Benjamin Peterson | 90f5ba5 | 2010-03-11 22:53:45 +0000 | [diff] [blame] | 1 | #!/usr/bin/env python3 |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 2 | # |
| 3 | # test_multibytecodec_support.py |
| 4 | # Common Unittest Routines for CJK codecs |
| 5 | # |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 6 | |
Victor Stinner | 033a543 | 2011-05-16 16:43:38 +0200 | [diff] [blame] | 7 | import codecs |
| 8 | import os |
| 9 | import re |
| 10 | import sys |
| 11 | import unittest |
Antoine Pitrou | 1a305fd | 2009-11-26 12:38:23 +0000 | [diff] [blame] | 12 | from http.client import HTTPException |
Benjamin Peterson | ee8712c | 2008-05-20 21:35:26 +0000 | [diff] [blame] | 13 | from test import support |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 14 | from io import BytesIO |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 15 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 16 | class TestBase: |
| 17 | encoding = '' # codec name |
| 18 | codec = None # codec tuple (with 4 elements) |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 19 | tstring = None # must set. 2 strings to test StreamReader |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 20 | |
| 21 | codectests = None # must set. codec test tuple |
| 22 | roundtriptest = 1 # set if roundtrip is possible with unicode |
| 23 | has_iso10646 = 0 # set if this encoding contains whole iso10646 map |
| 24 | xmlcharnametest = None # string to test xmlcharrefreplace |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 25 | unmappedunicode = '\udeee' # a unicode codepoint that is not mapped. |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 26 | |
| 27 | def setUp(self): |
| 28 | if self.codec is None: |
| 29 | self.codec = codecs.lookup(self.encoding) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 30 | self.encode = self.codec.encode |
| 31 | self.decode = self.codec.decode |
| 32 | self.reader = self.codec.streamreader |
| 33 | self.writer = self.codec.streamwriter |
| 34 | self.incrementalencoder = self.codec.incrementalencoder |
| 35 | self.incrementaldecoder = self.codec.incrementaldecoder |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 36 | |
| 37 | def test_chunkcoding(self): |
Guido van Rossum | 49f2174 | 2007-10-10 00:12:46 +0000 | [diff] [blame] | 38 | tstring_lines = [] |
| 39 | for b in self.tstring: |
| 40 | lines = b.split(b"\n") |
| 41 | last = lines.pop() |
| 42 | assert last == b"" |
| 43 | lines = [line + b"\n" for line in lines] |
| 44 | tstring_lines.append(lines) |
| 45 | for native, utf8 in zip(*tstring_lines): |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 46 | u = self.decode(native)[0] |
| 47 | self.assertEqual(u, utf8.decode('utf-8')) |
| 48 | if self.roundtriptest: |
| 49 | self.assertEqual(native, self.encode(u)[0]) |
| 50 | |
| 51 | def test_errorhandle(self): |
| 52 | for source, scheme, expected in self.codectests: |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 53 | if isinstance(source, bytes): |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 54 | func = self.decode |
| 55 | else: |
| 56 | func = self.encode |
| 57 | if expected: |
| 58 | result = func(source, scheme)[0] |
Guido van Rossum | 98297ee | 2007-11-06 21:34:58 +0000 | [diff] [blame] | 59 | if func is self.decode: |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 60 | self.assertTrue(type(result) is str, type(result)) |
Victor Stinner | f5048a4 | 2011-06-03 23:44:39 +0200 | [diff] [blame] | 61 | self.assertEqual(result, expected, |
| 62 | '%a.decode(%r, %r)=%a != %a' |
| 63 | % (source, self.encoding, scheme, result, |
| 64 | expected)) |
Guido van Rossum | 98297ee | 2007-11-06 21:34:58 +0000 | [diff] [blame] | 65 | else: |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 66 | self.assertTrue(type(result) is bytes, type(result)) |
Victor Stinner | f5048a4 | 2011-06-03 23:44:39 +0200 | [diff] [blame] | 67 | self.assertEqual(result, expected, |
| 68 | '%a.encode(%r, %r)=%a != %a' |
| 69 | % (source, self.encoding, scheme, result, |
| 70 | expected)) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 71 | else: |
| 72 | self.assertRaises(UnicodeError, func, source, scheme) |
| 73 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 74 | def test_xmlcharrefreplace(self): |
