1"""Test the binascii C module."""
2
3import unittest
4import binascii
5import array
6import re
7from test.support import warnings_helper
8
9
10# Note: "*_hex" functions are aliases for "(un)hexlify"
11b2a_functions = ['b2a_base64', 'b2a_hex', 'b2a_qp', 'b2a_uu',
12                 'hexlify']
13a2b_functions = ['a2b_base64', 'a2b_hex', 'a2b_qp', 'a2b_uu',
14                 'unhexlify']
15all_functions = a2b_functions + b2a_functions + ['crc32', 'crc_hqx']
16
17
18class BinASCIITest(unittest.TestCase):
19
20    type2test = bytes
21    # Create binary test data
22    rawdata = b"The quick brown fox jumps over the lazy dog.\r\n"
23    # Be slow so we don't depend on other modules
24    rawdata += bytes(range(256))
25    rawdata += b"\r\nHello world.\n"
26
27    def setUp(self):
28        self.data = self.type2test(self.rawdata)
29
30    def test_exceptions(self):
31        # Check module exceptions
32        self.assertTrue(issubclass(binascii.Error, Exception))
33        self.assertTrue(issubclass(binascii.Incomplete, Exception))
34
35    def test_functions(self):
36        # Check presence of all functions
37        for name in all_functions:
38            self.assertTrue(hasattr(getattr(binascii, name), '__call__'))
39            self.assertRaises(TypeError, getattr(binascii, name))
40
41    def test_returned_value(self):
42        # Limit to the minimum of all limits (b2a_uu)
43        MAX_ALL = 45
44        raw = self.rawdata[:MAX_ALL]
45        for fa, fb in zip(a2b_functions, b2a_functions):
46            a2b = getattr(binascii, fa)
47            b2a = getattr(binascii, fb)
48            try:
49                a = b2a(self.type2test(raw))
50                res = a2b(self.type2test(a))
51            except Exception as err:
52                self.fail("{}/{} conversion raises {!r}".format(fb, fa, err))
53            self.assertEqual(res, raw, "{}/{} conversion: "
54                             "{!r} != {!r}".format(fb, fa, res, raw))
55            self.assertIsInstance(res, bytes)
56            self.assertIsInstance(a, bytes)
57            self.assertLess(max(a), 128)
58        self.assertIsInstance(binascii.crc_hqx(raw, 0), int)
59        self.assertIsInstance(binascii.crc32(raw), int)
60
61    def test_base64valid(self):
62        # Test base64 with valid data
63        MAX_BASE64 = 57
64        lines = []
65        for i in range(0, len(self.rawdata), MAX_BASE64):
66            b = self.type2test(self.rawdata[i:i+MAX_BASE64])
67            a = binascii.b2a_base64(b)
68            lines.append(a)
69        res = bytes()
70        for line in lines:
71            a = self.type2test(line)
72            b = binascii.a2b_base64(a)
73            res += b
74        self.assertEqual(res, self.rawdata)
75
76    def test_base64invalid(self):
77        # Test base64 with random invalid characters sprinkled throughout
78        # (This requires a new version of binascii.)
79        MAX_BASE64 = 57
80        lines = []
81        for i in range(0, len(self.data), MAX_BASE64):
82            b = self.type2test(self.rawdata[i:i+MAX_BASE64])
83            a = binascii.b2a_base64(b)
84            lines.append(a)
85
86        fillers = bytearray()
87        valid = b"abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789+/"
88        for i in range(256):
89            if i not in valid:
90                fillers.append(i)
91        def addnoise(line):
92            noise = fillers
93            ratio = len(line) // len(noise)
94            res = bytearray()
95            while line and noise:
96                if len(line) // len(noise) > ratio:
97                    c, line = line[0], line[1:]
98                else:
99                    c, noise = noise[0], noise[1:]
100                res.append(c)
101            return res + noise + line
102        res = bytearray()
103        for line in map(addnoise, lines):
104            a = self.type2test(line)
105            b = binascii.a2b_base64(a)
106            res += b
107        self.assertEqual(res, self.rawdata)
108
109        # Test base64 with just invalid characters, which should return
110        # empty strings. TBD: shouldn't it raise an exception instead ?
