Lib/test/test_bz2.py

   1 #!/usr/bin/env python
   2 from test import test_support
   3 from test.test_support import TESTFN, import_module
   4
   5 import unittest
   6 from cStringIO import StringIO
   7 import os
   8 import subprocess
   9 import sys
  10
  11 try:
  12     import threading
  13 except ImportError:
  14     threading = None
  15
  16 bz2 = import_module('bz2')
  17 from bz2 import BZ2File, BZ2Compressor, BZ2Decompressor
  18
  19 has_cmdline_bunzip2 = sys.platform not in ("win32", "os2emx", "riscos")
  20
  21 class BaseTest(unittest.TestCase):
  22     "Base for other testcases."
  23     TEXT = 'root:x:0:0:root:/root:/bin/bash\nbin:x:1:1:bin:/bin:\ndaemon:x:2:2:daemon:/sbin:\nadm:x:3:4:adm:/var/adm:\nlp:x:4:7:lp:/var/spool/lpd:\nsync:x:5:0:sync:/sbin:/bin/sync\nshutdown:x:6:0:shutdown:/sbin:/sbin/shutdown\nhalt:x:7:0:halt:/sbin:/sbin/halt\nmail:x:8:12:mail:/var/spool/mail:\nnews:x:9:13:news:/var/spool/news:\nuucp:x:10:14:uucp:/var/spool/uucp:\noperator:x:11:0:operator:/root:\ngames:x:12:100:games:/usr/games:\ngopher:x:13:30:gopher:/usr/lib/gopher-data:\nftp:x:14:50:FTP User:/var/ftp:/bin/bash\nnobody:x:65534:65534:Nobody:/home:\npostfix:x:100:101:postfix:/var/spool/postfix:\nniemeyer:x:500:500::/home/niemeyer:/bin/bash\npostgres:x:101:102:PostgreSQL Server:/var/lib/pgsql:/bin/bash\nmysql:x:102:103:MySQL server:/var/lib/mysql:/bin/bash\nwww:x:103:104::/var/www:/bin/false\n'
  24     DATA = '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`'
  25     DATA_CRLF = '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'
  26
  27     if has_cmdline_bunzip2:
  28         def decompress(self, data):
  29             pop = subprocess.Popen("bunzip2", shell=True,
  30                                    stdin=subprocess.PIPE,
  31                                    stdout=subprocess.PIPE,
  32                                    stderr=subprocess.STDOUT)
  33             pop.stdin.write(data)
  34             pop.stdin.close()
  35             ret = pop.stdout.read()
  36             pop.stdout.close()
  37             if pop.wait() != 0:
  38                 ret = bz2.decompress(data)
  39             return ret
  40
  41     else:
  42         # bunzip2 isn't available to run on Windows.
  43         def decompress(self, data):
  44             return bz2.decompress(data)
  45
  46 class BZ2FileTest(BaseTest):
  47     "Test BZ2File type miscellaneous methods."
  48
  49     def setUp(self):
  50         self.filename = TESTFN
  51
  52     def tearDown(self):
  53         if os.path.isfile(self.filename):
  54             os.unlink(self.filename)
  55
  56     def createTempFile(self, crlf=0):
  57         f = open(self.filename, "wb")
  58         if crlf:
  59             data = self.DATA_CRLF
  60         else:
  61             data = self.DATA
  62         f.write(data)
  63         f.close()
  64
  65     def testRead(self):
  66         # "Test BZ2File.read()"
  67         self.createTempFile()
  68         bz2f = BZ2File(self.filename)
  69         self.assertRaises(TypeError, bz2f.read, None)
  70         self.assertEqual(bz2f.read(), self.TEXT)
  71         bz2f.close()
  72
  73     def testReadChunk10(self):
  74         # "Test BZ2File.read() in chunks of 10 bytes"
  75         self.createTempFile()
  76         bz2f = BZ2File(self.filename)
  77         text = ''
  78         while 1:
  79             str = bz2f.read(10)
  80             if not str:
  81                 break
  82             text += str
  83         self.assertEqual(text, text)
  84         bz2f.close()
  85
  86     def testRead100(self):
  87         # "Test BZ2File.read(100)"
  88         self.createTempFile()
  89         bz2f = BZ2File(self.filename)
  90         self.assertEqual(bz2f.read(100), self.TEXT[:100])
  91         bz2f.close()
  92
  93     def testReadLine(self):
  94         # "Test BZ2File.readline()"
  95         self.createTempFile()
  96         bz2f = BZ2File(self.filename)
  97         self.assertRaises(TypeError, bz2f.readline, None)
  98         sio = StringIO(self.TEXT)
  99         for line in sio.readlines():
 100             self.assertEqual(bz2f.readline(), line)
 101         bz2f.close()
 102
 103     def testReadLines(self):
 104         # "Test BZ2File.readlines()"
 105         self.createTempFile()
 106         bz2f = BZ2File(self.filename)
 107         self.assertRaises(TypeError, bz2f.readlines, None)
 108         sio = StringIO(self.TEXT)
 109         self.assertEqual(bz2f.readlines(), sio.readlines())
 110         bz2f.close()
 111
 112     def testIterator(self):
 113         # "Test iter(BZ2File)"
 114         self.createTempFile()
 115         bz2f = BZ2File(self.filename)
 116         sio = StringIO(self.TEXT)
 117         self.assertEqual(list(iter(bz2f)), sio.readlines())
 118         bz2f.close()
 119
 120     def testClosedIteratorDeadlock(self):
 121         # "Test that iteration on a closed bz2file releases the lock."
