kangshifu-board/scripts/summarize-product-review-samples.py

"""Summarize exact product identity, observed likes and task-specific bills."""
from collections import Counter
import json
from pathlib import Path
import re


PLACEHOLDER = re.compile(r'未填写评价|该用户觉得商品|此用户没有填写|系统默认')


def payload(response):
    return response.get('structuredContent', {})


def sample(folder):
    reservation = json.loads((folder / 'reservation.json').read_text(encoding='utf-8'))
    result_file = folder / 'mcp-result.json'
    if not result_file.is_file():
        return {'label': folder.name, 'status': 'pending', 'reservationCny': 0.30}
    result = payload(json.loads(result_file.read_text(encoding='utf-8'))['response'])
    family = reservation['endpoint'].split('.')[0]
    data = result.get('data', {}).get('result', {})
    rows = data.get('data', []) if family == 'jd' else data.get('comments', [])
    matched = []
    for index, row in enumerate(rows):
        iid = str(row.get('productId')) if family == 'jd' else str(row.get('auction', {}).get('aucNumId'))
        rid = str(row.get('commentId')) if family == 'jd' else str(row.get('rateId'))
        text = row.get('commentData', '') if family == 'jd' else row.get('content', '')
        likes = row.get('praiseCnt') if family == 'jd' else row.get('likeCount')
        matched.append({'itemId': iid, 'reviewId': rid, 'content': text, 'likes': likes,
            'isPlaceholder': bool(PLACEHOLDER.search(text)), 'index': index,
            'matchesRequestedProduct': iid == reservation['params']['itemId']})
    usages = []
    for file in sorted(folder.glob('usage-*.json')):
        data_usage = payload(json.loads(file.read_text(encoding='utf-8'))['response'])
        if data_usage.get('status') == 'ok':
            usages.append(data_usage)
    bill = usages[-1].get('summary') if usages else None
    billed = bool(bill and bill.get('billedCalls', 0))
    return {'label': folder.name, 'productId': family + ':' + reservation['params']['itemId'],
        'page': reservation['params']['page'], 'order': reservation['params'].get('orderType', 'upstream_default'),
        'status': result.get('status'), 'httpStatus': result.get('summary', {}).get('httpStatus'),
        'sourceTime': result.get('summary', {}).get('recordTime'),
        'historicalCountText': reservation.get('historicalCountDisplay'),
        'upstreamCounts': data.get('pageInfo') if family == 'jd' else {k: data.get(k) for k in
            ('total', 'totalFuzzy', 'maxPage', 'hasNextPage')},
        'returnedReviews': len(rows), 'matchingReviews': sum(r['matchesRequestedProduct'] for r in matched),
        'meaningfulMatchingReviews': sum(r['matchesRequestedProduct'] and not r['isPlaceholder'] for r in matched),
        'positiveLikeMatchingReviews': sum(r['matchesRequestedProduct'] and not r['isPlaceholder'] and
            type(r['likes']) is int and r['likes'] > 0 for r in matched),
        'likeDistribution': dict(Counter(str(r['likes']) for r in matched)),
        'maxObservedLikes': max((r['likes'] for r in matched if type(r['likes']) is int), default=None),
        'billing': bill, 'confirmedCostCny': bill.get('costCny', 0) if bill else 0,
        'unsettledReservationCny': 0 if billed else 0.30,
        'usageTraceId': reservation['usageTraceId'], 'records': matched}


def summarize(root):
    records = [sample(folder) for folder in sorted(Path(root).iterdir()) if (folder / 'reservation.json').is_file()]
    return {'collectionCalls': len(records), 'successfulCalls': sum(r['status'] == 'ok' for r in records),
        'confirmedCostCny': round(sum(r.get('confirmedCostCny', 0) for r in records), 2),
        'unsettledReservationCny': round(sum(r.get('unsettledReservationCny', r.get('reservationCny', 0)) for r in records), 2),
        'records': records}


if __name__ == '__main__':
    report = summarize('work/product-reviews/live-r1')
    print(json.dumps({**report, 'records': [{k: v for k, v in row.items() if k != 'records'} for row in report['records']]},
                     ensure_ascii=False, indent=2))