Update app.py

app.py

@@ -293,81 +293,48 @@ async def transcribe_voice_with_openai(file_path: str) -> str:
         system_prompt = """
 You are transcribing voice messages for Apex Biotical Veterinary WhatsApp Assistant. This is a professional veterinary products chatbot.
 
-CRITICAL LANGUAGE RESTRICTION: ONLY ENGLISH OR URDU - NOTHING ELSE
-
-CONTEXT: Users can speak product names, menu selections, numbers, and general queries in English or Urdu ONLY.
-
-STRICT LANGUAGE RULES: 
-- ONLY transcribe English (en) or Urdu (ur) speech
-- REJECT all other languages completely
-- If you hear any language other than English or Urdu, transcribe as English
-- Never transcribe German, French, Spanish, Italian, Portuguese, Russian, Chinese, Japanese, Korean, or any other language
-- Always default to English if uncertain
-- This is a veterinary assistant - users will speak in English or Urdu only
-
-PRODUCT NAMES (Veterinary Products):
-- Hydropex (electrolyte supplement)
-- Respira Aid Plus (respiratory support)
-- Heposel (liver tonic)
-- Bromacid (respiratory/mucolytic)
-- Hexatox (liver & kidney support)
-- APMA Fort (mycotoxin binder)
-- Para C.E (heat stress support)
-- Tribiotic (antibiotic)
-- PHYTO-SAL (phytogenic supplement)
-- Mycopex Super (mycotoxin binder)
-- Eflin KT-20 (antibiotic)
-- Salcozine ST-30 (anticoccidial)
-- Oftilex UA-10 (antibiotic)
-- Biscomin 10 (injectable antibiotic)
-- Apvita Plus (vitamin supplement)
-- B-G Aspro-C (aspirin + vitamin C)
-- EC-Immune (immune booster)
-- Liverpex (liver tonic)
-- Symodex (multivitamin)
-- Respira Aid (respiratory support)
-- Adek Gold (multivitamin)
-- Immuno DX (immune enhancer)
-
-MENU SELECTIONS:
-- Main menu options: 1, 2, 3, 4
-- Product numbers: 1-23
-- Category numbers: 1-10
-- Navigation: main, menu, back, home, start
+CRITICAL: TRANSCRIBE ONLY ENGLISH OR URDU SPEECH - NOTHING ELSE
 
-NUMBERS (English & Urdu):
-English: one, two, three, four, five, six, seven, eight, nine, ten, eleven, twelve, thirteen, fourteen, fifteen, sixteen, seventeen, eighteen, nineteen, twenty, twenty-one, twenty-two, twenty-three
-Urdu (Roman): aik, ek, do, teen, char, panch, che, sat, ath, nau, das, gyara, bara, tera, choda, pandra, sola, satara, athara, unnees, bees, ikkees, baees, tees
-Urdu (Script): ایک, دو, تین, چار, پانچ, چھ, سات, آٹھ, نو, دس, گیارہ, بارہ, تیرہ, چودہ, پندرہ, سولہ, سترہ, اٹھارہ, انیس, بیس, اکیس, بائیس, تئیس
+IMPORTANT RULES:
+1. ONLY transcribe English or Urdu speech
+2. If you hear unclear audio, transcribe as English
+3. If you hear mixed languages, transcribe as English
+4. Never transcribe gibberish or random characters
+5. If audio is unclear, transcribe as "unclear audio"
+6. Keep transcriptions simple and clean
 
-COMMON GREETINGS:
-English: hi, hello, hey, good morning, good afternoon, good evening, how are you
-Urdu: salam, assalamu alaikum, adaab, namaste, khuda hafiz
+PRODUCT NAMES (exact spelling required):
+- Hydropex, Respira Aid Plus, Heposel, Bromacid, Hexatox
+- APMA Fort, Para C.E, Tribiotic, PHYTO-SAL, Mycopex Super
+- Eflin KT-20, Salcozine ST-30, Oftilex UA-10, Biscomin 10
+- Apvita Plus, B-G Aspro-C, EC-Immune, Liverpex, Symodex
+- Respira Aid, Adek Gold, Immuno DX
 
 MENU COMMANDS:
-English: search, browse, download, catalog, contact, availability, main menu, option, number, choice
-Urdu: تلاش, براؤز, ڈاؤن لوڈ, کیٹلاگ, رابطہ, دستیابی, مین مینو, آپشن, نمبر, اختیار
+- Numbers: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
+- Navigation: main, menu, back, home, start
+- Options: option, number, choice, select
+
+GREETINGS:
+- English: hi, hello, hey, good morning, good afternoon, good evening
+- Urdu: salam, assalamu alaikum, adaab, namaste, khuda hafiz
 
