Benjamin Peterson | 46a9900 | 2010-01-09 18:45:30 +0000 | [diff] [blame] | 1 | # Copyright (C) 2001-2010 Python Software Foundation |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2 | # Contact: email-sig@python.org |
| 3 | # email package unit tests |
| 4 | |
R. David Murray | 719a449 | 2010-11-21 16:53:48 +0000 | [diff] [blame] | 5 | import re |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 6 | import time |
| 7 | import base64 |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 8 | import unittest |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 9 | import textwrap |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 10 | |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 11 | from io import StringIO, BytesIO |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 12 | from itertools import chain |
| 13 | |
| 14 | import email |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 15 | import email.policy |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 16 | |
| 17 | from email.charset import Charset |
| 18 | from email.header import Header, decode_header, make_header |
| 19 | from email.parser import Parser, HeaderParser |
R David Murray | 638d40b | 2012-08-24 11:14:13 -0400 | [diff] [blame] | 20 | from email.generator import Generator, DecodedGenerator, BytesGenerator |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 21 | from email.message import Message |
| 22 | from email.mime.application import MIMEApplication |
| 23 | from email.mime.audio import MIMEAudio |
| 24 | from email.mime.text import MIMEText |
| 25 | from email.mime.image import MIMEImage |
| 26 | from email.mime.base import MIMEBase |
| 27 | from email.mime.message import MIMEMessage |
| 28 | from email.mime.multipart import MIMEMultipart |
| 29 | from email import utils |
| 30 | from email import errors |
| 31 | from email import encoders |
| 32 | from email import iterators |
| 33 | from email import base64mime |
| 34 | from email import quoprimime |
| 35 | |
R David Murray | 965794e | 2013-03-07 18:16:47 -0500 | [diff] [blame] | 36 | from test.support import unlink |
R David Murray | a256bac | 2011-03-31 12:20:23 -0400 | [diff] [blame] | 37 | from test.test_email import openfile, TestEmailBase |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 38 | |
R David Murray | 612528d | 2013-03-15 20:38:15 -0400 | [diff] [blame] | 39 | # These imports are documented to work, but we are testing them using a |
| 40 | # different path, so we import them here just to make sure they are importable. |
| 41 | from email.parser import FeedParser, BytesFeedParser |
| 42 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 43 | NL = '\n' |
| 44 | EMPTYSTRING = '' |
| 45 | SPACE = ' ' |
| 46 | |
| 47 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 48 | # Test various aspects of the Message class's API |
| 49 | class TestMessageAPI(TestEmailBase): |
| 50 | def test_get_all(self): |
| 51 | eq = self.assertEqual |
| 52 | msg = self._msgobj('msg_20.txt') |
| 53 | eq(msg.get_all('cc'), ['ccc@zzz.org', 'ddd@zzz.org', 'eee@zzz.org']) |
| 54 | eq(msg.get_all('xx', 'n/a'), 'n/a') |
| 55 | |
R. David Murray | e5db263 | 2010-11-20 15:10:13 +0000 | [diff] [blame] | 56 | def test_getset_charset(self): |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 57 | eq = self.assertEqual |
| 58 | msg = Message() |
| 59 | eq(msg.get_charset(), None) |
| 60 | charset = Charset('iso-8859-1') |
| 61 | msg.set_charset(charset) |
| 62 | eq(msg['mime-version'], '1.0') |
| 63 | eq(msg.get_content_type(), 'text/plain') |
| 64 | eq(msg['content-type'], 'text/plain; charset="iso-8859-1"') |
| 65 | eq(msg.get_param('charset'), 'iso-8859-1') |
| 66 | eq(msg['content-transfer-encoding'], 'quoted-printable') |
| 67 | eq(msg.get_charset().input_charset, 'iso-8859-1') |
| 68 | # Remove the charset |
| 69 | msg.set_charset(None) |
| 70 | eq(msg.get_charset(), None) |
| 71 | eq(msg['content-type'], 'text/plain') |
| 72 | # Try adding a charset when there's already MIME headers present |
| 73 | msg = Message() |
| 74 | msg['MIME-Version'] = '2.0' |
| 75 | msg['Content-Type'] = 'text/x-weird' |
| 76 | msg['Content-Transfer-Encoding'] = 'quinted-puntable' |
| 77 | msg.set_charset(charset) |
| 78 | eq(msg['mime-version'], '2.0') |
| 79 | eq(msg['content-type'], 'text/x-weird; charset="iso-8859-1"') |
| 80 | eq(msg['content-transfer-encoding'], 'quinted-puntable') |
| 81 | |
| 82 | def test_set_charset_from_string(self): |
| 83 | eq = self.assertEqual |
| 84 | msg = Message() |
| 85 | msg.set_charset('us-ascii') |
| 86 | eq(msg.get_charset().input_charset, 'us-ascii') |
| 87 | eq(msg['content-type'], 'text/plain; charset="us-ascii"') |
| 88 | |
| 89 | def test_set_payload_with_charset(self): |
| 90 | msg = Message() |
| 91 | charset = Charset('iso-8859-1') |
| 92 | msg.set_payload('This is a string payload', charset) |
| 93 | self.assertEqual(msg.get_charset().input_charset, 'iso-8859-1') |
| 94 | |
| 95 | def test_get_charsets(self): |
| 96 | eq = self.assertEqual |
| 97 | |
| 98 | msg = self._msgobj('msg_08.txt') |
| 99 | charsets = msg.get_charsets() |
| 100 | eq(charsets, [None, 'us-ascii', 'iso-8859-1', 'iso-8859-2', 'koi8-r']) |
| 101 | |
| 102 | msg = self._msgobj('msg_09.txt') |
| 103 | charsets = msg.get_charsets('dingbat') |
| 104 | eq(charsets, ['dingbat', 'us-ascii', 'iso-8859-1', 'dingbat', |
| 105 | 'koi8-r']) |
| 106 | |
| 107 | msg = self._msgobj('msg_12.txt') |
| 108 | charsets = msg.get_charsets() |
| 109 | eq(charsets, [None, 'us-ascii', 'iso-8859-1', None, 'iso-8859-2', |
| 110 | 'iso-8859-3', 'us-ascii', 'koi8-r']) |
| 111 | |
| 112 | def test_get_filename(self): |
| 113 | eq = self.assertEqual |
| 114 | |
| 115 | msg = self._msgobj('msg_04.txt') |
| 116 | filenames = [p.get_filename() for p in msg.get_payload()] |
| 117 | eq(filenames, ['msg.txt', 'msg.txt']) |
| 118 | |
| 119 | msg = self._msgobj('msg_07.txt') |
| 120 | subpart = msg.get_payload(1) |
| 121 | eq(subpart.get_filename(), 'dingusfish.gif') |
| 122 | |
| 123 | def test_get_filename_with_name_parameter(self): |
| 124 | eq = self.assertEqual |
| 125 | |
| 126 | msg = self._msgobj('msg_44.txt') |
| 127 | filenames = [p.get_filename() for p in msg.get_payload()] |
| 128 | eq(filenames, ['msg.txt', 'msg.txt']) |
| 129 | |
| 130 | def test_get_boundary(self): |
| 131 | eq = self.assertEqual |
| 132 | msg = self._msgobj('msg_07.txt') |
| 133 | # No quotes! |
| 134 | eq(msg.get_boundary(), 'BOUNDARY') |
| 135 | |
| 136 | def test_set_boundary(self): |
| 137 | eq = self.assertEqual |
| 138 | # This one has no existing boundary parameter, but the Content-Type: |
| 139 | # header appears fifth. |
| 140 | msg = self._msgobj('msg_01.txt') |
| 141 | msg.set_boundary('BOUNDARY') |
| 142 | header, value = msg.items()[4] |
| 143 | eq(header.lower(), 'content-type') |
| 144 | eq(value, 'text/plain; charset="us-ascii"; boundary="BOUNDARY"') |
| 145 | # This one has a Content-Type: header, with a boundary, stuck in the |
| 146 | # middle of its headers. Make sure the order is preserved; it should |
| 147 | # be fifth. |
| 148 | msg = self._msgobj('msg_04.txt') |
| 149 | msg.set_boundary('BOUNDARY') |
| 150 | header, value = msg.items()[4] |
| 151 | eq(header.lower(), 'content-type') |
| 152 | eq(value, 'multipart/mixed; boundary="BOUNDARY"') |
| 153 | # And this one has no Content-Type: header at all. |
| 154 | msg = self._msgobj('msg_03.txt') |
| 155 | self.assertRaises(errors.HeaderParseError, |
| 156 | msg.set_boundary, 'BOUNDARY') |
| 157 | |
R. David Murray | 73a559d | 2010-12-21 18:07:59 +0000 | [diff] [blame] | 158 | def test_make_boundary(self): |
| 159 | msg = MIMEMultipart('form-data') |
| 160 | # Note that when the boundary gets created is an implementation |
| 161 | # detail and might change. |
| 162 | self.assertEqual(msg.items()[0][1], 'multipart/form-data') |
| 163 | # Trigger creation of boundary |
| 164 | msg.as_string() |
| 165 | self.assertEqual(msg.items()[0][1][:33], |
| 166 | 'multipart/form-data; boundary="==') |
| 167 | # XXX: there ought to be tests of the uniqueness of the boundary, too. |
| 168 | |
R. David Murray | 57c45ac | 2010-02-21 04:39:40 +0000 | [diff] [blame] | 169 | def test_message_rfc822_only(self): |
| 170 | # Issue 7970: message/rfc822 not in multipart parsed by |
| 171 | # HeaderParser caused an exception when flattened. |
R David Murray | 28346b8 | 2011-03-31 11:40:20 -0400 | [diff] [blame] | 172 | with openfile('msg_46.txt') as fp: |
Brett Cannon | 384917a | 2010-10-29 23:08:36 +0000 | [diff] [blame] | 173 | msgdata = fp.read() |
R. David Murray | 57c45ac | 2010-02-21 04:39:40 +0000 | [diff] [blame] | 174 | parser = HeaderParser() |
| 175 | msg = parser.parsestr(msgdata) |
| 176 | out = StringIO() |
| 177 | gen = Generator(out, True, 0) |
| 178 | gen.flatten(msg, False) |
| 179 | self.assertEqual(out.getvalue(), msgdata) |
| 180 | |
R David Murray | b35c850 | 2011-04-13 16:46:05 -0400 | [diff] [blame] | 181 | def test_byte_message_rfc822_only(self): |
| 182 | # Make sure new bytes header parser also passes this. |
Terry Jan Reedy | 740d6b6 | 2013-08-31 17:12:21 -0400 | [diff] [blame^] | 183 | with openfile('msg_46.txt') as fp: |
| 184 | msgdata = fp.read().encode('ascii') |
R David Murray | b35c850 | 2011-04-13 16:46:05 -0400 | [diff] [blame] | 185 | parser = email.parser.BytesHeaderParser() |
| 186 | msg = parser.parsebytes(msgdata) |
| 187 | out = BytesIO() |
| 188 | gen = email.generator.BytesGenerator(out) |
| 189 | gen.flatten(msg) |
| 190 | self.assertEqual(out.getvalue(), msgdata) |
| 191 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 192 | def test_get_decoded_payload(self): |
| 193 | eq = self.assertEqual |
| 194 | msg = self._msgobj('msg_10.txt') |
| 195 | # The outer message is a multipart |
| 196 | eq(msg.get_payload(decode=True), None) |
| 197 | # Subpart 1 is 7bit encoded |
| 198 | eq(msg.get_payload(0).get_payload(decode=True), |
| 199 | b'This is a 7bit encoded message.\n') |
| 200 | # Subpart 2 is quopri |
| 201 | eq(msg.get_payload(1).get_payload(decode=True), |
| 202 | b'\xa1This is a Quoted Printable encoded message!\n') |
| 203 | # Subpart 3 is base64 |
| 204 | eq(msg.get_payload(2).get_payload(decode=True), |
| 205 | b'This is a Base64 encoded message.') |
R. David Murray | 57a4b98 | 2010-03-08 02:17:03 +0000 | [diff] [blame] | 206 | # Subpart 4 is base64 with a trailing newline, which |
| 207 | # used to be stripped (issue 7143). |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 208 | eq(msg.get_payload(3).get_payload(decode=True), |
R. David Murray | 57a4b98 | 2010-03-08 02:17:03 +0000 | [diff] [blame] | 209 | b'This is a Base64 encoded message.\n') |
| 210 | # Subpart 5 has no Content-Transfer-Encoding: header. |
| 211 | eq(msg.get_payload(4).get_payload(decode=True), |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 212 | b'This has no Content-Transfer-Encoding: header.\n') |
| 213 | |
| 214 | def test_get_decoded_uu_payload(self): |
| 215 | eq = self.assertEqual |
| 216 | msg = Message() |
| 217 | msg.set_payload('begin 666 -\n+:&5L;&\\@=V]R;&0 \n \nend\n') |
| 218 | for cte in ('x-uuencode', 'uuencode', 'uue', 'x-uue'): |
| 219 | msg['content-transfer-encoding'] = cte |
| 220 | eq(msg.get_payload(decode=True), b'hello world') |
| 221 | # Now try some bogus data |
| 222 | msg.set_payload('foo') |
| 223 | eq(msg.get_payload(decode=True), b'foo') |
| 224 | |
R David Murray | a2860e8 | 2011-04-16 09:20:30 -0400 | [diff] [blame] | 225 | def test_get_payload_n_raises_on_non_multipart(self): |
| 226 | msg = Message() |
| 227 | self.assertRaises(TypeError, msg.get_payload, 1) |
| 228 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 229 | def test_decoded_generator(self): |
| 230 | eq = self.assertEqual |
| 231 | msg = self._msgobj('msg_07.txt') |
| 232 | with openfile('msg_17.txt') as fp: |
| 233 | text = fp.read() |
| 234 | s = StringIO() |
| 235 | g = DecodedGenerator(s) |
| 236 | g.flatten(msg) |
| 237 | eq(s.getvalue(), text) |
| 238 | |
| 239 | def test__contains__(self): |
| 240 | msg = Message() |
| 241 | msg['From'] = 'Me' |
| 242 | msg['to'] = 'You' |
| 243 | # Check for case insensitivity |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 244 | self.assertTrue('from' in msg) |
| 245 | self.assertTrue('From' in msg) |
| 246 | self.assertTrue('FROM' in msg) |
| 247 | self.assertTrue('to' in msg) |
| 248 | self.assertTrue('To' in msg) |
| 249 | self.assertTrue('TO' in msg) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 250 | |
| 251 | def test_as_string(self): |
| 252 | eq = self.ndiffAssertEqual |
| 253 | msg = self._msgobj('msg_01.txt') |
| 254 | with openfile('msg_01.txt') as fp: |
| 255 | text = fp.read() |
| 256 | eq(text, str(msg)) |
| 257 | fullrepr = msg.as_string(unixfrom=True) |
| 258 | lines = fullrepr.split('\n') |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 259 | self.assertTrue(lines[0].startswith('From ')) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 260 | eq(text, NL.join(lines[1:])) |
| 261 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 262 | # test_headerregistry.TestContentTypeHeader.bad_params |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 263 | def test_bad_param(self): |
| 264 | msg = email.message_from_string("Content-Type: blarg; baz; boo\n") |
| 265 | self.assertEqual(msg.get_param('baz'), '') |
| 266 | |
| 267 | def test_missing_filename(self): |
| 268 | msg = email.message_from_string("From: foo\n") |
| 269 | self.assertEqual(msg.get_filename(), None) |
| 270 | |
| 271 | def test_bogus_filename(self): |
| 272 | msg = email.message_from_string( |
| 273 | "Content-Disposition: blarg; filename\n") |
| 274 | self.assertEqual(msg.get_filename(), '') |
| 275 | |
| 276 | def test_missing_boundary(self): |
| 277 | msg = email.message_from_string("From: foo\n") |
| 278 | self.assertEqual(msg.get_boundary(), None) |
| 279 | |
| 280 | def test_get_params(self): |
| 281 | eq = self.assertEqual |
| 282 | msg = email.message_from_string( |
| 283 | 'X-Header: foo=one; bar=two; baz=three\n') |
| 284 | eq(msg.get_params(header='x-header'), |
| 285 | [('foo', 'one'), ('bar', 'two'), ('baz', 'three')]) |
| 286 | msg = email.message_from_string( |
| 287 | 'X-Header: foo; bar=one; baz=two\n') |
| 288 | eq(msg.get_params(header='x-header'), |
| 289 | [('foo', ''), ('bar', 'one'), ('baz', 'two')]) |
| 290 | eq(msg.get_params(), None) |
| 291 | msg = email.message_from_string( |
| 292 | 'X-Header: foo; bar="one"; baz=two\n') |
| 293 | eq(msg.get_params(header='x-header'), |
| 294 | [('foo', ''), ('bar', 'one'), ('baz', 'two')]) |
| 295 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 296 | # test_headerregistry.TestContentTypeHeader.spaces_around_param_equals |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 297 | def test_get_param_liberal(self): |
| 298 | msg = Message() |
| 299 | msg['Content-Type'] = 'Content-Type: Multipart/mixed; boundary = "CPIMSSMTPC06p5f3tG"' |
| 300 | self.assertEqual(msg.get_param('boundary'), 'CPIMSSMTPC06p5f3tG') |
| 301 | |
| 302 | def test_get_param(self): |
| 303 | eq = self.assertEqual |
| 304 | msg = email.message_from_string( |
| 305 | "X-Header: foo=one; bar=two; baz=three\n") |
| 306 | eq(msg.get_param('bar', header='x-header'), 'two') |
| 307 | eq(msg.get_param('quuz', header='x-header'), None) |
| 308 | eq(msg.get_param('quuz'), None) |
| 309 | msg = email.message_from_string( |
| 310 | 'X-Header: foo; bar="one"; baz=two\n') |
| 311 | eq(msg.get_param('foo', header='x-header'), '') |
| 312 | eq(msg.get_param('bar', header='x-header'), 'one') |
| 313 | eq(msg.get_param('baz', header='x-header'), 'two') |
| 314 | # XXX: We are not RFC-2045 compliant! We cannot parse: |
| 315 | # msg["Content-Type"] = 'text/plain; weird="hey; dolly? [you] @ <\\"home\\">?"' |
| 316 | # msg.get_param("weird") |
| 317 | # yet. |
| 318 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 319 | # test_headerregistry.TestContentTypeHeader.spaces_around_semis |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 320 | def test_get_param_funky_continuation_lines(self): |
| 321 | msg = self._msgobj('msg_22.txt') |
| 322 | self.assertEqual(msg.get_payload(1).get_param('name'), 'wibble.JPG') |
| 323 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 324 | # test_headerregistry.TestContentTypeHeader.semis_inside_quotes |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 325 | def test_get_param_with_semis_in_quotes(self): |
| 326 | msg = email.message_from_string( |
| 327 | 'Content-Type: image/pjpeg; name="Jim&&Jill"\n') |
| 328 | self.assertEqual(msg.get_param('name'), 'Jim&&Jill') |
| 329 | self.assertEqual(msg.get_param('name', unquote=False), |
| 330 | '"Jim&&Jill"') |
| 331 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 332 | # test_headerregistry.TestContentTypeHeader.quotes_inside_rfc2231_value |
R. David Murray | d48739f | 2010-04-14 18:59:18 +0000 | [diff] [blame] | 333 | def test_get_param_with_quotes(self): |
| 334 | msg = email.message_from_string( |
| 335 | 'Content-Type: foo; bar*0="baz\\"foobar"; bar*1="\\"baz"') |
| 336 | self.assertEqual(msg.get_param('bar'), 'baz"foobar"baz') |
| 337 | msg = email.message_from_string( |
| 338 | "Content-Type: foo; bar*0=\"baz\\\"foobar\"; bar*1=\"\\\"baz\"") |
| 339 | self.assertEqual(msg.get_param('bar'), 'baz"foobar"baz') |
| 340 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 341 | def test_field_containment(self): |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 342 | unless = self.assertTrue |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 343 | msg = email.message_from_string('Header: exists') |
| 344 | unless('header' in msg) |
| 345 | unless('Header' in msg) |
| 346 | unless('HEADER' in msg) |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 347 | self.assertFalse('headerx' in msg) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 348 | |
| 349 | def test_set_param(self): |
| 350 | eq = self.assertEqual |
| 351 | msg = Message() |
| 352 | msg.set_param('charset', 'iso-2022-jp') |
| 353 | eq(msg.get_param('charset'), 'iso-2022-jp') |
| 354 | msg.set_param('importance', 'high value') |
| 355 | eq(msg.get_param('importance'), 'high value') |
| 356 | eq(msg.get_param('importance', unquote=False), '"high value"') |
| 357 | eq(msg.get_params(), [('text/plain', ''), |
| 358 | ('charset', 'iso-2022-jp'), |
| 359 | ('importance', 'high value')]) |
| 360 | eq(msg.get_params(unquote=False), [('text/plain', ''), |
| 361 | ('charset', '"iso-2022-jp"'), |
| 362 | ('importance', '"high value"')]) |
| 363 | msg.set_param('charset', 'iso-9999-xx', header='X-Jimmy') |
| 364 | eq(msg.get_param('charset', header='X-Jimmy'), 'iso-9999-xx') |
| 365 | |
| 366 | def test_del_param(self): |
| 367 | eq = self.assertEqual |
| 368 | msg = self._msgobj('msg_05.txt') |
| 369 | eq(msg.get_params(), |
| 370 | [('multipart/report', ''), ('report-type', 'delivery-status'), |
| 371 | ('boundary', 'D1690A7AC1.996856090/mail.example.com')]) |
| 372 | old_val = msg.get_param("report-type") |
| 373 | msg.del_param("report-type") |
| 374 | eq(msg.get_params(), |
| 375 | [('multipart/report', ''), |
| 376 | ('boundary', 'D1690A7AC1.996856090/mail.example.com')]) |
| 377 | msg.set_param("report-type", old_val) |
| 378 | eq(msg.get_params(), |
| 379 | [('multipart/report', ''), |
| 380 | ('boundary', 'D1690A7AC1.996856090/mail.example.com'), |
| 381 | ('report-type', old_val)]) |
| 382 | |
| 383 | def test_del_param_on_other_header(self): |
| 384 | msg = Message() |
| 385 | msg.add_header('Content-Disposition', 'attachment', filename='bud.gif') |
| 386 | msg.del_param('filename', 'content-disposition') |
| 387 | self.assertEqual(msg['content-disposition'], 'attachment') |
| 388 | |
R David Murray | a2860e8 | 2011-04-16 09:20:30 -0400 | [diff] [blame] | 389 | def test_del_param_on_nonexistent_header(self): |
| 390 | msg = Message() |
R David Murray | 271ade8 | 2013-07-25 12:11:55 -0400 | [diff] [blame] | 391 | # Deleting param on empty msg should not raise exception. |
R David Murray | a2860e8 | 2011-04-16 09:20:30 -0400 | [diff] [blame] | 392 | msg.del_param('filename', 'content-disposition') |
| 393 | |
| 394 | def test_del_nonexistent_param(self): |
| 395 | msg = Message() |
| 396 | msg.add_header('Content-Type', 'text/plain', charset='utf-8') |
| 397 | existing_header = msg['Content-Type'] |
| 398 | msg.del_param('foobar', header='Content-Type') |
R David Murray | 271ade8 | 2013-07-25 12:11:55 -0400 | [diff] [blame] | 399 | self.assertEqual(msg['Content-Type'], existing_header) |
R David Murray | a2860e8 | 2011-04-16 09:20:30 -0400 | [diff] [blame] | 400 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 401 | def test_set_type(self): |
| 402 | eq = self.assertEqual |
| 403 | msg = Message() |
| 404 | self.assertRaises(ValueError, msg.set_type, 'text') |
| 405 | msg.set_type('text/plain') |
| 406 | eq(msg['content-type'], 'text/plain') |
| 407 | msg.set_param('charset', 'us-ascii') |
| 408 | eq(msg['content-type'], 'text/plain; charset="us-ascii"') |
| 409 | msg.set_type('text/html') |
| 410 | eq(msg['content-type'], 'text/html; charset="us-ascii"') |
| 411 | |
| 412 | def test_set_type_on_other_header(self): |
| 413 | msg = Message() |
| 414 | msg['X-Content-Type'] = 'text/plain' |
| 415 | msg.set_type('application/octet-stream', 'X-Content-Type') |
| 416 | self.assertEqual(msg['x-content-type'], 'application/octet-stream') |
| 417 | |
| 418 | def test_get_content_type_missing(self): |
| 419 | msg = Message() |
| 420 | self.assertEqual(msg.get_content_type(), 'text/plain') |
| 421 | |
| 422 | def test_get_content_type_missing_with_default_type(self): |
| 423 | msg = Message() |
| 424 | msg.set_default_type('message/rfc822') |
| 425 | self.assertEqual(msg.get_content_type(), 'message/rfc822') |
| 426 | |
| 427 | def test_get_content_type_from_message_implicit(self): |
| 428 | msg = self._msgobj('msg_30.txt') |
| 429 | self.assertEqual(msg.get_payload(0).get_content_type(), |
| 430 | 'message/rfc822') |
| 431 | |
| 432 | def test_get_content_type_from_message_explicit(self): |
| 433 | msg = self._msgobj('msg_28.txt') |
| 434 | self.assertEqual(msg.get_payload(0).get_content_type(), |
| 435 | 'message/rfc822') |
| 436 | |
| 437 | def test_get_content_type_from_message_text_plain_implicit(self): |
| 438 | msg = self._msgobj('msg_03.txt') |
| 439 | self.assertEqual(msg.get_content_type(), 'text/plain') |
| 440 | |
| 441 | def test_get_content_type_from_message_text_plain_explicit(self): |
| 442 | msg = self._msgobj('msg_01.txt') |
| 443 | self.assertEqual(msg.get_content_type(), 'text/plain') |
| 444 | |
| 445 | def test_get_content_maintype_missing(self): |
| 446 | msg = Message() |
| 447 | self.assertEqual(msg.get_content_maintype(), 'text') |
| 448 | |
| 449 | def test_get_content_maintype_missing_with_default_type(self): |
| 450 | msg = Message() |
| 451 | msg.set_default_type('message/rfc822') |
| 452 | self.assertEqual(msg.get_content_maintype(), 'message') |
| 453 | |
| 454 | def test_get_content_maintype_from_message_implicit(self): |
| 455 | msg = self._msgobj('msg_30.txt') |
| 456 | self.assertEqual(msg.get_payload(0).get_content_maintype(), 'message') |
| 457 | |
| 458 | def test_get_content_maintype_from_message_explicit(self): |
| 459 | msg = self._msgobj('msg_28.txt') |
| 460 | self.assertEqual(msg.get_payload(0).get_content_maintype(), 'message') |
| 461 | |
| 462 | def test_get_content_maintype_from_message_text_plain_implicit(self): |
| 463 | msg = self._msgobj('msg_03.txt') |
| 464 | self.assertEqual(msg.get_content_maintype(), 'text') |
| 465 | |
| 466 | def test_get_content_maintype_from_message_text_plain_explicit(self): |
| 467 | msg = self._msgobj('msg_01.txt') |
| 468 | self.assertEqual(msg.get_content_maintype(), 'text') |
| 469 | |
| 470 | def test_get_content_subtype_missing(self): |
| 471 | msg = Message() |
| 472 | self.assertEqual(msg.get_content_subtype(), 'plain') |
| 473 | |
| 474 | def test_get_content_subtype_missing_with_default_type(self): |
| 475 | msg = Message() |
| 476 | msg.set_default_type('message/rfc822') |
| 477 | self.assertEqual(msg.get_content_subtype(), 'rfc822') |
| 478 | |
| 479 | def test_get_content_subtype_from_message_implicit(self): |
| 480 | msg = self._msgobj('msg_30.txt') |
| 481 | self.assertEqual(msg.get_payload(0).get_content_subtype(), 'rfc822') |
| 482 | |
| 483 | def test_get_content_subtype_from_message_explicit(self): |
| 484 | msg = self._msgobj('msg_28.txt') |
| 485 | self.assertEqual(msg.get_payload(0).get_content_subtype(), 'rfc822') |
| 486 | |
| 487 | def test_get_content_subtype_from_message_text_plain_implicit(self): |
| 488 | msg = self._msgobj('msg_03.txt') |
| 489 | self.assertEqual(msg.get_content_subtype(), 'plain') |
| 490 | |
| 491 | def test_get_content_subtype_from_message_text_plain_explicit(self): |
| 492 | msg = self._msgobj('msg_01.txt') |
| 493 | self.assertEqual(msg.get_content_subtype(), 'plain') |
| 494 | |
| 495 | def test_get_content_maintype_error(self): |
| 496 | msg = Message() |
| 497 | msg['Content-Type'] = 'no-slash-in-this-string' |
| 498 | self.assertEqual(msg.get_content_maintype(), 'text') |
| 499 | |
| 500 | def test_get_content_subtype_error(self): |
| 501 | msg = Message() |
| 502 | msg['Content-Type'] = 'no-slash-in-this-string' |
| 503 | self.assertEqual(msg.get_content_subtype(), 'plain') |
| 504 | |
| 505 | def test_replace_header(self): |
| 506 | eq = self.assertEqual |
| 507 | msg = Message() |
| 508 | msg.add_header('First', 'One') |
| 509 | msg.add_header('Second', 'Two') |
| 510 | msg.add_header('Third', 'Three') |
| 511 | eq(msg.keys(), ['First', 'Second', 'Third']) |
| 512 | eq(msg.values(), ['One', 'Two', 'Three']) |
| 513 | msg.replace_header('Second', 'Twenty') |
| 514 | eq(msg.keys(), ['First', 'Second', 'Third']) |
| 515 | eq(msg.values(), ['One', 'Twenty', 'Three']) |
| 516 | msg.add_header('First', 'Eleven') |
| 517 | msg.replace_header('First', 'One Hundred') |
| 518 | eq(msg.keys(), ['First', 'Second', 'Third', 'First']) |
| 519 | eq(msg.values(), ['One Hundred', 'Twenty', 'Three', 'Eleven']) |
| 520 | self.assertRaises(KeyError, msg.replace_header, 'Fourth', 'Missing') |
| 521 | |
R David Murray | 80e0aee | 2012-05-27 21:23:34 -0400 | [diff] [blame] | 522 | # test_defect_handling:test_invalid_chars_in_base64_payload |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 523 | def test_broken_base64_payload(self): |
| 524 | x = 'AwDp0P7//y6LwKEAcPa/6Q=9' |
| 525 | msg = Message() |
| 526 | msg['content-type'] = 'audio/x-midi' |
| 527 | msg['content-transfer-encoding'] = 'base64' |
| 528 | msg.set_payload(x) |
| 529 | self.assertEqual(msg.get_payload(decode=True), |
R David Murray | 80e0aee | 2012-05-27 21:23:34 -0400 | [diff] [blame] | 530 | (b'\x03\x00\xe9\xd0\xfe\xff\xff.\x8b\xc0' |
| 531 | b'\xa1\x00p\xf6\xbf\xe9\x0f')) |
| 532 | self.assertIsInstance(msg.defects[0], |
| 533 | errors.InvalidBase64CharactersDefect) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 534 | |
R David Murray | a2860e8 | 2011-04-16 09:20:30 -0400 | [diff] [blame] | 535 | def test_broken_unicode_payload(self): |
| 536 | # This test improves coverage but is not a compliance test. |
| 537 | # The behavior in this situation is currently undefined by the API. |
| 538 | x = 'this is a br\xf6ken thing to do' |
| 539 | msg = Message() |
| 540 | msg['content-type'] = 'text/plain' |
| 541 | msg['content-transfer-encoding'] = '8bit' |
| 542 | msg.set_payload(x) |
| 543 | self.assertEqual(msg.get_payload(decode=True), |
| 544 | bytes(x, 'raw-unicode-escape')) |
| 545 | |
| 546 | def test_questionable_bytes_payload(self): |
| 547 | # This test improves coverage but is not a compliance test, |
| 548 | # since it involves poking inside the black box. |
| 549 | x = 'this is a quéstionable thing to do'.encode('utf-8') |
| 550 | msg = Message() |
| 551 | msg['content-type'] = 'text/plain; charset="utf-8"' |
| 552 | msg['content-transfer-encoding'] = '8bit' |
| 553 | msg._payload = x |
| 554 | self.assertEqual(msg.get_payload(decode=True), x) |
| 555 | |
R. David Murray | 7ec754b | 2010-12-13 23:51:19 +0000 | [diff] [blame] | 556 | # Issue 1078919 |
| 557 | def test_ascii_add_header(self): |
| 558 | msg = Message() |
| 559 | msg.add_header('Content-Disposition', 'attachment', |
| 560 | filename='bud.gif') |
| 561 | self.assertEqual('attachment; filename="bud.gif"', |
| 562 | msg['Content-Disposition']) |
| 563 | |
| 564 | def test_noascii_add_header(self): |
| 565 | msg = Message() |
| 566 | msg.add_header('Content-Disposition', 'attachment', |
| 567 | filename="Fußballer.ppt") |
| 568 | self.assertEqual( |
R. David Murray | dfd7eb0 | 2010-12-24 22:36:49 +0000 | [diff] [blame] | 569 | 'attachment; filename*=utf-8\'\'Fu%C3%9Fballer.ppt', |
R. David Murray | 7ec754b | 2010-12-13 23:51:19 +0000 | [diff] [blame] | 570 | msg['Content-Disposition']) |
| 571 | |
| 572 | def test_nonascii_add_header_via_triple(self): |
| 573 | msg = Message() |
| 574 | msg.add_header('Content-Disposition', 'attachment', |
| 575 | filename=('iso-8859-1', '', 'Fußballer.ppt')) |
| 576 | self.assertEqual( |
R. David Murray | dfd7eb0 | 2010-12-24 22:36:49 +0000 | [diff] [blame] | 577 | 'attachment; filename*=iso-8859-1\'\'Fu%DFballer.ppt', |
| 578 | msg['Content-Disposition']) |
| 579 | |
| 580 | def test_ascii_add_header_with_tspecial(self): |
| 581 | msg = Message() |
| 582 | msg.add_header('Content-Disposition', 'attachment', |
| 583 | filename="windows [filename].ppt") |
| 584 | self.assertEqual( |
| 585 | 'attachment; filename="windows [filename].ppt"', |
| 586 | msg['Content-Disposition']) |
| 587 | |
| 588 | def test_nonascii_add_header_with_tspecial(self): |
| 589 | msg = Message() |
| 590 | msg.add_header('Content-Disposition', 'attachment', |
| 591 | filename="Fußballer [filename].ppt") |
| 592 | self.assertEqual( |
| 593 | "attachment; filename*=utf-8''Fu%C3%9Fballer%20%5Bfilename%5D.ppt", |
R. David Murray | 7ec754b | 2010-12-13 23:51:19 +0000 | [diff] [blame] | 594 | msg['Content-Disposition']) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 595 | |
R David Murray | 00ae435 | 2013-08-21 21:10:31 -0400 | [diff] [blame] | 596 | def test_binary_quopri_payload(self): |
| 597 | for charset in ('latin-1', 'ascii'): |
| 598 | msg = Message() |
| 599 | msg['content-type'] = 'text/plain; charset=%s' % charset |
| 600 | msg['content-transfer-encoding'] = 'quoted-printable' |
| 601 | msg.set_payload(b'foo=e6=96=87bar') |
| 602 | self.assertEqual( |
| 603 | msg.get_payload(decode=True), |
| 604 | b'foo\xe6\x96\x87bar', |
| 605 | 'get_payload returns wrong result with charset %s.' % charset) |
| 606 | |
| 607 | def test_binary_base64_payload(self): |
| 608 | for charset in ('latin-1', 'ascii'): |
| 609 | msg = Message() |
| 610 | msg['content-type'] = 'text/plain; charset=%s' % charset |
| 611 | msg['content-transfer-encoding'] = 'base64' |
| 612 | msg.set_payload(b'Zm9v5paHYmFy') |
| 613 | self.assertEqual( |
| 614 | msg.get_payload(decode=True), |
| 615 | b'foo\xe6\x96\x87bar', |
| 616 | 'get_payload returns wrong result with charset %s.' % charset) |
| 617 | |
| 618 | def test_binary_uuencode_payload(self): |
| 619 | for charset in ('latin-1', 'ascii'): |
| 620 | for encoding in ('x-uuencode', 'uuencode', 'uue', 'x-uue'): |
| 621 | msg = Message() |
| 622 | msg['content-type'] = 'text/plain; charset=%s' % charset |
| 623 | msg['content-transfer-encoding'] = encoding |
| 624 | msg.set_payload(b"begin 666 -\n)9F]OYI:'8F%R\n \nend\n") |
| 625 | self.assertEqual( |
| 626 | msg.get_payload(decode=True), |
| 627 | b'foo\xe6\x96\x87bar', |
| 628 | str(('get_payload returns wrong result ', |
| 629 | 'with charset {0} and encoding {1}.')).\ |
| 630 | format(charset, encoding)) |
| 631 | |
R David Murray | a2860e8 | 2011-04-16 09:20:30 -0400 | [diff] [blame] | 632 | def test_add_header_with_name_only_param(self): |
| 633 | msg = Message() |
| 634 | msg.add_header('Content-Disposition', 'inline', foo_bar=None) |
| 635 | self.assertEqual("inline; foo-bar", msg['Content-Disposition']) |
| 636 | |
