#!/usr/bin/env python2.4
"""Sets vs. Regexp! Who will win?"""

import re, random, time

ints = (0x20, 0x21) + tuple(xrange(0x23, 0x5B)) + tuple(xrange(0x5D, 0x7E))
bytes = frozenset([chr(i) for i in ints])
r_safe = re.compile(r'[\x20\x21\x23-\x5B\x5D-\x7E]+')

teststring = ''
for i in xrange(50000): 
   rand = random.random()
   if rand < 0.95: 
      teststring += random.choice(tuple(bytes))
   else: teststring += random.choice('\x02\t\n\r"\\\x80\x8f\xff')

print 'teststring[:50]', teststring[:50]
print 'teststring[50:100]', teststring[50:100]

t = time.clock()
safe = 0
for c in teststring: 
   if c in bytes: safe += 1
print "sets: Found %s safe in %s seconds" % (safe, time.clock() - t)

t = time.clock()
safe = len(''.join(r_safe.findall(teststring)))
print "regexp: Found %s safe in %s seconds" % (safe, time.clock() - t)

r_safe = re.compile(r'[\x20\x21\x23-\x5B\x5D-\x7E]')
t = time.clock()
safe = 0
for c in teststring: 
   if r_safe.match(c): safe += 1
print "regexpiter: Found %s safe in %s seconds" % (safe, time.clock() - t)