1#!/usr/bin/env python 2from test import test_support 3from test.test_support import TESTFN, _4G, bigmemtest, import_module, findfile 4 5import unittest 6from cStringIO import StringIO 7import os 8import subprocess 9import sys 10 11try: 12 import threading 13except ImportError: 14 threading = None 15 16bz2 = import_module('bz2') 17from bz2 import BZ2File, BZ2Compressor, BZ2Decompressor 18 19has_cmdline_bunzip2 = sys.platform not in ("win32", "os2emx", "riscos") 20 21class 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 EMPTY_DATA = 'BZh9\x17rE8P\x90\x00\x00\x00\x00' 27 28 if has_cmdline_bunzip2: 29 def decompress(self, data): 30 pop = subprocess.Popen("bunzip2", shell=True, 31 stdin=subprocess.PIPE, 32 stdout=subprocess.PIPE, 33 stderr=subprocess.STDOUT) 34 pop.stdin.write(data) 35 pop.stdin.close() 36 ret = pop.stdout.read() 37 pop.stdout.close() 38 if pop.wait() != 0: 39 ret = bz2.decompress(data) 40 return ret 41 42 else: 43 # bunzip2 isn't available to run on Windows. 44 def decompress(self, data): 45 return bz2.decompress(data) 46 47 48class BZ2FileTest(BaseTest): 49 "Test BZ2File type miscellaneous methods." 50 51 def setUp(self): 52 self.filename = TESTFN 53 54 def tearDown(self): 55 if os.path.isfile(self.filename): 56 os.unlink(self.filename) 57 58 def createTempFile(self, crlf=0): 59 with open(self.filename, "wb") as f: 60 if crlf: 61 data = self.DATA_CRLF 62 else: 63 data = self.DATA 64 f.write(data) 65 66 def testRead(self): 67 # "Test BZ2File.read()" 68 self.createTempFile() 69 with BZ2File(self.filename) as bz2f: 70 self.assertRaises(TypeError, bz2f.read, None) 71 self.assertEqual(bz2f.read(), self.TEXT) 72 73 def testRead0(self): 74 # Test BBZ2File.read(0)" 75 self.createTempFile() 76 with BZ2File(self.filename) as bz2f: 77 self.assertRaises(TypeError, bz2f.read, None) 78 self.assertEqual(bz2f.read(0), "") 79 80 def testReadChunk10(self): 81 # "Test BZ2File.read() in chunks of 10 bytes" 82 self.createTempFile() 83 with BZ2File(self.filename) as bz2f: 84 text = '' 85 while 1: 86 str = bz2f.read(10) 87 if not str: 88 break 89 text += str 90 self.assertEqual(text, self.TEXT) 91 92 def testRead100(self): 93 # "Test BZ2File.read(100)" 94 self.createTempFile() 95 with BZ2File(self.filename) as bz2f: 96 self.assertEqual(bz2f.read(100), self.TEXT[:100]) 97 98 def testReadLine(self): 99 # "Test BZ2File.readline()" 100 self.createTempFile() 101 with BZ2File(self.filename) as bz2f: 102 self.assertRaises(TypeError, bz2f.readline, None) 103 sio = StringIO(self.TEXT) 104 for line in sio.readlines(): 105 self.assertEqual(bz2f.readline(), line) 106 107 def testReadLines(self): 108 # "Test BZ2File.readlines()" 109 self.createTempFile() 110 with BZ2File(self.filename) as bz2f: 111 self.assertRaises(TypeError, bz2f.readlines, None) 112 sio = StringIO(self.TEXT) 113 self.assertEqual(bz2f.readlines(), sio.readlines()) 114 115 def testIterator(self): 116 # "Test iter(BZ2File)" 117 self.createTempFile() 118 with BZ2File(self.filename) as bz2f: 119 sio = StringIO(self.TEXT) 120 self.assertEqual(list(iter(bz2f)), sio.readlines()) 121 122 def testClosedIteratorDeadlock(self): 123 # "Test that iteration on a closed bz2file releases the lock." 124 # http://bugs.python.org/issue3309 125 self.createTempFile() 126 bz2f = BZ2File(self.filename) 127 bz2f.close() 128 self.assertRaises(ValueError, bz2f.next) 129 # This call will deadlock of the above .next call failed to 130 # release the lock. 131 self.assertRaises(ValueError, bz2f.readlines) 132 133 def testXReadLines(self): 134 # "Test BZ2File.xreadlines()" 135 self.createTempFile() 136 bz2f = BZ2File(self.filename) 137 sio = StringIO(self.TEXT) 138 self.assertEqual(list(bz2f.xreadlines()), sio.readlines()) 139 bz2f.close() 140 141 def testUniversalNewlinesLF(self): 142 # "Test BZ2File.read() with universal newlines (\\n)" 143 self.createTempFile() 144 bz2f = BZ2File(self.filename, "rU") 145 self.assertEqual(bz2f.read(), self.TEXT) 146 self.assertEqual(bz2f.newlines, "\n") 147 bz2f.close() 148 149 def testUniversalNewlinesCRLF(self): 150 # "Test BZ2File.read() with universal newlines (\\r\\n)" 151 self.createTempFile(crlf=1) 152 bz2f = BZ2File(self.filename, "rU") 153 self.assertEqual(bz2f.read(), self.TEXT) 154 self.assertEqual(bz2f.newlines, "\r\n") 155 bz2f.close() 156 157 def testWrite(self): 158 # "Test BZ2File.write()" 159 with BZ2File(self.filename, "w") as bz2f: 160 self.assertRaises(TypeError, bz2f.write) 161 bz2f.write(self.TEXT) 162 with open(self.filename, 'rb') as f: 163 self.assertEqual(self.decompress(f.read()), self.TEXT) 164 165 def testWriteChunks10(self): 166 # "Test BZ2File.write() with chunks of 10 bytes" 167 with BZ2File(self.filename, "w") as bz2f: 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 with open(self.filename, 'rb') as f: 176 self.assertEqual(self.decompress(f.read()), self.TEXT) 177 178 def testWriteLines(self): 179 # "Test BZ2File.writelines()" 180 with BZ2File(self.filename, "w") as bz2f: 181 self.assertRaises(TypeError, bz2f.writelines) 182 sio = StringIO(self.TEXT) 183 bz2f.writelines(sio.readlines()) 184 # patch #1535500 185 self.assertRaises(ValueError, bz2f.writelines, ["a"]) 186 with open(self.filename, 'rb') as f: 187 self.assertEqual(self.decompress(f.read()), self.TEXT) 188 189 def testWriteMethodsOnReadOnlyFile(self): 190 with BZ2File(self.filename, "w") as bz2f: 191 bz2f.write("abc") 192 193 with BZ2File(self.filename, "r") as bz2f: 194 self.assertRaises(IOError, bz2f.write, "a") 195 self.assertRaises(IOError, bz2f.writelines, ["a"]) 196 197 def testSeekForward(self): 198 # "Test BZ2File.seek(150, 0)" 199 self.createTempFile() 200 with BZ2File(self.filename) as bz2f: 201 self.assertRaises(TypeError, bz2f.seek) 202 bz2f.seek(150) 203 self.assertEqual(bz2f.read(), self.TEXT[150:]) 204 205 def testSeekBackwards(self): 206 # "Test BZ2File.seek(-150, 1)" 207 self.createTempFile() 208 with BZ2File(self.filename) as bz2f: 209 bz2f.read(500) 210 bz2f.seek(-150, 1) 211 self.assertEqual(bz2f.read(), self.TEXT[500-150:]) 212 213 def testSeekBackwardsFromEnd(self): 214 # "Test BZ2File.seek(-150, 2)" 215 self.createTempFile() 216 with BZ2File(self.filename) as bz2f: 217 bz2f.seek(-150, 2) 218 self.assertEqual(bz2f.read(), self.TEXT[len(self.TEXT)-150:]) 219 220 def testSeekPostEnd(self): 221 # "Test BZ2File.seek(150000)" 222 self.createTempFile() 223 with BZ2File(self.filename) as bz2f: 224 bz2f.seek(150000) 225 self.assertEqual(bz2f.tell(), len(self.TEXT)) 226 self.assertEqual(bz2f.read(), "") 227 228 def testSeekPostEndTwice(self): 229 # "Test BZ2File.seek(150000) twice" 230 self.createTempFile() 231 with BZ2File(self.filename) as bz2f: 232 bz2f.seek(150000) 233 bz2f.seek(150000) 234 self.assertEqual(bz2f.tell(), len(self.TEXT)) 235 self.assertEqual(bz2f.read(), "") 236 237 def testSeekPreStart(self): 238 # "Test BZ2File.seek(-150, 0)" 239 self.createTempFile() 240 with BZ2File(self.filename) as bz2f: 241 bz2f.seek(-150) 242 self.assertEqual(bz2f.tell(), 0) 243 self.assertEqual(bz2f.read(), self.TEXT) 244 245 def testOpenDel(self): 246 # "Test opening and deleting a file many times" 247 self.createTempFile() 248 for i in xrange(10000): 249 o = BZ2File(self.filename) 250 del