pgp-words.py

#!/usr/bin/env python
# -*- coding: utf-8 -*- (because 0xFF, even : "Yucatán")

import os, re, sys, getopt

_words = [
    ["aardvark", "adroitness"], ["absurd", "adviser"], ["accrue", "aftermath"], ["acme", "aggregate"], ["adrift", "alkali"], ["adult", "almighty"], 
    ["afflict", "amulet"], ["ahead", "amusement"], ["aimless", "antenna"], ["Algol", "applicant"], ["allow", "Apollo"], ["alone", "armistice"], 
    ["ammo", "article"], ["ancient", "asteroid"], ["apple", "Atlantic"], ["artist", "atmosphere"], ["assume", "autopsy"], ["Athens", "Babylon"], 
    ["atlas", "backwater"], ["Aztec", "barbecue"], ["baboon", "belowground"], ["backfield", "bifocals"], ["backward", "bodyguard"], 
    ["banjo", "bookseller"], ["beaming", "borderline"], ["bedlamp", "bottomless"], ["beehive", "Bradbury"], ["beeswax", "bravado"],
    ["befriend", "Brazilian"], ["Belfast", "breakaway"], ["berserk", "Burlington"], ["billiard", "businessman"], ["bison", "butterfat"], 
    ["blackjack", "Camelot"], ["blockade", "candidate"], ["blowtorch", "cannonball"], ["bluebird", "Capricorn"], ["bombast", "caravan"], ["bookshelf", "caretaker"],
    ["brackish", "celebrate"], ["breadline", "cellulose"], ["breakup", "certify"], ["brickyard", "chambermaid"], ["briefcase", "Cherokee"], ["Burbank", "Chicago"], 
    ["button", "clergyman"], ["buzzard", "coherence"], ["cement", "combustion"], ["chairlift", "commando"], ["chatter", "company"], ["checkup", "component"],
    ["chisel", "concurrent"], ["choking", "confidence"], ["chopper", "conformist"], ["Christmas", "congregate"], ["clamshell", "consensus"], 
    ["classic", "consulting"], ["classroom", "corporate"], ["cleanup", "corrosion"], ["clockwork", "councilman"], ["cobra", "crossover"], ["commence", "crucifix"],
    ["concert", "cumbersome"], ["cowbell", "customer"], ["crackdown", "Dakota"], ["cranky", "decadence"], ["crowfoot", "December"], ["crucial", "decimal"], 
    ["crumpled", "designing"], ["crusade", "detector"], ["cubic", "detergent"], ["dashboard", "determine"], ["deadbolt", "dictator"], ["deckhand", "dinosaur"], 
    ["dogsled", "direction"], ["dragnet", "disable"], ["drainage", "disbelief"], ["dreadful", "disruptive"], ["drifter", "distortion"], ["dropper", "document"],
    ["drumbeat", "embezzle"], ["drunken", "enchanting"], ["Dupont", "enrollment"], ["dwelling", "enterprise"], ["eating", "equation"], ["edict", "equipment"], 
    ["egghead", "escapade"], ["eightball", "Eskimo"], ["endorse", "everyday"], ["endow", "examine"], ["enlist", "existence"], ["erase", "exodus"], ["escape", "fascinate"],
    ["exceed", "filament"], ["eyeglass", "finicky"], ["eyetooth", "forever"], ["facial", "fortitude"], ["fallout", "frequency"], ["flagpole", "gadgetry"],
    ["flatfoot", "Galveston"], ["flytrap", "getaway"], ["fracture", "glossary"], ["framework", "gossamer"], ["freedom", "graduate"], ["frighten", "gravity"],
    ["gazelle", "guitarist"], ["Geiger", "hamburger"], ["glitter", "Hamilton"], ["glucose", "handiwork"], ["goggles", "hazardous"], ["goldfish", "headwaters"],
    ["gremlin", "hemisphere"], ["guidance", "hesitate"], ["hamlet", "hideaway"], ["highchair", "holiness"], ["hockey", "hurricane"], ["indoors", "hydraulic"],
    ["indulge", "impartial"], ["inverse", "impetus"], ["involve", "inception"], ["island", "indigo"], ["jawbone", "inertia"], ["keyboard", "infancy"], 
