fix: rotated pages (missing commits) (#219)

Signed-off-by: Peter Staar <taa@zurich.ibm.com>

.github/workflows/wheels.yml

@@ -111,8 +111,9 @@ jobs:
           echo "Building cp: ${{ env.python_cp_version }}"
           echo "Building cache_tag: ${PY_CACHE_TAG}"
           echo "Building platform_id: ${{ matrix.os.platform_id }}"
+          echo "Building with BUILD_THREADS: ${BUILD_THREADS:-not set}"
           uv run python --version
-          uv run python --version | grep ${{ matrix.python-version }} 
+          uv run python --version | grep ${{ matrix.python-version }}
           cat ./pyproject.toml
           uv pip install --group build
           rm -rf ./build || true
@@ -163,8 +164,9 @@ jobs:
           echo "Building cp: ${{ env.python_cp_version }}"
           echo "Building cache_tag: ${PY_CACHE_TAG}"
           echo "Building platform_id: ${{ matrix.os.platform_id }}"
+          echo "Building with BUILD_THREADS: ${BUILD_THREADS:-not set}"
           uv run python --version
-          uv run python --version | grep ${{ matrix.python-version }} 
+          uv run python --version | grep ${{ matrix.python-version }}
           cat ./pyproject.toml
           uv pip install --group build
           rm -rf ./build || true
@@ -204,14 +206,18 @@ jobs:
           CIBW_SKIP: "pp* *musllinux_* *_i686* *_s390* *pypy*"
           CIBW_PROJECT_REQUIRES_PYTHON: "~=${{ matrix.python-version }}.0"
           CIBW_BUILD_VERBOSITY: 3
-          BUILD_THREADS: "8"
+          # Multi-threading: pass BUILD_THREADS into the cibuildwheel Docker container.
+          # Comment out the next line to disable parallel builds.
+          CIBW_ENVIRONMENT_PASS_LINUX: "BUILD_THREADS"
+          BUILD_THREADS: "4"
         run: |
           PY_CACHE_TAG=$(uv run python -c 'import sys;print(sys.implementation.cache_tag)')
           echo "Building cp: ${{ env.python_cp_version }}"
           echo "Building cache_tag: ${PY_CACHE_TAG}"
           echo "Building platform_id: ${{ matrix.os.platform_id }}"
+          echo "Building with BUILD_THREADS: ${BUILD_THREADS:-not set}"
           uv run python --version
-          uv run python --version | grep ${{ matrix.python-version }} 
+          uv run python --version | grep ${{ matrix.python-version }}
           cat ./pyproject.toml
           uv pip install --group build
           rm -rf ./build || true
@@ -275,7 +281,7 @@ jobs:
           CMAKE_LIBRARY_PATH: "C:/msys64/mingw64/lib;C:/mingw64/lib;C:/windows-libs/external-libs/lib-mingw64"
           CMAKE_INCLUDE_PATH: "C:/msys64/mingw64/include;C:/mingw64/include;C:/windows-libs/external-libs/include"
           CMAKE_GENERATOR: "MSYS Makefiles"
-          BUILD_THREADS: 1
+          BUILD_THREADS: 4
           ASM_NASM: "C:/nasm/nasm.exe"
         shell: pwsh
         run: |

local_build.py

@@ -70,4 +70,5 @@ def build_local(num_threads: int):
 
 if "__main__" == __name__:
     num_threads = int(os.getenv("BUILD_THREADS", "4"))
+    print(f"Building with {num_threads} threads (BUILD_THREADS={os.getenv('BUILD_THREADS', 'not set, defaulting to 4')})")
     build_local(num_threads=num_threads)

src/parse.h

@@ -14,6 +14,7 @@
 #include <regex>
 #include <fstream>
 #include <iostream>
+#include <cmath>
 
 #ifdef _WIN32
 #include <share.h> // to define _SH_DENYNO for loguru

src/parse/page_items/page_dimension.h

@@ -111,94 +111,80 @@ namespace pdflib
 
   std::pair<double, double> page_item<PAGE_DIMENSION>::rotate(int my_angle)
   {
-    angle -= my_angle;
-
     LOG_S(INFO) << "my_angle: " << my_angle;
     
-    utils::values::rotate_inplace(my_angle, media_bbox);
+    // normalize the angle (after rotation, it should be zero)
+    angle -= my_angle;
 