| 75 | if self.has_iso10646: |
| 76 | return |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 77 | |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 78 | s = "\u0b13\u0b23\u0b60 nd eggs" |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 79 | self.assertEqual( |
| 80 | self.encode(s, "xmlcharrefreplace")[0], |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 81 | b"ଓଣୠ nd eggs" |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 82 | ) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 83 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 84 | def test_customreplace_encode(self): |
| 85 | if self.has_iso10646: |
| 86 | return |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 87 | |
Fred Drake | 3c50ea4 | 2008-05-17 22:02:32 +0000 | [diff] [blame] | 88 | from html.entities import codepoint2name |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 89 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 90 | def xmlcharnamereplace(exc): |
| 91 | if not isinstance(exc, UnicodeEncodeError): |
| 92 | raise TypeError("don't know how to handle %r" % exc) |
| 93 | l = [] |
| 94 | for c in exc.object[exc.start:exc.end]: |
| 95 | if ord(c) in codepoint2name: |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 96 | l.append("&%s;" % codepoint2name[ord(c)]) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 97 | else: |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 98 | l.append("&#%d;" % ord(c)) |
| 99 | return ("".join(l), exc.end) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 100 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 101 | codecs.register_error("test.xmlcharnamereplace", xmlcharnamereplace) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 102 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 103 | if self.xmlcharnametest: |
| 104 | sin, sout = self.xmlcharnametest |
| 105 | else: |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 106 | sin = "\xab\u211c\xbb = \u2329\u1234\u232a" |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 107 | sout = b"«ℜ» = ⟨ሴ⟩" |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 108 | self.assertEqual(self.encode(sin, |
| 109 | "test.xmlcharnamereplace")[0], sout) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 110 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 111 | def test_callback_wrong_objects(self): |
| 112 | def myreplace(exc): |
| 113 | return (ret, exc.end) |
| 114 | codecs.register_error("test.cjktest", myreplace) |
| 115 | |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 116 | for ret in ([1, 2, 3], [], None, object(), b'string', b''): |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 117 | self.assertRaises(TypeError, self.encode, self.unmappedunicode, |
| 118 | 'test.cjktest') |
| 119 | |
| 120 | def test_callback_long_index(self): |
| 121 | def myreplace(exc): |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 122 | return ('x', int(exc.end)) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 123 | codecs.register_error("test.cjktest", myreplace) |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 124 | self.assertEqual(self.encode('abcd' + self.unmappedunicode + 'efgh', |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 125 | 'test.cjktest'), (b'abcdxefgh', 9)) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 126 | |
| 127 | def myreplace(exc): |
Christian Heimes | a37d4c6 | 2007-12-04 23:02:19 +0000 | [diff] [blame] | 128 | return ('x', sys.maxsize + 1) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 129 | codecs.register_error("test.cjktest", myreplace) |
| 130 | self.assertRaises(IndexError, self.encode, self.unmappedunicode, |
| 131 | 'test.cjktest') |
| 132 | |
| 133 | def test_callback_None_index(self): |
| 134 | def myreplace(exc): |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 135 | return ('x', None) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 136 | codecs.register_error("test.cjktest", myreplace) |
| 137 | self.assertRaises(TypeError, self.encode, self.unmappedunicode, |
| 138 | 'test.cjktest') |
| 139 | |
| 140 | def test_callback_backward_index(self): |
| 141 | def myreplace(exc): |
| 142 | if myreplace.limit > 0: |
