blob: 4d6684030bb5a0ca4b722a239e57d3385fced9ac [file] [log] [blame]
Benjamin Peterson90f5ba52010-03-11 22:53:45 +00001#!/usr/bin/env python3
Benjamin Petersonee8712c2008-05-20 21:35:26 +00002from test import support
Nadeem Vawdaea4b46f2011-04-12 23:02:42 +02003from test.support import TESTFN, precisionbigmemtest, _4G
Tim Peters499d09a2002-11-09 06:31:56 +00004
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +00005import unittest
Guido van Rossum34d19282007-08-09 01:03:29 +00006from io import BytesIO
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +00007import os
Nadeem Vawdaea4b46f2011-04-12 23:02:42 +02008import random
Guido van Rossum360e4b82007-05-14 22:51:27 +00009import subprocess
Tim Peters499d09a2002-11-09 06:31:56 +000010import sys
Victor Stinner45df8202010-04-28 22:31:17 +000011
12try:
13 import threading
14except ImportError:
15 threading = None
Tim Peters499d09a2002-11-09 06:31:56 +000016
R. David Murraya21e4ca2009-03-31 23:16:50 +000017# Skip tests if the bz2 module doesn't exist.
18bz2 = support.import_module('bz2')
Tim Peters499d09a2002-11-09 06:31:56 +000019from bz2 import BZ2File, BZ2Compressor, BZ2Decompressor
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +000020
Skip Montanaro289bc052007-08-17 02:30:27 +000021has_cmdline_bunzip2 = sys.platform not in ("win32", "os2emx")
Tim Peters9de06bd2002-11-09 06:45:31 +000022
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +000023class BaseTest(unittest.TestCase):
Tim Peters3de75262002-11-09 05:26:15 +000024 "Base for other testcases."
Antoine Pitrou37dc5f82011-04-03 17:05:46 +020025 TEXT_LINES = [
26 b'root:x:0:0:root:/root:/bin/bash\n',
27 b'bin:x:1:1:bin:/bin:\n',
28 b'daemon:x:2:2:daemon:/sbin:\n',
29 b'adm:x:3:4:adm:/var/adm:\n',
30 b'lp:x:4:7:lp:/var/spool/lpd:\n',
31 b'sync:x:5:0:sync:/sbin:/bin/sync\n',
32 b'shutdown:x:6:0:shutdown:/sbin:/sbin/shutdown\n',
33 b'halt:x:7:0:halt:/sbin:/sbin/halt\n',
34 b'mail:x:8:12:mail:/var/spool/mail:\n',
35 b'news:x:9:13:news:/var/spool/news:\n',
36 b'uucp:x:10:14:uucp:/var/spool/uucp:\n',
37 b'operator:x:11:0:operator:/root:\n',
38 b'games:x:12:100:games:/usr/games:\n',
39 b'gopher:x:13:30:gopher:/usr/lib/gopher-data:\n',
40 b'ftp:x:14:50:FTP User:/var/ftp:/bin/bash\n',
41 b'nobody:x:65534:65534:Nobody:/home:\n',
42 b'postfix:x:100:101:postfix:/var/spool/postfix:\n',
43 b'niemeyer:x:500:500::/home/niemeyer:/bin/bash\n',
44 b'postgres:x:101:102:PostgreSQL Server:/var/lib/pgsql:/bin/bash\n',
45 b'mysql:x:102:103:MySQL server:/var/lib/mysql:/bin/bash\n',
46 b'www:x:103:104::/var/www:/bin/false\n',
47 ]
48 TEXT = b''.join(TEXT_LINES)
