Update Session

translate_epub.py

@@ -32,6 +32,12 @@ MAX_CHUNK = 4000
 DELAY = 0.3
 SENTENCE_RE = re.compile(r'(?<=[.!?])\s+')
 
+# Patterns to protect from translation
+ROMAN_NUMERAL = re.compile(r'\b([IVXLCDM]+)\b', re.IGNORECASE)
+CHAPTER_PREFIX = re.compile(r'(Chapter\s+)([IVXLCDM]+|\d+)', re.IGNORECASE)
+ROMAN_STANDALONE = re.compile(r'^\s*([IVXLCDM]+)[.\s]*$', re.IGNORECASE)
+PLACEHOLDER_RE = re.compile(r'xxKEEP(\d+)xx')
+
 BASE_DIR = os.path.dirname(os.path.abspath(__file__))
 SOURCE_DIR = os.path.join(BASE_DIR, 'original')
 TRANSLATED_DIR = os.path.join(BASE_DIR, 'translated')
@@ -41,11 +47,38 @@ def log(msg):
     print(msg, flush=True)
 
 
+def protect_numerals(text):
+    """Replace chapter numbers / Roman numerals with placeholders."""
+    protected = []
+    def _store(val):
+        protected.append(val)
+        return f'xxKEEP{len(protected)-1}xx'
+
+    # Protect "Chapter X" - only replace the numeral, keep "Chapter" for translation
+    text = CHAPTER_PREFIX.sub(lambda m: m.group(1) + _store(m.group(2)), text)
+    # Protect standalone Roman numerals (chapter headings)
+    text = ROMAN_STANDALONE.sub(lambda m: _store(m.group(1)), text)
+    # Protect remaining Roman numerals in text
+    text = ROMAN_NUMERAL.sub(lambda m: _store(m.group(1)), text)
+    return text, protected
+
+
+def restore_numerals(text, protected):
+    """Restore original numerals from placeholders."""
+    def _restore(m):
+        idx = int(m.group(1))
+        return protected[idx] if idx < len(protected) else m.group(0)
+    return PLACEHOLDER_RE.sub(_restore, text)
+
+
 def translate_text(text):
-    """Translate using deep_translator."""
+    """Translate using deep_translator, preserving chapter numbers."""
+    clean_text, protected = protect_numerals(text)
     try:
         translator = GoogleTranslator(source='en', target='id')
-        result = translator.translate(text)
+        result = translator.translate(clean_text)
+        if result and protected:
+            result = restore_numerals(result, protected)
         return result if result else None
     except Exception:
         return None