Merge branch 'bugger' into dev

# Conflicts: # include/tesseract/tprintf.h # src/api/baseapi.cpp # src/ccmain/control.cpp # src/ccmain/paramsd.cpp # src/ccmain/tesseractclass.cpp # src/ccmain/tesseractclass.h # src/ccstruct/debugpixa.cpp # src/ccstruct/pageres.cpp # src/ccutil/ambigs.cpp # src/ccutil/errcode.h # src/ccutil/params.cpp # src/ccutil/params.h # src/ccutil/unicharset.cpp # src/lstm/lstmrecognizer.h # src/tesseract.cpp # src/training/wordlist2dawg.cpp
GerHobbelt · Jul 9, 2024 · f05dcf2 · f05dcf2
2 parents da9d4d1 + 8dc3cdd
commit f05dcf2
Show file tree

Hide file tree

Showing 335 changed files with 3,143 additions and 1,142 deletions.
diff --git a/include/tesseract/baseapi.h b/include/tesseract/baseapi.h
@@ -17,9 +17,7 @@
 #ifndef TESSERACT_API_BASEAPI_H_
 #define TESSERACT_API_BASEAPI_H_
 
-#ifdef HAVE_TESSERACT_CONFIG_H
-#  include "config_auto.h" // DISABLED_LEGACY_ENGINE
-#endif
+#include <tesseract/preparation.h> // compiler config, etc.
 
 #include "export.h"
 #include "pageiterator.h"
@@ -66,8 +64,12 @@ class Tesseract;
 // Returns false on failure.
 using FileReader = bool (*)(const char *filename, std::vector<char> *data);
 
-using DictFunc = int (Dict::*)(void *, const UNICHARSET &, UNICHAR_ID,
+enum PermuterType : int;
+// function prototype:
+//   PermuterType Dict::letter_is_okay(void *void_dawg_args, const UNICHARSET &unicharset, UNICHAR_ID unichar_id, bool word_end)
+using DictFunc = PermuterType (Dict::*)(void *, const UNICHARSET &, UNICHAR_ID,
                                bool) const;
+
 using ProbabilityInContextFunc = double (Dict::*)(const char *, const char *,
                                                   int, const char *, int);
 
@@ -287,7 +289,11 @@ class TESS_API TessBaseAPI {
   bool GetVariableAsString(const char *name, std::string *val) const;
 
   /**
-   * Take all the internally gathered diagnostics data (including the tprintError/Warn/Info/Debug/Trace messages issued thus far, plus all collected image snapshots representing the intermediate state of the tesseract process at that time) and produce a HTML report from it for human consumption.
+   * Take all the internally gathered diagnostics data (including the
+   * tprintError/Warn/Info/Debug/Trace messages issued thus far, plus all
+   * collected image snapshots representing the intermediate state of the
+   * tesseract process at that time) and produce a HTML report from it
+   * for human consumption.
    */
   void FinalizeAndWriteDiagnosticsReport(); //  --> ReportDebugInfo()
 

diff --git a/include/tesseract/fmt-support.h b/include/tesseract/fmt-support.h
@@ -17,9 +17,7 @@
 #ifndef TESSERACT_FMT_SUPPORT_H_
 #define TESSERACT_FMT_SUPPORT_H_
 
-#ifdef HAVE_TESSERACT_CONFIG_H
-#  include "config_auto.h" // DISABLED_LEGACY_ENGINE
-#endif
+#include <tesseract/preparation.h> // compiler config, etc.
 
 #include <fmt/base.h>
 #include <fmt/format.h>

diff --git a/include/tesseract/ltrresultiterator.h b/include/tesseract/ltrresultiterator.h
@@ -63,7 +63,7 @@ class TESS_API LTRResultIterator : public PageIterator {
                     int scaled_yres, int rect_left, int rect_top,
                     int rect_width, int rect_height);
 
-  ~LTRResultIterator() override;
+  virtual ~LTRResultIterator() override;
 
   // LTRResultIterators may be copied! This makes it possible to iterate over
   // all the objects at a lower level, while maintaining an iterator to

diff --git a/include/tesseract/preparation.h b/include/tesseract/preparation.h
@@ -0,0 +1,17 @@
+
+#define _USE_MATH_DEFINES // for M_PI, when you load math.h
+
+// Include automatically generated configuration file if running autoconf.
+#ifdef HAVE_TESSERACT_CONFIG_H
+#  include "config_auto.h"
+#endif
+
+#if defined(_MSC_VER)
+#include <crtdbg.h>
+#if 0
+#include <winsock2.h>
+#include <windows.h>
+#endif
+#endif
+
+#include <tesseract/debugheap.h>
diff --git a/include/tesseract/resultiterator.h b/include/tesseract/resultiterator.h
@@ -37,7 +37,7 @@ class TESS_API ResultIterator : public LTRResultIterator {
    * ResultIterator is copy constructible!
    * The default copy constructor works just fine for us.
    */
-  ~ResultIterator() override = default;
+  virtual ~ResultIterator() override = default;
 
