВ" />
+    <Word from="КИСЛОТЭМИ" to="КИСЛОТАМИ" />
+    <Word from="ОНЭ" to="ОНА" />
+    <Word from="ПРЗКТИЧЄСКИ" to="ПРАКТИЧЕСКИ" />
+    <Word from="ЛЄСУ" to="ЛЕСУ" />
+    <Word from="УСЛОБИЯМ" to="УСЛОВИЯМ" />
+    <Word from="СПЗСТИСЬ" to="СПАСТИСЬ" />
+    <Word from="РЗЗВИВЗЮЩИЄСЯ" to="РАЗВИВАЮЩИЕСЯ" />
+    <Word from="ШЭПКИ" to="ШАПКИ" />
+    <Word from="ЗНЗЄМ" to="ЗНАЕМ" />
+    <Word from="СООИРЭЄМСЯ" to="СОБИРАЕМСЯ" />
+    <Word from="БЫЯСНИТЬ" to="ВЫЯСНИТЬ" />
+    <Word from="СЗМ" to="САМ" />
+    <Word from="РЗСПОЗНЗТЬ" to="РАСПОЗНАТЬ" />
+    <Word from="УЗНЗТЬ" to="УЗНАТЬ" />
+    <Word from="КЭЖЄТСЯ" to="КАЖЕТСЯ" />
+    <Word from="ОРЄИТЗЛЬНЬІЄ" to="ОРБИТАЛЬНЫЕ" />
+    <Word from="ЛЄТЭТЄЛЬНЬІЄ" to="ЛЕТАТЕЛЬНЫЕ" />
+    <Word from="ЗППЗРЕТЬІ" to="АППАРАТЫ" />
+    <Word from="ЖЄ" to="ЖЕ" />
+    <Word from="ТЗКЗЯ" to="ТАКАЯ" />
+    <Word from="МЗЛЄНЬКЗЯ" to="МАЛЕНЬКАЯ" />
+    <Word from="ПЛЭНЄТЗ" to="ПЛАНЕТА" />
+    <Word from="СПЗІІЬКО" to="СТОЛЬКО" />
+    <Word from="бЬІЛ3" to="БЫЛА" />
+    <Word from="ЁЕСЧИСЛЄННОЄ" to="БЕСЧИСЛЕННОЕ" />
+    <Word from="МЗГНИїНЬІХ" to="МАГНИТНЫХ" />
+    <Word from="ПОСТраД3Л" to="ПОСТРАДАЛ" />
+    <Word from="ДЗЖЄ" to="ДАЖЕ" />
+    <Word from="РЗЗНЬІМИ" to="РАЗНЫМИ" />
+    <Word from="СУЩЄСТБОВЭНИЄ" to="СУЩЕСТВОВАНИЕ" />
+    <Word from="ПЛаНЄїЬІ" to="ПЛАНЕТЫ" />
+    <Word from="ПОДВЄРГЛЗСЬ" to="ПОДВЕРГЛАСЬ" />
+    <Word from="ОПЗСІ-ІОСТИ" to="ОПАСНОСТИ" />
+    <Word from="ПЛЗНЄТЄ" to="ПЛАНЕТЕ" />
+    <Word from="Н0" to="НО" />
+    <Word from="бЬІ" to="БЫ" />
+    <Word from="ОТДЗЛЄННЫЄ" to="ОТДАЛЁННЫЕ" />
+    <Word from="ПОЛЯРНЬІЄ" to="ПОЛЯРНЫЕ" />
+    <Word from="ЦЄЛЬІ-О" to="ЦЕЛЬЮ" />
+    <Word from="ПЄЩЄРЗХ" to="ПЕЩЕРАХ" />
+    <Word from="НЗПОЛНЄННЬІХ" to="НАПОЛНЕННЫХ" />
+    <Word from="ИСПЗРЄНИЯМИ" to="ИСПАРЕНИЯМИ" />
+    <Word from="МИНИЗТЮРНЬІЄ" to="МИНИАТЮРНЫЕ" />
+    <Word from="ТЭКЗЯ" to="ТАКАЯ" />
+    <Word from="ПрИСП0СОбИТЬСЯ" to="ПРИСПОСОБИТЬСЯ" />
+    <Word from="НЄОЄХОДИМЬІЄ" to="НЕОБХОДИМЫЕ" />
+    <Word from="ОРГВНИЧЄСКИЄ" to="ОРГАНИЧЕСКИЕ" />
+    <Word from="МЗРСИЗНСКИЄ" to="МАРСИАНСКИЕ" />
+    <Word from="МЄСТЄ" to="МЕСТЕ" />
+    <Word from="І\/ІАККЕЙШ" to="МАККЕЙН" />
+    <Word from="НЗХОДЯЩИЄСЯ" to="НАХОДЯЩИЕСЯ" />
+    <Word from="НЄЗКТИВНОМ" to="НЕАКТИВНОМ" />
+    <Word from="ЗЭСНЯТЬ" to="ЗАСНЯТЬ" />
+    <Word from="ОРГЗНИЗМЬІ" to="ОРГАНИЗМЫ" />
+    <Word from="ВЗЕИМОДЄЙСТВОВЕТЬ" to="ВЗАИМОДЕЙСТВОВАТЬ" />
+    <Word from="ПУТЄШЄСТБИЄ" to="ПУТЕШЕСТВИЕ" />
+    <Word from="ПуСїЬІННЫХ" to="ПУСТЫННЫХ" />
+    <Word from="ТЗКИХ" to="ТАКИХ" />
+    <Word from="ПЄРЄТЗСКИВЗЄМ" to="ПЕРЕТАСКИВАЕМ" />
+    <Word from="ЧТ0" to="ЧТО" />
+    <Word from="ВЄСЬМЗ" to="ВЕСЬМА" />
+    <Word from="ПОЛОСЗМИ" to="ПОЛОСАМИ" />
+    <Word from="ОрїЭНИЗМЬІ" to="ОРГАНИЗМЫ" />
+    <Word from="ОЁЛЗСТИ" to="ОБЛАСТИ" />
+    <Word from="ЯБЛЯЮТСЯ" to="ЯВЛЯЮТСЯ" />
+    <Word from="ЦЄЛЬЮ" to="ЦЕЛЬЮ" />
+    <Word from="ПОИСКОБ" to="ПОИСКОВ" />
+    <Word from="ДОКЗЗЗТЄІІЬСТВ" to="ДОКАЗАТЕЛЬСТВ" />
+    <Word from="МОЖЄТ" to="МОЖЕТ" />
+    <Word from="НЭХОДИТЬСЯ" to="НАХОДИТЬСЯ" />
+    <Word from="ОЧЄНЬ" to="ОЧЕНЬ" />
+    <Word from="СРЗВНИТЬ" to="СРАВНИТЬ" />
+    <Word from="ОЄНЗРУЖИЛ" to="ОБНАРУЖИЛ" />
+    <Word from="ЛЬДЗ" to="ЛЬДА" />
+    <Word from="ПОТЄПЛЄНИЄІИ" to="ПОТЕПЛЕНИЕМ" />
+    <Word from="ПОХОЛОДЗНИЄБД" to="ПОХОЛОДАНИЕМ" />
+    <Word from="КЭК" to="КАК" />
+    <Word from="ТЄЛО" to="ТЕЛО" />
+    <Word from="бОЛЬШЄ" to="БОЛЬШЕ" />
+    <Word from="НЭКЛОНЯЄТСЯ" to="НАКЛОНЯЕТСЯ" />
+    <Word from="СОІІНЦУ" to="СОЛНЦУ" />
+    <Word from="СТ3бИЛИЗИрОБЗТЬ" to="СТАБИЛИЗИРОВАТЬ" />
+    <Word from="СТЭБИЛЬНЭ" to="СТАБИЛЬНА" />
+    <Word from="МИЛІІИОНОВ" to="МИЛЛИОНОВ" />
+    <Word from="НЗЗЭД" to="НАЗАД" />
+    <Word from="ТЄПЛ0" to="ТЕПЛО" />
+    <Word from="ПОІІЯРНЫХ" to="ПОЛЯРНЫХ" />
+    <Word from="СОІІЕНЫМИ" to="СОЛЕНЫМИ" />
+    <Word from="КЕКИМИ" to="КАКИМИ" />
+    <Word from="кислютнюсггь" to="кислотность" />
+    <Word from="ТЗМ" to="ТАМ" />
+    <Word from="ОРГЗНИЗМЫ" to="ОРГАНИЗМЫ" />
+    <Word from="СУЩЄСТВОВЄТЬ" to="СУЩЕСТВОВАТЬ" />
+    <Word from="ВНИМЗНИЄ" to="ВНИМАНИЕ" />
+    <Word from="СДЄЛЗЄТ" to="СДЕЛАЕТ" />
+    <Word from="ПОЗНЭКОМИТЬСЯ" to="ПОЗНАКОМИТЬСЯ" />
+    <Word from="НЭШИМ" to="НАШИМ" />
+    <Word from="ДОКЗЗЭТЄЛЬСТБО" to="ДОКАЗАТЕЛЬСТВО" />
+    <Word from="ЩЗЗЩЄНИЯ" to="ВРАЩЕНИЯ" />
+    <Word from="бЬІЛ0" to="БЫЛО" />
+    <Word from="ОЄЛЕСТЯХ" to="ОБЛАСТЯХ" />
+    <Word from="бЬІЛИ" to="БЫЛИ" />
+    <Word from="РЭЗМЬІШЛЯІІИ" to="РАЗМЫШЛЯЛИ" />
+    <Word from="КОЛИЧЄСТБЄ" to="КОЛИЧЕСТВЕ" />
+    <Word from="ЩЄІІОЧНЫЄ" to="ЩЕЛОЧНЫЕ" />
+    <Word from="НЄКОТЩЗЬІЄ" to="НЕКОТОРЫЕ" />
+    <Word from="ПрИБІ1ЕКуї" to="ПРИВЛЕКУТ" />
+    <Word from="НЗЗЬІВЭЄМЫЄ" to="НАЗЫВАЕМЫЕ" />
+    <Word from="Чї06Ы" to="ЧТОБЫ" />
+  </WholeWords>
+  <PartialWordsAlways />
+  <PartialWords>
+    <WordPart from="Є" to="Е" />
+    <WordPart from="ЬІ" to="Ы" />
+    <WordPart from="КЗ" to="КА" />
+    <WordPart from="ЛЗ" to="ЛА" />
+    <WordPart from="НЗ" to="НА" />
+    <WordPart from="ШЗ" to="ША" />
+    <WordPart from="І\/І" to="М" />
+  </PartialWords>
+  <PartialLines />
+  <PartialLinesAlways />
+  <BeginLines />
+  <EndLines />
+  <WholeLines />
+  <RegularExpressions />
+</OCRFixReplaceList>

Contents/Libraries/Shared/subzero/modification/dictionaries/xml/spa_OCRFixReplaceList.xml

