part 2 of: refactor/remove the tesseract->lstmrecognizer->ccutil inte…

…rnal chain for unicode char array access and parameter settings.

src/training/classifier_tester.cpp

@@ -47,9 +47,9 @@ static const char *names[] = {"pruner", "full"};
 
 FZ_HEAPDBG_TRACKER_SECTION_END_MARKER(_)
 
-static tesseract::ShapeClassifier *InitializeClassifier(const char *classifer_name,
+static tesseract::ShapeClassifier *InitializeClassifier(tesseract::TessBaseAPI &api, const char *classifer_name,
                                                         const UNICHARSET &unicharset, int argc,
-                                                        const char **argv, tesseract::TessBaseAPI **api) {
+                                                        const char **argv) {
   // Decode the classifier string.
   ClassifierName classifier = CN_COUNT;
   for (int c = 0; c < CN_COUNT; ++c) {
@@ -64,17 +64,15 @@ static tesseract::ShapeClassifier *InitializeClassifier(const char *classifer_na
   }
 
   // We need to initialize tesseract to test.
-  *api = new tesseract::TessBaseAPI;
   tesseract::OcrEngineMode engine_mode = tesseract::OEM_TESSERACT_ONLY;
   tesseract::Tesseract *tesseract = nullptr;
   tesseract::Classify *classify = nullptr;
   if (classifier == CN_PRUNER || classifier == CN_FULL) {
-    tesseract::TessBaseAPI *tess = *api;
-    if (tess->InitOem(test_tessdata_dir.c_str(), test_lang.c_str(), engine_mode) < 0) {
+    if (api.InitOem(test_tessdata_dir.c_str(), test_lang.c_str(), engine_mode) < 0) {
       tprintError("Tesseract initialization failed!\n");
       return nullptr;
     }
-    tesseract = &tess->tesseract();
+    tesseract = &api.tesseract();
     classify = static_cast<tesseract::Classify *>(tesseract);
     if (classify->shape_table() == nullptr) {
       tprintError("Tesseract must contain a ShapeTable!\n");
@@ -120,16 +118,17 @@ extern "C" TESS_API int tesseract_classifier_tester_main(int argc, const char**
   tesseract::CheckSharedLibraryVersion();
   (void)tesseract::SetConsoleModeToUTF8();
 
-  int rv = ParseArguments(&argc, &argv);
+  tesseract::TessBaseAPI api;
+
+  int rv = ParseArguments(api, &argc, &argv);
   if (rv >= 0) {
     return rv;
   }
   std::string file_prefix;
   auto trainer = tesseract::LoadTrainingData(argv + 1, false, nullptr, file_prefix);
-  tesseract::TessBaseAPI *api;
   // Decode the classifier string.
   tesseract::ShapeClassifier *shape_classifier =
-      InitializeClassifier(test_classifier.c_str(), trainer->unicharset(), argc, argv, &api);
+      InitializeClassifier(api, test_classifier.c_str(), trainer->unicharset(), argc, argv);
   if (shape_classifier == nullptr) {
     tprintError("Classifier init failed!:{}\n", test_classifier.c_str());
     return EXIT_FAILURE;
@@ -144,7 +143,6 @@ extern "C" TESS_API int tesseract_classifier_tester_main(int argc, const char**
                                    test_report_level, false,
                                    shape_classifier, nullptr);
   delete shape_classifier;
-  delete api;
 
   return EXIT_SUCCESS;
 } /* main */

src/training/cntraining.cpp

@@ -116,6 +116,8 @@ extern "C" TESS_API int tesseract_cn_training_main(int argc, const char** argv)
   tesseract::CheckSharedLibraryVersion();
   (void)tesseract::SetConsoleModeToUTF8();
 
+  tesseract::TessBaseAPI api;
+
   // Set the global Config parameters before parsing the command line.
   Config = CNConfig;
 
@@ -129,7 +131,7 @@ extern "C" TESS_API int tesseract_cn_training_main(int argc, const char** argv)
   FEATURE_DEFS_STRUCT FeatureDefs;
   InitFeatureDefs(&FeatureDefs);
 
-  rv = ParseArguments(&argc, &argv);
+  rv = ParseArguments(api, &argc, &argv);
   if (rv >= 0) {
     return rv;
   }

src/training/common/commontraining.cpp

@@ -82,6 +82,7 @@ void ParseArguments(int* argc, const char ***argv) {
 #  include "tessdatamanager.h"
 #  include <tesseract/tprintf.h>
 #  include "unicity_table.h"
+#  include "tesseractclass.h"
 
