X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?p=ypp-sc-tools.db-test.git;a=blobdiff_plain;f=pctb%2Focr.c;h=fd4fcf4f8de99642291fb3e9c5410a2884c642fb;hp=be479f83b8d626442899c77f08ffdab049e26747;hb=8787ee59f6840de63bac432b516a30d0dfe22c84;hpb=cd6a4f773c32f73aff27f97e8994c6b7c1019bb7

diff --git a/pctb/ocr.c b/pctb/ocr.c
index be479f8..fd4fcf4 100644
--- a/pctb/ocr.c
+++ b/pctb/ocr.c
@@ -5,58 +5,115 @@
 
 typedef struct {
   Pixcol col;
-  struct OCRDatabaseNode *then;
-} OCRDatabaseLink;
+  struct DatabaseNode *then;
+} DatabaseLink;
 
 #define MAXGLYPHCHRS 3
 
-typedef struct OCRDatabaseNode {
+typedef struct DatabaseNode {
   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]))
+
+struct OcrReader {
+  int h;
+  DatabaseNode contexts[NCONTEXTS];
+  OcrResultGlyph *results;
+  int aresults, nresults;
+};
 
 static FILE *resolver;
 static pid_t resolver_pid;
 static int resolver_done;
 
-static void ocr_readdb(void) {
-  int ctx,nchrs;
-  OCRDatabaseNode *current, *additional;
+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 readdb(OcrReader *rd) {
+  int nchrs;
+  DatabaseNode *current, *additional;
   char chrs[MAXGLYPHCHRS+1];
   Pixcol cv;
-  int r,i,j;
+  int r,j,ctxi;
+  int h;
+  char lbuf[100];
+  FILE *db;
+
+  char *dbfname=0;
+  asprintf(&dbfname,"%s/charset-%d.txt",get_vardir(),rd->h);
+  eassert(dbfname);
+  
+  db= fopen(dbfname,"r");  eassert(db);
+  free(dbfname);
+
+  FGETSLINE(db,lbuf);
+  eassert(!strcmp(lbuf,"# ypp-sc-tools pctb font v1"));
+
+  r= fscanf(db, "%d", &h);
+  eassert(r==1);
+  eassert(h==rd->h);
 
-  assert(!db);
-  db= fopen("database","r");  eassert(db);
+  for (ctxi=0; ctxi<NCONTEXTS; ctxi++)
+    cleardb_node(&rd->contexts[ctxi]);
 
   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;
 
-    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;
@@ -71,7 +128,7 @@ static void ocr_readdb(void) {
 	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;
@@ -86,12 +143,11 @@ static void ocr_readdb(void) {
     strcpy(current->s, chrs);
   }
   eassert(!ferror(db));
-  eassert(feof(db));
+  eassert(!fclose(db));
 }      
 
-static void callout_unknown(int w, int h, Pixcol cols[],
-			    int unk_l, int unk_r, int unk_ctx,
-			    const OcrResultGlyph *sofar, int nsofar) {
+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;
@@ -119,10 +175,18 @@ static void callout_unknown(int w, int h, Pixcol cols[],
     resolver= fdopen(jobpipe[1],"w"); eassert(resolver);
     resolver_done= donepipe[0];
   }
-  fprintf(resolver,"%d %d %d",unk_l,unk_r,unk_ctx);
-  for (i=0, s=sofar; i<nsofar; i++, s++) {
+  fprintf(resolver,"%d %d {",unk_l,unk_r);
+  const char *spc="";
+  int ctxi;
+  for (ctxi=0; ctxi<NCONTEXTS; ctxi++) {
+    if (!(unk_ctxmap & (1u << ctxi))) continue;
+    fprintf(resolver,"%s%s",spc,context_names[ctxi]);
+    spc=" ";
+  }
+  fprintf(resolver,"}");
+  for (i=0, s=rd->results; i<rd->nresults; i++, s++) {
     if (!strcmp(s->s," ")) continue;
-    fprintf(resolver," %d %d %d ",s->l,s->r,s->ctx);
+    fprintf(resolver," %d %d %s ",s->l,s->r,context_names[s->ctx]);
     for (p=s->s; (c= *p); p++) {
       if (c=='\\') fprintf(resolver,"\\%c",c);
       else if (c>=33 && c<=126) fputc(c,resolver);
@@ -138,8 +202,8 @@ static void callout_unknown(int w, int h, Pixcol cols[],
 	  "\"%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);
@@ -180,32 +244,30 @@ static void callout_unknown(int w, int h, Pixcol cols[],
     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, int ctx) {
+  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].ctx= ctx;
+  rd->nresults++;
 }
 
