TEMP

Escapes HTML as described in:
https://html.spec.whatwg.org/multipage/parsing.html#escapingString

. codecov.yml

@@ -0,0 +1,5 @@
+codecov:
+  branch: develop
+
+ignore:
+  - "Tests/"

.jazzy.yaml

@@ -1,5 +1,5 @@
 module: HTMLKit
-module_version: 1.0.0
+module_version: 2.1.4
 author: Iskandar Abudiab 
 author_url: https://twitter.com/iabudiab
 github_url: https://github.com/iabudiab/HTMLKit
@@ -27,10 +27,12 @@ custom_categories:
     - HTMLDocumentType
     - HTMLDocumentFragment
     - HTMLDocumentPosition
+    - HTMLCharacterData
     - HTMLText
     - HTMLComment
     - HTMLTemplate
     - HTMLDOMTokenList
+    - HTMLRange
 
   - name: Iteration & Filtering
     children:

.travis.yml

@@ -1,5 +1,5 @@
 language: objective-c
-osx_image: xcode8
+osx_image: xcode9
 
 branches:
   except:
@@ -17,29 +17,29 @@ env:
     - MACOS_FRAMEWORK_SCHEME=HTMLKit-macOS
     - WATCHOS_FRAMEWORK_SCHEME="HTMLKit-watchOS"
     - TVOS_FRAMEWORK_SCHEME="HTMLKit-tvOS"
-    - IOS_SDK=iphonesimulator10.0
-    - MACOS_SDK=macosx10.12
-    - WATCHOS_SDK=watchsimulator3.0
-    - TVOS_SDK=appletvsimulator10.0
+    - IOS_SDK=iphonesimulator11.0
+    - MACOS_SDK=macosx10.13
+    - WATCHOS_SDK=watchsimulator4.0
+    - TVOS_SDK=appletvsimulator11.0
   matrix:
-    - DESTINATION="arch=x86_64"                     SIMULATOR=""                          SCHEME="$MACOS_FRAMEWORK_SCHEME"    SDK="$MACOS_SDK"
-    - DESTINATION="OS=9.0,name=iPhone 6"            SIMULATOR="iPhone 6 (9.0)"            SCHEME="$IOS_FRAMEWORK_SCHEME"      SDK="$IOS_SDK"
-    - DESTINATION="OS=9.1,name=iPhone 6 Plus"       SIMULATOR="iPhone 6 Plus (9.1)"       SCHEME="$IOS_FRAMEWORK_SCHEME"      SDK="$IOS_SDK"
-    - DESTINATION="OS=9.2,name=iPhone 6S"           SIMULATOR="iPhone 6S (9.2)"           SCHEME="$IOS_FRAMEWORK_SCHEME"      SDK="$IOS_SDK"
-    - DESTINATION="OS=9.3,name=iPhone 6S Plus"      SIMULATOR="iPhone 6S Plus (9.3)"      SCHEME="$IOS_FRAMEWORK_SCHEME"      SDK="$IOS_SDK"
-    - DESTINATION="OS=10.0,name=iPhone 7 Plus"      SIMULATOR="iPhone 7 Plus (10.0)"      SCHEME="$IOS_FRAMEWORK_SCHEME"      SDK="$IOS_SDK"
-    - DESTINATION="OS=2.0,name=Apple Watch - 42mm"  SIMULATOR="Apple Watch - 42mm (2.2)"  SCHEME="$WATCHOS_FRAMEWORK_SCHEME"  SDK="$WATCHOS_SDK"
-    - DESTINATION="OS=3.0,name=Apple Watch - 42mm"  SIMULATOR="Apple Watch - 42mm (3.0)"  SCHEME="$WATCHOS_FRAMEWORK_SCHEME"  SDK="$WATCHOS_SDK"
-    - DESTINATION="OS=9.0,name=Apple TV 1080p"      SIMULATOR="Apple TV 1080p (9.2)"      SCHEME="$TVOS_FRAMEWORK_SCHEME"     SDK="$TVOS_SDK"
-    - DESTINATION="OS=10.0,name=Apple TV 1080p"     SIMULATOR="Apple TV 1080p (10.0)"     SCHEME="$TVOS_FRAMEWORK_SCHEME"     SDK="$TVOS_SDK"
+    - DESTINATION="arch=x86_64"                                SCHEME="$MACOS_FRAMEWORK_SCHEME"      SDK="$MACOS_SDK"
+    - DESTINATION="OS=9.3,name=iPhone 6s Plus"                 SCHEME="$IOS_FRAMEWORK_SCHEME"        SDK="$IOS_SDK"
+    - DESTINATION="OS=10.3.1,name=iPhone 7 Plus"               SCHEME="$IOS_FRAMEWORK_SCHEME"        SDK="$IOS_SDK"
+    - DESTINATION="OS=11.0,name=iPhone X"                      SCHEME="$IOS_FRAMEWORK_SCHEME"        SDK="$IOS_SDK"
+    - DESTINATION="OS=3.2,name=Apple Watch Series 2 - 42mm"    SCHEME="$WATCHOS_FRAMEWORK_SCHEME"    SDK="$WATCHOS_SDK"
+    - DESTINATION="OS=4.0,name=Apple Watch Series 3 - 42mm"    SCHEME="$WATCHOS_FRAMEWORK_SCHEME"    SDK="$WATCHOS_SDK"
+    - DESTINATION="OS=10.2,name=Apple TV 1080p"                SCHEME="$TVOS_FRAMEWORK_SCHEME"       SDK="$TVOS_SDK"
+    - DESTINATION="OS=11.0,name=Apple TV 4K"                   SCHEME="$TVOS_FRAMEWORK_SCHEME"       SDK="$TVOS_SDK"
 
 script:
   - set -o pipefail
   - xcodebuild -version
   - xcodebuild -showsdks
-  - SIMULATOR_ID=$(xcrun instruments -s devices | grep -io "$SIMULATOR \[.*\]" | grep -o "\[.*\]" | sed "s/^\[\(.*\)\]$/\1/")
-  - open -b com.apple.iphonesimulator --args -CurrentDeviceUDID $SIMULATOR_ID
-  - xcodebuild -workspace "$WORKSPACE" -scheme "$SCHEME" -sdk "$SDK" -destination "$DESTINATION" -configuration Debug ONLY_ACTIVE_ARCH=NO clean build | xcpretty -c
+  - travis_retry xcodebuild -workspace "$WORKSPACE" -scheme "$SCHEME" -sdk "$SDK" -destination "$DESTINATION" -configuration Debug ONLY_ACTIVE_ARCH=NO clean build | xcpretty -c
   - if [ "$SDK" != "$WATCHOS_SDK" ]; then
-      xcodebuild -workspace "$WORKSPACE" -scheme "$SCHEME" -sdk "$SDK" -destination "$DESTINATION" -configuration Debug ONLY_ACTIVE_ARCH=NO test | xcpretty -c;
+      travis_retry xcodebuild -workspace "$WORKSPACE" -scheme "$SCHEME" -sdk "$SDK" -destination "$DESTINATION" -configuration Debug ONLY_ACTIVE_ARCH=NO GCC_INSTRUMENT_PROGRAM_FLOW_ARCS=YES GCC_GENERATE_TEST_COVERAGE_FILES=YES test | xcpretty -c;
     fi
+
+after_success:
+  - bash <(curl -s https://codecov.io/bash)
+

CHANGELOG.md

@@ -1,5 +1,172 @@
 # Change Log
 
+## [2.1.4](https://github.com/iabudiab/HTMLKit/releases/tag/2.1.4) 
+
+Released on 2018.05.01
+
+### Fixes
+
+- `gt(n)`, `lt(n)` and `eq(n)` selectors would select wrong elements for the zero-index (issue #25)
+
+
+## [2.1.3](https://github.com/iabudiab/HTMLKit/releases/tag/2.1.3) 
+
+Released on 2018.03.21
+
+### Fixes
+
+- `HTMLElement` clone would return an immutable dictionary for attributes (issue #20)
+	- Fixed by @CRivlaldo in PR #24
+- `HTMLNodeFilterBlock` would behave differently on simulator and device (issue #22)
+	- Fixed by @CRivlaldo in PR #23
+
+
+## [2.1.2](https://github.com/iabudiab/HTMLKit/releases/tag/2.1.2) 
+
+Released on 2017.11.6
+
+### Fixes
+
+- `HTMLText` serialization (issue #16)
+- `HTMLElement` attribute value serialization (issue #17)
+
+
+## [2.1.1](https://github.com/iabudiab/HTMLKit/releases/tag/2.1.1) 
+
+Released on 2017.10.13
+
+### Hotfix
+
+- Fixed documentation comments
+	- Should fix CocoaDocs generation and percentage
+
+
+## [2.1.0](https://github.com/iabudiab/HTMLKit/releases/tag/2.1.0) 
+
+Released on 2017.10.12
+
+### Added
+
+- Standarized tokenizer error codes:
+	- [whatwg/html#2701](https://github.com/whatwg/html/pull/2701)
+	- [html5lib/html5lib-tests#92](https://github.com/html5lib/html5lib-tests/pull/92)
+
+### Updated
+
+- Project for Xcode 9
+- Travis config for iOS 11.0, macOS 10.13, tvOS 11.0 and watchOS 4.0
+- Updated HTML5Lib-Tests submodule (cbafeba)
+
+## [2.0.6](https://github.com/iabudiab/HTMLKit/releases/tag/2.0.6) 
+
+Released on 2017.05.02
+
+### Added
+
+- Memory consumption improvements (issue #10)
+	- Allocate `childNodes` collection in `HTMLNode` only when inserting child nodes
+	- Replace `NSStringFromSelector` calls with constants in `HTMLNode` validations
+	- Improve `reverseObjectEnumerator` usage while parsing HTML
+	- Rewrite internal logic of the `HTMLStackOfOpenElements` to prevent excessive allocations
+
+
+## [2.0.5](https://github.com/iabudiab/HTMLKit/releases/tag/2.0.5) 
+
+Released on 2017.04.19
+
+### Fixed
+
+- Xcode 8.3 issue with modulemaps
+	- Temporary workaround (renamed modulemap file)
+- Memory Leaks in `CSSInputStream`
+
+### Added
+
+- Minor memory consumption improvements
+	- Collections for child nodes or attributes of HTML Nodes or Elements are allocated lazily
+	- Underyling data string of `CharacterData` is allocated on first access
+	- Autorelease pool for the main `HTMLTokenizer` loop
+
+
+## [2.0.4](https://github.com/iabudiab/HTMLKit/releases/tag/2.0.4) 
+
+Released on 2017.04.2
+
+### Fixed
+
+- Testing with Swift 3.1
+	- Fixed by @tali in PR #8
+
+### Deprecated
+
+- `HTMLRange` initializers with typo
+	- `initWithDowcument:startContainer:startOffset:endContainer:endOffset:`
+
+
+## [2.0.3](https://github.com/iabudiab/HTMLKit/releases/tag/2.0.3) 
+
+Released on 2017.03.6
+
+### Fixed
+
+- Compilation for Swift 3.1
+	- Fixed by @tali in PR #6
+
+
+## [2.0.2](https://github.com/iabudiab/HTMLKit/releases/tag/2.0.2) 
+
+Released on 2017.02.26
+
+### Fixed
+
+- Retain cycles in `HTMLNodeIterator` (issue #4)
+- Retain cycles in `HTMLRange` (issue #5)
+- The layout of `HTMLKit` tests module for Swift Package Manager
+
+
+## [2.0.1](https://github.com/iabudiab/HTMLKit/releases/tag/2.0.1) 
+
+Released on 2017.02.20
+
+### Hotifx
+
+- Set `INSTALL_PATH` and `DYLIB_INSTALL_NAME_BASE` to `@rpath` for macOS target
+	- This fixes embedding `HTMLKit` in a Cocoa application
+
+
+## [2.0.0](https://github.com/iabudiab/HTMLKit/releases/tag/2.0.0) 
+
+Released on 2017.02.11
+
+### Spec Change
+
+- Make `<menuitem>` parse like an unkonwn element. See:
+	- [whatwg/html#2319](https://github.com/whatwg/html/pull/2319)
+	- [html5lib/html5lib-tests#88](https://github.com/html5lib/html5lib-tests/pull/88)
+
+### Updated
+
+- Updated HTML5Lib-Tests submodule (13f1805)
+
+
+## [1.1.0](https://github.com/iabudiab/HTMLKit/releases/tag/1.1.0) 
+
+Released on 2017.01.14
+
+### Added
+
+- `DOM Ranges` implementation ([spec](https://dom.spec.whatwg.org/#ranges))
+- `HTMLChatacterData` as base class for `HTMLText` & `HTMLComment`
+	- `HTMLText` and `HTMLComment` no longer extend `HTMLNode` directly
+- `splitText` implementation for `HTMLText` nodes
+- `index` property for `HTMLNode`
+- `cloneNodeDeep` method for `HTMLNode`
+
+### Deprecated
+
+- `appendString` method in `HTMLText` in favor of `appendData` from the supperclass `HTMLCharacterData`
+
+
 ## [1.0.0](https://github.com/iabudiab/HTMLKit/releases/tag/1.0.0) 
 
 Released on 2016.09.28
@@ -61,10 +228,7 @@ This release passes all tokenizer and tree construction html5lib-tests as of 201
 - `<isindex>` is completely removed from the spec now, therefore it is dropped from the implementation
 - `Tokenizer` and `Tree-Construction` tests are now generated dynamically
 - Test failures are collected by a `XCTestObservation` for better reporting
-	
-- `<isindex>` is completely removed from the spec now, therefore it is dropped from the implementation
-- `Tokenizer` and `Tree-Construction` tests are now generated dynamically
-- Test failures are collected by a `XCTestObservation` for better reporting
+
 
 ### Fixed
 

Example/HTMLKitExample/HTMLKitExample.xcodeproj/project.pbxproj

@@ -89,7 +89,7 @@
 			isa = PBXProject;
 			attributes = {
 				LastSwiftUpdateCheck = 0800;
-				LastUpgradeCheck = 0800;
+				LastUpgradeCheck = 0900;
 				ORGANIZATIONNAME = iabudiab;
 				TargetAttributes = {
 					629A63C81D9AFE0E0089679F = {
@@ -136,7 +136,9 @@
 				CLANG_CXX_LIBRARY = "libc++";
 				CLANG_ENABLE_MODULES = YES;
 				CLANG_ENABLE_OBJC_ARC = YES;
+				CLANG_WARN_BLOCK_CAPTURE_AUTORELEASING = YES;
 				CLANG_WARN_BOOL_CONVERSION = YES;
+				CLANG_WARN_COMMA = YES;
 				CLANG_WARN_CONSTANT_CONVERSION = YES;
 				CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
 				CLANG_WARN_DOCUMENTATION_COMMENTS = YES;
@@ -144,7 +146,12 @@
 				CLANG_WARN_ENUM_CONVERSION = YES;
 				CLANG_WARN_INFINITE_RECURSION = YES;
 				CLANG_WARN_INT_CONVERSION = YES;
+				CLANG_WARN_NON_LITERAL_NULL_CONVERSION = YES;
+				CLANG_WARN_OBJC_LITERAL_CONVERSION = YES;
 				CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
+				CLANG_WARN_RANGE_LOOP_ANALYSIS = YES;
+				CLANG_WARN_STRICT_PROTOTYPES = YES;
+				CLANG_WARN_SUSPICIOUS_MOVE = YES;
 				CLANG_WARN_SUSPICIOUS_MOVES = YES;
 				CLANG_WARN_UNREACHABLE_CODE = YES;
 				CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
@@ -184,7 +191,9 @@
 				CLANG_CXX_LIBRARY = "libc++";
 				CLANG_ENABLE_MODULES = YES;
 				CLANG_ENABLE_OBJC_ARC = YES;
+				CLANG_WARN_BLOCK_CAPTURE_AUTORELEASING = YES;
 				CLANG_WARN_BOOL_CONVERSION = YES;
+				CLANG_WARN_COMMA = YES;
 				CLANG_WARN_CONSTANT_CONVERSION = YES;
 				CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
 				CLANG_WARN_DOCUMENTATION_COMMENTS = YES;
@@ -192,7 +201,12 @@
 				CLANG_WARN_ENUM_CONVERSION = YES;
 				CLANG_WARN_INFINITE_RECURSION = YES;
 				CLANG_WARN_INT_CONVERSION = YES;
+				CLANG_WARN_NON_LITERAL_NULL_CONVERSION = YES;
+				CLANG_WARN_OBJC_LITERAL_CONVERSION = YES;
 				CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
+				CLANG_WARN_RANGE_LOOP_ANALYSIS = YES;
+				CLANG_WARN_STRICT_PROTOTYPES = YES;
+				CLANG_WARN_SUSPICIOUS_MOVE = YES;
 				CLANG_WARN_SUSPICIOUS_MOVES = YES;
 				CLANG_WARN_UNREACHABLE_CODE = YES;
 				CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
@@ -212,6 +226,7 @@
 				MACOSX_DEPLOYMENT_TARGET = 10.11;
 				MTL_ENABLE_DEBUG_INFO = NO;
 				SDKROOT = macosx;
+				SWIFT_OPTIMIZATION_LEVEL = "-Owholemodule";
 			};
 			name = Release;
 		};
@@ -250,6 +265,7 @@
 				629A63D21D9AFE0E0089679F /* Release */,
 			);
 			defaultConfigurationIsVisible = 0;
+			defaultConfigurationName = Release;
 		};
 /* End XCConfigurationList section */
 	};

Example/HTMLKitExample/HTMLKitExample.xcodeproj/xcshareddata/xcschemes/HTMLKitExample.xcscheme

@@ -1,6 +1,6 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <Scheme
-   LastUpgradeVersion = "0800"
+   LastUpgradeVersion = "0900"
    version = "1.3">
    <BuildAction
       parallelizeBuildables = "YES"
@@ -26,6 +26,7 @@
       buildConfiguration = "Debug"
       selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
       selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
+      language = ""
       shouldUseLaunchSchemeArgsEnv = "YES">
       <Testables>
       </Testables>
@@ -45,6 +46,7 @@
       buildConfiguration = "Debug"
       selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
       selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
+      language = ""
       launchStyle = "0"
       useCustomWorkingDirectory = "NO"
       ignoresPersistentStateOnLaunch = "NO"

HTMLKit.playground/Pages/Sanitizing HTML.xcplaygroundpage/Contents.swift

@@ -0,0 +1,29 @@
+
+import HTMLKit
+
+HTMLSanitizingPolicy { (builder) in
+	builder
+		.allowCommonBlockElements()
+		.allowCommonInlineFormattingElements()
+		.allowElements(["p", "div"])
+		.allow(HTMLElementPolicy.identity(), onElements: ["b", "p"])
+		.allow(HTMLAttributePolicy.init(), onElements: [])
+		.disallowText(inElements: ["a"])
+}
+
+HTMLElementPolicy { (str) -> String in
+	return str
+}
+
+
+HTMLSanitizer { (builder) in
+	builder
+		.allowCommonBlockElements()
+		.allowCommonInlineFormattingElements()
+		.allowElements(["p", "div"])
+		.allow(HTMLElementPolicy.identity() , onElements: ["b", "p"])
+		.allow(HTMLAttributePolicy.init(), onElements: [])
+		.disallowText(inElements: ["a"])
+}
+
+

HTMLKit.playground/contents.xcplayground

@@ -1,10 +1,11 @@
 <?xml version="1.0" encoding="UTF-8" standalone="yes"?>
-<playground version='6.0' target-platform='macos' display-mode='rendered' last-migration='0800'>
+<playground version='6.0' target-platform='macos' display-mode='raw' last-migration='0800'>
     <pages>
         <page name='Intro'/>
         <page name='Parsing Documents'/>
         <page name='Parsing Fragments'/>
         <page name='The DOM'/>
         <page name='CSS Selectors'/>
+        <page name='Sanitizing HTML'/>
     </pages>
 </playground>

HTMLKit.podspec

@@ -1,6 +1,6 @@
 Pod::Spec.new do |s|
   s.name               = "HTMLKit"
-  s.version            = "1.0.0"
+  s.version            = "2.1.4"
   s.summary            = "HTMLKit, an Objective-C framework for your everyday HTML needs."
   s.license            = "MIT"
   s.homepage           = "https://github.com/iabudiab/HTMLKit"
@@ -16,7 +16,7 @@ Pod::Spec.new do |s|
 
   s.source_files         = "Sources", "Sources/**/*.{h,m}"
   s.private_header_files = [
-    'Sources/**/*{HTMLToken,HTMLTokens,HTMLTagToken,HTMLCharacterToken,HTMLCommentToken,HTMLDOCTYPEToken,HTMLEOFToken,HTMLTokenizer,HTMLTokenizerCharacters,HTMLTokenizerEntities,HTMLTokenizerStates,HTMLElementAdjustment,HTMLElementTypes,HTMLInputStreamReader,HTMLListOfActiveFormattingElements,HTMLNodeTraversal,HTMLParseErrorToken,HTMLParserInsertionModes,HTMLStackOfOpenElements,HTMLNode+Private,HTMLMarker,CSSCodePoints,CSSInputStream}.h'
+    'Sources/**/*{HTMLToken,HTMLTokens,HTMLTagToken,HTMLCharacterToken,HTMLCommentToken,HTMLDOCTYPEToken,HTMLEOFToken,HTMLTokenizer,HTMLTokenizerCharacters,HTMLTokenizerEntities,HTMLTokenizerStates,HTMLElementAdjustment,HTMLElementTypes,HTMLInputStreamReader,HTMLListOfActiveFormattingElements,HTMLParseErrorToken,HTMLParserInsertionModes,HTMLStackOfOpenElements,HTMLMarker,HTMLNode+Private,HTMLDocument+Private,HTMLCharacterData+Private,HTMLRange+Private,HTMLParser+Private,HTMLNodeIterator+Private,HTMLNodeTraversal,HTMLDOMUtils,CSSCodePoints,CSSInputStream}.h'
   ]
 
   s.requires_arc = true

HTMLKit.xcodeproj/xcshareddata/xcschemes/HTMLKit-iOS.xcscheme

@@ -1,6 +1,6 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <Scheme
-   LastUpgradeVersion = "0800"
+   LastUpgradeVersion = "0900"
    version = "1.3">
    <BuildAction
       parallelizeBuildables = "YES"
@@ -26,6 +26,7 @@
       buildConfiguration = "Debug"
       selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
       selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
+      language = ""
       shouldUseLaunchSchemeArgsEnv = "YES"
       codeCoverageEnabled = "YES">
       <Testables>
@@ -70,6 +71,7 @@
       buildConfiguration = "Debug"
       selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
       selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
+      language = ""
       launchStyle = "0"
       useCustomWorkingDirectory = "NO"
       ignoresPersistentStateOnLaunch = "NO"

HTMLKit.xcodeproj/xcshareddata/xcschemes/HTMLKit-macOS.xcscheme

@@ -1,6 +1,6 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <Scheme
-   LastUpgradeVersion = "0800"
+   LastUpgradeVersion = "0900"
    version = "1.3">
    <BuildAction
       parallelizeBuildables = "YES"
@@ -26,7 +26,9 @@
       buildConfiguration = "Debug"
       selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
       selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
-      shouldUseLaunchSchemeArgsEnv = "YES">
+      language = ""
+      shouldUseLaunchSchemeArgsEnv = "YES"
+      codeCoverageEnabled = "YES">
       <Testables>
          <TestableReference
             skipped = "NO">
@@ -69,6 +71,7 @@
       buildConfiguration = "Debug"
       selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
       selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
+      language = ""
       launchStyle = "0"
       useCustomWorkingDirectory = "NO"
       ignoresPersistentStateOnLaunch = "NO"

HTMLKit.xcodeproj/xcshareddata/xcschemes/HTMLKit-tvOS.xcscheme

@@ -1,6 +1,6 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <Scheme
-   LastUpgradeVersion = "0800"
+   LastUpgradeVersion = "0900"
    version = "1.3">
    <BuildAction
       parallelizeBuildables = "YES"
@@ -26,7 +26,9 @@
       buildConfiguration = "Debug"
       selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
       selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
-      shouldUseLaunchSchemeArgsEnv = "YES">
+      language = ""
+      shouldUseLaunchSchemeArgsEnv = "YES"
+      codeCoverageEnabled = "YES">
       <Testables>
          <TestableReference
             skipped = "NO">
@@ -63,6 +65,7 @@
       buildConfiguration = "Debug"
       selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
       selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
+      language = ""
       launchStyle = "0"
       useCustomWorkingDirectory = "NO"
       ignoresPersistentStateOnLaunch = "NO"

HTMLKit.xcodeproj/xcshareddata/xcschemes/HTMLKit-watchOS.xcscheme

@@ -1,6 +1,6 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <Scheme
-   LastUpgradeVersion = "0800"
+   LastUpgradeVersion = "0900"
    version = "1.3">
    <BuildAction
       parallelizeBuildables = "YES"
@@ -26,7 +26,9 @@
       buildConfiguration = "Debug"
       selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
       selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
-      shouldUseLaunchSchemeArgsEnv = "YES">
+      language = ""
+      shouldUseLaunchSchemeArgsEnv = "YES"
+      codeCoverageEnabled = "YES">
       <Testables>
       </Testables>
       <AdditionalOptions>
@@ -36,6 +38,7 @@
       buildConfiguration = "Debug"
       selectedDebuggerIdentifier = "Xcode.DebuggerFoundation.Debugger.LLDB"
       selectedLauncherIdentifier = "Xcode.DebuggerFoundation.Launcher.LLDB"
+      language = ""
       launchStyle = "0"
       useCustomWorkingDirectory = "NO"
       ignoresPersistentStateOnLaunch = "NO"

Package.swift

@@ -1,5 +1,6 @@
 import PackageDescription
 
 let package = Package(
-	name: "HTMLKit"
+	name: "HTMLKit",
+	exclude: ["Tests/Fixtures", "Tests/css-tests", "Tests/html5lib-tests"]
 )

README.md

@@ -4,13 +4,20 @@
 
 An Objective-C framework for your everyday HTML needs.
 
