rationalise debugging output

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

diff --git a/pctb/ocr.c b/pctb/ocr.c
index 87bf80842a510280e30826d8fe36c877911b59e8..7eb871a68844acc1d802747e67346aed760b3729 100644 (file)
--- a/pctb/ocr.c
+++ b/pctb/ocr.c
@@ -5,57 +5,121 @@
 
 typedef struct {
   Pixcol col;
 
 typedef struct {
   Pixcol col;

-  struct OCRDatabaseNode *then;
-} OCRDatabaseLink;
+  struct DatabaseNode *then;
+} DatabaseLink;

 
 #define MAXGLYPHCHRS 3
 
 
 #define MAXGLYPHCHRS 3
 

-typedef struct OCRDatabaseNode {
+typedef struct DatabaseNode {

   char s[MAXGLYPHCHRS+1]; /* null-terminated; "" means no match here */
   int nlinks, alinks;
   char s[MAXGLYPHCHRS+1]; /* null-terminated; "" means no match here */
   int nlinks, alinks;

-  OCRDatabaseLink *links;
-} OCRDatabaseNode;
+  DatabaseLink *links;
+} DatabaseNode;

 
 

-#define N_OCR_CONTEXTS 2
+static const char *context_names[]= {
+  "Lower",
+  "Upper",
+  "Digit"
+};

 
 

-static OCRDatabaseNode ocr_contexts[N_OCR_CONTEXTS];
-static FILE *db;
-static OcrResultGlyph *results;
-static int aresults, nresults;
+#define NCONTEXTS (sizeof(context_names)/sizeof(context_names[0]))
+
+#define SPACE_SPACES 3
+
+struct OcrReader {
+  int h;
+  DatabaseNode contexts[NCONTEXTS];
+  OcrResultGlyph *results;
+  int aresults, nresults;
+};

 
 static FILE *resolver;
 static pid_t resolver_pid;
 
 static FILE *resolver;
 static pid_t resolver_pid;

+static int resolver_done;
+
+static void fgetsline(FILE *f, char *lbuf, size_t lbufsz) {
+  char *s= fgets(lbuf,lbufsz,f);
+  eassert(s);
+  int l= strlen(lbuf);
+  eassert(l>0);  eassert(lbuf[--l]='\n');
+  lbuf[l]= 0;
+}
+#define FGETSLINE(f,buf) (fgetsline(f,buf,sizeof(buf)))
+
+static void cleardb_node(DatabaseNode *n) {
+  int i;
+  n->s[0]= 0;
+  for (i=0; i<n->nlinks; i++)
+    cleardb_node(n->links[i].then);
+}

 
 

-static void ocr_readdb(void) {
-  int ctx,nchrs;
-  OCRDatabaseNode *current, *additional;
+static void readdb(OcrReader *rd) {
+  int nchrs;
+  DatabaseNode *current, *additional;

   char chrs[MAXGLYPHCHRS+1];
   Pixcol cv;
   char chrs[MAXGLYPHCHRS+1];
   Pixcol cv;

-  int r,i,j;
+  int r,j,ctxi;
+  int h;
+  char lbuf[100];
+  FILE *db;

 
 

-  assert(!db);
-  db= fopen("database","r");  eassert(db);
+  for (ctxi=0; ctxi<NCONTEXTS; ctxi++)
+    cleardb_node(&rd->contexts[ctxi]);
+
+  char *dbfname=0;
+  asprintf(&dbfname,"%s/charset-%d.txt",get_vardir(),rd->h);
+  eassert(dbfname);
+  
+  db= fopen(dbfname,"r");
+  free(dbfname);
+  if (!db) {
+    eassert(errno==ENOENT);
+    return;
+  }
+
+  FGETSLINE(db,lbuf);
+  eassert(!strcmp(lbuf,"# ypp-sc-tools pctb font v1"));
+
+  r= fscanf(db, "%d", &h);
+  eassert(r==1);
+  eassert(h==rd->h);