111        self.assertEqual(binascii.a2b_base64(self.type2test(fillers)), b'')
112
113    def test_base64_strict_mode(self):
114        # Test base64 with strict mode on
115        def _assertRegexTemplate(assert_regex: str, data: bytes, non_strict_mode_expected_result: bytes):
116            with self.assertRaisesRegex(binascii.Error, assert_regex):
117                binascii.a2b_base64(self.type2test(data), strict_mode=True)
118            self.assertEqual(binascii.a2b_base64(self.type2test(data), strict_mode=False),
119                             non_strict_mode_expected_result)
120            self.assertEqual(binascii.a2b_base64(self.type2test(data)),
121                             non_strict_mode_expected_result)
122
123        def assertExcessData(data, non_strict_mode_expected_result: bytes):
124            _assertRegexTemplate(r'(?i)Excess data', data, non_strict_mode_expected_result)
125
126        def assertNonBase64Data(data, non_strict_mode_expected_result: bytes):
127            _assertRegexTemplate(r'(?i)Only base64 data', data, non_strict_mode_expected_result)
128
129        def assertLeadingPadding(data, non_strict_mode_expected_result: bytes):
130            _assertRegexTemplate(r'(?i)Leading padding', data, non_strict_mode_expected_result)
131
132        def assertDiscontinuousPadding(data, non_strict_mode_expected_result: bytes):
133            _assertRegexTemplate(r'(?i)Discontinuous padding', data, non_strict_mode_expected_result)
134
135        # Test excess data exceptions
136        assertExcessData(b'ab==a', b'i')
137        assertExcessData(b'ab===', b'i')
138        assertExcessData(b'ab==:', b'i')
139        assertExcessData(b'abc=a', b'i\xb7')
140        assertExcessData(b'abc=:', b'i\xb7')
141        assertExcessData(b'ab==\n', b'i')
142
143        # Test non-base64 data exceptions
144        assertNonBase64Data(b'\nab==', b'i')
145        assertNonBase64Data(b'ab:(){:|:&};:==', b'i')
146        assertNonBase64Data(b'a\nb==', b'i')
147        assertNonBase64Data(b'a\x00b==', b'i')
148
149        # Test malformed padding
150        assertLeadingPadding(b'=', b'')
151        assertLeadingPadding(b'==', b'')
152        assertLeadingPadding(b'===', b'')
153        assertDiscontinuousPadding(b'ab=c=', b'i\xb7')
154        assertDiscontinuousPadding(b'ab=ab==', b'i\xb6\x9b')
155
156
157    def test_base64errors(self):
158        # Test base64 with invalid padding
159        def assertIncorrectPadding(data):
160            with self.assertRaisesRegex(binascii.Error, r'(?i)Incorrect padding'):
161                binascii.a2b_base64(self.type2test(data))
162
163        assertIncorrectPadding(b'ab')
164        assertIncorrectPadding(b'ab=')
165        assertIncorrectPadding(b'abc')
166        assertIncorrectPadding(b'abcdef')
167        assertIncorrectPadding(b'abcdef=')
168        assertIncorrectPadding(b'abcdefg')
169        assertIncorrectPadding(b'a=b=')
170        assertIncorrectPadding(b'a\nb=')
171
172        # Test base64 with invalid number of valid characters (1 mod 4)
173        def assertInvalidLength(data):
174            n_data_chars = len(re.sub(br'[^A-Za-z0-9/+]', br'', data))
175            expected_errmsg_re = \
176                r'(?i)Invalid.+number of data characters.+' + str(n_data_chars)
177            with self.assertRaisesRegex(binascii.Error, expected_errmsg_re):
178                binascii.a2b_base64(self.type2test(data))
179
180        assertInvalidLength(b'a')
181        assertInvalidLength(b'a=')
182        assertInvalidLength(b'a==')
183        assertInvalidLength(b'a===')
184        assertInvalidLength(b'a' * 5)
185        assertInvalidLength(b'a' * (4 * 87 + 1))
186        assertInvalidLength(b'A\tB\nC ??DE')  # only 5 valid characters
187
188    def test_uu(self):
189        MAX_UU = 45
190        for backtick in (True, False):
191            lines = []
192            for i in range(0, len(self.data), MAX_UU):
193                b = self.type2test(self.rawdata[i:i+MAX_UU])
194                a = binascii.b2a_uu(b, backtick=backtick)
195                lines.append(a)
196            res = bytes()
197            for line in lines:
198                a = self.type2test(line)
199                b = binascii.a2b_uu(a)
200                res += b
201            self.assertEqual(res, self.rawdata)
202
203        self.assertEqual(binascii.a2b_uu(b"\x7f"), b"\x00"*31)
204        self.assertEqual(binascii.a2b_uu(b"\x80"), b"\x00"*32)
205        self.assertEqual(binascii.a2b_uu(b"\xff"), b"\x00"*31)
206        self.assertRaises(binascii.Error, binascii.a2b_uu, b"\xff\x00")
207        self.assertRaises(binascii.Error, binascii.a2b_uu, b"!!!!")
