Support image width and height larger than 32767

Builds now can define LARGE_IMAGES to use 32 bit for
image dimensions instead of 16 bit.

Signed-off-by: Stefan Weil <sw@weilnetz.de>

src/ccstruct/blobs.h

@@ -28,7 +28,7 @@
 
 #include <tesseract/publictypes.h> // for OcrEngineMode
 
-#include <cstdint> // for int16_t
+#include "tesstypes.h" // for TDimension
 
 struct Pix;
 
@@ -46,8 +46,8 @@ class WERD;
 ----------------------------------------------------------------------*/
 
 struct TPOINT {
-  TPOINT() : x(0), y(0) {}
-  TPOINT(int16_t vx, int16_t vy) : x(vx), y(vy) {}
+  TPOINT() = default;
+  TPOINT(TDimension vx, TDimension vy) : x(vx), y(vy) {}
   TPOINT(const ICOORD &ic) : x(ic.x()), y(ic.y()) {}
 
   void operator+=(const TPOINT &other) {
@@ -86,8 +86,8 @@ struct TPOINT {
     return x * x + y * y;
   }
 
-  int16_t x; // absolute x coord.
-  int16_t y; // absolute y coord.
+  TDimension x = 0; // absolute x coord.
+  TDimension y = 0; // absolute y coord.
 };
 
 using VECTOR = TPOINT; // structure for coordinates.

src/ccstruct/mod128.cpp

@@ -20,7 +20,7 @@
 
 namespace tesseract {
 
-static const int16_t idirtab[] = {
+static const TDimension idirtab[] = {
     1000, 0,    998,  49,   995,  98,   989,  146,  980,   195,  970,  242,  956,  290,   941,
     336,  923,  382,  903,  427,  881,  471,  857,  514,   831,  555,  803,  595,  773,   634,
     740,  671,  707,  707,  671,  740,  634,  773,  595,   803,  555,  831,  514,  857,   471,

src/ccstruct/ocrblock.cpp

@@ -31,13 +31,14 @@ namespace tesseract {
  *
  * Constructor for a simple rectangular block.
  */
-BLOCK::BLOCK(const char *name,           ///< filename
-             bool prop,                  ///< proportional
-             int16_t kern,               ///< kerning
-             int16_t space,              ///< spacing
-             int16_t xmin,               ///< bottom left
-             int16_t ymin, int16_t xmax, ///< top right
-             int16_t ymax)
+BLOCK::BLOCK(const char *name, ///< filename
+             bool prop,        ///< proportional
+             int16_t kern,     ///< kerning
+             int16_t space,    ///< spacing
+             TDimension xmin,  ///< bottom left
+             TDimension ymin,
+             TDimension xmax,  ///< top right
+             TDimension ymax)
     : pdblk(xmin, ymin, xmax, ymax)
     , filename(name)
     , re_rotation_(1.0f, 0.0f)

src/ccstruct/ocrblock.h

@@ -39,10 +39,10 @@ class TESS_API BLOCK : public ELIST_LINK
         bool prop,        ///< proportional
         int16_t kern,     ///< kerning
         int16_t space,    ///< spacing
-        int16_t xmin,     ///< bottom left
-        int16_t ymin,
-        int16_t xmax, ///< top right
-        int16_t ymax);
+        TDimension xmin,  ///< bottom left
+        TDimension ymin,
+        TDimension xmax,  ///< top right
+        TDimension ymax);
 
   ~BLOCK() = default;
 

src/ccstruct/pdblock.cpp

@@ -42,9 +42,10 @@ constexpr ERRCODE LOSTBLOCKLINE("Can't find rectangle for line");
  * Constructor for a simple rectangular block.
  **********************************************************************/
 PDBLK::PDBLK(                   // rectangular block
-    int16_t xmin,               // bottom left
-    int16_t ymin, int16_t xmax, // top right
-    int16_t ymax)
+    TDimension xmin,            // bottom left
+    TDimension ymin,
+    TDimension xmax,            // top right
+    TDimension ymax)
     : box(ICOORD(xmin, ymin), ICOORD(xmax, ymax)) {
   // boundaries
   ICOORDELT_IT left_it = &leftside;
@@ -349,9 +350,9 @@ void BLOCK_RECT_IT::forward() { // next rectangle
  * Get the the start and width of a line in the block.
  **********************************************************************/
 