 #if defined(HAVE_MUPDF)
 #include "mupdf/assertions.h"     // for ASSERT
@@ -98,8 +99,6 @@ FZ_HEAPDBG_TRACKER_SECTION_START_MARKER(_)
 CLUSTERCONFIG Config = {elliptical, 0.625, 0.05, 1.0, 1e-6, 0};
 FEATURE_DEFS_STRUCT feature_defs;
 
-static CCUtil *ccutil = nullptr;
-
 INT_VAR(trainer_debug_level, 0, "Level of Trainer debugging");
 INT_VAR(trainer_load_images, 0, "Load images with tr files");
 STRING_VAR(trainer_configfile, "", "File to load more configs from");
@@ -134,10 +133,7 @@ FZ_HEAPDBG_TRACKER_SECTION_END_MARKER(_)
  * @param argc number of command line arguments to parse
  * @param argv command line arguments
  */
-int ParseArguments(int *argc, const char ***argv) {
-	if (!ccutil)
-		ccutil = new CCUtil();
-
+int ParseArguments(TessBaseAPI &api, int *argc, const char ***argv) {
   std::string usage;
   if (*argc) {
     usage += (*argv)[0];
@@ -156,8 +152,7 @@ int ParseArguments(int *argc, const char ***argv) {
   Config.Confidence = std::max(0.0, std::min(1.0, double(clusterconfig_confidence)));
   // Set additional parameters from config file if specified.
   if (!trainer_configfile.empty()) {
-    ASSERT0(ccutil != nullptr);
-    tesseract::ParamUtils::ReadParamsFile(trainer_configfile.c_str(), tesseract::SET_PARAM_CONSTRAINT_NON_INIT_ONLY, ccutil->params());
+    tesseract::ParamUtils::ReadParamsFile(trainer_configfile.c_str(), tesseract::SET_PARAM_CONSTRAINT_NON_INIT_ONLY, api.tesseract().params());
   }
   return rv;
 }

src/training/common/commontraining.h

@@ -28,7 +28,7 @@
 namespace tesseract {
 
 TESS_COMMON_TRAINING_API
-int ParseArguments(int* argc, const char ***argv);
+int ParseArguments(TessBaseAPI &api, int *argc, const char ***argv);
 
 // Check whether the shared tesseract library is the right one.
 // This function must be inline because otherwise it would be part of

src/training/lstmeval.cpp

@@ -45,7 +45,9 @@ extern "C" int tesseract_lstm_eval_main(int argc, const char** argv)
   tesseract::CheckSharedLibraryVersion();
   (void)tesseract::SetConsoleModeToUTF8();
 
-  int rv = ParseArguments(&argc, &argv);
+  tesseract::TessBaseAPI api;
+
+  int rv = ParseArguments(api, &argc, &argv);
   if (rv >= 0) {
     return rv;
   }

src/training/lstmtraining.cpp

@@ -90,7 +90,9 @@ extern "C" int tesseract_lstm_training_main(int argc, const char** argv)
   tesseract::CheckSharedLibraryVersion();
   (void)tesseract::SetConsoleModeToUTF8();
 
-  int rv = ParseArguments(&argc, &argv);
+  tesseract::TessBaseAPI api;
+
+  int rv = ParseArguments(api, &argc, &argv);
   if (rv >= 0) {
     return rv;
   }
@@ -139,7 +141,6 @@ extern "C" int tesseract_lstm_training_main(int argc, const char** argv)
   }
 
   // Setup the trainer.
-  tesseract::TessBaseAPI api;
   std::string checkpoint_file = training_model_output;
   checkpoint_file += "_checkpoint";
   std::string checkpoint_bak = checkpoint_file + ".bak";

src/training/mftraining.cpp

@@ -201,7 +201,9 @@ extern "C" TESS_API int tesseract_mf_training_main(int argc, const char** argv)
 {
   tesseract::CheckSharedLibraryVersion();
 
-  int rv = ParseArguments(&argc, &argv);
+  tesseract::TessBaseAPI api;
+
+  int rv = ParseArguments(api, &argc, &argv);
   if (rv >= 0) {
     return rv;
   }

src/training/shapeclustering.cpp

@@ -55,7 +55,9 @@ extern "C" TESS_API int tesseract_shape_clustering_main(int argc, const char** a
 {
   tesseract::CheckSharedLibraryVersion();
 
-  ParseArguments(&argc, &argv);
+  tesseract::TessBaseAPI api;
+
+  ParseArguments(api, &argc, &argv);
 
   std::string file_prefix;
   auto trainer = tesseract::LoadTrainingData(argv + 1, false, nullptr, file_prefix);