| 143 | myreplace.limit -= 1 |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 144 | return ('REPLACED', 0) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 145 | else: |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 146 | return ('TERMINAL', exc.end) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 147 | myreplace.limit = 3 |
| 148 | codecs.register_error("test.cjktest", myreplace) |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 149 | self.assertEqual(self.encode('abcd' + self.unmappedunicode + 'efgh', |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 150 | 'test.cjktest'), |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 151 | (b'abcdREPLACEDabcdREPLACEDabcdREPLACEDabcdTERMINALefgh', 9)) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 152 | |
| 153 | def test_callback_forward_index(self): |
| 154 | def myreplace(exc): |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 155 | return ('REPLACED', exc.end + 2) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 156 | codecs.register_error("test.cjktest", myreplace) |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 157 | self.assertEqual(self.encode('abcd' + self.unmappedunicode + 'efgh', |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 158 | 'test.cjktest'), (b'abcdREPLACEDgh', 9)) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 159 | |
| 160 | def test_callback_index_outofbound(self): |
| 161 | def myreplace(exc): |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 162 | return ('TERM', 100) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 163 | codecs.register_error("test.cjktest", myreplace) |
| 164 | self.assertRaises(IndexError, self.encode, self.unmappedunicode, |
| 165 | 'test.cjktest') |
| 166 | |
| 167 | def test_incrementalencoder(self): |
| 168 | UTF8Reader = codecs.getreader('utf-8') |
Guido van Rossum | 805365e | 2007-05-07 22:24:25 +0000 | [diff] [blame] | 169 | for sizehint in [None] + list(range(1, 33)) + \ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 170 | [64, 128, 256, 512, 1024]: |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 171 | istream = UTF8Reader(BytesIO(self.tstring[1])) |
| 172 | ostream = BytesIO() |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 173 | encoder = self.incrementalencoder() |
| 174 | while 1: |
| 175 | if sizehint is not None: |
| 176 | data = istream.read(sizehint) |
| 177 | else: |
| 178 | data = istream.read() |
| 179 | |
| 180 | if not data: |
| 181 | break |
| 182 | e = encoder.encode(data) |
| 183 | ostream.write(e) |
| 184 | |
| 185 | self.assertEqual(ostream.getvalue(), self.tstring[0]) |
| 186 | |
| 187 | def test_incrementaldecoder(self): |
| 188 | UTF8Writer = codecs.getwriter('utf-8') |
Guido van Rossum | 805365e | 2007-05-07 22:24:25 +0000 | [diff] [blame] | 189 | for sizehint in [None, -1] + list(range(1, 33)) + \ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 190 | [64, 128, 256, 512, 1024]: |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 191 | istream = BytesIO(self.tstring[0]) |
| 192 | ostream = UTF8Writer(BytesIO()) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 193 | decoder = self.incrementaldecoder() |
| 194 | while 1: |
| 195 | data = istream.read(sizehint) |
| 196 | if not data: |
| 197 | break |
| 198 | else: |
| 199 | u = decoder.decode(data) |
| 200 | ostream.write(u) |
| 201 | |
| 202 | self.assertEqual(ostream.getvalue(), self.tstring[1]) |
| 203 | |
| 204 | def test_incrementalencoder_error_callback(self): |
| 205 | inv = self.unmappedunicode |
| 206 | |
| 207 | e = self.incrementalencoder() |
| 208 | self.assertRaises(UnicodeEncodeError, e.encode, inv, True) |
| 209 | |
| 210 | e.errors = 'ignore' |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 211 | self.assertEqual(e.encode(inv, True), b'') |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 212 | |
| 213 | e.reset() |
| 214 | def tempreplace(exc): |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 215 | return ('called', exc.end) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 216 | codecs.register_error('test.incremental_error_callback', tempreplace) |
| 217 | e.errors = 'test.incremental_error_callback' |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 218 | self.assertEqual(e.encode(inv, True), b'called') |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 219 | |
| 220 | # again |
