示例#1
0
def get_city_news(topic, start_ts, end_ts, unit=MinInterval, limit=TOP_NEWS_LIMIT):
    weibos = []
    if (end_ts - start_ts < unit):
        upbound = int(math.ceil(end_ts / (unit * 1.0)) * unit)
        item = db.session.query(CityNews).filter(CityNews.end==upbound, \
                                                       CityNews.topic==topic, \
                                                       CityNews.range==unit, \
                                                       CityNews.limit==limit).first()
        if item:
            news = _json_loads(item.news)
            for news_item in news:
                weibos.append((news_item['timestamp'],news_item))
    else:
        upbound = int(math.ceil(end_ts / (unit * 1.0)) * unit)
        lowbound = (start_ts / unit) * unit
        items = db.session.query(CityNews).filter(CityNews.end>lowbound, \
                                                         CityNews.end<=upbound, \
                                                         CityNews.topic==topic, \
                                                         CityNews.range==unit, \
                                                         CityNews.limit==limit).all()
        for item in items:
            news = _json_loads(item.news)
            try:
                for news_item in news:
                    weibos.append((news_item['timestamp'],news_item))
            except:
                continue
    sorted_weibos = sorted(weibos, key=lambda k: k[0], reverse=True)

    city_dict = {}
    k = 0
    for reposts_count, result in sorted_weibos:
        k += 1
        if (k > 1000):
            break
        """
        uid = result['user']
        user_info = acquire_user_by_id(uid)
        if user_info:
            result['username'] = user_info['name']
        else:
            result['username'] = '******'
        time = ts2date(result['timestamp'])
        result['time'] = time
        """
        try:
            city = result['source_from_area'].split('\t')[1]
        except:
            city = ''
        result['_id'] = deal_with(result['_id'])

        if city in province_list:
            try:
                city_dict[city].append(result)
            except:
                city_dict[city] = [result]
    return city_dict
示例#2
0
def _top_weibos(weibos_dict, top=TOP_READ):
    results_list = []
    results_list_new = []
    if weibos_dict != {}:
        results = sorted(weibos_dict.iteritems(), key=lambda (k, v): v[0], reverse=False)
        results = results[len(results) - top :]
        for k, v in results:
            results_list.extend(v[1])
        for i in range(len(results_list)):
            results_list_new.append(results_list[len(results_list) - 1 - i])

    for r in results_list_new:
        r["_id"] = deal_with(r["_id"])

    return results_list_new
示例#3
0
def _top_weibos(weibos_dict, top=TOP_READ):
    results_list = []
    results_list_new = []
    if weibos_dict != {}:
        results = sorted(weibos_dict.iteritems(),
                         key=lambda (k, v): v[0],
                         reverse=False)
        results = results[len(results) - top:]
        for k, v in results:
            results_list.extend(v[1])
        for i in range(len(results_list)):
            results_list_new.append(results_list[len(results_list) - 1 - i])

    for r in results_list_new:
        r['_id'] = deal_with(r['_id'])

    return results_list_new
示例#4
0
def parseNews(news):
    news_dict = {}
    news = _json_loads(news)

    if not news:
        return {}

    for weibo in news:
        try:
            _id = deal_with(weibo['_id'])
            replies = 1
            weibo['timestamp'] = ts2date(weibo['timestamp'])
            weibo['content168'] = weibo['content168']
            news_dict[_id] = [replies, weibo]
        except:
            continue

    return news_dict
示例#5
0
def get_city_news(topic,
                  start_ts,
                  end_ts,
                  unit=MinInterval,
                  limit=TOP_NEWS_LIMIT):
    weibos = []
    if (end_ts - start_ts < unit):
        upbound = int(math.ceil(end_ts / (unit * 1.0)) * unit)
        item = db.session.query(CityNews).filter(CityNews.end==upbound, \
                                                       CityNews.topic==topic, \
                                                       CityNews.range==unit, \
                                                       CityNews.limit==limit).first()
        if item:
            news = _json_loads(item.news)
            for news_item in news:
                weibos.append((news_item['timestamp'], news_item))
    else:
        upbound = int(math.ceil(end_ts / (unit * 1.0)) * unit)
        lowbound = (start_ts / unit) * unit
        items = db.session.query(CityNews).filter(CityNews.end>lowbound, \
                                                         CityNews.end<=upbound, \
                                                         CityNews.topic==topic, \
                                                         CityNews.range==unit, \
                                                         CityNews.limit==limit).all()
        for item in items:
            news = _json_loads(item.news)
            try:
                for news_item in news:
                    weibos.append((news_item['timestamp'], news_item))
            except:
                continue
    sorted_weibos = sorted(weibos, key=lambda k: k[0], reverse=True)

    city_dict = {}
    k = 0
    for reposts_count, result in sorted_weibos:
        k += 1
        if (k > 1000):
            break
        """
        uid = result['user']
        user_info = acquire_user_by_id(uid)
        if user_info:
            result['username'] = user_info['name']
        else:
            result['username'] = '******'
        time = ts2date(result['timestamp'])
        result['time'] = time
        """
        try:
            city = result['source_from_area'].split('\t')[1]
        except:
            city = ''
        result['_id'] = deal_with(result['_id'])

        if city in province_list:
            try:
                city_dict[city].append(result)
            except:
                city_dict[city] = [result]
    return city_dict