X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?p=ypp-sc-tools.db-test.git;a=blobdiff_plain;f=pctb%2Focr.c;h=ecfac976b2d025176bd4ee82675c31a108c3ba18;hp=2c3a16772ec8ac0a90f5a9be420c9db188c1e6da;hb=3a820e9871e6c4607b34939cc4d55c6925a7de29;hpb=3063e05a93fb97a5eca7f26c38da94fa4000406e;ds=sidebyside

diff --git a/pctb/ocr.c b/pctb/ocr.c
index 2c3a167..ecfac97 100644
--- a/pctb/ocr.c
+++ b/pctb/ocr.c
@@ -277,10 +277,44 @@ static void add_result(OcrReader *rd, const char *s, int l, int r,
 
 const char *ocr_celltype_name(OcrCellType ct) { return ct->name; }
 
+static DatabaseNode *findchar(OcrReader *rd, int w, Pixcol cols[],
+			      int x, int ctxi, int *matchx_r) {
+  DatabaseNode *current= &rd->contexts[ctxi];
+  DatabaseNode *bestmatch= 0;
+  int i;
+
+  for (;;) {
+    debug_flush();
+    debugf(" | x=%d",x);
+    if (x>w) break;
+    Pixcol cv= 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;
+}  
+
 OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
   int nspaces;
   unsigned ctxmap;
-  int ctxi, i, x;
+  int ctxi, x;
 
  restart:
 
@@ -314,63 +348,30 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
       add_result(rd," ",x-nspaces,x+1,0);
     nspaces=0;
 
-    /* find character */
-    int lx=x;
-
     DatabaseNode *uniquematch= 0;
     int uniquematch_rx=-1;
     
-    debugf("OCR  lx=%d ctxmap=%x  ",lx,ctxmap);
+    debugf("OCR  lx=%d ctxmap=%x  ",x,ctxmap);
 
     for (ctxi=0; ctxi<NCONTEXTS; ctxi++) {
-      DatabaseNode *current= &rd->contexts[ctxi];;
-      DatabaseNode *bestmatch= 0;
-      int bestmatch_rx=-1;
-
-      x= lx;
       if (!(ctxmap & (1u << ctxi))) continue;
       debugf(" || %s",context_names[ctxi]);
 
-      for (;;) {
-	debug_flush();
-	debugf(" | x=%d",x);
-	if (x>w) break;
-	Pixcol cv= 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;
-	  bestmatch_rx= x;
-	} else {
-	  debugf(" ...");
-	}
+      DatabaseNode *match=
+	findchar(rd,w,cols, x,ctxi, &uniquematch_rx);
+      if (!match) continue;
 
-	x++;
-      }
-      
-      if (bestmatch) {
-	if (uniquematch && strcmp(bestmatch->str, uniquematch->str)) {
-	  debugf( " ambiguous");
-	  uniquematch= 0;
-	  break;
-	}
-	uniquematch= bestmatch;
-	uniquematch_rx= bestmatch_rx;
+      if (uniquematch && strcmp(match->str, uniquematch->str)) {
+	debugf( " ambiguous");
+	uniquematch= 0;
+      } else {
+	uniquematch= match;
       }
     }
 
     if (uniquematch) {
       debugf(" || YES");
-      add_result(rd, uniquematch->str, lx, uniquematch_rx, ctxmap);
+      add_result(rd, uniquematch->str, x, uniquematch_rx, ctxmap);
       x= uniquematch_rx+1;
       if (uniquematch->match) ctxmap= ct->midword;
       else debugf(" (empty)");
@@ -383,12 +384,13 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
     } else {
       int rx;
       debugf(" || UNKNOWN");
-      for (rx=lx; rx<w && cols[rx]; rx++);
-      debugf(" x=%d ctxmap=%x %d..%d\n",x, ctxmap, lx,rx);
+      for (rx=x; rx<w && cols[rx]; rx++);
+      debugf(" x=%d ctxmap=%x %d..%d\n",x, ctxmap, x,rx);
       debug_flush();
-      callout_unknown(rd, w,cols, lx,rx-1, ctxmap);
+      callout_unknown(rd, w,cols, x,rx-1, ctxmap);
       goto restart;
     }
+
   }
   add_result(rd, 0,-1,-1,0);
   debugf("OCR  finished %d glyphs\n",rd->nresults);