test_hindi_stt.py

"""
test_hindi_stt.py — Test Hindi speech-to-text support
"""
import sys
from pathlib import Path
SCRIPT_DIR = Path(__file__).parent
sys.path.insert(0, str(SCRIPT_DIR))

from utils.logger import logger
import inspect

def test_hindi_stt_fixed():
    """Test that Hindi STT support is now fixed"""
    logger.info("=" * 70)
    logger.info("TESTING: Hindi Speech-to-Text Support")
    logger.info("=" * 70)

    # Test 1: Check transcriber1.py has Hindi support
    logger.info("\n1. Checking transcriber1.py for Hindi support parameters...")
    try:
        from speech_module.transcriber1 import SpeechTranscriber, WhisperTranscriber

        # Check SpeechTranscriber.transcribe signature
        sig = inspect.signature(SpeechTranscriber.transcribe)
        params = list(sig.parameters.keys())

        logger.info(f"   SpeechTranscriber.transcribe() parameters: {params}")

        if 'language' in params and 'task' in params:
            logger.info("   [PASS] FIXED: language and task parameters present")
        else:
            logger.error("   [FAIL] FAILED: language or task parameters missing")
            return False

        # Check WhisperTranscriber.transcribe signature
        sig_whisper = inspect.signature(WhisperTranscriber.transcribe)
        params_whisper = list(sig_whisper.parameters.keys())

        logger.info(f"   WhisperTranscriber.transcribe() parameters: {params_whisper}")

        if 'language' in params_whisper and 'task' in params_whisper:
            logger.info("   [PASS] FIXED: WhisperTranscriber has Hindi support")
        else:
            logger.error("   [FAIL] FAILED: WhisperTranscriber missing parameters")
            return False

    except Exception as e:
        logger.error(f"   [FAIL] FAILED: {e}")
        return False

    # Test 2: Check __init__.py imports from transcriber1
    logger.info("\n2. Checking speech_module/__init__.py imports...")
    try:
        init_path = SCRIPT_DIR / "speech_module" / "__init__.py"
        with open(init_path, "r") as f:
            init_content = f.read()

        if "transcriber1" in init_content:
            logger.info("   [PASS] __init__.py imports from transcriber1.py")
        else:
            logger.error("   [FAIL] __init__.py does not import from transcriber1.py")
            return False

    except Exception as e:
        logger.error(f"   [FAIL] FAILED: {e}")
        return False

    # Test 3: Check app.py has language selection (Hindi audio tab)
    logger.info("\n3. Checking app.py for Hindi language support...")
    try:
        app_path = SCRIPT_DIR / "app.py"
        with open(app_path, "r", encoding="utf-8") as f:
            app_content = f.read()

        checks = {
            "transcribe_audio function has language parameter": 'language: str = None' in app_content,
            "analyze_hindi_audio function exists": 'def analyze_hindi_audio(audio_path):' in app_content,
            "Hindi audio tab exists": 'Hindi audio' in app_content,
            "task=translate for Hindi": 'task="translate"' in app_content,
            "language=hi for Hindi": 'language="hi"' in app_content,
            "analyze_english_audio function exists": 'def analyze_english_audio(audio_path):' in app_content,
        }

        all_passed = True
        for check_name, result in checks.items():
            status = "[PASS]" if result else "[FAIL]"
            logger.info(f"   {status} {check_name}")
            if not result:
                all_passed = False

        if not all_passed:
            return False

    except Exception as e:
        logger.error(f"   [FAIL] FAILED: {e}")
        return False

    # Test 4: Import and verify the updated modules work
    logger.info("\n4. Testing import and initialization...")
    try:
        from speech_module import SpeechTranscriber
        logger.info("   [PASS] SpeechTranscriber imported successfully")

        # Try to instantiate
        transcriber = SpeechTranscriber()
        logger.info("   [PASS] SpeechTranscriber instantiated successfully")

        # Check method exists and has right signature
        method = getattr(transcriber, 'transcribe')
        sig = inspect.signature(method)
        if 'language' in sig.parameters:
            logger.info("   [PASS] transcribe method accepts language parameter")
        else:
            logger.error("   [FAIL] transcribe method missing language parameter")
            return False

    except Exception as e:
        logger.error(f"   [FAIL] FAILED: {e}")
        import traceback
        traceback.print_exc()
        return False

    # Test 5: Verify anti-hallucination parameters in transcriber1.py
    logger.info("\n5. Checking anti-hallucination decode parameters...")
    try:
        t1_path = SCRIPT_DIR / "speech_module" / "transcriber1.py"
        with open(t1_path, "r", encoding="utf-8") as f:
            t1_content = f.read()

        checks = {
            "temperature=0.0 set": '"temperature": 0.0' in t1_content,
            "condition_on_previous_text=False": '"condition_on_previous_text": False' in t1_content,
            "initial_prompt set": '"initial_prompt"' in t1_content,
            "without_timestamps=True": '"without_timestamps": True' in t1_content,
            "suppress_tokens set": '"suppress_tokens"' in t1_content,
        }

        all_passed = True
        for check_name, result in checks.items():
            status = "[PASS]" if result else "[FAIL]"
            logger.info(f"   {status} {check_name}")
            if not result:
                all_passed = False

        if not all_passed:
            return False

    except Exception as e:
        logger.error(f"   [FAIL] FAILED: {e}")
        return False

    return True

if __name__ == "__main__":
    logger.info("\n[HINDI STT CONFIGURATION TEST]")
    logger.info("This test verifies that Hindi speech-to-text support is properly configured.")

    success = test_hindi_stt_fixed()

    logger.info("\n" + "=" * 70)
    if success:
        logger.info("[PASS] ALL TESTS PASSED - Hindi STT support is now fully configured!")
        logger.info("\nYou can now:")
        logger.info("  1. Upload/record Hindi audio")
        logger.info("  2. Select 'Hindi audio' tab")
        logger.info("  3. Click 'Transcribe Hindi & analyze'")
        logger.info("  4. Whisper will transcribe and translate to English automatically")
    else:
        logger.error("[FAIL] SOME TESTS FAILED - Please review the errors above")
        sys.exit(1)
    logger.info("=" * 70)