Lib/test/test_bytes.py

"""Unit tests for the bytes type."""

import os
import re
import sys
import tempfile
import unittest
import test.test_support


class BytesTest(unittest.TestCase):

    def test_basics(self):
        b = bytes()
        self.assertEqual(type(b), bytes)
        self.assertEqual(b.__class__, bytes)

    def test_empty_sequence(self):
        b = bytes()
        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 = bytes(i for i in ints)
        self.assertEqual(len(b), 256)
        self.assertEqual(list(b), ints)

    def test_from_index(self):
        class C:
            def __init__(self, i=0):
                self.i = i
            def __index__(self):
                return self.i
        b = bytes([C(), C(1), C(254), C(255)])
        self.assertEqual(list(b), [0, 1, 254, 255])
        self.assertRaises(ValueError, bytes, [C(-1)])
        self.assertRaises(ValueError, bytes, [C(256)])

    def test_constructor_type_errors(self):
        self.assertRaises(TypeError, bytes, 0.0)
        class C:
            pass
        self.assertRaises(TypeError, bytes, ["0"])
        self.assertRaises(TypeError, bytes, [0.0])
        self.assertRaises(TypeError, bytes, [None])
        self.assertRaises(TypeError, bytes, [C()])

    def test_constructor_value_errors(self):
        self.assertRaises(ValueError, bytes, [-1])
        self.assertRaises(ValueError, bytes, [-sys.maxint])
        self.assertRaises(ValueError, bytes, [-sys.maxint-1])
        self.assertRaises(ValueError, bytes, [-sys.maxint-2])
        self.assertRaises(ValueError, bytes, [-10**100])
        self.assertRaises(ValueError, bytes, [256])
        self.assertRaises(ValueError, bytes, [257])
        self.assertRaises(ValueError, bytes, [sys.maxint])
        self.assertRaises(ValueError, bytes, [sys.maxint+1])
        self.assertRaises(ValueError, bytes, [10**100])

    def test_repr(self):
        self.assertEqual(repr(bytes()), "b''")
        self.assertEqual(repr(bytes([0])), "b'\\0'")
        self.assertEqual(repr(bytes([0, 1, 254, 255])), "b'\\0\\x01\\xfe\\xff'")
        self.assertEqual(repr(bytes('abc')), "b'abc'")
        self.assertEqual(repr(bytes("'")), "b'\\''")

    def test_compare(self):
        b1 = bytes([1, 2, 3])
        b2 = bytes([1, 2, 3])
        b3 = bytes([1, 3])

        self.failUnless(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_nohash(self):
        self.assertRaises(TypeError, hash, bytes())

    def test_doc(self):
        self.failUnless(bytes.__doc__ != None)
        self.failUnless(bytes.__doc__.startswith("bytes("))

    def test_buffer_api(self):
        short_sample = "Hello world\n"
        sample = short_sample + "x"*(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 = bytes([ord('x')]*20)
                n = f.readinto(b)
            self.assertEqual(n, len(short_sample))
            self.assertEqual(list(b), map(ord, 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)
            # Test writing in text mode
            with open(tfn, "w") as f:
                f.write(b)
            with open(tfn, "r") as f:
                self.assertEqual(f.read(), sample)
            # Can't use readinto in text mode
            with open(tfn, "r") as f:
                self.assertRaises(TypeError, f.readinto, b)
        finally:
            try:
                os.remove(tfn)
            except os.error:
                pass

    def test_reversed(self):
        input = map(ord, "Hello")
        b = bytes(input)
        output = list(reversed(b))
        input.reverse()
        self.assertEqual(output, input)

    def test_getslice(self):
        def by(s):
            return bytes(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 = bytes(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], bytes(L[start:stop:step]))
        
    def test_regexps(self):
        def by(s):
            return bytes(map(ord, s))
        b = by("Hello, world")
        self.assertEqual(re.findall(r"\w+", b), [by("Hello"), by("world")])

    def test_setitem(self):
        b = bytes([1, 2, 3])
        b[1] = 100
        self.assertEqual(b, bytes([1, 100, 3]))
        b[-1] = 200
        self.assertEqual(b, bytes([1, 100, 200]))
        class C:
            def __init__(self, i=0):
                self.i = i
            def __index__(self):
                return self.i
        b[0] = C(10)
        self.assertEqual(b, bytes([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:
            b[0] = C(-1)
            self.fail("Didn't raise ValueError")
        except ValueError:
            pass
        try:
            b[0] = None
            self.fail("Didn't raise TypeError")
        except TypeError:
            pass

