Modernize code

Makefile.am

@@ -318,7 +318,6 @@ libtesseract_la_SOURCES += src/ccstruct/blobs.cpp
 libtesseract_la_SOURCES += src/ccstruct/blread.cpp
 libtesseract_la_SOURCES += src/ccstruct/boxread.cpp
 libtesseract_la_SOURCES += src/ccstruct/boxword.cpp
-libtesseract_la_SOURCES += src/ccstruct/ccstruct.cpp
 libtesseract_la_SOURCES += src/ccstruct/coutln.cpp
 libtesseract_la_SOURCES += src/ccstruct/detlinefit.cpp
 libtesseract_la_SOURCES += src/ccstruct/dppoint.cpp

cmake/SourceLists.cmake

@@ -91,7 +91,6 @@ set(TESSERACT_SRC_CCSTRUCT
     src/ccstruct/blread.cpp
     src/ccstruct/boxread.cpp
     src/ccstruct/boxword.cpp
-    src/ccstruct/ccstruct.cpp
     src/ccstruct/coutln.cpp
     src/ccstruct/detlinefit.cpp
     src/ccstruct/dppoint.cpp

src/api/baseapi.cpp

@@ -71,6 +71,7 @@
 #include <memory>   // for std::unique_ptr
 #include <set>      // for std::pair
 #include <sstream>  // for std::stringstream
+#include <string_view>
 #include <vector>   // for std::vector
 
 #include <allheaders.h> // for pixDestroy, boxCreate, boxaAddBox, box...
@@ -125,18 +126,21 @@ static STRING_VAR(classify_font_name, kUnknownFontName,
 // /path/to/dir/[lang].[fontname].exp[num]
 // The [lang], [fontname] and [num] fields should not have '.' characters.
 // If the global parameter classify_font_name is set, its value is used instead.
-static void ExtractFontName(const char* filename, std::string* fontname) {
+static void ExtractFontName(std::string_view filename, std::string* fontname) {
   *fontname = classify_font_name;
   if (*fontname == kUnknownFontName) {
     // filename is expected to be of the form [lang].[fontname].exp[num]
     // The [lang], [fontname] and [num] fields should not have '.' characters.
-    const char *basename = strrchr(filename, '/');
-    const char *firstdot = strchr(basename ? basename : filename, '.');
-    const char *lastdot  = strrchr(filename, '.');
-    if (firstdot != lastdot && firstdot != nullptr && lastdot != nullptr) {
+    auto basename_pos = filename.find_last_of('/');
+    auto view = (basename_pos != std::string_view::npos)
+                    ? filename.substr(basename_pos + 1)
+                    : filename;
+    auto firstdot = view.find_first_of('.');
+    auto lastdot = view.find_last_of('.');
+    if (firstdot != lastdot && firstdot != std::string_view::npos &&
+        lastdot != std::string_view::npos) {
       ++firstdot;
-      *fontname = firstdot;
-      fontname->resize(lastdot - firstdot);
+      *fontname = view.substr(firstdot, lastdot - firstdot);
     }
   }
 }
@@ -225,7 +229,7 @@ bool TessBaseAPI::GetIntVariable(const char *name, int *value) const {
   if (p == nullptr) {
     return false;
   }
-  *value = (int32_t)(*p);
+  *value = static_cast<int32_t>(*p);
   return true;
 }
 
@@ -251,7 +255,7 @@ bool TessBaseAPI::GetDoubleVariable(const char *name, double *value) const {
   if (p == nullptr) {
     return false;
   }
-  *value = (double)(*p);
+  *value = static_cast<double>(*p);
   return true;
 }
 