| 221 | e.errors = 'ignore' |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 222 | self.assertEqual(e.encode(inv, True), b'') |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 223 | |
| 224 | def test_streamreader(self): |
| 225 | UTF8Writer = codecs.getwriter('utf-8') |
| 226 | for name in ["read", "readline", "readlines"]: |
Guido van Rossum | 805365e | 2007-05-07 22:24:25 +0000 | [diff] [blame] | 227 | for sizehint in [None, -1] + list(range(1, 33)) + \ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 228 | [64, 128, 256, 512, 1024]: |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 229 | istream = self.reader(BytesIO(self.tstring[0])) |
| 230 | ostream = UTF8Writer(BytesIO()) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 231 | func = getattr(istream, name) |
| 232 | while 1: |
| 233 | data = func(sizehint) |
| 234 | if not data: |
| 235 | break |
| 236 | if name == "readlines": |
| 237 | ostream.writelines(data) |
| 238 | else: |
| 239 | ostream.write(data) |
| 240 | |
| 241 | self.assertEqual(ostream.getvalue(), self.tstring[1]) |
| 242 | |
| 243 | def test_streamwriter(self): |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 244 | readfuncs = ('read', 'readline', 'readlines') |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 245 | UTF8Reader = codecs.getreader('utf-8') |
| 246 | for name in readfuncs: |
Guido van Rossum | 805365e | 2007-05-07 22:24:25 +0000 | [diff] [blame] | 247 | for sizehint in [None] + list(range(1, 33)) + \ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 248 | [64, 128, 256, 512, 1024]: |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 249 | istream = UTF8Reader(BytesIO(self.tstring[1])) |
| 250 | ostream = self.writer(BytesIO()) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 251 | func = getattr(istream, name) |
| 252 | while 1: |
| 253 | if sizehint is not None: |
| 254 | data = func(sizehint) |
| 255 | else: |
| 256 | data = func() |
| 257 | |
| 258 | if not data: |
| 259 | break |
| 260 | if name == "readlines": |
| 261 | ostream.writelines(data) |
| 262 | else: |
| 263 | ostream.write(data) |
| 264 | |
| 265 | self.assertEqual(ostream.getvalue(), self.tstring[0]) |
| 266 | |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 267 | if len('\U00012345') == 2: # ucs2 build |
Guido van Rossum | 84fc66d | 2007-05-03 17:18:26 +0000 | [diff] [blame] | 268 | _unichr = chr |
| 269 | def chr(v): |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 270 | if v >= 0x10000: |
| 271 | return _unichr(0xd800 + ((v - 0x10000) >> 10)) + \ |
| 272 | _unichr(0xdc00 + ((v - 0x10000) & 0x3ff)) |
| 273 | else: |
| 274 | return _unichr(v) |
| 275 | _ord = ord |
| 276 | def ord(c): |
| 277 | if len(c) == 2: |
| 278 | return 0x10000 + ((_ord(c[0]) - 0xd800) << 10) + \ |
| 279 | (ord(c[1]) - 0xdc00) |
| 280 | else: |
| 281 | return _ord(c) |
| 282 | |
| 283 | class TestBase_Mapping(unittest.TestCase): |
| 284 | pass_enctest = [] |
| 285 | pass_dectest = [] |
| 286 | supmaps = [] |
Victor Stinner | f5048a4 | 2011-06-03 23:44:39 +0200 | [diff] [blame] | 287 | codectests = [] |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 288 | |
| 289 | def __init__(self, *args, **kw): |
| 290 | unittest.TestCase.__init__(self, *args, **kw) |
Martin v. Löwis | 0f95ba9 | 2008-03-08 11:59:08 +0000 | [diff] [blame] | 291 | try: |
Antoine Pitrou | 92f60ed | 2010-10-14 22:11:44 +0000 | [diff] [blame] | 292 | self.open_mapping_file().close() # test it to report the error early |
Antoine Pitrou | 1a305fd | 2009-11-26 12:38:23 +0000 | [diff] [blame] | 293 | except (IOError, HTTPException): |
Antoine Pitrou | c9e9216 | 2009-11-01 15:59:11 +0000 | [diff] [blame] | 294 | self.skipTest("Could not retrieve "+self.mapfileurl) |
Hye-Shik Chang | aaa2f1d | 2005-12-10 17:44:27 +0000 | [diff] [blame] | 295 | |
| 296 | def open_mapping_file(self): |
Benjamin Peterson | ee8712c | 2008-05-20 21:35:26 +0000 | [diff] [blame] | 297 | return support.open_urlresource(self.mapfileurl) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 298 | |
| 299 | def test_mapping_file(self): |
Guido van Rossum | 806c246 | 2007-08-06 23:33:07 +0000 | [diff] [blame] | 300 | if self.mapfileurl.endswith('.xml'): |
| 301 | self._test_mapping_file_ucm() |
| 302 | else: |
| 303 | self._test_mapping_file_plain() |