    def test_delitem(self):
        b = bytes(range(10))
        del b[0]
        self.assertEqual(b, bytes(range(1, 10)))
        del b[-1]
        self.assertEqual(b, bytes(range(1, 9)))
        del b[4]
        self.assertEqual(b, bytes([1, 2, 3, 4, 6, 7, 8]))

    def test_setslice(self):
        b = bytes(range(10))
        self.assertEqual(list(b), list(range(10)))

        b[0:5] = bytes([1, 1, 1, 1, 1])
        self.assertEqual(b, bytes([1, 1, 1, 1, 1, 5, 6, 7, 8, 9]))

        del b[0:-5]
        self.assertEqual(b, bytes([5, 6, 7, 8, 9]))

        b[0:0] = bytes([0, 1, 2, 3, 4])
        self.assertEqual(b, bytes(range(10)))

        b[-7:-3] = bytes([100, 101])
        self.assertEqual(b, bytes([0, 1, 2, 100, 101, 7, 8, 9]))

        b[3:5] = [3, 4, 5, 6]
        self.assertEqual(b, bytes(range(10)))
        
        b[3:0] = [42, 42, 42]
        self.assertEqual(b, bytes([0, 1, 2, 42, 42, 42, 3, 4, 5, 6, 7, 8, 9]))

    def test_extended_set_del_slice(self):
        indices = (0, None, 1, 3, 19, 300, -1, -2, -31, -300)
        for start in indices:
            for stop in indices:
                # Skip invalid step 0
                for step in indices[1:]:
                    L = list(range(255))
                    b = bytes(L)
                    # Make sure we have a slice of exactly the right length,
                    # but with different data.
                    data = L[start:stop:step]
                    data.reverse()
                    L[start:stop:step] = data
                    b[start:stop:step] = data
                    self.assertEquals(b, bytes(L))
                    
                    del L[start:stop:step]
                    del b[start:stop:step]
                    self.assertEquals(b, bytes(L))

    def test_setslice_trap(self):
        # This test verifies that we correctly handle assigning self
        # to a slice of self (the old Lambert Meertens trap).
        b = bytes(range(256))
        b[8:] = b
        self.assertEqual(b, bytes(list(range(8)) + list(range(256))))

    def test_encoding(self):
        sample = u"Hello world\n\u1234\u5678\u9abc\udef0"
        for enc in ("utf8", "utf16"):
            b = bytes(sample, enc)
            self.assertEqual(b, bytes(map(ord, sample.encode(enc))))
        self.assertRaises(UnicodeEncodeError, bytes, sample, "latin1")
        b = bytes(sample, "latin1", "ignore")
        self.assertEqual(b, bytes(sample[:-4]))

    def test_decode(self):
        sample = u"Hello world\n\u1234\u5678\u9abc\def0\def0"
        for enc in ("utf8", "utf16"):
            b = bytes(sample, enc)
            self.assertEqual(b.decode(enc), sample)
        sample = u"Hello world\n\x80\x81\xfe\xff"
        b = bytes(sample, "latin1")
        self.assertRaises(UnicodeDecodeError, b.decode, "utf8")
        self.assertEqual(b.decode("utf8", "ignore"), "Hello world\n")

    def test_from_buffer(self):
        sample = "Hello world\n\x80\x81\xfe\xff"
        buf = buffer(sample)
        b = bytes(buf)
        self.assertEqual(b, bytes(map(ord, sample)))

    def test_to_str(self):
        sample = "Hello world\n\x80\x81\xfe\xff"
        b = bytes(sample)
        self.assertEqual(str(b), sample)

    def test_from_int(self):
        b = bytes(0)
        self.assertEqual(b, bytes())
        b = bytes(10)
        self.assertEqual(b, bytes([0]*10))
        b = bytes(10000)
        self.assertEqual(b, bytes([0]*10000))

    def test_concat(self):
        b1 = bytes("abc")
        b2 = bytes("def")
        self.assertEqual(b1 + b2, bytes("abcdef"))
        self.assertRaises(TypeError, lambda: b1 + "def")
        self.assertRaises(TypeError, lambda: "abc" + b2)

    def test_repeat(self):
        b = bytes("abc")
        self.assertEqual(b * 3, bytes("abcabcabc"))
        self.assertEqual(b * 0, bytes())
        self.assertEqual(b * -1, bytes())
        self.assertRaises(TypeError, lambda: b * 3.14)
        self.assertRaises(TypeError, lambda: 3.14 * b)
        self.assertRaises(MemoryError, lambda: b * sys.maxint)

