Andrew M. Kuchling | a6f68e1 | 2005-06-09 14:12:36 +0000 | [diff] [blame] | 1 | """Test script for the gzip module. |
| 2 | """ |
| 3 | |
Stéphane Wirtel | 84eec11 | 2018-10-09 23:16:43 +0200 | [diff] [blame] | 4 | import array |
| 5 | import functools |
| 6 | import io |
Christian Heimes | 05e8be1 | 2008-02-23 18:30:17 +0000 | [diff] [blame] | 7 | import os |
Berker Peksag | 03020cf | 2016-10-02 13:47:58 +0300 | [diff] [blame] | 8 | import pathlib |
Antoine Pitrou | 42db3ef | 2009-01-04 21:37:59 +0000 | [diff] [blame] | 9 | import struct |
Stéphane Wirtel | 84eec11 | 2018-10-09 23:16:43 +0200 | [diff] [blame] | 10 | import sys |
| 11 | import unittest |
| 12 | from subprocess import PIPE, Popen |
| 13 | from test import support |
| 14 | from test.support import _4G, bigmemtest |
Stéphane Wirtel | 3e28eed | 2018-11-03 16:24:23 +0100 | [diff] [blame] | 15 | from test.support.script_helper import assert_python_ok, assert_python_failure |
Stéphane Wirtel | 84eec11 | 2018-10-09 23:16:43 +0200 | [diff] [blame] | 16 | |
Ezio Melotti | 78ea202 | 2009-09-12 18:41:20 +0000 | [diff] [blame] | 17 | gzip = support.import_module('gzip') |
Andrew M. Kuchling | 605ebdd | 1999-03-25 21:50:27 +0000 | [diff] [blame] | 18 | |
Walter Dörwald | 5b1284d | 2007-06-06 16:43:59 +0000 | [diff] [blame] | 19 | data1 = b""" int length=DEFAULTALLOC, err = Z_OK; |
Andrew M. Kuchling | 605ebdd | 1999-03-25 21:50:27 +0000 | [diff] [blame] | 20 | PyObject *RetVal; |
| 21 | int flushmode = Z_FINISH; |
| 22 | unsigned long start_total_out; |
| 23 | |
| 24 | """ |
| 25 | |
Walter Dörwald | 5b1284d | 2007-06-06 16:43:59 +0000 | [diff] [blame] | 26 | data2 = b"""/* zlibmodule.c -- gzip-compatible data compression */ |
Neal Norwitz | 014f103 | 2004-07-29 03:55:56 +0000 | [diff] [blame] | 27 | /* See http://www.gzip.org/zlib/ |
Andrew M. Kuchling | 605ebdd | 1999-03-25 21:50:27 +0000 | [diff] [blame] | 28 | /* See http://www.winimage.com/zLibDll for Windows */ |
| 29 | """ |
| 30 | |
Andrew M. Kuchling | 605ebdd | 1999-03-25 21:50:27 +0000 | [diff] [blame] | 31 | |
Stéphane Wirtel | 84eec11 | 2018-10-09 23:16:43 +0200 | [diff] [blame] | 32 | TEMPDIR = os.path.abspath(support.TESTFN) + '-gzdir' |
| 33 | |
| 34 | |
Antoine Pitrou | 7b96984 | 2010-09-23 16:22:51 +0000 | [diff] [blame] | 35 | class UnseekableIO(io.BytesIO): |
| 36 | def seekable(self): |
| 37 | return False |
| 38 | |
| 39 | def tell(self): |
| 40 | raise io.UnsupportedOperation |
| 41 | |
| 42 | def seek(self, *args): |
| 43 | raise io.UnsupportedOperation |
| 44 | |
| 45 | |
Nadeem Vawda | 1b8a14d | 2012-05-06 15:17:52 +0200 | [diff] [blame] | 46 | class BaseTest(unittest.TestCase): |
Benjamin Peterson | ee8712c | 2008-05-20 21:35:26 +0000 | [diff] [blame] | 47 | filename = support.TESTFN |
Tim Peters | 5cfb05e | 2004-07-27 21:02:02 +0000 | [diff] [blame] | 48 | |
Georg Brandl | b533e26 | 2008-05-25 18:19:30 +0000 | [diff] [blame] | 49 | def setUp(self): |
Benjamin Peterson | ee8712c | 2008-05-20 21:35:26 +0000 | [diff] [blame] | 50 | support.unlink(self.filename) |
Andrew M. Kuchling | 605ebdd | 1999-03-25 21:50:27 +0000 | [diff] [blame] | 51 | |
Georg Brandl | b533e26 | 2008-05-25 18:19:30 +0000 | [diff] [blame] | 52 | def tearDown(self): |
Benjamin Peterson | ee8712c | 2008-05-20 21:35:26 +0000 | [diff] [blame] | 53 | support.unlink(self.filename) |
Andrew M. Kuchling | 605ebdd | 1999-03-25 21:50:27 +0000 | [diff] [blame] | 54 | |
Andrew M. Kuchling | 85ab738 | 2000-07-29 20:18:34 +0000 | [diff] [blame] | 55 | |
Nadeem Vawda | 1b8a14d | 2012-05-06 15:17:52 +0200 | [diff] [blame] | 56 | class TestGzip(BaseTest): |
Serhiy Storchaka | bca63b3 | 2015-03-23 14:59:48 +0200 | [diff] [blame] | 57 | def write_and_read_back(self, data, mode='b'): |
| 58 | b_data = bytes(data) |
| 59 | with gzip.GzipFile(self.filename, 'w'+mode) as f: |
| 60 | l = f.write(data) |
| 61 | self.assertEqual(l, len(b_data)) |
| 62 | with gzip.GzipFile(self.filename, 'r'+mode) as f: |
| 63 | self.assertEqual(f.read(), b_data) |
| 64 | |
Georg Brandl | b533e26 | 2008-05-25 18:19:30 +0000 | [diff] [blame] | 65 | def test_write(self): |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 66 | with gzip.GzipFile(self.filename, 'wb') as f: |
| 67 | f.write(data1 * 50) |
Andrew M. Kuchling | 85ab738 | 2000-07-29 20:18:34 +0000 | [diff] [blame] | 68 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 69 | # Try flush and fileno. |
| 70 | f.flush() |
| 71 | f.fileno() |
| 72 | if hasattr(os, 'fsync'): |
| 73 | os.fsync(f.fileno()) |
| 74 | f.close() |
Andrew M. Kuchling | 85ab738 | 2000-07-29 20:18:34 +0000 | [diff] [blame] | 75 | |
Georg Brandl | b533e26 | 2008-05-25 18:19:30 +0000 | [diff] [blame] | 76 | # Test multiple close() calls. |
| 77 | f.close() |
| 78 | |
Berker Peksag | 03020cf | 2016-10-02 13:47:58 +0300 | [diff] [blame] | 79 | def test_write_read_with_pathlike_file(self): |
| 80 | filename = pathlib.Path(self.filename) |
| 81 | with gzip.GzipFile(filename, 'w') as f: |
| 82 | f.write(data1 * 50) |
| 83 | self.assertIsInstance(f.name, str) |
| 84 | with gzip.GzipFile(filename, 'a') as f: |
| 85 | f.write(data1) |
| 86 | with gzip.GzipFile(filename) as f: |
| 87 | d = f.read() |
| 88 | self.assertEqual(d, data1 * 51) |
| 89 | self.assertIsInstance(f.name, str) |
| 90 | |
Serhiy Storchaka | bca63b3 | 2015-03-23 14:59:48 +0200 | [diff] [blame] | 91 | # The following test_write_xy methods test that write accepts |
| 92 | # the corresponding bytes-like object type as input |
| 93 | # and that the data written equals bytes(xy) in all cases. |
| 94 | def test_write_memoryview(self): |
| 95 | self.write_and_read_back(memoryview(data1 * 50)) |
