SHOW:
|
|
- or go back to the newest paste.
| 1 | #! /usr/bin/python | |
| 2 | #bcc.py | |
| 3 | # | |
| 4 | #A script to strip email addresses from a file and convert them to a format suitable for pasting into a bcc field | |
| 5 | ||
| 6 | import sys, string | |
| 7 | ||
| 8 | def main(): | |
| 9 | ||
| 10 | #infile = open("raw.txt", "r") #to call from within IDE for debugging
| |
| 11 | infile = open(sys.argv[1], 'r') #open the file given to the script as an argument | |
| 12 | rawString = infile.read() | |
| 13 | ||
| 14 | ||
| 15 | rawString = stripOut(rawString, "\n") #replace newlines with spaces | |
| 16 | - | rawString = stripOut(rawString, ",") #replace commas with spaces |
| 16 | + | |
| 17 | rawString = stripOut(rawString, "\t") #replace tabs with spaces | |
| 18 | ||
| 19 | for char in "\n\t\"\/\'^?\/]}[{":
| |
| 20 | rawString = stripOut(rawString, char) | |
| 21 | ||
| 22 | ||
| 23 | ||
| 24 | ||
| 25 | ||
| 26 | ||
| 27 | addressList = [] #create an empty list to put the addresses in | |
| 28 | for address in string.split(rawString):#break it up into items separated by whitespace | |
| 29 | ||
| 30 | address = wellFormed(address) #strip it of forbidden elements | |
| 31 | if address !="": #check the result still contains an email address | |
| 32 | addressList.append(address) #if so, add it to the list | |
| 33 | ||
| 34 | bccString = ", ".join(addressList) #join the addresses together separated by commas | |
| 35 | infile.close() #close the input file | |
| 36 | outfile = open("bcc.txt", "w") #open file to write
| |
| 37 | outfile.write(bccString) #write | |
| 38 | outfile.close() #close | |
| 39 | #print bccString #IDE debugging | |
| 40 | ||
| 41 | ||
| 42 | def wellFormed(possAddress): #turn input into well-formed email address, or empty string | |
| 43 | ||
| 44 | at = False #make sure "@" is present | |
| 45 | for char in possAddress: | |
| 46 | if char == "@": | |
| 47 | at = True | |
| 48 | if at == False: | |
| 49 | return "" #if not, return an empty string | |
| 50 | ||
| 51 | user, domain = possAddress.split("@") #divide into user and domain
| |
| 52 | user, domain = stripLeft(user), stripRight(domain) #strip off forbidden characters from beginning and end respectively | |
| 53 | ||
| 54 | if user == "" or domain == "": #if either is empty, return empty string | |
| 55 | return "" | |
| 56 | else: | |
| 57 | return user + "@" + domain | |
| 58 | ||
| 59 | ||
| 60 | def stripLeft(text): #strip inadmissible characters off the neginning of the username | |
| 61 | if text == "": | |
| 62 | return text | |
| 63 | elif alphaNum(text[len(text)-1]) == False: | |
| 64 | return "" #throw away illegal character and anything to its left | |
| 65 | else: | |
| 66 | return stripLeft(text[:-1]) + text[-1] #recursively move leftward if character is legal | |
| 67 | ||
| 68 | def stripRight(text): #strip inadmissible characters off the end of the domain name | |
| 69 | if text == "": | |
| 70 | return text | |
| 71 | elif alphaNum(text[0]) == False: | |
| 72 | return "" | |
| 73 | else: | |
| 74 | return text[0] + stripRight(text[1:]) | |
| 75 | ||
| 76 | ||
| 77 | def alphaNum(char): #check if character permitted | |
| 78 | if char == "": | |
| 79 | return False #if empty string | |
| 80 | ascii = ord(char) #get ASCII value of character | |
| 81 | if ascii == 46: #is it a full stop? | |
| 82 | return True | |
| 83 | if ascii == 95: #or an underscore? | |
| 84 | return True | |
| 85 | if ascii == 45: #or a minus sign? | |
| 86 | return True | |
| 87 | elif ascii > 64 and ascii < 91: #or a capital letter? | |
| 88 | return True | |
| 89 | elif ascii > 96 and ascii < 123: #lower case? | |
| 90 | return True | |
| 91 | elif ascii > 47 and ascii < 58: #a numeral? | |
| 92 | return True | |
| 93 | else: | |
| 94 | return False #not a permitted character then ... | |
| 95 | ||
| 96 | def stripOut(mainString, subString): | |
| 97 | return " ".join(mainString.split(subString))#replace occurrences of subString with spaces | |
| 98 | ||
| 99 | main() |