-    LOG_S(INFO) << "media: "
+    LOG_S(INFO) << "media (before): "
 		<< media_bbox[0] << ", "
 		<< media_bbox[1] << ", "
 		<< media_bbox[2] << ", "
 		<< media_bbox[3];
     
-    utils::values::rotate_inplace(my_angle, crop_bbox);
-
-    LOG_S(INFO) << "crop: "
+    LOG_S(INFO) << "crop (before): "
 		<< crop_bbox[0] << ", "
 		<< crop_bbox[1] << ", "
 		<< crop_bbox[2] << ", "
 		<< crop_bbox[3];
-	
-    utils::values::rotate_inplace(my_angle, bleed_bbox);
-    utils::values::rotate_inplace(my_angle, trim_bbox);
-    utils::values::rotate_inplace(my_angle, art_bbox);
 
-    utils::values::rotate_inplace(my_angle, bbox);
+    std::pair<double, double> delta = {0.0, 0.0};
     
-    // std::pair<double, double> delta = {0.0, std::abs(media_bbox[3])};
-
-    std::pair<double, double> delta = {0.0, 0.0}; //std::abs(media_bbox[3])};
-
     switch(my_angle)
       {
       case 0:
-	{}
+	{
+	  delta = {0.0, 0.0};
+	}
 	break;
 	
       case 90:
 	{
-	  delta = {0.0, std::abs(media_bbox[3])};
-
-	  media_bbox[3] += 2*delta.second;
-	  crop_bbox[3] += 2*delta.second;
-	  bleed_bbox[3] += 2*delta.second;
-	  trim_bbox[3] += 2*delta.second;
-	  art_bbox[3] += 2*delta.second;
-	  
-	  bbox[3] += 2*delta.second;	  
+	  delta = {
+	    0.0,
+	    std::abs(media_bbox[2])
+	  };
 	}
 	break;
 
       case 180:
 	{
-	  delta = {std::abs(media_bbox[2]), std::abs(media_bbox[3])};
-
-	  media_bbox[2] += 2*delta.first;
-	  media_bbox[3] += 2*delta.second;
-
-	  crop_bbox[2] += 2*delta.first;
-	  crop_bbox[3] += 2*delta.second;
-
-	  bleed_bbox[2] += 2*delta.first;
-	  bleed_bbox[3] += 2*delta.second;
-
-	  trim_bbox[2] += 2*delta.first;
-	  trim_bbox[3] += 2*delta.second;
-
-	  art_bbox[2] += 2*delta.first;
-	  art_bbox[3] += 2*delta.second;
-
-	  bbox[2] += 2*delta.first;	  	  
-	  bbox[3] += 2*delta.second;	  	  
+	  delta = {
+	    std::abs(media_bbox[2]),
+	    std::abs(media_bbox[3])
+	  };
 	}
 	break;
 
+      case 270:
+	{
+	  delta = {
+	    std::abs(media_bbox[3]),
+	    0.0	    
+	  };	  
+	}
+	break;
+	
       default:
 	{
 	  LOG_S(WARNING) << "might be unsupported rotation angle: " << my_angle;
 	}
       }
-    
 
-    LOG_S(INFO) << "crop: "
+    utils::values::transform_bottomleft_bbox_inplace(my_angle, delta, crop_bbox);
+    utils::values::transform_bottomleft_bbox_inplace(my_angle, delta, media_bbox);
+    utils::values::transform_bottomleft_bbox_inplace(my_angle, delta, trim_bbox);
+    utils::values::transform_bottomleft_bbox_inplace(my_angle, delta, bleed_bbox);
+    utils::values::transform_bottomleft_bbox_inplace(my_angle, delta, art_bbox);
+    utils::values::transform_bottomleft_bbox_inplace(my_angle, delta, bbox);
+    
+    LOG_S(INFO) << "crop (after): "
 		<< crop_bbox[0] << ", "
 		<< crop_bbox[1] << ", "
 		<< crop_bbox[2] << ", "
 		<< crop_bbox[3];
 