| 96 | m = memoryview(bytes(range(256))) |
| 97 | data = m.cast('B', shape=[8,8,4]) |
| 98 | self.write_and_read_back(data) |
| 99 | |
| 100 | def test_write_bytearray(self): |
| 101 | self.write_and_read_back(bytearray(data1 * 50)) |
| 102 | |
| 103 | def test_write_array(self): |
| 104 | self.write_and_read_back(array.array('I', data1 * 40)) |
| 105 | |
| 106 | def test_write_incompatible_type(self): |
| 107 | # Test that non-bytes-like types raise TypeError. |
| 108 | # Issue #21560: attempts to write incompatible types |
| 109 | # should not affect the state of the fileobject |
| 110 | with gzip.GzipFile(self.filename, 'wb') as f: |
| 111 | with self.assertRaises(TypeError): |
| 112 | f.write('') |
| 113 | with self.assertRaises(TypeError): |
| 114 | f.write([]) |
| 115 | f.write(data1) |
| 116 | with gzip.GzipFile(self.filename, 'rb') as f: |
| 117 | self.assertEqual(f.read(), data1) |
| 118 | |
Andrew M. Kuchling | a6f68e1 | 2005-06-09 14:12:36 +0000 | [diff] [blame] | 119 | def test_read(self): |
| 120 | self.test_write() |
| 121 | # Try reading. |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 122 | with gzip.GzipFile(self.filename, 'r') as f: |
| 123 | d = f.read() |
Andrew M. Kuchling | a6f68e1 | 2005-06-09 14:12:36 +0000 | [diff] [blame] | 124 | self.assertEqual(d, data1*50) |
Andrew M. Kuchling | 85ab738 | 2000-07-29 20:18:34 +0000 | [diff] [blame] | 125 | |
Antoine Pitrou | 4ec4b0c | 2011-04-04 21:00:37 +0200 | [diff] [blame] | 126 | def test_read1(self): |
| 127 | self.test_write() |
| 128 | blocks = [] |
| 129 | nread = 0 |
| 130 | with gzip.GzipFile(self.filename, 'r') as f: |
| 131 | while True: |
| 132 | d = f.read1() |
| 133 | if not d: |
| 134 | break |
| 135 | blocks.append(d) |
| 136 | nread += len(d) |
| 137 | # Check that position was updated correctly (see issue10791). |
| 138 | self.assertEqual(f.tell(), nread) |
| 139 | self.assertEqual(b''.join(blocks), data1 * 50) |
| 140 | |
Martin Panter | e99e977 | 2015-11-20 08:13:35 +0000 | [diff] [blame] | 141 | @bigmemtest(size=_4G, memuse=1) |
| 142 | def test_read_large(self, size): |
| 143 | # Read chunk size over UINT_MAX should be supported, despite zlib's |
| 144 | # limitation per low-level call |
| 145 | compressed = gzip.compress(data1, compresslevel=1) |
| 146 | f = gzip.GzipFile(fileobj=io.BytesIO(compressed), mode='rb') |
| 147 | self.assertEqual(f.read(size), data1) |
| 148 | |
Antoine Pitrou | 7980eaa | 2010-10-06 21:21:18 +0000 | [diff] [blame] | 149 | def test_io_on_closed_object(self): |
| 150 | # Test that I/O operations on closed GzipFile objects raise a |
| 151 | # ValueError, just like the corresponding functions on file objects. |
| 152 | |
| 153 | # Write to a file, open it for reading, then close it. |
| 154 | self.test_write() |
| 155 | f = gzip.GzipFile(self.filename, 'r') |
Antoine Pitrou | 2dbc6e6 | 2015-04-11 00:31:01 +0200 | [diff] [blame] | 156 | fileobj = f.fileobj |
| 157 | self.assertFalse(fileobj.closed) |
Antoine Pitrou | 7980eaa | 2010-10-06 21:21:18 +0000 | [diff] [blame] | 158 | f.close() |
Antoine Pitrou | 2dbc6e6 | 2015-04-11 00:31:01 +0200 | [diff] [blame] | 159 | self.assertTrue(fileobj.closed) |
Antoine Pitrou | 7980eaa | 2010-10-06 21:21:18 +0000 | [diff] [blame] | 160 | with self.assertRaises(ValueError): |
| 161 | f.read(1) |
| 162 | with self.assertRaises(ValueError): |
| 163 | f.seek(0) |
| 164 | with self.assertRaises(ValueError): |
| 165 | f.tell() |
| 166 | # Open the file for writing, then close it. |
| 167 | f = gzip.GzipFile(self.filename, 'w') |
Antoine Pitrou | 2dbc6e6 | 2015-04-11 00:31:01 +0200 | [diff] [blame] | 168 | fileobj = f.fileobj |
| 169 | self.assertFalse(fileobj.closed) |
Antoine Pitrou | 7980eaa | 2010-10-06 21:21:18 +0000 | [diff] [blame] | 170 | f.close() |
Antoine Pitrou | 2dbc6e6 | 2015-04-11 00:31:01 +0200 | [diff] [blame] | 171 | self.assertTrue(fileobj.closed) |
Antoine Pitrou | 7980eaa | 2010-10-06 21:21:18 +0000 | [diff] [blame] | 172 | with self.assertRaises(ValueError): |
| 173 | f.write(b'') |
| 174 | with self.assertRaises(ValueError): |
| 175 | f.flush() |
| 176 | |
Andrew M. Kuchling | a6f68e1 | 2005-06-09 14:12:36 +0000 | [diff] [blame] | 177 | def test_append(self): |
| 178 | self.test_write() |
| 179 | # Append to the previous file |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 180 | with gzip.GzipFile(self.filename, 'ab') as f: |
| 181 | f.write(data2 * 15) |
Andrew M. Kuchling | 85ab738 | 2000-07-29 20:18:34 +0000 | [diff] [blame] | 182 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 183 | with gzip.GzipFile(self.filename, 'rb') as f: |
| 184 | d = f.read() |
Andrew M. Kuchling | a6f68e1 | 2005-06-09 14:12:36 +0000 | [diff] [blame] | 185 | self.assertEqual(d, (data1*50) + (data2*15)) |
Andrew M. Kuchling | 85ab738 | 2000-07-29 20:18:34 +0000 | [diff] [blame] | 186 | |
Andrew M. Kuchling | 01cb47b | 2005-06-09 14:19:32 +0000 | [diff] [blame] | 187 | def test_many_append(self): |
| 188 | # Bug #1074261 was triggered when reading a file that contained |
| 189 | # many, many members. Create such a file and verify that reading it |
| 190 | # works. |
Nadeem Vawda | 1b8a14d | 2012-05-06 15:17:52 +0200 | [diff] [blame] | 191 | with gzip.GzipFile(self.filename, 'wb', 9) as f: |
Walter Dörwald | 5b1284d | 2007-06-06 16:43:59 +0000 | [diff] [blame] | 192 | f.write(b'a') |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 193 | for i in range(0, 200): |
Nadeem Vawda | 1b8a14d | 2012-05-06 15:17:52 +0200 | [diff] [blame] | 194 | with gzip.GzipFile(self.filename, "ab", 9) as f: # append |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 195 | f.write(b'a') |
