コード例 #1
0
    def setattr(self, attr, v):
        if attr == 'created_at':
            str_original = None
            if isinstance(v, int):  #Epoch Time as int or datetime.datetime
                str_original = util.time2str(v)
            elif isinstance(v, datetime):
                v = util.time2epoch(v)
                str_original = util.time2str(v)
            else:  # str
                str_original = v
                v = util.time2epoch(v)  #创建日期,转换为基于东八区的Epoch整数格式
            #print '%25s\t:=\t%s' % ('c_at_or',v)
            self.__setattr__('c_at_or', str_original)

        #print '%25s\t:=\t%s' % (attr,v)
        self.__setattr__(attr, v)
コード例 #2
0
ファイル: repost.py プロジェクト: CCPLab/miner
    def setattr(self,attr,v):
        if attr == 'created_at':
            str_original = None
            if isinstance(v,int):   #Epoch Time as int or datetime.datetime
                str_original = util.time2str(v)
            elif isinstance(v,datetime):
                v = util.time2epoch(v)
                str_original = util.time2str(v)
            else:   # str
                str_original = v
                v = util.time2epoch(v)  #创建日期,转换为基于东八区的Epoch整数格式
            #print '%25s\t:=\t%s' % ('c_at_or',v)
            self.__setattr__('c_at_or',str_original)

        #print '%25s\t:=\t%s' % (attr,v)
        self.__setattr__(attr,v)
コード例 #3
0
ファイル: profile.py プロジェクト: CCPLab/miner
    def setattr(self, attr, v):
        if attr in ['province', 'city', 'verified_type'] and not isinstance(v, str):
            # 认证类别、省、市ID用用str存储
            v = str(v)
        elif attr == 'idstr' and v is None:
            v = str(id)
        elif attr == 'created_at':
            str_original = None
            if isinstance(v, int):   # Epoch Time as int or datetime.datetime
                str_original = util.time2str(v)
            elif isinstance(v, datetime):
                v = util.time2epoch(v)
                str_original = util.time2str(v)
            else:   # str
                str_original = v
                v = util.time2epoch(v)  # 创建日期,转换为基于东八区的Epoch整数格式
            # print '%25s\t:=\t%s' % ('created_at_or',v)
            self.__setattr__('created_at_or',str_original)

        # print '%25s\t:=\t%s' % (attr,v)
        self.__setattr__(attr,v)
コード例 #4
0
def log(addr, _type, _dir, name="", data={}, err="", pg="", tag=""):
    """记录消息日志"""
    try:
        today = util.time2str(fmt="%Y-%m-%d")
        # 日期发生变化
        if MsgLog.log_date != today:
            # 关闭
            if MsgLog.log_file:
                MsgLog.log_file.close()
                MsgLog.log_file = None
            # 打开
            MsgLog.log_file = open("log/msg.%s.log" % today, "a")
            MsgLog.log_date = today

        MsgLog.log_file.write(
            "%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\n" %
            (util.time2str(fmt="%H:%M:%S"), pg, tag, addr, name, _dir, _type,
             data if data else "", err if err else ""))
    except:
        logging.fatal(
            "act.log(addr=%r, _type=%r, pg=%r, uid=%r, data=%r, err=%r)\nException: \n%s",
            addr, _type, pg, name, data, err, format_exc())
コード例 #5
0
ファイル: profile.py プロジェクト: zhang-zhan/crawlproject
    def setattr(self, attr, v):
        if attr in ['province', 'city', 'verified_type'
                    ] and not isinstance(v, str):
            #认证类别、省、市ID用用str存储
            v = str(v)
        elif attr == 'idstr' and v is None:
            v = str(id)
        elif attr == 'created_at':
            str_original = None
            if isinstance(v, int):  #Epoch Time as int or datetime.datetime
                str_original = util.time2str(v)
            elif isinstance(v, datetime):
                v = util.time2epoch(v)
                str_original = util.time2str(v)
            else:  # str
                str_original = v
                v = util.time2epoch(v)  #创建日期,转换为基于东八区的Epoch整数格式
            #print '%25s\t:=\t%s' % ('created_at_or',v)
            self.__setattr__('created_at_or', str_original)

        #print '%25s\t:=\t%s' % (attr,v)
        self.__setattr__(attr, v)
コード例 #6
0
def log(addr, _type, conn_id, name, data={}, err="", pg=""):
    """记录消息日志"""
    try:
        today = util.time2str(fmt="%Y-%m-%d")
        # 日期发生变化
        if ConnLog.log_date != today:
            # 关闭
            if ConnLog.log_file:
                ConnLog.log_file.close()
                ConnLog.log_file = None
            # 打开
            ConnLog.log_file = open("log/connect.%s.log" % today, "a")
            ConnLog.log_date = today