 122         # http://bugs.python.org/issue3309
 123         self.createTempFile()
 124         bz2f = BZ2File(self.filename)
 125         bz2f.close()
 126         self.assertRaises(ValueError, bz2f.next)
 127         # This call will deadlock of the above .next call failed to
 128         # release the lock.
 129         self.assertRaises(ValueError, bz2f.readlines)
 130
 131     def testXReadLines(self):
 132         # "Test BZ2File.xreadlines()"
 133         self.createTempFile()
 134         bz2f = BZ2File(self.filename)
 135         sio = StringIO(self.TEXT)
 136         self.assertEqual(list(bz2f.xreadlines()), sio.readlines())
 137         bz2f.close()
 138
 139     def testUniversalNewlinesLF(self):
 140         # "Test BZ2File.read() with universal newlines (\\n)"
 141         self.createTempFile()
 142         bz2f = BZ2File(self.filename, "rU")
 143         self.assertEqual(bz2f.read(), self.TEXT)
 144         self.assertEqual(bz2f.newlines, "\n")
 145         bz2f.close()
 146
 147     def testUniversalNewlinesCRLF(self):
 148         # "Test BZ2File.read() with universal newlines (\\r\\n)"
 149         self.createTempFile(crlf=1)
 150         bz2f = BZ2File(self.filename, "rU")
 151         self.assertEqual(bz2f.read(), self.TEXT)
 152         self.assertEqual(bz2f.newlines, "\r\n")
 153         bz2f.close()
 154
 155     def testWrite(self):
 156         # "Test BZ2File.write()"
 157         bz2f = BZ2File(self.filename, "w")
 158         self.assertRaises(TypeError, bz2f.write)
 159         bz2f.write(self.TEXT)
 160         bz2f.close()
 161         f = open(self.filename, 'rb')
 162         self.assertEqual(self.decompress(f.read()), self.TEXT)
 163         f.close()
 164
 165     def testWriteChunks10(self):
 166         # "Test BZ2File.write() with chunks of 10 bytes"
 167         bz2f = BZ2File(self.filename, "w")
 168         n = 0
 169         while 1:
 170             str = self.TEXT[n*10:(n+1)*10]
 171             if not str:
 172                 break
 173             bz2f.write(str)
 174             n += 1
 175         bz2f.close()
 176         f = open(self.filename, 'rb')
 177         self.assertEqual(self.decompress(f.read()), self.TEXT)
 178         f.close()
 179
 180     def testWriteLines(self):
 181         # "Test BZ2File.writelines()"
 182         bz2f = BZ2File(self.filename, "w")
 183         self.assertRaises(TypeError, bz2f.writelines)
 184         sio = StringIO(self.TEXT)
 185         bz2f.writelines(sio.readlines())
 186         bz2f.close()
 187         # patch #1535500
 188         self.assertRaises(ValueError, bz2f.writelines, ["a"])
 189         f = open(self.filename, 'rb')
 190         self.assertEqual(self.decompress(f.read()), self.TEXT)
 191         f.close()
 192
 193     def testWriteMethodsOnReadOnlyFile(self):
 194         bz2f = BZ2File(self.filename, "w")
 195         bz2f.write("abc")
 196         bz2f.close()
 197
 198         bz2f = BZ2File(self.filename, "r")
 199         self.assertRaises(IOError, bz2f.write, "a")
 200         self.assertRaises(IOError, bz2f.writelines, ["a"])
 201
 202     def testSeekForward(self):
 203         # "Test BZ2File.seek(150, 0)"
 204         self.createTempFile()
 205         bz2f = BZ2File(self.filename)
 206         self.assertRaises(TypeError, bz2f.seek)
 207         bz2f.seek(150)
 208         self.assertEqual(bz2f.read(), self.TEXT[150:])
 209         bz2f.close()
 210
 211     def testSeekBackwards(self):
 212         # "Test BZ2File.seek(-150, 1)"
 213         self.createTempFile()
 214         bz2f = BZ2File(self.filename)
 215         bz2f.read(500)
 216         bz2f.seek(-150, 1)
