From 752e3d1b783fc1c12a28e05a93aa73ac7c6b751c Mon Sep 17 00:00:00 2001 From: tailhook Date: Fri, 15 Apr 2011 17:36:17 +0300 Subject: Implemented encoding for strings * Packer by default uses `utf-8` encoding by default * Unpacker uses `None` by default, so no decoding is done * Both pack and unpack has `encoding` and `unicode_errors` arguments, if `encoding` is `None` no encoding/decoding is done, otherwise it is python codec. `unicode_errors` is supplied as `errors` parameter to codec --- test/test_pack.py | 57 +++++++++++++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 53 insertions(+), 4 deletions(-) (limited to 'test') diff --git a/test/test_pack.py b/test/test_pack.py index 5dec068..2aef588 100644 --- a/test/test_pack.py +++ b/test/test_pack.py @@ -15,14 +15,63 @@ def testPack(): 0, 1, 127, 128, 255, 256, 65535, 65536, -1, -32, -33, -128, -129, -32768, -32769, 1.0, - "", "a", "a"*31, "a"*32, + b"", b"a", b"a"*31, b"a"*32, None, True, False, - (), ((),), ((), None,), - {None: 0}, - (1<<23), + (), ((),), ((), None,), + {None: 0}, + (1<<23), ] for td in test_data: check(td) +def testPackUnicode(): + test_data = [ + u"", u"abcd", (u"defgh",), u"Русский текст", + ] + for td in test_data: + re = unpacks(packs(td, encoding='utf-8'), encoding='utf-8') + assert_equal(re, td) + +def testPackUTF32(): + test_data = [ + u"", u"abcd", (u"defgh",), u"Русский текст", + ] + for td in test_data: + print(packs(td, encoding='utf-32')) + re = unpacks(packs(td, encoding='utf-32'), encoding='utf-32') + assert_equal(re, td) + +def testPackBytes(): + test_data = [ + b"", b"abcd", (b"defgh",), + ] + for td in test_data: + check(td) + +def testIgnoreUnicodeErrors(): + re = unpacks(packs(b'abc\xeddef'), + encoding='utf-8', unicode_errors='ignore') + assert_equal(re, "abcdef") + +@raises(UnicodeDecodeError) +def testStrictUnicodeUnpack(): + unpacks(packs(b'abc\xeddef'), encoding='utf-8') + +@raises(UnicodeEncodeError) +def testStrictUnicodePack(): + packs(u"abc\xeddef", encoding='ascii', unicode_errors='strict') + +def testIgnoreErrorsPack(): + re = unpacks(packs(u"abcФФФdef", encoding='ascii', unicode_errors='ignore'), encoding='utf-8') + assert_equal(re, u"abcdef") + +@raises(TypeError) +def testNoEncoding(): + packs(u"abc", encoding=None) + +def testDecodeBinary(): + re = unpacks(packs(u"abc"), encoding=None) + assert_equal(re, b"abc") + if __name__ == '__main__': main() -- cgit v1.2.1