tesseract 4.1.0

include/leptonica/array.h

@@ -136,24 +136,23 @@ typedef struct L_Bytea L_BYTEA;
 /*------------------------------------------------------------------------*
  *                              Array flags                               *
  *------------------------------------------------------------------------*/
-    /*! Flags for interpolation in Numa */
+/*! Numa Interpolation */
 enum {
     L_LINEAR_INTERP = 1,        /*!< linear     */
     L_QUADRATIC_INTERP = 2      /*!< quadratic  */
 };
 
-    /*! Flags for added borders in Numa and Fpix */
+/*! Border Adding */
 enum {
     L_CONTINUED_BORDER = 1,    /*!< extended with same value                  */
     L_SLOPE_BORDER = 2,        /*!< extended with constant normal derivative  */
     L_MIRRORED_BORDER = 3      /*!< mirrored                                  */
 };
 
-    /*! Flags for data type converted from Numa */
+/*! Numa Data Conversion */
 enum {
     L_INTEGER_VALUE = 1,        /*!< convert to integer  */
     L_FLOAT_VALUE = 2           /*!< convert to float    */
 };
 
-
 #endif  /* LEPTONICA_ARRAY_H */

include/leptonica/arrayaccess.h

@@ -62,21 +62,27 @@
  *  defined for SET macros.  If SET_DATA_QBIT were defined as a
  *  compound macro, in analogy to l_setDataQbit(), it requires
  *  surrounding braces:
+ * \code
  *     #define  SET_DATA_QBIT(pdata, n, val) \
  *        {l_uint32 *_TEMP_WORD_PTR_; \
  *         _TEMP_WORD_PTR_ = (l_uint32 *)(pdata) + ((n) >> 3); \
  *         *_TEMP_WORD_PTR_ &= ~(0xf0000000 >> (4 * ((n) & 7))); \
  *         *_TEMP_WORD_PTR_ |= (((val) & 15) << (28 - 4 * ((n) & 7)));}
+ * \endcode
  *  but if used in an if/else
+ * \code
  *      if (x)
  *         SET_DATA_QBIT(...);
  *      else
  *         ...
+ * \endcode
  *  the compiler sees
+ * \code
  *      if (x)
  *         {......};
  *      else
  *         ...
+ * \endcode
  *  The semicolon comes after the brace and will not compile.
  *  This can be fixed in the call by either omitting the semicolon
  *  or requiring another set of braces around SET_DATA_QBIT(), but

include/leptonica/bmf.h

@@ -34,6 +34,7 @@
  */
 
     /*! Constants for deciding when text block is divided into paragraphs */
+/*! Split Text */
 enum {
     SPLIT_ON_LEADING_WHITE = 1,    /*!< tab or space at beginning of line   */
     SPLIT_ON_BLANK_LINE    = 2,    /*!< newline with optional white space   */

include/leptonica/bmp.h

@@ -46,20 +46,36 @@
  *                       BMP file header                       *
  *-------------------------------------------------------------*/
 
-/*! BMP file header */
+/*! BMP file header
+ *
+ * Notes:
+ *  (1) The bfSize field is stored as a 32 bit integer and includes
+ *      the size of the BMP_FileHeader, BMP_InfoHeader, the color
+ *      table (if any), and the size of the DIB bits.
+ *  (2) The bfOffBits field is also stored as a 32 bit integer and
+ *      contains the absolute offset in bytes of the image data
+ *      in this file. Some bmp files have additional data after the
+ *      BMP_InfoHeader and before the color table (if it exists).
+ *      However, enabling reading of these files makes the reader
+ *      vulnerable to various malware attacks.  Therefore we do not
+ *      read bmp files with extra data, and require that the size
+ *      of the color table in bytes is
+ *         offset - sizeof(BMP_FileHeader) - sizeof(BMP_InfoHeader)
+ *  (3) Use arrays of l_uint8[] to make an endianness agnostic
+ *      access to the BMP_FileHeader easier.
+ */
 struct BMP_FileHeader
 {
-    l_int16        bfType;         /*!< file type; must be "BM"            */
-    l_int16        bfSize;         /*!< length of the file;
+    l_uint8        bfType[2];      /*!< file type; must be "BM"            */
+    l_uint8        bfSize[4];      /*!< length of the file;
                                        sizeof(BMP_FileHeader) +
                                        sizeof(BMP_InfoHeader) +
+                                       size of optional extra data +
                                        size of color table +
                                        size of DIB bits                    */
-    l_int16        bfFill1;        /*!< remainder of the bfSize field      */
-    l_int16        bfReserved1;    /*!< don't care (set to 0)              */
-    l_int16        bfReserved2;    /*!< don't care (set to 0)              */
-    l_int16        bfOffBits;      /*!< offset from beginning of file      */
-    l_int16        bfFill2;        /*!< remainder of the bfOffBits field   */
+    l_uint8        bfReserved1[2]; /*!< don't care (set to 0)              */
+    l_uint8        bfReserved2[2]; /*!< don't care (set to 0)              */
+    l_uint8        bfOffBits[4];   /*!< offset from beginning of file      */
 };
 typedef struct BMP_FileHeader  BMP_FH;
 
