compare_cunit.py

#!/usr/bin/env python3
"""
Script to compare original transcription with c-unit segmented transcription
"""
import json
import os


def compare_transcriptions(session_id):
    """Compare original and c-unit transcriptions"""
    
    session_dir = f"session_data/{session_id}"
    original_file = os.path.join(session_dir, "transcription.json")
    cunit_file = os.path.join(session_dir, "transcription_cunit.json")
    
    # Check if files exist
    if not os.path.exists(original_file):
        print(f"Original transcription file not found: {original_file}")
        return
    
    if not os.path.exists(cunit_file):
        print(f"C-unit transcription file not found: {cunit_file}")
        return
    
    # Load original transcription
    with open(original_file, 'r', encoding='utf-8') as f:
        data = json.load(f)
        # Handle both old and new format
        if "segments" in data:
            original_data = data["segments"]
        else:
            original_data = data
    
    # Load c-unit transcription
    with open(cunit_file, 'r', encoding='utf-8') as f:
        data = json.load(f)
        # Handle both old and new format
        if "segments" in data:
            cunit_data = data["segments"]
        else:
            cunit_data = data
    
    print(f"Session ID: {session_id}")
    print("=" * 80)
    print(f"Original utterances: {len(original_data)}")
    print(f"C-unit segments: {len(cunit_data)}")
    print("=" * 80)
    
    print("\nORIGINAL TRANSCRIPTION:")
    print("-" * 40)
    for i, utterance in enumerate(original_data):
        print(f"Utterance {i+1}: {utterance['text']}")
    
    print("\nC-UNIT SEGMENTED TRANSCRIPTION:")
    print("-" * 40)
    for i, cunit in enumerate(cunit_data):
        print(f"C-unit {i+1}: {cunit['text']}")
    
    print("\nSIDE-BY-SIDE COMPARISON:")
    print("-" * 40)
    
    # Reconstruct original text for comparison
    original_full_text = " ".join([utterance['text'] for utterance in original_data])
    cunit_full_text = " ".join([cunit['text'] for cunit in cunit_data])
    
    print(f"Original:  {original_full_text}")
    print(f"C-units:   {cunit_full_text}")
    
    print(f"\nSegmentation ratio: {len(cunit_data) / len(original_data):.2f} c-units per utterance")


if __name__ == "__main__":
    session_id = "000020"
    compare_transcriptions(session_id)