Adding --print-fonts-table parameter & tessedit_font_id configuration option

include/tesseract/baseapi.h

@@ -150,6 +150,11 @@ public:
    */
   const char *GetStringVariable(const char *name) const;
 
+  /**
+   * Print Tesseract fonts table to the given file.
+   */
+  void PrintFontsTable(FILE* fp) const;
+
   /**
    * Print Tesseract parameters to the given file.
    */

src/api/baseapi.cpp

@@ -330,6 +330,22 @@ bool TessBaseAPI::GetVariableAsString(const char *name, std::string *val) const
   return ParamUtils::GetParamAsString(name, tesseract_->params(), val);
 }
 
+/** Print Tesseract fonts table to the given file. */
+void TessBaseAPI::PrintFontsTable(FILE *fp) const {
+  const int fontinfo_size = tesseract_->get_fontinfo_table().size();
+  for (int font_index = 1; font_index < fontinfo_size; ++font_index) {
+    FontInfo font = tesseract_->get_fontinfo_table().at(font_index);
+    fprintf(fp, "ID=%3d: %s is_italic=%s is_bold=%s"
+                " is_fixed_pitch=%s is_serif=%s is_fraktur=%s\n",
+                font_index, font.name,
+                font.is_italic() ? "true" : "false",
+                font.is_bold() ? "true" : "false",
+                font.is_fixed_pitch() ? "true" : "false",
+                font.is_serif() ? "true" : "false",
+                font.is_fraktur() ? "true" : "false");
+  }
+}
+
 /** Print Tesseract parameters to the given file. */
 void TessBaseAPI::PrintVariables(FILE *fp) const {
   ParamUtils::PrintParams(fp, tesseract_->params());

src/api/tesseractmain.cpp

@@ -226,6 +226,9 @@ static void PrintHelpExtra(const char *program) {
 #endif
       "--version\n"
       "  %s --list-langs [--tessdata-dir PATH]\n"
+#ifndef DISABLED_LEGACY_ENGINE
+      "  %s --print-fonts-table [options...] [configfile...]\n"
+#endif  // ndef DISABLED_LEGACY_ENGINE
       "  %s --print-parameters [options...] [configfile...]\n"
       "  %s imagename|imagelist|stdin outputbase|stdout [options...] "
       "[configfile...]\n"
@@ -244,7 +247,11 @@ static void PrintHelpExtra(const char *program) {
 #endif
       "NOTE: These options must occur before any configfile.\n"
       "\n",
-      program, program, program, program);
+      program, program, program, program
+#ifndef DISABLED_LEGACY_ENGINE
+      , program
+#endif  // ndef DISABLED_LEGACY_ENGINE
+  );
 
   PrintHelpForPSM();
 #ifndef DISABLED_LEGACY_ENGINE
@@ -263,6 +270,9 @@ static void PrintHelpExtra(const char *program) {
 #endif
       "  -v, --version         Show version information.\n"
       "  --list-langs          List available languages for tesseract engine.\n"
+#ifndef DISABLED_LEGACY_ENGINE
+      "  --print-fonts-table   Print tesseract fonts table.\n"
+#endif  // ndef DISABLED_LEGACY_ENGINE
       "  --print-parameters    Print tesseract parameters.\n");
 }
 
@@ -358,7 +368,7 @@ static bool checkArgValues(int arg, const char *mode, int count) {
 // NOTE: arg_i is used here to avoid ugly *i so many times in this function
 static bool ParseArgs(int argc, char **argv, const char **lang, const char **image,
                       const char **outputbase, const char **datapath, l_int32 *dpi,
-                      bool *list_langs, bool *print_parameters, std::vector<std::string> *vars_vec,
+                      bool *list_langs, bool *print_parameters, bool* print_fonts_table, std::vector<std::string> *vars_vec,
                       std::vector<std::string> *vars_values, l_int32 *arg_i,
                       tesseract::PageSegMode *pagesegmode, tesseract::OcrEngineMode *enginemode) {
   bool noocr = false;
@@ -422,6 +432,11 @@ static bool ParseArgs(int argc, char **argv, const char **lang, const char **ima
     } else if (strcmp(argv[i], "--print-parameters") == 0) {
       noocr = true;
       *print_parameters = true;
+#ifndef DISABLED_LEGACY_ENGINE
+    } else if (strcmp(argv[i], "--print-fonts-table") == 0) {
+      noocr = true;
+      *print_fonts_table = true;
+#endif  // ndef DISABLED_LEGACY_ENGINE
     } else if (strcmp(argv[i], "-c") == 0 && i + 1 < argc) {
       // handled properly after api init
       ++i;
@@ -608,6 +623,7 @@ int main(int argc, char **argv) {
   const char *datapath = nullptr;
   bool list_langs = false;
   bool print_parameters = false;
+  bool print_fonts_table = false;
   l_int32 dpi = 0;
   int arg_i = 1;
   tesseract::PageSegMode pagesegmode = tesseract::PSM_AUTO;
@@ -634,7 +650,7 @@ int main(int argc, char **argv) {
 #endif // HAVE_TIFFIO_H && _WIN32
 