Guido van Rossum522a6c62007-05-22 23:13:45 +000049 DATA = b'BZh91AY&SY.\xc8N\x18\x00\x01>_\x80\x00\x10@\x02\xff\xf0\x01\x07n\x00?\xe7\xff\xe00\x01\x99\xaa\x00\xc0\x03F\x86\x8c#&\x83F\x9a\x03\x06\xa6\xd0\xa6\x93M\x0fQ\xa7\xa8\x06\x804hh\x12$\x11\xa4i4\xf14S\xd2<Q\xb5\x0fH\xd3\xd4\xdd\xd5\x87\xbb\xf8\x94\r\x8f\xafI\x12\xe1\xc9\xf8/E\x00pu\x89\x12]\xc9\xbbDL\nQ\x0e\t1\x12\xdf\xa0\xc0\x97\xac2O9\x89\x13\x94\x0e\x1c7\x0ed\x95I\x0c\xaaJ\xa4\x18L\x10\x05#\x9c\xaf\xba\xbc/\x97\x8a#C\xc8\xe1\x8cW\xf9\xe2\xd0\xd6M\xa7\x8bXa<e\x84t\xcbL\xb3\xa7\xd9\xcd\xd1\xcb\x84.\xaf\xb3\xab\xab\xad`n}\xa0lh\tE,\x8eZ\x15\x17VH>\x88\xe5\xcd9gd6\x0b\n\xe9\x9b\xd5\x8a\x99\xf7\x08.K\x8ev\xfb\xf7xw\xbb\xdf\xa1\x92\xf1\xdd|/";\xa2\xba\x9f\xd5\xb1#A\xb6\xf6\xb3o\xc9\xc5y\\\xebO\xe7\x85\x9a\xbc\xb6f8\x952\xd5\xd7"%\x89>V,\xf7\xa6z\xe2\x9f\xa3\xdf\x11\x11"\xd6E)I\xa9\x13^\xca\xf3r\xd0\x03U\x922\xf26\xec\xb6\xed\x8b\xc3U\x13\x9d\xc5\x170\xa4\xfa^\x92\xacDF\x8a\x97\xd6\x19\xfe\xdd\xb8\xbd\x1a\x9a\x19\xa3\x80ankR\x8b\xe5\xd83]\xa9\xc6\x08\x82f\xf6\xb9"6l$\xb8j@\xc0\x8a\xb0l1..\xbak\x83ls\x15\xbc\xf4\xc1\x13\xbe\xf8E\xb8\x9d\r\xa8\x9dk\x84\xd3n\xfa\xacQ\x07\xb1%y\xaav\xb4\x08\xe0z\x1b\x16\xf5\x04\xe9\xcc\xb9\x08z\x1en7.G\xfc]\xc9\x14\xe1B@\xbb!8`'
50 DATA_CRLF = b'BZh91AY&SY\xaez\xbbN\x00\x01H\xdf\x80\x00\x12@\x02\xff\xf0\x01\x07n\x00?\xe7\xff\xe0@\x01\xbc\xc6`\x86*\x8d=M\xa9\x9a\x86\xd0L@\x0fI\xa6!\xa1\x13\xc8\x88jdi\x8d@\x03@\x1a\x1a\x0c\x0c\x83 \x00\xc4h2\x19\x01\x82D\x84e\t\xe8\x99\x89\x19\x1ah\x00\r\x1a\x11\xaf\x9b\x0fG\xf5(\x1b\x1f?\t\x12\xcf\xb5\xfc\x95E\x00ps\x89\x12^\xa4\xdd\xa2&\x05(\x87\x04\x98\x89u\xe40%\xb6\x19\'\x8c\xc4\x89\xca\x07\x0e\x1b!\x91UIFU%C\x994!DI\xd2\xfa\xf0\xf1N8W\xde\x13A\xf5\x9cr%?\x9f3;I45A\xd1\x8bT\xb1<l\xba\xcb_\xc00xY\x17r\x17\x88\x08\x08@\xa0\ry@\x10\x04$)`\xf2\xce\x89z\xb0s\xec\x9b.iW\x9d\x81\xb5-+t\x9f\x1a\'\x97dB\xf5x\xb5\xbe.[.\xd7\x0e\x81\xe7\x08\x1cN`\x88\x10\xca\x87\xc3!"\x80\x92R\xa1/\xd1\xc0\xe6mf\xac\xbd\x99\xcca\xb3\x8780>\xa4\xc7\x8d\x1a\\"\xad\xa1\xabyBg\x15\xb9l\x88\x88\x91k"\x94\xa4\xd4\x89\xae*\xa6\x0b\x10\x0c\xd6\xd4m\xe86\xec\xb5j\x8a\x86j\';\xca.\x01I\xf2\xaaJ\xe8\x88\x8cU+t3\xfb\x0c\n\xa33\x13r2\r\x16\xe0\xb3(\xbf\x1d\x83r\xe7M\xf0D\x1365\xd8\x88\xd3\xa4\x92\xcb2\x06\x04\\\xc1\xb0\xea//\xbek&\xd8\xe6+t\xe5\xa1\x13\xada\x16\xder5"w]\xa2i\xb7[\x97R \xe2IT\xcd;Z\x04dk4\xad\x8a\t\xd3\x81z\x10\xf1:^`\xab\x1f\xc5\xdc\x91N\x14$+\x9e\xae\xd3\x80'
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +000051
Tim Peters9de06bd2002-11-09 06:45:31 +000052 if has_cmdline_bunzip2:
53 def decompress(self, data):
Guido van Rossum360e4b82007-05-14 22:51:27 +000054 pop = subprocess.Popen("bunzip2", shell=True,
55 stdin=subprocess.PIPE,
56 stdout=subprocess.PIPE,
57 stderr=subprocess.STDOUT)
58 pop.stdin.write(data)
59 pop.stdin.close()
60 ret = pop.stdout.read()
61 pop.stdout.close()
Tim Peters9de06bd2002-11-09 06:45:31 +000062 if pop.wait() != 0:
63 ret = bz2.decompress(data)
64 return ret
Tim Peters499d09a2002-11-09 06:31:56 +000065
Tim Peters9de06bd2002-11-09 06:45:31 +000066 else:
Guido van Rossum360e4b82007-05-14 22:51:27 +000067 # bunzip2 isn't available to run on Windows.
Tim Peters9de06bd2002-11-09 06:45:31 +000068 def decompress(self, data):
69 return bz2.decompress(data)
70
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +000071class BZ2FileTest(BaseTest):
Georg Brandl33a5f2a2005-08-21 14:16:04 +000072 "Test BZ2File type miscellaneous methods."