   // ============= Moving around within the page ============.
   /**

diff --git a/src/api/altorenderer.cpp b/src/api/altorenderer.cpp
@@ -13,7 +13,8 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include <tesseract/debugheap.h>
+#include <tesseract/preparation.h> // compiler config, etc.
+
 #include "errcode.h" // for ASSERT_HOST
 #include "helpers.h" // for copy_string
 
@@ -189,6 +190,7 @@ char *TessBaseAPI::GetAltoText(ETEXT_DESC *monitor, int page_number) {
         continue;
       case PT_NOISE:
         ASSERT_HOST_MSG(false, "TODO: Please report image which triggers the noise case.\n");
+        break;
       default:
         break;
     }

diff --git a/src/api/baseapi.cpp b/src/api/baseapi.cpp
@@ -16,12 +16,8 @@
  *
  **********************************************************************/
 
-#define _USE_MATH_DEFINES // for M_PI
-
 // Include automatically generated configuration file if running autoconf.
-#ifdef HAVE_TESSERACT_CONFIG_H
-#  include "config_auto.h"
-#endif
+#include <tesseract/preparation.h> // compiler config, etc.
 
 #include <tesseract/debugheap.h>
 #include "boxword.h"    // for BoxWord
@@ -264,23 +260,10 @@ TessBaseAPI::TessBaseAPI()
       rect_height_(0),
       image_width_(0),
       image_height_(0) {
-  // make sure the debug_all preset is set up BEFORE any command-line arguments
-  // direct tesseract to set some arbitrary parameters just below,
-  // for otherwise those `-c xyz=v` commands may be overruled by the
-  // debug_all preset!
-  debug_all.set_on_modify_handler([this](decltype(debug_all) &target,
-                                         const int32_t old_value,
-                                         int32_t &new_value,
-                                         const int32_t default_value,
-                                         ParamSetBySourceType source_type,
-                                         ParamPtr optional_setter) {
-    this->SetupDebugAllPreset();
-  });
 }
 
 TessBaseAPI::~TessBaseAPI() {
   End();
-  debug_all.set_on_modify_handler(0);
 }
 
 /**
@@ -1996,7 +1979,7 @@ bool TessBaseAPI::ProcessPage(Pix *pix, const char *filename,
       } else {
         p1 = GetInputImage();
       }
-      tess.AddPixDebugPage(p1, fmt::format("(normalized) image to process @ graynorm_mode = {}", graynorm_mode));
+      tess.AddPixDebugPage(p1, fmt::format("Greyscale normalized image to process @ graynorm_mode = {}", graynorm_mode));
     }
   }
 
@@ -3004,7 +2987,17 @@ bool TessBaseAPI::Threshold(Pix **pix) {
         const char *sequence = "c1.1 + d3.3";
         const int dispsep = 0;
         Image pix_post = pixMorphSequence(pix_binary, sequence, dispsep);
-        tesseract_->AddPixCompedOverOrigDebugPage(pix_post, fmt::format("Otsu (tesseract) : post-processed: {} (just an example to showcase what leptonica can do for us!)", sequence));
+        tesseract_->AddPixCompedOverOrigDebugPage(pix_post, fmt::format("Otsu (tesseract) : post-processed: {} -- just an example to showcase what leptonica can do for us!", sequence));
+
+        l_int32 w, h, d;
+        Image composite = tesseract_->pix_grey().copy();
+        pixGetDimensions(composite, &w, &h, &d);
+        Image mask = pixConvert1To8(nullptr, pix_post, 255, 0);
+        pixRasterop(composite, 0, 0, w, h, PIX_PAINT, mask, 0, 0);
+        tesseract_->AddPixCompedOverOrigDebugPage(composite, fmt::format("post-processed & masked with: {} -- this should remove all image noise that's not very close to the text, i.e. is considered *not part of the text to OCR*.", sequence));
+
+        mask.destroy();
+        composite.destroy();
         pix_post.destroy();
       }
 	  } else {

diff --git a/src/api/capi.cpp b/src/api/capi.cpp
@@ -15,7 +15,8 @@
 //
 ///////////////////////////////////////////////////////////////////////
 
-#include <tesseract/debugheap.h>
+#include <tesseract/preparation.h> // compiler config, etc.
+
 #include <tesseract/capi.h>
 
 

diff --git a/src/api/hocrrenderer.cpp b/src/api/hocrrenderer.cpp
@@ -17,7 +17,8 @@
  *
  **********************************************************************/
 
