From 344df47a4f31ad647d31152a56a6cc8b004ce087 Mon Sep 17 00:00:00 2001
From: Ian Jackson <ian@liberator.relativity.greenend.org.uk>
Date: Sun, 7 Jun 2009 19:43:31 +0100
Subject: [PATCH] Digit fields have different spacing

---
 pctb/ocr.c | 41 ++++++++++++++++++++++-------------------
 1 file changed, 22 insertions(+), 19 deletions(-)

diff --git a/pctb/ocr.c b/pctb/ocr.c
index fd4cba9..d12218f 100644
--- a/pctb/ocr.c
+++ b/pctb/ocr.c
@@ -46,10 +46,27 @@ static const char *context_names[]= {
   "Upper",
   "Digit"
 };
+struct OcrCellTypeInfo {
+  /* bitmaps of indices into context_names: */
+  unsigned initial, nextword, midword;
+  int space_spaces;
+  const char *name;
+};
+const struct OcrCellTypeInfo ocr_celltype_number= {
+  4,4,4,
+  .space_spaces= 5,
+  .name= "number"
+};
+const struct OcrCellTypeInfo ocr_celltype_text= {
+  .initial=2, /* Uppercase */
+  .nextword=3, /* Either */
+  .midword=1, /* Lower only */
+  .space_spaces= 4,
+  .name= "text"
+};
 
-#define NCONTEXTS (sizeof(context_names)/sizeof(context_names[0]))
 
-#define SPACE_SPACES 4
+#define NCONTEXTS (sizeof(context_names)/sizeof(context_names[0]))
 
 struct OcrReader {
   int h;
@@ -320,20 +337,6 @@ static void add_result(OcrReader *rd, const char *s, int l, int r,
   rd->nresults++;
 }
 
-struct OcrCellTypeInfo {
-  unsigned initial, nextword, midword;
-  const char *name;
-};
-const struct OcrCellTypeInfo ocr_celltype_number= {
-  4,4,4,
-  .name= "number"
-};
-const struct OcrCellTypeInfo ocr_celltype_text= {
-  .initial=2, /* Uppercase */
-  .nextword=3, /* Either */
-  .midword=1, /* Lower only */
-  .name= "text"
-};
 
 const char *ocr_celltype_name(OcrCellType ct) { return ct->name; }
 
@@ -362,7 +365,7 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
     if (!cols[x]) {
       nspaces++;
       x++;
-      if (nspaces==SPACE_SPACES) {
+      if (nspaces == ct->space_spaces) {
 	debugf("OCR  x=%x nspaces=%d space\n",x,nspaces);
 	ctxmap= ct->nextword;
       }
@@ -370,7 +373,7 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
     }
 
     /* something here, so we need to add the spaces */
-    if (nspaces>=SPACE_SPACES)
+    if (nspaces >= ct->space_spaces)
       add_result(rd," ",x-nspaces,x+1,0);
     nspaces=0;
 
@@ -435,7 +438,7 @@ OcrResultGlyph *ocr(OcrReader *rd, OcrCellType ct, int w, Pixcol cols[]) {
       if (uniquematch->s[0]) ctxmap= ct->midword;
       else debugf(" (empty)");
       if (uniquematch->endsword) {
-	nspaces= SPACE_SPACES;
+	nspaces= ct->space_spaces;
 	debugf("_");
 	ctxmap= ct->nextword;
       }
-- 
2.30.2