o 251 252 def testOpenNonexistent(self): 253 # "Test opening a nonexistent file" 254 self.assertRaises(IOError, BZ2File, "/non/existent") 255 256 def testModeU(self): 257 # Bug #1194181: bz2.BZ2File opened for write with mode "U" 258 self.createTempFile() 259 bz2f = BZ2File(self.filename, "U") 260 bz2f.close() 261 f = file(self.filename) 262 f.seek(0, 2) 263 self.assertEqual(f.tell(), len(self.DATA)) 264 f.close() 265 266 def testBug1191043(self): 267 # readlines() for files containing no newline 268 data = 'BZh91AY&SY\xd9b\x89]\x00\x00\x00\x03\x80\x04\x00\x02\x00\x0c\x00 \x00!\x9ah3M\x13<]\xc9\x14\xe1BCe\x8a%t' 269 with open(self.filename, "wb") as f: 270 f.write(data) 271 with BZ2File(self.filename) as bz2f: 272 lines = bz2f.readlines() 273 self.assertEqual(lines, ['Test']) 274 with BZ2File(self.filename) as bz2f: 275 xlines = list(bz2f.readlines()) 276 self.assertEqual(xlines, ['Test']) 277 278 def testContextProtocol(self): 279 # BZ2File supports the context management protocol 280 f = None 281 with BZ2File(self.filename, "wb") as f: 282 f.write(b"xxx") 283 f = BZ2File(self.filename, "rb") 284 f.close() 285 try: 286 with f: 287 pass 288 except ValueError: 289 pass 290 else: 291 self.fail("__enter__ on a closed file didn't raise an exception") 292 try: 293 with BZ2File(self.filename, "wb") as f: 294 1 // 0 295 except ZeroDivisionError: 296 pass 297 else: 298 self.fail("1 // 0 didn't raise an exception") 299 300 @unittest.skipUnless(threading, 'Threading required for this test.') 301 def testThreading(self): 302 # Using a BZ2File from several threads doesn't deadlock (issue #7205). 303 data = "1" * 2**20 304 nthreads = 10 305 with bz2.BZ2File(self.filename, 'wb') as f: 306 def comp(): 307 for i in range(5): 308 f.write(data) 309 threads = [threading.Thread(target=comp) for i in range(nthreads)] 310 for t in threads: 311 t.start() 312 for t in threads: 313 t.join() 314 315 def testMixedIterationReads(self): 316 # Issue #8397: mixed iteration and reads should be forbidden. 317 with bz2.BZ2File(self.filename, 'wb') as f: 318 # The internal buffer size is hard-wired to 8192 bytes, we must 319 # write out more than that for the test to stop half through 320 # the buffer. 321 f.write(self.TEXT * 100) 322 with bz2.BZ2File(self.filename, 'rb') as f: 323 next(f) 324 self.assertRaises(ValueError, f.read) 325 self.assertRaises(ValueError, f.readline) 326 self.assertRaises(ValueError, f.readlines) 327 328class BZ2CompressorTest(BaseTest): 329 def testCompress(self): 330 # "Test BZ2Compressor.compress()/flush()" 331 bz2c = BZ2Compressor() 332 self.assertRaises(TypeError, bz2c.compress) 333 data = bz2c.compress(self.TEXT) 334 data += bz2c.flush() 335 self.assertEqual(self.decompress(data), self.TEXT) 336 337 def testCompressEmptyString(self): 338 # "Test BZ2Compressor.compress()/flush() of empty string" 339 bz2c = BZ2Compressor() 340 data = bz2c.compress('') 341 data += bz2c.flush() 342 self.assertEqual(data, self.EMPTY_DATA) 343 344 def testCompressChunks10(self): 345 # "Test BZ2Compressor.compress()/flush() with chunks of 10 bytes" 346 bz2c = BZ2Compressor() 347 n = 0 348 data = '' 349 while 1: 350 str = self.TEXT[n*10:(n+1)*10] 351 if not str: 352 break 353 data += bz2c.compress(str) 354 n += 1 355 data += bz2c.flush() 356 self.assertEqual(self.decompress(data), self.TEXT) 357 358 @bigmemtest(_4G, memuse=1.25) 359 def testBigmem(self, size): 360 text = "a" * size 361 bz2c = bz2.BZ2Compressor() 362 data = bz2c.compress(text) + bz2c.flush() 363 del text 364 text = self.decompress(data) 365 self.assertEqual(len(text), size) 366 