R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1 | """Header value parser implementing various email-related RFC parsing rules. |
| 2 | |
| 3 | The parsing methods defined in this module implement various email related |
| 4 | parsing rules. Principal among them is RFC 5322, which is the followon |
| 5 | to RFC 2822 and primarily a clarification of the former. It also implements |
| 6 | RFC 2047 encoded word decoding. |
| 7 | |
| 8 | RFC 5322 goes to considerable trouble to maintain backward compatibility with |
| 9 | RFC 822 in the parse phase, while cleaning up the structure on the generation |
| 10 | phase. This parser supports correct RFC 5322 generation by tagging white space |
| 11 | as folding white space only when folding is allowed in the non-obsolete rule |
| 12 | sets. Actually, the parser is even more generous when accepting input than RFC |
| 13 | 5322 mandates, following the spirit of Postel's Law, which RFC 5322 encourages. |
| 14 | Where possible deviations from the standard are annotated on the 'defects' |
| 15 | attribute of tokens that deviate. |
| 16 | |
| 17 | The general structure of the parser follows RFC 5322, and uses its terminology |
| 18 | where there is a direct correspondence. Where the implementation requires a |
| 19 | somewhat different structure than that used by the formal grammar, new terms |
| 20 | that mimic the closest existing terms are used. Thus, it really helps to have |
| 21 | a copy of RFC 5322 handy when studying this code. |
| 22 | |
| 23 | Input to the parser is a string that has already been unfolded according to |
| 24 | RFC 5322 rules. According to the RFC this unfolding is the very first step, and |
| 25 | this parser leaves the unfolding step to a higher level message parser, which |
| 26 | will have already detected the line breaks that need unfolding while |
| 27 | determining the beginning and end of each header. |
| 28 | |
| 29 | The output of the parser is a TokenList object, which is a list subclass. A |
| 30 | TokenList is a recursive data structure. The terminal nodes of the structure |
| 31 | are Terminal objects, which are subclasses of str. These do not correspond |
| 32 | directly to terminal objects in the formal grammar, but are instead more |
| 33 | practical higher level combinations of true terminals. |
| 34 | |
| 35 | All TokenList and Terminal objects have a 'value' attribute, which produces the |
| 36 | semantically meaningful value of that part of the parse subtree. The value of |
| 37 | all whitespace tokens (no matter how many sub-tokens they may contain) is a |
| 38 | single space, as per the RFC rules. This includes 'CFWS', which is herein |
| 39 | included in the general class of whitespace tokens. There is one exception to |
| 40 | the rule that whitespace tokens are collapsed into single spaces in values: in |
| 41 | the value of a 'bare-quoted-string' (a quoted-string with no leading or |
| 42 | trailing whitespace), any whitespace that appeared between the quotation marks |
| 43 | is preserved in the returned value. Note that in all Terminal strings quoted |
| 44 | pairs are turned into their unquoted values. |
| 45 | |
| 46 | All TokenList and Terminal objects also have a string value, which attempts to |
| 47 | be a "canonical" representation of the RFC-compliant form of the substring that |
| 48 | produced the parsed subtree, including minimal use of quoted pair quoting. |
| 49 | Whitespace runs are not collapsed. |
| 50 | |
| 51 | Comment tokens also have a 'content' attribute providing the string found |
| 52 | between the parens (including any nested comments) with whitespace preserved. |
| 53 | |
| 54 | All TokenList and Terminal objects have a 'defects' attribute which is a |
| 55 | possibly empty list all of the defects found while creating the token. Defects |
| 56 | may appear on any token in the tree, and a composite list of all defects in the |
| 57 | subtree is available through the 'all_defects' attribute of any node. (For |
| 58 | Terminal notes x.defects == x.all_defects.) |
| 59 | |
| 60 | Each object in a parse tree is called a 'token', and each has a 'token_type' |
| 61 | attribute that gives the name from the RFC 5322 grammar that it represents. |
| 62 | Not all RFC 5322 nodes are produced, and there is one non-RFC 5322 node that |
| 63 | may be produced: 'ptext'. A 'ptext' is a string of printable ascii characters. |
| 64 | It is returned in place of lists of (ctext/quoted-pair) and |
| 65 | (qtext/quoted-pair). |
| 66 | |
| 67 | XXX: provide complete list of token types. |
| 68 | """ |
| 69 | |
| 70 | import re |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 71 | import urllib # For urllib.parse.unquote |
R David Murray | 65171b2 | 2013-07-11 15:52:57 -0400 | [diff] [blame] | 72 | from string import hexdigits |
R David Murray | 7d0325d | 2015-03-29 21:53:05 -0400 | [diff] [blame] | 73 | from operator import itemgetter |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 74 | from email import _encoded_words as _ew |
| 75 | from email import errors |
| 76 | from email import utils |
| 77 | |
| 78 | # |
| 79 | # Useful constants and functions |
| 80 | # |
| 81 | |
| 82 | WSP = set(' \t') |
| 83 | CFWS_LEADER = WSP | set('(') |
| 84 | SPECIALS = set(r'()<>@,:;.\"[]') |
| 85 | ATOM_ENDS = SPECIALS | WSP |
| 86 | DOT_ATOM_ENDS = ATOM_ENDS - set('.') |
| 87 | # '.', '"', and '(' do not end phrases in order to support obs-phrase |
| 88 | PHRASE_ENDS = SPECIALS - set('."(') |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 89 | TSPECIALS = (SPECIALS | set('/?=')) - set('.') |
| 90 | TOKEN_ENDS = TSPECIALS | WSP |
| 91 | ASPECIALS = TSPECIALS | set("*'%") |
| 92 | ATTRIBUTE_ENDS = ASPECIALS | WSP |
| 93 | EXTENDED_ATTRIBUTE_ENDS = ATTRIBUTE_ENDS - set('%') |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 94 | |
| 95 | def quote_string(value): |
| 96 | return '"'+str(value).replace('\\', '\\\\').replace('"', r'\"')+'"' |
| 97 | |
| 98 | # |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 99 | # TokenList and its subclasses |
| 100 | # |
| 101 | |
| 102 | class TokenList(list): |
| 103 | |
| 104 | token_type = None |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 105 | syntactic_break = True |
| 106 | ew_combine_allowed = True |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 107 | |
| 108 | def __init__(self, *args, **kw): |
| 109 | super().__init__(*args, **kw) |
| 110 | self.defects = [] |
| 111 | |
| 112 | def __str__(self): |
| 113 | return ''.join(str(x) for x in self) |
| 114 | |
| 115 | def __repr__(self): |
| 116 | return '{}({})'.format(self.__class__.__name__, |
| 117 | super().__repr__()) |
| 118 | |
| 119 | @property |
| 120 | def value(self): |
| 121 | return ''.join(x.value for x in self if x.value) |
| 122 | |
| 123 | @property |
| 124 | def all_defects(self): |
| 125 | return sum((x.all_defects for x in self), self.defects) |
| 126 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 127 | def startswith_fws(self): |
| 128 | return self[0].startswith_fws() |
| 129 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 130 | @property |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 131 | def as_ew_allowed(self): |
| 132 | """True if all top level tokens of this part may be RFC2047 encoded.""" |
| 133 | return all(part.as_ew_allowed for part in self) |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 134 | |
| 135 | @property |
| 136 | def comments(self): |
| 137 | comments = [] |
| 138 | for token in self: |
| 139 | comments.extend(token.comments) |
| 140 | return comments |
| 141 | |
| 142 | def fold(self, *, policy): |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 143 | return _refold_parse_tree(self, policy=policy) |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 144 | |
| 145 | def pprint(self, indent=''): |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 146 | print(self.ppstr(indent=indent)) |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 147 | |
| 148 | def ppstr(self, indent=''): |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 149 | return '\n'.join(self._pp(indent=indent)) |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 150 | |
| 151 | def _pp(self, indent=''): |
| 152 | yield '{}{}/{}('.format( |
| 153 | indent, |
| 154 | self.__class__.__name__, |
| 155 | self.token_type) |
| 156 | for token in self: |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 157 | if not hasattr(token, '_pp'): |
| 158 | yield (indent + ' !! invalid element in token ' |
| 159 | 'list: {!r}'.format(token)) |
| 160 | else: |
Philip Jenvey | 4993cc0 | 2012-10-01 12:53:43 -0700 | [diff] [blame] | 161 | yield from token._pp(indent+' ') |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 162 | if self.defects: |
| 163 | extra = ' Defects: {}'.format(self.defects) |
| 164 | else: |
| 165 | extra = '' |
| 166 | yield '{}){}'.format(indent, extra) |
| 167 | |
| 168 | |
| 169 | class WhiteSpaceTokenList(TokenList): |
| 170 | |
| 171 | @property |
| 172 | def value(self): |
| 173 | return ' ' |
| 174 | |
| 175 | @property |
| 176 | def comments(self): |
| 177 | return [x.content for x in self if x.token_type=='comment'] |
| 178 | |
| 179 | |
| 180 | class UnstructuredTokenList(TokenList): |
| 181 | |
| 182 | token_type = 'unstructured' |
| 183 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 184 | |
| 185 | class Phrase(TokenList): |
| 186 | |
| 187 | token_type = 'phrase' |
| 188 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 189 | class Word(TokenList): |
| 190 | |
| 191 | token_type = 'word' |
| 192 | |
| 193 | |
| 194 | class CFWSList(WhiteSpaceTokenList): |
| 195 | |
| 196 | token_type = 'cfws' |
| 197 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 198 | |
| 199 | class Atom(TokenList): |
| 200 | |
| 201 | token_type = 'atom' |
| 202 | |
| 203 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 204 | class Token(TokenList): |
| 205 | |
| 206 | token_type = 'token' |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 207 | encode_as_ew = False |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 208 | |
| 209 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 210 | class EncodedWord(TokenList): |
| 211 | |
| 212 | token_type = 'encoded-word' |
| 213 | cte = None |
| 214 | charset = None |
| 215 | lang = None |
| 216 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 217 | |
| 218 | class QuotedString(TokenList): |
| 219 | |
| 220 | token_type = 'quoted-string' |
| 221 | |
| 222 | @property |
| 223 | def content(self): |
| 224 | for x in self: |
| 225 | if x.token_type == 'bare-quoted-string': |
| 226 | return x.value |
| 227 | |
| 228 | @property |
| 229 | def quoted_value(self): |
| 230 | res = [] |
| 231 | for x in self: |
| 232 | if x.token_type == 'bare-quoted-string': |
| 233 | res.append(str(x)) |
| 234 | else: |
| 235 | res.append(x.value) |
| 236 | return ''.join(res) |
| 237 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 238 | @property |
| 239 | def stripped_value(self): |
| 240 | for token in self: |
| 241 | if token.token_type == 'bare-quoted-string': |
| 242 | return token.value |
| 243 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 244 | |
| 245 | class BareQuotedString(QuotedString): |
| 246 | |
| 247 | token_type = 'bare-quoted-string' |
| 248 | |
| 249 | def __str__(self): |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 250 | return quote_string(''.join(str(x) for x in self)) |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 251 | |
| 252 | @property |
| 253 | def value(self): |
| 254 | return ''.join(str(x) for x in self) |
| 255 | |
| 256 | |
| 257 | class Comment(WhiteSpaceTokenList): |
| 258 | |
| 259 | token_type = 'comment' |
| 260 | |
| 261 | def __str__(self): |
| 262 | return ''.join(sum([ |
| 263 | ["("], |
| 264 | [self.quote(x) for x in self], |
| 265 | [")"], |
| 266 | ], [])) |
| 267 | |
| 268 | def quote(self, value): |
| 269 | if value.token_type == 'comment': |
| 270 | return str(value) |
| 271 | return str(value).replace('\\', '\\\\').replace( |
R David Murray | 44b548d | 2016-09-08 13:59:53 -0400 | [diff] [blame] | 272 | '(', r'\(').replace( |
| 273 | ')', r'\)') |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 274 | |
| 275 | @property |
| 276 | def content(self): |
| 277 | return ''.join(str(x) for x in self) |
| 278 | |
| 279 | @property |
| 280 | def comments(self): |
| 281 | return [self.content] |
| 282 | |
| 283 | class AddressList(TokenList): |
| 284 | |
| 285 | token_type = 'address-list' |
| 286 | |
| 287 | @property |
| 288 | def addresses(self): |
| 289 | return [x for x in self if x.token_type=='address'] |