| 637 | def test_add_header_with_no_value(self): |
| 638 | msg = Message() |
| 639 | msg.add_header('X-Status', None) |
| 640 | self.assertEqual('', msg['X-Status']) |
| 641 | |
R. David Murray | 5b2d9dd | 2011-01-09 02:35:24 +0000 | [diff] [blame] | 642 | # Issue 5871: reject an attempt to embed a header inside a header value |
| 643 | # (header injection attack). |
| 644 | def test_embeded_header_via_Header_rejected(self): |
| 645 | msg = Message() |
| 646 | msg['Dummy'] = Header('dummy\nX-Injected-Header: test') |
| 647 | self.assertRaises(errors.HeaderParseError, msg.as_string) |
| 648 | |
| 649 | def test_embeded_header_via_string_rejected(self): |
| 650 | msg = Message() |
| 651 | msg['Dummy'] = 'dummy\nX-Injected-Header: test' |
| 652 | self.assertRaises(errors.HeaderParseError, msg.as_string) |
| 653 | |
R David Murray | 7441a7a | 2012-03-14 02:59:51 -0400 | [diff] [blame] | 654 | def test_unicode_header_defaults_to_utf8_encoding(self): |
| 655 | # Issue 14291 |
| 656 | m = MIMEText('abc\n') |
| 657 | m['Subject'] = 'É test' |
| 658 | self.assertEqual(str(m),textwrap.dedent("""\ |
| 659 | Content-Type: text/plain; charset="us-ascii" |
| 660 | MIME-Version: 1.0 |
| 661 | Content-Transfer-Encoding: 7bit |
| 662 | Subject: =?utf-8?q?=C3=89_test?= |
| 663 | |
| 664 | abc |
| 665 | """)) |
| 666 | |
R David Murray | 8680bcc | 2012-03-22 22:17:51 -0400 | [diff] [blame] | 667 | def test_unicode_body_defaults_to_utf8_encoding(self): |
| 668 | # Issue 14291 |
| 669 | m = MIMEText('É testabc\n') |
| 670 | self.assertEqual(str(m),textwrap.dedent("""\ |
R David Murray | 8680bcc | 2012-03-22 22:17:51 -0400 | [diff] [blame] | 671 | Content-Type: text/plain; charset="utf-8" |
R David Murray | 42243c4 | 2012-03-22 22:40:44 -0400 | [diff] [blame] | 672 | MIME-Version: 1.0 |
R David Murray | 8680bcc | 2012-03-22 22:17:51 -0400 | [diff] [blame] | 673 | Content-Transfer-Encoding: base64 |
| 674 | |
| 675 | w4kgdGVzdGFiYwo= |
| 676 | """)) |
| 677 | |
| 678 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 679 | # Test the email.encoders module |
| 680 | class TestEncoders(unittest.TestCase): |
R David Murray | 6d94bd4 | 2011-03-16 15:52:22 -0400 | [diff] [blame] | 681 | |
| 682 | def test_EncodersEncode_base64(self): |
| 683 | with openfile('PyBanner048.gif', 'rb') as fp: |
| 684 | bindata = fp.read() |
| 685 | mimed = email.mime.image.MIMEImage(bindata) |
| 686 | base64ed = mimed.get_payload() |
| 687 | # the transfer-encoded body lines should all be <=76 characters |
| 688 | lines = base64ed.split('\n') |
| 689 | self.assertLessEqual(max([ len(x) for x in lines ]), 76) |
| 690 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 691 | def test_encode_empty_payload(self): |
| 692 | eq = self.assertEqual |
| 693 | msg = Message() |
| 694 | msg.set_charset('us-ascii') |
| 695 | eq(msg['content-transfer-encoding'], '7bit') |
| 696 | |
| 697 | def test_default_cte(self): |
| 698 | eq = self.assertEqual |
Ezio Melotti | c303c12 | 2010-04-22 11:57:12 +0000 | [diff] [blame] | 699 | # 7bit data and the default us-ascii _charset |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 700 | msg = MIMEText('hello world') |
| 701 | eq(msg['content-transfer-encoding'], '7bit') |
Ezio Melotti | c303c12 | 2010-04-22 11:57:12 +0000 | [diff] [blame] | 702 | # Similar, but with 8bit data |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 703 | msg = MIMEText('hello \xf8 world') |
R David Murray | 8680bcc | 2012-03-22 22:17:51 -0400 | [diff] [blame] | 704 | eq(msg['content-transfer-encoding'], 'base64') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 705 | # And now with a different charset |
| 706 | msg = MIMEText('hello \xf8 world', _charset='iso-8859-1') |
| 707 | eq(msg['content-transfer-encoding'], 'quoted-printable') |
| 708 | |
R. David Murray | e85200d | 2010-05-06 01:41:14 +0000 | [diff] [blame] | 709 | def test_encode7or8bit(self): |
| 710 | # Make sure a charset whose input character set is 8bit but |
| 711 | # whose output character set is 7bit gets a transfer-encoding |
| 712 | # of 7bit. |
| 713 | eq = self.assertEqual |
R. David Murray | 850fc85 | 2010-06-03 01:58:28 +0000 | [diff] [blame] | 714 | msg = MIMEText('文', _charset='euc-jp') |
R. David Murray | e85200d | 2010-05-06 01:41:14 +0000 | [diff] [blame] | 715 | eq(msg['content-transfer-encoding'], '7bit') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 716 | |
R David Murray | f581b37 | 2013-02-05 10:49:49 -0500 | [diff] [blame] | 717 | def test_qp_encode_latin1(self): |
| 718 | msg = MIMEText('\xe1\xf6\n', 'text', 'ISO-8859-1') |
| 719 | self.assertEqual(str(msg), textwrap.dedent("""\ |
| 720 | MIME-Version: 1.0 |
| 721 | Content-Type: text/text; charset="iso-8859-1" |
| 722 | Content-Transfer-Encoding: quoted-printable |
| 723 | |
| 724 | =E1=F6 |
| 725 | """)) |
| 726 | |
| 727 | def test_qp_encode_non_latin1(self): |
| 728 | # Issue 16948 |
| 729 | msg = MIMEText('\u017c\n', 'text', 'ISO-8859-2') |
| 730 | self.assertEqual(str(msg), textwrap.dedent("""\ |
| 731 | MIME-Version: 1.0 |
| 732 | Content-Type: text/text; charset="iso-8859-2" |
| 733 | Content-Transfer-Encoding: quoted-printable |
| 734 | |
| 735 | =BF |
| 736 | """)) |
| 737 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 738 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 739 | # Test long header wrapping |
| 740 | class TestLongHeaders(TestEmailBase): |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 741 | |
| 742 | maxDiff = None |
| 743 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 744 | def test_split_long_continuation(self): |
| 745 | eq = self.ndiffAssertEqual |
| 746 | msg = email.message_from_string("""\ |
| 747 | Subject: bug demonstration |
| 748 | \t12345678911234567892123456789312345678941234567895123456789612345678971234567898112345678911234567892123456789112345678911234567892123456789 |
| 749 | \tmore text |
| 750 | |
| 751 | test |
| 752 | """) |
| 753 | sfp = StringIO() |
| 754 | g = Generator(sfp) |
| 755 | g.flatten(msg) |
| 756 | eq(sfp.getvalue(), """\ |
| 757 | Subject: bug demonstration |
| 758 | \t12345678911234567892123456789312345678941234567895123456789612345678971234567898112345678911234567892123456789112345678911234567892123456789 |
| 759 | \tmore text |
| 760 | |
| 761 | test |
| 762 | """) |
| 763 | |
| 764 | def test_another_long_almost_unsplittable_header(self): |
| 765 | eq = self.ndiffAssertEqual |
| 766 | hstr = """\ |
| 767 | bug demonstration |
| 768 | \t12345678911234567892123456789312345678941234567895123456789612345678971234567898112345678911234567892123456789112345678911234567892123456789 |
| 769 | \tmore text""" |
| 770 | h = Header(hstr, continuation_ws='\t') |
| 771 | eq(h.encode(), """\ |
| 772 | bug demonstration |
| 773 | \t12345678911234567892123456789312345678941234567895123456789612345678971234567898112345678911234567892123456789112345678911234567892123456789 |
| 774 | \tmore text""") |
| 775 | h = Header(hstr.replace('\t', ' ')) |
| 776 | eq(h.encode(), """\ |
| 777 | bug demonstration |
| 778 | 12345678911234567892123456789312345678941234567895123456789612345678971234567898112345678911234567892123456789112345678911234567892123456789 |
| 779 | more text""") |
| 780 | |
| 781 | def test_long_nonstring(self): |
| 782 | eq = self.ndiffAssertEqual |
| 783 | g = Charset("iso-8859-1") |
| 784 | cz = Charset("iso-8859-2") |
| 785 | utf8 = Charset("utf-8") |
| 786 | g_head = (b'Die Mieter treten hier ein werden mit einem Foerderband ' |
| 787 | b'komfortabel den Korridor entlang, an s\xfcdl\xfcndischen ' |
| 788 | b'Wandgem\xe4lden vorbei, gegen die rotierenden Klingen ' |
| 789 | b'bef\xf6rdert. ') |
| 790 | cz_head = (b'Finan\xe8ni metropole se hroutily pod tlakem jejich ' |
| 791 | b'd\xf9vtipu.. ') |
| 792 | utf8_head = ('\u6b63\u78ba\u306b\u8a00\u3046\u3068\u7ffb\u8a33\u306f' |
| 793 | '\u3055\u308c\u3066\u3044\u307e\u305b\u3093\u3002\u4e00' |
| 794 | '\u90e8\u306f\u30c9\u30a4\u30c4\u8a9e\u3067\u3059\u304c' |
| 795 | '\u3001\u3042\u3068\u306f\u3067\u305f\u3089\u3081\u3067' |
| 796 | '\u3059\u3002\u5b9f\u969b\u306b\u306f\u300cWenn ist das ' |
| 797 | 'Nunstuck git und Slotermeyer? Ja! Beiherhund das Oder ' |
| 798 | 'die Flipperwaldt gersput.\u300d\u3068\u8a00\u3063\u3066' |
| 799 | '\u3044\u307e\u3059\u3002') |
| 800 | h = Header(g_head, g, header_name='Subject') |
| 801 | h.append(cz_head, cz) |
| 802 | h.append(utf8_head, utf8) |
| 803 | msg = Message() |
| 804 | msg['Subject'] = h |
| 805 | sfp = StringIO() |
| 806 | g = Generator(sfp) |
| 807 | g.flatten(msg) |
| 808 | eq(sfp.getvalue(), """\ |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 809 | Subject: =?iso-8859-1?q?Die_Mieter_treten_hier_ein_werden_mit_einem_Foerderb?= |
| 810 | =?iso-8859-1?q?and_komfortabel_den_Korridor_entlang=2C_an_s=FCdl=FCndischen?= |
| 811 | =?iso-8859-1?q?_Wandgem=E4lden_vorbei=2C_gegen_die_rotierenden_Klingen_bef?= |
| 812 | =?iso-8859-1?q?=F6rdert=2E_?= =?iso-8859-2?q?Finan=E8ni_metropole_se_hrouti?= |
| 813 | =?iso-8859-2?q?ly_pod_tlakem_jejich_d=F9vtipu=2E=2E_?= =?utf-8?b?5q2j56K6?= |
| 814 | =?utf-8?b?44Gr6KiA44GG44Go57+76Kiz44Gv44GV44KM44Gm44GE44G+44Gb44KT44CC5LiA?= |
| 815 | =?utf-8?b?6YOo44Gv44OJ44Kk44OE6Kqe44Gn44GZ44GM44CB44GC44Go44Gv44Gn44Gf44KJ?= |
| 816 | =?utf-8?b?44KB44Gn44GZ44CC5a6f6Zqb44Gr44Gv44CMV2VubiBpc3QgZGFzIE51bnN0dWNr?= |
| 817 | =?utf-8?b?IGdpdCB1bmQgU2xvdGVybWV5ZXI/IEphISBCZWloZXJodW5kIGRhcyBPZGVyIGRp?= |
| 818 | =?utf-8?b?ZSBGbGlwcGVyd2FsZHQgZ2Vyc3B1dC7jgI3jgajoqIDjgaPjgabjgYTjgb7jgZk=?= |
| 819 | =?utf-8?b?44CC?= |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 820 | |
| 821 | """) |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 822 | eq(h.encode(maxlinelen=76), """\ |
| 823 | =?iso-8859-1?q?Die_Mieter_treten_hier_ein_werden_mit_einem_Foerde?= |
| 824 | =?iso-8859-1?q?rband_komfortabel_den_Korridor_entlang=2C_an_s=FCdl=FCndis?= |
| 825 | =?iso-8859-1?q?chen_Wandgem=E4lden_vorbei=2C_gegen_die_rotierenden_Klinge?= |
| 826 | =?iso-8859-1?q?n_bef=F6rdert=2E_?= =?iso-8859-2?q?Finan=E8ni_metropole_se?= |
| 827 | =?iso-8859-2?q?_hroutily_pod_tlakem_jejich_d=F9vtipu=2E=2E_?= |
| 828 | =?utf-8?b?5q2j56K644Gr6KiA44GG44Go57+76Kiz44Gv44GV44KM44Gm44GE44G+44Gb?= |
| 829 | =?utf-8?b?44KT44CC5LiA6YOo44Gv44OJ44Kk44OE6Kqe44Gn44GZ44GM44CB44GC44Go?= |
| 830 | =?utf-8?b?44Gv44Gn44Gf44KJ44KB44Gn44GZ44CC5a6f6Zqb44Gr44Gv44CMV2VubiBp?= |
| 831 | =?utf-8?b?c3QgZGFzIE51bnN0dWNrIGdpdCB1bmQgU2xvdGVybWV5ZXI/IEphISBCZWlo?= |
| 832 | =?utf-8?b?ZXJodW5kIGRhcyBPZGVyIGRpZSBGbGlwcGVyd2FsZHQgZ2Vyc3B1dC7jgI0=?= |
| 833 | =?utf-8?b?44Go6KiA44Gj44Gm44GE44G+44GZ44CC?=""") |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 834 | |
| 835 | def test_long_header_encode(self): |
| 836 | eq = self.ndiffAssertEqual |
| 837 | h = Header('wasnipoop; giraffes="very-long-necked-animals"; ' |
| 838 | 'spooge="yummy"; hippos="gargantuan"; marshmallows="gooey"', |
| 839 | header_name='X-Foobar-Spoink-Defrobnit') |
| 840 | eq(h.encode(), '''\ |
| 841 | wasnipoop; giraffes="very-long-necked-animals"; |
| 842 | spooge="yummy"; hippos="gargantuan"; marshmallows="gooey"''') |
| 843 | |
| 844 | def test_long_header_encode_with_tab_continuation_is_just_a_hint(self): |
| 845 | eq = self.ndiffAssertEqual |
| 846 | h = Header('wasnipoop; giraffes="very-long-necked-animals"; ' |
| 847 | 'spooge="yummy"; hippos="gargantuan"; marshmallows="gooey"', |
| 848 | header_name='X-Foobar-Spoink-Defrobnit', |
| 849 | continuation_ws='\t') |
| 850 | eq(h.encode(), '''\ |
| 851 | wasnipoop; giraffes="very-long-necked-animals"; |
| 852 | spooge="yummy"; hippos="gargantuan"; marshmallows="gooey"''') |
| 853 | |
| 854 | def test_long_header_encode_with_tab_continuation(self): |
| 855 | eq = self.ndiffAssertEqual |
| 856 | h = Header('wasnipoop; giraffes="very-long-necked-animals";\t' |
| 857 | 'spooge="yummy"; hippos="gargantuan"; marshmallows="gooey"', |
| 858 | header_name='X-Foobar-Spoink-Defrobnit', |
| 859 | continuation_ws='\t') |
| 860 | eq(h.encode(), '''\ |
| 861 | wasnipoop; giraffes="very-long-necked-animals"; |
| 862 | \tspooge="yummy"; hippos="gargantuan"; marshmallows="gooey"''') |
| 863 | |
R David Murray | 3a6152f | 2011-03-14 21:13:03 -0400 | [diff] [blame] | 864 | def test_header_encode_with_different_output_charset(self): |
| 865 | h = Header('文', 'euc-jp') |
| 866 | self.assertEqual(h.encode(), "=?iso-2022-jp?b?GyRCSjgbKEI=?=") |
| 867 | |
| 868 | def test_long_header_encode_with_different_output_charset(self): |
| 869 | h = Header(b'test-ja \xa4\xd8\xc5\xea\xb9\xc6\xa4\xb5\xa4\xec\xa4' |
| 870 | b'\xbf\xa5\xe1\xa1\xbc\xa5\xeb\xa4\xcf\xbb\xca\xb2\xf1\xbc\xd4' |
| 871 | b'\xa4\xce\xbe\xb5\xc7\xa7\xa4\xf2\xc2\xd4\xa4\xc3\xa4\xc6\xa4' |
| 872 | b'\xa4\xa4\xde\xa4\xb9'.decode('euc-jp'), 'euc-jp') |
| 873 | res = """\ |
| 874 | =?iso-2022-jp?b?dGVzdC1qYSAbJEIkWEVqOUYkNSRsJD8lYSE8JWskTztKMnE8VCROPjUbKEI=?= |
| 875 | =?iso-2022-jp?b?GyRCRyckckJUJEMkRiQkJF4kORsoQg==?=""" |
| 876 | self.assertEqual(h.encode(), res) |
| 877 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 878 | def test_header_splitter(self): |
| 879 | eq = self.ndiffAssertEqual |
| 880 | msg = MIMEText('') |
| 881 | # It'd be great if we could use add_header() here, but that doesn't |
| 882 | # guarantee an order of the parameters. |
| 883 | msg['X-Foobar-Spoink-Defrobnit'] = ( |
| 884 | 'wasnipoop; giraffes="very-long-necked-animals"; ' |
| 885 | 'spooge="yummy"; hippos="gargantuan"; marshmallows="gooey"') |
| 886 | sfp = StringIO() |
| 887 | g = Generator(sfp) |
| 888 | g.flatten(msg) |
| 889 | eq(sfp.getvalue(), '''\ |
| 890 | Content-Type: text/plain; charset="us-ascii" |
| 891 | MIME-Version: 1.0 |
| 892 | Content-Transfer-Encoding: 7bit |
| 893 | X-Foobar-Spoink-Defrobnit: wasnipoop; giraffes="very-long-necked-animals"; |
| 894 | spooge="yummy"; hippos="gargantuan"; marshmallows="gooey" |
| 895 | |
| 896 | ''') |
| 897 | |
| 898 | def test_no_semis_header_splitter(self): |
| 899 | eq = self.ndiffAssertEqual |
| 900 | msg = Message() |
| 901 | msg['From'] = 'test@dom.ain' |
| 902 | msg['References'] = SPACE.join('<%d@dom.ain>' % i for i in range(10)) |
| 903 | msg.set_payload('Test') |
| 904 | sfp = StringIO() |
| 905 | g = Generator(sfp) |
| 906 | g.flatten(msg) |
| 907 | eq(sfp.getvalue(), """\ |
| 908 | From: test@dom.ain |
| 909 | References: <0@dom.ain> <1@dom.ain> <2@dom.ain> <3@dom.ain> <4@dom.ain> |
| 910 | <5@dom.ain> <6@dom.ain> <7@dom.ain> <8@dom.ain> <9@dom.ain> |
| 911 | |
| 912 | Test""") |
| 913 | |
R David Murray | 7da4db1 | 2011-04-07 20:37:17 -0400 | [diff] [blame] | 914 | def test_last_split_chunk_does_not_fit(self): |
| 915 | eq = self.ndiffAssertEqual |
| 916 | h = Header('Subject: the first part of this is short, but_the_second' |
| 917 | '_part_does_not_fit_within_maxlinelen_and_thus_should_be_on_a_line' |
| 918 | '_all_by_itself') |
| 919 | eq(h.encode(), """\ |
| 920 | Subject: the first part of this is short, |
| 921 | but_the_second_part_does_not_fit_within_maxlinelen_and_thus_should_be_on_a_line_all_by_itself""") |
| 922 | |
| 923 | def test_splittable_leading_char_followed_by_overlong_unsplitable(self): |
| 924 | eq = self.ndiffAssertEqual |
| 925 | h = Header(', but_the_second' |
| 926 | '_part_does_not_fit_within_maxlinelen_and_thus_should_be_on_a_line' |
| 927 | '_all_by_itself') |
| 928 | eq(h.encode(), """\ |
| 929 | , |
| 930 | but_the_second_part_does_not_fit_within_maxlinelen_and_thus_should_be_on_a_line_all_by_itself""") |
| 931 | |
| 932 | def test_multiple_splittable_leading_char_followed_by_overlong_unsplitable(self): |
| 933 | eq = self.ndiffAssertEqual |
| 934 | h = Header(', , but_the_second' |
| 935 | '_part_does_not_fit_within_maxlinelen_and_thus_should_be_on_a_line' |
| 936 | '_all_by_itself') |
| 937 | eq(h.encode(), """\ |
| 938 | , , |
| 939 | but_the_second_part_does_not_fit_within_maxlinelen_and_thus_should_be_on_a_line_all_by_itself""") |
| 940 | |
| 941 | def test_trailing_splitable_on_overlong_unsplitable(self): |
| 942 | eq = self.ndiffAssertEqual |
| 943 | h = Header('this_part_does_not_fit_within_maxlinelen_and_thus_should_' |
| 944 | 'be_on_a_line_all_by_itself;') |
| 945 | eq(h.encode(), "this_part_does_not_fit_within_maxlinelen_and_thus_should_" |
| 946 | "be_on_a_line_all_by_itself;") |
| 947 | |
| 948 | def test_trailing_splitable_on_overlong_unsplitable_with_leading_splitable(self): |
| 949 | eq = self.ndiffAssertEqual |
| 950 | h = Header('; ' |
| 951 | 'this_part_does_not_fit_within_maxlinelen_and_thus_should_' |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 952 | 'be_on_a_line_all_by_itself; ') |
R David Murray | 7da4db1 | 2011-04-07 20:37:17 -0400 | [diff] [blame] | 953 | eq(h.encode(), """\ |
| 954 | ; |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 955 | this_part_does_not_fit_within_maxlinelen_and_thus_should_be_on_a_line_all_by_itself; """) |
R David Murray | 7da4db1 | 2011-04-07 20:37:17 -0400 | [diff] [blame] | 956 | |
R David Murray | e1292a2 | 2011-04-07 20:54:03 -0400 | [diff] [blame] | 957 | def test_long_header_with_multiple_sequential_split_chars(self): |
R David Murray | e1292a2 | 2011-04-07 20:54:03 -0400 | [diff] [blame] | 958 | eq = self.ndiffAssertEqual |
| 959 | h = Header('This is a long line that has two whitespaces in a row. ' |
| 960 | 'This used to cause truncation of the header when folded') |
| 961 | eq(h.encode(), """\ |
| 962 | This is a long line that has two whitespaces in a row. This used to cause |
| 963 | truncation of the header when folded""") |
| 964 | |
Ezio Melotti | 1c4810b | 2013-08-10 18:57:12 +0300 | [diff] [blame] | 965 | def test_splitter_split_on_punctuation_only_if_fws_with_header(self): |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 966 | eq = self.ndiffAssertEqual |
| 967 | h = Header('thisverylongheaderhas;semicolons;and,commas,but' |
| 968 | 'they;arenotlegal;fold,points') |
| 969 | eq(h.encode(), "thisverylongheaderhas;semicolons;and,commas,butthey;" |
| 970 | "arenotlegal;fold,points") |
| 971 | |
| 972 | def test_leading_splittable_in_the_middle_just_before_overlong_last_part(self): |
| 973 | eq = self.ndiffAssertEqual |
| 974 | h = Header('this is a test where we need to have more than one line ' |
| 975 | 'before; our final line that is just too big to fit;; ' |
| 976 | 'this_part_does_not_fit_within_maxlinelen_and_thus_should_' |
| 977 | 'be_on_a_line_all_by_itself;') |
| 978 | eq(h.encode(), """\ |
| 979 | this is a test where we need to have more than one line before; |
| 980 | our final line that is just too big to fit;; |
| 981 | this_part_does_not_fit_within_maxlinelen_and_thus_should_be_on_a_line_all_by_itself;""") |
| 982 | |
| 983 | def test_overlong_last_part_followed_by_split_point(self): |
| 984 | eq = self.ndiffAssertEqual |
| 985 | h = Header('this_part_does_not_fit_within_maxlinelen_and_thus_should_' |
| 986 | 'be_on_a_line_all_by_itself ') |
| 987 | eq(h.encode(), "this_part_does_not_fit_within_maxlinelen_and_thus_" |
| 988 | "should_be_on_a_line_all_by_itself ") |
| 989 | |
| 990 | def test_multiline_with_overlong_parts_separated_by_two_split_points(self): |
| 991 | eq = self.ndiffAssertEqual |
| 992 | h = Header('this_is_a__test_where_we_need_to_have_more_than_one_line_' |
| 993 | 'before_our_final_line_; ; ' |
| 994 | 'this_part_does_not_fit_within_maxlinelen_and_thus_should_' |
| 995 | 'be_on_a_line_all_by_itself; ') |
| 996 | eq(h.encode(), """\ |
| 997 | this_is_a__test_where_we_need_to_have_more_than_one_line_before_our_final_line_; |
| 998 | ; |
| 999 | this_part_does_not_fit_within_maxlinelen_and_thus_should_be_on_a_line_all_by_itself; """) |
| 1000 | |
| 1001 | def test_multiline_with_overlong_last_part_followed_by_split_point(self): |
| 1002 | eq = self.ndiffAssertEqual |
| 1003 | h = Header('this is a test where we need to have more than one line ' |
| 1004 | 'before our final line; ; ' |
| 1005 | 'this_part_does_not_fit_within_maxlinelen_and_thus_should_' |
| 1006 | 'be_on_a_line_all_by_itself; ') |
| 1007 | eq(h.encode(), """\ |
| 1008 | this is a test where we need to have more than one line before our final line; |
| 1009 | ; |
| 1010 | this_part_does_not_fit_within_maxlinelen_and_thus_should_be_on_a_line_all_by_itself; """) |
| 1011 | |
| 1012 | def test_long_header_with_whitespace_runs(self): |
| 1013 | eq = self.ndiffAssertEqual |
| 1014 | msg = Message() |
| 1015 | msg['From'] = 'test@dom.ain' |
| 1016 | msg['References'] = SPACE.join(['<foo@dom.ain> '] * 10) |
| 1017 | msg.set_payload('Test') |
| 1018 | sfp = StringIO() |
| 1019 | g = Generator(sfp) |
| 1020 | g.flatten(msg) |
| 1021 | eq(sfp.getvalue(), """\ |
| 1022 | From: test@dom.ain |
| 1023 | References: <foo@dom.ain> <foo@dom.ain> <foo@dom.ain> <foo@dom.ain> |
| 1024 | <foo@dom.ain> <foo@dom.ain> <foo@dom.ain> <foo@dom.ain> |
| 1025 | <foo@dom.ain> <foo@dom.ain>\x20\x20 |
| 1026 | |
| 1027 | Test""") |
| 1028 | |
| 1029 | def test_long_run_with_semi_header_splitter(self): |
| 1030 | eq = self.ndiffAssertEqual |
| 1031 | msg = Message() |
| 1032 | msg['From'] = 'test@dom.ain' |
| 1033 | msg['References'] = SPACE.join(['<foo@dom.ain>'] * 10) + '; abc' |
| 1034 | msg.set_payload('Test') |
| 1035 | sfp = StringIO() |
| 1036 | g = Generator(sfp) |
| 1037 | g.flatten(msg) |
| 1038 | eq(sfp.getvalue(), """\ |
| 1039 | From: test@dom.ain |
| 1040 | References: <foo@dom.ain> <foo@dom.ain> <foo@dom.ain> <foo@dom.ain> |
| 1041 | <foo@dom.ain> <foo@dom.ain> <foo@dom.ain> <foo@dom.ain> <foo@dom.ain> |
| 1042 | <foo@dom.ain>; abc |
| 1043 | |
| 1044 | Test""") |
| 1045 | |
| 1046 | def test_splitter_split_on_punctuation_only_if_fws(self): |
| 1047 | eq = self.ndiffAssertEqual |
| 1048 | msg = Message() |
| 1049 | msg['From'] = 'test@dom.ain' |
| 1050 | msg['References'] = ('thisverylongheaderhas;semicolons;and,commas,but' |
| 1051 | 'they;arenotlegal;fold,points') |
| 1052 | msg.set_payload('Test') |
| 1053 | sfp = StringIO() |
| 1054 | g = Generator(sfp) |
| 1055 | g.flatten(msg) |
| 1056 | # XXX the space after the header should not be there. |
| 1057 | eq(sfp.getvalue(), """\ |
| 1058 | From: test@dom.ain |
| 1059 | References:\x20 |
| 1060 | thisverylongheaderhas;semicolons;and,commas,butthey;arenotlegal;fold,points |
| 1061 | |
| 1062 | Test""") |
| 1063 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1064 | def test_no_split_long_header(self): |
| 1065 | eq = self.ndiffAssertEqual |
| 1066 | hstr = 'References: ' + 'x' * 80 |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 1067 | h = Header(hstr) |
| 1068 | # These come on two lines because Headers are really field value |
| 1069 | # classes and don't really know about their field names. |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1070 | eq(h.encode(), """\ |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 1071 | References: |
| 1072 | xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx""") |
| 1073 | h = Header('x' * 80) |
| 1074 | eq(h.encode(), 'xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1075 | |
| 1076 | def test_splitting_multiple_long_lines(self): |
| 1077 | eq = self.ndiffAssertEqual |
| 1078 | hstr = """\ |
| 1079 | from babylon.socal-raves.org (localhost [127.0.0.1]); by babylon.socal-raves.org (Postfix) with ESMTP id B570E51B81; for <mailman-admin@babylon.socal-raves.org>; Sat, 2 Feb 2002 17:00:06 -0800 (PST) |
| 1080 | \tfrom babylon.socal-raves.org (localhost [127.0.0.1]); by babylon.socal-raves.org (Postfix) with ESMTP id B570E51B81; for <mailman-admin@babylon.socal-raves.org>; Sat, 2 Feb 2002 17:00:06 -0800 (PST) |
| 1081 | \tfrom babylon.socal-raves.org (localhost [127.0.0.1]); by babylon.socal-raves.org (Postfix) with ESMTP id B570E51B81; for <mailman-admin@babylon.socal-raves.org>; Sat, 2 Feb 2002 17:00:06 -0800 (PST) |
| 1082 | """ |
| 1083 | h = Header(hstr, continuation_ws='\t') |
| 1084 | eq(h.encode(), """\ |
| 1085 | from babylon.socal-raves.org (localhost [127.0.0.1]); |
| 1086 | by babylon.socal-raves.org (Postfix) with ESMTP id B570E51B81; |
| 1087 | for <mailman-admin@babylon.socal-raves.org>; |
| 1088 | Sat, 2 Feb 2002 17:00:06 -0800 (PST) |
| 1089 | \tfrom babylon.socal-raves.org (localhost [127.0.0.1]); |
| 1090 | by babylon.socal-raves.org (Postfix) with ESMTP id B570E51B81; |
| 1091 | for <mailman-admin@babylon.socal-raves.org>; |
| 1092 | Sat, 2 Feb 2002 17:00:06 -0800 (PST) |
| 1093 | \tfrom babylon.socal-raves.org (localhost [127.0.0.1]); |
| 1094 | by babylon.socal-raves.org (Postfix) with ESMTP id B570E51B81; |
| 1095 | for <mailman-admin@babylon.socal-raves.org>; |
| 1096 | Sat, 2 Feb 2002 17:00:06 -0800 (PST)""") |
| 1097 | |
| 1098 | def test_splitting_first_line_only_is_long(self): |
| 1099 | eq = self.ndiffAssertEqual |
| 1100 | hstr = """\ |
| 1101 | from modemcable093.139-201-24.que.mc.videotron.ca ([24.201.139.93] helo=cthulhu.gerg.ca) |
| 1102 | \tby kronos.mems-exchange.org with esmtp (Exim 4.05) |
| 1103 | \tid 17k4h5-00034i-00 |
| 1104 | \tfor test@mems-exchange.org; Wed, 28 Aug 2002 11:25:20 -0400""" |
| 1105 | h = Header(hstr, maxlinelen=78, header_name='Received', |
| 1106 | continuation_ws='\t') |
| 1107 | eq(h.encode(), """\ |
| 1108 | from modemcable093.139-201-24.que.mc.videotron.ca ([24.201.139.93] |
| 1109 | helo=cthulhu.gerg.ca) |
| 1110 | \tby kronos.mems-exchange.org with esmtp (Exim 4.05) |
| 1111 | \tid 17k4h5-00034i-00 |
| 1112 | \tfor test@mems-exchange.org; Wed, 28 Aug 2002 11:25:20 -0400""") |
| 1113 | |
| 1114 | def test_long_8bit_header(self): |
| 1115 | eq = self.ndiffAssertEqual |
| 1116 | msg = Message() |
| 1117 | h = Header('Britische Regierung gibt', 'iso-8859-1', |
| 1118 | header_name='Subject') |
| 1119 | h.append('gr\xfcnes Licht f\xfcr Offshore-Windkraftprojekte') |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 1120 | eq(h.encode(maxlinelen=76), """\ |
| 1121 | =?iso-8859-1?q?Britische_Regierung_gibt_gr=FCnes_Licht_f=FCr_Offs?= |
| 1122 | =?iso-8859-1?q?hore-Windkraftprojekte?=""") |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1123 | msg['Subject'] = h |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 1124 | eq(msg.as_string(maxheaderlen=76), """\ |
| 1125 | Subject: =?iso-8859-1?q?Britische_Regierung_gibt_gr=FCnes_Licht_f=FCr_Offs?= |
| 1126 | =?iso-8859-1?q?hore-Windkraftprojekte?= |
| 1127 | |
| 1128 | """) |
| 1129 | eq(msg.as_string(maxheaderlen=0), """\ |
| 1130 | Subject: =?iso-8859-1?q?Britische_Regierung_gibt_gr=FCnes_Licht_f=FCr_Offshore-Windkraftprojekte?= |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1131 | |
| 1132 | """) |
| 1133 | |
| 1134 | def test_long_8bit_header_no_charset(self): |
| 1135 | eq = self.ndiffAssertEqual |
| 1136 | msg = Message() |
Barry Warsaw | 8c57104 | 2007-08-30 19:17:18 +0000 | [diff] [blame] | 1137 | header_string = ('Britische Regierung gibt gr\xfcnes Licht ' |
| 1138 | 'f\xfcr Offshore-Windkraftprojekte ' |
| 1139 | '<a-very-long-address@example.com>') |
| 1140 | msg['Reply-To'] = header_string |
R David Murray | 7441a7a | 2012-03-14 02:59:51 -0400 | [diff] [blame] | 1141 | eq(msg.as_string(maxheaderlen=78), """\ |
| 1142 | Reply-To: =?utf-8?q?Britische_Regierung_gibt_gr=C3=BCnes_Licht_f=C3=BCr_Offs?= |
| 1143 | =?utf-8?q?hore-Windkraftprojekte_=3Ca-very-long-address=40example=2Ecom=3E?= |
| 1144 | |
| 1145 | """) |
Barry Warsaw | 8c57104 | 2007-08-30 19:17:18 +0000 | [diff] [blame] | 1146 | msg = Message() |
R David Murray | 7441a7a | 2012-03-14 02:59:51 -0400 | [diff] [blame] | 1147 | msg['Reply-To'] = Header(header_string, |
Barry Warsaw | 8c57104 | 2007-08-30 19:17:18 +0000 | [diff] [blame] | 1148 | header_name='Reply-To') |
| 1149 | eq(msg.as_string(maxheaderlen=78), """\ |
| 1150 | Reply-To: =?utf-8?q?Britische_Regierung_gibt_gr=C3=BCnes_Licht_f=C3=BCr_Offs?= |
| 1151 | =?utf-8?q?hore-Windkraftprojekte_=3Ca-very-long-address=40example=2Ecom=3E?= |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1152 | |
| 1153 | """) |
| 1154 | |
| 1155 | def test_long_to_header(self): |
| 1156 | eq = self.ndiffAssertEqual |
| 1157 | to = ('"Someone Test #A" <someone@eecs.umich.edu>,' |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 1158 | '<someone@eecs.umich.edu>, ' |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1159 | '"Someone Test #B" <someone@umich.edu>, ' |
| 1160 | '"Someone Test #C" <someone@eecs.umich.edu>, ' |
| 1161 | '"Someone Test #D" <someone@eecs.umich.edu>') |
| 1162 | msg = Message() |
| 1163 | msg['To'] = to |
| 1164 | eq(msg.as_string(maxheaderlen=78), '''\ |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 1165 | To: "Someone Test #A" <someone@eecs.umich.edu>,<someone@eecs.umich.edu>, |
Barry Warsaw | 70d61ce | 2009-03-30 23:12:30 +0000 | [diff] [blame] | 1166 | "Someone Test #B" <someone@umich.edu>, |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 1167 | "Someone Test #C" <someone@eecs.umich.edu>, |
| 1168 | "Someone Test #D" <someone@eecs.umich.edu> |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1169 | |
| 1170 | ''') |
| 1171 | |
| 1172 | def test_long_line_after_append(self): |
| 1173 | eq = self.ndiffAssertEqual |
| 1174 | s = 'This is an example of string which has almost the limit of header length.' |
| 1175 | h = Header(s) |
| 1176 | h.append('Add another line.') |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 1177 | eq(h.encode(maxlinelen=76), """\ |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1178 | This is an example of string which has almost the limit of header length. |
| 1179 | Add another line.""") |
| 1180 | |
| 1181 | def test_shorter_line_with_append(self): |
| 1182 | eq = self.ndiffAssertEqual |
| 1183 | s = 'This is a shorter line.' |
| 1184 | h = Header(s) |
| 1185 | h.append('Add another sentence. (Surprise?)') |
| 1186 | eq(h.encode(), |
| 1187 | 'This is a shorter line. Add another sentence. (Surprise?)') |
| 1188 | |
| 1189 | def test_long_field_name(self): |
| 1190 | eq = self.ndiffAssertEqual |
| 1191 | fn = 'X-Very-Very-Very-Long-Header-Name' |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 1192 | gs = ('Die Mieter treten hier ein werden mit einem Foerderband ' |
| 1193 | 'komfortabel den Korridor entlang, an s\xfcdl\xfcndischen ' |
| 1194 | 'Wandgem\xe4lden vorbei, gegen die rotierenden Klingen ' |
| 1195 | 'bef\xf6rdert. ') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1196 | h = Header(gs, 'iso-8859-1', header_name=fn) |
| 1197 | # BAW: this seems broken because the first line is too long |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 1198 | eq(h.encode(maxlinelen=76), """\ |
| 1199 | =?iso-8859-1?q?Die_Mieter_treten_hier_e?= |
| 1200 | =?iso-8859-1?q?in_werden_mit_einem_Foerderband_komfortabel_den_Korridor_e?= |
| 1201 | =?iso-8859-1?q?ntlang=2C_an_s=FCdl=FCndischen_Wandgem=E4lden_vorbei=2C_ge?= |
| 1202 | =?iso-8859-1?q?gen_die_rotierenden_Klingen_bef=F6rdert=2E_?=""") |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1203 | |
| 1204 | def test_long_received_header(self): |
| 1205 | h = ('from FOO.TLD (vizworld.acl.foo.tld [123.452.678.9]) ' |
| 1206 | 'by hrothgar.la.mastaler.com (tmda-ofmipd) with ESMTP; ' |
| 1207 | 'Wed, 05 Mar 2003 18:10:18 -0700') |
| 1208 | msg = Message() |
| 1209 | msg['Received-1'] = Header(h, continuation_ws='\t') |
| 1210 | msg['Received-2'] = h |
Barry Warsaw | bef9d21 | 2007-08-31 10:55:37 +0000 | [diff] [blame] | 1211 | # This should be splitting on spaces not semicolons. |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1212 | self.ndiffAssertEqual(msg.as_string(maxheaderlen=78), """\ |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 1213 | Received-1: from FOO.TLD (vizworld.acl.foo.tld [123.452.678.9]) by |
| 1214 | hrothgar.la.mastaler.com (tmda-ofmipd) with ESMTP; |
Barry Warsaw | bef9d21 | 2007-08-31 10:55:37 +0000 | [diff] [blame] | 1215 | Wed, 05 Mar 2003 18:10:18 -0700 |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 1216 | Received-2: from FOO.TLD (vizworld.acl.foo.tld [123.452.678.9]) by |
| 1217 | hrothgar.la.mastaler.com (tmda-ofmipd) with ESMTP; |
Barry Warsaw | bef9d21 | 2007-08-31 10:55:37 +0000 | [diff] [blame] | 1218 | Wed, 05 Mar 2003 18:10:18 -0700 |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1219 | |