Andrew M. Kuchling | 01cb47b | 2005-06-09 14:19:32 +0000 | [diff] [blame] | 196 | |
| 197 | # Try reading the file |
Nadeem Vawda | 1b8a14d | 2012-05-06 15:17:52 +0200 | [diff] [blame] | 198 | with gzip.GzipFile(self.filename, "rb") as zgfile: |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 199 | contents = b"" |
| 200 | while 1: |
| 201 | ztxt = zgfile.read(8192) |
| 202 | contents += ztxt |
| 203 | if not ztxt: break |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 204 | self.assertEqual(contents, b'a'*201) |
Andrew M. Kuchling | 01cb47b | 2005-06-09 14:19:32 +0000 | [diff] [blame] | 205 | |
Nadeem Vawda | ee1be99 | 2013-10-19 00:11:13 +0200 | [diff] [blame] | 206 | def test_exclusive_write(self): |
| 207 | with gzip.GzipFile(self.filename, 'xb') as f: |
| 208 | f.write(data1 * 50) |
| 209 | with gzip.GzipFile(self.filename, 'rb') as f: |
| 210 | self.assertEqual(f.read(), data1 * 50) |
| 211 | with self.assertRaises(FileExistsError): |
| 212 | gzip.GzipFile(self.filename, 'xb') |
| 213 | |
Antoine Pitrou | b1f8835 | 2010-01-03 22:37:40 +0000 | [diff] [blame] | 214 | def test_buffered_reader(self): |
| 215 | # Issue #7471: a GzipFile can be wrapped in a BufferedReader for |
| 216 | # performance. |
| 217 | self.test_write() |
| 218 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 219 | with gzip.GzipFile(self.filename, 'rb') as f: |
| 220 | with io.BufferedReader(f) as r: |
| 221 | lines = [line for line in r] |
Antoine Pitrou | b1f8835 | 2010-01-03 22:37:40 +0000 | [diff] [blame] | 222 | |
Ezio Melotti | d8b509b | 2011-09-28 17:37:55 +0300 | [diff] [blame] | 223 | self.assertEqual(lines, 50 * data1.splitlines(keepends=True)) |
Andrew M. Kuchling | 01cb47b | 2005-06-09 14:19:32 +0000 | [diff] [blame] | 224 | |
Andrew M. Kuchling | a6f68e1 | 2005-06-09 14:12:36 +0000 | [diff] [blame] | 225 | def test_readline(self): |
| 226 | self.test_write() |
| 227 | # Try .readline() with varying line lengths |
Martin v. Löwis | 8cc965c | 2001-08-09 07:21:56 +0000 | [diff] [blame] | 228 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 229 | with gzip.GzipFile(self.filename, 'rb') as f: |
| 230 | line_length = 0 |
| 231 | while 1: |
| 232 | L = f.readline(line_length) |
| 233 | if not L and line_length != 0: break |
| 234 | self.assertTrue(len(L) <= line_length) |
| 235 | line_length = (line_length + 1) % 50 |
Martin v. Löwis | 8cc965c | 2001-08-09 07:21:56 +0000 | [diff] [blame] | 236 | |
Andrew M. Kuchling | a6f68e1 | 2005-06-09 14:12:36 +0000 | [diff] [blame] | 237 | def test_readlines(self): |
| 238 | self.test_write() |
| 239 | # Try .readlines() |
Andrew M. Kuchling | 605ebdd | 1999-03-25 21:50:27 +0000 | [diff] [blame] | 240 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 241 | with gzip.GzipFile(self.filename, 'rb') as f: |
| 242 | L = f.readlines() |
Skip Montanaro | 12424bc | 2002-05-23 01:43:05 +0000 | [diff] [blame] | 243 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 244 | with gzip.GzipFile(self.filename, 'rb') as f: |
| 245 | while 1: |
| 246 | L = f.readlines(150) |
| 247 | if L == []: break |
Andrew M. Kuchling | a6f68e1 | 2005-06-09 14:12:36 +0000 | [diff] [blame] | 248 | |
| 249 | def test_seek_read(self): |
| 250 | self.test_write() |
| 251 | # Try seek, read test |
| 252 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 253 | with gzip.GzipFile(self.filename) as f: |
| 254 | while 1: |
| 255 | oldpos = f.tell() |
| 256 | line1 = f.readline() |
| 257 | if not line1: break |
| 258 | newpos = f.tell() |
| 259 | f.seek(oldpos) # negative seek |
| 260 | if len(line1)>10: |
| 261 | amount = 10 |
| 262 | else: |
| 263 | amount = len(line1) |
| 264 | line2 = f.read(amount) |
| 265 | self.assertEqual(line1[:amount], line2) |
| 266 | f.seek(newpos) # positive seek |
Andrew M. Kuchling | a6f68e1 | 2005-06-09 14:12:36 +0000 | [diff] [blame] | 267 | |
Thomas Wouters | 89f507f | 2006-12-13 04:49:30 +0000 | [diff] [blame] | 268 | def test_seek_whence(self): |
| 269 | self.test_write() |
| 270 | # Try seek(whence=1), read test |
| 271 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 272 | with gzip.GzipFile(self.filename) as f: |
| 273 | f.read(10) |
| 274 | f.seek(10, whence=1) |
| 275 | y = f.read(10) |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 276 | self.assertEqual(y, data1[20:30]) |
Thomas Wouters | 9fe394c | 2007-02-05 01:24:16 +0000 | [diff] [blame] | 277 | |
Andrew M. Kuchling | a6f68e1 | 2005-06-09 14:12:36 +0000 | [diff] [blame] | 278 | def test_seek_write(self): |
| 279 | # Try seek, write test |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 280 | with gzip.GzipFile(self.filename, 'w') as f: |
| 281 | for pos in range(0, 256, 16): |
| 282 | f.seek(pos) |
| 283 | f.write(b'GZ\n') |
Andrew M. Kuchling | a6f68e1 | 2005-06-09 14:12:36 +0000 | [diff] [blame] | 284 | |
| 285 | def test_mode(self): |
| 286 | self.test_write() |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 287 | with gzip.GzipFile(self.filename, 'r') as f: |
| 288 | self.assertEqual(f.myfileobj.mode, 'rb') |
Nadeem Vawda | ee1be99 | 2013-10-19 00:11:13 +0200 | [diff] [blame] | 289 | support.unlink(self.filename) |
| 290 | with gzip.GzipFile(self.filename, 'x') as f: |
| 291 | self.assertEqual(f.myfileobj.mode, 'xb') |
Andrew M. Kuchling | a6f68e1 | 2005-06-09 14:12:36 +0000 | [diff] [blame] | 292 | |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 293 | def test_1647484(self): |
| 294 | for mode in ('wb', 'rb'): |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 295 | with gzip.GzipFile(self.filename, mode) as f: |
| 296 | self.assertTrue(hasattr(f, "name")) |
