X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?a=blobdiff_plain;f=pctb%2Fconvert.c;h=f90ca4a450456d3b5d437e4a093da3569fa6f975;hb=5a2c03e2e4f52b8329f45cf67afc3edec1f2c65b;hp=95e75b3e613b55813df459168da9c55bd939a171;hpb=f9c00eb882d1758e95422d8f40ef5606422cffa3;p=ypp-sc-tools.web-live.git

diff --git a/pctb/convert.c b/pctb/convert.c
index 95e75b3..f90ca4a 100644
--- a/pctb/convert.c
+++ b/pctb/convert.c
@@ -1,11 +1,5 @@
-#include <pam.h>
-#include <stdint.h>
-#include <inttypes.h>
-#include <assert.h>
-#include <string.h>
 
-#define eassert assert
-#define debug stdout
+#include "ocr.h"
 
 typedef struct {
   unsigned long rgb; /* on screen */
@@ -15,7 +9,7 @@ typedef struct {
 static int height, width;
 static char *image;
 
-static void debug_flush(void) {
+void debug_flush(void) {
   eassert(!fflush(debug));
   eassert(!ferror(debug));
 }
@@ -33,6 +27,18 @@ 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 TEXT_COLUMNS        2
+#define MAX_COLUMNS         7
+
+static Rect mainr = { START_MAIN,START_MAIN };
+static int commbasey, comminty;
+static int colrightx[INTERESTING_COLUMNS];
+static int text_h;
+static OcrReader *rd;
+
 static const CanonColourInfo canoncolourinfos[]= {
   { 0x475A5E, '*' }, /* edge */
   { 0x2C5F7A, '*' }, /* edge just under box heading shadow */
@@ -45,16 +51,17 @@ static const CanonColourInfo canoncolourinfos[]= {
   { 0,0 }
 };
 
-#define START_MAIN {200,200}
-
-static void require_rectangle(int tlx, int tly, int brx, int bry, char c) {
+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++)
-      eassert(get(x,y) == c);
+    for (y=tly; y<=bry; y++) {
+      int c= get(x,y);
+      assert(strchr(ok,c));
+    }
 }
-static void require_rectangle_r(Rect rr, char c) {
-  require_rectangle(rr.tl.x,rr.tl.y, rr.br.x,rr.br.y, 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) {
@@ -63,7 +70,7 @@ static void debug_rect(const char *what, int whati, Rect rr) {
 	  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 %4d ",y,y-rr.tl.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);
@@ -72,26 +79,31 @@ static void debug_rect(const char *what, int whati, Rect rr) {
   debug_flush();
 }
 
-static void find_main_rectangle(void) {
-  Rect whole = { {0,0}, {width-1,height-1} };
-  Rect mainr = { START_MAIN,START_MAIN };
-
-#define WALK_UNTIL(point,coord,increm,stop,edge)			\
+#define WALK_UNTIL(point,coord,increm,last,edge)			\
   for (;;) {								\
+    if ((point).coord == (last)+(increm)) break;			\
     if (get_p((point)) == (edge)) { (point).coord -= (increm); break; }	\
-    eassert((point).coord != (stop));				\
     (point).coord += (increm);						\
   }
 
-  WALK_UNTIL(mainr.tl, x,-1, whole.tl.x, '*');
-  WALK_UNTIL(mainr.tl, y,-1, whole.tl.y, '*');
-  WALK_UNTIL(mainr.br, x,+1, whole.br.x, '*');
-  WALK_UNTIL(mainr.br, y,+1, whole.br.y, '*');
+#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, '*');
+  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 {						\
@@ -103,7 +115,7 @@ static void find_main_rectangle(void) {
       csb_r= mainr;				\
       csb_r.tl.xy= csb_p.xy;			\
       csb_r.br.xy= csb_p.xy;			\
-      require_rectangle_r(csb_r, '+');		\
+      require_rectangle_r(csb_r, "+");		\
       mainr.tlbr.xy += increm;			\
     }						\
   } while(0)
@@ -116,8 +128,81 @@ static void find_main_rectangle(void) {
   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);
+
+  text_h = comminty - 1;
 }		    
 
+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;
   unsigned char rgb[3];
@@ -154,11 +239,57 @@ static void load_image_and_canonify(void) {
   debug_flush();
 }
 
+static void ocr_rectangle(Rect r, const OcrCellType ct) {
+  OcrResultGlyph *results, *res;
+
+  int w= r.br.x - r.tl.x + 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<text_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");
+      }
+    }
+    cols[x]= cx;
+  }
+  cols[w]= 0;
+
+  results= ocr(rd,ct,w,cols);
+  printf("YES! \"");
+  for (res=results; res->s; res++)
+    printf("%s",res->s);
+  printf("\"\n");
+  eassert(!ferror(stdout));
+  eassert(!fflush(stdout));
+}
+
 int main(void) {
+  Rect thisr, entryr;
+  int tryrect, colno;
+
   load_image_and_canonify();
-  find_main_rectangle();
-  /*
-  repeatedly_find_top_thing();
-  */
+  find_structure();
+  rd= ocr_init(text_h);
+
+  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,
+		    colno<TEXT_COLUMNS
+		    ? &ocr_celltype_text
+		    : &ocr_celltype_number);
+    }
+  }
   return 0;
 }
+
+const char *get_vardir(void) { return "."; }