bcc2.py

#! /usr/bin/python
#bcc.py
#
#A script to strip email addresses from a file and convert them to a format suitable for pasting into a bcc field

import sys, string

def main():

    #infile = open("raw.txt", "r")         #to call from within IDE for debugging
    infile = open(sys.argv[1], 'r')        #open the file given to the script as an argument
    rawString = infile.read()


    rawString = stripOut(rawString, "\n")   #replace newlines with spaces

    rawString = stripOut(rawString, "\t")   #replace tabs with spaces

    for char in "\n\t\"\/\'^?\/]}[{":
        rawString = stripOut(rawString, char)






    addressList = []                       #create an empty list to put the addresses in
    for address in string.split(rawString):#break it up into items separated by whitespace

        address = wellFormed(address)      #strip it of forbidden elements
        if address !="":                   #check the result still contains an email address
            addressList.append(address)    #if so, add it to the list

    bccString = ", ".join(addressList)     #join the addresses together separated by commas
    infile.close()                         #close the input file
    outfile = open("bcc.txt", "w")         #open file to write
    outfile.write(bccString)               #write
    outfile.close()                        #close
    #print bccString                        #IDE debugging


def wellFormed(possAddress):               #turn input into well-formed email address, or empty string

    at = False                             #make sure "@" is present
    for char in possAddress:
        if char == "@":
            at = True
    if at == False:
        return ""                          #if not, return an empty string

    user, domain = possAddress.split("@")  #divide into user and domain
    user, domain = stripLeft(user), stripRight(domain) #strip off forbidden characters from beginning and end respectively

    if user == "" or domain == "":         #if either is empty, return empty string
        return ""
    else:
        return user + "@" + domain


def stripLeft(text):                       #strip inadmissible characters off the neginning of the username
    if text == "":
        return text
    elif alphaNum(text[len(text)-1]) == False:
        return ""                          #throw away illegal character and anything to its left
    else:
        return stripLeft(text[:-1]) + text[-1] #recursively move leftward if character is legal

def stripRight(text):                      #strip inadmissible characters off the end of the domain name
    if text == "":
        return text
    elif alphaNum(text[0]) == False:
        return ""
    else:
        return text[0] + stripRight(text[1:])


def alphaNum(char):                        #check if character permitted
    if char == "":
        return False                       #if empty string
    ascii = ord(char)                      #get ASCII value of character
    if ascii == 46:                        #is it a full stop?
        return True
    if ascii == 95:                        #or an underscore?
        return True
    if ascii == 45:                        #or a minus sign?
        return True
    elif ascii > 64 and ascii < 91:        #or a capital letter?
        return True
    elif ascii > 96 and ascii < 123:       #lower case?
        return True
    elif ascii > 47 and ascii < 58:        #a numeral?
        return True
    else:
        return False                       #not a permitted character then ...

def stripOut(mainString, subString):
    return " ".join(mainString.split(subString))#replace occurrences of subString with spaces

main()