-[![Build Status](https://img.shields.io/travis/iabudiab/HTMLKit/master.svg?style=flat)](https://travis-ci.org/iabudiab/HTMLKit)
+[![Build Status](https://img.shields.io/travis/iabudiab/HTMLKit/develop.svg?style=flat)](https://travis-ci.org/iabudiab/HTMLKit)
+[![codecov](https://codecov.io/gh/iabudiab/HTMLKit/branch/develop/graph/badge.svg)](https://codecov.io/gh/iabudiab/HTMLKit)
 [![Carthage Compatible](https://img.shields.io/badge/Carthage-compatible-4BC51D.svg?style=flat)](https://github.com/Carthage/Carthage)
 [![CocoaPods Compatible](https://img.shields.io/cocoapods/v/HTMLKit.svg?style=flat)](https://cocoapods.org/pods/HTMLKit)
 [![CocoaDocs](https://img.shields.io/cocoapods/metrics/doc-percent/HTMLKit.svg?style=flat)](http://cocoadocs.org/docsets/HTMLKit)
 [![Platform](https://img.shields.io/cocoapods/p/HTMLKit.svg?style=flat)](http://cocoadocs.org/docsets/HTMLKit)
 [![License MIT](https://img.shields.io/badge/license-MIT-4481C7.svg?style=flat)](https://opensource.org/licenses/MIT)
 
+- [Quick Overview](#overview)
+- [Installation](#installation)
+- [Parsing](#parsing)
+- [The DOM](#the-dom)
+- [CSS3 Selectors](#css3-selectors)
+
 # Quick Overview
 
 HTMLKit is a [WHATWG specification](https://html.spec.whatwg.org/multipage/)-compliant framework for parsing and serializing HTML documents and document fragments for iOS and OSX. HTMLKit parses real-world HTML the same way modern web browsers would.
@@ -33,7 +40,7 @@ Check out the playground!
 
 # Installation
 
-### Carthage
+## Carthage
 
 [Carthage](https://github.com/Carthage/Carthage) is a decentralized dependency manager that builds your dependencies and provides you with binary frameworks.
 
@@ -56,7 +63,7 @@ Then run the following command to build the framework and drag the built `HTMLKi
 $ carthage update
 ```
 
-### CocoaPods
+## CocoaPods
 
 [CocoaPods](http://cocoapods.org) is a dependency manager for Cocoa projects.
 
@@ -69,10 +76,8 @@ $ gem install cocoapods
 To add `HTMLKit` as a dependency into your project using CocoaPods just add the following in your `Podfile`:
 
 ```ruby
-use_frameworks!
-
 target 'MyTarget' do
-  pod 'HTMLKit', '~> 0.9'
+  pod 'HTMLKit', '~> 2.1'
 end
 ```
 
@@ -82,14 +87,14 @@ Then, run the following command:
 $ pod install
 ```
 
-### Swift Package Manager
+## Swift Package Manager
 
 [Swift Package Manager](https://github.com/apple/swift-package-manager) is the package manager for the Swift programming language.
 
 Add `HTMLKit` to your `Package.swift` dependecies:
 
 ```swift
-.Package(url: "https://github.com/iabudiab/HTMLKit", majorVersion: 0, minor: 9)
+.Package(url: "https://github.com/iabudiab/HTMLKit", majorVersion: 2)
 ```
 
 Then run:
@@ -98,7 +103,7 @@ Then run:
 $ swift build
 ```
 
-### Manually
+## Manually
 
 1- Add `HTMLKit` as git submodule
 
@@ -110,9 +115,9 @@ $ git submodule add https://github.com/iabudiab/HTMLKit.git
 
 3- In the General panel of your target add `HTMLKit.framework` under the `Embedded Binaries` 
 
-# Features
+# Parsing
 
-# Parsing Documents
+## Parsing Documents
 
 Given some HTML content, you can parse it either via the `HTMLParser` or instatiate a `HTMLDocument` directly:
 
@@ -127,7 +132,7 @@ HTMLDocument *document = [parser parseDocument];
 HTMLDocument *document = [HTMLDocument documentWithString:htmlString];
 ```
 
-# Parsing Fragments
+## Parsing Fragments
 
 You can also prase HTML content as a document fragment with a specified context element:
 
@@ -150,7 +155,7 @@ nodes = [parser parseFragmentWithContextElement:bodyContext];
 
 # The DOM
 
-Here are some of the things you can do:
+The DOM tree can be manipulated in several ways, here are just a few:
 
 * Create new elements and assign attributes
 
@@ -174,7 +179,7 @@ NSArray *nodes = @[
 [body appendNodes:nodes];
 ```
 
-* Enumerate child elements and perform DOM manipulation
+* Enumerate child elements and perform DOM editing
 
 ```objective-c
 [body enumerateChildElementsUsingBlock:^(HTMLElement *element, NSUInteger idx, BOOL *stop) {
@@ -194,6 +199,12 @@ NSArray *nodes = @[
 [body.lastChild removeFromParentNode];
 ```
 
+* Manipulate the HTML directly
+
+```objective-c
+greenDiv.innerHTML = @"<ul><li>item 1<li>item 2";
+```
+
 * Navigate to child and sibling nodes
 
 ```objective-c
@@ -201,12 +212,6 @@ HTMLNode *firstChild = body.firstChild;
 HTMLNode *greenDiv = firstChild.nextSibling;
 ```
 
-* Manipulate the HTML directly
-
-```objective-c
-greenDiv.innerHTML = @"<ul><li>item 1<li>item 2";
-```
-
 * Iterate the DOM tree with custom filters
 
