pctb/ocr.h

   1 #ifndef OCR_H
   2 #define OCR_H
   3
   4
   5 #define DEBUG_RECTANGLES
   6 // #define DEBUG_OCR
   7
   8
   9
  10 #define _GNU_SOURCE
  11
  12 #include <pam.h>
  13 #include <stdint.h>
  14 #include <inttypes.h>
  15 #include <assert.h>
  16 #include <string.h>
  17 #include <stdlib.h>
  18 #include <stdio.h>
  19 #include <stdarg.h>
  20 #include <time.h>
  21 #include <limits.h>
  22
  23 #include <sys/types.h>
  24 #include <sys/wait.h>
  25 #include <sys/time.h>
  26
  27
  28 typedef struct {
  29   int w,h;
  30   char d[];
  31 } CanonImage;
  32
  33
  34 typedef uint32_t Pixcol;
  35 #define PSPIXCOL(priscan) priscan##32
  36
  37 typedef struct {
  38   const char *s; /* valid until next call to ocr() */
  39   int l,r; /* column numbers */
  40   unsigned ctxmap; /* match context index */
  41 } OcrResultGlyph;
  42
  43 typedef const struct OcrCellTypeInfo *OcrCellType;
  44 extern const struct OcrCellTypeInfo ocr_celltype_text;
  45 extern const struct OcrCellTypeInfo ocr_celltype_number;
  46
  47 typedef struct OcrReader OcrReader;
  48 OcrReader *ocr_init(int h);
  49
  50 OcrResultGlyph *ocr(OcrReader *rd, OcrCellType, int w, Pixcol cols[]);
  51   /* return value is array terminated by {0,-1,-1}
  52    * array is valid until next call to ocr()
  53    */
  54
  55 void debug_flush(void);
  56
  57 void find_structure(CanonImage *im);
  58
  59 #define eassert assert
  60 #define debug stdout
  61
  62 const char *get_vardir(void);
  63
  64 CanonImage *file_read_image(FILE *f);
  65 int main_test(void);
  66
  67 #define MAX_PAGES 100
  68 extern CanonImage *page_images[MAX_PAGES];
  69 extern int npages;
  70
  71
  72 typedef struct {
  73   unsigned long rgb; /* on screen */
  74   char c; /* canonical */
  75 } CanonColourInfo;
  76
  77 extern const CanonColourInfo canoncolourinfos[];
  78
  79 CanonImage *alloc_canon_image(int w, int h);
  80
  81 #ifdef DEBUG_RECTANGLES
  82 # define CANIMG_DEBUG_RECTANGLE_1LINE(im,w,h)   \
  83       fprintf(debug, "%4d ",y);                 \
  84       r= fwrite(im->d + y*w, 1,w, debug);       \
  85       eassert(r==w);                            \
  86       fputc('\n',debug);
  87 #else
  88 # define CANIMG_DEBUG_RECTANGLE_1LINE(im,y,h) /* nothing */
  89 #endif
  90
  91 #define CANONICALISE_IMAGE(im,w,h, COMPUTE_RGB) do{             \
  92     /* compute_rgb should be a number of statements, or         \
  93      * a block, which assigns to                                \
  94      *   unsigned long rgb;                                     \
  95      * given the values of                                      \
  96      *   int x,y;                                               \
  97      * all of which are anamorphic.  Result is stored in im.    \
  98      */                                                         \
  99     (im)= alloc_canon_image((w), (h));                          \
 100                                                                 \
 101     int x,y,r;                                                  \
 102     for (y=0; y<(h); y++) {                                     \
 103       for (x=0; x<(w); x++) {                                   \
 104         const CanonColourInfo *cci;                             \
 105         unsigned long rgb;                                      \
 106         COMPUTE_RGB;                                            \
 107         for (cci=canoncolourinfos; cci->c; cci++)               \
 108           if (cci->rgb == rgb) {                                \
 109             (im)->d[y*(w) + x]= cci->c;                         \
 110             break;                                              \
 111           }                                                     \
 112       }                                                         \
 113       CANIMG_DEBUG_RECTANGLE_1LINE((im),(w),(h))                \
 114     }                                                           \
 115     debug_flush();                                              \
 116   }while(0)
 117
 118
 119 #endif /*OCR_H*/