 217         self.assertEqual(bz2f.read(), self.TEXT[500-150:])
 218         bz2f.close()
 219
 220     def testSeekBackwardsFromEnd(self):
 221         # "Test BZ2File.seek(-150, 2)"
 222         self.createTempFile()
 223         bz2f = BZ2File(self.filename)
 224         bz2f.seek(-150, 2)
 225         self.assertEqual(bz2f.read(), self.TEXT[len(self.TEXT)-150:])
 226         bz2f.close()
 227
 228     def testSeekPostEnd(self):
 229         # "Test BZ2File.seek(150000)"
 230         self.createTempFile()
 231         bz2f = BZ2File(self.filename)
 232         bz2f.seek(150000)
 233         self.assertEqual(bz2f.tell(), len(self.TEXT))
 234         self.assertEqual(bz2f.read(), "")
 235         bz2f.close()
 236
 237     def testSeekPostEndTwice(self):
 238         # "Test BZ2File.seek(150000) twice"
 239         self.createTempFile()
 240         bz2f = BZ2File(self.filename)
 241         bz2f.seek(150000)
 242         bz2f.seek(150000)
 243         self.assertEqual(bz2f.tell(), len(self.TEXT))
 244         self.assertEqual(bz2f.read(), "")
 245         bz2f.close()
 246
 247     def testSeekPreStart(self):
 248         # "Test BZ2File.seek(-150, 0)"
 249         self.createTempFile()
 250         bz2f = BZ2File(self.filename)
 251         bz2f.seek(-150)
 252         self.assertEqual(bz2f.tell(), 0)
 253         self.assertEqual(bz2f.read(), self.TEXT)
 254         bz2f.close()
 255
 256     def testOpenDel(self):
 257         # "Test opening and deleting a file many times"
 258         self.createTempFile()
 259         for i in xrange(10000):
 260             o = BZ2File(self.filename)
 261             del o
 262
 263     def testOpenNonexistent(self):
 264         # "Test opening a nonexistent file"
 265         self.assertRaises(IOError, BZ2File, "/non/existent")
 266
 267     def testModeU(self):
 268         # Bug #1194181: bz2.BZ2File opened for write with mode "U"
 269         self.createTempFile()
 270         bz2f = BZ2File(self.filename, "U")
 271         bz2f.close()
 272         f = file(self.filename)
 273         f.seek(0, 2)
 274         self.assertEqual(f.tell(), len(self.DATA))
 275         f.close()
 276
 277     def testBug1191043(self):
 278         # readlines() for files containing no newline
 279         data = 'BZh91AY&SY\xd9b\x89]\x00\x00\x00\x03\x80\x04\x00\x02\x00\x0c\x00 \x00!\x9ah3M\x13<]\xc9\x14\xe1BCe\x8a%t'
 280         f = open(self.filename, "wb")
 281         f.write(data)
 282         f.close()
 283         bz2f = BZ2File(self.filename)
 284         lines = bz2f.readlines()
 285         bz2f.close()
 286         self.assertEqual(lines, ['Test'])
 287         bz2f = BZ2File(self.filename)
 288         xlines = list(bz2f.xreadlines())
 289         bz2f.close()
 290         self.assertEqual(xlines, ['Test'])
 291
 292     def testContextProtocol(self):
 293         # BZ2File supports the context management protocol
 294         f = None
 295         with BZ2File(self.filename, "wb") as f:
 296             f.write(b"xxx")
 297         f = BZ2File(self.filename, "rb")
 298         f.close()
 299         try:
 300             with f:
 301                 pass
 302         except ValueError:
 303             pass
 304         else:
 305             self.fail("__enter__ on a closed file didn't raise an exception")
 306         try:
 307             with BZ2File(self.filename, "wb") as f:
 308                 1 // 0
 309         except ZeroDivisionError:
 310             pass
 311         else:
 312             self.fail("1 // 0 didn't raise an exception")
 313
 314     @unittest.skipUnless(threading, 'Threading required for this test.')
