Tesseract 4.1.1 plus ucrt libs

Author: jeroen

include/leptonica/allheaders.h

@@ -121,7 +121,7 @@
      *--------------------------------------------------*/
 /*! 1 bit access - get */
 #define  GET_DATA_BIT(pdata, n) \
-    ((*((l_uint32 *)(pdata) + ((n) >> 5)) >> (31 - ((n) & 31))) & 1)
+    ((*((const l_uint32 *)(pdata) + ((n) >> 5)) >> (31 - ((n) & 31))) & 1)
 
 /*! 1 bit access - set */
 #define  SET_DATA_BIT(pdata, n) \
@@ -136,21 +136,21 @@
      *((l_uint32 *)(pdata) + ((n) >> 5)) = \
         ((*((l_uint32 *)(pdata) + ((n) >> 5)) \
         & (~(0x80000000 >> ((n) & 31)))) \
-        | ((val) << (31 - ((n) & 31))))
+        | ((l_uint32)(val) << (31 - ((n) & 31))))
 
     /*--------------------------------------------------*
      *                     2 bit access                 *
      *--------------------------------------------------*/
 /*! 2 bit access - get */
 #define  GET_DATA_DIBIT(pdata, n) \
-    ((*((l_uint32 *)(pdata) + ((n) >> 4)) >> (2 * (15 - ((n) & 15)))) & 3)
+    ((*((const l_uint32 *)(pdata) + ((n) >> 4)) >> (2 * (15 - ((n) & 15)))) & 3)
 
 /*! 2 bit access - set value (0 ... 3) */
 #define  SET_DATA_DIBIT(pdata, n, val) \
      *((l_uint32 *)(pdata) + ((n) >> 4)) = \
         ((*((l_uint32 *)(pdata) + ((n) >> 4)) \
         & (~(0xc0000000 >> (2 * ((n) & 15))))) \
-        | (((val) & 3) << (30 - 2 * ((n) & 15))))
+        | ((l_uint32)((val) & 3) << (30 - 2 * ((n) & 15))))
 
 /*! 2 bit access - clear */
 #define  CLEAR_DATA_DIBIT(pdata, n) \
@@ -162,14 +162,14 @@
      *--------------------------------------------------*/
 /*! 4 bit access - get */
 #define  GET_DATA_QBIT(pdata, n) \
-     ((*((l_uint32 *)(pdata) + ((n) >> 3)) >> (4 * (7 - ((n) & 7)))) & 0xf)
+     ((*((const l_uint32 *)(pdata) + ((n) >> 3)) >> (4 * (7 - ((n) & 7)))) & 0xf)
 
 /*! 4 bit access - set value (0 ... 15) */
 #define  SET_DATA_QBIT(pdata, n, val) \
      *((l_uint32 *)(pdata) + ((n) >> 3)) = \
         ((*((l_uint32 *)(pdata) + ((n) >> 3)) \
         & (~(0xf0000000 >> (4 * ((n) & 7))))) \
-        | (((val) & 15) << (28 - 4 * ((n) & 7))))
+        | ((l_uint32)((val) & 15) << (28 - 4 * ((n) & 7))))
 
 /*! 4 bit access - clear */
 #define  CLEAR_DATA_QBIT(pdata, n) \
@@ -182,11 +182,11 @@
 #ifdef  L_BIG_ENDIAN
 /*! 8 bit access - get */
 #define  GET_DATA_BYTE(pdata, n) \
-             (*((l_uint8 *)(pdata) + (n)))
+             (*((const l_uint8 *)(pdata) + (n)))
 #else  /* L_LITTLE_ENDIAN */
 /*! 8 bit access - get */
 #define  GET_DATA_BYTE(pdata, n) \
-             (*(l_uint8 *)((l_uintptr_t)((l_uint8 *)(pdata) + (n)) ^ 3))
+             (*(l_uint8 *)((l_uintptr_t)((const l_uint8 *)(pdata) + (n)) ^ 3))
 #endif  /* L_BIG_ENDIAN */
 