| 304 | |
| 305 | def _test_mapping_file_plain(self): |
Guido van Rossum | 84fc66d | 2007-05-03 17:18:26 +0000 | [diff] [blame] | 306 | unichrs = lambda s: ''.join(map(chr, map(eval, s.split('+')))) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 307 | urt_wa = {} |
| 308 | |
Antoine Pitrou | 92f60ed | 2010-10-14 22:11:44 +0000 | [diff] [blame] | 309 | with self.open_mapping_file() as f: |
| 310 | for line in f: |
| 311 | if not line: |
| 312 | break |
| 313 | data = line.split('#')[0].strip().split() |
| 314 | if len(data) != 2: |
| 315 | continue |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 316 | |
Antoine Pitrou | 92f60ed | 2010-10-14 22:11:44 +0000 | [diff] [blame] | 317 | csetval = eval(data[0]) |
| 318 | if csetval <= 0x7F: |
| 319 | csetch = bytes([csetval & 0xff]) |
| 320 | elif csetval >= 0x1000000: |
| 321 | csetch = bytes([(csetval >> 24), ((csetval >> 16) & 0xff), |
| 322 | ((csetval >> 8) & 0xff), (csetval & 0xff)]) |
| 323 | elif csetval >= 0x10000: |
| 324 | csetch = bytes([(csetval >> 16), ((csetval >> 8) & 0xff), |
| 325 | (csetval & 0xff)]) |
| 326 | elif csetval >= 0x100: |
| 327 | csetch = bytes([(csetval >> 8), (csetval & 0xff)]) |
| 328 | else: |
| 329 | continue |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 330 | |
Antoine Pitrou | 92f60ed | 2010-10-14 22:11:44 +0000 | [diff] [blame] | 331 | unich = unichrs(data[1]) |
| 332 | if ord(unich) == 0xfffd or unich in urt_wa: |
| 333 | continue |
| 334 | urt_wa[unich] = csetch |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 335 | |
Antoine Pitrou | 92f60ed | 2010-10-14 22:11:44 +0000 | [diff] [blame] | 336 | self._testpoint(csetch, unich) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 337 | |
Guido van Rossum | 806c246 | 2007-08-06 23:33:07 +0000 | [diff] [blame] | 338 | def _test_mapping_file_ucm(self): |
Antoine Pitrou | 92f60ed | 2010-10-14 22:11:44 +0000 | [diff] [blame] | 339 | with self.open_mapping_file() as f: |
| 340 | ucmdata = f.read() |
Guido van Rossum | 806c246 | 2007-08-06 23:33:07 +0000 | [diff] [blame] | 341 | uc = re.findall('<a u="([A-F0-9]{4})" b="([0-9A-F ]+)"/>', ucmdata) |
| 342 | for uni, coded in uc: |
| 343 | unich = chr(int(uni, 16)) |
| 344 | codech = bytes(int(c, 16) for c in coded.split()) |
| 345 | self._testpoint(codech, unich) |
| 346 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 347 | def test_mapping_supplemental(self): |
| 348 | for mapping in self.supmaps: |
| 349 | self._testpoint(*mapping) |
| 350 | |
| 351 | def _testpoint(self, csetch, unich): |
| 352 | if (csetch, unich) not in self.pass_enctest: |
| 353 | self.assertEqual(unich.encode(self.encoding), csetch) |
| 354 | if (csetch, unich) not in self.pass_dectest: |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 355 | self.assertEqual(str(csetch, self.encoding), unich) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 356 | |
Victor Stinner | f5048a4 | 2011-06-03 23:44:39 +0200 | [diff] [blame] | 357 | def test_errorhandle(self): |
| 358 | for source, scheme, expected in self.codectests: |
| 359 | if isinstance(source, bytes): |
| 360 | func = source.decode |
| 361 | else: |
| 362 | func = source.encode |
| 363 | if expected: |
| 364 | if isinstance(source, bytes): |
| 365 | result = func(self.encoding, scheme) |
| 366 | self.assertTrue(type(result) is str, type(result)) |
| 367 | self.assertEqual(result, expected, |
| 368 | '%a.decode(%r, %r)=%a != %a' |
| 369 | % (source, self.encoding, scheme, result, |
| 370 | expected)) |
| 371 | else: |
| 372 | result = func(self.encoding, scheme) |
| 373 | self.assertTrue(type(result) is bytes, type(result)) |
| 374 | self.assertEqual(result, expected, |
| 375 | '%a.encode(%r, %r)=%a != %a' |
| 376 | % (source, self.encoding, scheme, result, |
| 377 | expected)) |
| 378 | else: |
| 379 | self.assertRaises(UnicodeError, func, self.encoding, scheme) |
| 380 | |
Victor Stinner | 033a543 | 2011-05-16 16:43:38 +0200 | [diff] [blame] | 381 | def load_teststring(name): |
| 382 | dir = os.path.join(os.path.dirname(__file__), 'cjkencodings') |
| 383 | with open(os.path.join(dir, name + '.txt'), 'rb') as f: |
| 384 | encoded = f.read() |
| 385 | with open(os.path.join(dir, name + '-utf8.txt'), 'rb') as f: |
| 386 | utf8 = f.read() |
| 387 | return encoded, utf8 |