| 297 | self.assertEqual(f.name, self.filename) |
Thomas Wouters | cf297e4 | 2007-02-23 15:07:44 +0000 | [diff] [blame] | 298 | |
Georg Brandl | 9f1c1dc | 2010-11-20 11:25:01 +0000 | [diff] [blame] | 299 | def test_paddedfile_getattr(self): |
| 300 | self.test_write() |
| 301 | with gzip.GzipFile(self.filename, 'rb') as f: |
| 302 | self.assertTrue(hasattr(f.fileobj, "name")) |
| 303 | self.assertEqual(f.fileobj.name, self.filename) |
| 304 | |
Antoine Pitrou | 42db3ef | 2009-01-04 21:37:59 +0000 | [diff] [blame] | 305 | def test_mtime(self): |
| 306 | mtime = 123456789 |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 307 | with gzip.GzipFile(self.filename, 'w', mtime = mtime) as fWrite: |
| 308 | fWrite.write(data1) |
| 309 | with gzip.GzipFile(self.filename) as fRead: |
Antoine Pitrou | 2dbc6e6 | 2015-04-11 00:31:01 +0200 | [diff] [blame] | 310 | self.assertTrue(hasattr(fRead, 'mtime')) |
| 311 | self.assertIsNone(fRead.mtime) |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 312 | dataRead = fRead.read() |
| 313 | self.assertEqual(dataRead, data1) |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 314 | self.assertEqual(fRead.mtime, mtime) |
Antoine Pitrou | 42db3ef | 2009-01-04 21:37:59 +0000 | [diff] [blame] | 315 | |
| 316 | def test_metadata(self): |
| 317 | mtime = 123456789 |
| 318 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 319 | with gzip.GzipFile(self.filename, 'w', mtime = mtime) as fWrite: |
| 320 | fWrite.write(data1) |
Antoine Pitrou | 42db3ef | 2009-01-04 21:37:59 +0000 | [diff] [blame] | 321 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 322 | with open(self.filename, 'rb') as fRead: |
| 323 | # see RFC 1952: http://www.faqs.org/rfcs/rfc1952.html |
Antoine Pitrou | 42db3ef | 2009-01-04 21:37:59 +0000 | [diff] [blame] | 324 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 325 | idBytes = fRead.read(2) |
| 326 | self.assertEqual(idBytes, b'\x1f\x8b') # gzip ID |
Antoine Pitrou | 42db3ef | 2009-01-04 21:37:59 +0000 | [diff] [blame] | 327 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 328 | cmByte = fRead.read(1) |
| 329 | self.assertEqual(cmByte, b'\x08') # deflate |
Antoine Pitrou | 42db3ef | 2009-01-04 21:37:59 +0000 | [diff] [blame] | 330 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 331 | flagsByte = fRead.read(1) |
| 332 | self.assertEqual(flagsByte, b'\x08') # only the FNAME flag is set |
Antoine Pitrou | 42db3ef | 2009-01-04 21:37:59 +0000 | [diff] [blame] | 333 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 334 | mtimeBytes = fRead.read(4) |
| 335 | self.assertEqual(mtimeBytes, struct.pack('<i', mtime)) # little-endian |
Antoine Pitrou | 42db3ef | 2009-01-04 21:37:59 +0000 | [diff] [blame] | 336 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 337 | xflByte = fRead.read(1) |
| 338 | self.assertEqual(xflByte, b'\x02') # maximum compression |
Antoine Pitrou | 42db3ef | 2009-01-04 21:37:59 +0000 | [diff] [blame] | 339 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 340 | osByte = fRead.read(1) |
| 341 | self.assertEqual(osByte, b'\xff') # OS "unknown" (OS-independent) |
Antoine Pitrou | 42db3ef | 2009-01-04 21:37:59 +0000 | [diff] [blame] | 342 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 343 | # Since the FNAME flag is set, the zero-terminated filename follows. |
| 344 | # RFC 1952 specifies that this is the name of the input file, if any. |
| 345 | # However, the gzip module defaults to storing the name of the output |
| 346 | # file in this field. |
| 347 | expected = self.filename.encode('Latin-1') + b'\x00' |
| 348 | nameBytes = fRead.read(len(expected)) |
| 349 | self.assertEqual(nameBytes, expected) |
Antoine Pitrou | 42db3ef | 2009-01-04 21:37:59 +0000 | [diff] [blame] | 350 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 351 | # Since no other flags were set, the header ends here. |
| 352 | # Rather than process the compressed data, let's seek to the trailer. |
| 353 | fRead.seek(os.stat(self.filename).st_size - 8) |
Antoine Pitrou | 42db3ef | 2009-01-04 21:37:59 +0000 | [diff] [blame] | 354 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 355 | crc32Bytes = fRead.read(4) # CRC32 of uncompressed data [data1] |
| 356 | self.assertEqual(crc32Bytes, b'\xaf\xd7d\x83') |
Antoine Pitrou | 42db3ef | 2009-01-04 21:37:59 +0000 | [diff] [blame] | 357 | |
Brian Curtin | 28f96b5 | 2010-10-13 02:21:42 +0000 | [diff] [blame] | 358 | isizeBytes = fRead.read(4) |
| 359 | self.assertEqual(isizeBytes, struct.pack('<i', len(data1))) |
Antoine Pitrou | 42db3ef | 2009-01-04 21:37:59 +0000 | [diff] [blame] | 360 | |
Antoine Pitrou | 308705e | 2009-01-10 16:22:51 +0000 | [diff] [blame] | 361 | def test_with_open(self): |
| 362 | # GzipFile supports the context management protocol |
| 363 | with gzip.GzipFile(self.filename, "wb") as f: |
| 364 | f.write(b"xxx") |
| 365 | f = gzip.GzipFile(self.filename, "rb") |
| 366 | f.close() |
| 367 | try: |
| 368 | with f: |
| 369 | pass |
| 370 | except ValueError: |
| 371 | pass |
| 372 | else: |
| 373 | self.fail("__enter__ on a closed file didn't raise an exception") |
| 374 | try: |
| 375 | with gzip.GzipFile(self.filename, "wb") as f: |
| 376 | 1/0 |
| 377 | except ZeroDivisionError: |
| 378 | pass |
| 379 | else: |
| 380 | self.fail("1/0 didn't raise an exception") |
| 381 | |
Antoine Pitrou | 8e33fd7 | 2010-01-13 14:37:26 +0000 | [diff] [blame] | 382 | def test_zero_padded_file(self): |
| 383 | with gzip.GzipFile(self.filename, "wb") as f: |
| 384 | f.write(data1 * 50) |
| 385 | |
| 386 | # Pad the file with zeroes |
| 387 | with open(self.filename, "ab") as f: |