@@ -0,0 +1,946 @@
+<OCRFixReplaceList>
+  <WholeWords>
+    <!-- Abreviaturas simples -->
+    <Word from="KBs" to="kB" />
+    <Word from="Vd" to="Ud" />
+    <Word from="N°" to="N.°" />
+    <Word from="n°" to="n.°" />
+    <Word from="nro." to="n.°" />
+    <Word from="Nro." to="N.°" />
+    <!-- Ortografía básica -->
+    <Word from="aca" to="acá" />
+    <Word from="actuas" to="actúas" />
+    <Word from="actues" to="actúes" />
+    <Word from="adios" to="adiós" />
+    <Word from="agarrenla" to="agárrenla" />
+    <Word from="agarrenlo" to="agárrenlo" />
+    <Word from="agarrandose" to="agarrándose" />
+    <Word from="algun" to="algún" />
+    <Word from="alli" to="allí" />
+    <Word from="alla" to="allá" />
+    <Word from="alejate" to="aléjate" />
+    <Word from="ahi" to="ahí" />
+    <Word from="angel" to="ángel" />
+    <Word from="angeles" to="ángeles" />
+    <Word from="apagala" to="apágala" />
+    <Word from="aqui" to="aquí" />
+    <Word from="asi" to="así" />
+    <Word from="bahia" to="bahía" />
+    <Word from="busqueda" to="búsqueda" />
+    <Word from="busquedas" to="búsquedas" />
+    <Word from="callate" to="cállate" />
+    <Word from="carcel" to="cárcel" />
+    <Word from="camara" to="cámara" />
+    <Word from="caido" to="caído" />
+    <Word from="cabron" to="cabrón" />
+    <Word from="camion" to="camión" />
+    <Word from="codigo" to="código" />
+    <Word from="codigos" to="códigos" />
+    <Word from="comence" to="comencé" />
+    <Word from="comprate" to="cómprate" />
+    <Word from="consegui" to="conseguí" />
+    <Word from="confias" to="confías" />
+    <Word from="convertira" to="convertirá" />
+    <Word from="corazon" to="corazón" />
+    <Word from="crei" to="creí" />
+    <Word from="creia" to="creía" />
+    <Word from="creido" to="creído" />
+    <Word from="creiste" to="creíste" />
+    <Word from="cubrenos" to="cúbrenos" />
+    <Word from="comio" to="comió" />
+    <Word from="dara" to="dará" />
+    <Word from="dia" to="día" />
+    <Word from="dias" to="días" />
+    <Word from="debio" to="debió" />
+    <Word from="demelo" to="démelo" />
+    <Word from="dimelo" to="dímelo" />
+    <Word from="denoslo" to="dénoslo" />
+    <Word from="deselo" to="déselo" />
+    <Word from="decia" to="decía" />
+    <Word from="decian" to="decían" />
+    <Word from="detras" to="detrás" />
+    <Word from="deberia" to="debería" />
+    <Word from="deberas" to="deberás" />
+    <Word from="deberias" to="deberías" />
+    <Word from="deberian" to="deberían" />
+    <Word from="deberiamos" to="deberíamos" />
+    <Word from="dejame" to="déjame" />
+    <Word from="dejate" to="déjate" />
+    <Word from="dejalo" to="déjalo" />
+    <Word from="dejarian" to="dejarían" />
+    <Word from="damela" to="dámela" />
+    <Word from="despues" to="después" />
+    <Word from="diciendome" to="diciéndome" />
+    <Word from="dificil" to="difícil" />
+    <Word from="dificiles" to="difíciles" />
+    <Word from="disculpate" to="discúlpate" />
+    <Word from="dolares" to="dólares" />
+    <Word from="hechar" to="echar" />
+    <Word from="examenes" to="exámenes" />
+    <Word from="empezo" to="empezó" />
+    <Word from="empujon" to="empujón" />
+    <Word from="empujalo" to="empújalo" />
+    <Word from="escondanme" to="escóndanme" />
+    <Word from="esperame" to="espérame" />
+    <Word from="estara" to="estará" />
+    <Word from="estare" to="estaré" />
+    <Word from="estaria" to="estaría" />
+    <Word from="estan" to="están" />
+    <Word from="estaran" to="estarán" />
+    <Word from="estabamos" to="estábamos" />
+    <Word from="estuvieramos" to="estuviéramos" />
+    <Word from="exito" to="éxito" />
+    <Word from="facil" to="fácil" />
+    <Word from="fiscalia" to="fiscalía" />
+    <Word from="fragil" to="frágil" />
+    <Word from="fragiles" to="frágiles" />
+    <Word from="frances" to="francés" />
+    <Word from="gustaria" to="gustaría" />
+    <Word from="habia" to="había" />
+    <Word from="habias" to="habías" />
+    <Word from="habian" to="habían" />
+    <Word from="habrian" to="habrían" />
+    <Word from="habrias" to="habrías" />
+    <Word from="hagalo" to="hágalo" />
+    <Word from="haria" to="haría" />
+    <Word from="increible" to="increíble" />
+    <Word from="incredulo" to="incrédulo" />
+    <Word from="intentalo" to="inténtalo" />
+    <Word from="ire" to="iré" />
+    <Word from="jovenes" to="jóvenes" />
+    <Word from="ladron" to="ladrón" />
+    <Word from="linea" to="línea" />
+    <Word from="llamame" to="llámame" />
+    <Word from="llevalo" to="llévalo" />
+    <Word from="mama" to="mamá" />
+    <Word from="maricon" to="maricón" />
+    <Word from="mayoria" to="mayoría" />
+    <Word from="metodo" to="método" />
+    <Word from="metodos" to="métodos" />
+    <Word from="mio" to="mío" />
+    <Word from="mostro" to="mostró" />
+    <Word from="morira" to="morirá" />
+    <Word from="muevete" to="muévete" />
+    <Word from="murio" to="murió" />
+    <Word from="numero" to="número" />
+    <Word from="numeros" to="números" />
+    <Word from="ningun" to="ningún" />
+    <Word from="oido" to="oído" />
+    <Word from="oidos" to="oídos" />
+    <Word from="oimos" to="oímos" />
+    <Word from="oiste" to="oíste" />
+    <Word from="pasale" to="pásale" />
+    <Word from="pasame" to="pásame" />
+    <Word from="paraiso" to="paraíso" />
+    <Word from="parate" to="párate" />
+    <Word from="pense" to="pensé" />
+    <Word from="peluqueria" to="peluquería" />
+    <Word from="platano" to="plátano" />
+    <Word from="plastico" to="plástico" />
+    <Word from="plasticos" to="plásticos" />
+    <Word from="policia" to="policía" />
+    <Word from="policias" to="policías" />
+    <Word from="poster" to="póster" />
+    <Word from="podia" to="podía" />
+    <Word from="podias" to="podías" />
+    <Word from="podria" to="podría" />
+    <Word from="podrian" to="podrían" />
+    <Word from="podrias" to="podrías" />
+    <Word from="podriamos" to="podríamos" />
+    <Word from="prometio" to="prometió" />
+    <Word from="proposito" to="propósito" />
+    <Word from="pideselo" to="pídeselo" />
+    <Word from="ponganse" to="pónganse" />
+    <Word from="prometeme" to="prométeme" />
+    <Word from="publico" to="público" />
+    <Word from="publicos" to="públicos" />
+    <Word from="publicamente" to="públicamente" />
+    <Word from="quedate" to="quédate" />
+    <Word from="queria" to="quería" />
+    <Word from="querrias" to="querrías" />
+    <Word from="querian" to="querían" />
+    <Word from="rapido" to="rápido" />
+    <Word from="rapidamente" to="rápidamente" />
+    <Word from="razon" to="razón" />
+    <Word from="rehusen" to="rehúsen" />
+    <Word from="rie" to="ríe" />
+    <Word from="rias" to="rías" />
+    <Word from="rindete" to="ríndete" />
+    <Word from="sacame" to="sácame" />
+    <Word from="sentian" to="sentían" />
+    <Word from="sientate" to="siéntate" />
+    <Word from="sera" to="será" />
+    <Word from="soplon" to="soplón" />
+    <Word from="sueltalo" to="suéltalo" />
+    <Word from="tambien" to="también" />
+    <Word from="teoria" to="teoría" />
+    <Word from="tendra" to="tendrá" />
+    <Word from="telefono" to="teléfono" />
+    <Word from="tipica" to="típica" />
+    <Word from="todavia" to="todavía" />
+    <Word from="tomalo" to="tómalo" />
+    <Word from="tonterias" to="tonterías" />
+    <Word from="torci" to="torcí" />
+    <Word from="traelos" to="tráelos" />
+    <Word from="traiganlo" to="tráiganlo" />
+    <Word from="traiganlos" to="tráiganlos" />
+    <Word from="trio" to="trío" />
+    <Word from="tuvieramos" to="tuviéramos" />
+    <Word from="union" to="unión" />
+    <Word from="ultimo" to="último" />
+    <Word from="ultima" to="última" />
+    <Word from="ultimos" to="últimos" />
+    <Word from="ultimas" to="últimas" />
+    <Word from="unica" to="única" />
+    <Word from="unico" to="único" />
+    <Word from="vamonos" to="vámonos" />
+    <Word from="vayanse" to="váyanse" />
+    <Word from="victima" to="víctima" />
+    <Word from="vivira" to="vivirá" />
+    <Word from="volvio" to="volvió" />
+    <Word from="volvia" to="volvía" />
+    <Word from="volvian" to="volvían" />
+    <!-- Palabras con eír/oír más usadas -->
+    <Word from="reir" to="reír" />
+    <Word from="freir" to="freír" />
+    <Word from="sonreir" to="sonreír" />
+    <Word from="hazmerreir" to="hazmerreír" />
+    <Word from="oir" to="oír" />
+    <Word from="oirlo" to="oírlo" />
+    <Word from="oirte" to="oírte" />
+    <Word from="oirse" to="oírse" />
+    <Word from="oirme" to="oírme" />
+    <Word from="oirle" to="oírle" />
+    <Word from="oirla" to="oírla" />
+    <Word from="oirles" to="oírles" />
+    <Word from="oirnos" to="oírnos" />
+    <Word from="oirlas" to="oírlas" />
+    <!-- Palabras que no llevan acento -->
+    <Word from="bién" to="bien" />
+    <Word from="crímen" to="crimen" />
+    <Word from="fué" to="fue" />
+    <Word from="fuí" to="fui" />
+    <Word from="quiéres" to="quieres" />
+    <Word from="tí" to="ti" />
+    <Word from="dí" to="di" />
+    <Word from="vá" to="va" />
+    <Word from="vé" to="ve" />
+    <Word from="ví" to="vi" />
+    <Word from="vió" to="vio" />
+    <Word from="ó" to="o" />
+    <Word from="clón" to="clon" />
+    <Word from="dió" to="dio" />
+    <Word from="guión" to="guion" />
+    <Word from="dón" to="don" />
+    <Word from="fé" to="fe" />
+    <Word from="áquel" to="aquel" />
+    <!-- Palabras donde se puede prescindir de la tilde diacrítica -->
+    <Word from="éste" to="este" />
+    <Word from="ésta" to="esta" />
+    <Word from="éstos" to="estos" />
+    <Word from="éstas" to="estas" />
+    <Word from="ése" to="ese" />
+    <Word from="ésa" to="esa" />
+    <Word from="ésos" to="esos" />
+    <Word from="ésas" to="esas" />
+    <Word from="sólo" to="solo" />
+    <!-- Errores no relacionados con los tildes -->
+    <Word from="coktel" to="cóctel" />
+    <Word from="cocktel" to="cóctel" />
+    <Word from="conciente" to="consciente" />
+    <Word from="comenzé" to="comencé" />
+    <Word from="desilucionarte" to="desilusionarte" />
+    <Word from="dijieron" to="dijeron" />
+    <Word from="empezé" to="empecé" />
+    <Word from="hize" to="hice" />
+    <Word from="ilucionarte" to="ilusionarte" />
+    <Word from="inconciente" to="inconsciente" />
+    <Word from="quize" to="quise" />
+    <Word from="quizo" to="quiso" />
+    <Word from="verguenza" to="vergüenza" />
+    <!-- Errores en nombres propios o de países -->
+    <Word from="Nuñez" to="Núñez" />
+    <Word from="Ivan" to="Iván" />
+    <Word from="Japon" to="Japón" />
+    <Word from="Monica" to="Mónica" />
+    <Word from="Maria" to="María" />
+    <Word from="Jose" to="José" />
+    <Word from="Ramon" to="Ramón" />
+    <Word from="Garcia" to="García" />
+    <Word from="Gonzalez" to="González" />
+    <Word from="Jesus" to="Jesús" />
+    <Word from="Alvarez" to="Álvarez" />
+    <Word from="Damian" to="Damián" />
+    <Word from="Rene" to="René" />
+    <Word from="Nicolas" to="Nicolás" />
+    <Word from="Jonas" to="Jonás" />
+    <Word from="Lopez" to="López" />
+    <Word from="Hernandez" to="Hernández" />
+    <Word from="Bermudez" to="Bermúdez" />
+    <Word from="Fernandez" to="Fernández" />
+    <Word from="Suarez" to="Suárez" />
+    <Word from="Sofia" to="Sofía" />
+    <Word from="Seneca" to="Séneca" />
+    <Word from="Tokyo" to="Tokio" />
+    <Word from="Canada" to="Canadá" />
+    <Word from="Paris" to="París" />
+    <Word from="Turquia" to="Turquía" />
+    <Word from="Mexico" to="México" />
+    <Word from="Mejico" to="México" />
+    <Word from="Matias" to="Matías" />
+    <Word from="Valentin" to="Valentín" />
+    <Word from="mejicano" to="mexicano" />
+    <Word from="mejicanos" to="mexicanos" />
+    <Word from="mejicana" to="mexicana" />
+    <Word from="mejicanas" to="mexicanas" />
+    <!-- Creados por SE -->
+    <Word from="io" to="lo" />
+    <Word from="ia" to="la" />
+    <Word from="ie" to="le" />
+    <Word from="Io" to="lo" />
+    <Word from="Ia" to="la" />
+    <Word from="AI" to="Al" />
+    <Word from="Ie" to="le" />
+    <Word from="EI" to="El" />
+    <Word from="subafluente" to="subafluente" />
+    <Word from="aflójalo" to="aflójalo" />
+    <Word from="Aflójalo" to="Aflójalo" />
+    <Word from="perdi" to="perdí" />
+    <Word from="Podria" to="Podría" />
+    <Word from="confia" to="confía" />
+    <Word from="pasaria" to="pasaría" />
+    <Word from="Podias" to="Podías" />
+    <Word from="responsabke" to="responsable" />
+    <Word from="Todavia" to="Todavía" />
+    <Word from="envien" to="envíen" />
+    <Word from="Queria" to="Quería" />
+    <Word from="tio" to="tío" />
+    <Word from="traido" to="traído" />
+    <Word from="Asi" to="Así" />
+    <Word from="elegi" to="elegí" />
+    <Word from="habria" to="habría" />
+    <Word from="encantaria" to="encantaría" />
+    <Word from="leido" to="leído" />
+    <Word from="conocias" to="conocías" />
+    <Word from="harias" to="harías" />
+    <Word from="Aqui" to="Aquí" />
+    <Word from="decidi" to="decidí" />
+    <Word from="mia" to="mía" />
+    <Word from="Crei" to="Creí" />
+    <Word from="podiamos" to="podíamos" />
+    <Word from="avisame" to="avísame" />
+    <Word from="debia" to="debía" />
+    <Word from="pensarias" to="pensarías" />
+    <Word from="reuniamos" to="reuníamos" />
+    <Word from="POÏ" to="por" />
+    <Word from="vendria" to="vendría" />
+    <Word from="caida" to="caída" />
+    <Word from="venian" to="venían" />
+    <Word from="compañias" to="compañías" />
+    <Word from="leiste" to="leíste" />
+    <Word from="Leiste" to="Leíste" />
+    <Word from="fiaria" to="fiaría" />
+    <Word from="Hungria" to="Hungría" />
+    <Word from="fotografia" to="fotografía" />
+    <Word from="cafeteria" to="cafetería" />
+    <Word from="Digame" to="Dígame" />
+    <Word from="debias" to="debías" />
+    <Word from="tendria" to="tendría" />
+    <Word from="CÏGO" to="creo" />
+    <Word from="anteg" to="antes" />
+    <Word from="SóIo" to="Solo" />
+    <Word from="Ilamándola" to="llamándola" />
+    <Word from="Cáflaté" to="Cállate" />
+    <Word from="Ilamaste" to="llamaste" />
+    <Word from="daria" to="daría" />
+    <Word from="Iargaba" to="largaba" />
+    <Word from="Yati" to="Y a ti" />
+    <Word from="querias" to="querías" />
+    <Word from="Iimpiarlo" to="limpiarlo" />
+    <Word from="Iargado" to="largado" />
+    <Word from="galeria" to="galería" />
+    <Word from="Bartomeu" to="Bertomeu" />
+    <Word from="Iocalizarlo" to="localizarlo" />
+    <Word from="Ilámame" to="llámame" />
+  </WholeWords>
+  <PartialWordsAlways />
+  <PartialWords />
+  <PartialLines>
+    <!-- Varios -->
+    <LinePart from="de gratis" to="gratis" />
+    <LinePart from="si quiera" to="siquiera" />
+    <LinePart from="Cada una de los" to="Cada uno de los" />
+    <LinePart from="Cada uno de las" to="Cada una de las" />
+    <!-- Uso incorrecto de haber / a ver -->
+    <LinePart from="haber que" to="a ver qué" />
+    <LinePart from="haber qué" to="a ver qué" />
+    <LinePart from="Haber si" to="A ver si" />
+    <!-- Ponombres exclamativos o interrogativos Parte 1 -->
+    <LinePart from=" que hora" to=" qué hora" />
+    <LinePart from="yo que se" to="yo qué sé" />
