wip new ctxmap passing

diff --git a/pctb/TODO b/pctb/TODO
index 6a603c61c61989ea6b94a3042f3bc785a14fffdb..042feb1a5d0553bc69281cea17e52aa772e313f8 100644 (file)
--- a/pctb/TODO
+++ b/pctb/TODO
@@ -1,3 +1,5 @@
+glyphsdone in dictionary-manager must cope with new entry in each entry
+
 add UI option to dictionary-manager to make user specify which dictionary
   to add multi-context entries to
 install/test dictionary upload/approval

diff --git a/pctb/ocr.c b/pctb/ocr.c
index 1a8ac9a771e0104206ab6ff0454f3f813d6c373b..27271c766495406dedce819b8dfc2feea431cacd 100644 (file)
--- a/pctb/ocr.c
+++ b/pctb/ocr.c
@@ -244,7 +244,9 @@ static void callout_unknown(OcrReader *rd, int w, Pixcol cols[],
   for (i=0, s=rd->results; i<rd->nresults; 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, 1u << s->ctxmap);
     fprintf(resolver," ");
     for (p=s->s; (c= *p); p++) {
       if (c=='\\') fprintf(resolver,"\\%c",c);
@@ -274,7 +276,8 @@ 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 +285,8 @@ 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 +429,7 @@ 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 +440,7 @@ 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 +461,7 @@ 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;

diff --git a/pctb/ocr.h b/pctb/ocr.h
index 389a7d27a8224b4925081de24bff63fb568ce5eb..64739068fc51b23ba06ff9395699ffab1055b982 100644 (file)
--- a/pctb/ocr.h
+++ b/pctb/ocr.h
@@ -38,7 +38,8 @@ typedef uint32_t Pixcol;
 typedef struct {
   const char *s; /* valid until next call to ocr() */
   int l,r; /* column numbers */
-  int ctxi; /* match context index */
+  int match; /* match context index */
+  unsigned ctxmap; /* possible match contexts */
 } OcrResultGlyph;