| 290 | |
| 291 | @property |
| 292 | def mailboxes(self): |
| 293 | return sum((x.mailboxes |
| 294 | for x in self if x.token_type=='address'), []) |
| 295 | |
| 296 | @property |
| 297 | def all_mailboxes(self): |
| 298 | return sum((x.all_mailboxes |
| 299 | for x in self if x.token_type=='address'), []) |
| 300 | |
| 301 | |
| 302 | class Address(TokenList): |
| 303 | |
| 304 | token_type = 'address' |
| 305 | |
| 306 | @property |
| 307 | def display_name(self): |
| 308 | if self[0].token_type == 'group': |
| 309 | return self[0].display_name |
| 310 | |
| 311 | @property |
| 312 | def mailboxes(self): |
| 313 | if self[0].token_type == 'mailbox': |
| 314 | return [self[0]] |
| 315 | elif self[0].token_type == 'invalid-mailbox': |
| 316 | return [] |
| 317 | return self[0].mailboxes |
| 318 | |
| 319 | @property |
| 320 | def all_mailboxes(self): |
| 321 | if self[0].token_type == 'mailbox': |
| 322 | return [self[0]] |
| 323 | elif self[0].token_type == 'invalid-mailbox': |
| 324 | return [self[0]] |
| 325 | return self[0].all_mailboxes |
| 326 | |
| 327 | class MailboxList(TokenList): |
| 328 | |
| 329 | token_type = 'mailbox-list' |
| 330 | |
| 331 | @property |
| 332 | def mailboxes(self): |
| 333 | return [x for x in self if x.token_type=='mailbox'] |
| 334 | |
| 335 | @property |
| 336 | def all_mailboxes(self): |
| 337 | return [x for x in self |
| 338 | if x.token_type in ('mailbox', 'invalid-mailbox')] |
| 339 | |
| 340 | |
| 341 | class GroupList(TokenList): |
| 342 | |
| 343 | token_type = 'group-list' |
| 344 | |
| 345 | @property |
| 346 | def mailboxes(self): |
| 347 | if not self or self[0].token_type != 'mailbox-list': |
| 348 | return [] |
| 349 | return self[0].mailboxes |
| 350 | |
| 351 | @property |
| 352 | def all_mailboxes(self): |
| 353 | if not self or self[0].token_type != 'mailbox-list': |
| 354 | return [] |
| 355 | return self[0].all_mailboxes |
| 356 | |
| 357 | |
| 358 | class Group(TokenList): |
| 359 | |
| 360 | token_type = "group" |
| 361 | |
| 362 | @property |
| 363 | def mailboxes(self): |
| 364 | if self[2].token_type != 'group-list': |
| 365 | return [] |
| 366 | return self[2].mailboxes |
| 367 | |
| 368 | @property |
| 369 | def all_mailboxes(self): |
| 370 | if self[2].token_type != 'group-list': |
| 371 | return [] |
| 372 | return self[2].all_mailboxes |
| 373 | |
| 374 | @property |
| 375 | def display_name(self): |
| 376 | return self[0].display_name |
| 377 | |
| 378 | |
| 379 | class NameAddr(TokenList): |
| 380 | |
| 381 | token_type = 'name-addr' |
| 382 | |
| 383 | @property |
| 384 | def display_name(self): |
| 385 | if len(self) == 1: |
| 386 | return None |
| 387 | return self[0].display_name |
| 388 | |
| 389 | @property |
| 390 | def local_part(self): |
| 391 | return self[-1].local_part |
| 392 | |
| 393 | @property |
| 394 | def domain(self): |
| 395 | return self[-1].domain |
| 396 | |
| 397 | @property |
| 398 | def route(self): |
| 399 | return self[-1].route |
| 400 | |
| 401 | @property |
| 402 | def addr_spec(self): |
| 403 | return self[-1].addr_spec |
| 404 | |
| 405 | |
| 406 | class AngleAddr(TokenList): |
| 407 | |
| 408 | token_type = 'angle-addr' |
| 409 | |
| 410 | @property |
| 411 | def local_part(self): |
| 412 | for x in self: |
| 413 | if x.token_type == 'addr-spec': |
| 414 | return x.local_part |
| 415 | |
| 416 | @property |
| 417 | def domain(self): |
| 418 | for x in self: |
| 419 | if x.token_type == 'addr-spec': |
| 420 | return x.domain |
| 421 | |
| 422 | @property |
| 423 | def route(self): |
| 424 | for x in self: |
| 425 | if x.token_type == 'obs-route': |
| 426 | return x.domains |
| 427 | |
| 428 | @property |
| 429 | def addr_spec(self): |
| 430 | for x in self: |
| 431 | if x.token_type == 'addr-spec': |
jayyyin | aa218d1 | 2018-01-29 13:07:44 -0500 | [diff] [blame] | 432 | if x.local_part: |
| 433 | return x.addr_spec |
| 434 | else: |
| 435 | return quote_string(x.local_part) + x.addr_spec |
R David Murray | 032eed3 | 2012-05-26 14:31:12 -0400 | [diff] [blame] | 436 | else: |
| 437 | return '<>' |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 438 | |
| 439 | |
| 440 | class ObsRoute(TokenList): |
| 441 | |
| 442 | token_type = 'obs-route' |
| 443 | |
| 444 | @property |
| 445 | def domains(self): |
| 446 | return [x.domain for x in self if x.token_type == 'domain'] |
| 447 | |
| 448 | |
| 449 | class Mailbox(TokenList): |
| 450 | |
| 451 | token_type = 'mailbox' |
| 452 | |
| 453 | @property |
| 454 | def display_name(self): |
| 455 | if self[0].token_type == 'name-addr': |
| 456 | return self[0].display_name |
| 457 | |
| 458 | @property |
| 459 | def local_part(self): |
| 460 | return self[0].local_part |
| 461 | |
| 462 | @property |
| 463 | def domain(self): |
| 464 | return self[0].domain |
| 465 | |
| 466 | @property |
| 467 | def route(self): |
| 468 | if self[0].token_type == 'name-addr': |
| 469 | return self[0].route |
| 470 | |
| 471 | @property |
| 472 | def addr_spec(self): |
| 473 | return self[0].addr_spec |
| 474 | |
| 475 | |
| 476 | class InvalidMailbox(TokenList): |
| 477 | |
| 478 | token_type = 'invalid-mailbox' |
| 479 | |
| 480 | @property |
| 481 | def display_name(self): |
| 482 | return None |
| 483 | |
| 484 | local_part = domain = route = addr_spec = display_name |
| 485 | |
| 486 | |
| 487 | class Domain(TokenList): |
| 488 | |
| 489 | token_type = 'domain' |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 490 | as_ew_allowed = False |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 491 | |
| 492 | @property |
| 493 | def domain(self): |
| 494 | return ''.join(super().value.split()) |
| 495 | |
| 496 | |
| 497 | class DotAtom(TokenList): |
| 498 | |
| 499 | token_type = 'dot-atom' |
| 500 | |
| 501 | |
| 502 | class DotAtomText(TokenList): |
| 503 | |
| 504 | token_type = 'dot-atom-text' |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 505 | as_ew_allowed = True |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 506 | |
| 507 | |
| 508 | class AddrSpec(TokenList): |
| 509 | |
| 510 | token_type = 'addr-spec' |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 511 | as_ew_allowed = False |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 512 | |
| 513 | @property |
| 514 | def local_part(self): |
| 515 | return self[0].local_part |
| 516 | |
| 517 | @property |
| 518 | def domain(self): |
| 519 | if len(self) < 3: |
| 520 | return None |
| 521 | return self[-1].domain |
| 522 | |
| 523 | @property |
| 524 | def value(self): |
| 525 | if len(self) < 3: |
| 526 | return self[0].value |
| 527 | return self[0].value.rstrip()+self[1].value+self[2].value.lstrip() |
| 528 | |
| 529 | @property |
| 530 | def addr_spec(self): |
| 531 | nameset = set(self.local_part) |
| 532 | if len(nameset) > len(nameset-DOT_ATOM_ENDS): |
| 533 | lp = quote_string(self.local_part) |
| 534 | else: |
| 535 | lp = self.local_part |
| 536 | if self.domain is not None: |
| 537 | return lp + '@' + self.domain |
| 538 | return lp |
| 539 | |
| 540 | |
| 541 | class ObsLocalPart(TokenList): |
| 542 | |
| 543 | token_type = 'obs-local-part' |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 544 | as_ew_allowed = False |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 545 | |
| 546 | |
| 547 | class DisplayName(Phrase): |
| 548 | |
| 549 | token_type = 'display-name' |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 550 | ew_combine_allowed = False |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 551 | |
| 552 | @property |
| 553 | def display_name(self): |
| 554 | res = TokenList(self) |
| 555 | if res[0].token_type == 'cfws': |
| 556 | res.pop(0) |
| 557 | else: |
| 558 | if res[0][0].token_type == 'cfws': |
| 559 | res[0] = TokenList(res[0][1:]) |
| 560 | if res[-1].token_type == 'cfws': |
| 561 | res.pop() |
| 562 | else: |
| 563 | if res[-1][-1].token_type == 'cfws': |
| 564 | res[-1] = TokenList(res[-1][:-1]) |
| 565 | return res.value |
| 566 | |
| 567 | @property |
| 568 | def value(self): |
| 569 | quote = False |
| 570 | if self.defects: |
| 571 | quote = True |
| 572 | else: |
| 573 | for x in self: |
| 574 | if x.token_type == 'quoted-string': |
| 575 | quote = True |
| 576 | if quote: |
| 577 | pre = post = '' |
| 578 | if self[0].token_type=='cfws' or self[0][0].token_type=='cfws': |
| 579 | pre = ' ' |
| 580 | if self[-1].token_type=='cfws' or self[-1][-1].token_type=='cfws': |
| 581 | post = ' ' |
| 582 | return pre+quote_string(self.display_name)+post |
| 583 | else: |
| 584 | return super().value |
| 585 | |
| 586 | |
| 587 | class LocalPart(TokenList): |
| 588 | |
| 589 | token_type = 'local-part' |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 590 | as_ew_allowed = False |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 591 | |
| 592 | @property |
| 593 | def value(self): |
| 594 | if self[0].token_type == "quoted-string": |
| 595 | return self[0].quoted_value |
| 596 | else: |
| 597 | return self[0].value |
| 598 | |
| 599 | @property |
| 600 | def local_part(self): |
| 601 | # Strip whitespace from front, back, and around dots. |
| 602 | res = [DOT] |
| 603 | last = DOT |
| 604 | last_is_tl = False |
| 605 | for tok in self[0] + [DOT]: |
| 606 | if tok.token_type == 'cfws': |
| 607 | continue |
| 608 | if (last_is_tl and tok.token_type == 'dot' and |
| 609 | last[-1].token_type == 'cfws'): |
| 610 | res[-1] = TokenList(last[:-1]) |
| 611 | is_tl = isinstance(tok, TokenList) |
| 612 | if (is_tl and last.token_type == 'dot' and |
| 613 | tok[0].token_type == 'cfws'): |
| 614 | res.append(TokenList(tok[1:])) |
| 615 | else: |
| 616 | res.append(tok) |
| 617 | last = res[-1] |
| 618 | last_is_tl = is_tl |
| 619 | res = TokenList(res[1:-1]) |
| 620 | return res.value |
| 621 | |
| 622 | |
| 623 | class DomainLiteral(TokenList): |
| 624 | |
| 625 | token_type = 'domain-literal' |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 626 | as_ew_allowed = False |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 627 | |
| 628 | @property |
| 629 | def domain(self): |
| 630 | return ''.join(super().value.split()) |
| 631 | |
| 632 | @property |
| 633 | def ip(self): |
| 634 | for x in self: |
| 635 | if x.token_type == 'ptext': |
| 636 | return x.value |
| 637 | |
| 638 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 639 | class MIMEVersion(TokenList): |
| 640 | |
| 641 | token_type = 'mime-version' |
| 642 | major = None |
| 643 | minor = None |
| 644 | |
| 645 | |
| 646 | class Parameter(TokenList): |
| 647 | |
| 648 | token_type = 'parameter' |
| 649 | sectioned = False |
| 650 | extended = False |
| 651 | charset = 'us-ascii' |
| 652 | |
| 653 | @property |
| 654 | def section_number(self): |
| 655 | # Because the first token, the attribute (name) eats CFWS, the second |
| 656 | # token is always the section if there is one. |
| 657 | return self[1].number if self.sectioned else 0 |
| 658 | |
| 659 | @property |
| 660 | def param_value(self): |
| 661 | # This is part of the "handle quoted extended parameters" hack. |
| 662 | for token in self: |
| 663 | if token.token_type == 'value': |
| 664 | return token.stripped_value |
| 665 | if token.token_type == 'quoted-string': |
| 666 | for token in token: |
| 667 | if token.token_type == 'bare-quoted-string': |
| 668 | for token in token: |
| 669 | if token.token_type == 'value': |
| 670 | return token.stripped_value |
| 671 | return '' |
| 672 | |
| 673 | |
| 674 | class InvalidParameter(Parameter): |
| 675 | |
| 676 | token_type = 'invalid-parameter' |
| 677 | |
| 678 | |
| 679 | class Attribute(TokenList): |
| 680 | |
| 681 | token_type = 'attribute' |
| 682 | |
| 683 | @property |
| 684 | def stripped_value(self): |
| 685 | for token in self: |
| 686 | if token.token_type.endswith('attrtext'): |
| 687 | return token.value |
| 688 | |
| 689 | class Section(TokenList): |
| 690 | |
| 691 | token_type = 'section' |
| 692 | number = None |
| 693 | |
| 694 | |
| 695 | class Value(TokenList): |
| 696 | |
| 697 | token_type = 'value' |
| 698 | |
| 699 | @property |
| 700 | def stripped_value(self): |
| 701 | token = self[0] |
| 702 | if token.token_type == 'cfws': |
| 703 | token = self[1] |
| 704 | if token.token_type.endswith( |
| 705 | ('quoted-string', 'attribute', 'extended-attribute')): |
| 706 | return token.stripped_value |
| 707 | return self.value |
| 708 | |
| 709 | |
| 710 | class MimeParameters(TokenList): |
| 711 | |
| 712 | token_type = 'mime-parameters' |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 713 | syntactic_break = False |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 714 | |
