2005-06-09 11:12:36 -03:00
|
|
|
#! /usr/bin/env python
|
|
|
|
"""Test script for the gzip module.
|
|
|
|
"""
|
|
|
|
|
|
|
|
import unittest
|
|
|
|
from test import test_support
|
2008-02-23 13:40:11 -04:00
|
|
|
import os
|
2010-01-03 18:29:56 -04:00
|
|
|
import io
|
2009-01-04 17:29:23 -04:00
|
|
|
import struct
|
2009-09-12 11:43:43 -03:00
|
|
|
gzip = test_support.import_module('gzip')
|
1999-03-25 17:50:27 -04:00
|
|
|
|
|
|
|
data1 = """ int length=DEFAULTALLOC, err = Z_OK;
|
|
|
|
PyObject *RetVal;
|
|
|
|
int flushmode = Z_FINISH;
|
|
|
|
unsigned long start_total_out;
|
|
|
|
|
|
|
|
"""
|
|
|
|
|
|
|
|
data2 = """/* zlibmodule.c -- gzip-compatible data compression */
|
2004-07-29 00:55:56 -03:00
|
|
|
/* See http://www.gzip.org/zlib/
|
1999-03-25 17:50:27 -04:00
|
|
|
/* See http://www.winimage.com/zLibDll for Windows */
|
|
|
|
"""
|
|
|
|
|
2005-06-09 11:12:36 -03:00
|
|
|
|
|
|
|
class TestGzip(unittest.TestCase):
|
|
|
|
filename = test_support.TESTFN
|
|
|
|
|
2008-05-25 05:07:37 -03:00
|
|
|
def setUp(self):
|
2008-04-10 02:46:39 -03:00
|
|
|
test_support.unlink(self.filename)
|
2005-06-09 11:12:36 -03:00
|
|
|
|
2008-05-25 05:07:37 -03:00
|
|
|
def tearDown(self):
|
2008-04-10 02:46:39 -03:00
|
|
|
test_support.unlink(self.filename)
|
2005-06-09 11:12:36 -03:00
|
|
|
|
|
|
|
|
2008-05-25 05:07:37 -03:00
|
|
|
def test_write(self):
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.GzipFile(self.filename, 'wb') as f:
|
|
|
|
f.write(data1 * 50)
|
2005-06-09 11:12:36 -03:00
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
# Try flush and fileno.
|
|
|
|
f.flush()
|
|
|
|
f.fileno()
|
|
|
|
if hasattr(os, 'fsync'):
|
|
|
|
os.fsync(f.fileno())
|
|
|
|
f.close()
|
2005-06-09 11:12:36 -03:00
|
|
|
|
2008-05-25 05:07:37 -03:00
|
|
|
# Test multiple close() calls.
|
|
|
|
f.close()
|
|
|
|
|
2005-06-09 11:12:36 -03:00
|
|
|
def test_read(self):
|
|
|
|
self.test_write()
|
|
|
|
# Try reading.
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.GzipFile(self.filename, 'r') as f:
|
|
|
|
d = f.read()
|
2005-06-09 11:12:36 -03:00
|
|
|
self.assertEqual(d, data1*50)
|
|
|
|
|
2010-10-06 18:26:52 -03:00
|
|
|
def test_io_on_closed_object(self):
|
|
|
|
# Test that I/O operations on closed GzipFile objects raise a
|
|
|
|
# ValueError, just like the corresponding functions on file objects.
|
|
|
|
|
|
|
|
# Write to a file, open it for reading, then close it.
|
|
|
|
self.test_write()
|
|
|
|
f = gzip.GzipFile(self.filename, 'r')
|
|
|
|
f.close()
|
|
|
|
with self.assertRaises(ValueError):
|
|
|
|
f.read(1)
|
|
|
|
with self.assertRaises(ValueError):
|
|
|
|
f.seek(0)
|
|
|
|
with self.assertRaises(ValueError):
|
|
|
|
f.tell()