    ["kickoff", "inferno"], ["kiwi", "informant"], ["klaxon", "insincere"], ["locale", "insurgent"], ["lockup", "integrate"], ["merit", "intention"], 
    ["minnow", "inventive"], ["miser", "Istanbul"], ["Mohawk", "Jamaica"], ["mural", "Jupiter"], ["music", "leprosy"], ["necklace", "letterhead"],
    ["Neptune", "liberty"], ["newborn", "maritime"], ["nightbird", "matchmaker"], ["Oakland", "maverick"], ["obtuse", "Medusa"], ["offload", "megaton"],
    ["optic", "microscope"], ["orca", "microwave"], ["payday", "midsummer"], ["peachy", "millionaire"], ["pheasant", "miracle"], ["physique", "misnomer"],
    ["playhouse", "molasses"], ["Pluto", "molecule"], ["preclude", "Montana"], ["prefer", "monument"], ["preshrunk", "mosquito"], ["printer", "narrative"],
    ["prowler", "nebula"], ["pupil", "newsletter"], ["puppy", "Norwegian"], ["python", "October"], ["quadrant", "Ohio"], ["quiver", "onlooker"], ["quota", "opulent"],
    ["ragtime", "Orlando"], ["ratchet", "outfielder"], ["rebirth", "Pacific"], ["reform", "pandemic"], ["regain", "Pandora"], ["reindeer", "paperweight"], 
    ["rematch", "paragon"], ["repay", "paragraph"], ["retouch", "paramount"], ["revenge", "passenger"], ["reward", "pedigree"], ["rhythm", "Pegasus"], 
    ["ribcage", "penetrate"], ["ringbolt", "perceptive"], ["robust", "performance"], ["rocker", "pharmacy"], ["ruffled", "phonetic"], ["sailboat", "photograph"],
    ["sawdust", "pioneer"], ["scallion", "pocketful"], ["scenic", "politeness"], ["scorecard", "positive"], ["Scotland", "potato"], ["seabird", "processor"],
    ["select", "provincial"], ["sentence", "proximate"], ["shadow", "puberty"], ["shamrock", "publisher"], ["showgirl", "pyramid"], ["skullcap", "quantity"],
    ["skydive", "racketeer"], ["slingshot", "rebellion"], ["slowdown", "recipe"], ["snapline", "recover"], ["snapshot", "repellent"], ["snowcap", "replica"],
    ["snowslide", "reproduce"], ["solo", "resistor"], ["southward", "responsive"], ["soybean", "retraction"], ["spaniel", "retrieval"], ["spearhead", "retrospect"],
    ["spellbind", "revenue"], ["spheroid", "revival"], ["spigot", "revolver"], ["spindle", "sandalwood"], ["spyglass", "sardonic"], ["stagehand", "Saturday"],
    ["stagnate", "savagery"], ["stairway", "scavenger"], ["standard", "sensation"], ["stapler", "sociable"], ["steamship", "souvenir"], ["sterling", "specialist"],
    ["stockman", "speculate"], ["stopwatch", "stethoscope"], ["stormy", "stupendous"], ["sugar", "supportive"], ["surmount", "surrender"], 
    ["suspense", "suspicious"], ["sweatband", "sympathy"], ["swelter", "tambourine"], ["tactics", "telephone"], ["talon", "therapist"], ["tapeworm", "tobacco"],
    ["tempest", "tolerance"], ["tiger", "tomorrow"], ["tissue", "torpedo"], ["tonic", "tradition"], ["topmost", "travesty"], ["tracker", "trombonist"], 
    ["transit", "truncated"], ["trauma", "typewriter"], ["treadmill", "ultimate"], ["Trojan", "undaunted"], ["trouble", "underfoot"], ["tumor", "unicorn"],
    ["tunnel", "unify"], ["tycoon", "universe"], ["uncut", "unravel"], ["unearth", "upcoming"], ["unwind", "vacancy"], ["uproot", "vagabond"], ["upset", "vertigo"],
    ["upshot", "Virginia"], ["vapor", "visitor"], ["village", "vocalist"], ["virus", "voyager"], ["Vulcan", "warranty"], ["waffle", "Waterloo"], 
    ["wallet", "whimsical"], ["watchword", "Wichita"], ["wayside", "Wilmington"], ["willow", "Wyoming"], ["woodlark", "yesteryear"], ["Zulu", "Yucatán"]]

