-#include <pam.h>
-#include <stdint.h>
-#include <inttypes.h>
-#include <assert.h>
-#include <string.h>
-#include <stdlib.h>
-#define eassert assert
-#define debug stdout
+#include "ocr.h"
-typedef struct {
- unsigned long rgb; /* on screen */
- char c; /* canonical */
-} CanonColourInfo;
-
-static int height, width;
-static char *image;
+static CanonImage *cim;
-static void debug_flush(void) {
+void debug_flush(void) {
eassert(!fflush(debug));
eassert(!ferror(debug));
}
Point br;
} Rect;
-static inline char get(int x, int y) { return image[y * width + x]; }
+static inline char get(int x, int y) { return cim->d[y * cim->w + x]; }
static inline char get_p(Point p) { return get(p.x,p.y); }
#define START_MAIN {200,200}
#define MIN_COLUMNS 6
#define INTERESTING_COLUMNS 6
+#define TEXT_COLUMNS 2
#define MAX_COLUMNS 7
static Rect mainr = { START_MAIN,START_MAIN };
static int commbasey, comminty;
static int colrightx[INTERESTING_COLUMNS];
+static int text_h;
+static OcrReader *rd;
-
-static const CanonColourInfo canoncolourinfos[]= {
+const CanonColourInfo canoncolourinfos[]= {
{ 0x475A5E, '*' }, /* edge */
{ 0x2C5F7A, '*' }, /* edge just under box heading shadow */
{ 0x7D9094, '+' }, /* interbox */
- { 0xBDC5BF, ' ' }, /* background - pale */
- { 0xADB5AF, ' ' }, /* background - dark */
+
+ { 0xBDC5BF, ' ' }, /* background - pale Sugar cane, etc. */
+ { 0xADB5AF, ' ' }, /* background - dark */
+ { 0xC7E1C3, ' ' }, /* background - pale Swill, etc. */
+ { 0xB5CFB1, ' ' }, /* background - dark */
+ { 0xD6CEB0, ' ' }, /* background - pale Madder, etc. */
+ { 0xC8C0A2, ' ' }, /* background - dark */
+ { 0xE0E1D3, ' ' }, /* background - pale Lorandite, etc. */
+ { 0xD0D1C3, ' ' }, /* background - dark */
+ { 0xE5E6C1, ' ' }, /* background - pale Cloth */
+ { 0xD7D8B3, ' ' }, /* background - dark */
+ { 0xEDDED9, ' ' }, /* background - pale Dye */
+ { 0xDACBC6, ' ' }, /* background - dark */
+ { 0xD3DEDF, ' ' }, /* background - pale Paint */
+ { 0xC5D0D1, ' ' }, /* background - dark */
+ { 0xDCD1CF, ' ' }, /* background - pale Enamel */
+ { 0xCEC3C1, ' ' }, /* background - dark */
+ { 0xF3F6F5, ' ' }, /* background - pale fruit */
+ { 0xE2E7E5, ' ' }, /* background - dark */
+
{ 0x000000, 'o' }, /* foreground */
{ 0xD4B356, ' ' }, /* background (cursor) */
{ 0xFFFFFF, 'o' }, /* foreground (cursor) */
+
+ { 0x5B93BF, '_' }, /* selector dropdown background */
+ { 0xD7C94F, 'X' }, /* selector dropdown foreground */
{ 0,0 }
};
}
static void debug_rect(const char *what, int whati, Rect rr) {
+#ifdef DEBUG_RECTANGLES
int y,r,w;
fprintf(debug, "%s %d: %d,%d..%d,%d:\n", what, whati,
rr.tl.x,rr.tl.y, rr.br.x,rr.br.y);
w= rr.br.x - rr.tl.x + 1;
for (y=rr.tl.y; y<=rr.br.y; y++) {
fprintf(debug, "%4d%*s|", y, rr.tl.x,"");
- r= fwrite(image + y*width + rr.tl.x, 1, w, debug);
+ r= fwrite(cim->d + y*cim->w + rr.tl.x, 1, w, debug);
eassert(r==w);
fputc('|',debug);
fputc('\n',debug);
}
+#endif
debug_flush();
}
} while(0)
static void find_structure(void) {
- Rect whole = { {0,0}, {width-1,height-1} };
+ Rect whole = { {0,0}, {cim->w-1,cim->h-1} };
WALK_UNTIL_MUST(mainr.tl, x,-1, whole.tl.x, '*');
WALK_UNTIL_MUST(mainr.tl, y,-1, whole.tl.y, '*');
down.y++;
