X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?p=ypp-sc-tools.db-test.git;a=blobdiff_plain;f=pctb%2Focr.c;h=f174c68117c8d31e29da62fe94f153882eb8d14f;hp=1a8ac9a771e0104206ab6ff0454f3f813d6c373b;hb=4b550c8a3c77210340f2eaac09fa63c51a06e679;hpb=a6b59a7c170cb7e1f3940ea90c22580adae57451 diff --git a/pctb/ocr.c b/pctb/ocr.c index 1a8ac9a..f174c68 100644 --- a/pctb/ocr.c +++ b/pctb/ocr.c @@ -36,7 +36,7 @@ typedef struct { typedef struct DatabaseNode { char *str; int nlinks, alinks; - unsigned match:1, defined:1, endsword:1; + unsigned match:1, defined:1, endsword:1, local:1; DatabaseLink *links; } DatabaseNode; @@ -99,7 +99,7 @@ static void cleardb_node(DatabaseNode *n) { cleardb_node(n->links[i].then); } -static void readdb1(OcrReader *rd, const char *which); +static void readdb1(OcrReader *rd, const char *which, int local); static void readdb(OcrReader *rd) { int ctxi; @@ -107,11 +107,11 @@ static void readdb(OcrReader *rd) { for (ctxi=0; ctxicontexts[ctxi]); - readdb1(rd, "master"); - readdb1(rd, "local"); + readdb1(rd, "master", 0); + readdb1(rd, "local", 1); } -static void readdb1(OcrReader *rd, const char *which) { +static void readdb1(OcrReader *rd, const char *which, int local) { int nchrs; DatabaseNode *current, *additional; char chrs[100]; @@ -197,6 +197,7 @@ static void readdb1(OcrReader *rd, const char *which) { current->str= 0; current->defined= 1; current->match= 0; + current->local= local; if (nchrs) { current->str= mmalloc(nchrs+1); @@ -244,7 +245,9 @@ 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, s->ctxmap); fprintf(resolver," "); for (p=s->s; (c= *p); p++) { if (c=='\\') fprintf(resolver,"\\%c",c); @@ -274,7 +277,8 @@ 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); @@ -282,7 +286,8 @@ 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++; } @@ -425,7 +430,7 @@ 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; @@ -436,7 +441,7 @@ 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)"); @@ -457,7 +462,7 @@ 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; @@ -479,3 +484,119 @@ OcrReader *ocr_init(int h) { readdb(rd); return rd; } + +/*---------- character set dump ----------*/ + +static void show_recurse(const DatabaseNode *t, int *count, + const DatabaseNode **store_ary) { + if (t->defined) { + if (store_ary) store_ary[*count]= t; + (*count)++; + } + int l; + for (l=0; lnlinks; l++) + show_recurse(t->links[l].then, count,store_ary); +} + +static int show_char_compar(const void *av, const void *bv) { + const DatabaseNode *const *ap= av; const DatabaseNode *a= *ap; + const DatabaseNode *const *bp= bv; const DatabaseNode *b= *bp; + return strcmp(a->str, b->str) ?: + ((int)a->match - (int)b->match) ?: + ((int)a->endsword - (int)b->endsword) ?: + ((int)a->local - (int)b->local) ?: + 0; +} + +void ocr_showcharsets(void) { + DIR *d; + struct dirent *de; + char found[32]; + pcre *fnpat; + int matchvec[10]; + char hbuf[10]; + const char *pcre_err; + int pcre_erroffset; + + memset(found,0,sizeof(found)); + + fnpat= pcre_compile("\\#(?:master|local)\\-char([1-9]\\d{0,2})\\#\\.txt$", + PCRE_ANCHORED|PCRE_DOLLAR_ENDONLY, + &pcre_err,&pcre_erroffset, 0); + debugf("pcre_compile %p %s\n",fnpat,pcre_err); + assert(fnpat); + + sysassert( d= opendir(get_vardir()) ); + for (;;) { + errno=0; de= readdir(d); if (!de) break; + + int rer= pcre_exec(fnpat,0, de->d_name,strlen(de->d_name), 0,0, + matchvec,ARRAYSIZE(matchvec)); + debugf("pcre_exec `%s' => %d\n", de->d_name,rer); + + if (rer==PCRE_ERROR_NOMATCH || rer==PCRE_ERROR_BADUTF8) continue; + assert(rer==2); + + rer= pcre_copy_substring(de->d_name,matchvec,rer, 1, hbuf,sizeof(hbuf)); + debugf("pcre_copy_substring => %d\n", rer); + assert(rer>0); + + int h= atoi(hbuf); + if (h >= ARRAYSIZE(found)) continue; + + found[h]= 1; + } + + int h; + for (h=0; hcontexts[ctxi], &nchars, 0); + const DatabaseNode **chars= mmalloc(sizeof(*chars) * nchars); + int chari= 0; + show_recurse(&rd->contexts[ctxi], &chari, chars); + assert(chari==nchars); + qsort(chars, nchars, sizeof(*chars), show_char_compar); + + int local; + for (local=0; local<2; local++) { + printf("%2d %-6s %-6s ", h, context_names[ctxi], + local?"local":"master"); + for (chari=0; chari"; + + if (t->local != local) continue; + + if (!t->match) + printf(" [nomatch]"); + else if (!t->endsword && strspn(t->str, accept) == strlen(t->str)) + printf(" %s",t->str); + else { + printf(" \""); + char *p= t->str; + int c; + while ((c=*p++)) { + if (c=='"' || c=='\\') printf("\\%c",c); + else if (c>=' ' && c<=126) putchar(c); + else printf("\\x%02x", (unsigned char)c); + } + if (t->endsword) putchar(' '); + putchar('"'); + } + } + putchar('\n'); + } + free(chars); + } + } +}