blob: 3567b3625748828e6cebadba6f0037896bfc7a43 [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
87 def createTempFile(self, crlf=False):
Brian Curtind2e948d2010-10-13 02:56:26 +000088 with open(self.filename, "wb") as f:
Antoine Pitrou37dc5f82011-04-03 17:05:46 +020089 f.write(self.getData(crlf))
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
Guido van Rossum75c26bc2007-08-07 23:29:20 +000098 def testRead0(self):
Antoine Pitrou37dc5f82011-04-03 17:05:46 +020099 # "Test BBZ2File.read(0)"
Guido van Rossum75c26bc2007-08-07 23:29:20 +0000100 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000101 with BZ2File(self.filename) as bz2f:
102 self.assertRaises(TypeError, bz2f.read, None)
103 self.assertEqual(bz2f.read(0), b"")
Guido van Rossum75c26bc2007-08-07 23:29:20 +0000104
Tim Peters3de75262002-11-09 05:26:15 +0000105 def testReadChunk10(self):
106 # "Test BZ2File.read() in chunks of 10 bytes"
107 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000108 with BZ2File(self.filename) as bz2f:
109 text = b''
110 while 1:
111 str = bz2f.read(10)
112 if not str:
113 break
114 text += str
Antoine Pitrou528d9f62011-01-26 23:29:28 +0000115 self.assertEqual(text, self.TEXT)
Tim Peters230a60c2002-11-09 05:08:07 +0000116
Tim Peters3de75262002-11-09 05:26:15 +0000117 def testRead100(self):
118 # "Test BZ2File.read(100)"
119 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000120 with BZ2File(self.filename) as bz2f:
121 self.assertEqual(bz2f.read(100), self.TEXT[:100])
Tim Peters230a60c2002-11-09 05:08:07 +0000122
Antoine Pitrou37dc5f82011-04-03 17:05:46 +0200123 def testPeek(self):
124 # "Test BZ2File.peek()"
125 self.createTempFile()
126 with BZ2File(self.filename) as bz2f:
127 pdata = bz2f.peek()
128 self.assertNotEqual(len(pdata), 0)
129 self.assertTrue(self.TEXT.startswith(pdata))
130 self.assertEqual(bz2f.read(), self.TEXT)
131
132 def testReadInto(self):
133 # "Test BZ2File.readinto()"
134 self.createTempFile()
135 with BZ2File(self.filename) as bz2f:
136 n = 128
137 b = bytearray(n)
138 self.assertEqual(bz2f.readinto(b), n)
139 self.assertEqual(b, self.TEXT[:n])
140 n = len(self.TEXT) - n
141 b = bytearray(len(self.TEXT))
142 self.assertEqual(bz2f.readinto(b), n)
143 self.assertEqual(b[:n], self.TEXT[-n:])
144
Tim Peters3de75262002-11-09 05:26:15 +0000145 def testReadLine(self):
146 # "Test BZ2File.readline()"
147 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000148 with BZ2File(self.filename) as bz2f:
149 self.assertRaises(TypeError, bz2f.readline, None)
150 sio = BytesIO(self.TEXT)
151 for line in sio.readlines():
152 self.assertEqual(bz2f.readline(), line)
Tim Peters230a60c2002-11-09 05:08:07 +0000153
Tim Peters3de75262002-11-09 05:26:15 +0000154 def testReadLines(self):
155 # "Test BZ2File.readlines()"
156 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000157 with BZ2File(self.filename) as bz2f:
158 self.assertRaises(TypeError, bz2f.readlines, None)
159 sio = BytesIO(self.TEXT)
160 self.assertEqual(bz2f.readlines(), sio.readlines())
Tim Peters230a60c2002-11-09 05:08:07 +0000161
Tim Peters3de75262002-11-09 05:26:15 +0000162 def testIterator(self):
163 # "Test iter(BZ2File)"
164 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000165 with BZ2File(self.filename) as bz2f:
166 sio = BytesIO(self.TEXT)
167 self.assertEqual(list(iter(bz2f)), sio.readlines())
Tim Peters230a60c2002-11-09 05:08:07 +0000168
Georg Brandl86b2fb92008-07-16 03:43:04 +0000169 def testClosedIteratorDeadlock(self):
170 # "Test that iteration on a closed bz2file releases the lock."
