# HG changeset patch
# User Bernhard Reiter <bernhard@intevation.de>
# Date 1455892126 -3600
# Node ID 7558ecd1cbf1f947930917e7dccc9f89186fd955

Initial version.

diff -r 000000000000 -r 7558ecd1cbf1 ppgen.py
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/ppgen.py	Fri Feb 19 15:28:46 2016 +0100
@@ -0,0 +1,138 @@
+#!/usr/bin/env python3
+"""Create a passphrase from a few random words. DRAFT
+
+Relies on the entropy of python's
+    random.SystemRandom class
+        which (according to the documentation) calls os.urandom()
+        which (according to the documentation) calls the operating system
+           specific randomness source which "should be unpredictable 
+            enough for cryptographic applications"
+
+Requires:
+  * Python v>=3.2
+  * a dictionary, Ding's trans-de-en by default.
+    E.g. on a Debian/Ubuntu system in package "trans-de-en".
+    or from http://ftp.tu-chemnitz.de/pub/Local/urz/ding/de-en/
+
+Uses a hardcodes filepath and selected language.
+Search for **customize** below to change it.
+
+Copyright 2016 by Intevation GmbH.
+Author: 2016-01-21 Bernhard E. Reiter <bernhard@intevation.de>
+
+This file is Free Software under the Apache 2.0 license and thus
+comes without any warranty (to extend permissible under applicable law).
+"""
+
+import math
+import re
+import sys
+
+from random import SystemRandom
+_srandom = SystemRandom()
+
+tainted = False   # to be set if we find a hint that the passphrase may be weak
+
+def buildDictionary():
+    """Build up a dictionary of unique words, calculate stats."""
+    global tainted
+    d = []
+
+    # dictionary for testing
+    #d = ["abc", "aBc", "cde", "efg", "hij", "blubber", "jikf", "zug", "lmf", "opq"]
+
+    # Using the dictionary from Ding **customize**
+    d = readDingDict(filename="/usr/share/trans/de-en", useLeft=True)
+
+    ## for debugging purpuses, dump dictionary
+    #dumpfilename = "ddump.txt"
+    #print("Writing out {}.".format(dumpfilename))
+    #with open(dumpfilename, "w") as f:
+    #    for i in d:
+    #        f.write("{}\n".format(i))
+
+    # Print some stats on the dictionary to be used
+    dl = len(d)
+    print("Found {:d} dictionary entries".format(dl))
+    if dl < 2000:
+        print("!Your dictionary is below 2k entries, that is quite small!")
+        tainted = True
+
+    print("|= Number of words |= possibilities |")
+    for i in range(1,5):
+        print("|               {:2d} |    2^{:4.1f}      |".format(
+                                    i, math.log(dl**i,2)))
+    return d
+
+
+def readDingDict(filename = "/usr/share/trans/de-en", useLeft=False):
+    """Read dictionary with unique words from file in Ding format.
+
+    useLeft: Boolean to control which language to use
+
+    TODO: add option to use both languages for people that speak them both?
+    """
+
+    dset = set() #using the datatype 'set' to aviod duplicates
+
+    splitter = re.compile(r"""\ \|\  # first pattern  ' | '
+                           |;\       # second pattern '; '
+                           |(?<=\S)/(?=\S)  # 3.:    '\' surrounded by chars
+                           |\s+      # by whitespace
+                           """,re.VERBOSE)
+
+    print("Reading entries from {}.".format(filename), end='')
+    counter = 0 # for progress or stopping early
+    with open(filename, "r") as f:
+        for line in f:
+            if line[0] == '#': continue
+
+            # languages are separated by " :: "
+            p = line.partition(" :: ")
+            languageEntry = p[0] if useLeft else p[2]
+
+            for word in splitter.split(languageEntry):
+                word = word.strip('(",.)\'!:;').rstrip('/')
+                if len(word) > 2 and not word[0] in '[{/':
+                    dset.add(word)
+
+            #TODO: check for very common words and remove them?
+
+            counter += 1
+            ## stop early when debugging
+            #if counter > 10: break
+            if not counter % 10000:
+                    print('.', end='')
+                    sys.stdout.flush()
+        print()
+
+    return list(dset)
+
+def main():
+    global tainted
+    dictionary = buildDictionary()
+
+    howMany = 4
+
+    # use a dictionary with lower case words for a simple check if
+    # our random source is okay
+    print("\nGenerated passphrase with {} randomly selected words:\n".format(
+          howMany))
+    print("    ", end='')
+    words = {}
+    for x in range(howMany):
+        word = _srandom.choice(dictionary)
+        words[word.lower]= True
+        print(word, end='\n    ')
+    print("\n")
+
+    if len(words) < howMany:
+        print("! Your random generator is weak")
+        print("! or you are being very lucky.")
+        tainted = True
+
+    if tainted:
+        print("!!! Don't use the resulting passphrase !!!")
+
+if __name__ == "__main__":
+    main()