Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1 | #!/usr/bin/env python |
| 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 | |
Christian Heimes | 05e8be1 | 2008-02-23 18:30:17 +0000 | [diff] [blame] | 7 | import sys, codecs |
Guido van Rossum | 806c246 | 2007-08-06 23:33:07 +0000 | [diff] [blame] | 8 | import unittest, re |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 9 | from test import test_support |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 10 | from io import BytesIO |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 11 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 12 | class TestBase: |
| 13 | encoding = '' # codec name |
| 14 | codec = None # codec tuple (with 4 elements) |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 15 | tstring = None # must set. 2 strings to test StreamReader |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 16 | |
| 17 | codectests = None # must set. codec test tuple |
| 18 | roundtriptest = 1 # set if roundtrip is possible with unicode |
| 19 | has_iso10646 = 0 # set if this encoding contains whole iso10646 map |
| 20 | xmlcharnametest = None # string to test xmlcharrefreplace |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 21 | unmappedunicode = '\udeee' # a unicode codepoint that is not mapped. |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 22 | |
| 23 | def setUp(self): |
| 24 | if self.codec is None: |
| 25 | self.codec = codecs.lookup(self.encoding) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 26 | self.encode = self.codec.encode |
| 27 | self.decode = self.codec.decode |
| 28 | self.reader = self.codec.streamreader |
| 29 | self.writer = self.codec.streamwriter |
| 30 | self.incrementalencoder = self.codec.incrementalencoder |
| 31 | self.incrementaldecoder = self.codec.incrementaldecoder |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 32 | |
| 33 | def test_chunkcoding(self): |
Guido van Rossum | 49f2174 | 2007-10-10 00:12:46 +0000 | [diff] [blame] | 34 | tstring_lines = [] |
| 35 | for b in self.tstring: |
| 36 | lines = b.split(b"\n") |
| 37 | last = lines.pop() |
| 38 | assert last == b"" |
| 39 | lines = [line + b"\n" for line in lines] |
| 40 | tstring_lines.append(lines) |
| 41 | for native, utf8 in zip(*tstring_lines): |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 42 | u = self.decode(native)[0] |
| 43 | self.assertEqual(u, utf8.decode('utf-8')) |
| 44 | if self.roundtriptest: |
| 45 | self.assertEqual(native, self.encode(u)[0]) |
| 46 | |
| 47 | def test_errorhandle(self): |
| 48 | for source, scheme, expected in self.codectests: |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 49 | if isinstance(source, bytes): |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 50 | func = self.decode |
| 51 | else: |
| 52 | func = self.encode |
| 53 | if expected: |
| 54 | result = func(source, scheme)[0] |
Guido van Rossum | 98297ee | 2007-11-06 21:34:58 +0000 | [diff] [blame] | 55 | if func is self.decode: |
| 56 | self.assert_(type(result) is str, type(result)) |
| 57 | else: |
| 58 | self.assert_(type(result) is bytes, type(result)) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 59 | self.assertEqual(result, expected) |
| 60 | else: |
| 61 | self.assertRaises(UnicodeError, func, source, scheme) |
| 62 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 63 | def test_xmlcharrefreplace(self): |
| 64 | if self.has_iso10646: |
| 65 | return |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 66 | |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 67 | s = "\u0b13\u0b23\u0b60 nd eggs" |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 68 | self.assertEqual( |
| 69 | self.encode(s, "xmlcharrefreplace")[0], |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 70 | b"ଓଣୠ nd eggs" |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 71 | ) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 72 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 73 | def test_customreplace_encode(self): |
| 74 | if self.has_iso10646: |
| 75 | return |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 76 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 77 | from htmlentitydefs import codepoint2name |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 78 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 79 | def xmlcharnamereplace(exc): |
| 80 | if not isinstance(exc, UnicodeEncodeError): |
| 81 | raise TypeError("don't know how to handle %r" % exc) |
| 82 | l = [] |
| 83 | for c in exc.object[exc.start:exc.end]: |
