X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~yarrgweb/git?a=blobdiff_plain;f=pctb%2Fstructure.c;h=49385a2fb8b89e822562ea7408118aeafc892a82;hb=ae5d4e36af4f4a5567e1231958179d3a50eb4466;hp=8141feb3db461c60099841db25d3a4954d018dec;hpb=8b296a5f4da05a4dcf356f71e1ce4e63d8d282cf;p=ypp-sc-tools.db-live.git

diff --git a/pctb/structure.c b/pctb/structure.c
index 8141feb..49385a2 100644
--- a/pctb/structure.c
+++ b/pctb/structure.c
@@ -47,6 +47,7 @@ static OcrReader *rd;
 const CanonColourInfo canoncolourinfos[]= {
   { 0x475A5E, '*' }, /* edge */
   { 0x2C5F7A, '*' }, /* edge just under box heading shadow */
+  { 0xC5C7Ae, '*' }, /* blank area of partial commodities list */
   { 0x7D9094, '+' }, /* interbox */
 
   { 0xBDC5BF, ' ' }, /* background - pale  Sugar cane, etc. */
@@ -111,20 +112,23 @@ static void mustfail2(void) {
                       #v,(v).tl.x,(v).tl.y,(v).br.x,(v).br.y)
 
 
+#define REQUIRE_RECTANGLE(tlx,tly,brx,bry,ok) \
+ require_rectangle(tlx, tly, brx, bry, ok, __LINE__);
+
 static void require_rectangle(int tlx, int tly, int brx, int bry,
-			      const char *ok) {
+			      const char *ok, int lineno) {
   Point p;
   for (p.x=tlx; p.x<=brx; p.x++)
     for (p.y=tly; p.y<=bry; p.y++) {
       int c= get_p(p);
       MUST( strchr(ok,c), ({
 	     Rect rm={{tlx,tly},{brx,bry}};
-	     MR(rm);MP(p);MS(ok);
+	     MI(lineno),MR(rm);MP(p);MS(ok);
       }));
     }
 }
-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 require_rectangle_r(Rect rr, const char *ok, int lineno) {
+  require_rectangle(rr.tl.x,rr.tl.y, rr.br.x,rr.br.y, ok, lineno);
 }
 
 static void debug_rect(const char *what, int whati, Rect rr) {
@@ -167,24 +171,24 @@ void find_structure(CanonImage *im) {
   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;			\
-    }						\
+  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, "+", __LINE__);	\
+      mainr.tlbr.xy += increm;				\
+    }							\
   } while(0)
 
   debug_rect("mainr",0, mainr);
@@ -235,7 +239,7 @@ void find_structure(CanonImage *im) {
       break;
 
     across.x++;
-    require_rectangle(across.x,mainr.tl.y, across.x,mainr.br.y, "+");
+    REQUIRE_RECTANGLE(across.x,mainr.tl.y, across.x,mainr.br.y, "+");
     across.x++;
   }
   MUST( colno >= MIN_COLUMNS, MI(colno);MR(mainr);MP(across); );
@@ -280,7 +284,7 @@ static void file_read_image_ppm(FILE *f) {
 	               (rgb_buf[2]);
   });
 
-  sysassert(!ferror(screenshots_file));
+  sysassert(!ferror(screenshot_file));
 
   if (!(npages < MAX_PAGES))
     fatal("Too many images in screenshots file; max is %d.\n", MAX_PAGES);
@@ -290,28 +294,28 @@ static void file_read_image_ppm(FILE *f) {
 
 void read_one_screenshot(void) {
   progress("reading screenshot...");
-  file_read_image_ppm(screenshots_file);
+  file_read_image_ppm(screenshot_file);
   progress_log("read screenshot.");
 }
 
 void read_screenshots(void) {
   struct stat stab;
   
-  sysassert(! fstat(fileno(screenshots_file), &stab) );
+  sysassert(! fstat(fileno(screenshot_file), &stab) );
   
   for (;;) {
     if (S_ISREG(stab.st_mode)) {
-      long pos= ftell(screenshots_file);
+      long pos= ftell(screenshot_file);
       if (pos == stab.st_size) break;
     } else {
-      int c= fgetc(screenshots_file);
+      int c= fgetc(screenshot_file);
       if (c==EOF) break;
-      ungetc(c, screenshots_file);
+      ungetc(c, screenshot_file);
     }
     progress("reading screenshot %d...",npages);
-    file_read_image_ppm(screenshots_file);
+    file_read_image_ppm(screenshot_file);
   }
-  sysassert(!ferror(screenshots_file));
+  sysassert(!ferror(screenshot_file));
   progress_log("read %d screenshots.",npages);
 }
 
@@ -325,9 +329,9 @@ static void find_commodity(int offset, Rect *rr) {
   rr->br.x= mainr.br.x;
 
   if (rr->tl.y > mainr.tl.y)
-    require_rectangle(rr->tl.x,rr->tl.y-1, rr->br.x,rr->tl.y-1, "+");
+    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, "+");
+    REQUIRE_RECTANGLE(rr->tl.x,rr->br.y+1, rr->br.x,rr->br.y+1, "+");
 }
 
 static void find_table_entry(Rect commod, int colno, Rect *cellr) {
@@ -336,7 +340,7 @@ static void find_table_entry(Rect commod, int colno, Rect *cellr) {
   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");
+  require_rectangle_r(*cellr, " o", __LINE__);
 }
 
 static void ocr_rectangle(Rect r, const OcrCellType ct, FILE *tsv_output) {