"""Unit tests for the bytes and bytearray types.

XXX This is a mess.  Common tests should be moved to buffer_tests.py,

which itself ought to be unified with string_tests.py (and the latter

should be modernized).

"""

import os

import re

import sys

import copy

import pickle

import tempfile

import unittest

import warnings

import test.test_support

import test.string_tests

import test.buffer_tests

class Indexable:

    def __init__(self, value=0):

        self.value = value

    def __index__(self):

        return self.value

class BaseBytesTest(unittest.TestCase):

    def setUp(self):

        self.warning_filters = warnings.filters[:]

    def tearDown(self):

        warnings.filters = self.warning_filters

    def test_basics(self):

        b = self.type2test()

        self.assertEqual(type(b), self.type2test)

        self.assertEqual(b.__class__, self.type2test)

    def test_empty_sequence(self):

        b = self.type2test()

        self.assertEqual(len(b), 0)

        self.assertRaises(IndexError, lambda: b[0])

        self.assertRaises(IndexError, lambda: b[1])

        self.assertRaises(IndexError, lambda: b[sys.maxint])

        self.assertRaises(IndexError, lambda: b[sys.maxint+1])

        self.assertRaises(IndexError, lambda: b[10**100])

        self.assertRaises(IndexError, lambda: b[-1])

        self.assertRaises(IndexError, lambda: b[-2])

        self.assertRaises(IndexError, lambda: b[-sys.maxint])

        self.assertRaises(IndexError, lambda: b[-sys.maxint-1])

        self.assertRaises(IndexError, lambda: b[-sys.maxint-2])

        self.assertRaises(IndexError, lambda: b[-10**100])

    def test_from_list(self):

        ints = list(range(256))

        b = self.type2test(i for i in ints)

        self.assertEqual(len(b), 256)

        self.assertEqual(list(b), ints)

    def test_from_index(self):

        b = self.type2test([Indexable(), Indexable(1), Indexable(254),

                            Indexable(255)])

        self.assertEqual(list(b), [0, 1, 254, 255])

        self.assertRaises(ValueError, bytearray, [Indexable(-1)])

        self.assertRaises(ValueError, bytearray, [Indexable(256)])

    def test_from_ssize(self):

        self.assertEqual(bytearray(0), b'')

        self.assertEqual(bytearray(1), b'\x00')

        self.assertEqual(bytearray(5), b'\x00\x00\x00\x00\x00')

        self.assertRaises(ValueError, bytearray, -1)

        self.assertEqual(bytearray('0', 'ascii'), b'0')

        self.assertEqual(bytearray(b'0'), b'0')

    def test_constructor_type_errors(self):

        self.assertRaises(TypeError, self.type2test, 0.0)

        class C:

            pass

        # allowed in 2.6

        #self.assertRaises(TypeError, self.type2test, ["0"])

        self.assertRaises(TypeError, self.type2test, [0.0])

        self.assertRaises(TypeError, self.type2test, [None])

        self.assertRaises(TypeError, self.type2test, [C()])

    def test_constructor_value_errors(self):

        self.assertRaises(ValueError, self.type2test, [-1])

        self.assertRaises(ValueError, self.type2test, [-sys.maxint])

        self.assertRaises(ValueError, self.type2test, [-sys.maxint-1])

        self.assertRaises(ValueError, self.type2test, [-sys.maxint-2])

        self.assertRaises(ValueError, self.type2test, [-10**100])

        self.assertRaises(ValueError, self.type2test, [256])

        self.assertRaises(ValueError, self.type2test, [257])

        self.assertRaises(ValueError, self.type2test, [sys.maxint])

        self.assertRaises(ValueError, self.type2test, [sys.maxint+1])

        self.assertRaises(ValueError, self.type2test, [10**100])

    def test_compare(self):

        b1 = self.type2test([1, 2, 3])

        b2 = self.type2test([1, 2, 3])

        b3 = self.type2test([1, 3])

        self.assertEqual(b1, b2)

        self.failUnless(b2 != b3)

        self.failUnless(b1 <= b2)