 ```objective-c
@@ -222,6 +227,17 @@ for (HTMLElement *element in [body nodeIteratorWithShowOptions:HTMLNodeFilterSho
 }
 ```
 
+* Create and manipulate DOM Ranges
+
+```objective-c
+HTMLDocument *document = [HTMLDocument documentWithString:@"<div><h1>HTMLKit</h1><p id='foo'>Hello there!</p></div>"];
+HTMLRange *range = [[HTMLRange alloc] initWithDocument:document];
+
+HTMLNode *paragraph = [document querySelector:@"#foo"];
+[range selectNode:paragraph];
+[range extractContents];
+```
+
 # CSS3 Selectors
 
 All CSS3 Selectors are supported except for the pseudo-elements (`::first-line`, `::first-letter`, ...etc.). You can use them the way you always have:

Sources/CSSInputStream.m

@@ -24,14 +24,14 @@
 
 - (NSString *)consumeIdentifier
 {
-	CFMutableStringRef value = CFStringCreateMutable(kCFAllocatorDefault, 0);
-
 	if (!isValidIdentifierStart([self inputCharacterPointAtOffset:0],
 								[self inputCharacterPointAtOffset:1],
 								[self inputCharacterPointAtOffset:2])) {
 		return nil;
 	}
 
+	CFMutableStringRef value = CFStringCreateMutable(kCFAllocatorDefault, 0);
+
 	while (YES) {
 		UTF32Char codePoint = [self consumeNextInputCharacter];
 		if (codePoint == EOF) {
@@ -47,7 +47,12 @@
 		}
 	}
 
-	return (__bridge NSString *)(CFStringGetLength(value) > 0 ? value : nil);
+	if (CFStringGetLength(value) > 0) {
+		return (__bridge_transfer NSString *)value;
+	}
+
+	CFRelease(value);
+	return nil;
 }
 
 - (NSString *)consumeStringWithEndingCodePoint:(UTF32Char)endingCodePoint
@@ -85,7 +90,12 @@
 		}
 	}
 
-	return (__bridge NSString *)(CFStringGetLength(value) > 0 ? value : nil);
+	if (CFStringGetLength(value) > 0) {
+		return (__bridge_transfer NSString *)value;
+	}
+
+	CFRelease(value);
+	return nil;
 }
 
 - (UTF32Char)consumeEscapedCodePoint
@@ -105,7 +115,7 @@
 			[self consumeNextInputCharacter];
 		}
 
-		NSScanner *scanner = [NSScanner scannerWithString:(__bridge NSString *)(hexString)];
+		NSScanner *scanner = [NSScanner scannerWithString:(__bridge_transfer NSString *)(hexString)];
 		unsigned int number;
 		[scanner scanHexInt:&number];
 

Sources/CSSNthExpressionSelector.m

@@ -38,7 +38,7 @@ NSString * _Nonnull NSStringFromNthExpression(CSSNthExpression expression)
 
 #pragma mark - Implementation
 
-NSInteger computeIndex(NSEnumerator *enumerator, HTMLElement *element)
+NS_INLINE NSInteger computeIndex(NSEnumerator *enumerator, HTMLElement *element)
 {
 	NSInteger index = 0;
 	for (HTMLNode *node in enumerator) {

Sources/CSSStructuralPseudoSelectors.m

@@ -276,7 +276,7 @@ CSSSelector * ltSelector(NSInteger index)
 	return namedBlockSelector(name, ^BOOL(HTMLElement * _Nonnull element) {
 		NSUInteger elementIndex = [element.parentElement indexOfChildNode:element];
 
-		if (index > 0) {
+		if (index >= 0) {
 			return elementIndex < index;
 		} else {
 			return elementIndex < element.parentElement.childNodesCount - index - 1;
@@ -290,7 +290,7 @@ CSSSelector * gtSelector(NSInteger index)
 	return namedBlockSelector(name, ^BOOL(HTMLElement * _Nonnull element) {
 		NSUInteger elementIndex = [element.parentElement indexOfChildNode:element];
 
-		if (index > 0) {
+		if (index >= 0) {
 			return elementIndex > index;
 		} else {
 			return elementIndex > element.parentElement.childNodesCount - index - 1;
@@ -304,7 +304,7 @@ CSSSelector * eqSelector(NSInteger index)
 	return namedBlockSelector(name, ^BOOL(HTMLElement * _Nonnull element) {
 		NSUInteger elementIndex = [element.parentElement indexOfChildNode:element];
 
-		if (index > 0) {
+		if (index >= 0) {
 			return elementIndex == index;
 		} else {
 			return elementIndex == element.parentElement.childNodesCount - index - 1;

Sources/HTMLAttributePolicy.h

@@ -0,0 +1,25 @@
+//
+//  HTMLAttributePolicy.h
+//  HTMLKit
+//
+//  Created by Iska on 28.05.18.
+//  Copyright © 2018 BrainCookie. All rights reserved.
+//
+
+#import <Foundation/Foundation.h>
+
+NS_ASSUME_NONNULL_BEGIN
+
+typedef NSString * _Nullable (^ SanitizeAttrbiuteValueBlock) (NSString * _Nullable value, NSString * key);
+
+@interface HTMLAttributePolicy : NSObject
+
++ (instancetype)identity;
++ (instancetype)rejectAll;
++ (instancetype)policyWithBlock:(SanitizeAttrbiuteValueBlock)block;
+
+- (NSString *)sanitizeValue:(NSString *)value forKey:(NSString *)key;
+
+@end
+
+NS_ASSUME_NONNULL_END

Sources/HTMLAttributePolicy.m

@@ -0,0 +1,56 @@
+//
+//  HTMLAttributePolicy.m
+//  HTMLKit
+//
+//  Created by Iska on 28.05.18.
+//  Copyright © 2018 BrainCookie. All rights reserved.
+//
+
+#import "HTMLAttributePolicy.h"
+
+@interface HTMLAttributePolicy()
+{
+	SanitizeAttrbiuteValueBlock _block;
+}
+@end
+
+@implementation HTMLAttributePolicy
+
++ (instancetype)identity
+{
+	return [self policyWithBlock:^NSString * _Nullable (NSString * _Nullable value, NSString * _Nonnull key) {
+		return value;
+	}];
+}
+
++ (instancetype)rejectAll
+{
+	return [self policyWithBlock:^NSString * _Nullable (NSString * _Nullable value, NSString * _Nonnull key) {
+		return nil;
+	}];
+}
+
++ (instancetype)policyWithBlock:(SanitizeAttrbiuteValueBlock)block
+{
+	HTMLAttributePolicy *policy = [[HTMLAttributePolicy alloc] initWithBlock:block];
+	return policy;
+}
+
+- (instancetype)initWithBlock:(SanitizeAttrbiuteValueBlock)block
+{
+	self = [super init];
+	if (self) {
+		_block = block;
+	}
+	return self;
+}
+
+- (NSString *)sanitizeValue:(NSString *)value forKey:(NSString *)key
+{
+	if (_block) {
+		return _block(value, key);
+	}
+	return value;
+}
+
+@end

Sources/HTMLCharacterData.m

@@ -0,0 +1,114 @@
+//
+//  HTMLCharacterData.m
+//  HTMLKit
+//
+//  Created by Iska on 26/11/16.
+//  Copyright © 2016 BrainCookie. All rights reserved.
+//
+
+#import "HTMLCharacterData.h"
+#import "HTMLNode+Private.h"
+#import "HTMLDocument+Private.h"
+#import "HTMLKitDOMExceptions.h"
+
+@interface HTMLCharacterData ()
+{
+	NSMutableString *_data;
+}
+@end
+
+@implementation HTMLCharacterData
+@synthesize data = _data;
+
+- (instancetype)initWithName:(NSString *)name type:(HTMLNodeType)type data:(NSString *)data
+{
+	self = [super initWithName:name type:type];
+	if (self) {
+		if (data) {
+			_data = [[NSMutableString alloc] initWithString:data];
+		}
+	}
+	return self;
+}
+
+- (NSString *)data
+{
+	if (_data == nil) {
+		_data = [[NSMutableString alloc] initWithString:@""];
+	}
+
+	return _data;
+}
+
+- (NSString *)textContent
+{
+	return [self.data copy];
+}
+
+- (void)setTextContent:(NSString *)textContent
+{
+	[self setData:textContent];
+}
+
+- (NSUInteger)length
+{
+	return self.data.length;
+}
+
+#pragma mark - Data
+
+NS_INLINE void CheckValidOffset(HTMLCharacterData *node, NSUInteger offset, NSString *cmd)
+{
+	if (offset > node.length) {
+		[NSException raise:HTMLKitIndexSizeError
+					format:@"%@: Index Size Error, invalid index %lu for character data node %@.",
+		 cmd, (unsigned long)offset, node];
+	}
+}
+
+- (void)setData:(NSString *)data
+{
+	[self replaceDataInRange:NSMakeRange(0, self.length) withData:data];
+}
+
+- (void)appendData:(NSString *)data
+{
+	[self replaceDataInRange:NSMakeRange(self.length, 0) withData:data];
+}
+
+- (void)insertData:(NSString *)data atOffset:(NSUInteger)offset
+{
+	[self replaceDataInRange:NSMakeRange(offset, 0) withData:data];
+}
+
+- (void)deleteDataInRange:(NSRange)range
+{
+	[self replaceDataInRange:range withData:@""];
+}
+
+- (void)replaceDataInRange:(NSRange)range withData:(NSString *)data
+{
+	CheckValidOffset(self, range.location, NSStringFromSelector(_cmd));
+
+	range.length = MIN(range.length, self.length - range.location);
+
+	[(NSMutableString *)self.data replaceCharactersInRange:range withString:data];
+	[self.ownerDocument didRemoveCharacterDataInNode:self atOffset:range.location withLength:range.length];
+	[self.ownerDocument didAddCharacterDataToNode:self atOffset:range.location withLength:data.length];
+}
+
+- (NSString *)substringDataWithRange:(NSRange)range
+{
+	return [_data substringWithRange:range];
+}
+
+#pragma mark - NSCopying
+
+- (id)copyWithZone:(NSZone *)zone
+{
+	HTMLCharacterData *copy = [super copyWithZone:zone];
+	copy->_data = [_data mutableCopy];
+	return copy;
+}
+
+@end

Sources/HTMLComment.m

@@ -7,7 +7,7 @@
 //
 
 #import "HTMLComment.h"
-#import "HTMLNode+Private.h"
+#import "HTMLCharacterData+Private.h"
 
 @implementation HTMLComment
 
@@ -18,30 +18,7 @@
 
 - (instancetype)initWithData:(NSString *)data
 {
-	self = [super initWithName:@"#comment" type:HTMLNodeComment];
-	if (self) {
-		self.data = data ?: @"";
-	}
-	return self;
-}
-
-- (NSString *)textContent
-{
-	return self.data;
-}
-
-- (void)setTextContent:(NSString *)textContent
-{
-	self.data = textContent ?: @"";
-}
-
-#pragma mark - NSCopying
-
-- (id)copyWithZone:(NSZone *)zone
-{
-	HTMLComment *copy = [super copyWithZone:zone];
-	copy.data = self.data;
-	return copy;
+	return [super initWithName:@"#comment" type:HTMLNodeComment data:data];
 }
 
 #pragma mark - Serialization

Sources/HTMLDOMUtils.m

@@ -0,0 +1,32 @@
+//
+//  HTMLDOMUtils.m
+//  HTMLKit
+//
+//  Created by Iska on 03/12/16.
+//  Copyright © 2016 BrainCookie. All rights reserved.
+//
+
+#import "HTMLDOMUtils.h"
+#import "HTMLNode.h"
+
+extern HTMLNode * GetCommonAncestorContainer(HTMLNode *nodeA, HTMLNode *nodeB)
+{
+	for (HTMLNode *parentA = nodeA; parentA != nil; parentA = parentA.parentNode) {
+		for (HTMLNode *parentB = nodeB; parentB != nil; parentB = parentB.parentNode) {
+			if (parentA == parentB) {
+				return parentA;
+			}
+		}
+	}
+
+	return nil;
+}
+
+extern NSArray<HTMLNode *> * GetAncestorNodes(HTMLNode *node)
+{
+	NSMutableArray *ancestors = [NSMutableArray array];
+	for (HTMLNode *it = node; it; it = it.parentNode) {
+		[ancestors addObject:it];
+	}
+	return ancestors;
+}

Sources/HTMLDocument.m

@@ -9,19 +9,19 @@
 #import "HTMLDocument.h"
 #import "HTMLParser.h"
 #import "HTMLNodeIterator.h"
+#import "HTMLRange.h"
+#import "HTMLCharacterData.h"
+#import "HTMLText.h"
 #import "HTMLKitDOMExceptions.h"
 #import "HTMLNode+Private.h"
-
-@interface HTMLNodeIterator (Private)
-- (void)runRemovingStepsForNode:(HTMLNode *)oldNode
-				  withOldParent:(HTMLNode *)oldParent
-		  andOldPreviousSibling:(HTMLNode *)oldPreviousSibling;
-@end
+#import "HTMLNodeIterator+Private.h"
+#import "HTMLRange+Private.h"
 
 @interface HTMLDocument ()
 {
 	HTMLDocument *_inertTemplateDocument;
-	NSMutableArray *_nodeIterators;
+	NSHashTable *_nodeIterators;
+	NSHashTable *_ranges;
 }
 @property (nonatomic, assign) HTMLDocumentReadyState readyState;
 @end
@@ -41,7 +41,8 @@
 	self = [super initWithName:@"#document" type:HTMLNodeDocument];
 	if (self) {
 		_readyState = HTMLDocumentLoading;
-		_nodeIterators = [NSMutableArray new];
+		_nodeIterators = [[NSHashTable alloc] initWithOptions:NSHashTableWeakMemory capacity:10];
+		_ranges = [[NSHashTable alloc] initWithOptions:NSHashTableWeakMemory capacity:10];
 	}
 	return self;
 }
@@ -143,14 +144,62 @@
 	[_nodeIterators removeObject:iterator];
 }
 
+#pragma mark - Ranges
+
+- (void)attachRange:(HTMLRange *)range
+{
+	[_ranges addObject:range];
+}
+
+- (void)detachRange:(HTMLRange *)range
+{
+	[_ranges removeObject:range];
+}
+
+- (void)didRemoveCharacterDataInNode:(HTMLCharacterData *)node atOffset:(NSUInteger)offset withLength:(NSUInteger)length
+{
+	for (HTMLRange *range in _ranges) {
+		[range didRemoveCharacterDataInNode:node atOffset:offset withLength:length];
+	}
+}
+
+- (void)didAddCharacterDataToNode:(HTMLCharacterData *)node atOffset:(NSUInteger)offset withLength:(NSUInteger)length
+{
+	for (HTMLRange *range in _ranges) {
+		[range didAddCharacterDataToNode:node atOffset:offset withLength:length];
+	}
+}
+
+- (void)didInsertNewTextNode:(HTMLText *)newNode intoParent:(HTMLNode *)parent afterSplittingTextNode:(HTMLText *)node atOffset:(NSUInteger)offset
+{
+	for (HTMLRange *range in _ranges) {
+		[range didInsertNewTextNode:newNode intoParent:parent afterSplittingTextNode:node atOffset:offset];
+	}
+}
+
+- (void)clampRangesAfterSplittingTextNode:(HTMLText *)node atOffset:(NSUInteger)offset
+{
+	for (HTMLRange *range in _ranges) {
+		[range clampRangesAfterSplittingTextNode:node atOffset:offset];
+	}
+}
+
+#pragma mark - Mutation Callback
+
 - (void)runRemovingStepsForNode:(HTMLNode *)oldNode
 				  withOldParent:(HTMLNode *)oldParent
 		  andOldPreviousSibling:(HTMLNode *)oldPreviousSibling
 {
+	for (HTMLRange *range in _ranges) {
+		[range runRemovingStepsForNode:oldNode
+						 withOldParent:oldParent
+				 andOldPreviousSibling:oldPreviousSibling];
+	}
+
 	for (HTMLNodeIterator *iterator in _nodeIterators) {
 		[iterator runRemovingStepsForNode:oldNode
-							 withOldParent:oldParent
-					 andOldPreviousSibling:oldPreviousSibling];
+							withOldParent:oldParent
+					andOldPreviousSibling:oldPreviousSibling];
 	}
 }
 

Sources/HTMLDocumentType.m

@@ -129,13 +129,18 @@ NS_INLINE BOOL nilOrEqual(id first, id second) {
 	return HTMLQuirksModeNoQuirks;
 }
 
+- (NSUInteger)length
+{
+	return 0;
+}
+
 #pragma mark - NSCopying
 
 - (id)copyWithZone:(NSZone *)zone
 {
 	HTMLDocumentType *copy = [super copyWithZone:zone];
-	copy->_publicIdentifier = self.publicIdentifier;
-	copy->_systemIdentifier = self.systemIdentifier;
+	copy->_publicIdentifier = [_publicIdentifier copy];
+	copy->_systemIdentifier = [_systemIdentifier copy];
 	return copy;
 }
 

Sources/HTMLElement.m

@@ -32,7 +32,7 @@
 
 - (instancetype)initWithTagName:(NSString *)tagName
 {
-	return [self initWithTagName:tagName attributes:@{}];
+	return [self initWithTagName:tagName attributes:nil];
 }
 
 - (instancetype)initWithTagName:(NSString *)tagName attributes:(NSDictionary *)attributes
@@ -45,8 +45,9 @@
 	self = [super initWithName:tagName type:HTMLNodeElement];
 	if (self) {
 		_tagName = [tagName copy];
-		_attributes = [HTMLOrderedDictionary new];
+		_attributes = nil;
 		if (attributes != nil) {
+			_attributes = [HTMLOrderedDictionary new];
 			[_attributes addEntriesFromDictionary:attributes];
 		}
 		_htmlNamespace = htmlNamespace;
@@ -56,24 +57,33 @@
 
 #pragma mark - Special Attributes
 
+- (NSMutableDictionary<NSString *,NSString *> *)attributes
+{
+	if (_attributes == nil) {
+		_attributes = [HTMLOrderedDictionary new];
+	}
+
+	return _attributes;
+}
+
 - (NSString *)elementId
 {
-	return _attributes[@"id"] ?: @"";
+	return self.attributes[@"id"] ?: @"";
 }
 
 - (void)setElementId:(NSString *)elementId
 {
-	_attributes[@"id"] = elementId;
+	self.attributes[@"id"] = elementId;
 }
 
 - (NSString *)className
 {
-	return _attributes[@"class"] ?: @"";
+	return self.attributes[@"class"] ?: @"";
 }
 
 - (void)setClassName:(NSString *)className
 {
-	_attributes[@"class"] = className;
+	self.attributes[@"class"] = className;
 }
 
 - (HTMLDOMTokenList *)classList
@@ -85,22 +95,22 @@
 
 - (BOOL)hasAttribute:(NSString *)name
 {
-	return _attributes[name] != nil;
+	return self.attributes[name] != nil;
 }
 
 - (NSString *)objectForKeyedSubscript:(NSString *)name;
 {
-	return _attributes[name];
+	return self.attributes[name];
 }
 
 - (void)setObject:(NSString *)value forKeyedSubscript:(NSString *)attribute
 {
-	_attributes[attribute] = value;
+	self.attributes[attribute] = value;
 }
 
 - (void)removeAttribute:(NSString *)name
 {
-	[_attributes removeObjectForKey:name];
+	[self.attributes removeObjectForKey:name];
 }
 
 - (NSString *)textContent
@@ -134,7 +144,7 @@
 {
 	HTMLElement *copy = [super copyWithZone:zone];
 	copy->_tagName = [_tagName copy];
-	copy->_attributes = [_attributes copy];
+	copy->_attributes = [_attributes mutableCopy];
 	copy->_htmlNamespace = _htmlNamespace;
 	return copy;
 }
@@ -147,11 +157,10 @@
 
 	[result appendFormat:@"<%@", self.tagName];
 	[self.attributes enumerateKeysAndObjectsUsingBlock:^(NSString *key, NSString *value, BOOL *stop) {
-		NSRange range = NSMakeRange(0, value.length);
 		NSMutableString *escaped = [value mutableCopy];
-		[escaped replaceOccurrencesOfString:@"&" withString:@"&amp;" options:0 range:range];
-		[escaped replaceOccurrencesOfString:@"\00A0" withString:@"&nbsp;" options:0 range:range];
-		[escaped replaceOccurrencesOfString:@"\"" withString:@"&quot;" options:0 range:range];
+		[escaped replaceOccurrencesOfString:@"&" withString:@"&amp;" options:0 range:NSMakeRange(0, escaped.length)];
+		[escaped replaceOccurrencesOfString:@"0x00A0" withString:@"&nbsp;" options:0 range:NSMakeRange(0, escaped.length)];
+		[escaped replaceOccurrencesOfString:@"\"" withString:@"&quot;" options:0 range:NSMakeRange(0, escaped.length)];
 
 		[result appendFormat:@" %@=\"%@\"", key, escaped];
 	}];

Sources/HTMLElementPolicy.h

@@ -0,0 +1,31 @@
+//
+//  HTMLElementPolicy.h
+//  HTMLKit
+//
+//  Created by Iska on 26.05.18.
+//  Copyright © 2018 BrainCookie. All rights reserved.
+//
+
+#import <Foundation/Foundation.h>
+#import "HTMLSanitizingPolicy.h"
+#import "HTMLOrderedDictionary.h"
+
+NS_ASSUME_NONNULL_BEGIN
+
+typedef NSString * _Nullable (^ SanitizeElementNameBlock) (NSString *);
+typedef HTMLOrderedDictionary * _Nullable (^ SanitizeElementAttributesBlock) (HTMLOrderedDictionary * _Nullable);
+
+@interface HTMLElementPolicy : NSObject
+
++ (instancetype)identity;
++ (instancetype)rejectAll;
++ (instancetype)policyWithNameBlock:(SanitizeElementNameBlock)nameBlock;
++ (instancetype)policyWithNameBlock:(SanitizeElementNameBlock)nameBlock
+					attributesBlock:(nullable SanitizeElementAttributesBlock)attributesBlock;
+
+- (NSString *)sanitizeName:(NSString *)name;
+- (HTMLOrderedDictionary *)sanitzeAttributes:(HTMLOrderedDictionary *)attributes;
+
+@end
+
+NS_ASSUME_NONNULL_END

Sources/HTMLElementPolicy.m

@@ -0,0 +1,73 @@
+//
+//  HTMLElementPolicy.m
+//  HTMLKit
+//
+//  Created by Iska on 26.05.18.
+//  Copyright © 2018 BrainCookie. All rights reserved.
+//
+
+#import "HTMLElementPolicy.h"
+
+@interface HTMLElementPolicy()
+{
+	SanitizeElementNameBlock _nameBlock;
+	SanitizeElementAttributesBlock _attributesBlock;
+}
+@end
+
+@implementation HTMLElementPolicy
+
++ (instancetype)identity
+{
+	return [self policyWithNameBlock:^NSString * _Nullable (NSString * _Nonnull name) {
+		return name;
+	}];
+}
+
++ (instancetype)rejectAll
+{
+	return [self policyWithNameBlock:^NSString *_Nullable (NSString * _Nonnull name) {
+		return nil;
+	}];
+}
+
++ (instancetype)policyWithNameBlock:(NSString * _Nonnull (^)(NSString * _Nonnull))nameBlock
+{
+	return [self policyWithNameBlock:nameBlock attributesBlock:nil];
+}
+
++ (instancetype)policyWithNameBlock:(SanitizeElementNameBlock)nameBlock
+					attributesBlock:(SanitizeElementAttributesBlock)attributesBlock
+{
+	HTMLElementPolicy *policy = [[HTMLElementPolicy alloc] initWithNameBlock:nameBlock attributesBlock:attributesBlock];
+	return policy;
+}
+
+- (instancetype)initWithNameBlock:(SanitizeElementNameBlock)nameBlock
+				  attributesBlock:(SanitizeElementAttributesBlock)attributesBlock
+{
+	self = [super init];
+	if (self) {
+		_nameBlock = nameBlock;
+		_attributesBlock = attributesBlock;
+	}
+	return self;
+}
+
+- (NSString *)sanitizeName:(NSString *)name
+{
+	if (_nameBlock) {
+		return _nameBlock(name);
+	}
+	return name;
+}
+
+- (HTMLOrderedDictionary *)sanitzeAttributes:(HTMLOrderedDictionary *)attributes
+{
+	if (_attributesBlock) {
+		return _attributesBlock(attributes);
+	}
+	return attributes;
+}
+
+@end

Sources/HTMLInputStreamReader.m

@@ -48,10 +48,10 @@
 
 #pragma mark - Errors
 
-- (void)emitParseError:(NSString *)reason
+- (void)emitParseError:(NSString *)code details:(NSString *)details
 {
 	if (self.errorCallback) {
-		self.errorCallback(reason);
+		self.errorCallback(code, details);
 	}
 }
 
@@ -82,16 +82,16 @@
 		return LINE_FEED;
 	}
 	if (CFStringIsSurrogateLowCharacter(nextInputCharacter)) {
-		NSString *reason = [NSString stringWithFormat:@"Non-Unicode character found (an isolated low surrogate: 0x%X)", (unsigned int)nextInputCharacter];
-		[self emitParseError:reason];
+		NSString *details = [NSString stringWithFormat:@"Non-Unicode character found (an isolated low surrogate: 0x%X)", (unsigned int)nextInputCharacter];
+		[self emitParseError:@"surrogate-in-input-stream" details:details];
 		return nextInputCharacter;
 	}
 
 	if (CFStringIsSurrogateHighCharacter(nextInputCharacter)) {
 		UniChar surrogateLow = CFStringGetCharacterFromInlineBuffer(&_buffer, _location + 1);
 		if (CFStringIsSurrogateLowCharacter(surrogateLow) == NO) {
-			NSString *reason = [NSString stringWithFormat:@"Non-Unicode character found (an isolated high surrogate: 0x%X)", (unsigned int)nextInputCharacter];
-			[self emitParseError:reason];
+			NSString *details = [NSString stringWithFormat:@"Non-Unicode character found (an isolated high surrogate: 0x%X)", (unsigned int)nextInputCharacter];
+			[self emitParseError:@"surrogate-in-input-stream" details:details];
 			return nextInputCharacter;
 		}
 
@@ -99,9 +99,14 @@
 		nextInputCharacter = CFStringGetLongCharacterForSurrogatePair(nextInputCharacter, surrogateLow);
 	}
 
-	if (isControlOrUndefinedCharacter(nextInputCharacter)) {
-		NSString *reason = [NSString stringWithFormat:@"A control/undefined character found: (0x%X)", (unsigned int)nextInputCharacter];
-		[self emitParseError:reason];
+	if (isControlCharacter(nextInputCharacter)) {
+		NSString *details = [NSString stringWithFormat:@"A control character found: (0x%X)", (unsigned int)nextInputCharacter];
+		[self emitParseError:@"control-character-in-input-stream" details:details];
+	}
+
+	if (isNoncharacter(nextInputCharacter)) {
+		NSString *details = [NSString stringWithFormat:@"A noncharacter found: (0x%X)", (unsigned int)nextInputCharacter];
+		[self emitParseError:@"noncharacter-in-input-stream" details:details];
 	}
 
 	return nextInputCharacter;

Sources/HTMLKit-Info.plist

@@ -17,7 +17,7 @@
 	<key>CFBundlePackageType</key>
 	<string>FMWK</string>
 	<key>CFBundleShortVersionString</key>
-	<string>1.0.0</string>
+	<string>2.1.4</string>
 	<key>CFBundleSignature</key>
 	<string>????</string>
 	<key>CFBundleVersion</key>

Sources/HTMLKitDOMExceptions.m

@@ -14,3 +14,7 @@ NSString * const HTMLKitNotFoundError = @"NotFoundError";
 NSString * const HTMLKitNotSupportedError = @"NotSupportedError";
 NSString * const HTMLKitSyntaxError = @"SyntaxError";
 NSString * const HTMLKitInvalidCharacterError = @"InvalidCharacterError";
+NSString * const HTMLKitInvalidNodeTypeError = @"InvalidNodeTypeError";
+NSString * const HTMLKitIndexSizeError = @"IndexSizeError";
+NSString * const HTMLKitWrongDocumentError = @"WrongDocumentError";
+NSString * const HTMLKitInvalidStateError = @"InvalidStateError";

Sources/HTMLListOfActiveFormattingElements.m

@@ -46,7 +46,7 @@
 - (void)addElement:(HTMLElement *)element
 {
 	NSUInteger existing = 0;
-	for (HTMLElement *node in _list.reverseObjectEnumerator.allObjects) {
+	for (HTMLElement *node in _list.reverseObjectEnumerator) {
 		if ([node isEqual:[HTMLMarker marker]]) {
 			break;
 		}

Sources/HTMLNode.m

@@ -16,12 +16,12 @@
 #import "HTMLKitDOMExceptions.h"