WALK_UNTIL_MUST(down, y,+1, mainr.br.y, '+');
+#ifdef DEBUG_RECTANGLES
int xscaleunit, y,x;
for (y=0, xscaleunit=1; y<4; y++, xscaleunit*=10) {
fprintf(debug," ");
- for (x=0; x<=width; x++) {
+ for (x=0; x<=cim->w; x++) {
if (x % xscaleunit) fputc(' ',debug);
else fprintf(debug,"%d",(x / xscaleunit)%10);
}
fputc('\n',debug);
}
+#endif
commbasey= up.y;
comminty= down.y - up.y + 2;
- fprintf(debug, "up.y=%d down.y=%d commbasey=%d comminty=%d\n",
- up.y,down.y, commbasey,comminty);
Point across= { mainr.tl.x, commbasey };
int colno=0;
eassert(colno < MAX_COLUMNS);
int colrx= across.x;
if (colrx > mainr.br.x) colrx= mainr.br.x;
- if (colno < INTERESTING_COLUMNS) {
+ if (colno < INTERESTING_COLUMNS)
colrightx[colno]= colrx;
- fprintf(debug,"colrightx[%d]= %d\n",colno,colrx);
- } else {
- fprintf(debug,"extra colr %d %d\n",colno,colrx);
- }
colno++;
across.x++;
}
eassert(colno >= MIN_COLUMNS);
+
+ text_h = comminty - 1;
}
static void find_commodity(int offset, Rect *rr) {
require_rectangle_r(*cellr, " o");
}
-static void load_image_and_canonify(void) {
+CanonImage *alloc_canon_image(int w, int h) {
+ CanonImage *im= malloc(sizeof(CanonImage) + w*h);
+ eassert(im);
+ im->w= w;
+ im->h= h;
+ memset(im->d,'?',w*h);
+ return im;
+}
+
+CanonImage *file_read_image(FILE *f) {
struct pam inpam;
- unsigned char rgb[3];
- int x,y,r;
- const CanonColourInfo *cci;
+ unsigned char rgb_buf[3];
+ CanonImage *im;
- pnm_readpaminit(stdin, &inpam, sizeof(inpam));
- height= inpam.height;
- width= inpam.width;
+ pnm_readpaminit(f, &inpam, sizeof(inpam));
eassert(inpam.maxval == 255);
eassert(inpam.bytes_per_sample == 1);
- image= malloc(width*height);
- eassert(image);
- memset(image,'?',width*height);
-
- for (y=0; y<height; y++) {
- for (x=0; x<width; x++) {
- r= fread(&rgb,1,3,stdin); eassert(r==3);
- unsigned long rgb_l=
- ((unsigned long)rgb[0]<<16) |
- ((unsigned long)rgb[1]<<8) |
- (rgb[2]);
- for (cci=canoncolourinfos; cci->c; cci++)
- if (cci->rgb == rgb_l) {
- image[y*width + x]= cci->c;
- break;
- }
- }
- fprintf(debug, "%4d ",y);
- r= fwrite(image + y*width, 1,width, debug); eassert(r==width);
- fputc('\n',debug);
- }
- debug_flush();
-}
-
-typedef uint32_t Pixcol;
-#define PSPIXCOL(priscan) priscan##32
-
-typedef struct {
- Pixcol col;
- struct OCRDatabaseNode *then;
-} OCRDatabaseLink;
-
-#define MAXGLYPHCHRS 3
-
-typedef struct OCRDatabaseNode {
- char s[MAXGLYPHCHRS+1]; /* null-terminated; "" means no match here */
- int nlinks, alinks;
- OCRDatabaseLink *links;
-} OCRDatabaseNode;
-
-#define N_OCR_CONTEXTS 2
-static OCRDatabaseNode ocr_contexts[N_OCR_CONTEXTS];
+ CANONICALISE_IMAGE(im, inpam.width, inpam.height, {
+ r= fread(&rgb_buf,1,3,f); eassert(r==3);
-static void load_ocr_database(void) {
- int ctx,nchrs;
- OCRDatabaseNode *current, *additional;
- char chrs[MAXGLYPHCHRS+1];
- Pixcol cv;
- int r,i,j;
+ rgb=
+ ((unsigned long)rgb_buf[0]<<16) |
+ ((unsigned long)rgb_buf[1]<<8) |
+ (rgb_buf[2]);
+ });
- FILE *db= fopen("database","r"); eassert(db);
-
- for (;;) {
- r= fscanf(db, "%d %d", &ctx, &nchrs);
- if (r==EOF) break;
- eassert(r==2);
- eassert(ctx>=0 && ctx<N_OCR_CONTEXTS);
- eassert(nchrs>0 && nchrs<=MAXGLYPHCHRS);
-
- for (i=0; i<nchrs; i++) {
