1*2abb3134SXin Li#!/usr/bin/python 2*2abb3134SXin Li# 3*2abb3134SXin Li# Copyright 2014 Google Inc. All rights reserved. 4*2abb3134SXin Li# 5*2abb3134SXin Li# Licensed under the Apache License, Version 2.0 (the "License"); 6*2abb3134SXin Li# you may not use this file except in compliance with the License. 7*2abb3134SXin Li# You may obtain a copy of the License at 8*2abb3134SXin Li# 9*2abb3134SXin Li# http://www.apache.org/licenses/LICENSE-2.0 10*2abb3134SXin Li# 11*2abb3134SXin Li# Unless required by applicable law or agreed to in writing, software 12*2abb3134SXin Li# distributed under the License is distributed on an "AS IS" BASIS, 13*2abb3134SXin Li# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14*2abb3134SXin Li# See the License for the specific language governing permissions and 15*2abb3134SXin Li# limitations under the License. 16*2abb3134SXin Li 17*2abb3134SXin Li""" 18*2abb3134SXin LiGiven a list of candidates on stdin, produce a file of hashes ("map file"). 19*2abb3134SXin Li""" 20*2abb3134SXin Li 21*2abb3134SXin Liimport csv 22*2abb3134SXin Liimport sys 23*2abb3134SXin Li 24*2abb3134SXin Liimport rappor 25*2abb3134SXin Li 26*2abb3134SXin Li 27*2abb3134SXin Lidef HashCandidates(params, stdin, stdout): 28*2abb3134SXin Li num_bloombits = params.num_bloombits 29*2abb3134SXin Li csv_out = csv.writer(stdout) 30*2abb3134SXin Li 31*2abb3134SXin Li for line in stdin: 32*2abb3134SXin Li word = line.strip() 33*2abb3134SXin Li row = [word] 34*2abb3134SXin Li for cohort in xrange(params.num_cohorts): 35*2abb3134SXin Li bloom_bits = rappor.get_bloom_bits(word, cohort, params.num_hashes, 36*2abb3134SXin Li num_bloombits) 37*2abb3134SXin Li for bit_to_set in bloom_bits: 38*2abb3134SXin Li # bits are indexed from 1. Add a fixed offset for each cohort. 39*2abb3134SXin Li # NOTE: This detail could be omitted from the map file format, and done 40*2abb3134SXin Li # in R. 41*2abb3134SXin Li row.append(cohort * num_bloombits + (bit_to_set + 1)) 42*2abb3134SXin Li csv_out.writerow(row) 43*2abb3134SXin Li 44*2abb3134SXin Li 45*2abb3134SXin Lidef main(argv): 46*2abb3134SXin Li try: 47*2abb3134SXin Li filename = argv[1] 48*2abb3134SXin Li except IndexError: 49*2abb3134SXin Li raise RuntimeError('Usage: hash_candidates.py <params file>') 50*2abb3134SXin Li with open(filename) as f: 51*2abb3134SXin Li try: 52*2abb3134SXin Li params = rappor.Params.from_csv(f) 53*2abb3134SXin Li except rappor.Error as e: 54*2abb3134SXin Li raise RuntimeError(e) 55*2abb3134SXin Li 56*2abb3134SXin Li HashCandidates(params, sys.stdin, sys.stdout) 57*2abb3134SXin Li 58*2abb3134SXin Li 59*2abb3134SXin Liif __name__ == '__main__': 60*2abb3134SXin Li try: 61*2abb3134SXin Li main(sys.argv) 62*2abb3134SXin Li except RuntimeError, e: 63*2abb3134SXin Li print >>sys.stderr, e.args[0] 64*2abb3134SXin Li sys.exit(1) 65