        self.failUnless(b1 <= b3)

        self.failUnless(b1 <  b3)

        self.failUnless(b1 >= b2)

        self.failUnless(b3 >= b2)

        self.failUnless(b3 >  b2)

        self.failIf(b1 != b2)

        self.failIf(b2 == b3)

        self.failIf(b1 >  b2)

        self.failIf(b1 >  b3)

        self.failIf(b1 >= b3)

        self.failIf(b1 <  b2)

        self.failIf(b3 <  b2)

        self.failIf(b3 <= b2)

    def test_compare_to_str(self):

        warnings.simplefilter('ignore', BytesWarning)

        # Byte comparisons with unicode should always fail!

        # Test this for all expected byte orders and Unicode character sizes

        self.assertEqual(self.type2test(b"\0a\0b\0c") == u"abc", False)

        self.assertEqual(self.type2test(b"\0\0\0a\0\0\0b\0\0\0c") == u"abc", False)

        self.assertEqual(self.type2test(b"a\0b\0c\0") == u"abc", False)

        self.assertEqual(self.type2test(b"a\0\0\0b\0\0\0c\0\0\0") == u"abc", False)

        self.assertEqual(self.type2test() == unicode(), False)

        self.assertEqual(self.type2test() != unicode(), True)

    def test_reversed(self):

        input = list(map(ord, "Hello"))

        b = self.type2test(input)

        output = list(reversed(b))

        input.reverse()

        self.assertEqual(output, input)

    def test_getslice(self):

        def by(s):

            return self.type2test(map(ord, s))

        b = by("Hello, world")

        self.assertEqual(b[:5], by("Hello"))

        self.assertEqual(b[1:5], by("ello"))

        self.assertEqual(b[5:7], by(", "))

        self.assertEqual(b[7:], by("world"))

        self.assertEqual(b[7:12], by("world"))

        self.assertEqual(b[7:100], by("world"))

        self.assertEqual(b[:-7], by("Hello"))

        self.assertEqual(b[-11:-7], by("ello"))

        self.assertEqual(b[-7:-5], by(", "))

        self.assertEqual(b[-5:], by("world"))

        self.assertEqual(b[-5:12], by("world"))

        self.assertEqual(b[-5:100], by("world"))

        self.assertEqual(b[-100:5], by("Hello"))

    def test_extended_getslice(self):

        # Test extended slicing by comparing with list slicing.

        L = list(range(255))

        b = self.type2test(L)

        indices = (0, None, 1, 3, 19, 100, -1, -2, -31, -100)

        for start in indices:

            for stop in indices:

                # Skip step 0 (invalid)

                for step in indices[1:]:

                    self.assertEqual(b[start:stop:step], self.type2test(L[start:stop:step]))

    def test_encoding(self):

        sample = u"Hello world\n\u1234\u5678\u9abc\udef0"

        for enc in ("utf8", "utf16"):

            b = self.type2test(sample, enc)

            self.assertEqual(b, self.type2test(sample.encode(enc)))

        self.assertRaises(UnicodeEncodeError, self.type2test, sample, "latin1")

        b = self.type2test(sample, "latin1", "ignore")

        self.assertEqual(b, self.type2test(sample[:-4], "utf-8"))

    def test_decode(self):

        sample = u"Hello world\n\u1234\u5678\u9abc\def0\def0"

        for enc in ("utf8", "utf16"):

            b = self.type2test(sample, enc)

            self.assertEqual(b.decode(enc), sample)

        sample = u"Hello world\n\x80\x81\xfe\xff"

        b = self.type2test(sample, "latin1")

        self.assertRaises(UnicodeDecodeError, b.decode, "utf8")

        self.assertEqual(b.decode("utf8", "ignore"), "Hello world\n")

    def test_from_int(self):

        b = self.type2test(0)

        self.assertEqual(b, self.type2test())

        b = self.type2test(10)

        self.assertEqual(b, self.type2test([0]*10))

        b = self.type2test(10000)

        self.assertEqual(b, self.type2test([0]*10000))

    def test_concat(self):