        ConnLog.log_file.write(
            "%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\n" %
            (util.time2str(fmt="%H:%M:%S"), _type, conn_id, pg, addr, name,
             data if data else "", err if err else ""))
    except:
        logging.fatal(
            "Conn.log(addr=%r, _type=%r, pg=%r, name=%r, conn_id=%r, data=%r, err=%r)\nException: \n%s",
            addr, _type, pg, name, conn_id, data, err, format_exc())
コード例 #7
0
ファイル: triage.py プロジェクト: zezhen/brick
def query_topdiffspender(config, start, end, param):
    today_start = calendar.timegm(start.timetuple())
    ystd_start = calendar.timegm((start - timedelta(days=1)).timetuple())
    lastwk_start = calendar.timegm((start - timedelta(days=7)).timetuple())

    db_end = util.query_end_time(param['table'],
                                 config,
                                 timezone=param['timezone'])
    end = min(end, db_end)
    today_end = calendar.timegm(end.timetuple()) / 900 * 900

    ystd_end = calendar.timegm((end - timedelta(days=1)).timetuple())
    lastwk_end = calendar.timegm((end - timedelta(days=7)).timetuple())

    param['today_start'] = util.time2str(today_start, DATE_FORMAT)
    param['ystd_start'] = util.time2str(ystd_start, DATE_FORMAT)
    param['lastwk_start'] = util.time2str(lastwk_start, DATE_FORMAT)
    param['today_end'] = util.time2str(today_end, DATE_FORMAT)
    param['ystd_end'] = util.time2str(ystd_end, DATE_FORMAT)
    param['lastwk_end'] = util.time2str(lastwk_end, DATE_FORMAT)

    fields = 'time,id,today_spend,ystd_spend,lastwk_spend,today_click,ystd_click,lastwk_click,today_imp,ystd_imp,lastwk_imp,today_serve,ystd_serve,lastwk_serve,dod_delta,wow_delta,tot_3day_spend,adv'.split(
        ',')

    tabid = param.get('tabid')
    if tabid == 'native_topdiffsection':
        command = SECTION_DIFF_TEMPLATE.format(**param)
    else:
        param['today_query'] = iterate_day(today_start, today_end, param,
                                           TODAY_TEMPLATE)
        param['ystd_query'] = iterate_day(ystd_start, ystd_end, param,
                                          YEST_TEMPLATE)
        param['lastwk_query'] = iterate_day(lastwk_start, lastwk_end, param,
                                            LASTWK_TEMPLATE)
        command = CMPGN_ADV_DIFF_TEMPLATE.format(**param)

    return util.fetch_sql_data(config, fields, command,
                               -util.total_seconds(timedelta(minutes=5)))
コード例 #8
0
    def fetch_data0(self, appconfig, query, args, start, end, offset, data,
                    _type):

        query = copy.deepcopy(query)

        dateformat = util.get(query, 'dateformat', util.DATE_ISO)
        query['dateformat'] = util.convert_date_format(dateformat)

        # timezone_switch_aware_offset is used to make time align in graph
        adjusted_offset = util.timezone_switch_aware_offset(
            start, offset, query['timezone'])
        groupby = query.get('groupby')
        fields = query['fields'].split(',')
        if groupby:
            query['fields'] = ",".join(
                map(
                    lambda f: 'sum(%s) as %s' % (f, f)
                    if (f not in groupby and not re.match(
                        " as ", f, re.IGNORECASE)) else f, fields))
            query['groupby'] = "group by " + groupby
        else:
            query['groupby'] = ""

        query['orderby'] = util.get(query, 'orderby', '',
                                    lambda x: 'order by ' + x)
        query['sort'] = util.get(query, 'sort', '')
        query['limit'] = util.get(query, 'limit', '', lambda x: 'limit ' + x)

        # support parameters in conditions
        query['conditions'] = query['conditions'].format(**query)

        query['partition'] = util.get(query, 'partition', 'ds')

        # support parameters to join 2 tables
        join = util.get(query, 'join', None)
        join_to = None
        if join:
            join_to = util.get(join, 'join_to', None)
            if join_to:
                query['table1'] = util.get(query, 'table', None)
                query['groupby_t1'] = "group by " + util.get(
                    join, 'groupby', '')
                query['conditions_t1'] = self.get_additional_conditions(
                    args, join)
                query['raw_fields_t1'] = ",".join(
                    map(
                        lambda f: 'sum(%s) as %s' % (f, f)
                        if (f not in query['groupby_t1'] and not re.match(
                            " as ", f, re.IGNORECASE)) else f,
                        join['raw_fields'].split(',')))
                query['join_field_t1'] = util.get(join, 'on', None)

                query['table2'] = util.get(join_to, 'table', None)
                query['groupby_t2'] = "group by " + util.get(
                    join_to, 'groupby', '')
                query['conditions_t2'] = self.get_additional_conditions(
                    args, join_to)
                query['raw_fields_t2'] = ",".join(
                    map(
                        lambda f: 'sum(%s) as %s' % (f, f)
                        if (f not in query['groupby_t2'] and not re.match(
                            " as ", f, re.IGNORECASE)) else f,
                        join_to['raw_fields'].split(',')))
                query['join_field_t2'] = util.get(join_to, 'on', None)

        offset_w_timezone = adjusted_offset + util.getOffsetSeconds(
            start, query['timezone'])
        if util.get(query, 'source', None) == 'hive':
            query['start'] = util.time_to_utc(start - offset, dateformat,
                                              query['timezone'])
            query['end'] = util.time_to_utc(end - offset, dateformat,
                                            query['timezone'])
            sql = HIVE_TEMPLATE.format(**query)
            res = util.fetch_hive_data(appconfig, fields, sql,
                                       offset_w_timezone)
        else:
            #query['start'] & query['end'] need to be in timezone-free format,
            #because SQL_TEMPLATE will convert it to timezone-specific time
            query['start'] = util.time2str(start - offset, dateformat)
            query['end'] = util.time2str(end - offset, dateformat)
            if join and join_to:
                sql = SQL_JOIN_TEMPLATE.format(**query)
            else:
                sql = SQL_TEMPLATE.format(**query)
            res = util.fetch_sql_data(appconfig, fields, sql,
                                      offset_w_timezone)

        data[_type] = res