| 84 | if ord(c) in codepoint2name: |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 85 | l.append("&%s;" % codepoint2name[ord(c)]) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 86 | else: |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 87 | l.append("&#%d;" % ord(c)) |
| 88 | return ("".join(l), exc.end) |
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 | codecs.register_error("test.xmlcharnamereplace", xmlcharnamereplace) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 91 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 92 | if self.xmlcharnametest: |
| 93 | sin, sout = self.xmlcharnametest |
| 94 | else: |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 95 | sin = "\xab\u211c\xbb = \u2329\u1234\u232a" |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 96 | sout = b"«ℜ» = ⟨ሴ⟩" |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 97 | self.assertEqual(self.encode(sin, |
| 98 | "test.xmlcharnamereplace")[0], sout) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 99 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 100 | def test_callback_wrong_objects(self): |
| 101 | def myreplace(exc): |
| 102 | return (ret, exc.end) |
| 103 | codecs.register_error("test.cjktest", myreplace) |
| 104 | |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 105 | for ret in ([1, 2, 3], [], None, object(), b'string', b''): |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 106 | self.assertRaises(TypeError, self.encode, self.unmappedunicode, |
| 107 | 'test.cjktest') |
| 108 | |
| 109 | def test_callback_long_index(self): |
| 110 | def myreplace(exc): |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 111 | return ('x', int(exc.end)) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 112 | codecs.register_error("test.cjktest", myreplace) |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 113 | self.assertEqual(self.encode('abcd' + self.unmappedunicode + 'efgh', |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 114 | 'test.cjktest'), (b'abcdxefgh', 9)) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 115 | |
| 116 | def myreplace(exc): |
Christian Heimes | a37d4c6 | 2007-12-04 23:02:19 +0000 | [diff] [blame] | 117 | return ('x', sys.maxsize + 1) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 118 | codecs.register_error("test.cjktest", myreplace) |
| 119 | self.assertRaises(IndexError, self.encode, self.unmappedunicode, |
| 120 | 'test.cjktest') |
| 121 | |
| 122 | def test_callback_None_index(self): |
| 123 | def myreplace(exc): |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 124 | return ('x', None) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 125 | codecs.register_error("test.cjktest", myreplace) |
| 126 | self.assertRaises(TypeError, self.encode, self.unmappedunicode, |
| 127 | 'test.cjktest') |
| 128 | |
| 129 | def test_callback_backward_index(self): |
| 130 | def myreplace(exc): |
| 131 | if myreplace.limit > 0: |
| 132 | myreplace.limit -= 1 |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 133 | return ('REPLACED', 0) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 134 | else: |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 135 | return ('TERMINAL', exc.end) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 136 | myreplace.limit = 3 |
| 137 | codecs.register_error("test.cjktest", myreplace) |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 138 | self.assertEqual(self.encode('abcd' + self.unmappedunicode + 'efgh', |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 139 | 'test.cjktest'), |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 140 | (b'abcdREPLACEDabcdREPLACEDabcdREPLACEDabcdTERMINALefgh', 9)) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 141 | |
| 142 | def test_callback_forward_index(self): |
| 143 | def myreplace(exc): |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 144 | return ('REPLACED', exc.end + 2) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 145 | codecs.register_error("test.cjktest", myreplace) |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 146 | self.assertEqual(self.encode('abcd' + self.unmappedunicode + 'efgh', |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 147 | 'test.cjktest'), (b'abcdREPLACEDgh', 9)) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 148 | |
| 149 | def test_callback_index_outofbound(self): |
| 150 | def myreplace(exc): |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 151 | return ('TERM', 100) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 152 | codecs.register_error("test.cjktest", myreplace) |
| 153 | self.assertRaises(IndexError, self.encode, self.unmappedunicode, |
| 154 | 'test.cjktest') |
| 155 | |
| 156 | def test_incrementalencoder(self): |
| 157 | UTF8Reader = codecs.getreader('utf-8') |