-int16_t BLOCK_LINE_IT::get_line( // get a line
-    int16_t y,                   // line to get
-    int16_t &xext                // output extent
+TDimension BLOCK_LINE_IT::get_line( // get a line
+    TDimension y,                   // line to get
+    TDimension &xext                // output extent
 ) {
   ICOORD bleft;  // bounding box
   ICOORD tright; // of block & rect

src/ccstruct/pdblock.h

@@ -41,10 +41,10 @@ class PDBLK {
     index_ = 0;
   }
   /// simple constructor
-  PDBLK(int16_t xmin, ///< bottom left
-        int16_t ymin,
-        int16_t xmax, ///< top right
-        int16_t ymax);
+  PDBLK(TDimension xmin, ///< bottom left
+        TDimension ymin,
+        TDimension xmax, ///< top right
+        TDimension ymax);
 
   /// set vertex lists
   ///@param left list of left vertices
@@ -145,8 +145,8 @@ class BLOCK_RECT_IT // rectangle iterator
   }
 
 private:
-  int16_t ymin = 0;       ///< bottom of rectangle
-  int16_t ymax = 0;       ///< top of rectangle
+  TDimension ymin = 0;    ///< bottom of rectangle
+  TDimension ymax = 0;    ///< top of rectangle
   PDBLK *block = nullptr; ///< block to iterate
   ICOORDELT_IT left_it;   ///< boundary iterators
   ICOORDELT_IT right_it;
@@ -172,7 +172,7 @@ class BLOCK_LINE_IT {
   /// get a line
   ///@param y line to get
   ///@param xext output extent
-  int16_t get_line(int16_t y, int16_t &xext);
+  TDimension get_line(TDimension y, TDimension &xext);
 
 private:
   PDBLK *block;          ///< block to iterate

src/ccstruct/points.cpp

@@ -40,12 +40,32 @@ bool FCOORD::normalise() { // Convert to unit vec
   return true;
 }
 
+// Deserialize an ICOORD.
+// For compatibility reasons it uses unsigned 16 bit coordinates
+// instead of 32 bit coordinates.
 bool ICOORD::DeSerialize(TFile *f) {
-  return f->DeSerialize(&xcoord) && f->DeSerialize(&ycoord);
+  bool success = false;
+  uint16_t coord;
+  if (f->DeSerialize(&coord)) {
+    xcoord = coord;
+    if (f->DeSerialize(&coord)) {
+      ycoord = coord;
+      success = true;
+    }
+  }
+  return success;
 }
 
+// Serialize an ICOORD.
+// For compatibility reasons it uses unsigned 16 bit coordinates
+// instead of 32 bit coordinates.
 bool ICOORD::Serialize(TFile *f) const {
-  return f->Serialize(&xcoord) && f->Serialize(&ycoord);
+  uint16_t coord;
+  coord = xcoord;
+  auto success = f->Serialize(&coord);
+  coord = ycoord;
+  success = success && f->Serialize(&coord);
+  return success;
 }
 
 // Set from the given x,y, shrinking the vector to fit if needed.

src/ccstruct/points.h

@@ -21,6 +21,7 @@
 
 #include "elst.h"
 #include "errcode.h" // for ASSERT_HOST
+#include "tesstypes.h" // for TDimension
 
 #include <tesseract/export.h> // for DLLSYM
 
@@ -43,7 +44,7 @@ class ICOORD {
   /// constructor
   ///@param xin x value
   ///@param yin y value
-  ICOORD(int16_t xin, int16_t yin) {
+  ICOORD(TDimension xin, TDimension yin) {
     xcoord = xin;
     ycoord = yin;
   }
@@ -54,20 +55,20 @@ class ICOORD {
   bool Serialize(TFile *f) const;
 
   /// access function
-  int16_t x() const {
+  TDimension x() const {
     return xcoord;
   }
   /// access_function
-  int16_t y() const {
+  TDimension y() const {
     return ycoord;
   }
 