+    <LinePart from="Yo que se" to="Yo qué sé" />
+    <!-- Acentos al final de los signos de exclamación -->
+    <LinePart from=" tu!" to=" tú!" />
+    <LinePart from=" si!" to=" sí!" />
+    <LinePart from=" mi!" to=" mí!" />
+    <LinePart from=" el!" to=" él!" />
+    <!-- Acentos al final de los signos de interrogación -->
+    <LinePart from=" tu?" to=" tú?" />
+    <LinePart from=" si?" to=" sí?" />
+    <LinePart from=" mi?" to=" mí?" />
+    <LinePart from=" el?" to=" él?" />
+    <LinePart from=" aun?" to=" aún?" />
+    <LinePart from=" mas?" to=" más?" />
+    <LinePart from=" que?" to=" qué?" />
+    <LinePart from=" paso?" to=" pasó?" />
+    <LinePart from=" cuando?" to=" cuándo?" />
+    <LinePart from=" cuanto?" to=" cuánto?" />
+    <LinePart from=" cuanta?" to=" cuánta?" />
+    <LinePart from=" cuantas?" to=" cuántas?" />
+    <LinePart from=" cuantos?" to=" cuántos?" />
+    <LinePart from=" donde?" to=" dónde?" />
+    <LinePart from=" quien?" to=" quién?" />
+    <LinePart from=" como?" to=" cómo?" />
+    <LinePart from=" adonde?" to=" adónde?" />
+    <LinePart from=" cual?" to=" cuál?" />
+    <!-- Acentos en los signos de interrogación completos -->
+    <LinePart from="¿Si?" to="¿Sí?" />
+    <LinePart from="¿esta bien?" to="¿está bien?" />
+    <!-- Enunciados que son a la vez interrogativos y exclamativos -->
+    <LinePart from="¿Pero qué haces?" to="¡¿Pero qué haces?!" />
+    <LinePart from="¿pero qué haces?" to="¡¿pero qué haces?!" />
+    <LinePart from="¿Es que no me has escuchado?" to="¡¿Es que no me has escuchado?!" />
+    <LinePart from="¡¿es que no me has escuchado?!" to="¡¿es que no me has escuchado?!" />
+    <!-- Acentos al principio de los signos de interrogación con minúsculas -->
+    <LinePart from="¿aun" to="¿aún" />
+    <LinePart from="¿tu " to="¿tú " />
+    <LinePart from="¿que " to="¿qué " />
+    <LinePart from="¿sabes que" to="¿sabes qué" />
+    <LinePart from="¿sabes adonde" to="¿sabes adónde" />
+    <LinePart from="¿sabes cual" to="¿sabes cuál" />
+    <LinePart from="¿sabes quien" to="¿sabes quién" />
+    <LinePart from="¿sabes como" to="¿sabes cómo" />
+    <LinePart from="¿sabes cuan" to="¿sabes cuán" />
+    <LinePart from="¿sabes cuanto" to="¿sabes cuánto" />
+    <LinePart from="¿sabes cuanta" to="¿sabes cuánta" />
+    <LinePart from="¿sabes cuantos" to="¿sabes cuántos" />
+    <LinePart from="¿sabes cuantas" to="¿sabes cuántas" />
+    <LinePart from="¿sabes cuando" to="¿sabes cuándo" />
+    <LinePart from="¿sabes donde" to="¿sabes dónde" />
+    <LinePart from="¿sabe que" to="¿sabe qué" />
+    <LinePart from="¿sabe adonde" to="¿sabe adónde" />
+    <LinePart from="¿sabe cual" to="¿sabe cuál" />
+    <LinePart from="¿sabe quien" to="¿sabe quién" />
+    <LinePart from="¿sabe como" to="¿sabe cómo" />
+    <LinePart from="¿sabe cuan" to="¿sabe cuán" />
+    <LinePart from="¿sabe cuanto" to="¿sabe cuánto" />
+    <LinePart from="¿sabe cuanta" to="¿sabe cuánta" />
+    <LinePart from="¿sabe cuantos" to="¿sabe cuántos" />
+    <LinePart from="¿sabe cuantas" to="¿sabe cuántas" />
+    <LinePart from="¿sabe cuando" to="¿sabe cuándo" />
+    <LinePart from="¿sabe donde" to="¿sabe dónde" />
+    <LinePart from="¿saben que" to="¿saben qué" />
+    <LinePart from="¿saben adonde" to="¿saben adónde" />
+    <LinePart from="¿saben cual" to="¿saben cuál" />
+    <LinePart from="¿saben quien" to="¿saben quién" />
+    <LinePart from="¿saben como" to="¿saben cómo" />
+    <LinePart from="¿saben cuan" to="¿saben cuán" />
+    <LinePart from="¿saben cuanto" to="¿saben cuánto" />
+    <LinePart from="¿saben cuanta" to="¿saben cuánta" />
+    <LinePart from="¿saben cuantos" to="¿saben cuántos" />
+    <LinePart from="¿saben cuantas" to="¿saben cuántas" />
+    <LinePart from="¿saben cuando" to="¿saben cuándo" />
+    <LinePart from="¿saben donde" to="¿saben dónde" />
+    <LinePart from="¿de que" to="¿de qué" />
+    <LinePart from="¿de donde" to="¿de dónde" />
+    <LinePart from="¿de cual" to="¿de cuál" />
+    <LinePart from="¿de quien" to="¿de quién" />
+    <LinePart from="¿de cuanto" to="¿de cuánto" />
+    <LinePart from="¿de cuanta" to="¿de cuánta" />
+    <LinePart from="¿de cuantos" to="¿de cuántos" />
+    <LinePart from="¿de cuantas" to="¿de cuántas" />
+    <LinePart from="¿de cuando" to="¿de cuándo" />
+    <LinePart from="¿sobre que" to="¿sobre qué" />
+    <LinePart from="¿como " to="¿cómo " />
+    <LinePart from="¿cual " to="¿cuál " />
+    <LinePart from="¿en cual" to="¿en cuál" />
+    <LinePart from="¿cuando" to="¿cuándo" />
+    <LinePart from="¿hasta cual" to="¿hasta cuál" />
+    <LinePart from="¿hasta quien" to="¿hasta quién" />
+    <LinePart from="¿hasta cuanto" to="¿hasta cuánto" />
+    <LinePart from="¿hasta cuantas" to="¿hasta cuántas" />
+    <LinePart from="¿hasta cuantos" to="¿hasta cuántos" />
+    <LinePart from="¿hasta cuando" to="¿hasta cuándo" />
+    <LinePart from="¿hasta donde" to="¿hasta dónde" />
+    <LinePart from="¿hasta que" to="¿hasta qué" />
+    <LinePart from="¿hasta adonde" to="¿hasta adónde" />
+    <LinePart from="¿desde que" to="¿desde qué" />
+    <LinePart from="¿desde cuando" to="¿desde cuándo" />
+    <LinePart from="¿desde quien" to="¿desde quién" />
+    <LinePart from="¿desde donde" to="¿desde dónde" />
+    <LinePart from="¿cuanto" to="¿cuánto" />
+    <LinePart from="¿cuantos" to="¿cuántos" />
+    <LinePart from="¿donde" to="¿dónde" />
+    <LinePart from="¿adonde" to="¿adónde" />
+    <LinePart from="¿con que" to="¿con qué" />
+    <LinePart from="¿con cual" to="¿con cuál" />
+    <LinePart from="¿con quien" to="¿con quién" />
+    <LinePart from="¿con cuantos" to="¿con cuántos" />
+    <LinePart from="¿con cuantas" to="¿con cuántas" />
+    <LinePart from="¿con cuanta" to="¿con cuánta" />
+    <LinePart from="¿con cuanto" to="¿con cuánto" />
+    <LinePart from="¿para donde" to="¿para dónde" />
+    <LinePart from="¿para adonde" to="¿para adónde" />
+    <LinePart from="¿para cuando" to="¿para cuándo" />
+    <LinePart from="¿para que" to="¿para qué" />
+    <LinePart from="¿para quien" to="¿para quién" />
+    <LinePart from="¿para cuanto" to="¿para cuánto" />
+    <LinePart from="¿para cuanta" to="¿para cuánta" />
+    <LinePart from="¿para cuantos" to="¿para cuántos" />
+    <LinePart from="¿para cuantas" to="¿para cuántas" />
+    <LinePart from="¿a donde" to="¿a dónde" />
+    <LinePart from="¿a que" to="¿a qué" />
+    <LinePart from="¿a cual" to="¿a cuál" />
+    <LinePart from="¿a quien" to="¿a quien" />
+    <LinePart from="¿a como" to="¿a cómo" />
+    <LinePart from="¿a cuanto" to="¿a cuánto" />
+    <LinePart from="¿a cuanta" to="¿a cuánta" />
+    <LinePart from="¿a cuantos" to="¿a cuántos" />
+    <LinePart from="¿a cuantas" to="¿a cuántas" />
+    <LinePart from="¿por que" to="¿por qué" />
+    <LinePart from="¿por cual" to="¿por cuál" />
+    <LinePart from="¿por quien" to="¿por quién" />
+    <LinePart from="¿por cuanto" to="¿por cuánto" />
+    <LinePart from="¿por cuanta" to="¿por cuánta" />
+    <LinePart from="¿por cuantos" to="¿por cuántos" />
+    <LinePart from="¿por cuantas" to="¿por cuántas" />
+    <LinePart from="¿por donde" to="¿por dónde" />
+    <LinePart from="¿porque" to="¿por qué" />
+    <LinePart from="¿porqué" to="¿por qué" />
+    <LinePart from="¿y que" to="¿y qué" />
+    <LinePart from="¿y como" to="¿y cómo" />
+    <LinePart from="¿y cuando" to="¿y cuándo" />
+    <LinePart from="¿y cual" to="¿y cuál" />
+    <LinePart from="¿y quien" to="¿y quién" />
+    <LinePart from="¿y cuanto" to="¿y cuánto" />
+    <LinePart from="¿y cuanta" to="¿y cuánta" />
+    <LinePart from="¿y cuantos" to="¿y cuántos" />
+    <LinePart from="¿y cuantas" to="¿y cuántas" />
+    <LinePart from="¿y donde" to="¿y dónde" />
+    <LinePart from="¿y adonde" to="¿y adónde" />
+    <LinePart from="¿quien " to="¿quién " />
+    <LinePart from="¿esta " to="¿está " />
+    <LinePart from="¿estas " to="¿estás " />
+    <!-- Acentos al principio de los signos de interrogación con mayúsculas -->
+    <LinePart from="¿Aun" to="¿Aún" />
+    <LinePart from="¿Que " to="¿Qué " />
+    <LinePart from="¿Sabes que" to="¿Sabes qué" />
+    <LinePart from="¿Sabes adonde" to="¿Sabes adónde" />
+    <LinePart from="¿Sabes cual" to="¿Sabes cuál" />
+    <LinePart from="¿Sabes quien" to="¿Sabes quién" />
+    <LinePart from="¿Sabes como" to="¿Sabes cómo" />
+    <LinePart from="¿Sabes cuan" to="¿Sabes cuán" />
+    <LinePart from="¿Sabes cuanto" to="¿Sabes cuánto" />
+    <LinePart from="¿Sabes cuanta" to="¿Sabes cuánta" />
+    <LinePart from="¿Sabes cuantos" to="¿Sabes cuántos" />
+    <LinePart from="¿Sabes cuantas" to="¿Sabes cuántas" />
+    <LinePart from="¿Sabes cuando" to="¿Sabes cuándo" />
+    <LinePart from="¿Sabes donde" to="¿Sabes dónde" />
+    <LinePart from="¿Sabe que" to="¿Sabe qué" />
+    <LinePart from="¿Sabe adonde" to="¿Sabe adónde" />
+    <LinePart from="¿Sabe cual" to="¿Sabe cuál" />
+    <LinePart from="¿Sabe quien" to="¿Sabe quién" />
+    <LinePart from="¿Sabe como" to="¿Sabe cómo" />
+    <LinePart from="¿Sabe cuan" to="¿Sabe cuán" />
+    <LinePart from="¿Sabe cuanto" to="¿Sabe cuánto" />
+    <LinePart from="¿Sabe cuanta" to="¿Sabe cuánta" />
+    <LinePart from="¿Sabe cuantos" to="¿Sabe cuántos" />
+    <LinePart from="¿Sabe cuantas" to="¿Sabe cuántas" />
+    <LinePart from="¿Sabe cuando" to="¿Sabe cuándo" />
+    <LinePart from="¿Sabe donde" to="¿Sabe dónde" />
+    <LinePart from="¿Saben que" to="¿Saben qué" />
+    <LinePart from="¿Saben adonde" to="¿Saben adónde" />
+    <LinePart from="¿Saben cual" to="¿Saben cuál" />
+    <LinePart from="¿Saben quien" to="¿Saben quién" />
+    <LinePart from="¿Saben como" to="¿Saben cómo" />
+    <LinePart from="¿Saben cuan" to="¿Saben cuán" />
+    <LinePart from="¿Saben cuanto" to="¿Saben cuánto" />
+    <LinePart from="¿Saben cuanta" to="¿Saben cuánta" />
+    <LinePart from="¿Saben cuantos" to="¿Saben cuántos" />
+    <LinePart from="¿Saben cuantas" to="¿Saben cuántas" />
+    <LinePart from="¿Saben cuando" to="¿Saben cuándo" />
+    <LinePart from="¿Saben donde" to="¿Saben dónde" />
+    <LinePart from="¿De que" to="¿De qué" />
+    <LinePart from="¿De donde" to="¿De dónde" />
+    <LinePart from="¿De cual" to="¿De cuál" />
+    <LinePart from="¿De quien" to="¿De quién" />
+    <LinePart from="¿De cuanto" to="¿De cuánto" />
+    <LinePart from="¿De cuanta" to="¿De cuánta" />
+    <LinePart from="¿De cuantos" to="¿De cuántos" />
+    <LinePart from="¿De cuantas" to="¿De cuántas" />
+    <LinePart from="¿De cuando" to="¿De cuándo" />
+    <LinePart from="¿Desde que" to="¿Desde qué" />
+    <LinePart from="¿Desde cuando" to="¿Desde cuándo" />
+    <LinePart from="¿Desde quien" to="¿Desde quién" />
+    <LinePart from="¿Desde donde" to="¿Desde dónde" />
+    <LinePart from="¿Sobre que" to="¿Sobre qué" />
+    <LinePart from="¿Como " to="¿Cómo " />
+    <LinePart from="¿Cual " to="¿Cuál " />
+    <LinePart from="¿En cual" to="¿En cuál" />
+    <LinePart from="¿Cuando" to="¿Cuándo" />
+    <LinePart from="¿Hasta cual" to="¿Hasta cuál" />
+    <LinePart from="¿Hasta quien" to="¿Hasta quién" />
+    <LinePart from="¿Hasta cuanto" to="¿Hasta cuánto" />
+    <LinePart from="¿Hasta cuantas" to="¿Hasta cuántas" />
+    <LinePart from="¿Hasta cuantos" to="¿Hasta cuántos" />
+    <LinePart from="¿Hasta cuando" to="¿Hasta cuándo" />
+    <LinePart from="¿Hasta donde" to="¿Hasta dónde" />
+    <LinePart from="¿Hasta que" to="¿Hasta qué" />
+    <LinePart from="¿Hasta adonde" to="¿Hasta adónde" />
+    <LinePart from="¿Cuanto" to="¿Cuánto" />
+    <LinePart from="¿Cuantos" to="¿Cuántos" />
+    <LinePart from="¿Donde" to="¿Dónde" />
+    <LinePart from="¿Adonde" to="¿Adónde" />
+    <LinePart from="¿Con que" to="¿Con qué" />
+    <LinePart from="¿Con cual" to="¿Con cuál" />
+    <LinePart from="¿Con quien" to="¿Con quién" />
+    <LinePart from="¿Con cuantos" to="¿Con cuántos" />
+    <LinePart from="¿Con cuanta" to="¿Con cuántas" />
+    <LinePart from="¿Con cuanta" to="¿Con cuánta" />
+    <LinePart from="¿Con cuanto" to="¿Con cuánto" />
+    <LinePart from="¿Para donde" to="¿Para dónde" />
+    <LinePart from="¿Para adonde" to="¿Para adónde" />
+    <LinePart from="¿Para cuando" to="¿Para cuándo" />
+    <LinePart from="¿Para que" to="¿Para qué" />
+    <LinePart from="¿Para quien" to="¿Para quién" />
+    <LinePart from="¿Para cuanto" to="¿Para cuánto" />
+    <LinePart from="¿Para cuanta" to="¿Para cuánta" />
+    <LinePart from="¿Para cuantos" to="¿Para cuántos" />
+    <LinePart from="¿Para cuantas" to="¿Para cuántas" />
+    <LinePart from="¿A donde" to="¿A dónde" />
+    <LinePart from="¿A que" to="¿A qué" />
+    <LinePart from="¿A cual" to="¿A cuál" />
+    <LinePart from="¿A quien" to="¿A quien" />
+    <LinePart from="¿A como" to="¿A cómo" />
+    <LinePart from="¿A cuanto" to="¿A cuánto" />
+    <LinePart from="¿A cuanta" to="¿A cuánta" />
+    <LinePart from="¿A cuantos" to="¿A cuántos" />
+    <LinePart from="¿A cuantas" to="¿A cuántas" />
+    <LinePart from="¿Por que" to="¿Por qué" />
+    <LinePart from="¿Por cual" to="¿Por cuál" />
+    <LinePart from="¿Por quien" to="¿Por quién" />
+    <LinePart from="¿Por cuanto" to="¿Por cuánto" />
+    <LinePart from="¿Por cuanta" to="¿Por cuánta" />
+    <LinePart from="¿Por cuantos" to="¿Por cuántos" />
+    <LinePart from="¿Por cuantas" to="¿Por cuántas" />
+    <LinePart from="¿Por donde" to="¿Por dónde" />
+    <LinePart from="¿Porque" to="¿Por qué" />
+    <LinePart from="¿Porqué" to="¿Por qué" />
+    <LinePart from="¿Y que" to="¿Y qué" />
+    <LinePart from="¿Y como" to="¿Y cómo" />
+    <LinePart from="¿Y cuando" to="¿Y cuándo" />
+    <LinePart from="¿Y cual" to="¿Y cuál" />
+    <LinePart from="¿Y quien" to="¿Y quién" />
+    <LinePart from="¿Y cuanto" to="¿Y cuánto" />
+    <LinePart from="¿Y cuanta" to="¿Y cuánta" />
+    <LinePart from="¿Y cuantos" to="¿Y cuántos" />
+    <LinePart from="¿Y cuantas" to="¿Y cuántas" />
+    <LinePart from="¿Y donde" to="¿Y dónde" />
+    <LinePart from="¿Y adonde" to="¿Y adónde" />
+    <LinePart from="¿Quien " to="¿Quién " />
+    <LinePart from="¿Esta " to="¿Está " />
+    <!-- Tilde diacrítica en oraciones interrogativas o exclamativas indirectas -->
+    <LinePart from="el porque" to="el porqué" />
+    <LinePart from="su porque" to="su porqué" />
+    <LinePart from="los porqués" to="los porqués" />
+    <!-- aún -->
+    <LinePart from="aun," to="aún," />
+    <LinePart from="aun no" to="aún no" />
+    <!-- dé -->
+    <LinePart from=" de y " to=" dé y " />
+    <LinePart from=" nos de " to=" nos dé " />
+    <!-- tú -->
+    <LinePart from=" tu ya " to=" tú ya " />
+    <LinePart from="Tu ya " to="Tú ya " />
+    <!-- casos específicos antes de la coma -->
+    <LinePart from=" de, " to=" dé," />
+    <LinePart from=" mi, " to=" mí," />
+    <LinePart from=" tu, " to=" tú," />
+    <LinePart from=" el, " to=" él," />
+    <LinePart from=" te, " to=" té," />
+    <LinePart from=" mas, " to=" más," />
+    <LinePart from=" quien, " to=" quién," />
+    <LinePart from=" cual," to=" cuál," />
+    <LinePart from="porque, " to="porqué," />
+    <LinePart from="cuanto, " to="cuánto," />
+    <LinePart from="cuando, " to="cuándo," />
+    <!-- sé -->
+    <LinePart from=" se," to=" sé," />
+    <LinePart from="se donde" to="sé dónde" />
