Remaining changes for Unicodeization project

git-svn-id: https://tesseract-ocr.googlecode.com/svn/trunk@87 d0cd1f9f-072b-0410-8dd7-cf729c803f20
lsfjGitHub · Jul 18, 2007 · 570af48 · 570af48
1 parent 627368d
commit 570af48
Show file tree

Hide file tree

Showing 97 changed files with 9,091 additions and 10,032 deletions.
diff --git a/Makefile.in b/Makefile.in
diff --git a/ccmain/Makefile.in b/ccmain/Makefile.in
diff --git a/ccmain/adaptions.cpp b/ccmain/adaptions.cpp
@@ -137,7 +137,9 @@ BOOL8 word_adaptable(  //should we adapt?
 
 //  if (flags.bit (CHECK_AMBIG_WERD) && test_ambig_word (word))
   if (flags.bit (CHECK_AMBIG_WERD) &&
-      !NoDangerousAmbig(word->best_choice->string().string(), NULL))
+      !NoDangerousAmbig(word->best_choice->string().string(),
+                        word->best_choice->lengths().string(),
+                        NULL))
     return FALSE;
 
   return status;

diff --git a/ccmain/applybox.h b/ccmain/applybox.h
@@ -24,6 +24,7 @@
 #include          "ocrblock.h"
 #include          "ocrrow.h"
 #include          "notdll.h"
+#include          "unichar.h"
 
 extern BOOL_VAR_H (applybox_rebalance, TRUE, "Drop dead");
 extern INT_VAR_H (applybox_debug, 0, "Debug level");
@@ -37,7 +38,7 @@ void clear_any_old_text(                        //remove correct text
                        );
 BOOL8 read_next_box(FILE* box_file,  //
                     BOX *box,
-                    char *ch);
+                    UNICHAR_ID *uch_id);
 ROW *find_row_of_box(                         //
                      BLOCK_LIST *block_list,  //real blocks
                      BOX box,                 //from boxfile
@@ -46,7 +47,7 @@ ROW *find_row_of_box(                         //
 INT16 resegment_box(  //
                     ROW *row,
                     BOX box,
-                    char *ch,
+                    UNICHAR_ID uch_id,
                     INT16 block_id,
                     INT16 row_id,
                     INT16 boxfile_lineno,
@@ -58,13 +59,13 @@ void tidy_up(                         //
              INT16 &unlabelled_words,
              INT16 *tgt_char_counts,
              INT16 &rebalance_count,
-             char &min_char,
+             UNICHAR_ID *min_uch_id,
              INT16 &min_samples,
              INT16 &final_labelled_blob_count);
 void report_failed_box(INT16 boxfile_lineno,
                        INT16 boxfile_charno,
                        BOX box,
-                       char *box_ch,
+                       const char *box_ch,
                        const char *err_msg);
 void apply_box_training(BLOCK_LIST *block_list);
 void apply_box_testing(BLOCK_LIST *block_list);

diff --git a/ccmain/blobcmp.cpp b/ccmain/blobcmp.cpp
@@ -27,7 +27,7 @@
 #include "const.h"
 #include "tessvars.h"
 
-#define CMP_CLASS       'x'
+#define CMP_CLASS       1
 
 /**********************************************************************
  * compare_tess_blobs

diff --git a/ccmain/control.h b/ccmain/control.h
@@ -155,11 +155,11 @@ void fix_rep_char(                //Repeated char word
                   WERD_RES *word  //word to do
                  );
 void fix_quotes(               //make double quotes
-                char *string,  //string to fix
+                WERD_CHOICE *choice,  //string to fix
                 WERD *word,    //word to do //char choices
                 BLOB_CHOICE_LIST_CLIST *blob_choices);
 void fix_hyphens(               //crunch double hyphens
-                 char *string,  //string to fix
+                 WERD_CHOICE *choice,  //string to fix
                  WERD *word,    //word to do //char choices
                  BLOB_CHOICE_LIST_CLIST *blob_choices);
 void merge_blobs(               //combine 2 blobs
@@ -175,7 +175,8 @@ void choice_dump_tester(                           //dump chars in word
                         BLOB_CHOICE_LIST *ratings  //list of results
                        );
 WERD *make_bln_copy(WERD *src_word, ROW *row, float x_height, DENORM *denorm);
-ACCEPTABLE_WERD_TYPE acceptable_word_string(const char *s);
+ACCEPTABLE_WERD_TYPE acceptable_word_string(const char *s,
+                                            const char *lengths);
 BOOL8 check_debug_pt(WERD_RES *word, int location);
 void set_word_fonts(                 //good chars in word
                     WERD_RES *word,  //word to adapt to //detailed results