-    LOG_S(INFO) << "bbox: "
+    LOG_S(INFO) << "bbox (after): "
 		<< bbox[0] << ", "
 		<< bbox[1] << ", "
 		<< bbox[2] << ", "
@@ -206,7 +192,8 @@ namespace pdflib
     
     return delta;
   }
-
+  
+  
   std::array<double, 4> page_item<PAGE_DIMENSION>::normalize_page_boundaries(std::array<double, 4> bbox, std::string name)
   {
     LOG_S(INFO) << __FUNCTION__;

src/parse/page_items/page_image.h

@@ -132,7 +132,7 @@ namespace pdflib
 
   void page_item<PAGE_IMAGE>::rotate(int angle, std::pair<double, double> delta)
   {
-    LOG_S(INFO) << "rotate & translate image: (x0: " << x0 << ", y0: " << y0 << ", x1: " << x1 << ", y1: " << y1 << ")"; 
+    // LOG_S(INFO) << "rotate & translate image: (x0: " << x0 << ", y0: " << y0 << ", x1: " << x1 << ", y1: " << y1 << ")"; 
     
     utils::values::rotate_inplace(angle, x0, y0);
     utils::values::rotate_inplace(angle, x1, y1);
@@ -140,7 +140,7 @@ namespace pdflib
     utils::values::translate_inplace(delta, x0, y0);
     utils::values::translate_inplace(delta, x1, y1);
 
-    LOG_S(INFO) << "into (x0: " << x0 << ", y0: " << y0 << ", x1: " << x1 << ", y1: " << y1 << ")"; 
+    // LOG_S(INFO) << "into (x0: " << x0 << ", y0: " << y0 << ", x1: " << x1 << ", y1: " << y1 << ")"; 
 
     // The bounding box always needs to have x0<x1 and y0<y1. If you want
     // to keep the orientation of the image, we will need to add the rectangle

src/parse/utils/values.h

@@ -76,7 +76,8 @@ namespace utils
     
     void rotate_inplace(int angle, double& x0, double& y0)
     {
-      double phi = -3.141592*angle/180.0;
+      //double phi = -3.141592*angle/180.0;
+      double phi = -M_PI*angle/180.0;
       
       double x1 = x0, y1 = y0;
 
@@ -102,6 +103,47 @@ namespace utils
       bbox[2] = std::cos(phi)*tmp[2] - std::sin(phi)*tmp[3];
       bbox[3] = std::sin(phi)*tmp[2] + std::cos(phi)*tmp[3];
     }
+
+    void transform_bottomleft_bbox_inplace(const int& angle,
+					   const std::pair<double, double>& delta,
+					   std::array<double, 4>& bbox)
+    {
+      // LOG_S(INFO) << "delta-x1: " << delta.first << ", delta-y1: " << delta.second;
+      
+      double& x0 = bbox.at(0);
+      double& y0 = bbox.at(1);
+
+      double& x1 = bbox.at(2);
+      double& y1 = bbox.at(3);
+
+      // LOG_S(INFO) << "before x1: " << x1 << ", y1: " << y1;
+      
+      utils::values::rotate_inplace(angle, x0, y0);
+      utils::values::rotate_inplace(angle, x1, y1);
+
+      // LOG_S(INFO) << "rotate  x1: " << x1 << ", y1: " << y1;
+      
+      utils::values::translate_inplace(delta, x0, y0);
+      utils::values::translate_inplace(delta, x1, y1);
+
+      // LOG_S(INFO) << "translate x1: " << x1 << ", y1: " << y1;
+      
+      // The bounding box always needs to have x0<x1 and y0<y1. If you want
+      // to keep the orientation of the image, we will need to add the rectangle
+      // as demonstrated in the page_cell.
+    
+      double x_min = std::min(x0, x1);
+      double x_max = std::max(x0, x1);
+      
+      double y_min = std::min(y0, y1);
+      double y_max = std::max(y0, y1);
+
+      x0 = x_min;
+      x1 = x_max;
+      
+      y0 = y_min;
+      y1 = y_max;      
+    }
     
   }
   

tests/test_parse.py

@@ -418,10 +418,9 @@ def test_reference_documents_from_filenames():
 
                     _fname = fname + f".{unit}.txt"
 
-                    lines = []
                     with open(_fname, "r") as fr:
                         content = fr.read()
-                        lines = content.split(SPECIAL_SEPERATOR)
+                        lines = content.split(SPECIAL_SEPERATOR) if content else []
 
                     assert len(lines) == len(
                         _lines