+    <LinePart from="se cuando" to="sé cuándo" />
+    <LinePart from="se adonde" to="sé adónde" />
+    <LinePart from="se como" to="sé cómo" />
+    <LinePart from="se cual" to="sé cuál" />
+    <LinePart from="se quien" to="sé quién" />
+    <LinePart from="se cuanto" to="sé cuánto" />
+    <LinePart from="se cuanta" to="sé cuánta" />
+    <LinePart from="se cuantos" to="sé cuántos" />
+    <LinePart from="se cuantas" to="sé cuántas" />
+    <LinePart from="se cuan" to="sé cuán" />
+    <!-- si/sí -->
+    <LinePart from=" el si " to=" el sí " />
+    <LinePart from="si mismo" to="sí mismo" />
+    <LinePart from="si misma" to="sí misma" />
+    <!-- Errores de "l" en vez de "i" en casos específicos -->
+    <LinePart from=" llegal" to=" ilegal" />
+    <LinePart from=" lluminar" to=" iluminar" />
+    <LinePart from="sllbato" to="silbato" />
+    <LinePart from="sllenclo" to="silencio" />
+    <LinePart from="clemencla" to="clemencia" />
+    <LinePart from="socledad" to="sociedad" />
+    <LinePart from="tlene" to="tiene" />
+    <LinePart from="tlempo" to="tiempo" />
+    <LinePart from="equlvocaba" to="equivocaba" />
+    <LinePart from="qulnce" to="quince" />
+    <LinePart from="comlen" to="comien" />
+    <LinePart from="historl" to="histori" />
+    <LinePart from="misterl" to="misteri" />
+    <LinePart from="vivencl" to="vivenci" />
+  </PartialLines>
+  <PartialLinesAlways />
+  <BeginLines />
+  <EndLines>
+    <Ending from=".»." to="»." />
+  </EndLines>
+  <WholeLines>
+    <!-- Todas las líneas -->
+    <Line from="No" to="No." />
+  </WholeLines>
+  <RegularExpressions>
+    <!-- Abreviaturas compuestas -->
+    <RegEx find="\b[Ss](r|ra|rta)\b\.?" replaceWith="S$1." />
+    <RegEx find="\b[Dd](r|ra)\b\.?" replaceWith="D$1." />
+    <RegEx find="\b[Uu](d|ds)\b\.?" replaceWith="U$1." />
+    <RegEx find="(\d)(\s){0,1}([Aa])(\.){0,1}([Mm])(\.){0,1}(\W){0,1}" replaceWith="$1 a. m.$7" />
+    <RegEx find="(\d)(\s){0,1}([Pp])(\.){0,1}([Mm])(\.){0,1}(\W){0,1}" replaceWith="$1 p. m.$7" />
+    <RegEx find="(\d)(\s){0,1}(h)(s\b|r\b|rs\b){0,1}(\.){0,1}(\W){0,1}" replaceWith="$1 $3$6" />
+    <RegEx find="(\d)(\s){0,1}([Kk])(m\b|ms\b)(\.){0,1}(\W){0,1}" replaceWith="$1 km$6" />
+    <RegEx find="(\d)(\s){0,1}(s)(g\b|eg\b){0,1}(\.){0,1}(\W){0,1}" replaceWith="$1 s$6" />
+    <RegEx find="(\d)(\s){0,1}([Kk])(g\b|gs\b)(\.){0,1}(\W){0,1}" replaceWith="$1 kg$6" />
+    <RegEx find="(\d)(\s){0,1}(m)(t\b|ts\b){0,1}(\.){0,1}(\W){0,1}" replaceWith="$1 m$6" />
+    <RegEx find="(\d)KBs(\W){0,1}" replaceWith="$1 kB$2" />
+    <RegEx find="([Nn])°(\s){0,1}(\d)" replaceWith="$1.° $3" />
+    <RegEx find="([Nn])ro(\.){0,1}(\s){0,1}(\d)" replaceWith="$1.° $4" />
+    <!-- Signos invertidos -->
+    <RegEx find="\?¿(\W|\w)" replaceWith="? ¿$1" />
+    <RegEx find="\!¡(\W|\w)" replaceWith="! ¡$1" />
+    <RegEx find="\?¿¿(\W|\w)" replaceWith="? ¿$1" />
+    <RegEx find="\!¡¡(\W|\w)" replaceWith="! ¡$1" />
+    <!-- Inicio de línea -->
+    <RegEx find="^_(\s)" replaceWith="-$1" />
+    <RegEx find="^_(\w)" replaceWith="- $1" />
+    <!-- Uso de comillas según la recomendación de la RAE y la Wikipedia -->
+    <RegEx find="(«[^“«»]+)«" replaceWith="$1“" />
+    <RegEx find="(“[^«»”]+)»" replaceWith="$1”" />
+    <RegEx find="`" replaceWith="‘" />
+    <RegEx find="´" replaceWith="’" />
+    <RegEx find="([\wá-ú])(\.)(«|»)" replaceWith="$1»." />
+    <RegEx find="«(\?)" replaceWith="»?" />
+    <RegEx find="«(\!)" replaceWith="»!" />
+    <RegEx find="«\s" replaceWith="» " />
+    <RegEx find="«(\))" replaceWith="»)" />
+    <RegEx find="(\?)«" replaceWith="?»" />
+    <RegEx find="(\!)«" replaceWith="!»" />
+    <RegEx find="«(,)" replaceWith="»," />
+    <RegEx find="«(;)" replaceWith="»;" />
+    <RegEx find="«(:)" replaceWith="»:" />
+    <RegEx find="(¿)»" replaceWith="¿«" />
+    <RegEx find="(¡)»" replaceWith="¡«" />
+    <!-- Uso de comillas (ANSI) según la recomendación de la RAE («\x22» es el carácter «"») -->
+    <RegEx find="([\wá-ú])([\.,]) ?[\x22»]" replaceWith="$1»$2" />
+    <RegEx find="([\wá-ú])\?[\x22»](\s|$)" replaceWith="$1?».$2" />
+    <RegEx find="^(\.\.\.)(\s){0,1}\x22" replaceWith="$1«" />
+    <RegEx find="«\x22" replaceWith="«" />
+    <RegEx find="\x22»" replaceWith="»" />
+    <RegEx find="^\x22{2,}" replaceWith="«" />
+    <RegEx find="\x22{2,}$" replaceWith="»" />
+    <RegEx find="\x22\r" replaceWith="»" />
+    <RegEx find="^\x22" replaceWith="«" />
+    <RegEx find="\x22$" replaceWith="»." />
+    <RegEx find="([\wá-ú])\.[\x22»]" replaceWith="$1»." />
+    <RegEx find="\s\x22" replaceWith=" «" />
+    <RegEx find="\x22\s" replaceWith="» " />
+    <RegEx find="\x22(,)" replaceWith="»," />
+    <RegEx find="\x22(\.)" replaceWith="»." />
+    <RegEx find="\x22(;)" replaceWith="»;" />
+    <RegEx find="\x22(:)" replaceWith="»:" />
+    <RegEx find="(\!)\x22" replaceWith="!»" />
+    <RegEx find="\x22(\!)" replaceWith="»!" />
+    <RegEx find="(\?)\x22" replaceWith="?»" />
+    <RegEx find="\x22(\?)" replaceWith="»?" />
+    <RegEx find="\x22(¿)" replaceWith="«¿" />
+    <RegEx find="(¿)\x22" replaceWith="¿«" />
+    <RegEx find="\x22(¡)" replaceWith="«¡" />
+    <RegEx find="(¡)\x22" replaceWith="¡«" />
+    <RegEx find="\x22(\))" replaceWith="»)" />
+    <RegEx find="(\))\x22" replaceWith=")»" />
+    <RegEx find="(\()\x22" replaceWith="(«" />
+    <!-- Uso de comillas (Unicode) según la recomendación de la RAE («\u0022» es el carácter «"») -->
+    <RegEx find="^(\.\.\.)(\s){0,1}\u0022" replaceWith="$1«" />
+    <RegEx find="^\u0022{2,}" replaceWith="«" />
+    <RegEx find="\u0022{2,}$" replaceWith="»" />
+    <RegEx find="\u0022\r" replaceWith="»" />
+    <RegEx find="^\u0022" replaceWith="«" />
+    <RegEx find="\u0022$" replaceWith="»" />
+    <RegEx find="(\w)(\.)\u0022" replaceWith="$1»." />
+    <RegEx find="\s\u0022" replaceWith=" «" />
+    <RegEx find="\u0022\s" replaceWith="» " />
+    <RegEx find="\u0022(,)" replaceWith="»," />
+    <RegEx find="\u0022(\.)" replaceWith="»." />
+    <RegEx find="\u0022(;)" replaceWith="»;" />
+    <RegEx find="\u0022(:)" replaceWith="»:" />
+    <RegEx find="(\!)\u0022" replaceWith="!»" />
+    <RegEx find="\u0022(\!)" replaceWith="»!" />
+    <RegEx find="(\?)\u0022" replaceWith="?»" />
+    <RegEx find="\u0022(\?)" replaceWith="»?" />
+    <RegEx find="\u0022(¿)" replaceWith="«¿" />
+    <RegEx find="(¿)\u0022" replaceWith="¿«" />
+    <RegEx find="\u0022(¡)" replaceWith="«¡" />
+    <RegEx find="(¡)\u0022" replaceWith="¡«" />
+    <RegEx find="\u0022(\))" replaceWith="»)" />
+    <RegEx find="(\))\u0022" replaceWith=")»" />
+    <RegEx find="(\()\u0022" replaceWith="(«" />
+    <!-- Numeración -->
+    <RegEx find="([0-9])\.([0-9])\b" replaceWith="$1,$2" />
+    <RegEx find="(^|\s|[¡¿«])([0-9])(,|\.)?([0-9]{3})\b" replaceWith="$1$2$4" />
+    <RegEx find="(\d)\s(?=\d{2}\b)" replaceWith="$1-" />
+    <!-- "1 :", "2 :"... "n :" a "n:" -->
+    <RegEx find="(\d) ([:;])" replaceWith="$1$2" />
+    <!-- Corregir las comas y puntos por ej. «, ,» por «,» & «,,,» o similar por «...» -->
+    <RegEx find="(\.\.\.+)$" replaceWith="..." />
+    <RegEx find="(, ,+)$" replaceWith="," />
+    <RegEx find="(,\s),+\s" replaceWith="$1" />
+    <RegEx find="(\.\.\.),$" replaceWith="$1" />
+    <RegEx find="([\wá-ú])(\.\.)$" replaceWith="$1." />
+    <!-- Puntos innecesarios (complemento) -->
+    <RegEx find="([\w\W]\.{3})([¡¿])" replaceWith="$1 $2" />
+    <RegEx find="(\w)\.\.(\s)" replaceWith="$1.$2" />
+    <RegEx find="([\wá-ú\x22»])\.([\?\!])" replaceWith="$1$2" />
+    <RegEx find="([\:\;])\." replaceWith="$1" />
+    <RegEx find="\.([\:\;])" replaceWith="$1" />
+    <RegEx find="\:+" replaceWith=":" />
+    <!-- Terminaciones ción/sión -->
+    <RegEx find="([sc]i)o(n)\b" replaceWith="$1ó$2" />
+    <RegEx find="([SC]I)O(N)\b" replaceWith="$1Ó$2" />
+    <!-- "i" en vez de "l" en terminaciones «clón» -->
+    <RegEx find="clón\b" replaceWith="ción" />
+    <!-- "si" en vez de "sl" -->
+    <RegEx find="\b([Ss])(l)\b" replaceWith="$1i" />
+    <!-- Para corregir por ej. raclones, perforaclones, opclones, etc -->
+    <RegEx find="([Rr]ac)l(o)" replaceWith="$1i$2" />
+    <RegEx find="([Oo]pc)l(o)" replaceWith="$1i$2" />
+    <!-- Para corregir por ej. tenldo, víctlmas, olvldarlo, legítlmo, etc -->
+    <RegEx find="([BbCcDdFfHhMmNnRrSsTtVv])l([bcdhmnrstv])" replaceWith="$1i$2" />
+    <!-- Corrige los errores en el ripeo de la «o» mayúscula por el cero «0» y viceversa -->
+    <RegEx find="(\d)O" replaceWith="$1 0" />
+    <RegEx find="(\d)[,\.]O" replaceWith="$1.0" />
+    <RegEx find="([A-Z])0" replaceWith="$1O" />
+    <RegEx find="\b0([A-Za-z])" replaceWith="O$1" />
+    <!-- Signos musicales -->
+    <RegEx find="[♪♫☺☹♥©☮☯Σ∞≡⇒π#](\r\n)[♪♫☺☹♥©☮☯Σ∞≡⇒π#]" replaceWith="$1" />
+    <!-- Tilde diacrítica antes del punto -->
+    <RegEx find="(\s)([dst])e\.(\s|\$)" replaceWith="$1$2é.$3" />
+    <RegEx find="(\s)mi\.(\s|\$)" replaceWith="$1mí.$2" />
+    <RegEx find="(\s)el\.(\s|\$)" replaceWith="$1él.$2" />
+    <RegEx find="(\s)tu\.(\s|\$)" replaceWith="$1tú.$2" />
+    <RegEx find="(\s)si\.(\s|\$)" replaceWith="$1sí.$2" />
+    <RegEx find="(\s)aun\.(\s|\$)" replaceWith="$1aún.$2" />
+    <RegEx find="(\s)mas\.(\s|\$)" replaceWith="$1más.$2" />
+    <RegEx find="(\s)quien\.(\s|\$)" replaceWith="$1quién.$2" />
+    <RegEx find="(\s)cual\.(\s|\$)" replaceWith="$1cuál.$2" />
+    <RegEx find="(\s)que\.(\s|\$)" replaceWith="$1qué.$2" />
+    <RegEx find="(\s)porque\.(\s|\$)" replaceWith="$1porqué.$2" />
+    <RegEx find="(\s)cuanto\.(\s|\$)" replaceWith="$1cuánto.$2" />
+    <RegEx find="(\s)cuando\.(\s|\$)" replaceWith="$1cuándo.$2" />
+    <!-- Prefijos; palabras compuestas (simple) -->
+    <RegEx find="(\b[Ee]x|\b[Ss]uper|\b[Aa]nti|\b[Pp]os|\b[Pp]re|\b[Pp]ro|\b[Vv]ice)[\s\x2D]([a-zá-ú]{3,20})(\b)" replaceWith="$1$2" />
+    <!-- Prefijos; palabras compuestas (números) -->
+    <RegEx find="(\b[Ss]ub|\b[Ss]uper)[\s\x2D](\d{2})(\b)" replaceWith="$1-$2$3" />
+    <!-- Prefijos; palabras compuestas (mayúsculas) -->
+    <RegEx find="(\b[Aa]nti|\b[Mm]ini|\b[Pp]os|\b[Pp]ro)\s([A-Z]{1,10})([A-Z][a-zá-ú]){0,10}(\b)" replaceWith="$1-$2$3" />
+    <!-- Casos de mayúsculas con dos puntos -->
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(a)" replaceWith="$1A" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(b)" replaceWith="$1B" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(c)" replaceWith="$1C" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(d)" replaceWith="$1D" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(e)" replaceWith="$1E" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(f)" replaceWith="$1F" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(g)" replaceWith="$1G" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(h)" replaceWith="$1H" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(i)" replaceWith="$1I" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(j)" replaceWith="$1J" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(k)" replaceWith="$1K" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(l)" replaceWith="$1L" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(m)" replaceWith="$1M" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(n)" replaceWith="$1N" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(ñ)" replaceWith="$1Ñ" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(o)" replaceWith="$1O" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(p)" replaceWith="$1P" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(q)" replaceWith="$1Q" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(r)" replaceWith="$1R" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(s)" replaceWith="$1S" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(t)" replaceWith="$1T" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(u)" replaceWith="$1U" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(v)" replaceWith="$1V" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(w)" replaceWith="$1W" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(x)" replaceWith="$1X" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(y)" replaceWith="$1Y" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(z)" replaceWith="$1Z" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(á)" replaceWith="$1Á" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(é)" replaceWith="$1É" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(í)" replaceWith="$1Í" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(ó)" replaceWith="$1Ó" />
+    <RegEx find="([\wá-ú]:\s[«\x22]?)(ú)" replaceWith="$1Ú" />
+    <!-- Usos correctos de la coma -->
+    <RegEx find="(\b[Pp]ero),(\s)([¡¿])" replaceWith="$1$2$3" />
+    <RegEx find="(\b[Aa]unque),(\s|$)" replaceWith="$1$2" />
+    <!-- Vocativos -->
+    <RegEx find="(\bHola|\bBueno|\bBien|\bVen|\bVen acá|\besto|\bBuenos días|\bFeliz cumpleaños|\bsiento)\s([A-Z][a-zá-ú]{3,12}\b|seño(r|ra|rita)\b|hij(o|a) mío\b|amig(o|a)\b)" replaceWith="$1, $2" />
+    <!-- «aún» cuando son sinónimos de «incluso» o «hasta» -->
+    <RegEx find="(\W|^)(\b[Aa])ú(n)(\s)(así\b|cuando\b|los\b|las\b|negar(te|se)\b)" replaceWith="$1$2u$3$4$5" />
+    <RegEx find="(\b[Nn]i)(\s)(a)ú(n)(\W|$)" replaceWith="$1$2$3u$4$5" />
+    <!-- «sí» -->
+    <RegEx find="\b([Ss])i(:|;|\.)" replaceWith="$1í$2" />
+    <!-- «sé» -->
+    <RegEx find="(\b[Ll]o|\b[Ll]a|\b[Ll]e)(\s)se(\W|$)" replaceWith="$1$2sé$3" />
+    <RegEx find="[Ss]e\s(dónde\b|cuándo\b|adónde\b|cómo\b|cuál\b|quién\b|cuánto\b|cuánta\b|cuántos\b|cuántas\b|cuán\b)" replaceWith="sé $1" />
+    <!-- «té» -->
+    <RegEx find="\b([Tt])e\s(verde\b|negro\b|perla\b|de manzanilla\b|de lim[óo]n\b|de jazm[íi]n\b)" replaceWith="$1é $2" />
+    <!-- Apóstrofo -->
+    <RegEx find="(\b[A-Z][a-zá-ú]{3,12})\s(’|')(\d\d(\s|$))" replaceWith="$1 $3" />
+    <RegEx find="(\b[A-Z]{2,5})(’|')(s)" replaceWith="(Ej. Devedés)$1$3" />
+    <RegEx find="(\b\d{1,2})(’|')(\d{2})\s(s|m)(\W|$)" replaceWith="$1,$3 $4$5" />
+    <RegEx find="(\b\d{1,2})(’|')(\d{2})\s(h)(\W|$)" replaceWith="$1:$3 $4$5" />
+    <!-- Porcentaje (debe llevar espacio) -->
+    <RegEx find="(\b\d{1,3})%(\W)" replaceWith="$1 %$2" />
+    <!-- Haz/has -->
+    <RegEx find="(\b)([Hh])as\s(la\b|lo\b|clic\b)(\W)" replaceWith="$1$2az $3$4" />
+    <RegEx find="(\b)([Hh])az\s(de\b)(\W)" replaceWith="$1$2as $3$4" />
+    <RegEx find="(\b)([Hh])as(le\b|nos\b|me\b)(\W)" replaceWith="$1$2az$3$4" />
+    <!-- Quitar itálicas en 3 o menos letras -->
+    <RegEx find="\x3ci\x3e(.{1,3})\x3c\/i\x3e" replaceWith="$1" />
+    <!-- Miscelánea -->
+    <RegEx find="(\b[Cc]erca|\b[Ee]ncima|\b[Dd]ebajo|\b[Dd]etrás|\b[Dd]elante)(\s)mío" replaceWith="$1 de mí" />
+    <RegEx find="(\b[Cc]erca|\b[Ee]ncima|\b[Dd]ebajo|\b[Dd]etrás|\b[Dd]elante)(\s)tuyo" replaceWith="$1 de ti" />
+    <!-- Punto antes de «¿» y «¡» -->
+    <RegEx find="([\wá-ú»])\s(?=(¿|¡)[A-ZÁ-Ú])" replaceWith="$1. " />
+    <!-- Espacios después del guión -->
+    <RegEx find="(^|\n)(-)([^\s])" replaceWith="$1$2 $3" />
+    <!-- Punto antes del guión -->
+    <RegEx find="([^\.\?\!]) - " replaceWith="$1. - " />
+    <!-- Terminaciones en «ólogo», «ílogo» y «álogo» -->
+    <RegEx find="\Bo(log[ao]s?\b)" replaceWith="ó$1" />
+    <RegEx find="\Ba(log[ao]s?\b)" replaceWith="á$1" />
+    <RegEx find="\Bi(log[ao]s?\b)" replaceWith="í$1" />
+  </RegularExpressions>
+</OCRFixReplaceList>