@@ -527,7 +531,7 @@ void TessBaseAPI::SetImage(Pix *pix) {
       // remove alpha channel from png
       Pix *p1 = pixRemoveAlpha(pix);
       pixSetSpp(p1, 3);
-      (void)pixCopy(pix, p1);
+      static_cast<void>(pixCopy(pix, p1));
       pixDestroy(&p1);
     }
     thresholder_->SetImage(pix);
@@ -813,7 +817,7 @@ int TessBaseAPI::Recognize(ETEXT_DESC *monitor) {
 #ifndef DISABLED_LEGACY_ENGINE
   } else if (tesseract_->tessedit_train_from_boxes) {
     std::string fontname;
-    ExtractFontName(output_file_.c_str(), &fontname);
+    ExtractFontName(output_file_, &fontname);
     tesseract_->ApplyBoxTraining(fontname, page_res_);
   } else if (tesseract_->tessedit_ambigs_training) {
     FILE *training_output_file = tesseract_->init_recog_training(input_file_.c_str());

src/ccmain/osdetect.cpp

@@ -112,7 +112,7 @@ int OSResults::get_best_script(int orientation_id) const {
 }
 
 // Print the script scores for all possible orientations.
-void OSResults::print_scores(void) const {
+void OSResults::print_scores() const {
   for (int i = 0; i < 4; ++i) {
     tprintf("Orientation id #%d", i);
     print_scores(i);

src/ccmain/pgedit.cpp

@@ -778,10 +778,10 @@ bool Tesseract::word_display(PAGE_RES_IT *pr_it) {
   // display bounding box
   if (word->display_flag(DF_BOX)) {
     word->bounding_box().plot(image_win,
-                              static_cast<ScrollView::Color>((int32_t)editor_image_word_bb_color),
-                              static_cast<ScrollView::Color>((int32_t)editor_image_word_bb_color));
+                              static_cast<ScrollView::Color>(static_cast<int32_t>(editor_image_word_bb_color)),
+                              static_cast<ScrollView::Color>(static_cast<int32_t>(editor_image_word_bb_color)));
 
-    auto c = static_cast<ScrollView::Color>((int32_t)editor_image_blob_bb_color);
+    auto c = static_cast<ScrollView::Color>(static_cast<int32_t>(editor_image_blob_bb_color));
     image_win->Pen(c);
     // cblob iterator
     C_BLOB_IT c_it(word->cblob_list());
@@ -859,8 +859,8 @@ bool Tesseract::word_display(PAGE_RES_IT *pr_it) {
 
   if (!displayed_something) { // display BBox anyway
     word->bounding_box().plot(image_win,
-                              static_cast<ScrollView::Color>((int32_t)editor_image_word_bb_color),
-                              static_cast<ScrollView::Color>((int32_t)editor_image_word_bb_color));
+                              static_cast<ScrollView::Color>(static_cast<int32_t>(editor_image_word_bb_color)),
+                              static_cast<ScrollView::Color>(static_cast<int32_t>(editor_image_word_bb_color)));
   }
   return true;
 }

src/ccmain/resultiterator.cpp

@@ -43,7 +43,7 @@ ResultIterator::ResultIterator(const LTRResultIterator &resit) : LTRResultIterat
   auto *p = ParamUtils::FindParam<BoolParam>(
       "preserve_interword_spaces", GlobalParams()->bool_params, tesseract_->params()->bool_params);
   if (p != nullptr) {
-    preserve_interword_spaces_ = (bool)(*p);
+    preserve_interword_spaces_ = static_cast<bool>(*p);
   }
 
   current_paragraph_is_ltr_ = CurrentParagraphIsLtr();
@@ -781,7 +781,7 @@ bool ResultIterator::BidiDebug(int min_level) const {
   auto *p = ParamUtils::FindParam<IntParam>("bidi_debug", GlobalParams()->int_params,
                                             tesseract_->params()->int_params);
   if (p != nullptr) {
-    debug_level = (int32_t)(*p);
+    debug_level = static_cast<int32_t>(*p);
   }
   return debug_level >= min_level;
 }

src/ccmain/thresholder.cpp

@@ -242,8 +242,8 @@ std::tuple<bool, Image, Image, Image> ImageThresholder::Threshold(
     }
 
