bugfixes for Hindi text generation

README.md

@@ -173,6 +173,15 @@ Output
 ![29](samples/29.jpg "2")
 
 
+## Create images with Hindi text 
+
+It is simple! Just do `trdg -l hi -c 1000 -w 5`!
+
+Output 
+
+![30](samples/30.jpg)
+
+
 ## Add new fonts
 
 The script picks a font at random from the *fonts* directory.
@@ -184,6 +193,7 @@ The script picks a font at random from the *fonts* directory.
 | fonts/ko | Korean |
 | fonts/ja | Japanese |
 | fonts/th | Thai |
+| fonts/hi | Hindi |
 
 Simply add/remove fonts until you get the desired output.
 

trdg/data_generator.py

@@ -268,23 +268,27 @@ def generate(
         tess_box_name = "{}.box".format(name)
 
         # Save the image
-        if out_dir is not None:
-            final_image.save(os.path.join(out_dir, image_name))
-            if output_mask == 1:
-                final_mask.save(os.path.join(out_dir, mask_name))
-            if output_bboxes == 1:
-                bboxes = mask_to_bboxes(final_mask)
-                with open(os.path.join(out_dir, box_name), "w") as f:
-                    for bbox in bboxes:
-                        f.write(" ".join([str(v) for v in bbox]) + "\n")
-            if output_bboxes == 2:
-                bboxes = mask_to_bboxes(final_mask, tess=True)
-                with open(os.path.join(out_dir, tess_box_name), "w") as f:
-                    for bbox, char in zip(bboxes, text):
-                        f.write(
-                            " ".join([char] + [str(v) for v in bbox] + ["0"]) + "\n"
-                        )
-        else:
-            if output_mask == 1:
-                return final_image, final_mask
-            return final_image
+        try:
+            if out_dir is not None:
+                final_image.save(os.path.join(out_dir, image_name))
+                if output_mask == 1:
+                    final_mask.save(os.path.join(out_dir, mask_name))
+                if output_bboxes == 1:
+                    bboxes = mask_to_bboxes(final_mask)
+                    with open(os.path.join(out_dir, box_name), "w") as f:
+                        for bbox in bboxes:
+                            f.write(" ".join([str(v) for v in bbox]) + "\n")
+                if output_bboxes == 2:
+                    bboxes = mask_to_bboxes(final_mask, tess=True)
+                    with open(os.path.join(out_dir, tess_box_name), "w") as f:
+                        for bbox, char in zip(bboxes, text):
+                            f.write(
+                                " ".join([char] + [str(v) for v in bbox] + ["0"]) + "\n"
+                            )
+            else:
+                if output_mask == 1:
+                    return final_image, final_mask
+                return final_image
+
+        except Exception as e:
+            print("Error while saving the image: {}".format(e))

trdg/utils.py

@@ -111,26 +111,14 @@ def draw_bounding_boxes(
 
 
 def make_filename_valid(value: str, allow_unicode: bool = False) -> str:
-    """
-    Code adapted from: https://docs.djangoproject.com/en/4.0/_modules/django/utils/text/#slugify
 
-    Convert to ASCII if 'allow_unicode' is False. Convert spaces or repeated
-    dashes to single dashes. Remove characters that aren't alphanumerics,
-    underscores, or hyphens. Convert to lowercase. Also strip leading and
-    trailing whitespace, dashes, and underscores.
-    """
-    value = str(value)
-    if allow_unicode:
-        value = unicodedata.normalize("NFKC", value)
-    else:
-        value = (
-            unicodedata.normalize("NFKD", value)
-            .encode("ascii", "ignore")
-            .decode("ascii")
-        )
-    value = re.sub(r"[^\w\s-]", "", value)
-
-    # Image names will be shortened to avoid exceeding the max filename length
+    #remove spaces if in last character only
+    value = re.sub(r'\s+$', '', value)
+
+    #replacing invalid characters with underscores
+    value = re.sub(r'[:?<>|*$`]', '', value)
+
+    # Truncate the resulting string to a maximum length of 200 characters
     return value[:200]