Contents/Libraries/Shared/subzero/modification/dictionaries/xml/srp_OCRFixReplaceList.xml

@@ -0,0 +1,234 @@
+<!-- Credit goes to: MilanRS [http://www.prijevodi-online.org] -->
+<OCRFixReplaceList>
+  <WholeWords>
+    <Word from="ču" to="ću" />
+    <Word from="češ" to="ćeš" />
+    <Word from="če" to="će" />
+    <Word from="ćš" to="ćeš" />
+    <Word from="ćmo" to="ćemo" />
+    <Word from="ćte" to="ćete" />
+    <Word from="čemo" to="ćemo" />
+    <Word from="čete" to="čete" />
+    <Word from="djete" to="dijete" />
+    <Word from="Hey" to="Hej" />
+    <Word from="hey" to="hej" />
+    <Word from="htjeo" to="htio" />
+    <Word from="Hočeš" to="Hoćeš" />
+    <Word from="hočeš" to="hoćeš" />
+    <Word from="iči" to="ići" />
+    <Word from="jel" to="je l'" />
+    <Word from="Jel" to="Je l'" />
+    <Word from="nedaj" to="ne daj" />
+    <Word from="Rješit" to="Riješit" />
+    <Word from="smjeo" to="smio" />
+    <Word from="uopče" to="uopće" />
+    <Word from="valda" to="valjda" />
+    <Word from="želila" to="željela" />
+  </WholeWords>
+  <PartialWordsAlways />
+  <PartialWords>
+    <WordPart from="¤" to="o" />
+    <WordPart from="vv" to="w" />
+    <WordPart from="IVI" to="M" />
+    <WordPart from="lVI" to="M" />
+    <WordPart from="IVl" to="M" />
+    <WordPart from="lVl" to="M" />
+  </PartialWords>
+  <PartialLines>
+    <LinePart from="bi smo" to="bismo" />
+    <LinePart from="dali je" to="da li je" />
+    <LinePart from="dali si" to="da li si" />
+    <LinePart from="Dali si" to="Da li si" />
+    <LinePart from="Jel sam ti" to="Jesam li ti" />
+    <LinePart from="Jel si" to="Jesi li" />
+    <LinePart from="Jel' si" to="Jesi li" />
+    <LinePart from="Je I'" to="Jesi li" />
+    <LinePart from="Jel si to" to="Jesi li to" />
+    <LinePart from="Jel' si to" to="Da li si to" />
+    <LinePart from="jel si to" to="da li si to" />
+    <LinePart from="jel' si to" to="jesi li to" />
+    <LinePart from="Jel si ti" to="Da li si ti" />
+    <LinePart from="Jel' si ti" to="Da li si ti" />
+    <LinePart from="jel si ti" to="da li si ti" />
+    <LinePart from="jel' si ti" to="da li si ti" />
+    <LinePart from="jel ste " to="jeste li " />
+    <LinePart from="Jel ste" to="Jeste li" />
+    <LinePart from="jel' ste " to="jeste li " />
+    <LinePart from="Jel' ste " to="Jeste li " />
+    <LinePart from="Jel su " to="Jesu li " />
+    <LinePart from="Jel da " to="Zar ne" />
+    <LinePart from="jel da " to="zar ne" />
+    <LinePart from="jel'da " to="zar ne" />
+    <LinePart from="Jeli sve " to="Je li sve" />
+    <LinePart from="Jeli on " to="Je li on" />
+    <LinePart from="Jeli ti " to="Je li ti" />
+    <LinePart from="jeli ti " to="je li ti" />
+    <LinePart from="Jeli to " to="Je li to" />
+    <LinePart from="Nebrini" to="Ne brini" />
+    <LinePart from="nedaj" to="ne daj" />
+    <LinePart from="ne ću" to="neću" />
+    <LinePart from="Nemogu" to="Ne mogu" />
+    <LinePart from="ne mogu" to="ne mogu" />
+    <LinePart from="Nemoraš" to="Ne moraš" />
+    <LinePart from="od kako" to="otkako" />
+    <LinePart from="Si dobro" to="Jesi li dobro" />
+    <LinePart from="Svo vreme" to="Sve vrijeme" />
+    <LinePart from="Svo vrijeme" to="Sve vrijeme" />
+    <LinePart from="Cijelo vrijeme" to="Sve vrijeme" />
+  </PartialLines>
+  <PartialLinesAlways />
+  <BeginLines />
+  <EndLines />
+  <WholeLines />
+  <RegularExpressions>
+    <RegEx find="đž" replaceWith="dž" />
+    <RegEx find="ajsmiješnij" replaceWith="ajsmješnij" />
+    <RegEx find="boži[čć]([aeiu]|em|ima)?\b" replaceWith="Božić$1" />
+    <RegEx find=" g-dine\.$" replaceWith=" gospodine." />
+    <RegEx find=" g-dine +(?=[A-ZČĐŠŽ])" replaceWith=" g. " />
+    <RegEx find="([gG])dine? +(?=[A-ZČĐŠŽ])" replaceWith="$1. " />
+    <RegEx find="([gG])-đo +(?=[A-ZČĐŠŽ])" replaceWith="$1gđo " />
+    <RegEx find="gdina +(?=[A-ZČĐŠŽ])" replaceWith="g. " />
+    <RegEx find=" gosp +" replaceWith=" g. " />
+    <RegEx find="Jel si sigur" replaceWith="Jesi li sigur" />
+    <RegEx find="Jel' si sigur" replaceWith="Jesi li sigur" />
+    <RegEx find="\b([jJ])el\?" replaceWith="$1e l'?" />
+    <RegEx find="\bJel'" replaceWith="Je l'" />
+    <RegEx find="([kK]alib(?:ar|r[aeui]))\. *([0-9])" replaceWith="$1 .$2" />
+    <RegEx find="([mM])jenjati" replaceWith="$1ijenjati" />
+    <RegEx find="([mM])oguč" replaceWith="$1oguć" />
+    <RegEx find="\b([nN])ebih?" replaceWith="$1e bi" />
+    <RegEx find="\b([nN])eč([ue]š?|emo|ete)\b" replaceWith="$1eć$2" />
+    <RegEx find="\b([nN])emože(mo|š|te)?\b" replaceWith="$1e može$2" />
+    <RegEx find="\b([nN])ezna([šm]o?|t[ei]|ju|jući|vši)?\b" replaceWith="$1e zna$2" />
+    <RegEx find="najcijenjen" replaceWith="najcjenjen" />
+    <RegEx find="N[jJ]u Jork" replaceWith="Njujork" />
+    <RegEx find="([oO])d([kp])" replaceWith="$1t$2" />
+    <RegEx find="([oO])ružij([aeu])" replaceWith="$1ružj$2" />
+    <RegEx find="([oO])sječa" replaceWith="$1sjeća" />
+    <RegEx find="([pPdD])onje([lt])" replaceWith="$1onije$2" />
+    <RegEx find="([pP])objedi([mšto])" replaceWith="$1obijedi$2" />
+    <RegEx find="redamnom" replaceWith="reda mnom" />
+    <RegEx find="redpostav" replaceWith="retpostav" />
+    <RegEx find="([pP])rimjeti" replaceWith="$1rimijeti" />
+    <RegEx find="([pP])romjeni([mštol])" replaceWith="$1romijeni$2" />
+    <RegEx find="([rR])azumijeć" replaceWith="$1azumjeć" />
+    <RegEx find="rascjepljen" replaceWith="rascijepljen" />
+    <RegEx find="redhodn" replaceWith="rethodn" />
+    <RegEx find="rimjenjen" replaceWith="rimijenjen" />
+    <RegEx find="([^d])rješit" replaceWith="$1riješit" />
+    <RegEx find="([sSzZ])amnom" replaceWith="$1a mnom" />
+    <RegEx find="([sS])lijede[čć]([aeiu]|e[mg])" replaceWith="$1ljedeć$2" />
+    <RegEx find="([sS])mješno" replaceWith="$1miješno" />
+    <RegEx find="([uU])mijesto" replaceWith="$1mjesto" />
+    <RegEx find="([uU])spijeh" replaceWith="$1spjeh" />
+    <RegEx find="([uU])spiješ(an|n[aeiou]|no[mgj])" replaceWith="$1spješ$2" />
+    <RegEx find="([uU])vjek" replaceWith="$1vijek" />
+    <RegEx find="\b([vV])eč([aeiou])" replaceWith="$1eć$2" />
+    <RegEx find="([zZ])ahtijeva" replaceWith="$1ahtjeva" />
+    <RegEx find="([zZ])ahtjeva([ojlmšt])" replaceWith="$1ahtijeva$2" />
+    <RegEx find="([ks]ao)\.:" replaceWith="$1:" />
+    <RegEx find="(?&lt;=[a-zčđšž])Ij(?=[a-zčđšž])" replaceWith="lj" />
+    <RegEx find="(?&lt;=[^A-ZČĐŠŽa-zčđšž])Iju(?=bav|d|t)" replaceWith="lju" />
+    <!-- kad ima razmak između tagova </i> <i> -->
+    <!-- <RegEx find="(&gt;) +(&lt;)" replaceWith="$1$2" /> -->
+    <!-- ',"' to '",' -->
+    <RegEx find="(?&lt;=\w),&quot;(?=\s|$)" replaceWith="&quot;," />
+    <RegEx find=",\.{3}|\.{3},|\.{2} \." replaceWith="..." />
+    <!-- "1 :", "2 :"... "n :" to "n:" -->
+    <RegEx find="([0-9]) +: +(\D)" replaceWith="$1: $2" />
+    <!-- Two or more consecutive "," to "..." -->
+    <RegEx find=",{2,}" replaceWith="..." />
+    <!-- Two or more consecutive "-" to "..." -->
+    <RegEx find="-{2,}" replaceWith="..." />
+    <RegEx find="([^().])\.{2}([^().:])" replaceWith="$1...$2" />
+    <!-- separator stotica i decimalnog ostatka 1,499,000.00 -> 1.499.000,00 -->
+    <RegEx find="([0-9]{3})\.([0-9]{2}[^0-9])" replaceWith="$1,$2" />
+    <RegEx find="([0-9]),([0-9]{3}\D)" replaceWith="$1.$2" />
+    <!-- Apostrophes -->
+    <RegEx find="´´" replaceWith="&quot;" />
+    <!-- <RegEx find="[´`]" replaceWith="'" /> -->
+    <!-- <RegEx find="[“”]" replaceWith="&quot;" /> -->
+    <RegEx find="''" replaceWith="&quot;" />
+    <!-- Two or more consecutive '"' to one '"' -->
+    <RegEx find="&quot;{2,}" replaceWith="&quot;" />
+    <!-- Fix zero and capital 'o' ripping mistakes -->
+    <RegEx find="(?&lt;=[0-9]\.?)O" replaceWith="0" />
+    <RegEx find="\b0(?=[A-ZČĐŠŽa-zčđšž])" replaceWith="O" />
+    <!-- Brisanje crte - na početku 1. reda (i kada ima dva reda) -->
+    <RegEx find="\A- ?([A-ZČĐŠŽa-zčđšž0-9„'&quot;]|\.{3})" replaceWith="$1" />
+    <RegEx find="\A(&lt;[ibu]&gt;)- ?" replaceWith="$1" />
+    <RegEx find="  - " replaceWith=" -" />
+    <!-- Brisanje razmaka iza crte - na početku 2. reda -->
+    <RegEx find="(?&lt;=\n(&lt;[ibu]&gt;)?)- (?=[A-ZČĐŠŽčš0-9„'&quot;&lt;])" replaceWith="-" />
+    <!-- Korigovanje crte - kad je u sredini prvog reda -->
+    <RegEx find="([.!?&quot;&gt;]) - ([A-ZČĐŠŽčš'&quot;&lt;])" replaceWith="$1 -$2" />
+    <!-- Zatvoren tag pa razmak poslije crtice -->
+    <RegEx find="(&gt;) - ([A-ZČĐŠŽčš„'&quot;])" replaceWith="$1 -$2" />
+    <!-- Zatvoren tag pa crtica razmak -->
+    <RegEx find="(&gt;)- ([A-ZČĐŠŽčš„'&quot;])" replaceWith="$1-$2" />
+    <!-- Zagrada pa crtica razmak -->
+    <RegEx find="\(- ([A-ZČĐŠŽčš„'&quot;])" replaceWith="(-$1" />
+    <!-- Smart space after dot -->
+    <!-- osim kad je zadnje t (riječ kolt) -->
+    <RegEx find="(?&lt;=[a-su-zá-úñä-ü])\.(?=[^\s\n().:?!*^“”'&quot;&lt;])" replaceWith=". " />
+    <!-- Oznaka za kalibar. Npr. "Colt .45" -->
+    <!-- Da bi radilo, da bi ovaj razmak bio dozvoljen, odčekirajte "Razmaci ispred tačke" -->
+    <RegEx find="t\.(?=[0-9]{2})" replaceWith="t ." />
+    <!-- Joey(j)a -->
+    <RegEx find="(?&lt;=\b[A-Z][a-z])eyj(?=[a-z])" replaceWith="ey" />
+    <!-- Sređuje zarez sa razmakom -->
+    <RegEx find="(?&lt;=[A-ZČĐŠŽa-zčđšžá-úñä-ü&quot;]),(?=[^\s(),?!“&lt;])" replaceWith=", " />
+    <RegEx find=" +,(?=[A-ZČĐŠŽa-zčđšž])" replaceWith=", " />
+    <RegEx find=" +, +" replaceWith=", " />
+    <RegEx find=" +,$" replaceWith="," />
+    <RegEx find="([?!])-" replaceWith="$1 -" />
+    <!-- Space after last of some consecutive dots (eg. "...") -->
+    <RegEx find="(?&lt;=[a-zčđšž])(\.{3}|!)(?=[a-zčđšž])" replaceWith="$1 " />
+    <!-- Delete space after "..." that is at the beginning of the line. You may delete this line if you don't like it -->
+    <!-- <RegEx find="^\.{3} +" replaceWith="..." /> -->
+    <!-- "tekst ... tekst" mijenja u "tekst... tekst" -->
+    <RegEx find="(?&lt;=[A-ZČĐŠŽa-zčđšž]) +\.{3} +" replaceWith="... " />
+    <RegEx find="(?&lt;=\S)\. +&quot;" replaceWith=".&quot;" />
+    <RegEx find="&quot; +\." replaceWith="&quot;." />
+    <RegEx find="(?&lt;=\S\.{3}) +&quot;(?=\s|$)" replaceWith="&quot;" />
+    <RegEx find=" +\.{3}$" replaceWith="..." />
+    <RegEx find="(?&lt;=[a-zčđšž])(?: +\.{3}|\.{2}$)" replaceWith="..." />
+    <!-- Razmak ispred zagrade -->
+    <RegEx find="(?&lt;=[A-ZČĐŠŽa-zčđšž])\(" replaceWith=" (" />
+    <!-- Razmak iza upitnika -->
+    <RegEx find="\?(?=[A-ZČĐŠŽčš])" replaceWith="? " />
+    <RegEx find="(?&lt;=^|&gt;)\.{3} +(?=[A-ZČĐŠŽčš])" replaceWith="..." />
+    <!-- Brise ... kad je na poč. reda "... -->
+    <RegEx find="^&quot;\.{3} +" replaceWith="&quot;" />
+    <RegEx find="(?&lt;=[0-9])\$" replaceWith=" $$" />
+    <!-- ti š -> t š by Strider -->
+    <!-- Zamijeni sva "**ti šu*" s "**t šu*" i "**ti še*" s "**t še*" -->
+    <!-- <RegEx find="([a-z])ti (š+[eu])" replaceWith="$1t $2" /> -->
+    <!-- <RegEx find="([A-Za-z])ti( |\r?\n)(š[eu])" replaceWith="$1t$2$3" /> -->
+    <!-- <RegEx find="(?i)\b(ni)t (š[eu])" replaceWith="$1ti $2" /> -->
+    <!-- <RegEx find="\. +Mr. " replaceWith=". G. " /> -->
+    <!-- <RegEx find="\. +Mrs. " replaceWith=". Gđa " /> -->
+    <!-- <RegEx find="\. +Miss " replaceWith=". Gđica " /> -->
+    <!-- <RegEx find=", +Mrs. " replaceWith=", gđo " /> -->
+    <!-- <RegEx find=", +Miss " replaceWith=", gđice " /> -->
+    <!-- Razmak poslije <i> i poslije .. -->
+    <RegEx find="^(&lt;[ibu]&gt;) +" replaceWith="$1" />
+    <RegEx find="^\.{2} +" replaceWith="..." />
+    <!-- Razmak ? "</i> -->
+    <RegEx find="([.?!]) +(&quot;&lt;)" replaceWith="$1$2" />
+    <!-- Bez razmaka kod Npr.: -->
+    <RegEx find="(?&lt;=[Nn]pr\.) *: *" replaceWith=": " />
+    <RegEx find="\. ," replaceWith=".," />
+    <RegEx find="([?!])\." replaceWith="$1" />
+    <!-- Da ne kvari potpise sa ..:: -->
+    <RegEx find="\.{3}::" replaceWith="..::" />
+    <RegEx find="::\.{3}" replaceWith="::.." />
+    <RegEx find="\.{2} +::" replaceWith="..::" />
+    <!-- Skracenice bez razmaka -->
+    <RegEx find="d\. o\.o\." replaceWith="d.o.o." />
+    <!-- Kad red počinje sa ...pa malo slovo -->
+    <!-- <RegEx find="^\.{3}([a-zčđšž&quot;&lt;])" replaceWith="$1" /> -->
+    <!-- <RegEx find=" +([.?!])" replaceWith="$1" /> -->
+  </RegularExpressions>
+</OCRFixReplaceList>

