PDF Renderer: allow to specify an alternate image or resolution progr…

…ammatically.

Support new rendering_dpi api params.
Add pdf renderer tests.
Install pdf font in cmake tool chain.

resolves tesseract-ocr#210
resolves tesseract-ocr#3798

CMakeLists.txt

@@ -100,6 +100,7 @@ option(DISABLE_TIFF "Disable build with libtiff (if available)" OFF)
 option(DISABLE_ARCHIVE "Disable build with libarchive (if available)" OFF)
 option(DISABLE_CURL "Disable build with libcurl (if available)" OFF)
 option(INSTALL_CONFIGS "Install tesseract configs" ON)
+option(INSTALL_PDF_TTF "Install pdf font file" ON)
 
 if(NOT ${CMAKE_VERSION} VERSION_LESS "3.15.0")
   if(WIN32 AND MSVC)
@@ -573,6 +574,8 @@ endif(ENABLE_OPENCL)
 message(STATUS "Use system ICU Library [USE_SYSTEM_ICU]: ${USE_SYSTEM_ICU}")
 message(
   STATUS "Install tesseract configs [INSTALL_CONFIGS]: ${INSTALL_CONFIGS}")
+message(
+  STATUS "Install tesseract pdf font [INSTALL_PDF_TTF]: ${INSTALL_PDF_TTF}")
 message(STATUS "--------------------------------------------------------")
 message(STATUS)
 
@@ -984,6 +987,10 @@ if(INSTALL_CONFIGS)
   install(FILES ${TESSERACT_TESSCONFIGS}
           DESTINATION ${CMAKE_INSTALL_PREFIX}/share/tessdata/tessconfigs)
 endif()
+if (INSTALL_PDF_TTF)
+  install(FILES tessdata/pdf.ttf
+          DESTINATION ${CMAKE_INSTALL_PREFIX}/share/tessdata)
+endif ()
 
 # ##############################################################################
 # uninstall target

Makefile.am

@@ -1269,6 +1269,7 @@ check_PROGRAMS += paragraphs_test
 if !DISABLED_LEGACY_ENGINE
 check_PROGRAMS += params_model_test
 endif # !DISABLED_LEGACY_ENGINE
+check_PROGRAMS += pdfrenderer_test
 check_PROGRAMS += progress_test
 check_PROGRAMS += qrsequence_test
 check_PROGRAMS += recodebeam_test
@@ -1497,6 +1498,10 @@ progress_test_CPPFLAGS = $(unittest_CPPFLAGS)
 progress_test_LDFLAGS = $(OPENCL_LDFLAGS) $(LEPTONICA_LIBS)
 progress_test_LDADD = $(GTEST_LIBS) $(GMOCK_LIBS) $(TESS_LIBS) $(LEPTONICA_LIBS)
 
+pdfrenderer_test_SOURCES = unittest/pdfrenderer_test.cc
+pdfrenderer_test_CPPFLAGS = $(unittest_CPPFLAGS)
+pdfrenderer_test_LDADD = $(TESS_LIBS) $(TRAINING_LIBS)
+
 qrsequence_test_SOURCES = unittest/qrsequence_test.cc
 qrsequence_test_CPPFLAGS = $(unittest_CPPFLAGS)
 qrsequence_test_LDADD = $(TESS_LIBS)

include/tesseract/renderer.h

@@ -106,6 +106,23 @@ class TESS_API TessResultRenderer {
     return imagenum_;
   }
 
+  /**
+   * Specifies an alternate image to render with the extracted text.
+   * It must be called after BeginDocument and before AddImage.
+   */
+  void SetRenderingImage(Pix *rendering_image) {
+    rendering_image_ = rendering_image;
+  }
+
+  /**
+   * Specifies the expected rendering resolution.
+   * If not set, rendering_dpi api params will be used, else the source image
+   * resolution.
+   */
+  void SetRenderingResolution(int rendering_dpi) {
+    rendering_dpi_ = rendering_dpi;
+  }
+
 protected:
   /**
    * Called by concrete classes.
@@ -139,12 +156,29 @@ class TESS_API TessResultRenderer {
   // This method will grow the output buffer if needed.
   void AppendData(const char *s, int len);
 
