made its first tsv!

[ypp-sc-tools.main.git] / pctb / ocr.c

diff --git a/pctb/ocr.c b/pctb/ocr.c
index 266ede2a32fa197299aaff620e974e6a3cc4c834..712f90df77c21e239b8f47c256564aae512acaf4 100644 (file)
--- a/pctb/ocr.c
+++ b/pctb/ocr.c
@@ -13,6 +13,7 @@ typedef struct {
 typedef struct DatabaseNode {
   char s[MAXGLYPHCHRS+1]; /* null-terminated; "" means no match here */
   int nlinks, alinks;
 typedef struct DatabaseNode {
   char s[MAXGLYPHCHRS+1]; /* null-terminated; "" means no match here */
   int nlinks, alinks;

+  unsigned endsword:1;

   DatabaseLink *links;
 } DatabaseNode;
 
   DatabaseLink *links;
 } DatabaseNode;
 


@@ -61,7 +62,7 @@ static void readdb(OcrReader *rd) {
   char chrs[MAXGLYPHCHRS+1];
   Pixcol cv;
   int r,j,ctxi;
   char chrs[MAXGLYPHCHRS+1];
   Pixcol cv;
   int r,j,ctxi;

-  int h;
+  int h, endsword;

   char lbuf[100];
   FILE *db;
 
   char lbuf[100];
   FILE *db;
 


@@ -112,6 +113,11 @@ static void readdb(OcrReader *rd) {
       }
       chrs[nchrs++]= c;
     }
       }
       chrs[nchrs++]= c;
     }

+    endsword= 0;
+    if (nchrs>1 && chrs[nchrs-1]==' ') {
+      endsword= 1;
+      nchrs--;
+    }

     chrs[nchrs]= 0;
 
     current= &rd->contexts[ctxi];
     chrs[nchrs]= 0;
 
     current= &rd->contexts[ctxi];


@@ -149,6 +155,7 @@ static void readdb(OcrReader *rd) {
 
     eassert(!current->s[0]);
     strcpy(current->s, chrs);
 
     eassert(!current->s[0]);
     strcpy(current->s, chrs);

+    current->endsword= endsword;

   }
   eassert(!ferror(db));
   eassert(!fclose(db));
   }
   eassert(!ferror(db));
   eassert(!fclose(db));


@@ -360,7 +367,7 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
       found:
        current= current->links[i].then;
        if (current->s[0]) {
       found:
        current= current->links[i].then;
        if (current->s[0]) {

-         debugf(" \"%s\"",current->s);
+         debugf(" \"%s\"%s",current->s,current->endsword?"_":"");

          bestmatch= current;
          bestmatch_rx= x;
        } else {
          bestmatch= current;
          bestmatch_rx= x;
        } else {


@@ -382,10 +389,17 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
     }
 
     if (uniquematch) {
     }
 
     if (uniquematch) {

-      debugf(" || YES\n");
+      debugf(" || YES");

       add_result(rd, uniquematch->s, lx, uniquematch_rx, ctxmap);
       x= uniquematch_rx+1;
       add_result(rd, uniquematch->s, lx, uniquematch_rx, ctxmap);
       x= uniquematch_rx+1;

-      ctxmap= ct->midword;
+      if (uniquematch->s[0]) ctxmap= ct->midword;
+      else debugf(" (empty)");
+      if (uniquematch->endsword) {
+       nspaces= SPACE_SPACES;
+       debugf("_");
+       ctxmap= ct->nextword;
+      }
+      debugf("\n");

     } else {
       int rx;
       debugf(" || UNKNOWN");
     } else {
       int rx;
       debugf(" || UNKNOWN");