| 715 | @property |
| 716 | def params(self): |
| 717 | # The RFC specifically states that the ordering of parameters is not |
| 718 | # guaranteed and may be reordered by the transport layer. So we have |
| 719 | # to assume the RFC 2231 pieces can come in any order. However, we |
| 720 | # output them in the order that we first see a given name, which gives |
| 721 | # us a stable __str__. |
Inada Naoki | c95404f | 2019-02-05 17:05:43 +0900 | [diff] [blame] | 722 | params = {} # Using order preserving dict from Python 3.7+ |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 723 | for token in self: |
| 724 | if not token.token_type.endswith('parameter'): |
| 725 | continue |
| 726 | if token[0].token_type != 'attribute': |
| 727 | continue |
| 728 | name = token[0].value.strip() |
| 729 | if name not in params: |
| 730 | params[name] = [] |
| 731 | params[name].append((token.section_number, token)) |
| 732 | for name, parts in params.items(): |
R David Murray | 7d0325d | 2015-03-29 21:53:05 -0400 | [diff] [blame] | 733 | parts = sorted(parts, key=itemgetter(0)) |
| 734 | first_param = parts[0][1] |
| 735 | charset = first_param.charset |
| 736 | # Our arbitrary error recovery is to ignore duplicate parameters, |
| 737 | # to use appearance order if there are duplicate rfc 2231 parts, |
| 738 | # and to ignore gaps. This mimics the error recovery of get_param. |
| 739 | if not first_param.extended and len(parts) > 1: |
| 740 | if parts[1][0] == 0: |
| 741 | parts[1][1].defects.append(errors.InvalidHeaderDefect( |
| 742 | 'duplicate parameter name; duplicate(s) ignored')) |
| 743 | parts = parts[:1] |
| 744 | # Else assume the *0* was missing...note that this is different |
| 745 | # from get_param, but we registered a defect for this earlier. |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 746 | value_parts = [] |
R David Murray | 7d0325d | 2015-03-29 21:53:05 -0400 | [diff] [blame] | 747 | i = 0 |
| 748 | for section_number, param in parts: |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 749 | if section_number != i: |
R David Murray | 7d0325d | 2015-03-29 21:53:05 -0400 | [diff] [blame] | 750 | # We could get fancier here and look for a complete |
| 751 | # duplicate extended parameter and ignore the second one |
| 752 | # seen. But we're not doing that. The old code didn't. |
| 753 | if not param.extended: |
| 754 | param.defects.append(errors.InvalidHeaderDefect( |
| 755 | 'duplicate parameter name; duplicate ignored')) |
| 756 | continue |
| 757 | else: |
| 758 | param.defects.append(errors.InvalidHeaderDefect( |
| 759 | "inconsistent RFC2231 parameter numbering")) |
| 760 | i += 1 |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 761 | value = param.param_value |
| 762 | if param.extended: |
| 763 | try: |
| 764 | value = urllib.parse.unquote_to_bytes(value) |
| 765 | except UnicodeEncodeError: |
| 766 | # source had surrogate escaped bytes. What we do now |
| 767 | # is a bit of an open question. I'm not sure this is |
| 768 | # the best choice, but it is what the old algorithm did |
| 769 | value = urllib.parse.unquote(value, encoding='latin-1') |
| 770 | else: |
| 771 | try: |
| 772 | value = value.decode(charset, 'surrogateescape') |
| 773 | except LookupError: |
| 774 | # XXX: there should really be a custom defect for |
| 775 | # unknown character set to make it easy to find, |
| 776 | # because otherwise unknown charset is a silent |
| 777 | # failure. |
| 778 | value = value.decode('us-ascii', 'surrogateescape') |
| 779 | if utils._has_surrogates(value): |
| 780 | param.defects.append(errors.UndecodableBytesDefect()) |
| 781 | value_parts.append(value) |
| 782 | value = ''.join(value_parts) |
| 783 | yield name, value |
| 784 | |
| 785 | def __str__(self): |
| 786 | params = [] |
| 787 | for name, value in self.params: |
| 788 | if value: |
| 789 | params.append('{}={}'.format(name, quote_string(value))) |
| 790 | else: |
| 791 | params.append(name) |
| 792 | params = '; '.join(params) |
| 793 | return ' ' + params if params else '' |
| 794 | |
| 795 | |
| 796 | class ParameterizedHeaderValue(TokenList): |
| 797 | |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 798 | # Set this false so that the value doesn't wind up on a new line even |
| 799 | # if it and the parameters would fit there but not on the first line. |
| 800 | syntactic_break = False |
| 801 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 802 | @property |
| 803 | def params(self): |
| 804 | for token in reversed(self): |
| 805 | if token.token_type == 'mime-parameters': |
| 806 | return token.params |
| 807 | return {} |
| 808 | |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 809 | |
| 810 | class ContentType(ParameterizedHeaderValue): |
| 811 | |
| 812 | token_type = 'content-type' |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 813 | as_ew_allowed = False |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 814 | maintype = 'text' |
| 815 | subtype = 'plain' |
| 816 | |
| 817 | |
| 818 | class ContentDisposition(ParameterizedHeaderValue): |
| 819 | |
| 820 | token_type = 'content-disposition' |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 821 | as_ew_allowed = False |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 822 | content_disposition = None |
| 823 | |
| 824 | |
| 825 | class ContentTransferEncoding(TokenList): |
| 826 | |
| 827 | token_type = 'content-transfer-encoding' |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 828 | as_ew_allowed = False |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 829 | cte = '7bit' |
| 830 | |
| 831 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 832 | class HeaderLabel(TokenList): |
| 833 | |
| 834 | token_type = 'header-label' |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 835 | as_ew_allowed = False |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 836 | |
| 837 | |
| 838 | class Header(TokenList): |
| 839 | |
| 840 | token_type = 'header' |
| 841 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 842 | |
| 843 | # |
| 844 | # Terminal classes and instances |
| 845 | # |
| 846 | |
| 847 | class Terminal(str): |
| 848 | |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 849 | as_ew_allowed = True |
| 850 | ew_combine_allowed = True |
| 851 | syntactic_break = True |
| 852 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 853 | def __new__(cls, value, token_type): |
| 854 | self = super().__new__(cls, value) |
| 855 | self.token_type = token_type |
| 856 | self.defects = [] |
| 857 | return self |
| 858 | |
| 859 | def __repr__(self): |
| 860 | return "{}({})".format(self.__class__.__name__, super().__repr__()) |
| 861 | |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 862 | def pprint(self): |
| 863 | print(self.__class__.__name__ + '/' + self.token_type) |
| 864 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 865 | @property |
| 866 | def all_defects(self): |
| 867 | return list(self.defects) |
| 868 | |
| 869 | def _pp(self, indent=''): |
| 870 | return ["{}{}/{}({}){}".format( |
| 871 | indent, |
| 872 | self.__class__.__name__, |
| 873 | self.token_type, |
| 874 | super().__repr__(), |
| 875 | '' if not self.defects else ' {}'.format(self.defects), |
| 876 | )] |
| 877 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 878 | def pop_trailing_ws(self): |
| 879 | # This terminates the recursion. |
| 880 | return None |
| 881 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 882 | @property |
| 883 | def comments(self): |
| 884 | return [] |
| 885 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 886 | def __getnewargs__(self): |
| 887 | return(str(self), self.token_type) |
| 888 | |
| 889 | |
| 890 | class WhiteSpaceTerminal(Terminal): |
| 891 | |
| 892 | @property |
| 893 | def value(self): |
| 894 | return ' ' |
| 895 | |
| 896 | def startswith_fws(self): |
| 897 | return True |
| 898 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 899 | |
| 900 | class ValueTerminal(Terminal): |
| 901 | |
| 902 | @property |
| 903 | def value(self): |
| 904 | return self |
| 905 | |
| 906 | def startswith_fws(self): |
| 907 | return False |
| 908 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 909 | |
| 910 | class EWWhiteSpaceTerminal(WhiteSpaceTerminal): |
| 911 | |
| 912 | @property |
| 913 | def value(self): |
| 914 | return '' |
| 915 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 916 | def __str__(self): |
| 917 | return '' |
| 918 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 919 | |
| 920 | # XXX these need to become classes and used as instances so |
| 921 | # that a program can't change them in a parse tree and screw |
| 922 | # up other parse trees. Maybe should have tests for that, too. |
| 923 | DOT = ValueTerminal('.', 'dot') |
| 924 | ListSeparator = ValueTerminal(',', 'list-separator') |
| 925 | RouteComponentMarker = ValueTerminal('@', 'route-component-marker') |
| 926 | |
| 927 | # |
| 928 | # Parser |
| 929 | # |
| 930 | |
Victor Stinner | 765531d | 2013-03-26 01:11:54 +0100 | [diff] [blame] | 931 | # Parse strings according to RFC822/2047/2822/5322 rules. |
| 932 | # |
| 933 | # This is a stateless parser. Each get_XXX function accepts a string and |
| 934 | # returns either a Terminal or a TokenList representing the RFC object named |
| 935 | # by the method and a string containing the remaining unparsed characters |
| 936 | # from the input. Thus a parser method consumes the next syntactic construct |
| 937 | # of a given type and returns a token representing the construct plus the |
| 938 | # unparsed remainder of the input string. |
| 939 | # |
| 940 | # For example, if the first element of a structured header is a 'phrase', |
| 941 | # then: |
| 942 | # |
| 943 | # phrase, value = get_phrase(value) |
| 944 | # |
| 945 | # returns the complete phrase from the start of the string value, plus any |
| 946 | # characters left in the string after the phrase is removed. |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 947 | |
| 948 | _wsp_splitter = re.compile(r'([{}]+)'.format(''.join(WSP))).split |
| 949 | _non_atom_end_matcher = re.compile(r"[^{}]+".format( |
Serhiy Storchaka | 05cb728 | 2017-11-16 12:38:26 +0200 | [diff] [blame] | 950 | re.escape(''.join(ATOM_ENDS)))).match |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 951 | _non_printable_finder = re.compile(r"[\x00-\x20\x7F]").findall |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 952 | _non_token_end_matcher = re.compile(r"[^{}]+".format( |
Serhiy Storchaka | 05cb728 | 2017-11-16 12:38:26 +0200 | [diff] [blame] | 953 | re.escape(''.join(TOKEN_ENDS)))).match |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 954 | _non_attribute_end_matcher = re.compile(r"[^{}]+".format( |
Serhiy Storchaka | 05cb728 | 2017-11-16 12:38:26 +0200 | [diff] [blame] | 955 | re.escape(''.join(ATTRIBUTE_ENDS)))).match |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 956 | _non_extended_attribute_end_matcher = re.compile(r"[^{}]+".format( |
Serhiy Storchaka | 05cb728 | 2017-11-16 12:38:26 +0200 | [diff] [blame] | 957 | re.escape(''.join(EXTENDED_ATTRIBUTE_ENDS)))).match |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 958 | |
| 959 | def _validate_xtext(xtext): |
| 960 | """If input token contains ASCII non-printables, register a defect.""" |
| 961 | |
| 962 | non_printables = _non_printable_finder(xtext) |
| 963 | if non_printables: |
| 964 | xtext.defects.append(errors.NonPrintableDefect(non_printables)) |
| 965 | if utils._has_surrogates(xtext): |
| 966 | xtext.defects.append(errors.UndecodableBytesDefect( |
| 967 | "Non-ASCII characters found in header token")) |
| 968 | |
| 969 | def _get_ptext_to_endchars(value, endchars): |
| 970 | """Scan printables/quoted-pairs until endchars and return unquoted ptext. |
| 971 | |
| 972 | This function turns a run of qcontent, ccontent-without-comments, or |
| 973 | dtext-with-quoted-printables into a single string by unquoting any |
| 974 | quoted printables. It returns the string, the remaining value, and |
| 975 | a flag that is True iff there were any quoted printables decoded. |
| 976 | |
| 977 | """ |
| 978 | fragment, *remainder = _wsp_splitter(value, 1) |
| 979 | vchars = [] |
| 980 | escape = False |
| 981 | had_qp = False |
| 982 | for pos in range(len(fragment)): |
| 983 | if fragment[pos] == '\\': |
| 984 | if escape: |
| 985 | escape = False |
| 986 | had_qp = True |
| 987 | else: |
| 988 | escape = True |
| 989 | continue |
| 990 | if escape: |
| 991 | escape = False |
| 992 | elif fragment[pos] in endchars: |
| 993 | break |
| 994 | vchars.append(fragment[pos]) |
| 995 | else: |