 315     def testThreading(self):
 316         # Using a BZ2File from several threads doesn't deadlock (issue #7205).
 317         data = "1" * 2**20
 318         nthreads = 10
 319         f = bz2.BZ2File(self.filename, 'wb')
 320         try:
 321             def comp():
 322                 for i in range(5):
 323                     f.write(data)
 324             threads = [threading.Thread(target=comp) for i in range(nthreads)]
 325             for t in threads:
 326                 t.start()
 327             for t in threads:
 328                 t.join()
 329         finally:
 330             f.close()
 331
 332
 333 class BZ2CompressorTest(BaseTest):
 334     def testCompress(self):
 335         # "Test BZ2Compressor.compress()/flush()"
 336         bz2c = BZ2Compressor()
 337         self.assertRaises(TypeError, bz2c.compress)
 338         data = bz2c.compress(self.TEXT)
 339         data += bz2c.flush()
 340         self.assertEqual(self.decompress(data), self.TEXT)
 341
 342     def testCompressChunks10(self):
 343         # "Test BZ2Compressor.compress()/flush() with chunks of 10 bytes"
 344         bz2c = BZ2Compressor()
 345         n = 0
 346         data = ''
 347         while 1:
 348             str = self.TEXT[n*10:(n+1)*10]
 349             if not str:
 350                 break
 351             data += bz2c.compress(str)
 352             n += 1
 353         data += bz2c.flush()
 354         self.assertEqual(self.decompress(data), self.TEXT)
 355
 356 class BZ2DecompressorTest(BaseTest):
 357     def test_Constructor(self):
 358         self.assertRaises(TypeError, BZ2Decompressor, 42)
 359
 360     def testDecompress(self):
 361         # "Test BZ2Decompressor.decompress()"
 362         bz2d = BZ2Decompressor()
 363         self.assertRaises(TypeError, bz2d.decompress)
 364         text = bz2d.decompress(self.DATA)
 365         self.assertEqual(text, self.TEXT)
 366
 367     def testDecompressChunks10(self):
 368         # "Test BZ2Decompressor.decompress() with chunks of 10 bytes"
 369         bz2d = BZ2Decompressor()
 370         text = ''
 371         n = 0
 372         while 1:
 373             str = self.DATA[n*10:(n+1)*10]
 374             if not str:
 375                 break
 376             text += bz2d.decompress(str)
 377             n += 1
 378         self.assertEqual(text, self.TEXT)
 379
 380     def testDecompressUnusedData(self):
 381         # "Test BZ2Decompressor.decompress() with unused data"
 382         bz2d = BZ2Decompressor()
 383         unused_data = "this is unused data"
 384         text = bz2d.decompress(self.DATA+unused_data)
 385         self.assertEqual(text, self.TEXT)
 386         self.assertEqual(bz2d.unused_data, unused_data)
 387
 388     def testEOFError(self):
 389         # "Calling BZ2Decompressor.decompress() after EOS must raise EOFError"
 390         bz2d = BZ2Decompressor()
 391         text = bz2d.decompress(self.DATA)
 392         self.assertRaises(EOFError, bz2d.decompress, "anything")
 393
 394
 395 class FuncTest(BaseTest):
 396     "Test module functions"
 397
 398     def testCompress(self):
 399         # "Test compress() function"
 400         data = bz2.compress(self.TEXT)
 401         self.assertEqual(self.decompress(data), self.TEXT)
 402
 403     def testDecompress(self):
 404         # "Test decompress() function"
 405         text = bz2.decompress(self.DATA)
 406         self.assertEqual(text, self.TEXT)
 407
 408     def testDecompressEmpty(self):
 409         # "Test decompress() function with empty string"
 410         text = bz2.decompress("")
 411         self.assertEqual(text, "")
 412
 413     def testDecompressIncomplete(self):
 414         # "Test decompress() function with incomplete data"
 415         self.assertRaises(ValueError, bz2.decompress, self.DATA[:-10])
 416
 417 def test_main():
 418     test_support.run_unittest(
 419         BZ2FileTest,
 420         BZ2CompressorTest,
 421         BZ2DecompressorTest,
 422         FuncTest
 423     )
 424     test_support.reap_children()
 425
 426 if __name__ == '__main__':
 427     test_main()
 428
 429 # vim:ts=4:sw=4