Guido van Rossum | 805365e | 2007-05-07 22:24:25 +0000 | [diff] [blame] | 158 | for sizehint in [None] + list(range(1, 33)) + \ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 159 | [64, 128, 256, 512, 1024]: |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 160 | istream = UTF8Reader(BytesIO(self.tstring[1])) |
| 161 | ostream = BytesIO() |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 162 | encoder = self.incrementalencoder() |
| 163 | while 1: |
| 164 | if sizehint is not None: |
| 165 | data = istream.read(sizehint) |
| 166 | else: |
| 167 | data = istream.read() |
| 168 | |
| 169 | if not data: |
| 170 | break |
| 171 | e = encoder.encode(data) |
| 172 | ostream.write(e) |
| 173 | |
| 174 | self.assertEqual(ostream.getvalue(), self.tstring[0]) |
| 175 | |
| 176 | def test_incrementaldecoder(self): |
| 177 | UTF8Writer = codecs.getwriter('utf-8') |
Guido van Rossum | 805365e | 2007-05-07 22:24:25 +0000 | [diff] [blame] | 178 | for sizehint in [None, -1] + list(range(1, 33)) + \ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 179 | [64, 128, 256, 512, 1024]: |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 180 | istream = BytesIO(self.tstring[0]) |
| 181 | ostream = UTF8Writer(BytesIO()) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 182 | decoder = self.incrementaldecoder() |
| 183 | while 1: |
| 184 | data = istream.read(sizehint) |
| 185 | if not data: |
| 186 | break |
| 187 | else: |
| 188 | u = decoder.decode(data) |
| 189 | ostream.write(u) |
| 190 | |
| 191 | self.assertEqual(ostream.getvalue(), self.tstring[1]) |
| 192 | |
| 193 | def test_incrementalencoder_error_callback(self): |
| 194 | inv = self.unmappedunicode |
| 195 | |
| 196 | e = self.incrementalencoder() |
| 197 | self.assertRaises(UnicodeEncodeError, e.encode, inv, True) |
| 198 | |
| 199 | e.errors = 'ignore' |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 200 | self.assertEqual(e.encode(inv, True), b'') |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 201 | |
| 202 | e.reset() |
| 203 | def tempreplace(exc): |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 204 | return ('called', exc.end) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 205 | codecs.register_error('test.incremental_error_callback', tempreplace) |
| 206 | e.errors = 'test.incremental_error_callback' |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 207 | self.assertEqual(e.encode(inv, True), b'called') |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 208 | |
| 209 | # again |
| 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'') |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 212 | |
| 213 | def test_streamreader(self): |
| 214 | UTF8Writer = codecs.getwriter('utf-8') |
| 215 | for name in ["read", "readline", "readlines"]: |
Guido van Rossum | 805365e | 2007-05-07 22:24:25 +0000 | [diff] [blame] | 216 | for sizehint in [None, -1] + list(range(1, 33)) + \ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 217 | [64, 128, 256, 512, 1024]: |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 218 | istream = self.reader(BytesIO(self.tstring[0])) |
| 219 | ostream = UTF8Writer(BytesIO()) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 220 | func = getattr(istream, name) |
| 221 | while 1: |
| 222 | data = func(sizehint) |
| 223 | if not data: |
| 224 | break |
| 225 | if name == "readlines": |
| 226 | ostream.writelines(data) |
| 227 | else: |
| 228 | ostream.write(data) |
| 229 | |
| 230 | self.assertEqual(ostream.getvalue(), self.tstring[1]) |
| 231 | |
| 232 | def test_streamwriter(self): |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 233 | readfuncs = ('read', 'readline', 'readlines') |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 234 | UTF8Reader = codecs.getreader('utf-8') |
| 235 | for name in readfuncs: |
Guido van Rossum | 805365e | 2007-05-07 22:24:25 +0000 | [diff] [blame] | 236 | for sizehint in [None] + list(range(1, 33)) + \ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 237 | [64, 128, 256, 512, 1024]: |
Guido van Rossum | 024da5c | 2007-05-17 23:59:11 +0000 | [diff] [blame] | 238 | istream = UTF8Reader(BytesIO(self.tstring[1])) |
| 239 | ostream = self.writer(BytesIO()) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 240 | func = getattr(istream, name) |
| 241 | while 1: |
| 242 | if sizehint is not None: |
| 243 | data = func(sizehint) |
| 244 | else: |
| 245 | data = func() |
| 246 | |
| 247 | if not data: |
| 248 | break |
| 249 | if name == "readlines": |
| 250 | ostream.writelines(data) |
| 251 | else: |
| 252 | ostream.write(data) |
| 253 | |
| 254 | self.assertEqual(ostream.getvalue(), self.tstring[0]) |