self.assertEqual(text.strip("a"), "") 367 368 369class BZ2DecompressorTest(BaseTest): 370 def test_Constructor(self): 371 self.assertRaises(TypeError, BZ2Decompressor, 42) 372 373 def testDecompress(self): 374 # "Test BZ2Decompressor.decompress()" 375 bz2d = BZ2Decompressor() 376 self.assertRaises(TypeError, bz2d.decompress) 377 text = bz2d.decompress(self.DATA) 378 self.assertEqual(text, self.TEXT) 379 380 def testDecompressChunks10(self): 381 # "Test BZ2Decompressor.decompress() with chunks of 10 bytes" 382 bz2d = BZ2Decompressor() 383 text = '' 384 n = 0 385 while 1: 386 str = self.DATA[n*10:(n+1)*10] 387 if not str: 388 break 389 text += bz2d.decompress(str) 390 n += 1 391 self.assertEqual(text, self.TEXT) 392 393 def testDecompressUnusedData(self): 394 # "Test BZ2Decompressor.decompress() with unused data" 395 bz2d = BZ2Decompressor() 396 unused_data = "this is unused data" 397 text = bz2d.decompress(self.DATA+unused_data) 398 self.assertEqual(text, self.TEXT) 399 self.assertEqual(bz2d.unused_data, unused_data) 400 401 def testEOFError(self): 402 # "Calling BZ2Decompressor.decompress() after EOS must raise EOFError" 403 bz2d = BZ2Decompressor() 404 text = bz2d.decompress(self.DATA) 405 self.assertRaises(EOFError, bz2d.decompress, "anything") 406 self.assertRaises(EOFError, bz2d.decompress, "") 407 408 @bigmemtest(_4G, memuse=1.25) 409 def testBigmem(self, size): 410 # Issue #14398: decompression fails when output data is >=2GB. 411 if size < _4G: 412 self.skipTest("Test needs 5GB of memory to run.") 413 compressed = bz2.compress("a" * _4G) 414 text = bz2.BZ2Decompressor().decompress(compressed) 415 self.assertEqual(len(text), _4G) 416 self.assertEqual(text.strip("a"), "") 417 418 419class FuncTest(BaseTest): 420 "Test module functions" 421 422 def testCompress(self): 423 # "Test compress() function" 424 data = bz2.compress(self.TEXT) 425 self.assertEqual(self.decompress(data), self.TEXT) 426 427 def testCompressEmptyString(self): 428 # "Test compress() of empty string" 429 text = bz2.compress('') 430 self.assertEqual(text, self.EMPTY_DATA) 431 432 def testDecompress(self): 433 # "Test decompress() function" 434 text = bz2.decompress(self.DATA) 435 self.assertEqual(text, self.TEXT) 436 437 def testDecompressEmpty(self): 438 # "Test decompress() function with empty string" 439 text = bz2.decompress("") 440 self.assertEqual(text, "") 441 442 def testDecompressToEmptyString(self): 443 # "Test decompress() of minimal bz2 data to empty string" 444 text = bz2.decompress(self.EMPTY_DATA) 445 self.assertEqual(text, '') 446 447 def testDecompressIncomplete(self): 448 # "Test decompress() function with incomplete data" 449 self.assertRaises(ValueError, bz2.decompress, self.DATA[:-10]) 450 451 @bigmemtest(_4G, memuse=1.25) 452 def testCompressBigmem(self, size): 453 text = "a" * size 454 data = bz2.compress(text) 455 del text 456 text = self.decompress(data) 457 self.assertEqual(len(text), size) 458 self.assertEqual(text.strip("a"), "") 459 460 @bigmemtest(_4G, memuse=1.25) 461 def testDecompressBigmem(self, size): 462 # Issue #14398: decompression fails when output data is >=2GB. 463 if size < _4G: 464 self.skipTest("Test needs 5GB of memory to run.") 465 compressed = bz2.compress("a" * _4G) 466 text = bz2.decompress(compressed) 467 self.assertEqual(len(text), _4G) 468 self.assertEqual(text.strip("a"), "") 469 470def test_main(): 471 test_support.run_unittest( 472 BZ2FileTest, 473 BZ2CompressorTest, 474 BZ2DecompressorTest, 475 FuncTest 476 ) 477 test_support.reap_children() 478 479if __name__ == '__main__': 480 test_main() 481 482# vim:ts=4:sw=4 483