| 388 | f.write(b"\x00" * 50) |
| 389 | |
| 390 | with gzip.GzipFile(self.filename, "rb") as f: |
| 391 | d = f.read() |
| 392 | self.assertEqual(d, data1 * 50, "Incorrect data in file") |
| 393 | |
Zackery Spytz | cf599f6 | 2019-05-13 01:50:52 -0600 | [diff] [blame] | 394 | def test_gzip_BadGzipFile_exception(self): |
| 395 | self.assertTrue(issubclass(gzip.BadGzipFile, OSError)) |
| 396 | |
| 397 | def test_bad_gzip_file(self): |
| 398 | with open(self.filename, 'wb') as file: |
| 399 | file.write(data1 * 50) |
| 400 | with gzip.GzipFile(self.filename, 'r') as file: |
| 401 | self.assertRaises(gzip.BadGzipFile, file.readlines) |
| 402 | |
Antoine Pitrou | 7b96984 | 2010-09-23 16:22:51 +0000 | [diff] [blame] | 403 | def test_non_seekable_file(self): |
| 404 | uncompressed = data1 * 50 |
| 405 | buf = UnseekableIO() |
| 406 | with gzip.GzipFile(fileobj=buf, mode="wb") as f: |
| 407 | f.write(uncompressed) |
| 408 | compressed = buf.getvalue() |
| 409 | buf = UnseekableIO(compressed) |
| 410 | with gzip.GzipFile(fileobj=buf, mode="rb") as f: |
| 411 | self.assertEqual(f.read(), uncompressed) |
| 412 | |
Antoine Pitrou | c3ed2e7 | 2010-09-29 10:49:46 +0000 | [diff] [blame] | 413 | def test_peek(self): |
| 414 | uncompressed = data1 * 200 |
| 415 | with gzip.GzipFile(self.filename, "wb") as f: |
| 416 | f.write(uncompressed) |
| 417 | |
| 418 | def sizes(): |
| 419 | while True: |
| 420 | for n in range(5, 50, 10): |
| 421 | yield n |
| 422 | |
| 423 | with gzip.GzipFile(self.filename, "rb") as f: |
| 424 | f.max_read_chunk = 33 |
| 425 | nread = 0 |
| 426 | for n in sizes(): |
| 427 | s = f.peek(n) |
| 428 | if s == b'': |
| 429 | break |
| 430 | self.assertEqual(f.read(len(s)), s) |
| 431 | nread += len(s) |
| 432 | self.assertEqual(f.read(100), b'') |
| 433 | self.assertEqual(nread, len(uncompressed)) |
| 434 | |
Antoine Pitrou | 4ec4b0c | 2011-04-04 21:00:37 +0200 | [diff] [blame] | 435 | def test_textio_readlines(self): |
| 436 | # Issue #10791: TextIOWrapper.readlines() fails when wrapping GzipFile. |
Ezio Melotti | d8b509b | 2011-09-28 17:37:55 +0300 | [diff] [blame] | 437 | lines = (data1 * 50).decode("ascii").splitlines(keepends=True) |
Antoine Pitrou | 4ec4b0c | 2011-04-04 21:00:37 +0200 | [diff] [blame] | 438 | self.test_write() |
| 439 | with gzip.GzipFile(self.filename, 'r') as f: |
| 440 | with io.TextIOWrapper(f, encoding="ascii") as t: |
| 441 | self.assertEqual(t.readlines(), lines) |
| 442 | |
Nadeem Vawda | 892b0b9 | 2012-01-18 09:25:58 +0200 | [diff] [blame] | 443 | def test_fileobj_from_fdopen(self): |
| 444 | # Issue #13781: Opening a GzipFile for writing fails when using a |
| 445 | # fileobj created with os.fdopen(). |
| 446 | fd = os.open(self.filename, os.O_WRONLY | os.O_CREAT) |
| 447 | with os.fdopen(fd, "wb") as f: |
| 448 | with gzip.GzipFile(fileobj=f, mode="w") as g: |
| 449 | pass |
| 450 | |
Serhiy Storchaka | bcbdd2f | 2017-10-22 13:18:21 +0300 | [diff] [blame] | 451 | def test_fileobj_mode(self): |
| 452 | gzip.GzipFile(self.filename, "wb").close() |
| 453 | with open(self.filename, "r+b") as f: |
| 454 | with gzip.GzipFile(fileobj=f, mode='r') as g: |
| 455 | self.assertEqual(g.mode, gzip.READ) |
| 456 | with gzip.GzipFile(fileobj=f, mode='w') as g: |
| 457 | self.assertEqual(g.mode, gzip.WRITE) |
| 458 | with gzip.GzipFile(fileobj=f, mode='a') as g: |
| 459 | self.assertEqual(g.mode, gzip.WRITE) |
| 460 | with gzip.GzipFile(fileobj=f, mode='x') as g: |
| 461 | self.assertEqual(g.mode, gzip.WRITE) |
| 462 | with self.assertRaises(ValueError): |
| 463 | gzip.GzipFile(fileobj=f, mode='z') |
| 464 | for mode in "rb", "r+b": |
| 465 | with open(self.filename, mode) as f: |
| 466 | with gzip.GzipFile(fileobj=f) as g: |
| 467 | self.assertEqual(g.mode, gzip.READ) |
| 468 | for mode in "wb", "ab", "xb": |
| 469 | if "x" in mode: |
| 470 | support.unlink(self.filename) |
| 471 | with open(self.filename, mode) as f: |
| 472 | with gzip.GzipFile(fileobj=f) as g: |
| 473 | self.assertEqual(g.mode, gzip.WRITE) |
| 474 | |
Nadeem Vawda | 103e811 | 2012-06-20 01:35:22 +0200 | [diff] [blame] | 475 | def test_bytes_filename(self): |
| 476 | str_filename = self.filename |
| 477 | try: |
| 478 | bytes_filename = str_filename.encode("ascii") |
| 479 | except UnicodeEncodeError: |
| 480 | self.skipTest("Temporary file name needs to be ASCII") |
| 481 | with gzip.GzipFile(bytes_filename, "wb") as f: |
| 482 | f.write(data1 * 50) |
| 483 | with gzip.GzipFile(bytes_filename, "rb") as f: |
| 484 | self.assertEqual(f.read(), data1 * 50) |
| 485 | # Sanity check that we are actually operating on the right file. |
| 486 | with gzip.GzipFile(str_filename, "rb") as f: |
| 487 | self.assertEqual(f.read(), data1 * 50) |
| 488 | |
Antoine Pitrou | 2dbc6e6 | 2015-04-11 00:31:01 +0200 | [diff] [blame] | 489 | def test_decompress_limited(self): |
| 490 | """Decompressed data buffering should be limited""" |
Serhiy Storchaka | 5f1a518 | 2016-09-11 14:41:02 +0300 | [diff] [blame] | 491 | bomb = gzip.compress(b'\0' * int(2e6), compresslevel=9) |
Antoine Pitrou | 2dbc6e6 | 2015-04-11 00:31:01 +0200 | [diff] [blame] | 492 | self.assertLess(len(bomb), io.DEFAULT_BUFFER_SIZE) |
| 493 | |
| 494 | bomb = io.BytesIO(bomb) |
| 495 | decomp = gzip.GzipFile(fileobj=bomb) |
Serhiy Storchaka | 5f1a518 | 2016-09-11 14:41:02 +0300 | [diff] [blame] | 496 | self.assertEqual(decomp.read(1), b'\0') |
Antoine Pitrou | 2dbc6e6 | 2015-04-11 00:31:01 +0200 | [diff] [blame] | 497 | max_decomp = 1 + io.DEFAULT_BUFFER_SIZE |
| 498 | self.assertLessEqual(decomp._buffer.raw.tell(), max_decomp, |
| 499 | "Excessive amount of data was decompressed") |
| 500 | |