    def test_repeat_1char(self):
        self.assertEqual(bytes('x')*100, bytes('x'*100))

    def test_iconcat(self):
        b = bytes("abc")
        b1 = b
        b += bytes("def")
        self.assertEqual(b, bytes("abcdef"))
        self.assertEqual(b, b1)
        self.failUnless(b is b1)

    def test_irepeat(self):
        b = bytes("abc")
        b1 = b
        b *= 3
        self.assertEqual(b, bytes("abcabcabc"))
        self.assertEqual(b, b1)
        self.failUnless(b is b1)

    def test_irepeat_1char(self):
        b = bytes("x")
        b1 = b
        b *= 100
        self.assertEqual(b, bytes("x"*100))
        self.assertEqual(b, b1)
        self.failUnless(b is b1)

    def test_contains(self):
        b = bytes("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: "a" in b)
        self.failUnless(bytes("") in b)
        self.failUnless(bytes("a") in b)
        self.failUnless(bytes("b") in b)
        self.failUnless(bytes("c") in b)
        self.failUnless(bytes("ab") in b)
        self.failUnless(bytes("bc") in b)
        self.failUnless(bytes("abc") in b)
        self.failIf(bytes("ac") in b)
        self.failIf(bytes("d") in b)
        self.failIf(bytes("dab") in b)
        self.failIf(bytes("abd") in b)

    def test_alloc(self):
        b = bytes()
        alloc = b.__alloc__()
        self.assert_(alloc >= 0)
        seq = [alloc]
        for i in range(100):
            b += bytes("x")
            alloc = b.__alloc__()
            self.assert_(alloc >= len(b))
            if alloc not in seq:
                seq.append(alloc)
        #print seq

    def test_fromhex(self):
        self.assertRaises(TypeError, bytes.fromhex)
        self.assertRaises(TypeError, bytes.fromhex, 1)
        self.assertEquals(bytes.fromhex(''), bytes())
        b = bytes([0x1a, 0x2b, 0x30])
        self.assertEquals(bytes.fromhex('1a2B30'), b)
        self.assertEquals(bytes.fromhex('  1A 2B  30   '), b)
        self.assertEquals(bytes.fromhex(buffer('')), bytes())
        self.assertEquals(bytes.fromhex(buffer('0000')), bytes([0, 0]))
        self.assertRaises(ValueError, bytes.fromhex, 'a')
        self.assertRaises(ValueError, bytes.fromhex, 'rt')
        self.assertRaises(ValueError, bytes.fromhex, '1a b cd')
        self.assertRaises(ValueError, bytes.fromhex, '\x00')
        self.assertRaises(ValueError, bytes.fromhex, '12   \x00   34')

    def test_join(self):
        self.assertEqual(bytes.join([]), bytes())
        self.assertEqual(bytes.join([bytes()]), bytes())
        for part in [("abc",), ("a", "bc"), ("ab", "c"), ("a", "b", "c")]:
            lst = map(bytes, part)
            self.assertEqual(bytes.join(lst), bytes("abc"))
            self.assertEqual(bytes.join(tuple(lst)), bytes("abc"))
            self.assertEqual(bytes.join(iter(lst)), bytes("abc"))
        # XXX more...

    def test_literal(self):
        tests =  [
            (b"Wonderful spam", u"Wonderful spam"),
            (br"Wonderful spam too", u"Wonderful spam too"),
            (b"\xaa\x00\000\200", u"\xaa\x00\000\200"),
            (br"\xaa\x00\000\200", ur"\xaa\x00\000\200"),
        ]
        for b, s in tests:
            self.assertEqual(b, bytes(s, 'latin-1'))
        for c in range(128, 256):
            self.assertRaises(SyntaxError, eval,
                              'b"%s"' % chr(c))

    # Optimizations:
    # __iter__? (optimization)
    # __reversed__? (optimization)

    # XXX Some list methods?
    # extended slicing
    # extended slice assignment
    # extend (same as b[len(b):] = src)
    # reverse (in-place)
    # remove
    # pop
    # NOT sort!
    # With int arg:
    # index
    # count
    # append
    # insert

    # XXX Some string methods?  (Those that don't use character properties)
    # startswith
    # endswidth
    # find, rfind
    # index, rindex (bytes arg)
    # join
    # replace
    # translate
    # split, rsplit
    # lstrip, rstrip, strip??

    # XXX pickle and marshal support?


def test_main():
    test.test_support.run_unittest(BytesTest)


if __name__ == "__main__":
    test_main()
    ##unittest.main()