171 # http://bugs.python.org/issue3309
172 self.createTempFile()
173 bz2f = BZ2File(self.filename)
174 bz2f.close()
175 self.assertRaises(ValueError, bz2f.__next__)
Antoine Pitrou37dc5f82011-04-03 17:05:46 +0200176 # This call will deadlock if the above .__next__ call failed to
Georg Brandl86b2fb92008-07-16 03:43:04 +0000177 # release the lock.
178 self.assertRaises(ValueError, bz2f.readlines)
179
Tim Peters3de75262002-11-09 05:26:15 +0000180 def testWrite(self):
181 # "Test BZ2File.write()"
Brian Curtind2e948d2010-10-13 02:56:26 +0000182 with BZ2File(self.filename, "w") as bz2f:
183 self.assertRaises(TypeError, bz2f.write)
184 bz2f.write(self.TEXT)
185 with open(self.filename, 'rb') as f:
186 self.assertEqual(self.decompress(f.read()), self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000187
Tim Peters3de75262002-11-09 05:26:15 +0000188 def testWriteChunks10(self):
189 # "Test BZ2File.write() with chunks of 10 bytes"
Brian Curtind2e948d2010-10-13 02:56:26 +0000190 with BZ2File(self.filename, "w") as bz2f:
191 n = 0
192 while 1:
193 str = self.TEXT[n*10:(n+1)*10]
194 if not str:
195 break
196 bz2f.write(str)
197 n += 1
198 with open(self.filename, 'rb') as f:
199 self.assertEqual(self.decompress(f.read()), self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000200
Tim Peters3de75262002-11-09 05:26:15 +0000201 def testWriteLines(self):
202 # "Test BZ2File.writelines()"
Brian Curtind2e948d2010-10-13 02:56:26 +0000203 with BZ2File(self.filename, "w") as bz2f:
204 self.assertRaises(TypeError, bz2f.writelines)
205 sio = BytesIO(self.TEXT)
206 bz2f.writelines(sio.readlines())
Thomas Wouters00ee7ba2006-08-21 19:07:27 +0000207 # patch #1535500
208 self.assertRaises(ValueError, bz2f.writelines, ["a"])
Brian Curtind2e948d2010-10-13 02:56:26 +0000209 with open(self.filename, 'rb') as f:
210 self.assertEqual(self.decompress(f.read()), self.TEXT)
Tim Peters230a60c2002-11-09 05:08:07 +0000211
Thomas Wouters00ee7ba2006-08-21 19:07:27 +0000212 def testWriteMethodsOnReadOnlyFile(self):
Brian Curtind2e948d2010-10-13 02:56:26 +0000213 with BZ2File(self.filename, "w") as bz2f:
214 bz2f.write(b"abc")
Thomas Wouters00ee7ba2006-08-21 19:07:27 +0000215
Brian Curtind2e948d2010-10-13 02:56:26 +0000216 with BZ2File(self.filename, "r") as bz2f:
217 self.assertRaises(IOError, bz2f.write, b"a")
218 self.assertRaises(IOError, bz2f.writelines, [b"a"])
Thomas Wouters00ee7ba2006-08-21 19:07:27 +0000219
Tim Peters3de75262002-11-09 05:26:15 +0000220 def testSeekForward(self):
221 # "Test BZ2File.seek(150, 0)"
222 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000223 with BZ2File(self.filename) as bz2f:
224 self.assertRaises(TypeError, bz2f.seek)
225 bz2f.seek(150)
226 self.assertEqual(bz2f.read(), self.TEXT[150:])
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000227
Tim Peters3de75262002-11-09 05:26:15 +0000228 def testSeekBackwards(self):
229 # "Test BZ2File.seek(-150, 1)"
230 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000231 with BZ2File(self.filename) as bz2f:
232 bz2f.read(500)
233 bz2f.seek(-150, 1)
234 self.assertEqual(bz2f.read(), self.TEXT[500-150:])
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000235
Tim Peters3de75262002-11-09 05:26:15 +0000236 def testSeekBackwardsFromEnd(self):
237 # "Test BZ2File.seek(-150, 2)"
238 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000239 with BZ2File(self.filename) as bz2f:
240 bz2f.seek(-150, 2)
241 self.assertEqual(bz2f.read(), self.TEXT[len(self.TEXT)-150:])
