def do_fetch(city: City, communities_queue: Queue):
    # 抓取, 直到没东西可抓

    # http://docs.sqlalchemy.org/en/latest/orm/session_basics.html#is-the-session-thread-safe
    db_session = Session()

    while not communities_queue.empty():

        a_community = communities_queue.get()

        try:
            fetch_page(city, a_community.id)
            a_community.page_fetched_at = datetime.now()
        except Exception as e:
            _counts['failed'] += 1
            logging.error(
                f'# 抓取失败, community_id={a_community.id}, message="{e}"')

        else:
            db_session.add(a_community)
            db_session.commit()

        _counts['completed'] += 1

        if _counts['completed'] % 10 == 0:
            count_remaining = _counts["total"] - _counts["completed"]
            logging.info(
                f'进度={_counts["completed"]}/{_counts["total"]}, 剩余={count_remaining}'
            )

        communities_queue.task_done()

    db_session.close()
Example #2
0
def update_city(city_id):
    """
    初始化/更新城市信息
    """
    logging.info('初始化/更新城市信息... city_id={}'.format(city_id))

    city_info = get_city_info(city_id)
    city = City(city_info)

    db_session = Session()
    db_session.merge(city)

    for district_info in city_info['district']:
        district = District(city.id, district_info)
        logging.info(
            '城市={}, 区域={}, 商圈数={}'.format(
                city.name, district.name, district.biz_circles_count
            )
        )
        DISTRICT_MAP[district.name] = district.id
        db_session.merge(district)

        for biz_circle_info in district_info['bizcircle']:
            biz_circle = db_session.query(BizCircle).filter(
                BizCircle.id == int(biz_circle_info['bizcircle_id'])
            ).first()

            if biz_circle:
                # 记录已存在,可能需要更新 district_id
                if district.id not in biz_circle.district_id:
                    # biz_circle.district_id.append()、district_id += 等方式都不能更新表
                    biz_circle.district_id = biz_circle.district_id + [
                        district.id
                    ]
            else:
                biz_circle = BizCircle(city.id, district.id, biz_circle_info)
                db_session.add(biz_circle)

    db_session.commit()
    db_session.close()

    logging.info('初始化/更新城市信息结束.')
Example #3
0
def parse_all_communities(city_id):
    db_session = Session()

    communities = db_session.query(Community).filter(
        Community.city_id == city_id, Community.detail == None,
        Community.page_fetched_at != None).all()

    total_count = len(communities)
    logging.info(f'city_id={city_id}, 待分析={total_count}')

    for i, a_community in enumerate(communities):
        detail = parse_community_detail(a_community.id)

        if detail:
            a_community.detail = detail

        if (i + 1) % 100 == 0 or (i == total_count - 1):
            logging.info(f'进度={i + 1}/{total_count}, 剩余={total_count - i - 1}')
            db_session.commit()

    logging.info('已全部分析完成.')
    db_session.close()