        b1 = self.type2test(b"abc")

        b2 = self.type2test(b"def")

        self.assertEqual(b1 + b2, b"abcdef")

        self.assertEqual(b1 + bytes(b"def"), b"abcdef")

        self.assertEqual(bytes(b"def") + b1, b"defabc")

        self.assertRaises(TypeError, lambda: b1 + u"def")

        self.assertRaises(TypeError, lambda: u"abc" + b2)

    def test_repeat(self):

        for b in b"abc", self.type2test(b"abc"):

            self.assertEqual(b * 3, b"abcabcabc")

            self.assertEqual(b * 0, b"")

            self.assertEqual(b * -1, b"")

            self.assertRaises(TypeError, lambda: b * 3.14)

            self.assertRaises(TypeError, lambda: 3.14 * b)

            # XXX Shouldn't bytes and bytearray agree on what to raise?

            self.assertRaises((OverflowError, MemoryError),

                              lambda: b * sys.maxsize)

    def test_repeat_1char(self):

        self.assertEqual(self.type2test(b'x')*100, self.type2test([ord('x')]*100))

    def test_contains(self):

        b = self.type2test(b"abc")

        self.failUnless(ord('a') in b)

        self.failUnless(int(ord('a')) in b)

        self.failIf(200 in b)

        self.failIf(200 in b)

        self.assertRaises(ValueError, lambda: 300 in b)

        self.assertRaises(ValueError, lambda: -1 in b)

        self.assertRaises(TypeError, lambda: None in b)

        self.assertRaises(TypeError, lambda: float(ord('a')) in b)

        self.assertRaises(TypeError, lambda: u"a" in b)

        for f in bytes, bytearray:

            self.failUnless(f(b"") in b)

            self.failUnless(f(b"a") in b)

            self.failUnless(f(b"b") in b)

            self.failUnless(f(b"c") in b)

            self.failUnless(f(b"ab") in b)

            self.failUnless(f(b"bc") in b)

            self.failUnless(f(b"abc") in b)

            self.failIf(f(b"ac") in b)

            self.failIf(f(b"d") in b)

            self.failIf(f(b"dab") in b)

            self.failIf(f(b"abd") in b)

    def test_fromhex(self):

        self.assertRaises(TypeError, self.type2test.fromhex)

        self.assertRaises(TypeError, self.type2test.fromhex, 1)

        self.assertEquals(self.type2test.fromhex(u''), self.type2test())

        b = bytearray([0x1a, 0x2b, 0x30])

        self.assertEquals(self.type2test.fromhex(u'1a2B30'), b)

        self.assertEquals(self.type2test.fromhex(u'  1A 2B  30   '), b)

        self.assertEquals(self.type2test.fromhex(u'0000'), b'\0\0')

        self.assertRaises(TypeError, self.type2test.fromhex, b'1B')

        self.assertRaises(ValueError, self.type2test.fromhex, u'a')

        self.assertRaises(ValueError, self.type2test.fromhex, u'rt')

        self.assertRaises(ValueError, self.type2test.fromhex, u'1a b cd')

        self.assertRaises(ValueError, self.type2test.fromhex, u'\x00')

        self.assertRaises(ValueError, self.type2test.fromhex, u'12   \x00   34')

    def test_join(self):

        self.assertEqual(self.type2test(b"").join([]), b"")

        self.assertEqual(self.type2test(b"").join([b""]), b"")

        for lst in [[b"abc"], [b"a", b"bc"], [b"ab", b"c"], [b"a", b"b", b"c"]]:

            lst = list(map(self.type2test, lst))

            self.assertEqual(self.type2test(b"").join(lst), b"abc")

            self.assertEqual(self.type2test(b"").join(tuple(lst)), b"abc")

            self.assertEqual(self.type2test(b"").join(iter(lst)), b"abc")

        self.assertEqual(self.type2test(b".").join([b"ab", b"cd"]), b"ab.cd")

