X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?a=blobdiff_plain;f=pctb%2Focr.c;h=27271c766495406dedce819b8dfc2feea431cacd;hb=bc6ff7bc24f1839450325f2dbb46ef287ea2fb64;hp=1f315c8d91c2d5fe1fafdae6474fb8ab145abc9a;hpb=7ca1164cb2e18550cbdf1684802b6573a6165a8c;p=ypp-sc-tools.db-test.git diff --git a/pctb/ocr.c b/pctb/ocr.c index 1f315c8..27271c7 100644 --- a/pctb/ocr.c +++ b/pctb/ocr.c @@ -53,13 +53,22 @@ typedef struct { int rx; } FindCharResults; -enum { ct_Lower, ct_Upper, ct_Word, ct_Digit }; -static const char *context_names[]= { - "Lower", /* bit 0, value 001 */ - "Upper", /* bit 1, value 002 */ - "Word", /* bit 2, value 004 */ - "Digit", /* bit 3, value 010 */ +#define FOR_EACH_CONTEXT(EACH) \ + EACH(Word) \ + EACH(Upper) \ + EACH(Lower) \ + EACH(Digit) + +#define FEC_ENUM(Context) ct_##Context, +#define FEC_BIT(Context) ctf_##Context = 1 << ct_##Context, +enum { + FOR_EACH_CONTEXT(FEC_ENUM) + FOR_EACH_CONTEXT(FEC_BIT) }; + +#define FEC_STRINGS(Context) #Context, +static const char *context_names[]= { FOR_EACH_CONTEXT(FEC_STRINGS) }; + struct OcrCellTypeInfo { /* bitmaps of indices into context_names: */ unsigned initial, nextword, midword; @@ -235,7 +244,9 @@ static void callout_unknown(OcrReader *rd, int w, Pixcol cols[], for (i=0, s=rd->results; inresults; i++, s++) { if (!strcmp(s->s," ")) continue; fprintf(resolver," %d %d ",s->l,s->r); - cu_pr_ctxmap(resolver, 1u << s->ctxi); + cu_pr_ctxmap(resolver, 1u << s->match); + fprintf(resolver," "); + cu_pr_ctxmap(resolver, 1u << s->ctxmap); fprintf(resolver," "); for (p=s->s; (c= *p); p++) { if (c=='\\') fprintf(resolver,"\\%c",c); @@ -265,7 +276,8 @@ static void callout_unknown(OcrReader *rd, int w, Pixcol cols[], readdb(rd); } -static void add_result(OcrReader *rd, const char *s, int l, int r, int ctxi) { +static void add_result(OcrReader *rd, const char *s, int l, int r, + int match, unsigned ctxmap) { if (rd->nresults >= rd->aresults) { rd->aresults++; rd->aresults<<=1; rd->results= mrealloc(rd->results, sizeof(*rd->results)*rd->aresults); @@ -273,7 +285,8 @@ static void add_result(OcrReader *rd, const char *s, int l, int r, int ctxi) { rd->results[rd->nresults].s= s; rd->results[rd->nresults].l= l; rd->results[rd->nresults].r= r; - rd->results[rd->nresults].ctxi= ctxi; + rd->results[rd->nresults].match= match; + rd->results[rd->nresults].ctxmap= ctxmap; rd->nresults++; } @@ -359,15 +372,15 @@ static int findchar_select_text(const FindCharArgs *fca, } const struct OcrCellTypeInfo ocr_celltype_number= { - 010,010,010, + ctf_Digit, ctf_Digit, ctf_Digit, .space_spaces= 5, .name= "number", .findchar_select= 0 }; const struct OcrCellTypeInfo ocr_celltype_text= { - .initial= 012, /* Digit|Upper */ - .nextword= 017, /* Digit|Upper|Lower|Word */ - .midword= 011, /* Digit|Lower */ + .initial= ctf_Digit | ctf_Upper, + .nextword= ctf_Digit | ctf_Upper | ctf_Lower | ctf_Word, + .midword= ctf_Digit | ctf_Lower, .space_spaces= 4, .name= "text", .findchar_select= findchar_select_text @@ -416,7 +429,7 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) { /* something here, so we need to add the spaces */ if (nspaces >= ct->space_spaces) - add_result(rd," ",x-nspaces,x+1,-1); + add_result(rd," ",x-nspaces,x+1,-1,0); nspaces=0; fca.x= x; @@ -427,7 +440,7 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) { if (match) { debugf(" || YES"); - add_result(rd, match->str, x, match_rx, match_ctxi); + add_result(rd, match->str, x, match_rx, match_ctxi, fca.ctxmap); x= match_rx+1; if (match->match) fca.ctxmap= ct->midword; else debugf(" (empty)"); @@ -448,7 +461,7 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) { } } - add_result(rd, 0,-1,-1,-1); + add_result(rd, 0,-1,-1,-1,0); debugf("OCR finished %d glyphs\n",rd->nresults); debug_flush(); return rd->results;