#!/usr/bin/env python

import sys
import matplotlib
matplotlib.use('Agg')
import pylab
import re
import datetime

if len(sys.argv) <= 1:
    print "Usage: ./makegraph.py <file1> <file2> <file3> ..."
    print "(where the files are the output of adddates.py"

files = sys.argv[1:]
data = []
for n in files:
    f = open(n, 'r')
    data.append(f.read())
    f.close()

class CVE(object):
    def toDate(self, str):
        # Discard the time portion
        if str == 'None':
            return None
        date = str.split(' ')[0]
        y,m,d = [int(x) for x in date.split('-')]
        return datetime.date(y,m,d)
    def __init__(self, name, assign, commit, public, source):
        self.name = name
        self.commit = commit
        self.source = source
        if self.commit == '':
            print >>sys.stderr, "%s is missing a git commit!" % self.name
        if self.source == '':
            print >>sys.stderr, "%s is missing a source for its public disclosure date!" % self.name

        self.assign = self.toDate(assign)
        self.public = self.toDate(public)
    def __cmp__(self, other):
        return cmp(self.name, other.name)

# Strip out comments, break into entries
cves = []
for d in data:
    d2 = re.sub(r'\#.*\n', r'', d)
    d3 = re.sub(r'\n{2,}', r'\n\n', d2)
    for entry in d3.strip().split('\n\n'):
        lines = entry.split("\n")
        name = lines[0].strip()
        assignstr = lines[1].split(':', 1)[1].strip()
        commit = lines[2].split(':', 1)[1].strip()
        publicstr = lines[3].split(':', 1)[1].strip()
        source = lines[4].split(':', 1)[1].strip()

        if assignstr == '' or assignstr == 'None':
            print "Discarding %s... no assigned date" % name
            continue
        if publicstr == '' or publicstr == 'None':
            print "Discarding %s... no public disclosure date" % name
            continue

        cves.append(CVE(name,
                        assignstr,
                        commit,
                        publicstr,
                        source))

hist = []
vuln = []
for day in range(0, 365*3):
    vuln.append(0)

cves.sort()
before = 0
after = 0
after2wk = 0
after4wk = 0
after8wk = 0
start = datetime.date(2006, 01, 01)
for c in cves:
    diff = c.assign - c.public
    for day in range(0, 365*3):
        if (start.toordinal() + day < c.assign.toordinal()) and \
           (start.toordinal() + day >= c.public.toordinal()):
#            print c.public.toordinal(), start.toordinal() + day, c.assign.toordinal()
            vuln[day] = vuln[day] + 1

#    print c.assign
#    print "%s: Public announcement predates CVE by %s days" % (c.name,diff.days)

#    if diff.days >= 350:
#        hist.append(50)
    if diff.days >= 14:
        hist.append(diff.days / 7)
    if diff.days >= 0:
        after += 1
    else:
        before += 1
    if diff.days >= 14:
        after2wk += 1
    if diff.days >= 28:
        after4wk += 1
    if diff.days >= 56:
        after8wk += 1


#print vuln

print
print before+after, "CVEs processed"
print after, "had their CVEs assigned after public disclosure"
print "after: %0.2f%%" % (float(after)/float(before+after) * 100)
print "2 weeks after: %0.2f%% (%i)" % (float(after2wk)/float(before+after) * 100, after2wk)
print "4 weeks after: %0.2f%% (%i)" % (float(after4wk)/float(before+after) * 100, after4wk)
print "8 weeks after: %0.2f%% (%i)" % (float(after8wk)/float(before+after) * 100, after8wk)
print

LABELFONTSIZE=20
AXISFONTSIZE=20

pylab.figure()
pylab.hist(hist, bins=50)
pylab.xlabel('Number of weeks of impact delay', fontsize=LABELFONTSIZE)
pylab.ylabel('Number of bugs', fontsize=LABELFONTSIZE)
ax = pylab.axes()
for xlabel_i in ax.get_xticklabels():
    xlabel_i.set_fontsize(AXISFONTSIZE)
for ylabel_i in ax.get_yticklabels():
    ylabel_i.set_fontsize(AXISFONTSIZE)
img = 'histograph.png'
pylab.savefig(img)
print "Histogram output to %s" % img

pylab.figure()
pylab.plot(vuln)
pylab.xlabel('Day between January 2006 and December 2008', fontsize=LABELFONTSIZE)
pylab.ylabel('Number of bugs', fontsize=LABELFONTSIZE)
pylab.axis([0, 365*3, 0, 16])
ax = pylab.axes()
for xlabel_i in ax.get_xticklabels():
    xlabel_i.set_fontsize(AXISFONTSIZE)
for ylabel_i in ax.get_yticklabels():
    ylabel_i.set_fontsize(AXISFONTSIZE)
img = 'vuln-window.png'
pylab.savefig(img)
print "Line graph output to %s" % img