        # XXX more...

    def test_index(self):

        b = self.type2test(b'parrot')

        self.assertEqual(b.index('p'), 0)

        self.assertEqual(b.index('rr'), 2)

        self.assertEqual(b.index('t'), 5)

        self.assertRaises(ValueError, lambda: b.index('w'))

    def test_count(self):

        b = self.type2test(b'mississippi')

        self.assertEqual(b.count(b'i'), 4)

        self.assertEqual(b.count(b'ss'), 2)

        self.assertEqual(b.count(b'w'), 0)

    def test_startswith(self):

        b = self.type2test(b'hello')

        self.assertFalse(self.type2test().startswith(b"anything"))

        self.assertTrue(b.startswith(b"hello"))

        self.assertTrue(b.startswith(b"hel"))

        self.assertTrue(b.startswith(b"h"))

        self.assertFalse(b.startswith(b"hellow"))

        self.assertFalse(b.startswith(b"ha"))

    def test_endswith(self):

        b = self.type2test(b'hello')

        self.assertFalse(bytearray().endswith(b"anything"))

        self.assertTrue(b.endswith(b"hello"))

        self.assertTrue(b.endswith(b"llo"))

        self.assertTrue(b.endswith(b"o"))

        self.assertFalse(b.endswith(b"whello"))

        self.assertFalse(b.endswith(b"no"))

    def test_find(self):

        b = self.type2test(b'mississippi')

        self.assertEqual(b.find(b'ss'), 2)

        self.assertEqual(b.find(b'ss', 3), 5)

        self.assertEqual(b.find(b'ss', 1, 7), 2)

        self.assertEqual(b.find(b'ss', 1, 3), -1)

        self.assertEqual(b.find(b'w'), -1)

        self.assertEqual(b.find(b'mississippian'), -1)

    def test_rfind(self):

        b = self.type2test(b'mississippi')

        self.assertEqual(b.rfind(b'ss'), 5)

        self.assertEqual(b.rfind(b'ss', 3), 5)

        self.assertEqual(b.rfind(b'ss', 0, 6), 2)

        self.assertEqual(b.rfind(b'w'), -1)

        self.assertEqual(b.rfind(b'mississippian'), -1)

    def test_index(self):

        b = self.type2test(b'world')

        self.assertEqual(b.index(b'w'), 0)

        self.assertEqual(b.index(b'orl'), 1)

        self.assertRaises(ValueError, b.index, b'worm')

        self.assertRaises(ValueError, b.index, b'ldo')

    def test_rindex(self):

        # XXX could be more rigorous

        b = self.type2test(b'world')

        self.assertEqual(b.rindex(b'w'), 0)

        self.assertEqual(b.rindex(b'orl'), 1)

        self.assertRaises(ValueError, b.rindex, b'worm')

        self.assertRaises(ValueError, b.rindex, b'ldo')

    def test_replace(self):

        b = self.type2test(b'mississippi')

        self.assertEqual(b.replace(b'i', b'a'), b'massassappa')

        self.assertEqual(b.replace(b'ss', b'x'), b'mixixippi')

    def test_split(self):

        b = self.type2test(b'mississippi')

        self.assertEqual(b.split(b'i'), [b'm', b'ss', b'ss', b'pp', b''])

        self.assertEqual(b.split(b'ss'), [b'mi', b'i', b'ippi'])

        self.assertEqual(b.split(b'w'), [b])

    def test_split_whitespace(self):

        for b in (b'  arf  barf  ', b'arf\tbarf', b'arf\nbarf', b'arf\rbarf',

                  b'arf\fbarf', b'arf\vbarf'):

            b = self.type2test(b)

            self.assertEqual(b.split(), [b'arf', b'barf'])

            self.assertEqual(b.split(None), [b'arf', b'barf'])

            self.assertEqual(b.split(None, 2), [b'arf', b'barf'])

        for b in (b'a\x1Cb', b'a\x1Db', b'a\x1Eb', b'a\x1Fb'):

            b = self.type2test(b)

            self.assertEqual(b.split(), [b])

        self.assertEqual(self.type2test(b'  a  bb  c  ').split(None, 0), [b'a  bb  c  '])

