X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?a=blobdiff_plain;f=pctb%2Focr.c;h=b7934c983e2f407a362183c0e5aad6756653165c;hb=1d0e4ca6694c19fadd97452eed1451248b156997;hp=ecfac976b2d025176bd4ee82675c31a108c3ba18;hpb=3a820e9871e6c4607b34939cc4d55c6925a7de29;p=ypp-sc-tools.db-test.git

diff --git a/pctb/ocr.c b/pctb/ocr.c
index ecfac97..b7934c9 100644
--- a/pctb/ocr.c
+++ b/pctb/ocr.c
@@ -275,19 +275,26 @@ static void add_result(OcrReader *rd, const char *s, int l, int r,
 }
 
 
-const char *ocr_celltype_name(OcrCellType ct) { return ct->name; }
+typedef struct {
+  OcrReader *rd;
+  int w;
+  Pixcol *cols;
+  int x;
+  unsigned ctxmap;
+} FindCharArgs;
 
-static DatabaseNode *findchar(OcrReader *rd, int w, Pixcol cols[],
-			      int x, int ctxi, int *matchx_r) {
-  DatabaseNode *current= &rd->contexts[ctxi];
+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>w) break;
-    Pixcol cv= cols[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)
@@ -311,15 +318,46 @@ static DatabaseNode *findchar(OcrReader *rd, int w, Pixcol cols[],
   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; }
+
 OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
   int nspaces;
-  unsigned ctxmap;
-  int ctxi, x;
+  int x;
+
+  FindCharArgs fca;
+  fca.rd= rd;
+  fca.w= w;
+  fca.cols= cols;
+  fca.x= -1;
 
  restart:
 
   nspaces=- w;
-  ctxmap= ct->initial;
+  fca.ctxmap= ct->initial;
   rd->nresults=0;
   debugf("OCR h=%d w=%d",rd->h,w);
   for (x=0; x<w; x++) debugf(" %"PSPIXCOL(PRIx),cols[x]);
@@ -338,7 +376,7 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
       x++;
       if (nspaces == ct->space_spaces) {
 	debugf("OCR  x=%x nspaces=%d space\n",x,nspaces);
-	ctxmap= ct->nextword;
+	fca.ctxmap= ct->nextword;
       }
       continue;
     }
@@ -348,46 +386,30 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
       add_result(rd," ",x-nspaces,x+1,0);
     nspaces=0;
 
-    DatabaseNode *uniquematch= 0;
-    int uniquematch_rx=-1;
-    
-    debugf("OCR  lx=%d ctxmap=%x  ",x,ctxmap);
+    fca.x= x;
 
-    for (ctxi=0; ctxi<NCONTEXTS; ctxi++) {
-      if (!(ctxmap & (1u << ctxi))) continue;
-      debugf(" || %s",context_names[ctxi]);
-
-      DatabaseNode *match=
-	findchar(rd,w,cols, x,ctxi, &uniquematch_rx);
-      if (!match) continue;
-
-      if (uniquematch && strcmp(match->str, uniquematch->str)) {
-	debugf( " ambiguous");
-	uniquematch= 0;
-      } else {
-	uniquematch= match;
-      }
-    }
-
-    if (uniquematch) {
+    int match_rx=-1;
+    DatabaseNode *match= findchar_unique(&fca, &match_rx);
+    
+    if (match) {
       debugf(" || YES");
-      add_result(rd, uniquematch->str, x, uniquematch_rx, ctxmap);
-      x= uniquematch_rx+1;
-      if (uniquematch->match) ctxmap= ct->midword;
+      add_result(rd, match->str, x, match_rx, fca.ctxmap);
+      x= match_rx+1;
+      if (match->match) fca.ctxmap= ct->midword;
       else debugf(" (empty)");
-      if (uniquematch->endsword) {
+      if (match->endsword) {
 	nspaces= ct->space_spaces;
 	debugf("_");
-	ctxmap= ct->nextword;
+	fca.ctxmap= ct->nextword;
       }
       debugf("\n");
     } else {
       int rx;
       debugf(" || UNKNOWN");
       for (rx=x; rx<w && cols[rx]; rx++);
-      debugf(" x=%d ctxmap=%x %d..%d\n",x, ctxmap, x,rx);
+      debugf(" x=%d ctxmap=%x %d..%d\n",x, fca.ctxmap, x,rx);
       debug_flush();
-      callout_unknown(rd, w,cols, x,rx-1, ctxmap);
+      callout_unknown(rd, w,cols, x,rx-1, fca.ctxmap);
       goto restart;
     }