   /// rewrite function
-  void set_x(int16_t xin) {
+  void set_x(TDimension xin) {
     xcoord = xin; // write new value
   }
   /// rewrite function
-  void set_y(int16_t yin) { // value to set
+  void set_y(TDimension yin) { // value to set
     ycoord = yin;
   }
 
@@ -128,15 +129,15 @@ class ICOORD {
   /// cross product
   friend int32_t operator*(const ICOORD &, const ICOORD &);
   /// multiply
-  friend ICOORD operator*(const ICOORD &, int16_t);
+  friend ICOORD operator*(const ICOORD &, TDimension);
   /// multiply
-  friend ICOORD operator*(int16_t, const ICOORD &);
+  friend ICOORD operator*(TDimension, const ICOORD &);
   /// multiply
-  friend ICOORD &operator*=(ICOORD &, int16_t);
+  friend ICOORD &operator*=(ICOORD &, TDimension);
   /// divide
-  friend ICOORD operator/(const ICOORD &, int16_t);
+  friend ICOORD operator/(const ICOORD &, TDimension);
   /// divide
-  friend ICOORD &operator/=(ICOORD &, int16_t);
+  friend ICOORD &operator/=(ICOORD &, TDimension);
   /// rotate
   ///@param vec by vector
   void rotate(const FCOORD &vec);
@@ -155,8 +156,8 @@ class ICOORD {
   bool DeSerialize(bool swap, FILE *fp);
 
 protected:
-  int16_t xcoord; ///< x value
-  int16_t ycoord; ///< y value
+  TDimension xcoord; ///< x value
+  TDimension ycoord; ///< y value
 };
 
 class ICOORDELT : public ELIST_LINK,
@@ -171,7 +172,7 @@ class ICOORDELT : public ELIST_LINK,
   /// constructor
   ///@param xin x value
   ///@param yin y value
-  ICOORDELT(int16_t xin, int16_t yin) {
+  ICOORDELT(TDimension xin, TDimension yin) {
     xcoord = xin;
     ycoord = yin;
   }
@@ -438,7 +439,7 @@ inline int32_t operator*( // cross product
 
 inline ICOORD operator*( // scalar multiply
     const ICOORD &op1,   // operands
-    int16_t scale) {
+    TDimension scale) {
   ICOORD result; // output
 
   result.xcoord = op1.xcoord * scale;
@@ -447,7 +448,7 @@ inline ICOORD operator*( // scalar multiply
 }
 
 inline ICOORD operator*( // scalar multiply
-    int16_t scale,
+    TDimension scale,
     const ICOORD &op1 // operands
 ) {
   ICOORD result; // output
@@ -465,7 +466,7 @@ inline ICOORD operator*( // scalar multiply
 
 inline ICOORD &operator*=( // scalar multiply
     ICOORD &op1,           // operands
-    int16_t scale) {
+    TDimension scale) {
   op1.xcoord *= scale;
   op1.ycoord *= scale;
   return op1;
@@ -479,7 +480,7 @@ inline ICOORD &operator*=( // scalar multiply
 
 inline ICOORD operator/( // scalar divide
     const ICOORD &op1,   // operands
-    int16_t scale) {
+    TDimension scale) {
   ICOORD result; // output
 
   result.xcoord = op1.xcoord / scale;
@@ -495,7 +496,7 @@ inline ICOORD operator/( // scalar divide
 
 inline ICOORD &operator/=( // scalar divide
     ICOORD &op1,           // operands
-    int16_t scale) {
+    TDimension scale) {
   op1.xcoord /= scale;
   op1.ycoord /= scale;
   return op1;
@@ -509,8 +510,8 @@ inline ICOORD &operator/=( // scalar divide
 
 inline void ICOORD::rotate( // rotate by vector
     const FCOORD &vec) {
-  auto tmp = static_cast<int16_t>(std::floor(xcoord * vec.x() - ycoord * vec.y() + 0.5f));
-  ycoord = static_cast<int16_t>(std::floor(ycoord * vec.x() + xcoord * vec.y() + 0.5f));
+  auto tmp = static_cast<TDimension>(std::floor(xcoord * vec.x() - ycoord * vec.y() + 0.5f));
+  ycoord = static_cast<TDimension>(std::floor(ycoord * vec.x() + xcoord * vec.y() + 0.5f));
   xcoord = tmp;
 }
 

