"""Run the canonical scorer, then attach measured route observations separately."""
import json
from pathlib import Path
import re
import subprocess
import sys

lab = Path(__file__).resolve().parent
root = lab.parents[2]
subprocess.run([sys.executable, str(root / 'tools/lab_score.py'), '--lab', str(lab),
                '--out', str(lab / 'scores.json'), '--markdown', str(lab / 'scores.md')], check=True)
scores = json.loads((lab / 'scores.json').read_text())
experiment = json.loads((lab / 'media-experiment.json').read_text())
observations = []
for source in experiment['sources']:
    name = Path(source['file']).stem
    lines = source['stream_lines']
    # Input stream descriptions precede output mapping and indented output streams.
    descriptions = []
    for line in lines:
        if line not in descriptions:
            descriptions.append(line)
    observations.append({'sample': name,
        'input_video_stream_present': any('Video:' in x for x in descriptions),
        'input_audio_stream_present': any('Audio:' in x for x in descriptions),
        'input_subtitle_stream_present': any('Subtitle:' in x for x in descriptions),
        'markitdown_output_characters': len((lab / 'markitdown' / (name + '.md')).read_text()),
        'ffmpeg_extracted_subtitle_cues': len(re.findall(r'^\d+\n\d\d:\d\d:',
                (lab / 'ffmpeg' / (name + '.md')).read_text(), re.M))})
scores['media_observations'] = observations
scores['media_boundary'] = experiment['boundary']
scores['media_method'] = 'Input stream descriptions from FFmpeg inspection; raw UTF-8 Markdown character count; SRT cue number followed by timestamp. Not ASR accuracy.'
(lab / 'scores.json').write_text(json.dumps(scores, indent=2) + '\n')
print(json.dumps({'cells': len(scores['cells']), 'media_observations': observations}, indent=2))