@@ -92,4 +108,17 @@ typedef struct BMP_InfoHeader  BMP_IH;
 #define BMP_IHBYTES  sizeof(BMP_IH)
 
 
+/*-------------------------------------------------------------*
+ *           Align BMP headers on 4 byte boundaries            *
+ *-------------------------------------------------------------*/
+
+/*! BMP_IH is misaligned, causing crashes on some big-endians.
+ *  A packed struct forces alignment. */
+#if defined(__GNUC__)
+typedef struct __attribute__((__packed__)) {
+  BMP_FH bmpfh;
+  BMP_IH bmpih;
+} BMP_HEADER;
+#endif
+
 #endif  /* LEPTONICA_BMP_H */

include/leptonica/ccbord.h

@@ -37,12 +37,14 @@
  */
 
     /*! Use in ccbaStepChainsToPixCoords() */
+/*! CCB Coords */
 enum {
       CCB_LOCAL_COORDS = 1,
       CCB_GLOBAL_COORDS = 2
 };
 
     /*! Use in ccbaGenerateSPGlobalLocs() */
+/*! CCB Points */
 enum {
       CCB_SAVE_ALL_PTS = 1,
       CCB_SAVE_TURNING_PTS = 2

include/leptonica/environ.h

@@ -97,21 +97,39 @@ typedef uintptr_t l_uintptr_t;
  *  non-functioning stubs to be linked.
  */
 #if !defined(HAVE_CONFIG_H) && !defined(ANDROID_BUILD) && !defined(OS_IOS)
-#define  HAVE_LIBJPEG       1
-#define  HAVE_LIBTIFF       1
-#define  HAVE_LIBPNG        1
-#define  HAVE_LIBZ          1
-#define  HAVE_LIBGIF        0
-#define  HAVE_LIBUNGIF      0
-#define  HAVE_LIBWEBP       0
-#define  HAVE_LIBJP2K       0
 
-/*-------------------------------------------------------------------------*
- * 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      *
- * header in angle brackets here.                                          *
- *-------------------------------------------------------------------------*/
-#define  LIBJP2K_HEADER   <openjpeg-2.3/openjpeg.h>
+  #if !defined(HAVE_LIBJPEG)
+  #define  HAVE_LIBJPEG       1
+  #endif
+  #if !defined(HAVE_LIBTIFF)
+  #define  HAVE_LIBTIFF       1
+  #endif
+  #if !defined(HAVE_LIBPNG)
+  #define  HAVE_LIBPNG        1
+  #endif
+  #if !defined(HAVE_LIBZ)
+  #define  HAVE_LIBZ          1
+  #endif
+  #if !defined(HAVE_LIBGIF)
+  #define  HAVE_LIBGIF        0
+  #endif
+  #if !defined(HAVE_LIBUNGIF)
+  #define  HAVE_LIBUNGIF      0
+  #endif
+  #if !defined(HAVE_LIBWEBP)
+  #define  HAVE_LIBWEBP       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    *
+   * header in angle brackets here.                                        *
+   *-----------------------------------------------------------------------*/
+  #define  LIBJP2K_HEADER   <openjpeg-2.3/openjpeg.h>
+
 #endif  /* ! HAVE_CONFIG_H etc. */
 
 /*--------------------------------------------------------------------*
@@ -171,6 +189,7 @@ typedef uintptr_t l_uintptr_t;
 /*--------------------------------------------------------------------*
  *                          Built-in types                            *
  *--------------------------------------------------------------------*/
+typedef int                     l_ok;    /*!< return type 0 if OK, 1 on error */
 typedef signed char             l_int8;     /*!< signed 8-bit value */
 typedef unsigned char           l_uint8;    /*!< unsigned 8-bit value */
 typedef short                   l_int16;    /*!< signed 16-bit value */
@@ -207,16 +226,16 @@ LEPT_DLL extern l_int32  LeptDebugOK;  /* default is 0 */
  *------------------------------------------------------------------------*/
 #ifndef L_MIN
 /*! Minimum of %x and %y */
-#define L_MIN(x,y)   (((x) < (y)) ? (x) : (y))
+#define L_MIN(x, y)   (((x) < (y)) ? (x) : (y))
 #endif
 
 #ifndef L_MAX
 /*! Maximum of %x and %y */
-#define L_MAX(x,y)   (((x) > (y)) ? (x) : (y))
+#define L_MAX(x, y)   (((x) > (y)) ? (x) : (y))
 #endif
 
 #ifndef L_ABS
-/*! Absoulute value of %x */
+/*! Absolute value of %x */
 #define L_ABS(x)     (((x) < 0) ? (-1 * (x)) : (x))
 #endif
 