- int c;
- r= fscanf(db, "%x", &c); eassert(r==1);
- eassert(c>0 && c<=255);
- chrs[i]= c;
- }
- chrs[nchrs]= 0;
-
- int twidth;
- r= fscanf(db, "%d", &twidth); eassert(r==1);
- current= &ocr_contexts[ctx];
- for (i=0; i<twidth; i++) {
- r= fscanf(db, "%"PSPIXCOL(SCNx), &cv); eassert(r==1);
- for (j=0; j<current->nlinks; j++)
- if (current->links[j].col == cv) {
- current= current->links[j].then;
- goto found_link;
- }
-
- additional= malloc(sizeof(*additional)); eassert(additional);
- additional->s[0]= 0;
- additional->nlinks= additional->alinks= 0;
- additional->links= 0;
- if (current->nlinks==current->alinks) {
- current->alinks++;
- current->alinks<<=1;
- current->links= realloc(current->links,
- sizeof(*current->links) * current->alinks);
- eassert(current->links);
- }
- current->links[current->nlinks].col= cv;
- current->links[current->nlinks].then= additional;
- current->nlinks++;
- current= additional;
+ return im;
+}
- found_link:;
- }
+static void load_image_and_canonify(void) {
+ cim= file_read_image(stdin);
+}
- eassert(!current->s[0]);
- strcpy(current->s, chrs);
- }
- eassert(!ferror(db));
- eassert(feof(db));
- fclose(db);
-}
+static void ocr_rectangle(Rect r, const OcrCellType ct) {
+ OcrResultGlyph *results, *res;
-static void ocr_rectangle(Rect r) {
int w= r.br.x - r.tl.x + 1;
- int h= r.br.y - r.tl.y + 1;
Pixcol cols[w+1];
int x,y;
for (x=0; x<w; x++) {
Pixcol cx, rv;
- for (y=0, cx=0, rv=1; y<h; y++, rv<<=1) {
+ for (y=0, cx=0, rv=1; y<text_h; y++, rv<<=1) {
switch (get(x+r.tl.x, y+r.tl.y)) {
case ' ': break;
case 'o': cx |= rv; break;
}
cols[w]= 0;
- int nspaces=0;
- int ctx=1,i;
- x=0;
-
- for (;;) {
- if (x>w) break;
-
- if (!cols[x]) {
- nspaces++;
- x++;
- if (nspaces>3) ctx=1;
- continue;
- }
-
- OCRDatabaseNode *current=0, *lastmatch=0;
- int startx=x;
- int afterlastmatchx=-1;
- current= &ocr_contexts[ctx];
- for (;;) {
- if (x>w) break;
- Pixcol cv= cols[x];
- for (i=0; i<current->nlinks; i++)
- if (current->links[i].col == cv)
- goto found;
- /* not found */
- break;
- found:
- x++;
- current= current->links[i].then;
- if (current->s[0]) { lastmatch=current; afterlastmatchx=x; }
- }
-
- if (!lastmatch) {
- int x2;
- for (x2=x+1; x2<w && cols[x2]; x2++);
- printf("UNKNOWN x=%d ctx=%d %d..%d\n",x, ctx, startx,x2);
- x++;
- } else {
- printf("OUTPUT x=%d `%s'\n", x, lastmatch->s);
- x= afterlastmatchx;
- ctx= 0;
- }
- }
+ results= ocr(rd,ct,w,cols);
+ printf("YES! \"");
+ for (res=results; res->s; res++)
+ printf("%s",res->s);
+ printf("\"\n");
+ eassert(!ferror(stdout));
+ eassert(!fflush(stdout));
}
-int main(void) {
+int main_test(void) {
Rect thisr, entryr;
int tryrect, colno;
- load_ocr_database();
load_image_and_canonify();
find_structure();
+ rd= ocr_init(text_h);
- for (tryrect= +height; tryrect >= -height; tryrect--) {
+ for (tryrect= +cim->h; tryrect >= -cim->h; tryrect--) {
find_commodity(tryrect, &thisr);
if (thisr.tl.x < 0)
continue;
for (colno=0; colno<MIN_COLUMNS; colno++) {
find_table_entry(thisr,colno,&entryr);
- ocr_rectangle(entryr);
+ ocr_rectangle(entryr,
+ colno<TEXT_COLUMNS
+ ? &ocr_celltype_text
+ : &ocr_celltype_number);
}
}
return 0;
}
+
+const char *get_vardir(void) { return "."; }