X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?a=blobdiff_plain;f=pctb%2Fstructure.c;h=200b86ea43c28b5e3a0bd58a5f299be8ed2be941;hb=9c4e6ae34ea06c776ebbf5410f7bc1ce78edad2f;hp=a3cca0b4408fc3da115bd91f069589956c0fb0f7;hpb=9b746cc0a0512dd570aa5d4b6185bc1007f74b03;p=ypp-sc-tools.db-live.git

diff --git a/pctb/structure.c b/pctb/structure.c
index a3cca0b..200b86e 100644
--- a/pctb/structure.c
+++ b/pctb/structure.c
@@ -96,7 +96,6 @@ static void mustfail1(const char *file, int line, const char *what) {
  "Unable to figure out contents of YPP client display.\n"
  "Please check the following:\n"
  "   * YPP client is showing commodity listing screen\n"
- "   * YPP client has antialiased text turned off (Ye / Options / General)\n"
  "   * YPP client window is on top (we try to raise it but your window\n"
  "      manager might have prevented that from succeeding)\n"
  "   * Your X display is 24bpp (NB some VNC servers use 16bpp by default)\n"
@@ -320,6 +319,8 @@ void find_structure(CanonImage *im, int *max_relevant_y_r) {
   SET_ONCE(text_h, comminty - 1);
   if (max_relevant_y_r)
     SET_ONCE(*max_relevant_y_r, mainr.br.y + 10);
+
+  MUST( text_h <= OCR_MAX_H, MI(text_h) );
 }		    
 
 void check_correct_commodities(void) {
@@ -460,28 +461,38 @@ static void find_table_entry(Rect commod, int colno, Rect *cell) {
   const RgbImage *ri= cim->rgb;
   
   Rgb background= ri_rgb(ri, cell->br.x, cell->br.y);
-  long bg_count=0, white_count=0, black_count=0;
+  long bg_count=0, light_count=0, dark_count=0;
   Point p;
   FOR_P_RECT(p,*cell) {
     Rgb here= ri_rgb(ri, p.x, p.y);
     if (here == background) bg_count++;
-    else if (here == 0) black_count++;
-    else if (here == 0xffffffU) white_count++;
+    else if (here < background) dark_count++;
+    else if (here > background) light_count++;
   }
   long total_count= RECT_W(*cell) * RECT_H(*cell);
-  MUST( bg_count > 8*total_count / 10,
-	MR(*cell);MIL(bg_count);MIL(white_count);MIL(black_count) );
+  MUST( bg_count > total_count / 2,
+	MR(*cell);MIL(total_count);MIL(bg_count);
+	MIL(light_count);MIL(dark_count) );
   if (bg_count == total_count)
     return;
 
-  MUST( !!black_count != !!white_count,
-	MR(*cell);MIL(bg_count);MIL(white_count);MIL(black_count) );
+  MUST( !!dark_count != !!light_count,
+	MR(*cell);MIL(total_count);MIL(bg_count);
+	MIL(light_count);MIL(dark_count) );
 
-  debugf("TABLEENTRY col=%d %d,%d..%d,%d bg=%ld white=%ld black=%ld\n",
+  debugf("TABLEENTRY col=%d %d,%d..%d,%d bg=%ld light=%ld dark=%ld\n",
 	 colno, cell->tl.x,cell->tl.y, cell->br.x,cell->br.y,
-	 bg_count, white_count, black_count);
+	 bg_count, light_count, dark_count);
   
-  Rgb foreground= white_count ? 0xffffffU : 0;
+  Rgb foreground;
+  double fg_extra;
+  if (light_count) {
+    foreground= 0xffffffU;
+    fg_extra= +1;
+  } else {
+    foreground= 0;
+    fg_extra= -1;
+  }
   int monochrome= 1;
 
   FOR_P_RECT(p,*cell) {
@@ -493,28 +504,30 @@ static void find_table_entry(Rect commod, int colno, Rect *cell) {
       unsigned char bg_chan=   background >> (i*8);
       unsigned char fg_chan=   foreground >> (i*8);
       double alpha_chan=
-	((double)here_chan - (double)bg_chan) /
-	((double)fg_chan   - (double)bg_chan);
+	((double)here_chan    - (double)bg_chan) /
+	((fg_chan + fg_extra) - (double)bg_chan);
       alpha[i]= alpha_chan;
-      alpha_mean += alpha_chan / 3;
+      alpha_mean += alpha_chan * (1/3.0);
     }
 
-    double alpha_min = alpha_mean - 0.13;
-    double alpha_max = alpha_mean + 0.13;
+    double thresh= 1.5/AAMAXVAL;
+    double alpha_min= alpha_mean - thresh;
+    double alpha_max= alpha_mean + thresh;
     for (i=0; i<3; i++)
       MUST( alpha_min <= alpha[i] && alpha[i] <= alpha_max,
 	    MI(i);MRGB(here);MRGB(background);MRGB(foreground);
 	    MF(alpha_min);MF(alpha[i]);MF(alpha_max) );
 
-    MUST( 0 <= alpha_mean && alpha_mean <= 1,
+    MUST( 0 <= alpha_mean && alpha_mean < 0.999,
 	  MRGB(here);MRGB(background);MRGB(foreground);
-	  MF(alpha_mean) );
-    int here_int= 7*alpha_mean;
-    if (!(here_int==0 || here_int==7)) monochrome=0;
+	  MF(alpha_mean); MF(alpha[0]);MF(alpha[1]);MF(alpha[2]); );
+    int here_int= floor((AAMAXVAL+1)*alpha_mean);
+    assert(here_int <= AAMAXVAL);
+    if (!(here_int==0 || here_int==AAMAXVAL)) monochrome=0;
     cim->d[p.y * cim->w + p.x]= '0' + here_int;
   }
 
-  debug_rect("cell07", colno, *cell);
+  debug_rect("cell0M", colno, *cell);
 
   require_rectangle_r(*cell, "0123456789", __LINE__);
 }
@@ -526,22 +539,17 @@ static void ocr_rectangle(Rect r, const OcrCellType ct, FILE *tsv_output) {
   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) {
+    FILLZERO(cols[x]);
+    for (y=0; y<text_h; y++) {
       Point here= { x+r.tl.x, y+r.tl.y };
       int pixel= get_p(here);
-      switch (pixel) {
-      case ' ':           break;
-      case '0':           break;
-      case '7': cx |= rv; break;
-      default:
-	MUST(!"wrong pixel",
-	     MC(pixel);MP(here);MSB(ocr_celltype_name(ct));MR(r); );
-      }
+      if (pixel==' ') pixel= '0';
+      MUST( pixel >= '0' && pixel <= '0'+AAMAXVAL,
+	    MC(pixel);MP(here);MSB(ocr_celltype_name(ct));MR(r); );
+      pixcol_p_add(&cols[x], y, pixel-'0');
     }
-    cols[x]= cx;
   }
-  cols[w]= 0;
+  FILLZERO(cols[w]);
 
   results= ocr(rd,ct,w,cols);
   for (res=results; res->s; res++)