Antoine Pitrou | 79c5ef1 | 2010-08-17 21:10:05 +0000 | [diff] [blame] | 501 | # Testing compress/decompress shortcut functions |
| 502 | |
| 503 | def test_compress(self): |
| 504 | for data in [data1, data2]: |
| 505 | for args in [(), (1,), (6,), (9,)]: |
| 506 | datac = gzip.compress(data, *args) |
| 507 | self.assertEqual(type(datac), bytes) |
| 508 | with gzip.GzipFile(fileobj=io.BytesIO(datac), mode="rb") as f: |
| 509 | self.assertEqual(f.read(), data) |
| 510 | |
guoci | 0e7497c | 2018-11-07 04:50:23 -0500 | [diff] [blame] | 511 | def test_compress_mtime(self): |
| 512 | mtime = 123456789 |
| 513 | for data in [data1, data2]: |
| 514 | for args in [(), (1,), (6,), (9,)]: |
| 515 | with self.subTest(data=data, args=args): |
| 516 | datac = gzip.compress(data, *args, mtime=mtime) |
| 517 | self.assertEqual(type(datac), bytes) |
| 518 | with gzip.GzipFile(fileobj=io.BytesIO(datac), mode="rb") as f: |
| 519 | f.read(1) # to set mtime attribute |
| 520 | self.assertEqual(f.mtime, mtime) |
| 521 | |
Antoine Pitrou | 79c5ef1 | 2010-08-17 21:10:05 +0000 | [diff] [blame] | 522 | def test_decompress(self): |
| 523 | for data in (data1, data2): |
| 524 | buf = io.BytesIO() |
| 525 | with gzip.GzipFile(fileobj=buf, mode="wb") as f: |
| 526 | f.write(data) |
| 527 | self.assertEqual(gzip.decompress(buf.getvalue()), data) |
| 528 | # Roundtrip with compress |
| 529 | datac = gzip.compress(data) |
| 530 | self.assertEqual(gzip.decompress(datac), data) |
| 531 | |
Serhiy Storchaka | 7c3922f | 2013-01-22 17:01:59 +0200 | [diff] [blame] | 532 | def test_read_truncated(self): |
| 533 | data = data1*50 |
| 534 | # Drop the CRC (4 bytes) and file size (4 bytes). |
| 535 | truncated = gzip.compress(data)[:-8] |
| 536 | with gzip.GzipFile(fileobj=io.BytesIO(truncated)) as f: |
| 537 | self.assertRaises(EOFError, f.read) |
| 538 | with gzip.GzipFile(fileobj=io.BytesIO(truncated)) as f: |
| 539 | self.assertEqual(f.read(len(data)), data) |
| 540 | self.assertRaises(EOFError, f.read, 1) |
| 541 | # Incomplete 10-byte header. |
| 542 | for i in range(2, 10): |
| 543 | with gzip.GzipFile(fileobj=io.BytesIO(truncated[:i])) as f: |
| 544 | self.assertRaises(EOFError, f.read, 1) |
| 545 | |
Serhiy Storchaka | 7e69f00 | 2013-04-08 22:35:02 +0300 | [diff] [blame] | 546 | def test_read_with_extra(self): |
| 547 | # Gzip data with an extra field |
| 548 | gzdata = (b'\x1f\x8b\x08\x04\xb2\x17cQ\x02\xff' |
| 549 | b'\x05\x00Extra' |
| 550 | b'\x0bI-.\x01\x002\xd1Mx\x04\x00\x00\x00') |
| 551 | with gzip.GzipFile(fileobj=io.BytesIO(gzdata)) as f: |
| 552 | self.assertEqual(f.read(), b'Test') |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 553 | |
Ned Deily | 6120739 | 2014-03-09 14:44:34 -0700 | [diff] [blame] | 554 | def test_prepend_error(self): |
| 555 | # See issue #20875 |
| 556 | with gzip.open(self.filename, "wb") as f: |
| 557 | f.write(data1) |
| 558 | with gzip.open(self.filename, "rb") as f: |
Antoine Pitrou | 2dbc6e6 | 2015-04-11 00:31:01 +0200 | [diff] [blame] | 559 | f._buffer.raw._fp.prepend() |
Ned Deily | 6120739 | 2014-03-09 14:44:34 -0700 | [diff] [blame] | 560 | |
Nadeem Vawda | 1b8a14d | 2012-05-06 15:17:52 +0200 | [diff] [blame] | 561 | class TestOpen(BaseTest): |
| 562 | def test_binary_modes(self): |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 563 | uncompressed = data1 * 50 |
Nadeem Vawda | ee1be99 | 2013-10-19 00:11:13 +0200 | [diff] [blame] | 564 | |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 565 | with gzip.open(self.filename, "wb") as f: |
| 566 | f.write(uncompressed) |
| 567 | with open(self.filename, "rb") as f: |
| 568 | file_data = gzip.decompress(f.read()) |
| 569 | self.assertEqual(file_data, uncompressed) |
Nadeem Vawda | ee1be99 | 2013-10-19 00:11:13 +0200 | [diff] [blame] | 570 | |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 571 | with gzip.open(self.filename, "rb") as f: |
| 572 | self.assertEqual(f.read(), uncompressed) |
Nadeem Vawda | ee1be99 | 2013-10-19 00:11:13 +0200 | [diff] [blame] | 573 | |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 574 | with gzip.open(self.filename, "ab") as f: |
| 575 | f.write(uncompressed) |
| 576 | with open(self.filename, "rb") as f: |
| 577 | file_data = gzip.decompress(f.read()) |
| 578 | self.assertEqual(file_data, uncompressed * 2) |
| 579 | |
Nadeem Vawda | ee1be99 | 2013-10-19 00:11:13 +0200 | [diff] [blame] | 580 | with self.assertRaises(FileExistsError): |
| 581 | gzip.open(self.filename, "xb") |
| 582 | support.unlink(self.filename) |
| 583 | with gzip.open(self.filename, "xb") as f: |
| 584 | f.write(uncompressed) |
| 585 | with open(self.filename, "rb") as f: |
| 586 | file_data = gzip.decompress(f.read()) |
| 587 | self.assertEqual(file_data, uncompressed) |
| 588 | |
Berker Peksag | 03020cf | 2016-10-02 13:47:58 +0300 | [diff] [blame] | 589 | def test_pathlike_file(self): |
| 590 | filename = pathlib.Path(self.filename) |
| 591 | with gzip.open(filename, "wb") as f: |
| 592 | f.write(data1 * 50) |
| 593 | with gzip.open(filename, "ab") as f: |
| 594 | f.write(data1) |
| 595 | with gzip.open(filename) as f: |
| 596 | self.assertEqual(f.read(), data1 * 51) |
| 597 | |
Nadeem Vawda | 1b8a14d | 2012-05-06 15:17:52 +0200 | [diff] [blame] | 598 | def test_implicit_binary_modes(self): |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 599 | # Test implicit binary modes (no "b" or "t" in mode string). |
| 600 | uncompressed = data1 * 50 |
Nadeem Vawda | ee1be99 | 2013-10-19 00:11:13 +0200 | [diff] [blame] | 601 | |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 602 | with gzip.open(self.filename, "w") as f: |