Contents/Libraries/Shared/subzero/modification/dictionaries/xml/swe_OCRFixReplaceList.xml

@@ -0,0 +1,405 @@
+<OCRFixReplaceList>
+  <WholeWords>
+    <Word from="lârt" to="lärt" />
+    <Word from="hedervårda" to="hedervärda" />
+    <Word from="stormâstare" to="stormästare" />
+    <Word from="Avfârd" to="Avfärd" />
+    <Word from="tâlten" to="tälten" />
+    <Word from="ârjag" to="är jag" />
+    <Word from="ärjag" to="är jag" />
+    <Word from="jâmlikar" to="jämlikar" />
+    <Word from="Riskakofl" to="Riskakor" />
+    <Word from="Karamellen/" to="Karamellen" />
+    <Word from="Lngenüng" to="Ingenting" />
+    <Word from="ärju" to="är ju" />
+    <Word from="Sá" to="Så" />
+    <Word from="närjag" to="när jag" />
+    <Word from="alltjag" to="allt jag" />
+    <Word from="görjag" to="gör jag" />
+    <Word from="trorjag" to="tror jag" />
+    <Word from="varju" to="var ju" />
+    <Word from="görju" to="gör ju" />
+    <Word from="kanju" to="kan ju" />
+    <Word from="blirjag" to="blir jag" />
+    <Word from="sägerjag" to="säger jag" />
+    <Word from="behållerjag" to="behåller jag" />
+    <Word from="prøblem" to="problem" />
+    <Word from="räddadeju" to="räddade ju" />
+    <Word from="honøm" to="honom" />
+    <Word from="Ln" to="In" />
+    <Word from="svårflörtad" to="svårflörtad" />
+    <Word from="øch" to="och" />
+    <Word from="flörtar" to="flörtar" />
+    <Word from="kännerjag" to="känner jag" />
+    <Word from="flickan" to="flickan" />
+    <Word from="snø" to="snö" />
+    <Word from="gerju" to="ger ju" />
+    <Word from="køntakter" to="kontakter" />
+    <Word from="ølycka" to="olycka" />
+    <Word from="nølla" to="nolla" />
+    <Word from="sinnenajublar" to="sinnena jublar" />
+    <Word from="ijobbet" to="i jobbet" />
+    <Word from="Fårjag" to="Får jag" />
+    <Word from="Ar" to="Är" />
+    <Word from="liggerju" to="ligger ju" />
+    <Word from="um" to="om" />
+    <Word from="lbland" to="Ibland" />
+    <Word from="skjuterjag" to="skjuter jag" />
+    <Word from="Vaddå" to="Vad då" />
+    <Word from="pratarjämt" to="pratar jämt" />
+    <Word from="harju" to="har ju" />
+    <Word from="sitterjag" to="sitter jag" />
+    <Word from="häfla" to="härja" />
+    <Word from="sfiäl" to="stjäl" />
+    <Word from="FÖU" to="Följ" />
+    <Word from="varförjag" to="varför jag" />
+    <Word from="sfiärna" to="stjärna" />
+    <Word from="böflar" to="börjar" />
+    <Word from="böflan" to="början" />
+    <Word from="stäri" to="står" />
+    <Word from="pä" to="på" />
+    <Word from="harjag" to="har jag" />
+    <Word from="attjag" to="att jag" />
+    <Word from="Verkarjag" to="Verkar jag" />
+    <Word from="Kännerjag" to="Känner jag" />
+    <Word from="därjag" to="där jag" />
+    <Word from="tufi" to="tuff" />
+    <Word from="lurarjag" to="lurar jag" />
+    <Word from="varjättebra" to="var jättebra" />
+    <Word from="allvan" to="allvar" />
+    <Word from="dethär" to="det här" />
+    <Word from="vafle" to="varje" />
+    <Word from="FöUer" to="Följer" />
+    <Word from="personalmötetl" to="personalmötet!" />
+    <Word from="harjust" to="har just" />
+    <Word from="ärjätteduktig" to="är jätteduktig" />
+    <Word from="därja" to="där ja" />
+    <Word from="lngenüng" to="lngenting" />
+    <Word from="iluften" to="i luften" />
+    <Word from="ösen" to="öser" />
+    <Word from="tvâ" to="två" />
+    <Word from="Uejerna" to="Tjejerna" />
+    <Word from="hån*" to="hårt" />
+    <Word from="Ärjag" to="Är jag" />
+    <Word from="keL" to="Okej" />
+    <Word from="Förjag" to="För jag" />
+    <Word from="varjättekul" to="var jättekul" />
+    <Word from="kämpan" to="kämpar" />
+    <Word from="mycketjobb" to="mycket jobb" />
+    <Word from="Uus" to="ljus" />
+    <Word from="serjag" to="ser jag" />
+    <Word from="vetjag" to="vet jag" />
+    <Word from="fårjag" to="får jag" />
+    <Word from="hurjag" to="hur jag" />
+    <Word from="försökerjag" to="försöker jag" />
+    <Word from="tánagel" to="tånagel" />
+    <Word from="vaüe" to="varje" />
+    <Word from="Uudet" to="ljudet" />
+    <Word from="amhopa" to="allihopa" />
+    <Word from="Väü" to="Välj" />
+    <Word from="gäri" to="går" />
+    <Word from="rödüus" to="rödljus" />
+    <Word from="Uuset" to="ljuset" />
+    <Word from="Ridàn" to="Ridån" />
+    <Word from="viüa" to="vilja" />
+    <Word from="gåri" to="går i" />
+    <Word from="Hurdå" to="Hur då" />
+    <Word from="inter\/juar" to="intervjuar" />
+    <Word from="menarjag" to="menar jag" />
+    <Word from="spyrjag" to="spyr jag" />
+    <Word from="briüera" to="briljera" />
+    <Word from="Närjag" to="När jag" />
+    <Word from="ner\/ös" to="nervös" />
+    <Word from="ilivets" to="i livets" />
+    <Word from="nägot" to="något" />
+    <Word from="pà" to="på" />
+    <Word from="Lnnan" to="Innan" />
+    <Word from="Uf" to="Ut" />
+    <Word from="lnnan" to="Innan" />
+    <Word from="Dàren" to="Dåren" />
+    <Word from="Fàrjag" to="Får jag" />
+    <Word from="VadärdetdäL" to="Vad är det där" />
+    <Word from="smàtjuv" to="småtjuv" />
+    <Word from="tàgrånare" to="tågrånare" />
+    <Word from="ditàt" to="ditåt" />
+    <Word from="sä" to="så" />
+    <Word from="vàrdslösa" to="vårdslösa" />
+    <Word from="nàn" to="nån" />
+    <Word from="kommerjag" to="kommer jag" />
+    <Word from="ärjättebra" to="är jättebra" />
+    <Word from="ärjävligt" to="är jävligt" />
+    <Word from="àkerjag" to="åker jag" />
+    <Word from="ellerjapaner" to="eller japaner" />
+    <Word from="attjaga" to="att jaga" />
+    <Word from="eften" to="efter" />
+    <Word from="hästan" to="hästar" />
+    <Word from="Lntensivare" to="Intensivare" />
+    <Word from="fràgarjag" to="frågar jag" />
+    <Word from="pen/ers" to="pervers" />
+    <Word from="ràbarkade" to="råbarkade" />
+    <Word from="styrkon" to="styrkor" />
+    <Word from="Difåf" to="Ditåt" />
+    <Word from="händen" to="händer" />
+    <Word from="föfia" to="följa" />
+    <Word from="Idioten/" to="Idioter!" />
+    <Word from="Varförjagade" to="Varför jagade" />
+    <Word from="därförjag" to="därför jag" />
+    <Word from="forjag" to="for jag" />
+    <Word from="Iivsgladje" to="livsglädje" />
+    <Word from="narjag" to="när jag" />
+    <Word from="sajag" to="sa jag" />
+    <Word from="genastja" to="genast ja" />
+    <Word from="rockumentàren" to="rockumentären" />
+    <Word from="turne" to="turné" />
+    <Word from="fickjag" to="fick jag" />
+    <Word from="sager" to="säger" />
+    <Word from="Ijushårig" to="ljushårig" />
+    <Word from="tradgårdsolycka" to="trädgårdsolycka" />
+    <Word from="kvavdes" to="kvävdes" />
+    <Word from="dàrja" to="där ja" />
+    <Word from="hedersgaster" to="hedersgäster" />
+    <Word from="Nar" to="När" />
+    <Word from="smakiösa" to="smaklösa" />
+    <Word from="lan" to="Ian" />
+    <Word from="Lan" to="Ian" />
+    <Word from="eri" to="er i" />
+    <Word from="universitetsamne" to="universitetsämne" />
+    <Word from="garna" to="gärna" />
+    <Word from="ar" to="är" />
+    <Word from="baltdjur" to="bältdjur" />
+    <Word from="varjag" to="var jag" />
+    <Word from="àr" to="är" />
+    <Word from="förförstàrkare" to="förförstärkare" />
+    <Word from="arjattespeciell" to="är jättespeciell" />
+    <Word from="hàrgår" to="här går" />
+    <Word from="Ia" to="la" />
+    <Word from="Iimousinen" to="limousinen" />
+    <Word from="krickettra" to="kricketträ" />
+    <Word from="hårdrockvàrlden" to="hårdrockvärlden" />
+    <Word from="tràbit" to="träbit" />
+    <Word from="Mellanvastern" to="Mellanvästern" />
+    <Word from="arju" to="är ju" />
+    <Word from="turnen" to="turnén" />
+    <Word from="kanns" to="känns" />
+    <Word from="battre" to="bättre" />
+    <Word from="vàrldsturne" to="världsturne" />
+    <Word from="dar" to="där" />
+    <Word from="sjàlvantànder" to="självantänder" />
+    <Word from="jattelange" to="jättelänge" />
+    <Word from="berattade" to="berättade" />
+    <Word from="Sä" to="Så" />
+    <Word from="vandpunkten" to="vändpunkten" />
+    <Word from="Nàrjag" to="När jag" />
+    <Word from="lasa" to="läsa" />
+    <Word from="skitlàskigt" to="skitläskigt" />
+    <Word from="sambandsvàg" to="sambandsväg" />
+    <Word from="valdigt" to="väldigt" />
+    <Word from="Stamgafiel" to="Stämgaffel" />
+    <Word from="àrjag" to="är jag" />
+    <Word from="tajming" to="tajmning" />
+    <Word from="utgäng" to="utgång" />
+    <Word from="Hàråt" to="Häråt" />
+    <Word from="hàråt" to="häråt" />
+    <Word from="anvander" to="använder" />
+    <Word from="harjobbat" to="har jobbat" />
+    <Word from="imageide" to="imageidé" />
+    <Word from="klafien" to="klaffen" />
+    <Word from="sjalv" to="själv" />
+    <Word from="dvarg" to="dvärg" />
+    <Word from="detjag" to="det jag" />
+    <Word from="dvargarna" to="dvärgarna" />
+    <Word from="fantasivàrld" to="fantasivärld" />
+    <Word from="fiolliga" to="Fjolliga" />
+    <Word from="mandoiinstràngar" to="mandollnsträngar" />
+    <Word from="mittjobb" to="mitt jobb" />
+    <Word from="Skajag" to="Ska jag" />
+    <Word from="landari" to="landar i" />
+    <Word from="gang" to="gäng" />
+    <Word from="Detjag" to="Det jag" />
+    <Word from="Narmre" to="Närmre" />
+    <Word from="Iåtjavelni" to="låtjäveln" />
+    <Word from="Hållerjag" to="Håller jag" />
+    <Word from="visionarer" to="visionärer" />
+    <Word from="Tülvad" to="Till vad" />
+    <Word from="militàrbas" to="militärbas" />
+    <Word from="jattegiada" to="jätteglada" />
+    <Word from="Fastjag" to="Fast jag" />
+    <Word from="såjag" to="så jag" />
+    <Word from="rockvarlden" to="rockvärlden" />
+    <Word from="saknarjag" to="saknar jag" />
+    <Word from="allafall" to="alla fall" />
+    <Word from="fianta" to="fjanta" />
+    <Word from="Kràma" to="Kräma" />
+    <Word from="stammer" to="stämmer" />
+    <Word from="budbàrare" to="budbärare" />
+    <Word from="Iivsfiiosofi" to="livsfiiosofi" />
+    <Word from="förjämnan" to="för jämnan" />
+    <Word from="gillarjag" to="gillar jag" />
+    <Word from="Iarvat" to="larvat" />
+    <Word from="klararjag" to="klarar jag" />
+    <Word from="hattafi'àr" to="hattaffär" />
+    <Word from="Dà" to="Då" />
+    <Word from="uppfinna" to="uppfinna" />
+    <Word from="Ràttfåglar" to="Råttfåglar" />
+    <Word from="Sväüboda" to="Sväljboda" />
+    <Word from="Påböflar" to="Påbörjar" />
+    <Word from="slutarju" to="slutar ju" />
+    <Word from="nifiskebuüken" to="i fiskebutiken" />
+    <Word from="härjäkeln" to="här jäkeln" />
+    <Word from="Hßppa" to="Hoppa" />
+    <Word from="förstörds" to="förstördes" />
+    <Word from="varjättegoda" to="var jättegoda" />
+    <Word from="Kor\/" to="Korv" />
+    <Word from="brüléel" to="brülée!" />
+    <Word from="Hei" to="Hej" />
+    <Word from="älskarjordgubbsglass" to="älskar jordgubbsglass" />
+    <Word from="Snöbom" to="Snöboll" />
+    <Word from="SnöboH" to="Snöboll" />
+    <Word from="Snöbol" to="Snöboll" />
+    <Word from="snöboH" to="snöboll" />
+    <Word from="Läggerpå" to="Lägger på" />
+    <Word from="lngefl" to="lnget!" />
+    <Word from="Sägerjättesmarta" to="Säger jättesmarta" />
+    <Word from="dopplen/äderradar" to="dopplerväderradar" />
+    <Word from="säkertjättefin" to="säkert jättefin" />
+    <Word from="ärjättefin" to="är jättefin" />
+    <Word from="verkarju" to="verkar ju" />
+    <Word from="blirju" to="blir ju" />
+    <Word from="kor\/" to="korv" />
+    <Word from="naturkatastrofi" to="naturkatastrof!" />
+    <Word from="stickerjag" to="stickerj ag" />
+    <Word from="jättebufié" to="jättebuffé" />
+    <Word from="befinner" to="befinner" />
+    <Word from="Spflng" to="Spring" />
+    <Word from="trecfie" to="tredje" />
+    <Word from="ryckerjag" to="rycker jag" />
+    <Word from="skullejag" to="skulle jag" />
+    <Word from="vetju" to="vet ju" />
+    <Word from="afljag" to="att jag" />
+    <Word from="flnns" to="finns" />
+    <Word from="ärlång" to="är lång" />
+    <Word from="kåra" to="kära" />
+    <Word from="ärfina" to="är fina" />
+    <Word from="äri" to="är i" />
+    <Word from="hörden" to="hör den" />
+    <Word from="ättjäg" to="att jäg" />
+    <Word from="gär" to="går" />
+    <Word from="föri" to="för i" />
+    <Word from="Hurvisste" to="Hur visste" />
+    <Word from="fick" to="fick" />
+    <Word from="finns" to="finns" />
+    <Word from="fin" to="fin" />
+    <Word from="Fa" to="Bra." />
+    <Word from="bori" to="bor i" />
+    <Word from="fiendeplanl" to="fiendeplan!" />
+    <Word from="iförnamn" to="i förnamn" />
+    <Word from="detju" to="det ju" />
+    <Word from="Nüd" to="Niki" />
+    <Word from="hatarjag" to="hatar jag" />
+    <Word from="Klararjag" to="Klarar jag" />
+    <Word from="detafier" to="detaljer" />
+    <Word from="vä/" to="väl" />
+    <Word from="smakarju" to="smakar ju" />
+    <Word from="Teachefl" to="Teacher!" />
+    <Word from="imorse" to="i morse" />
+    <Word from="drickerjag" to="dricker jag" />
+    <Word from="ståri" to="står i" />
+    <Word from="Harjag" to="Har jag" />
+    <Word from="Talarjag" to="Talar jag" />
+    <Word from="undrarjag" to="undrar jag" />
+    <Word from="ålderjag" to="ålder jag" />
+    <Word from="vafie" to="varje" />
+    <Word from="förfalskningl" to="förfalskning!" />
+    <Word from="Vifiiiiam" to="William" />
+    <Word from="V\filliams" to="Williams" />
+    <Word from="attjobba" to="att jobba" />
+    <Word from="intei" to="inte i" />
+    <Word from="närV\filliam" to="när William" />
+    <Word from="V\filliam" to="William" />
+    <Word from="Efiersom" to="Eftersom" />
+    <Word from="Vlfilliam" to="William" />
+    <Word from="Iängejag" to="länge jag" />
+    <Word from="'fidigare" to="Tidigare" />
+    <Word from="börjadei" to="började i" />
+    <Word from="merjust" to="mer just" />
+    <Word from="efieråt" to="efteråt" />
+    <Word from="gjordejag" to="gjorde jag" />
+    <Word from="hadeju" to="hade ju" />
+    <Word from="gårvi" to="går vi" />
+    <Word from="köperjag" to="köper jag" />
+    <Word from="Måstejag" to="Måste jag" />
+    <Word from="kännerju" to="känner ju" />
+    <Word from="fln" to="fin" />
+    <Word from="treviig" to="trevlig" />
+    <Word from="Grattisl" to="Grattis!" />
+    <Word from="kande" to="kände" />
+    <Word from="'llden" to="Tiden" />
+    <Word from="sakjag" to="sak jag" />
+    <Word from="klartjag" to="klart jag" />
+    <Word from="häfiigt" to="häftigt" />
+    <Word from="Iämnarjag" to="lämnar jag" />
+    <Word from="gickju" to="gick ju" />
+    <Word from="skajag" to="ska jag" />
+    <Word from="Görjag" to="Gör jag" />
+    <Word from="måstejag" to="måste jag" />
+    <Word from="gra\/iditet" to="graviditet" />
+    <Word from="hittadqdin" to="hittade din" />
+    <Word from="ärjobbigt" to="är jobbigt" />
+    <Word from="Overdrivet" to="Överdrivet" />
+    <Word from="hOgtidlig" to="högtidlig" />
+    <Word from="Overtyga" to="Övertyga" />
+    <Word from="SKILSMASSA" to="SKILSMÄSSA" />
+    <Word from="brukarju" to="brukar ju" />
+    <Word from="lsabel" to="Isabel" />
+    <Word from="kundejag" to="kunde jag" />
+    <Word from="ärläget" to="är läget" />
+    <Word from="blirinte" to="blir inte" />
+    <Word from="l'm" to="I'm" />
+    <Word from="lt's" to="It's" />
+    <Word from="ijakt" to="i jakt" />
+    <Word from="avjordens" to="av jordens" />
+  </WholeWords>
+  <PartialWordsAlways />
+  <PartialWords>
+    <!-- Will be used to check words not in dictionary -->
+    <!-- If new word(s) exists in spelling dictionary, it(they) is accepted -->
+    <WordPart from="¤" to="o" />
+    <WordPart from="fi" to="fi" />
+    <WordPart from="â" to="ä" />
+    <WordPart from="/" to="l" />
+    <WordPart from="vv" to="w" />
+    <WordPart from="IVI" to="M" />
+    <WordPart from="lVI" to="M" />
+    <WordPart from="IVl" to="M" />
+    <WordPart from="lVl" to="M" />
+    <WordPart from="m" to="rn" />
+    <WordPart from="l" to="i" />
+    <WordPart from="€" to="e" />
+    <WordPart from="I" to="l" />
+    <WordPart from="c" to="o" />
+    <WordPart from="i" to="t" />
+    <WordPart from="cc" to="oo" />
+    <WordPart from="ii" to="tt" />
+    <WordPart from="n/" to="ry" />
+    <WordPart from="ae" to="æ" />
+    <!-- "f " will be two words -->
+    <WordPart from="f" to="f " />
+    <WordPart from="c" to="e" />
+    <WordPart from="o" to="e" />
+    <WordPart from="I" to="t" />
+    <WordPart from="n" to="o" />
+    <WordPart from="s" to="e" />
+    <WordPart from="å" to="ä" />
+    <WordPart from="à" to="å" />
+    <WordPart from="n/" to="rv" />
+  </PartialWords>
+  <PartialLines />
+  <PartialLinesAlways />
+  <BeginLines>
+    <Beginning from="Ln " to="In " />
+    <Beginning from="U ppfattat" to="Uppfattat" />
+  </BeginLines>
+  <EndLines />
+  <WholeLines />
+  <RegularExpressions />
+</OCRFixReplaceList>

