X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?a=blobdiff_plain;f=pctb%2Focr.c;h=712f90df77c21e239b8f47c256564aae512acaf4;hb=4d7eb93b7c82da31f928349ac7c6cd551d78ca57;hp=bb9576769e10a1d633ed6ebb7786d75e3bfcf585;hpb=8b3b006869bfdc6c2ddbf58d6709a73433abe2f9;p=ypp-sc-tools.web-live.git

diff --git a/pctb/ocr.c b/pctb/ocr.c
index bb95767..712f90d 100644
--- a/pctb/ocr.c
+++ b/pctb/ocr.c
@@ -8,11 +8,12 @@ typedef struct {
   struct DatabaseNode *then;
 } DatabaseLink;
 
-#define MAXGLYPHCHRS 3
+#define MAXGLYPHCHRS 7
 
 typedef struct DatabaseNode {
   char s[MAXGLYPHCHRS+1]; /* null-terminated; "" means no match here */
   int nlinks, alinks;
+  unsigned endsword:1;
   DatabaseLink *links;
 } DatabaseNode;
 
@@ -24,7 +25,7 @@ static const char *context_names[]= {
 
 #define NCONTEXTS (sizeof(context_names)/sizeof(context_names[0]))
 
-#define SPACE_SPACES 3
+#define SPACE_SPACES 4
 
 struct OcrReader {
   int h;
@@ -37,6 +38,8 @@ static FILE *resolver;
 static pid_t resolver_pid;
 static int resolver_done;
 
+DEBUG_DEFINE_DEBUGF(ocr)
+
 static void fgetsline(FILE *f, char *lbuf, size_t lbufsz) {
   char *s= fgets(lbuf,lbufsz,f);
   eassert(s);
@@ -59,7 +62,7 @@ static void readdb(OcrReader *rd) {
   char chrs[MAXGLYPHCHRS+1];
   Pixcol cv;
   int r,j,ctxi;
-  int h;
+  int h, endsword;
   char lbuf[100];
   FILE *db;
 
@@ -110,6 +113,11 @@ static void readdb(OcrReader *rd) {
       }
       chrs[nchrs++]= c;
     }
+    endsword= 0;
+    if (nchrs>1 && chrs[nchrs-1]==' ') {
+      endsword= 1;
+      nchrs--;
+    }
     chrs[nchrs]= 0;
 
     current= &rd->contexts[ctxi];
@@ -147,6 +155,7 @@ static void readdb(OcrReader *rd) {
 
     eassert(!current->s[0]);
     strcpy(current->s, chrs);
+    current->endsword= endsword;
   }
   eassert(!ferror(db));
   eassert(!fclose(db));
@@ -185,7 +194,9 @@ static void callout_unknown(OcrReader *rd, int w, Pixcol cols[],
        * so we aren't in any danger of overwriting some other fd 4: */
       r= dup2(donepipe[1],4); eassert(r==4);
       execlp("./show-thing.tcl", "./show-thing.tcl",
-	     "--automatic","1",(char*)0);
+	     DEBUGP(callout) ? "--debug" : "--noop-arg",
+	     "--automatic-1",
+	     (char*)0);
       eassert(!"execlp failed");
     }
     r= close(jobpipe[0]); eassert(!r);
@@ -296,9 +307,9 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
   nspaces=- w;
   ctxmap= ct->initial;
   rd->nresults=0;
-  fprintf(debug,"OCR h=%d w=%d",rd->h,w);
-  for (x=0; x<w; x++) fprintf(debug," %"PSPIXCOL(PRIx),cols[x]);
-  fprintf(debug,"\n");
+  debugf("OCR h=%d w=%d",rd->h,w);
+  for (x=0; x<w; x++) debugf(" %"PSPIXCOL(PRIx),cols[x]);
+  debugf("\n");
   debug_flush();
 
   x=0;
@@ -312,7 +323,7 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
       nspaces++;
       x++;
       if (nspaces==SPACE_SPACES) {
-	fprintf(debug,"OCR  x=%x nspaces=%d space\n",x,nspaces);
+	debugf("OCR  x=%x nspaces=%d space\n",x,nspaces);
 	ctxmap= ct->nextword;
       }
       continue;
@@ -329,7 +340,7 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
     DatabaseNode *uniquematch= 0;
     int uniquematch_rx=-1;
     
-    fprintf(debug,"OCR  lx=%d ctxmap=%x  ",lx,ctxmap);
+    debugf("OCR  lx=%d ctxmap=%x  ",lx,ctxmap);
 
     for (ctxi=0; ctxi<NCONTEXTS; ctxi++) {
       DatabaseNode *current= &rd->contexts[ctxi];;
@@ -338,29 +349,29 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
 
       x= lx;
       if (!(ctxmap & (1u << ctxi))) continue;
-      fprintf(debug," || %s",context_names[ctxi]);
+      debugf(" || %s",context_names[ctxi]);
 
       for (;;) {
 	debug_flush();
-	fprintf(debug," | x=%d",x);
+	debugf(" | x=%d",x);
 	if (x>w) break;
 	Pixcol cv= cols[x];
-	fprintf(debug," cv=%"PSPIXCOL(PRIx),cv);
+	debugf(" cv=%"PSPIXCOL(PRIx),cv);
 	for (i=0; i<current->nlinks; i++)
 	  if (current->links[i].col == cv)
 	    goto found;
 	/* not found */
-	fprintf(debug," ?");
+	debugf(" ?");
 	break;
 
       found:
 	current= current->links[i].then;
 	if (current->s[0]) {
-	  fprintf(debug," \"%s\"",current->s);
+	  debugf(" \"%s\"%s",current->s,current->endsword?"_":"");
 	  bestmatch= current;
 	  bestmatch_rx= x;
 	} else {
-	  fprintf(debug," ...");
+	  debugf(" ...");
 	}
 
 	x++;
@@ -368,7 +379,7 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
       
       if (bestmatch) {
 	if (uniquematch && strcmp(bestmatch->s, uniquematch->s)) {
-	  fprintf(debug, " ambiguous");
+	  debugf( " ambiguous");
 	  uniquematch= 0;
 	  break;
 	}
@@ -378,22 +389,29 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
     }
 
     if (uniquematch) {
-      fprintf(debug," || YES\n");
+      debugf(" || YES");
       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;
-      fprintf(debug," || UNKNOWN");
+      debugf(" || UNKNOWN");
       for (rx=lx; rx<w && cols[rx]; rx++);
-      fprintf(debug," x=%d ctxmap=%x %d..%d\n",x, ctxmap, lx,rx);
+      debugf(" x=%d ctxmap=%x %d..%d\n",x, ctxmap, lx,rx);
       debug_flush();
       callout_unknown(rd, w,cols, lx,rx-1, ctxmap);
       goto restart;
     }
   }
   add_result(rd, 0,-1,-1,0);
-  fprintf(debug,"OCR  finished %d glyphs\n",rd->nresults);
+  debugf("OCR  finished %d glyphs\n",rd->nresults);
   debug_flush();
   return rd->results;
 }