| 603 | f.write(uncompressed) |
| 604 | with open(self.filename, "rb") as f: |
| 605 | file_data = gzip.decompress(f.read()) |
| 606 | self.assertEqual(file_data, uncompressed) |
Nadeem Vawda | ee1be99 | 2013-10-19 00:11:13 +0200 | [diff] [blame] | 607 | |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 608 | with gzip.open(self.filename, "r") as f: |
| 609 | self.assertEqual(f.read(), uncompressed) |
Nadeem Vawda | ee1be99 | 2013-10-19 00:11:13 +0200 | [diff] [blame] | 610 | |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 611 | with gzip.open(self.filename, "a") as f: |
| 612 | f.write(uncompressed) |
| 613 | with open(self.filename, "rb") as f: |
| 614 | file_data = gzip.decompress(f.read()) |
| 615 | self.assertEqual(file_data, uncompressed * 2) |
| 616 | |
Nadeem Vawda | ee1be99 | 2013-10-19 00:11:13 +0200 | [diff] [blame] | 617 | with self.assertRaises(FileExistsError): |
| 618 | gzip.open(self.filename, "x") |
| 619 | support.unlink(self.filename) |
| 620 | with gzip.open(self.filename, "x") as f: |
| 621 | f.write(uncompressed) |
| 622 | with open(self.filename, "rb") as f: |
| 623 | file_data = gzip.decompress(f.read()) |
| 624 | self.assertEqual(file_data, uncompressed) |
| 625 | |
Nadeem Vawda | 1b8a14d | 2012-05-06 15:17:52 +0200 | [diff] [blame] | 626 | def test_text_modes(self): |
Nadeem Vawda | 11328e4 | 2012-05-06 19:24:18 +0200 | [diff] [blame] | 627 | uncompressed = data1.decode("ascii") * 50 |
| 628 | uncompressed_raw = uncompressed.replace("\n", os.linesep) |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 629 | with gzip.open(self.filename, "wt") as f: |
| 630 | f.write(uncompressed) |
| 631 | with open(self.filename, "rb") as f: |
| 632 | file_data = gzip.decompress(f.read()).decode("ascii") |
Nadeem Vawda | 11328e4 | 2012-05-06 19:24:18 +0200 | [diff] [blame] | 633 | self.assertEqual(file_data, uncompressed_raw) |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 634 | with gzip.open(self.filename, "rt") as f: |
| 635 | self.assertEqual(f.read(), uncompressed) |
| 636 | with gzip.open(self.filename, "at") as f: |
| 637 | f.write(uncompressed) |
| 638 | with open(self.filename, "rb") as f: |
| 639 | file_data = gzip.decompress(f.read()).decode("ascii") |
Nadeem Vawda | 11328e4 | 2012-05-06 19:24:18 +0200 | [diff] [blame] | 640 | self.assertEqual(file_data, uncompressed_raw * 2) |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 641 | |
Nadeem Vawda | 6872101 | 2012-06-04 23:21:38 +0200 | [diff] [blame] | 642 | def test_fileobj(self): |
| 643 | uncompressed_bytes = data1 * 50 |
| 644 | uncompressed_str = uncompressed_bytes.decode("ascii") |
| 645 | compressed = gzip.compress(uncompressed_bytes) |
| 646 | with gzip.open(io.BytesIO(compressed), "r") as f: |
| 647 | self.assertEqual(f.read(), uncompressed_bytes) |
| 648 | with gzip.open(io.BytesIO(compressed), "rb") as f: |
| 649 | self.assertEqual(f.read(), uncompressed_bytes) |
| 650 | with gzip.open(io.BytesIO(compressed), "rt") as f: |
| 651 | self.assertEqual(f.read(), uncompressed_str) |
| 652 | |
Nadeem Vawda | 1b8a14d | 2012-05-06 15:17:52 +0200 | [diff] [blame] | 653 | def test_bad_params(self): |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 654 | # Test invalid parameter combinations. |
Nadeem Vawda | 6872101 | 2012-06-04 23:21:38 +0200 | [diff] [blame] | 655 | with self.assertRaises(TypeError): |
| 656 | gzip.open(123.456) |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 657 | with self.assertRaises(ValueError): |
| 658 | gzip.open(self.filename, "wbt") |
| 659 | with self.assertRaises(ValueError): |
Nadeem Vawda | ee1be99 | 2013-10-19 00:11:13 +0200 | [diff] [blame] | 660 | gzip.open(self.filename, "xbt") |
| 661 | with self.assertRaises(ValueError): |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 662 | gzip.open(self.filename, "rb", encoding="utf-8") |
| 663 | with self.assertRaises(ValueError): |
| 664 | gzip.open(self.filename, "rb", errors="ignore") |
| 665 | with self.assertRaises(ValueError): |
| 666 | gzip.open(self.filename, "rb", newline="\n") |
| 667 | |
Nadeem Vawda | 1b8a14d | 2012-05-06 15:17:52 +0200 | [diff] [blame] | 668 | def test_encoding(self): |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 669 | # Test non-default encoding. |
Nadeem Vawda | 11328e4 | 2012-05-06 19:24:18 +0200 | [diff] [blame] | 670 | uncompressed = data1.decode("ascii") * 50 |
| 671 | uncompressed_raw = uncompressed.replace("\n", os.linesep) |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 672 | with gzip.open(self.filename, "wt", encoding="utf-16") as f: |
| 673 | f.write(uncompressed) |
| 674 | with open(self.filename, "rb") as f: |
| 675 | file_data = gzip.decompress(f.read()).decode("utf-16") |
Nadeem Vawda | 11328e4 | 2012-05-06 19:24:18 +0200 | [diff] [blame] | 676 | self.assertEqual(file_data, uncompressed_raw) |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 677 | with gzip.open(self.filename, "rt", encoding="utf-16") as f: |
| 678 | self.assertEqual(f.read(), uncompressed) |
| 679 | |
Nadeem Vawda | 1b8a14d | 2012-05-06 15:17:52 +0200 | [diff] [blame] | 680 | def test_encoding_error_handler(self): |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 681 | # Test with non-default encoding error handler. |
| 682 | with gzip.open(self.filename, "wb") as f: |
| 683 | f.write(b"foo\xffbar") |
| 684 | with gzip.open(self.filename, "rt", encoding="ascii", errors="ignore") \ |
| 685 | as f: |
| 686 | self.assertEqual(f.read(), "foobar") |
| 687 | |
Nadeem Vawda | 1b8a14d | 2012-05-06 15:17:52 +0200 | [diff] [blame] | 688 | def test_newline(self): |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 689 | # Test with explicit newline (universal newline mode disabled). |