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +000073
Tim Peters3de75262002-11-09 05:26:15 +000074 def setUp(self):
Tim Petersc2937042002-11-14 16:23:29 +000075 self.filename = TESTFN
Tim Peters230a60c2002-11-09 05:08:07 +000076
Tim Peters3de75262002-11-09 05:26:15 +000077 def tearDown(self):
78 if os.path.isfile(self.filename):
79 os.unlink(self.filename)
Tim Peters230a60c2002-11-09 05:08:07 +000080
Antoine Pitrou37dc5f82011-04-03 17:05:46 +020081 def getData(self, crlf=False):
82 if crlf:
83 return self.DATA_CRLF
84 else:
85 return self.DATA
86
Nadeem Vawda55b43382011-05-27 01:52:15 +020087 def createTempFile(self, crlf=False, streams=1):
Brian Curtind2e948d2010-10-13 02:56:26 +000088 with open(self.filename, "wb") as f:
Nadeem Vawda55b43382011-05-27 01:52:15 +020089 f.write(self.getData(crlf) * streams)
Tim Peters230a60c2002-11-09 05:08:07 +000090
Tim Peters3de75262002-11-09 05:26:15 +000091 def testRead(self):
92 # "Test BZ2File.read()"
93 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +000094 with BZ2File(self.filename) as bz2f:
95 self.assertRaises(TypeError, bz2f.read, None)
96 self.assertEqual(bz2f.read(), self.TEXT)
Tim Peters230a60c2002-11-09 05:08:07 +000097
Nadeem Vawda55b43382011-05-27 01:52:15 +020098 def testReadMultiStream(self):
99 # "Test BZ2File.read() with a multi stream archive"
100 self.createTempFile(streams=5)
101 with BZ2File(self.filename) as bz2f:
102 self.assertRaises(TypeError, bz2f.read, None)
103 self.assertEqual(bz2f.read(), self.TEXT * 5)
104
105 def testReadMonkeyMultiStream(self):
106 # "Test BZ2File.read() with a multi stream archive in which stream"
107 # "end is alined with internal buffer size"
108 buffer_size = bz2._BUFFER_SIZE
109 bz2._BUFFER_SIZE = len(self.DATA)
110 try:
111 self.createTempFile(streams=5)
112 with BZ2File(self.filename) as bz2f:
113 self.assertRaises(TypeError, bz2f.read, None)
114 self.assertEqual(bz2f.read(), self.TEXT * 5)
115 finally:
116 bz2._BUFFER_SIZE = buffer_size
117
Guido van Rossum75c26bc2007-08-07 23:29:20 +0000118 def testRead0(self):
Antoine Pitrou37dc5f82011-04-03 17:05:46 +0200119 # "Test BBZ2File.read(0)"
Guido van Rossum75c26bc2007-08-07 23:29:20 +0000120 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000121 with BZ2File(self.filename) as bz2f:
122 self.assertRaises(TypeError, bz2f.read, None)
123 self.assertEqual(bz2f.read(0), b"")
Guido van Rossum75c26bc2007-08-07 23:29:20 +0000124
Tim Peters3de75262002-11-09 05:26:15 +0000125 def testReadChunk10(self):
126 # "Test BZ2File.read() in chunks of 10 bytes"
127 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000128 with BZ2File(self.filename) as bz2f:
129 text = b''
130 while 1:
131 str = bz2f.read(10)
132 if not str:
133 break
134 text += str
Antoine Pitrou528d9f62011-01-26 23:29:28 +0000135 self.assertEqual(text, self.TEXT)
Tim Peters230a60c2002-11-09 05:08:07 +0000136
Nadeem Vawda55b43382011-05-27 01:52:15 +0200137 def testReadChunk10MultiStream(self):
138 # "Test BZ2File.read() in chunks of 10 bytes with a multi stream archive"
139 self.createTempFile(streams=5)
140 with BZ2File(self.filename) as bz2f:
141 text = b''
142 while 1:
143 str = bz2f.read(10)
144 if not str:
145 break
146 text += str
147 self.assertEqual(text, self.TEXT * 5)
148
Tim Peters3de75262002-11-09 05:26:15 +0000149 def testRead100(self):
150 # "Test BZ2File.read(100)"
151 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000152 with BZ2File(self.filename) as bz2f:
153 self.assertEqual(bz2f.read(100), self.TEXT[:100])
Tim Peters230a60c2002-11-09 05:08:07 +0000154
Antoine Pitrou37dc5f82011-04-03 17:05:46 +0200155 def testPeek(self):
156 # "Test BZ2File.peek()"
157 self.createTempFile()
158 with BZ2File(self.filename) as bz2f:
159 pdata = bz2f.peek()
160 self.assertNotEqual(len(pdata), 0)
161 self.assertTrue(self.TEXT.startswith(pdata))
162 self.assertEqual(bz2f.read(), self.TEXT)
163
164 def testReadInto(self):
165 # "Test BZ2File.readinto()"
166 self.createTempFile()
167 with BZ2File(self.filename) as bz2f:
168 n = 128
169 b = bytearray(n)
170 self.assertEqual(bz2f.readinto(b), n)
171 self.assertEqual(b, self.TEXT[:n])
172 n = len(self.TEXT) - n
173 b = bytearray(len(self.TEXT))
174 self.assertEqual(bz2f.readinto(b), n)
