Some LGTM fixes and potential bugfixes

Author: bharatr21

src/ccutil/unicharmap.cpp

@@ -39,7 +39,7 @@ UNICHAR_ID UNICHARMAP::unichar_to_id(const char* const unichar_repr,
   assert(length > 0 && length <= UNICHAR_LEN);
 
   int index = 0;
-  if (index >= length || unichar_repr[index] == '\0') return INVALID_UNICHAR_ID;
+  if (unichar_repr[index] == '\0') return INVALID_UNICHAR_ID;
   do {
     if (index + 1 >= length || unichar_repr[index + 1] == '\0')
       return current_nodes[static_cast<unsigned char>(unichar_repr[index])].id;

src/classify/trainingsampleset.cpp

@@ -618,6 +618,7 @@ void TrainingSampleSet::ComputeCanonicalSamples(const IntFeatureMap& map,
           if (dist > max_dist) {
             max_dist = dist;
             if (dist > max_max_dist) {
+              max_max_dist = dist;
               max_s1 = s1;
               max_s2 = s2;
             }

src/training/tesstrain_utils.py

@@ -56,6 +56,16 @@ def __init__(self):
         self.extract_font_properties = True
         self.distort_image = False
 
+    def __eq__(self, other):
+        return (argparse.Namespace.__eq__(self, other) and
+        self.uname = other.uname and self.lang_code = other.lang_code and
+        self.timestamp = other.timestamp and self.font_config_cache = other.font_config_cache and
+        self.fonts_dir = other.fonts_dir and self.max_pages = other.max_pages and
+        self.save_box_tiff = other.save_box_tiff and self.overwrite = other.overwrite and
+        self.linedata = other.linedata and self.run_shape_clustering = other.run_shape_clustering and
+        self.extract_font_properties = other.extract_font_properties and
+        self.distort_image = other.distort_image)
+
 
 def err_exit(msg):
     log.critical(msg)
@@ -356,7 +366,7 @@ def generate_font_image(ctx, font, exposure, char_spacing):
 
 
 # Phase I : Generate (I)mages from training text for each font.
-def phase_I_generate_image(ctx, par_factor):
+def phase_I_generate_image(ctx, par_factor=None):
     if not par_factor or par_factor <= 0:
         par_factor = 1
 
@@ -386,7 +396,7 @@ def phase_I_generate_image(ctx, par_factor):
 
         with tqdm(
                 total=len(ctx.fonts)
-        ) as pbar, concurrent.futures.ThreadPoolExecutor(max_workers=8) as executor:
+        ) as pbar, concurrent.futures.ThreadPoolExecutor(max_workers=par_factor) as executor:
             futures = [
                 executor.submit(generate_font_image, ctx, font, exposure, char_spacing)
                 for font in ctx.fonts