X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?a=blobdiff_plain;f=pctb%2Fconvert.c;h=f90ca4a450456d3b5d437e4a093da3569fa6f975;hb=e684cf18e3fc8ce4024a6a258033d38f67f5aef0;hp=b518dbabf6913e202761dfc9b34cd524d77d4d55;hpb=2f9b824c735e9174d7cc7d8aea829a85d2276d21;p=ypp-sc-tools.db-test.git

diff --git a/pctb/convert.c b/pctb/convert.c
index b518dba..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));
 }
@@ -34,13 +28,16 @@ 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
+#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 */
@@ -178,7 +175,9 @@ static void find_structure(void) {
     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) {
@@ -195,6 +194,15 @@ static void find_commodity(int offset, Rect *rr) {
   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];
@@ -231,16 +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_structure();
+  rd= ocr_init(text_h);
 
-  Rect thisr;
-  int tryrect;
   for (tryrect= +height; tryrect >= -height; tryrect--) {
     find_commodity(tryrect, &thisr);
-    if (thisr.tl.x >= 0)
-      debug_rect("commod",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 "."; }