Tim Peters230a60c2002-11-09 05:08:07 +0000242
Tim Peters3de75262002-11-09 05:26:15 +0000243 def testSeekPostEnd(self):
244 # "Test BZ2File.seek(150000)"
245 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000246 with BZ2File(self.filename) as bz2f:
247 bz2f.seek(150000)
248 self.assertEqual(bz2f.tell(), len(self.TEXT))
249 self.assertEqual(bz2f.read(), b"")
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000250
Tim Peters3de75262002-11-09 05:26:15 +0000251 def testSeekPostEndTwice(self):
252 # "Test BZ2File.seek(150000) twice"
253 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000254 with BZ2File(self.filename) as bz2f:
255 bz2f.seek(150000)
256 bz2f.seek(150000)
257 self.assertEqual(bz2f.tell(), len(self.TEXT))
258 self.assertEqual(bz2f.read(), b"")
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000259
Tim Peters3de75262002-11-09 05:26:15 +0000260 def testSeekPreStart(self):
261 # "Test BZ2File.seek(-150, 0)"
262 self.createTempFile()
Brian Curtind2e948d2010-10-13 02:56:26 +0000263 with BZ2File(self.filename) as bz2f:
264 bz2f.seek(-150)
265 self.assertEqual(bz2f.tell(), 0)
266 self.assertEqual(bz2f.read(), self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000267
Antoine Pitrou37dc5f82011-04-03 17:05:46 +0200268 def testFileno(self):
269 # "Test BZ2File.fileno()"
270 self.createTempFile()
271 with open(self.filename) as rawf:
272 with BZ2File(fileobj=rawf) as bz2f:
273 self.assertEqual(bz2f.fileno(), rawf.fileno())
274
Neal Norwitz7cbd2472003-02-12 23:09:53 +0000275 def testOpenDel(self):
Gustavo Niemeyer572f5232003-04-29 14:53:08 +0000276 # "Test opening and deleting a file many times"
Neal Norwitz7cbd2472003-02-12 23:09:53 +0000277 self.createTempFile()
Guido van Rossum805365e2007-05-07 22:24:25 +0000278 for i in range(10000):
Neal Norwitz7cbd2472003-02-12 23:09:53 +0000279 o = BZ2File(self.filename)
280 del o
281
Gustavo Niemeyer572f5232003-04-29 14:53:08 +0000282 def testOpenNonexistent(self):
283 # "Test opening a nonexistent file"
284 self.assertRaises(IOError, BZ2File, "/non/existent")
285
Georg Brandl33a5f2a2005-08-21 14:16:04 +0000286 def testBug1191043(self):
287 # readlines() for files containing no newline
Guido van Rossum522a6c62007-05-22 23:13:45 +0000288 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 +0000289 with open(self.filename, "wb") as f:
290 f.write(data)
291 with BZ2File(self.filename) as bz2f:
292 lines = bz2f.readlines()
Guido van Rossum522a6c62007-05-22 23:13:45 +0000293 self.assertEqual(lines, [b'Test'])
Brian Curtind2e948d2010-10-13 02:56:26 +0000294 with BZ2File(self.filename) as bz2f:
295 xlines = list(bz2f.readlines())
Guido van Rossum522a6c62007-05-22 23:13:45 +0000296 self.assertEqual(xlines, [b'Test'])
Georg Brandl33a5f2a2005-08-21 14:16:04 +0000297
Antoine Pitrou308705e2009-01-10 16:22:51 +0000298 def testContextProtocol(self):
299 # BZ2File supports the context management protocol
300 f = None
301 with BZ2File(self.filename, "wb") as f:
302 f.write(b"xxx")
303 f = BZ2File(self.filename, "rb")
304 f.close()
305 try:
306 with f:
307 pass
308 except ValueError:
309 pass
310 else:
311 self.fail("__enter__ on a closed file didn't raise an exception")
312 try:
313 with BZ2File(self.filename, "wb") as f:
314 1/0
315 except ZeroDivisionError:
316 pass
317 else:
318 self.fail("1/0 didn't raise an exception")
319
Victor Stinner45df8202010-04-28 22:31:17 +0000320 @unittest.skipUnless(threading, 'Threading required for this test.')