+  // Renderers can call this to get the actual image to render with extracted
+  // text. This method returns:
+  //  - the rendering image set by the caller or
+  //  - the input image scaled to the rendering_dpi field if defined or
+  //  - the input image from the api otherwise
+  Pix *GetRenderingImage(TessBaseAPI *api);
+
+  // Resolution of the rendering image either set manually by the caller or with
+  // the rendering_dpi api parameter.
+  int GetRenderingResolution(TessBaseAPI *api);
+
+  // Reset rendering image and dpi to previous state. Destroy scaled rendered
+  // image if exists.
+  void ResetRenderingState(Pix *rendering_image_prev, int rendering_dpi_prev);
+
 private:
   TessResultRenderer *next_;   // Can link multiple renderers together
   FILE *fout_;                 // output file pointer
   const char *file_extension_; // standard extension for generated output
   std::string title_;          // title of document being rendered
   int imagenum_;               // index of last image added
+  Pix *rendering_image_;       // Image to render with the extracted text
+  int rendering_dpi_;          // Resolution of the rendering_image
   bool happy_;                 // I get grumpy when the disk fills up, etc.
 };
 

src/api/pdfrenderer.cpp

@@ -319,7 +319,12 @@ static bool CodepointToUtf16be(int code, char utf16[kMaxBytesPerCodepoint]) {
 }
 
 char *TessPDFRenderer::GetPDFTextObjects(TessBaseAPI *api, double width, double height) {
-  double ppi = api->GetSourceYResolution();
+  double input_image_ppi = api->GetSourceYResolution();
+  double ppi = GetRenderingResolution(api);
+  double scale = 1;
+  if (input_image_ppi > 0) {
+    scale = ppi / input_image_ppi;
+  }
 
   // These initial conditions are all arbitrary and will be overwritten
   double old_x = 0.0, old_y = 0.0;
@@ -369,6 +374,7 @@ char *TessPDFRenderer::GetPDFTextObjects(TessBaseAPI *api, double width, double
     if (res_it->IsAtBeginningOf(RIL_TEXTLINE)) {
       int x1, y1, x2, y2;
       res_it->Baseline(RIL_TEXTLINE, &x1, &y1, &x2, &y2);
+      x1 *= scale; y1 *= scale; x2 *= scale; y2 *= scale;
       ClipBaseline(ppi, x1, y1, x2, y2, &line_x1, &line_y1, &line_x2, &line_y2);
     }
 
@@ -403,6 +409,7 @@ char *TessPDFRenderer::GetPDFTextObjects(TessBaseAPI *api, double width, double
     {
       int word_x1, word_y1, word_x2, word_y2;
       res_it->Baseline(RIL_WORD, &word_x1, &word_y1, &word_x2, &word_y2);
+      word_x1 *= scale; word_y1 *= scale; word_x2 *= scale; word_y2 *= scale;
       GetWordBaseline(writing_direction, ppi, height, word_x1, word_y1, word_x2, word_y2, line_x1,
                       line_y1, line_x2, line_y2, &x, &y, &word_length);
     }
@@ -809,9 +816,9 @@ bool TessPDFRenderer::imageToPDFObj(Pix *pix, const char *filename, long int obj
 }
 
 bool TessPDFRenderer::AddImageHandler(TessBaseAPI *api) {
-  Pix *pix = api->GetInputImage();
+  Pix *pix = GetRenderingImage(api);
   const char *filename = api->GetInputName();
-  int ppi = api->GetSourceYResolution();
+  int ppi = GetRenderingResolution(api);
   if (!pix || ppi <= 0) {
     return false;
   }

src/api/renderer.cpp

@@ -18,12 +18,14 @@
 #ifdef HAVE_CONFIG_H
 #  include "config_auto.h"
 #endif
+#include <allheaders.h>
 #include <tesseract/baseapi.h>
 #include <tesseract/renderer.h>
 #include <cstring>
 #include <memory>     // std::unique_ptr
 #include <string>     // std::string
 #include "serialis.h" // Serialize
+#include "tprintf.h"
 