|
|
|
|
# Open the file for writing, then close it.
|
|
|
|
f = gzip.GzipFile(self.filename, 'w')
|
|
|
|
f.close()
|
|
|
|
with self.assertRaises(ValueError):
|
2010-10-13 20:51:19 -03:00
|
|
|
f.write('')
|
2010-10-06 18:26:52 -03:00
|
|
|
with self.assertRaises(ValueError):
|
|
|
|
f.flush()
|
|
|
|
|
2005-06-09 11:12:36 -03:00
|
|
|
def test_append(self):
|
|
|
|
self.test_write()
|
|
|
|
# Append to the previous file
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.GzipFile(self.filename, 'ab') as f:
|
|
|
|
f.write(data2 * 15)
|
2005-06-09 11:12:36 -03:00
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.GzipFile(self.filename, 'rb') as f:
|
|
|
|
d = f.read()
|
2005-06-09 11:12:36 -03:00
|
|
|
self.assertEqual(d, (data1*50) + (data2*15))
|
|
|
|
|
2005-06-09 11:19:32 -03:00
|
|
|
def test_many_append(self):
|
|
|
|
# Bug #1074261 was triggered when reading a file that contained
|
|
|
|
# many, many members. Create such a file and verify that reading it
|
|
|
|
# works.
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.open(self.filename, 'wb', 9) as f:
|
2005-06-09 11:19:32 -03:00
|
|
|
f.write('a')
|
2010-10-13 20:51:19 -03:00
|
|
|
for i in range(0, 200):
|
|
|
|
with gzip.open(self.filename, "ab", 9) as f: # append
|
|
|
|
f.write('a')
|
2005-06-09 11:19:32 -03:00
|
|
|
|
|
|
|
# Try reading the file
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.open(self.filename, "rb") as zgfile:
|
|
|
|
contents = ""
|
|
|
|
while 1:
|
|
|
|
ztxt = zgfile.read(8192)
|
|
|
|
contents += ztxt
|
|
|
|
if not ztxt: break
|
2010-11-21 09:34:58 -04:00
|
|
|
self.assertEqual(contents, 'a'*201)
|
2005-06-09 11:19:32 -03:00
|
|
|
|
2010-01-03 18:29:56 -04:00
|
|
|
def test_buffered_reader(self):
|
|
|
|
# Issue #7471: a GzipFile can be wrapped in a BufferedReader for
|
|
|
|
# performance.
|
|
|
|
self.test_write()
|
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.GzipFile(self.filename, 'rb') as f:
|
|
|
|
with io.BufferedReader(f) as r:
|
|
|
|
lines = [line for line in r]
|
2010-01-03 18:29:56 -04:00
|
|
|
|
|
|
|
self.assertEqual(lines, 50 * data1.splitlines(True))
|
2005-06-09 11:19:32 -03:00
|
|
|
|
2005-06-09 11:12:36 -03:00
|
|
|
def test_readline(self):
|
|
|
|
self.test_write()
|
|
|
|
# Try .readline() with varying line lengths
|
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.GzipFile(self.filename, 'rb') as f:
|
|
|
|
line_length = 0
|
|
|
|
while 1:
|
|
|
|
L = f.readline(line_length)
|
|
|
|
if not L and line_length != 0: break
|
|
|
|
self.assertTrue(len(L) <= line_length)
|
|
|
|
line_length = (line_length + 1) % 50
|
2005-06-09 11:12:36 -03:00
|
|
|
|
|
|
|
def test_readlines(self):
|
|
|
|
self.test_write()
|
|
|
|
# Try .readlines()
|
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.GzipFile(self.filename, 'rb') as f:
|
|
|
|
L = f.readlines()
|
2005-06-09 11:12:36 -03:00
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.GzipFile(self.filename, 'rb') as f:
|
|
|
|
while 1:
|
|
|
|
L = f.readlines(150)
|
|
|
|
if L == []: break
|
2005-06-09 11:12:36 -03:00
|
|
|
|
|
|
|
def test_seek_read(self):
|
|
|
|
self.test_write()
|
|
|
|
# Try seek, read test
|
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.GzipFile(self.filename) as f:
|
|
|
|
while 1:
|
|
|
|
oldpos = f.tell()
|
|
|
|
line1 = f.readline()
|
|
|
|
if not line1: break
|
|
|
|
newpos = f.tell()
|
|
|
|
f.seek(oldpos) # negative seek