| 996 | pos = pos + 1 |
| 997 | return ''.join(vchars), ''.join([fragment[pos:]] + remainder), had_qp |
| 998 | |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 999 | def get_fws(value): |
| 1000 | """FWS = 1*WSP |
| 1001 | |
| 1002 | This isn't the RFC definition. We're using fws to represent tokens where |
| 1003 | folding can be done, but when we are parsing the *un*folding has already |
| 1004 | been done so we don't need to watch out for CRLF. |
| 1005 | |
| 1006 | """ |
| 1007 | newvalue = value.lstrip() |
| 1008 | fws = WhiteSpaceTerminal(value[:len(value)-len(newvalue)], 'fws') |
| 1009 | return fws, newvalue |
| 1010 | |
| 1011 | def get_encoded_word(value): |
| 1012 | """ encoded-word = "=?" charset "?" encoding "?" encoded-text "?=" |
| 1013 | |
| 1014 | """ |
| 1015 | ew = EncodedWord() |
| 1016 | if not value.startswith('=?'): |
| 1017 | raise errors.HeaderParseError( |
| 1018 | "expected encoded word but found {}".format(value)) |
| 1019 | tok, *remainder = value[2:].split('?=', 1) |
| 1020 | if tok == value[2:]: |
| 1021 | raise errors.HeaderParseError( |
| 1022 | "expected encoded word but found {}".format(value)) |
| 1023 | remstr = ''.join(remainder) |
R David Murray | 65171b2 | 2013-07-11 15:52:57 -0400 | [diff] [blame] | 1024 | if len(remstr) > 1 and remstr[0] in hexdigits and remstr[1] in hexdigits: |
| 1025 | # The ? after the CTE was followed by an encoded word escape (=XX). |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1026 | rest, *remainder = remstr.split('?=', 1) |
| 1027 | tok = tok + '?=' + rest |
| 1028 | if len(tok.split()) > 1: |
| 1029 | ew.defects.append(errors.InvalidHeaderDefect( |
| 1030 | "whitespace inside encoded word")) |
| 1031 | ew.cte = value |
| 1032 | value = ''.join(remainder) |
| 1033 | try: |
| 1034 | text, charset, lang, defects = _ew.decode('=?' + tok + '?=') |
| 1035 | except ValueError: |
| 1036 | raise errors.HeaderParseError( |
| 1037 | "encoded word format invalid: '{}'".format(ew.cte)) |
| 1038 | ew.charset = charset |
| 1039 | ew.lang = lang |
| 1040 | ew.defects.extend(defects) |
| 1041 | while text: |
| 1042 | if text[0] in WSP: |
| 1043 | token, text = get_fws(text) |
| 1044 | ew.append(token) |
| 1045 | continue |
| 1046 | chars, *remainder = _wsp_splitter(text, 1) |
| 1047 | vtext = ValueTerminal(chars, 'vtext') |
| 1048 | _validate_xtext(vtext) |
| 1049 | ew.append(vtext) |
| 1050 | text = ''.join(remainder) |
| 1051 | return ew, value |
| 1052 | |
| 1053 | def get_unstructured(value): |
| 1054 | """unstructured = (*([FWS] vchar) *WSP) / obs-unstruct |
| 1055 | obs-unstruct = *((*LF *CR *(obs-utext) *LF *CR)) / FWS) |
| 1056 | obs-utext = %d0 / obs-NO-WS-CTL / LF / CR |
| 1057 | |
| 1058 | obs-NO-WS-CTL is control characters except WSP/CR/LF. |
| 1059 | |
| 1060 | So, basically, we have printable runs, plus control characters or nulls in |
| 1061 | the obsolete syntax, separated by whitespace. Since RFC 2047 uses the |
| 1062 | obsolete syntax in its specification, but requires whitespace on either |
| 1063 | side of the encoded words, I can see no reason to need to separate the |
| 1064 | non-printable-non-whitespace from the printable runs if they occur, so we |
| 1065 | parse this into xtext tokens separated by WSP tokens. |
| 1066 | |
| 1067 | Because an 'unstructured' value must by definition constitute the entire |
| 1068 | value, this 'get' routine does not return a remaining value, only the |
| 1069 | parsed TokenList. |
| 1070 | |
| 1071 | """ |
| 1072 | # XXX: but what about bare CR and LF? They might signal the start or |
R David Murray | 65171b2 | 2013-07-11 15:52:57 -0400 | [diff] [blame] | 1073 | # end of an encoded word. YAGNI for now, since our current parsers |
| 1074 | # will never send us strings with bare CR or LF. |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1075 | |
| 1076 | unstructured = UnstructuredTokenList() |
| 1077 | while value: |
| 1078 | if value[0] in WSP: |
| 1079 | token, value = get_fws(value) |
| 1080 | unstructured.append(token) |
| 1081 | continue |
| 1082 | if value.startswith('=?'): |
| 1083 | try: |
| 1084 | token, value = get_encoded_word(value) |
| 1085 | except errors.HeaderParseError: |
R David Murray | 65171b2 | 2013-07-11 15:52:57 -0400 | [diff] [blame] | 1086 | # XXX: Need to figure out how to register defects when |
| 1087 | # appropriate here. |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1088 | pass |
| 1089 | else: |
| 1090 | have_ws = True |
| 1091 | if len(unstructured) > 0: |
| 1092 | if unstructured[-1].token_type != 'fws': |
| 1093 | unstructured.defects.append(errors.InvalidHeaderDefect( |
| 1094 | "missing whitespace before encoded word")) |
| 1095 | have_ws = False |
| 1096 | if have_ws and len(unstructured) > 1: |
| 1097 | if unstructured[-2].token_type == 'encoded-word': |
| 1098 | unstructured[-1] = EWWhiteSpaceTerminal( |
| 1099 | unstructured[-1], 'fws') |
| 1100 | unstructured.append(token) |
| 1101 | continue |
| 1102 | tok, *remainder = _wsp_splitter(value, 1) |
| 1103 | vtext = ValueTerminal(tok, 'vtext') |
| 1104 | _validate_xtext(vtext) |
| 1105 | unstructured.append(vtext) |
| 1106 | value = ''.join(remainder) |
| 1107 | return unstructured |
| 1108 | |
| 1109 | def get_qp_ctext(value): |
R David Murray | 44b548d | 2016-09-08 13:59:53 -0400 | [diff] [blame] | 1110 | r"""ctext = <printable ascii except \ ( )> |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1111 | |
| 1112 | This is not the RFC ctext, since we are handling nested comments in comment |
| 1113 | and unquoting quoted-pairs here. We allow anything except the '()' |
| 1114 | characters, but if we find any ASCII other than the RFC defined printable |
Serhiy Storchaka | 6a7b3a7 | 2016-04-17 08:32:47 +0300 | [diff] [blame] | 1115 | ASCII, a NonPrintableDefect is added to the token's defects list. Since |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1116 | quoted pairs are converted to their unquoted values, what is returned is |
| 1117 | a 'ptext' token. In this case it is a WhiteSpaceTerminal, so it's value |
| 1118 | is ' '. |
| 1119 | |
| 1120 | """ |
| 1121 | ptext, value, _ = _get_ptext_to_endchars(value, '()') |
| 1122 | ptext = WhiteSpaceTerminal(ptext, 'ptext') |
| 1123 | _validate_xtext(ptext) |
| 1124 | return ptext, value |
| 1125 | |
| 1126 | def get_qcontent(value): |
| 1127 | """qcontent = qtext / quoted-pair |
| 1128 | |
| 1129 | We allow anything except the DQUOTE character, but if we find any ASCII |
Serhiy Storchaka | 6a7b3a7 | 2016-04-17 08:32:47 +0300 | [diff] [blame] | 1130 | other than the RFC defined printable ASCII, a NonPrintableDefect is |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1131 | added to the token's defects list. Any quoted pairs are converted to their |
| 1132 | unquoted values, so what is returned is a 'ptext' token. In this case it |
| 1133 | is a ValueTerminal. |
| 1134 | |
| 1135 | """ |
| 1136 | ptext, value, _ = _get_ptext_to_endchars(value, '"') |
| 1137 | ptext = ValueTerminal(ptext, 'ptext') |
| 1138 | _validate_xtext(ptext) |
| 1139 | return ptext, value |
| 1140 | |
| 1141 | def get_atext(value): |
| 1142 | """atext = <matches _atext_matcher> |
| 1143 | |
| 1144 | We allow any non-ATOM_ENDS in atext, but add an InvalidATextDefect to |
| 1145 | the token's defects list if we find non-atext characters. |
| 1146 | """ |
| 1147 | m = _non_atom_end_matcher(value) |
| 1148 | if not m: |
| 1149 | raise errors.HeaderParseError( |
| 1150 | "expected atext but found '{}'".format(value)) |
| 1151 | atext = m.group() |
| 1152 | value = value[len(atext):] |
| 1153 | atext = ValueTerminal(atext, 'atext') |
| 1154 | _validate_xtext(atext) |
| 1155 | return atext, value |
| 1156 | |
| 1157 | def get_bare_quoted_string(value): |
| 1158 | """bare-quoted-string = DQUOTE *([FWS] qcontent) [FWS] DQUOTE |
| 1159 | |
| 1160 | A quoted-string without the leading or trailing white space. Its |
| 1161 | value is the text between the quote marks, with whitespace |
| 1162 | preserved and quoted pairs decoded. |
| 1163 | """ |
| 1164 | if value[0] != '"': |
| 1165 | raise errors.HeaderParseError( |
| 1166 | "expected '\"' but found '{}'".format(value)) |
| 1167 | bare_quoted_string = BareQuotedString() |
| 1168 | value = value[1:] |
jayyyin | aa218d1 | 2018-01-29 13:07:44 -0500 | [diff] [blame] | 1169 | if value[0] == '"': |
| 1170 | token, value = get_qcontent(value) |
| 1171 | bare_quoted_string.append(token) |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1172 | while value and value[0] != '"': |
| 1173 | if value[0] in WSP: |
| 1174 | token, value = get_fws(value) |
R David Murray | 0400d33 | 2014-02-08 13:12:00 -0500 | [diff] [blame] | 1175 | elif value[:2] == '=?': |
| 1176 | try: |
| 1177 | token, value = get_encoded_word(value) |
| 1178 | bare_quoted_string.defects.append(errors.InvalidHeaderDefect( |
| 1179 | "encoded word inside quoted string")) |
| 1180 | except errors.HeaderParseError: |
| 1181 | token, value = get_qcontent(value) |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1182 | else: |
| 1183 | token, value = get_qcontent(value) |
| 1184 | bare_quoted_string.append(token) |
| 1185 | if not value: |
| 1186 | bare_quoted_string.defects.append(errors.InvalidHeaderDefect( |
| 1187 | "end of header inside quoted string")) |
| 1188 | return bare_quoted_string, value |
| 1189 | return bare_quoted_string, value[1:] |
| 1190 | |
| 1191 | def get_comment(value): |
| 1192 | """comment = "(" *([FWS] ccontent) [FWS] ")" |
| 1193 | ccontent = ctext / quoted-pair / comment |
| 1194 | |
| 1195 | We handle nested comments here, and quoted-pair in our qp-ctext routine. |
| 1196 | """ |
| 1197 | if value and value[0] != '(': |
| 1198 | raise errors.HeaderParseError( |
| 1199 | "expected '(' but found '{}'".format(value)) |
| 1200 | comment = Comment() |
| 1201 | value = value[1:] |
| 1202 | while value and value[0] != ")": |
| 1203 | if value[0] in WSP: |
| 1204 | token, value = get_fws(value) |
| 1205 | elif value[0] == '(': |
| 1206 | token, value = get_comment(value) |
| 1207 | else: |
| 1208 | token, value = get_qp_ctext(value) |
| 1209 | comment.append(token) |
| 1210 | if not value: |
| 1211 | comment.defects.append(errors.InvalidHeaderDefect( |
| 1212 | "end of header inside comment")) |
| 1213 | return comment, value |
| 1214 | return comment, value[1:] |
| 1215 | |
| 1216 | def get_cfws(value): |
| 1217 | """CFWS = (1*([FWS] comment) [FWS]) / FWS |
| 1218 | |
| 1219 | """ |
| 1220 | cfws = CFWSList() |
| 1221 | while value and value[0] in CFWS_LEADER: |
| 1222 | if value[0] in WSP: |
| 1223 | token, value = get_fws(value) |
| 1224 | else: |
| 1225 | token, value = get_comment(value) |
| 1226 | cfws.append(token) |
| 1227 | return cfws, value |
| 1228 | |
| 1229 | def get_quoted_string(value): |
| 1230 | """quoted-string = [CFWS] <bare-quoted-string> [CFWS] |
| 1231 | |
| 1232 | 'bare-quoted-string' is an intermediate class defined by this |
| 1233 | parser and not by the RFC grammar. It is the quoted string |
| 1234 | without any attached CFWS. |
| 1235 | """ |
| 1236 | quoted_string = QuotedString() |
| 1237 | if value and value[0] in CFWS_LEADER: |
| 1238 | token, value = get_cfws(value) |
| 1239 | quoted_string.append(token) |
| 1240 | token, value = get_bare_quoted_string(value) |
| 1241 | quoted_string.append(token) |
| 1242 | if value and value[0] in CFWS_LEADER: |
| 1243 | token, value = get_cfws(value) |
| 1244 | quoted_string.append(token) |
| 1245 | return quoted_string, value |
| 1246 | |
| 1247 | def get_atom(value): |
| 1248 | """atom = [CFWS] 1*atext [CFWS] |
| 1249 | |
R David Murray | 923512f | 2013-07-12 16:00:28 -0400 | [diff] [blame] | 1250 | An atom could be an rfc2047 encoded word. |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1251 | """ |
| 1252 | atom = Atom() |
| 1253 | if value and value[0] in CFWS_LEADER: |
| 1254 | token, value = get_cfws(value) |
| 1255 | atom.append(token) |
| 1256 | if value and value[0] in ATOM_ENDS: |
| 1257 | raise errors.HeaderParseError( |
| 1258 | "expected atom but found '{}'".format(value)) |
R David Murray | 923512f | 2013-07-12 16:00:28 -0400 | [diff] [blame] | 1259 | if value.startswith('=?'): |
| 1260 | try: |
| 1261 | token, value = get_encoded_word(value) |
| 1262 | except errors.HeaderParseError: |
| 1263 | # XXX: need to figure out how to register defects when |
| 1264 | # appropriate here. |
| 1265 | token, value = get_atext(value) |
| 1266 | else: |
| 1267 | token, value = get_atext(value) |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1268 | atom.append(token) |
| 1269 | if value and value[0] in CFWS_LEADER: |