208        self.assertRaises(binascii.Error, binascii.b2a_uu, 46*b"!")
209
210        # Issue #7701 (crash on a pydebug build)
211        self.assertEqual(binascii.b2a_uu(b'x'), b'!>   \n')
212
213        self.assertEqual(binascii.b2a_uu(b''), b' \n')
214        self.assertEqual(binascii.b2a_uu(b'', backtick=True), b'`\n')
215        self.assertEqual(binascii.a2b_uu(b' \n'), b'')
216        self.assertEqual(binascii.a2b_uu(b'`\n'), b'')
217        self.assertEqual(binascii.b2a_uu(b'\x00Cat'), b'$ $-A=   \n')
218        self.assertEqual(binascii.b2a_uu(b'\x00Cat', backtick=True),
219                         b'$`$-A=```\n')
220        self.assertEqual(binascii.a2b_uu(b'$`$-A=```\n'),
221                         binascii.a2b_uu(b'$ $-A=   \n'))
222        with self.assertRaises(TypeError):
223            binascii.b2a_uu(b"", True)
224
225    def test_crc_hqx(self):
226        crc = binascii.crc_hqx(self.type2test(b"Test the CRC-32 of"), 0)
227        crc = binascii.crc_hqx(self.type2test(b" this string."), crc)
228        self.assertEqual(crc, 14290)
229
230        self.assertRaises(TypeError, binascii.crc_hqx)
231        self.assertRaises(TypeError, binascii.crc_hqx, self.type2test(b''))
232
233        for crc in 0, 1, 0x1234, 0x12345, 0x12345678, -1:
234            self.assertEqual(binascii.crc_hqx(self.type2test(b''), crc),
235                             crc & 0xffff)
236
237    def test_crc32(self):
238        crc = binascii.crc32(self.type2test(b"Test the CRC-32 of"))
239        crc = binascii.crc32(self.type2test(b" this string."), crc)
240        self.assertEqual(crc, 1571220330)
241
242        self.assertRaises(TypeError, binascii.crc32)
243
244    def test_hex(self):
245        # test hexlification
246        s = b'{s\005\000\000\000worldi\002\000\000\000s\005\000\000\000helloi\001\000\000\0000'
247        t = binascii.b2a_hex(self.type2test(s))
248        u = binascii.a2b_hex(self.type2test(t))
249        self.assertEqual(s, u)
250        self.assertRaises(binascii.Error, binascii.a2b_hex, t[:-1])
251        self.assertRaises(binascii.Error, binascii.a2b_hex, t[:-1] + b'q')
252        self.assertRaises(binascii.Error, binascii.a2b_hex, bytes([255, 255]))
253        self.assertRaises(binascii.Error, binascii.a2b_hex, b'0G')
254        self.assertRaises(binascii.Error, binascii.a2b_hex, b'0g')
255        self.assertRaises(binascii.Error, binascii.a2b_hex, b'G0')
256        self.assertRaises(binascii.Error, binascii.a2b_hex, b'g0')
257
258        # Confirm that b2a_hex == hexlify and a2b_hex == unhexlify
259        self.assertEqual(binascii.hexlify(self.type2test(s)), t)
260        self.assertEqual(binascii.unhexlify(self.type2test(t)), u)
261
262    def test_hex_separator(self):
263        """Test that hexlify and b2a_hex are binary versions of bytes.hex."""
264        # Logic of separators is tested in test_bytes.py.  This checks that
265        # arg parsing works and exercises the direct to bytes object code
266        # path within pystrhex.c.
267        s = b'{s\005\000\000\000worldi\002\000\000\000s\005\000\000\000helloi\001\000\000\0000'
268        self.assertEqual(binascii.hexlify(self.type2test(s)), s.hex().encode('ascii'))
269        expected8 = s.hex('.', 8).encode('ascii')
270        self.assertEqual(binascii.hexlify(self.type2test(s), '.', 8), expected8)
271        expected1 = s.hex(':').encode('ascii')
272        self.assertEqual(binascii.b2a_hex(self.type2test(s), ':'), expected1)
273
274    def test_qp(self):
275        type2test = self.type2test
276        a2b_qp = binascii.a2b_qp
277        b2a_qp = binascii.b2a_qp
278
279        a2b_qp(data=b"", header=False)  # Keyword arguments allowed
280
281        # A test for SF bug 534347 (segfaults without the proper fix)
282        try:
283            a2b_qp(b"", **{1:1})
284        except TypeError:
285            pass
286        else:
287            self.fail("binascii.a2b_qp(**{1:1}) didn't raise TypeError")
288
289        self.assertEqual(a2b_qp(type2test(b"=")), b"")
290        self.assertEqual(a2b_qp(type2test(b"= ")), b"= ")
291        self.assertEqual(a2b_qp(type2test(b"==")), b"=")
292        self.assertEqual(a2b_qp(type2test(b"=\nAB")), b"AB")
293        self.assertEqual(a2b_qp(type2test(b"=\r\nAB")), b"AB")
294        self.assertEqual(a2b_qp(type2test(b"=\rAB")), b"")  # ?
