User:Statsrick/PYTHONUDF code

Text Cleaner UDF

#/usr/bin/python

import urllib
import re


@outputSchema("input:chararray")
def URLDECODE(input):
        return urllib.unquote(input).decode('utf8')

@outputSchema("input:chararray")
def URLENCODE(input):
        return urllib.unquote(input).encode('utf8')

@outputSchema("input:chararray")
def TXTCLEANER(input):
        tmp = re.sub("[^A-Za-z0-9\\s]+", "",input)
        return tmp

@outputSchema("input:chararray")
def AUTHORPARSE(input):
        tmp = ""
        indexP1 = input.find("(")
        hrefInx = input.find("a href")

        if hrefInx<0 :

                if indexP1>=0 :

                        tmp = re.sub("[^A-Za-z\\s]+", "", input[indexP1+1:]).lower()

                elif input.find("@")>0 :

                        tmp = re.sub("[^A-Za-z\\s]+","",input[:input.find("@")]).lower()


                else :

                        tmp = re.sub("[^A-Za-z\\s]+", "",input).lower()
        return tmp

Customer Targeting

#/usr/bin/python

@outputSchema("input:chararray")
def CTPARSE(input):
        tmp = ""
        try:
                f1 = input.find("site=")
                if f1>0:
                        tsite = input[input.find("site=")+5:]
                        site =  tsite[:tsite.find(";")]
                else:
                        site = ""

                f2 = input.find("sub=")
                if f2>0:
                        tsub = input[input.find("sub=")+4:]
                        sub =  tsub[:tsub.find(";")]
                else:
                        sub = ""

                f3 = input.find("subsub=")
                if f3>0:
                        tsubsub = input[input.find("subsub=")+7:]
                        f33 = tsubsub.find(";")
                        if f33>0:
                                subsub =  tsubsub[:tsubsub.find(";")]
                        else:
                                subsub =  tsubsub
                else:
                        subsub = ""

                f4 = input.find("page=")
                if f4>0:
                        tpage = input[input.find("page=")+5:]
                        page =  tpage[:tpage.find(";")]
                else:
                        page = ""

                if site!="" and sub!="" and subsub!="" and page!="":
                        tmp = "http://www." +  site.strip()  + ".com" + sub.strip() + "/" + subsub.strip() + "/" + page.strip()
        except:
                tmp=""
        return tmp

@outputSchema("input:chararray")
def ARTICLE(input):
        tmp = ""
        try:
                f5 = input.find("artid=")
                if f5>0:
                        tartid = input[input.find("artid=")+6:]
                        tmp =  tartid[:tartid.find(";")]
                else:
                        tmp = ""
        except:
                tmp=""
        return tmp

@outputSchema("input:chararray")
def TOOL(input):
        tmp = ""
        try:
                f6 = input.find("tool=")
                if f6>0:
                        ttool = input[input.find("tool=")+5:]
                        f66 = ttool.find(";")
                        f67 = ttool.find(",")
                        if f66>0:
                                tmp =  ttool[:ttool.find(";")]

                        elif f67>0:
                                tmp =  ttool[:ttool.find(",")]
                        else:
                                tmp = ""
                else:
                        tmp = ""
        except:
                tmp=""
        return tmp

Client Parser UDF

#/usr/bin/python

 #implement CASE FOR BO Field
@outputSchema("client:chararray")
def BU_parse(client):
 hmgList = ['10048','10049','10050','1782','10051','10052','10053','10054','10055','10056','10057','10058','10059','10060','10061','10062','10063','10064','10065','10066','10067','10069','10070','10071','10047']
 hnpList = ['9','10012','6','8','10005','10006','10007','10004','10015','7','1845','10013','10008','10009','10010','10014','4','5','10011','1888','1887']
 htvList = ['10041','10038','10042','10039','10045','10033','10044','10037','10046','10019','10017','10022','10040','10024','10023','10036','10020','10028','10034','10035','10031','10027','10032','10029','10018','10021','10043','10030',$
 if client in hmgList : return 'HMG'
 elif client in hnpList : return 'HNP'
 elif client in htvList : return 'HTV'
 else : return 'HMI'


