[Python-3000-checkins] r61558 - in python/branches/py3k/Lib: sre_compile.py sre_parse.py test/test_re.py
thomas.wouters
python-3000-checkins at python.org
Tue Mar 18 21:19:54 CET 2008
Author: thomas.wouters
Date: Tue Mar 18 21:19:54 2008
New Revision: 61558
Modified:
python/branches/py3k/Lib/sre_compile.py
python/branches/py3k/Lib/sre_parse.py
python/branches/py3k/Lib/test/test_re.py
Log:
Fix 're' to work on bytes. It could do with a few more tests, though.
Modified: python/branches/py3k/Lib/sre_compile.py
==============================================================================
--- python/branches/py3k/Lib/sre_compile.py (original)
+++ python/branches/py3k/Lib/sre_compile.py Tue Mar 18 21:19:54 2008
@@ -472,7 +472,7 @@
code[skip] = len(code) - skip
def isstring(obj):
- return isinstance(obj, str)
+ return isinstance(obj, (str, bytes))
def _code(p, flags):
Modified: python/branches/py3k/Lib/sre_parse.py
==============================================================================
--- python/branches/py3k/Lib/sre_parse.py (original)
+++ python/branches/py3k/Lib/sre_parse.py Tue Mar 18 21:19:54 2008
@@ -192,8 +192,8 @@
char = self.string[self.index:self.index+1]
# Special case for the str8, since indexing returns a integer
# XXX This is only needed for test_bug_926075 in test_re.py
- if isinstance(self.string, bytes):
- char = chr(char)
+ if char and isinstance(char, bytes):
+ char = chr(char[0])
if char == "\\":
try:
c = self.string[self.index + 1]
Modified: python/branches/py3k/Lib/test/test_re.py
==============================================================================
--- python/branches/py3k/Lib/test/test_re.py (original)
+++ python/branches/py3k/Lib/test/test_re.py Tue Mar 18 21:19:54 2008
@@ -83,33 +83,22 @@
self.assertEqual(re.sub('\r\n', '\n', 'abc\r\ndef\r\n'),
'abc\ndef\n')
-# This test makes no sense until re supports bytes, and should then probably
-# test for the *in*ability to mix bytes and str this way :)
-#
-# def test_bug_1140(self):
-# # re.sub(x, y, b'') should return b'', not '', and
-# # re.sub(x, y, '') should return '', not b''.
-# # Also:
-# # re.sub(x, y, str(x)) should return str(y), and
-# # re.sub(x, y, bytes(x)) should return
-# # str(y) if isinstance(y, str) else unicode(y).
-# for x in 'x', u'x':
-# for y in 'y', u'y':
-# z = re.sub(x, y, u'')
-# self.assertEqual(z, u'')
-# self.assertEqual(type(z), unicode)
-# #
-# z = re.sub(x, y, '')
-# self.assertEqual(z, '')
-# self.assertEqual(type(z), str)
-# #
-# z = re.sub(x, y, unicode(x))
-# self.assertEqual(z, y)
-# self.assertEqual(type(z), unicode)
-# #
-# z = re.sub(x, y, str(x))
-# self.assertEqual(z, y)
-# self.assertEqual(type(z), type(y))
+ def test_bug_1140(self):
+ # re.sub(x, y, b'') should return b'', not '', and
+ # re.sub(x, y, '') should return '', not b''.
+ # Also:
+ # re.sub(x, y, str(x)) should return str(y), and
+ # re.sub(x, y, bytes(x)) should return
+ # str(y) if isinstance(y, str) else unicode(y).
+ for x in 'x', b'x':
+ for y in 'y', b'y':
+ z = re.sub(x, y, b'')
+ self.assertEqual(z, b'')
+ self.assertEqual(type(z), bytes)
+ #
+ z = re.sub(x, y, '')
+ self.assertEqual(z, '')
+ self.assertEqual(type(z), str)
def test_bug_1661(self):
# Verify that flags do not get silently ignored with compiled patterns
@@ -599,10 +588,9 @@
self.assertEqual([item.group(0) for item in iter],
[":", "::", ":::"])
- # XXX This needs to be restored for str vs. bytes.
-## def test_bug_926075(self):
-## self.assert_(re.compile('bug_926075') is not
-## re.compile(str8('bug_926075')))
+ def test_bug_926075(self):
+ self.assert_(re.compile('bug_926075') is not
+ re.compile(b'bug_926075'))
def test_bug_931848(self):
pattern = eval('"[\u002E\u3002\uFF0E\uFF61]"')
More information about the Python-3000-checkins
mailing list