Update app.py

app.py

@@ -70,29 +70,91 @@ def contains_sinhala_roman(text):
             return True
     return False
 
+def contains_tamil_roman(text):
+    """Check if text contains Roman Tamil patterns"""
+    # Common Roman Tamil patterns
+    tamil_roman_patterns = [
+        r'\b[a-z]+[aeiou][a-z]*\b',  # Basic Tamil roman patterns
+        r'\b(amma|appa|akka|anna|thambi|thangai|paapa|amma|appa|akka|anna|thambi|thangai|paapa)\b',  # Common Tamil words
+        r'\b(naan|neenga|avan|aval|adhu|idhu|edhu|yaaru|eppadi|enna|yaen|kaalam|vaaram|maasam|varusham)\b',  # Tamil pronouns/words
+    ]
+    
+    for pattern in tamil_roman_patterns:
+        if re.search(pattern, text.lower()):
+            return True
+    return False
+
 def transliterate_sinhala_roman_to_sinhala(text):
-    """Use Gemini to convert Roman Sinhala to Sinhala script"""
+    """Use Gemini to convert Roman Sinhala to Sinhala script with enhanced context"""
     if not gemini_api_key or not contains_sinhala_roman(text):
         return text
     
     try:
-        prompt = f"""Convert this Roman Sinhala text to Sinhala script. Only convert if it's actually Sinhala words in Roman script. If it's English or other language, return as is.
+        prompt = f"""You are a language expert specializing in Sri Lankan languages. Convert this Roman Sinhala text (Sinhala words written in English letters) to proper Sinhala script.
+
+IMPORTANT CONTEXT:
+- This is for a Sri Lankan budget proposals search system
+- The user is likely searching for government policies, economic proposals, or budget information
+- Use formal Sinhala appropriate for policy discussions
+- Only convert if it's actually Sinhala words in Roman script
+- If it's English or other language, return as is
+- Be accurate with Sri Lankan Sinhala terminology
 
-Text: "{text}"
+Text to convert: "{text}"
 
-Sinhala script:"""
+Converted Sinhala script:"""
         
         response = gemini_model.generate_content(prompt)
         result = response.text.strip()
         
-        # Clean up the response
+        # Clean up the response - remove any extra text that might be added
         if result and len(result) > 0:
+            # Remove common prefixes that Gemini might add
+            result = result.replace("Converted Sinhala script:", "").strip()
+            result = result.replace("Sinhala script:", "").strip()
             return result
         else:
             return text
             
     except Exception as e:
-        logger.warning(f"Transliteration failed: {e}")
+        logger.warning(f"Sinhala transliteration failed: {e}")
+        return text
+
+def transliterate_tamil_roman_to_tamil(text):
+    """Use Gemini to convert Roman Tamil to Tamil script with enhanced context"""
+    if not gemini_api_key or not contains_tamil_roman(text):
+        return text
+    
+    try:
+        prompt = f"""You are a language expert specializing in Sri Lankan languages. Convert this Roman Tamil text (Tamil words written in English letters) to proper Tamil script.
+
+IMPORTANT CONTEXT:
+- This is for a Sri Lankan budget proposals search system
+- The user is likely searching for government policies, economic proposals, or budget information
+- Use formal Tamil appropriate for policy discussions
+- Use Sri Lankan Tamil dialect and terminology
+- Only convert if it's actually Tamil words in Roman script
+- If it's English or other language, return as is
+- Be accurate with Sri Lankan Tamil terminology and context
+
+Text to convert: "{text}"
+
+Converted Tamil script:"""
+        
+        response = gemini_model.generate_content(prompt)
+        result = response.text.strip()
+        
+        # Clean up the response - remove any extra text that might be added
+        if result and len(result) > 0:
+            # Remove common prefixes that Gemini might add
+            result = result.replace("Converted Tamil script:", "").strip()
+            result = result.replace("Tamil script:", "").strip()
+            return result
+        else:
+            return text
+            
+    except Exception as e:
+        logger.warning(f"Tamil transliteration failed: {e}")
         return text
 
 def preprocess_query(query, language):
@@ -102,6 +164,11 @@ def preprocess_query(query, language):
         transliterated = transliterate_sinhala_roman_to_sinhala(query)
         logger.info(f"Transliterated to: {transliterated}")
         return transliterated
+    elif language == 'ta' and contains_tamil_roman(query):
+        logger.info(f"Transliterating Roman Tamil: {query}")
+        transliterated = transliterate_tamil_roman_to_tamil(query)
+        logger.info(f"Transliterated to: {transliterated}")
+        return transliterated
     return query
 
 # Load dynamic metadata