X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?a=blobdiff_plain;f=pctb%2Fconvert.c;h=a0ecca16703749cd47ef6dbd81c03437e2868a33;hb=54f10d7fe3898d21ea03d544c9c3d3f1c0561969;hp=80bc894559656714c48e6c90b2c193a44309d57a;hpb=0051984495aa2a5ed61224a7618f5bbd04311f7a;p=ypp-sc-tools.db-live.git

diff --git a/pctb/convert.c b/pctb/convert.c
index 80bc894..a0ecca1 100644
--- a/pctb/convert.c
+++ b/pctb/convert.c
@@ -3,16 +3,50 @@
 #include <inttypes.h>
 #include <assert.h>
 #include <string.h>
+#include <stdlib.h>
 
 #define eassert assert
+#define debug stdout
 
 typedef struct {
   unsigned long rgb; /* on screen */
   char c; /* canonical */
 } CanonColourInfo;
 
+static int height, width;
+static char *image;
+
+static void debug_flush(void) {
+  eassert(!fflush(debug));
+  eassert(!ferror(debug));
+}
+
+typedef struct {
+  int x, y;
+} Point;
+
+typedef struct { /* both inclusive */
+  Point tl;
+  Point br;
+} Rect;
+
+static inline char get(int x, int y) { return image[y * width + x]; }
+static inline char get_p(Point p) { return get(p.x,p.y); }
+
+
+#define START_MAIN {200,200}
+#define MIN_COLUMNS         6
+#define INTERESTING_COLUMNS 6
+#define MAX_COLUMNS         7
+
+static Rect mainr = { START_MAIN,START_MAIN };
+static int commbasey, comminty;
+static int colrightx[INTERESTING_COLUMNS];
+
+
 static const CanonColourInfo canoncolourinfos[]= {
   { 0x475A5E, '*' }, /* edge */
+  { 0x2C5F7A, '*' }, /* edge just under box heading shadow */
   { 0x7D9094, '+' }, /* interbox */
   { 0xBDC5BF, ' ' }, /* background - pale */
   { 0xADB5AF, ' ' }, /* background - dark */
@@ -22,8 +56,155 @@ static const CanonColourInfo canoncolourinfos[]= {
   { 0,0 }
 };
 