175 self.assertEqual(b[:n], self.TEXT[-n:])
176
Tim Peters3de75262002-11-09 05:26:15 +0000177 def testReadLine(self):
178 # "Test BZ2File.readline()"
179 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000180 with BZ2File(self.filename) as bz2f:
181 self.assertRaises(TypeError, bz2f.readline, None)
182 sio = BytesIO(self.TEXT)
183 for line in sio.readlines():
184 self.assertEqual(bz2f.readline(), line)
Tim Peters230a60c2002-11-09 05:08:07 +0000185
Nadeem Vawda55b43382011-05-27 01:52:15 +0200186 def testReadLineMultiStream(self):
187 # "Test BZ2File.readline() with a multi stream archive"
188 self.createTempFile(streams=5)
189 with BZ2File(self.filename) as bz2f:
190 self.assertRaises(TypeError, bz2f.readline, None)
191 sio = BytesIO(self.TEXT * 5)
192 for line in sio.readlines():
193 self.assertEqual(bz2f.readline(), line)
194
Tim Peters3de75262002-11-09 05:26:15 +0000195 def testReadLines(self):
196 # "Test BZ2File.readlines()"
197 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000198 with BZ2File(self.filename) as bz2f:
199 self.assertRaises(TypeError, bz2f.readlines, None)
200 sio = BytesIO(self.TEXT)
201 self.assertEqual(bz2f.readlines(), sio.readlines())
Tim Peters230a60c2002-11-09 05:08:07 +0000202
Nadeem Vawda55b43382011-05-27 01:52:15 +0200203 def testReadLinesMultiStream(self):
204 # "Test BZ2File.readlines() with a multi stream archive"
205 self.createTempFile(streams=5)
206 with BZ2File(self.filename) as bz2f:
207 self.assertRaises(TypeError, bz2f.readlines, None)
208 sio = BytesIO(self.TEXT * 5)
209 self.assertEqual(bz2f.readlines(), sio.readlines())
210
Tim Peters3de75262002-11-09 05:26:15 +0000211 def testIterator(self):
212 # "Test iter(BZ2File)"
213 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000214 with BZ2File(self.filename) as bz2f:
215 sio = BytesIO(self.TEXT)
216 self.assertEqual(list(iter(bz2f)), sio.readlines())
Tim Peters230a60c2002-11-09 05:08:07 +0000217
Nadeem Vawda55b43382011-05-27 01:52:15 +0200218 def testIteratorMultiStream(self):
219 # "Test iter(BZ2File) with a multi stream archive"
220 self.createTempFile(streams=5)
221 with BZ2File(self.filename) as bz2f:
222 sio = BytesIO(self.TEXT * 5)
223 self.assertEqual(list(iter(bz2f)), sio.readlines())
224
Georg Brandl86b2fb92008-07-16 03:43:04 +0000225 def testClosedIteratorDeadlock(self):
226 # "Test that iteration on a closed bz2file releases the lock."
227 # http://bugs.python.org/issue3309
228 self.createTempFile()
229 bz2f = BZ2File(self.filename)
230 bz2f.close()
231 self.assertRaises(ValueError, bz2f.__next__)
Antoine Pitrou37dc5f82011-04-03 17:05:46 +0200232 # This call will deadlock if the above .__next__ call failed to
Georg Brandl86b2fb92008-07-16 03:43:04 +0000233 # release the lock.
234 self.assertRaises(ValueError, bz2f.readlines)
235
Tim Peters3de75262002-11-09 05:26:15 +0000236 def testWrite(self):
237 # "Test BZ2File.write()"
Brian Curtind2e948d2010-10-13 02:56:26 +0000238 with BZ2File(self.filename, "w") as bz2f:
239 self.assertRaises(TypeError, bz2f.write)
240 bz2f.write(self.TEXT)
241 with open(self.filename, 'rb') as f:
242 self.assertEqual(self.decompress(f.read()), self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000243
Tim Peters3de75262002-11-09 05:26:15 +0000244 def testWriteChunks10(self):
245 # "Test BZ2File.write() with chunks of 10 bytes"
Brian Curtind2e948d2010-10-13 02:56:26 +0000246 with BZ2File(self.filename, "w") as bz2f:
247 n = 0
248 while 1:
249 str = self.TEXT[n*10:(n+1)*10]
250 if not str:
251 break
252 bz2f.write(str)
253 n += 1
254 with open(self.filename, 'rb') as f:
255 self.assertEqual(self.decompress(f.read()), self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000256
Tim Peters3de75262002-11-09 05:26:15 +0000257 def testWriteLines(self):
258 # "Test BZ2File.writelines()"
Brian Curtind2e948d2010-10-13 02:56:26 +0000259 with BZ2File(self.filename, "w") as bz2f:
260 self.assertRaises(TypeError, bz2f.writelines)
261 sio = BytesIO(self.TEXT)
262 bz2f.writelines(sio.readlines())
Thomas Wouters00ee7ba2006-08-21 19:07:27 +0000263 # patch #1535500
264 self.assertRaises(ValueError, bz2f.writelines, ["a"])
Brian Curtind2e948d2010-10-13 02:56:26 +0000265 with open(self.filename, 'rb') as f:
266 self.assertEqual(self.decompress(f.read()), self.TEXT)