 #implement CASE for client_cd_1 field
@outputSchema("client:chararray")
def client_cd1_parse(client):
 if client=='10047' : return 'USM1'
 elif client=='10047' : return 'USM1'
 elif client=='10048' : return 'USM2'
 elif client=='10049' : return 'USM3'
 elif client=='10050' : return 'USM4'
 elif client=='1782' : return 'USM5'
 elif client=='10051' : return 'USM6'
 elif client=='10052' : return 'USM7'
 elif client=='10054' : return 'USM8'
 elif client=='10055' : return 'USM9'
 elif client=='10056' : return 'USM10'
 elif client=='10057' : return 'USM11'
 elif client=='10058' : return 'USM12'
 elif client=='10059' : return 'USM13'
 elif client=='10060' : return 'USM14'
 elif client=='10061' : return 'USM15'
 elif client=='10062' : return 'USM16'
 elif client=='10063' : return 'USM17'
 elif client=='10064' : return 'USM18'
 elif client=='10065' : return 'USM19'
 elif client=='10066' : return 'USM20'
 elif client=='10067' : return 'USM21'
 elif client=='10069' : return 'USM22'
 elif client=='10070' : return 'USM23'
 elif client=='10041' : return 'UST1'
 elif client=='10038' : return 'UST2'
 elif client=='10042' : return 'UST3'
 elif client=='10039' : return 'UST4'
 elif client=='10045' : return 'UST5'
 elif client=='10033' : return 'UST6'
 elif client=='10044' : return 'UST7'
 elif client=='10037' : return 'UST8'
 elif client=='10046' : return 'UST9'
 elif client=='10019' : return 'UST10'
 elif client=='10017' : return 'UST11'
 elif client=='10022' : return 'UST12'
 elif client=='10040' : return 'UST13'
 elif client=='10024' : return 'UST14'
 elif client=='10023' : return 'UST15'
 elif client=='10036' : return 'UST16'
 elif client=='10020' : return 'UST17'
 elif client=='10028' : return 'UST18'
 elif client=='10034' : return 'UST19'
 elif client=='10035' : return 'UST20'
 elif client=='10031' : return 'UST21'



 elif client=='10027' : return 'UST22'
 elif client=='10032' : return 'UST23'
 elif client=='10029' : return 'UST24'
 elif client=='10018' : return 'UST25'
 elif client=='10021' : return 'UST26'
 elif client=='10043' : return 'UST27'
 elif client=='10030' : return 'UST28'
 elif client=='10026' : return 'UST29'
 elif client=='9' : return 'USN1'
 elif client=='10012' : return 'USN2'
 elif client=='6' : return 'USN3'
 elif client=='8' : return 'USN4'
 elif client=='10005' : return 'USN5'
 elif client=='10006' : return 'USN6'
 elif client=='10007' : return 'USN7'
 elif client=='10004' : return 'USN8'
 elif client=='10015' : return 'USN9'
 elif client=='7' : return 'USN10'
 elif client=='1845' : return 'USN11'
 elif client=='10013' : return 'USN12'
 elif client=='10008' : return 'USN13'
 elif client=='10009' : return 'USN14'
 elif client=='10010' : return 'USN15'
 elif client=='10014' : return 'USN16'
 elif client=='4' : return 'USN17'
 elif client=='5' : return 'USN18'
 elif client=='10011' : return 'USN19'
 elif client=='1887' : return 'USN20'
 elif client=='1888' : return 'USN21'
 elif client=='1941' : return 'ITH1'
 elif client=='1942' : return 'ITH2'
 elif client=='1943' : return 'ITH3'
 elif client=='1944' : return 'ITH4'
 elif client=='1945' : return 'ITH5'
 elif client=='1946' : return 'NLH1'
 elif client=='1947' : return 'NLH2'
 elif client=='1948' : return 'NLH3'
 elif client=='1949' : return 'NLH4'
 elif client=='1950' : return 'NLH5'
 elif client=='1951' : return 'NLH6'
 elif client=='1952' : return 'NLH7'
 elif client=='1953' : return 'NLH8'
 elif client=='1984' : return 'NLH9'
 elif client=='1954' : return 'ESH1'
 elif client=='1955' : return 'ESH2'
 elif client=='1956' : return 'ESH3'
 elif client=='1957' : return 'ESH4'
 elif client=='1958' : return 'ESH5'
 elif client=='1959' : return 'ESH6'
 elif client=='1960' : return 'ESH7'
 elif client=='1961' : return 'ESH8'
 elif client=='1962' : return 'ESH9'
 elif client=='1963' : return 'ESH10'
 elif client=='1964' : return 'ESH11'
 elif client=='1965' : return 'ESH12'
 elif client=='1966' : return 'ESH13'
 elif client=='1967' : return 'ESH14'
 elif client=='1968' : return 'ESH15'
 elif client=='1969' : return 'ESH16'


 elif client=='1985' : return 'ESH17'
 elif client=='1915' : return 'UKH1'
 elif client=='1916' : return 'UKH2'
 elif client=='1917' : return 'UKH3'
 elif client=='1918' : return 'UKH4'
 elif client=='1919' : return 'UKH5'
 elif client=='1920' : return 'UKH6'
 elif client=='1921' : return 'UKH7'
 elif client=='1922' : return 'UKH8'
 elif client=='1923' : return 'UKH9'
 elif client=='1924' : return 'UKH10'
 elif client=='1925' : return 'UKH11'
 elif client=='1926' : return 'UKH12'
 elif client=='1927' : return 'UKH13'
 elif client=='1928' : return 'UKH14'
 elif client=='1929' : return 'UKH15'
 elif client=='1930' : return 'UKH16'
 elif client=='1931' : return 'UKH17'
 elif client=='1932' : return 'UKH18'
 elif client=='1933' : return 'UKH19'
 elif client=='1934' : return 'UKH20'
 elif client=='1935' : return 'UKH21'
 elif client=='1936' : return 'UKH22'
 elif client=='1937' : return 'UKH23'
 elif client=='1938' : return 'UKH24'
 elif client=='1939' : return 'UKH25'
 elif client=='1940' : return 'UKH26'