_unwords="".join(["|%-15s"%x[0].lower() for x in _words]+["|%-15s"%x[1].lower() for x in _words])
    
hexre = re.compile("[a-fA-F0-9]+")

def wordify(seq):
    """Convert seq (hexadecimal string) to PGP words """
    seq = filter(lambda x: x not in (' ', '\n', '\t'), seq)
    seq = "".join(seq) # Python3 compatibility

    if not hexre.match(seq):
        raise Exception("Input is not a valid hexadecimal value.")

    if len(seq) % 2:
        raise Exception("Input contains an odd number of bytes.")

    ret = []
    for i in range(0, len(seq), 2):
        ret.append(_words[int(seq[i:i+2], 16)][(i//2)%2])
    return ret

def wordifyByteLong(longIn):
    """Convert longIn (long) to PGP words """
    return wordify(hex(longIn)[2:][:-1])

def hexify(seq):
    """Convert word sequence of PGP words to hex string"""
    #cleanup seq first
    seq=re.sub("\s\s+", " ",seq.lower())
    tokens=seq.split(" ")
    hexes=[[y / 265, y % 256] for y in [_unwords.find("|%s"%x)/16 for x in tokens]]
    ret = []    
    for idx,hh in enumerate(hexes):
        ret.append(hh[1])
        if  hh[0]<0:
            raise Exception("Input contains bad pgp word at position %d: %s" % (idx,tokens[idx]))
        elif hh[0]!=idx%2:
            raise Exception("Odd-Even word choice error in input at position %d: %s" % (idx,tokens[idx]))
    return "".join(["%02x"%rr for rr in ret])

def usage():
    print("Usage:")
    print("  {0} [fingerprint...]   ==> encode hex to pgpwords".format(os.path.basename(sys.argv[0])))
    print("  {0} -d [\"wordlist...\"]   ==> decode pgpwords to hex".format(os.path.basename(sys.argv[0])))
    print("  {0} -w                 ==> print hex to word map".format(os.path.basename(sys.argv[0])))
    print("  {0} -h                 ==> print word to hex map".format(os.path.basename(sys.argv[0])))
    print("")
    print("If called with multiple arguments, they will be concatenated")
    print("and treated as a single fingerprint.")
    print("")
    print("If called with no arguments, input is read from stdin,")
    print("and each line is treated as a single fingerprint.  In this")
    print("mode, invalid values are silently ignored.")
    exit(1)

if __name__ == '__main__':
    skip=0
    if 1 == len(sys.argv):
        fps = sys.stdin.readlines()
    else:
        fps = [" ".join(sys.argv[1:])]
        try:
            opts, args = getopt.getopt(argv, "whd:", ["decode="])
        except getopt.GetoptError:
            usage()
            sys.exit(2)
        for opt, arg in opts:
            if opt == '-w' or opt == '-h':
                for i,w in enumerate(_words):
                        print("%6s %28s "%(hex(i),str(w))),
                        if (i+1) % 3 ==0:
                            print(" ")
                sys.exit()
            elif opt in ("-d", "--decode"):
                decode = True
                fps = [" ".join(sys.argv[2:])]

    for fp in fps:
        try:
            if decode:
                print("TODO: Implement Decoding")
            else:
                words = wordify(fp)
                print("\n{0}: ".format(fp.strip()))
                sys.stdout.write("\t")
                for i in range(0, len(words)):
                    sys.stdout.write(words[i] + " ")
                    if (not (i+1) % 4) and  not i == len(words)-1:
                        sys.stdout.write("\n\t")
                print("")

        except Exception as e:
            if len(fps) == 1:
                print (e)
                usage()

    print("")