 
   for (;;) {
 
   for (;;) {

-    r= fscanf(db, "%d %d", &ctx, &nchrs);
-    if (r==EOF) break;
-    eassert(r==2);
-    eassert(ctx>=0 && ctx<N_OCR_CONTEXTS);
-    eassert(nchrs>0 && nchrs<=MAXGLYPHCHRS);
-
-    for (i=0; i<nchrs; i++) {
-      int c;
-      r= fscanf(db, "%x", &c);  eassert(r==1);
-      eassert(c>0 && c<=255);
-      chrs[i]= c;
+    FGETSLINE(db,lbuf);
+    if (!lbuf || 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; }
+    continue;
+
+  found_ctx:
+    for (nchrs=0;;) {
+      int c= fgetc(db);  eassert(c!=EOF);
+      if (c=='\n') { eassert(nchrs); break; }
+      eassert(nchrs<MAXGLYPHCHRS);
+      if (c=='\\') {
+       unsigned cr;
+       c= fgetc(db);  eassert(c=='x');
+       r= fscanf(db, "%2x", &cr);  eassert(r==1);
+       assert(cr>0 && cr<=255);
+       c= cr;
+      }
+      chrs[nchrs++]= c;

     }
     chrs[nchrs]= 0;
 
     }
     chrs[nchrs]= 0;
 

-    int twidth;
-    r= fscanf(db, "%d", &twidth);  eassert(r==1);
-    current= &ocr_contexts[ctx];
-    for (i=0; i<twidth; i++) {
-      r= fscanf(db, "%"PSPIXCOL(SCNx), &cv);  eassert(r==1);
+    current= &rd->contexts[ctxi];
+    for (;;) {
+      FGETSLINE(db,lbuf);
+      if (!lbuf[0]) { eassert(current != &rd->contexts[ctxi]); break; }
+      char *ep;
+      cv= strtoul(lbuf,&ep,16);  eassert(!*ep);
+      eassert(!(cv & ~((1UL << rd->h)-1)));
+      

       for (j=0; j<current->nlinks; j++)
        if (current->links[j].col == cv) {
          current= current->links[j].then;
       for (j=0; j<current->nlinks; j++)
        if (current->links[j].col == cv) {
          current= current->links[j].then;


@@ -70,7 +134,7 @@ static void ocr_readdb(void) {
        current->alinks++;
        current->alinks<<=1;
        current->links= realloc(current->links,
        current->alinks++;
        current->alinks<<=1;
        current->links= realloc(current->links,

-            sizeof(*current->links) * current->alinks);
+                               sizeof(*current->links) * current->alinks);

        eassert(current->links);
       }
       current->links[current->nlinks].col= cv;
        eassert(current->links);
       }
       current->links[current->nlinks].col= cv;


@@ -85,32 +149,57 @@ static void ocr_readdb(void) {
     strcpy(current->s, chrs);
   }
   eassert(!ferror(db));
     strcpy(current->s, chrs);
   }
   eassert(!ferror(db));

-  eassert(feof(db));
-}      
+  eassert(!fclose(db));
+}
+
+static void cu_pr_ctxmap(unsigned ctxmap) {
+  fprintf(resolver,"{");
+  const char *spc="";
+  int ctxi;
+  for (ctxi=0; ctxi<NCONTEXTS; ctxi++) {
+    if (!(ctxmap & (1u << ctxi))) continue;
+    fprintf(resolver,"%s%s",spc,context_names[ctxi]);
+    spc=" ";
+  }
+  fprintf(resolver,"}");
+}

 
 

-static void callout_unknown(int w, int h, Pixcol cols[], int unk_l, int unk_r,
-                           const OcrResultGlyph *sofar, int nsofar) {
-  int pfd[2], c, r,i, x,y;
+static void callout_unknown(OcrReader *rd, int w, Pixcol cols[],
+                           int unk_l, int unk_r, unsigned unk_ctxmap) {
+  int jobpipe[2],donepipe[2], c, r,i, x,y;

   const OcrResultGlyph *s;
   const char *p;
   const OcrResultGlyph *s;
   const char *p;

+  char cb;

   Pixcol pv;
   
   if (!resolver) {
   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) {
     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");
     }
       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);
-  for (i=0, s=sofar; i<nsofar; i++, s++) {
-    fprintf(resolver," %d %d %d ",s->l,s->r,s->ctx);
+  fprintf(resolver,"%d %d ",unk_l,unk_r);
+  cu_pr_ctxmap(unk_ctxmap);
+  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(s->ctxmap);
+    fprintf(resolver," ");

     for (p=s->s; (c= *p); p++) {
       if (c=='\\') fprintf(resolver,"\\%c",c);
       else if (c>=33 && c<=126) fputc(c,resolver);
     for (p=s->s; (c= *p); p++) {
       if (c=='\\') fprintf(resolver,"\\%c",c);
       else if (c>=33 && c<=126) fputc(c,resolver);


