-struct OcrCellTypeInfo {
- unsigned initial, nextword, midword;
-};
-const struct OcrCellTypeInfo ocr_celltype_number= {
- 4,4,4
-};
-const struct OcrCellTypeInfo ocr_celltype_text= {
- .initial=2 /* Uppercase */,
- .nextword=3 /* Either */,
- .midword=1 /* Lower only */
-};
+
+typedef struct {
+ OcrReader *rd;
+ int w;
+ Pixcol *cols;
+ int x;
+ unsigned ctxmap;
+} FindCharArgs;
+
+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_unique(const FindCharArgs *fca, int *match_rx) {
+ DatabaseNode *uniquematch= 0;
+ int ctxi;
+
+ debugf("OCR lx=%d ct_state=%x ", fca->x, fca->ctxmap);
+ for (ctxi=0; ctxi<NCONTEXTS; ctxi++) {
+ if (!(fca->ctxmap & (1u << ctxi))) continue;
+ debugf(" || %s",context_names[ctxi]);
+
+ DatabaseNode *match=
+ findchar_1ctx(fca, &fca->rd->contexts[ctxi], match_rx);
+ if (!match) continue;
+
+ if (uniquematch && strcmp(match->str, uniquematch->str)) {
+ debugf( " ambiguous");
+ uniquematch= 0;
+ } else {
+ uniquematch= match;
+ }
+ }
+
+ return uniquematch;
+}
+
+const char *ocr_celltype_name(OcrCellType ct) { return ct->name; }