Fixed the dawg crash (edge_char_of/letter_is_okay) issue 128 and dupl…

…icates git-svn-id: https://tesseract-ocr.googlecode.com/svn/trunk@205 d0cd1f9f-072b-0410-8dd7-cf729c803f20
lzpfmh · Dec 24, 2008 · 04c4620 · 04c4620
1 parent cb3b9b4
commit 04c4620
Show file tree

Hide file tree

Showing 4 changed files with 14 additions and 21 deletions.
diff --git a/dict/dawg.cpp b/dict/dawg.cpp
@@ -144,18 +144,6 @@ inT32 def_letter_is_okay(EDGE_ARRAY dawg,
   edge = edge_char_of(dawg, *node,
                       static_cast<unsigned char>(dummy_word[char_index]),
                       word_end);
-#if 0
-  // TODO(rays) Turn dawg to a class so this can be used properly.
-  // The maximum value of edge is unknown to this function, but could
-  // easily be stored in a dawg class.
-  if ((edge < 0 || edge >= 10000000)  // MAX_NUM_EDGES_IN_SQUISHED_DAWG_FILE)
-      && edge != NO_EDGE) {
-    tprintf("Impossible return from edge_char_of: node=" REFFORMAT
-            ", ch=%d, edge=" REFFORMAT "\n",
-            *node, static_cast<unsigned char>(dummy_word[char_index]), edge);
-    edge = NO_EDGE;
-  }
-#endif
 
   if (edge != NO_EDGE) {         /* Normal edge in DAWG */
     if (case_sensative || case_is_okay (dummy_word, char_index)) {

diff --git a/dict/permdawg.cpp b/dict/permdawg.cpp
@@ -199,7 +199,12 @@ void append_next_choice(  /*previous option */
     better_choice = new_choice (word, unichar_lengths, rating, certainty, -1, permuter);
     adjust_word(better_choice, certainty_array);
     push_on(*result, better_choice);
+    if (permuter == SYSTEM_DAWG_PERM) {
+      // Until dawg is a class, it is only safe to use set_hyphen_word
+      // on a single dawg type, as it saves permuter state to use on another
+      // word in hyphen_state, and this *must* apply to the same dawg.
     set_hyphen_word(word, unichar_lengths, unichar_offsets, rating, node);
+    }
   }
   /* Look up char in DAWG */
   else {
@@ -319,8 +324,7 @@ void dawg_permute_and_select(const char *string,
                              EDGE_ARRAY dawg,
                              char permuter,
                              CHOICES_LIST character_choices,
-                             A_CHOICE *best_choice,
-                             inT16 system_words) {
+                             A_CHOICE *best_choice) {
   CHOICES result = NIL;
   char word[UNICHAR_LEN * MAX_WERD_LENGTH + 1];
   char unichar_lengths[MAX_WERD_LENGTH + 1];
@@ -352,7 +356,10 @@ void dawg_permute_and_select(const char *string,
         unichar_offsets[hyphen_base_size() - 1] +
         unichar_lengths[hyphen_base_size() - 1];
     char_index = strlen (hyphen_unichar_lengths);
-    if (system_words)
+    if (permuter == SYSTEM_DAWG_PERM)
+      // Until dawg is a class, it is only safe to use set_hyphen_word
+      // on a single dawg type, as it saves permuter state to use on another
+      // word in hyphen_state, and this *must* apply to the same dawg.
       dawg_node = hyphen_state;
   }
 

diff --git a/dict/permdawg.h b/dict/permdawg.h
@@ -88,8 +88,7 @@ void dawg_permute_and_select(const char *string,
                              EDGE_ARRAY dawg,
                              char permuter,
                              CHOICES_LIST character_choices,
-                             A_CHOICE *best_choice,
-                             inT16 system_words);
+                             A_CHOICE *best_choice);
 
 void init_permdawg_vars();
 void init_permdawg();

diff --git a/dict/permute.cpp b/dict/permute.cpp
@@ -1670,14 +1670,13 @@ A_CHOICE *permute_words(CHOICES_LIST char_choices, float rating_limit) {
   else {
 
     dawg_permute_and_select ("system words:", word_dawg, SYSTEM_DAWG_PERM,
-      char_choices, best_choice, TRUE);
+      char_choices, best_choice);
 
     dawg_permute_and_select ("document_words", document_words,
-      DOC_DAWG_PERM, char_choices, best_choice,
-      FALSE);
+      DOC_DAWG_PERM, char_choices, best_choice);
 
     dawg_permute_and_select ("user words", user_words, USER_DAWG_PERM,
-      char_choices, best_choice, FALSE);
+      char_choices, best_choice);
   }
 
   return (best_choice);