Tim Peters230a60c2002-11-09 05:08:07 +0000267
Thomas Wouters00ee7ba2006-08-21 19:07:27 +0000268 def testWriteMethodsOnReadOnlyFile(self):
Brian Curtind2e948d2010-10-13 02:56:26 +0000269 with BZ2File(self.filename, "w") as bz2f:
270 bz2f.write(b"abc")
Thomas Wouters00ee7ba2006-08-21 19:07:27 +0000271
Brian Curtind2e948d2010-10-13 02:56:26 +0000272 with BZ2File(self.filename, "r") as bz2f:
273 self.assertRaises(IOError, bz2f.write, b"a")
274 self.assertRaises(IOError, bz2f.writelines, [b"a"])
Thomas Wouters00ee7ba2006-08-21 19:07:27 +0000275
Nadeem Vawda55b43382011-05-27 01:52:15 +0200276 def testAppend(self):
277 # "Test BZ2File.write()"
278 with BZ2File(self.filename, "w") as bz2f:
279 self.assertRaises(TypeError, bz2f.write)
280 bz2f.write(self.TEXT)
281 with BZ2File(self.filename, "a") as bz2f:
282 self.assertRaises(TypeError, bz2f.write)
283 bz2f.write(self.TEXT)
284 with open(self.filename, 'rb') as f:
285 self.assertEqual(self.decompress(f.read()), self.TEXT * 2)
286
Tim Peters3de75262002-11-09 05:26:15 +0000287 def testSeekForward(self):
288 # "Test BZ2File.seek(150, 0)"
289 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000290 with BZ2File(self.filename) as bz2f:
291 self.assertRaises(TypeError, bz2f.seek)
292 bz2f.seek(150)
293 self.assertEqual(bz2f.read(), self.TEXT[150:])
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000294
Nadeem Vawda55b43382011-05-27 01:52:15 +0200295 def testSeekForwardMultiStream(self):
296 # "Test BZ2File.seek(150, 0) across stream boundaries"
297 self.createTempFile(streams=2)
298 with BZ2File(self.filename) as bz2f:
299 self.assertRaises(TypeError, bz2f.seek)
300 bz2f.seek(len(self.TEXT) + 150)
301 self.assertEqual(bz2f.read(), self.TEXT[150:])
302
Tim Peters3de75262002-11-09 05:26:15 +0000303 def testSeekBackwards(self):
304 # "Test BZ2File.seek(-150, 1)"
305 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000306 with BZ2File(self.filename) as bz2f:
307 bz2f.read(500)
308 bz2f.seek(-150, 1)
309 self.assertEqual(bz2f.read(), self.TEXT[500-150:])
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000310
Nadeem Vawda55b43382011-05-27 01:52:15 +0200311 def testSeekBackwardsMultiStream(self):
312 # "Test BZ2File.seek(-150, 1) across stream boundaries"
313 self.createTempFile(streams=2)
314 with BZ2File(self.filename) as bz2f:
315 readto = len(self.TEXT) + 100
316 while readto > 0:
317 readto -= len(bz2f.read(readto))
318 bz2f.seek(-150, 1)
319 self.assertEqual(bz2f.read(), self.TEXT[100-150:] + self.TEXT)
320
Tim Peters3de75262002-11-09 05:26:15 +0000321 def testSeekBackwardsFromEnd(self):
322 # "Test BZ2File.seek(-150, 2)"
323 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000324 with BZ2File(self.filename) as bz2f:
325 bz2f.seek(-150, 2)
326 self.assertEqual(bz2f.read(), self.TEXT[len(self.TEXT)-150:])
Tim Peters230a60c2002-11-09 05:08:07 +0000327
Nadeem Vawda55b43382011-05-27 01:52:15 +0200328 def testSeekBackwardsFromEndMultiStream(self):
329 # "Test BZ2File.seek(-1000, 2) across stream boundaries"
330 self.createTempFile(streams=2)
331 with BZ2File(self.filename) as bz2f:
332 bz2f.seek(-1000, 2)
333 self.assertEqual(bz2f.read(), (self.TEXT * 2)[-1000:])
334
Tim Peters3de75262002-11-09 05:26:15 +0000335 def testSeekPostEnd(self):
336 # "Test BZ2File.seek(150000)"
337 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000338 with BZ2File(self.filename) as bz2f:
339 bz2f.seek(150000)
340 self.assertEqual(bz2f.tell(), len(self.TEXT))
341 self.assertEqual(bz2f.read(), b"")
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000342
Nadeem Vawda55b43382011-05-27 01:52:15 +0200343 def testSeekPostEndMultiStream(self):
344 # "Test BZ2File.seek(150000)"
345 self.createTempFile(streams=5)
346 with BZ2File(self.filename) as bz2f:
347 bz2f.seek(150000)
348 self.assertEqual(bz2f.tell(), len(self.TEXT) * 5)
349 self.assertEqual(bz2f.read(), b"")
350
Tim Peters3de75262002-11-09 05:26:15 +0000351 def testSeekPostEndTwice(self):
352 # "Test BZ2File.seek(150000) twice"
353 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000354 with BZ2File(self.filename) as bz2f:
355 bz2f.seek(150000)
356 bz2f.seek(150000)
357 self.assertEqual(bz2f.tell(), len(self.TEXT))
358 self.assertEqual(bz2f.read(), b"")