   if (!ParseArgs(argc, argv, &lang, &image, &outputbase, &datapath, &dpi, &list_langs,
-                 &print_parameters, &vars_vec, &vars_values, &arg_i, &pagesegmode, &enginemode)) {
+                 &print_parameters, &print_fonts_table, &vars_vec, &vars_values, &arg_i, &pagesegmode, &enginemode)) {
     return EXIT_FAILURE;
   }
 
@@ -643,7 +659,7 @@ int main(int argc, char **argv) {
     lang = "eng";
   }
 
-  if (image == nullptr && !list_langs && !print_parameters) {
+  if (image == nullptr && !list_langs && !print_parameters && !print_fonts_table) {
     return EXIT_SUCCESS;
   }
 
@@ -684,6 +700,16 @@ int main(int argc, char **argv) {
     return EXIT_SUCCESS;
   }
 
+#ifndef DISABLED_LEGACY_ENGINE
+  if (print_fonts_table) {
+    FILE* fout = stdout;
+    fprintf(stdout, "Tesseract fonts table:\n");
+    api.PrintFontsTable(fout);
+    api.End();
+    return EXIT_SUCCESS;
+  }
+#endif  // ndef DISABLED_LEGACY_ENGINE
+
   FixPageSegMode(api, pagesegmode);
 
   if (dpi) {

src/ccmain/control.cpp

@@ -1922,10 +1922,22 @@ void Tesseract::set_word_fonts(WERD_RES *word) {
   ASSERT_HOST(word->best_choice != nullptr);
 
 #ifndef DISABLED_LEGACY_ENGINE
-  const int fontinfo_size = get_fontinfo_table().size();
+  const int fontinfo_size = fontinfo_table_.size();
   if (fontinfo_size == 0) {
     return;
   }
+  if (tessedit_font_id > 0) {
+    if (tessedit_font_id >= fontinfo_size) {
+      tprintf("Error, invalid font ID provided: must be below %d.\n"
+              "Falling back to font auto-detection.\n", fontinfo_size);
+    } else {
+      word->fontinfo = &fontinfo_table_.at(tessedit_font_id);
+      word->fontinfo2 = nullptr;
+      word->fontinfo_id_count = INT8_MAX;
+      word->fontinfo_id2_count = 0;
+      return;
+    }
+  }
   std::vector<int> font_total_score(fontinfo_size);
 
   // Compute the font scores for the word

src/ccmain/tesseractclass.cpp

@@ -129,6 +129,7 @@ Tesseract::Tesseract()
     , BOOL_MEMBER(tessedit_enable_doc_dict, true, "Add words to the document dictionary",
                   this->params())
     , BOOL_MEMBER(tessedit_debug_fonts, false, "Output font info per char", this->params())
+    , INT_MEMBER(tessedit_font_id, 0, "Font ID to use or zero", this->params())
     , BOOL_MEMBER(tessedit_debug_block_rejection, false, "Block and Row stats", this->params())
     , BOOL_MEMBER(tessedit_enable_bigram_correction, true,
                   "Enable correction based on the word bigram dictionary.", this->params())

src/ccmain/tesseractclass.h

@@ -798,6 +798,8 @@ public:
   BOOL_VAR_H(tessedit_fix_hyphens, true, "Crunch double hyphens?");
   BOOL_VAR_H(tessedit_enable_doc_dict, true, "Add words to the document dictionary");
   BOOL_VAR_H(tessedit_debug_fonts, false, "Output font info per char");
+  INT_VAR_H(tessedit_font_id, 0, "Disable font detection and use the font"
+                                 " corresponding to the ID specified instead");
   BOOL_VAR_H(tessedit_debug_block_rejection, false, "Block and Row stats");
   BOOL_VAR_H(tessedit_enable_bigram_correction, true,
              "Enable correction based on the word bigram dictionary.");