-#include <tesseract/debugheap.h>
+#include <tesseract/preparation.h> // compiler config, etc.
+
 #include <tesseract/baseapi.h> // for TessBaseAPI
 #include <locale>              // for std::locale::classic
 #include <memory>              // for std::unique_ptr
@@ -194,6 +195,7 @@ char *TessBaseAPI::GetHOCRText(ETEXT_DESC *monitor, int page_number) {
         continue;
       case PT_NOISE:
         ASSERT_HOST_MSG(false, "TODO: Please report image which triggers the noise case.\n");
+        break;
       default:
         break;
     }

diff --git a/src/api/lstmboxrenderer.cpp b/src/api/lstmboxrenderer.cpp
@@ -16,7 +16,8 @@
  *
  **********************************************************************/
 
-#include <tesseract/debugheap.h>
+#include <tesseract/preparation.h> // compiler config, etc.
+
 #include <tesseract/baseapi.h> // for TessBaseAPI
 #include <tesseract/renderer.h>
 #include "helpers.h"        // for copy_string

diff --git a/src/api/pagerenderer.cpp b/src/api/pagerenderer.cpp
@@ -15,7 +15,8 @@
 // limitations under the License.
  **********************************************************************/
 
-#include <tesseract/debugheap.h>
+#include <tesseract/preparation.h> // compiler config, etc.
+
 #include "errcode.h" // for ASSERT_HOST
 #include "helpers.h" // for copy_string
 #include <tesseract/baseapi.h> // for TessBaseAPI
@@ -807,6 +808,7 @@ char *TessBaseAPI::GetPAGEText(ETEXT_DESC *monitor, int page_number) {
         continue;
       case PT_NOISE:
         ASSERT_HOST_MSG(false, "TODO: Please report image which triggers the noise case.\n");
+        break;
       default:
         break;
     }
@@ -927,8 +929,10 @@ char *TessBaseAPI::GetPAGEText(ETEXT_DESC *monitor, int page_number) {
 
     if (LEVELFLAG > 0 || (POLYGONFLAG && !skewed_flag)) {
       // Sort wordpolygons
-      word_top_pts = RecalcPolygonline(word_top_pts, 1 - ttb_flag);
-      word_bottom_pts = RecalcPolygonline(word_bottom_pts, 0 + ttb_flag);
+      // 
+      // warning C4800: Implicit conversion from 'int' to bool. Possible information loss
+      word_top_pts = RecalcPolygonline(word_top_pts, !ttb_flag);
+      word_bottom_pts = RecalcPolygonline(word_bottom_pts, ttb_flag);
 
       // AppendLinePolygon
       AppendLinePolygon(line_top_ltr_pts, line_top_rtl_pts, word_top_pts,
@@ -1010,13 +1014,13 @@ char *TessBaseAPI::GetPAGEText(ETEXT_DESC *monitor, int page_number) {
       }
       if ((POLYGONFLAG && !skewed_flag) || LEVELFLAG > 0) {
         // Recalc Polygonlines
-        line_top_ltr_pts = RecalcPolygonline(line_top_ltr_pts, 1 - ttb_flag);
+        line_top_ltr_pts = RecalcPolygonline(line_top_ltr_pts, !ttb_flag);
         line_bottom_ltr_pts =
-            RecalcPolygonline(line_bottom_ltr_pts, 0 + ttb_flag);
+            RecalcPolygonline(line_bottom_ltr_pts, ttb_flag);
 
         // Smooth the polygonline
-        SimplifyLinePolygon(line_top_ltr_pts, 5, 1 - ttb_flag);
-        SimplifyLinePolygon(line_bottom_ltr_pts, 5, 0 + ttb_flag);
+        SimplifyLinePolygon(line_top_ltr_pts, 5, !ttb_flag);
+        SimplifyLinePolygon(line_bottom_ltr_pts, 5, ttb_flag);
 
         // Fit linepolygon matching the baselinepoints
         line_baseline_pts = SortBaseline(line_baseline_pts, writing_direction);

diff --git a/src/api/pdfrenderer.cpp b/src/api/pdfrenderer.cpp
@@ -16,11 +16,7 @@
 ///////////////////////////////////////////////////////////////////////
 
 // Include automatically generated configuration file if running autoconf.
-#ifdef HAVE_TESSERACT_CONFIG_H
-#  include "config_auto.h"
-#endif
-
-#include <tesseract/debugheap.h>
+#include <tesseract/preparation.h> // compiler config, etc.
 