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000359
Nadeem Vawda55b43382011-05-27 01:52:15 +0200360 def testSeekPostEndTwiceMultiStream(self):
361 # "Test BZ2File.seek(150000) twice with a multi stream archive"
362 self.createTempFile(streams=5)
363 with BZ2File(self.filename) as bz2f:
364 bz2f.seek(150000)
365 bz2f.seek(150000)
366 self.assertEqual(bz2f.tell(), len(self.TEXT) * 5)
367 self.assertEqual(bz2f.read(), b"")
368
Tim Peters3de75262002-11-09 05:26:15 +0000369 def testSeekPreStart(self):
370 # "Test BZ2File.seek(-150, 0)"
371 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000372 with BZ2File(self.filename) as bz2f:
373 bz2f.seek(-150)
374 self.assertEqual(bz2f.tell(), 0)
375 self.assertEqual(bz2f.read(), self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000376
Nadeem Vawda55b43382011-05-27 01:52:15 +0200377 def testSeekPreStartMultiStream(self):
378 # "Test BZ2File.seek(-150, 0) with a multi stream archive"
379 self.createTempFile(streams=2)
380 with BZ2File(self.filename) as bz2f:
381 bz2f.seek(-150)
382 self.assertEqual(bz2f.tell(), 0)
383 self.assertEqual(bz2f.read(), self.TEXT * 2)
384
Antoine Pitrou37dc5f82011-04-03 17:05:46 +0200385 def testFileno(self):
386 # "Test BZ2File.fileno()"
387 self.createTempFile()
388 with open(self.filename) as rawf:
389 with BZ2File(fileobj=rawf) as bz2f:
390 self.assertEqual(bz2f.fileno(), rawf.fileno())
391
Neal Norwitz7cbd2472003-02-12 23:09:53 +0000392 def testOpenDel(self):
Gustavo Niemeyer572f5232003-04-29 14:53:08 +0000393 # "Test opening and deleting a file many times"
Neal Norwitz7cbd2472003-02-12 23:09:53 +0000394 self.createTempFile()
Guido van Rossum805365e2007-05-07 22:24:25 +0000395 for i in range(10000):
Neal Norwitz7cbd2472003-02-12 23:09:53 +0000396 o = BZ2File(self.filename)
397 del o
398
Gustavo Niemeyer572f5232003-04-29 14:53:08 +0000399 def testOpenNonexistent(self):
400 # "Test opening a nonexistent file"
401 self.assertRaises(IOError, BZ2File, "/non/existent")
402
Georg Brandl33a5f2a2005-08-21 14:16:04 +0000403 def testBug1191043(self):
404 # readlines() for files containing no newline
Guido van Rossum522a6c62007-05-22 23:13:45 +0000405 data = b'BZh91AY&SY\xd9b\x89]\x00\x00\x00\x03\x80\x04\x00\x02\x00\x0c\x00 \x00!\x9ah3M\x13<]\xc9\x14\xe1BCe\x8a%t'
Brian Curtind2e948d2010-10-13 02:56:26 +0000406 with open(self.filename, "wb") as f:
407 f.write(data)
408 with BZ2File(self.filename) as bz2f:
409 lines = bz2f.readlines()
Guido van Rossum522a6c62007-05-22 23:13:45 +0000410 self.assertEqual(lines, [b'Test'])
Brian Curtind2e948d2010-10-13 02:56:26 +0000411 with BZ2File(self.filename) as bz2f:
412 xlines = list(bz2f.readlines())
Guido van Rossum522a6c62007-05-22 23:13:45 +0000413 self.assertEqual(xlines, [b'Test'])
Georg Brandl33a5f2a2005-08-21 14:16:04 +0000414
Antoine Pitrou308705e2009-01-10 16:22:51 +0000415 def testContextProtocol(self):
416 # BZ2File supports the context management protocol
417 f = None
418 with BZ2File(self.filename, "wb") as f:
419 f.write(b"xxx")
420 f = BZ2File(self.filename, "rb")
421 f.close()
422 try:
423 with f:
424 pass
425 except ValueError:
426 pass
427 else:
428 self.fail("__enter__ on a closed file didn't raise an exception")
429 try:
430 with BZ2File(self.filename, "wb") as f:
431 1/0
432 except ZeroDivisionError:
433 pass
434 else:
435 self.fail("1/0 didn't raise an exception")
436
Victor Stinner45df8202010-04-28 22:31:17 +0000437 @unittest.skipUnless(threading, 'Threading required for this test.')
Antoine Pitrouc66363f2009-10-27 17:47:14 +0000438 def testThreading(self):
439 # Using a BZ2File from several threads doesn't deadlock (issue #7205).
440 data = b"1" * 2**20
441 nthreads = 10
Brian Curtind2e948d2010-10-13 02:56:26 +0000442 with bz2.BZ2File(self.filename, 'wb') as f:
Antoine Pitrouc66363f2009-10-27 17:47:14 +0000443 def comp():
444 for i in range(5):
445 f.write(data)
446 threads = [threading.Thread(target=comp) for i in range(nthreads)]
447 for t in threads:
448 t.start()
449 for t in threads:
450 t.join()
Antoine Pitrouc66363f2009-10-27 17:47:14 +0000451
Antoine Pitrou7ffa1962010-08-01 20:08:46 +0000452 def testMixedIterationReads(self):
Antoine Pitrou37dc5f82011-04-03 17:05:46 +0200453 # "Test mixed iteration and reads."