295        self.assertEqual(a2b_qp(type2test(b"=\rAB\nCD")), b"CD")  # ?
296        self.assertEqual(a2b_qp(type2test(b"=AB")), b"\xab")
297        self.assertEqual(a2b_qp(type2test(b"=ab")), b"\xab")
298        self.assertEqual(a2b_qp(type2test(b"=AX")), b"=AX")
299        self.assertEqual(a2b_qp(type2test(b"=XA")), b"=XA")
300        self.assertEqual(a2b_qp(type2test(b"=AB")[:-1]), b"=A")
301
302        self.assertEqual(a2b_qp(type2test(b'_')), b'_')
303        self.assertEqual(a2b_qp(type2test(b'_'), header=True), b' ')
304
305        self.assertRaises(TypeError, b2a_qp, foo="bar")
306        self.assertEqual(a2b_qp(type2test(b"=00\r\n=00")), b"\x00\r\n\x00")
307        self.assertEqual(b2a_qp(type2test(b"\xff\r\n\xff\n\xff")),
308                         b"=FF\r\n=FF\r\n=FF")
309        self.assertEqual(b2a_qp(type2test(b"0"*75+b"\xff\r\n\xff\r\n\xff")),
310                         b"0"*75+b"=\r\n=FF\r\n=FF\r\n=FF")
311
312        self.assertEqual(b2a_qp(type2test(b'\x7f')), b'=7F')
313        self.assertEqual(b2a_qp(type2test(b'=')), b'=3D')
314
315        self.assertEqual(b2a_qp(type2test(b'_')), b'_')
316        self.assertEqual(b2a_qp(type2test(b'_'), header=True), b'=5F')
317        self.assertEqual(b2a_qp(type2test(b'x y'), header=True), b'x_y')
318        self.assertEqual(b2a_qp(type2test(b'x '), header=True), b'x=20')
319        self.assertEqual(b2a_qp(type2test(b'x y'), header=True, quotetabs=True),
320                         b'x=20y')
321        self.assertEqual(b2a_qp(type2test(b'x\ty'), header=True), b'x\ty')
322
323        self.assertEqual(b2a_qp(type2test(b' ')), b'=20')
324        self.assertEqual(b2a_qp(type2test(b'\t')), b'=09')
325        self.assertEqual(b2a_qp(type2test(b' x')), b' x')
326        self.assertEqual(b2a_qp(type2test(b'\tx')), b'\tx')
327        self.assertEqual(b2a_qp(type2test(b' x')[:-1]), b'=20')
328        self.assertEqual(b2a_qp(type2test(b'\tx')[:-1]), b'=09')
329        self.assertEqual(b2a_qp(type2test(b'\0')), b'=00')
330
331        self.assertEqual(b2a_qp(type2test(b'\0\n')), b'=00\n')
332        self.assertEqual(b2a_qp(type2test(b'\0\n'), quotetabs=True), b'=00\n')
333
334        self.assertEqual(b2a_qp(type2test(b'x y\tz')), b'x y\tz')
335        self.assertEqual(b2a_qp(type2test(b'x y\tz'), quotetabs=True),
336                         b'x=20y=09z')
337        self.assertEqual(b2a_qp(type2test(b'x y\tz'), istext=False),
338                         b'x y\tz')
339        self.assertEqual(b2a_qp(type2test(b'x \ny\t\n')),
340                         b'x=20\ny=09\n')
341        self.assertEqual(b2a_qp(type2test(b'x \ny\t\n'), quotetabs=True),
342                         b'x=20\ny=09\n')
343        self.assertEqual(b2a_qp(type2test(b'x \ny\t\n'), istext=False),
344                         b'x =0Ay\t=0A')
345        self.assertEqual(b2a_qp(type2test(b'x \ry\t\r')),
346                         b'x \ry\t\r')
347        self.assertEqual(b2a_qp(type2test(b'x \ry\t\r'), quotetabs=True),
348                         b'x=20\ry=09\r')
349        self.assertEqual(b2a_qp(type2test(b'x \ry\t\r'), istext=False),
350                         b'x =0Dy\t=0D')
351        self.assertEqual(b2a_qp(type2test(b'x \r\ny\t\r\n')),
352                         b'x=20\r\ny=09\r\n')
353        self.assertEqual(b2a_qp(type2test(b'x \r\ny\t\r\n'), quotetabs=True),
354                         b'x=20\r\ny=09\r\n')
355        self.assertEqual(b2a_qp(type2test(b'x \r\ny\t\r\n'), istext=False),
356                         b'x =0D=0Ay\t=0D=0A')
357
358        self.assertEqual(b2a_qp(type2test(b'x \r\n')[:-1]), b'x \r')
359        self.assertEqual(b2a_qp(type2test(b'x\t\r\n')[:-1]), b'x\t\r')