| 1270 | token, value = get_cfws(value) |
| 1271 | atom.append(token) |
| 1272 | return atom, value |
| 1273 | |
| 1274 | def get_dot_atom_text(value): |
| 1275 | """ dot-text = 1*atext *("." 1*atext) |
| 1276 | |
| 1277 | """ |
| 1278 | dot_atom_text = DotAtomText() |
| 1279 | if not value or value[0] in ATOM_ENDS: |
| 1280 | raise errors.HeaderParseError("expected atom at a start of " |
| 1281 | "dot-atom-text but found '{}'".format(value)) |
| 1282 | while value and value[0] not in ATOM_ENDS: |
| 1283 | token, value = get_atext(value) |
| 1284 | dot_atom_text.append(token) |
| 1285 | if value and value[0] == '.': |
| 1286 | dot_atom_text.append(DOT) |
| 1287 | value = value[1:] |
| 1288 | if dot_atom_text[-1] is DOT: |
| 1289 | raise errors.HeaderParseError("expected atom at end of dot-atom-text " |
| 1290 | "but found '{}'".format('.'+value)) |
| 1291 | return dot_atom_text, value |
| 1292 | |
| 1293 | def get_dot_atom(value): |
| 1294 | """ dot-atom = [CFWS] dot-atom-text [CFWS] |
| 1295 | |
R David Murray | 923512f | 2013-07-12 16:00:28 -0400 | [diff] [blame] | 1296 | Any place we can have a dot atom, we could instead have an rfc2047 encoded |
| 1297 | word. |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1298 | """ |
| 1299 | dot_atom = DotAtom() |
| 1300 | if value[0] in CFWS_LEADER: |
| 1301 | token, value = get_cfws(value) |
| 1302 | dot_atom.append(token) |
R David Murray | 923512f | 2013-07-12 16:00:28 -0400 | [diff] [blame] | 1303 | if value.startswith('=?'): |
| 1304 | try: |
| 1305 | token, value = get_encoded_word(value) |
| 1306 | except errors.HeaderParseError: |
| 1307 | # XXX: need to figure out how to register defects when |
| 1308 | # appropriate here. |
| 1309 | token, value = get_dot_atom_text(value) |
| 1310 | else: |
| 1311 | token, value = get_dot_atom_text(value) |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1312 | dot_atom.append(token) |
| 1313 | if value and value[0] in CFWS_LEADER: |
| 1314 | token, value = get_cfws(value) |
| 1315 | dot_atom.append(token) |
| 1316 | return dot_atom, value |
| 1317 | |
| 1318 | def get_word(value): |
| 1319 | """word = atom / quoted-string |
| 1320 | |
| 1321 | Either atom or quoted-string may start with CFWS. We have to peel off this |
| 1322 | CFWS first to determine which type of word to parse. Afterward we splice |
| 1323 | the leading CFWS, if any, into the parsed sub-token. |
| 1324 | |
| 1325 | If neither an atom or a quoted-string is found before the next special, a |
| 1326 | HeaderParseError is raised. |
| 1327 | |
| 1328 | The token returned is either an Atom or a QuotedString, as appropriate. |
| 1329 | This means the 'word' level of the formal grammar is not represented in the |
| 1330 | parse tree; this is because having that extra layer when manipulating the |
| 1331 | parse tree is more confusing than it is helpful. |
| 1332 | |
| 1333 | """ |
| 1334 | if value[0] in CFWS_LEADER: |
| 1335 | leader, value = get_cfws(value) |
| 1336 | else: |
| 1337 | leader = None |
| 1338 | if value[0]=='"': |
| 1339 | token, value = get_quoted_string(value) |
| 1340 | elif value[0] in SPECIALS: |
| 1341 | raise errors.HeaderParseError("Expected 'atom' or 'quoted-string' " |
| 1342 | "but found '{}'".format(value)) |
| 1343 | else: |
| 1344 | token, value = get_atom(value) |
| 1345 | if leader is not None: |
| 1346 | token[:0] = [leader] |
| 1347 | return token, value |
| 1348 | |
| 1349 | def get_phrase(value): |
| 1350 | """ phrase = 1*word / obs-phrase |
| 1351 | obs-phrase = word *(word / "." / CFWS) |
| 1352 | |
| 1353 | This means a phrase can be a sequence of words, periods, and CFWS in any |
| 1354 | order as long as it starts with at least one word. If anything other than |
| 1355 | words is detected, an ObsoleteHeaderDefect is added to the token's defect |
| 1356 | list. We also accept a phrase that starts with CFWS followed by a dot; |
| 1357 | this is registered as an InvalidHeaderDefect, since it is not supported by |
| 1358 | even the obsolete grammar. |
| 1359 | |
| 1360 | """ |
| 1361 | phrase = Phrase() |
| 1362 | try: |
| 1363 | token, value = get_word(value) |
| 1364 | phrase.append(token) |
| 1365 | except errors.HeaderParseError: |
| 1366 | phrase.defects.append(errors.InvalidHeaderDefect( |
| 1367 | "phrase does not start with word")) |
| 1368 | while value and value[0] not in PHRASE_ENDS: |
| 1369 | if value[0]=='.': |
| 1370 | phrase.append(DOT) |
| 1371 | phrase.defects.append(errors.ObsoleteHeaderDefect( |
| 1372 | "period in 'phrase'")) |
| 1373 | value = value[1:] |
| 1374 | else: |
| 1375 | try: |
| 1376 | token, value = get_word(value) |
| 1377 | except errors.HeaderParseError: |
| 1378 | if value[0] in CFWS_LEADER: |
| 1379 | token, value = get_cfws(value) |
| 1380 | phrase.defects.append(errors.ObsoleteHeaderDefect( |
| 1381 | "comment found without atom")) |
| 1382 | else: |
| 1383 | raise |
| 1384 | phrase.append(token) |
| 1385 | return phrase, value |
| 1386 | |
| 1387 | def get_local_part(value): |
| 1388 | """ local-part = dot-atom / quoted-string / obs-local-part |
| 1389 | |
| 1390 | """ |
| 1391 | local_part = LocalPart() |
| 1392 | leader = None |
| 1393 | if value[0] in CFWS_LEADER: |
| 1394 | leader, value = get_cfws(value) |
| 1395 | if not value: |
| 1396 | raise errors.HeaderParseError( |
| 1397 | "expected local-part but found '{}'".format(value)) |
| 1398 | try: |
| 1399 | token, value = get_dot_atom(value) |
| 1400 | except errors.HeaderParseError: |
| 1401 | try: |
| 1402 | token, value = get_word(value) |
| 1403 | except errors.HeaderParseError: |
| 1404 | if value[0] != '\\' and value[0] in PHRASE_ENDS: |
| 1405 | raise |
| 1406 | token = TokenList() |
| 1407 | if leader is not None: |
| 1408 | token[:0] = [leader] |
| 1409 | local_part.append(token) |
| 1410 | if value and (value[0]=='\\' or value[0] not in PHRASE_ENDS): |
| 1411 | obs_local_part, value = get_obs_local_part(str(local_part) + value) |
| 1412 | if obs_local_part.token_type == 'invalid-obs-local-part': |
| 1413 | local_part.defects.append(errors.InvalidHeaderDefect( |
| 1414 | "local-part is not dot-atom, quoted-string, or obs-local-part")) |
| 1415 | else: |
| 1416 | local_part.defects.append(errors.ObsoleteHeaderDefect( |
| 1417 | "local-part is not a dot-atom (contains CFWS)")) |
| 1418 | local_part[0] = obs_local_part |
| 1419 | try: |
| 1420 | local_part.value.encode('ascii') |
| 1421 | except UnicodeEncodeError: |
| 1422 | local_part.defects.append(errors.NonASCIILocalPartDefect( |
| 1423 | "local-part contains non-ASCII characters)")) |
| 1424 | return local_part, value |
| 1425 | |
| 1426 | def get_obs_local_part(value): |
| 1427 | """ obs-local-part = word *("." word) |
| 1428 | """ |
| 1429 | obs_local_part = ObsLocalPart() |
| 1430 | last_non_ws_was_dot = False |
| 1431 | while value and (value[0]=='\\' or value[0] not in PHRASE_ENDS): |
| 1432 | if value[0] == '.': |
| 1433 | if last_non_ws_was_dot: |
| 1434 | obs_local_part.defects.append(errors.InvalidHeaderDefect( |
| 1435 | "invalid repeated '.'")) |
| 1436 | obs_local_part.append(DOT) |
| 1437 | last_non_ws_was_dot = True |
| 1438 | value = value[1:] |
| 1439 | continue |
| 1440 | elif value[0]=='\\': |
| 1441 | obs_local_part.append(ValueTerminal(value[0], |
| 1442 | 'misplaced-special')) |
| 1443 | value = value[1:] |
| 1444 | obs_local_part.defects.append(errors.InvalidHeaderDefect( |
| 1445 | "'\\' character outside of quoted-string/ccontent")) |
| 1446 | last_non_ws_was_dot = False |
| 1447 | continue |
| 1448 | if obs_local_part and obs_local_part[-1].token_type != 'dot': |
| 1449 | obs_local_part.defects.append(errors.InvalidHeaderDefect( |
| 1450 | "missing '.' between words")) |
| 1451 | try: |
| 1452 | token, value = get_word(value) |
| 1453 | last_non_ws_was_dot = False |
| 1454 | except errors.HeaderParseError: |
| 1455 | if value[0] not in CFWS_LEADER: |
| 1456 | raise |
| 1457 | token, value = get_cfws(value) |
| 1458 | obs_local_part.append(token) |
| 1459 | if (obs_local_part[0].token_type == 'dot' or |
| 1460 | obs_local_part[0].token_type=='cfws' and |
| 1461 | obs_local_part[1].token_type=='dot'): |
| 1462 | obs_local_part.defects.append(errors.InvalidHeaderDefect( |
| 1463 | "Invalid leading '.' in local part")) |
| 1464 | if (obs_local_part[-1].token_type == 'dot' or |
| 1465 | obs_local_part[-1].token_type=='cfws' and |
| 1466 | obs_local_part[-2].token_type=='dot'): |
| 1467 | obs_local_part.defects.append(errors.InvalidHeaderDefect( |
| 1468 | "Invalid trailing '.' in local part")) |
| 1469 | if obs_local_part.defects: |
| 1470 | obs_local_part.token_type = 'invalid-obs-local-part' |
| 1471 | return obs_local_part, value |
| 1472 | |
| 1473 | def get_dtext(value): |
R David Murray | 44b548d | 2016-09-08 13:59:53 -0400 | [diff] [blame] | 1474 | r""" dtext = <printable ascii except \ [ ]> / obs-dtext |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1475 | obs-dtext = obs-NO-WS-CTL / quoted-pair |
| 1476 | |
Terry Jan Reedy | 0f84764 | 2013-03-11 18:34:00 -0400 | [diff] [blame] | 1477 | We allow anything except the excluded characters, but if we find any |
Serhiy Storchaka | 6a7b3a7 | 2016-04-17 08:32:47 +0300 | [diff] [blame] | 1478 | ASCII other than the RFC defined printable ASCII, a NonPrintableDefect is |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1479 | added to the token's defects list. Quoted pairs are converted to their |
| 1480 | unquoted values, so what is returned is a ptext token, in this case a |
| 1481 | ValueTerminal. If there were quoted-printables, an ObsoleteHeaderDefect is |
| 1482 | added to the returned token's defect list. |
| 1483 | |
| 1484 | """ |
| 1485 | ptext, value, had_qp = _get_ptext_to_endchars(value, '[]') |
| 1486 | ptext = ValueTerminal(ptext, 'ptext') |
| 1487 | if had_qp: |
| 1488 | ptext.defects.append(errors.ObsoleteHeaderDefect( |
| 1489 | "quoted printable found in domain-literal")) |
| 1490 | _validate_xtext(ptext) |
| 1491 | return ptext, value |
| 1492 | |
| 1493 | def _check_for_early_dl_end(value, domain_literal): |
| 1494 | if value: |
| 1495 | return False |
| 1496 | domain_literal.append(errors.InvalidHeaderDefect( |
| 1497 | "end of input inside domain-literal")) |
| 1498 | domain_literal.append(ValueTerminal(']', 'domain-literal-end')) |
| 1499 | return True |
| 1500 | |
| 1501 | def get_domain_literal(value): |
| 1502 | """ domain-literal = [CFWS] "[" *([FWS] dtext) [FWS] "]" [CFWS] |
| 1503 | |
| 1504 | """ |
| 1505 | domain_literal = DomainLiteral() |
| 1506 | if value[0] in CFWS_LEADER: |
| 1507 | token, value = get_cfws(value) |
| 1508 | domain_literal.append(token) |
| 1509 | if not value: |
| 1510 | raise errors.HeaderParseError("expected domain-literal") |
| 1511 | if value[0] != '[': |
| 1512 | raise errors.HeaderParseError("expected '[' at start of domain-literal " |
| 1513 | "but found '{}'".format(value)) |
| 1514 | value = value[1:] |
| 1515 | if _check_for_early_dl_end(value, domain_literal): |
| 1516 | return domain_literal, value |
| 1517 | domain_literal.append(ValueTerminal('[', 'domain-literal-start')) |
| 1518 | if value[0] in WSP: |
| 1519 | token, value = get_fws(value) |
| 1520 | domain_literal.append(token) |
| 1521 | token, value = get_dtext(value) |
| 1522 | domain_literal.append(token) |
| 1523 | if _check_for_early_dl_end(value, domain_literal): |
| 1524 | return domain_literal, value |
| 1525 | if value[0] in WSP: |
| 1526 | token, value = get_fws(value) |
| 1527 | domain_literal.append(token) |
| 1528 | if _check_for_early_dl_end(value, domain_literal): |
| 1529 | return domain_literal, value |
| 1530 | if value[0] != ']': |
| 1531 | raise errors.HeaderParseError("expected ']' at end of domain-literal " |
| 1532 | "but found '{}'".format(value)) |
| 1533 | domain_literal.append(ValueTerminal(']', 'domain-literal-end')) |
| 1534 | value = value[1:] |
| 1535 | if value and value[0] in CFWS_LEADER: |
| 1536 | token, value = get_cfws(value) |
| 1537 | domain_literal.append(token) |
| 1538 | return domain_literal, value |
| 1539 | |
| 1540 | def get_domain(value): |
| 1541 | """ domain = dot-atom / domain-literal / obs-domain |
| 1542 | obs-domain = atom *("." atom)) |
| 1543 | |
| 1544 | """ |
| 1545 | domain = Domain() |
| 1546 | leader = None |
| 1547 | if value[0] in CFWS_LEADER: |
| 1548 | leader, value = get_cfws(value) |
| 1549 | if not value: |
| 1550 | raise errors.HeaderParseError( |
| 1551 | "expected domain but found '{}'".format(value)) |
| 1552 | if value[0] == '[': |
| 1553 | token, value = get_domain_literal(value) |
| 1554 | if leader is not None: |
| 1555 | token[:0] = [leader] |