 #import "HTMLNodeFilter.h"
 #import "CSSSelector.h"
+#import "HTMLDocument+Private.h"
+#import "HTMLDOMUtils.h"
 
-@interface HTMLDocument (Private)
-- (void)runRemovingStepsForNode:(HTMLNode *)oldNode
-				  withOldParent:(HTMLNode *)oldParent
-		  andOldPreviousSibling:(HTMLNode *)oldPreviousSibling;
-@end
+NSString * const ValidationNodePreInsertion = @"-ensurePreInsertionValidityOfNode:beforeChildNode:";
+NSString * const ValidationNodeReplacement = @"-ensureReplacementValidityOfChildNode:withNode:";
+NSString * const RemoveChildNode = @"-removeChildNode:";
 
 @interface HTMLNode ()
 {
@@ -40,13 +40,22 @@
 	if (self) {
 		_name = name;
 		_nodeType = type;
-		_childNodes = [NSMutableOrderedSet new];
+		_childNodes = nil;
 	}
 	return self;
 }
 
 #pragma mark - Properties
 
+- (NSOrderedSet<HTMLNode *> *)childNodes
+{
+	if (_childNodes == nil) {
+		_childNodes = [NSMutableOrderedSet new];
+	}
+
+	return _childNodes;
+}
+
 - (HTMLDocument *)ownerDocument
 {
 	if (_nodeType == HTMLNodeDocument) {
@@ -59,7 +68,14 @@
 - (void)setOwnerDocument:(HTMLDocument *)ownerDocument
 {
 	_ownerDocument = ownerDocument;
-	[self.childNodes.array makeObjectsPerformSelector:@selector(setOwnerDocument:) withObject:ownerDocument];
+	for (HTMLNode *child in _childNodes) {
+		[child setOwnerDocument:ownerDocument];
+	}
+}
+
+- (HTMLNode *)rootNode
+{
+	return _parentNode == nil ? self : _parentNode.rootNode;
 }
 
 - (void)setParentNode:(HTMLNode *)parentNode
@@ -74,12 +90,12 @@
 
 - (HTMLNode *)firstChild
 {
-	return self.childNodes.firstObject;
+	return _childNodes.firstObject;
 }
 
 - (HTMLNode *)lastChild
 {
-	return self.childNodes.lastObject;
+	return _childNodes.lastObject;
 }
 
 - (HTMLNode *)previousSibling
@@ -111,18 +127,28 @@
 
 - (HTMLElement *)nextSiblingElement
 {
-	HTMLNode *node = self.previousSibling;
+	HTMLNode *node = self.nextSibling;
 	while (node && node.nodeType != HTMLNodeElement) {
 		node = node.nextSibling;
 	}
 	return node.asElement;
 }
 
+- (NSUInteger)index
+{
+	return [_parentNode indexOfChildNode:self];
+}
+
 - (NSString *)textContent
 {
 	return nil;
 }
 
+- (NSUInteger)length
+{
+	return self.childNodesCount;
+}
+
 #pragma mark - Cast
 
 - (HTMLElement *)asElement
@@ -134,12 +160,16 @@
 
 - (BOOL)hasChildNodes
 {
-	return self.childNodes.count > 0;
+	return _childNodes.count > 0;
 }
 
 - (BOOL)hasChildNodeOfType:(HTMLNodeType)type
 {
-	NSUInteger index = [self.childNodes indexOfObjectPassingTest:^BOOL(id obj, NSUInteger idx, BOOL *stop) {
+	if (_childNodes == nil) {
+		return NO;
+	}
+
+	NSUInteger index = [_childNodes indexOfObjectPassingTest:^BOOL(id obj, NSUInteger idx, BOOL *stop) {
 		if ([(HTMLNode *)obj nodeType] == type) {
 			*stop = YES;
 			return YES;
@@ -152,30 +182,35 @@
 
 - (NSUInteger)childNodesCount
 {
-	return self.childNodes.count;
+	return _childNodes.count;
+}
+
+- (BOOL)isEmpty
+{
+	return self.length == 0;
 }
 
 - (HTMLNode *)childNodeAtIndex:(NSUInteger)index
 {
-	return [self.childNodes objectAtIndex:index];
+	return [_childNodes objectAtIndex:index];
 }
 
 - (NSUInteger)childElementsCount
 {
-	return [self.childNodes indexesOfObjectsPassingTest:^BOOL(HTMLNode *  _Nonnull node, NSUInteger idx, BOOL * _Nonnull stop) {
+	return [_childNodes indexesOfObjectsPassingTest:^BOOL(HTMLNode *  _Nonnull node, NSUInteger idx, BOOL * _Nonnull stop) {
 		return node.nodeType == HTMLNodeElement;
 	}].count;
 }
 
 - (NSUInteger)indexOfChildNode:(HTMLNode *)node
 {
-	return [self.childNodes indexOfObject:node];
+	return [_childNodes indexOfObject:node];
 }
 
 - (HTMLElement *)childElementAtIndex:(NSUInteger)index
 {
 	NSUInteger counter = 0;
-	for (HTMLNode *node in self.childNodes) {
+	for (HTMLNode *node in _childNodes) {
 		if (node.nodeType == HTMLNodeElement) {
 			if (counter == index) {
 				return node.asElement;
@@ -189,7 +224,7 @@
 - (NSUInteger)indexOfChildElement:(HTMLElement *)element
 {
 	NSUInteger counter = 0;
-	for (HTMLNode *node in self.childNodes) {
+	for (HTMLNode *node in _childNodes) {
 		if (node.nodeType == HTMLNodeElement) {
 			if (node == element) {
 				return counter;
@@ -246,7 +281,9 @@
 		[node removeAllChildNodes];
 	}
 
-	[nodes makeObjectsPerformSelector:@selector(setParentNode:) withObject:self];
+	for (HTMLNode *node in nodes) {
+		[node setParentNode:self];
+	}
 
 	return node;
 }
@@ -274,7 +311,7 @@
 
 - (void)removeFromParentNode
 {
-	[self.parentNode removeChildNode:self];
+	[_parentNode removeChildNode:self];
 }
 
 - (HTMLNode *)removeChildNode:(HTMLNode *)child
@@ -282,7 +319,7 @@
 	if (child.parentNode != self) {
 		[NSException raise:HTMLKitNotFoundError
 					format:@"%@: Not Fount Error, removing non-child node %@. The object can not be found here.",
-		 NSStringFromSelector(_cmd), child];
+		 RemoveChildNode, child];
 	}
 
 	HTMLNode *oldNode = child;
@@ -306,16 +343,18 @@
 
 - (void)reparentChildNodesIntoNode:(HTMLNode *)node
 {
-	for (HTMLNode *child in self.childNodes.array) {
+	for (HTMLNode *child in _childNodes) {
 		[node appendNode:child];
 	}
-	[(NSMutableOrderedSet *)self.childNodes removeAllObjects];
+	[(NSMutableOrderedSet *)_childNodes removeAllObjects];
 }
 
 - (void)removeAllChildNodes
 {
-	[self.childNodes.array makeObjectsPerformSelector:@selector(setParentNode:) withObject:nil];
-	[(NSMutableOrderedSet *)self.childNodes removeAllObjects];
+	for (HTMLNode *child in _childNodes) {
+		[child setParentNode:nil];
+	}
+	[(NSMutableOrderedSet *)_childNodes removeAllObjects];
 }
 
 - (HTMLDocumentPosition)compareDocumentPositionWithNode:(HTMLNode *)otherNode
@@ -328,41 +367,43 @@
 		return HTMLDocumentPositionEquivalent;
 	}
 
-	NSArray * (^ ancestorNodes) (HTMLNode *) = ^ NSArray * (HTMLNode *node) {
-		NSMutableArray *ancestors = [NSMutableArray array];
-		for (HTMLNode *node = self; node; node = node.parentNode) {
-			[ancestors addObject:node];
-		}
-		return ancestors;
-	};
-
-	NSArray *ancestors1 = ancestorNodes(self);
-	NSArray *ancestors2 = ancestorNodes(otherNode);
-
-	if (ancestors1.lastObject != ancestors2.lastObject) {
-		return HTMLDocumentPositionDisconnected |
-		HTMLDocumentPositionImplementationSpecific |
-		HTMLDocumentPositionFollowing;
+	if (self.ownerDocument != otherNode.ownerDocument) {
+		return (HTMLDocumentPositionDisconnected | HTMLDocumentPositionImplementationSpecific |
+				self.hash < otherNode.hash ? HTMLDocumentPositionPreceding : HTMLDocumentPositionFollowing);
 	}
 
-	for (NSUInteger i = MIN(ancestors1.count - 1, ancestors2.count - 1); i; --i) {
-		HTMLNode *child1 = ancestors1[i];
-		HTMLNode *child2 = ancestors2[i];
+	NSArray *ancestors1 = GetAncestorNodes(self);
+	NSArray *ancestors2 = GetAncestorNodes(otherNode);
+
+	if (ancestors1.lastObject != ancestors2.lastObject) {
+		return (HTMLDocumentPositionDisconnected | HTMLDocumentPositionImplementationSpecific |
+				self.hash < otherNode.hash ? HTMLDocumentPositionPreceding : HTMLDocumentPositionFollowing);
+	}
+
+	NSUInteger index1 = ancestors1.count;
+	NSUInteger index2 = ancestors2.count;
+
+	for (NSUInteger i = MIN(index1, index2); i; --i) {
+		index1 -= 1;
+		index2 -= 1;
+
+		HTMLNode *child1 = ancestors1[index1];
+		HTMLNode *child2 = ancestors2[index2];
 
 		if (child1 != child2) {
 			for (HTMLNode *sibling = child1.nextSibling; sibling; sibling = sibling.nextSibling) {
 				if (sibling == child2) {
-					return HTMLDocumentPositionFollowing;
+					return HTMLDocumentPositionPreceding;
 				}
 			}
-			return HTMLDocumentPositionPreceding;
+			return HTMLDocumentPositionFollowing;
 		}
 	}
 
 	if (ancestors1.count < ancestors2.count) {
-		return HTMLDocumentPositionContainedBy | HTMLDocumentPositionFollowing;
-	} else {
 		return HTMLDocumentPositionContains | HTMLDocumentPositionPreceding;
+	} else {
+		return HTMLDocumentPositionContainedBy | HTMLDocumentPositionFollowing;
 	}
 }
 
@@ -384,7 +425,7 @@
 		return self.nodeType != HTMLNodeDocument && self.ownerDocument == otherNode;
 	}
 
-	for (HTMLNode *parentNode = self.parentNode; parentNode; parentNode = parentNode.parentNode) {
+	for (HTMLNode *parentNode = _parentNode; parentNode; parentNode = parentNode.parentNode) {
 		if (parentNode == otherNode) {
 			return YES;
 		}
@@ -406,7 +447,7 @@
 		return;
 	}
 
-	[self.childNodes enumerateObjectsUsingBlock:^(id obj, NSUInteger idx, BOOL *stop) {
+	[_childNodes enumerateObjectsUsingBlock:^(id obj, NSUInteger idx, BOOL *stop) {
 		block(obj, idx, stop);
 	}];
 }
@@ -417,7 +458,7 @@
 		return;
 	}
 
-	[self.childNodes enumerateObjectsUsingBlock:^(id obj, NSUInteger idx, BOOL *stop) {
+	[_childNodes enumerateObjectsUsingBlock:^(id obj, NSUInteger idx, BOOL *stop) {
 		if ([obj isKindOfClass:[HTMLElement class]]) {
 			block([obj asElement], idx, stop);
 		}
@@ -540,18 +581,18 @@ NS_INLINE void CheckInvalidCombination(HTMLNode *parent, HTMLNode *node, NSStrin
 
 - (void)ensurePreInsertionValidityOfNode:(HTMLNode *)node beforeChildNode:(HTMLNode *)child
 {
-	CheckParentValid(self, NSStringFromSelector(_cmd));
+	CheckParentValid(self, ValidationNodePreInsertion);
 
-	CheckChildsParent(self, child, NSStringFromSelector(_cmd));
+	CheckChildsParent(self, child, ValidationNodePreInsertion);
 
-	CheckInsertedNodeValid(node, NSStringFromSelector(_cmd));
+	CheckInsertedNodeValid(node, ValidationNodePreInsertion);
 
-	CheckInvalidCombination(self, node, NSStringFromSelector(_cmd));
+	CheckInvalidCombination(self, node, ValidationNodePreInsertion);
 
-	void (^ hierarchyError)() = ^{
+	void (^ hierarchyError)(void) = ^{
 		[NSException raise:HTMLKitHierarchyRequestError
 					format:@"%@: Hierarchy Request Error, inserting (%@) into (%@). The operation would yield an incorrect node tree.",
-		 NSStringFromSelector(_cmd), self, node];
+		 ValidationNodePreInsertion, self, node];
 	};
 
 	if (self.nodeType == HTMLNodeDocument) {
@@ -590,18 +631,18 @@ NS_INLINE void CheckInvalidCombination(HTMLNode *parent, HTMLNode *node, NSStrin
 
 - (void)ensureReplacementValidityOfChildNode:(HTMLNode *)child withNode:(HTMLNode *)node
 {
-	CheckParentValid(self, NSStringFromSelector(_cmd));
+	CheckParentValid(self, ValidationNodeReplacement);
 
-	CheckChildsParent(self, child, NSStringFromSelector(_cmd));
+	CheckChildsParent(self, child, ValidationNodeReplacement);
 
-	CheckInsertedNodeValid(node, NSStringFromSelector(_cmd));
+	CheckInsertedNodeValid(node, ValidationNodeReplacement);
 
-	CheckInvalidCombination(self, node, NSStringFromSelector(_cmd));
+	CheckInvalidCombination(self, node, ValidationNodeReplacement);
 
-	void (^ hierarchyError)() = ^{
+	void (^ hierarchyError)(void) = ^{
 		[NSException raise:HTMLKitHierarchyRequestError
 					format:@"%@: Hierarchy Request Error. The operation would yield an incorrect node tree.",
-		 NSStringFromSelector(_cmd)];
+		 ValidationNodeReplacement];
 	};
 
 	void (^ checkParentHasAnotherChildOfType)(HTMLNodeType) = ^ void (HTMLNodeType type) {
@@ -650,6 +691,21 @@ NS_INLINE void CheckInvalidCombination(HTMLNode *parent, HTMLNode *node, NSStrin
 
 #endif
 
+#pragma mark - Clone
+
+- (instancetype)cloneNodeDeep:(BOOL)deep
+{
+	HTMLNode *copy = [self copy];
+
+	if (deep) {
+		for (HTMLNode *child in _childNodes) {
+			[copy appendNode:[child cloneNodeDeep:YES]];
+		}
+	}
+
+	return copy;
+}
+
 #pragma mark - NSCopying
 
 - (id)copyWithZone:(NSZone *)zone

Sources/HTMLNodeFilter.m

@@ -15,7 +15,7 @@
 
 @interface HTMLNodeFilterBlock ()
 {
-	BOOL (^ _block)(HTMLNode *);
+	HTMLNodeFilterValue (^ _block)(HTMLNode *);
 }
 @end
 

Sources/HTMLNodeIterator.m

@@ -11,6 +11,7 @@
 #import "HTMLNode.h"
 #import "HTMLNodeFilter.h"
 #import "HTMLNodeTraversal.h"
+#import "HTMLDocument+Private.h"
 
 typedef NS_ENUM(short, TraverseDirection)
 {
@@ -18,11 +19,6 @@ typedef NS_ENUM(short, TraverseDirection)
 	TraverseDirectionPrevious
 };
 
-@interface HTMLDocument (Private)
-- (void)attachNodeIterator:(HTMLNodeIterator *)iterator;
-- (void)detachNodeIterator:(HTMLNodeIterator *)iterator;
-@end
-
 @interface HTMLNodeIterator ()
 {
 	HTMLNode *_root;

Sources/HTMLOrderedDictionary.m

@@ -154,4 +154,11 @@
 	return [_keys countByEnumeratingWithState:state objects:buffer count:len];
 }
 
+#pragma mark - Copying
+
+- (id)mutableCopy
+{
+	return [[HTMLOrderedDictionary alloc] initWithDictionary:self];
+}
+
 @end

Sources/HTMLParseErrorToken.m

@@ -10,29 +10,46 @@
 
 @interface HTMLParseErrorToken ()
 {
-	NSString *_reason;
+	NSString *_code;
+	NSString *_details;
 	NSUInteger _location;
 }
 @end
 
 @implementation HTMLParseErrorToken
-@synthesize reason = _reason;
+@synthesize code = _code;
+@synthesize details = _details;
 @synthesize location = _location;
 
-- (instancetype)initWithReasonMessage:(NSString *)reason andStreamLocation:(NSUInteger)location
+- (instancetype)initWithCode:(NSString *)code details:(NSString *)details location:(NSUInteger)location
 {
 	self = [super init];
 	if (self) {
 		self.type = HTMLTokenTypeParseError;
-		_reason = [reason copy];
+		_code = [code copy];
+		_details = [details copy];
 		_location = location;
 	}
 	return self;
 }
 
+- (BOOL)isEqual:(id)other
+{
+	if ([other isKindOfClass:[self class]]) {
+		HTMLParseErrorToken *token = (HTMLParseErrorToken *)other;
+		return bothNilOrEqual(self.code, token.code);
+	}
+	return NO;
+}
+
+- (NSUInteger)hash
+{
+	return self.code.hash + self.code.hash;
+}
+
 - (NSString *)description
 {
-	return [NSString stringWithFormat:@"<%@: %p Reason='%@' Location='%lu'>", self.class, self, _reason, (unsigned long)_location];
+	return [NSString stringWithFormat:@"<%@: %p Code='%@' Details='%@' Location='%lu'>", self.class, self, _code, _details, (unsigned long)_location];
 }
 
 @end

Sources/HTMLParser.m

@@ -18,14 +18,7 @@
 #import "HTMLMarker.h"
 #import "NSString+HTMLKit.h"
 #import "CSSSelectors.h"
-
-@interface HTMLTokenizer (Private)
-@property (nonatomic, weak) HTMLParser *parser;
-@end
-
-@interface HTMLDocument (Private)
-@property (nonatomic, assign) HTMLDocumentReadyState readyState;
-@end
+#import "HTMLDocument+Private.h"
 
 @interface HTMLParser ()
 {
@@ -80,6 +73,10 @@
 
 		_tokenizer = [[HTMLTokenizer alloc] initWithString:string ?: @""];
 		_tokenizer.parser = self;
+		__weak HTMLParser *weakSelf = self;
+		_tokenizer.parseErrorCallback = ^(HTMLParseErrorToken *token) {
+			[weakSelf emitParseError:@"Tokenization error: %@", token.asParseError];
+		};
 
 		_pendingTableCharacterTokens = [[HTMLCharacterToken alloc] initWithString:@""];
 
@@ -137,6 +134,10 @@
 	[self initializeDocument];
 	_tokenizer = [[HTMLTokenizer alloc] initWithString:_tokenizer.string];
 	_tokenizer.parser = self;
+	__weak HTMLParser *weakSelf = self;
+	_tokenizer.parseErrorCallback = ^(HTMLParseErrorToken *token) {
+		[weakSelf emitParseError:@"Tokenization error: %@", token.asParseError];
+	};
 
 	_contextElement = contextElement;
 	_fragmentParsingAlgorithm = YES;
@@ -233,11 +234,6 @@
 		return NO;
 	};
 
-	if (token.isParseError) {
-		[self emitParseError:@"Tokenizer Parser Error: %@", token.asParseError];
-		return;
-	}
-
 	if (_ignoreNextLineFeedCharacterToken) {
 		_ignoreNextLineFeedCharacterToken = NO;
 		if (token.isCharacterToken) {
@@ -457,9 +453,11 @@
 																					beforeChildNode:&child];
 	if (adjustedInsertionLocation.nodeType != HTMLNodeDocument) {
 		if (child != nil && child.previousSibling.nodeType == HTMLNodeText) {
-			[(HTMLText *)child.previousSibling appendString:data];
+			HTMLText *textNode = (HTMLText *)child.previousSibling;
+			[textNode appendData:data];
 		} else if (adjustedInsertionLocation.lastChild.nodeType == HTMLNodeText) {
-			[(HTMLText *)adjustedInsertionLocation.lastChild appendString:data];
+			HTMLText *textNode = (HTMLText *)adjustedInsertionLocation.lastChild;
+			[textNode appendData:data];
 		} else {
 			HTMLText *text = [[HTMLText alloc] initWithData:data];
 			[adjustedInsertionLocation insertNode:text beforeChildNode:child];
@@ -519,7 +517,7 @@
 
 - (void)generateImpliedEndTagsExceptForElement:(NSString *)tagName
 {
-	while ([self.currentNode.tagName isEqualToAny:@"dd", @"dt", @"li", @"menuitem", @"option", @"optgroup", @"p", @"rb", @"rp", @"rt", @"rtc", nil] &&
+	while ([self.currentNode.tagName isEqualToAny:@"dd", @"dt", @"li", @"option", @"optgroup", @"p", @"rb", @"rp", @"rt", @"rtc", nil] &&
 		   ![self.currentNode.tagName isEqualToString:tagName]) {
 		[_stackOfOpenElements popCurrentNode];
 	}
@@ -1167,8 +1165,8 @@
 				[self HTMLInsertionModeInTemplate:token];
 			} else {
 				for (HTMLElement *node in _stackOfOpenElements) {
-					if ([node.tagName isEqualToAny:@"dd", @"dt", @"li", @"menuitem", @"optgroup", @"option", @"p", @"rb",
-						 @"rp", @"rt", @"rtc", @"tbody", @"td", @"tfoot", @"th", @"thead", @"tr", @"body", @"html", nil]) {
+					if ([node.tagName isEqualToAny:@"dd", @"dt", @"li", @"optgroup", @"option", @"p", @"rb", @"rp",
+						 @"rt", @"rtc", @"tbody", @"td", @"tfoot", @"th", @"thead", @"tr", @"body", @"html", nil]) {
 						[self emitParseError:@"EOF reached with unclosed element <%@> in <body>", node.tagName];
 						break;
 					}
@@ -1231,20 +1229,12 @@
 		[self switchInsertionMode:HTMLInsertionModeInFrameset];
 	} else if ([tagName isEqualToAny:@"address", @"article", @"aside", @"blockquote", @"center",
 				@"details", @"dialog", @"dir", @"div", @"dl", @"fieldset", @"figcaption", @"figure",
-				@"footer", @"header", @"hgroup", @"main", @"nav", @"ol", @"p", @"section",
+				@"footer", @"header", @"hgroup", @"main", @"menu", @"nav", @"ol", @"p", @"section",
 				@"summary", @"ul", nil]) {
 		if ([_stackOfOpenElements hasElementInButtonScopeWithTagName:@"p"]) {
 			[self closePElement];
 		}
 		[self insertElementForToken:token];
-	} else if ([tagName isEqualToString:@"menu"]) {
-		if ([_stackOfOpenElements hasElementInButtonScopeWithTagName:@"p"]) {
-			[self closePElement];
-		}
-		if ([self.currentNode.tagName isEqualToString:@"menuitem"]) {
-			[_stackOfOpenElements popCurrentNode];
-		}
-		[self insertElementForToken:token];
 	} else if ([tagName isEqualToAny:@"h1", @"h2", @"h3", @"h4", @"h5", @"h6", nil]) {
 		if ([_stackOfOpenElements hasElementInButtonScopeWithTagName:@"p"]) {
 			[self closePElement];
@@ -1285,7 +1275,7 @@
 								  @"dd": @[@"dd", @"dt"],
 								  @"dt": @[@"dd", @"dt"]};
 
-		for (HTMLElement *node in _stackOfOpenElements.reverseObjectEnumerator.allObjects) {
+		for (HTMLElement *node in _stackOfOpenElements.reverseObjectEnumerator) {
 			if ([map[tagName] containsObject:node.tagName]) {
 				[self generateImpliedEndTagsExceptForElement:node.tagName];
 				if (![self.currentNode.tagName isEqualToString:node.tagName]) {
@@ -1388,9 +1378,6 @@
 		if ([_stackOfOpenElements hasElementInButtonScopeWithTagName:@"p"]) {
 			[self closePElement];
 		}
-		if ([self.currentNode.tagName isEqualToString:@"menuitem"]) {
-			[_stackOfOpenElements popCurrentNode];
-		}
 		[self insertElementForToken:token];
 		[_stackOfOpenElements popCurrentNode];
 		_framesetOkFlag = NO;
@@ -1436,12 +1423,6 @@
 		}
 		[self reconstructActiveFormattingElements];
 		[self insertElementForToken:token];
-	} else if ([tagName isEqualToString:@"menuitem"]) {
-		if ([self.currentNode.tagName isEqualToString:@"menuitem"]) {
-			[_stackOfOpenElements popCurrentNode];
-		}
-		[self reconstructActiveFormattingElements];
-		[self insertElementForToken:token];
 	} else if ([tagName isEqualToAny:@"rb", @"rtc", nil]) {
 		if ([_stackOfOpenElements hasElementInScopeWithTagName:@"ruby"]) {
 			[self generateImpliedEndTagsExceptForElement:nil];
@@ -1462,7 +1443,6 @@
 	} else if ([tagName isEqualToString:@"math"]) {
 		[self reconstructActiveFormattingElements];
 		AdjustMathMLAttributes(token);
-		AdjustForeignAttributes(token);
 		[self insertForeignElementForToken:token inNamespace:HTMLNamespaceMathML];
 		if (token.isSelfClosing) {
 			[_stackOfOpenElements popCurrentNode];
@@ -1470,7 +1450,6 @@
 	} else if ([tagName isEqualToString:@"svg"]) {
 		[self reconstructActiveFormattingElements];
 		AdjustSVGAttributes(token);
-		AdjustForeignAttributes(token);
 		[self insertForeignElementForToken:token inNamespace:HTMLNamespaceSVG];
 		if (token.isSelfClosing) {
 			[_stackOfOpenElements popCurrentNode];
@@ -1496,8 +1475,8 @@
 			[self emitParseError:@"Unexpected end tag </body> without body element in scope in <body>"];
 		}
 		for (HTMLElement *node in _stackOfOpenElements) {
-			if ([node.tagName isEqualToAny:@"dd", @"dt", @"li", @"menuitem", @"optgroup", @"option", @"p", @"rb", @"rp",
-				 @"rt", @"rtc", @"tbody", @"td", @"tfoot", @"th", @"thead", @"tr", @"body", @"html", nil]) {
+			if ([node.tagName isEqualToAny:@"dd", @"dt", @"li", @"optgroup", @"option", @"p", @"rb", @"rp", @"rt",
+				 @"rtc", @"tbody", @"td", @"tfoot", @"th", @"thead", @"tr", @"body", @"html", nil]) {
 				[self emitParseError:@"Misnested end tag </%@> with open element <%@> in <body>", tagName, node.tagName];
 				break;
 			}
@@ -1551,7 +1530,7 @@
 		}
 		[self closePElement];
 	} else if ([tagName isEqualToString:@"li"]) {
-		if (![_stackOfOpenElements hasElementInListItemScopeWithTagName:@"li"]) {
+		if (![_stackOfOpenElements hasElementInListItemScopeWithTagName:tagName]) {
 			[self emitParseError:@"Unexpected <li> element in <body>"];
 			return;
 		}
@@ -1571,7 +1550,7 @@
 		}
 		[_stackOfOpenElements popElementsUntilElementPoppedWithTagName:tagName];
 	} else if ([tagName isEqualToAny:@"h1", @"h2", @"h3", @"h4", @"h5", @"h6", nil]) {
-		if (![_stackOfOpenElements hasAnyElementInScopeWithAnyOfTagNames:@[@"h1", @"h2", @"h3", @"h4", @"h5", @"h6"]]) {
+		if (![_stackOfOpenElements hasHeaderElementInScope]) {
 			[self emitParseError:@"Unexpected <%@> element in <body>", tagName];
 			return;
 		}
@@ -1591,7 +1570,7 @@
 			return;
 		}
 	} else if ([tagName isEqualToAny:@"applet", @"marquee", @"object", nil]) {
-		if (![_stackOfOpenElements hasAnyElementInScopeWithAnyOfTagNames:@[@"applet", @"marquee", @"object"]]) {
+		if (![_stackOfOpenElements hasElementInScopeWithTagName:tagName]) {
 			[self emitParseError:@"Unexpected <%@> element in <body>", tagName];
 			return;
 		}
@@ -1612,7 +1591,7 @@
 
 - (void)processAnyOtherEndTagTokenInBody:(HTMLTagToken *)token
 {
-	for (HTMLElement *node in _stackOfOpenElements.reverseObjectEnumerator.allObjects) {
+	for (HTMLElement *node in _stackOfOpenElements.reverseObjectEnumerator) {
 		if ([node.tagName isEqualToString:token.tagName]) {
 			[self generateImpliedEndTagsExceptForElement:token.tagName];
 			if (![node.tagName isEqualToString:self.currentNode.tagName]) {
@@ -2552,7 +2531,7 @@
 			return;
 		case HTMLTokenTypeStartTag:
 		{
-			void (^ anythingElse)() = ^ {
+			void (^ anythingElse)(void) = ^ {
 				if (self.adjustedCurrentNode.htmlNamespace == HTMLNamespaceMathML) {
 					AdjustMathMLAttributes(token.asTagToken);
 				}
@@ -2560,14 +2539,13 @@
 					AdjustSVGNameCase(token.asTagToken);
 					AdjustSVGAttributes(token.asTagToken);
 				}