|
|
|
|
if len(line1)>10:
|
|
|
|
amount = 10
|
|
|
|
else:
|
|
|
|
amount = len(line1)
|
|
|
|
line2 = f.read(amount)
|
|
|
|
self.assertEqual(line1[:amount], line2)
|
|
|
|
f.seek(newpos) # positive seek
|
2005-06-09 11:12:36 -03:00
|
|
|
|
2006-11-12 06:41:39 -04:00
|
|
|
def test_seek_whence(self):
|
|
|
|
self.test_write()
|
|
|
|
# Try seek(whence=1), read test
|
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.GzipFile(self.filename) as f:
|
|
|
|
f.read(10)
|
|
|
|
f.seek(10, whence=1)
|
|
|
|
y = f.read(10)
|
2010-11-21 09:34:58 -04:00
|
|
|
self.assertEqual(y, data1[20:30])
|
2007-01-29 23:03:46 -04:00
|
|
|
|
2005-06-09 11:12:36 -03:00
|
|
|
def test_seek_write(self):
|
|
|
|
# Try seek, write test
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.GzipFile(self.filename, 'w') as f:
|
|
|
|
for pos in range(0, 256, 16):
|
|
|
|
f.seek(pos)
|
|
|
|
f.write('GZ\n')
|
2005-06-09 11:12:36 -03:00
|
|
|
|
|
|
|
def test_mode(self):
|
|
|
|
self.test_write()
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.GzipFile(self.filename, 'r') as f:
|
|
|
|
self.assertEqual(f.myfileobj.mode, 'rb')
|
2005-06-09 11:12:36 -03:00
|
|
|
|
2007-02-13 12:09:24 -04:00
|
|
|
def test_1647484(self):
|
|
|
|
for mode in ('wb', 'rb'):
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.GzipFile(self.filename, mode) as f:
|
|
|
|
self.assertTrue(hasattr(f, "name"))
|
|
|
|
self.assertEqual(f.name, self.filename)
|
2007-02-13 12:09:24 -04:00
|
|
|
|
2009-01-04 17:29:23 -04:00
|
|
|
def test_mtime(self):
|
|
|
|
mtime = 123456789
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.GzipFile(self.filename, 'w', mtime = mtime) as fWrite:
|
|
|
|
fWrite.write(data1)
|
|
|
|
with gzip.GzipFile(self.filename) as fRead:
|
|
|
|
dataRead = fRead.read()
|
|
|
|
self.assertEqual(dataRead, data1)
|
|
|
|
self.assertTrue(hasattr(fRead, 'mtime'))
|
|
|
|
self.assertEqual(fRead.mtime, mtime)
|
2009-01-04 17:29:23 -04:00
|
|
|
|
|
|
|
def test_metadata(self):
|
|
|
|
mtime = 123456789
|
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
with gzip.GzipFile(self.filename, 'w', mtime = mtime) as fWrite:
|
|
|
|
fWrite.write(data1)
|
2009-01-04 17:29:23 -04:00
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
with open(self.filename, 'rb') as fRead:
|
|
|
|
# see RFC 1952: http://www.faqs.org/rfcs/rfc1952.html
|
2009-01-04 17:29:23 -04:00
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
idBytes = fRead.read(2)
|
|
|
|
self.assertEqual(idBytes, '\x1f\x8b') # gzip ID
|
2009-01-04 17:29:23 -04:00
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
cmByte = fRead.read(1)
|
|
|
|
self.assertEqual(cmByte, '\x08') # deflate
|
2009-01-04 17:29:23 -04:00
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
flagsByte = fRead.read(1)
|
|
|
|
self.assertEqual(flagsByte, '\x08') # only the FNAME flag is set
|
2009-01-04 17:29:23 -04:00
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
mtimeBytes = fRead.read(4)
|
|
|
|
self.assertEqual(mtimeBytes, struct.pack('<i', mtime)) # little-endian
|
2009-01-04 17:29:23 -04:00
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
xflByte = fRead.read(1)
|
|
|
|
self.assertEqual(xflByte, '\x02') # maximum compression
|
2009-01-04 17:29:23 -04:00
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
osByte = fRead.read(1)
|
|
|
|
self.assertEqual(osByte, '\xff') # OS "unknown" (OS-independent)