| 1556 | domain.append(token) |
| 1557 | return domain, value |
| 1558 | try: |
| 1559 | token, value = get_dot_atom(value) |
| 1560 | except errors.HeaderParseError: |
| 1561 | token, value = get_atom(value) |
| 1562 | if leader is not None: |
| 1563 | token[:0] = [leader] |
| 1564 | domain.append(token) |
| 1565 | if value and value[0] == '.': |
| 1566 | domain.defects.append(errors.ObsoleteHeaderDefect( |
| 1567 | "domain is not a dot-atom (contains CFWS)")) |
| 1568 | if domain[0].token_type == 'dot-atom': |
| 1569 | domain[:] = domain[0] |
| 1570 | while value and value[0] == '.': |
| 1571 | domain.append(DOT) |
| 1572 | token, value = get_atom(value[1:]) |
| 1573 | domain.append(token) |
| 1574 | return domain, value |
| 1575 | |
| 1576 | def get_addr_spec(value): |
| 1577 | """ addr-spec = local-part "@" domain |
| 1578 | |
| 1579 | """ |
| 1580 | addr_spec = AddrSpec() |
| 1581 | token, value = get_local_part(value) |
| 1582 | addr_spec.append(token) |
| 1583 | if not value or value[0] != '@': |
| 1584 | addr_spec.defects.append(errors.InvalidHeaderDefect( |
| 1585 | "add-spec local part with no domain")) |
| 1586 | return addr_spec, value |
| 1587 | addr_spec.append(ValueTerminal('@', 'address-at-symbol')) |
| 1588 | token, value = get_domain(value[1:]) |
| 1589 | addr_spec.append(token) |
| 1590 | return addr_spec, value |
| 1591 | |
| 1592 | def get_obs_route(value): |
| 1593 | """ obs-route = obs-domain-list ":" |
| 1594 | obs-domain-list = *(CFWS / ",") "@" domain *("," [CFWS] ["@" domain]) |
| 1595 | |
| 1596 | Returns an obs-route token with the appropriate sub-tokens (that is, |
| 1597 | there is no obs-domain-list in the parse tree). |
| 1598 | """ |
| 1599 | obs_route = ObsRoute() |
| 1600 | while value and (value[0]==',' or value[0] in CFWS_LEADER): |
| 1601 | if value[0] in CFWS_LEADER: |
| 1602 | token, value = get_cfws(value) |
| 1603 | obs_route.append(token) |
| 1604 | elif value[0] == ',': |
| 1605 | obs_route.append(ListSeparator) |
| 1606 | value = value[1:] |
| 1607 | if not value or value[0] != '@': |
| 1608 | raise errors.HeaderParseError( |
| 1609 | "expected obs-route domain but found '{}'".format(value)) |
| 1610 | obs_route.append(RouteComponentMarker) |
| 1611 | token, value = get_domain(value[1:]) |
| 1612 | obs_route.append(token) |
| 1613 | while value and value[0]==',': |
| 1614 | obs_route.append(ListSeparator) |
| 1615 | value = value[1:] |
| 1616 | if not value: |
| 1617 | break |
| 1618 | if value[0] in CFWS_LEADER: |
| 1619 | token, value = get_cfws(value) |
| 1620 | obs_route.append(token) |
| 1621 | if value[0] == '@': |
| 1622 | obs_route.append(RouteComponentMarker) |
| 1623 | token, value = get_domain(value[1:]) |
| 1624 | obs_route.append(token) |
| 1625 | if not value: |
| 1626 | raise errors.HeaderParseError("end of header while parsing obs-route") |
| 1627 | if value[0] != ':': |
| 1628 | raise errors.HeaderParseError( "expected ':' marking end of " |
| 1629 | "obs-route but found '{}'".format(value)) |
| 1630 | obs_route.append(ValueTerminal(':', 'end-of-obs-route-marker')) |
| 1631 | return obs_route, value[1:] |
| 1632 | |
| 1633 | def get_angle_addr(value): |
| 1634 | """ angle-addr = [CFWS] "<" addr-spec ">" [CFWS] / obs-angle-addr |
| 1635 | obs-angle-addr = [CFWS] "<" obs-route addr-spec ">" [CFWS] |
| 1636 | |
| 1637 | """ |
| 1638 | angle_addr = AngleAddr() |
| 1639 | if value[0] in CFWS_LEADER: |
| 1640 | token, value = get_cfws(value) |
| 1641 | angle_addr.append(token) |
| 1642 | if not value or value[0] != '<': |
| 1643 | raise errors.HeaderParseError( |
| 1644 | "expected angle-addr but found '{}'".format(value)) |
| 1645 | angle_addr.append(ValueTerminal('<', 'angle-addr-start')) |
| 1646 | value = value[1:] |
R David Murray | 032eed3 | 2012-05-26 14:31:12 -0400 | [diff] [blame] | 1647 | # Although it is not legal per RFC5322, SMTP uses '<>' in certain |
| 1648 | # circumstances. |
| 1649 | if value[0] == '>': |
| 1650 | angle_addr.append(ValueTerminal('>', 'angle-addr-end')) |
| 1651 | angle_addr.defects.append(errors.InvalidHeaderDefect( |
| 1652 | "null addr-spec in angle-addr")) |
| 1653 | value = value[1:] |
| 1654 | return angle_addr, value |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1655 | try: |
| 1656 | token, value = get_addr_spec(value) |
| 1657 | except errors.HeaderParseError: |
| 1658 | try: |
| 1659 | token, value = get_obs_route(value) |
| 1660 | angle_addr.defects.append(errors.ObsoleteHeaderDefect( |
| 1661 | "obsolete route specification in angle-addr")) |
| 1662 | except errors.HeaderParseError: |
| 1663 | raise errors.HeaderParseError( |
R David Murray | 032eed3 | 2012-05-26 14:31:12 -0400 | [diff] [blame] | 1664 | "expected addr-spec or obs-route but found '{}'".format(value)) |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1665 | angle_addr.append(token) |
| 1666 | token, value = get_addr_spec(value) |
| 1667 | angle_addr.append(token) |
| 1668 | if value and value[0] == '>': |
| 1669 | value = value[1:] |
| 1670 | else: |
| 1671 | angle_addr.defects.append(errors.InvalidHeaderDefect( |
| 1672 | "missing trailing '>' on angle-addr")) |
| 1673 | angle_addr.append(ValueTerminal('>', 'angle-addr-end')) |
| 1674 | if value and value[0] in CFWS_LEADER: |
| 1675 | token, value = get_cfws(value) |
| 1676 | angle_addr.append(token) |
| 1677 | return angle_addr, value |
| 1678 | |
| 1679 | def get_display_name(value): |
| 1680 | """ display-name = phrase |
| 1681 | |
| 1682 | Because this is simply a name-rule, we don't return a display-name |
| 1683 | token containing a phrase, but rather a display-name token with |
| 1684 | the content of the phrase. |
| 1685 | |
| 1686 | """ |
| 1687 | display_name = DisplayName() |
| 1688 | token, value = get_phrase(value) |
| 1689 | display_name.extend(token[:]) |
| 1690 | display_name.defects = token.defects[:] |
| 1691 | return display_name, value |
| 1692 | |
| 1693 | |
| 1694 | def get_name_addr(value): |
| 1695 | """ name-addr = [display-name] angle-addr |
| 1696 | |
| 1697 | """ |
| 1698 | name_addr = NameAddr() |
| 1699 | # Both the optional display name and the angle-addr can start with cfws. |
| 1700 | leader = None |
| 1701 | if value[0] in CFWS_LEADER: |
| 1702 | leader, value = get_cfws(value) |
| 1703 | if not value: |
| 1704 | raise errors.HeaderParseError( |
| 1705 | "expected name-addr but found '{}'".format(leader)) |
| 1706 | if value[0] != '<': |
| 1707 | if value[0] in PHRASE_ENDS: |
| 1708 | raise errors.HeaderParseError( |
| 1709 | "expected name-addr but found '{}'".format(value)) |
| 1710 | token, value = get_display_name(value) |
| 1711 | if not value: |
| 1712 | raise errors.HeaderParseError( |
| 1713 | "expected name-addr but found '{}'".format(token)) |
| 1714 | if leader is not None: |
| 1715 | token[0][:0] = [leader] |
| 1716 | leader = None |
| 1717 | name_addr.append(token) |
| 1718 | token, value = get_angle_addr(value) |
| 1719 | if leader is not None: |
| 1720 | token[:0] = [leader] |
| 1721 | name_addr.append(token) |
| 1722 | return name_addr, value |
| 1723 | |
| 1724 | def get_mailbox(value): |
| 1725 | """ mailbox = name-addr / addr-spec |
| 1726 | |
| 1727 | """ |
| 1728 | # The only way to figure out if we are dealing with a name-addr or an |
| 1729 | # addr-spec is to try parsing each one. |
| 1730 | mailbox = Mailbox() |
| 1731 | try: |
| 1732 | token, value = get_name_addr(value) |
| 1733 | except errors.HeaderParseError: |
| 1734 | try: |
| 1735 | token, value = get_addr_spec(value) |
| 1736 | except errors.HeaderParseError: |
| 1737 | raise errors.HeaderParseError( |
| 1738 | "expected mailbox but found '{}'".format(value)) |
| 1739 | if any(isinstance(x, errors.InvalidHeaderDefect) |
| 1740 | for x in token.all_defects): |
| 1741 | mailbox.token_type = 'invalid-mailbox' |
| 1742 | mailbox.append(token) |
| 1743 | return mailbox, value |
| 1744 | |
| 1745 | def get_invalid_mailbox(value, endchars): |
| 1746 | """ Read everything up to one of the chars in endchars. |
| 1747 | |
| 1748 | This is outside the formal grammar. The InvalidMailbox TokenList that is |
| 1749 | returned acts like a Mailbox, but the data attributes are None. |
| 1750 | |
| 1751 | """ |
| 1752 | invalid_mailbox = InvalidMailbox() |
| 1753 | while value and value[0] not in endchars: |
| 1754 | if value[0] in PHRASE_ENDS: |
| 1755 | invalid_mailbox.append(ValueTerminal(value[0], |
| 1756 | 'misplaced-special')) |
| 1757 | value = value[1:] |
| 1758 | else: |
| 1759 | token, value = get_phrase(value) |
| 1760 | invalid_mailbox.append(token) |
| 1761 | return invalid_mailbox, value |
| 1762 | |
| 1763 | def get_mailbox_list(value): |
| 1764 | """ mailbox-list = (mailbox *("," mailbox)) / obs-mbox-list |
| 1765 | obs-mbox-list = *([CFWS] ",") mailbox *("," [mailbox / CFWS]) |
| 1766 | |
| 1767 | For this routine we go outside the formal grammar in order to improve error |
| 1768 | handling. We recognize the end of the mailbox list only at the end of the |
| 1769 | value or at a ';' (the group terminator). This is so that we can turn |
| 1770 | invalid mailboxes into InvalidMailbox tokens and continue parsing any |
| 1771 | remaining valid mailboxes. We also allow all mailbox entries to be null, |
| 1772 | and this condition is handled appropriately at a higher level. |
| 1773 | |
| 1774 | """ |
| 1775 | mailbox_list = MailboxList() |
| 1776 | while value and value[0] != ';': |
| 1777 | try: |
| 1778 | token, value = get_mailbox(value) |
| 1779 | mailbox_list.append(token) |
| 1780 | except errors.HeaderParseError: |
| 1781 | leader = None |
| 1782 | if value[0] in CFWS_LEADER: |
| 1783 | leader, value = get_cfws(value) |
| 1784 | if not value or value[0] in ',;': |
| 1785 | mailbox_list.append(leader) |
| 1786 | mailbox_list.defects.append(errors.ObsoleteHeaderDefect( |
| 1787 | "empty element in mailbox-list")) |
| 1788 | else: |
| 1789 | token, value = get_invalid_mailbox(value, ',;') |
| 1790 | if leader is not None: |
| 1791 | token[:0] = [leader] |
| 1792 | mailbox_list.append(token) |
| 1793 | mailbox_list.defects.append(errors.InvalidHeaderDefect( |
| 1794 | "invalid mailbox in mailbox-list")) |
| 1795 | elif value[0] == ',': |
| 1796 | mailbox_list.defects.append(errors.ObsoleteHeaderDefect( |
| 1797 | "empty element in mailbox-list")) |
| 1798 | else: |
| 1799 | token, value = get_invalid_mailbox(value, ',;') |
| 1800 | if leader is not None: |
| 1801 | token[:0] = [leader] |
| 1802 | mailbox_list.append(token) |
| 1803 | mailbox_list.defects.append(errors.InvalidHeaderDefect( |
| 1804 | "invalid mailbox in mailbox-list")) |
| 1805 | if value and value[0] not in ',;': |
| 1806 | # Crap after mailbox; treat it as an invalid mailbox. |
| 1807 | # The mailbox info will still be available. |
| 1808 | mailbox = mailbox_list[-1] |
| 1809 | mailbox.token_type = 'invalid-mailbox' |
| 1810 | token, value = get_invalid_mailbox(value, ',;') |
| 1811 | mailbox.extend(token) |
| 1812 | mailbox_list.defects.append(errors.InvalidHeaderDefect( |
| 1813 | "invalid mailbox in mailbox-list")) |
| 1814 | if value and value[0] == ',': |
| 1815 | mailbox_list.append(ListSeparator) |
| 1816 | value = value[1:] |
| 1817 | return mailbox_list, value |
| 1818 | |
| 1819 | |
| 1820 | def get_group_list(value): |
| 1821 | """ group-list = mailbox-list / CFWS / obs-group-list |
| 1822 | obs-group-list = 1*([CFWS] ",") [CFWS] |
| 1823 | |
| 1824 | """ |
| 1825 | group_list = GroupList() |
| 1826 | if not value: |
| 1827 | group_list.defects.append(errors.InvalidHeaderDefect( |
| 1828 | "end of header before group-list")) |
| 1829 | return group_list, value |
| 1830 | leader = None |
| 1831 | if value and value[0] in CFWS_LEADER: |
| 1832 | leader, value = get_cfws(value) |
| 1833 | if not value: |
| 1834 | # This should never happen in email parsing, since CFWS-only is a |
| 1835 | # legal alternative to group-list in a group, which is the only |
| 1836 | # place group-list appears. |
| 1837 | group_list.defects.append(errors.InvalidHeaderDefect( |
| 1838 | "end of header in group-list")) |
| 1839 | group_list.append(leader) |
| 1840 | return group_list, value |
| 1841 | if value[0] == ';': |
| 1842 | group_list.append(leader) |
| 1843 | return group_list, value |
| 1844 | token, value = get_mailbox_list(value) |
| 1845 | if len(token.all_mailboxes)==0: |
| 1846 | if leader is not None: |
| 1847 | group_list.append(leader) |
| 1848 | group_list.extend(token) |
| 1849 | group_list.defects.append(errors.ObsoleteHeaderDefect( |
| 1850 | "group-list with empty entries")) |
| 1851 | return group_list, value |
| 1852 | if leader is not None: |
| 1853 | token[:0] = [leader] |
