Find connected components in speaker-conversation graphs

split-convs.py

import json
import sys
from collections import defaultdict

speakers = defaultdict(list)
conversations = defaultdict(list)

for filename in sys.argv[1:]:
    with open(filename) as fp:
        metadata = json.loads(fp.read())

    for speaker in metadata['speakers']:
        speakers[speaker['user_id']].append(metadata['id'])
        conversations[metadata['id']].append(speaker['user_id'])

parts = defaultdict(set)
speaker_colors = {}

def mark_reachable(target, color):
    if target not in speaker_colors:
        speaker_colors[target] = color
        for conversation in speakers[target]:
            parts[color].add(conversation)
            for speaker in conversations[conversation]:
                if speaker != target and speaker not in speaker_colors:
                    mark_reachable(speaker, color)


for color, target in enumerate(speakers.keys()):
    mark_reachable(target, color)

# TODO: drop conversations to cut cliques
for i, part in enumerate(sorted(parts.values(), key=len, reverse=True)):
    print(i, len(part), ' '.join(part))