| 1220 | """) |
| 1221 | |
| 1222 | def test_string_headerinst_eq(self): |
| 1223 | h = ('<15975.17901.207240.414604@sgigritzmann1.mathematik.' |
| 1224 | 'tu-muenchen.de> (David Bremner\'s message of ' |
| 1225 | '"Thu, 6 Mar 2003 13:58:21 +0100")') |
| 1226 | msg = Message() |
| 1227 | msg['Received-1'] = Header(h, header_name='Received-1', |
| 1228 | continuation_ws='\t') |
| 1229 | msg['Received-2'] = h |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 1230 | # XXX The space after the ':' should not be there. |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1231 | self.ndiffAssertEqual(msg.as_string(maxheaderlen=78), """\ |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 1232 | Received-1:\x20 |
| 1233 | <15975.17901.207240.414604@sgigritzmann1.mathematik.tu-muenchen.de> (David |
| 1234 | Bremner's message of \"Thu, 6 Mar 2003 13:58:21 +0100\") |
| 1235 | Received-2:\x20 |
| 1236 | <15975.17901.207240.414604@sgigritzmann1.mathematik.tu-muenchen.de> (David |
| 1237 | Bremner's message of \"Thu, 6 Mar 2003 13:58:21 +0100\") |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1238 | |
| 1239 | """) |
| 1240 | |
| 1241 | def test_long_unbreakable_lines_with_continuation(self): |
| 1242 | eq = self.ndiffAssertEqual |
| 1243 | msg = Message() |
| 1244 | t = """\ |
| 1245 | iVBORw0KGgoAAAANSUhEUgAAADAAAAAwBAMAAAClLOS0AAAAGFBMVEUAAAAkHiJeRUIcGBi9 |
| 1246 | locQDQ4zJykFBAXJfWDjAAACYUlEQVR4nF2TQY/jIAyFc6lydlG5x8Nyp1Y69wj1PN2I5gzp""" |
| 1247 | msg['Face-1'] = t |
| 1248 | msg['Face-2'] = Header(t, header_name='Face-2') |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 1249 | msg['Face-3'] = ' ' + t |
Barry Warsaw | bef9d21 | 2007-08-31 10:55:37 +0000 | [diff] [blame] | 1250 | # XXX This splitting is all wrong. It the first value line should be |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 1251 | # snug against the field name or the space after the header not there. |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1252 | eq(msg.as_string(maxheaderlen=78), """\ |
Barry Warsaw | c5a6a30 | 2007-08-31 11:19:21 +0000 | [diff] [blame] | 1253 | Face-1:\x20 |
Barry Warsaw | 70d61ce | 2009-03-30 23:12:30 +0000 | [diff] [blame] | 1254 | iVBORw0KGgoAAAANSUhEUgAAADAAAAAwBAMAAAClLOS0AAAAGFBMVEUAAAAkHiJeRUIcGBi9 |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1255 | locQDQ4zJykFBAXJfWDjAAACYUlEQVR4nF2TQY/jIAyFc6lydlG5x8Nyp1Y69wj1PN2I5gzp |
Barry Warsaw | c5a6a30 | 2007-08-31 11:19:21 +0000 | [diff] [blame] | 1256 | Face-2:\x20 |
Barry Warsaw | bef9d21 | 2007-08-31 10:55:37 +0000 | [diff] [blame] | 1257 | iVBORw0KGgoAAAANSUhEUgAAADAAAAAwBAMAAAClLOS0AAAAGFBMVEUAAAAkHiJeRUIcGBi9 |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1258 | locQDQ4zJykFBAXJfWDjAAACYUlEQVR4nF2TQY/jIAyFc6lydlG5x8Nyp1Y69wj1PN2I5gzp |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 1259 | Face-3:\x20 |
| 1260 | iVBORw0KGgoAAAANSUhEUgAAADAAAAAwBAMAAAClLOS0AAAAGFBMVEUAAAAkHiJeRUIcGBi9 |
| 1261 | locQDQ4zJykFBAXJfWDjAAACYUlEQVR4nF2TQY/jIAyFc6lydlG5x8Nyp1Y69wj1PN2I5gzp |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1262 | |
| 1263 | """) |
| 1264 | |
| 1265 | def test_another_long_multiline_header(self): |
| 1266 | eq = self.ndiffAssertEqual |
| 1267 | m = ('Received: from siimage.com ' |
| 1268 | '([172.25.1.3]) by zima.siliconimage.com with ' |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 1269 | 'Microsoft SMTPSVC(5.0.2195.4905); ' |
| 1270 | 'Wed, 16 Oct 2002 07:41:11 -0700') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1271 | msg = email.message_from_string(m) |
| 1272 | eq(msg.as_string(maxheaderlen=78), '''\ |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 1273 | Received: from siimage.com ([172.25.1.3]) by zima.siliconimage.com with |
| 1274 | Microsoft SMTPSVC(5.0.2195.4905); Wed, 16 Oct 2002 07:41:11 -0700 |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1275 | |
| 1276 | ''') |
| 1277 | |
| 1278 | def test_long_lines_with_different_header(self): |
| 1279 | eq = self.ndiffAssertEqual |
| 1280 | h = ('List-Unsubscribe: ' |
| 1281 | '<http://lists.sourceforge.net/lists/listinfo/spamassassin-talk>,' |
| 1282 | ' <mailto:spamassassin-talk-request@lists.sourceforge.net' |
| 1283 | '?subject=unsubscribe>') |
| 1284 | msg = Message() |
| 1285 | msg['List'] = h |
| 1286 | msg['List'] = Header(h, header_name='List') |
| 1287 | eq(msg.as_string(maxheaderlen=78), """\ |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 1288 | List: List-Unsubscribe: |
| 1289 | <http://lists.sourceforge.net/lists/listinfo/spamassassin-talk>, |
Barry Warsaw | bef9d21 | 2007-08-31 10:55:37 +0000 | [diff] [blame] | 1290 | <mailto:spamassassin-talk-request@lists.sourceforge.net?subject=unsubscribe> |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 1291 | List: List-Unsubscribe: |
| 1292 | <http://lists.sourceforge.net/lists/listinfo/spamassassin-talk>, |
Barry Warsaw | bef9d21 | 2007-08-31 10:55:37 +0000 | [diff] [blame] | 1293 | <mailto:spamassassin-talk-request@lists.sourceforge.net?subject=unsubscribe> |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1294 | |
| 1295 | """) |
| 1296 | |
R. David Murray | 6f0022d | 2011-01-07 21:57:25 +0000 | [diff] [blame] | 1297 | def test_long_rfc2047_header_with_embedded_fws(self): |
| 1298 | h = Header(textwrap.dedent("""\ |
| 1299 | We're going to pretend this header is in a non-ascii character set |
| 1300 | \tto see if line wrapping with encoded words and embedded |
| 1301 | folding white space works"""), |
| 1302 | charset='utf-8', |
| 1303 | header_name='Test') |
| 1304 | self.assertEqual(h.encode()+'\n', textwrap.dedent("""\ |
| 1305 | =?utf-8?q?We=27re_going_to_pretend_this_header_is_in_a_non-ascii_chara?= |
| 1306 | =?utf-8?q?cter_set?= |
| 1307 | =?utf-8?q?_to_see_if_line_wrapping_with_encoded_words_and_embedded?= |
| 1308 | =?utf-8?q?_folding_white_space_works?=""")+'\n') |
| 1309 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1310 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 1311 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1312 | # Test mangling of "From " lines in the body of a message |
| 1313 | class TestFromMangling(unittest.TestCase): |
| 1314 | def setUp(self): |
| 1315 | self.msg = Message() |
| 1316 | self.msg['From'] = 'aaa@bbb.org' |
| 1317 | self.msg.set_payload("""\ |
| 1318 | From the desk of A.A.A.: |
| 1319 | Blah blah blah |
| 1320 | """) |
| 1321 | |
| 1322 | def test_mangled_from(self): |
| 1323 | s = StringIO() |
| 1324 | g = Generator(s, mangle_from_=True) |
| 1325 | g.flatten(self.msg) |
| 1326 | self.assertEqual(s.getvalue(), """\ |
| 1327 | From: aaa@bbb.org |
| 1328 | |
| 1329 | >From the desk of A.A.A.: |
| 1330 | Blah blah blah |
| 1331 | """) |
| 1332 | |
| 1333 | def test_dont_mangle_from(self): |
| 1334 | s = StringIO() |
| 1335 | g = Generator(s, mangle_from_=False) |
| 1336 | g.flatten(self.msg) |
| 1337 | self.assertEqual(s.getvalue(), """\ |
| 1338 | From: aaa@bbb.org |
| 1339 | |
| 1340 | From the desk of A.A.A.: |
| 1341 | Blah blah blah |
| 1342 | """) |
| 1343 | |
R David Murray | 6a31bc6 | 2012-07-22 21:47:53 -0400 | [diff] [blame] | 1344 | def test_mangle_from_in_preamble_and_epilog(self): |
| 1345 | s = StringIO() |
| 1346 | g = Generator(s, mangle_from_=True) |
| 1347 | msg = email.message_from_string(textwrap.dedent("""\ |
| 1348 | From: foo@bar.com |
| 1349 | Mime-Version: 1.0 |
| 1350 | Content-Type: multipart/mixed; boundary=XXX |
| 1351 | |
| 1352 | From somewhere unknown |
| 1353 | |
| 1354 | --XXX |
| 1355 | Content-Type: text/plain |
| 1356 | |
| 1357 | foo |
| 1358 | |
| 1359 | --XXX-- |
| 1360 | |
| 1361 | From somewhere unknowable |
| 1362 | """)) |
| 1363 | g.flatten(msg) |
| 1364 | self.assertEqual(len([1 for x in s.getvalue().split('\n') |
| 1365 | if x.startswith('>From ')]), 2) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1366 | |
R David Murray | 638d40b | 2012-08-24 11:14:13 -0400 | [diff] [blame] | 1367 | def test_mangled_from_with_bad_bytes(self): |
| 1368 | source = textwrap.dedent("""\ |
| 1369 | Content-Type: text/plain; charset="utf-8" |
| 1370 | MIME-Version: 1.0 |
| 1371 | Content-Transfer-Encoding: 8bit |
| 1372 | From: aaa@bbb.org |
| 1373 | |
| 1374 | """).encode('utf-8') |
| 1375 | msg = email.message_from_bytes(source + b'From R\xc3\xb6lli\n') |
| 1376 | b = BytesIO() |
| 1377 | g = BytesGenerator(b, mangle_from_=True) |
| 1378 | g.flatten(msg) |
| 1379 | self.assertEqual(b.getvalue(), source + b'>From R\xc3\xb6lli\n') |
| 1380 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 1381 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1382 | # Test the basic MIMEAudio class |
| 1383 | class TestMIMEAudio(unittest.TestCase): |
| 1384 | def setUp(self): |
R David Murray | 28346b8 | 2011-03-31 11:40:20 -0400 | [diff] [blame] | 1385 | with openfile('audiotest.au', 'rb') as fp: |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1386 | self._audiodata = fp.read() |
| 1387 | self._au = MIMEAudio(self._audiodata) |
| 1388 | |
| 1389 | def test_guess_minor_type(self): |
| 1390 | self.assertEqual(self._au.get_content_type(), 'audio/basic') |
| 1391 | |
| 1392 | def test_encoding(self): |
| 1393 | payload = self._au.get_payload() |
R. David Murray | 7da8f06 | 2010-06-04 16:11:08 +0000 | [diff] [blame] | 1394 | self.assertEqual(base64.decodebytes(bytes(payload, 'ascii')), |
| 1395 | self._audiodata) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1396 | |
| 1397 | def test_checkSetMinor(self): |
| 1398 | au = MIMEAudio(self._audiodata, 'fish') |
| 1399 | self.assertEqual(au.get_content_type(), 'audio/fish') |
| 1400 | |
| 1401 | def test_add_header(self): |
| 1402 | eq = self.assertEqual |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 1403 | unless = self.assertTrue |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1404 | self._au.add_header('Content-Disposition', 'attachment', |
| 1405 | filename='audiotest.au') |
| 1406 | eq(self._au['content-disposition'], |
| 1407 | 'attachment; filename="audiotest.au"') |
| 1408 | eq(self._au.get_params(header='content-disposition'), |
| 1409 | [('attachment', ''), ('filename', 'audiotest.au')]) |
| 1410 | eq(self._au.get_param('filename', header='content-disposition'), |
| 1411 | 'audiotest.au') |
| 1412 | missing = [] |
| 1413 | eq(self._au.get_param('attachment', header='content-disposition'), '') |
| 1414 | unless(self._au.get_param('foo', failobj=missing, |
| 1415 | header='content-disposition') is missing) |
| 1416 | # Try some missing stuff |
| 1417 | unless(self._au.get_param('foobar', missing) is missing) |
| 1418 | unless(self._au.get_param('attachment', missing, |
| 1419 | header='foobar') is missing) |
| 1420 | |
| 1421 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 1422 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1423 | # Test the basic MIMEImage class |
| 1424 | class TestMIMEImage(unittest.TestCase): |
| 1425 | def setUp(self): |
| 1426 | with openfile('PyBanner048.gif', 'rb') as fp: |
| 1427 | self._imgdata = fp.read() |
| 1428 | self._im = MIMEImage(self._imgdata) |
| 1429 | |
| 1430 | def test_guess_minor_type(self): |
| 1431 | self.assertEqual(self._im.get_content_type(), 'image/gif') |
| 1432 | |
| 1433 | def test_encoding(self): |
| 1434 | payload = self._im.get_payload() |
R. David Murray | 7da8f06 | 2010-06-04 16:11:08 +0000 | [diff] [blame] | 1435 | self.assertEqual(base64.decodebytes(bytes(payload, 'ascii')), |
| 1436 | self._imgdata) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1437 | |
| 1438 | def test_checkSetMinor(self): |
| 1439 | im = MIMEImage(self._imgdata, 'fish') |
| 1440 | self.assertEqual(im.get_content_type(), 'image/fish') |
| 1441 | |
| 1442 | def test_add_header(self): |
| 1443 | eq = self.assertEqual |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 1444 | unless = self.assertTrue |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1445 | self._im.add_header('Content-Disposition', 'attachment', |
| 1446 | filename='dingusfish.gif') |
| 1447 | eq(self._im['content-disposition'], |
| 1448 | 'attachment; filename="dingusfish.gif"') |
| 1449 | eq(self._im.get_params(header='content-disposition'), |
| 1450 | [('attachment', ''), ('filename', 'dingusfish.gif')]) |
| 1451 | eq(self._im.get_param('filename', header='content-disposition'), |
| 1452 | 'dingusfish.gif') |
| 1453 | missing = [] |
| 1454 | eq(self._im.get_param('attachment', header='content-disposition'), '') |
| 1455 | unless(self._im.get_param('foo', failobj=missing, |
| 1456 | header='content-disposition') is missing) |
| 1457 | # Try some missing stuff |
| 1458 | unless(self._im.get_param('foobar', missing) is missing) |
| 1459 | unless(self._im.get_param('attachment', missing, |
| 1460 | header='foobar') is missing) |
| 1461 | |
| 1462 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 1463 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1464 | # Test the basic MIMEApplication class |
| 1465 | class TestMIMEApplication(unittest.TestCase): |
| 1466 | def test_headers(self): |
| 1467 | eq = self.assertEqual |
Barry Warsaw | 8b2af27 | 2007-08-31 03:04:26 +0000 | [diff] [blame] | 1468 | msg = MIMEApplication(b'\xfa\xfb\xfc\xfd\xfe\xff') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1469 | eq(msg.get_content_type(), 'application/octet-stream') |
| 1470 | eq(msg['content-transfer-encoding'], 'base64') |
| 1471 | |
| 1472 | def test_body(self): |
| 1473 | eq = self.assertEqual |
R David Murray | 6d94bd4 | 2011-03-16 15:52:22 -0400 | [diff] [blame] | 1474 | bytesdata = b'\xfa\xfb\xfc\xfd\xfe\xff' |
| 1475 | msg = MIMEApplication(bytesdata) |
| 1476 | # whitespace in the cte encoded block is RFC-irrelevant. |
| 1477 | eq(msg.get_payload().strip(), '+vv8/f7/') |
| 1478 | eq(msg.get_payload(decode=True), bytesdata) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1479 | |
R David Murray | ec317a8 | 2013-02-11 10:51:28 -0500 | [diff] [blame] | 1480 | def test_binary_body_with_encode_7or8bit(self): |
| 1481 | # Issue 17171. |
| 1482 | bytesdata = b'\xfa\xfb\xfc\xfd\xfe\xff' |
| 1483 | msg = MIMEApplication(bytesdata, _encoder=encoders.encode_7or8bit) |
| 1484 | # Treated as a string, this will be invalid code points. |
| 1485 | self.assertEqual(msg.get_payload(), '\uFFFD' * len(bytesdata)) |
| 1486 | self.assertEqual(msg.get_payload(decode=True), bytesdata) |
| 1487 | self.assertEqual(msg['Content-Transfer-Encoding'], '8bit') |
| 1488 | s = BytesIO() |
| 1489 | g = BytesGenerator(s) |
| 1490 | g.flatten(msg) |
| 1491 | wireform = s.getvalue() |
| 1492 | msg2 = email.message_from_bytes(wireform) |
| 1493 | self.assertEqual(msg.get_payload(), '\uFFFD' * len(bytesdata)) |
| 1494 | self.assertEqual(msg2.get_payload(decode=True), bytesdata) |
| 1495 | self.assertEqual(msg2['Content-Transfer-Encoding'], '8bit') |
| 1496 | |
| 1497 | def test_binary_body_with_encode_noop(self): |
R David Murray | ceaa8b1 | 2013-02-09 13:02:58 -0500 | [diff] [blame] | 1498 | # Issue 16564: This does not produce an RFC valid message, since to be |
| 1499 | # valid it should have a CTE of binary. But the below works in |
| 1500 | # Python2, and is documented as working this way. |
| 1501 | bytesdata = b'\xfa\xfb\xfc\xfd\xfe\xff' |
| 1502 | msg = MIMEApplication(bytesdata, _encoder=encoders.encode_noop) |
| 1503 | # Treated as a string, this will be invalid code points. |
| 1504 | self.assertEqual(msg.get_payload(), '\uFFFD' * len(bytesdata)) |
| 1505 | self.assertEqual(msg.get_payload(decode=True), bytesdata) |
| 1506 | s = BytesIO() |
| 1507 | g = BytesGenerator(s) |
| 1508 | g.flatten(msg) |
| 1509 | wireform = s.getvalue() |
| 1510 | msg2 = email.message_from_bytes(wireform) |
| 1511 | self.assertEqual(msg.get_payload(), '\uFFFD' * len(bytesdata)) |
| 1512 | self.assertEqual(msg2.get_payload(decode=True), bytesdata) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1513 | |
R David Murray | f6069f9 | 2013-06-27 18:37:00 -0400 | [diff] [blame] | 1514 | def test_binary_body_with_encode_quopri(self): |
| 1515 | # Issue 14360. |
| 1516 | bytesdata = b'\xfa\xfb\xfc\xfd\xfe\xff ' |
| 1517 | msg = MIMEApplication(bytesdata, _encoder=encoders.encode_quopri) |
| 1518 | self.assertEqual(msg.get_payload(), '=FA=FB=FC=FD=FE=FF=20') |
| 1519 | self.assertEqual(msg.get_payload(decode=True), bytesdata) |
| 1520 | self.assertEqual(msg['Content-Transfer-Encoding'], 'quoted-printable') |
| 1521 | s = BytesIO() |
| 1522 | g = BytesGenerator(s) |
| 1523 | g.flatten(msg) |
| 1524 | wireform = s.getvalue() |
| 1525 | msg2 = email.message_from_bytes(wireform) |
| 1526 | self.assertEqual(msg.get_payload(), '=FA=FB=FC=FD=FE=FF=20') |
| 1527 | self.assertEqual(msg2.get_payload(decode=True), bytesdata) |
| 1528 | self.assertEqual(msg2['Content-Transfer-Encoding'], 'quoted-printable') |
| 1529 | |
| 1530 | def test_binary_body_with_encode_base64(self): |
| 1531 | bytesdata = b'\xfa\xfb\xfc\xfd\xfe\xff' |
| 1532 | msg = MIMEApplication(bytesdata, _encoder=encoders.encode_base64) |
| 1533 | self.assertEqual(msg.get_payload(), '+vv8/f7/\n') |
| 1534 | self.assertEqual(msg.get_payload(decode=True), bytesdata) |
| 1535 | s = BytesIO() |
| 1536 | g = BytesGenerator(s) |
| 1537 | g.flatten(msg) |
| 1538 | wireform = s.getvalue() |
| 1539 | msg2 = email.message_from_bytes(wireform) |
| 1540 | self.assertEqual(msg.get_payload(), '+vv8/f7/\n') |
| 1541 | self.assertEqual(msg2.get_payload(decode=True), bytesdata) |
| 1542 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 1543 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1544 | # Test the basic MIMEText class |
| 1545 | class TestMIMEText(unittest.TestCase): |
| 1546 | def setUp(self): |
| 1547 | self._msg = MIMEText('hello there') |
| 1548 | |
| 1549 | def test_types(self): |
| 1550 | eq = self.assertEqual |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 1551 | unless = self.assertTrue |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1552 | eq(self._msg.get_content_type(), 'text/plain') |
| 1553 | eq(self._msg.get_param('charset'), 'us-ascii') |
| 1554 | missing = [] |
| 1555 | unless(self._msg.get_param('foobar', missing) is missing) |
| 1556 | unless(self._msg.get_param('charset', missing, header='foobar') |
| 1557 | is missing) |
| 1558 | |
| 1559 | def test_payload(self): |
| 1560 | self.assertEqual(self._msg.get_payload(), 'hello there') |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 1561 | self.assertTrue(not self._msg.is_multipart()) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1562 | |
| 1563 | def test_charset(self): |
| 1564 | eq = self.assertEqual |
| 1565 | msg = MIMEText('hello there', _charset='us-ascii') |
| 1566 | eq(msg.get_charset().input_charset, 'us-ascii') |
| 1567 | eq(msg['content-type'], 'text/plain; charset="us-ascii"') |
| 1568 | |
R. David Murray | 850fc85 | 2010-06-03 01:58:28 +0000 | [diff] [blame] | 1569 | def test_7bit_input(self): |
| 1570 | eq = self.assertEqual |
| 1571 | msg = MIMEText('hello there', _charset='us-ascii') |
| 1572 | eq(msg.get_charset().input_charset, 'us-ascii') |
| 1573 | eq(msg['content-type'], 'text/plain; charset="us-ascii"') |
| 1574 | |
| 1575 | def test_7bit_input_no_charset(self): |
| 1576 | eq = self.assertEqual |
| 1577 | msg = MIMEText('hello there') |
| 1578 | eq(msg.get_charset(), 'us-ascii') |
| 1579 | eq(msg['content-type'], 'text/plain; charset="us-ascii"') |
| 1580 | self.assertTrue('hello there' in msg.as_string()) |
| 1581 | |
| 1582 | def test_utf8_input(self): |
| 1583 | teststr = '\u043a\u0438\u0440\u0438\u043b\u0438\u0446\u0430' |
| 1584 | eq = self.assertEqual |
| 1585 | msg = MIMEText(teststr, _charset='utf-8') |
| 1586 | eq(msg.get_charset().output_charset, 'utf-8') |
| 1587 | eq(msg['content-type'], 'text/plain; charset="utf-8"') |
| 1588 | eq(msg.get_payload(decode=True), teststr.encode('utf-8')) |
| 1589 | |
| 1590 | @unittest.skip("can't fix because of backward compat in email5, " |
| 1591 | "will fix in email6") |
| 1592 | def test_utf8_input_no_charset(self): |
| 1593 | teststr = '\u043a\u0438\u0440\u0438\u043b\u0438\u0446\u0430' |
| 1594 | self.assertRaises(UnicodeEncodeError, MIMEText, teststr) |
| 1595 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1596 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 1597 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1598 | # Test complicated multipart/* messages |
| 1599 | class TestMultipart(TestEmailBase): |
| 1600 | def setUp(self): |
| 1601 | with openfile('PyBanner048.gif', 'rb') as fp: |
| 1602 | data = fp.read() |
| 1603 | container = MIMEBase('multipart', 'mixed', boundary='BOUNDARY') |
| 1604 | image = MIMEImage(data, name='dingusfish.gif') |
| 1605 | image.add_header('content-disposition', 'attachment', |
| 1606 | filename='dingusfish.gif') |
| 1607 | intro = MIMEText('''\ |
| 1608 | Hi there, |
| 1609 | |
| 1610 | This is the dingus fish. |
| 1611 | ''') |
| 1612 | container.attach(intro) |
| 1613 | container.attach(image) |
| 1614 | container['From'] = 'Barry <barry@digicool.com>' |
| 1615 | container['To'] = 'Dingus Lovers <cravindogs@cravindogs.com>' |
| 1616 | container['Subject'] = 'Here is your dingus fish' |
| 1617 | |
| 1618 | now = 987809702.54848599 |
| 1619 | timetuple = time.localtime(now) |
| 1620 | if timetuple[-1] == 0: |
| 1621 | tzsecs = time.timezone |
| 1622 | else: |
| 1623 | tzsecs = time.altzone |
| 1624 | if tzsecs > 0: |
| 1625 | sign = '-' |
| 1626 | else: |
| 1627 | sign = '+' |
| 1628 | tzoffset = ' %s%04d' % (sign, tzsecs / 36) |
| 1629 | container['Date'] = time.strftime( |
| 1630 | '%a, %d %b %Y %H:%M:%S', |
| 1631 | time.localtime(now)) + tzoffset |
| 1632 | self._msg = container |
| 1633 | self._im = image |
| 1634 | self._txt = intro |
| 1635 | |
| 1636 | def test_hierarchy(self): |
| 1637 | # convenience |
| 1638 | eq = self.assertEqual |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 1639 | unless = self.assertTrue |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1640 | raises = self.assertRaises |
| 1641 | # tests |
| 1642 | m = self._msg |
| 1643 | unless(m.is_multipart()) |
| 1644 | eq(m.get_content_type(), 'multipart/mixed') |
| 1645 | eq(len(m.get_payload()), 2) |
| 1646 | raises(IndexError, m.get_payload, 2) |
| 1647 | m0 = m.get_payload(0) |
| 1648 | m1 = m.get_payload(1) |
| 1649 | unless(m0 is self._txt) |
| 1650 | unless(m1 is self._im) |
| 1651 | eq(m.get_payload(), [m0, m1]) |
| 1652 | unless(not m0.is_multipart()) |
| 1653 | unless(not m1.is_multipart()) |
| 1654 | |
| 1655 | def test_empty_multipart_idempotent(self): |
| 1656 | text = """\ |
| 1657 | Content-Type: multipart/mixed; boundary="BOUNDARY" |
| 1658 | MIME-Version: 1.0 |
| 1659 | Subject: A subject |
| 1660 | To: aperson@dom.ain |
| 1661 | From: bperson@dom.ain |
| 1662 | |
| 1663 | |
| 1664 | --BOUNDARY |
| 1665 | |
| 1666 | |
| 1667 | --BOUNDARY-- |
| 1668 | """ |
| 1669 | msg = Parser().parsestr(text) |
| 1670 | self.ndiffAssertEqual(text, msg.as_string()) |
| 1671 | |
| 1672 | def test_no_parts_in_a_multipart_with_none_epilogue(self): |
| 1673 | outer = MIMEBase('multipart', 'mixed') |
| 1674 | outer['Subject'] = 'A subject' |
| 1675 | outer['To'] = 'aperson@dom.ain' |
| 1676 | outer['From'] = 'bperson@dom.ain' |
| 1677 | outer.set_boundary('BOUNDARY') |
| 1678 | self.ndiffAssertEqual(outer.as_string(), '''\ |
| 1679 | Content-Type: multipart/mixed; boundary="BOUNDARY" |
| 1680 | MIME-Version: 1.0 |
| 1681 | Subject: A subject |
| 1682 | To: aperson@dom.ain |
| 1683 | From: bperson@dom.ain |
| 1684 | |
| 1685 | --BOUNDARY |
| 1686 | |
| 1687 | --BOUNDARY--''') |
| 1688 | |
| 1689 | def test_no_parts_in_a_multipart_with_empty_epilogue(self): |
| 1690 | outer = MIMEBase('multipart', 'mixed') |
| 1691 | outer['Subject'] = 'A subject' |
| 1692 | outer['To'] = 'aperson@dom.ain' |
| 1693 | outer['From'] = 'bperson@dom.ain' |
| 1694 | outer.preamble = '' |
| 1695 | outer.epilogue = '' |
| 1696 | outer.set_boundary('BOUNDARY') |
| 1697 | self.ndiffAssertEqual(outer.as_string(), '''\ |
| 1698 | Content-Type: multipart/mixed; boundary="BOUNDARY" |
| 1699 | MIME-Version: 1.0 |
| 1700 | Subject: A subject |
| 1701 | To: aperson@dom.ain |
| 1702 | From: bperson@dom.ain |
| 1703 | |
| 1704 | |
| 1705 | --BOUNDARY |
| 1706 | |
| 1707 | --BOUNDARY-- |
| 1708 | ''') |
| 1709 | |
| 1710 | def test_one_part_in_a_multipart(self): |
| 1711 | eq = self.ndiffAssertEqual |
| 1712 | outer = MIMEBase('multipart', 'mixed') |
| 1713 | outer['Subject'] = 'A subject' |
| 1714 | outer['To'] = 'aperson@dom.ain' |
| 1715 | outer['From'] = 'bperson@dom.ain' |
| 1716 | outer.set_boundary('BOUNDARY') |
| 1717 | msg = MIMEText('hello world') |
| 1718 | outer.attach(msg) |
| 1719 | eq(outer.as_string(), '''\ |
| 1720 | Content-Type: multipart/mixed; boundary="BOUNDARY" |
| 1721 | MIME-Version: 1.0 |
| 1722 | Subject: A subject |
| 1723 | To: aperson@dom.ain |
| 1724 | From: bperson@dom.ain |
| 1725 | |
| 1726 | --BOUNDARY |
| 1727 | Content-Type: text/plain; charset="us-ascii" |
| 1728 | MIME-Version: 1.0 |
| 1729 | Content-Transfer-Encoding: 7bit |
| 1730 | |
| 1731 | hello world |
| 1732 | --BOUNDARY--''') |
| 1733 | |
| 1734 | def test_seq_parts_in_a_multipart_with_empty_preamble(self): |
| 1735 | eq = self.ndiffAssertEqual |
| 1736 | outer = MIMEBase('multipart', 'mixed') |
| 1737 | outer['Subject'] = 'A subject' |
| 1738 | outer['To'] = 'aperson@dom.ain' |
| 1739 | outer['From'] = 'bperson@dom.ain' |
| 1740 | outer.preamble = '' |
| 1741 | msg = MIMEText('hello world') |
| 1742 | outer.attach(msg) |
| 1743 | outer.set_boundary('BOUNDARY') |
| 1744 | eq(outer.as_string(), '''\ |
| 1745 | Content-Type: multipart/mixed; boundary="BOUNDARY" |
| 1746 | MIME-Version: 1.0 |
| 1747 | Subject: A subject |
| 1748 | To: aperson@dom.ain |
| 1749 | From: bperson@dom.ain |
| 1750 | |
| 1751 | |
| 1752 | --BOUNDARY |
| 1753 | Content-Type: text/plain; charset="us-ascii" |
| 1754 | MIME-Version: 1.0 |
| 1755 | Content-Transfer-Encoding: 7bit |
| 1756 | |
| 1757 | hello world |
| 1758 | --BOUNDARY--''') |
| 1759 | |
| 1760 | |
| 1761 | def test_seq_parts_in_a_multipart_with_none_preamble(self): |
| 1762 | eq = self.ndiffAssertEqual |
| 1763 | outer = MIMEBase('multipart', 'mixed') |
| 1764 | outer['Subject'] = 'A subject' |
| 1765 | outer['To'] = 'aperson@dom.ain' |
| 1766 | outer['From'] = 'bperson@dom.ain' |
| 1767 | outer.preamble = None |
| 1768 | msg = MIMEText('hello world') |
| 1769 | outer.attach(msg) |
| 1770 | outer.set_boundary('BOUNDARY') |
| 1771 | eq(outer.as_string(), '''\ |
| 1772 | Content-Type: multipart/mixed; boundary="BOUNDARY" |
| 1773 | MIME-Version: 1.0 |
| 1774 | Subject: A subject |
| 1775 | To: aperson@dom.ain |
| 1776 | From: bperson@dom.ain |
| 1777 | |
| 1778 | --BOUNDARY |
| 1779 | Content-Type: text/plain; charset="us-ascii" |
| 1780 | MIME-Version: 1.0 |
| 1781 | Content-Transfer-Encoding: 7bit |
| 1782 | |
| 1783 | hello world |
| 1784 | --BOUNDARY--''') |
| 1785 | |
| 1786 | |
| 1787 | def test_seq_parts_in_a_multipart_with_none_epilogue(self): |
| 1788 | eq = self.ndiffAssertEqual |
| 1789 | outer = MIMEBase('multipart', 'mixed') |
| 1790 | outer['Subject'] = 'A subject' |
| 1791 | outer['To'] = 'aperson@dom.ain' |
| 1792 | outer['From'] = 'bperson@dom.ain' |
| 1793 | outer.epilogue = None |
| 1794 | msg = MIMEText('hello world') |
| 1795 | outer.attach(msg) |
| 1796 | outer.set_boundary('BOUNDARY') |
| 1797 | eq(outer.as_string(), '''\ |
| 1798 | Content-Type: multipart/mixed; boundary="BOUNDARY" |
| 1799 | MIME-Version: 1.0 |
| 1800 | Subject: A subject |
| 1801 | To: aperson@dom.ain |
| 1802 | From: bperson@dom.ain |
| 1803 | |
| 1804 | --BOUNDARY |
| 1805 | Content-Type: text/plain; charset="us-ascii" |
| 1806 | MIME-Version: 1.0 |
| 1807 | Content-Transfer-Encoding: 7bit |
| 1808 | |
| 1809 | hello world |
| 1810 | --BOUNDARY--''') |
| 1811 | |
| 1812 | |
| 1813 | def test_seq_parts_in_a_multipart_with_empty_epilogue(self): |
| 1814 | eq = self.ndiffAssertEqual |
| 1815 | outer = MIMEBase('multipart', 'mixed') |
| 1816 | outer['Subject'] = 'A subject' |
| 1817 | outer['To'] = 'aperson@dom.ain' |
| 1818 | outer['From'] = 'bperson@dom.ain' |
| 1819 | outer.epilogue = '' |
| 1820 | msg = MIMEText('hello world') |
| 1821 | outer.attach(msg) |
| 1822 | outer.set_boundary('BOUNDARY') |
| 1823 | eq(outer.as_string(), '''\ |
| 1824 | Content-Type: multipart/mixed; boundary="BOUNDARY" |
| 1825 | MIME-Version: 1.0 |
| 1826 | Subject: A subject |
| 1827 | To: aperson@dom.ain |
| 1828 | From: bperson@dom.ain |
| 1829 | |
| 1830 | --BOUNDARY |
| 1831 | Content-Type: text/plain; charset="us-ascii" |
| 1832 | MIME-Version: 1.0 |
| 1833 | Content-Transfer-Encoding: 7bit |
| 1834 | |
| 1835 | hello world |
| 1836 | --BOUNDARY-- |
| 1837 | ''') |
| 1838 | |
| 1839 | |
| 1840 | def test_seq_parts_in_a_multipart_with_nl_epilogue(self): |
| 1841 | eq = self.ndiffAssertEqual |
| 1842 | outer = MIMEBase('multipart', 'mixed') |
| 1843 | outer['Subject'] = 'A subject' |
| 1844 | outer['To'] = 'aperson@dom.ain' |
| 1845 | outer['From'] = 'bperson@dom.ain' |
| 1846 | outer.epilogue = '\n' |
| 1847 | msg = MIMEText('hello world') |
| 1848 | outer.attach(msg) |
| 1849 | outer.set_boundary('BOUNDARY') |
| 1850 | eq(outer.as_string(), '''\ |
| 1851 | Content-Type: multipart/mixed; boundary="BOUNDARY" |
| 1852 | MIME-Version: 1.0 |
| 1853 | Subject: A subject |
| 1854 | To: aperson@dom.ain |
| 1855 | From: bperson@dom.ain |
| 1856 | |
| 1857 | --BOUNDARY |
| 1858 | Content-Type: text/plain; charset="us-ascii" |
| 1859 | MIME-Version: 1.0 |
| 1860 | Content-Transfer-Encoding: 7bit |
| 1861 | |
| 1862 | hello world |
| 1863 | --BOUNDARY-- |
| 1864 | |
| 1865 | ''') |
| 1866 | |
| 1867 | def test_message_external_body(self): |
| 1868 | eq = self.assertEqual |
| 1869 | msg = self._msgobj('msg_36.txt') |
| 1870 | eq(len(msg.get_payload()), 2) |
| 1871 | msg1 = msg.get_payload(1) |
| 1872 | eq(msg1.get_content_type(), 'multipart/alternative') |
| 1873 | eq(len(msg1.get_payload()), 2) |
| 1874 | for subpart in msg1.get_payload(): |
| 1875 | eq(subpart.get_content_type(), 'message/external-body') |
| 1876 | eq(len(subpart.get_payload()), 1) |
| 1877 | subsubpart = subpart.get_payload(0) |
| 1878 | eq(subsubpart.get_content_type(), 'text/plain') |
| 1879 | |
| 1880 | def test_double_boundary(self): |
| 1881 | # msg_37.txt is a multipart that contains two dash-boundary's in a |
| 1882 | # row. Our interpretation of RFC 2046 calls for ignoring the second |
| 1883 | # and subsequent boundaries. |
| 1884 | msg = self._msgobj('msg_37.txt') |
| 1885 | self.assertEqual(len(msg.get_payload()), 3) |
| 1886 | |
| 1887 | def test_nested_inner_contains_outer_boundary(self): |
| 1888 | eq = self.ndiffAssertEqual |
| 1889 | # msg_38.txt has an inner part that contains outer boundaries. My |
| 1890 | # interpretation of RFC 2046 (based on sections 5.1 and 5.1.2) say |
| 1891 | # these are illegal and should be interpreted as unterminated inner |
| 1892 | # parts. |
| 1893 | msg = self._msgobj('msg_38.txt') |
| 1894 | sfp = StringIO() |
| 1895 | iterators._structure(msg, sfp) |
| 1896 | eq(sfp.getvalue(), """\ |
| 1897 | multipart/mixed |
| 1898 | multipart/mixed |
| 1899 | multipart/alternative |
| 1900 | text/plain |
| 1901 | text/plain |
| 1902 | text/plain |
| 1903 | text/plain |
| 1904 | """) |
| 1905 | |
| 1906 | def test_nested_with_same_boundary(self): |
| 1907 | eq = self.ndiffAssertEqual |
| 1908 | # msg 39.txt is similarly evil in that it's got inner parts that use |
| 1909 | # the same boundary as outer parts. Again, I believe the way this is |
| 1910 | # parsed is closest to the spirit of RFC 2046 |
| 1911 | msg = self._msgobj('msg_39.txt') |
| 1912 | sfp = StringIO() |
| 1913 | iterators._structure(msg, sfp) |
| 1914 | eq(sfp.getvalue(), """\ |
| 1915 | multipart/mixed |
| 1916 | multipart/mixed |
| 1917 | multipart/alternative |
| 1918 | application/octet-stream |
| 1919 | application/octet-stream |
| 1920 | text/plain |
| 1921 | """) |
| 1922 | |
| 1923 | def test_boundary_in_non_multipart(self): |
| 1924 | msg = self._msgobj('msg_40.txt') |
| 1925 | self.assertEqual(msg.as_string(), '''\ |