 TRANSCRIPTION RULES:
-1. ONLY transcribe English or Urdu speech
-2. Transcribe product names exactly as listed above
-3. Convert spoken numbers to digits (1, 2, 3, etc.)
-4. Handle both English and Urdu speech
-5. Preserve exact spelling for product names
-6. Convert menu selections to numbers
-7. Handle common transcription errors (opium->option, numara->number)
-8. Maintain context for veterinary domain
-9. If unsure about language, default to English
+1. Transcribe exactly what you hear in English or Urdu
+2. Convert numbers to digits (one->1, two->2, etc.)
+3. Preserve product names exactly
+4. If unclear, transcribe as "unclear audio"
+5. Keep it simple and clean
+6. No random characters or mixed languages
 
 EXAMPLES:
 - "hydropex" -> "hydropex"
 - "respira aid plus" -> "respira aid plus"
-- "option number one" -> "1"
-- "aik" -> "1"
-- "do" -> "2"
+- "option one" -> "1"
 - "main menu" -> "main"
 - "salam" -> "salam"
 - "search products" -> "search products"
+- Unclear audio -> "unclear audio"
 """
         
         # First attempt with comprehensive system prompt
@@ -474,10 +441,21 @@ TRANSCRIPTION RULES:
                 transcribed_text = transcript.text.strip()
                 logger.info(f"[Transcribe] Third attempt (mixed) transcribed: '{transcribed_text}'")
         
-        # Final check for empty transcription
+        # Final check for empty transcription or unclear audio
         if not transcribed_text or len(transcribed_text.strip()) < 2:
             logger.warning(f"[Transcribe] Very short or empty transcription: '{transcribed_text}'")
-            return ""
+            return "unclear audio"
+        
+        # Check for gibberish or mixed characters
+        if len(transcribed_text) > 10 and not re.search(r'[a-zA-Z\u0600-\u06FF]', transcribed_text):
+            logger.warning(f"[Transcribe] Gibberish detected: '{transcribed_text}'")
+            return "unclear audio"
+        
+        # Check for too many special characters
+        special_char_ratio = len(re.findall(r'[^\w\s]', transcribed_text)) / len(transcribed_text)
+        if special_char_ratio > 0.3:
+            logger.warning(f"[Transcribe] Too many special characters: '{transcribed_text}'")
+            return "unclear audio"
         
         return transcribed_text
         
@@ -2599,8 +2577,8 @@ Response:
         if reply_language == 'ur':
             try:
                 # Get all product and category names
-                product_names = [p.get('Product Name', '') for p in all_products if p.get('Product Name')]
-                category_names = list(set([p.get('Category', '') for p in all_products if p.get('Category')]))
+                product_names = [str(p.get('Product Name', '')) for p in all_products if p.get('Product Name')]
+                category_names = list(set([str(p.get('Category', '')) for p in all_products if p.get('Category')]))
                 translated_response = GoogleTranslator(source='auto', target='ur').translate(ai_response)
                 # Restore English terms
                 translated_response = restore_english_terms(translated_response, ai_response, product_names, category_names)
@@ -3431,21 +3409,23 @@ async def handle_voice_message_complete(from_number: str, msg: dict):
         except:
             pass
         
-        # Handle empty or failed transcription
-        if not transcribed_text or transcribed_text.strip() == "":
-            logger.warning(f"[Voice] Empty transcription for {from_number}")
+        # Handle empty, failed, or unclear transcription
+        if not transcribed_text or transcribed_text.strip() == "" or transcribed_text.lower() == "unclear audio":
+            logger.warning(f"[Voice] Empty or unclear transcription for {from_number}: '{transcribed_text}'")
             send_whatsjet_message(from_number, 
                 "🎤 *Voice Message Issue*\n\n"
-                "I couldn't hear anything in your voice message. This can happen due to:\n"
+                "I couldn't understand your voice message clearly. This can happen due to:\n"
                 "• Very short voice note\n"
                 "• Background noise\n"
                 "• Microphone too far away\n"
-                "• Audio quality issues\n\n"
+                "• Audio quality issues\n"
+                "• Speaking too fast\n\n"
                 "💡 *Tips for better voice notes:*\n"
                 "• Speak clearly and slowly\n"
                 "• Keep phone close to mouth\n"
                 "• Record in quiet environment\n"
-                "• Make voice note at least 2-3 seconds\n\n"
+                "• Make voice note at least 2-3 seconds\n"
+                "• Speak in English or Urdu only\n\n"
                 "💬 *You can also:*\n"
                 "• Send a text message\n"
                 "• Type 'main' to see menu options\n"