-				AdjustForeignAttributes(token.asTagToken);
 				[self insertForeignElementForToken:token.asTagToken inNamespace:self.adjustedCurrentNode.htmlNamespace];
 				if (token.asTagToken.selfClosing) {
 					[_stackOfOpenElements popCurrentNode];
 				}
 			};
 
-			void (^ matchedCase)() = ^ {
+			void (^ matchedCase)(void) = ^ {
 				[self emitParseError:@"Unexpected start tag <%@> in foreign content", token.asTagToken.tagName];
 				if (_fragmentParsingAlgorithm) {
 					anythingElse();

Sources/HTMLRange.m

@@ -0,0 +1,767 @@
+//
+//  HTMLRange.m
+//  HTMLKit
+//
+//  Created by Iska on 20/11/16.
+//  Copyright © 2016 BrainCookie. All rights reserved.
+//
+
+#import "HTMLRange.h"
+#import "HTMLDocument.h"
+#import "HTMLKitDOMExceptions.h"
+#import "HTMLDocument+Private.h"
+#import "HTMLDOMUtils.h"
+#import "HTMLNodeTraversal.h"
+
+@interface HTMLRange ()
+{
+	HTMLDocument *_ownerDocument;
+}
+@end
+
+@implementation HTMLRange
+
+#pragma mark - Lifecycle
+
+- (instancetype)initWithDocument:(HTMLDocument *)document
+{
+	return [self initWithDocument:document startContainer:document startOffset:0 endContainer:document endOffset:0];
+}
+
+- (instancetype)initWithDowcument:(HTMLDocument *)document
+{
+	return [self initWithDocument:document startContainer:document startOffset:0 endContainer:document endOffset:0];
+}
+
+- (instancetype)initWithDowcument:(HTMLDocument *)document
+				   startContainer:(HTMLNode *)startContainer startOffset:(NSUInteger)startOffset
+					 endContainer:(HTMLNode *)endContainer endOffset:(NSUInteger)endOffset
+{
+	return [self initWithDocument:document
+				   startContainer:startContainer startOffset:startOffset
+					 endContainer:endContainer endOffset:endOffset];
+}
+
+- (instancetype)initWithDocument:(HTMLDocument *)document
+				  startContainer:(HTMLNode *)startContainer startOffset:(NSUInteger)startOffset
+					endContainer:(HTMLNode *)endContainer endOffset:(NSUInteger)endOffset
+{
+	self = [super init];
+	if (self) {
+		_ownerDocument = document;
+		[_ownerDocument attachRange:self];
+		[self setStartNode:startContainer startOffset:startOffset];
+		[self setEndNode:endContainer endOffset:endOffset];
+	}
+	return self;
+}
+
+- (void)dealloc
+{
+	[_ownerDocument detachRange:self];
+}
+
+#pragma mark - Properties
+
+- (BOOL)isCollapsed
+{
+	return _startContainer == _endContainer && _startOffset == _endOffset;
+}
+
+- (HTMLNode *)commonAncestorContainer
+{
+	return GetCommonAncestorContainer(_startContainer, _endContainer);
+}
+
+- (HTMLNode *)rootNode
+{
+	return _startContainer.rootNode;
+}
+
+#pragma mark - Boundaries
+
+NS_INLINE void CheckValidBoundaryNode(HTMLDocument *document, HTMLNode *node, NSString *cmd)
+{
+	if (node.ownerDocument != document) {
+		[NSException raise:HTMLKitWrongDocumentError
+					format:@"%@: Invalid Node Error, %@ is not in the same document.",
+		 cmd, node];
+	}
+}
+
+NS_INLINE void CheckValidBoundaryNodeType(HTMLNode *node, NSString *cmd)
+{
+	if (node == nil || node.nodeType == HTMLNodeDocumentType) {
+		[NSException raise:HTMLKitInvalidNodeTypeError
+					format:@"%@: Invalid Node Type Error, %@ is not a valid range boundary node.",
+		 cmd, node];
+	}
+}
+
+NS_INLINE void CheckValidBoundaryOffset(HTMLNode *node, NSUInteger offset, NSString *cmd)
+{
+	if (node.length < offset) {
+		[NSException raise:HTMLKitIndexSizeError
+					format:@"%@: Index Size Error, invalid index %lu for range boundary node %@.",
+		 cmd, (unsigned long)offset, node];
+	}
+}
+
+NS_INLINE void CheckValidDocument(HTMLRange *lhs, HTMLRange *rhs, NSString *cmd)
+{
+	if (lhs.rootNode != rhs.rootNode) {
+		[NSException raise:HTMLKitWrongDocumentError
+					format:@"%@: Wrong Document Error, ranges %@ and %@ are not in the same document.",
+		 cmd, lhs, rhs];
+	}
+}
+
+NS_INLINE NSComparisonResult CompareBoundaries(HTMLNode *startNode, NSUInteger startOffset, HTMLNode *endNode, NSUInteger endOffset)
+{
+	if (startNode == endNode) {
+		if (startOffset == endOffset) {
+			return NSOrderedSame;
+		} else if (startOffset < endOffset) {
+			return NSOrderedAscending;
+		} else {
+			return NSOrderedDescending;
+		}
+	}
+
+	HTMLDocumentPosition position = [startNode compareDocumentPositionWithNode:endNode];
+	if ((position & HTMLDocumentPositionFollowing) == HTMLDocumentPositionFollowing) {
+		if (CompareBoundaries(endNode, endOffset, startNode, startOffset) == NSOrderedAscending) {
+			return NSOrderedDescending;
+		} else {
+			return NSOrderedAscending;
+		}
+	}
+
+	if ((position & HTMLDocumentPositionContains) == HTMLDocumentPositionContains) {
+		HTMLNode *child = endNode;
+		while (child.parentNode != startNode) {
+			child = child.parentNode;
+		}
+		if (child.index < startOffset) {
+			return NSOrderedDescending;
+		}
+	}
+
+	return NSOrderedAscending;
+}
+
+- (void)setStartNode:(HTMLNode *)node startOffset:(NSUInteger)offset
+{
+	CheckValidBoundaryNode(_ownerDocument, node, NSStringFromSelector(_cmd));
+
+	CheckValidBoundaryNodeType(node, NSStringFromSelector(_cmd));
+
+	CheckValidBoundaryOffset(node, offset, NSStringFromSelector(_cmd));
+
+	if (self.rootNode != node.rootNode ||
+		CompareBoundaries(node, offset, _endContainer, _endOffset) == NSOrderedDescending) {
+		_endContainer = node;
+		_endOffset = offset;
+	}
+
+	_startContainer = node;
+	_startOffset = offset;
+}
+
+- (void)setEndNode:(HTMLNode *)node endOffset:(NSUInteger)offset
+{
+	CheckValidBoundaryNode(_ownerDocument, node, NSStringFromSelector(_cmd));
+
+	CheckValidBoundaryNodeType(node, NSStringFromSelector(_cmd));
+
+	CheckValidBoundaryOffset(node, offset, NSStringFromSelector(_cmd));
+
+	if (self.rootNode != node.rootNode ||
+		CompareBoundaries(node, offset, _startContainer, _startOffset) == NSOrderedAscending) {
+		_startContainer = node;
+		_startOffset = offset;
+	}
+
+	_endContainer = node;
+	_endOffset = offset;
+}
+
+- (void)setStartBeforeNode:(HTMLNode *)node
+{
+	HTMLNode *parent = node.parentNode;
+	[self setStartNode:parent startOffset:node.index];
+}
+
+- (void)setStartAfterNode:(HTMLNode *)node
+{
+	HTMLNode *parent = node.parentNode;
+	[self setStartNode:parent startOffset:node.index + 1];
+}
+
+- (void)setEndBeforeNode:(HTMLNode *)node
+{
+	HTMLNode *parent = node.parentNode;
+	[self setEndNode:parent endOffset:node.index];
+}
+
+- (void)setEndAfterNode:(HTMLNode *)node
+{
+	HTMLNode *parent = node.parentNode;
+	[self setEndNode:parent endOffset:node.index + 1];
+}
+
+- (void)collapseToStart
+{
+	[self setEndNode:_startContainer endOffset:_startOffset];
+}
+
+- (void)collapseToEnd
+{
+	[self setStartNode:_endContainer startOffset:_endOffset];
+}
+
+- (void)selectNode:(HTMLNode *)node
+{
+	HTMLNode *parent = node.parentNode;
+	[self setStartNode:parent startOffset:node.index];
+	[self setEndNode:parent endOffset:node.index + 1];
+}
+
+- (void)selectNodeContents:(HTMLNode *)node
+{
+	[self setStartNode:node startOffset:0];
+	[self setEndNode:node endOffset:node.length];
+}
+
+- (NSComparisonResult)compareBoundaryPoints:(HTMLRangeComparisonMethod)method sourceRange:(HTMLRange *)sourceRange
+{
+	CheckValidDocument(self, sourceRange, NSStringFromSelector(_cmd));
+
+	switch (method) {
+		case HTMLRangeComparisonMethodStartToStart:
+			return CompareBoundaries(_startContainer, _startOffset, sourceRange.startContainer, sourceRange.startOffset);
+		case HTMLRangeComparisonMethodStartToEnd:
+			return CompareBoundaries(_endContainer, _endOffset, sourceRange.startContainer, sourceRange.startOffset);
+		case HTMLRangeComparisonMethodEndToEnd:
+			return CompareBoundaries(_endContainer, _endOffset, sourceRange.endContainer, sourceRange.endOffset);
+		case HTMLRangeComparisonMethodEndToStart:
+			return CompareBoundaries(_startContainer, _startOffset, sourceRange.endContainer, sourceRange.endOffset);
+	}
+}
+
+#pragma mark - Containment
+
+- (NSComparisonResult)comparePoint:(HTMLNode *)node offset:(NSUInteger)offset
+{
+	CheckValidBoundaryNode(_ownerDocument, node, NSStringFromSelector(_cmd));
+
+	CheckValidBoundaryNodeType(node, NSStringFromSelector(_cmd));
+
+	CheckValidBoundaryOffset(node, offset, NSStringFromSelector(_cmd));
+
+	if (CompareBoundaries(node, offset, _startContainer, _startOffset) == NSOrderedAscending) {
+		return NSOrderedAscending;
+	}
+
+	if (CompareBoundaries(node, offset, _endContainer, _endOffset) == NSOrderedDescending) {
+		return NSOrderedDescending;
+	}
+
+	return NSOrderedSame;
+}
+
+- (BOOL)containsPoint:(HTMLNode *)node offset:(NSUInteger)offset
+{
+	return [self comparePoint:node offset:offset] == NSOrderedSame;
+}
+
+- (BOOL)intersectsNode:(HTMLNode *)node
+{
+	if (self.rootNode != node.rootNode) {
+		return NO;
+	}
+
+	HTMLNode *parent = node.parentNode;
+	if (parent == nil) {
+		return YES;
+	}
+
+	NSUInteger offset = node.index;
+	if (CompareBoundaries(parent, offset, _endContainer, _endOffset) == NSOrderedAscending &&
+		CompareBoundaries(parent, offset + 1, _startContainer, _startOffset) == NSOrderedDescending) {
+		return YES;
+	}
+
+	return NO;
+}
+
+- (BOOL)containsNode:(HTMLNode *)node
+{
+	return CompareBoundaries(_startContainer, _startOffset, node, 0) == NSOrderedAscending &&
+	CompareBoundaries(_endContainer, _endOffset, node, node.length) == NSOrderedDescending;
+}
+
+- (BOOL)partiallyContainsNode:(HTMLNode *)node
+{
+	return [GetAncestorNodes(_startContainer) containsObject:node] || [GetAncestorNodes(_endContainer) containsObject:node];
+}
+
+- (NSArray *)containedNodes:(HTMLNode *)commonAncestor
+{
+	NSMutableArray *containedNodes = [NSMutableArray array];
+	[commonAncestor.childNodes enumerateObjectsUsingBlock:^(HTMLNode * _Nonnull node, NSUInteger idx, BOOL * _Nonnull stop) {
+		if (node.nodeType == HTMLNodeDocumentType) {
+			[NSException raise:HTMLKitHierarchyRequestError format:@"Hierarchy Request Error, encountered a DOCTYPE contained in range: %@", self];
+		}
+		if ([self containsNode:node]) {
+			[containedNodes addObject:node];
+		}
+	}];
+
+	return containedNodes;
+}
+
+#pragma mark - Update Callbacks
+
+- (void)didRemoveCharacterDataInNode:(HTMLCharacterData *)node atOffset:(NSUInteger)offset withLength:(NSUInteger)length
+{
+	if (_startContainer == node && _startOffset > offset) {
+		if (_startOffset <= offset + length) {
+			_startOffset = offset;
+		} else {
+			_startOffset = _startOffset - length;
+		}
+	}
+	if (_endContainer == node && _endOffset > offset) {
+		if (_endOffset <= offset + length) {
+			_endOffset = offset;
+		} else {
+			_endOffset = _endOffset - length;
+		}
+	}
+}
+
+- (void)didAddCharacterDataToNode:(HTMLCharacterData *)node atOffset:(NSUInteger)offset withLength:(NSUInteger)length
+{
+	if (_startContainer == node && _startOffset > offset) {
+		_startOffset = _startOffset + length;
+	}
+	if (_endContainer == node && _endOffset > offset) {
+		_endOffset = _endOffset + length;
+	}
+}
+
+- (void)didInsertNewTextNode:(HTMLText *)newNode intoParent:(HTMLNode *)parent afterSplittingTextNode:(HTMLText *)node atOffset:(NSUInteger)offset
+{
+	if (_startContainer == node && _startOffset > offset) {
+		_startContainer = newNode;
+		_startOffset -= offset;
+	}
+
+	if (_endContainer == node && _endOffset > offset) {
+		_endContainer = newNode;
+		_endOffset -= offset;
+	}
+
+	if (_startContainer == parent && _startOffset == node.index + 1) {
+		_startOffset += 1;
+	}
+
+	if (_endContainer == parent && _endOffset == node.index + 1) {
+		_endOffset += 1;
+	}
+}
+
+- (void)clampRangesAfterSplittingTextNode:(HTMLText *)node atOffset:(NSUInteger)offset
+{
+	if (_startContainer == node && _startOffset > offset) {
+		_startOffset = offset;
+	}
+
+	if (_endContainer == node && _endOffset > offset) {
+		_endOffset = offset;
+	}
+}
+
+- (void)runRemovingStepsForNode:(HTMLNode *)oldNode withOldParent:(HTMLNode *)oldParent andOldPreviousSibling:(HTMLNode *)oldPreviousSibling
+{
+	NSUInteger oldIndex = oldPreviousSibling.index + 1;
+
+	if ([_startContainer containsNode:oldNode]) {
+		[self setStartNode:oldNode startOffset:oldIndex];
+	}
+
+	if ([_endContainer containsNode:oldNode]) {
+		[self setEndNode:oldNode endOffset:oldIndex];
+	}
+
+	if (_startContainer == oldParent && _startOffset > oldIndex) {
+		_startOffset -= 1;
+	}
+
+	if (_endContainer == oldParent && _endOffset > oldIndex) {
+		_endOffset -= 1;
+	}
+}
+
+#pragma mark - Mutations
+
+NS_INLINE HTMLNode * GetHighestPartiallyContainedChild(HTMLNode *node, HTMLNode *root)
+{
+	if (node == root) {
+		return nil;
+	}
+
+	while (node.parentNode != root) {
+		node = node.parentNode;
+	}
+	return node;
+}
+
+NS_INLINE HTMLCharacterData * CloneCharachterData(HTMLNode *node, NSUInteger start, NSUInteger length, BOOL delete)
+{
+	HTMLCharacterData *clone = (HTMLCharacterData *)[node copy];
+	NSRange range = NSMakeRange(start, length);
+	[clone setData:[clone.data substringWithRange:range]];
+
+	if (delete) {
+		[(HTMLCharacterData *)node deleteDataInRange:range];
+	}
+
+	return clone;
+}
+
+- (void)deleteContents
+{
+	if (self.isCollapsed) {
+		return;
+	}
+
+	if (_startContainer == _endContainer && [_startContainer isKindOfClass:[HTMLCharacterData class]]) {
+		[(HTMLCharacterData *)_startContainer deleteDataInRange:NSMakeRange(_startOffset, _endOffset - _startOffset)];
+		return;
+	}
+
+	HTMLNode *commonAncestor = self.commonAncestorContainer;
+
+	NSMutableArray *containedNodes = [NSMutableArray array];
+
+	HTMLNode *node = FollowingNode(_startContainer, commonAncestor);
+	while (node) {
+		if ([self containsNode:node]) {
+			[containedNodes addObject:node];
+			node = FollowingNodeSkippingChildren(node, commonAncestor);
+		} else {
+			node = FollowingNode(node, commonAncestor);
+		}
+	}
+
+	HTMLNode *newNode = _startContainer;
+	NSUInteger newOffset = _startOffset;
+
+	if (![_startContainer containsNode:_endContainer]) {
+		HTMLNode *referenceNode = _startContainer;
+		while (referenceNode.parentNode) {
+			if ([referenceNode.parentNode containsNode:_endContainer]) {
+				newNode = referenceNode.parentNode;
+				newOffset = referenceNode.index + 1;
+				break;
+			}
+			referenceNode = referenceNode.parentNode;
+		}
+	}
+
+	if ([_startContainer isKindOfClass:[HTMLCharacterData class]]) {
+		[(HTMLCharacterData *)_startContainer deleteDataInRange:NSMakeRange(_startOffset, _startContainer.length - _startOffset)];
+	}
+
+	for (HTMLNode *node in containedNodes) {
+		[node removeFromParentNode];
+	}
+
+	if ([_endContainer isKindOfClass:[HTMLCharacterData class]]) {
+		[(HTMLCharacterData *)_endContainer deleteDataInRange:NSMakeRange(0, _endOffset)];
+	}
+
+	[self setStartNode:newNode startOffset:newOffset];
+	[self setEndNode:newNode endOffset:newOffset];
+}
+
+- (HTMLDocumentFragment *)extractContents
+{
+	HTMLDocumentFragment *fragment = [[HTMLDocumentFragment alloc] initWithDocument:_ownerDocument];
+
+		// Nothing todo
+	if (self.isCollapsed) {
+		return fragment;
+	}
+
+		// Same character data container, handle that and return
+	if (_startContainer == _endContainer && [_startContainer isKindOfClass:[HTMLCharacterData class]]) {
+		HTMLCharacterData *clone = CloneCharachterData(_startContainer, _startOffset, _endOffset - _startOffset, YES);
+		[fragment appendNode:clone];
+
+		return fragment;
+	}
+
+	HTMLNode *commonAncestor = self.commonAncestorContainer;
+	HTMLNode *firstPartiallyContainedChild = GetHighestPartiallyContainedChild(_startContainer, commonAncestor);
+	HTMLNode *lastPartiallyContainedChild = GetHighestPartiallyContainedChild(_endContainer, commonAncestor);
+	NSArray *containedNodes = [self containedNodes:commonAncestor];
+
+	HTMLNode *newNode = _startContainer;
+	NSUInteger newOffset = _startOffset;
+
+	if (![_startContainer containsNode:_endContainer]) {
+		HTMLNode *referenceNode = _startContainer;
+		while (referenceNode.parentNode) {
+			if ([referenceNode.parentNode containsNode:_endContainer]) {
+				newNode = referenceNode.parentNode;
+				newOffset = referenceNode.index + 1;
+				break;
+			}
+			referenceNode = referenceNode.parentNode;
+		}
+	}
+
+	if ([firstPartiallyContainedChild isKindOfClass:[HTMLCharacterData class]]) {
+		HTMLCharacterData *clone = CloneCharachterData(_startContainer, _startOffset, _startContainer.length - _startOffset, YES);
+		[fragment appendNode:clone];
+	} else if (firstPartiallyContainedChild != nil) {
+		HTMLNode *clone = [firstPartiallyContainedChild copy];
+		[fragment appendNode:clone];
+
+		HTMLRange *subRange = [[HTMLRange alloc] initWithDocument:_ownerDocument
+													startContainer:_startContainer
+													   startOffset:_startOffset
+													  endContainer:firstPartiallyContainedChild
+														 endOffset:firstPartiallyContainedChild.length];
+		HTMLDocumentFragment *subFragment = [subRange extractContents];
+		[clone appendNode:subFragment];
+	}
+
+	for (HTMLNode *node in containedNodes) {
+		[fragment appendNode:node];
+	}
+
+	if ([lastPartiallyContainedChild isKindOfClass:[HTMLCharacterData class]]) {
+		HTMLCharacterData *clone = CloneCharachterData(_endContainer, 0, _endOffset, YES);
+		[fragment appendNode:clone];
+	} else if (lastPartiallyContainedChild != nil) {
+		HTMLNode *clone = [lastPartiallyContainedChild copy];
+		[fragment appendNode:clone];
+
+		HTMLRange *subRange = [[HTMLRange alloc] initWithDocument:_ownerDocument
+													startContainer:lastPartiallyContainedChild
+													   startOffset:0
+													  endContainer:_endContainer
+														 endOffset:_endOffset];
+		HTMLDocumentFragment *subFragment = [subRange extractContents];
+		[clone appendNode:subFragment];
+	}
+
+	[self setStartNode:newNode startOffset:newOffset];
+	[self setEndNode:newNode endOffset:newOffset];
+
+	return fragment;
+}
+
+- (HTMLDocumentFragment *)cloneContents
+{
+	HTMLDocumentFragment *fragment = [[HTMLDocumentFragment alloc] initWithDocument:_ownerDocument];
+
+	// Nothing todo
+	if (self.isCollapsed) {
+		return fragment;
+	}
+
+	// Same character data container, handle that and return
+	if (_startContainer == _endContainer && [_startContainer isKindOfClass:[HTMLCharacterData class]]) {
+		HTMLCharacterData *clone = CloneCharachterData(_startContainer, _startOffset, _endOffset - _startOffset, NO);
+		[fragment appendNode:clone];
+		return fragment;
+	}
+
+	HTMLNode *commonAncestor = self.commonAncestorContainer;
+	HTMLNode *firstPartiallyContainedChild = GetHighestPartiallyContainedChild(_startContainer, commonAncestor);
+	HTMLNode *lastPartiallyContainedChild = GetHighestPartiallyContainedChild(_endContainer, commonAncestor);
+	NSArray *containedNodes = [self containedNodes:commonAncestor];
+
+	if ([firstPartiallyContainedChild isKindOfClass:[HTMLCharacterData class]]) {
+		HTMLCharacterData *clone = CloneCharachterData(_startContainer, _startOffset, _startContainer.length - _startOffset, NO);
+		[fragment appendNode:clone];
+	} else if (firstPartiallyContainedChild != nil) {
+		HTMLNode *clone = [firstPartiallyContainedChild copy];
+		[fragment appendNode:clone];
+
+		HTMLRange *subRange = [[HTMLRange alloc] initWithDocument:_ownerDocument
+													startContainer:_startContainer
+													   startOffset:_startOffset
+													  endContainer:firstPartiallyContainedChild
+														 endOffset:firstPartiallyContainedChild.length];
+		HTMLDocumentFragment *subFragment = [subRange cloneContents];
+		[clone appendNode:subFragment];
+	}
+
+	for (HTMLNode *node in containedNodes) {
+		HTMLNode *clone = [node cloneNodeDeep:YES];
+		[fragment appendNode:clone];
+	}
+
+	if ([lastPartiallyContainedChild isKindOfClass:[HTMLCharacterData class]]) {
+		HTMLCharacterData *clone = CloneCharachterData(_endContainer, 0, _endOffset, NO);
+		[fragment appendNode:clone];
+	} else if (lastPartiallyContainedChild != nil) {
+		HTMLNode *clone = [lastPartiallyContainedChild copy];
+		[fragment appendNode:clone];
+
+		HTMLRange *subRange = [[HTMLRange alloc] initWithDocument:_ownerDocument
+													startContainer:lastPartiallyContainedChild
+													   startOffset:0
+													  endContainer:_endContainer
+														 endOffset:_endOffset];
+		HTMLDocumentFragment *subFragment = [subRange cloneContents];
+		[clone appendNode:subFragment];
+	}
+
+	return fragment;
+}
+
+#pragma mark - Insertion & Surround
+
+NS_INLINE void CheckValidInsertionNode(HTMLNode *startContainer, HTMLNode *node, NSString *cmd)
+{
+	if (startContainer == node || startContainer.nodeType == HTMLNodeComment ||
+		(startContainer.nodeType == HTMLNodeText && startContainer.parentNode == nil)) {
+		[NSException raise:HTMLKitHierarchyRequestError
+					format:@"%@: Hierarchy Request Error, cannot insert node into range: %@", cmd, node];
+	}
+}
+
+- (void)insertNode:(HTMLNode *)node
+{
+	CheckValidInsertionNode(_startContainer, node, NSStringFromSelector(_cmd));
+
+	HTMLNode *referenceNode = nil;
+
+	if (_startContainer.nodeType == HTMLNodeText) {
+		referenceNode = _startContainer;
+	} else {
+		referenceNode = [_startContainer childNodeAtIndex:_startOffset];
+	}
+
+	HTMLNode *parent = _startContainer;
+	if (referenceNode != nil) {
+		parent = referenceNode.parentNode;
+	}
+
+	if (_startContainer.nodeType == HTMLNodeText) {
+		referenceNode = [(HTMLText *)_startContainer splitTextAtOffset:_startOffset];
+	}
+
+	if (node == referenceNode) {
+		referenceNode = referenceNode.nextSibling;
+	}
+
+	[node removeFromParentNode];
+
+	NSUInteger newOffset = referenceNode ? referenceNode.index : parent.length;
+	newOffset += (node.nodeType == HTMLNodeDocumentFragment) ? node.length : 1;
+
+	[parent insertNode:node beforeChildNode:referenceNode];
+
+	if (self.isCollapsed) {
+		[self setEndNode:parent endOffset:newOffset];
+	}
+}
+
+NS_INLINE void CheckValidSurroundState(HTMLRange *range, NSString *cmd)
+{
+	for (HTMLNode *node in GetAncestorNodes(range.startContainer)) {
+		if ([node containsNode:range.endContainer]) {
+			return;
+		}
+
+		if (node.nodeType != HTMLNodeText) {
+			[NSException raise:HTMLKitInvalidStateError
+						format:@"%@: Invalid State Error, cannot surround range with a partially-contaied non-text node.", cmd];
+		}
+	};
+
+	for (HTMLNode *node in GetAncestorNodes(range.endContainer)) {
+		if ([node containsNode:range.startContainer]) {
+			return;
+		}
+
+		if (node.nodeType != HTMLNodeText) {
+			[NSException raise:HTMLKitInvalidNodeTypeError
+						format:@"%@: Invalid State Error, cannot surround range with a partially-contaied non-text node.", cmd];
+		}
+	};
+}
+
+NS_INLINE void CheckValidSurroundNodeType(HTMLNode *node, NSString *cmd)
+{
+	if (node == nil || node.nodeType == HTMLNodeDocumentType || node.nodeType == HTMLNodeDocument ||
+		node.nodeType == HTMLNodeDocumentFragment) {
+		[NSException raise:HTMLKitInvalidNodeTypeError
+					format:@"%@: Invalid Node Type Error, %@ is not a valid new parent for a range.",
+		 cmd, node];
+	}
+}
+
+- (void)surroundContents:(HTMLNode *)newParent
+{
+	CheckValidSurroundState(self, NSStringFromSelector(_cmd));
+