| 1926 | MIME-Version: 1.0 |
| 1927 | Content-Type: text/html; boundary="--961284236552522269" |
| 1928 | |
| 1929 | ----961284236552522269 |
| 1930 | Content-Type: text/html; |
| 1931 | Content-Transfer-Encoding: 7Bit |
| 1932 | |
| 1933 | <html></html> |
| 1934 | |
| 1935 | ----961284236552522269-- |
| 1936 | ''') |
| 1937 | |
| 1938 | def test_boundary_with_leading_space(self): |
| 1939 | eq = self.assertEqual |
| 1940 | msg = email.message_from_string('''\ |
| 1941 | MIME-Version: 1.0 |
| 1942 | Content-Type: multipart/mixed; boundary=" XXXX" |
| 1943 | |
| 1944 | -- XXXX |
| 1945 | Content-Type: text/plain |
| 1946 | |
| 1947 | |
| 1948 | -- XXXX |
| 1949 | Content-Type: text/plain |
| 1950 | |
| 1951 | -- XXXX-- |
| 1952 | ''') |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 1953 | self.assertTrue(msg.is_multipart()) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1954 | eq(msg.get_boundary(), ' XXXX') |
| 1955 | eq(len(msg.get_payload()), 2) |
| 1956 | |
| 1957 | def test_boundary_without_trailing_newline(self): |
| 1958 | m = Parser().parsestr("""\ |
| 1959 | Content-Type: multipart/mixed; boundary="===============0012394164==" |
| 1960 | MIME-Version: 1.0 |
| 1961 | |
| 1962 | --===============0012394164== |
| 1963 | Content-Type: image/file1.jpg |
| 1964 | MIME-Version: 1.0 |
| 1965 | Content-Transfer-Encoding: base64 |
| 1966 | |
| 1967 | YXNkZg== |
| 1968 | --===============0012394164==--""") |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 1969 | self.assertEqual(m.get_payload(0).get_payload(), 'YXNkZg==') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1970 | |
| 1971 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 1972 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1973 | # Test some badly formatted messages |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 1974 | class TestNonConformant(TestEmailBase): |
R David Murray | 3edd22a | 2011-04-18 13:59:37 -0400 | [diff] [blame] | 1975 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1976 | def test_parse_missing_minor_type(self): |
| 1977 | eq = self.assertEqual |
| 1978 | msg = self._msgobj('msg_14.txt') |
| 1979 | eq(msg.get_content_type(), 'text/plain') |
| 1980 | eq(msg.get_content_maintype(), 'text') |
| 1981 | eq(msg.get_content_subtype(), 'plain') |
| 1982 | |
R David Murray | 80e0aee | 2012-05-27 21:23:34 -0400 | [diff] [blame] | 1983 | # test_defect_handling |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1984 | def test_same_boundary_inner_outer(self): |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 1985 | unless = self.assertTrue |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1986 | msg = self._msgobj('msg_15.txt') |
| 1987 | # XXX We can probably eventually do better |
| 1988 | inner = msg.get_payload(0) |
| 1989 | unless(hasattr(inner, 'defects')) |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 1990 | self.assertEqual(len(inner.defects), 1) |
| 1991 | unless(isinstance(inner.defects[0], |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1992 | errors.StartBoundaryNotFoundDefect)) |
| 1993 | |
R David Murray | 80e0aee | 2012-05-27 21:23:34 -0400 | [diff] [blame] | 1994 | # test_defect_handling |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1995 | def test_multipart_no_boundary(self): |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 1996 | unless = self.assertTrue |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 1997 | msg = self._msgobj('msg_25.txt') |
| 1998 | unless(isinstance(msg.get_payload(), str)) |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 1999 | self.assertEqual(len(msg.defects), 2) |
| 2000 | unless(isinstance(msg.defects[0], |
R David Murray | 3edd22a | 2011-04-18 13:59:37 -0400 | [diff] [blame] | 2001 | errors.NoBoundaryInMultipartDefect)) |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 2002 | unless(isinstance(msg.defects[1], |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2003 | errors.MultipartInvariantViolationDefect)) |
| 2004 | |
R David Murray | 749073a | 2011-06-22 13:47:53 -0400 | [diff] [blame] | 2005 | multipart_msg = textwrap.dedent("""\ |
| 2006 | Date: Wed, 14 Nov 2007 12:56:23 GMT |
| 2007 | From: foo@bar.invalid |
| 2008 | To: foo@bar.invalid |
| 2009 | Subject: Content-Transfer-Encoding: base64 and multipart |
| 2010 | MIME-Version: 1.0 |
| 2011 | Content-Type: multipart/mixed; |
| 2012 | boundary="===============3344438784458119861=="{} |
| 2013 | |
| 2014 | --===============3344438784458119861== |
| 2015 | Content-Type: text/plain |
| 2016 | |
| 2017 | Test message |
| 2018 | |
| 2019 | --===============3344438784458119861== |
| 2020 | Content-Type: application/octet-stream |
| 2021 | Content-Transfer-Encoding: base64 |
| 2022 | |
| 2023 | YWJj |
| 2024 | |
| 2025 | --===============3344438784458119861==-- |
| 2026 | """) |
| 2027 | |
R David Murray | 80e0aee | 2012-05-27 21:23:34 -0400 | [diff] [blame] | 2028 | # test_defect_handling |
R David Murray | 749073a | 2011-06-22 13:47:53 -0400 | [diff] [blame] | 2029 | def test_multipart_invalid_cte(self): |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 2030 | msg = self._str_msg( |
| 2031 | self.multipart_msg.format("\nContent-Transfer-Encoding: base64")) |
| 2032 | self.assertEqual(len(msg.defects), 1) |
| 2033 | self.assertIsInstance(msg.defects[0], |
R David Murray | 749073a | 2011-06-22 13:47:53 -0400 | [diff] [blame] | 2034 | errors.InvalidMultipartContentTransferEncodingDefect) |
| 2035 | |
R David Murray | 80e0aee | 2012-05-27 21:23:34 -0400 | [diff] [blame] | 2036 | # test_defect_handling |
R David Murray | 749073a | 2011-06-22 13:47:53 -0400 | [diff] [blame] | 2037 | def test_multipart_no_cte_no_defect(self): |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 2038 | msg = self._str_msg(self.multipart_msg.format('')) |
| 2039 | self.assertEqual(len(msg.defects), 0) |
R David Murray | 749073a | 2011-06-22 13:47:53 -0400 | [diff] [blame] | 2040 | |
R David Murray | 80e0aee | 2012-05-27 21:23:34 -0400 | [diff] [blame] | 2041 | # test_defect_handling |
R David Murray | 749073a | 2011-06-22 13:47:53 -0400 | [diff] [blame] | 2042 | def test_multipart_valid_cte_no_defect(self): |
| 2043 | for cte in ('7bit', '8bit', 'BINary'): |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 2044 | msg = self._str_msg( |
R David Murray | 749073a | 2011-06-22 13:47:53 -0400 | [diff] [blame] | 2045 | self.multipart_msg.format( |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 2046 | "\nContent-Transfer-Encoding: {}".format(cte))) |
| 2047 | self.assertEqual(len(msg.defects), 0) |
R David Murray | 749073a | 2011-06-22 13:47:53 -0400 | [diff] [blame] | 2048 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 2049 | # test_headerregistry.TestContentTyopeHeader invalid_1 and invalid_2. |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2050 | def test_invalid_content_type(self): |
| 2051 | eq = self.assertEqual |
| 2052 | neq = self.ndiffAssertEqual |
| 2053 | msg = Message() |
| 2054 | # RFC 2045, $5.2 says invalid yields text/plain |
| 2055 | msg['Content-Type'] = 'text' |
| 2056 | eq(msg.get_content_maintype(), 'text') |
| 2057 | eq(msg.get_content_subtype(), 'plain') |
| 2058 | eq(msg.get_content_type(), 'text/plain') |
| 2059 | # Clear the old value and try something /really/ invalid |
| 2060 | del msg['content-type'] |
| 2061 | msg['Content-Type'] = 'foo' |
| 2062 | eq(msg.get_content_maintype(), 'text') |
| 2063 | eq(msg.get_content_subtype(), 'plain') |
| 2064 | eq(msg.get_content_type(), 'text/plain') |
| 2065 | # Still, make sure that the message is idempotently generated |
| 2066 | s = StringIO() |
| 2067 | g = Generator(s) |
| 2068 | g.flatten(msg) |
| 2069 | neq(s.getvalue(), 'Content-Type: foo\n\n') |
| 2070 | |
| 2071 | def test_no_start_boundary(self): |
| 2072 | eq = self.ndiffAssertEqual |
| 2073 | msg = self._msgobj('msg_31.txt') |
| 2074 | eq(msg.get_payload(), """\ |
| 2075 | --BOUNDARY |
| 2076 | Content-Type: text/plain |
| 2077 | |
| 2078 | message 1 |
| 2079 | |
| 2080 | --BOUNDARY |
| 2081 | Content-Type: text/plain |
| 2082 | |
| 2083 | message 2 |
| 2084 | |
| 2085 | --BOUNDARY-- |
| 2086 | """) |
| 2087 | |
| 2088 | def test_no_separating_blank_line(self): |
| 2089 | eq = self.ndiffAssertEqual |
| 2090 | msg = self._msgobj('msg_35.txt') |
| 2091 | eq(msg.as_string(), """\ |
| 2092 | From: aperson@dom.ain |
| 2093 | To: bperson@dom.ain |
| 2094 | Subject: here's something interesting |
| 2095 | |
| 2096 | counter to RFC 2822, there's no separating newline here |
| 2097 | """) |
| 2098 | |
R David Murray | 80e0aee | 2012-05-27 21:23:34 -0400 | [diff] [blame] | 2099 | # test_defect_handling |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2100 | def test_lying_multipart(self): |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 2101 | unless = self.assertTrue |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2102 | msg = self._msgobj('msg_41.txt') |
| 2103 | unless(hasattr(msg, 'defects')) |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 2104 | self.assertEqual(len(msg.defects), 2) |
| 2105 | unless(isinstance(msg.defects[0], |
R David Murray | 3edd22a | 2011-04-18 13:59:37 -0400 | [diff] [blame] | 2106 | errors.NoBoundaryInMultipartDefect)) |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 2107 | unless(isinstance(msg.defects[1], |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2108 | errors.MultipartInvariantViolationDefect)) |
| 2109 | |
R David Murray | 80e0aee | 2012-05-27 21:23:34 -0400 | [diff] [blame] | 2110 | # test_defect_handling |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2111 | def test_missing_start_boundary(self): |
| 2112 | outer = self._msgobj('msg_42.txt') |
| 2113 | # The message structure is: |
| 2114 | # |
| 2115 | # multipart/mixed |
| 2116 | # text/plain |
| 2117 | # message/rfc822 |
| 2118 | # multipart/mixed [*] |
| 2119 | # |
| 2120 | # [*] This message is missing its start boundary |
| 2121 | bad = outer.get_payload(1).get_payload(0) |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 2122 | self.assertEqual(len(bad.defects), 1) |
| 2123 | self.assertTrue(isinstance(bad.defects[0], |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2124 | errors.StartBoundaryNotFoundDefect)) |
| 2125 | |
R David Murray | 80e0aee | 2012-05-27 21:23:34 -0400 | [diff] [blame] | 2126 | # test_defect_handling |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2127 | def test_first_line_is_continuation_header(self): |
| 2128 | eq = self.assertEqual |
R David Murray | adbdcdb | 2012-05-27 20:45:01 -0400 | [diff] [blame] | 2129 | m = ' Line 1\nSubject: test\n\nbody' |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 2130 | msg = email.message_from_string(m) |
R David Murray | adbdcdb | 2012-05-27 20:45:01 -0400 | [diff] [blame] | 2131 | eq(msg.keys(), ['Subject']) |
| 2132 | eq(msg.get_payload(), 'body') |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 2133 | eq(len(msg.defects), 1) |
R David Murray | adbdcdb | 2012-05-27 20:45:01 -0400 | [diff] [blame] | 2134 | self.assertDefectsEqual(msg.defects, |
| 2135 | [errors.FirstHeaderLineIsContinuationDefect]) |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 2136 | eq(msg.defects[0].line, ' Line 1\n') |
R David Murray | 3edd22a | 2011-04-18 13:59:37 -0400 | [diff] [blame] | 2137 | |
R David Murray | d41595b | 2012-05-28 20:14:10 -0400 | [diff] [blame] | 2138 | # test_defect_handling |
R David Murray | adbdcdb | 2012-05-27 20:45:01 -0400 | [diff] [blame] | 2139 | def test_missing_header_body_separator(self): |
| 2140 | # Our heuristic if we see a line that doesn't look like a header (no |
| 2141 | # leading whitespace but no ':') is to assume that the blank line that |
| 2142 | # separates the header from the body is missing, and to stop parsing |
| 2143 | # headers and start parsing the body. |
| 2144 | msg = self._str_msg('Subject: test\nnot a header\nTo: abc\n\nb\n') |
| 2145 | self.assertEqual(msg.keys(), ['Subject']) |
| 2146 | self.assertEqual(msg.get_payload(), 'not a header\nTo: abc\n\nb\n') |
| 2147 | self.assertDefectsEqual(msg.defects, |
| 2148 | [errors.MissingHeaderBodySeparatorDefect]) |
| 2149 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 2150 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2151 | # Test RFC 2047 header encoding and decoding |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 2152 | class TestRFC2047(TestEmailBase): |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2153 | def test_rfc2047_multiline(self): |
| 2154 | eq = self.assertEqual |
| 2155 | s = """Re: =?mac-iceland?q?r=8Aksm=9Arg=8Cs?= baz |
| 2156 | foo bar =?mac-iceland?q?r=8Aksm=9Arg=8Cs?=""" |
| 2157 | dh = decode_header(s) |
| 2158 | eq(dh, [ |
R David Murray | 07ea53c | 2012-06-02 17:56:49 -0400 | [diff] [blame] | 2159 | (b'Re: ', None), |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2160 | (b'r\x8aksm\x9arg\x8cs', 'mac-iceland'), |
R David Murray | 07ea53c | 2012-06-02 17:56:49 -0400 | [diff] [blame] | 2161 | (b' baz foo bar ', None), |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2162 | (b'r\x8aksm\x9arg\x8cs', 'mac-iceland')]) |
| 2163 | header = make_header(dh) |
| 2164 | eq(str(header), |
| 2165 | 'Re: r\xe4ksm\xf6rg\xe5s baz foo bar r\xe4ksm\xf6rg\xe5s') |
Barry Warsaw | 00b3422 | 2007-08-31 02:35:00 +0000 | [diff] [blame] | 2166 | self.ndiffAssertEqual(header.encode(maxlinelen=76), """\ |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 2167 | Re: =?mac-iceland?q?r=8Aksm=9Arg=8Cs?= baz foo bar =?mac-iceland?q?r=8Aksm?= |
| 2168 | =?mac-iceland?q?=9Arg=8Cs?=""") |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2169 | |
R David Murray | 07ea53c | 2012-06-02 17:56:49 -0400 | [diff] [blame] | 2170 | def test_whitespace_keeper_unicode(self): |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2171 | eq = self.assertEqual |
| 2172 | s = '=?ISO-8859-1?Q?Andr=E9?= Pirard <pirard@dom.ain>' |
| 2173 | dh = decode_header(s) |
| 2174 | eq(dh, [(b'Andr\xe9', 'iso-8859-1'), |
R David Murray | 07ea53c | 2012-06-02 17:56:49 -0400 | [diff] [blame] | 2175 | (b' Pirard <pirard@dom.ain>', None)]) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2176 | header = str(make_header(dh)) |
| 2177 | eq(header, 'Andr\xe9 Pirard <pirard@dom.ain>') |
| 2178 | |
R David Murray | 07ea53c | 2012-06-02 17:56:49 -0400 | [diff] [blame] | 2179 | def test_whitespace_keeper_unicode_2(self): |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2180 | eq = self.assertEqual |
| 2181 | s = 'The =?iso-8859-1?b?cXVpY2sgYnJvd24gZm94?= jumped over the =?iso-8859-1?b?bGF6eSBkb2c=?=' |
| 2182 | dh = decode_header(s) |
R David Murray | 07ea53c | 2012-06-02 17:56:49 -0400 | [diff] [blame] | 2183 | eq(dh, [(b'The ', None), (b'quick brown fox', 'iso-8859-1'), |
| 2184 | (b' jumped over the ', None), (b'lazy dog', 'iso-8859-1')]) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2185 | hu = str(make_header(dh)) |
| 2186 | eq(hu, 'The quick brown fox jumped over the lazy dog') |
| 2187 | |
| 2188 | def test_rfc2047_missing_whitespace(self): |
| 2189 | s = 'Sm=?ISO-8859-1?B?9g==?=rg=?ISO-8859-1?B?5Q==?=sbord' |
| 2190 | dh = decode_header(s) |
R David Murray | 07ea53c | 2012-06-02 17:56:49 -0400 | [diff] [blame] | 2191 | self.assertEqual(dh, [(b'Sm', None), (b'\xf6', 'iso-8859-1'), |
| 2192 | (b'rg', None), (b'\xe5', 'iso-8859-1'), |
| 2193 | (b'sbord', None)]) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2194 | |
| 2195 | def test_rfc2047_with_whitespace(self): |
| 2196 | s = 'Sm =?ISO-8859-1?B?9g==?= rg =?ISO-8859-1?B?5Q==?= sbord' |
| 2197 | dh = decode_header(s) |
R David Murray | 07ea53c | 2012-06-02 17:56:49 -0400 | [diff] [blame] | 2198 | self.assertEqual(dh, [(b'Sm ', None), (b'\xf6', 'iso-8859-1'), |
| 2199 | (b' rg ', None), (b'\xe5', 'iso-8859-1'), |
| 2200 | (b' sbord', None)]) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2201 | |
R. David Murray | c4e69cc | 2010-08-03 22:14:10 +0000 | [diff] [blame] | 2202 | def test_rfc2047_B_bad_padding(self): |
| 2203 | s = '=?iso-8859-1?B?%s?=' |
| 2204 | data = [ # only test complete bytes |
| 2205 | ('dm==', b'v'), ('dm=', b'v'), ('dm', b'v'), |
| 2206 | ('dmk=', b'vi'), ('dmk', b'vi') |
| 2207 | ] |
| 2208 | for q, a in data: |
| 2209 | dh = decode_header(s % q) |
| 2210 | self.assertEqual(dh, [(a, 'iso-8859-1')]) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2211 | |
R. David Murray | 31e984c | 2010-10-01 15:40:20 +0000 | [diff] [blame] | 2212 | def test_rfc2047_Q_invalid_digits(self): |
| 2213 | # issue 10004. |
| 2214 | s = '=?iso-8659-1?Q?andr=e9=zz?=' |
| 2215 | self.assertEqual(decode_header(s), |
| 2216 | [(b'andr\xe9=zz', 'iso-8659-1')]) |
| 2217 | |
R David Murray | 07ea53c | 2012-06-02 17:56:49 -0400 | [diff] [blame] | 2218 | def test_rfc2047_rfc2047_1(self): |
| 2219 | # 1st testcase at end of rfc2047 |
| 2220 | s = '(=?ISO-8859-1?Q?a?=)' |
| 2221 | self.assertEqual(decode_header(s), |
| 2222 | [(b'(', None), (b'a', 'iso-8859-1'), (b')', None)]) |
| 2223 | |
| 2224 | def test_rfc2047_rfc2047_2(self): |
| 2225 | # 2nd testcase at end of rfc2047 |
| 2226 | s = '(=?ISO-8859-1?Q?a?= b)' |
| 2227 | self.assertEqual(decode_header(s), |
| 2228 | [(b'(', None), (b'a', 'iso-8859-1'), (b' b)', None)]) |
| 2229 | |
| 2230 | def test_rfc2047_rfc2047_3(self): |
| 2231 | # 3rd testcase at end of rfc2047 |
| 2232 | s = '(=?ISO-8859-1?Q?a?= =?ISO-8859-1?Q?b?=)' |
| 2233 | self.assertEqual(decode_header(s), |
| 2234 | [(b'(', None), (b'ab', 'iso-8859-1'), (b')', None)]) |
| 2235 | |
| 2236 | def test_rfc2047_rfc2047_4(self): |
| 2237 | # 4th testcase at end of rfc2047 |
| 2238 | s = '(=?ISO-8859-1?Q?a?= =?ISO-8859-1?Q?b?=)' |
| 2239 | self.assertEqual(decode_header(s), |
| 2240 | [(b'(', None), (b'ab', 'iso-8859-1'), (b')', None)]) |
| 2241 | |
| 2242 | def test_rfc2047_rfc2047_5a(self): |
| 2243 | # 5th testcase at end of rfc2047 newline is \r\n |
| 2244 | s = '(=?ISO-8859-1?Q?a?=\r\n =?ISO-8859-1?Q?b?=)' |
| 2245 | self.assertEqual(decode_header(s), |
| 2246 | [(b'(', None), (b'ab', 'iso-8859-1'), (b')', None)]) |
| 2247 | |
| 2248 | def test_rfc2047_rfc2047_5b(self): |
| 2249 | # 5th testcase at end of rfc2047 newline is \n |
| 2250 | s = '(=?ISO-8859-1?Q?a?=\n =?ISO-8859-1?Q?b?=)' |
| 2251 | self.assertEqual(decode_header(s), |
| 2252 | [(b'(', None), (b'ab', 'iso-8859-1'), (b')', None)]) |
| 2253 | |
| 2254 | def test_rfc2047_rfc2047_6(self): |
| 2255 | # 6th testcase at end of rfc2047 |
| 2256 | s = '(=?ISO-8859-1?Q?a_b?=)' |
| 2257 | self.assertEqual(decode_header(s), |
| 2258 | [(b'(', None), (b'a b', 'iso-8859-1'), (b')', None)]) |
| 2259 | |
| 2260 | def test_rfc2047_rfc2047_7(self): |
| 2261 | # 7th testcase at end of rfc2047 |
| 2262 | s = '(=?ISO-8859-1?Q?a?= =?ISO-8859-2?Q?_b?=)' |
| 2263 | self.assertEqual(decode_header(s), |
| 2264 | [(b'(', None), (b'a', 'iso-8859-1'), (b' b', 'iso-8859-2'), |
| 2265 | (b')', None)]) |
| 2266 | self.assertEqual(make_header(decode_header(s)).encode(), s.lower()) |
| 2267 | self.assertEqual(str(make_header(decode_header(s))), '(a b)') |
| 2268 | |
R David Murray | 82ffabd | 2012-06-03 12:27:07 -0400 | [diff] [blame] | 2269 | def test_multiline_header(self): |
| 2270 | s = '=?windows-1252?q?=22M=FCller_T=22?=\r\n <T.Mueller@xxx.com>' |
| 2271 | self.assertEqual(decode_header(s), |
| 2272 | [(b'"M\xfcller T"', 'windows-1252'), |
| 2273 | (b'<T.Mueller@xxx.com>', None)]) |
| 2274 | self.assertEqual(make_header(decode_header(s)).encode(), |
| 2275 | ''.join(s.splitlines())) |
| 2276 | self.assertEqual(str(make_header(decode_header(s))), |
| 2277 | '"Müller T" <T.Mueller@xxx.com>') |
| 2278 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 2279 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2280 | # Test the MIMEMessage class |
| 2281 | class TestMIMEMessage(TestEmailBase): |
| 2282 | def setUp(self): |
| 2283 | with openfile('msg_11.txt') as fp: |
| 2284 | self._text = fp.read() |
| 2285 | |
| 2286 | def test_type_error(self): |
| 2287 | self.assertRaises(TypeError, MIMEMessage, 'a plain string') |
| 2288 | |
| 2289 | def test_valid_argument(self): |
| 2290 | eq = self.assertEqual |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 2291 | unless = self.assertTrue |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2292 | subject = 'A sub-message' |
| 2293 | m = Message() |
| 2294 | m['Subject'] = subject |
| 2295 | r = MIMEMessage(m) |
| 2296 | eq(r.get_content_type(), 'message/rfc822') |
| 2297 | payload = r.get_payload() |
| 2298 | unless(isinstance(payload, list)) |
| 2299 | eq(len(payload), 1) |
| 2300 | subpart = payload[0] |
| 2301 | unless(subpart is m) |
| 2302 | eq(subpart['subject'], subject) |
| 2303 | |
| 2304 | def test_bad_multipart(self): |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2305 | msg1 = Message() |
| 2306 | msg1['Subject'] = 'subpart 1' |
| 2307 | msg2 = Message() |
| 2308 | msg2['Subject'] = 'subpart 2' |
| 2309 | r = MIMEMessage(msg1) |
| 2310 | self.assertRaises(errors.MultipartConversionError, r.attach, msg2) |
| 2311 | |
| 2312 | def test_generate(self): |
| 2313 | # First craft the message to be encapsulated |
| 2314 | m = Message() |
| 2315 | m['Subject'] = 'An enclosed message' |
| 2316 | m.set_payload('Here is the body of the message.\n') |
| 2317 | r = MIMEMessage(m) |
| 2318 | r['Subject'] = 'The enclosing message' |
| 2319 | s = StringIO() |
| 2320 | g = Generator(s) |
| 2321 | g.flatten(r) |
| 2322 | self.assertEqual(s.getvalue(), """\ |
| 2323 | Content-Type: message/rfc822 |
| 2324 | MIME-Version: 1.0 |
| 2325 | Subject: The enclosing message |
| 2326 | |
| 2327 | Subject: An enclosed message |
| 2328 | |
| 2329 | Here is the body of the message. |
| 2330 | """) |
| 2331 | |
| 2332 | def test_parse_message_rfc822(self): |
| 2333 | eq = self.assertEqual |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 2334 | unless = self.assertTrue |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2335 | msg = self._msgobj('msg_11.txt') |
| 2336 | eq(msg.get_content_type(), 'message/rfc822') |
| 2337 | payload = msg.get_payload() |
| 2338 | unless(isinstance(payload, list)) |
| 2339 | eq(len(payload), 1) |
| 2340 | submsg = payload[0] |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 2341 | self.assertTrue(isinstance(submsg, Message)) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2342 | eq(submsg['subject'], 'An enclosed message') |
| 2343 | eq(submsg.get_payload(), 'Here is the body of the message.\n') |
| 2344 | |
| 2345 | def test_dsn(self): |
| 2346 | eq = self.assertEqual |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 2347 | unless = self.assertTrue |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2348 | # msg 16 is a Delivery Status Notification, see RFC 1894 |
| 2349 | msg = self._msgobj('msg_16.txt') |
| 2350 | eq(msg.get_content_type(), 'multipart/report') |
| 2351 | unless(msg.is_multipart()) |
| 2352 | eq(len(msg.get_payload()), 3) |
| 2353 | # Subpart 1 is a text/plain, human readable section |
| 2354 | subpart = msg.get_payload(0) |
| 2355 | eq(subpart.get_content_type(), 'text/plain') |
| 2356 | eq(subpart.get_payload(), """\ |
| 2357 | This report relates to a message you sent with the following header fields: |
| 2358 | |
| 2359 | Message-id: <002001c144a6$8752e060$56104586@oxy.edu> |
| 2360 | Date: Sun, 23 Sep 2001 20:10:55 -0700 |
| 2361 | From: "Ian T. Henry" <henryi@oxy.edu> |
| 2362 | To: SoCal Raves <scr@socal-raves.org> |
| 2363 | Subject: [scr] yeah for Ians!! |
| 2364 | |
| 2365 | Your message cannot be delivered to the following recipients: |
| 2366 | |
| 2367 | Recipient address: jangel1@cougar.noc.ucla.edu |
| 2368 | Reason: recipient reached disk quota |
| 2369 | |
| 2370 | """) |
| 2371 | # Subpart 2 contains the machine parsable DSN information. It |
| 2372 | # consists of two blocks of headers, represented by two nested Message |
| 2373 | # objects. |
| 2374 | subpart = msg.get_payload(1) |
| 2375 | eq(subpart.get_content_type(), 'message/delivery-status') |
| 2376 | eq(len(subpart.get_payload()), 2) |
| 2377 | # message/delivery-status should treat each block as a bunch of |
| 2378 | # headers, i.e. a bunch of Message objects. |
| 2379 | dsn1 = subpart.get_payload(0) |
| 2380 | unless(isinstance(dsn1, Message)) |
| 2381 | eq(dsn1['original-envelope-id'], '0GK500B4HD0888@cougar.noc.ucla.edu') |
| 2382 | eq(dsn1.get_param('dns', header='reporting-mta'), '') |
| 2383 | # Try a missing one <wink> |
| 2384 | eq(dsn1.get_param('nsd', header='reporting-mta'), None) |
| 2385 | dsn2 = subpart.get_payload(1) |
| 2386 | unless(isinstance(dsn2, Message)) |
| 2387 | eq(dsn2['action'], 'failed') |
| 2388 | eq(dsn2.get_params(header='original-recipient'), |
| 2389 | [('rfc822', ''), ('jangel1@cougar.noc.ucla.edu', '')]) |
| 2390 | eq(dsn2.get_param('rfc822', header='final-recipient'), '') |
| 2391 | # Subpart 3 is the original message |
| 2392 | subpart = msg.get_payload(2) |
| 2393 | eq(subpart.get_content_type(), 'message/rfc822') |
| 2394 | payload = subpart.get_payload() |
| 2395 | unless(isinstance(payload, list)) |
| 2396 | eq(len(payload), 1) |
| 2397 | subsubpart = payload[0] |
| 2398 | unless(isinstance(subsubpart, Message)) |
| 2399 | eq(subsubpart.get_content_type(), 'text/plain') |
| 2400 | eq(subsubpart['message-id'], |
| 2401 | '<002001c144a6$8752e060$56104586@oxy.edu>') |
| 2402 | |
| 2403 | def test_epilogue(self): |
| 2404 | eq = self.ndiffAssertEqual |
| 2405 | with openfile('msg_21.txt') as fp: |
| 2406 | text = fp.read() |
| 2407 | msg = Message() |
| 2408 | msg['From'] = 'aperson@dom.ain' |
| 2409 | msg['To'] = 'bperson@dom.ain' |
| 2410 | msg['Subject'] = 'Test' |
| 2411 | msg.preamble = 'MIME message' |
| 2412 | msg.epilogue = 'End of MIME message\n' |
| 2413 | msg1 = MIMEText('One') |
| 2414 | msg2 = MIMEText('Two') |
| 2415 | msg.add_header('Content-Type', 'multipart/mixed', boundary='BOUNDARY') |
| 2416 | msg.attach(msg1) |
| 2417 | msg.attach(msg2) |
| 2418 | sfp = StringIO() |
| 2419 | g = Generator(sfp) |
| 2420 | g.flatten(msg) |
| 2421 | eq(sfp.getvalue(), text) |
| 2422 | |
| 2423 | def test_no_nl_preamble(self): |
| 2424 | eq = self.ndiffAssertEqual |
| 2425 | msg = Message() |
| 2426 | msg['From'] = 'aperson@dom.ain' |
| 2427 | msg['To'] = 'bperson@dom.ain' |
| 2428 | msg['Subject'] = 'Test' |
| 2429 | msg.preamble = 'MIME message' |
| 2430 | msg.epilogue = '' |
| 2431 | msg1 = MIMEText('One') |
| 2432 | msg2 = MIMEText('Two') |
| 2433 | msg.add_header('Content-Type', 'multipart/mixed', boundary='BOUNDARY') |
| 2434 | msg.attach(msg1) |
| 2435 | msg.attach(msg2) |
| 2436 | eq(msg.as_string(), """\ |
| 2437 | From: aperson@dom.ain |
| 2438 | To: bperson@dom.ain |
| 2439 | Subject: Test |
| 2440 | Content-Type: multipart/mixed; boundary="BOUNDARY" |
| 2441 | |
| 2442 | MIME message |
| 2443 | --BOUNDARY |
| 2444 | Content-Type: text/plain; charset="us-ascii" |
| 2445 | MIME-Version: 1.0 |
| 2446 | Content-Transfer-Encoding: 7bit |
| 2447 | |
| 2448 | One |
| 2449 | --BOUNDARY |
| 2450 | Content-Type: text/plain; charset="us-ascii" |
| 2451 | MIME-Version: 1.0 |
| 2452 | Content-Transfer-Encoding: 7bit |
| 2453 | |
| 2454 | Two |
| 2455 | --BOUNDARY-- |
| 2456 | """) |
| 2457 | |
| 2458 | def test_default_type(self): |
| 2459 | eq = self.assertEqual |
| 2460 | with openfile('msg_30.txt') as fp: |
| 2461 | msg = email.message_from_file(fp) |
| 2462 | container1 = msg.get_payload(0) |
| 2463 | eq(container1.get_default_type(), 'message/rfc822') |
| 2464 | eq(container1.get_content_type(), 'message/rfc822') |
| 2465 | container2 = msg.get_payload(1) |
| 2466 | eq(container2.get_default_type(), 'message/rfc822') |
| 2467 | eq(container2.get_content_type(), 'message/rfc822') |
| 2468 | container1a = container1.get_payload(0) |
| 2469 | eq(container1a.get_default_type(), 'text/plain') |
| 2470 | eq(container1a.get_content_type(), 'text/plain') |
| 2471 | container2a = container2.get_payload(0) |
| 2472 | eq(container2a.get_default_type(), 'text/plain') |
| 2473 | eq(container2a.get_content_type(), 'text/plain') |
| 2474 | |
| 2475 | def test_default_type_with_explicit_container_type(self): |
| 2476 | eq = self.assertEqual |
| 2477 | with openfile('msg_28.txt') as fp: |
| 2478 | msg = email.message_from_file(fp) |
| 2479 | container1 = msg.get_payload(0) |
| 2480 | eq(container1.get_default_type(), 'message/rfc822') |
| 2481 | eq(container1.get_content_type(), 'message/rfc822') |
| 2482 | container2 = msg.get_payload(1) |
| 2483 | eq(container2.get_default_type(), 'message/rfc822') |
| 2484 | eq(container2.get_content_type(), 'message/rfc822') |
| 2485 | container1a = container1.get_payload(0) |
| 2486 | eq(container1a.get_default_type(), 'text/plain') |
| 2487 | eq(container1a.get_content_type(), 'text/plain') |
| 2488 | container2a = container2.get_payload(0) |
| 2489 | eq(container2a.get_default_type(), 'text/plain') |
| 2490 | eq(container2a.get_content_type(), 'text/plain') |
| 2491 | |
| 2492 | def test_default_type_non_parsed(self): |
| 2493 | eq = self.assertEqual |
| 2494 | neq = self.ndiffAssertEqual |
| 2495 | # Set up container |
| 2496 | container = MIMEMultipart('digest', 'BOUNDARY') |
| 2497 | container.epilogue = '' |
| 2498 | # Set up subparts |
| 2499 | subpart1a = MIMEText('message 1\n') |
| 2500 | subpart2a = MIMEText('message 2\n') |
| 2501 | subpart1 = MIMEMessage(subpart1a) |
| 2502 | subpart2 = MIMEMessage(subpart2a) |
| 2503 | container.attach(subpart1) |
| 2504 | container.attach(subpart2) |
| 2505 | eq(subpart1.get_content_type(), 'message/rfc822') |
| 2506 | eq(subpart1.get_default_type(), 'message/rfc822') |
| 2507 | eq(subpart2.get_content_type(), 'message/rfc822') |
| 2508 | eq(subpart2.get_default_type(), 'message/rfc822') |
| 2509 | neq(container.as_string(0), '''\ |
| 2510 | Content-Type: multipart/digest; boundary="BOUNDARY" |
| 2511 | MIME-Version: 1.0 |
| 2512 | |
| 2513 | --BOUNDARY |
| 2514 | Content-Type: message/rfc822 |
| 2515 | MIME-Version: 1.0 |
| 2516 | |
| 2517 | Content-Type: text/plain; charset="us-ascii" |
| 2518 | MIME-Version: 1.0 |
| 2519 | Content-Transfer-Encoding: 7bit |
| 2520 | |
| 2521 | message 1 |
| 2522 | |
| 2523 | --BOUNDARY |
| 2524 | Content-Type: message/rfc822 |
| 2525 | MIME-Version: 1.0 |
| 2526 | |
| 2527 | Content-Type: text/plain; charset="us-ascii" |
| 2528 | MIME-Version: 1.0 |
| 2529 | Content-Transfer-Encoding: 7bit |
| 2530 | |
| 2531 | message 2 |
| 2532 | |
| 2533 | --BOUNDARY-- |
| 2534 | ''') |
| 2535 | del subpart1['content-type'] |
| 2536 | del subpart1['mime-version'] |
| 2537 | del subpart2['content-type'] |
| 2538 | del subpart2['mime-version'] |
| 2539 | eq(subpart1.get_content_type(), 'message/rfc822') |
| 2540 | eq(subpart1.get_default_type(), 'message/rfc822') |
| 2541 | eq(subpart2.get_content_type(), 'message/rfc822') |
| 2542 | eq(subpart2.get_default_type(), 'message/rfc822') |
| 2543 | neq(container.as_string(0), '''\ |
| 2544 | Content-Type: multipart/digest; boundary="BOUNDARY" |
| 2545 | MIME-Version: 1.0 |
| 2546 | |
| 2547 | --BOUNDARY |
| 2548 | |
| 2549 | Content-Type: text/plain; charset="us-ascii" |
| 2550 | MIME-Version: 1.0 |
| 2551 | Content-Transfer-Encoding: 7bit |
| 2552 | |
| 2553 | message 1 |
| 2554 | |
| 2555 | --BOUNDARY |
| 2556 | |
| 2557 | Content-Type: text/plain; charset="us-ascii" |
| 2558 | MIME-Version: 1.0 |
| 2559 | Content-Transfer-Encoding: 7bit |
| 2560 | |
| 2561 | message 2 |
| 2562 | |
| 2563 | --BOUNDARY-- |
| 2564 | ''') |
| 2565 | |
| 2566 | def test_mime_attachments_in_constructor(self): |
| 2567 | eq = self.assertEqual |
| 2568 | text1 = MIMEText('') |
| 2569 | text2 = MIMEText('') |
| 2570 | msg = MIMEMultipart(_subparts=(text1, text2)) |
| 2571 | eq(len(msg.get_payload()), 2) |
| 2572 | eq(msg.get_payload(0), text1) |
| 2573 | eq(msg.get_payload(1), text2) |
| 2574 | |
Christian Heimes | 587c2bf | 2008-01-19 16:21:02 +0000 | [diff] [blame] | 2575 | def test_default_multipart_constructor(self): |
| 2576 | msg = MIMEMultipart() |
