| 1 |
nino.borges |
621 |
"""
|
| 2 |
|
|
|
| 3 |
|
|
EmailDomainExtractor
|
| 4 |
|
|
|
| 5 |
|
|
Created by
|
| 6 |
|
|
Emanuel Borges
|
| 7 |
|
|
11.14.2014
|
| 8 |
|
|
|
| 9 |
|
|
This program will take a delimited file with fields containing email address data and will parse out the domains,
|
| 10 |
|
|
outputting a list of unique values.
|
| 11 |
|
|
|
| 12 |
|
|
"""
|
| 13 |
|
|
|
| 14 |
|
|
import os
|
| 15 |
|
|
|
| 16 |
|
|
class ExtractDomains():
|
| 17 |
|
|
version = "0.1.0"
|
| 18 |
|
|
def __init__(self):
|
| 19 |
|
|
self.delimFile = ""
|
| 20 |
|
|
self.outputFile = ""
|
| 21 |
|
|
self.domainMatrix = {}
|
| 22 |
|
|
self.domainList = []
|
| 23 |
|
|
|
| 24 |
|
|
def TestValues(self):
|
| 25 |
|
|
"""An internal method to first test that delimFile and outputFile have been selected and are allowed"""
|
| 26 |
|
|
everythingOk = True
|
| 27 |
|
|
if self.delimFile:
|
| 28 |
|
|
if os.path.exists(self.delimFile):
|
| 29 |
|
|
pass
|
| 30 |
|
|
else:
|
| 31 |
|
|
everythingOK = False
|
| 32 |
|
|
else:
|
| 33 |
|
|
everythingOK = False
|
| 34 |
|
|
|
| 35 |
|
|
if self.outPutFile:
|
| 36 |
|
|
pass
|
| 37 |
|
|
else:
|
| 38 |
|
|
everythingOK = False
|
| 39 |
|
|
|
| 40 |
|
|
return everythingOk
|
| 41 |
|
|
|
| 42 |
|
|
def ParseDomain(self, emailAddress):
|
| 43 |
|
|
"""Interal method that will parse 1 email address, returning only the domain"""
|
| 44 |
|
|
if "@" in emailAddress:
|
| 45 |
|
|
domain = emailAddress.split("@")[1]
|
| 46 |
|
|
domain = "@" + domain
|
| 47 |
|
|
else:
|
| 48 |
|
|
domain = None
|
| 49 |
|
|
return domain
|
| 50 |
|
|
|
| 51 |
|
|
def ExtractDomains(self):
|
| 52 |
|
|
"""Main method that goes through the file and does the parsing."""
|
| 53 |
|
|
everythingOk = self.TestValues()
|
| 54 |
|
|
if everythingOk:
|
| 55 |
|
|
contents = open(self.delimFile).readlines()
|
| 56 |
|
|
for line in contents:
|
| 57 |
|
|
line = line.replace("\n","")
|
| 58 |
|
|
fields = line.split("|")
|
| 59 |
|
|
for field in fields:
|
| 60 |
|
|
if ";" in field:
|
| 61 |
|
|
values = field.split(";")
|
| 62 |
|
|
else:
|
| 63 |
|
|
values = [field]
|
| 64 |
|
|
for emailAddress in values:
|
| 65 |
|
|
domain = self.ParseDomain(emailAddress)
|
| 66 |
|
|
if domain:
|
| 67 |
|
|
self.domainMatrix[domain] = 1
|
| 68 |
|
|
self.domainList = domainMatrix.keys()
|
| 69 |
|
|
sucessResult = True
|
| 70 |
|
|
else:
|
| 71 |
|
|
sucessResult = False
|
| 72 |
|
|
return sucessResult
|
| 73 |
|
|
|
| 74 |
|
|
def WriteValuesToFile(self):
|
| 75 |
|
|
"""This method unpacks the domainList to the outputFile"""
|
| 76 |
|
|
everythingOk = self.TestValues()
|
| 77 |
|
|
if everythingOk:
|
| 78 |
|
|
outputFile = open(self.outputFile,'w')
|
| 79 |
|
|
if self.domainList:
|
| 80 |
|
|
outputFile.write("List of Domains\n")
|
| 81 |
|
|
for domain in domainList:
|
| 82 |
|
|
outputFile.write("%s\n"%domain)
|
| 83 |
|
|
outputFile.close()
|
| 84 |
|
|
sucessResult = True
|
| 85 |
|
|
else:
|
| 86 |
|
|
sucessResult = False
|
| 87 |
|
|
|
| 88 |
|
|
|
| 89 |
|
|
|