+	CheckValidSurroundNodeType(newParent, NSStringFromSelector(_cmd));
+
+	HTMLDocumentFragment *fragment = [self extractContents];
+	[newParent removeAllChildNodes];
+
+	[self insertNode:newParent];
+	[newParent appendNode:fragment];
+	[self selectNode:newParent];
+}
+
+#pragma mark - Description & Stringifier
+
+- (NSString *)description
+{
+	return [NSString stringWithFormat:@"<%@: %p (%@, %lu), (%@, %lu)>", self.class, self,
+			_startContainer, (unsigned long)_startOffset,
+			_endContainer, (unsigned long)_endOffset];
+}
+
+- (NSString *)textContent
+{
+	HTMLNode *lastNode = nil;
+	if ([_endContainer isKindOfClass:[HTMLCharacterData class]]) {
+		lastNode = FollowingNodeSkippingChildren(_endContainer, _ownerDocument);
+	} else if (_endContainer.childNodesCount > _endOffset) {
+		lastNode = [_endContainer childNodeAtIndex:_endOffset];
+	} else {
+		lastNode = FollowingNodeSkippingChildren(_endContainer, _ownerDocument);
+	}
+
+	NSMutableString *content = [NSMutableString string];
+	for (HTMLNode *node = _startContainer; node != lastNode; node = FollowingNode(node, _ownerDocument)) {
+		if (node.nodeType == HTMLNodeText) {
+			HTMLText *text = (HTMLText *)node;
+
+			if (node == _startContainer) {
+				NSString *string = [text substringDataWithRange:NSMakeRange(_startOffset, _startContainer.length - _startOffset)];
+				[content appendString:string];
+			} else if (node == _endContainer) {
+				NSString *string = [text substringDataWithRange:NSMakeRange(0, _endOffset)];
+				[content appendString:string];
+			} else {
+				[content appendString:text.data];
+			}
+		}
+	}
+
+	return content;
+}
+
+@end

Sources/HTMLSanitizer.h

@@ -0,0 +1,21 @@
+//
+//  HTMLSanitizer.h
+//  HTMLKit
+//
+//  Created by Iska on 26.05.18.
+//  Copyright © 2018 BrainCookie. All rights reserved.
+//
+
+#import <Foundation/Foundation.h>
+#import "HTMLSanitizingPolicyBuilder.h"
+#import "HTMLSanitizingPolicy.h"
+
+NS_ASSUME_NONNULL_BEGIN
+
+@interface HTMLSanitizer : NSObject
+
++ (instancetype)sanitizerWithPolicy:(void (^)(HTMLSanitizingPolicyBuilder *))block;
+
+@end
+
+NS_ASSUME_NONNULL_END

Sources/HTMLSanitizer.m

@@ -0,0 +1,45 @@
+//
+//  HTMLSanitizer.m
+//  HTMLKit
+//
+//  Created by Iska on 26.05.18.
+//  Copyright © 2018 BrainCookie. All rights reserved.
+//
+
+#import "HTMLSanitizer.h"
+#import "HTMLTokenizer.h"
+#import "HTMLTokens.h"
+
+@interface HTMLSanitizer()
+{
+	HTMLTokenizer *_tokenizer;
+}
+
+@end
+
+@implementation HTMLSanitizer
+
++ (instancetype)sanitizerWithPolicy:(void (^)(HTMLSanitizingPolicyBuilder *))block
+{
+	HTMLSanitizingPolicyBuilder *builder = [HTMLSanitizingPolicyBuilder new];
+	block(builder);
+	return nil; //[[HTMLSanitizingPolicy alloc] initWithBuilder:builder];
+}
+
+- (instancetype)initWithString:(NSString *)string
+{
+	self = [super init];
+	if (self) {
+		_tokenizer = [[HTMLTokenizer alloc] initWithString:string ?: @""];
+	}
+	return self;
+}
+
+- (void)sanitize
+{
+//	for (HTMLToken *token in _tokenizer) {
+//		
+//	}
+}
+
+@end

Sources/HTMLSanitizingPolicy.h

@@ -0,0 +1,20 @@
+//
+//  HTMLSanitizingPolicy.h
+//  HTMLKit
+//
+//  Created by Iska on 26.05.18.
+//  Copyright © 2018 BrainCookie. All rights reserved.
+//
+
+#import <Foundation/Foundation.h>
+
+NS_ASSUME_NONNULL_BEGIN
+
+@interface HTMLSanitizingPolicy : NSObject
+
+- (HTMLSanitizingPolicy *)combineWith:(nullable HTMLSanitizingPolicy *)other;
+
+@end
+
+NS_ASSUME_NONNULL_END
+

Sources/HTMLSanitizingPolicy.m

@@ -0,0 +1,36 @@
+//
+//  HTMLSanitizingPolicy.m
+//  HTMLKit
+//
+//  Created by Iska on 26.05.18.
+//  Copyright © 2018 BrainCookie. All rights reserved.
+//
+
+#import "HTMLSanitizingPolicy.h"
+
+@interface HTMLSanitizingPolicy()
+{
+	NSMutableArray *_policies;
+}
+@end
+
+@implementation HTMLSanitizingPolicy
+
+- (instancetype)init
+{
+	self = [super init];
+	if (self) {
+		_policies = [NSMutableArray new];
+	}
+	return self;
+}
+
+- (HTMLSanitizingPolicy *)combineWith:(HTMLSanitizingPolicy *)other
+{
+	if (other) {
+		[_policies addObject:other];
+	}
+	return self;
+}
+
+@end

Sources/HTMLSanitizingPolicyBuilder.h

@@ -0,0 +1,39 @@
+//
+//  HTMLSanitizingPolicyBuilder.h
+//  HTMLKit
+//
+//  Created by Iska on 26.05.18.
+//  Copyright © 2018 BrainCookie. All rights reserved.
+//
+
+#import <Foundation/Foundation.h>
+
+#import "HTMLElementPolicy.h"
+#import "HTMLTokenPolicy.h"
+#import "HTMLAttributePolicy.h"
+
+NS_ASSUME_NONNULL_BEGIN
+
+@interface HTMLSanitizingPolicyBuilder : NSObject
+
+- (HTMLSanitizingPolicyBuilder *)allowElements:(NSArray<NSString *> *)elementNames;
+- (HTMLSanitizingPolicyBuilder *)disallowElements:(NSArray<NSString *> *)elementNames;
+- (HTMLSanitizingPolicyBuilder *)allowPolicy:(HTMLElementPolicy *)policy onElements:(NSArray<NSString *> *)elementNames;
+- (HTMLSanitizingPolicyBuilder *)allowCommonInlineFormattingElements;
+- (HTMLSanitizingPolicyBuilder *)allowCommonBlockElements;
+- (HTMLSanitizingPolicyBuilder *)allowTextInElements:(NSArray<NSString *> *)elementNames;
+- (HTMLSanitizingPolicyBuilder *)disallowTextInElements:(NSArray<NSString *> *)elementNames;
+
+//- (HTMLSanitizingPolicyBuilder *)allowAttributes:(NSArray<NSString *> *)attributeName
+//									  onElements:(NSArray<NSString *> *)elementNames;
+//- (HTMLSanitizingPolicyBuilder *)disallowAttributes:(NSArray<NSString *> *)attributeName
+//										 onElements:(NSArray<NSString *> *)elementNames;
+//
+//- (HTMLSanitizingPolicyBuilder *)allowAttributePolicy:(HTMLAttributePolicy *)policy
+//										   onElements:(NSArray<NSString *> *)elementNames;
+//- (HTMLSanitizingPolicyBuilder *)disallowAttributePolicy:(HTMLAttributePolicy *)policy
+//											  onElements:(NSArray<NSString *> *)elementNames;
+
+@end
+
+NS_ASSUME_NONNULL_END

Sources/HTMLSanitizingPolicyBuilder.m

@@ -0,0 +1,94 @@
+//
+//  HTMLSanitizingPolicyBuilder.m
+//  HTMLKit
+//
+//  Created by Iska on 26.05.18.
+//  Copyright © 2018 BrainCookie. All rights reserved.
+//
+
+#import "HTMLSanitizingPolicyBuilder.h"
+
+@interface HTMLSanitizingPolicyBuilder()
+{
+	NSMutableDictionary<NSString *, NSMutableArray<HTMLElementPolicy *> *> * elementPolicies;
+	NSMutableDictionary<NSString *, HTMLAttributePolicy *> * attributePolicies;
+	NSMutableDictionary<NSString *, NSNumber *> * textContainers;
+}
+@end
+
+@implementation HTMLSanitizingPolicyBuilder
+
+- (HTMLSanitizingPolicyBuilder *)allowElements:(NSArray<NSString *> *)elementNames
+{
+	return [self allowPolicy:HTMLElementPolicy.identity onElements:elementNames];
+}
+
+- (HTMLSanitizingPolicyBuilder *)disallowElements:(NSArray<NSString *> *)elementNames
+{
+	return [self allowPolicy:HTMLElementPolicy.rejectAll onElements:elementNames];
+}
+
+- (HTMLSanitizingPolicyBuilder *)allowPolicy:(HTMLElementPolicy *)policy onElements:(NSArray<NSString *> *)elementNames
+{
+	for (NSString *name in elementNames) {
+		NSMutableArray<HTMLElementPolicy *> *list = elementPolicies[name];
+		if (list == nil) {
+			list = [NSMutableArray new];
+		}
+		[list addObject:policy];
+		elementPolicies[name] = list;
+	}
+	return self;
+}
+
+- (HTMLSanitizingPolicyBuilder *)allowCommonInlineFormattingElements
+{
+	return [self allowElements:@[@"b", @"i", @"font", @"s", @"u", @"o", @"sup", @"sub", @"ins", @"del",
+								 @"strong", @"strike", @"tt", @"code", @"big", @"small", @"br", @"span", @"em"]];
+}
+
+- (HTMLSanitizingPolicyBuilder *)allowCommonBlockElements
+{
+	return [self allowElements:@[@"p", @"div", @"h1", @"h2", @"h3", @"h4", @"h5", @"h6", @"ul", @"ol", @"li",
+								 @"blockquote"]];
+}
+
+- (HTMLSanitizingPolicyBuilder *)allowTextInElements:(NSArray<NSString *> *)elementNames;
+{
+	for (NSString *name in elementNames) {
+		textContainers[name] = @YES;
+	}
+	return self;
+}
+
+- (HTMLSanitizingPolicyBuilder *)disallowTextInElements:(NSArray<NSString *> *)elementNames
+{
+	for (NSString *name in elementNames) {
+		textContainers[name] = @NO;
+	}
+	return self;
+}
+
+//- (HTMLSanitizingPolicyBuilder *)allowAttributes:(NSArray<NSString *> *)attributeNames
+//									  onElements:(NSArray<NSString *> *)elementNames
+//{
+//	for (NSString *attribute in attributeNames) {
+//		[self allowAttributePolicy:HTMLAttributePolicy.identity onElements:elementNames];
+//	}
+//	return self;
+//}
+//
+//- (HTMLSanitizingPolicyBuilder *)disallowAttributes:(NSArray<NSString *> *)attributeName
+//										 onElements:(NSArray<NSString *> *)elementNames;
+//
+//- (HTMLSanitizingPolicyBuilder *)allowAttributePolicy:(HTMLAttributePolicy *)policy onElements:(NSArray<NSString *> *)elementNames
+//{
+//	return self;
+//}
+//
+//- (HTMLSanitizingPolicyBuilder *)disallowAttributePolicy:(HTMLAttributePolicy *)policy onElements:(NSArray<NSString *> *)elementNames
+//{
+//	return self;
+//}
+
+@end

Sources/HTMLStackOfOpenElements.m

@@ -14,7 +14,6 @@
 @interface HTMLStackOfOpenElements ()
 {
 	NSMutableArray *_stack;
-	NSDictionary *_specificScopeElementTypes;
 }
 @end
 
@@ -27,26 +26,6 @@
 	self = [super init];
 	if (self) {
 		_stack = [NSMutableArray new];
-		_specificScopeElementTypes = @{
-									   @"applet": @(HTMLNamespaceHTML),
-									   @"caption": @(HTMLNamespaceHTML),
-									   @"html": @(HTMLNamespaceHTML),
-									   @"table": @(HTMLNamespaceHTML),
-									   @"td": @(HTMLNamespaceHTML),
-									   @"th": @(HTMLNamespaceHTML),
-									   @"marquee": @(HTMLNamespaceHTML),
-									   @"object": @(HTMLNamespaceHTML),
-									   @"template": @(HTMLNamespaceHTML),
-									   @"mi": @(HTMLNamespaceMathML),
-									   @"mo": @(HTMLNamespaceMathML),
-									   @"mn": @(HTMLNamespaceMathML),
-									   @"ms": @(HTMLNamespaceMathML),
-									   @"mtext": @(HTMLNamespaceMathML),
-									   @"annotation-xml": @(HTMLNamespaceMathML),
-									   @"foreignObject": @(HTMLNamespaceSVG),
-									   @"desc": @(HTMLNamespaceSVG),
-									   @"title": @(HTMLNamespaceSVG)
-									   };
 	}
 	return self;
 }
@@ -195,82 +174,148 @@
 
 #pragma mark - Element Scope
 
+NS_INLINE BOOL IsSpecificScopeElement(HTMLElement *element)
+{
+	switch (element.htmlNamespace) {
+		case HTMLNamespaceHTML:
+			return [element.tagName isEqualToAny:@"applet", @"caption", @"html", @"table", @"td", @"th", @"marquee", @"object", @"template", nil];
+		case HTMLNamespaceMathML:
+			return [element.tagName isEqualToAny:@"mi", @"mo", @"mn", @"ms", @"mtext", @"annotation-xml", nil];
+		case HTMLNamespaceSVG:
+			return [element.tagName isEqualToAny:@"foreignObject", @"desc", @"title", nil];
+	}
+}
+
+NS_INLINE BOOL IsHeaderElement(HTMLElement *element)
+{
+	if (element.htmlNamespace != HTMLNamespaceHTML) {
+		return NO;
+	}
+
+	return [element.tagName isEqualToAny:@"h1", @"h2", @"h3", @"h4", @"h5", @"h6", nil];
+}
+
+NS_INLINE BOOL IsTableScopeElement(HTMLElement *element)
+{
+	if (element.htmlNamespace != HTMLNamespaceHTML) {
+		return NO;
+	}
+
+	return [element.tagName isEqualToAny:@"html", @"table", @"template", nil];
+}
+
+NS_INLINE BOOL IsListItemScopeElement(HTMLElement *element)
+{
+	if (element.htmlNamespace != HTMLNamespaceHTML) {
+		return NO;
+	}
+
+	return [element.tagName isEqualToAny:@"ol", @"ul", nil];
+}
+
+NS_INLINE BOOL IsSelectScopeElement(HTMLElement *element)
+{
+	if (element.htmlNamespace != HTMLNamespaceHTML) {
+		return NO;
+	}
+
+	return ![element.tagName isEqualToString:@"optgroup"] && ![element.tagName isEqualToString:@"option"];
+}
+
+NS_INLINE BOOL IsButtonScopeElement(HTMLElement *element)
+{
+	if (element.htmlNamespace != HTMLNamespaceHTML) {
+		return NO;
+	}
+
+	return [element.tagName isEqualToString:@"button"];
+}
+
 - (HTMLElement *)hasElementInScopeWithTagName:(NSString *)tagName;
-{
-	return [self hasAnyElementInSpecificScopeWithTagNames:@[tagName] andElementTypes:_specificScopeElementTypes];
-}
-
-- (HTMLElement *)hasAnyElementInScopeWithAnyOfTagNames:(NSArray *)tagNames
-{
-	return [self hasAnyElementInSpecificScopeWithTagNames:tagNames andElementTypes:_specificScopeElementTypes];
-}
-
-- (HTMLElement *)hasElementInListItemScopeWithTagName:(NSString *)tagName
-{
-	NSMutableDictionary *elementTypes = [NSMutableDictionary dictionaryWithDictionary:_specificScopeElementTypes];
-	[elementTypes addEntriesFromDictionary:@{@"ol": @(HTMLNamespaceHTML),
-											 @"ul": @(HTMLNamespaceHTML)}];
-
-	return [self hasElementInSpecificScopeWithTagName:tagName
-									  andElementTypes:elementTypes];
-}
-
-- (HTMLElement *)hasElementInButtonScopeWithTagName:(NSString *)tagName
-{
-	NSMutableDictionary *elementTypes = [NSMutableDictionary dictionaryWithDictionary:_specificScopeElementTypes];
-	[elementTypes addEntriesFromDictionary:@{@"button": @(HTMLNamespaceHTML)}];
-
-	return [self hasElementInSpecificScopeWithTagName:tagName
-									  andElementTypes:elementTypes];
-}
-
-- (HTMLElement *)hasElementInTableScopeWithTagName:(NSString *)tagName
-{
-	return [self hasElementInSpecificScopeWithTagName:tagName
-									  andElementTypes:@{@"html": @(HTMLNamespaceHTML),
-														@"table": @(HTMLNamespaceHTML),
-														@"template": @(HTMLNamespaceHTML)}];
-}
-
-- (HTMLElement *)hasElementInTableScopeWithAnyOfTagNames:(NSArray *)tagNames
-{
-	return [self hasAnyElementInSpecificScopeWithTagNames:tagNames
-										  andElementTypes:@{@"html": @(HTMLNamespaceHTML),
-															@"table": @(HTMLNamespaceHTML),
-															@"template": @(HTMLNamespaceHTML)}];
-}
-
-- (HTMLElement *)hasElementInSelectScopeWithTagName:(NSString *)tagName
 {
 	for (HTMLElement *node in _stack.reverseObjectEnumerator) {
-		if ([node.tagName isEqualToString:tagName]) {
+		if (node.htmlNamespace == HTMLNamespaceHTML && [tagName isEqualToString:node.tagName]) {
 			return node;
 		}
-		if (!(node.htmlNamespace == HTMLNamespaceHTML &&
-			  [node.tagName isEqualToAny:@"optgroup", @"option", nil])) {
+		if (IsSpecificScopeElement(node)) {
 			return nil;
 		}
 	}
 	return nil;
 }
 
-- (HTMLElement *)hasElementInSpecificScopeWithTagName:(NSString *)tagName
-									  andElementTypes:(NSDictionary *)elementTypes
-{
-	return [self hasAnyElementInSpecificScopeWithTagNames:@[tagName] andElementTypes:elementTypes];
-}
-
-- (HTMLElement *)hasAnyElementInSpecificScopeWithTagNames:(NSArray *)tagNames
-										  andElementTypes:(NSDictionary *)elementTypes
+- (HTMLElement *)hasHeaderElementInScope
 {
 	for (HTMLElement *node in _stack.reverseObjectEnumerator) {
-		if ([tagNames containsObject:node.tagName]) {
-			NSNumber *namespace = elementTypes[node.tagName] ?: @(HTMLNamespaceHTML);
-			if ([namespace isEqual:@(node.htmlNamespace)]) {
-				return node;
-			}
+		if (IsHeaderElement(node)) {
+			return node;
 		}
-		if ([elementTypes[node.tagName] isEqual:@(node.htmlNamespace)]) {
+		if (IsSpecificScopeElement(node)) {
+			return nil;
+		}
+	}
+	return nil;
+}
+
+- (HTMLElement *)hasElementInTableScopeWithTagName:(NSString *)tagName
+{
+	for (HTMLElement *node in _stack.reverseObjectEnumerator) {
+		if (node.htmlNamespace == HTMLNamespaceHTML && [tagName isEqualToString:node.tagName]) {
+			return node;
+		}
+		if (IsTableScopeElement(node)) {
+			return nil;
+		}
+	}
+	return nil;
+}
+
+- (HTMLElement *)hasElementInTableScopeWithAnyOfTagNames:(NSArray *)tagNames
+{
+	for (HTMLElement *node in _stack.reverseObjectEnumerator) {
+		if (node.htmlNamespace == HTMLNamespaceHTML && [tagNames containsObject:node.tagName]) {
+			return node;
+		}
+		if (IsTableScopeElement(node)) {
+			return nil;
+		}
+	}
+	return nil;
+}
+
+- (HTMLElement *)hasElementInListItemScopeWithTagName:(NSString *)tagName
+{
+	for (HTMLElement *node in _stack.reverseObjectEnumerator) {
+		if (node.htmlNamespace == HTMLNamespaceHTML && [tagName isEqualToString:node.tagName]) {
+			return node;
+		}
+		if (IsSpecificScopeElement(node) || IsListItemScopeElement(node)) {
+			return nil;
+		}
+	}
+	return nil;
+}
+
+- (HTMLElement *)hasElementInButtonScopeWithTagName:(NSString *)tagName
+{
+	for (HTMLElement *node in _stack.reverseObjectEnumerator) {
+		if (node.htmlNamespace == HTMLNamespaceHTML && [tagName isEqualToString:node.tagName]) {
+			return node;
+		}
+		if (IsSpecificScopeElement(node) || IsButtonScopeElement(node)) {
+			return nil;
+		}
+	}
+	return nil;
+}
+
+- (HTMLElement *)hasElementInSelectScopeWithTagName:(NSString *)tagName
+{
+	for (HTMLElement *node in _stack.reverseObjectEnumerator) {
+		if (node.htmlNamespace == HTMLNamespaceHTML && [tagName isEqualToString:node.tagName]) {
+			return node;
+		}
+		if (IsSelectScopeElement(node)) {
 			return nil;
 		}
 	}

Sources/HTMLTemplate.m

@@ -8,10 +8,7 @@
 
 #import "HTMLTemplate.h"
 #import "HTMLDocument.h"
-
-@interface HTMLNode (Private)
-@property (nonatomic, weak) HTMLDocument *ownerDocument;
-@end
+#import "HTMLNode+Private.h"
 
 @implementation HTMLTemplate
 

Sources/HTMLText.m

@@ -9,7 +9,9 @@
 #import "HTMLText.h"
 #import "HTMLElement.h"
 #import "NSString+HTMLKit.h"
-#import "HTMLNode+Private.h"
+#import "HTMLCharacterData+Private.h"
+#import "HTMLKitDOMExceptions.h"
+#import "HTMLDocument+Private.h"
 
 @implementation HTMLText
 
@@ -20,35 +22,48 @@
 
 - (instancetype)initWithData:(NSString *)data
 {
-	self = [super initWithName:@"#text" type:HTMLNodeText];
-	if (self) {
-		_data = [[NSMutableString alloc] initWithString:data ?: @""];
-	}
-	return self;
-}
-
-- (NSString *)textContent
-{
-	return [self.data copy];
-}
-
-- (void)setTextContent:(NSString *)textContent
-{
-	[self.data setString:textContent ?: @""];
+	return [super initWithName:@"#text" type:HTMLNodeText data:data];
 }
 
 - (void)appendString:(NSString *)string
 {
-	[self.data appendString:string];
+	[self appendData:string];
 }
 
-#pragma mark - NSCopying
-
-- (id)copyWithZone:(NSZone *)zone
+NS_INLINE void CheckValidOffset(HTMLNode *node, NSUInteger offset, NSString *cmd)
 {
-	HTMLText *copy = [super copyWithZone:zone];
-	copy.data = self.data;
-	return copy;
+	if (offset > node.length) {
+		[NSException raise:HTMLKitIndexSizeError
+					format:@"%@: Index Size Error, invalid offset %lu for splitting text node %@.",
+		 cmd, (unsigned long)offset, node];
+	}
+}
+
+- (HTMLText *)splitTextAtOffset:(NSUInteger)offset
+{
+	CheckValidOffset(self, offset, NSStringFromSelector(_cmd));
+
+	NSUInteger length = self.length;
+	NSUInteger count = length - offset;
+	NSRange range = NSMakeRange(offset, count);
+
+	NSString *newData = [self.data substringWithRange:range];
+	HTMLText *newNode = [[HTMLText alloc] initWithData:newData];
+	[self.ownerDocument adoptNode:newNode];
+
+	HTMLNode *parent = self.parentNode;
+	if (parent != nil) {
+		[parent insertNode:newNode beforeChildNode:self.nextSibling];
+		[self.ownerDocument didInsertNewTextNode:newNode intoParent:parent afterSplittingTextNode:self atOffset:offset];
+	}
+
+	[self deleteDataInRange:range];
+
+	if (parent != nil) {
+		[self.ownerDocument clampRangesAfterSplittingTextNode:self atOffset:offset];
+	}
+
+	return newNode;
 }
 
 #pragma mark - Serialization
@@ -59,12 +74,11 @@
 		 @"plaintext", @"noscript", nil]) {
 		return self.data;
 	} else {
-		NSRange range = NSMakeRange(0, self.data.length);
 		NSMutableString *escaped = [self.data mutableCopy];
-		[escaped replaceOccurrencesOfString:@"&" withString:@"&" options:0 range:range];
-		[escaped replaceOccurrencesOfString:@"\00A0" withString:@" " options:0 range:range];
-		[escaped replaceOccurrencesOfString:@"<" withString:@"&lt;" options:0 range:range];
-		[escaped replaceOccurrencesOfString:@">" withString:@"&gt;" options:0 range:range];
+		[escaped replaceOccurrencesOfString:@"&" withString:@"&amp;" options:0 range:NSMakeRange(0, escaped.length)];
+		[escaped replaceOccurrencesOfString:@"\00A0" withString:@"&nbsp;" options:0 range:NSMakeRange(0, escaped.length)];
+		[escaped replaceOccurrencesOfString:@"<" withString:@"&lt;" options:0 range:NSMakeRange(0, escaped.length)];
+		[escaped replaceOccurrencesOfString:@">" withString:@"&gt;" options:0 range:NSMakeRange(0, escaped.length)];
 		return escaped;
 	}
 }

Sources/HTMLTokenPolicy.h

@@ -0,0 +1,22 @@
+//
+//  HTMLTokenPolicy.h
+//  HTMLKit
+//
+//  Created by Iska on 01.06.18.
+//  Copyright © 2018 BrainCookie. All rights reserved.
+//
+
+#import <Foundation/Foundation.h>
+#import "HTMLTokens.h"
+
+NS_ASSUME_NONNULL_BEGIN
+
+@interface HTMLTokenPolicy : NSObject
+
++ (instancetype)policy:(HTMLToken * _Nullable (^)(HTMLToken *))block;
+
+- (nullable HTMLToken *)apply:(HTMLToken *)token;
+
+@end
+
+NS_ASSUME_NONNULL_END

Sources/HTMLTokenPolicy.m

@@ -0,0 +1,38 @@
+//
+//  HTMLTokenPolicy.m
+//  HTMLKit
+//
+//  Created by Iska on 01.06.18.
+//  Copyright © 2018 BrainCookie. All rights reserved.
+//
+
+#import "HTMLTokenPolicy.h"
+
+@interface HTMLTokenPolicy()
+{
+	HTMLToken * (^ _policyBlock)(HTMLToken *);
+}
+@end
+
+@implementation HTMLTokenPolicy
+
++ (instancetype)policy:(HTMLToken * _Nullable (^)(HTMLToken * _Nonnull))block
+{
+	return [[HTMLTokenPolicy alloc] initWithBlock:block];
+}
+
+- (instancetype)initWithBlock:(HTMLToken * _Nullable (^)(HTMLToken * _Nonnull))block
+{
+	self = [super init];
+	if (self) {
+		_policyBlock = block;
+	}
+	return self;
+}
+
+- (HTMLToken *)apply:(HTMLToken *)token
+{
+	return _policyBlock(token);
+}
+
+@end

Sources/NSString+HTMLKit.m

@@ -58,4 +58,15 @@ NS_INLINE BOOL isHtmlWhitespaceChar(unichar c)
 	return idx;
 }
 
+- (NSString *)stringByEscapingForHTML
+{
+	NSMutableString *escaped = [self mutableCopy];
+	[escaped replaceOccurrencesOfString:@"&" withString:@"&" options:0 range:NSMakeRange(0, escaped.length)];
+	[escaped replaceOccurrencesOfString:@"0x00A0" withString:@" " options:0 range:NSMakeRange(0, escaped.length)];
+	[escaped replaceOccurrencesOfString:@"\"" withString:@""" options:0 range:NSMakeRange(0, escaped.length)];
+	[escaped replaceOccurrencesOfString:@"<" withString:@"&lt;" options:0 range:NSMakeRange(0, escaped.length)];
+	[escaped replaceOccurrencesOfString:@">" withString:@"&gt;" options:0 range:NSMakeRange(0, escaped.length)];
+	return escaped;
+}
+
 @end

Sources/include/CSSPseudoFunctionSelector.h

@@ -26,7 +26,7 @@ NS_ASSUME_NONNULL_BEGIN
 /**
  Initializes and returns a CSS has-descendant selector, e.g. 'div:has(p)'
 
- @discussion 'div:has(p)' matches all <div> elements which have a descendant <p> element.
+ @discussion 'div:has(p)' matches all &lt;div&gt; elements which have a descendant &lt;p&gt; element.
 
  @param selector The selector matching a descendant element.
  @return A new instance of the has-descendant selector.

Sources/include/CSSSelector.h

@@ -76,7 +76,7 @@ extern NSString * _Nonnull NSStringFromNthExpression(CSSNthExpression expression
  @param string The selector string which will be parsed.
  @return A new instance of a parsed CSS Selector, `nil` if the string is not a valid selector string.
  */
-+ (nullable instancetype)selectorWithString:(NSString *)stirng;
++ (nullable instancetype)selectorWithString:(NSString *)string;
 
 /**