 #include "pdf_ttf.h"
 #include <tesseract/tprintf.h>

diff --git a/src/api/renderer.cpp b/src/api/renderer.cpp
@@ -15,10 +15,7 @@
 //
 ///////////////////////////////////////////////////////////////////////
 
-#ifdef HAVE_TESSERACT_CONFIG_H
-#  include "config_auto.h"
-#endif
-#include <tesseract/debugheap.h>
+#include <tesseract/preparation.h> // compiler config, etc.
 #include <tesseract/baseapi.h>
 #include <tesseract/renderer.h>
 #include <cstring>
@@ -136,7 +133,7 @@ void TessResultRenderer::AppendData(const char *s, int len) {
   if (!tesseract::Serialize(fout_, s, len)) {
     happy_ = false;
   }
-  fflush(fout_);
+  //fflush(fout_);  -- only slows down performance.
 }
 
 bool TessResultRenderer::BeginDocumentHandler() {

diff --git a/src/api/wordstrboxrenderer.cpp b/src/api/wordstrboxrenderer.cpp
@@ -16,7 +16,8 @@
  *
  **********************************************************************/
 
-#include <tesseract/debugheap.h>
+#include <tesseract/preparation.h> // compiler config, etc.
+
 #include <tesseract/baseapi.h> // for TessBaseAPI
 #include <tesseract/renderer.h>
 #include "helpers.h"        // for copy_string

diff --git a/src/arch/dotproduct.cpp b/src/arch/dotproduct.cpp
@@ -14,6 +14,8 @@
 // limitations under the License.
 ///////////////////////////////////////////////////////////////////////
 
+#include <tesseract/preparation.h> // compiler config, etc.
+
 #include "dotproduct.h"
 
 namespace tesseract {

diff --git a/src/arch/dotproductavx.cpp b/src/arch/dotproductavx.cpp
@@ -15,6 +15,8 @@
 // limitations under the License.
 ///////////////////////////////////////////////////////////////////////
 
+#include <tesseract/preparation.h> // compiler config, etc.
+
 #include "dotproduct.h"
 #include "intsimdmatrix.h"
 

diff --git a/src/arch/dotproductavx512.cpp b/src/arch/dotproductavx512.cpp
@@ -15,6 +15,8 @@
 // limitations under the License.
 ///////////////////////////////////////////////////////////////////////
 
+#include <tesseract/preparation.h> // compiler config, etc.
+
 // General Notice:
 // 
 // This is not about whether the compiler is optimizing **the rest of your code using FMA instructions**.

diff --git a/src/arch/dotproductfma.cpp b/src/arch/dotproductfma.cpp
@@ -15,6 +15,8 @@
 // limitations under the License.
 ///////////////////////////////////////////////////////////////////////
 
+#include <tesseract/preparation.h> // compiler config, etc.
+
 #include "dotproduct.h"
 
 // General Notice:

diff --git a/src/arch/dotproductneon.cpp b/src/arch/dotproductneon.cpp
@@ -14,6 +14,8 @@
 // limitations under the License.
 ///////////////////////////////////////////////////////////////////////
 
+#include <tesseract/preparation.h> // compiler config, etc.
+
 #include "dotproduct.h"
 #include "tesstypes.h"
 

diff --git a/src/arch/dotproductsse.cpp b/src/arch/dotproductsse.cpp
@@ -15,6 +15,8 @@
 // limitations under the License.
 ///////////////////////////////////////////////////////////////////////
 
+#include <tesseract/preparation.h> // compiler config, etc.
+
 #include "dotproduct.h"
 
 // General Notice:

diff --git a/src/arch/intsimdmatrix.cpp b/src/arch/intsimdmatrix.cpp
@@ -15,6 +15,8 @@
 // limitations under the License.
 ///////////////////////////////////////////////////////////////////////
 
+#include <tesseract/preparation.h> // compiler config, etc.
+
 #include "intsimdmatrix.h"
 #include "matrix.h"     // for GENERIC_2D_ARRAY
 #include "simddetect.h" // for SIMDDetect

diff --git a/src/arch/intsimdmatrixavx2.cpp b/src/arch/intsimdmatrixavx2.cpp
@@ -15,6 +15,8 @@
 // limitations under the License.
 ///////////////////////////////////////////////////////////////////////
 
+#include <tesseract/preparation.h> // compiler config, etc.
+
 #include "intsimdmatrix.h"
 
 // General Notice:

diff --git a/src/arch/intsimdmatrixavx512vnni.cpp b/src/arch/intsimdmatrixavx512vnni.cpp
@@ -15,6 +15,8 @@
 // limitations under the License.
 ///////////////////////////////////////////////////////////////////////
 
+#include <tesseract/preparation.h> // compiler config, etc.
+
 #include "intsimdmatrix.h"
 
 // General Notice:

diff --git a/src/arch/intsimdmatrixneon.cpp b/src/arch/intsimdmatrixneon.cpp
@@ -16,6 +16,8 @@
 // limitations under the License.
 ///////////////////////////////////////////////////////////////////////
 
+#include <tesseract/preparation.h> // compiler config, etc.
+
 #include "intsimdmatrix.h"
 #include "tesstypes.h"