     r = pixSauvolaBinarizeTiled(pix_grey, half_window_size, kfactor, nx, ny,
-                               (PIX**)pix_thresholds,
-                                (PIX**)pix_binary);
+                                static_cast<PIX **>(pix_thresholds),
+                                static_cast<PIX **>(pix_binary));
   } else { // if (method == ThresholdMethod::LeptonicaOtsu)
     int tile_size;
     double tile_size_factor;
@@ -269,8 +269,8 @@ std::tuple<bool, Image, Image, Image> ImageThresholder::Threshold(
     r = pixOtsuAdaptiveThreshold(pix_grey, tile_size, tile_size,
                                  half_smooth_size, half_smooth_size,
                                  score_fraction,
-                                 (PIX**)pix_thresholds,
-                                 (PIX**)pix_binary);
+                                 static_cast<PIX **>(pix_thresholds),
+                                 static_cast<PIX **>(pix_binary));
   }
 
   bool ok = (r == 0);

src/ccstruct/blobbox.h

@@ -34,6 +34,7 @@
 #include "tprintf.h"    // for tprintf
 #include "werd.h"       // for WERD_LIST
 
+#include <array>
 #include <cinttypes> // for PRId32
 #include <cmath>     // for std::sqrt
 #include <cstdint>   // for int16_t, int32_t
@@ -537,11 +538,11 @@ class BLOBNBOX : public ELIST<BLOBNBOX>::LINK {
   int32_t line_crossings_;                   // Number of line intersections touched.
   BLOBNBOX *base_char_blob_;                 // The blob that was the base char.
   tesseract::ColPartition *owner_;           // Who will delete me when I am not needed
-  BLOBNBOX *neighbours_[BND_COUNT];
+  std::array<BLOBNBOX *, BND_COUNT> neighbours_;
   float horz_stroke_width_ = 0.0f; // Median horizontal stroke width
   float vert_stroke_width_ = 0.0f; // Median vertical stroke width
   float area_stroke_width_ = 0.0f; // Stroke width from area/perimeter ratio.
-  bool good_stroke_neighbours_[BND_COUNT];
+  std::array<bool, BND_COUNT> good_stroke_neighbours_;
   bool horz_possible_;   // Could be part of horizontal flow.
   bool vert_possible_;   // Could be part of vertical flow.
   bool leader_on_left_;  // There is a leader to the left.

src/ccstruct/ccstruct.h

@@ -28,11 +28,11 @@ class CCStruct : public CCUtil {
   // Globally accessible constants.
   // APPROXIMATIONS of the fractions of the character cell taken by
   // the descenders, ascenders, and x-height.
-  static const double kDescenderFraction; // = 0.25;
-  static const double kXHeightFraction;   // = 0.5;
-  static const double kAscenderFraction;  // = 0.25;
+  static constexpr double kDescenderFraction = 0.25;
+  static constexpr double kXHeightFraction = 0.5;
+  static constexpr double kAscenderFraction = 0.25;
   // Derived value giving the x-height as a fraction of cap-height.
-  static const double kXHeightCapRatio; // = XHeight/(XHeight + Ascender).
+  static constexpr double kXHeightCapRatio = kXHeightFraction / (kXHeightFraction + kAscenderFraction);
 };
 
 } // namespace tesseract

src/ccstruct/coutln.cpp

@@ -42,7 +42,7 @@
 
 namespace tesseract {
 
-ICOORD C_OUTLINE::step_coords[4] = {ICOORD(-1, 0), ICOORD(0, -1), ICOORD(1, 0), ICOORD(0, 1)};
+std::array<ICOORD, 4> C_OUTLINE::step_coords = {ICOORD(-1, 0), ICOORD(0, -1), ICOORD(1, 0), ICOORD(0, 1)};
 
 /**
  * @name C_OUTLINE::C_OUTLINE

src/ccstruct/coutln.h

@@ -27,8 +27,9 @@
 
 #include <tesseract/export.h> // for DLLSYM
 
-#include <cstdint> // for int16_t, int32_t
+#include <array>
 #include <bitset>  // for std::bitset<16>
+#include <cstdint> // for int16_t, int32_t
 
 struct Pix;
 