Antoine Pitrouc66363f2009-10-27 17:47:14 +0000321 def testThreading(self):
322 # Using a BZ2File from several threads doesn't deadlock (issue #7205).
323 data = b"1" * 2**20
324 nthreads = 10
Brian Curtind2e948d2010-10-13 02:56:26 +0000325 with bz2.BZ2File(self.filename, 'wb') as f:
Antoine Pitrouc66363f2009-10-27 17:47:14 +0000326 def comp():
327 for i in range(5):
328 f.write(data)
329 threads = [threading.Thread(target=comp) for i in range(nthreads)]
330 for t in threads:
331 t.start()
332 for t in threads:
333 t.join()
Antoine Pitrouc66363f2009-10-27 17:47:14 +0000334
Antoine Pitrou7ffa1962010-08-01 20:08:46 +0000335 def testMixedIterationReads(self):
Antoine Pitrou37dc5f82011-04-03 17:05:46 +0200336 # "Test mixed iteration and reads."
337 self.createTempFile()
338 linelen = len(self.TEXT_LINES[0])
339 halflen = linelen // 2
340 with bz2.BZ2File(self.filename) as bz2f:
341 bz2f.read(halflen)
342 self.assertEqual(next(bz2f), self.TEXT_LINES[0][halflen:])
343 self.assertEqual(bz2f.read(), self.TEXT[linelen:])
344 with bz2.BZ2File(self.filename) as bz2f:
345 bz2f.readline()
346 self.assertEqual(next(bz2f), self.TEXT_LINES[1])
347 self.assertEqual(bz2f.readline(), self.TEXT_LINES[2])
348 with bz2.BZ2File(self.filename) as bz2f:
349 bz2f.readlines()
350 with self.assertRaises(StopIteration):
351 next(bz2f)
352 self.assertEqual(bz2f.readlines(), [])
353
354 def testReadBytesIO(self):
355 # "Test BZ2File.read() with BytesIO source"
356 with BytesIO(self.getData()) as bio:
357 with BZ2File(fileobj=bio) as bz2f:
358 self.assertRaises(TypeError, bz2f.read, None)
359 self.assertEqual(bz2f.read(), self.TEXT)
360 self.assertFalse(bio.closed)
361
362 def testPeekBytesIO(self):
363 # "Test BZ2File.peek() with BytesIO source"
364 with BytesIO(self.getData()) as bio:
365 with BZ2File(fileobj=bio) as bz2f:
366 pdata = bz2f.peek()
367 self.assertNotEqual(len(pdata), 0)
368 self.assertTrue(self.TEXT.startswith(pdata))
369 self.assertEqual(bz2f.read(), self.TEXT)
370
371 def testWriteBytesIO(self):
372 # "Test BZ2File.write() with BytesIO destination"
373 with BytesIO() as bio:
374 with BZ2File(fileobj=bio, mode="w") as bz2f:
375 self.assertRaises(TypeError, bz2f.write)
376 bz2f.write(self.TEXT)
377 self.assertEqual(self.decompress(bio.getvalue()), self.TEXT)
378 self.assertFalse(bio.closed)
379
380 def testSeekForwardBytesIO(self):
381 # "Test BZ2File.seek(150, 0) with BytesIO source"
382 with BytesIO(self.getData()) as bio:
383 with BZ2File(fileobj=bio) as bz2f:
384 self.assertRaises(TypeError, bz2f.seek)
385 bz2f.seek(150)
386 self.assertEqual(bz2f.read(), self.TEXT[150:])
387
388 def testSeekBackwardsBytesIO(self):
389 # "Test BZ2File.seek(-150, 1) with BytesIO source"
390 with BytesIO(self.getData()) as bio:
391 with BZ2File(fileobj=bio) as bz2f:
392 bz2f.read(500)
393 bz2f.seek(-150, 1)
394 self.assertEqual(bz2f.read(), self.TEXT[500-150:])
Georg Brandl33a5f2a2005-08-21 14:16:04 +0000395
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000396class BZ2CompressorTest(BaseTest):