| 1854 | group_list.append(token) |
| 1855 | return group_list, value |
| 1856 | |
| 1857 | def get_group(value): |
| 1858 | """ group = display-name ":" [group-list] ";" [CFWS] |
| 1859 | |
| 1860 | """ |
| 1861 | group = Group() |
| 1862 | token, value = get_display_name(value) |
| 1863 | if not value or value[0] != ':': |
| 1864 | raise errors.HeaderParseError("expected ':' at end of group " |
| 1865 | "display name but found '{}'".format(value)) |
| 1866 | group.append(token) |
| 1867 | group.append(ValueTerminal(':', 'group-display-name-terminator')) |
| 1868 | value = value[1:] |
| 1869 | if value and value[0] == ';': |
| 1870 | group.append(ValueTerminal(';', 'group-terminator')) |
| 1871 | return group, value[1:] |
| 1872 | token, value = get_group_list(value) |
| 1873 | group.append(token) |
| 1874 | if not value: |
| 1875 | group.defects.append(errors.InvalidHeaderDefect( |
| 1876 | "end of header in group")) |
Dong-hee Na | 8fe9eed | 2018-07-28 21:55:11 +0900 | [diff] [blame] | 1877 | elif value[0] != ';': |
R David Murray | 0b6f6c8 | 2012-05-25 18:42:14 -0400 | [diff] [blame] | 1878 | raise errors.HeaderParseError( |
| 1879 | "expected ';' at end of group but found {}".format(value)) |
| 1880 | group.append(ValueTerminal(';', 'group-terminator')) |
| 1881 | value = value[1:] |
| 1882 | if value and value[0] in CFWS_LEADER: |
| 1883 | token, value = get_cfws(value) |
| 1884 | group.append(token) |
| 1885 | return group, value |
| 1886 | |
| 1887 | def get_address(value): |
| 1888 | """ address = mailbox / group |
| 1889 | |
| 1890 | Note that counter-intuitively, an address can be either a single address or |
| 1891 | a list of addresses (a group). This is why the returned Address object has |
| 1892 | a 'mailboxes' attribute which treats a single address as a list of length |
| 1893 | one. When you need to differentiate between to two cases, extract the single |
| 1894 | element, which is either a mailbox or a group token. |
| 1895 | |
| 1896 | """ |
| 1897 | # The formal grammar isn't very helpful when parsing an address. mailbox |
| 1898 | # and group, especially when allowing for obsolete forms, start off very |
| 1899 | # similarly. It is only when you reach one of @, <, or : that you know |
| 1900 | # what you've got. So, we try each one in turn, starting with the more |
| 1901 | # likely of the two. We could perhaps make this more efficient by looking |
| 1902 | # for a phrase and then branching based on the next character, but that |
| 1903 | # would be a premature optimization. |
| 1904 | address = Address() |
| 1905 | try: |
| 1906 | token, value = get_group(value) |
| 1907 | except errors.HeaderParseError: |
| 1908 | try: |
| 1909 | token, value = get_mailbox(value) |
| 1910 | except errors.HeaderParseError: |
| 1911 | raise errors.HeaderParseError( |
| 1912 | "expected address but found '{}'".format(value)) |
| 1913 | address.append(token) |
| 1914 | return address, value |
| 1915 | |
| 1916 | def get_address_list(value): |
| 1917 | """ address_list = (address *("," address)) / obs-addr-list |
| 1918 | obs-addr-list = *([CFWS] ",") address *("," [address / CFWS]) |
| 1919 | |
| 1920 | We depart from the formal grammar here by continuing to parse until the end |
| 1921 | of the input, assuming the input to be entirely composed of an |
| 1922 | address-list. This is always true in email parsing, and allows us |
| 1923 | to skip invalid addresses to parse additional valid ones. |
| 1924 | |
| 1925 | """ |
| 1926 | address_list = AddressList() |
| 1927 | while value: |
| 1928 | try: |
| 1929 | token, value = get_address(value) |
| 1930 | address_list.append(token) |
| 1931 | except errors.HeaderParseError as err: |
| 1932 | leader = None |
| 1933 | if value[0] in CFWS_LEADER: |
| 1934 | leader, value = get_cfws(value) |
| 1935 | if not value or value[0] == ',': |
| 1936 | address_list.append(leader) |
| 1937 | address_list.defects.append(errors.ObsoleteHeaderDefect( |
| 1938 | "address-list entry with no content")) |
| 1939 | else: |
| 1940 | token, value = get_invalid_mailbox(value, ',') |
| 1941 | if leader is not None: |
| 1942 | token[:0] = [leader] |
| 1943 | address_list.append(Address([token])) |
| 1944 | address_list.defects.append(errors.InvalidHeaderDefect( |
| 1945 | "invalid address in address-list")) |
| 1946 | elif value[0] == ',': |
| 1947 | address_list.defects.append(errors.ObsoleteHeaderDefect( |
| 1948 | "empty element in address-list")) |
| 1949 | else: |
| 1950 | token, value = get_invalid_mailbox(value, ',') |
| 1951 | if leader is not None: |
| 1952 | token[:0] = [leader] |
| 1953 | address_list.append(Address([token])) |
| 1954 | address_list.defects.append(errors.InvalidHeaderDefect( |
| 1955 | "invalid address in address-list")) |
| 1956 | if value and value[0] != ',': |
| 1957 | # Crap after address; treat it as an invalid mailbox. |
| 1958 | # The mailbox info will still be available. |
| 1959 | mailbox = address_list[-1][0] |
| 1960 | mailbox.token_type = 'invalid-mailbox' |
| 1961 | token, value = get_invalid_mailbox(value, ',') |
| 1962 | mailbox.extend(token) |
| 1963 | address_list.defects.append(errors.InvalidHeaderDefect( |
| 1964 | "invalid address in address-list")) |
| 1965 | if value: # Must be a , at this point. |
| 1966 | address_list.append(ValueTerminal(',', 'list-separator')) |
| 1967 | value = value[1:] |
| 1968 | return address_list, value |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 1969 | |
| 1970 | # |
| 1971 | # XXX: As I begin to add additional header parsers, I'm realizing we probably |
| 1972 | # have two level of parser routines: the get_XXX methods that get a token in |
| 1973 | # the grammar, and parse_XXX methods that parse an entire field value. So |
| 1974 | # get_address_list above should really be a parse_ method, as probably should |
| 1975 | # be get_unstructured. |
| 1976 | # |
| 1977 | |
| 1978 | def parse_mime_version(value): |
| 1979 | """ mime-version = [CFWS] 1*digit [CFWS] "." [CFWS] 1*digit [CFWS] |
| 1980 | |
| 1981 | """ |
| 1982 | # The [CFWS] is implicit in the RFC 2045 BNF. |
| 1983 | # XXX: This routine is a bit verbose, should factor out a get_int method. |
| 1984 | mime_version = MIMEVersion() |
| 1985 | if not value: |
| 1986 | mime_version.defects.append(errors.HeaderMissingRequiredValue( |
| 1987 | "Missing MIME version number (eg: 1.0)")) |
| 1988 | return mime_version |
| 1989 | if value[0] in CFWS_LEADER: |
| 1990 | token, value = get_cfws(value) |
| 1991 | mime_version.append(token) |
| 1992 | if not value: |
| 1993 | mime_version.defects.append(errors.HeaderMissingRequiredValue( |
| 1994 | "Expected MIME version number but found only CFWS")) |
| 1995 | digits = '' |
| 1996 | while value and value[0] != '.' and value[0] not in CFWS_LEADER: |
| 1997 | digits += value[0] |
| 1998 | value = value[1:] |
| 1999 | if not digits.isdigit(): |
| 2000 | mime_version.defects.append(errors.InvalidHeaderDefect( |
| 2001 | "Expected MIME major version number but found {!r}".format(digits))) |
| 2002 | mime_version.append(ValueTerminal(digits, 'xtext')) |
| 2003 | else: |
| 2004 | mime_version.major = int(digits) |
| 2005 | mime_version.append(ValueTerminal(digits, 'digits')) |
| 2006 | if value and value[0] in CFWS_LEADER: |
| 2007 | token, value = get_cfws(value) |
| 2008 | mime_version.append(token) |
| 2009 | if not value or value[0] != '.': |
| 2010 | if mime_version.major is not None: |
| 2011 | mime_version.defects.append(errors.InvalidHeaderDefect( |
| 2012 | "Incomplete MIME version; found only major number")) |
| 2013 | if value: |
| 2014 | mime_version.append(ValueTerminal(value, 'xtext')) |
| 2015 | return mime_version |
| 2016 | mime_version.append(ValueTerminal('.', 'version-separator')) |
| 2017 | value = value[1:] |
| 2018 | if value and value[0] in CFWS_LEADER: |
| 2019 | token, value = get_cfws(value) |
| 2020 | mime_version.append(token) |
| 2021 | if not value: |
| 2022 | if mime_version.major is not None: |
| 2023 | mime_version.defects.append(errors.InvalidHeaderDefect( |
| 2024 | "Incomplete MIME version; found only major number")) |
| 2025 | return mime_version |
| 2026 | digits = '' |
| 2027 | while value and value[0] not in CFWS_LEADER: |
| 2028 | digits += value[0] |
| 2029 | value = value[1:] |
| 2030 | if not digits.isdigit(): |
| 2031 | mime_version.defects.append(errors.InvalidHeaderDefect( |
| 2032 | "Expected MIME minor version number but found {!r}".format(digits))) |
| 2033 | mime_version.append(ValueTerminal(digits, 'xtext')) |
| 2034 | else: |
| 2035 | mime_version.minor = int(digits) |
| 2036 | mime_version.append(ValueTerminal(digits, 'digits')) |
| 2037 | if value and value[0] in CFWS_LEADER: |
| 2038 | token, value = get_cfws(value) |
| 2039 | mime_version.append(token) |
| 2040 | if value: |
| 2041 | mime_version.defects.append(errors.InvalidHeaderDefect( |
| 2042 | "Excess non-CFWS text after MIME version")) |
| 2043 | mime_version.append(ValueTerminal(value, 'xtext')) |
| 2044 | return mime_version |
| 2045 | |
| 2046 | def get_invalid_parameter(value): |
| 2047 | """ Read everything up to the next ';'. |
| 2048 | |
| 2049 | This is outside the formal grammar. The InvalidParameter TokenList that is |
| 2050 | returned acts like a Parameter, but the data attributes are None. |
| 2051 | |
| 2052 | """ |
| 2053 | invalid_parameter = InvalidParameter() |
| 2054 | while value and value[0] != ';': |
| 2055 | if value[0] in PHRASE_ENDS: |
| 2056 | invalid_parameter.append(ValueTerminal(value[0], |
| 2057 | 'misplaced-special')) |
| 2058 | value = value[1:] |
| 2059 | else: |
| 2060 | token, value = get_phrase(value) |
| 2061 | invalid_parameter.append(token) |
| 2062 | return invalid_parameter, value |
| 2063 | |
| 2064 | def get_ttext(value): |
| 2065 | """ttext = <matches _ttext_matcher> |
| 2066 | |
| 2067 | We allow any non-TOKEN_ENDS in ttext, but add defects to the token's |
| 2068 | defects list if we find non-ttext characters. We also register defects for |
| 2069 | *any* non-printables even though the RFC doesn't exclude all of them, |
| 2070 | because we follow the spirit of RFC 5322. |
| 2071 | |
| 2072 | """ |
| 2073 | m = _non_token_end_matcher(value) |
| 2074 | if not m: |
| 2075 | raise errors.HeaderParseError( |
| 2076 | "expected ttext but found '{}'".format(value)) |
| 2077 | ttext = m.group() |
| 2078 | value = value[len(ttext):] |
| 2079 | ttext = ValueTerminal(ttext, 'ttext') |
| 2080 | _validate_xtext(ttext) |
| 2081 | return ttext, value |
| 2082 | |
| 2083 | def get_token(value): |
| 2084 | """token = [CFWS] 1*ttext [CFWS] |
| 2085 | |
| 2086 | The RFC equivalent of ttext is any US-ASCII chars except space, ctls, or |
| 2087 | tspecials. We also exclude tabs even though the RFC doesn't. |
| 2088 | |
| 2089 | The RFC implies the CFWS but is not explicit about it in the BNF. |
| 2090 | |
| 2091 | """ |
| 2092 | mtoken = Token() |
| 2093 | if value and value[0] in CFWS_LEADER: |
| 2094 | token, value = get_cfws(value) |
| 2095 | mtoken.append(token) |
| 2096 | if value and value[0] in TOKEN_ENDS: |
| 2097 | raise errors.HeaderParseError( |
| 2098 | "expected token but found '{}'".format(value)) |
| 2099 | token, value = get_ttext(value) |
| 2100 | mtoken.append(token) |
| 2101 | if value and value[0] in CFWS_LEADER: |
| 2102 | token, value = get_cfws(value) |
| 2103 | mtoken.append(token) |
| 2104 | return mtoken, value |
| 2105 | |
| 2106 | def get_attrtext(value): |
| 2107 | """attrtext = 1*(any non-ATTRIBUTE_ENDS character) |
| 2108 | |
| 2109 | We allow any non-ATTRIBUTE_ENDS in attrtext, but add defects to the |
| 2110 | token's defects list if we find non-attrtext characters. We also register |
| 2111 | defects for *any* non-printables even though the RFC doesn't exclude all of |
| 2112 | them, because we follow the spirit of RFC 5322. |
| 2113 | |
| 2114 | """ |
| 2115 | m = _non_attribute_end_matcher(value) |
| 2116 | if not m: |
| 2117 | raise errors.HeaderParseError( |
| 2118 | "expected attrtext but found {!r}".format(value)) |
| 2119 | attrtext = m.group() |
| 2120 | value = value[len(attrtext):] |
| 2121 | attrtext = ValueTerminal(attrtext, 'attrtext') |
| 2122 | _validate_xtext(attrtext) |
| 2123 | return attrtext, value |
| 2124 | |
| 2125 | def get_attribute(value): |
| 2126 | """ [CFWS] 1*attrtext [CFWS] |
| 2127 | |
| 2128 | This version of the BNF makes the CFWS explicit, and as usual we use a |
| 2129 | value terminal for the actual run of characters. The RFC equivalent of |
| 2130 | attrtext is the token characters, with the subtraction of '*', "'", and '%'. |