| 690 | uncompressed = data1.decode("ascii") * 50 |
Nadeem Vawda | 9d9dc8e | 2012-05-06 16:25:35 +0200 | [diff] [blame] | 691 | with gzip.open(self.filename, "wt", newline="\n") as f: |
Nadeem Vawda | 7e12620 | 2012-05-06 15:04:01 +0200 | [diff] [blame] | 692 | f.write(uncompressed) |
| 693 | with gzip.open(self.filename, "rt", newline="\r") as f: |
| 694 | self.assertEqual(f.readlines(), [uncompressed]) |
| 695 | |
Stéphane Wirtel | 84eec11 | 2018-10-09 23:16:43 +0200 | [diff] [blame] | 696 | |
| 697 | def create_and_remove_directory(directory): |
| 698 | def decorator(function): |
| 699 | @functools.wraps(function) |
| 700 | def wrapper(*args, **kwargs): |
| 701 | os.makedirs(directory) |
| 702 | try: |
| 703 | return function(*args, **kwargs) |
| 704 | finally: |
| 705 | support.rmtree(directory) |
| 706 | return wrapper |
| 707 | return decorator |
| 708 | |
| 709 | |
| 710 | class TestCommandLine(unittest.TestCase): |
| 711 | data = b'This is a simple test with gzip' |
| 712 | |
| 713 | def test_decompress_stdin_stdout(self): |
| 714 | with io.BytesIO() as bytes_io: |
| 715 | with gzip.GzipFile(fileobj=bytes_io, mode='wb') as gzip_file: |
| 716 | gzip_file.write(self.data) |
| 717 | |
| 718 | args = sys.executable, '-m', 'gzip', '-d' |
| 719 | with Popen(args, stdin=PIPE, stdout=PIPE, stderr=PIPE) as proc: |
| 720 | out, err = proc.communicate(bytes_io.getvalue()) |
| 721 | |
| 722 | self.assertEqual(err, b'') |
| 723 | self.assertEqual(out, self.data) |
| 724 | |
| 725 | @create_and_remove_directory(TEMPDIR) |
| 726 | def test_decompress_infile_outfile(self): |
| 727 | gzipname = os.path.join(TEMPDIR, 'testgzip.gz') |
| 728 | self.assertFalse(os.path.exists(gzipname)) |
| 729 | |
| 730 | with gzip.open(gzipname, mode='wb') as fp: |
| 731 | fp.write(self.data) |
| 732 | rc, out, err = assert_python_ok('-m', 'gzip', '-d', gzipname) |
| 733 | |
| 734 | with open(os.path.join(TEMPDIR, "testgzip"), "rb") as gunziped: |
| 735 | self.assertEqual(gunziped.read(), self.data) |
| 736 | |
| 737 | self.assertTrue(os.path.exists(gzipname)) |
| 738 | self.assertEqual(rc, 0) |
| 739 | self.assertEqual(out, b'') |
| 740 | self.assertEqual(err, b'') |
| 741 | |
| 742 | def test_decompress_infile_outfile_error(self): |
| 743 | rc, out, err = assert_python_ok('-m', 'gzip', '-d', 'thisisatest.out') |
| 744 | self.assertIn(b"filename doesn't end in .gz:", out) |
| 745 | self.assertEqual(rc, 0) |
| 746 | self.assertEqual(err, b'') |
| 747 | |
| 748 | @create_and_remove_directory(TEMPDIR) |
| 749 | def test_compress_stdin_outfile(self): |
| 750 | args = sys.executable, '-m', 'gzip' |
| 751 | with Popen(args, stdin=PIPE, stdout=PIPE, stderr=PIPE) as proc: |
| 752 | out, err = proc.communicate(self.data) |
| 753 | |
| 754 | self.assertEqual(err, b'') |
| 755 | self.assertEqual(out[:2], b"\x1f\x8b") |
| 756 | |
| 757 | @create_and_remove_directory(TEMPDIR) |
Gregory P. Smith | cd46655 | 2019-04-14 10:32:07 -0700 | [diff] [blame] | 758 | def test_compress_infile_outfile_default(self): |
Stéphane Wirtel | 84eec11 | 2018-10-09 23:16:43 +0200 | [diff] [blame] | 759 | local_testgzip = os.path.join(TEMPDIR, 'testgzip') |
| 760 | gzipname = local_testgzip + '.gz' |
| 761 | self.assertFalse(os.path.exists(gzipname)) |
| 762 | |
| 763 | with open(local_testgzip, 'wb') as fp: |
| 764 | fp.write(self.data) |
| 765 | |
| 766 | rc, out, err = assert_python_ok('-m', 'gzip', local_testgzip) |
| 767 | |
| 768 | self.assertTrue(os.path.exists(gzipname)) |
Stéphane Wirtel | 84eec11 | 2018-10-09 23:16:43 +0200 | [diff] [blame] | 769 | self.assertEqual(out, b'') |
| 770 | self.assertEqual(err, b'') |
| 771 | |
Stéphane Wirtel | 3e28eed | 2018-11-03 16:24:23 +0100 | [diff] [blame] | 772 | @create_and_remove_directory(TEMPDIR) |
| 773 | def test_compress_infile_outfile(self): |
| 774 | for compress_level in ('--fast', '--best'): |
| 775 | with self.subTest(compress_level=compress_level): |
| 776 | local_testgzip = os.path.join(TEMPDIR, 'testgzip') |
| 777 | gzipname = local_testgzip + '.gz' |
| 778 | self.assertFalse(os.path.exists(gzipname)) |
| 779 | |
| 780 | with open(local_testgzip, 'wb') as fp: |
| 781 | fp.write(self.data) |
| 782 | |
| 783 | rc, out, err = assert_python_ok('-m', 'gzip', compress_level, local_testgzip) |
| 784 | |
| 785 | self.assertTrue(os.path.exists(gzipname)) |
| 786 | self.assertEqual(out, b'') |
| 787 | self.assertEqual(err, b'') |
| 788 | os.remove(gzipname) |
| 789 | self.assertFalse(os.path.exists(gzipname)) |
| 790 | |
| 791 | def test_compress_fast_best_are_exclusive(self): |
| 792 | rc, out, err = assert_python_failure('-m', 'gzip', '--fast', '--best') |
| 793 | self.assertIn(b"error: argument --best: not allowed with argument --fast", err) |
| 794 | self.assertEqual(out, b'') |
| 795 | |
| 796 | def test_decompress_cannot_have_flags_compression(self): |
| 797 | rc, out, err = assert_python_failure('-m', 'gzip', '--fast', '-d') |
| 798 | self.assertIn(b'error: argument -d/--decompress: not allowed with argument --fast', err) |
| 799 | self.assertEqual(out, b'') |
| 800 | |
Stéphane Wirtel | 84eec11 | 2018-10-09 23:16:43 +0200 | [diff] [blame] | 801 | |
Andrew M. Kuchling | a6f68e1 | 2005-06-09 14:12:36 +0000 | [diff] [blame] | 802 | def test_main(verbose=None): |
Stéphane Wirtel | 84eec11 | 2018-10-09 23:16:43 +0200 | [diff] [blame] | 803 | support.run_unittest(TestGzip, TestOpen, TestCommandLine) |
| 804 | |
Andrew M. Kuchling | a6f68e1 | 2005-06-09 14:12:36 +0000 | [diff] [blame] | 805 | |
| 806 | if __name__ == "__main__": |
| 807 | test_main(verbose=True) |