Added all

.gitignore

@@ -23,6 +23,7 @@ Icon
 *.jpg
 *.jpeg
 *.json
+*.webp
 
 # Files that might appear in the root of a volume
 .DocumentRevisions-V100

manga-translator.py

@@ -302,15 +302,10 @@ def split_bubble_if_multiple_columns(indices, ocr, bid=None, use_aggressive_thre
 
 
 def split_bubble_if_multiple_rows(indices, ocr, bid=None):
-    """
-    Splits a bubble horizontally (top/bottom) if there is a massive vertical gap
-    between text lines, indicating two separate bubbles were merged.
-    """
     if len(indices) < 2:
         return None
         
     boxes = [quad_bbox(ocr[i][0]) for i in indices]
-    # Sort by Y-coordinate (top to bottom)
     sorted_items = sorted(zip(indices, boxes), key=lambda x: x[1][1])
     
     gaps = []
@@ -326,23 +321,19 @@ def split_bubble_if_multiple_rows(indices, ocr, bid=None):
     if not gaps:
         return None
     
-    # Find the largest vertical gap
     max_gap_idx, max_gap_size, _, _ = max(gaps, key=lambda x: x[1])
     
-    # Calculate median line height to define what a "large" gap is
     hs = [b[3] - b[1] for b in boxes]
     med_h = float(np.median(hs)) if hs else 15.0
     
-    # If the vertical gap is more than 2.5x the height of a text line, it's a split!
     threshold = med_h * 2.5
-    min_gap = 40.0 # Absolute minimum pixel gap to prevent micro-splits
+    min_gap = 40.0
     
     if max_gap_size > threshold and max_gap_size > min_gap:
         split_idx = max_gap_idx
         top_indices = [item[0] for item in sorted_items[:split_idx]]
         bottom_indices = [item[0] for item in sorted_items[split_idx:]]
         
-        # Ensure we don't just split off a single noise character
         if len(top_indices) >= 1 and len(bottom_indices) >= 1:
             return top_indices, bottom_indices
     
@@ -354,10 +345,26 @@ def split_bubble_if_multiple_rows(indices, ocr, bid=None):
 # ============================================================
 class MacVisionDetector:
     def __init__(self, source_lang="en"):
-        lang_map = {"en": "en-US", "es": "es-ES", "ca": "ca-ES", "fr": "fr-FR", "ja": "ja-JP"}
-        apple_lang = lang_map.get(source_lang, "en-US")
+        # 1. Normalize the input language string
+        lang_key = source_lang.lower().strip()
+        
+        # 2. Comprehensive mapping to Apple Vision BCP-47 language codes
+        lang_map = {
+            "en": "en-US", "english": "en-US",
+            "es": "es-ES", "spanish": "es-ES",
+            "ca": "ca-ES", "catalan": "ca-ES",
+            "fr": "fr-FR", "french": "fr-FR",
+            "ja": "ja-JP", "japanese": "ja-JP",
+            "it": "it-IT", "italian": "it-IT",
+            "de": "de-DE", "german": "de-DE",
+            "ko": "ko-KR", "korean": "ko-KR",
+            "zh": "ko-KR", "chinese": "zh-Hans" # Simplified Chinese
+        }
+        
+        # 3. Resolve the language code
+        apple_lang = lang_map.get(lang_key, "en-US")
         self.langs = [apple_lang]
-        print(f"⚡ Using Apple Vision OCR (Language: {self.langs})")
+        print(f"⚡ Using Apple Vision OCR (Language: {self.langs[0]})")
 
     def read(self, image_path_or_array):
         if isinstance(image_path_or_array, str):
@@ -1108,8 +1115,8 @@ def translate_manga_text(
 
 if __name__ == "__main__":
     translate_manga_text(
-        image_path="004.png",
-        source_lang="en",
+        image_path="003.jpg",
+        source_lang="es",
         target_lang="ca",
         confidence_threshold=0.05,
         min_text_length=1,
@@ -1118,6 +1125,6 @@ if __name__ == "__main__":
         quality_threshold=0.62,
         export_to_file="output.txt",
         export_bubbles_to="bubbles.json",
-        reading_mode="ltr",
+        reading_mode="rtl", # Changed to RTL for Japanese Manga
         debug=True
-    )
+    )