| 2577 | self.assertTrue(msg.is_multipart()) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2578 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 2579 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2580 | # A general test of parser->model->generator idempotency. IOW, read a message |
| 2581 | # in, parse it into a message object tree, then without touching the tree, |
| 2582 | # regenerate the plain text. The original text and the transformed text |
| 2583 | # should be identical. Note: that we ignore the Unix-From since that may |
| 2584 | # contain a changed date. |
| 2585 | class TestIdempotent(TestEmailBase): |
R. David Murray | 719a449 | 2010-11-21 16:53:48 +0000 | [diff] [blame] | 2586 | |
| 2587 | linesep = '\n' |
| 2588 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2589 | def _msgobj(self, filename): |
| 2590 | with openfile(filename) as fp: |
| 2591 | data = fp.read() |
| 2592 | msg = email.message_from_string(data) |
| 2593 | return msg, data |
| 2594 | |
R. David Murray | 719a449 | 2010-11-21 16:53:48 +0000 | [diff] [blame] | 2595 | def _idempotent(self, msg, text, unixfrom=False): |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2596 | eq = self.ndiffAssertEqual |
| 2597 | s = StringIO() |
| 2598 | g = Generator(s, maxheaderlen=0) |
R. David Murray | 719a449 | 2010-11-21 16:53:48 +0000 | [diff] [blame] | 2599 | g.flatten(msg, unixfrom=unixfrom) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2600 | eq(text, s.getvalue()) |
| 2601 | |
| 2602 | def test_parse_text_message(self): |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 2603 | eq = self.assertEqual |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2604 | msg, text = self._msgobj('msg_01.txt') |
| 2605 | eq(msg.get_content_type(), 'text/plain') |
| 2606 | eq(msg.get_content_maintype(), 'text') |
| 2607 | eq(msg.get_content_subtype(), 'plain') |
| 2608 | eq(msg.get_params()[1], ('charset', 'us-ascii')) |
| 2609 | eq(msg.get_param('charset'), 'us-ascii') |
| 2610 | eq(msg.preamble, None) |
| 2611 | eq(msg.epilogue, None) |
| 2612 | self._idempotent(msg, text) |
| 2613 | |
| 2614 | def test_parse_untyped_message(self): |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 2615 | eq = self.assertEqual |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2616 | msg, text = self._msgobj('msg_03.txt') |
| 2617 | eq(msg.get_content_type(), 'text/plain') |
| 2618 | eq(msg.get_params(), None) |
| 2619 | eq(msg.get_param('charset'), None) |
| 2620 | self._idempotent(msg, text) |
| 2621 | |
| 2622 | def test_simple_multipart(self): |
| 2623 | msg, text = self._msgobj('msg_04.txt') |
| 2624 | self._idempotent(msg, text) |
| 2625 | |
| 2626 | def test_MIME_digest(self): |
| 2627 | msg, text = self._msgobj('msg_02.txt') |
| 2628 | self._idempotent(msg, text) |
| 2629 | |
| 2630 | def test_long_header(self): |
| 2631 | msg, text = self._msgobj('msg_27.txt') |
| 2632 | self._idempotent(msg, text) |
| 2633 | |
| 2634 | def test_MIME_digest_with_part_headers(self): |
| 2635 | msg, text = self._msgobj('msg_28.txt') |
| 2636 | self._idempotent(msg, text) |
| 2637 | |
| 2638 | def test_mixed_with_image(self): |
| 2639 | msg, text = self._msgobj('msg_06.txt') |
| 2640 | self._idempotent(msg, text) |
| 2641 | |
| 2642 | def test_multipart_report(self): |
| 2643 | msg, text = self._msgobj('msg_05.txt') |
| 2644 | self._idempotent(msg, text) |
| 2645 | |
| 2646 | def test_dsn(self): |
| 2647 | msg, text = self._msgobj('msg_16.txt') |
| 2648 | self._idempotent(msg, text) |
| 2649 | |
| 2650 | def test_preamble_epilogue(self): |
| 2651 | msg, text = self._msgobj('msg_21.txt') |
| 2652 | self._idempotent(msg, text) |
| 2653 | |
| 2654 | def test_multipart_one_part(self): |
| 2655 | msg, text = self._msgobj('msg_23.txt') |
| 2656 | self._idempotent(msg, text) |
| 2657 | |
| 2658 | def test_multipart_no_parts(self): |
| 2659 | msg, text = self._msgobj('msg_24.txt') |
| 2660 | self._idempotent(msg, text) |
| 2661 | |
| 2662 | def test_no_start_boundary(self): |
| 2663 | msg, text = self._msgobj('msg_31.txt') |
| 2664 | self._idempotent(msg, text) |
| 2665 | |
| 2666 | def test_rfc2231_charset(self): |
| 2667 | msg, text = self._msgobj('msg_32.txt') |
| 2668 | self._idempotent(msg, text) |
| 2669 | |
| 2670 | def test_more_rfc2231_parameters(self): |
| 2671 | msg, text = self._msgobj('msg_33.txt') |
| 2672 | self._idempotent(msg, text) |
| 2673 | |
| 2674 | def test_text_plain_in_a_multipart_digest(self): |
| 2675 | msg, text = self._msgobj('msg_34.txt') |
| 2676 | self._idempotent(msg, text) |
| 2677 | |
| 2678 | def test_nested_multipart_mixeds(self): |
| 2679 | msg, text = self._msgobj('msg_12a.txt') |
| 2680 | self._idempotent(msg, text) |
| 2681 | |
| 2682 | def test_message_external_body_idempotent(self): |
| 2683 | msg, text = self._msgobj('msg_36.txt') |
| 2684 | self._idempotent(msg, text) |
| 2685 | |
R. David Murray | 719a449 | 2010-11-21 16:53:48 +0000 | [diff] [blame] | 2686 | def test_message_delivery_status(self): |
| 2687 | msg, text = self._msgobj('msg_43.txt') |
| 2688 | self._idempotent(msg, text, unixfrom=True) |
| 2689 | |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 2690 | def test_message_signed_idempotent(self): |
| 2691 | msg, text = self._msgobj('msg_45.txt') |
| 2692 | self._idempotent(msg, text) |
| 2693 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2694 | def test_content_type(self): |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 2695 | eq = self.assertEqual |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 2696 | unless = self.assertTrue |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2697 | # Get a message object and reset the seek pointer for other tests |
| 2698 | msg, text = self._msgobj('msg_05.txt') |
| 2699 | eq(msg.get_content_type(), 'multipart/report') |
| 2700 | # Test the Content-Type: parameters |
| 2701 | params = {} |
| 2702 | for pk, pv in msg.get_params(): |
| 2703 | params[pk] = pv |
| 2704 | eq(params['report-type'], 'delivery-status') |
| 2705 | eq(params['boundary'], 'D1690A7AC1.996856090/mail.example.com') |
R. David Murray | 719a449 | 2010-11-21 16:53:48 +0000 | [diff] [blame] | 2706 | eq(msg.preamble, 'This is a MIME-encapsulated message.' + self.linesep) |
| 2707 | eq(msg.epilogue, self.linesep) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2708 | eq(len(msg.get_payload()), 3) |
| 2709 | # Make sure the subparts are what we expect |
| 2710 | msg1 = msg.get_payload(0) |
| 2711 | eq(msg1.get_content_type(), 'text/plain') |
R. David Murray | 719a449 | 2010-11-21 16:53:48 +0000 | [diff] [blame] | 2712 | eq(msg1.get_payload(), 'Yadda yadda yadda' + self.linesep) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2713 | msg2 = msg.get_payload(1) |
| 2714 | eq(msg2.get_content_type(), 'text/plain') |
R. David Murray | 719a449 | 2010-11-21 16:53:48 +0000 | [diff] [blame] | 2715 | eq(msg2.get_payload(), 'Yadda yadda yadda' + self.linesep) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2716 | msg3 = msg.get_payload(2) |
| 2717 | eq(msg3.get_content_type(), 'message/rfc822') |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 2718 | self.assertTrue(isinstance(msg3, Message)) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2719 | payload = msg3.get_payload() |
| 2720 | unless(isinstance(payload, list)) |
| 2721 | eq(len(payload), 1) |
| 2722 | msg4 = payload[0] |
| 2723 | unless(isinstance(msg4, Message)) |
R. David Murray | 719a449 | 2010-11-21 16:53:48 +0000 | [diff] [blame] | 2724 | eq(msg4.get_payload(), 'Yadda yadda yadda' + self.linesep) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2725 | |
| 2726 | def test_parser(self): |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 2727 | eq = self.assertEqual |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 2728 | unless = self.assertTrue |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2729 | msg, text = self._msgobj('msg_06.txt') |
| 2730 | # Check some of the outer headers |
| 2731 | eq(msg.get_content_type(), 'message/rfc822') |
| 2732 | # Make sure the payload is a list of exactly one sub-Message, and that |
| 2733 | # that submessage has a type of text/plain |
| 2734 | payload = msg.get_payload() |
| 2735 | unless(isinstance(payload, list)) |
| 2736 | eq(len(payload), 1) |
| 2737 | msg1 = payload[0] |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 2738 | self.assertTrue(isinstance(msg1, Message)) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2739 | eq(msg1.get_content_type(), 'text/plain') |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 2740 | self.assertTrue(isinstance(msg1.get_payload(), str)) |
R. David Murray | 719a449 | 2010-11-21 16:53:48 +0000 | [diff] [blame] | 2741 | eq(msg1.get_payload(), self.linesep) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2742 | |
| 2743 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 2744 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2745 | # Test various other bits of the package's functionality |
| 2746 | class TestMiscellaneous(TestEmailBase): |
| 2747 | def test_message_from_string(self): |
| 2748 | with openfile('msg_01.txt') as fp: |
| 2749 | text = fp.read() |
| 2750 | msg = email.message_from_string(text) |
| 2751 | s = StringIO() |
| 2752 | # Don't wrap/continue long headers since we're trying to test |
| 2753 | # idempotency. |
| 2754 | g = Generator(s, maxheaderlen=0) |
| 2755 | g.flatten(msg) |
| 2756 | self.assertEqual(text, s.getvalue()) |
| 2757 | |
| 2758 | def test_message_from_file(self): |
| 2759 | with openfile('msg_01.txt') as fp: |
| 2760 | text = fp.read() |
| 2761 | fp.seek(0) |
| 2762 | msg = email.message_from_file(fp) |
| 2763 | s = StringIO() |
| 2764 | # Don't wrap/continue long headers since we're trying to test |
| 2765 | # idempotency. |
| 2766 | g = Generator(s, maxheaderlen=0) |
| 2767 | g.flatten(msg) |
| 2768 | self.assertEqual(text, s.getvalue()) |
| 2769 | |
| 2770 | def test_message_from_string_with_class(self): |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 2771 | unless = self.assertTrue |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2772 | with openfile('msg_01.txt') as fp: |
| 2773 | text = fp.read() |
| 2774 | |
| 2775 | # Create a subclass |
| 2776 | class MyMessage(Message): |
| 2777 | pass |
| 2778 | |
| 2779 | msg = email.message_from_string(text, MyMessage) |
| 2780 | unless(isinstance(msg, MyMessage)) |
| 2781 | # Try something more complicated |
| 2782 | with openfile('msg_02.txt') as fp: |
| 2783 | text = fp.read() |
| 2784 | msg = email.message_from_string(text, MyMessage) |
| 2785 | for subpart in msg.walk(): |
| 2786 | unless(isinstance(subpart, MyMessage)) |
| 2787 | |
| 2788 | def test_message_from_file_with_class(self): |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 2789 | unless = self.assertTrue |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2790 | # Create a subclass |
| 2791 | class MyMessage(Message): |
| 2792 | pass |
| 2793 | |
| 2794 | with openfile('msg_01.txt') as fp: |
| 2795 | msg = email.message_from_file(fp, MyMessage) |
| 2796 | unless(isinstance(msg, MyMessage)) |
| 2797 | # Try something more complicated |
| 2798 | with openfile('msg_02.txt') as fp: |
| 2799 | msg = email.message_from_file(fp, MyMessage) |
| 2800 | for subpart in msg.walk(): |
| 2801 | unless(isinstance(subpart, MyMessage)) |
| 2802 | |
R David Murray | c27e522 | 2012-05-25 15:01:48 -0400 | [diff] [blame] | 2803 | def test_custom_message_does_not_require_arguments(self): |
| 2804 | class MyMessage(Message): |
| 2805 | def __init__(self): |
| 2806 | super().__init__() |
| 2807 | msg = self._str_msg("Subject: test\n\ntest", MyMessage) |
| 2808 | self.assertTrue(isinstance(msg, MyMessage)) |
| 2809 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2810 | def test__all__(self): |
| 2811 | module = __import__('email') |
R David Murray | 1b6c724 | 2012-03-16 22:43:05 -0400 | [diff] [blame] | 2812 | self.assertEqual(sorted(module.__all__), [ |
| 2813 | 'base64mime', 'charset', 'encoders', 'errors', 'feedparser', |
| 2814 | 'generator', 'header', 'iterators', 'message', |
| 2815 | 'message_from_binary_file', 'message_from_bytes', |
| 2816 | 'message_from_file', 'message_from_string', 'mime', 'parser', |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2817 | 'quoprimime', 'utils', |
| 2818 | ]) |
| 2819 | |
| 2820 | def test_formatdate(self): |
| 2821 | now = time.time() |
| 2822 | self.assertEqual(utils.parsedate(utils.formatdate(now))[:6], |
| 2823 | time.gmtime(now)[:6]) |
| 2824 | |
| 2825 | def test_formatdate_localtime(self): |
| 2826 | now = time.time() |
| 2827 | self.assertEqual( |
| 2828 | utils.parsedate(utils.formatdate(now, localtime=True))[:6], |
| 2829 | time.localtime(now)[:6]) |
| 2830 | |
| 2831 | def test_formatdate_usegmt(self): |
| 2832 | now = time.time() |
| 2833 | self.assertEqual( |
| 2834 | utils.formatdate(now, localtime=False), |
| 2835 | time.strftime('%a, %d %b %Y %H:%M:%S -0000', time.gmtime(now))) |
| 2836 | self.assertEqual( |
| 2837 | utils.formatdate(now, localtime=False, usegmt=True), |
| 2838 | time.strftime('%a, %d %b %Y %H:%M:%S GMT', time.gmtime(now))) |
| 2839 | |
Georg Brandl | 1aca31e | 2012-09-22 09:03:56 +0200 | [diff] [blame] | 2840 | # parsedate and parsedate_tz will become deprecated interfaces someday |
| 2841 | def test_parsedate_returns_None_for_invalid_strings(self): |
| 2842 | self.assertIsNone(utils.parsedate('')) |
| 2843 | self.assertIsNone(utils.parsedate_tz('')) |
| 2844 | self.assertIsNone(utils.parsedate('0')) |
| 2845 | self.assertIsNone(utils.parsedate_tz('0')) |
| 2846 | self.assertIsNone(utils.parsedate('A Complete Waste of Time')) |
| 2847 | self.assertIsNone(utils.parsedate_tz('A Complete Waste of Time')) |
| 2848 | # Not a part of the spec but, but this has historically worked: |
| 2849 | self.assertIsNone(utils.parsedate(None)) |
| 2850 | self.assertIsNone(utils.parsedate_tz(None)) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2851 | |
| 2852 | def test_parsedate_compact(self): |
| 2853 | # The FWS after the comma is optional |
| 2854 | self.assertEqual(utils.parsedate('Wed,3 Apr 2002 14:58:26 +0800'), |
| 2855 | utils.parsedate('Wed, 3 Apr 2002 14:58:26 +0800')) |
| 2856 | |
| 2857 | def test_parsedate_no_dayofweek(self): |
| 2858 | eq = self.assertEqual |
| 2859 | eq(utils.parsedate_tz('25 Feb 2003 13:47:26 -0800'), |
| 2860 | (2003, 2, 25, 13, 47, 26, 0, 1, -1, -28800)) |
| 2861 | |
| 2862 | def test_parsedate_compact_no_dayofweek(self): |
| 2863 | eq = self.assertEqual |
| 2864 | eq(utils.parsedate_tz('5 Feb 2003 13:47:26 -0800'), |
| 2865 | (2003, 2, 5, 13, 47, 26, 0, 1, -1, -28800)) |
| 2866 | |
R. David Murray | 4a62e89 | 2010-12-23 20:35:46 +0000 | [diff] [blame] | 2867 | def test_parsedate_no_space_before_positive_offset(self): |
| 2868 | self.assertEqual(utils.parsedate_tz('Wed, 3 Apr 2002 14:58:26+0800'), |
| 2869 | (2002, 4, 3, 14, 58, 26, 0, 1, -1, 28800)) |
| 2870 | |
| 2871 | def test_parsedate_no_space_before_negative_offset(self): |
| 2872 | # Issue 1155362: we already handled '+' for this case. |
| 2873 | self.assertEqual(utils.parsedate_tz('Wed, 3 Apr 2002 14:58:26-0800'), |
| 2874 | (2002, 4, 3, 14, 58, 26, 0, 1, -1, -28800)) |
| 2875 | |
| 2876 | |
R David Murray | accd1c0 | 2011-03-13 20:06:23 -0400 | [diff] [blame] | 2877 | def test_parsedate_accepts_time_with_dots(self): |
| 2878 | eq = self.assertEqual |
| 2879 | eq(utils.parsedate_tz('5 Feb 2003 13.47.26 -0800'), |
| 2880 | (2003, 2, 5, 13, 47, 26, 0, 1, -1, -28800)) |
| 2881 | eq(utils.parsedate_tz('5 Feb 2003 13.47 -0800'), |
| 2882 | (2003, 2, 5, 13, 47, 0, 0, 1, -1, -28800)) |
| 2883 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2884 | def test_parsedate_acceptable_to_time_functions(self): |
| 2885 | eq = self.assertEqual |
| 2886 | timetup = utils.parsedate('5 Feb 2003 13:47:26 -0800') |
| 2887 | t = int(time.mktime(timetup)) |
| 2888 | eq(time.localtime(t)[:6], timetup[:6]) |
| 2889 | eq(int(time.strftime('%Y', timetup)), 2003) |
| 2890 | timetup = utils.parsedate_tz('5 Feb 2003 13:47:26 -0800') |
| 2891 | t = int(time.mktime(timetup[:9])) |
| 2892 | eq(time.localtime(t)[:6], timetup[:6]) |
| 2893 | eq(int(time.strftime('%Y', timetup[:9])), 2003) |
| 2894 | |
Alexander Belopolsky | a07548e | 2012-06-21 20:34:09 -0400 | [diff] [blame] | 2895 | def test_mktime_tz(self): |
| 2896 | self.assertEqual(utils.mktime_tz((1970, 1, 1, 0, 0, 0, |
| 2897 | -1, -1, -1, 0)), 0) |
| 2898 | self.assertEqual(utils.mktime_tz((1970, 1, 1, 0, 0, 0, |
| 2899 | -1, -1, -1, 1234)), -1234) |
| 2900 | |
R. David Murray | 219d1c8 | 2010-08-25 00:45:55 +0000 | [diff] [blame] | 2901 | def test_parsedate_y2k(self): |
| 2902 | """Test for parsing a date with a two-digit year. |
| 2903 | |
| 2904 | Parsing a date with a two-digit year should return the correct |
| 2905 | four-digit year. RFC822 allows two-digit years, but RFC2822 (which |
| 2906 | obsoletes RFC822) requires four-digit years. |
| 2907 | |
| 2908 | """ |
| 2909 | self.assertEqual(utils.parsedate_tz('25 Feb 03 13:47:26 -0800'), |
| 2910 | utils.parsedate_tz('25 Feb 2003 13:47:26 -0800')) |
| 2911 | self.assertEqual(utils.parsedate_tz('25 Feb 71 13:47:26 -0800'), |
| 2912 | utils.parsedate_tz('25 Feb 1971 13:47:26 -0800')) |
| 2913 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2914 | def test_parseaddr_empty(self): |
| 2915 | self.assertEqual(utils.parseaddr('<>'), ('', '')) |
| 2916 | self.assertEqual(utils.formataddr(utils.parseaddr('<>')), '') |
| 2917 | |
| 2918 | def test_noquote_dump(self): |
| 2919 | self.assertEqual( |
| 2920 | utils.formataddr(('A Silly Person', 'person@dom.ain')), |
| 2921 | 'A Silly Person <person@dom.ain>') |
| 2922 | |
| 2923 | def test_escape_dump(self): |
| 2924 | self.assertEqual( |
| 2925 | utils.formataddr(('A (Very) Silly Person', 'person@dom.ain')), |
R David Murray | b53319f | 2012-03-14 15:31:47 -0400 | [diff] [blame] | 2926 | r'"A (Very) Silly Person" <person@dom.ain>') |
| 2927 | self.assertEqual( |
| 2928 | utils.parseaddr(r'"A \(Very\) Silly Person" <person@dom.ain>'), |
| 2929 | ('A (Very) Silly Person', 'person@dom.ain')) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2930 | a = r'A \(Special\) Person' |
| 2931 | b = 'person@dom.ain' |
| 2932 | self.assertEqual(utils.parseaddr(utils.formataddr((a, b))), (a, b)) |
| 2933 | |
| 2934 | def test_escape_backslashes(self): |
| 2935 | self.assertEqual( |
| 2936 | utils.formataddr(('Arthur \Backslash\ Foobar', 'person@dom.ain')), |
| 2937 | r'"Arthur \\Backslash\\ Foobar" <person@dom.ain>') |
| 2938 | a = r'Arthur \Backslash\ Foobar' |
| 2939 | b = 'person@dom.ain' |
| 2940 | self.assertEqual(utils.parseaddr(utils.formataddr((a, b))), (a, b)) |
| 2941 | |
R David Murray | 8debacb | 2011-04-06 09:35:57 -0400 | [diff] [blame] | 2942 | def test_quotes_unicode_names(self): |
| 2943 | # issue 1690608. email.utils.formataddr() should be rfc2047 aware. |
| 2944 | name = "H\u00e4ns W\u00fcrst" |
| 2945 | addr = 'person@dom.ain' |
| 2946 | utf8_base64 = "=?utf-8?b?SMOkbnMgV8O8cnN0?= <person@dom.ain>" |
| 2947 | latin1_quopri = "=?iso-8859-1?q?H=E4ns_W=FCrst?= <person@dom.ain>" |
| 2948 | self.assertEqual(utils.formataddr((name, addr)), utf8_base64) |
| 2949 | self.assertEqual(utils.formataddr((name, addr), 'iso-8859-1'), |
| 2950 | latin1_quopri) |
| 2951 | |
| 2952 | def test_accepts_any_charset_like_object(self): |
| 2953 | # issue 1690608. email.utils.formataddr() should be rfc2047 aware. |
| 2954 | name = "H\u00e4ns W\u00fcrst" |
| 2955 | addr = 'person@dom.ain' |
| 2956 | utf8_base64 = "=?utf-8?b?SMOkbnMgV8O8cnN0?= <person@dom.ain>" |
| 2957 | foobar = "FOOBAR" |
| 2958 | class CharsetMock: |
| 2959 | def header_encode(self, string): |
| 2960 | return foobar |
| 2961 | mock = CharsetMock() |
| 2962 | mock_expected = "%s <%s>" % (foobar, addr) |
| 2963 | self.assertEqual(utils.formataddr((name, addr), mock), mock_expected) |
| 2964 | self.assertEqual(utils.formataddr((name, addr), Charset('utf-8')), |
| 2965 | utf8_base64) |
| 2966 | |
| 2967 | def test_invalid_charset_like_object_raises_error(self): |
| 2968 | # issue 1690608. email.utils.formataddr() should be rfc2047 aware. |
| 2969 | name = "H\u00e4ns W\u00fcrst" |
| 2970 | addr = 'person@dom.ain' |
| 2971 | # A object without a header_encode method: |
| 2972 | bad_charset = object() |
| 2973 | self.assertRaises(AttributeError, utils.formataddr, (name, addr), |
| 2974 | bad_charset) |
| 2975 | |
| 2976 | def test_unicode_address_raises_error(self): |
| 2977 | # issue 1690608. email.utils.formataddr() should be rfc2047 aware. |
| 2978 | addr = 'pers\u00f6n@dom.in' |
| 2979 | self.assertRaises(UnicodeError, utils.formataddr, (None, addr)) |
| 2980 | self.assertRaises(UnicodeError, utils.formataddr, ("Name", addr)) |
| 2981 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 2982 | def test_name_with_dot(self): |
| 2983 | x = 'John X. Doe <jxd@example.com>' |
| 2984 | y = '"John X. Doe" <jxd@example.com>' |
| 2985 | a, b = ('John X. Doe', 'jxd@example.com') |
| 2986 | self.assertEqual(utils.parseaddr(x), (a, b)) |
| 2987 | self.assertEqual(utils.parseaddr(y), (a, b)) |
| 2988 | # formataddr() quotes the name if there's a dot in it |
| 2989 | self.assertEqual(utils.formataddr((a, b)), y) |
| 2990 | |
R. David Murray | 5397e86 | 2010-10-02 15:58:26 +0000 | [diff] [blame] | 2991 | def test_parseaddr_preserves_quoted_pairs_in_addresses(self): |
| 2992 | # issue 10005. Note that in the third test the second pair of |
| 2993 | # backslashes is not actually a quoted pair because it is not inside a |
| 2994 | # comment or quoted string: the address being parsed has a quoted |
| 2995 | # string containing a quoted backslash, followed by 'example' and two |
| 2996 | # backslashes, followed by another quoted string containing a space and |
| 2997 | # the word 'example'. parseaddr copies those two backslashes |
| 2998 | # literally. Per rfc5322 this is not technically correct since a \ may |
| 2999 | # not appear in an address outside of a quoted string. It is probably |
| 3000 | # a sensible Postel interpretation, though. |
| 3001 | eq = self.assertEqual |
| 3002 | eq(utils.parseaddr('""example" example"@example.com'), |
| 3003 | ('', '""example" example"@example.com')) |
| 3004 | eq(utils.parseaddr('"\\"example\\" example"@example.com'), |
| 3005 | ('', '"\\"example\\" example"@example.com')) |
| 3006 | eq(utils.parseaddr('"\\\\"example\\\\" example"@example.com'), |
| 3007 | ('', '"\\\\"example\\\\" example"@example.com')) |
| 3008 | |
R. David Murray | 63563cd | 2010-12-18 18:25:38 +0000 | [diff] [blame] | 3009 | def test_parseaddr_preserves_spaces_in_local_part(self): |
| 3010 | # issue 9286. A normal RFC5322 local part should not contain any |
| 3011 | # folding white space, but legacy local parts can (they are a sequence |
| 3012 | # of atoms, not dotatoms). On the other hand we strip whitespace from |
| 3013 | # before the @ and around dots, on the assumption that the whitespace |
| 3014 | # around the punctuation is a mistake in what would otherwise be |
| 3015 | # an RFC5322 local part. Leading whitespace is, usual, stripped as well. |
| 3016 | self.assertEqual(('', "merwok wok@xample.com"), |
| 3017 | utils.parseaddr("merwok wok@xample.com")) |
| 3018 | self.assertEqual(('', "merwok wok@xample.com"), |
| 3019 | utils.parseaddr("merwok wok@xample.com")) |
| 3020 | self.assertEqual(('', "merwok wok@xample.com"), |
| 3021 | utils.parseaddr(" merwok wok @xample.com")) |
| 3022 | self.assertEqual(('', 'merwok"wok" wok@xample.com'), |
| 3023 | utils.parseaddr('merwok"wok" wok@xample.com')) |
| 3024 | self.assertEqual(('', 'merwok.wok.wok@xample.com'), |
| 3025 | utils.parseaddr('merwok. wok . wok@xample.com')) |
| 3026 | |
R David Murray | b53319f | 2012-03-14 15:31:47 -0400 | [diff] [blame] | 3027 | def test_formataddr_does_not_quote_parens_in_quoted_string(self): |
| 3028 | addr = ("'foo@example.com' (foo@example.com)", |
| 3029 | 'foo@example.com') |
| 3030 | addrstr = ('"\'foo@example.com\' ' |
| 3031 | '(foo@example.com)" <foo@example.com>') |
| 3032 | self.assertEqual(utils.parseaddr(addrstr), addr) |
| 3033 | self.assertEqual(utils.formataddr(addr), addrstr) |
| 3034 | |
| 3035 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3036 | def test_multiline_from_comment(self): |
| 3037 | x = """\ |
| 3038 | Foo |
| 3039 | \tBar <foo@example.com>""" |
| 3040 | self.assertEqual(utils.parseaddr(x), ('Foo Bar', 'foo@example.com')) |
| 3041 | |
| 3042 | def test_quote_dump(self): |
| 3043 | self.assertEqual( |
| 3044 | utils.formataddr(('A Silly; Person', 'person@dom.ain')), |
| 3045 | r'"A Silly; Person" <person@dom.ain>') |
| 3046 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3047 | def test_charset_richcomparisons(self): |
| 3048 | eq = self.assertEqual |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 3049 | ne = self.assertNotEqual |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3050 | cset1 = Charset() |
| 3051 | cset2 = Charset() |
| 3052 | eq(cset1, 'us-ascii') |
| 3053 | eq(cset1, 'US-ASCII') |
| 3054 | eq(cset1, 'Us-AsCiI') |
| 3055 | eq('us-ascii', cset1) |
| 3056 | eq('US-ASCII', cset1) |
| 3057 | eq('Us-AsCiI', cset1) |
| 3058 | ne(cset1, 'usascii') |
| 3059 | ne(cset1, 'USASCII') |
| 3060 | ne(cset1, 'UsAsCiI') |
| 3061 | ne('usascii', cset1) |
| 3062 | ne('USASCII', cset1) |
| 3063 | ne('UsAsCiI', cset1) |
| 3064 | eq(cset1, cset2) |
| 3065 | eq(cset2, cset1) |
| 3066 | |
| 3067 | def test_getaddresses(self): |
| 3068 | eq = self.assertEqual |
| 3069 | eq(utils.getaddresses(['aperson@dom.ain (Al Person)', |
| 3070 | 'Bud Person <bperson@dom.ain>']), |
| 3071 | [('Al Person', 'aperson@dom.ain'), |
| 3072 | ('Bud Person', 'bperson@dom.ain')]) |
| 3073 | |
| 3074 | def test_getaddresses_nasty(self): |
| 3075 | eq = self.assertEqual |
| 3076 | eq(utils.getaddresses(['foo: ;']), [('', '')]) |
| 3077 | eq(utils.getaddresses( |
| 3078 | ['[]*-- =~$']), |
| 3079 | [('', ''), ('', ''), ('', '*--')]) |
| 3080 | eq(utils.getaddresses( |
| 3081 | ['foo: ;', '"Jason R. Mastaler" <jason@dom.ain>']), |
| 3082 | [('', ''), ('Jason R. Mastaler', 'jason@dom.ain')]) |
| 3083 | |
| 3084 | def test_getaddresses_embedded_comment(self): |
| 3085 | """Test proper handling of a nested comment""" |
| 3086 | eq = self.assertEqual |
| 3087 | addrs = utils.getaddresses(['User ((nested comment)) <foo@bar.com>']) |
| 3088 | eq(addrs[0][1], 'foo@bar.com') |
| 3089 | |
| 3090 | def test_utils_quote_unquote(self): |
| 3091 | eq = self.assertEqual |
| 3092 | msg = Message() |
| 3093 | msg.add_header('content-disposition', 'attachment', |
| 3094 | filename='foo\\wacky"name') |
| 3095 | eq(msg.get_filename(), 'foo\\wacky"name') |
| 3096 | |
| 3097 | def test_get_body_encoding_with_bogus_charset(self): |
| 3098 | charset = Charset('not a charset') |
| 3099 | self.assertEqual(charset.get_body_encoding(), 'base64') |
| 3100 | |
| 3101 | def test_get_body_encoding_with_uppercase_charset(self): |
| 3102 | eq = self.assertEqual |
| 3103 | msg = Message() |
| 3104 | msg['Content-Type'] = 'text/plain; charset=UTF-8' |
| 3105 | eq(msg['content-type'], 'text/plain; charset=UTF-8') |
| 3106 | charsets = msg.get_charsets() |
| 3107 | eq(len(charsets), 1) |
| 3108 | eq(charsets[0], 'utf-8') |
| 3109 | charset = Charset(charsets[0]) |
| 3110 | eq(charset.get_body_encoding(), 'base64') |
Martin v. Löwis | 15b16a3 | 2008-12-02 06:00:15 +0000 | [diff] [blame] | 3111 | msg.set_payload(b'hello world', charset=charset) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3112 | eq(msg.get_payload(), 'aGVsbG8gd29ybGQ=\n') |
| 3113 | eq(msg.get_payload(decode=True), b'hello world') |
| 3114 | eq(msg['content-transfer-encoding'], 'base64') |
| 3115 | # Try another one |
| 3116 | msg = Message() |
| 3117 | msg['Content-Type'] = 'text/plain; charset="US-ASCII"' |
| 3118 | charsets = msg.get_charsets() |
| 3119 | eq(len(charsets), 1) |
| 3120 | eq(charsets[0], 'us-ascii') |
| 3121 | charset = Charset(charsets[0]) |
| 3122 | eq(charset.get_body_encoding(), encoders.encode_7or8bit) |
| 3123 | msg.set_payload('hello world', charset=charset) |
| 3124 | eq(msg.get_payload(), 'hello world') |
| 3125 | eq(msg['content-transfer-encoding'], '7bit') |
| 3126 | |
| 3127 | def test_charsets_case_insensitive(self): |
| 3128 | lc = Charset('us-ascii') |
| 3129 | uc = Charset('US-ASCII') |
| 3130 | self.assertEqual(lc.get_body_encoding(), uc.get_body_encoding()) |
| 3131 | |
| 3132 | def test_partial_falls_inside_message_delivery_status(self): |
| 3133 | eq = self.ndiffAssertEqual |
| 3134 | # The Parser interface provides chunks of data to FeedParser in 8192 |
| 3135 | # byte gulps. SF bug #1076485 found one of those chunks inside |
| 3136 | # message/delivery-status header block, which triggered an |
| 3137 | # unreadline() of NeedMoreData. |
| 3138 | msg = self._msgobj('msg_43.txt') |
| 3139 | sfp = StringIO() |
| 3140 | iterators._structure(msg, sfp) |
| 3141 | eq(sfp.getvalue(), """\ |
| 3142 | multipart/report |
| 3143 | text/plain |
| 3144 | message/delivery-status |
| 3145 | text/plain |
| 3146 | text/plain |
| 3147 | text/plain |
| 3148 | text/plain |
| 3149 | text/plain |
| 3150 | text/plain |
| 3151 | text/plain |
| 3152 | text/plain |
| 3153 | text/plain |
| 3154 | text/plain |
| 3155 | text/plain |
| 3156 | text/plain |
| 3157 | text/plain |
| 3158 | text/plain |
| 3159 | text/plain |
| 3160 | text/plain |
| 3161 | text/plain |
| 3162 | text/plain |
| 3163 | text/plain |
| 3164 | text/plain |
| 3165 | text/plain |
| 3166 | text/plain |
| 3167 | text/plain |
| 3168 | text/plain |
| 3169 | text/plain |
| 3170 | text/plain |
| 3171 | text/rfc822-headers |
| 3172 | """) |
| 3173 | |
R. David Murray | a0b44b5 | 2010-12-02 21:47:19 +0000 | [diff] [blame] | 3174 | def test_make_msgid_domain(self): |
| 3175 | self.assertEqual( |
| 3176 | email.utils.make_msgid(domain='testdomain-string')[-19:], |
| 3177 | '@testdomain-string>') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3178 | |
R David Murray | e67c6c5 | 2013-03-07 16:38:03 -0500 | [diff] [blame] | 3179 | def test_Generator_linend(self): |
| 3180 | # Issue 14645. |
| 3181 | with openfile('msg_26.txt', newline='\n') as f: |
| 3182 | msgtxt = f.read() |
| 3183 | msgtxt_nl = msgtxt.replace('\r\n', '\n') |
| 3184 | msg = email.message_from_string(msgtxt) |
| 3185 | s = StringIO() |
| 3186 | g = email.generator.Generator(s) |
| 3187 | g.flatten(msg) |
| 3188 | self.assertEqual(s.getvalue(), msgtxt_nl) |
| 3189 | |
| 3190 | def test_BytesGenerator_linend(self): |
| 3191 | # Issue 14645. |
| 3192 | with openfile('msg_26.txt', newline='\n') as f: |
| 3193 | msgtxt = f.read() |
| 3194 | msgtxt_nl = msgtxt.replace('\r\n', '\n') |
| 3195 | msg = email.message_from_string(msgtxt_nl) |
| 3196 | s = BytesIO() |
| 3197 | g = email.generator.BytesGenerator(s) |
| 3198 | g.flatten(msg, linesep='\r\n') |
| 3199 | self.assertEqual(s.getvalue().decode('ascii'), msgtxt) |
| 3200 | |
| 3201 | def test_BytesGenerator_linend_with_non_ascii(self): |
| 3202 | # Issue 14645. |
| 3203 | with openfile('msg_26.txt', 'rb') as f: |
| 3204 | msgtxt = f.read() |
| 3205 | msgtxt = msgtxt.replace(b'with attachment', b'fo\xf6') |
| 3206 | msgtxt_nl = msgtxt.replace(b'\r\n', b'\n') |
| 3207 | msg = email.message_from_bytes(msgtxt_nl) |
| 3208 | s = BytesIO() |
| 3209 | g = email.generator.BytesGenerator(s) |
| 3210 | g.flatten(msg, linesep='\r\n') |
| 3211 | self.assertEqual(s.getvalue(), msgtxt) |
| 3212 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 3213 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3214 | # Test the iterator/generators |
| 3215 | class TestIterators(TestEmailBase): |
| 3216 | def test_body_line_iterator(self): |
| 3217 | eq = self.assertEqual |
| 3218 | neq = self.ndiffAssertEqual |
| 3219 | # First a simple non-multipart message |
| 3220 | msg = self._msgobj('msg_01.txt') |
| 3221 | it = iterators.body_line_iterator(msg) |
| 3222 | lines = list(it) |
| 3223 | eq(len(lines), 6) |
| 3224 | neq(EMPTYSTRING.join(lines), msg.get_payload()) |
| 3225 | # Now a more complicated multipart |
| 3226 | msg = self._msgobj('msg_02.txt') |