454 self.createTempFile()
455 linelen = len(self.TEXT_LINES[0])
456 halflen = linelen // 2
457 with bz2.BZ2File(self.filename) as bz2f:
458 bz2f.read(halflen)
459 self.assertEqual(next(bz2f), self.TEXT_LINES[0][halflen:])
460 self.assertEqual(bz2f.read(), self.TEXT[linelen:])
461 with bz2.BZ2File(self.filename) as bz2f:
462 bz2f.readline()
463 self.assertEqual(next(bz2f), self.TEXT_LINES[1])
464 self.assertEqual(bz2f.readline(), self.TEXT_LINES[2])
465 with bz2.BZ2File(self.filename) as bz2f:
466 bz2f.readlines()
467 with self.assertRaises(StopIteration):
468 next(bz2f)
469 self.assertEqual(bz2f.readlines(), [])
470
471 def testReadBytesIO(self):
472 # "Test BZ2File.read() with BytesIO source"
473 with BytesIO(self.getData()) as bio:
474 with BZ2File(fileobj=bio) as bz2f:
475 self.assertRaises(TypeError, bz2f.read, None)
476 self.assertEqual(bz2f.read(), self.TEXT)
477 self.assertFalse(bio.closed)
478
479 def testPeekBytesIO(self):
480 # "Test BZ2File.peek() with BytesIO source"
481 with BytesIO(self.getData()) as bio:
482 with BZ2File(fileobj=bio) as bz2f:
483 pdata = bz2f.peek()
484 self.assertNotEqual(len(pdata), 0)
485 self.assertTrue(self.TEXT.startswith(pdata))
486 self.assertEqual(bz2f.read(), self.TEXT)
487
488 def testWriteBytesIO(self):
489 # "Test BZ2File.write() with BytesIO destination"
490 with BytesIO() as bio:
491 with BZ2File(fileobj=bio, mode="w") as bz2f:
492 self.assertRaises(TypeError, bz2f.write)
493 bz2f.write(self.TEXT)
494 self.assertEqual(self.decompress(bio.getvalue()), self.TEXT)
495 self.assertFalse(bio.closed)
496
497 def testSeekForwardBytesIO(self):
498 # "Test BZ2File.seek(150, 0) with BytesIO source"
499 with BytesIO(self.getData()) as bio:
500 with BZ2File(fileobj=bio) as bz2f:
501 self.assertRaises(TypeError, bz2f.seek)
502 bz2f.seek(150)
503 self.assertEqual(bz2f.read(), self.TEXT[150:])
504
505 def testSeekBackwardsBytesIO(self):
506 # "Test BZ2File.seek(-150, 1) with BytesIO source"
507 with BytesIO(self.getData()) as bio:
508 with BZ2File(fileobj=bio) as bz2f:
509 bz2f.read(500)
510 bz2f.seek(-150, 1)
511 self.assertEqual(bz2f.read(), self.TEXT[500-150:])
Georg Brandl33a5f2a2005-08-21 14:16:04 +0000512
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000513class BZ2CompressorTest(BaseTest):
Tim Peters3de75262002-11-09 05:26:15 +0000514 def testCompress(self):
515 # "Test BZ2Compressor.compress()/flush()"
516 bz2c = BZ2Compressor()
Walter Dörwaldbb9c7392004-11-01 17:10:19 +0000517 self.assertRaises(TypeError, bz2c.compress)
Tim Peters3de75262002-11-09 05:26:15 +0000518 data = bz2c.compress(self.TEXT)
519 data += bz2c.flush()
520 self.assertEqual(self.decompress(data), self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000521
Tim Peters3de75262002-11-09 05:26:15 +0000522 def testCompressChunks10(self):
523 # "Test BZ2Compressor.compress()/flush() with chunks of 10 bytes"
524 bz2c = BZ2Compressor()
525 n = 0
Guido van Rossum522a6c62007-05-22 23:13:45 +0000526 data = b''
Tim Peters3de75262002-11-09 05:26:15 +0000527 while 1:
528 str = self.TEXT[n*10:(n+1)*10]
529 if not str:
530 break
531 data += bz2c.compress(str)
532 n += 1
533 data += bz2c.flush()
534 self.assertEqual(self.decompress(data), self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000535
Nadeem Vawdaea4b46f2011-04-12 23:02:42 +0200536 @precisionbigmemtest(size=_4G + 100, memuse=2)
537 def testCompress4G(self, size):
538 # "Test BZ2Compressor.compress()/flush() with >4GiB input"
539 bz2c = BZ2Compressor()
540 data = b"x" * size
541 try:
542 compressed = bz2c.compress(data)
543 compressed += bz2c.flush()
544 finally:
545 data = None # Release memory
546 data = bz2.decompress(compressed)
547 try:
548 self.assertEqual(len(data), size)
549 self.assertEqual(len(data.strip(b"x")), 0)
550 finally:
551 data = None
552
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000553class BZ2DecompressorTest(BaseTest):
Walter Dörwaldbb9c7392004-11-01 17:10:19 +0000554 def test_Constructor(self):
555 self.assertRaises(TypeError, BZ2Decompressor, 42)
556
Tim Peters3de75262002-11-09 05:26:15 +0000557 def testDecompress(self):
558 # "Test BZ2Decompressor.decompress()"
559 bz2d = BZ2Decompressor()
Walter Dörwaldbb9c7392004-11-01 17:10:19 +0000560 self.assertRaises(TypeError, bz2d.decompress)
