WIP island determination; can recognise a pixmap!

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

diff --git a/pctb/ocr.c b/pctb/ocr.c
index 6a91935769ecfc7da0e70681ea80db627c49327e..fafdba8eec2ac74f54c4ffc05f096f62c0c5d479 100644 (file)
--- a/pctb/ocr.c
+++ b/pctb/ocr.c
@@ -81,22 +81,7 @@ static int resolver_done;
 
 DEBUG_DEFINE_DEBUGF(ocr)
 
-#define dbassert(x)                                                    \
-  ((x) ? (void)0 :                                                     \
-     fatal("Error in character set database.\n"                                \
-          " Requirement not met: %s:%d: %s", __FILE__,__LINE__, #x))
-
-static void fgetsline(FILE *f, char *lbuf, size_t lbufsz) {
-  errno=0;
-  char *s= fgets(lbuf,lbufsz,f);
-  sysassert(!ferror(f));
-  dbassert(!feof(f));
-  assert(s);
-  int l= strlen(lbuf);
-  dbassert(l>0);  dbassert(lbuf[--l]='\n');
-  lbuf[l]= 0;
-}
-#define FGETSLINE(f,buf) (fgetsline(f,buf,sizeof(buf)))
+#define FGETSLINE (dbfile_getsline(lbuf,sizeof(lbuf),__FILE__,__LINE__))
 
 static void cleardb_node(DatabaseNode *n) {
   int i;
@@ -110,10 +95,9 @@ static void readdb(OcrReader *rd) {
   DatabaseNode *current, *additional;
   char chrs[MAXGLYPHCHRS+1];
   Pixcol cv;
-  int r,j,ctxi;
+  int j,ctxi;
   int h, endsword;
   char lbuf[100];
-  FILE *db;
 
   for (ctxi=0; ctxi<NCONTEXTS; ctxi++)
     cleardb_node(&rd->contexts[ctxi]);
@@ -122,41 +106,36 @@ static void readdb(OcrReader *rd) {
   asprintf(&dbfname,"%s/charset-%d.txt",get_vardir(),rd->h);
   sysassert(dbfname);
   
-  db= fopen(dbfname,"r");
-  free(dbfname);
-  if (!db) {
-    sysassert(errno==ENOENT);
-    return;
-  }
+  if (!dbfile_open(dbfname))
+    goto x;
 
-  FGETSLINE(db,lbuf);
+  FGETSLINE;
   dbassert(!strcmp(lbuf,"# ypp-sc-tools pctb font v1"));
 
-  r= fscanf(db, "%d", &h);
-  dbassert(r==1);
+  dbassert( dbfile_scanf("%d", &h) == 1);
   dbassert(h==rd->h);
 
   for (;;) {
-    FGETSLINE(db,lbuf);
-    if (!lbuf || lbuf[0]=='#') continue;
+    FGETSLINE;
+    if (!lbuf[0] || lbuf[0]=='#') continue;
     if (!strcmp(lbuf,".")) break;
 
     for (ctxi=0; ctxi<NCONTEXTS; ctxi++)
       if (!strcmp(lbuf,context_names[ctxi]))
        goto found_ctx;
     /* not found, just skip */
-    for (;;) { FGETSLINE(db,lbuf); if (!lbuf[0]) break; }
+    for (;;) { FGETSLINE; if (!lbuf[0]) break; }
     continue;
 
   found_ctx:
     for (nchrs=0;;) {
-      int c= fgetc(db); sysassert(!ferror(db)); dbassert(c!=EOF);
+      int c= fgetc(dbfile); sysassert(!ferror(dbfile)); dbassert(c!=EOF);
       if (c=='\n') { dbassert(nchrs); break; }
       dbassert(nchrs<MAXGLYPHCHRS);
       if (c=='\\') {
        unsigned cr;
-       c= fgetc(db); sysassert(!ferror(db)); dbassert(c=='x');
-       r= fscanf(db, "%2x", &cr); sysassert(!ferror(db)); dbassert(r==1);
+       c= fgetc(dbfile); sysassert(!ferror(dbfile)); dbassert(c=='x');
+       dbassert( dbfile_scanf("%2x", &cr) == 1);
        assert(cr>0 && cr<=255);
        c= cr;
       }
@@ -171,7 +150,7 @@ static void readdb(OcrReader *rd) {
 
     current= &rd->contexts[ctxi];
     for (;;) {
-      FGETSLINE(db,lbuf);
+      FGETSLINE;
       if (!lbuf[0]) { dbassert(current != &rd->contexts[ctxi]); break; }
       char *ep;
       cv= strtoul(lbuf,&ep,16);  dbassert(!*ep);
@@ -205,8 +184,9 @@ static void readdb(OcrReader *rd) {
     strcpy(current->s, chrs);
     current->endsword= endsword;
   }
-  sysassert(!ferror(db));
-  sysassert(!fclose(db));
+ x:
+  dbfile_close();
+  free(dbfname);
 }
 
 static void cu_pr_ctxmap(unsigned ctxmap) {
@@ -228,6 +208,11 @@ static void callout_unknown(OcrReader *rd, int w, Pixcol cols[],
   const char *p;
   char cb;
   Pixcol pv;
+
+  if (!o_resolver)
+    fatal("OCR failed - unrecognised characters or ligatures.\n"
+         "Character set database needs to be updated or augmented.\n"
+         "See README.charset.\n");
   
   if (!resolver) {
     sysassert(! pipe(jobpipe) );
@@ -241,7 +226,7 @@ static void callout_unknown(OcrReader *rd, int w, Pixcol cols[],
       /* we know donepipe[1] is >= 4 and we have dealt with all the others
        * so we aren't in any danger of overwriting some other fd 4: */
       sysassert( dup2(donepipe[1],4) ==4 );
-      execlp("./yppsc-ocr-resolver", "yppsc-ocr-resolver",
+      execlp(o_resolver, o_resolver,
             DEBUGP(callout) ? "--debug" : "--noop-arg",
             "--automatic-1",
             (char*)0);