| 3227 | it = iterators.body_line_iterator(msg) |
| 3228 | lines = list(it) |
| 3229 | eq(len(lines), 43) |
| 3230 | with openfile('msg_19.txt') as fp: |
| 3231 | neq(EMPTYSTRING.join(lines), fp.read()) |
| 3232 | |
| 3233 | def test_typed_subpart_iterator(self): |
| 3234 | eq = self.assertEqual |
| 3235 | msg = self._msgobj('msg_04.txt') |
| 3236 | it = iterators.typed_subpart_iterator(msg, 'text') |
| 3237 | lines = [] |
| 3238 | subparts = 0 |
| 3239 | for subpart in it: |
| 3240 | subparts += 1 |
| 3241 | lines.append(subpart.get_payload()) |
| 3242 | eq(subparts, 2) |
| 3243 | eq(EMPTYSTRING.join(lines), """\ |
| 3244 | a simple kind of mirror |
| 3245 | to reflect upon our own |
| 3246 | a simple kind of mirror |
| 3247 | to reflect upon our own |
| 3248 | """) |
| 3249 | |
| 3250 | def test_typed_subpart_iterator_default_type(self): |
| 3251 | eq = self.assertEqual |
| 3252 | msg = self._msgobj('msg_03.txt') |
| 3253 | it = iterators.typed_subpart_iterator(msg, 'text', 'plain') |
| 3254 | lines = [] |
| 3255 | subparts = 0 |
| 3256 | for subpart in it: |
| 3257 | subparts += 1 |
| 3258 | lines.append(subpart.get_payload()) |
| 3259 | eq(subparts, 1) |
| 3260 | eq(EMPTYSTRING.join(lines), """\ |
| 3261 | |
| 3262 | Hi, |
| 3263 | |
| 3264 | Do you like this message? |
| 3265 | |
| 3266 | -Me |
| 3267 | """) |
| 3268 | |
R. David Murray | 45bf773f | 2010-07-17 01:19:57 +0000 | [diff] [blame] | 3269 | def test_pushCR_LF(self): |
| 3270 | '''FeedParser BufferedSubFile.push() assumed it received complete |
| 3271 | line endings. A CR ending one push() followed by a LF starting |
| 3272 | the next push() added an empty line. |
| 3273 | ''' |
| 3274 | imt = [ |
| 3275 | ("a\r \n", 2), |
| 3276 | ("b", 0), |
| 3277 | ("c\n", 1), |
| 3278 | ("", 0), |
| 3279 | ("d\r\n", 1), |
| 3280 | ("e\r", 0), |
| 3281 | ("\nf", 1), |
| 3282 | ("\r\n", 1), |
| 3283 | ] |
| 3284 | from email.feedparser import BufferedSubFile, NeedMoreData |
| 3285 | bsf = BufferedSubFile() |
| 3286 | om = [] |
| 3287 | nt = 0 |
| 3288 | for il, n in imt: |
| 3289 | bsf.push(il) |
| 3290 | nt += n |
| 3291 | n1 = 0 |
| 3292 | while True: |
| 3293 | ol = bsf.readline() |
| 3294 | if ol == NeedMoreData: |
| 3295 | break |
| 3296 | om.append(ol) |
| 3297 | n1 += 1 |
| 3298 | self.assertTrue(n == n1) |
| 3299 | self.assertTrue(len(om) == nt) |
| 3300 | self.assertTrue(''.join([il for il, n in imt]) == ''.join(om)) |
| 3301 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3302 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 3303 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3304 | class TestParsers(TestEmailBase): |
R David Murray | b35c850 | 2011-04-13 16:46:05 -0400 | [diff] [blame] | 3305 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3306 | def test_header_parser(self): |
| 3307 | eq = self.assertEqual |
| 3308 | # Parse only the headers of a complex multipart MIME document |
| 3309 | with openfile('msg_02.txt') as fp: |
| 3310 | msg = HeaderParser().parse(fp) |
| 3311 | eq(msg['from'], 'ppp-request@zzz.org') |
| 3312 | eq(msg['to'], 'ppp@zzz.org') |
| 3313 | eq(msg.get_content_type(), 'multipart/mixed') |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 3314 | self.assertFalse(msg.is_multipart()) |
| 3315 | self.assertTrue(isinstance(msg.get_payload(), str)) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3316 | |
R David Murray | b35c850 | 2011-04-13 16:46:05 -0400 | [diff] [blame] | 3317 | def test_bytes_header_parser(self): |
| 3318 | eq = self.assertEqual |
| 3319 | # Parse only the headers of a complex multipart MIME document |
| 3320 | with openfile('msg_02.txt', 'rb') as fp: |
| 3321 | msg = email.parser.BytesHeaderParser().parse(fp) |
| 3322 | eq(msg['from'], 'ppp-request@zzz.org') |
| 3323 | eq(msg['to'], 'ppp@zzz.org') |
| 3324 | eq(msg.get_content_type(), 'multipart/mixed') |
| 3325 | self.assertFalse(msg.is_multipart()) |
| 3326 | self.assertTrue(isinstance(msg.get_payload(), str)) |
| 3327 | self.assertTrue(isinstance(msg.get_payload(decode=True), bytes)) |
| 3328 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3329 | def test_whitespace_continuation(self): |
| 3330 | eq = self.assertEqual |
| 3331 | # This message contains a line after the Subject: header that has only |
| 3332 | # whitespace, but it is not empty! |
| 3333 | msg = email.message_from_string("""\ |
| 3334 | From: aperson@dom.ain |
| 3335 | To: bperson@dom.ain |
| 3336 | Subject: the next line has a space on it |
| 3337 | \x20 |
| 3338 | Date: Mon, 8 Apr 2002 15:09:19 -0400 |
| 3339 | Message-ID: spam |
| 3340 | |
| 3341 | Here's the message body |
| 3342 | """) |
| 3343 | eq(msg['subject'], 'the next line has a space on it\n ') |
| 3344 | eq(msg['message-id'], 'spam') |
| 3345 | eq(msg.get_payload(), "Here's the message body\n") |
| 3346 | |
| 3347 | def test_whitespace_continuation_last_header(self): |
| 3348 | eq = self.assertEqual |
| 3349 | # Like the previous test, but the subject line is the last |
| 3350 | # header. |
| 3351 | msg = email.message_from_string("""\ |
| 3352 | From: aperson@dom.ain |
| 3353 | To: bperson@dom.ain |
| 3354 | Date: Mon, 8 Apr 2002 15:09:19 -0400 |
| 3355 | Message-ID: spam |
| 3356 | Subject: the next line has a space on it |
| 3357 | \x20 |
| 3358 | |
| 3359 | Here's the message body |
| 3360 | """) |
| 3361 | eq(msg['subject'], 'the next line has a space on it\n ') |
| 3362 | eq(msg['message-id'], 'spam') |
| 3363 | eq(msg.get_payload(), "Here's the message body\n") |
| 3364 | |
| 3365 | def test_crlf_separation(self): |
| 3366 | eq = self.assertEqual |
Guido van Rossum | 98297ee | 2007-11-06 21:34:58 +0000 | [diff] [blame] | 3367 | with openfile('msg_26.txt', newline='\n') as fp: |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3368 | msg = Parser().parse(fp) |
| 3369 | eq(len(msg.get_payload()), 2) |
| 3370 | part1 = msg.get_payload(0) |
| 3371 | eq(part1.get_content_type(), 'text/plain') |
| 3372 | eq(part1.get_payload(), 'Simple email with attachment.\r\n\r\n') |
| 3373 | part2 = msg.get_payload(1) |
| 3374 | eq(part2.get_content_type(), 'application/riscos') |
| 3375 | |
R. David Murray | 8451c4b | 2010-10-23 22:19:56 +0000 | [diff] [blame] | 3376 | def test_crlf_flatten(self): |
| 3377 | # Using newline='\n' preserves the crlfs in this input file. |
| 3378 | with openfile('msg_26.txt', newline='\n') as fp: |
| 3379 | text = fp.read() |
| 3380 | msg = email.message_from_string(text) |
| 3381 | s = StringIO() |
| 3382 | g = Generator(s) |
| 3383 | g.flatten(msg, linesep='\r\n') |
| 3384 | self.assertEqual(s.getvalue(), text) |
| 3385 | |
| 3386 | maxDiff = None |
| 3387 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3388 | def test_multipart_digest_with_extra_mime_headers(self): |
| 3389 | eq = self.assertEqual |
| 3390 | neq = self.ndiffAssertEqual |
| 3391 | with openfile('msg_28.txt') as fp: |
| 3392 | msg = email.message_from_file(fp) |
| 3393 | # Structure is: |
| 3394 | # multipart/digest |
| 3395 | # message/rfc822 |
| 3396 | # text/plain |
| 3397 | # message/rfc822 |
| 3398 | # text/plain |
| 3399 | eq(msg.is_multipart(), 1) |
| 3400 | eq(len(msg.get_payload()), 2) |
| 3401 | part1 = msg.get_payload(0) |
| 3402 | eq(part1.get_content_type(), 'message/rfc822') |
| 3403 | eq(part1.is_multipart(), 1) |
| 3404 | eq(len(part1.get_payload()), 1) |
| 3405 | part1a = part1.get_payload(0) |
| 3406 | eq(part1a.is_multipart(), 0) |
| 3407 | eq(part1a.get_content_type(), 'text/plain') |
| 3408 | neq(part1a.get_payload(), 'message 1\n') |
| 3409 | # next message/rfc822 |
| 3410 | part2 = msg.get_payload(1) |
| 3411 | eq(part2.get_content_type(), 'message/rfc822') |
| 3412 | eq(part2.is_multipart(), 1) |
| 3413 | eq(len(part2.get_payload()), 1) |
| 3414 | part2a = part2.get_payload(0) |
| 3415 | eq(part2a.is_multipart(), 0) |
| 3416 | eq(part2a.get_content_type(), 'text/plain') |
| 3417 | neq(part2a.get_payload(), 'message 2\n') |
| 3418 | |
| 3419 | def test_three_lines(self): |
| 3420 | # A bug report by Andrew McNamara |
| 3421 | lines = ['From: Andrew Person <aperson@dom.ain', |
| 3422 | 'Subject: Test', |
| 3423 | 'Date: Tue, 20 Aug 2002 16:43:45 +1000'] |
| 3424 | msg = email.message_from_string(NL.join(lines)) |
| 3425 | self.assertEqual(msg['date'], 'Tue, 20 Aug 2002 16:43:45 +1000') |
| 3426 | |
| 3427 | def test_strip_line_feed_and_carriage_return_in_headers(self): |
| 3428 | eq = self.assertEqual |
| 3429 | # For [ 1002475 ] email message parser doesn't handle \r\n correctly |
| 3430 | value1 = 'text' |
| 3431 | value2 = 'more text' |
| 3432 | m = 'Header: %s\r\nNext-Header: %s\r\n\r\nBody\r\n\r\n' % ( |
| 3433 | value1, value2) |
| 3434 | msg = email.message_from_string(m) |
| 3435 | eq(msg.get('Header'), value1) |
| 3436 | eq(msg.get('Next-Header'), value2) |
| 3437 | |
| 3438 | def test_rfc2822_header_syntax(self): |
| 3439 | eq = self.assertEqual |
| 3440 | m = '>From: foo\nFrom: bar\n!"#QUX;~: zoo\n\nbody' |
| 3441 | msg = email.message_from_string(m) |
| 3442 | eq(len(msg), 3) |
| 3443 | eq(sorted(field for field in msg), ['!"#QUX;~', '>From', 'From']) |
| 3444 | eq(msg.get_payload(), 'body') |
| 3445 | |
| 3446 | def test_rfc2822_space_not_allowed_in_header(self): |
| 3447 | eq = self.assertEqual |
| 3448 | m = '>From foo@example.com 11:25:53\nFrom: bar\n!"#QUX;~: zoo\n\nbody' |
| 3449 | msg = email.message_from_string(m) |
| 3450 | eq(len(msg.keys()), 0) |
| 3451 | |
| 3452 | def test_rfc2822_one_character_header(self): |
| 3453 | eq = self.assertEqual |
| 3454 | m = 'A: first header\nB: second header\nCC: third header\n\nbody' |
| 3455 | msg = email.message_from_string(m) |
| 3456 | headers = msg.keys() |
| 3457 | headers.sort() |
| 3458 | eq(headers, ['A', 'B', 'CC']) |
| 3459 | eq(msg.get_payload(), 'body') |
| 3460 | |
R. David Murray | 45e0e14 | 2010-06-16 02:19:40 +0000 | [diff] [blame] | 3461 | def test_CRLFLF_at_end_of_part(self): |
| 3462 | # issue 5610: feedparser should not eat two chars from body part ending |
| 3463 | # with "\r\n\n". |
| 3464 | m = ( |
| 3465 | "From: foo@bar.com\n" |
| 3466 | "To: baz\n" |
| 3467 | "Mime-Version: 1.0\n" |
| 3468 | "Content-Type: multipart/mixed; boundary=BOUNDARY\n" |
| 3469 | "\n" |
| 3470 | "--BOUNDARY\n" |
| 3471 | "Content-Type: text/plain\n" |
| 3472 | "\n" |
| 3473 | "body ending with CRLF newline\r\n" |
| 3474 | "\n" |
| 3475 | "--BOUNDARY--\n" |
| 3476 | ) |
| 3477 | msg = email.message_from_string(m) |
| 3478 | self.assertTrue(msg.get_payload(0).get_payload().endswith('\r\n')) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3479 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 3480 | |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3481 | class Test8BitBytesHandling(unittest.TestCase): |
| 3482 | # In Python3 all input is string, but that doesn't work if the actual input |
| 3483 | # uses an 8bit transfer encoding. To hack around that, in email 5.1 we |
| 3484 | # decode byte streams using the surrogateescape error handler, and |
| 3485 | # reconvert to binary at appropriate places if we detect surrogates. This |
| 3486 | # doesn't allow us to transform headers with 8bit bytes (they get munged), |
| 3487 | # but it does allow us to parse and preserve them, and to decode body |
| 3488 | # parts that use an 8bit CTE. |
| 3489 | |
| 3490 | bodytest_msg = textwrap.dedent("""\ |
| 3491 | From: foo@bar.com |
| 3492 | To: baz |
| 3493 | Mime-Version: 1.0 |
| 3494 | Content-Type: text/plain; charset={charset} |
| 3495 | Content-Transfer-Encoding: {cte} |
| 3496 | |
| 3497 | {bodyline} |
| 3498 | """) |
| 3499 | |
| 3500 | def test_known_8bit_CTE(self): |
| 3501 | m = self.bodytest_msg.format(charset='utf-8', |
| 3502 | cte='8bit', |
| 3503 | bodyline='pöstal').encode('utf-8') |
| 3504 | msg = email.message_from_bytes(m) |
| 3505 | self.assertEqual(msg.get_payload(), "pöstal\n") |
| 3506 | self.assertEqual(msg.get_payload(decode=True), |
| 3507 | "pöstal\n".encode('utf-8')) |
| 3508 | |
| 3509 | def test_unknown_8bit_CTE(self): |
| 3510 | m = self.bodytest_msg.format(charset='notavalidcharset', |
| 3511 | cte='8bit', |
| 3512 | bodyline='pöstal').encode('utf-8') |
| 3513 | msg = email.message_from_bytes(m) |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3514 | self.assertEqual(msg.get_payload(), "p\uFFFD\uFFFDstal\n") |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3515 | self.assertEqual(msg.get_payload(decode=True), |
| 3516 | "pöstal\n".encode('utf-8')) |
| 3517 | |
| 3518 | def test_8bit_in_quopri_body(self): |
| 3519 | # This is non-RFC compliant data...without 'decode' the library code |
| 3520 | # decodes the body using the charset from the headers, and because the |
| 3521 | # source byte really is utf-8 this works. This is likely to fail |
| 3522 | # against real dirty data (ie: produce mojibake), but the data is |
| 3523 | # invalid anyway so it is as good a guess as any. But this means that |
| 3524 | # this test just confirms the current behavior; that behavior is not |
| 3525 | # necessarily the best possible behavior. With 'decode' it is |
| 3526 | # returning the raw bytes, so that test should be of correct behavior, |
| 3527 | # or at least produce the same result that email4 did. |
| 3528 | m = self.bodytest_msg.format(charset='utf-8', |
| 3529 | cte='quoted-printable', |
| 3530 | bodyline='p=C3=B6stál').encode('utf-8') |
| 3531 | msg = email.message_from_bytes(m) |
| 3532 | self.assertEqual(msg.get_payload(), 'p=C3=B6stál\n') |
| 3533 | self.assertEqual(msg.get_payload(decode=True), |
| 3534 | 'pöstál\n'.encode('utf-8')) |
| 3535 | |
| 3536 | def test_invalid_8bit_in_non_8bit_cte_uses_replace(self): |
| 3537 | # This is similar to the previous test, but proves that if the 8bit |
| 3538 | # byte is undecodeable in the specified charset, it gets replaced |
| 3539 | # by the unicode 'unknown' character. Again, this may or may not |
| 3540 | # be the ideal behavior. Note that if decode=False none of the |
| 3541 | # decoders will get involved, so this is the only test we need |
| 3542 | # for this behavior. |
| 3543 | m = self.bodytest_msg.format(charset='ascii', |
| 3544 | cte='quoted-printable', |
| 3545 | bodyline='p=C3=B6stál').encode('utf-8') |
| 3546 | msg = email.message_from_bytes(m) |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3547 | self.assertEqual(msg.get_payload(), 'p=C3=B6st\uFFFD\uFFFDl\n') |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3548 | self.assertEqual(msg.get_payload(decode=True), |
| 3549 | 'pöstál\n'.encode('utf-8')) |
| 3550 | |
R David Murray | 80e0aee | 2012-05-27 21:23:34 -0400 | [diff] [blame] | 3551 | # test_defect_handling:test_invalid_chars_in_base64_payload |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3552 | def test_8bit_in_base64_body(self): |
R David Murray | 80e0aee | 2012-05-27 21:23:34 -0400 | [diff] [blame] | 3553 | # If we get 8bit bytes in a base64 body, we can just ignore them |
| 3554 | # as being outside the base64 alphabet and decode anyway. But |
| 3555 | # we register a defect. |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3556 | m = self.bodytest_msg.format(charset='utf-8', |
| 3557 | cte='base64', |
| 3558 | bodyline='cMO2c3RhbAá=').encode('utf-8') |
| 3559 | msg = email.message_from_bytes(m) |
| 3560 | self.assertEqual(msg.get_payload(decode=True), |
R David Murray | 80e0aee | 2012-05-27 21:23:34 -0400 | [diff] [blame] | 3561 | 'pöstal'.encode('utf-8')) |
| 3562 | self.assertIsInstance(msg.defects[0], |
| 3563 | errors.InvalidBase64CharactersDefect) |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3564 | |
| 3565 | def test_8bit_in_uuencode_body(self): |
| 3566 | # Sticking an 8bit byte in a uuencode block makes it undecodable by |
| 3567 | # normal means, so the block is returned undecoded, but as bytes. |
| 3568 | m = self.bodytest_msg.format(charset='utf-8', |
| 3569 | cte='uuencode', |
| 3570 | bodyline='<,.V<W1A; á ').encode('utf-8') |
| 3571 | msg = email.message_from_bytes(m) |
| 3572 | self.assertEqual(msg.get_payload(decode=True), |
| 3573 | '<,.V<W1A; á \n'.encode('utf-8')) |
| 3574 | |
| 3575 | |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3576 | headertest_headers = ( |
| 3577 | ('From: foo@bar.com', ('From', 'foo@bar.com')), |
| 3578 | ('To: báz', ('To', '=?unknown-8bit?q?b=C3=A1z?=')), |
| 3579 | ('Subject: Maintenant je vous présente mon collègue, le pouf célèbre\n' |
| 3580 | '\tJean de Baddie', |
| 3581 | ('Subject', '=?unknown-8bit?q?Maintenant_je_vous_pr=C3=A9sente_mon_' |
| 3582 | 'coll=C3=A8gue=2C_le_pouf_c=C3=A9l=C3=A8bre?=\n' |
| 3583 | ' =?unknown-8bit?q?_Jean_de_Baddie?=')), |
| 3584 | ('From: göst', ('From', '=?unknown-8bit?b?Z8O2c3Q=?=')), |
| 3585 | ) |
| 3586 | headertest_msg = ('\n'.join([src for (src, _) in headertest_headers]) + |
| 3587 | '\nYes, they are flying.\n').encode('utf-8') |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3588 | |
| 3589 | def test_get_8bit_header(self): |
| 3590 | msg = email.message_from_bytes(self.headertest_msg) |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3591 | self.assertEqual(str(msg.get('to')), 'b\uFFFD\uFFFDz') |
| 3592 | self.assertEqual(str(msg['to']), 'b\uFFFD\uFFFDz') |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3593 | |
| 3594 | def test_print_8bit_headers(self): |
| 3595 | msg = email.message_from_bytes(self.headertest_msg) |
| 3596 | self.assertEqual(str(msg), |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3597 | textwrap.dedent("""\ |
| 3598 | From: {} |
| 3599 | To: {} |
| 3600 | Subject: {} |
| 3601 | From: {} |
| 3602 | |
| 3603 | Yes, they are flying. |
| 3604 | """).format(*[expected[1] for (_, expected) in |
| 3605 | self.headertest_headers])) |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3606 | |
| 3607 | def test_values_with_8bit_headers(self): |
| 3608 | msg = email.message_from_bytes(self.headertest_msg) |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3609 | self.assertListEqual([str(x) for x in msg.values()], |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3610 | ['foo@bar.com', |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3611 | 'b\uFFFD\uFFFDz', |
| 3612 | 'Maintenant je vous pr\uFFFD\uFFFDsente mon ' |
| 3613 | 'coll\uFFFD\uFFFDgue, le pouf ' |
| 3614 | 'c\uFFFD\uFFFDl\uFFFD\uFFFDbre\n' |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3615 | '\tJean de Baddie', |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3616 | "g\uFFFD\uFFFDst"]) |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3617 | |
| 3618 | def test_items_with_8bit_headers(self): |
| 3619 | msg = email.message_from_bytes(self.headertest_msg) |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3620 | self.assertListEqual([(str(x), str(y)) for (x, y) in msg.items()], |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3621 | [('From', 'foo@bar.com'), |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3622 | ('To', 'b\uFFFD\uFFFDz'), |
| 3623 | ('Subject', 'Maintenant je vous ' |
| 3624 | 'pr\uFFFD\uFFFDsente ' |
| 3625 | 'mon coll\uFFFD\uFFFDgue, le pouf ' |
| 3626 | 'c\uFFFD\uFFFDl\uFFFD\uFFFDbre\n' |
| 3627 | '\tJean de Baddie'), |
| 3628 | ('From', 'g\uFFFD\uFFFDst')]) |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3629 | |
| 3630 | def test_get_all_with_8bit_headers(self): |
| 3631 | msg = email.message_from_bytes(self.headertest_msg) |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3632 | self.assertListEqual([str(x) for x in msg.get_all('from')], |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3633 | ['foo@bar.com', |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3634 | 'g\uFFFD\uFFFDst']) |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3635 | |
R David Murray | a215023 | 2011-03-16 21:11:23 -0400 | [diff] [blame] | 3636 | def test_get_content_type_with_8bit(self): |
| 3637 | msg = email.message_from_bytes(textwrap.dedent("""\ |
| 3638 | Content-Type: text/pl\xA7in; charset=utf-8 |
| 3639 | """).encode('latin-1')) |
| 3640 | self.assertEqual(msg.get_content_type(), "text/pl\uFFFDin") |
| 3641 | self.assertEqual(msg.get_content_maintype(), "text") |
| 3642 | self.assertEqual(msg.get_content_subtype(), "pl\uFFFDin") |
| 3643 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 3644 | # test_headerregistry.TestContentTypeHeader.non_ascii_in_params |
R David Murray | a215023 | 2011-03-16 21:11:23 -0400 | [diff] [blame] | 3645 | def test_get_params_with_8bit(self): |
| 3646 | msg = email.message_from_bytes( |
| 3647 | 'X-Header: foo=\xa7ne; b\xa7r=two; baz=three\n'.encode('latin-1')) |
| 3648 | self.assertEqual(msg.get_params(header='x-header'), |
| 3649 | [('foo', '\uFFFDne'), ('b\uFFFDr', 'two'), ('baz', 'three')]) |
| 3650 | self.assertEqual(msg.get_param('Foo', header='x-header'), '\uFFFdne') |
| 3651 | # XXX: someday you might be able to get 'b\xa7r', for now you can't. |
| 3652 | self.assertEqual(msg.get_param('b\xa7r', header='x-header'), None) |
| 3653 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 3654 | # test_headerregistry.TestContentTypeHeader.non_ascii_in_rfc2231_value |
R David Murray | a215023 | 2011-03-16 21:11:23 -0400 | [diff] [blame] | 3655 | def test_get_rfc2231_params_with_8bit(self): |
| 3656 | msg = email.message_from_bytes(textwrap.dedent("""\ |
| 3657 | Content-Type: text/plain; charset=us-ascii; |
| 3658 | title*=us-ascii'en'This%20is%20not%20f\xa7n""" |
| 3659 | ).encode('latin-1')) |
| 3660 | self.assertEqual(msg.get_param('title'), |
| 3661 | ('us-ascii', 'en', 'This is not f\uFFFDn')) |
| 3662 | |
| 3663 | def test_set_rfc2231_params_with_8bit(self): |
| 3664 | msg = email.message_from_bytes(textwrap.dedent("""\ |
| 3665 | Content-Type: text/plain; charset=us-ascii; |
| 3666 | title*=us-ascii'en'This%20is%20not%20f\xa7n""" |
| 3667 | ).encode('latin-1')) |
| 3668 | msg.set_param('title', 'test') |
| 3669 | self.assertEqual(msg.get_param('title'), 'test') |
| 3670 | |
| 3671 | def test_del_rfc2231_params_with_8bit(self): |
| 3672 | msg = email.message_from_bytes(textwrap.dedent("""\ |
| 3673 | Content-Type: text/plain; charset=us-ascii; |
| 3674 | title*=us-ascii'en'This%20is%20not%20f\xa7n""" |
| 3675 | ).encode('latin-1')) |
| 3676 | msg.del_param('title') |
| 3677 | self.assertEqual(msg.get_param('title'), None) |
| 3678 | self.assertEqual(msg.get_content_maintype(), 'text') |
| 3679 | |
| 3680 | def test_get_payload_with_8bit_cte_header(self): |
| 3681 | msg = email.message_from_bytes(textwrap.dedent("""\ |
| 3682 | Content-Transfer-Encoding: b\xa7se64 |
| 3683 | Content-Type: text/plain; charset=latin-1 |
| 3684 | |
| 3685 | payload |
| 3686 | """).encode('latin-1')) |
| 3687 | self.assertEqual(msg.get_payload(), 'payload\n') |
| 3688 | self.assertEqual(msg.get_payload(decode=True), b'payload\n') |
| 3689 | |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3690 | non_latin_bin_msg = textwrap.dedent("""\ |
| 3691 | From: foo@bar.com |
| 3692 | To: báz |
| 3693 | Subject: Maintenant je vous présente mon collègue, le pouf célèbre |
| 3694 | \tJean de Baddie |
| 3695 | Mime-Version: 1.0 |
| 3696 | Content-Type: text/plain; charset="utf-8" |
| 3697 | Content-Transfer-Encoding: 8bit |
| 3698 | |
| 3699 | Да, они летят. |
| 3700 | """).encode('utf-8') |
| 3701 | |
| 3702 | def test_bytes_generator(self): |
| 3703 | msg = email.message_from_bytes(self.non_latin_bin_msg) |
| 3704 | out = BytesIO() |
| 3705 | email.generator.BytesGenerator(out).flatten(msg) |
| 3706 | self.assertEqual(out.getvalue(), self.non_latin_bin_msg) |
| 3707 | |
R. David Murray | 7372a07 | 2011-01-26 21:21:32 +0000 | [diff] [blame] | 3708 | def test_bytes_generator_handles_None_body(self): |
| 3709 | #Issue 11019 |
| 3710 | msg = email.message.Message() |
| 3711 | out = BytesIO() |
| 3712 | email.generator.BytesGenerator(out).flatten(msg) |
| 3713 | self.assertEqual(out.getvalue(), b"\n") |
| 3714 | |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3715 | non_latin_bin_msg_as7bit_wrapped = textwrap.dedent("""\ |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3716 | From: foo@bar.com |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3717 | To: =?unknown-8bit?q?b=C3=A1z?= |
| 3718 | Subject: =?unknown-8bit?q?Maintenant_je_vous_pr=C3=A9sente_mon_coll=C3=A8gue?= |
| 3719 | =?unknown-8bit?q?=2C_le_pouf_c=C3=A9l=C3=A8bre?= |
| 3720 | =?unknown-8bit?q?_Jean_de_Baddie?= |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3721 | Mime-Version: 1.0 |
| 3722 | Content-Type: text/plain; charset="utf-8" |
| 3723 | Content-Transfer-Encoding: base64 |
| 3724 | |
| 3725 | 0JTQsCwg0L7QvdC4INC70LXRgtGP0YIuCg== |
| 3726 | """) |
| 3727 | |
| 3728 | def test_generator_handles_8bit(self): |
| 3729 | msg = email.message_from_bytes(self.non_latin_bin_msg) |
| 3730 | out = StringIO() |
| 3731 | email.generator.Generator(out).flatten(msg) |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3732 | self.assertEqual(out.getvalue(), self.non_latin_bin_msg_as7bit_wrapped) |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3733 | |
| 3734 | def test_bytes_generator_with_unix_from(self): |
| 3735 | # The unixfrom contains a current date, so we can't check it |
| 3736 | # literally. Just make sure the first word is 'From' and the |
| 3737 | # rest of the message matches the input. |
| 3738 | msg = email.message_from_bytes(self.non_latin_bin_msg) |
| 3739 | out = BytesIO() |
| 3740 | email.generator.BytesGenerator(out).flatten(msg, unixfrom=True) |
| 3741 | lines = out.getvalue().split(b'\n') |
| 3742 | self.assertEqual(lines[0].split()[0], b'From') |
| 3743 | self.assertEqual(b'\n'.join(lines[1:]), self.non_latin_bin_msg) |
| 3744 | |
R. David Murray | 9253214 | 2011-01-07 23:25:30 +0000 | [diff] [blame] | 3745 | non_latin_bin_msg_as7bit = non_latin_bin_msg_as7bit_wrapped.split('\n') |
| 3746 | non_latin_bin_msg_as7bit[2:4] = [ |
| 3747 | 'Subject: =?unknown-8bit?q?Maintenant_je_vous_pr=C3=A9sente_mon_' |
| 3748 | 'coll=C3=A8gue=2C_le_pouf_c=C3=A9l=C3=A8bre?='] |
| 3749 | non_latin_bin_msg_as7bit = '\n'.join(non_latin_bin_msg_as7bit) |
| 3750 | |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3751 | def test_message_from_binary_file(self): |
| 3752 | fn = 'test.msg' |
| 3753 | self.addCleanup(unlink, fn) |
| 3754 | with open(fn, 'wb') as testfile: |
| 3755 | testfile.write(self.non_latin_bin_msg) |
Brett Cannon | 384917a | 2010-10-29 23:08:36 +0000 | [diff] [blame] | 3756 | with open(fn, 'rb') as testfile: |
| 3757 | m = email.parser.BytesParser().parse(testfile) |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3758 | self.assertEqual(str(m), self.non_latin_bin_msg_as7bit) |
| 3759 | |
| 3760 | latin_bin_msg = textwrap.dedent("""\ |
| 3761 | From: foo@bar.com |
| 3762 | To: Dinsdale |
| 3763 | Subject: Nudge nudge, wink, wink |
| 3764 | Mime-Version: 1.0 |
| 3765 | Content-Type: text/plain; charset="latin-1" |
| 3766 | Content-Transfer-Encoding: 8bit |
| 3767 | |
| 3768 | oh là là, know what I mean, know what I mean? |
| 3769 | """).encode('latin-1') |
| 3770 | |
| 3771 | latin_bin_msg_as7bit = textwrap.dedent("""\ |
| 3772 | From: foo@bar.com |
| 3773 | To: Dinsdale |
| 3774 | Subject: Nudge nudge, wink, wink |
| 3775 | Mime-Version: 1.0 |
| 3776 | Content-Type: text/plain; charset="iso-8859-1" |
| 3777 | Content-Transfer-Encoding: quoted-printable |
| 3778 | |
| 3779 | oh l=E0 l=E0, know what I mean, know what I mean? |
| 3780 | """) |
| 3781 | |
| 3782 | def test_string_generator_reencodes_to_quopri_when_appropriate(self): |
| 3783 | m = email.message_from_bytes(self.latin_bin_msg) |
| 3784 | self.assertEqual(str(m), self.latin_bin_msg_as7bit) |
| 3785 | |
| 3786 | def test_decoded_generator_emits_unicode_body(self): |
| 3787 | m = email.message_from_bytes(self.latin_bin_msg) |
| 3788 | out = StringIO() |
| 3789 | email.generator.DecodedGenerator(out).flatten(m) |
| 3790 | #DecodedHeader output contains an extra blank line compared |
| 3791 | #to the input message. RDM: not sure if this is a bug or not, |
| 3792 | #but it is not specific to the 8bit->7bit conversion. |
| 3793 | self.assertEqual(out.getvalue(), |
| 3794 | self.latin_bin_msg.decode('latin-1')+'\n') |
| 3795 | |
| 3796 | def test_bytes_feedparser(self): |
| 3797 | bfp = email.feedparser.BytesFeedParser() |
| 3798 | for i in range(0, len(self.latin_bin_msg), 10): |
| 3799 | bfp.feed(self.latin_bin_msg[i:i+10]) |
| 3800 | m = bfp.close() |
| 3801 | self.assertEqual(str(m), self.latin_bin_msg_as7bit) |
| 3802 | |
R. David Murray | 8451c4b | 2010-10-23 22:19:56 +0000 | [diff] [blame] | 3803 | def test_crlf_flatten(self): |
| 3804 | with openfile('msg_26.txt', 'rb') as fp: |
| 3805 | text = fp.read() |
| 3806 | msg = email.message_from_bytes(text) |
| 3807 | s = BytesIO() |
| 3808 | g = email.generator.BytesGenerator(s) |
| 3809 | g.flatten(msg, linesep='\r\n') |
| 3810 | self.assertEqual(s.getvalue(), text) |
R David Murray | c5c1472 | 2011-04-06 08:13:02 -0400 | [diff] [blame] | 3811 | |
| 3812 | def test_8bit_multipart(self): |
| 3813 | # Issue 11605 |
| 3814 | source = textwrap.dedent("""\ |
| 3815 | Date: Fri, 18 Mar 2011 17:15:43 +0100 |
| 3816 | To: foo@example.com |
| 3817 | From: foodwatch-Newsletter <bar@example.com> |
| 3818 | Subject: Aktuelles zu Japan, Klonfleisch und Smiley-System |
| 3819 | Message-ID: <76a486bee62b0d200f33dc2ca08220ad@localhost.localdomain> |
| 3820 | MIME-Version: 1.0 |
| 3821 | Content-Type: multipart/alternative; |
| 3822 | boundary="b1_76a486bee62b0d200f33dc2ca08220ad" |
| 3823 | |
| 3824 | --b1_76a486bee62b0d200f33dc2ca08220ad |
| 3825 | Content-Type: text/plain; charset="utf-8" |
| 3826 | Content-Transfer-Encoding: 8bit |
| 3827 | |
| 3828 | Guten Tag, , |
| 3829 | |
| 3830 | mit großer Betroffenheit verfolgen auch wir im foodwatch-Team die |
| 3831 | Nachrichten aus Japan. |
| 3832 | |
| 3833 | |
| 3834 | --b1_76a486bee62b0d200f33dc2ca08220ad |
| 3835 | Content-Type: text/html; charset="utf-8" |
| 3836 | Content-Transfer-Encoding: 8bit |
| 3837 | |
| 3838 | <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" |
| 3839 | "http://www.w3.org/TR/html4/loose.dtd"> |
| 3840 | <html lang="de"> |
| 3841 | <head> |
| 3842 | <title>foodwatch - Newsletter</title> |
| 3843 | </head> |
| 3844 | <body> |
| 3845 | <p>mit großer Betroffenheit verfolgen auch wir im foodwatch-Team |
| 3846 | die Nachrichten aus Japan.</p> |
| 3847 | </body> |
| 3848 | </html> |
| 3849 | --b1_76a486bee62b0d200f33dc2ca08220ad-- |
| 3850 | |
| 3851 | """).encode('utf-8') |
| 3852 | msg = email.message_from_bytes(source) |
| 3853 | s = BytesIO() |
| 3854 | g = email.generator.BytesGenerator(s) |
| 3855 | g.flatten(msg) |
| 3856 | self.assertEqual(s.getvalue(), source) |
| 3857 | |
R David Murray | 9fd170e | 2012-03-14 14:05:03 -0400 | [diff] [blame] | 3858 | def test_bytes_generator_b_encoding_linesep(self): |
| 3859 | # Issue 14062: b encoding was tacking on an extra \n. |
| 3860 | m = Message() |
| 3861 | # This has enough non-ascii that it should always end up b encoded. |
| 3862 | m['Subject'] = Header('žluťoučký kůň') |
| 3863 | s = BytesIO() |
| 3864 | g = email.generator.BytesGenerator(s) |
| 3865 | g.flatten(m, linesep='\r\n') |
| 3866 | self.assertEqual( |
| 3867 | s.getvalue(), |
| 3868 | b'Subject: =?utf-8?b?xb5sdcWlb3XEjWvDvSBrxa/FiA==?=\r\n\r\n') |
| 3869 | |
| 3870 | def test_generator_b_encoding_linesep(self): |
| 3871 | # Since this broke in ByteGenerator, test Generator for completeness. |
| 3872 | m = Message() |
| 3873 | # This has enough non-ascii that it should always end up b encoded. |
| 3874 | m['Subject'] = Header('žluťoučký kůň') |
| 3875 | s = StringIO() |
| 3876 | g = email.generator.Generator(s) |
| 3877 | g.flatten(m, linesep='\r\n') |
| 3878 | self.assertEqual( |
| 3879 | s.getvalue(), |
| 3880 | 'Subject: =?utf-8?b?xb5sdcWlb3XEjWvDvSBrxa/FiA==?=\r\n\r\n') |
| 3881 | |
R. David Murray | 8451c4b | 2010-10-23 22:19:56 +0000 | [diff] [blame] | 3882 | maxDiff = None |
| 3883 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 3884 | |
R. David Murray | 719a449 | 2010-11-21 16:53:48 +0000 | [diff] [blame] | 3885 | class BaseTestBytesGeneratorIdempotent: |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3886 | |
