X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?a=blobdiff_plain;f=pctb%2Focr.c;h=27271c766495406dedce819b8dfc2feea431cacd;hb=bc6ff7bc24f1839450325f2dbb46ef287ea2fb64;hp=f6104452dc1c1854e6ccd19aac77882caae0aba9;hpb=8d6cf0f224b5df9866eba9350343067edcee78dd;p=ypp-sc-tools.db-test.git

diff --git a/pctb/ocr.c b/pctb/ocr.c
index f610445..27271c7 100644
--- a/pctb/ocr.c
+++ b/pctb/ocr.c
@@ -53,13 +53,22 @@ typedef struct {
   int rx;
 } FindCharResults;
 
-enum { ct_Lower, ct_Upper, ct_Word, ct_Digit };
-static const char *context_names[]= {
-  "Lower",  /*  bit 0, value 001 */
-  "Upper",  /*  bit 1, value 002 */
-  "Word",   /*  bit 2, value 004 */
-  "Digit",  /*  bit 3, value 010 */
+#define FOR_EACH_CONTEXT(EACH)			\
+  EACH(Word)					\
+  EACH(Upper)					\
+  EACH(Lower)					\
+  EACH(Digit)
+
+#define FEC_ENUM(Context) ct_##Context,
+#define FEC_BIT(Context) ctf_##Context = 1 << ct_##Context,
+enum {
+  FOR_EACH_CONTEXT(FEC_ENUM)
+  FOR_EACH_CONTEXT(FEC_BIT)
 };
+
+#define FEC_STRINGS(Context) #Context,
+static const char *context_names[]= { FOR_EACH_CONTEXT(FEC_STRINGS) };
+
 struct OcrCellTypeInfo {
   /* bitmaps of indices into context_names: */
   unsigned initial, nextword, midword;
@@ -223,7 +232,7 @@ static void callout_unknown(OcrReader *rd, int w, Pixcol cols[],
   Pixcol pv;
 
   FILE *resolver= resolve_start();
-  if (!resolver)
+  if (!resolver || !(o_flags & ff_editcharset))
     fatal("OCR failed - unrecognised characters or ligatures.\n"
 	  "Character set database needs to be updated or augmented.\n"
 	  "See README.charset.\n");
@@ -235,7 +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,s->ctxmap);
+    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);
@@ -266,7 +277,7 @@ static void callout_unknown(OcrReader *rd, int w, Pixcol cols[],
 }
 
 static void add_result(OcrReader *rd, const char *s, int l, int r,
-		       unsigned ctxmap) {
+		       int match, unsigned ctxmap) {
   if (rd->nresults >= rd->aresults) {
     rd->aresults++; rd->aresults<<=1;
     rd->results= mrealloc(rd->results, sizeof(*rd->results)*rd->aresults);
@@ -274,6 +285,7 @@ static void add_result(OcrReader *rd, const char *s, int l, int r,
   rd->results[rd->nresults].s= s;
   rd->results[rd->nresults].l= l;
   rd->results[rd->nresults].r= r;
+  rd->results[rd->nresults].match= match;
   rd->results[rd->nresults].ctxmap= ctxmap;
   rd->nresults++;
 }
@@ -314,7 +326,8 @@ static DatabaseNode *findchar_1ctx(const FindCharArgs *fca,
   return bestmatch;
 }  
 
-static DatabaseNode *findchar(const FindCharArgs *fca, int *match_rx) {
+static DatabaseNode *findchar(const FindCharArgs *fca,
+			      int *match_rx, int *match_rctxi) {
   FindCharResults results[NCONTEXTS];
   int ctxi, match=-1, nmatches=0;
 
@@ -342,8 +355,9 @@ static DatabaseNode *findchar(const FindCharArgs *fca, int *match_rx) {
   if (match<0)
     return 0;
   
-  *match_rx= results[ctxi].rx;
-  return results[ctxi].match;
+  *match_rx= results[match].rx;
+  if (match_rctxi) *match_rctxi= match;
+  return results[match].match;
 }
 
 static int findchar_select_text(const FindCharArgs *fca,
@@ -358,15 +372,15 @@ static int findchar_select_text(const FindCharArgs *fca,
 }
 
 const struct OcrCellTypeInfo ocr_celltype_number= {
-  010,010,010,
+  ctf_Digit, ctf_Digit, ctf_Digit,
   .space_spaces= 5,
   .name= "number",
   .findchar_select= 0
 };
 const struct OcrCellTypeInfo ocr_celltype_text= {
-  .initial=  012, /* Digit|Upper */
-  .nextword= 017, /* Digit|Upper|Lower|Word */
-  .midword=  014, /* Digit|Lower */
+  .initial=  ctf_Digit | ctf_Upper,
+  .nextword= ctf_Digit | ctf_Upper | ctf_Lower | ctf_Word,
+  .midword=  ctf_Digit | ctf_Lower,
   .space_spaces= 4,
   .name= "text",
   .findchar_select= findchar_select_text
@@ -415,17 +429,18 @@ 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+1,0);
+      add_result(rd," ",x-nspaces,x+1,-1,0);
     nspaces=0;
 
     fca.x= x;
 
     int match_rx=-1;
-    DatabaseNode *match= findchar(&fca, &match_rx);
+    int match_ctxi=-1;
+    DatabaseNode *match= findchar(&fca, &match_rx, &match_ctxi);
     
     if (match) {
       debugf(" || YES");
-      add_result(rd, match->str, x, match_rx, fca.ctxmap);
+      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)");
@@ -446,7 +461,7 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
     }
 
   }
-  add_result(rd, 0,-1,-1,0);
+  add_result(rd, 0,-1,-1,-1,0);
   debugf("OCR  finished %d glyphs\n",rd->nresults);
   debug_flush();
   return rd->results;