        self.assertEqual(self.type2test(b'  a  bb  c  ').split(None, 1), [b'a', b'bb  c  '])

        self.assertEqual(self.type2test(b'  a  bb  c  ').split(None, 2), [b'a', b'bb', b'c  '])

        self.assertEqual(self.type2test(b'  a  bb  c  ').split(None, 3), [b'a', b'bb', b'c'])

    def test_split_string_error(self):

        self.assertRaises(TypeError, self.type2test(b'a b').split, u' ')

    def test_rsplit(self):

        b = self.type2test(b'mississippi')

        self.assertEqual(b.rsplit(b'i'), [b'm', b'ss', b'ss', b'pp', b''])

        self.assertEqual(b.rsplit(b'ss'), [b'mi', b'i', b'ippi'])

        self.assertEqual(b.rsplit(b'w'), [b])

    def test_rsplit_whitespace(self):

        for b in (b'  arf  barf  ', b'arf\tbarf', b'arf\nbarf', b'arf\rbarf',

                  b'arf\fbarf', b'arf\vbarf'):

            b = self.type2test(b)

            self.assertEqual(b.rsplit(), [b'arf', b'barf'])

            self.assertEqual(b.rsplit(None), [b'arf', b'barf'])

            self.assertEqual(b.rsplit(None, 2), [b'arf', b'barf'])

        self.assertEqual(self.type2test(b'  a  bb  c  ').rsplit(None, 0), [b'  a  bb  c'])

        self.assertEqual(self.type2test(b'  a  bb  c  ').rsplit(None, 1), [b'  a  bb', b'c'])

        self.assertEqual(self.type2test(b'  a  bb  c  ').rsplit(None, 2), [b'  a', b'bb', b'c'])

        self.assertEqual(self.type2test(b'  a  bb  c  ').rsplit(None, 3), [b'a', b'bb', b'c'])

    def test_rsplit_string_error(self):

        self.assertRaises(TypeError, self.type2test(b'a b').rsplit, u' ')

    def test_rsplit_unicodewhitespace(self):

        b = self.type2test(b"\x09\x0A\x0B\x0C\x0D\x1C\x1D\x1E\x1F")

        self.assertEqual(b.split(), [b'\x1c\x1d\x1e\x1f'])

        self.assertEqual(b.rsplit(), [b'\x1c\x1d\x1e\x1f'])

    def test_partition(self):

        b = self.type2test(b'mississippi')

        self.assertEqual(b.partition(b'ss'), (b'mi', b'ss', b'issippi'))

        self.assertEqual(b.rpartition(b'w'), (b'', b'', b'mississippi'))

    def test_rpartition(self):

        b = self.type2test(b'mississippi')

        self.assertEqual(b.rpartition(b'ss'), (b'missi', b'ss', b'ippi'))

        self.assertEqual(b.rpartition(b'i'), (b'mississipp', b'i', b''))

    def test_pickling(self):

        for proto in range(pickle.HIGHEST_PROTOCOL + 1):

            for b in b"", b"a", b"abc", b"\xffab\x80", b"\0\0\377\0\0":

                b = self.type2test(b)

                ps = pickle.dumps(b, proto)

                q = pickle.loads(ps)

                self.assertEqual(b, q)

    def test_strip(self):

        b = self.type2test(b'mississippi')

        self.assertEqual(b.strip(b'i'), b'mississipp')

        self.assertEqual(b.strip(b'm'), b'ississippi')

        self.assertEqual(b.strip(b'pi'), b'mississ')

        self.assertEqual(b.strip(b'im'), b'ssissipp')

        self.assertEqual(b.strip(b'pim'), b'ssiss')

        self.assertEqual(b.strip(b), b'')

    def test_lstrip(self):

        b = self.type2test(b'mississippi')

        self.assertEqual(b.lstrip(b'i'), b'mississippi')

        self.assertEqual(b.lstrip(b'm'), b'ississippi')

        self.assertEqual(b.lstrip(b'pi'), b'mississippi')

        self.assertEqual(b.lstrip(b'im'), b'ssissippi')