|
2009-01-04 17:29:23 -04:00
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
# Since the FNAME flag is set, the zero-terminated filename follows.
|
|
|
|
# RFC 1952 specifies that this is the name of the input file, if any.
|
|
|
|
# However, the gzip module defaults to storing the name of the output
|
|
|
|
# file in this field.
|
|
|
|
expected = self.filename.encode('Latin-1') + '\x00'
|
|
|
|
nameBytes = fRead.read(len(expected))
|
|
|
|
self.assertEqual(nameBytes, expected)
|
2009-01-04 17:29:23 -04:00
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
# Since no other flags were set, the header ends here.
|
|
|
|
# Rather than process the compressed data, let's seek to the trailer.
|
|
|
|
fRead.seek(os.stat(self.filename).st_size - 8)
|
2009-01-04 17:29:23 -04:00
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
crc32Bytes = fRead.read(4) # CRC32 of uncompressed data [data1]
|
|
|
|
self.assertEqual(crc32Bytes, '\xaf\xd7d\x83')
|
2009-01-04 17:29:23 -04:00
|
|
|
|
2010-10-13 20:51:19 -03:00
|
|
|
isizeBytes = fRead.read(4)
|
|
|
|
self.assertEqual(isizeBytes, struct.pack('<i', len(data1)))
|
2009-01-04 17:29:23 -04:00
|
|
|
|
2009-01-10 12:13:45 -04:00
|
|
|
def test_with_open(self):
|
|
|
|
# GzipFile supports the context management protocol
|
|
|
|
with gzip.GzipFile(self.filename, "wb") as f:
|
|
|
|
f.write(b"xxx")
|
|
|
|
f = gzip.GzipFile(self.filename, "rb")
|
|
|
|
f.close()
|
|
|
|
try:
|
|
|
|
with f:
|
|
|
|
pass
|
|
|
|
except ValueError:
|
|
|
|
pass
|
|
|
|
else:
|
|
|
|
self.fail("__enter__ on a closed file didn't raise an exception")
|
|
|
|
try:
|
|
|
|
with gzip.GzipFile(self.filename, "wb") as f:
|
2010-02-03 01:37:26 -04:00
|
|
|
1 // 0
|
2009-01-10 12:13:45 -04:00
|
|
|
except ZeroDivisionError:
|
|
|
|
pass
|
|
|
|
else:
|
2010-02-03 01:37:26 -04:00
|
|
|
self.fail("1 // 0 didn't raise an exception")
|
2009-01-10 12:13:45 -04:00
|
|
|
|
2010-01-13 10:32:10 -04:00
|
|
|
def test_zero_padded_file(self):
|
|
|
|
with gzip.GzipFile(self.filename, "wb") as f:
|
|
|
|
f.write(data1 * 50)
|
|
|
|
|
|
|
|
# Pad the file with zeroes
|
|
|
|
with open(self.filename, "ab") as f:
|
|
|
|
f.write("\x00" * 50)
|
|
|
|
|
|
|
|
with gzip.GzipFile(self.filename, "rb") as f:
|
|
|
|
d = f.read()
|
|
|
|
self.assertEqual(d, data1 * 50, "Incorrect data in file")
|
|
|
|
|
2005-06-09 11:12:36 -03:00
|
|
|
def test_main(verbose=None):
|
|
|
|
test_support.run_unittest(TestGzip)
|
|
|
|
|
|
|
|
if __name__ == "__main__":
|
|
|
|
test_main(verbose=True)
|