R. David Murray | e5db263 | 2010-11-20 15:10:13 +0000 | [diff] [blame] | 3887 | maxDiff = None |
| 3888 | |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3889 | def _msgobj(self, filename): |
| 3890 | with openfile(filename, 'rb') as fp: |
| 3891 | data = fp.read() |
R. David Murray | 719a449 | 2010-11-21 16:53:48 +0000 | [diff] [blame] | 3892 | data = self.normalize_linesep_regex.sub(self.blinesep, data) |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3893 | msg = email.message_from_bytes(data) |
| 3894 | return msg, data |
| 3895 | |
R. David Murray | 719a449 | 2010-11-21 16:53:48 +0000 | [diff] [blame] | 3896 | def _idempotent(self, msg, data, unixfrom=False): |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3897 | b = BytesIO() |
| 3898 | g = email.generator.BytesGenerator(b, maxheaderlen=0) |
R. David Murray | 719a449 | 2010-11-21 16:53:48 +0000 | [diff] [blame] | 3899 | g.flatten(msg, unixfrom=unixfrom, linesep=self.linesep) |
R David Murray | a46ed11 | 2011-03-31 13:11:40 -0400 | [diff] [blame] | 3900 | self.assertEqual(data, b.getvalue()) |
R. David Murray | 96fd54e | 2010-10-08 15:55:28 +0000 | [diff] [blame] | 3901 | |
| 3902 | |
R. David Murray | 719a449 | 2010-11-21 16:53:48 +0000 | [diff] [blame] | 3903 | class TestBytesGeneratorIdempotentNL(BaseTestBytesGeneratorIdempotent, |
| 3904 | TestIdempotent): |
| 3905 | linesep = '\n' |
| 3906 | blinesep = b'\n' |
| 3907 | normalize_linesep_regex = re.compile(br'\r\n') |
| 3908 | |
| 3909 | |
| 3910 | class TestBytesGeneratorIdempotentCRLF(BaseTestBytesGeneratorIdempotent, |
| 3911 | TestIdempotent): |
| 3912 | linesep = '\r\n' |
| 3913 | blinesep = b'\r\n' |
| 3914 | normalize_linesep_regex = re.compile(br'(?<!\r)\n') |
| 3915 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 3916 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3917 | class TestBase64(unittest.TestCase): |
| 3918 | def test_len(self): |
| 3919 | eq = self.assertEqual |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 3920 | eq(base64mime.header_length('hello'), |
Martin v. Löwis | 15b16a3 | 2008-12-02 06:00:15 +0000 | [diff] [blame] | 3921 | len(base64mime.body_encode(b'hello', eol=''))) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3922 | for size in range(15): |
| 3923 | if size == 0 : bsize = 0 |
| 3924 | elif size <= 3 : bsize = 4 |
| 3925 | elif size <= 6 : bsize = 8 |
| 3926 | elif size <= 9 : bsize = 12 |
| 3927 | elif size <= 12: bsize = 16 |
| 3928 | else : bsize = 20 |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 3929 | eq(base64mime.header_length('x' * size), bsize) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3930 | |
| 3931 | def test_decode(self): |
| 3932 | eq = self.assertEqual |
Barry Warsaw | 2cc1f6d | 2007-08-30 14:28:55 +0000 | [diff] [blame] | 3933 | eq(base64mime.decode(''), b'') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3934 | eq(base64mime.decode('aGVsbG8='), b'hello') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3935 | |
| 3936 | def test_encode(self): |
| 3937 | eq = self.assertEqual |
Martin v. Löwis | 15b16a3 | 2008-12-02 06:00:15 +0000 | [diff] [blame] | 3938 | eq(base64mime.body_encode(b''), b'') |
| 3939 | eq(base64mime.body_encode(b'hello'), 'aGVsbG8=\n') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3940 | # Test the binary flag |
Martin v. Löwis | 15b16a3 | 2008-12-02 06:00:15 +0000 | [diff] [blame] | 3941 | eq(base64mime.body_encode(b'hello\n'), 'aGVsbG8K\n') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3942 | # Test the maxlinelen arg |
Martin v. Löwis | 15b16a3 | 2008-12-02 06:00:15 +0000 | [diff] [blame] | 3943 | eq(base64mime.body_encode(b'xxxx ' * 20, maxlinelen=40), """\ |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3944 | eHh4eCB4eHh4IHh4eHggeHh4eCB4eHh4IHh4eHgg |
| 3945 | eHh4eCB4eHh4IHh4eHggeHh4eCB4eHh4IHh4eHgg |
| 3946 | eHh4eCB4eHh4IHh4eHggeHh4eCB4eHh4IHh4eHgg |
| 3947 | eHh4eCB4eHh4IA== |
| 3948 | """) |
| 3949 | # Test the eol argument |
Martin v. Löwis | 15b16a3 | 2008-12-02 06:00:15 +0000 | [diff] [blame] | 3950 | eq(base64mime.body_encode(b'xxxx ' * 20, maxlinelen=40, eol='\r\n'), |
Barry Warsaw | 7aa02e6 | 2007-08-31 03:26:19 +0000 | [diff] [blame] | 3951 | """\ |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3952 | eHh4eCB4eHh4IHh4eHggeHh4eCB4eHh4IHh4eHgg\r |
| 3953 | eHh4eCB4eHh4IHh4eHggeHh4eCB4eHh4IHh4eHgg\r |
| 3954 | eHh4eCB4eHh4IHh4eHggeHh4eCB4eHh4IHh4eHgg\r |
| 3955 | eHh4eCB4eHh4IA==\r |
| 3956 | """) |
| 3957 | |
| 3958 | def test_header_encode(self): |
| 3959 | eq = self.assertEqual |
| 3960 | he = base64mime.header_encode |
| 3961 | eq(he('hello'), '=?iso-8859-1?b?aGVsbG8=?=') |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 3962 | eq(he('hello\r\nworld'), '=?iso-8859-1?b?aGVsbG8NCndvcmxk?=') |
| 3963 | eq(he('hello\nworld'), '=?iso-8859-1?b?aGVsbG8Kd29ybGQ=?=') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3964 | # Test the charset option |
| 3965 | eq(he('hello', charset='iso-8859-2'), '=?iso-8859-2?b?aGVsbG8=?=') |
| 3966 | eq(he('hello\nworld'), '=?iso-8859-1?b?aGVsbG8Kd29ybGQ=?=') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3967 | |
| 3968 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 3969 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3970 | class TestQuopri(unittest.TestCase): |
| 3971 | def setUp(self): |
| 3972 | # Set of characters (as byte integers) that don't need to be encoded |
| 3973 | # in headers. |
| 3974 | self.hlit = list(chain( |
| 3975 | range(ord('a'), ord('z') + 1), |
| 3976 | range(ord('A'), ord('Z') + 1), |
| 3977 | range(ord('0'), ord('9') + 1), |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 3978 | (c for c in b'!*+-/'))) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3979 | # Set of characters (as byte integers) that do need to be encoded in |
| 3980 | # headers. |
| 3981 | self.hnon = [c for c in range(256) if c not in self.hlit] |
| 3982 | assert len(self.hlit) + len(self.hnon) == 256 |
| 3983 | # Set of characters (as byte integers) that don't need to be encoded |
| 3984 | # in bodies. |
| 3985 | self.blit = list(range(ord(' '), ord('~') + 1)) |
| 3986 | self.blit.append(ord('\t')) |
| 3987 | self.blit.remove(ord('=')) |
| 3988 | # Set of characters (as byte integers) that do need to be encoded in |
| 3989 | # bodies. |
| 3990 | self.bnon = [c for c in range(256) if c not in self.blit] |
| 3991 | assert len(self.blit) + len(self.bnon) == 256 |
| 3992 | |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 3993 | def test_quopri_header_check(self): |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3994 | for c in self.hlit: |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 3995 | self.assertFalse(quoprimime.header_check(c), |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 3996 | 'Should not be header quopri encoded: %s' % chr(c)) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 3997 | for c in self.hnon: |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 3998 | self.assertTrue(quoprimime.header_check(c), |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 3999 | 'Should be header quopri encoded: %s' % chr(c)) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4000 | |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4001 | def test_quopri_body_check(self): |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4002 | for c in self.blit: |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 4003 | self.assertFalse(quoprimime.body_check(c), |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4004 | 'Should not be body quopri encoded: %s' % chr(c)) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4005 | for c in self.bnon: |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 4006 | self.assertTrue(quoprimime.body_check(c), |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4007 | 'Should be body quopri encoded: %s' % chr(c)) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4008 | |
| 4009 | def test_header_quopri_len(self): |
| 4010 | eq = self.assertEqual |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4011 | eq(quoprimime.header_length(b'hello'), 5) |
| 4012 | # RFC 2047 chrome is not included in header_length(). |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4013 | eq(len(quoprimime.header_encode(b'hello', charset='xxx')), |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4014 | quoprimime.header_length(b'hello') + |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4015 | # =?xxx?q?...?= means 10 extra characters |
| 4016 | 10) |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4017 | eq(quoprimime.header_length(b'h@e@l@l@o@'), 20) |
| 4018 | # RFC 2047 chrome is not included in header_length(). |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4019 | eq(len(quoprimime.header_encode(b'h@e@l@l@o@', charset='xxx')), |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4020 | quoprimime.header_length(b'h@e@l@l@o@') + |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4021 | # =?xxx?q?...?= means 10 extra characters |
| 4022 | 10) |
| 4023 | for c in self.hlit: |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4024 | eq(quoprimime.header_length(bytes([c])), 1, |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4025 | 'expected length 1 for %r' % chr(c)) |
| 4026 | for c in self.hnon: |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4027 | # Space is special; it's encoded to _ |
| 4028 | if c == ord(' '): |
| 4029 | continue |
| 4030 | eq(quoprimime.header_length(bytes([c])), 3, |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4031 | 'expected length 3 for %r' % chr(c)) |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4032 | eq(quoprimime.header_length(b' '), 1) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4033 | |
| 4034 | def test_body_quopri_len(self): |
| 4035 | eq = self.assertEqual |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4036 | for c in self.blit: |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4037 | eq(quoprimime.body_length(bytes([c])), 1) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4038 | for c in self.bnon: |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4039 | eq(quoprimime.body_length(bytes([c])), 3) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4040 | |
| 4041 | def test_quote_unquote_idempotent(self): |
| 4042 | for x in range(256): |
| 4043 | c = chr(x) |
| 4044 | self.assertEqual(quoprimime.unquote(quoprimime.quote(c)), c) |
| 4045 | |
R David Murray | ec1b5b8 | 2011-03-23 14:19:05 -0400 | [diff] [blame] | 4046 | def _test_header_encode(self, header, expected_encoded_header, charset=None): |
| 4047 | if charset is None: |
| 4048 | encoded_header = quoprimime.header_encode(header) |
| 4049 | else: |
| 4050 | encoded_header = quoprimime.header_encode(header, charset) |
| 4051 | self.assertEqual(encoded_header, expected_encoded_header) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4052 | |
R David Murray | cafd79d | 2011-03-23 15:25:55 -0400 | [diff] [blame] | 4053 | def test_header_encode_null(self): |
| 4054 | self._test_header_encode(b'', '') |
| 4055 | |
R David Murray | ec1b5b8 | 2011-03-23 14:19:05 -0400 | [diff] [blame] | 4056 | def test_header_encode_one_word(self): |
| 4057 | self._test_header_encode(b'hello', '=?iso-8859-1?q?hello?=') |
| 4058 | |
| 4059 | def test_header_encode_two_lines(self): |
| 4060 | self._test_header_encode(b'hello\nworld', |
| 4061 | '=?iso-8859-1?q?hello=0Aworld?=') |
| 4062 | |
| 4063 | def test_header_encode_non_ascii(self): |
| 4064 | self._test_header_encode(b'hello\xc7there', |
| 4065 | '=?iso-8859-1?q?hello=C7there?=') |
| 4066 | |
| 4067 | def test_header_encode_alt_charset(self): |
| 4068 | self._test_header_encode(b'hello', '=?iso-8859-2?q?hello?=', |
| 4069 | charset='iso-8859-2') |
| 4070 | |
| 4071 | def _test_header_decode(self, encoded_header, expected_decoded_header): |
| 4072 | decoded_header = quoprimime.header_decode(encoded_header) |
| 4073 | self.assertEqual(decoded_header, expected_decoded_header) |
| 4074 | |
| 4075 | def test_header_decode_null(self): |
| 4076 | self._test_header_decode('', '') |
| 4077 | |
| 4078 | def test_header_decode_one_word(self): |
| 4079 | self._test_header_decode('hello', 'hello') |
| 4080 | |
| 4081 | def test_header_decode_two_lines(self): |
| 4082 | self._test_header_decode('hello=0Aworld', 'hello\nworld') |
| 4083 | |
| 4084 | def test_header_decode_non_ascii(self): |
| 4085 | self._test_header_decode('hello=C7there', 'hello\xc7there') |
| 4086 | |
Ezio Melotti | 2a99d5d | 2013-07-06 17:16:04 +0200 | [diff] [blame] | 4087 | def test_header_decode_re_bug_18380(self): |
| 4088 | # Issue 18380: Call re.sub with a positional argument for flags in the wrong position |
| 4089 | self.assertEqual(quoprimime.header_decode('=30' * 257), '0' * 257) |
| 4090 | |
R David Murray | ec1b5b8 | 2011-03-23 14:19:05 -0400 | [diff] [blame] | 4091 | def _test_decode(self, encoded, expected_decoded, eol=None): |
| 4092 | if eol is None: |
| 4093 | decoded = quoprimime.decode(encoded) |
| 4094 | else: |
| 4095 | decoded = quoprimime.decode(encoded, eol=eol) |
| 4096 | self.assertEqual(decoded, expected_decoded) |
| 4097 | |
| 4098 | def test_decode_null_word(self): |
| 4099 | self._test_decode('', '') |
| 4100 | |
| 4101 | def test_decode_null_line_null_word(self): |
| 4102 | self._test_decode('\r\n', '\n') |
| 4103 | |
| 4104 | def test_decode_one_word(self): |
| 4105 | self._test_decode('hello', 'hello') |
| 4106 | |
| 4107 | def test_decode_one_word_eol(self): |
| 4108 | self._test_decode('hello', 'hello', eol='X') |
| 4109 | |
| 4110 | def test_decode_one_line(self): |
| 4111 | self._test_decode('hello\r\n', 'hello\n') |
| 4112 | |
| 4113 | def test_decode_one_line_lf(self): |
| 4114 | self._test_decode('hello\n', 'hello\n') |
| 4115 | |
R David Murray | cafd79d | 2011-03-23 15:25:55 -0400 | [diff] [blame] | 4116 | def test_decode_one_line_cr(self): |
| 4117 | self._test_decode('hello\r', 'hello\n') |
| 4118 | |
| 4119 | def test_decode_one_line_nl(self): |
| 4120 | self._test_decode('hello\n', 'helloX', eol='X') |
| 4121 | |
| 4122 | def test_decode_one_line_crnl(self): |
| 4123 | self._test_decode('hello\r\n', 'helloX', eol='X') |
| 4124 | |
R David Murray | ec1b5b8 | 2011-03-23 14:19:05 -0400 | [diff] [blame] | 4125 | def test_decode_one_line_one_word(self): |
| 4126 | self._test_decode('hello\r\nworld', 'hello\nworld') |
| 4127 | |
| 4128 | def test_decode_one_line_one_word_eol(self): |
| 4129 | self._test_decode('hello\r\nworld', 'helloXworld', eol='X') |
| 4130 | |
| 4131 | def test_decode_two_lines(self): |
| 4132 | self._test_decode('hello\r\nworld\r\n', 'hello\nworld\n') |
| 4133 | |
R David Murray | cafd79d | 2011-03-23 15:25:55 -0400 | [diff] [blame] | 4134 | def test_decode_two_lines_eol(self): |
| 4135 | self._test_decode('hello\r\nworld\r\n', 'helloXworldX', eol='X') |
| 4136 | |
R David Murray | ec1b5b8 | 2011-03-23 14:19:05 -0400 | [diff] [blame] | 4137 | def test_decode_one_long_line(self): |
| 4138 | self._test_decode('Spam' * 250, 'Spam' * 250) |
| 4139 | |
| 4140 | def test_decode_one_space(self): |
| 4141 | self._test_decode(' ', '') |
| 4142 | |
| 4143 | def test_decode_multiple_spaces(self): |
| 4144 | self._test_decode(' ' * 5, '') |
| 4145 | |
| 4146 | def test_decode_one_line_trailing_spaces(self): |
| 4147 | self._test_decode('hello \r\n', 'hello\n') |
| 4148 | |
| 4149 | def test_decode_two_lines_trailing_spaces(self): |
| 4150 | self._test_decode('hello \r\nworld \r\n', 'hello\nworld\n') |
| 4151 | |
| 4152 | def test_decode_quoted_word(self): |
| 4153 | self._test_decode('=22quoted=20words=22', '"quoted words"') |
| 4154 | |
| 4155 | def test_decode_uppercase_quoting(self): |
| 4156 | self._test_decode('ab=CD=EF', 'ab\xcd\xef') |
| 4157 | |
| 4158 | def test_decode_lowercase_quoting(self): |
| 4159 | self._test_decode('ab=cd=ef', 'ab\xcd\xef') |
| 4160 | |
| 4161 | def test_decode_soft_line_break(self): |
| 4162 | self._test_decode('soft line=\r\nbreak', 'soft linebreak') |
| 4163 | |
| 4164 | def test_decode_false_quoting(self): |
| 4165 | self._test_decode('A=1,B=A ==> A+B==2', 'A=1,B=A ==> A+B==2') |
| 4166 | |
| 4167 | def _test_encode(self, body, expected_encoded_body, maxlinelen=None, eol=None): |
| 4168 | kwargs = {} |
| 4169 | if maxlinelen is None: |
| 4170 | # Use body_encode's default. |
| 4171 | maxlinelen = 76 |
| 4172 | else: |
| 4173 | kwargs['maxlinelen'] = maxlinelen |
| 4174 | if eol is None: |
| 4175 | # Use body_encode's default. |
| 4176 | eol = '\n' |
| 4177 | else: |
| 4178 | kwargs['eol'] = eol |
| 4179 | encoded_body = quoprimime.body_encode(body, **kwargs) |
| 4180 | self.assertEqual(encoded_body, expected_encoded_body) |
| 4181 | if eol == '\n' or eol == '\r\n': |
| 4182 | # We know how to split the result back into lines, so maxlinelen |
| 4183 | # can be checked. |
| 4184 | for line in encoded_body.splitlines(): |
| 4185 | self.assertLessEqual(len(line), maxlinelen) |
| 4186 | |
| 4187 | def test_encode_null(self): |
| 4188 | self._test_encode('', '') |
| 4189 | |
| 4190 | def test_encode_null_lines(self): |
| 4191 | self._test_encode('\n\n', '\n\n') |
| 4192 | |
| 4193 | def test_encode_one_line(self): |
| 4194 | self._test_encode('hello\n', 'hello\n') |
| 4195 | |
| 4196 | def test_encode_one_line_crlf(self): |
| 4197 | self._test_encode('hello\r\n', 'hello\n') |
| 4198 | |
| 4199 | def test_encode_one_line_eol(self): |
| 4200 | self._test_encode('hello\n', 'hello\r\n', eol='\r\n') |
| 4201 | |
| 4202 | def test_encode_one_space(self): |
| 4203 | self._test_encode(' ', '=20') |
| 4204 | |
| 4205 | def test_encode_one_line_one_space(self): |
| 4206 | self._test_encode(' \n', '=20\n') |
| 4207 | |
R David Murray | b938c8c | 2011-03-24 12:19:26 -0400 | [diff] [blame] | 4208 | # XXX: body_encode() expect strings, but uses ord(char) from these strings |
| 4209 | # to index into a 256-entry list. For code points above 255, this will fail. |
| 4210 | # Should there be a check for 8-bit only ord() values in body, or at least |
| 4211 | # a comment about the expected input? |
| 4212 | |
| 4213 | def test_encode_two_lines_one_space(self): |
| 4214 | self._test_encode(' \n \n', '=20\n=20\n') |
| 4215 | |
R David Murray | ec1b5b8 | 2011-03-23 14:19:05 -0400 | [diff] [blame] | 4216 | def test_encode_one_word_trailing_spaces(self): |
| 4217 | self._test_encode('hello ', 'hello =20') |
| 4218 | |
| 4219 | def test_encode_one_line_trailing_spaces(self): |
| 4220 | self._test_encode('hello \n', 'hello =20\n') |
| 4221 | |
| 4222 | def test_encode_one_word_trailing_tab(self): |
| 4223 | self._test_encode('hello \t', 'hello =09') |
| 4224 | |
| 4225 | def test_encode_one_line_trailing_tab(self): |
| 4226 | self._test_encode('hello \t\n', 'hello =09\n') |
| 4227 | |
| 4228 | def test_encode_trailing_space_before_maxlinelen(self): |
| 4229 | self._test_encode('abcd \n1234', 'abcd =\n\n1234', maxlinelen=6) |
| 4230 | |
R David Murray | b938c8c | 2011-03-24 12:19:26 -0400 | [diff] [blame] | 4231 | def test_encode_trailing_space_at_maxlinelen(self): |
| 4232 | self._test_encode('abcd \n1234', 'abcd=\n=20\n1234', maxlinelen=5) |
| 4233 | |
R David Murray | ec1b5b8 | 2011-03-23 14:19:05 -0400 | [diff] [blame] | 4234 | def test_encode_trailing_space_beyond_maxlinelen(self): |
R David Murray | b938c8c | 2011-03-24 12:19:26 -0400 | [diff] [blame] | 4235 | self._test_encode('abcd \n1234', 'abc=\nd=20\n1234', maxlinelen=4) |
| 4236 | |
| 4237 | def test_encode_whitespace_lines(self): |
| 4238 | self._test_encode(' \n' * 5, '=20\n' * 5) |
R David Murray | ec1b5b8 | 2011-03-23 14:19:05 -0400 | [diff] [blame] | 4239 | |
| 4240 | def test_encode_quoted_equals(self): |
| 4241 | self._test_encode('a = b', 'a =3D b') |
| 4242 | |
| 4243 | def test_encode_one_long_string(self): |
| 4244 | self._test_encode('x' * 100, 'x' * 75 + '=\n' + 'x' * 25) |
| 4245 | |
| 4246 | def test_encode_one_long_line(self): |
| 4247 | self._test_encode('x' * 100 + '\n', 'x' * 75 + '=\n' + 'x' * 25 + '\n') |
| 4248 | |
| 4249 | def test_encode_one_very_long_line(self): |
| 4250 | self._test_encode('x' * 200 + '\n', |
| 4251 | 2 * ('x' * 75 + '=\n') + 'x' * 50 + '\n') |
| 4252 | |
R David Murray | ec1b5b8 | 2011-03-23 14:19:05 -0400 | [diff] [blame] | 4253 | def test_encode_shortest_maxlinelen(self): |
| 4254 | self._test_encode('=' * 5, '=3D=\n' * 4 + '=3D', maxlinelen=4) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4255 | |
R David Murray | b938c8c | 2011-03-24 12:19:26 -0400 | [diff] [blame] | 4256 | def test_encode_maxlinelen_too_small(self): |
| 4257 | self.assertRaises(ValueError, self._test_encode, '', '', maxlinelen=3) |
| 4258 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4259 | def test_encode(self): |
| 4260 | eq = self.assertEqual |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4261 | eq(quoprimime.body_encode(''), '') |
| 4262 | eq(quoprimime.body_encode('hello'), 'hello') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4263 | # Test the binary flag |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4264 | eq(quoprimime.body_encode('hello\r\nworld'), 'hello\nworld') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4265 | # Test the maxlinelen arg |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4266 | eq(quoprimime.body_encode('xxxx ' * 20, maxlinelen=40), """\ |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4267 | xxxx xxxx xxxx xxxx xxxx xxxx xxxx xxxx= |
| 4268 | xxxx xxxx xxxx xxxx xxxx xxxx xxxx xxx= |
| 4269 | x xxxx xxxx xxxx xxxx=20""") |
| 4270 | # Test the eol argument |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4271 | eq(quoprimime.body_encode('xxxx ' * 20, maxlinelen=40, eol='\r\n'), |
| 4272 | """\ |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4273 | xxxx xxxx xxxx xxxx xxxx xxxx xxxx xxxx=\r |
| 4274 | xxxx xxxx xxxx xxxx xxxx xxxx xxxx xxx=\r |
| 4275 | x xxxx xxxx xxxx xxxx=20""") |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4276 | eq(quoprimime.body_encode("""\ |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4277 | one line |
| 4278 | |
| 4279 | two line"""), """\ |
| 4280 | one line |
| 4281 | |
| 4282 | two line""") |
| 4283 | |
| 4284 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 4285 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4286 | # Test the Charset class |
| 4287 | class TestCharset(unittest.TestCase): |
| 4288 | def tearDown(self): |
| 4289 | from email import charset as CharsetModule |
| 4290 | try: |
| 4291 | del CharsetModule.CHARSETS['fake'] |
| 4292 | except KeyError: |
| 4293 | pass |
| 4294 | |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4295 | def test_codec_encodeable(self): |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4296 | eq = self.assertEqual |
| 4297 | # Make sure us-ascii = no Unicode conversion |
| 4298 | c = Charset('us-ascii') |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4299 | eq(c.header_encode('Hello World!'), 'Hello World!') |
| 4300 | # Test 8-bit idempotency with us-ascii |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4301 | s = '\xa4\xa2\xa4\xa4\xa4\xa6\xa4\xa8\xa4\xaa' |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4302 | self.assertRaises(UnicodeError, c.header_encode, s) |
| 4303 | c = Charset('utf-8') |
| 4304 | eq(c.header_encode(s), '=?utf-8?b?wqTCosKkwqTCpMKmwqTCqMKkwqo=?=') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4305 | |
| 4306 | def test_body_encode(self): |
| 4307 | eq = self.assertEqual |
| 4308 | # Try a charset with QP body encoding |
| 4309 | c = Charset('iso-8859-1') |
Barry Warsaw | 7aa02e6 | 2007-08-31 03:26:19 +0000 | [diff] [blame] | 4310 | eq('hello w=F6rld', c.body_encode('hello w\xf6rld')) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4311 | # Try a charset with Base64 body encoding |
| 4312 | c = Charset('utf-8') |
Martin v. Löwis | 15b16a3 | 2008-12-02 06:00:15 +0000 | [diff] [blame] | 4313 | eq('aGVsbG8gd29ybGQ=\n', c.body_encode(b'hello world')) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4314 | # Try a charset with None body encoding |
| 4315 | c = Charset('us-ascii') |
Barry Warsaw | 7aa02e6 | 2007-08-31 03:26:19 +0000 | [diff] [blame] | 4316 | eq('hello world', c.body_encode('hello world')) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4317 | # Try the convert argument, where input codec != output codec |
| 4318 | c = Charset('euc-jp') |
| 4319 | # With apologies to Tokio Kikuchi ;) |
Barry Warsaw | bef9d21 | 2007-08-31 10:55:37 +0000 | [diff] [blame] | 4320 | # XXX FIXME |
| 4321 | ## try: |
| 4322 | ## eq('\x1b$B5FCO;~IW\x1b(B', |
| 4323 | ## c.body_encode('\xb5\xc6\xc3\xcf\xbb\xfe\xc9\xd7')) |
| 4324 | ## eq('\xb5\xc6\xc3\xcf\xbb\xfe\xc9\xd7', |
| 4325 | ## c.body_encode('\xb5\xc6\xc3\xcf\xbb\xfe\xc9\xd7', False)) |
| 4326 | ## except LookupError: |
| 4327 | ## # We probably don't have the Japanese codecs installed |
| 4328 | ## pass |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4329 | # Testing SF bug #625509, which we have to fake, since there are no |
| 4330 | # built-in encodings where the header encoding is QP but the body |
| 4331 | # encoding is not. |
| 4332 | from email import charset as CharsetModule |
R David Murray | 56a9d7e | 2011-03-15 12:20:02 -0400 | [diff] [blame] | 4333 | CharsetModule.add_charset('fake', CharsetModule.QP, None, 'utf-8') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4334 | c = Charset('fake') |
R David Murray | 56a9d7e | 2011-03-15 12:20:02 -0400 | [diff] [blame] | 4335 | eq('hello world', c.body_encode('hello world')) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4336 | |
| 4337 | def test_unicode_charset_name(self): |
| 4338 | charset = Charset('us-ascii') |
| 4339 | self.assertEqual(str(charset), 'us-ascii') |
| 4340 | self.assertRaises(errors.CharsetError, Charset, 'asc\xffii') |
| 4341 | |
| 4342 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 4343 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4344 | # Test multilingual MIME headers. |
| 4345 | class TestHeader(TestEmailBase): |
| 4346 | def test_simple(self): |
| 4347 | eq = self.ndiffAssertEqual |
| 4348 | h = Header('Hello World!') |
| 4349 | eq(h.encode(), 'Hello World!') |
| 4350 | h.append(' Goodbye World!') |
| 4351 | eq(h.encode(), 'Hello World! Goodbye World!') |
| 4352 | |
| 4353 | def test_simple_surprise(self): |
| 4354 | eq = self.ndiffAssertEqual |
| 4355 | h = Header('Hello World!') |
| 4356 | eq(h.encode(), 'Hello World!') |
| 4357 | h.append('Goodbye World!') |
| 4358 | eq(h.encode(), 'Hello World! Goodbye World!') |
| 4359 | |
| 4360 | def test_header_needs_no_decoding(self): |
| 4361 | h = 'no decoding needed' |
| 4362 | self.assertEqual(decode_header(h), [(h, None)]) |
| 4363 | |
| 4364 | def test_long(self): |
| 4365 | h = Header("I am the very model of a modern Major-General; I've information vegetable, animal, and mineral; I know the kings of England, and I quote the fights historical from Marathon to Waterloo, in order categorical; I'm very well acquainted, too, with matters mathematical; I understand equations, both the simple and quadratical; about binomial theorem I'm teeming with a lot o' news, with many cheerful facts about the square of the hypotenuse.", |
| 4366 | maxlinelen=76) |
| 4367 | for l in h.encode(splitchars=' ').split('\n '): |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 4368 | self.assertTrue(len(l) <= 76) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4369 | |
| 4370 | def test_multilingual(self): |
| 4371 | eq = self.ndiffAssertEqual |
| 4372 | g = Charset("iso-8859-1") |
| 4373 | cz = Charset("iso-8859-2") |
| 4374 | utf8 = Charset("utf-8") |
| 4375 | g_head = (b'Die Mieter treten hier ein werden mit einem ' |
| 4376 | b'Foerderband komfortabel den Korridor entlang, ' |
| 4377 | b'an s\xfcdl\xfcndischen Wandgem\xe4lden vorbei, ' |
| 4378 | b'gegen die rotierenden Klingen bef\xf6rdert. ') |
| 4379 | cz_head = (b'Finan\xe8ni metropole se hroutily pod tlakem jejich ' |
| 4380 | b'd\xf9vtipu.. ') |
| 4381 | utf8_head = ('\u6b63\u78ba\u306b\u8a00\u3046\u3068\u7ffb\u8a33\u306f' |
| 4382 | '\u3055\u308c\u3066\u3044\u307e\u305b\u3093\u3002\u4e00' |
| 4383 | '\u90e8\u306f\u30c9\u30a4\u30c4\u8a9e\u3067\u3059\u304c' |
| 4384 | '\u3001\u3042\u3068\u306f\u3067\u305f\u3089\u3081\u3067' |
| 4385 | '\u3059\u3002\u5b9f\u969b\u306b\u306f\u300cWenn ist das ' |
| 4386 | 'Nunstuck git und Slotermeyer? Ja! Beiherhund das Oder ' |
| 4387 | 'die Flipperwaldt gersput.\u300d\u3068\u8a00\u3063\u3066' |
| 4388 | '\u3044\u307e\u3059\u3002') |
| 4389 | h = Header(g_head, g) |
| 4390 | h.append(cz_head, cz) |
| 4391 | h.append(utf8_head, utf8) |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4392 | enc = h.encode(maxlinelen=76) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4393 | eq(enc, """\ |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4394 | =?iso-8859-1?q?Die_Mieter_treten_hier_ein_werden_mit_einem_Foerderband_kom?= |
| 4395 | =?iso-8859-1?q?fortabel_den_Korridor_entlang=2C_an_s=FCdl=FCndischen_Wand?= |
| 4396 | =?iso-8859-1?q?gem=E4lden_vorbei=2C_gegen_die_rotierenden_Klingen_bef=F6r?= |
| 4397 | =?iso-8859-1?q?dert=2E_?= =?iso-8859-2?q?Finan=E8ni_metropole_se_hroutily?= |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4398 | =?iso-8859-2?q?_pod_tlakem_jejich_d=F9vtipu=2E=2E_?= =?utf-8?b?5q2j56K6?= |
| 4399 | =?utf-8?b?44Gr6KiA44GG44Go57+76Kiz44Gv44GV44KM44Gm44GE44G+44Gb44KT44CC?= |
| 4400 | =?utf-8?b?5LiA6YOo44Gv44OJ44Kk44OE6Kqe44Gn44GZ44GM44CB44GC44Go44Gv44Gn?= |
| 4401 | =?utf-8?b?44Gf44KJ44KB44Gn44GZ44CC5a6f6Zqb44Gr44Gv44CMV2VubiBpc3QgZGFz?= |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4402 | =?utf-8?b?IE51bnN0dWNrIGdpdCB1bmQgU2xvdGVybWV5ZXI/IEphISBCZWloZXJodW5k?= |
| 4403 | =?utf-8?b?IGRhcyBPZGVyIGRpZSBGbGlwcGVyd2FsZHQgZ2Vyc3B1dC7jgI3jgajoqIA=?= |
| 4404 | =?utf-8?b?44Gj44Gm44GE44G+44GZ44CC?=""") |
| 4405 | decoded = decode_header(enc) |
| 4406 | eq(len(decoded), 3) |
| 4407 | eq(decoded[0], (g_head, 'iso-8859-1')) |
| 4408 | eq(decoded[1], (cz_head, 'iso-8859-2')) |
| 4409 | eq(decoded[2], (utf8_head.encode('utf-8'), 'utf-8')) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4410 | ustr = str(h) |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4411 | eq(ustr, |
| 4412 | (b'Die Mieter treten hier ein werden mit einem Foerderband ' |
| 4413 | b'komfortabel den Korridor entlang, an s\xc3\xbcdl\xc3\xbcndischen ' |
| 4414 | b'Wandgem\xc3\xa4lden vorbei, gegen die rotierenden Klingen ' |
| 4415 | b'bef\xc3\xb6rdert. Finan\xc4\x8dni metropole se hroutily pod ' |
| 4416 | b'tlakem jejich d\xc5\xafvtipu.. \xe6\xad\xa3\xe7\xa2\xba\xe3\x81' |
| 4417 | b'\xab\xe8\xa8\x80\xe3\x81\x86\xe3\x81\xa8\xe7\xbf\xbb\xe8\xa8\xb3' |
| 4418 | b'\xe3\x81\xaf\xe3\x81\x95\xe3\x82\x8c\xe3\x81\xa6\xe3\x81\x84\xe3' |
| 4419 | b'\x81\xbe\xe3\x81\x9b\xe3\x82\x93\xe3\x80\x82\xe4\xb8\x80\xe9\x83' |
| 4420 | b'\xa8\xe3\x81\xaf\xe3\x83\x89\xe3\x82\xa4\xe3\x83\x84\xe8\xaa\x9e' |
| 4421 | b'\xe3\x81\xa7\xe3\x81\x99\xe3\x81\x8c\xe3\x80\x81\xe3\x81\x82\xe3' |
| 4422 | b'\x81\xa8\xe3\x81\xaf\xe3\x81\xa7\xe3\x81\x9f\xe3\x82\x89\xe3\x82' |
| 4423 | b'\x81\xe3\x81\xa7\xe3\x81\x99\xe3\x80\x82\xe5\xae\x9f\xe9\x9a\x9b' |
| 4424 | b'\xe3\x81\xab\xe3\x81\xaf\xe3\x80\x8cWenn ist das Nunstuck git ' |
| 4425 | b'und Slotermeyer? Ja! Beiherhund das Oder die Flipperwaldt ' |
| 4426 | b'gersput.\xe3\x80\x8d\xe3\x81\xa8\xe8\xa8\x80\xe3\x81\xa3\xe3\x81' |
| 4427 | b'\xa6\xe3\x81\x84\xe3\x81\xbe\xe3\x81\x99\xe3\x80\x82' |
| 4428 | ).decode('utf-8')) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4429 | # Test make_header() |
| 4430 | newh = make_header(decode_header(enc)) |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4431 | eq(newh, h) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4432 | |
| 4433 | def test_empty_header_encode(self): |
| 4434 | h = Header() |
| 4435 | self.assertEqual(h.encode(), '') |
Barry Warsaw | 8b3d659 | 2007-08-30 02:10:49 +0000 | [diff] [blame] | 4436 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4437 | def test_header_ctor_default_args(self): |