Tim Peters3de75262002-11-09 05:26:15 +0000397 def testCompress(self):
398 # "Test BZ2Compressor.compress()/flush()"
399 bz2c = BZ2Compressor()
Walter Dörwaldbb9c7392004-11-01 17:10:19 +0000400 self.assertRaises(TypeError, bz2c.compress)
Tim Peters3de75262002-11-09 05:26:15 +0000401 data = bz2c.compress(self.TEXT)
402 data += bz2c.flush()
403 self.assertEqual(self.decompress(data), self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000404
Tim Peters3de75262002-11-09 05:26:15 +0000405 def testCompressChunks10(self):
406 # "Test BZ2Compressor.compress()/flush() with chunks of 10 bytes"
407 bz2c = BZ2Compressor()
408 n = 0
Guido van Rossum522a6c62007-05-22 23:13:45 +0000409 data = b''
Tim Peters3de75262002-11-09 05:26:15 +0000410 while 1:
411 str = self.TEXT[n*10:(n+1)*10]
412 if not str:
413 break
414 data += bz2c.compress(str)
415 n += 1
416 data += bz2c.flush()
417 self.assertEqual(self.decompress(data), self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000418
Nadeem Vawdaea4b46f2011-04-12 23:02:42 +0200419 @precisionbigmemtest(size=_4G + 100, memuse=2)
420 def testCompress4G(self, size):
421 # "Test BZ2Compressor.compress()/flush() with >4GiB input"
422 bz2c = BZ2Compressor()
423 data = b"x" * size
424 try:
425 compressed = bz2c.compress(data)
426 compressed += bz2c.flush()
427 finally:
428 data = None # Release memory
429 data = bz2.decompress(compressed)
430 try:
431 self.assertEqual(len(data), size)
432 self.assertEqual(len(data.strip(b"x")), 0)
433 finally:
434 data = None
435
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000436class BZ2DecompressorTest(BaseTest):
Walter Dörwaldbb9c7392004-11-01 17:10:19 +0000437 def test_Constructor(self):
438 self.assertRaises(TypeError, BZ2Decompressor, 42)
439
Tim Peters3de75262002-11-09 05:26:15 +0000440 def testDecompress(self):
441 # "Test BZ2Decompressor.decompress()"
442 bz2d = BZ2Decompressor()
Walter Dörwaldbb9c7392004-11-01 17:10:19 +0000443 self.assertRaises(TypeError, bz2d.decompress)
Tim Peters3de75262002-11-09 05:26:15 +0000444 text = bz2d.decompress(self.DATA)
445 self.assertEqual(text, self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000446
Tim Peters3de75262002-11-09 05:26:15 +0000447 def testDecompressChunks10(self):
448 # "Test BZ2Decompressor.decompress() with chunks of 10 bytes"
449 bz2d = BZ2Decompressor()
Guido van Rossum522a6c62007-05-22 23:13:45 +0000450 text = b''
Tim Peters3de75262002-11-09 05:26:15 +0000451 n = 0
452 while 1:
453 str = self.DATA[n*10:(n+1)*10]
454 if not str:
455 break
456 text += bz2d.decompress(str)
457 n += 1
458 self.assertEqual(text, self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000459
Tim Peters3de75262002-11-09 05:26:15 +0000460 def testDecompressUnusedData(self):
461 # "Test BZ2Decompressor.decompress() with unused data"
462 bz2d = BZ2Decompressor()
Guido van Rossum522a6c62007-05-22 23:13:45 +0000463 unused_data = b"this is unused data"
Tim Peters3de75262002-11-09 05:26:15 +0000464 text = bz2d.decompress(self.DATA+unused_data)