        self.assertEqual(b.lstrip(b'pim'), b'ssissippi')

    def test_rstrip(self):

        b = self.type2test(b'mississippi')

        self.assertEqual(b.rstrip(b'i'), b'mississipp')

        self.assertEqual(b.rstrip(b'm'), b'mississippi')

        self.assertEqual(b.rstrip(b'pi'), b'mississ')

        self.assertEqual(b.rstrip(b'im'), b'mississipp')

        self.assertEqual(b.rstrip(b'pim'), b'mississ')

    def test_strip_whitespace(self):

        b = self.type2test(b' \t\n\r\f\vabc \t\n\r\f\v')

        self.assertEqual(b.strip(), b'abc')

        self.assertEqual(b.lstrip(), b'abc \t\n\r\f\v')

        self.assertEqual(b.rstrip(), b' \t\n\r\f\vabc')

    def XXXtest_strip_bytearray(self):

        # XXX memoryview not available

        self.assertEqual(self.type2test(b'abc').strip(memoryview(b'ac')), b'b')

        self.assertEqual(self.type2test(b'abc').lstrip(memoryview(b'ac')), b'bc')

        self.assertEqual(self.type2test(b'abc').rstrip(memoryview(b'ac')), b'ab')

    def test_strip_string_error(self):

        self.assertRaises(TypeError, self.type2test(b'abc').strip, u'b')

        self.assertRaises(TypeError, self.type2test(b'abc').lstrip, u'b')

        self.assertRaises(TypeError, self.type2test(b'abc').rstrip, u'b')

    def test_ord(self):

        b = self.type2test(b'\0A\x7f\x80\xff')

        self.assertEqual([ord(b[i:i+1]) for i in range(len(b))],

                         [0, 65, 127, 128, 255])

class ByteArrayTest(BaseBytesTest):

    type2test = bytearray

    def test_nohash(self):

        self.assertRaises(TypeError, hash, bytearray())

    def test_bytearray_api(self):

        short_sample = b"Hello world\n"

        sample = short_sample + b"\0"*(20 - len(short_sample))

        tfn = tempfile.mktemp()

        try:

            # Prepare

            with open(tfn, "wb") as f:

                f.write(short_sample)

            # Test readinto

            with open(tfn, "rb") as f:

                b = bytearray(20)

                n = f.readinto(b)

            self.assertEqual(n, len(short_sample))

            # Python 2.x

            b_sample = (ord(s) for s in sample)

            self.assertEqual(list(b), list(b_sample))

            # Test writing in binary mode

            with open(tfn, "wb") as f:

                f.write(b)

            with open(tfn, "rb") as f:

                self.assertEqual(f.read(), sample)

            # Text mode is ambiguous; don't test

        finally:

            try:

                os.remove(tfn)

            except os.error:

                pass

    def test_reverse(self):

        b = bytearray(b'hello')

        self.assertEqual(b.reverse(), None)

        self.assertEqual(b, b'olleh')

        b = bytearray(b'hello1') # test even number of items

        b.reverse()

        self.assertEqual(b, b'1olleh')

        b = bytearray()

        b.reverse()

        self.assertFalse(b)

    def test_regexps(self):

        def by(s):

            return bytearray(map(ord, s))

        b = by("Hello, world")

        self.assertEqual(re.findall(r"\w+", b), [by("Hello"), by("world")])

    def test_setitem(self):

        b = bytearray([1, 2, 3])

        b[1] = 100

        self.assertEqual(b, bytearray([1, 100, 3]))

        b[-1] = 200

        self.assertEqual(b, bytearray([1, 100, 200]))

        b[0] = Indexable(10)

        self.assertEqual(b, bytearray([10, 100, 200]))

        try:

            b[3] = 0

            self.fail("Didn't raise IndexError")

        except IndexError:

            pass

        try:

            b[-10] = 0

            self.fail("Didn't raise IndexError")

        except IndexError:

            pass

        try:

            b[0] = 256

            self.fail("Didn't raise ValueError")

        except ValueError:

            pass

        try: