+static DatabaseNode *findchar_1ctx(const FindCharArgs *fca,
+ DatabaseNode *start, int *matchx_r) {
+ DatabaseNode *current= start;
+ DatabaseNode *bestmatch= 0;
+ int i;
+ int x= fca->x;
+
+ for (;;) {
+ debug_flush();
+ debugf(" | x=%d",x);
+ if (x > fca->w) break;
+ Pixcol cv= fca->cols[x];
+ debugf(" cv=%"PSPIXCOL(PRIx),cv);
+ for (i=0; i<current->nlinks; i++)
+ if (current->links[i].col == cv)
+ goto found;
+ /* not found */
+ debugf(" ?");
+ break;
+
+ found:
+ current= current->links[i].then;
+ if (current->match) {
+ debugf(" \"%s\"%s",current->str,current->endsword?"_":"");
+ bestmatch= current;
+ *matchx_r= x;
+ } else {
+ debugf(" ...");
+ }
+
+ x++;
+ }
+ return bestmatch;
+}
+
+static DatabaseNode *findchar(const FindCharArgs *fca,
+ int *match_rx, int *match_rctxi) {
+ FindCharResults results[NCONTEXTS];
+ int ctxi, match=-1, nmatches=0;
+
+ debugf("OCR lx=%d ct_state=%x ", fca->x, fca->ctxmap);
+ for (ctxi=0; ctxi<NCONTEXTS; ctxi++) {
+ results[ctxi].match= 0;
+ if (!(fca->ctxmap & (1u << ctxi))) continue;
+ debugf(" || %s",context_names[ctxi]);
+
+ results[ctxi].match= findchar_1ctx(fca, &fca->rd->contexts[ctxi],
+ &results[ctxi].rx);
+ if (!results[ctxi].match) continue;
+
+ match= ctxi;
+ nmatches++;
+ }
+ if (nmatches==1) {
+ debugf(" unique");
+ } else {
+ debugf(" ambiguous");
+ match= !fca->ct->findchar_select ? -1 :
+ fca->ct->findchar_select(fca,results);
+ debugf(" resolved %s", match<0 ? "<none>" : context_names[match]);
+ }
+ if (match<0)
+ return 0;
+
+ *match_rx= results[match].rx;
+ if (match_rctxi) *match_rctxi= match;
+ return results[match].match;
+}
+
+static int findchar_select_text(const FindCharArgs *fca,
+ const FindCharResults results[]) {
+ if (fca->ctxmap != 017) return -1;
+
+ dbassert(! results[ct_Digit].match );
+ if (results[ct_Word].match) return ct_Word;
+ if (results[ct_Lower].rx > results[ct_Upper].rx) return ct_Lower;
+ if (results[ct_Upper].rx > results[ct_Lower].rx) return ct_Upper;
+ return -1;
+}
+
+const struct OcrCellTypeInfo ocr_celltype_number= {
+ ctf_Digit, ctf_Digit, ctf_Digit,
+ .space_spaces= 5,
+ .name= "number",
+ .findchar_select= 0
+};
+const struct OcrCellTypeInfo ocr_celltype_text= {
+ .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
+};
+
+