 #ifdef  L_BIG_ENDIAN
@@ -206,11 +206,11 @@
 #ifdef  L_BIG_ENDIAN
 /*! 16 bit access - get */
 #define  GET_DATA_TWO_BYTES(pdata, n) \
-             (*((l_uint16 *)(pdata) + (n)))
+             (*((const l_uint16 *)(pdata) + (n)))
 #else  /* L_LITTLE_ENDIAN */
 /*! 16 bit access - get */
 #define  GET_DATA_TWO_BYTES(pdata, n) \
-             (*(l_uint16 *)((l_uintptr_t)((l_uint16 *)(pdata) + (n)) ^ 2))
+             (*(l_uint16 *)((l_uintptr_t)((const l_uint16 *)(pdata) + (n)) ^ 2))
 #endif  /* L_BIG_ENDIAN */
 
 #ifdef  L_BIG_ENDIAN
@@ -229,7 +229,7 @@
      *--------------------------------------------------*/
 /*! 32 bit access - get */
 #define  GET_DATA_FOUR_BYTES(pdata, n) \
-             (*((l_uint32 *)(pdata) + (n)))
+             (*((const l_uint32 *)(pdata) + (n)))
 
 /*! 32 bit access - set (0 ... 4294967295) */
 #define  SET_DATA_FOUR_BYTES(pdata, n, val) \

include/leptonica/arrayaccess.h

@@ -119,10 +119,14 @@ typedef uintptr_t l_uintptr_t;
   #if !defined(HAVE_LIBWEBP)
   #define  HAVE_LIBWEBP       0
   #endif
+  #if !defined(HAVE_LIBWEBP_ANIM)
+  #define  HAVE_LIBWEBP_ANIM  0
+  #endif
   #if !defined(HAVE_LIBJP2K)
   #define  HAVE_LIBJP2K       0
   #endif
 
+
   /*-----------------------------------------------------------------------*
    * Leptonica supports OpenJPEG 2.0+.  If you have a version of openjpeg  *
    * (HAVE_LIBJP2K == 1) that is >= 2.0, set the path to the openjpeg.h    *
@@ -330,6 +334,11 @@ typedef struct L_WallTimer  L_WALLTIMER;
 /*------------------------------------------------------------------------*
  *         Control printing of error, warning, and info messages          *
  *                                                                        *
+ *  Leptonica never sends output to stdout.  By default, all messages     *
+ *  go to stderr.  However, we provide a mechanism for runtime            *
+ *  redirection of output, using a custom stderr handler defined          *
+ *  by the user.  See utils1.c for details and examples.                  *
+ *                                                                        *
  *  To omit all messages to stderr, simply define NO_CONSOLE_IO on the    *
  *  command line.  For finer grained control, we have a mechanism         *
  *  based on the message severity level.  The following assumes that      *
@@ -428,7 +437,7 @@ enum {
 #endif
 
 
-/*!  The run-time message severity threshold is defined in utils.c.  */
+/*!  The run-time message severity threshold is defined in utils1.c.  */
 LEPT_DLL extern l_int32  LeptMsgSeverity;
 
 /*
@@ -508,32 +517,32 @@ LEPT_DLL extern l_int32  LeptMsgSeverity;
 
   #define L_ERROR(a, ...) \
       IF_SEV(L_SEVERITY_ERROR, \
-             (void)fprintf(stderr, "Error in %s: " a, __VA_ARGS__), \
+             (void)lept_stderr("Error in %s: " a, __VA_ARGS__), \
              (void)0)
   #define L_WARNING(a, ...) \
       IF_SEV(L_SEVERITY_WARNING, \
-             (void)fprintf(stderr, "Warning in %s: " a, __VA_ARGS__), \
+             (void)lept_stderr("Warning in %s: " a, __VA_ARGS__), \
              (void)0)
   #define L_INFO(a, ...) \
       IF_SEV(L_SEVERITY_INFO, \
-             (void)fprintf(stderr, "Info in %s: " a, __VA_ARGS__), \
+             (void)lept_stderr("Info in %s: " a, __VA_ARGS__), \
              (void)0)
 