-OcrResultGlyph *ocr(int w, int h, Pixcol cols[]) {
+OcrResultGlyph *ocr(OcrReader *rd, int w, Pixcol cols[]) {
   int nspaces=-w;
-  int ctx=1,i, x;
+  unsigned ctxmap=2; /* uppercase */
+  int ctxi, i, x;
 
-  nresults=0;
-  assert(db);
+  rd->nresults=0;
 
-  fprintf(debug,"OCR h=%d w=%d",w,h);
+  fprintf(debug,"OCR h=%d w=%d",rd->h,w);
   for (x=0; x<w; x++) fprintf(debug," %"PSPIXCOL(PRIx),cols[x]);
   fprintf(debug,"\n");
   debug_flush();
@@ -223,65 +285,95 @@ OcrResultGlyph *ocr(int w, int h, Pixcol cols[]) {
       x++;
       if (nspaces==3) {
 	fprintf(debug,"OCR  x=%x nspaces=%d space\n",x,nspaces);
-	add_result(" ",x-nspaces,x+1,0);
-	ctx=1;
+	add_result(rd," ",x-nspaces,x+1,0);
+	ctxmap=3; /* either */
       }
       continue;
     }
     nspaces=0;
 
     /* find character */
-    OCRDatabaseNode *current=0, *bestmatch=0;
     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, uniquematch_ctxi=-1;
+    
+    fprintf(debug,"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;
+
+      x= lx;
+      if (!(ctxmap & (1u << ctxi))) continue;
+      fprintf(debug," || %s",context_names[ctxi]);
+
+      for (;;) {
+	debug_flush();
+	fprintf(debug," | x=%d",x);
+	if (x>w) break;
+	Pixcol cv= cols[x];
+	fprintf(debug," cv=%"PSPIXCOL(PRIx),cv);
+	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]) {
+	  fprintf(debug," \"%s\"",current->s);
+	  bestmatch= current;
+	  bestmatch_rx= x;
+	} else {
+	  fprintf(debug," ...");
+	}
 
-    found:
-      current= current->links[i].then;
-      if (current->s[0]) {
-	fprintf(debug," \"%s\"",current->s);
-	bestmatch=current; bestmatch_rx=x;
-      } else {
-	fprintf(debug," ...");
+	x++;
+      }
+      
+      if (bestmatch) {
+	if (uniquematch) {
+	  fprintf(debug, " ambiguous");
+	  uniquematch= 0;
+	  break;
+	}
+	uniquematch= bestmatch;
+	uniquematch_rx= bestmatch_rx;
+	uniquematch_ctxi= ctxi;
       }
-      x++;
     }
 
-    if (bestmatch) {
-      fprintf(debug," YES\n");
-      add_result(bestmatch->s, lx, bestmatch_rx, ctx);
-      x= bestmatch_rx+1;
-      ctx= 0;
+    if (uniquematch) {
+      fprintf(debug," || YES\n");
+      add_result(rd, uniquematch->s, lx, uniquematch_rx, uniquematch_ctxi);
+      x= uniquematch_rx+1;
+      ctxmap= 1; /* Lower only */
     } else {
       int rx;
-      fprintf(debug," UNKNOWN");
+      fprintf(debug," || UNKNOWN");
       for (rx=lx; rx<w && cols[rx]; rx++);
-      fprintf(debug," x=%d ctx=%d %d..%d\n",x, ctx, lx,rx);
+      fprintf(debug," x=%d ctxmap=%x %d..%d\n",x, ctxmap, lx,rx);
       debug_flush();
-      callout_unknown(w,h,cols, lx,rx-1,ctx, results,nresults);
+      callout_unknown(rd, w,cols, lx,rx-1, ctxmap);
       goto restart;
     }
   }
-  add_result(0,-1,-1,0);
-  fprintf(debug,"OCR  finished %d glyphs\n",nresults);
+  add_result(rd, 0,-1,-1,0);
+  fprintf(debug,"OCR  finished %d glyphs\n",rd->nresults);
   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;
 }