@@ -126,8 +215,8 @@ static void callout_unknown(int w, int h, Pixcol cols[], int unk_l, int unk_r,
          "\"%d %d 2 1\",\n"
          "\"  c black\",\n"
          "\"o c white\",\n",
          "\"%d %d 2 1\",\n"
          "\"  c black\",\n"
          "\"o c white\",\n",

-         w,h);
-  for (y=0, pv=1; y<h; y++, pv<<=1) {
+         w,rd->h);
+  for (y=0, pv=1; y<rd->h; y++, pv<<=1) {

     fputc('"',resolver);
     for (x=0; x<w; x++)
       fputc(cols[x] & pv ? 'o' : ' ', resolver);
     fputc('"',resolver);
     for (x=0; x<w; x++)
       fputc(cols[x] & pv ? 'o' : ' ', resolver);


@@ -137,61 +226,90 @@ static void callout_unknown(int w, int h, Pixcol cols[], int unk_l, int unk_r,
   eassert(!ferror(resolver));
   eassert(!fflush(resolver));
 
   eassert(!ferror(resolver));
   eassert(!fflush(resolver));
 

+  eassert(resolver);
+

   for (;;) {
   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);
     eassert(pid==resolver_pid);
     if (WIFEXITED(r)) {
       eassert(!WEXITSTATUS(r));
       fclose(resolver);

+      close(resolver_done);

       resolver= 0;
       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");
     }
     } 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();
+
+  readdb(rd);

 }
 
 }
 

-static void add_result(const char *s, int l, int r, int ctx) {
-  if (nresults >= aresults) {
-    aresults++; aresults<<=1;
-    results= realloc(results,sizeof(*results)*aresults);
-    eassert(results);
+static void add_result(OcrReader *rd, const char *s, int l, int r,
+                      unsigned ctxmap) {
+  if (rd->nresults >= rd->aresults) {
+    rd->aresults++; rd->aresults<<=1;
+    rd->results= realloc(rd->results,sizeof(*rd->results)*rd->aresults);
+    eassert(rd->results);

   }
   }

-  results[nresults].s= s;
-  results[nresults].l= l;
-  results[nresults].r= r;
-  results[nresults].ctx= ctx;
-  nresults++;
+  rd->results[rd->nresults].s= s;
+  rd->results[rd->nresults].l= l;
+  rd->results[rd->nresults].r= r;
+  rd->results[rd->nresults].ctxmap= ctxmap;
+  rd->nresults++;

 }
 
 }
 

-OcrResultGlyph *ocr(int w, int h, Pixcol cols[]) {
-  int nspaces=0;
-  int ctx=1,i, x;
+struct OcrCellTypeInfo {
+  unsigned initial, nextword, midword;
+};
+const struct OcrCellTypeInfo ocr_celltype_number= {
+  4,4,4
+};
+const struct OcrCellTypeInfo ocr_celltype_text= {
+  .initial=2 /* Uppercase */,
+  .nextword=3 /* Either */,
+  .midword=1 /* Lower only */
+};

 
 

-  nresults=0;
-  assert(db);
+static void vdebugf(const char *fmt, va_list al) {
+#ifdef DEBUG_OCR
+  vfprintf(debug,fmt,al);
+#endif
+}
+static void debugf(const char *fmt, ...) {
+  va_list al;  va_start(al,fmt);  vdebugf(fmt,al);  va_end(al);
+}

 
 

-  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();
+OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
+  int nspaces;
+  unsigned ctxmap;
+  int ctxi, i, x;

 
  restart:
 
  restart:

+
+  nspaces=- w;
+  ctxmap= ct->initial;
+  rd->nresults=0;
+  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;
   for (;;) {
     debug_flush();
   x=0;
   for (;;) {
     debug_flush();


@@ -202,62 +320,99 @@ OcrResultGlyph *ocr(int w, int h, Pixcol cols[]) {
     if (!cols[x]) {
       nspaces++;
       x++;
     if (!cols[x]) {
       nspaces++;
       x++;