  Implementations should override this method to provide the selector-sprecific logic for matching elements.

Sources/include/CSSSelectors.h

@@ -26,7 +26,7 @@ NS_ASSUME_NONNULL_BEGIN
 
  @return The universal CSS selector.
  */
-extern CSSSelector * universalSelector();
+extern CSSSelector * universalSelector(void);
 
 /**
  CSS type selector, e.g. 'div', 'p', ...etc.
@@ -125,7 +125,7 @@ extern CSSSelector * nthLastOfTypeSelector(CSSNthExpression expression);
 
  @return Odd-Child selector.
  */
-extern CSSSelector * oddSelector();
+extern CSSSelector * oddSelector(void);
 
 /**
  CSS even-child selector: ':nth-child(even)'
@@ -134,49 +134,49 @@ extern CSSSelector * oddSelector();
 
  @return Even-Child selector.
  */
-extern CSSSelector * evenSlector();
+extern CSSSelector * evenSlector(void);
 
 /**
  CSS first-child selector: ':nth-child(1)'
 
  @return First-Child selector.
  */
-extern CSSSelector * firstChildSelector();
+extern CSSSelector * firstChildSelector(void);
 
 /**
  CSS first-child selector: ':nth-last-child(1)'
 
  @return First-Child selector.
  */
-extern CSSSelector * lastChildSelector();
+extern CSSSelector * lastChildSelector(void);
 
 /**
  CSS first-of-type selector: ':nth-first-of-type(1)'
 
  @return First-Of-Type selector.
  */
-extern CSSSelector * firstOfTypeSelector();
+extern CSSSelector * firstOfTypeSelector(void);
 
 /**
  CSS last-of-type selector: ':nth-last-of-type(1)'
 
  @return Last-Of-Type selector.
  */
-extern CSSSelector * lastOfTypeSelector();
+extern CSSSelector * lastOfTypeSelector(void);
 
 /**
  CSS only-child selector: ':first-child:last-child'
 
  @return Only-Child selector.
  */
-extern CSSSelector * onlyChildSelector();
+extern CSSSelector * onlyChildSelector(void);
 
 /**
  CSS only-of-type selector: ':first-of-type:last-of-type'
 
  @return Only-Of-Type selector.
  */
-extern CSSSelector * onlyOfTypeSelector();
+extern CSSSelector * onlyOfTypeSelector(void);
 
 #pragma mark - Combinators
 
@@ -225,7 +225,7 @@ extern CSSSelector * not(CSSSelector *selector);
 /**
  CSS has-descendant selector, e.g. 'div:has(p)'
 
- @discussion 'div:has(p)' matches all <div> elements which have a descendant <p> element.
+ @discussion 'div:has(p)' matches all &lt;div&gt; elements which have a descendant &lt;p&gt; element.
 
  @param selector The selector matching a descendant element.
  @return A has-descendant selector.

Sources/include/CSSStructuralPseudoSelectors.h

@@ -15,102 +15,102 @@ NS_ASSUME_NONNULL_BEGIN
 /**
  @return Root element selector: ':root'
  */
-extern CSSSelector * rootSelector();
+extern CSSSelector * rootSelector(void);
 
 /**
  @return Empy element selector: ':empty'
  */
-extern CSSSelector * emptySelector();
+extern CSSSelector * emptySelector(void);
 
 /**
  @return A parent element selector: ':parent'
  */
-extern CSSSelector * parentSelector();
+extern CSSSelector * parentSelector(void);
 
 /**
  @return A button element selector: ':button'
  */
-extern CSSSelector * buttonSelector();
+extern CSSSelector * buttonSelector(void);
 
 /**
  @return A checkbox element selector: ':checkbox'
  */
-extern CSSSelector * checkboxSelector();
+extern CSSSelector * checkboxSelector(void);
 
 /**
  @return A file element selector: ':file'
  */
-extern CSSSelector * fileSelector();
+extern CSSSelector * fileSelector(void);
 
 /**
  @return A header element selector: ':header'
  */
-extern CSSSelector * headerSelector();
+extern CSSSelector * headerSelector(void);
 
 /**
  @return An image element selector: ':image'
  */
-extern CSSSelector * imageSelector();
+extern CSSSelector * imageSelector(void);
 
 /**
  @return A parent element selector: ':parent'
  */
-extern CSSSelector * inputSelector();
+extern CSSSelector * inputSelector(void);
 
 /**
  @return A link element selector: ':link'
  */
-extern CSSSelector * linkSelector();
+extern CSSSelector * linkSelector(void);
 
 /**
  @return A password element selector: ':password'
  */
-extern CSSSelector * passwordSelector();
+extern CSSSelector * passwordSelector(void);
 
 /**
  @return A radio element selector: ':radio'
  */
-extern CSSSelector * radioSelector();
+extern CSSSelector * radioSelector(void);
 
 /**
  @return A reset element selector: ':reset'
  */
-extern CSSSelector * resetSelector();
+extern CSSSelector * resetSelector(void);
 
 /**
  @return A submit element selector: ':submit'
  */
-extern CSSSelector * submitSelector();
+extern CSSSelector * submitSelector(void);
 
 /**
  @return A text element selector: ':text'
  */
-extern CSSSelector * textSelector();
+extern CSSSelector * textSelector(void);
 
 /**
  @return An enabled element selector: ':enabled'
  */
-extern CSSSelector * enabledSelector();
+extern CSSSelector * enabledSelector(void);
 
 /**
  @return A disabled element selector: ':disabled'
  */
-extern CSSSelector * disabledSelector();
+extern CSSSelector * disabledSelector(void);
 
 /**
  @return A checked element selector: ':checked'
  */
-extern CSSSelector * checkedSelector();
+extern CSSSelector * checkedSelector(void);
 
 /**
  @return An optional element selector: ':optional'
  */
-extern CSSSelector * optionalSelector();
+extern CSSSelector * optionalSelector(void);
 
 /**
  @return A required element selector: ':required'
  */
-extern CSSSelector * requiredSelector();
+extern CSSSelector * requiredSelector(void);
 
 /**
  Less-than selector, e.g. 'lt(2)'

Sources/include/HTMLCharacterData+Private.h

@@ -0,0 +1,28 @@
+//
+//  HTMLCharacterData+Private.h
+//  HTMLKit
+//
+//  Created by Iska on 26/11/16.
+//  Copyright © 2016 BrainCookie. All rights reserved.
+//
+
+#import "HTMLCharacterData.h"
+#import "HTMLNode.h"
+
+/**
+ Private HTML Character Data methods which are not intended for public API.
+ */
+@interface HTMLCharacterData ()
+
+/**
+ Designated initializer of the HTML CharacterData, which, however, should not be used directly. It is intended to be 
+ called only by subclasses, i.e. HTMLText and HTMLComment.
+
+ @param name The node's name.
+ @param type The node's type.
+ @param data The node's data string.
+ @return A new instance of a HTML CharacterData.
+ */
+- (instancetype)initWithName:(NSString *)name type:(HTMLNodeType)type data:(NSString *)data NS_DESIGNATED_INITIALIZER;
+
+@end

Sources/include/HTMLCharacterData.h

@@ -0,0 +1,32 @@
+//
+//  HTMLCharacterData.h
+//  HTMLKit
+//
+//  Created by Iska on 26/11/16.
+//  Copyright © 2016 BrainCookie. All rights reserved.
+//
+
+#import "HTMLNode.h"
+
+NS_ASSUME_NONNULL_BEGIN
+
+/**
+ A HTML CharacterData
+ 
+ https://dom.spec.whatwg.org/#characterdata
+ */
+@interface HTMLCharacterData : HTMLNode
+
+/** @brief The associated data string. */
+@property (nonatomic, copy, readonly) NSString *data;
+
+- (void)setData:(NSString *)data;
+- (void)appendData:(NSString *)data;
+- (void)insertData:(NSString *)data atOffset:(NSUInteger)offset;
+- (void)deleteDataInRange:(NSRange)range;
+- (void)replaceDataInRange:(NSRange)range withData:(NSString *)data;
+- (NSString *)substringDataWithRange:(NSRange)range;
+
+@end
+
+NS_ASSUME_NONNULL_END

Sources/include/HTMLCharacterToken.h

@@ -6,10 +6,6 @@
 //  Copyright (c) 2014 BrainCookie. All rights reserved.
 //
 
-///------------------------------------------------------
-/// HTMLKit private header
-///------------------------------------------------------
-
 #import <Foundation/Foundation.h>
 #import "HTMLToken.h"
 

Sources/include/HTMLComment.h

@@ -6,17 +6,14 @@
 //  Copyright (c) 2015 BrainCookie. All rights reserved.
 //
 
-#import "HTMLNode.h"
+#import "HTMLCharacterData.h"
 
 NS_ASSUME_NONNULL_BEGIN
 
 /**
  A HTML Comment node
  */
-@interface HTMLComment : HTMLNode
-
-/** @brief The comment string. */
-@property (nonatomic, copy) NSString *data;
+@interface HTMLComment : HTMLCharacterData
 
 /**
  Initializes a new HTML comment node.

Sources/include/HTMLCommentToken.h

@@ -6,10 +6,6 @@
 //  Copyright (c) 2014 BrainCookie. All rights reserved.
 //
 
-///------------------------------------------------------
-/// HTMLKit private header
-///------------------------------------------------------
-
 #import <Foundation/Foundation.h>
 #import "HTMLToken.h"
 
@@ -24,7 +20,7 @@
 /**
  Initializes a new comment token.
 
- @param string The string with which to initialize the token.
+ @param data The string with which to initialize the token.
  @return A new instance of a comment token.
  */
 - (instancetype)initWithData:(NSString *)data;

Sources/include/HTMLDOCTYPEToken.h

@@ -6,10 +6,6 @@
 //  Copyright (c) 2014 BrainCookie. All rights reserved.
 //
 
-///------------------------------------------------------
-/// HTMLKit private header
-///------------------------------------------------------
-
 #import <Foundation/Foundation.h>
 #import "HTMLToken.h"
 

Sources/include/HTMLDOM.h

@@ -11,9 +11,11 @@
 #import "HTMLDocumentType.h"
 #import "HTMLDocumentFragment.h"
 #import "HTMLElement.h"
+#import "HTMLCharacterData.h"
 #import "HTMLComment.h"
 #import "HTMLText.h"
 #import "HTMLTemplate.h"
+#import "HTMLRange.h"
 #import "HTMLDOMTokenList.h"
 #import "HTMLNodeIterator.h"
 #import "HTMLTreeWalker.h"
@@ -22,3 +24,5 @@
 #import "HTMLKitDOMExceptions.h"
 #import "HTMLNamespaces.h"
 #import "HTMLQuirksMode.h"
+
+#import "HTMLOrderedDictionary.h"

Sources/include/HTMLDOMUtils.h

@@ -0,0 +1,15 @@
+//
+//  HTMLDOMUtils.h
+//  HTMLKit
+//
+//  Created by Iska on 03/12/16.
+//  Copyright © 2016 BrainCookie. All rights reserved.
+//
+
+#import <Foundation/Foundation.h>
+#import "HTMLDOM.h"
+
+@class HTMLNode;
+
+extern HTMLNode * GetCommonAncestorContainer(HTMLNode *nodeA, HTMLNode *nodeB);
+extern NSArray<HTMLNode *> * GetAncestorNodes(HTMLNode *node);

Sources/include/HTMLDocument+Private.h

@@ -0,0 +1,99 @@
+//
+//  HTMLDocument+Private.h
+//  HTMLKit
+//
+//  Created by Iska on 27/11/16.
+//  Copyright © 2016 BrainCookie. All rights reserved.
+//
+
+#import "HTMLDocument.h"
+#import "HTMLNode.h"
+#import "HTMLCharacterData.h"
+#import "HTMLNodeIterator.h"
+#import "HTMLRange.h"
+#import "HTMLText.h"
+
+/**
+ Private HTML Document methods which are not intended for public API.
+ */
+@interface HTMLDocument ()
+
+@property (nonatomic, assign) HTMLDocumentReadyState readyState;
+
+/**
+ Runs the necessary steps after removing a node from the DOM.
+
+ @param oldNode The old node that was removed.
+ @param oldParent The old parent of the node that was removed.
+ @param oldPreviousSibling The old previous sibling node of the node that was removed.
+ */
+- (void)runRemovingStepsForNode:(HTMLNode *)oldNode
+				  withOldParent:(HTMLNode *)oldParent
+		  andOldPreviousSibling:(HTMLNode *)oldPreviousSibling;
+
+/**
+ Attaches a node iterator to this document.
+
+ @param iterator The iterator to attach.
+ */
+- (void)attachNodeIterator:(HTMLNodeIterator *)iterator;
+
+/**
+ Detaches a node interator from this document.
+
+ @param iterator The iterator to detach.
+ */
+- (void)detachNodeIterator:(HTMLNodeIterator *)iterator;
+
+
+/**
+ Attaches a range to this document.
+
+ @param range The range to attach.
+ */
+- (void)attachRange:(HTMLRange *)range;
+
+/**
+ Detaches a range from this document.
+
+ @param range The range to detach.
+ */
+- (void)detachRange:(HTMLRange *)range;
+
+/**
+ Callback on removing text from a CharacterData node.
+
+ @param node The CharacterData node.
+ @param offset The offset at which the data was removed.
+ @param length The length of the data that was removed.
+ */
+- (void)didRemoveCharacterDataInNode:(HTMLCharacterData *)node atOffset:(NSUInteger)offset withLength:(NSUInteger)length;
+
+/**
+ Callback on adding text from a CharacterData node.
+
+ @param node The CharacterData node.
+ @param offset The offset at which the data was added.
+ @param length The length of the data that was added.
+ */
+- (void)didAddCharacterDataToNode:(HTMLCharacterData *)node atOffset:(NSUInteger)offset withLength:(NSUInteger)length;
+
+/**
+ Callback on inserting a new text node when an old text node is split.
+
+ @param newNode The new text node after splitting.
+ @param parent The parent where newNode was inserted.
+ @param node The old text node that was split.
+ @param offset The offset of splitting.
+ */
+- (void)didInsertNewTextNode:(HTMLText *)newNode intoParent:(HTMLNode *)parent afterSplittingTextNode:(HTMLText *)node atOffset:(NSUInteger)offset;
+
+/**
+ Callback for clamping current ranges whose end boundary is after the text node upon splitting it.
+
+ @param node The text node that was split.
+ @param offset The offset of splitting
+ */
+- (void)clampRangesAfterSplittingTextNode:(HTMLText *)node atOffset:(NSUInteger)offset;
+
+@end

Sources/include/HTMLEOFToken.h

@@ -6,10 +6,6 @@
 //  Copyright (c) 2015 BrainCookie. All rights reserved.
 //
 
-///------------------------------------------------------
-/// HTMLKit private header
-///------------------------------------------------------
-
 #import "HTMLToken.h"
 
 /**

Sources/include/HTMLElement.h

@@ -63,7 +63,7 @@ NS_ASSUME_NONNULL_BEGIN
 /**
  Initializes a new HTML element with the given tag name.
  
- @param tagname The tag name.
+ @param tagName The tag name.
  @return A new HTML element.
  */
 - (instancetype)initWithTagName:(NSString *)tagName;
@@ -71,21 +71,21 @@ NS_ASSUME_NONNULL_BEGIN
 /**
  Initializes a new HTML element with the given tag name and attributes.
 
- @param tagname The tag name.
+ @param tagName The tag name.
  @param attributes The attributes.
  @return A new HTML element.
  */
-- (instancetype)initWithTagName:(NSString *)tagName attributes:(NSDictionary<NSString *, NSString *> *)attributes;
+- (instancetype)initWithTagName:(NSString *)tagName attributes:(nullable NSDictionary<NSString *, NSString *> *)attributes;
 
 /**
  Initializes a new HTML element with the given tag name, namespace, and attributes.
 
- @param tagname The tag name.
- @param namespace The namespace.
+ @param tagName The tag name.
+ @param htmlNamespace The HTML namespace.
  @param attributes The attributes.
  @return A new HTML element.
  */
-- (instancetype)initWithTagName:(NSString *)tagName namespace:(HTMLNamespace)htmlNamespace attributes:(NSDictionary<NSString *, NSString *> *)attributes;
+- (instancetype)initWithTagName:(NSString *)tagName namespace:(HTMLNamespace)htmlNamespace attributes:(nullable NSDictionary<NSString *, NSString *> *)attributes;
 
 /**
  Checks whether this element has an attribute with the given name.
@@ -107,7 +107,7 @@ NS_ASSUME_NONNULL_BEGIN
  Set the value of the attribute with the given name.
 
  @param value The value to set.
- @param name The attribute's name.
+ @param attribute The attribute's name.
  */
 - (void)setObject:(NSString *)value forKeyedSubscript:(NSString *)attribute;
 

Sources/include/HTMLElementAdjustment.h

@@ -140,31 +140,3 @@ NS_INLINE void AdjustSVGNameCase(HTMLTagToken *token)
 	NSString *replacement = replacements[token.tagName] ?: token.tagName;
 	token.tagName = replacement;
 }
-
-NS_INLINE void AdjustForeignAttributes(HTMLTagToken *token)
-{
-	if (token.attributes == nil) {
-		return;
-	}
-
-	NSDictionary *replacements = @{ @"xlink:actuate": @"xlink actuate",
-									@"xlink:arcrole": @"xlink arcrole",
-									@"xlink:href": @"xlink href",
-									@"xlink:role": @"xlink role",
-									@"xlink:show": @"xlink show",
-									@"xlink:title": @"xlink title",
-									@"xlink:type": @"xlink type",
-									@"xml:base": @"xml base",
-									@"xml:lang": @"xml lang",
-									@"xml:space": @"xml space",
-									@"xmlns": @"xmlns",
-									@"xmlns:xlink": @"xmlns xlink"};
-
-	HTMLOrderedDictionary *adjusted = [HTMLOrderedDictionary new];
-	for (id key in token.attributes) {
-		NSString *replacement = replacements[key] ?: key;
-		adjusted[replacement] = token.attributes[key];
-	}
-	token.attributes = adjusted;
-}
-

Sources/include/HTMLInputStreamReader.h

@@ -15,9 +15,10 @@
 /** 
  Typedef for the error callback block.
 
- @param reason The string describing the reason of the reported error.
+ @param code The standarized error-code
+ @param details The string describing the reason of the reported error.
  */
-typedef void (^ HTMLStreamReaderErrorCallback)(NSString *reason);
+typedef void (^ HTMLStreamReaderErrorCallback)(NSString *code, NSString *details);
 
 /**
  * HTML Input Stream Reader processor conforming to the HTML standard

Sources/include/HTMLKit.h

@@ -16,6 +16,8 @@ extern const unsigned char HTMLKitVersionString[];
 
 #import "HTMLDOM.h"
 #import "HTMLParser.h"
+#import "HTMLSanitizer.h"
+
 #import "HTMLKitErrorDomain.h"
 #import "HTMLOrderedDictionary.h"
 

Sources/include/HTMLKitDOMExceptions.h

@@ -14,3 +14,8 @@ extern NSString * const HTMLKitNotSupportedError;
 
 extern NSString * const HTMLKitSyntaxError;
 extern NSString * const HTMLKitInvalidCharacterError;
+
+extern NSString * const HTMLKitInvalidNodeTypeError;
+extern NSString * const HTMLKitIndexSizeError;
+extern NSString * const HTMLKitWrongDocumentError;
+extern NSString * const HTMLKitInvalidStateError;

Sources/include/HTMLNode.h

@@ -28,7 +28,7 @@ typedef NS_ENUM(short, HTMLNodeType)
 /**
  A node's position in the HTML document when compared with other nodes.
  */
-typedef NS_ENUM(unsigned short, HTMLDocumentPosition)
+typedef NS_OPTIONS(unsigned short, HTMLDocumentPosition)
 {
 	HTMLDocumentPositionEquivalent = 0x0,
 	HTMLDocumentPositionDisconnected = 0x01,
@@ -71,6 +71,11 @@ typedef NS_ENUM(unsigned short, HTMLDocumentPosition)
  */
 @property (nonatomic, weak, readonly, nullable) HTMLDocument *ownerDocument;
 
+/**
+ The root node of this node, if any.
+ */
+@property (nonatomic, weak, readonly, nullable) HTMLNode *rootNode;
+
 /**
  The parent node of this node, if any.
  */
@@ -122,6 +127,11 @@ typedef NS_ENUM(unsigned short, HTMLDocumentPosition)
  */
 @property (nonatomic, strong, readonly, nullable) HTMLElement *nextSiblingElement;
 
+/**
+ The index of this node.
+ */
+@property (nonatomic, readonly, assign) NSUInteger index;
+
 /**
  The text content of this node.
  */
@@ -137,6 +147,11 @@ typedef NS_ENUM(unsigned short, HTMLDocumentPosition)
  */
 @property (nonatomic, copy)	NSString *innerHTML;
 
+/**
+ The length of the node as described in https://dom.spec.whatwg.org/#concept-node-length
+ */
+@property (nonatomic, assign) NSUInteger length;
+
 /**
  @abstract Use concrete subclasses of the HTML Node.
  */
@@ -164,6 +179,21 @@ typedef NS_ENUM(unsigned short, HTMLDocumentPosition)
  */
 - (NSUInteger)childNodesCount;
 
+/**
+ Checks whether the node is empty as described in https://dom.spec.whatwg.org/#concept-node-length
+
+ @return `YES` if the node is empty, `NO` otherwise.
+ */
+- (BOOL)isEmpty;
+
+/**
+ Clones this node.
+
+ @param deep If `YES` then also clones child nodes. Otherwise a shallow clone is returned, which behaves the same as `copy`.
+ @return A clone of this node.
+ */
+- (instancetype)cloneNodeDeep:(BOOL)deep;
+
 /**
  Returns the child node at a given index.
 
@@ -202,7 +232,7 @@ typedef NS_ENUM(unsigned short, HTMLDocumentPosition)
 /**
  Returns the index of the given child element in the set of child nodes.
 
- @param node The element.
+ @param element The element.
  @return The index of the given element in the children set.
  */
 - (NSUInteger)indexOfChildElement:(HTMLElement *)element;
@@ -218,7 +248,7 @@ typedef NS_ENUM(unsigned short, HTMLDocumentPosition)
 /**
  Prepends the given array of nodes to the set of child nodes.
 
- @param node The nodes to prepend.
+ @param nodes The nodes to prepend.
  */
 - (void)prependNodes:(NSArray<HTMLNode *> *)nodes;
 
@@ -318,7 +348,8 @@ typedef NS_ENUM(unsigned short, HTMLDocumentPosition)
 - (BOOL)isDescendantOfNode:(HTMLNode *)node;
 
 /**
- Checks whether this node contains the given node.
+ Checks whether this node contains the given node. This performs an `invlusive ancestor` check, i.e. it returns `YES`
+ if the given node is the same object as this node.
 
  @param node The node to check.
  @return `YES` if this node contains the given node, `NO` otherwsie.

Sources/include/HTMLNodeIterator+Private.h

@@ -0,0 +1,28 @@
+//
+//  HTMLNodeIterator+Private.h
+//  HTMLKit
+//
+//  Created by Iska on 27/11/16.
+//  Copyright © 2016 BrainCookie. All rights reserved.
+//
+
+#import "HTMLNodeIterator.h"
+#import "HTMLNode.h"
+
+/**
+ Private HTML Node Iterator methods which are not intended for public API.
+ */
+@interface HTMLNodeIterator (Private)
+
+/**
+ Runs the necessary steps after removing a node from the DOM.
+
+ @param oldNode The old node that was removed.
+ @param oldParent The old parent of the node that was removed.
+ @param oldPreviousSibling The old previous sibling node of the node that was removed.
+ */
+- (void)runRemovingStepsForNode:(HTMLNode *)oldNode
+				  withOldParent:(HTMLNode *)oldParent
+		  andOldPreviousSibling:(HTMLNode *)oldPreviousSibling;
+
+@end

Sources/include/HTMLParseErrorToken.h

@@ -6,10 +6,6 @@
 //  Copyright (c) 2014 BrainCookie. All rights reserved.
 //
 
-///------------------------------------------------------
-/// HTMLKit private header
-///------------------------------------------------------
-
 #import <Foundation/Foundation.h>
 #import "HTMLToken.h"
 
@@ -18,19 +14,22 @@
  */
 @interface HTMLParseErrorToken : HTMLToken
 
-/** @brief The error's reason message. */
-@property (nonatomic, copy) NSString *reason;
+/** @brief The parse error's code as specified at https://html.spec.whatwg.org/multipage/parsing.html#parse-errors. */
+@property (nonatomic, strong, readonly) NSString *code;
+
+/** @brief Additional detailed error information. */
+@property (nonatomic, strong, readonly) NSString *details;
 
 /** @brief The error's location in the stream. */
-@property (nonatomic, assign) NSUInteger location;
+@property (nonatomic, assign, readonly) NSUInteger location;
 
 /**
  Initializes a new Parse Error token.
 
- @param reason The error's reason message.
+ @param code The parse error's as specified at https://html.spec.whatwg.org/multipage/parsing.html#parse-errors.
  @param location The error's location in the stream.
  @return A new instance of a parse error token.
  */
-- (instancetype)initWithReasonMessage:(NSString *)reason andStreamLocation:(NSUInteger)location;
+- (instancetype)initWithCode:(NSString *)code details:(NSString *)details location:(NSUInteger)location;
 
 @end

Sources/include/HTMLParser+Private.h

@@ -0,0 +1,23 @@
+//
+//  HTLMLParser+Private.h
+//  HTMLKit
+//
+//  Created by Iska on 27/11/16.
+//  Copyright © 2016 BrainCookie. All rights reserved.
+//
+
+#import "HTMLParser.h"
+#import "HTMLElement.h"
+
+/**
+ Private HTML Parser properties & methods which are not intended for public API.
+ */
+@interface HTMLParser (Private)
+
+/**
+ The adjusted current node in the context of HTML parsing as described in:
+ https://html.spec.whatwg.org/#adjusted-current-node
+ */
+@property (nonatomic, strong, readonly) HTMLElement *adjustedCurrentNode;