| 255 | |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 256 | if len('\U00012345') == 2: # ucs2 build |
Guido van Rossum | 84fc66d | 2007-05-03 17:18:26 +0000 | [diff] [blame] | 257 | _unichr = chr |
| 258 | def chr(v): |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 259 | if v >= 0x10000: |
| 260 | return _unichr(0xd800 + ((v - 0x10000) >> 10)) + \ |
| 261 | _unichr(0xdc00 + ((v - 0x10000) & 0x3ff)) |
| 262 | else: |
| 263 | return _unichr(v) |
| 264 | _ord = ord |
| 265 | def ord(c): |
| 266 | if len(c) == 2: |
| 267 | return 0x10000 + ((_ord(c[0]) - 0xd800) << 10) + \ |
| 268 | (ord(c[1]) - 0xdc00) |
| 269 | else: |
| 270 | return _ord(c) |
| 271 | |
| 272 | class TestBase_Mapping(unittest.TestCase): |
| 273 | pass_enctest = [] |
| 274 | pass_dectest = [] |
| 275 | supmaps = [] |
| 276 | |
| 277 | def __init__(self, *args, **kw): |
| 278 | unittest.TestCase.__init__(self, *args, **kw) |
Hye-Shik Chang | aaa2f1d | 2005-12-10 17:44:27 +0000 | [diff] [blame] | 279 | self.open_mapping_file() # test it to report the error early |
| 280 | |
| 281 | def open_mapping_file(self): |
| 282 | return test_support.open_urlresource(self.mapfileurl) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 283 | |
| 284 | def test_mapping_file(self): |
Guido van Rossum | 806c246 | 2007-08-06 23:33:07 +0000 | [diff] [blame] | 285 | if self.mapfileurl.endswith('.xml'): |
| 286 | self._test_mapping_file_ucm() |
| 287 | else: |
| 288 | self._test_mapping_file_plain() |
| 289 | |
| 290 | def _test_mapping_file_plain(self): |
Guido van Rossum | 84fc66d | 2007-05-03 17:18:26 +0000 | [diff] [blame] | 291 | unichrs = lambda s: ''.join(map(chr, map(eval, s.split('+')))) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 292 | urt_wa = {} |
| 293 | |
Hye-Shik Chang | aaa2f1d | 2005-12-10 17:44:27 +0000 | [diff] [blame] | 294 | for line in self.open_mapping_file(): |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 295 | if not line: |
| 296 | break |
| 297 | data = line.split('#')[0].strip().split() |
| 298 | if len(data) != 2: |
| 299 | continue |
| 300 | |
| 301 | csetval = eval(data[0]) |
| 302 | if csetval <= 0x7F: |
Guido van Rossum | 005ebb1 | 2007-07-23 18:06:59 +0000 | [diff] [blame] | 303 | csetch = bytes([csetval & 0xff]) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 304 | elif csetval >= 0x1000000: |
Guido van Rossum | 005ebb1 | 2007-07-23 18:06:59 +0000 | [diff] [blame] | 305 | csetch = bytes([(csetval >> 24), ((csetval >> 16) & 0xff), |
| 306 | ((csetval >> 8) & 0xff), (csetval & 0xff)]) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 307 | elif csetval >= 0x10000: |
Guido van Rossum | 005ebb1 | 2007-07-23 18:06:59 +0000 | [diff] [blame] | 308 | csetch = bytes([(csetval >> 16), ((csetval >> 8) & 0xff), |
| 309 | (csetval & 0xff)]) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 310 | elif csetval >= 0x100: |
Guido van Rossum | 005ebb1 | 2007-07-23 18:06:59 +0000 | [diff] [blame] | 311 | csetch = bytes([(csetval >> 8), (csetval & 0xff)]) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 312 | else: |
| 313 | continue |
| 314 | |
| 315 | unich = unichrs(data[1]) |
Guido van Rossum | e2b70bc | 2006-08-18 22:13:04 +0000 | [diff] [blame] | 316 | if ord(unich) == 0xfffd or unich in urt_wa: |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 317 | continue |
| 318 | urt_wa[unich] = csetch |
| 319 | |
| 320 | self._testpoint(csetch, unich) |
| 321 | |
Guido van Rossum | 806c246 | 2007-08-06 23:33:07 +0000 | [diff] [blame] | 322 | def _test_mapping_file_ucm(self): |
| 323 | ucmdata = self.open_mapping_file().read() |
| 324 | uc = re.findall('<a u="([A-F0-9]{4})" b="([0-9A-F ]+)"/>', ucmdata) |
| 325 | for uni, coded in uc: |
| 326 | unich = chr(int(uni, 16)) |
| 327 | codech = bytes(int(c, 16) for c in coded.split()) |
| 328 | self._testpoint(codech, unich) |
| 329 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 330 | def test_mapping_supplemental(self): |
| 331 | for mapping in self.supmaps: |
| 332 | self._testpoint(*mapping) |
| 333 | |
| 334 | def _testpoint(self, csetch, unich): |
| 335 | if (csetch, unich) not in self.pass_enctest: |
| 336 | self.assertEqual(unich.encode(self.encoding), csetch) |
| 337 | if (csetch, unich) not in self.pass_dectest: |
Guido van Rossum | ef87d6e | 2007-05-02 19:09:54 +0000 | [diff] [blame] | 338 | self.assertEqual(str(csetch, self.encoding), unich) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 339 | |
| 340 | def load_teststring(encoding): |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 341 | from test import cjkencodings_test |
| 342 | return cjkencodings_test.teststring[encoding] |