OcrReader *rd;
OcrCellType ct;
int w;
- Pixcol *cols;
+ const Pixcol *cols;
int x;
unsigned ctxmap;
} FindCharArgs;
goto x;
FGETSLINE;
- dbassert(!strcmp(lbuf,"# ypp-sc-tools pctb font v1"));
+ dbassert(!strcmp(lbuf,"# ypp-sc-tools pctb font v2"));
dbassert( dbfile_scanf("%d", &h) == 1);
dbassert(h==rd->h);
for (;;) {
FGETSLINE;
if (!lbuf[0]) { dbassert(current != &rd->contexts[ctxi]); break; }
- char *ep;
- cv= strtoul(lbuf,&ep,16); dbassert(!*ep);
- dbassert(!(cv & ~((1UL << rd->h)-1)));
+ dbassert( strlen(lbuf) == rd->h );
+ FILLZERO(cv);
+ int y;
+ for (y=0; y<h; y++) {
+ char pb[2];
+ pb[0]= lbuf[y];
+ pb[1]= 0;
+ char *ep;
+ unsigned pv= strtoul(pb,&ep,16); dbassert(!*ep);
+ pixcol_p_add(&cv, y, pv);
+ }
for (j=0; j<current->nlinks; j++)
- if (current->links[j].col == cv) {
+ if (!pixcol_cmp(&cv, ¤t->links[j].col)) {
current= current->links[j].then;
goto found_link;
}
fprintf(resolver,"}");
}
-static void callout_unknown(OcrReader *rd, int w, Pixcol cols[],
+static void callout_unknown(OcrReader *rd, int w, const Pixcol cols[],
int unk_l, int unk_r, unsigned unk_ctxmap) {
int c,i, x,y;
const OcrResultGlyph *s;
const char *p;
- Pixcol pv;
FILE *resolver= resolve_start();
if (!resolver || !(o_flags & ff_editcharset))
fputc('\n',resolver);
fprintf(resolver,
- "/* XPM */\n"
- "static char *t[] = {\n"
- "/* columns rows colors chars-per-pixel */\n"
- "\"%d %d 2 1\",\n"
- "\" c black\",\n"
- "\"o c white\",\n",
- w,rd->h);
- for (y=0, pv=1; y<rd->h; y++, pv<<=1) {
- fputc('"',resolver);
+ "P2\n%d %d\n%d\n", w, rd->h, AAMAXVAL);
+ for (y=0; y<rd->h; y++) {
for (x=0; x<w; x++)
- fputc(cols[x] & pv ? 'o' : ' ', resolver);
- fputs("\",\n",resolver);
+ fprintf(resolver, " %x", pixcol_p_get(&cols[x], y));
+ fputc('\n',resolver);
}
- fputs("};\n",resolver);
resolve_finish();
readdb(rd);
debugf(" | x=%d",x);
if (x > fca->w) break;
Pixcol cv= fca->cols[x];
- debugf(" cv=%"PSPIXCOL(PRIx),cv);
+ debugf(" cv="PIXCOL_PRFMT, PIXCOL_PRVAL(cv));
for (i=0; i<current->nlinks; i++)
- if (current->links[i].col == cv)
+ if (!pixcol_cmp(&cv, ¤t->links[i].col))
goto found;
/* not found */
debugf(" ?");
const char *ocr_celltype_name(OcrCellType ct) { return ct->name; }
-OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
+OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w,
+ const Pixcol cols[]) {
int nspaces;
int x;
fca.ctxmap= ct->initial;
rd->nresults=0;
debugf("OCR h=%d w=%d",rd->h,w);
- for (x=0; x<w; x++) debugf(" %"PSPIXCOL(PRIx),cols[x]);
+ for (x=0; x<w; x++) debugf(" "PIXCOL_PRFMT, PIXCOL_PRVAL(cols[x]));
debugf("\n");
debug_flush();
if (x>=w)
break;
- if (!cols[x]) {
+ if (!pixcol_nonzero(&cols[x])) {
nspaces++;
x++;
if (nspaces == ct->space_spaces) {
} else {
int rx;
debugf(" || UNKNOWN");
- for (rx=x; rx<w && cols[rx]; rx++);
+ for (rx=x; rx<w && pixcol_nonzero(&cols[rx]); rx++);
debugf(" x=%d ctxmap=%x %d..%d\n",x, fca.ctxmap, x,rx);
debug_flush();
callout_unknown(rd, w,cols, x,rx-1, fca.ctxmap);
OcrReader *ocr_init(int h) {
OcrReader *rd;
+ assert(h <= OCR_MAX_H);