 #if 0  /* Alternative method for controlling L_* message output */
   #define L_ERROR(a, ...) \
     { if (L_SEVERITY_ERROR >= MINIMUM_SEVERITY && \
           L_SEVERITY_ERROR >= LeptMsgSeverity) \
-          fprintf(stderr, "Error in %s: " a, __VA_ARGS__) \
+          lept_stderr("Error in %s: " a, __VA_ARGS__) \
     }
   #define L_WARNING(a, ...) \
     { if (L_SEVERITY_WARNING >= MINIMUM_SEVERITY && \
           L_SEVERITY_WARNING >= LeptMsgSeverity) \
-          fprintf(stderr, "Warning in %s: " a, __VA_ARGS__) \
+          lept_stderr("Warning in %s: " a, __VA_ARGS__) \
     }
   #define L_INFO(a, ...) \
     { if (L_SEVERITY_INFO >= MINIMUM_SEVERITY && \
           L_SEVERITY_INFO >= LeptMsgSeverity) \
-             fprintf(stderr, "Info in %s: " a, __VA_ARGS__) \
+          lept_stderr("Info in %s: " a, __VA_ARGS__) \
     }
 #endif
 

include/leptonica/environ.h

@@ -43,7 +43,7 @@
 
 #define  GPLOT_VERSION_NUMBER    1
 
-#define  NUM_GPLOT_STYLES      5
+#define  NUM_GPLOT_STYLES  5
 enum GPLOT_STYLE {
     GPLOT_LINES       = 0,
     GPLOT_POINTS      = 1,
@@ -52,13 +52,14 @@ enum GPLOT_STYLE {
     GPLOT_DOTS        = 4
 };
 
-#define  NUM_GPLOT_OUTPUTS     5
+#define  NUM_GPLOT_OUTPUTS  6
 enum GPLOT_OUTPUT {
     GPLOT_NONE  = 0,
     GPLOT_PNG   = 1,
     GPLOT_PS    = 2,
     GPLOT_EPS   = 3,
-    GPLOT_LATEX = 4
+    GPLOT_LATEX = 4,
+    GPLOT_PNM   = 5,
 };
 
 enum GPLOT_SCALING {
@@ -79,7 +80,7 @@ struct GPlot
     struct Sarray *cmddata;    /*!< command file contents            */
     struct Sarray *datanames;  /*!< data file names                  */
     struct Sarray *plotdata;   /*!< plot data (1 string/file)        */
-    struct Sarray *plottitles; /*!< title for each individual plot   */
+    struct Sarray *plotlabels; /*!< label for each individual plot   */
     struct Numa   *plotstyles; /*!< plot style for individual plots  */
     l_int32        nplots;     /*!< current number of plots          */
     char          *outname;    /*!< output file name                 */

include/leptonica/gplot.h

@@ -111,6 +111,12 @@ enum {
     IFF_SPIX           = 19
 };
 
+/* Convenient macro for checking requested tiff output */
+#define  L_FORMAT_IS_TIFF(f)  ((f) == IFF_TIFF || (f) == IFF_TIFF_PACKBITS || \
+                               (f) == IFF_TIFF_RLE || (f) == IFF_TIFF_G3 || \
+                               (f) == IFF_TIFF_G4 || (f) == IFF_TIFF_LZW || \
+                               (f) == IFF_TIFF_ZIP || (f) == IFF_TIFF_JPEG)
+
 
 /* --------------------------------------------------------------- *
  *                         Format header ids                       *

include/leptonica/imageio.h

@@ -100,8 +100,9 @@
  *         Box size adjustment flags
  *         Flags for modifying box boundaries using a second box
  *         Handling overlapping bounding boxes in boxa
+ *         Selecting or making a box from two (intersecting) boxes
  *         Flags for replacing invalid boxes
- *         Flags for box corners
+ *         Flags for box corners and center
  *         Horizontal warp
  *         Pixel selection for resampling
  *         Thinning flags
@@ -1109,6 +1110,18 @@ enum {
 };
 