Contents/Libraries/Shared/subzero/modification/main.py

@@ -0,0 +1,130 @@
+# coding=utf-8
+
+import traceback
+import pysubs2
+import logging
+
+from registry import registry
+
+logger = logging.getLogger(__name__)
+
+
+class SubtitleModifications(object):
+    debug = False
+    language = None
+    initialized_mods = {}
+
+    def __init__(self, debug=False):
+        self.debug = debug
+        self.initialized_mods = {}
+
+    def load(self, fn=None, content=None, language=None):
+        """
+        
+        :param language: babelfish.Language language of the subtitle
+        :param fn:  filename
+        :param content: unicode 
+        :return: 
+        """
+        self.language = language
+        self.initialized_mods = {}
+        try:
+            if fn:
+                self.f = pysubs2.load(fn)
+            elif content:
+                self.f = pysubs2.SSAFile.from_string(content)
+        except (IOError,
+                UnicodeDecodeError,
+                pysubs2.exceptions.UnknownFPSError,
+                pysubs2.exceptions.UnknownFormatIdentifierError,
+                pysubs2.exceptions.FormatAutodetectionError):
+            if fn:
+                logger.exception("Couldn't load subtitle: %s: %s", fn, traceback.format_exc())
+            elif content:
+                logger.exception("Couldn't load subtitle: %s", traceback.format_exc())
+
+    @classmethod
+    def parse_identifier(cls, identifier):
+        # simple identifier
+        if identifier in registry.mods:
+            return identifier, {}
+
+        # identifier with params; identifier(param=value)
+        split_args = identifier[identifier.find("(")+1:-1].split(",")
+        args = dict((key, value) for key, value in [sub.split("=") for sub in split_args])
+        return identifier[:identifier.find("(")], args
+
+    @classmethod
+    def get_mod_class(cls, identifier):
+        identifier, args = cls.parse_identifier(identifier)
+        return registry.mods[identifier]
+
+    @classmethod
+    def get_mod_signature(cls, identifier, **kwargs):
+        return cls.get_mod_class(identifier).get_signature(**kwargs)
+
+    def modify(self, *mods):
+        new_f = []
+
+        parsed_mods = [SubtitleModifications.parse_identifier(mod) for mod in mods]
+        line_mods = []
+        non_line_mods = []
+
+        print parsed_mods
+        for identifier, args in parsed_mods:
+            if identifier not in registry.mods:
+                raise NotImplementedError("Mod %s not loaded" % identifier)
+
+            mod_cls = registry.mods[identifier]
+            if mod_cls.modifies_whole_file:
+                non_line_mods.append((identifier, args))
+            else:
+                line_mods.append((identifier, args))
+
+            if identifier not in self.initialized_mods:
+                self.initialized_mods[identifier] = mod_cls(self)
+
+        # apply file mods
+        if non_line_mods:
+            for identifier, args in non_line_mods:
+                mod = self.initialized_mods[identifier]
+                mod.modify(None, debug=self.debug, parent=self, **args)
+
+        # apply line mods
+        if line_mods:
+            for line in self.f:
+                applied_mods = []
+                for identifier, args in line_mods:
+                    mod = self.initialized_mods[identifier]
+
+                    # don't bother reapplying exclusive mods multiple times
+                    if mod.exclusive and identifier in applied_mods:
+                        continue
+
+                    if not mod.processors:
+                        continue
+
+                    new_content = mod.modify(line.text, debug=self.debug, parent=self, **args)
+                    if not new_content:
+                        if self.debug:
+                            logger.debug("%s: deleting %s", identifier, line)
+                        continue
+
+                    line.text = new_content
+                    applied_mods.append(identifier)
+                new_f.append(line)
+
+        self.f.events = new_f
+
+    def to_string(self, format="srt", encoding="utf-8"):
+        return self.f.to_string(format, encoding=encoding)
+
+    def save(self, fn):
+        self.f.save(fn)
+
+
+SubMod = SubtitleModifications
+
+
+
+