-      if (nspaces>3) ctx=1;
+      if (nspaces==SPACE_SPACES) {
+       debugf("OCR  x=%x nspaces=%d space\n",x,nspaces);
+       ctxmap= ct->nextword;
+      }

       continue;
     }
 
       continue;
     }
 

+    /* something here, so we need to add the spaces */
+    if (nspaces>=SPACE_SPACES)
+      add_result(rd," ",x-nspaces,x+1,0);
+    nspaces=0;
+

     /* find character */
     /* find character */

-    OCRDatabaseNode *current=0, *bestmatch=0;

     int lx=x;
     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;
+    DatabaseNode *uniquematch= 0;
+    int uniquematch_rx=-1;
+    
+    debugf("OCR  lx=%d ctxmap=%x  ",lx,ctxmap);
+
+    for (ctxi=0; ctxi<NCONTEXTS; ctxi++) {
+      DatabaseNode *current= &rd->contexts[ctxi];;
+      DatabaseNode *bestmatch= 0;
+      int bestmatch_rx=-1;

 
 

-    found:
-      current= current->links[i].then;
-      if (current->s[0]) {
-       fprintf(debug," \"%s\"",current->s);
-       bestmatch=current; bestmatch_rx=x;
-      } else {
-       fprintf(debug," ...");
+      x= lx;
+      if (!(ctxmap & (1u << ctxi))) continue;
+      debugf(" || %s",context_names[ctxi]);
+
+      for (;;) {
+       debug_flush();
+       debugf(" | x=%d",x);
+       if (x>w) break;
+       Pixcol cv= cols[x];
+       debugf(" cv=%"PSPIXCOL(PRIx),cv);
+       for (i=0; i<current->nlinks; i++)
+         if (current->links[i].col == cv)
+           goto found;
+       /* not found */
+       debugf(" ?");
+       break;
+
+      found:
+       current= current->links[i].then;
+       if (current->s[0]) {
+         debugf(" \"%s\"",current->s);
+         bestmatch= current;
+         bestmatch_rx= x;
+       } else {
+         debugf(" ...");
+       }
+
+       x++;
+      }
+      
+      if (bestmatch) {
+       if (uniquematch && strcmp(bestmatch->s, uniquematch->s)) {
+         debugf( " ambiguous");
+         uniquematch= 0;
+         break;
+       }
+       uniquematch= bestmatch;
+       uniquematch_rx= bestmatch_rx;

       }
       }

-      x++;

     }
 
     }
 

-    if (bestmatch) {
-      fprintf(debug," YES\n");
-      add_result(bestmatch->s, lx, bestmatch_rx, ctx);
-      x= bestmatch_rx+1;
-      ctx= 0;
+    if (uniquematch) {
+      debugf(" || YES\n");
+      add_result(rd, uniquematch->s, lx, uniquematch_rx, ctxmap);
+      x= uniquematch_rx+1;
+      ctxmap= ct->midword;

     } else {
       int rx;
     } else {
       int rx;

-      fprintf(debug," UNKNOWN");
-      for (rx=lx+1; rx<w && cols[rx]; rx++);
-      fprintf(debug," x=%d ctx=%d %d..%d\n",x, ctx, lx,rx);
+      debugf(" || UNKNOWN");
+      for (rx=lx; rx<w && cols[rx]; rx++);
+      debugf(" x=%d ctxmap=%x %d..%d\n",x, ctxmap, lx,rx);

       debug_flush();
       debug_flush();

-      callout_unknown(w,h,cols, lx,rx, results,nresults);
+      callout_unknown(rd, w,cols, lx,rx-1, ctxmap);

       goto restart;
     }
   }
       goto restart;
     }
   }

-  add_result(0,-1,-1,0);
-  fprintf(debug,"OCR  finished %d glyphs\n",nresults);
+  add_result(rd, 0,-1,-1,0);
+  debugf("OCR  finished %d glyphs\n",rd->nresults);

   debug_flush();
   debug_flush();

-  return results;
+  return rd->results;

 }
 
 }
 

-void ocr_init(void) {
-  ocr_readdb();
+OcrReader *ocr_init(int h) {
+  OcrReader *rd;
+
+  rd= malloc(sizeof(*rd));  eassert(rd);
+  memset(rd,0,sizeof(*rd));
+  rd->h= h;
+  readdb(rd);
+  return rd;

 }
 }