 
+/*-------------------------------------------------------------------------*
+ *        Selecting or making a box from two (intersecting) boxes          *
+ *-------------------------------------------------------------------------*/
+/*! Box Combine or Select */
+enum {
+    L_GEOMETRIC_UNION = 1,         /*!< use union of two boxes             */
+    L_GEOMETRIC_INTERSECTION = 2,  /*!< use intersection of two boxes      */
+    L_LARGEST_AREA = 3,            /*!< use box with largest area          */
+    L_SMALLEST_AREA = 4            /*!< use box with smallest area         */
+};
+
+
 /*-------------------------------------------------------------------------*
  *                    Flags for replacing invalid boxes                    *
  *-------------------------------------------------------------------------*/
@@ -1120,14 +1133,15 @@ enum {
 
 
 /*-------------------------------------------------------------------------*
- *                          Flags for box corners                          *
+ *                    Flags for box corners and center                     *
  *-------------------------------------------------------------------------*/
-/*! Box Corners */
+/*! Box Corners and Center */
 enum {
     L_UPPER_LEFT = 1,         /*!< UL corner                               */
     L_UPPER_RIGHT = 2,        /*!< UR corner                               */
     L_LOWER_LEFT = 3,         /*!< LL corner                               */
-    L_LOWER_RIGHT = 4         /*!< LR corner                               */
+    L_LOWER_RIGHT = 4,        /*!< LR corner                               */
+    L_BOX_CENTER = 5          /*!< center                                  */
 };
 
 

include/leptonica/pix.h

@@ -115,6 +115,7 @@ typedef enum TessPageSegMode {
   PSM_SINGLE_CHAR,
   PSM_SPARSE_TEXT,
   PSM_SPARSE_TEXT_OSD,
+  PSM_RAW_LINE,
   PSM_COUNT
 } TessPageSegMode;
 typedef enum TessPageIteratorLevel {

include/tesseract/capi.h

@@ -21,6 +21,8 @@
 
 #include <algorithm>
 #include <cassert>
+#include <climits>      // for LONG_MAX
+#include <cstdint>      // for uint32_t
 #include <cstdio>
 #include <cstdlib>
 
@@ -375,7 +377,7 @@ inline bool LoadDataFromFile(const char* filename, GenericVector<char>* data) {
   FILE* fp = fopen(filename, "rb");
   if (fp != nullptr) {
     fseek(fp, 0, SEEK_END);
-    long size = ftell(fp);
+    auto size = std::ftell(fp);
     fseek(fp, 0, SEEK_SET);
     // Trying to open a directory on Linux sets size to LONG_MAX. Catch it here.
     if (size > 0 && size < LONG_MAX) {
@@ -407,17 +409,6 @@ inline bool SaveDataToFile(const GenericVector<char>& data,
   fclose(fp);
   return result;
 }
-// Reads a file as a vector of STRING.
-inline bool LoadFileLinesToStrings(const STRING& filename,
-                                   GenericVector<STRING>* lines) {
-  GenericVector<char> data;
-  if (!LoadDataFromFile(filename.string(), &data)) {
-    return false;
-  }
-  STRING lines_str(&data[0], data.size());
-  lines_str.split('\n', lines);
-  return true;
-}
 
 template <typename T>
 bool cmp_eq(T const& t1, T const& t2) {

include/tesseract/genericvector.h

@@ -24,6 +24,7 @@
 #include <string>  // for std::string
 #include "genericvector.h"
 #include "platform.h"
+#include "strngs.h"             // for STRING
 
 struct Pix;
 

include/tesseract/renderer.h

@@ -20,11 +20,11 @@
 
 #define TESSERACT_MAJOR_VERSION 4
 #define TESSERACT_MINOR_VERSION 1
-#define TESSERACT_MICRO_VERSION 0
+#define TESSERACT_MICRO_VERSION 1
 #define TESSERACT_VERSION \
   (TESSERACT_MAJOR_VERSION << 16 | \
    TESSERACT_MINOR_VERSION << 8 | \
    TESSERACT_MICRO_VERSION)
-#define TESSERACT_VERSION_STR "4.1.0"
+#define TESSERACT_VERSION_STR "4.1.1"
 
 #endif  // TESSERACT_API_VERSION_H_