X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?p=ypp-sc-tools.db-test.git;a=blobdiff_plain;f=pctb%2Focr.c;h=df4aacfbb05e299e09642f66115abfbe0bd8b584;hp=a90ecea64e590641911a3faa5e250373ce16863c;hb=1ae1abc15aae822e996b23e2dc73612b4d401044;hpb=ad09cd7cce6584c63c275d7ed1106e66959b3f9d

diff --git a/pctb/ocr.c b/pctb/ocr.c
index a90ecea..df4aacf 100644
--- a/pctb/ocr.c
+++ b/pctb/ocr.c
@@ -25,6 +25,7 @@ static int aresults, nresults;
 
 static FILE *resolver;
 static pid_t resolver_pid;
+static int resolver_done;
 
 static void ocr_readdb(void) {
   int ctx,nchrs;
@@ -88,27 +89,37 @@ static void ocr_readdb(void) {
   eassert(feof(db));
 }      
 
-static void callout_unknown(int w, int h, Pixcol cols[], int unk_l, int unk_r,
+static void callout_unknown(int w, int h, Pixcol cols[],
+			    int unk_l, int unk_r, int unk_ctx,
 			    const OcrResultGlyph *sofar, int nsofar) {
-  int pfd[2], c, r,i, x,y;
+  int jobpipe[2],donepipe[2], c, r,i, x,y;
   const OcrResultGlyph *s;
   const char *p;
+  char cb;
   Pixcol pv;
   
   if (!resolver) {
-    r= pipe(pfd);  eassert(!r);
+    r= pipe(jobpipe);  eassert(!r);
+    r= pipe(donepipe);  eassert(!r);
     resolver_pid= fork();
     eassert(resolver_pid!=-1);
     if (!resolver_pid) {
-      r= dup2(pfd[0],0); eassert(!r);
-      r= close(pfd[1]); eassert(!r);
-      execlp("./show-thing.tcl", "./show-thing.tcl",(char*)0);
+      r= dup2(jobpipe[0],0); eassert(r==0);
+      r= close(jobpipe[1]); eassert(!r);
+      r= close(donepipe[0]); eassert(!r);
+      /* 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: */
+      r= dup2(donepipe[1],4); eassert(r==4);
+      execlp("./show-thing.tcl", "./show-thing.tcl",
+	     "--automatic","1",(char*)0);
       eassert(!"execlp failed");
     }
-    r= close(pfd[0]); eassert(!r);
-    resolver= fdopen(pfd[1],"w"); eassert(resolver);
+    r= close(jobpipe[0]); eassert(!r);
+    r= close(donepipe[1]); eassert(!r);
+    resolver= fdopen(jobpipe[1],"w"); eassert(resolver);
+    resolver_done= donepipe[0];
   }
-  fprintf(resolver,"%d %d",unk_l,unk_r);
+  fprintf(resolver,"%d %d %d",unk_l,unk_r,unk_ctx);
   for (i=0, s=sofar; i<nsofar; i++, s++) {
     fprintf(resolver," %d %d %d ",s->l,s->r,s->ctx);
     for (p=s->s; (c= *p); p++) {
@@ -137,30 +148,37 @@ static void callout_unknown(int w, int h, Pixcol cols[], int unk_l, int unk_r,
   eassert(!ferror(resolver));
   eassert(!fflush(resolver));
 
+  eassert(resolver);
+
   for (;;) {
-    eassert(resolver);
-    pid_t pid= waitpid(resolver_pid, &r, WUNTRACED);
-    if (pid==-1) { eassert(errno==EINTR); continue; }
+    r= read(resolver_done,&cb,1);
+    if (r==-1) { eassert(errno==EINTR); continue; }
+    break;
+  }
+
+  if (r==0) {
+    pid_t pid;
+    for (;;) {
+      pid= waitpid(resolver_pid, &r, 0);
+      if (pid==-1) { eassert(errno==EINTR); continue; }
+      break;
+    }
     eassert(pid==resolver_pid);
     if (WIFEXITED(r)) {
       eassert(!WEXITSTATUS(r));
       fclose(resolver);
+      close(resolver_done);
       resolver= 0;
-    } else if (WIFSTOPPED(r)) {
-      r= kill(resolver_pid,SIGCONT);
-      eassert(!r);
     } else if (WIFSIGNALED(r)) {
       eassert(!"resolver child died due to signal");
     } else {
       eassert(!"weird wait status");
     }
-    struct stat stab, fstab;
-    r= stat("database",&stab);  eassert(!r);
-    r= fstat(fileno(db),&fstab);  eassert(!r);
-    if (stab.st_ino != fstab.st_ino ||
-	stab.st_dev != fstab.st_dev)
-      break;
+  } else {
+    eassert(r==1);
+    eassert(cb==0);
   }
+
   fclose(db);
   db= 0;
   ocr_readdb();
@@ -186,11 +204,18 @@ OcrResultGlyph *ocr(int w, int h, Pixcol cols[]) {
   nresults=0;
   assert(db);
 
+  fprintf(debug,"OCR h=%d w=%d",w,h);
+  for (x=0; x<w; x++) fprintf(debug," %"PSPIXCOL(PRIx),cols[x]);
+  fprintf(debug,"\n");
+  debug_flush();
+
  restart:
   x=0;
   for (;;) {
+    debug_flush();
     /* skip spaces */
-    if (x>=w) break;
+    if (x>=w)
+      break;
 
     if (!cols[x]) {
       nspaces++;
@@ -204,34 +229,50 @@ OcrResultGlyph *ocr(int w, int h, Pixcol cols[]) {
     int lx=x;
     int bestmatch_rx=-1;
     current= &ocr_contexts[ctx];
+    fprintf(debug,"OCR  lx=%d ctx=%d  ",lx,ctx);
+
     for (;;) {
+      debug_flush();
+      fprintf(debug,"| x=%d",x);
       if (x>w) break;
       Pixcol cv= cols[x];
+      fprintf(debug," cv=%"PSPIXCOL(PRIx),x);
       for (i=0; i<current->nlinks; i++)
 	if (current->links[i].col == cv)
 	  goto found;
       /* not found */
+      fprintf(debug," ?");
       break;
+
     found:
       current= current->links[i].then;
-      if (current->s[0]) { bestmatch=current; bestmatch_rx=x; }
+      if (current->s[0]) {
+	fprintf(debug," \"%s\"",current->s);
+	bestmatch=current; bestmatch_rx=x;
+      } else {
+	fprintf(debug," ...");
+      }
       x++;
     }
 
-    if (!bestmatch) {
+    if (bestmatch) {
+      fprintf(debug," YES\n");
       add_result(bestmatch->s, lx, bestmatch_rx, ctx);
       x= bestmatch_rx+1;
       ctx= 0;
     } else {
-      
       int rx;
-      for (rx=lx+1; rx<w && cols[rx]; rx++);
-      printf("UNKNOWN x=%d ctx=%d %d..%d\n",x, ctx, lx,rx);
-      callout_unknown(w,h,cols, lx,rx, results,nresults);
+      fprintf(debug," UNKNOWN");
+      for (rx=lx; rx<w && cols[rx]; rx++);
+      fprintf(debug," x=%d ctx=%d %d..%d\n",x, ctx, lx,rx);
+      debug_flush();
+      callout_unknown(w,h,cols, lx,rx-1,ctx, results,nresults);
       goto restart;
     }
   }
   add_result(0,-1,-1,0);
+  fprintf(debug,"OCR  finished %d glyphs\n",nresults);
+  debug_flush();
   return results;
 }