Made some preliminary changes for improving xheights

git-svn-id: https://tesseract-ocr.googlecode.com/svn/trunk@107 d0cd1f9f-072b-0410-8dd7-cf729c803f20

ccmain/control.cpp

@@ -606,7 +606,7 @@ void classify_word_pass1(                 //recog one word
 
   check_debug_pt (word, 0);
   matcher_pass = 0;
-  bln_word = make_bln_copy (word->word, row, row->x_height (), &word->denorm);
+  bln_word = make_bln_copy (word->word, row, word->x_height, &word->denorm);
 
   word->best_choice = tess_segment_pass1 (bln_word, &word->denorm,
     tess_default_matcher,
@@ -744,14 +744,15 @@ void classify_word_pass2(  //word to do
   if (!word->done ||
     tessedit_training_tess ||
   tessedit_training_wiseowl || tessedit_dump_choices) {
-    word->x_height = row->x_height ();
     word->caps_height = 0.0;
+    if (word->x_height == 0.0f)
+      word->x_height = row->x_height();
     if (word->outword != NULL) {
       delete word->outword;      //get rid of junk
       delete word->best_choice;
       delete word->raw_choice;
     }
-    match_word_pass2 (word, row, row->x_height ());
+    match_word_pass2 (word, row, word->x_height);
     done_this_pass = TRUE;
     check_debug_pt (word, 40);
   }

ccmain/fixspace.cpp

@@ -431,6 +431,7 @@ void transform_to_next_perm(WERD_RES_LIST &words) {
               //deep copy
               combo = new WERD_RES (copy_word);
               combo->combination = TRUE;
+              combo->x_height = prev_word->x_height;
               prev_word->part_of_combo = TRUE;
               prev_word_it.add_before_then_move (combo);
             }

ccmain/tfacepp.cpp

@@ -348,8 +348,10 @@ LIST call_matcher(                  //call a matcher
   char choice_lengths[2] = {0, 0};
 
   blob = make_ed_blob (tessblob);//convert blob
-  if (blob == NULL)
+  if (blob == NULL) {
+    tprintf("Failed to convert blob for recognition!\n");
     return NULL;                 //can't do it
+  }
   pblob = ptblob != NULL ? make_ed_blob (ptblob) : NULL;
   nblob = ntblob != NULL ? make_ed_blob (ntblob) : NULL;
   (*tess_matcher) (pblob, blob, nblob, tess_word, tess_denorm, ratings);

ccstruct/pageres.cpp

@@ -34,8 +34,8 @@ CLISTIZE (BLOCK_RES) ELISTIZE (ROW_RES) ELISTIZE (WERD_RES)
 PAGE_RES::PAGE_RES(                            //recursive construct
                    BLOCK_LIST *the_block_list  //real page
                   ) {
-  BLOCK_IT block_it(the_block_list); 
-  BLOCK_RES_IT block_res_it(&block_res_list); 
+  BLOCK_IT block_it(the_block_list);
+  BLOCK_RES_IT block_res_it(&block_res_list);
 
   char_count = 0;
   rej_count = 0;
@@ -58,7 +58,7 @@ BLOCK_RES::BLOCK_RES(                  //recursive construct
                      BLOCK *the_block  //real BLOCK
                     ) {
   ROW_IT row_it (the_block->row_list ());
-  ROW_RES_IT row_res_it(&row_res_list); 
+  ROW_RES_IT row_res_it(&row_res_list);
 
   char_count = 0;
   rej_count = 0;
@@ -87,7 +87,7 @@ ROW_RES::ROW_RES(              //recursive construct
                  ROW *the_row  //real ROW
                 ) {
   WERD_IT word_it (the_row->word_list ());
-  WERD_RES_IT word_res_it(&word_res_list); 
+  WERD_RES_IT word_res_it(&word_res_list);
   WERD_RES *combo = NULL;        //current combination of fuzzies
   WERD_RES *word_res;            //current word
   WERD *copy_word;
@@ -104,6 +104,7 @@ ROW_RES::ROW_RES(              //recursive construct
 
   for (word_it.mark_cycle_pt (); !word_it.cycled_list (); word_it.forward ()) {
     word_res = new WERD_RES (word_it.data ());
+    word_res->x_height = the_row->x_height();
 
     if (word_res->word->flag (W_FUZZY_NON)) {
       ASSERT_HOST (combo != NULL);
@@ -116,6 +117,7 @@ ROW_RES::ROW_RES(              //recursive construct
                                  //deep copy
         *copy_word = *(word_it.data ());
         combo = new WERD_RES (copy_word);
+        combo->x_height = the_row->x_height();
         combo->combination = TRUE;
         word_res_it.add_to_end (combo);
       }
@@ -207,7 +209,7 @@ WERD_RES::~WERD_RES () {
  * Set things up at the start of the page
  *************************************************************************/
 
-WERD_RES *PAGE_RES_IT::restart_page() { 
+WERD_RES *PAGE_RES_IT::restart_page() {
   block_res_it.set_to_list (&page_res->block_res_list);
   block_res_it.mark_cycle_pt ();
   block_res = NULL;
@@ -216,7 +218,7 @@ WERD_RES *PAGE_RES_IT::restart_page() {
   next_block_res = NULL;
   next_row_res = NULL;
   next_word_res = NULL;
-  internal_forward(TRUE); 
+  internal_forward(TRUE);
   return internal_forward (FALSE);
 }
 
@@ -230,7 +232,7 @@ WERD_RES *PAGE_RES_IT::restart_page() {
  * boundaries. NULL values denote start and end of the page.
  *************************************************************************/
 
-WERD_RES *PAGE_RES_IT::internal_forward(BOOL8 new_block) { 
+WERD_RES *PAGE_RES_IT::internal_forward(BOOL8 new_block) {
   BOOL8 found_next_word = FALSE;
   BOOL8 new_row = FALSE;
 
@@ -291,7 +293,7 @@ WERD_RES *PAGE_RES_IT::internal_forward(BOOL8 new_block) {
  * the block, the prev block, row and word are all NULL.
  *************************************************************************/
 
-WERD_RES *PAGE_RES_IT::forward_block() { 
+WERD_RES *PAGE_RES_IT::forward_block() {
   if (block_res == next_block_res) {
     block_res_it.forward ();;
     block_res = NULL;
@@ -300,13 +302,13 @@ WERD_RES *PAGE_RES_IT::forward_block() {
     next_block_res = NULL;
     next_row_res = NULL;
     next_word_res = NULL;
-    internal_forward(TRUE); 
+    internal_forward(TRUE);
   }
   return internal_forward (FALSE);
 }
 
 
-void PAGE_RES_IT::rej_stat_word() { 
+void PAGE_RES_IT::rej_stat_word() {
   INT16 chars_in_word;
   INT16 rejects_in_word = 0;
 

wordrec/chopper.cpp

@@ -434,6 +434,8 @@ CHOICES_LIST chop_word_main(register TWERD *word,
       (CHOICES) classify_blob (pblob, blob, blob->next, NULL, fx,
       "chop_word:", Green, &chop_states[0],
       &best_state, matcher_pass, index);
+    if (match_result == NULL)
+      cprintf("Null classifier output!\n");
     char_choices = array_push (char_choices, match_result);
     pblob = blob;
   }