Contents/Libraries/Shared/subzero/modification/mods/__init__.py

@@ -0,0 +1,84 @@
+# coding=utf-8
+import re
+import logging
+
+from subzero.modification.processors.re_processor import ReProcessor, NReProcessor
+
+logger = logging.getLogger(__name__)
+
+
+class SubtitleModification(object):
+    identifier = None
+    description = None
+    long_description = None
+    exclusive = False
+    advanced = False  # has parameters
+    modifies_whole_file = False  # operates on the whole file, not individual entries
+    pre_processors = []
+    processors = []
+    post_processors = []
+
+    def __init__(self, parent):
+        return
+
+    def _process(self, content, processors, debug=False, parent=None, **kwargs):
+        if not content:
+            return
+
+        # processors may be a list or a callable
+        #if callable(processors):
+        #    _processors = processors()
+        #else:
+        #    _processors = processors
+        _processors = processors
+
+        new_content = content
+        for processor in _processors:
+            old_content = new_content
+            new_content = processor.process(new_content, debug=debug)
+            if not new_content:
+                if debug:
+                    logger.debug("Processor returned empty line: %s", processor)
+                break
+            if debug:
+                if old_content == new_content:
+                    continue
+                logger.debug("%s: %s -> %s", processor, old_content, new_content)
+        return new_content
+
+    def pre_process(self, content, debug=False, parent=None, **kwargs):
+        return self._process(content, self.pre_processors, debug=debug, parent=parent, **kwargs)
+
+    def process(self, content, debug=False, parent=None, **kwargs):
+        return self._process(content, self.processors, debug=debug, parent=parent, **kwargs)
+
+    def post_process(self, content, debug=False, parent=None, **kwargs):
+        return self._process(content, self.post_processors, debug=debug, parent=parent, **kwargs)
+
+    def modify(self, content, debug=False, parent=None, **kwargs):
+        new_content = content
+        for method in ("pre_process", "process", "post_process"):
+            new_content = getattr(self, method)(new_content, debug=debug, parent=parent, **kwargs)
+
+        return new_content
+
+    @classmethod
+    def get_signature(cls, **kwargs):
+        string_args = ",".join(["%s=%s" % (key, value) for key, value in kwargs.iteritems()])
+        return "%s(%s)" % (cls.identifier, string_args)
+
+
+class SubtitleTextModification(SubtitleModification):
+    post_processors = [
+        # empty tag
+        ReProcessor(re.compile(r'({\\\w+1})[\s.,-_!?]+({\\\w+0})'), "", name="empty_tag"),
+
+        # empty line (needed?)
+        NReProcessor(re.compile(r'^\s+$'), "", name="empty_line"),
+
+        # empty dash line (needed?)
+        NReProcessor(re.compile(r'(^[\s]*[\-]+[\s]*)$'), "", name="empty_dash_line"),
+
+        # clean whitespace at start and end
+        ReProcessor(re.compile(r'^\s*([^\s]+)\s*$'), r"\1", name="surrounding_whitespace"),
+    ]

Contents/Libraries/Shared/subzero/modification/mods/fps.py

@@ -0,0 +1,27 @@
+# coding=utf-8
+
+import logging
+
+from subzero.modification.mods import SubtitleModification
+from subzero.modification import registry
+
+logger = logging.getLogger(__name__)
+
+
+class ChangeFPS(SubtitleModification):
+    identifier = "change_FPS"
+    description = "Change the FPS of the subtitle"
+    exclusive = True
+    advanced = True
+    modifies_whole_file = True
+
+    long_description = """\
+    Re-syncs the subtitle to the framerate of the current media file. 
+    """
+
+    def modify(self, content, debug=False, parent=None, **kwargs):
+        fps_from = kwargs.get("from")
+        fps_to = kwargs.get("to")
+        parent.f.transform_framerate(float(fps_from), float(fps_to))
+
+registry.register(ChangeFPS)

Contents/Libraries/Shared/subzero/modification/mods/hearing_impaired.py

@@ -0,0 +1,33 @@
+# coding=utf-8
+import re
+
+from subzero.modification.mods import SubtitleTextModification
+from subzero.modification.processors.re_processor import NReProcessor
+from subzero.modification import registry
+
+
+class HearingImpaired(SubtitleTextModification):
+    identifier = "remove_HI"
+    description = "Remove Hearing Impaired tags"
+    exclusive = True
+
+    long_description = """\
+    Removes tags, text and characters from subtitles that are meant for hearing impaired people
+    """
+
+    processors = [
+        # brackets
+        NReProcessor(re.compile(r'(?sux)[([].+[)\]]'), "", name="HI_brackets"),
+
+        # text before colon (and possible dash in front), max 11 chars after the first whitespace (if any)
+        NReProcessor(re.compile(r'(?u)(^[A-z\-\'"_]+[\w\s]{0,11}:[^0-9{2}][\s]*)'), "", name="HI_before_colon"),
+
+        # all caps line (at least 3 chars)
+        NReProcessor(re.compile(r'(?u)(^[A-Z]{3,}$)'), "", name="HI_all_caps"),
+
+        # dash in front
+        NReProcessor(re.compile(r'(?u)^\s*-\s*'), "", name="HI_starting_dash"),
+    ]
+
+
+registry.register(HearingImpaired)

Contents/Libraries/Shared/subzero/modification/mods/ocr_fixes.py

@@ -0,0 +1,47 @@
+# coding=utf-8
+import logging
+
+from subzero.modification.mods import SubtitleTextModification
+from subzero.modification.processors.string_processor import MultipleLineProcessor
+from subzero.modification.processors.re_processor import MultipleWordReProcessor
+from subzero.modification import registry
+from subzero.modification.dictionaries.data import data as OCR_fix_data
+
+logger = logging.getLogger(__name__)
+
+
+class FixOCR(SubtitleTextModification):
+    identifier = "OCR_fixes"
+    description = "Fix common OCR issues"
+    exclusive = True
+    data_dict = None
+
+    long_description = """\
+    Fix issues that happen when a subtitle gets converted from bitmap to text through OCR
+    """
+
+    def __init__(self, parent):
+        super(FixOCR, self).__init__(parent)
+        data_dict = OCR_fix_data.get(parent.language.alpha3t)
+        if not data_dict:
+            logger.debug("No SnR-data available for language %s", parent.language)
+            return
+
+        self.data_dict = data_dict
+        self.processors = self.get_processors()
+
+    def get_processors(self):
+        if not self.data_dict:
+            return []
+
+        return [
+            MultipleLineProcessor(self.data_dict["WholeLines"], name="SE_replace_line"),
+            MultipleWordReProcessor(self.data_dict["WholeWords"], name="SE_replace_word"),
+            MultipleWordReProcessor(self.data_dict["BeginLines"], name="SE_replace_beginline"),
+            MultipleWordReProcessor(self.data_dict["EndLines"], name="SE_replace_endline"),
+            MultipleLineProcessor(self.data_dict["PartialLines"], name="SE_replace_partialline"),
+            MultipleLineProcessor(self.data_dict["PartialWordsAlways"], name="SE_replace_partialwordsalways")
+        ]
+
+
+registry.register(FixOCR)

Contents/Libraries/Shared/subzero/modification/mods/offset.py

@@ -0,0 +1,27 @@
+# coding=utf-8
+
+import logging
+
+from subzero.modification.mods import SubtitleModification
+from subzero.modification import registry
+
+logger = logging.getLogger(__name__)
+
+
+class ShiftOffset(SubtitleModification):
+    identifier = "shift_offset"
+    description = "Change the timing of the subtitle"
+    exclusive = True
+    advanced = True
+    modifies_whole_file = True
+
+    long_description = """\
+    Adds or substracts a certain amount of time from the whole subtitle to match your media 
+    """
+
+    def modify(self, content, debug=False, parent=None, **kwargs):
+        parent.f.shift(h=int(kwargs.get("h", 0)), m=int(kwargs.get("m", 0)), s=int(kwargs.get("s", 0)),
+                       ms=int(kwargs.get("ms", 0)))
+
+
+registry.register(ShiftOffset)

Contents/Libraries/Shared/subzero/modification/processors/__init__.py

@@ -0,0 +1,29 @@
+# coding=utf-8
+
+
+class Processor(object):
+    """
+    Processor base class
+    """
+    name = None
+    parent = None
+
+    def __init__(self, name=None, parent=None):
+        self.name = name
+        self.parent = parent
+
+    @property
+    def info(self):
+        return self.name
+
+    def process(self, content, debug=False):
+        return content
+
+    def __repr__(self):
+        return "Processor <%s %s>" % (self.__class__.__name__, self.info)
+
+    def __str__(self):
+        return repr(self)
+
+    def __unicode__(self):
+        return unicode(repr(self))

Contents/Libraries/Shared/subzero/modification/processors/re_processor.py

@@ -0,0 +1,62 @@
+# coding=utf-8
+import re
+import logging
+
+from subzero.modification.processors import Processor
+
+logger = logging.getLogger(__name__)
+
+
+class ReProcessor(Processor):
+    """
+    Regex processor
+    """
+    pattern = None
+    replace_with = None
+
+    def __init__(self, pattern, replace_with, name=None):
+        super(ReProcessor, self).__init__(name=name)
+        self.pattern = pattern
+        self.replace_with = replace_with
+
+    def process(self, content, debug=False):
+        return self.pattern.sub(self.replace_with, content)
+
+
+class NReProcessor(ReProcessor):
+    """
+    Single line regex processor
+    """
+
+    def process(self, content, debug=False):
+        lines = []
+        for line in content.split(r"\N"):
+            a = super(NReProcessor, self).process(line, debug=debug)
+            if not a:
+                continue
+            lines.append(a)
+        return r"\N".join(lines)
+
+
+class MultipleWordReProcessor(ReProcessor):
+    """
+    Expects a dictionary in the form of:
+    dict = {
+        "data": {"old_value": "new_value"},
+        "pattern": compiled re object that matches data.keys()
+    }
+    replaces found key in pattern with the corresponding value in data
+    """
+    def __init__(self, snr_dict, name=None, parent=None):
+        super(ReProcessor, self).__init__(name=name)
+        self.snr_dict = snr_dict
+
+    def process(self, content, debug=False):
+        if not self.snr_dict["data"]:
+            return content
+
+        out = []
+        for a in content.split(ur"\N"):
+            out.append(self.snr_dict["pattern"].sub(lambda x: self.snr_dict["data"][x.group(0)], a))
+        return ur"\N".join(out)
+

Contents/Libraries/Shared/subzero/modification/processors/string_processor.py

@@ -0,0 +1,75 @@
+# coding=utf-8
+
+import logging
+
+from subzero.modification.processors import Processor
+
+logger = logging.getLogger(__name__)
+
+
+class StringProcessor(Processor):
+    """
+    String replacement processor base
+    """
+
+    def __init__(self, search, replace, name=None, parent=None):
+        super(StringProcessor, self).__init__(name=name)
+        self.search = search
+        self.replace = replace
+
+    def process(self, content, debug=False):
+        return content.replace(self.search, self.replace)
+
+
+class MultipleLineProcessor(Processor):
+    """
+    replaces stuff in whole lines
+    
+    takes a search/replace dict as first argument
+    Expects a dictionary in the form of:
+    dict = {
+        "data": {"old_value": "new_value"}
+    }
+    """
+    def __init__(self, snr_dict, name=None, parent=None):
+        super(MultipleLineProcessor, self).__init__(name=name)
+        self.snr_dict = snr_dict
+
+    def process(self, content, debug=False):
+        if not self.snr_dict["data"]:
+            return content
+
+        out = []
+        for cnt in content.split(ur"\N"):
+            cnt_ = cnt
+            for key, value in self.snr_dict["data"].iteritems():
+                if debug and key in cnt_:
+                    logger.debug(u"Replacing '%s' with '%s' in '%s'", key, value, cnt_)
+
+                cnt_ = cnt_.replace(key, value)
+            out.append(cnt_)
+
+        return ur"\N".join(out)
+
+
+class MultipleWordProcessor(MultipleLineProcessor):
+    """
+    replaces words
+
+    takes a search/replace dict as first argument
+    Expects a dictionary in the form of:
+    dict = {
+        "data": {"old_value": "new_value"}
+    }
+    """
+    def process(self, content, debug=False):
+        new_lines = []
+        for line in content.split(u"\\N"):
+            words = line.split(u" ")
+            new_words = []
+            for word in words:
+                new_words.append(self.snr_dict.get(word, word))
+
+            new_lines.append(u" ".join(new_words))
+
+        return u"\\N".join(new_lines)

Contents/Libraries/Shared/subzero/modification/registry.py

@@ -0,0 +1,17 @@
+# coding=utf-8
+from collections import OrderedDict
+
+
+class SubtitleModRegistry(object):
+    mods = None
+    mods_available = None
+
+    def __init__(self):
+        self.mods = OrderedDict()
+        self.mods_available = []
+
+    def register(self, mod):
+        self.mods[mod.identifier] = mod
+        self.mods_available.append(mod.identifier)
+
+registry = SubtitleModRegistry()

Contents/Libraries/Shared/test.srt

@@ -14,19 +14,19 @@ of signal, drawing the Tardis off course.
 
 4
 00:00:16,099 --> 00:00:17,224
-Where are we?
+this is a subtitle test with a text before colons: Where are we?
 
 5
 00:00:17,225 --> 00:00:19,684
-Earth. Utah, North America.
+less text before colons: Earth. Utah, North America.
 
 6
 00:00:19,686 --> 00:00:21,103
-About half a mile underground.
+Ithinkyou're About half a mile underground.
 
 7
 00:00:21,103 --> 00:00:23,603
-And when are we?
+lrn gonna And when are we?
 
 8
 00:00:24,274 --> 00:00:26,649
@@ -55,7 +55,7 @@ ROSE: Like a great big museum.
 
 14
 00:00:40,414 --> 00:00:42,914
-DOCTOR: An alien museum.
+DOCTOR's MOM: An alien museum.
 
 15
 00:00:43,542 --> 00:00:46,042