Merge pull request #1750 from stweil/fix

Fix several compiler warnings (clang)

src/api/baseapi.cpp

@@ -456,7 +456,7 @@ void TessBaseAPI::GetAvailableLanguagesAsVector(
   }
 }
 
-//TODO(amit): Adapt to lstm 
+//TODO(amit): Adapt to lstm
 #ifndef DISABLED_LEGACY_ENGINE
 /**
  * Init only the lang model component of Tesseract. The only functions
@@ -833,8 +833,8 @@ int TessBaseAPI::Recognize(ETEXT_DESC* monitor) {
     page_res_ = tesseract_->ApplyBoxes(*input_file_, true, block_list_);
   } else if (tesseract_->tessedit_resegment_from_boxes) {
     page_res_ = tesseract_->ApplyBoxes(*input_file_, false, block_list_);
-  } else 
-#endif  // ndef DISABLED_LEGACY_ENGINE 
+  } else
+#endif  // ndef DISABLED_LEGACY_ENGINE
   {
     page_res_ = new PAGE_RES(tesseract_->AnyLSTMLang(),
                              block_list_, &tesseract_->prev_word_best_choice_);
@@ -1616,8 +1616,11 @@ char* TessBaseAPI::GetTSVText(int page_number) {
 
   STRING tsv_str("");
 
-  int page_num = page_id, block_num = 0, par_num = 0, line_num = 0,
-      word_num = 0;
+  int page_num = page_id;
+  int block_num = 0;
+  int par_num = 0;
+  int line_num = 0;
+  int word_num = 0;
 
   tsv_str.add_str_int("1\t", page_num);  // level 1 - page
   tsv_str.add_str_int("\t", block_num);
@@ -1639,7 +1642,10 @@ char* TessBaseAPI::GetTSVText(int page_number) {
 
     // Add rows for any new block/paragraph/textline.
     if (res_it->IsAtBeginningOf(RIL_BLOCK)) {
-      block_num++, par_num = 0, line_num = 0, word_num = 0;
+      block_num++;
+      par_num = 0;
+      line_num = 0;
+      word_num = 0;
       tsv_str.add_str_int("2\t", page_num);  // level 2 - block
       tsv_str.add_str_int("\t", block_num);
       tsv_str.add_str_int("\t", par_num);
@@ -1649,7 +1655,9 @@ char* TessBaseAPI::GetTSVText(int page_number) {
       tsv_str += "\t-1\t\n";  // end of row for block
     }
     if (res_it->IsAtBeginningOf(RIL_PARA)) {
-      par_num++, line_num = 0, word_num = 0;
+      par_num++;
+      line_num = 0;
+      word_num = 0;
       tsv_str.add_str_int("3\t", page_num);  // level 3 - paragraph
       tsv_str.add_str_int("\t", block_num);
       tsv_str.add_str_int("\t", par_num);
@@ -1659,7 +1667,8 @@ char* TessBaseAPI::GetTSVText(int page_number) {
       tsv_str += "\t-1\t\n";  // end of row for para
     }
     if (res_it->IsAtBeginningOf(RIL_TEXTLINE)) {
-      line_num++, word_num = 0;
+      line_num++;
+      word_num = 0;
       tsv_str.add_str_int("4\t", page_num);  // level 4 - line
       tsv_str.add_str_int("\t", block_num);
       tsv_str.add_str_int("\t", par_num);

src/ccstruct/coutln.cpp

@@ -141,14 +141,14 @@ int16_t length                     //length of loop
  */
 
 C_OUTLINE::C_OUTLINE(C_OUTLINE* srcline, FCOORD rotation) : offsets(nullptr) {
-  TBOX new_box;                   //easy bounding
-  int16_t stepindex;               //index to step
-  int16_t dirdiff;                 //direction change
+  TBOX new_box;                  //easy bounding
+  int16_t stepindex;             //index to step
+  int16_t dirdiff;               //direction change
   ICOORD pos;                    //current position
   ICOORD prevpos;                //previous dest point
 
   ICOORD destpos;                //destination point
-  int16_t destindex;               //index to step
+  int16_t destindex = INT16_MAX; //index to step
   DIR128 dir;                    //coded direction
   uint8_t new_step;
 

src/classify/trainingsampleset.cpp

@@ -35,8 +35,6 @@ const int kSquareLimit = 25;
 // Prime numbers for subsampling distances.
 const int kPrime1 = 17;
 const int kPrime2 = 13;
-// Min samples from which to start discarding outliers.
-const int kMinOutlierSamples = 5;
 
 TrainingSampleSet::FontClassInfo::FontClassInfo()
   : num_raw_samples(0), canonical_sample(-1), canonical_dist(0.0f) {

src/lstm/lstmrecognizer.cpp

@@ -41,8 +41,6 @@
 
 namespace tesseract {
 
-// Max number of blob choices to return in any given position.
-const int kMaxChoices = 4;
 // Default ratio between dict and non-dict words.
 const double kDictRatio = 2.25;
 // Default certainty offset to give the dictionary a chance.

src/training/dawg2wordlist.cpp

@@ -25,8 +25,6 @@
 #include "trie.h"
 #include "unicharset.h"
 
-const int kDictDebugLevel = 1;
-
 tesseract::Dawg *LoadSquishedDawg(const UNICHARSET &unicharset,
                                   const char *filename) {
   const int kDictDebugLevel = 1;

src/training/stringrenderer.cpp

@@ -47,7 +47,6 @@ static const int kDefaultOutputResolution = 300;
 // recommendation in http://unicode.org/reports/tr14/ to avoid line-breaks at
 // hyphens and other non-alpha characters.
 static const char* kWordJoinerUTF8 = "\u2060";
-static const char32 kWordJoiner = 0x2060;
 
 static bool IsCombiner(int ch) {
   const int char_type = u_charType(ch);

src/wordrec/language_model.cpp

@@ -1127,7 +1127,7 @@ void LanguageModel::FillConsistencyInfo(
     }
     if (!word_res->blob_widths.empty()) {  // if we have widths/gaps info
       bool expected_gap_found = false;
-      float expected_gap;
+      float expected_gap = 0.0f;
       int temp_gap;
       if (fontinfo_id >= 0) {  // found a common font
         ASSERT_HOST(fontinfo_id < fontinfo_table_->size());
@@ -1140,7 +1140,6 @@ void LanguageModel::FillConsistencyInfo(
         consistency_info->inconsistent_font = true;
         // Get an average of the expected gaps in each font
         int num_addends = 0;
-        expected_gap = 0;
         int temp_fid;
         for (int i = 0; i < 4; ++i) {
           if (i == 0) {
@@ -1159,9 +1158,9 @@ void LanguageModel::FillConsistencyInfo(
             num_addends++;
           }
         }
-        expected_gap_found = (num_addends > 0);
         if (num_addends > 0) {
           expected_gap /= static_cast<float>(num_addends);
+          expected_gap_found = true;
         }
       }
       if (expected_gap_found) {