@@ -289,7 +290,7 @@ class C_OUTLINE : public ELIST<C_OUTLINE>::LINK {
   std::vector<uint8_t> steps; // step array
   EdgeOffset *offsets;     // Higher precision edge.
   C_OUTLINE_LIST children; // child elements
-  static ICOORD step_coords[4];
+  static std::array<ICOORD, 4> step_coords;
 };
 
 } // namespace tesseract

src/ccstruct/debugpixa.h

@@ -15,7 +15,7 @@ class DebugPixa {
   DebugPixa() {
     pixa_ = pixaCreate(0);
 #ifdef TESSERACT_DISABLE_DEBUG_FONTS
-    fonts_ = NULL;
+    fonts_ = nullptr;
 #else
     fonts_ = bmfCreate(nullptr, 14);
 #endif

src/ccstruct/normalis.h

@@ -29,9 +29,9 @@ struct Pix;
 
 namespace tesseract {
 
-const int kBlnCellHeight = 256;    // Full-height for baseline normalization.
-const int kBlnXHeight = 128;       // x-height for baseline normalization.
-const int kBlnBaselineOffset = 64; // offset for baseline normalization.
+constexpr int kBlnCellHeight = 256;    // Full-height for baseline normalization.
+constexpr int kBlnXHeight = 128;       // x-height for baseline normalization.
+constexpr int kBlnBaselineOffset = 64; // offset for baseline normalization.
 
 class BLOCK;
 class FCOORD;

src/ccstruct/pageres.cpp

@@ -40,6 +40,7 @@
 #include <cassert> // for assert
 #include <cstdint> // for INT32_MAX
 #include <cstring> // for strlen
+#include <string_view> // for std::string_view
 
 struct Pix;
 
@@ -1021,23 +1022,27 @@ void WERD_RES::MergeAdjacentBlobs(unsigned index) {
 // Utility function for fix_quotes
 // Return true if the next character in the string (given the UTF8 length in
 // bytes) is a quote character.
-static int is_simple_quote(const char *signed_str, int length) {
-  const auto *str = reinterpret_cast<const unsigned char *>(signed_str);
+static int is_simple_quote(std::string_view str) {
   // Standard 1 byte quotes.
-  return (length == 1 && (*str == '\'' || *str == '`')) ||
-         // UTF-8 3 bytes curved quotes.
-         (length == 3 &&
-          ((*str == 0xe2 && *(str + 1) == 0x80 && *(str + 2) == 0x98) ||
-           (*str == 0xe2 && *(str + 1) == 0x80 && *(str + 2) == 0x99)));
+  if (str.size() == 1 && (str[0] == '\'' || str[0] == '`')) {
+    return true;
+  }
+  // UTF-8 3 bytes curved quotes.
+  if (str.size() == 3 && static_cast<unsigned char>(str[0]) == 0xe2 &&
+      static_cast<unsigned char>(str[1]) == 0x80 &&
+      (static_cast<unsigned char>(str[2]) == 0x98 ||
+       static_cast<unsigned char>(str[2]) == 0x99)) {
+    return true;
+  }
+  return false;
 }
 
 // Callback helper for fix_quotes returns a double quote if both
 // arguments are quote, otherwise INVALID_UNICHAR_ID.
 UNICHAR_ID WERD_RES::BothQuotes(UNICHAR_ID id1, UNICHAR_ID id2) {
   const char *ch = uch_set->id_to_unichar(id1);
   const char *next_ch = uch_set->id_to_unichar(id2);
-  if (is_simple_quote(ch, strlen(ch)) &&
-      is_simple_quote(next_ch, strlen(next_ch))) {
+  if (is_simple_quote(ch) && is_simple_quote(next_ch)) {
     return uch_set->unichar_to_id("\"");
   }
   return INVALID_UNICHAR_ID;