Handle null raw_choice - fixes #235, fixes #246

Author: tfmorris

src/ccmain/control.cpp

@@ -1300,14 +1300,25 @@ float Tesseract::ClassifyBlobAsWord(int pass_n, PAGE_RES_IT* pr_it,
   SetupWordPassN(1, &wd);
   classify_word_and_language(pass_n, &it, &wd);
   if (debug_noise_removal) {
-    tprintf("word xheight=%g, row=%g, range=[%g,%g]\n", word_res->x_height,
-            wd.row->x_height(), wd.word->raw_choice->min_x_height(),
-            wd.word->raw_choice->max_x_height());
-  }
-  const float cert = wd.word->raw_choice->certainty();
-  const float rat = wd.word->raw_choice->rating();
-  *c2 = rat > 0.0f ? cert * cert / rat : 0.0f;
-  *best_str = wd.word->raw_choice->unichar_string();
+    if (wd.word->raw_choice != NULL) {
+      tprintf("word xheight=%g, row=%g, range=[%g,%g]\n", word_res->x_height,
+              wd.row->x_height(), wd.word->raw_choice->min_x_height(),
+              wd.word->raw_choice->max_x_height());
+    } else {
+      tprintf("Got word with null raw choice xheight=%g, row=%g\n", word_res->x_height,
+              wd.row->x_height());
+    }
+  }
+  float cert = 0.0f;
+  if (wd.word->raw_choice != NULL) {  // This probably shouldn't happen, but...
+    cert = wd.word->raw_choice->certainty();
+    float rat = wd.word->raw_choice->rating();
+    *c2 = rat > 0.0f ? cert * cert / rat : 0.0f;
+    *best_str = wd.word->raw_choice->unichar_string();
+  } else {
+    *c2 = 0.0f;
+    *best_str = "";
+  }
   it.DeleteCurrentWord();
   pr_it->ResetWordIterator();
   return cert;

src/training/stringrenderer.cpp

@@ -543,7 +543,11 @@ void StringRenderer::ComputeClusterBoxes() {
   // pango.
   std::vector<std::string> cluster_text;
   if (GetClusterStrings(&cluster_text)) {
-    ASSERT_HOST(cluster_text.size() == start_byte_to_box.size());
+    tprintf("* %d, cluster_text.size(): %d\t", page_, cluster_text.size());
+    tprintf("start_byte_to_box.size(): %d\n", start_byte_to_box.size());
+    if (cluster_text.size() != start_byte_to_box.size())
+      tprintf(">%s<\n", cluster_text[0].c_str());
+	ASSERT_HOST(cluster_text.size() == start_byte_to_box.size());
     int ind = 0;
     for (std::map<int, BoxChar*>::iterator it = start_byte_to_box.begin();
          it != start_byte_to_box.end(); ++it, ++ind) {