-static int height, width;
-static char *image;
+static void require_rectangle(int tlx, int tly, int brx, int bry,
+			      const char *ok) {
+  int x,y;
+  for (x=tlx; x<=brx; x++)
+    for (y=tly; y<=bry; y++) {
+      int c= get(x,y);
+      assert(strchr(ok,c));
+    }
+}
+static void require_rectangle_r(Rect rr, const char *ok) {
+  require_rectangle(rr.tl.x,rr.tl.y, rr.br.x,rr.br.y, ok);
+}
+
+static void debug_rect(const char *what, int whati, Rect rr) {
+  int y,r,w;
+  fprintf(debug, "%s %d: %d,%d..%d,%d:\n", what, whati,
+	  rr.tl.x,rr.tl.y, rr.br.x,rr.br.y);
+  w= rr.br.x - rr.tl.x + 1;
+  for (y=rr.tl.y; y<=rr.br.y; y++) {
+    fprintf(debug, "%4d%*s|", y, rr.tl.x,"");
+    r= fwrite(image + y*width + rr.tl.x, 1, w, debug);
+    eassert(r==w);
+    fputc('|',debug);
+    fputc('\n',debug);
+  }
+  debug_flush();
+}
+
+#define WALK_UNTIL(point,coord,increm,last,edge)			\
+  for (;;) {								\
+    if ((point).coord == (last)+(increm)) break;			\
+    if (get_p((point)) == (edge)) { (point).coord -= (increm); break; }	\
+    (point).coord += (increm);						\
+  }
+
+#define WALK_UNTIL_MUST(point,coord,increm,last,edge)	\
+  do {							\
+    WALK_UNTIL(point,coord,increm,last,edge);		\
+    eassert((point).coord != (last)+(increm));		\
+  } while(0)
+
+static void find_structure(void) {
+  Rect whole = { {0,0}, {width-1,height-1} };
+
+  WALK_UNTIL_MUST(mainr.tl, x,-1, whole.tl.x, '*');
+  WALK_UNTIL_MUST(mainr.tl, y,-1, whole.tl.y, '*');
+  WALK_UNTIL_MUST(mainr.br, x,+1, whole.br.x, '*');
+  WALK_UNTIL_MUST(mainr.br, y,+1, whole.br.y, '*');
+
+  require_rectangle(mainr.tl.x-1, mainr.tl.y, mainr.tl.x-1, mainr.br.y, "*");
+  require_rectangle(mainr.br.x+1, mainr.tl.y, mainr.br.x+1, mainr.br.y, "*");
+  require_rectangle(mainr.tl.x, mainr.tl.y-1, mainr.br.x, mainr.tl.y-1, "*");
+  require_rectangle(mainr.tl.x, mainr.br.y+1, mainr.br.x, mainr.br.y+1, "*");
+
+#define CHECK_STRIP_BORDER(tlbr,xy,increm)	\
+  do {						\
+    Point csb_p;				\
+    Rect csb_r;					\
+    csb_p= mainr.tl;				\
+    csb_p.xy= mainr.tlbr.xy;			\
+    if (get_p(csb_p)=='+') {			\
+      csb_r= mainr;				\
+      csb_r.tl.xy= csb_p.xy;			\
+      csb_r.br.xy= csb_p.xy;			\
+      require_rectangle_r(csb_r, "+");		\
+      mainr.tlbr.xy += increm;			\
+    }						\
+  } while(0)
+
+  debug_rect("mainr",0, mainr);
+
+  CHECK_STRIP_BORDER(tl,x,+1);
+  CHECK_STRIP_BORDER(tl,y,+1);
+  CHECK_STRIP_BORDER(br,x,-1);
+  CHECK_STRIP_BORDER(br,y,-1);
+
+  debug_rect("mainr",1, mainr);
+
+  Point up = START_MAIN;
+  WALK_UNTIL_MUST(up, y,-1, mainr.tl.y, '+');
+
+  Point down = START_MAIN;
+  down.y++;
+  WALK_UNTIL_MUST(down, y,+1, mainr.br.y, '+');
+
+  int xscaleunit, y,x;
+  for (y=0, xscaleunit=1; y<4; y++, xscaleunit*=10) {
+    fprintf(debug,"     ");
+    for (x=0; x<=width; x++) {
+      if (x % xscaleunit) fputc(' ',debug);
+      else fprintf(debug,"%d",(x / xscaleunit)%10);
+    }
+    fputc('\n',debug);
+  }
+
+  commbasey= up.y;
+  comminty= down.y - up.y + 2;
+  fprintf(debug, "up.y=%d down.y=%d commbasey=%d comminty=%d\n",
+	  up.y,down.y, commbasey,comminty);
+
+  Point across= { mainr.tl.x, commbasey };
+  int colno=0;
+  for (;;) {
+    eassert(get_p(across) != '+');
+    WALK_UNTIL(across, x,+1, mainr.br.x, '+');
+    eassert(colno < MAX_COLUMNS);
+    int colrx= across.x;
+    if (colrx > mainr.br.x) colrx= mainr.br.x;
+    if (colno < INTERESTING_COLUMNS) {
+      colrightx[colno]= colrx;
+      fprintf(debug,"colrightx[%d]= %d\n",colno,colrx);
+    } else {
+      fprintf(debug,"extra colr %d  %d\n",colno,colrx);
+    }
+      
+    colno++;
+    
+    if (across.x >= mainr.br.x-1)
+      break;
+
+    across.x++;
+    require_rectangle(across.x,mainr.tl.y, across.x,mainr.br.y, "+");
+    across.x++;
+  }
+  eassert(colno >= MIN_COLUMNS);
+}		    
+
+static void find_commodity(int offset, Rect *rr) {
+  /* rr->tl.x==-1 if offset out of range */
+  rr->tl.y= commbasey - offset*comminty;
+  rr->br.y= rr->tl.y + comminty-2;
+  if (rr->tl.y < mainr.tl.y || rr->br.y > mainr.br.y) { rr->tl.x=-1; return; }
+  if (rr->tl.y > mainr.tl.y)
+    require_rectangle(rr->tl.x,rr->tl.y-1, rr->br.x,rr->tl.y-1, "+");
+  if (rr->br.y < mainr.tl.y)
+    require_rectangle(rr->tl.x,rr->br.y+1, rr->br.x,rr->br.y+1, "+");
+  
+  rr->tl.x= mainr.tl.x;
+  rr->br.x= mainr.br.x;
+}
+
+static void find_table_entry(Rect commod, int colno, Rect *cellr) {
+  cellr->tl.y= commod.tl.y;
+  cellr->br.y= commod.br.y;
+  cellr->tl.x= !colno ? commod.tl.x : colrightx[colno-1]+2;
+  cellr->br.x=                        colrightx[colno];
+  debug_rect("cell", colno, *cellr);
+  require_rectangle_r(*cellr, " o");
+}
 