 namespace tesseract {
 
@@ -36,6 +38,8 @@ TessResultRenderer::TessResultRenderer(const char *outputbase, const char *exten
     , file_extension_(extension)
     , title_("")
     , imagenum_(-1)
+    , rendering_image_(nullptr)
+    , rendering_dpi_(0)
     , happy_(true) {
   if (strcmp(outputbase, "-") && strcmp(outputbase, "stdout")) {
     std::string outfile = std::string(outputbase) + "." + extension;
@@ -90,13 +94,71 @@ bool TessResultRenderer::AddImage(TessBaseAPI *api) {
     return false;
   }
   ++imagenum_;
+  Pix *rendering_image_prev = rendering_image_;
+  int rendering_dpi_prev = rendering_dpi_;
   bool ok = AddImageHandler(api);
+  ResetRenderingState(rendering_image_prev, rendering_dpi_prev);
   if (next_) {
     ok = next_->AddImage(api) && ok;
   }
   return ok;
 }
 
+void TessResultRenderer::ResetRenderingState(Pix *rendering_image_prev,
+                                             int rendering_dpi_prev) {
+  if (rendering_image_ != rendering_image_prev) {
+    pixDestroy(&rendering_image_);
+    rendering_image_ = rendering_image_prev;
+  }
+  if (rendering_dpi_ != rendering_dpi_prev) {
+    rendering_dpi_ = rendering_dpi_prev;
+  }
+}
+
+Pix *TessResultRenderer::GetRenderingImage(TessBaseAPI *api) {
+  if (!rendering_image_) {
+    Pix *source_image = api->GetInputImage();
+    int source_dpi = api->GetSourceYResolution();
+    if (!source_image || source_dpi <= 0) {
+      happy_ = false;
+      return nullptr;
+    }
+
+    int rendering_dpi = GetRenderingResolution(api);
+    if (rendering_dpi != source_dpi) {
+      float scale = (float)rendering_dpi / (float)source_dpi;
+
+      rendering_image_ = pixScale(source_image, scale, scale);
+    } else {
+      return source_image;
+    }
+  }
+  return rendering_image_;
+}
+
+int TessResultRenderer::GetRenderingResolution(tesseract::TessBaseAPI *api) {
+  if (rendering_dpi_) {
+    return rendering_dpi_;
+  }
+  int source_dpi = api->GetSourceYResolution();
+  int rendering_dpi;
+  if (api->GetIntVariable("rendering_dpi", &rendering_dpi) &&
+      rendering_dpi > 0 && rendering_dpi != source_dpi) {
+    if (rendering_dpi < kMinCredibleResolution ||
+        rendering_dpi > kMaxCredibleResolution) {
+#if !defined(NDEBUG)
+      tprintf(
+          "Warning: User defined rendering dpi %d is outside of expected range "
+          "(%d - %d)!\n",
+          rendering_dpi, kMinCredibleResolution, kMaxCredibleResolution);
+#endif
+    }
+    rendering_dpi_ = rendering_dpi;
+    return rendering_dpi_;
+  }
+  return source_dpi;
+}
+
 bool TessResultRenderer::EndDocument() {
   if (!happy_) {
     return false;

src/ccmain/tesseractclass.cpp

@@ -349,6 +349,7 @@ Tesseract::Tesseract()
     , BOOL_MEMBER(textonly_pdf, false, "Create PDF with only one invisible text layer",
                   this->params())
     , INT_MEMBER(jpg_quality, 85, "Set JPEG quality level", this->params())
+    , INT_MEMBER(rendering_dpi, 0, "Scaled input image resolution before rendering", this->params())
     , INT_MEMBER(user_defined_dpi, 0, "Specify DPI for input image", this->params())
     , INT_MEMBER(min_characters_to_try, 50, "Specify minimum characters to try during OSD",
                  this->params())

src/ccmain/tesseractclass.h

@@ -903,6 +903,7 @@ class TESS_API Tesseract : public Wordrec {
   BOOL_VAR_H(tessedit_create_pdf);
   BOOL_VAR_H(textonly_pdf);
   INT_VAR_H(jpg_quality);
+  INT_VAR_H(rendering_dpi);
   INT_VAR_H(user_defined_dpi);
   INT_VAR_H(min_characters_to_try);
   STRING_VAR_H(unrecognised_char);