@@ -263,7 +282,7 @@ LEPT_DLL extern l_int32  LeptDebugOK;  /* default is 0 */
 /*------------------------------------------------------------------------*
  *                    Simple search state variables                       *
  *------------------------------------------------------------------------*/
-/*! Simple search state variables */
+/*! Search State */
 enum {
     L_NOT_FOUND = 0,
     L_FOUND = 1
@@ -273,7 +292,7 @@ enum {
 /*------------------------------------------------------------------------*
  *                     Path separator conversion                          *
  *------------------------------------------------------------------------*/
-/*! Path separator conversion */
+/*! Path Separators */
 enum {
     UNIX_PATH_SEPCHAR = 0,
     WIN_PATH_SEPCHAR = 1
@@ -346,7 +365,8 @@ typedef struct L_WallTimer  L_WALLTIMER;
  *  "Print messages that have an equal or greater severity than this."    *
  *------------------------------------------------------------------------*/
 
-/*! Control printing of error, warning and info messages */
+    /*! Control printing of error, warning and info messages */
+/*! Message Control */
 enum {
     L_SEVERITY_EXTERNAL = 0,   /* Get the severity from the environment   */
     L_SEVERITY_ALL      = 1,   /* Lowest severity: print all messages     */
@@ -466,51 +486,51 @@ LEPT_DLL extern l_int32  LeptMsgSeverity;
 #ifdef  NO_CONSOLE_IO
 
   #define PROCNAME(name)
-  #define ERROR_INT(a,b,c)            ((l_int32)(c))
-  #define ERROR_FLOAT(a,b,c)          ((l_float32)(c))
-  #define ERROR_PTR(a,b,c)            ((void *)(c))
-  #define L_ERROR(a,...)
-  #define L_WARNING(a,...)
-  #define L_INFO(a,...)
+  #define ERROR_INT(a, b, c)            ((l_int32)(c))
+  #define ERROR_FLOAT(a, b, c)          ((l_float32)(c))
+  #define ERROR_PTR(a, b, c)            ((void *)(c))
+  #define L_ERROR(a, ...)
+  #define L_WARNING(a, ...)
+  #define L_INFO(a, ...)
 
 #else
 
   #define PROCNAME(name)              static const char procName[] = name
-  #define IF_SEV(l,t,f) \
+  #define IF_SEV(l, t, f) \
       ((l) >= MINIMUM_SEVERITY && (l) >= LeptMsgSeverity ? (t) : (f))
 
-  #define ERROR_INT(a,b,c) \
-      IF_SEV(L_SEVERITY_ERROR, returnErrorInt((a),(b),(c)), (l_int32)(c))
-  #define ERROR_FLOAT(a,b,c) \
-      IF_SEV(L_SEVERITY_ERROR, returnErrorFloat((a),(b),(c)), (l_float32)(c))
-  #define ERROR_PTR(a,b,c) \
-      IF_SEV(L_SEVERITY_ERROR, returnErrorPtr((a),(b),(c)), (void *)(c))
+  #define ERROR_INT(a, b, c) \
+      IF_SEV(L_SEVERITY_ERROR, returnErrorInt((a), (b), (c)), (l_int32)(c))
+  #define ERROR_FLOAT(a, b, c) \
+      IF_SEV(L_SEVERITY_ERROR, returnErrorFloat((a), (b), (c)), (l_float32)(c))
+  #define ERROR_PTR(a, b, c) \
+      IF_SEV(L_SEVERITY_ERROR, returnErrorPtr((a), (b), (c)), (void *)(c))
 
-  #define L_ERROR(a,...) \
+  #define L_ERROR(a, ...) \
       IF_SEV(L_SEVERITY_ERROR, \
              (void)fprintf(stderr, "Error in %s: " a, __VA_ARGS__), \
              (void)0)
-  #define L_WARNING(a,...) \
+  #define L_WARNING(a, ...) \
       IF_SEV(L_SEVERITY_WARNING, \
              (void)fprintf(stderr, "Warning in %s: " a, __VA_ARGS__), \
              (void)0)
-  #define L_INFO(a,...) \
+  #define L_INFO(a, ...) \
       IF_SEV(L_SEVERITY_INFO, \
              (void)fprintf(stderr, "Info in %s: " a, __VA_ARGS__), \
              (void)0)
 
 #if 0  /* Alternative method for controlling L_* message output */
-  #define L_ERROR(a,...) \
+  #define L_ERROR(a, ...) \
     { if (L_SEVERITY_ERROR >= MINIMUM_SEVERITY && \
           L_SEVERITY_ERROR >= LeptMsgSeverity) \
           fprintf(stderr, "Error in %s: " a, __VA_ARGS__) \
     }
-  #define L_WARNING(a,...) \
+  #define L_WARNING(a, ...) \
     { if (L_SEVERITY_WARNING >= MINIMUM_SEVERITY && \
           L_SEVERITY_WARNING >= LeptMsgSeverity) \
           fprintf(stderr, "Warning in %s: " a, __VA_ARGS__) \
     }
-  #define L_INFO(a,...) \
+  #define L_INFO(a, ...) \
     { if (L_SEVERITY_INFO >= MINIMUM_SEVERITY && \
           L_SEVERITY_INFO >= LeptMsgSeverity) \
              fprintf(stderr, "Info in %s: " a, __VA_ARGS__) \