X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?a=blobdiff_plain;f=pctb%2Focr.c;h=fafdba8eec2ac74f54c4ffc05f096f62c0c5d479;hb=8360f12a6a457b73ebb18dbeedbb15e6ed91318b;hp=5fdc4575ec4f64540129d00055e3847478f36a4f;hpb=89dfaeec1540f73ba85dbd25dd5332416f98778e;p=ypp-sc-tools.db-test.git

diff --git a/pctb/ocr.c b/pctb/ocr.c
index 5fdc457..fafdba8 100644
--- a/pctb/ocr.c
+++ b/pctb/ocr.c
@@ -1,5 +1,29 @@
 /*
-  */
+ * Core OCR algorithm (first exact bitmap match)
+ */
+/*
+ *  This is part of ypp-sc-tools, a set of third-party tools for assisting
+ *  players of Yohoho Puzzle Pirates.
+ * 
+ *  Copyright (C) 2009 Ian Jackson <ijackson@chiark.greenend.org.uk>
+ * 
+ *  This program is free software: you can redistribute it and/or modify
+ *  it under the terms of the GNU General Public License as published by
+ *  the Free Software Foundation, either version 3 of the License, or
+ *  (at your option) any later version.
+ * 
+ *  This program is distributed in the hope that it will be useful,
+ *  but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *  GNU General Public License for more details.
+ * 
+ *  You should have received a copy of the GNU General Public License
+ *  along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ * 
+ *  Yohoho and Puzzle Pirates are probably trademarks of Three Rings and
+ *  are used without permission.  This program is not endorsed or
+ *  sponsored by Three Rings.
+ */
 
 #include "ocr.h"
 
@@ -22,10 +46,27 @@ static const char *context_names[]= {
   "Upper",
   "Digit"
 };
+struct OcrCellTypeInfo {
+  /* bitmaps of indices into context_names: */
+  unsigned initial, nextword, midword;
+  int space_spaces;
+  const char *name;
+};
+const struct OcrCellTypeInfo ocr_celltype_number= {
+  4,4,4,
+  .space_spaces= 5,
+  .name= "number"
+};
+const struct OcrCellTypeInfo ocr_celltype_text= {
+  .initial=2, /* Uppercase */
+  .nextword=3, /* Either */
+  .midword=1, /* Lower only */
+  .space_spaces= 4,
+  .name= "text"
+};
 
-#define NCONTEXTS (sizeof(context_names)/sizeof(context_names[0]))
 
-#define SPACE_SPACES 4
+#define NCONTEXTS (sizeof(context_names)/sizeof(context_names[0]))
 
 struct OcrReader {
   int h;
@@ -40,22 +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;
@@ -69,10 +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]);
@@ -81,41 +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;
       }
@@ -130,7 +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);
@@ -164,8 +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) {
@@ -187,6 +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) );
@@ -200,11 +226,11 @@ 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);
-      sysassert(!"execlp failed");
+      sysassert(!"execlp ocr-resolver failed");
     }
     sysassert(! close(jobpipe[0]) );
     sysassert(! close(donepipe[1]) );
@@ -254,27 +280,10 @@ static void callout_unknown(OcrReader *rd, int w, Pixcol cols[],
   }
 
   if (r==0) {
-    pid_t pid;
-    int st;
-    for (;;) {
-      pid= waitpid(resolver_pid, &st, 0);
-      if (pid==-1) { sysassert(errno==EINTR); continue; }
-      break;
-    }
-    sysassert(pid==resolver_pid);
-    if (WIFEXITED(st)) {
-      if (WEXITSTATUS(st))
-	fatal("character resolver failed with nonzero exit status %d",
-	      WEXITSTATUS(st));
-      fclose(resolver);
-      close(resolver_done);
-      resolver= 0;
-    } else if (WIFSIGNALED(st)) {
-      fatal("character resolver died due to signal %s%s",
-	    strsignal(WTERMSIG(st)), WCOREDUMP(st)?" (core dumped)":"");
-    } else {
-      fatal("character resolver gave strange wait status %d",st);
-    }
+    waitpid_check_exitstatus(resolver_pid, "character resolver");
+    fclose(resolver);
+    close(resolver_done);
+    resolver= 0;
   } else {
     assert(r==1);
     sysassert(cb==0);
@@ -296,20 +305,6 @@ static void add_result(OcrReader *rd, const char *s, int l, int r,
   rd->nresults++;
 }
 
-struct OcrCellTypeInfo {
-  unsigned initial, nextword, midword;
-  const char *name;
-};
-const struct OcrCellTypeInfo ocr_celltype_number= {
-  4,4,4,
-  .name= "number"
-};
-const struct OcrCellTypeInfo ocr_celltype_text= {
-  .initial=2, /* Uppercase */
-  .nextword=3, /* Either */
-  .midword=1, /* Lower only */
-  .name= "text"
-};
 
 const char *ocr_celltype_name(OcrCellType ct) { return ct->name; }
 
@@ -338,7 +333,7 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
     if (!cols[x]) {
       nspaces++;
       x++;
-      if (nspaces==SPACE_SPACES) {
+      if (nspaces == ct->space_spaces) {
 	debugf("OCR  x=%x nspaces=%d space\n",x,nspaces);
 	ctxmap= ct->nextword;
       }
@@ -346,7 +341,7 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
     }
 
     /* something here, so we need to add the spaces */
-    if (nspaces>=SPACE_SPACES)
+    if (nspaces >= ct->space_spaces)
       add_result(rd," ",x-nspaces,x+1,0);
     nspaces=0;
 
@@ -411,7 +406,7 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
       if (uniquematch->s[0]) ctxmap= ct->midword;
       else debugf(" (empty)");
       if (uniquematch->endsword) {
-	nspaces= SPACE_SPACES;
+	nspaces= ct->space_spaces;
 	debugf("_");
 	ctxmap= ct->nextword;
       }