示例#1
0
def save_item(item, name):
    if not item:
        return
    try:
        cursor = _conn.cursor()

        #record last_crawl time
        item['add_time'] = get_epoch_datetime()
        item['update_time'] = get_epoch_datetime()

        _upsert_item(cursor, item)
        _conn.commit()
    except Exception, e:
        log_error(e)
示例#2
0
    def adapt(self, item):
        super(BeiHangItemAdapter, self).adapt(item)
        if item.has_key('content'):
            self._get_detail(item)
            item['content'] = _adapt_content_str(item['content'])

        item['post_time'] = get_epoch_datetime()

        return item
示例#3
0
def report_link(source, catetory, link, description=''):
    try:
        cursor = _conn.cursor()

        insert_sql = "INSERT INTO %s (source, category, link, description, create_time) VALUES('%s', '%s', '%s', '%s', %s)" % \
                    (_link_monitor_table, source, catetory, link, description, get_epoch_datetime())
        cursor.execute(insert_sql)
        _conn.commit()
    except MySQLdb.Error, e:
        log_error(e)
示例#4
0
    def adapt(self, item):
        super(USTBItemAdapter, self).adapt(item)
        if item.has_key('address'):
            item['address'] = self._get_address(item['address'])
        if item.has_key('begin_time'):
            item['begin_time'], item['end_time'] = self._get_time(item['begin_time'])
        if item.has_key('content'):
            item['content'] = _adapt_content_str(item['content'])

        item['post_time'] = get_epoch_datetime()

        return item
示例#5
0
 def adapt(self, item):
     super(self.__class__, self).adapt(item)
     if item.has_key('address'):
         item['address'] = item['address'].strip()
     if item.has_key('begin_time'):
         item['begin_time'] = _adapt_datetime_str(item.get('begin_time').replace(u'\xa0\xa0', ' '))
     if item.has_key('post_time'):
         item['post_time'] = self._get_post_time(item.get('post_time'))
     else:
         item['post_time'] = get_epoch_datetime()
     if item.has_key('content'):
         item['content'] = _adapt_content_str(item.get('content'))
     item['end_time'] = 0
     return item
示例#6
0
def _adapt_datetime_str(datetime_str):
    if not datetime_str:
        return None;
    date = datetime.strptime(datetime_str, "%Y-%m-%d %H:%M")
    return get_epoch_datetime(date)