#!/usr/bin/python3
import math
from math import ceil
import matplotlib.pyplot as plt
from colors import *
from padme import *
import matplotlib
import utils as utils
utils.prepare_for_latex()
def getNextPowerOfTwo(x):
exp = math.ceil(math.log(x, 2))
return math.pow(2, exp)
xVals = []
padMeEffectiveSizes = []
pow2EffectiveSizes = []
L = 10
end = 1000 * 1000
while L < end:
a = 100 * float(getPadme(L) - L) / L
b = 100 * float(getNextPowerOfTwo(L) - L) / L
#print(L, a, b)
def createFigDatasetAnonymityCDF(i, input_file):
plt.figure(i)
utils.prepare_for_latex()
Ls = [] # the raw L's values
Bs = []
## Parse the input file, fills Ls, Ls_count, Ls_hist
with open(input_file, "r") as f:
for line in f:
L = int(line.strip())
if L == 0 or L == 1:
continue
Ls.append(L)
Bs = [getPadme(L) for L in Ls]
Ps = [math.pow(2, math.ceil(math.log(L, 2))) for L in Ls]
Qs = [L + (L % 256) for L in Ls]
Qs2 = [L + (L % (512 * 8)) for L in Ls]
legends = []
data = computeDensity(Ls)
p = 100. * np.arange(len(data)) / (len(data) - 1)
plt.plot(data,
p,
color=unpadded_color,
linestyle=unpadded_style,
linewidth=curve_width)
legends.append('Unpadded')
#data = computeDensity(Qs)
#p = 100. * np.arange(len(data)) / (len(data) - 1)
#plt.plot(data, p, color=blockcipher_color, linestyle=blockcipher_style, linewidth=curve_width)
#legends.append('Block cipher (256b)')
data = computeDensity(Qs2)
p = 100. * np.arange(len(data)) / (len(data) - 1)
plt.plot(data,
p,
color=blockcipher_color,
linestyle=blockcipher_style,
linewidth=curve_width)
legends.append('Tor cell (512B)')
data = computeDensity(Bs)
p = 100. * np.arange(len(data)) / (len(data) - 1)
plt.plot(data,
p,
color=padme_color,
linestyle=padme_style,
linewidth=curve_width)
legends.append('Padmé')
data = computeDensity(Ps)
p = 100. * np.arange(len(data)) / (len(data) - 1)
plt.plot(data,
p,
color=nextpow2_color,
linestyle=nextpow2_style,
linewidth=curve_width)
legends.append('Next power of 2')
#plt.title('CDF of anonymity in the dataset '+input_file)
plt.title('')
plt.ylabel('Percentile')
plt.xlabel('Anonymity set size')
plt.legend(legends, loc='lower right')
plt.xscale('log')
#plt.yscale('log')
plt.grid(color=grid_color, linestyle=grid_style, linewidth=grid_width)
plt.tight_layout()
dataset = input_file.replace('./', '').replace('.sizes', '')
plt.savefig('fig4-' + str(i) + '-' + dataset + '-anonymity-cdf.eps')
def plot(j, input_file):
plt.figure(j)
utils.prepare_for_latex()
Ls = [] # the raw L's values
Ls_count = {
} # a dictionary [packet size] -> [number of packets with this size]
with open(input_file, "r") as f:
for line in f:
L = int(line.strip())
if L == 0 or L == 1:
continue
Ls.append(L)
if str(L) not in Ls_count:
Ls_count[str(L)] = 0
Ls_count[str(L)] += 1
Bs = [getPadme(L) for L in Ls]
Ps = [int(math.pow(2, math.ceil(math.log(L, 2)))) for L in Ls]
Bs_count = {}
Ps_count = {}
for B in Bs:
if str(B) not in Bs_count:
Bs_count[str(B)] = 0
Bs_count[str(B)] += 1
for P in Ps:
if str(P) not in Ps_count:
Ps_count[str(P)] = 0
Ps_count[str(P)] += 1
Ls_buckets = bucketize(Ls_count)
Bs_buckets = bucketize(Bs_count)
Ps_buckets = bucketize(Ps_count)
minLsKey = min([int(x) for x in [*Ls_buckets.keys()]])
minBsKey = min([int(x) for x in [*Bs_buckets.keys()]])
minPsKey = min([int(x) for x in [*Ps_buckets.keys()]])
minKey = min(minLsKey, min(minBsKey, minPsKey))
maxLsKey = max([int(x) for x in [*Ls_buckets.keys()]])
maxBsKey = max([int(x) for x in [*Bs_buckets.keys()]])
maxPsKey = max([int(x) for x in [*Ps_buckets.keys()]])
maxKey = max(maxLsKey, max(maxBsKey, maxPsKey))
Ls_percentageUniquePackets = extractCurves(Ls_buckets, minKey, maxKey)
Bs_percentageUniquePackets = extractCurves(Bs_buckets, minKey, maxKey)
Ps_percentageUniquePackets = extractCurves(Ps_buckets, minKey, maxKey)
legends = []
xValues = []
i = minLsKey
while i <= maxKey:
xValues.append(math.pow(2, i))
i += 1
plt.plot(xValues,
Ls_percentageUniquePackets,
color=unpadded_color,
linestyle=unpadded_style,
linewidth=curve_width)
legends.append('Unpadded')
plt.plot(xValues,
Bs_percentageUniquePackets,
color=padme_color,
linestyle=padme_style,
linewidth=curve_width)
legends.append('Padmé')
plt.plot(xValues,
Ps_percentageUniquePackets,
color=nextpow2_color,
linestyle=nextpow2_style,
linewidth=curve_width)
legends.append('Pow2')
#plt.title('Anonymity set w.r.t size, '+input_file)
plt.xlabel('File sizes [bits]')
plt.ylabel('Percentage of unique objects [\\%]')
plt.legend(legends, loc='upper left')
plt.xscale('log', basex=10)
#plt.yscale('log')
plt.grid(color=grid_color, linestyle=grid_style, linewidth=grid_width)
plt.tight_layout()
dataset = input_file.replace('./', '').replace('.sizes', '')
plt.savefig('fig5-' + str(j) + '-' + dataset +
'-sizes-vs-anon-percentage.eps')
tags = build_tags(args)
save_tags(tags, args.json)
elif args.action == 'parse':
if args.json:
tags = load_tags(args.json)
else:
tags = build_tags(args)
sents = get_corpus(args.lexicon, 'unparsed', args.num_sents)
parses = parse_sents(tags, sents, args.words)
pretty_print(sents, parses)
elif args.action == 'analyze':
test_data = [(filename, load_tags(filename)) for filename in args.test]
truth_tags = load_tags(args.truth)
parsed_sents = get_corpus(args.lexicon, 'parsed', args.num_sents)
unparsed_sents = get_corpus(args.lexicon, 'unparsed', args.num_sents)
(fields, truth_results,
test_results) = analyze(parsed_sents, unparsed_sents, truth_tags,
test_data, args.words)
prepare_for_latex(args.truh, truth_results, test_results)
else:
ValueError(f'unrecognized command "{args.action}" (get)')
exit(1)