465 self.assertEqual(text, self.TEXT)
466 self.assertEqual(bz2d.unused_data, unused_data)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000467
Tim Peters3de75262002-11-09 05:26:15 +0000468 def testEOFError(self):
469 # "Calling BZ2Decompressor.decompress() after EOS must raise EOFError"
470 bz2d = BZ2Decompressor()
471 text = bz2d.decompress(self.DATA)
Guido van Rossum98297ee2007-11-06 21:34:58 +0000472 self.assertRaises(EOFError, bz2d.decompress, b"anything")
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000473
Nadeem Vawdaea4b46f2011-04-12 23:02:42 +0200474 @precisionbigmemtest(size=_4G + 100, memuse=3)
475 def testDecompress4G(self, size):
476 # "Test BZ2Decompressor.decompress() with >4GiB input"
477 blocksize = 10 * 1024 * 1024
478 block = random.getrandbits(blocksize * 8).to_bytes(blocksize, 'little')
479 try:
480 data = block * (size // blocksize + 1)
481 compressed = bz2.compress(data)
482 bz2d = BZ2Decompressor()
483 decompressed = bz2d.decompress(compressed)
484 self.assertTrue(decompressed == data)
485 finally:
486 data = None
487 compressed = None
488 decompressed = None
489
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000490
491class FuncTest(BaseTest):
Tim Peters3de75262002-11-09 05:26:15 +0000492 "Test module functions"
Tim Peters230a60c2002-11-09 05:08:07 +0000493
Tim Peters3de75262002-11-09 05:26:15 +0000494 def testCompress(self):
495 # "Test compress() function"
Tim Peters499d09a2002-11-09 06:31:56 +0000496 data = bz2.compress(self.TEXT)
Tim Peters3de75262002-11-09 05:26:15 +0000497 self.assertEqual(self.decompress(data), self.TEXT)
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000498
Tim Peters3de75262002-11-09 05:26:15 +0000499 def testDecompress(self):
500 # "Test decompress() function"
Tim Peters499d09a2002-11-09 06:31:56 +0000501 text = bz2.decompress(self.DATA)
Tim Peters3de75262002-11-09 05:26:15 +0000502 self.assertEqual(text, self.TEXT)
Tim Peters230a60c2002-11-09 05:08:07 +0000503
Tim Peters3de75262002-11-09 05:26:15 +0000504 def testDecompressEmpty(self):
505 # "Test decompress() function with empty string"
Guido van Rossum522a6c62007-05-22 23:13:45 +0000506 text = bz2.decompress(b"")
507 self.assertEqual(text, b"")
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000508
Tim Peters3de75262002-11-09 05:26:15 +0000509 def testDecompressIncomplete(self):
510 # "Test decompress() function with incomplete data"
Tim Peters499d09a2002-11-09 06:31:56 +0000511 self.assertRaises(ValueError, bz2.decompress, self.DATA[:-10])
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000512
513def test_main():
Benjamin Petersonee8712c2008-05-20 21:35:26 +0000514 support.run_unittest(
Walter Dörwald21d3a322003-05-01 17:45:56 +0000515 BZ2FileTest,
516 BZ2CompressorTest,
517 BZ2DecompressorTest,
518 FuncTest
519 )
Benjamin Petersonee8712c2008-05-20 21:35:26 +0000520 support.reap_children()
Gustavo Niemeyerf8ca8362002-11-05 16:50:05 +0000521
522if __name__ == '__main__':
523 test_main()
524
525# vim:ts=4:sw=4