+
if (o_flags & ff_dict_fetch) {
char *fetchfile= masprintf("char%d",h);
progress("Updating %s...",fetchfile);
void ocr_showcharsets(void) {
DIR *d;
struct dirent *de;
- char found[32];
+ char found[OCR_MAX_H];
pcre *fnpat;
int matchvec[10];
char hbuf[10];
#define AAMAXVAL ((1<<AADEPTH)-1)
-typedef uint32_t Pixcol;
-#define PSPIXCOL(priscan) priscan##32
+typedef uint32_t Pixcolv;
+#define PIXCOL_P_PER_WORD (32 / AADEPTH)
+
+#define OCR_MAX_H 30
+#define PIXCOL_WORDS ((OCR_MAX_H + PIXCOL_P_PER_WORD - 1) / PIXCOL_P_PER_WORD)
+
+typedef struct { Pixcolv w[PIXCOL_WORDS]; } Pixcol;
+
+static inline int pixcol_p_word(int y) {
+ return y / PIXCOL_P_PER_WORD;
+}
+static inline int pixcol_p_shift(int y) {
+ return (y % PIXCOL_P_PER_WORD)*AADEPTH;
+}
+static inline void pixcol_p_add(Pixcol *pixcol, int y, unsigned pixval) {
+ pixcol->w[pixcol_p_word(y)] |= pixval << pixcol_p_shift(y);
+}
+static inline unsigned int pixcol_p_get(const Pixcol *pixcol, int y) {
+ return (pixcol->w[pixcol_p_word(y)] >> pixcol_p_shift(y)) & AAMAXVAL;
+}
+static inline int pixcol_cmp(const Pixcol *pixcol1, const Pixcol *pixcol2) {
+ return memcmp(pixcol1, pixcol2, sizeof(*pixcol1));
+}
+static inline int pixcol_nonzero(const Pixcol *pixcol) {
+ static const Pixcol zero;
+ return pixcol_cmp(pixcol, &zero);
+}
+
+#if AADEPTH==3
+# define PRPIXCOL1 "%0*" PRIo32
+#else
+# error need to implement PRPIXCOL1 for this AADEPTH
+#endif
+
+#if PIXCOL_WORDS==3
+# define PIXCOL_PRFMT \
+ PRPIXCOL1 "-" \
+ PRPIXCOL1 "-" \
+ PRPIXCOL1
+# define PIXCOL_PRVAL(pixcol) \
+ PIXCOL_P_PER_WORD, (pixcol).w[2], \
+ PIXCOL_P_PER_WORD, (pixcol).w[1], \
+ PIXCOL_P_PER_WORD, (pixcol).w[0]
+#else
+# error need to implement PIXCOL_PR{FMT,VAL} for this PIXCOL_WORDS
+#endif
typedef struct {
const char *s; /* valid until next call to ocr() */
OcrReader *ocr_init(int h);
void ocr_showcharsets(void);
-OcrResultGlyph *ocr(OcrReader *rd, OcrCellType, int w, Pixcol cols[]);
+OcrResultGlyph *ocr(OcrReader *rd, OcrCellType, int w, const Pixcol cols[]);
/* return value is array terminated by {0,-1,-1}
* array is valid until next call to ocr()
*/
SET_ONCE(text_h, comminty - 1);
if (max_relevant_y_r)
SET_ONCE(*max_relevant_y_r, mainr.br.y + 10);
+
+ MUST( text_h <= OCR_MAX_H, MI(text_h) );
}
void check_correct_commodities(void) {
Pixcol cols[w+1];
int x,y;
for (x=0; x<w; x++) {
- Pixcol cx, rv;
- for (y=0, cx=0, rv=1; y<text_h; y++, rv<<=1) {
+ FILLZERO(cols[x]);
+ for (y=0; y<text_h; y++) {
Point here= { x+r.tl.x, y+r.tl.y };
int pixel= get_p(here);
- switch (pixel) {
- case ' ': break;
- case '0': break;
- case '0'+AAMAXVAL: cx |= rv; break;
- default:
- MUST(!"wrong pixel",
- MC(pixel);MP(here);MSB(ocr_celltype_name(ct));MR(r); );
- }
+ if (pixel==' ') pixel= '0';
+ MUST( pixel >= '0' && pixel <= '0'+AAMAXVAL,
+ MC(pixel);MP(here);MSB(ocr_celltype_name(ct));MR(r); );
+ pixcol_p_add(&cols[x], y, pixel-'0');
}
- cols[x]= cx;
}
- cols[w]= 0;
+ FILLZERO(cols[w]);
results= ocr(rd,ct,w,cols);
for (res=results; res->s; res++)