| 4438 | eq = self.ndiffAssertEqual |
| 4439 | h = Header() |
| 4440 | eq(h, '') |
| 4441 | h.append('foo', Charset('iso-8859-1')) |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4442 | eq(h, 'foo') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4443 | |
| 4444 | def test_explicit_maxlinelen(self): |
| 4445 | eq = self.ndiffAssertEqual |
| 4446 | hstr = ('A very long line that must get split to something other ' |
| 4447 | 'than at the 76th character boundary to test the non-default ' |
| 4448 | 'behavior') |
| 4449 | h = Header(hstr) |
| 4450 | eq(h.encode(), '''\ |
| 4451 | A very long line that must get split to something other than at the 76th |
| 4452 | character boundary to test the non-default behavior''') |
| 4453 | eq(str(h), hstr) |
| 4454 | h = Header(hstr, header_name='Subject') |
| 4455 | eq(h.encode(), '''\ |
| 4456 | A very long line that must get split to something other than at the |
| 4457 | 76th character boundary to test the non-default behavior''') |
| 4458 | eq(str(h), hstr) |
| 4459 | h = Header(hstr, maxlinelen=1024, header_name='Subject') |
| 4460 | eq(h.encode(), hstr) |
| 4461 | eq(str(h), hstr) |
| 4462 | |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4463 | def test_quopri_splittable(self): |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4464 | eq = self.ndiffAssertEqual |
| 4465 | h = Header(charset='iso-8859-1', maxlinelen=20) |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4466 | x = 'xxxx ' * 20 |
| 4467 | h.append(x) |
| 4468 | s = h.encode() |
| 4469 | eq(s, """\ |
| 4470 | =?iso-8859-1?q?xxx?= |
| 4471 | =?iso-8859-1?q?x_?= |
| 4472 | =?iso-8859-1?q?xx?= |
| 4473 | =?iso-8859-1?q?xx?= |
| 4474 | =?iso-8859-1?q?_x?= |
| 4475 | =?iso-8859-1?q?xx?= |
| 4476 | =?iso-8859-1?q?x_?= |
| 4477 | =?iso-8859-1?q?xx?= |
| 4478 | =?iso-8859-1?q?xx?= |
| 4479 | =?iso-8859-1?q?_x?= |
| 4480 | =?iso-8859-1?q?xx?= |
| 4481 | =?iso-8859-1?q?x_?= |
| 4482 | =?iso-8859-1?q?xx?= |
| 4483 | =?iso-8859-1?q?xx?= |
| 4484 | =?iso-8859-1?q?_x?= |
| 4485 | =?iso-8859-1?q?xx?= |
| 4486 | =?iso-8859-1?q?x_?= |
| 4487 | =?iso-8859-1?q?xx?= |
| 4488 | =?iso-8859-1?q?xx?= |
| 4489 | =?iso-8859-1?q?_x?= |
| 4490 | =?iso-8859-1?q?xx?= |
| 4491 | =?iso-8859-1?q?x_?= |
| 4492 | =?iso-8859-1?q?xx?= |
| 4493 | =?iso-8859-1?q?xx?= |
| 4494 | =?iso-8859-1?q?_x?= |
| 4495 | =?iso-8859-1?q?xx?= |
| 4496 | =?iso-8859-1?q?x_?= |
| 4497 | =?iso-8859-1?q?xx?= |
| 4498 | =?iso-8859-1?q?xx?= |
| 4499 | =?iso-8859-1?q?_x?= |
| 4500 | =?iso-8859-1?q?xx?= |
| 4501 | =?iso-8859-1?q?x_?= |
| 4502 | =?iso-8859-1?q?xx?= |
| 4503 | =?iso-8859-1?q?xx?= |
| 4504 | =?iso-8859-1?q?_x?= |
| 4505 | =?iso-8859-1?q?xx?= |
| 4506 | =?iso-8859-1?q?x_?= |
| 4507 | =?iso-8859-1?q?xx?= |
| 4508 | =?iso-8859-1?q?xx?= |
| 4509 | =?iso-8859-1?q?_x?= |
| 4510 | =?iso-8859-1?q?xx?= |
| 4511 | =?iso-8859-1?q?x_?= |
| 4512 | =?iso-8859-1?q?xx?= |
| 4513 | =?iso-8859-1?q?xx?= |
| 4514 | =?iso-8859-1?q?_x?= |
| 4515 | =?iso-8859-1?q?xx?= |
| 4516 | =?iso-8859-1?q?x_?= |
| 4517 | =?iso-8859-1?q?xx?= |
| 4518 | =?iso-8859-1?q?xx?= |
| 4519 | =?iso-8859-1?q?_?=""") |
| 4520 | eq(x, str(make_header(decode_header(s)))) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4521 | h = Header(charset='iso-8859-1', maxlinelen=40) |
| 4522 | h.append('xxxx ' * 20) |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4523 | s = h.encode() |
| 4524 | eq(s, """\ |
| 4525 | =?iso-8859-1?q?xxxx_xxxx_xxxx_xxxx_xxx?= |
| 4526 | =?iso-8859-1?q?x_xxxx_xxxx_xxxx_xxxx_?= |
| 4527 | =?iso-8859-1?q?xxxx_xxxx_xxxx_xxxx_xx?= |
| 4528 | =?iso-8859-1?q?xx_xxxx_xxxx_xxxx_xxxx?= |
| 4529 | =?iso-8859-1?q?_xxxx_xxxx_?=""") |
| 4530 | eq(x, str(make_header(decode_header(s)))) |
| 4531 | |
| 4532 | def test_base64_splittable(self): |
| 4533 | eq = self.ndiffAssertEqual |
| 4534 | h = Header(charset='koi8-r', maxlinelen=20) |
| 4535 | x = 'xxxx ' * 20 |
| 4536 | h.append(x) |
| 4537 | s = h.encode() |
| 4538 | eq(s, """\ |
| 4539 | =?koi8-r?b?eHh4?= |
| 4540 | =?koi8-r?b?eCB4?= |
| 4541 | =?koi8-r?b?eHh4?= |
| 4542 | =?koi8-r?b?IHh4?= |
| 4543 | =?koi8-r?b?eHgg?= |
| 4544 | =?koi8-r?b?eHh4?= |
| 4545 | =?koi8-r?b?eCB4?= |
| 4546 | =?koi8-r?b?eHh4?= |
| 4547 | =?koi8-r?b?IHh4?= |
| 4548 | =?koi8-r?b?eHgg?= |
| 4549 | =?koi8-r?b?eHh4?= |
| 4550 | =?koi8-r?b?eCB4?= |
| 4551 | =?koi8-r?b?eHh4?= |
| 4552 | =?koi8-r?b?IHh4?= |
| 4553 | =?koi8-r?b?eHgg?= |
| 4554 | =?koi8-r?b?eHh4?= |
| 4555 | =?koi8-r?b?eCB4?= |
| 4556 | =?koi8-r?b?eHh4?= |
| 4557 | =?koi8-r?b?IHh4?= |
| 4558 | =?koi8-r?b?eHgg?= |
| 4559 | =?koi8-r?b?eHh4?= |
| 4560 | =?koi8-r?b?eCB4?= |
| 4561 | =?koi8-r?b?eHh4?= |
| 4562 | =?koi8-r?b?IHh4?= |
| 4563 | =?koi8-r?b?eHgg?= |
| 4564 | =?koi8-r?b?eHh4?= |
| 4565 | =?koi8-r?b?eCB4?= |
| 4566 | =?koi8-r?b?eHh4?= |
| 4567 | =?koi8-r?b?IHh4?= |
| 4568 | =?koi8-r?b?eHgg?= |
| 4569 | =?koi8-r?b?eHh4?= |
| 4570 | =?koi8-r?b?eCB4?= |
| 4571 | =?koi8-r?b?eHh4?= |
| 4572 | =?koi8-r?b?IA==?=""") |
| 4573 | eq(x, str(make_header(decode_header(s)))) |
| 4574 | h = Header(charset='koi8-r', maxlinelen=40) |
| 4575 | h.append(x) |
| 4576 | s = h.encode() |
| 4577 | eq(s, """\ |
| 4578 | =?koi8-r?b?eHh4eCB4eHh4IHh4eHggeHh4?= |
| 4579 | =?koi8-r?b?eCB4eHh4IHh4eHggeHh4eCB4?= |
| 4580 | =?koi8-r?b?eHh4IHh4eHggeHh4eCB4eHh4?= |
| 4581 | =?koi8-r?b?IHh4eHggeHh4eCB4eHh4IHh4?= |
| 4582 | =?koi8-r?b?eHggeHh4eCB4eHh4IHh4eHgg?= |
| 4583 | =?koi8-r?b?eHh4eCB4eHh4IA==?=""") |
| 4584 | eq(x, str(make_header(decode_header(s)))) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4585 | |
| 4586 | def test_us_ascii_header(self): |
| 4587 | eq = self.assertEqual |
| 4588 | s = 'hello' |
| 4589 | x = decode_header(s) |
| 4590 | eq(x, [('hello', None)]) |
| 4591 | h = make_header(x) |
| 4592 | eq(s, h.encode()) |
| 4593 | |
| 4594 | def test_string_charset(self): |
| 4595 | eq = self.assertEqual |
| 4596 | h = Header() |
| 4597 | h.append('hello', 'iso-8859-1') |
Guido van Rossum | 9604e66 | 2007-08-30 03:46:43 +0000 | [diff] [blame] | 4598 | eq(h, 'hello') |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4599 | |
| 4600 | ## def test_unicode_error(self): |
| 4601 | ## raises = self.assertRaises |
| 4602 | ## raises(UnicodeError, Header, u'[P\xf6stal]', 'us-ascii') |
| 4603 | ## raises(UnicodeError, Header, '[P\xf6stal]', 'us-ascii') |
| 4604 | ## h = Header() |
| 4605 | ## raises(UnicodeError, h.append, u'[P\xf6stal]', 'us-ascii') |
| 4606 | ## raises(UnicodeError, h.append, '[P\xf6stal]', 'us-ascii') |
| 4607 | ## raises(UnicodeError, Header, u'\u83ca\u5730\u6642\u592b', 'iso-8859-1') |
| 4608 | |
| 4609 | def test_utf8_shortest(self): |
| 4610 | eq = self.assertEqual |
| 4611 | h = Header('p\xf6stal', 'utf-8') |
| 4612 | eq(h.encode(), '=?utf-8?q?p=C3=B6stal?=') |
| 4613 | h = Header('\u83ca\u5730\u6642\u592b', 'utf-8') |
| 4614 | eq(h.encode(), '=?utf-8?b?6I+K5Zyw5pmC5aSr?=') |
| 4615 | |
| 4616 | def test_bad_8bit_header(self): |
| 4617 | raises = self.assertRaises |
| 4618 | eq = self.assertEqual |
| 4619 | x = b'Ynwp4dUEbay Auction Semiar- No Charge \x96 Earn Big' |
| 4620 | raises(UnicodeError, Header, x) |
| 4621 | h = Header() |
| 4622 | raises(UnicodeError, h.append, x) |
| 4623 | e = x.decode('utf-8', 'replace') |
| 4624 | eq(str(Header(x, errors='replace')), e) |
| 4625 | h.append(x, errors='replace') |
| 4626 | eq(str(h), e) |
| 4627 | |
R David Murray | 041015c | 2011-03-25 15:10:55 -0400 | [diff] [blame] | 4628 | def test_escaped_8bit_header(self): |
| 4629 | x = b'Ynwp4dUEbay Auction Semiar- No Charge \x96 Earn Big' |
R David Murray | 6bdb176 | 2011-06-18 12:30:55 -0400 | [diff] [blame] | 4630 | e = x.decode('ascii', 'surrogateescape') |
| 4631 | h = Header(e, charset=email.charset.UNKNOWN8BIT) |
R David Murray | 041015c | 2011-03-25 15:10:55 -0400 | [diff] [blame] | 4632 | self.assertEqual(str(h), |
| 4633 | 'Ynwp4dUEbay Auction Semiar- No Charge \uFFFD Earn Big') |
| 4634 | self.assertEqual(email.header.decode_header(h), [(x, 'unknown-8bit')]) |
| 4635 | |
R David Murray | e5e366c | 2011-06-18 12:57:28 -0400 | [diff] [blame] | 4636 | def test_header_handles_binary_unknown8bit(self): |
| 4637 | x = b'Ynwp4dUEbay Auction Semiar- No Charge \x96 Earn Big' |
| 4638 | h = Header(x, charset=email.charset.UNKNOWN8BIT) |
| 4639 | self.assertEqual(str(h), |
| 4640 | 'Ynwp4dUEbay Auction Semiar- No Charge \uFFFD Earn Big') |
| 4641 | self.assertEqual(email.header.decode_header(h), [(x, 'unknown-8bit')]) |
| 4642 | |
| 4643 | def test_make_header_handles_binary_unknown8bit(self): |
| 4644 | x = b'Ynwp4dUEbay Auction Semiar- No Charge \x96 Earn Big' |
| 4645 | h = Header(x, charset=email.charset.UNKNOWN8BIT) |
| 4646 | h2 = email.header.make_header(email.header.decode_header(h)) |
| 4647 | self.assertEqual(str(h2), |
| 4648 | 'Ynwp4dUEbay Auction Semiar- No Charge \uFFFD Earn Big') |
| 4649 | self.assertEqual(email.header.decode_header(h2), [(x, 'unknown-8bit')]) |
| 4650 | |
R David Murray | 041015c | 2011-03-25 15:10:55 -0400 | [diff] [blame] | 4651 | def test_modify_returned_list_does_not_change_header(self): |
| 4652 | h = Header('test') |
| 4653 | chunks = email.header.decode_header(h) |
| 4654 | chunks.append(('ascii', 'test2')) |
| 4655 | self.assertEqual(str(h), 'test') |
| 4656 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4657 | def test_encoded_adjacent_nonencoded(self): |
| 4658 | eq = self.assertEqual |
| 4659 | h = Header() |
| 4660 | h.append('hello', 'iso-8859-1') |
| 4661 | h.append('world') |
| 4662 | s = h.encode() |
| 4663 | eq(s, '=?iso-8859-1?q?hello?= world') |
| 4664 | h = make_header(decode_header(s)) |
| 4665 | eq(h.encode(), s) |
| 4666 | |
R David Murray | 07ea53c | 2012-06-02 17:56:49 -0400 | [diff] [blame] | 4667 | def test_whitespace_keeper(self): |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4668 | eq = self.assertEqual |
| 4669 | s = 'Subject: =?koi8-r?b?8NLP18XSy8EgzsEgxsnOwczYztk=?= =?koi8-r?q?=CA?= zz.' |
| 4670 | parts = decode_header(s) |
R David Murray | 07ea53c | 2012-06-02 17:56:49 -0400 | [diff] [blame] | 4671 | eq(parts, [(b'Subject: ', None), (b'\xf0\xd2\xcf\xd7\xc5\xd2\xcb\xc1 \xce\xc1 \xc6\xc9\xce\xc1\xcc\xd8\xce\xd9\xca', 'koi8-r'), (b' zz.', None)]) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4672 | hdr = make_header(parts) |
| 4673 | eq(hdr.encode(), |
| 4674 | 'Subject: =?koi8-r?b?8NLP18XSy8EgzsEgxsnOwczYztnK?= zz.') |
| 4675 | |
| 4676 | def test_broken_base64_header(self): |
| 4677 | raises = self.assertRaises |
R. David Murray | c4e69cc | 2010-08-03 22:14:10 +0000 | [diff] [blame] | 4678 | s = 'Subject: =?EUC-KR?B?CSixpLDtKSC/7Liuvsax4iC6uLmwMcijIKHaILzSwd/H0SC8+LCjwLsgv7W/+Mj3I ?=' |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4679 | raises(errors.HeaderParseError, decode_header, s) |
| 4680 | |
R. David Murray | 477efb3 | 2011-01-05 01:39:32 +0000 | [diff] [blame] | 4681 | def test_shift_jis_charset(self): |
| 4682 | h = Header('文', charset='shift_jis') |
| 4683 | self.assertEqual(h.encode(), '=?iso-2022-jp?b?GyRCSjgbKEI=?=') |
| 4684 | |
R David Murray | de91276 | 2011-03-16 18:26:23 -0400 | [diff] [blame] | 4685 | def test_flatten_header_with_no_value(self): |
| 4686 | # Issue 11401 (regression from email 4.x) Note that the space after |
| 4687 | # the header doesn't reflect the input, but this is also the way |
| 4688 | # email 4.x behaved. At some point it would be nice to fix that. |
| 4689 | msg = email.message_from_string("EmptyHeader:") |
| 4690 | self.assertEqual(str(msg), "EmptyHeader: \n\n") |
| 4691 | |
R David Murray | 01581ee | 2011-04-18 10:04:34 -0400 | [diff] [blame] | 4692 | def test_encode_preserves_leading_ws_on_value(self): |
| 4693 | msg = Message() |
| 4694 | msg['SomeHeader'] = ' value with leading ws' |
| 4695 | self.assertEqual(str(msg), "SomeHeader: value with leading ws\n\n") |
| 4696 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4697 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 4698 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4699 | # Test RFC 2231 header parameters (en/de)coding |
| 4700 | class TestRFC2231(TestEmailBase): |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 4701 | |
| 4702 | # test_headerregistry.TestContentTypeHeader.rfc2231_encoded_with_double_quotes |
| 4703 | # test_headerregistry.TestContentTypeHeader.rfc2231_single_quote_inside_double_quotes |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4704 | def test_get_param(self): |
| 4705 | eq = self.assertEqual |
| 4706 | msg = self._msgobj('msg_29.txt') |
| 4707 | eq(msg.get_param('title'), |
| 4708 | ('us-ascii', 'en', 'This is even more ***fun*** isn\'t it!')) |
| 4709 | eq(msg.get_param('title', unquote=False), |
| 4710 | ('us-ascii', 'en', '"This is even more ***fun*** isn\'t it!"')) |
| 4711 | |
| 4712 | def test_set_param(self): |
| 4713 | eq = self.ndiffAssertEqual |
| 4714 | msg = Message() |
| 4715 | msg.set_param('title', 'This is even more ***fun*** isn\'t it!', |
| 4716 | charset='us-ascii') |
| 4717 | eq(msg.get_param('title'), |
| 4718 | ('us-ascii', '', 'This is even more ***fun*** isn\'t it!')) |
| 4719 | msg.set_param('title', 'This is even more ***fun*** isn\'t it!', |
| 4720 | charset='us-ascii', language='en') |
| 4721 | eq(msg.get_param('title'), |
| 4722 | ('us-ascii', 'en', 'This is even more ***fun*** isn\'t it!')) |
| 4723 | msg = self._msgobj('msg_01.txt') |
| 4724 | msg.set_param('title', 'This is even more ***fun*** isn\'t it!', |
| 4725 | charset='us-ascii', language='en') |
| 4726 | eq(msg.as_string(maxheaderlen=78), """\ |
| 4727 | Return-Path: <bbb@zzz.org> |
| 4728 | Delivered-To: bbb@zzz.org |
| 4729 | Received: by mail.zzz.org (Postfix, from userid 889) |
| 4730 | \tid 27CEAD38CC; Fri, 4 May 2001 14:05:44 -0400 (EDT) |
| 4731 | MIME-Version: 1.0 |
| 4732 | Content-Transfer-Encoding: 7bit |
| 4733 | Message-ID: <15090.61304.110929.45684@aaa.zzz.org> |
| 4734 | From: bbb@ddd.com (John X. Doe) |
| 4735 | To: bbb@zzz.org |
| 4736 | Subject: This is a test message |
| 4737 | Date: Fri, 4 May 2001 14:05:44 -0400 |
| 4738 | Content-Type: text/plain; charset=us-ascii; |
R. David Murray | dfd7eb0 | 2010-12-24 22:36:49 +0000 | [diff] [blame] | 4739 | title*=us-ascii'en'This%20is%20even%20more%20%2A%2A%2Afun%2A%2A%2A%20isn%27t%20it%21 |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4740 | |
| 4741 | |
| 4742 | Hi, |
| 4743 | |
| 4744 | Do you like this message? |
| 4745 | |
| 4746 | -Me |
| 4747 | """) |
| 4748 | |
R David Murray | a2860e8 | 2011-04-16 09:20:30 -0400 | [diff] [blame] | 4749 | def test_set_param_requote(self): |
| 4750 | msg = Message() |
| 4751 | msg.set_param('title', 'foo') |
| 4752 | self.assertEqual(msg['content-type'], 'text/plain; title="foo"') |
| 4753 | msg.set_param('title', 'bar', requote=False) |
| 4754 | self.assertEqual(msg['content-type'], 'text/plain; title=bar') |
| 4755 | # tspecial is still quoted. |
| 4756 | msg.set_param('title', "(bar)bell", requote=False) |
| 4757 | self.assertEqual(msg['content-type'], 'text/plain; title="(bar)bell"') |
| 4758 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4759 | def test_del_param(self): |
| 4760 | eq = self.ndiffAssertEqual |
| 4761 | msg = self._msgobj('msg_01.txt') |
| 4762 | msg.set_param('foo', 'bar', charset='us-ascii', language='en') |
| 4763 | msg.set_param('title', 'This is even more ***fun*** isn\'t it!', |
| 4764 | charset='us-ascii', language='en') |
| 4765 | msg.del_param('foo', header='Content-Type') |
| 4766 | eq(msg.as_string(maxheaderlen=78), """\ |
| 4767 | Return-Path: <bbb@zzz.org> |
| 4768 | Delivered-To: bbb@zzz.org |
| 4769 | Received: by mail.zzz.org (Postfix, from userid 889) |
| 4770 | \tid 27CEAD38CC; Fri, 4 May 2001 14:05:44 -0400 (EDT) |
| 4771 | MIME-Version: 1.0 |
| 4772 | Content-Transfer-Encoding: 7bit |
| 4773 | Message-ID: <15090.61304.110929.45684@aaa.zzz.org> |
| 4774 | From: bbb@ddd.com (John X. Doe) |
| 4775 | To: bbb@zzz.org |
| 4776 | Subject: This is a test message |
| 4777 | Date: Fri, 4 May 2001 14:05:44 -0400 |
| 4778 | Content-Type: text/plain; charset="us-ascii"; |
R. David Murray | dfd7eb0 | 2010-12-24 22:36:49 +0000 | [diff] [blame] | 4779 | title*=us-ascii'en'This%20is%20even%20more%20%2A%2A%2Afun%2A%2A%2A%20isn%27t%20it%21 |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4780 | |
| 4781 | |
| 4782 | Hi, |
| 4783 | |
| 4784 | Do you like this message? |
| 4785 | |
| 4786 | -Me |
| 4787 | """) |
| 4788 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 4789 | # test_headerregistry.TestContentTypeHeader.rfc2231_encoded_charset |
| 4790 | # I changed the charset name, though, because the one in the file isn't |
| 4791 | # a legal charset name. Should add a test for an illegal charset. |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4792 | def test_rfc2231_get_content_charset(self): |
| 4793 | eq = self.assertEqual |
| 4794 | msg = self._msgobj('msg_32.txt') |
| 4795 | eq(msg.get_content_charset(), 'us-ascii') |
| 4796 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 4797 | # test_headerregistry.TestContentTypeHeader.rfc2231_encoded_no_double_quotes |
R. David Murray | dfd7eb0 | 2010-12-24 22:36:49 +0000 | [diff] [blame] | 4798 | def test_rfc2231_parse_rfc_quoting(self): |
| 4799 | m = textwrap.dedent('''\ |
| 4800 | Content-Disposition: inline; |
| 4801 | \tfilename*0*=''This%20is%20even%20more%20; |
| 4802 | \tfilename*1*=%2A%2A%2Afun%2A%2A%2A%20; |
| 4803 | \tfilename*2="is it not.pdf" |
| 4804 | |
| 4805 | ''') |
| 4806 | msg = email.message_from_string(m) |
| 4807 | self.assertEqual(msg.get_filename(), |
| 4808 | 'This is even more ***fun*** is it not.pdf') |
| 4809 | self.assertEqual(m, msg.as_string()) |
| 4810 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 4811 | # test_headerregistry.TestContentTypeHeader.rfc2231_encoded_with_double_quotes |
R. David Murray | dfd7eb0 | 2010-12-24 22:36:49 +0000 | [diff] [blame] | 4812 | def test_rfc2231_parse_extra_quoting(self): |
| 4813 | m = textwrap.dedent('''\ |
| 4814 | Content-Disposition: inline; |
| 4815 | \tfilename*0*="''This%20is%20even%20more%20"; |
| 4816 | \tfilename*1*="%2A%2A%2Afun%2A%2A%2A%20"; |
| 4817 | \tfilename*2="is it not.pdf" |
| 4818 | |
| 4819 | ''') |
| 4820 | msg = email.message_from_string(m) |
| 4821 | self.assertEqual(msg.get_filename(), |
| 4822 | 'This is even more ***fun*** is it not.pdf') |
| 4823 | self.assertEqual(m, msg.as_string()) |
| 4824 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 4825 | # test_headerregistry.TestContentTypeHeader.rfc2231_no_language_or_charset |
| 4826 | # but new test uses *0* because otherwise lang/charset is not valid. |
| 4827 | # test_headerregistry.TestContentTypeHeader.rfc2231_segmented_normal_values |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4828 | def test_rfc2231_no_language_or_charset(self): |
| 4829 | m = '''\ |
| 4830 | Content-Transfer-Encoding: 8bit |
| 4831 | Content-Disposition: inline; filename="file____C__DOCUMENTS_20AND_20SETTINGS_FABIEN_LOCAL_20SETTINGS_TEMP_nsmail.htm" |
| 4832 | Content-Type: text/html; NAME*0=file____C__DOCUMENTS_20AND_20SETTINGS_FABIEN_LOCAL_20SETTINGS_TEM; NAME*1=P_nsmail.htm |
| 4833 | |
| 4834 | ''' |
| 4835 | msg = email.message_from_string(m) |
| 4836 | param = msg.get_param('NAME') |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 4837 | self.assertFalse(isinstance(param, tuple)) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4838 | self.assertEqual( |
| 4839 | param, |
| 4840 | 'file____C__DOCUMENTS_20AND_20SETTINGS_FABIEN_LOCAL_20SETTINGS_TEMP_nsmail.htm') |
| 4841 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 4842 | # test_headerregistry.TestContentTypeHeader.rfc2231_encoded_no_charset |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4843 | def test_rfc2231_no_language_or_charset_in_filename(self): |
| 4844 | m = '''\ |
| 4845 | Content-Disposition: inline; |
| 4846 | \tfilename*0*="''This%20is%20even%20more%20"; |
| 4847 | \tfilename*1*="%2A%2A%2Afun%2A%2A%2A%20"; |
| 4848 | \tfilename*2="is it not.pdf" |
| 4849 | |
| 4850 | ''' |
| 4851 | msg = email.message_from_string(m) |
| 4852 | self.assertEqual(msg.get_filename(), |
| 4853 | 'This is even more ***fun*** is it not.pdf') |
| 4854 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 4855 | # Duplicate of previous test? |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4856 | def test_rfc2231_no_language_or_charset_in_filename_encoded(self): |
| 4857 | m = '''\ |
| 4858 | Content-Disposition: inline; |
| 4859 | \tfilename*0*="''This%20is%20even%20more%20"; |
| 4860 | \tfilename*1*="%2A%2A%2Afun%2A%2A%2A%20"; |
| 4861 | \tfilename*2="is it not.pdf" |
| 4862 | |
| 4863 | ''' |
| 4864 | msg = email.message_from_string(m) |
| 4865 | self.assertEqual(msg.get_filename(), |
| 4866 | 'This is even more ***fun*** is it not.pdf') |
| 4867 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 4868 | # test_headerregistry.TestContentTypeHeader.rfc2231_partly_encoded, |
| 4869 | # but the test below is wrong (the first part should be decoded). |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4870 | def test_rfc2231_partly_encoded(self): |
| 4871 | m = '''\ |
| 4872 | Content-Disposition: inline; |
| 4873 | \tfilename*0="''This%20is%20even%20more%20"; |
| 4874 | \tfilename*1*="%2A%2A%2Afun%2A%2A%2A%20"; |
| 4875 | \tfilename*2="is it not.pdf" |
| 4876 | |
| 4877 | ''' |
| 4878 | msg = email.message_from_string(m) |
| 4879 | self.assertEqual( |
| 4880 | msg.get_filename(), |
| 4881 | 'This%20is%20even%20more%20***fun*** is it not.pdf') |
| 4882 | |
| 4883 | def test_rfc2231_partly_nonencoded(self): |
| 4884 | m = '''\ |
| 4885 | Content-Disposition: inline; |
| 4886 | \tfilename*0="This%20is%20even%20more%20"; |
| 4887 | \tfilename*1="%2A%2A%2Afun%2A%2A%2A%20"; |
| 4888 | \tfilename*2="is it not.pdf" |
| 4889 | |
| 4890 | ''' |
| 4891 | msg = email.message_from_string(m) |
| 4892 | self.assertEqual( |
| 4893 | msg.get_filename(), |
| 4894 | 'This%20is%20even%20more%20%2A%2A%2Afun%2A%2A%2A%20is it not.pdf') |
| 4895 | |
| 4896 | def test_rfc2231_no_language_or_charset_in_boundary(self): |
| 4897 | m = '''\ |
| 4898 | Content-Type: multipart/alternative; |
| 4899 | \tboundary*0*="''This%20is%20even%20more%20"; |
| 4900 | \tboundary*1*="%2A%2A%2Afun%2A%2A%2A%20"; |
| 4901 | \tboundary*2="is it not.pdf" |
| 4902 | |
| 4903 | ''' |
| 4904 | msg = email.message_from_string(m) |
| 4905 | self.assertEqual(msg.get_boundary(), |
| 4906 | 'This is even more ***fun*** is it not.pdf') |
| 4907 | |
| 4908 | def test_rfc2231_no_language_or_charset_in_charset(self): |
| 4909 | # This is a nonsensical charset value, but tests the code anyway |
| 4910 | m = '''\ |
| 4911 | Content-Type: text/plain; |
| 4912 | \tcharset*0*="This%20is%20even%20more%20"; |
| 4913 | \tcharset*1*="%2A%2A%2Afun%2A%2A%2A%20"; |
| 4914 | \tcharset*2="is it not.pdf" |
| 4915 | |
| 4916 | ''' |
| 4917 | msg = email.message_from_string(m) |
| 4918 | self.assertEqual(msg.get_content_charset(), |
| 4919 | 'this is even more ***fun*** is it not.pdf') |
| 4920 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 4921 | # test_headerregistry.TestContentTypeHeader.rfc2231_unknown_charset_treated_as_ascii |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4922 | def test_rfc2231_bad_encoding_in_filename(self): |
| 4923 | m = '''\ |
| 4924 | Content-Disposition: inline; |
| 4925 | \tfilename*0*="bogus'xx'This%20is%20even%20more%20"; |
| 4926 | \tfilename*1*="%2A%2A%2Afun%2A%2A%2A%20"; |
| 4927 | \tfilename*2="is it not.pdf" |
| 4928 | |
| 4929 | ''' |
| 4930 | msg = email.message_from_string(m) |
| 4931 | self.assertEqual(msg.get_filename(), |
| 4932 | 'This is even more ***fun*** is it not.pdf') |
| 4933 | |
| 4934 | def test_rfc2231_bad_encoding_in_charset(self): |
| 4935 | m = """\ |
| 4936 | Content-Type: text/plain; charset*=bogus''utf-8%E2%80%9D |
| 4937 | |
| 4938 | """ |
| 4939 | msg = email.message_from_string(m) |
| 4940 | # This should return None because non-ascii characters in the charset |
| 4941 | # are not allowed. |
| 4942 | self.assertEqual(msg.get_content_charset(), None) |
| 4943 | |
| 4944 | def test_rfc2231_bad_character_in_charset(self): |
| 4945 | m = """\ |
| 4946 | Content-Type: text/plain; charset*=ascii''utf-8%E2%80%9D |
| 4947 | |
| 4948 | """ |
| 4949 | msg = email.message_from_string(m) |
| 4950 | # This should return None because non-ascii characters in the charset |
| 4951 | # are not allowed. |
| 4952 | self.assertEqual(msg.get_content_charset(), None) |
| 4953 | |
| 4954 | def test_rfc2231_bad_character_in_filename(self): |
| 4955 | m = '''\ |
| 4956 | Content-Disposition: inline; |
| 4957 | \tfilename*0*="ascii'xx'This%20is%20even%20more%20"; |
| 4958 | \tfilename*1*="%2A%2A%2Afun%2A%2A%2A%20"; |
| 4959 | \tfilename*2*="is it not.pdf%E2" |
| 4960 | |
| 4961 | ''' |
| 4962 | msg = email.message_from_string(m) |
| 4963 | self.assertEqual(msg.get_filename(), |
| 4964 | 'This is even more ***fun*** is it not.pdf\ufffd') |
| 4965 | |
| 4966 | def test_rfc2231_unknown_encoding(self): |
| 4967 | m = """\ |
| 4968 | Content-Transfer-Encoding: 8bit |
| 4969 | Content-Disposition: inline; filename*=X-UNKNOWN''myfile.txt |
| 4970 | |
| 4971 | """ |
| 4972 | msg = email.message_from_string(m) |
| 4973 | self.assertEqual(msg.get_filename(), 'myfile.txt') |
| 4974 | |
| 4975 | def test_rfc2231_single_tick_in_filename_extended(self): |
| 4976 | eq = self.assertEqual |
| 4977 | m = """\ |
| 4978 | Content-Type: application/x-foo; |
| 4979 | \tname*0*=\"Frank's\"; name*1*=\" Document\" |
| 4980 | |
| 4981 | """ |
| 4982 | msg = email.message_from_string(m) |
| 4983 | charset, language, s = msg.get_param('name') |
| 4984 | eq(charset, None) |
| 4985 | eq(language, None) |
| 4986 | eq(s, "Frank's Document") |
| 4987 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 4988 | # test_headerregistry.TestContentTypeHeader.rfc2231_single_quote_inside_double_quotes |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4989 | def test_rfc2231_single_tick_in_filename(self): |
| 4990 | m = """\ |
| 4991 | Content-Type: application/x-foo; name*0=\"Frank's\"; name*1=\" Document\" |
| 4992 | |
| 4993 | """ |
| 4994 | msg = email.message_from_string(m) |
| 4995 | param = msg.get_param('name') |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 4996 | self.assertFalse(isinstance(param, tuple)) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 4997 | self.assertEqual(param, "Frank's Document") |
| 4998 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 4999 | # test_headerregistry.TestContentTypeHeader.rfc2231_single_quote_in_value_with_charset_and_lang |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 5000 | def test_rfc2231_tick_attack_extended(self): |
| 5001 | eq = self.assertEqual |
| 5002 | m = """\ |
| 5003 | Content-Type: application/x-foo; |
| 5004 | \tname*0*=\"us-ascii'en-us'Frank's\"; name*1*=\" Document\" |
| 5005 | |
| 5006 | """ |
| 5007 | msg = email.message_from_string(m) |
| 5008 | charset, language, s = msg.get_param('name') |
| 5009 | eq(charset, 'us-ascii') |
| 5010 | eq(language, 'en-us') |
| 5011 | eq(s, "Frank's Document") |
| 5012 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 5013 | # test_headerregistry.TestContentTypeHeader.rfc2231_single_quote_in_non_encoded_value |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 5014 | def test_rfc2231_tick_attack(self): |
| 5015 | m = """\ |
| 5016 | Content-Type: application/x-foo; |
| 5017 | \tname*0=\"us-ascii'en-us'Frank's\"; name*1=\" Document\" |
| 5018 | |
| 5019 | """ |
| 5020 | msg = email.message_from_string(m) |
| 5021 | param = msg.get_param('name') |
Benjamin Peterson | c9c0f20 | 2009-06-30 23:06:06 +0000 | [diff] [blame] | 5022 | self.assertFalse(isinstance(param, tuple)) |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 5023 | self.assertEqual(param, "us-ascii'en-us'Frank's Document") |
| 5024 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 5025 | # test_headerregistry.TestContentTypeHeader.rfc2231_single_quotes_inside_quotes |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 5026 | def test_rfc2231_no_extended_values(self): |
| 5027 | eq = self.assertEqual |
| 5028 | m = """\ |
| 5029 | Content-Type: application/x-foo; name=\"Frank's Document\" |
| 5030 | |
| 5031 | """ |
| 5032 | msg = email.message_from_string(m) |
| 5033 | eq(msg.get_param('name'), "Frank's Document") |
| 5034 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 5035 | # test_headerregistry.TestContentTypeHeader.rfc2231_encoded_then_unencoded_segments |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 5036 | def test_rfc2231_encoded_then_unencoded_segments(self): |
| 5037 | eq = self.assertEqual |
| 5038 | m = """\ |
| 5039 | Content-Type: application/x-foo; |
| 5040 | \tname*0*=\"us-ascii'en-us'My\"; |
| 5041 | \tname*1=\" Document\"; |
| 5042 | \tname*2*=\" For You\" |
| 5043 | |
| 5044 | """ |
| 5045 | msg = email.message_from_string(m) |
| 5046 | charset, language, s = msg.get_param('name') |
| 5047 | eq(charset, 'us-ascii') |
| 5048 | eq(language, 'en-us') |
| 5049 | eq(s, 'My Document For You') |
| 5050 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 5051 | # test_headerregistry.TestContentTypeHeader.rfc2231_unencoded_then_encoded_segments |
| 5052 | # test_headerregistry.TestContentTypeHeader.rfc2231_quoted_unencoded_then_encoded_segments |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 5053 | def test_rfc2231_unencoded_then_encoded_segments(self): |
| 5054 | eq = self.assertEqual |
| 5055 | m = """\ |
| 5056 | Content-Type: application/x-foo; |
| 5057 | \tname*0=\"us-ascii'en-us'My\"; |
| 5058 | \tname*1*=\" Document\"; |
| 5059 | \tname*2*=\" For You\" |
| 5060 | |
| 5061 | """ |
| 5062 | msg = email.message_from_string(m) |
| 5063 | charset, language, s = msg.get_param('name') |
| 5064 | eq(charset, 'us-ascii') |
| 5065 | eq(language, 'en-us') |
| 5066 | eq(s, 'My Document For You') |
| 5067 | |
| 5068 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 5069 | |
R. David Murray | a8f480f | 2010-01-16 18:30:03 +0000 | [diff] [blame] | 5070 | # Tests to ensure that signed parts of an email are completely preserved, as |
| 5071 | # required by RFC1847 section 2.1. Note that these are incomplete, because the |
| 5072 | # email package does not currently always preserve the body. See issue 1670765. |
| 5073 | class TestSigned(TestEmailBase): |
| 5074 | |
| 5075 | def _msg_and_obj(self, filename): |
R David Murray | 28346b8 | 2011-03-31 11:40:20 -0400 | [diff] [blame] | 5076 | with openfile(filename) as fp: |
R. David Murray | a8f480f | 2010-01-16 18:30:03 +0000 | [diff] [blame] | 5077 | original = fp.read() |
| 5078 | msg = email.message_from_string(original) |
| 5079 | return original, msg |
| 5080 | |
| 5081 | def _signed_parts_eq(self, original, result): |
| 5082 | # Extract the first mime part of each message |
| 5083 | import re |
| 5084 | repart = re.compile(r'^--([^\n]+)\n(.*?)\n--\1$', re.S | re.M) |
| 5085 | inpart = repart.search(original).group(2) |
| 5086 | outpart = repart.search(result).group(2) |
| 5087 | self.assertEqual(outpart, inpart) |
| 5088 | |
| 5089 | def test_long_headers_as_string(self): |
| 5090 | original, msg = self._msg_and_obj('msg_45.txt') |
| 5091 | result = msg.as_string() |
| 5092 | self._signed_parts_eq(original, result) |
| 5093 | |
| 5094 | def test_long_headers_as_string_maxheaderlen(self): |
| 5095 | original, msg = self._msg_and_obj('msg_45.txt') |
| 5096 | result = msg.as_string(maxheaderlen=60) |
| 5097 | self._signed_parts_eq(original, result) |
| 5098 | |
| 5099 | def test_long_headers_flatten(self): |
| 5100 | original, msg = self._msg_and_obj('msg_45.txt') |
| 5101 | fp = StringIO() |
| 5102 | Generator(fp).flatten(msg) |
| 5103 | result = fp.getvalue() |
| 5104 | self._signed_parts_eq(original, result) |
| 5105 | |
| 5106 | |
Ezio Melotti | b3aedd4 | 2010-11-20 19:04:17 +0000 | [diff] [blame] | 5107 | |
Guido van Rossum | 8b3febe | 2007-08-30 01:15:14 +0000 | [diff] [blame] | 5108 | if __name__ == '__main__': |
R David Murray | 9aaba78 | 2011-03-21 17:17:06 -0400 | [diff] [blame] | 5109 | unittest.main() |