+
+@end

Sources/include/HTMLRange+Private.h

@@ -0,0 +1,61 @@
+//
+//  HTMLRange+Private.h
+//  HTMLKit
+//
+//  Created by Iska on 27/11/16.
+//  Copyright © 2016 BrainCookie. All rights reserved.
+//
+
+#import "HTMLRange.h"
+#import "HTMLCharacterData.h"
+
+@interface HTMLRange ()
+
+/**
+ Runs the necessary steps after removing data from a character data node that may be a range's boundary.
+
+ @param node The character data node.
+ @param offset The offset at which the data was removed.
+ @param length The length of the data that was removed.
+ */
+- (void)didRemoveCharacterDataInNode:(HTMLCharacterData *)node atOffset:(NSUInteger)offset withLength:(NSUInteger)length;
+
+/**
+ Runs the necessary steps after adding data to a character data node that may be a range's boundary.
+
+ @param node The character data node.
+ @param offset The offset at which the data was added.
+ @param length The length of the data that was added.
+ */
+- (void)didAddCharacterDataToNode:(HTMLCharacterData *)node atOffset:(NSUInteger)offset withLength:(NSUInteger)length;
+
+/**
+ Runs the necessary steps after inserting a new text node when an old text node is split.
+
+ @param newNode The new text node after splitting.
+ @param parent The parent where newNode was inserted.
+ @param node The old text node that was split.
+ @param offset The offset of splitting.
+ */
+- (void)didInsertNewTextNode:(HTMLText *)newNode intoParent:(HTMLNode *)parent afterSplittingTextNode:(HTMLText *)node atOffset:(NSUInteger)offset;
+
+/**
+ Runs the necessary steps to clamp the range whose end boundary is after the text node upon splitting it.
+
+ @param node The text node that was split.
+ @param offset The offset of splitting
+ */
+- (void)clampRangesAfterSplittingTextNode:(HTMLText *)node atOffset:(NSUInteger)offset;
+
+/**
+ Runs the necessary steps after removing a node from the DOM.
+
+ @param oldNode The old node that was removed.
+ @param oldParent The old parent of the node that was removed.
+@param oldPreviousSibling The old previous sibling node of the node that was removed.
+ */
+- (void)runRemovingStepsForNode:(HTMLNode *)oldNode
+				  withOldParent:(HTMLNode *)oldParent
+		  andOldPreviousSibling:(HTMLNode *)oldPreviousSibling;
+
+@end

Sources/include/HTMLRange.h

@@ -0,0 +1,265 @@
+//
+//  HTMLRange.h
+//  HTMLKit
+//
+//  Created by Iska on 20/11/16.
+//  Copyright © 2016 BrainCookie. All rights reserved.
+//
+
+#import "HTMLNode.h"
+#import "HTMLDocumentFragment.h"
+
+NS_ASSUME_NONNULL_BEGIN
+
+/**
+ The comparison method of range boundaries
+
+ - HTMLRangeComparisonMethodStartToStart: Compares the start boundary-point of sourceRange to the start boundary-point of this range.
+ - HTMLRangeComparisonMethodStartToEnd: Compares the start boundary-point of sourceRange to the end boundary-point of this range.
+ - HTMLRangeComparisonMethodEndToEnd: Compares the end boundary-point of sourceRange to the end boundary-point of this range.
+ - HTMLRangeComparisonMethodEndToStart: Compares the end boundary-point of sourceRange to the start boundary-point of this range.
+ */
+typedef NS_ENUM(unsigned short, HTMLRangeComparisonMethod)
+{
+	HTMLRangeComparisonMethodStartToStart = 0,
+	HTMLRangeComparisonMethodStartToEnd = 1,
+	HTMLRangeComparisonMethodEndToEnd = 2,
+	HTMLRangeComparisonMethodEndToStart = 3
+};
+
+#pragma mark - DOM Range
+
+/**
+ A HTML Range, represents a sequence of content within a node tree. 
+ Each range has a start and an end which are boundary points. 
+ A boundary point is a tuple consisting of a node and a non-negative numeric offset.
+
+ https://dom.spec.whatwg.org/#ranges
+ */
+@interface HTMLRange : NSObject
+
+/**
+ The node of the start boundary point.
+ */
+@property (nonatomic, readonly, strong) HTMLNode *startContainer;
+
+/**
+  The offset of the start boundary point.
+ */
+@property (nonatomic, readonly, assign) NSUInteger startOffset;
+
+/**
+ The node of the end boundary point.
+ */
+@property (nonatomic, readonly, strong) HTMLNode *endContainer;
+
+/**
+ The offset of the end boundary point.
+ */
+@property (nonatomic, readonly, assign) NSUInteger endOffset;
+
+/**
+ Checks whether the range is collapsed, i.e. if start is the same as end.
+
+ @return `YES` if the range is collapsed, `NO` otherwise.
+ */
+@property (nonatomic, readonly, assign, getter=isCollapsed) BOOL collapsed;
+
+/**
+ The common container node that contains both start and end nodes.
+ */
+@property (nonatomic, readonly, weak) HTMLNode *commonAncestorContainer;
+
+/**
+ @abstract A range is always associated with a HTML Document. Use `initWithDocument:` initializer instead.
+ */
+- (instancetype)init NS_UNAVAILABLE;
+
+/**
+ Initializes a new range instance for the given document.
+
+ @param document The HTML doucment for which the range will be constructed.
+ @return A new HTML Range instance.
+ */
+- (instancetype)initWithDocument:(HTMLDocument *)document;
+
+/**
+ Deprecated due to typo.
+ */
+- (instancetype)initWithDowcument:(HTMLDocument *)document __attribute__((deprecated("Replaced by -initWithDocument:")));
+
+/**
+ Initializes a new range instance for the given document and boundaries.
+
+ @param document The HTML doucment for which the range will be constructed.
+ @param startContainer The node for the start boundary
+ @param startOffset The offset of the start boundary
+ @param endContainer The node for the end boundary
+ @param endOffset The offset of the end boundary
+ @return A new HTML Range instance.
+ */
+- (instancetype)initWithDocument:(HTMLDocument *)document
+				  startContainer:(HTMLNode *)startContainer startOffset:(NSUInteger)startOffset
+					endContainer:(HTMLNode *)endContainer endOffset:(NSUInteger)endOffset;
+
+/**
+ Deprecated due to typo.
+ */
+- (instancetype)initWithDowcument:(HTMLDocument *)document
+				   startContainer:(HTMLNode *)startContainer startOffset:(NSUInteger)startOffset
+					 endContainer:(HTMLNode *)endContainer endOffset:(NSUInteger)endOffset
+	__attribute__((deprecated("Replaced by -initWithDocument:startContainer:startOffset:endContainer:endOffset:")));
+
+/**
+ Sets the start boundary.
+
+ @param node The new node of the start boundary.
+ @param offset The new offset of the start boundary.
+ */
+- (void)setStartNode:(HTMLNode *)node startOffset:(NSUInteger)offset;
+
+/**
+ Sets the end boundary.
+
+ @param node The new node of the end boundary.
+ @param offset The new offset of the end boundary.
+ */
+- (void)setEndNode:(HTMLNode *)node endOffset:(NSUInteger)offset;
+
+/**
+ Sets the start boundary before the given node.
+
+ @param node The node before which the boundary will be set.
+ */
+- (void)setStartBeforeNode:(HTMLNode *)node;
+
+/**
+ Sets the start boundary after the given node.
+
+ @param node The node after which the boundary will be set.
+ */
+- (void)setStartAfterNode:(HTMLNode *)node;
+
+/**
+ Sets the end boundary before the given node.
+
+ @param node The node before which the boundary will be set.
+ */
+- (void)setEndBeforeNode:(HTMLNode *)node;
+
+/**
+ Sets the end boundary after the given node.
+
+ @param node The node after which the boundary will be set.
+ */
+- (void)setEndAfterNode:(HTMLNode *)node;
+
+/**
+ Collapses this range to its start.
+ */
+- (void)collapseToStart;
+
+/**
+ Collapses this range to its end.
+ */
+- (void)collapseToEnd;
+
+/**
+ Selects the given node in the range.
+
+ @param node The node to select
+ */
+- (void)selectNode:(HTMLNode *)node;
+
+/**
+ Selects the node's contents in the range.
+
+ @param node The node to select the contents.
+ */
+- (void)selectNodeContents:(HTMLNode *)node;
+
+/**
+ Compares the boundary points of the given range with this range.
+
+ @param method The comparison method.
+ @param sourceRange The source range for comparison.
+ @return `NSOrderedAscending` if ordered before, `NSOrderedSame` if ordered same, `NSOrderedDescending` otherwise.
+ 
+ @see HTMLRangeComparisonMethod
+ */
+- (NSComparisonResult)compareBoundaryPoints:(HTMLRangeComparisonMethod)method sourceRange:(HTMLRange *)sourceRange;
+
+/**
+ Compares the given point (reference node, offset) with this range.
+
+ @param node The node to compare with this range.
+ @param offset The offset inside the reference node.
+ @return `NSOrderedAscending`, `NSOrderedSame`, or `NSOrderedDescending` depending on whether the node is before, the same as, or after this range.
+ */
+- (NSComparisonResult)comparePoint:(HTMLNode *)node offset:(NSUInteger)offset;
+
+/**
+ Checks if the given point (reference node, offset) is in this range.
+
+ @param node The node to compare with this range.
+ @param offset The offset inside the reference node.
+ @return `YES` if the given point is in this range, `NO` otherwise.
+ */
+- (BOOL)containsPoint:(HTMLNode *)node offset:(NSUInteger)offset;
+
+/**
+ Checks if the given node intersects this range.
+
+ @param node The node to compare with this range.
+ @return `YES` if the given node intersects the range, `NO` otherwise.
+ */
+- (BOOL)intersectsNode:(HTMLNode *)node;
+
+/**
+ Deletes the contents represented by this range from the associated document.
+ */
+- (void)deleteContents;
+
+/**
+ Extracts the contents represented by this range from the associated document.
+
+ @return A document fragment with the extracted contents.
+ */
+- (HTMLDocumentFragment *)extractContents;
+
+/**
+ Clones the contents represented by this range in the associated document.
+
+ @return A document fragment with the cloned contents.
+ */
+- (HTMLDocumentFragment *)cloneContents;
+
+/**
+ Inserts the given node at the start of this range.
+
+ If the node is being added to a text node, then the text node is split at the insertion point and the given node
+ is inserted between the resulting text nodes.
+
+ @param node The node to insert.
+ */
+- (void)insertNode:(HTMLNode *)node;
+
+/**
+ Surrounds the contents of this range with the given node.
+
+ The range's boundaries will placed around the given node, i.e. start is before and end is after newParent.
+
+ @param newParent The new parent node which will surround the range.
+ */
+- (void)surroundContents:(HTMLNode *)newParent;
+
+/**
+ The stringifier of the range.
+
+ @return The text contents of the range.
+ */
+- (NSString *)textContent;
+
+@end
+
+NS_ASSUME_NONNULL_END

Sources/include/HTMLStackOfOpenElements.h

@@ -79,7 +79,7 @@
 /**
  Checks whether an element with the given tag name is in the stack.
 
- @param tagname The element's tag name.
+ @param tagName The element's tag name.
  @return `YES` if such an element is in the stack, `NO` otherwise.
  */
 - (BOOL)containsElementWithTagName:(NSString *)tagName;
@@ -163,11 +163,11 @@
  https://html.spec.whatwg.org/multipage/syntax.html#has-an-element-in-the-specific-scope
  */
 - (HTMLElement *)hasElementInScopeWithTagName:(NSString *)tagName;
-- (HTMLElement *)hasAnyElementInScopeWithAnyOfTagNames:(NSArray *)tagNames;
-- (HTMLElement *)hasElementInListItemScopeWithTagName:(NSString *)tagName;
-- (HTMLElement *)hasElementInButtonScopeWithTagName:(NSString *)tagName;
+- (HTMLElement *)hasHeaderElementInScope;
 - (HTMLElement *)hasElementInTableScopeWithTagName:(NSString *)tagName;
 - (HTMLElement *)hasElementInTableScopeWithAnyOfTagNames:(NSArray *)tagNames;
+- (HTMLElement *)hasElementInListItemScopeWithTagName:(NSString *)tagName;
+- (HTMLElement *)hasElementInButtonScopeWithTagName:(NSString *)tagName;
 - (HTMLElement *)hasElementInSelectScopeWithTagName:(NSString *)tagName;
 
 /**

Sources/include/HTMLTagToken.h

@@ -6,10 +6,6 @@
 //  Copyright (c) 2014 BrainCookie. All rights reserved.
 //
 
-///------------------------------------------------------
-/// HTMLKit private header
-///------------------------------------------------------
-
 #import <Foundation/Foundation.h>
 #import "HTMLToken.h"
 #import "HTMLOrderedDictionary.h"

Sources/include/HTMLText.h

@@ -6,17 +6,14 @@
 //  Copyright (c) 2015 BrainCookie. All rights reserved.
 //
 
-#import "HTMLNode.h"
+#import "HTMLCharacterData.h"
 
 NS_ASSUME_NONNULL_BEGIN
 
 /**
  A HTML Text node
  */
-@interface HTMLText : HTMLNode
-
-/** @brief The text string. */
-@property (nonatomic, copy) NSMutableString *data;
+@interface HTMLText : HTMLCharacterData
 
 /**
  Initializes a new HTML text node.
@@ -31,7 +28,9 @@ NS_ASSUME_NONNULL_BEGIN
  
  @param string The string to append.
  */
-- (void)appendString:(NSString *)string;
+- (void)appendString:(NSString *)string __attribute__((deprecated("Use `appendData:` instead.")));
+
+- (HTMLText *)splitTextAtOffset:(NSUInteger)offset;
 
 @end
 

Sources/include/HTMLToken.h

@@ -6,10 +6,6 @@
 //  Copyright (c) 2014 BrainCookie. All rights reserved.
 //
 
-///------------------------------------------------------
-/// HTMLKit private header
-///------------------------------------------------------
-
 #import <Foundation/Foundation.h>
 
 @class HTMLDOCTYPEToken;

Sources/include/HTMLTokenizer.h

@@ -16,6 +16,13 @@
 
 @class HTMLParser;
 
+/**
+ Typedef for the parse error callback block.
+
+ @param token The parse error token.
+ */
+typedef void (^ HTMLTokenizerParseErrorCallback)(HTMLParseErrorToken *token);
+
 /**
  * HTML Tokenizer
  * https://html.spec.whatwg.org/multipage/syntax.html#tokenization
@@ -37,7 +44,14 @@
  
  @see HTMLParser
  */
-@property (nonatomic, weak, readonly) HTMLParser *parser;
+@property (nonatomic, weak) HTMLParser *parser;
+
+/**
+ An error callback block, which gets called when encountering parse errors while tokenizing the stream
+ 
+ Parse error tokens are dropped if the callback is `nil`.
+ */
+@property (nonatomic, copy) HTMLTokenizerParseErrorCallback parseErrorCallback;
 
 /**
  Initializes a new Tokenizer with the given string.

Sources/include/HTMLTokenizerCharacters.h

@@ -23,11 +23,16 @@
 	CHAR( DIGIT_NINE, 0x0039 ) \
 	CHAR( LATIN_CAPITAL_LETTER_A, 0x0041 ) \
 	CHAR( LATIN_CAPITAL_LETTER_F, 0x0046 ) \
+	CHAR( LATIN_CAPITAL_LETTER_P, 0x0050 ) \
+	CHAR( LATIN_CAPITAL_LETTER_S, 0x0053 ) \
 	CHAR( LATIN_CAPITAL_LETTER_X, 0x0058 ) \
 	CHAR( LATIN_CAPITAL_LETTER_Z, 0x005A ) \
+	CHAR( RIGHT_SQUARE_BRACKET, 0x005D ) \
 	CHAR( GRAVE_ACCENT, 0x0060 ) \
 	CHAR( LATIN_SMALL_LETTER_A, 0x0061 ) \
 	CHAR( LATIN_SMALL_LETTER_F, 0x0066 ) \
+	CHAR( LATIN_SMALL_LETTER_P, 0x0070 ) \
+	CHAR( LATIN_SMALL_LETTER_S, 0x0073 ) \
 	CHAR( LATIN_SMALL_LETTER_X, 0x0078 ) \
 	CHAR( LATIN_SMALL_LETTER_Z, 0x007A ) \
 	CHAR( HYPHEN_MINUS, 0x002D ) \
@@ -82,13 +87,17 @@ NUMERIC_REPLACEMENT_CHARACTERS
 #undef CHAR
 };
 
-NS_INLINE BOOL isControlOrUndefinedCharacter(UTF32Char character)
+NS_INLINE BOOL isControlCharacter(unsigned long long character)
 {
 	return ((character >= 0x0001 && character <= 0x0008) ||
-			(character >= 0x000D && character <= 0x001F) ||
-			(character >= 0x007F && character <= 0x009F) ||
-			(character >= 0xFDD0 && character <= 0xFDEF) ||
 			character == 0x000B ||
+			(character >= 0x000E && character <= 0x001F) ||
+			(character >= 0x007F && character <= 0x009F));
+}
+
+NS_INLINE BOOL isNoncharacter(unsigned long long character)
+{
+	return ((character >= 0xFDD0 && character <= 0xFDEF) ||
 			character == 0xFFFE ||
 			character == 0xFFFF ||
 			character == 0x1FFFE ||
@@ -137,6 +146,18 @@ NS_INLINE BOOL isHexDigit(UTF32Char character)
 			(character >= LATIN_SMALL_LETTER_A && character <= LATIN_SMALL_LETTER_F));
 }
 
+NS_INLINE BOOL isUpperHexDigit(UTF32Char character)
+{
+	return ((character >= LATIN_CAPITAL_LETTER_A && character <= LATIN_CAPITAL_LETTER_F) ||
+			(character >= DIGIT_ZERO && character <= DIGIT_NINE));
+}
+
+NS_INLINE BOOL isLowerHexDigit(UTF32Char character)
+{
+	return ((character >= LATIN_SMALL_LETTER_A && character <= LATIN_SMALL_LETTER_F) ||
+			(character >= DIGIT_ZERO && character <= DIGIT_NINE));
+}
+
 NS_INLINE BOOL isAlphanumeric(UTF32Char character)
 {
 	return ((character >= DIGIT_ZERO && character <= DIGIT_NINE) ||
@@ -151,17 +172,14 @@ NS_INLINE BOOL isStringAlphanumeric(NSString *string)
 	return ([string rangeOfCharacterFromSet:set].location == NSNotFound);
 }
 
-NS_INLINE BOOL isInvalidNumericRange(unsigned long long numeric)
+NS_INLINE BOOL isSurrogate(unsigned long long character)
 {
-	return ((numeric >= 0xD800 && numeric <= 0xDFFF) ||
-			numeric > 0x10FFFF);
+	return (character >= 0xD800 && character <= 0xDFFF);
 }
 
 NS_INLINE unichar NumericReplacementCharacter(UTF32Char character)
 {
-	if (character == NULL_CHAR) {
-		return REPLACEMENT_CHAR;
-	} else if (character >= 0x0080 && character <= 0x009F) {
+	if (character >= 0x0080 && character <= 0x009F) {
 		return NumericReplacementTable[character - 0x0080];
 	} else {
 		return NULL_CHAR;

Sources/include/HTMLTokenizerStates.h

@@ -12,9 +12,7 @@
 
 #define TOKENIZER_STATES \
 	STATE_ENTRY( HTMLTokenizerStateData, = 0) \
-	STATE_ENTRY( HTMLTokenizerStateCharacterReferenceInData, ) \
 	STATE_ENTRY( HTMLTokenizerStateRCDATA, ) \
-	STATE_ENTRY( HTMLTokenizerStateCharacterReferenceInRCDATA, ) \
 	STATE_ENTRY( HTMLTokenizerStateRAWTEXT, ) \
 	STATE_ENTRY( HTMLTokenizerStateScriptData, ) \
 	STATE_ENTRY( HTMLTokenizerStatePLAINTEXT, ) \
@@ -59,6 +57,10 @@
 	STATE_ENTRY( HTMLTokenizerStateCommentStart, ) \
 	STATE_ENTRY( HTMLTokenizerStateCommentStartDash, ) \
 	STATE_ENTRY( HTMLTokenizerStateComment, ) \
+	STATE_ENTRY( HTMLTokenizerStateCommentLessThanSign, ) \
+	STATE_ENTRY( HTMLTokenizerStateCommentLessThanSignBang, ) \
+	STATE_ENTRY( HTMLTokenizerStateCommentLessThanSignBangDash, ) \
+	STATE_ENTRY( HTMLTokenizerStateCommentLessThanSignBangDashDash, ) \
 	STATE_ENTRY( HTMLTokenizerStateCommentEndDash, ) \
 	STATE_ENTRY( HTMLTokenizerStateCommentEnd, ) \
 	STATE_ENTRY( HTMLTokenizerStateCommentEndBang, ) \
@@ -78,7 +80,18 @@
 	STATE_ENTRY( HTMLTokenizerStateDOCTYPESystemIdentifierSingleQuoted, ) \
 	STATE_ENTRY( HTMLTokenizerStateAfterDOCTYPESystemIdentifier, ) \
 	STATE_ENTRY( HTMLTokenizerStateBogusDOCTYPE, ) \
-	STATE_ENTRY( HTMLTokenizerStateCDATASection, )
+	STATE_ENTRY( HTMLTokenizerStateCDATASection, ) \
+	STATE_ENTRY( HTMLTokenizerStateCDATASectionBracket, ) \
+	STATE_ENTRY( HTMLTokenizerStateCDATASectionEnd, ) \
+	STATE_ENTRY( HTMLTokenizerStateCharacterReference, ) \
+	STATE_ENTRY( HTMLTokenizerStateNamedCharacterReference, ) \
+	STATE_ENTRY( HTMLTokenizerStateAmbiguousAmpersand, ) \
+	STATE_ENTRY( HTMLTokenizerStateNumericCharacterReference, ) \
+	STATE_ENTRY( HTMLTokenizerStateHexadecimalCharacterReferenceStart, ) \
+	STATE_ENTRY( HTMLTokenizerStateDecimalCharacterReferenceStart, ) \
+	STATE_ENTRY( HTMLTokenizerStateHexadecimalCharacterReference, ) \
+	STATE_ENTRY( HTMLTokenizerStateDecimalCharacterReference, ) \
+	STATE_ENTRY( HTMLTokenizerStateNumericCharacterReferenceEnd, )
 
 typedef NS_ENUM(NSUInteger, HTMLTokenizerState)
 {

Sources/include/HTMLTokens.h

@@ -6,10 +6,6 @@
 //  Copyright (c) 2014 BrainCookie. All rights reserved.
 //
 
-///------------------------------------------------------
-/// HTMLKit private header
-///------------------------------------------------------
-
 #import	"HTMLToken.h"
 #import "HTMLCharacterToken.h"
 #import "HTMLCommentToken.h"

Sources/include/HTMLTreeWalker.h

@@ -49,7 +49,6 @@ NS_ASSUME_NONNULL_BEGIN
  Initializes a new tree walker with no filter and HTMLNodeFilterShowAll show options.
 
  @param node The root node.
- @param filter The node filter to use.
  @return A new instance of a tree walker.
  */
 - (instancetype)initWithNode:(HTMLNode *)node;

Sources/include/NSString+HTMLKit.h

@@ -48,6 +48,13 @@ NS_ASSUME_NONNULL_BEGIN
  */
 - (NSUInteger)leadingHTMLWhitespaceLength;
 
+/**
+Escapes this string as specified in `https://html.spec.whatwg.org/multipage/parsing.html#escapingString`
+
+ @return A copy of this string by HTML-escaping this stirng.
+ */
+- (NSString *)stringByEscapingForHTML;
+
 @end
 
 NS_ASSUME_NONNULL_END

Sources/include/module.modulemap

@@ -0,0 +1,38 @@
+module HTMLKit {
+	umbrella header "HTMLKit.h"
+
+	module * { export * }
+	export *
+
+	explicit module Private {
+		textual header "CSSCodePoints.h"
+		header "CSSInputStream.h"
+		header "HTMLCharacterToken.h"
+		header "HTMLCommentToken.h"
+		header "HTMLDOCTYPEToken.h"
+		header "HTMLElementAdjustment.h"
+		header "HTMLElementTypes.h"
+		header "HTMLEOFToken.h"
+		header "HTMLInputStreamReader.h"
+		header "HTMLListOfActiveFormattingElements.h"
+		header "HTMLMarker.h"
+		header "HTMLParseErrorToken.h"
+		header "HTMLParserInsertionModes.h"
+		header "HTMLStackOfOpenElements.h"
+		header "HTMLTagToken.h"
+		header "HTMLToken.h"
+		header "HTMLTokenizer.h"
+		textual header "HTMLTokenizerCharacters.h"
+		header "HTMLTokenizerEntities.h"
+		header "HTMLTokenizerStates.h"
+		header "HTMLTokens.h"
+		header "HTMLNode+Private.h"
+		header "HTMLDocument+Private.h"
+		header "HTMLCharacterData+Private.h"
+		header "HTMLRange+Private.h"
+		header "HTMLNodeIterator+Private.h"
+		header "HTMLParser+Private.h"
+		header "HTMLNodeTraversal.h"
+		header "HTMLDOMUtils.h"
+	}
+}

Sources/module.modulemap

@@ -1,30 +0,0 @@
-module HTMLKit {
-	umbrella header "HTMLKit.h"
-
-	private header "CSSCodePoints.h"
-	private header "CSSInputStream.h"
-	private header "HTMLCharacterToken.h"
-	private header "HTMLCommentToken.h"
-	private header "HTMLDOCTYPEToken.h"
-	private header "HTMLElementAdjustment.h"
-	private header "HTMLElementTypes.h"
-	private header "HTMLEOFToken.h"
-	private header "HTMLInputStreamReader.h"
-	private header "HTMLListOfActiveFormattingElements.h"
-	private header "HTMLMarker.h"
-	private header "HTMLNode+Private.h"
-	private header "HTMLNodeTraversal.h"
-	private header "HTMLParseErrorToken.h"
-	private header "HTMLParserInsertionModes.h"
-	private header "HTMLStackOfOpenElements.h"
-	private header "HTMLTagToken.h"
-	private header "HTMLToken.h"
-	private header "HTMLTokenizer.h"
-	private header "HTMLTokenizerCharacters.h"
-	private header "HTMLTokenizerEntities.h"
-	private header "HTMLTokenizerStates.h"
-	private header "HTMLTokens.h"
-
-	module * { export * }
-	export *
-}

Tests/HTMLKitTests/CSSSelectorParserTests.m

@@ -7,7 +7,7 @@
 //
 
 #import <XCTest/XCTest.h>
-#import <HTMLKit/HTMLKit.h>
+#import "HTMLKit.h"
 #import "CSSSelectorTest.h"
 #import "CSSSelectorParser.h"
 

Tests/HTMLKitTests/CSSSelectorTest.m

@@ -11,6 +11,7 @@
 #import "HTMLDocument.h"
 #import "HTMLElement.h"
 #import "CSSSelectors.h"
+#import "HTMLKitTestUtil.h"
 
 static NSString * const CSSTests = @"css-tests";
 
@@ -18,8 +19,7 @@ static NSString * const CSSTests = @"css-tests";
 
 + (NSArray *)loadCSSSelectorTests
 {
-	NSString *path = [[NSBundle bundleForClass:self.class] resourcePath];
-	path = [path stringByAppendingPathComponent:CSSTests];
+	NSString *path = [HTMLKitTestUtil pathForFixture:CSSTests ofType:nil inDirectory:nil];
 
 	NSMutableArray *tests = [NSMutableArray array];
 	NSArray *testFiles = [[NSFileManager defaultManager] contentsOfDirectoryAtPath:path error:nil];