360        self.assertEqual(b2a_qp(type2test(b'x \r\n')[:-1], quotetabs=True),
361                         b'x=20\r')
362        self.assertEqual(b2a_qp(type2test(b'x\t\r\n')[:-1], quotetabs=True),
363                         b'x=09\r')
364        self.assertEqual(b2a_qp(type2test(b'x \r\n')[:-1], istext=False),
365                         b'x =0D')
366        self.assertEqual(b2a_qp(type2test(b'x\t\r\n')[:-1], istext=False),
367                         b'x\t=0D')
368
369        self.assertEqual(b2a_qp(type2test(b'.')), b'=2E')
370        self.assertEqual(b2a_qp(type2test(b'.\n')), b'=2E\n')
371        self.assertEqual(b2a_qp(type2test(b'.\r')), b'=2E\r')
372        self.assertEqual(b2a_qp(type2test(b'.\0')), b'=2E=00')
373        self.assertEqual(b2a_qp(type2test(b'a.\n')), b'a.\n')
374        self.assertEqual(b2a_qp(type2test(b'.a')[:-1]), b'=2E')
375
376    def test_empty_string(self):
377        # A test for SF bug #1022953.  Make sure SystemError is not raised.
378        empty = self.type2test(b'')
379        for func in all_functions:
380            if func == 'crc_hqx':
381                # crc_hqx needs 2 arguments
382                binascii.crc_hqx(empty, 0)
383                continue
384            f = getattr(binascii, func)
385            try:
386                f(empty)
387            except Exception as err:
388                self.fail("{}({!r}) raises {!r}".format(func, empty, err))
389
390    def test_unicode_b2a(self):
391        # Unicode strings are not accepted by b2a_* functions.
392        for func in set(all_functions) - set(a2b_functions):
393            try:
394                self.assertRaises(TypeError, getattr(binascii, func), "test")
395            except Exception as err:
396                self.fail('{}("test") raises {!r}'.format(func, err))
397        # crc_hqx needs 2 arguments
398        self.assertRaises(TypeError, binascii.crc_hqx, "test", 0)
399
400    def test_unicode_a2b(self):
401        # Unicode strings are accepted by a2b_* functions.
402        MAX_ALL = 45
403        raw = self.rawdata[:MAX_ALL]
404        for fa, fb in zip(a2b_functions, b2a_functions):
405            a2b = getattr(binascii, fa)
406            b2a = getattr(binascii, fb)
407            try:
408                a = b2a(self.type2test(raw))
409                binary_res = a2b(a)
410                a = a.decode('ascii')
411                res = a2b(a)
412            except Exception as err:
413                self.fail("{}/{} conversion raises {!r}".format(fb, fa, err))
414            self.assertEqual(res, raw, "{}/{} conversion: "
415                             "{!r} != {!r}".format(fb, fa, res, raw))
416            self.assertEqual(res, binary_res)
417            self.assertIsInstance(res, bytes)
418            # non-ASCII string
419            self.assertRaises(ValueError, a2b, "\x80")
420
421    def test_b2a_base64_newline(self):
422        # Issue #25357: test newline parameter
423        b = self.type2test(b'hello')
424        self.assertEqual(binascii.b2a_base64(b),
425                         b'aGVsbG8=\n')
426        self.assertEqual(binascii.b2a_base64(b, newline=True),
427                         b'aGVsbG8=\n')
428        self.assertEqual(binascii.b2a_base64(b, newline=False),
429                         b'aGVsbG8=')
430
431
432class ArrayBinASCIITest(BinASCIITest):
433    def type2test(self, s):
434        return array.array('B', list(s))
435
436
437class BytearrayBinASCIITest(BinASCIITest):
438    type2test = bytearray
439
440
441class MemoryviewBinASCIITest(BinASCIITest):
442    type2test = memoryview
443
444
445if __name__ == "__main__":
446    unittest.main()
447