 static void load_image_and_canonify(void) {
   struct pam inpam;
@@ -53,23 +234,176 @@ static void load_image_and_canonify(void) {
 	  image[y*width + x]= cci->c;
 	  break;
 	}
-      if (y==234 && x==82) {
-	printf("y=%d/%d x=%d/%d rgb=%d,%d,%d rgb_l=%lx c=%c\n",
-	       y,height,x,width, rgb[0],rgb[1],rgb[2], rgb_l, image[y*width+x]);
+    }
+    fprintf(debug, "%4d ",y);
+    r= fwrite(image + y*width, 1,width, debug);  eassert(r==width);
+    fputc('\n',debug);
+  }
+  debug_flush();
+}
+
+typedef uint32_t Pixcol;
+#define PSPIXCOL(priscan) priscan##32
+
+typedef struct {
+  Pixcol col;
+  struct OCRDatabaseNode *then;
+} OCRDatabaseLink;
+
+#define MAXGLYPHCHRS 3
+
+typedef struct OCRDatabaseNode {
+  char s[MAXGLYPHCHRS+1]; /* null-terminated; "" means no match here */
+  int nlinks, alinks;
+  OCRDatabaseLink *links;
+} OCRDatabaseNode;
+
+#define N_OCR_CONTEXTS 2
+static OCRDatabaseNode ocr_contexts[N_OCR_CONTEXTS];
+
+static void load_ocr_database(void) {
+  int ctx,nchrs;
+  OCRDatabaseNode *current, *additional;
+  char chrs[MAXGLYPHCHRS+1];
+  Pixcol cv;
+  int r,i,j;
+
+  FILE *db= fopen("database","r");  eassert(db);
+
+  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;
+    }
+    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);
+      for (j=0; j<current->nlinks; j++)
+	if (current->links[j].col == cv) {
+	  current= current->links[j].then;
+	  goto found_link;
+	}
+
+      additional= malloc(sizeof(*additional)); eassert(additional);
+      additional->s[0]= 0;
+      additional->nlinks= additional->alinks= 0;
+      additional->links= 0;
+      if (current->nlinks==current->alinks) {
+	current->alinks++;
+	current->alinks<<=1;
+	current->links= realloc(current->links,
+            sizeof(*current->links) * current->alinks);
+	eassert(current->links);
+      }
+      current->links[current->nlinks].col= cv;
+      current->links[current->nlinks].then= additional;
+      current->nlinks++;
+      current= additional;
+
+    found_link:;
+    }
+
+    eassert(!current->s[0]);
+    strcpy(current->s, chrs);
+  }
+  eassert(!ferror(db));
+  eassert(feof(db));
+  fclose(db);
+}      
+
+static void ocr_rectangle(Rect r) {
+  int w= r.br.x - r.tl.x + 1;
+  int h= r.br.y - r.tl.y + 1;
+  Pixcol cols[w+1];
+  int x,y;
+  for (x=0; x<w; x++) {
+    Pixcol cx, rv;
+    for (y=0, cx=0, rv=1; y<h; y++, rv<<=1) {
+      switch (get(x+r.tl.x, y+r.tl.y)) {
+      case ' ':           break;
+      case 'o': cx |= rv; break;
+      default: eassert(!"wrong pixel");
       }
     }
-    r= fwrite(image + y*width, 1,width, stdout);  eassert(r==width);
-    putchar('\n');
+    cols[x]= cx;
+  }
+  cols[w]= 0;
+
+  int nspaces=0;
+  int ctx=1,i;
+  x=0;
+
+  for (;;) {
+    if (x>w) break;
+
+    if (!cols[x]) {
+      nspaces++;
+      x++;
+      if (nspaces>3) ctx=1;
+      continue;
+    }
+      
+    OCRDatabaseNode *current=0, *lastmatch=0;
+    int startx=x;
+    int afterlastmatchx=-1;
+    current= &ocr_contexts[ctx];
+    for (;;) {
+      if (x>w) break;
+      Pixcol cv= cols[x];
+      for (i=0; i<current->nlinks; i++)
+	if (current->links[i].col == cv)
+	  goto found;
+      /* not found */
+      break;
+    found:
+      x++;
+      current= current->links[i].then;
+      if (current->s[0]) { lastmatch=current; afterlastmatchx=x; }
+    }
+
+    if (!lastmatch) {
+      int x2;
+      for (x2=x+1; x2<w && cols[x2]; x2++);
+      printf("UNKNOWN x=%d ctx=%d %d..%d\n",x, ctx, startx,x2);
+      x++;
+    } else {
+      printf("OUTPUT x=%d `%s'\n", x, lastmatch->s);
+      x= afterlastmatchx;
+      ctx= 0;
+    }
   }
-  eassert(!fflush(stdout));
-  eassert(!ferror(stdout));
 }
- 
+
 int main(void) {
+  Rect thisr, entryr;
+  int tryrect, colno;
+
+  load_ocr_database();
   load_image_and_canonify();
-  /*
-  find_main_rectangle();
-  repeatedly_find_top_thing();
-  */
+  find_structure();
+
+  for (tryrect= +height; tryrect >= -height; tryrect--) {
+    find_commodity(tryrect, &thisr);
+    if (thisr.tl.x < 0)
+      continue;
+    debug_rect("commod",tryrect, thisr);
+    
+    for (colno=0; colno<MIN_COLUMNS; colno++) {
+      find_table_entry(thisr,colno,&entryr);
+      ocr_rectangle(entryr);
+    }
+  }
   return 0;
 }