Tim Peters3de75262002-11-09 05:26:15 +0000561 text = bz2d.decompress(self.DATA)
562 self.assertEqual(text, self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000563
Tim Peters3de75262002-11-09 05:26:15 +0000564 def testDecompressChunks10(self):
565 # "Test BZ2Decompressor.decompress() with chunks of 10 bytes"
566 bz2d = BZ2Decompressor()
Guido van Rossum522a6c62007-05-22 23:13:45 +0000567 text = b''
Tim Peters3de75262002-11-09 05:26:15 +0000568 n = 0
569 while 1:
570 str = self.DATA[n*10:(n+1)*10]
571 if not str:
572 break
573 text += bz2d.decompress(str)
574 n += 1
575 self.assertEqual(text, self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000576
Tim Peters3de75262002-11-09 05:26:15 +0000577 def testDecompressUnusedData(self):
578 # "Test BZ2Decompressor.decompress() with unused data"
579 bz2d = BZ2Decompressor()
Guido van Rossum522a6c62007-05-22 23:13:45 +0000580 unused_data = b"this is unused data"
Tim Peters3de75262002-11-09 05:26:15 +0000581 text = bz2d.decompress(self.DATA+unused_data)
582 self.assertEqual(text, self.TEXT)
583 self.assertEqual(bz2d.unused_data, unused_data)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000584
Tim Peters3de75262002-11-09 05:26:15 +0000585 def testEOFError(self):
586 # "Calling BZ2Decompressor.decompress() after EOS must raise EOFError"
587 bz2d = BZ2Decompressor()
588 text = bz2d.decompress(self.DATA)
Guido van Rossum98297ee2007-11-06 21:34:58 +0000589 self.assertRaises(EOFError, bz2d.decompress, b"anything")
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000590
Nadeem Vawdaea4b46f2011-04-12 23:02:42 +0200591 @precisionbigmemtest(size=_4G + 100, memuse=3)
592 def testDecompress4G(self, size):
593 # "Test BZ2Decompressor.decompress() with >4GiB input"
594 blocksize = 10 * 1024 * 1024
595 block = random.getrandbits(blocksize * 8).to_bytes(blocksize, 'little')
596 try:
597 data = block * (size // blocksize + 1)
598 compressed = bz2.compress(data)
599 bz2d = BZ2Decompressor()
600 decompressed = bz2d.decompress(compressed)
601 self.assertTrue(decompressed == data)
602 finally:
603 data = None
604 compressed = None
605 decompressed = None
606
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000607
608class FuncTest(BaseTest):
Tim Peters3de75262002-11-09 05:26:15 +0000609 "Test module functions"
Tim Peters230a60c2002-11-09 05:08:07 +0000610
Tim Peters3de75262002-11-09 05:26:15 +0000611 def testCompress(self):
612 # "Test compress() function"
Tim Peters499d09a2002-11-09 06:31:56 +0000613 data = bz2.compress(self.TEXT)
Tim Peters3de75262002-11-09 05:26:15 +0000614 self.assertEqual(self.decompress(data), self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000615
Tim Peters3de75262002-11-09 05:26:15 +0000616 def testDecompress(self):
617 # "Test decompress() function"
Tim Peters499d09a2002-11-09 06:31:56 +0000618 text = bz2.decompress(self.DATA)
Tim Peters3de75262002-11-09 05:26:15 +0000619 self.assertEqual(text, self.TEXT)
Tim Peters230a60c2002-11-09 05:08:07 +0000620
Tim Peters3de75262002-11-09 05:26:15 +0000621 def testDecompressEmpty(self):
622 # "Test decompress() function with empty string"
Guido van Rossum522a6c62007-05-22 23:13:45 +0000623 text = bz2.decompress(b"")
624 self.assertEqual(text, b"")
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000625
Tim Peters3de75262002-11-09 05:26:15 +0000626 def testDecompressIncomplete(self):
627 # "Test decompress() function with incomplete data"
Tim Peters499d09a2002-11-09 06:31:56 +0000628 self.assertRaises(ValueError, bz2.decompress, self.DATA[:-10])
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000629
Nadeem Vawda55b43382011-05-27 01:52:15 +0200630 def testDecompressMultiStream(self):
631 # "Test decompress() function for data with multiple streams"
632 text = bz2.decompress(self.DATA * 5)
633 self.assertEqual(text, self.TEXT * 5)
634
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000635def test_main():
Benjamin Petersonee8712c2008-05-20 21:35:26 +0000636 support.run_unittest(
Walter Dörwald21d3a322003-05-01 17:45:56 +0000637 BZ2FileTest,
638 BZ2CompressorTest,
639 BZ2DecompressorTest,
640 FuncTest
641 )
Benjamin Petersonee8712c2008-05-20 21:35:26 +0000642 support.reap_children()
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000643
644if __name__ == '__main__':
645 test_main()
646
647# vim:ts=4:sw=4