hive udf (python)

来源:互联网 发布:网络红歌2016火爆歌曲 编辑:程序博客网 时间:2024/05/17 18:47

searchstr_map.py

#!/usr/bin/env python#coding:utf-8import sysimport refor line in sys.stdin:    line = line.strip()    fields = line.split("\t")    cityid = str(fields[0])    searchstr = fields[1]    pv = str(fields[2])    qrys = re.findall(r'([a-z][\-_.0-9]+)', searchstr)    qrys.sort()    sorted_str = ''.join(qrys)    print '%s\t%s\t%s\t%s' % (cityid, searchstr, sorted_str, pv)

hive 命令行使用:

add file searchstr_map.pyselect    TRANSFORM (src.cityid, src.searchstr, src.pv)      USING 'python searchstr_map.py'      AS (cityid, searchstr, sorted_str, pv) from(select     if(cityid='1', 0, cast(cityid as int)) cityid,    a.searchstr,    count(1) pvfrom (select     split(request['cityid'], ',') cityid_arr,    request['searchstr'] searchstrfrom db.tablename where dt_ymd=20170731and request_ser like '%/search/carinfo%'and request['searchstr'] is not null and request['searchstr'] <> '') alateral view explode(a.cityid_arr)t as cityidwhere cityid rlike '^[0-9]+$'group by if(cityid='1', 0, cast(cityid as int)), a.searchstr) src
原创粉丝点击