src/ccstruct/polyblk.cpp

@@ -198,8 +198,8 @@ void POLY_BLOCK::rotate(FCOORD rotation) {
     pos.set_x(pt->x());
     pos.set_y(pt->y());
     pos.rotate(rotation);
-    pt->set_x(static_cast<int16_t>(floor(pos.x() + 0.5)));
-    pt->set_y(static_cast<int16_t>(floor(pos.y() + 0.5)));
+    pt->set_x(static_cast<TDimension>(floor(pos.x() + 0.5)));
+    pt->set_y(static_cast<TDimension>(floor(pos.y() + 0.5)));
     pts.forward();
   } while (!pts.at_first());
   compute_bb();
@@ -270,30 +270,25 @@ void POLY_BLOCK::plot(ScrollView *window, int32_t num) {
 }
 
 void POLY_BLOCK::fill(ScrollView *window, ScrollView::Color colour) {
-  int16_t y;
-  int16_t width;
-  PB_LINE_IT *lines;
   ICOORDELT_IT s_it;
 
-  lines = new PB_LINE_IT(this);
+  std::unique_ptr<PB_LINE_IT> lines(new PB_LINE_IT(this));
   window->Pen(colour);
 
-  for (y = this->bounding_box()->bottom(); y <= this->bounding_box()->top(); y++) {
+  for (auto y = this->bounding_box()->bottom(); y <= this->bounding_box()->top(); y++) {
     const std::unique_ptr</*non-const*/ ICOORDELT_LIST> segments(lines->get_line(y));
     if (!segments->empty()) {
       s_it.set_to_list(segments.get());
       for (s_it.mark_cycle_pt(); !s_it.cycled_list(); s_it.forward()) {
         // Note different use of ICOORDELT, x coord is x coord of pixel
         // at the start of line segment, y coord is length of line segment
         // Last pixel is start pixel + length.
-        width = s_it.data()->y();
+        auto width = s_it.data()->y();
         window->SetCursor(s_it.data()->x(), y);
         window->DrawTo(s_it.data()->x() + static_cast<float>(width), y);
       }
     }
   }
-
-  delete lines;
 }
 #endif
 
@@ -339,7 +334,7 @@ bool POLY_BLOCK::overlap(POLY_BLOCK *other) {
   return false;
 }
 
-ICOORDELT_LIST *PB_LINE_IT::get_line(int16_t y) {
+ICOORDELT_LIST *PB_LINE_IT::get_line(TDimension y) {
   ICOORDELT_IT v, r;
   ICOORDELT_LIST *result;
   ICOORDELT *x, *current, *previous;
@@ -356,7 +351,7 @@ ICOORDELT_LIST *PB_LINE_IT::get_line(int16_t y) {
       float fx =
           0.5f + previous->x() +
           (current->x() - previous->x()) * (fy - previous->y()) / (current->y() - previous->y());
-      x = new ICOORDELT(static_cast<int16_t>(fx), 0);
+      x = new ICOORDELT(static_cast<TDimension>(fx), 0);
       r.add_to_end(x);
     }
   }

src/ccstruct/polyblk.h

@@ -106,7 +106,7 @@ class PB_LINE_IT {
   // Each element of the returned list is the start (x) and extent(y) of
   // a run inside the region.
   // Delete the returned list after use.
-  ICOORDELT_LIST *get_line(int16_t y);
+  ICOORDELT_LIST *get_line(TDimension y);
 
 private:
   POLY_BLOCK *block;

src/ccstruct/rect.cpp

@@ -83,10 +83,10 @@ void TBOX::rotate_large(const FCOORD &vec) {
 
 TBOX TBOX::intersection( // shared area box
     const TBOX &box) const {
-  int16_t left;
-  int16_t bottom;
-  int16_t right;
-  int16_t top;
+  TDimension left;
+  TDimension bottom;
+  TDimension right;
+  TDimension top;
   if (overlap(box)) {
     if (box.bot_left.x() > bot_left.x()) {
       left = box.bot_left.x();