| 2131 | We include tab in the excluded set just as we do for token. |
| 2132 | |
| 2133 | """ |
| 2134 | attribute = Attribute() |
| 2135 | if value and value[0] in CFWS_LEADER: |
| 2136 | token, value = get_cfws(value) |
| 2137 | attribute.append(token) |
| 2138 | if value and value[0] in ATTRIBUTE_ENDS: |
| 2139 | raise errors.HeaderParseError( |
| 2140 | "expected token but found '{}'".format(value)) |
| 2141 | token, value = get_attrtext(value) |
| 2142 | attribute.append(token) |
| 2143 | if value and value[0] in CFWS_LEADER: |
| 2144 | token, value = get_cfws(value) |
| 2145 | attribute.append(token) |
| 2146 | return attribute, value |
| 2147 | |
| 2148 | def get_extended_attrtext(value): |
| 2149 | """attrtext = 1*(any non-ATTRIBUTE_ENDS character plus '%') |
| 2150 | |
| 2151 | This is a special parsing routine so that we get a value that |
| 2152 | includes % escapes as a single string (which we decode as a single |
| 2153 | string later). |
| 2154 | |
| 2155 | """ |
| 2156 | m = _non_extended_attribute_end_matcher(value) |
| 2157 | if not m: |
| 2158 | raise errors.HeaderParseError( |
| 2159 | "expected extended attrtext but found {!r}".format(value)) |
| 2160 | attrtext = m.group() |
| 2161 | value = value[len(attrtext):] |
| 2162 | attrtext = ValueTerminal(attrtext, 'extended-attrtext') |
| 2163 | _validate_xtext(attrtext) |
| 2164 | return attrtext, value |
| 2165 | |
| 2166 | def get_extended_attribute(value): |
| 2167 | """ [CFWS] 1*extended_attrtext [CFWS] |
| 2168 | |
| 2169 | This is like the non-extended version except we allow % characters, so that |
| 2170 | we can pick up an encoded value as a single string. |
| 2171 | |
| 2172 | """ |
| 2173 | # XXX: should we have an ExtendedAttribute TokenList? |
| 2174 | attribute = Attribute() |
| 2175 | if value and value[0] in CFWS_LEADER: |
| 2176 | token, value = get_cfws(value) |
| 2177 | attribute.append(token) |
| 2178 | if value and value[0] in EXTENDED_ATTRIBUTE_ENDS: |
| 2179 | raise errors.HeaderParseError( |
| 2180 | "expected token but found '{}'".format(value)) |
| 2181 | token, value = get_extended_attrtext(value) |
| 2182 | attribute.append(token) |
| 2183 | if value and value[0] in CFWS_LEADER: |
| 2184 | token, value = get_cfws(value) |
| 2185 | attribute.append(token) |
| 2186 | return attribute, value |
| 2187 | |
| 2188 | def get_section(value): |
| 2189 | """ '*' digits |
| 2190 | |
| 2191 | The formal BNF is more complicated because leading 0s are not allowed. We |
| 2192 | check for that and add a defect. We also assume no CFWS is allowed between |
| 2193 | the '*' and the digits, though the RFC is not crystal clear on that. |
| 2194 | The caller should already have dealt with leading CFWS. |
| 2195 | |
| 2196 | """ |
| 2197 | section = Section() |
| 2198 | if not value or value[0] != '*': |
| 2199 | raise errors.HeaderParseError("Expected section but found {}".format( |
| 2200 | value)) |
| 2201 | section.append(ValueTerminal('*', 'section-marker')) |
| 2202 | value = value[1:] |
| 2203 | if not value or not value[0].isdigit(): |
| 2204 | raise errors.HeaderParseError("Expected section number but " |
| 2205 | "found {}".format(value)) |
| 2206 | digits = '' |
| 2207 | while value and value[0].isdigit(): |
| 2208 | digits += value[0] |
| 2209 | value = value[1:] |
| 2210 | if digits[0] == '0' and digits != '0': |
Serhiy Storchaka | 34fd4c2 | 2018-11-05 16:20:25 +0200 | [diff] [blame] | 2211 | section.defects.append(errors.InvalidHeaderError( |
| 2212 | "section number has an invalid leading 0")) |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 2213 | section.number = int(digits) |
| 2214 | section.append(ValueTerminal(digits, 'digits')) |
| 2215 | return section, value |
| 2216 | |
| 2217 | |
| 2218 | def get_value(value): |
| 2219 | """ quoted-string / attribute |
| 2220 | |
| 2221 | """ |
| 2222 | v = Value() |
| 2223 | if not value: |
| 2224 | raise errors.HeaderParseError("Expected value but found end of string") |
| 2225 | leader = None |
| 2226 | if value[0] in CFWS_LEADER: |
| 2227 | leader, value = get_cfws(value) |
| 2228 | if not value: |
| 2229 | raise errors.HeaderParseError("Expected value but found " |
| 2230 | "only {}".format(leader)) |
| 2231 | if value[0] == '"': |
| 2232 | token, value = get_quoted_string(value) |
| 2233 | else: |
| 2234 | token, value = get_extended_attribute(value) |
| 2235 | if leader is not None: |
| 2236 | token[:0] = [leader] |
| 2237 | v.append(token) |
| 2238 | return v, value |
| 2239 | |
| 2240 | def get_parameter(value): |
| 2241 | """ attribute [section] ["*"] [CFWS] "=" value |
| 2242 | |
| 2243 | The CFWS is implied by the RFC but not made explicit in the BNF. This |
| 2244 | simplified form of the BNF from the RFC is made to conform with the RFC BNF |
| 2245 | through some extra checks. We do it this way because it makes both error |
| 2246 | recovery and working with the resulting parse tree easier. |
| 2247 | """ |
| 2248 | # It is possible CFWS would also be implicitly allowed between the section |
| 2249 | # and the 'extended-attribute' marker (the '*') , but we've never seen that |
| 2250 | # in the wild and we will therefore ignore the possibility. |
| 2251 | param = Parameter() |
| 2252 | token, value = get_attribute(value) |
| 2253 | param.append(token) |
| 2254 | if not value or value[0] == ';': |
| 2255 | param.defects.append(errors.InvalidHeaderDefect("Parameter contains " |
| 2256 | "name ({}) but no value".format(token))) |
| 2257 | return param, value |
| 2258 | if value[0] == '*': |
| 2259 | try: |
| 2260 | token, value = get_section(value) |
| 2261 | param.sectioned = True |
| 2262 | param.append(token) |
| 2263 | except errors.HeaderParseError: |
| 2264 | pass |
| 2265 | if not value: |
| 2266 | raise errors.HeaderParseError("Incomplete parameter") |
| 2267 | if value[0] == '*': |
| 2268 | param.append(ValueTerminal('*', 'extended-parameter-marker')) |
| 2269 | value = value[1:] |
| 2270 | param.extended = True |
| 2271 | if value[0] != '=': |
| 2272 | raise errors.HeaderParseError("Parameter not followed by '='") |
| 2273 | param.append(ValueTerminal('=', 'parameter-separator')) |
| 2274 | value = value[1:] |
| 2275 | leader = None |
| 2276 | if value and value[0] in CFWS_LEADER: |
| 2277 | token, value = get_cfws(value) |
| 2278 | param.append(token) |
| 2279 | remainder = None |
| 2280 | appendto = param |
| 2281 | if param.extended and value and value[0] == '"': |
| 2282 | # Now for some serious hackery to handle the common invalid case of |
| 2283 | # double quotes around an extended value. We also accept (with defect) |
| 2284 | # a value marked as encoded that isn't really. |
| 2285 | qstring, remainder = get_quoted_string(value) |
| 2286 | inner_value = qstring.stripped_value |
| 2287 | semi_valid = False |
| 2288 | if param.section_number == 0: |
| 2289 | if inner_value and inner_value[0] == "'": |
| 2290 | semi_valid = True |
| 2291 | else: |
| 2292 | token, rest = get_attrtext(inner_value) |
| 2293 | if rest and rest[0] == "'": |
| 2294 | semi_valid = True |
| 2295 | else: |
| 2296 | try: |
| 2297 | token, rest = get_extended_attrtext(inner_value) |
| 2298 | except: |
| 2299 | pass |
| 2300 | else: |
| 2301 | if not rest: |
| 2302 | semi_valid = True |
| 2303 | if semi_valid: |
| 2304 | param.defects.append(errors.InvalidHeaderDefect( |
| 2305 | "Quoted string value for extended parameter is invalid")) |
| 2306 | param.append(qstring) |
| 2307 | for t in qstring: |
| 2308 | if t.token_type == 'bare-quoted-string': |
| 2309 | t[:] = [] |
| 2310 | appendto = t |
| 2311 | break |
| 2312 | value = inner_value |
| 2313 | else: |
| 2314 | remainder = None |
| 2315 | param.defects.append(errors.InvalidHeaderDefect( |
| 2316 | "Parameter marked as extended but appears to have a " |
| 2317 | "quoted string value that is non-encoded")) |
| 2318 | if value and value[0] == "'": |
| 2319 | token = None |
| 2320 | else: |
| 2321 | token, value = get_value(value) |
| 2322 | if not param.extended or param.section_number > 0: |
| 2323 | if not value or value[0] != "'": |
| 2324 | appendto.append(token) |
| 2325 | if remainder is not None: |
| 2326 | assert not value, value |
| 2327 | value = remainder |
| 2328 | return param, value |
| 2329 | param.defects.append(errors.InvalidHeaderDefect( |
| 2330 | "Apparent initial-extended-value but attribute " |
| 2331 | "was not marked as extended or was not initial section")) |
| 2332 | if not value: |
| 2333 | # Assume the charset/lang is missing and the token is the value. |
| 2334 | param.defects.append(errors.InvalidHeaderDefect( |
| 2335 | "Missing required charset/lang delimiters")) |
| 2336 | appendto.append(token) |
| 2337 | if remainder is None: |
| 2338 | return param, value |
| 2339 | else: |
| 2340 | if token is not None: |
| 2341 | for t in token: |
| 2342 | if t.token_type == 'extended-attrtext': |
| 2343 | break |
| 2344 | t.token_type == 'attrtext' |
| 2345 | appendto.append(t) |
| 2346 | param.charset = t.value |
| 2347 | if value[0] != "'": |
| 2348 | raise errors.HeaderParseError("Expected RFC2231 char/lang encoding " |
| 2349 | "delimiter, but found {!r}".format(value)) |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 2350 | appendto.append(ValueTerminal("'", 'RFC2231-delimiter')) |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 2351 | value = value[1:] |
| 2352 | if value and value[0] != "'": |
| 2353 | token, value = get_attrtext(value) |
| 2354 | appendto.append(token) |
| 2355 | param.lang = token.value |
| 2356 | if not value or value[0] != "'": |
| 2357 | raise errors.HeaderParseError("Expected RFC2231 char/lang encoding " |
| 2358 | "delimiter, but found {}".format(value)) |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 2359 | appendto.append(ValueTerminal("'", 'RFC2231-delimiter')) |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 2360 | value = value[1:] |
| 2361 | if remainder is not None: |
| 2362 | # Treat the rest of value as bare quoted string content. |
| 2363 | v = Value() |
| 2364 | while value: |
| 2365 | if value[0] in WSP: |
| 2366 | token, value = get_fws(value) |
| 2367 | else: |
| 2368 | token, value = get_qcontent(value) |
| 2369 | v.append(token) |
| 2370 | token = v |
| 2371 | else: |
| 2372 | token, value = get_value(value) |
| 2373 | appendto.append(token) |
| 2374 | if remainder is not None: |
| 2375 | assert not value, value |
| 2376 | value = remainder |
| 2377 | return param, value |
| 2378 | |
| 2379 | def parse_mime_parameters(value): |
| 2380 | """ parameter *( ";" parameter ) |
| 2381 | |
| 2382 | That BNF is meant to indicate this routine should only be called after |
| 2383 | finding and handling the leading ';'. There is no corresponding rule in |
| 2384 | the formal RFC grammar, but it is more convenient for us for the set of |
| 2385 | parameters to be treated as its own TokenList. |
| 2386 | |
| 2387 | This is 'parse' routine because it consumes the reminaing value, but it |
| 2388 | would never be called to parse a full header. Instead it is called to |
| 2389 | parse everything after the non-parameter value of a specific MIME header. |
| 2390 | |
| 2391 | """ |
| 2392 | mime_parameters = MimeParameters() |
| 2393 | while value: |
| 2394 | try: |
| 2395 | token, value = get_parameter(value) |
| 2396 | mime_parameters.append(token) |
| 2397 | except errors.HeaderParseError as err: |
| 2398 | leader = None |
| 2399 | if value[0] in CFWS_LEADER: |
| 2400 | leader, value = get_cfws(value) |
| 2401 | if not value: |
| 2402 | mime_parameters.append(leader) |
| 2403 | return mime_parameters |
| 2404 | if value[0] == ';': |
| 2405 | if leader is not None: |
| 2406 | mime_parameters.append(leader) |
| 2407 | mime_parameters.defects.append(errors.InvalidHeaderDefect( |
| 2408 | "parameter entry with no content")) |
| 2409 | else: |
| 2410 | token, value = get_invalid_parameter(value) |
| 2411 | if leader: |
| 2412 | token[:0] = [leader] |
| 2413 | mime_parameters.append(token) |
| 2414 | mime_parameters.defects.append(errors.InvalidHeaderDefect( |
| 2415 | "invalid parameter {!r}".format(token))) |
| 2416 | if value and value[0] != ';': |
| 2417 | # Junk after the otherwise valid parameter. Mark it as |
| 2418 | # invalid, but it will have a value. |
| 2419 | param = mime_parameters[-1] |
| 2420 | param.token_type = 'invalid-parameter' |
| 2421 | token, value = get_invalid_parameter(value) |
| 2422 | param.extend(token) |
| 2423 | mime_parameters.defects.append(errors.InvalidHeaderDefect( |
| 2424 | "parameter with invalid trailing text {!r}".format(token))) |
| 2425 | if value: |
| 2426 | # Must be a ';' at this point. |
| 2427 | mime_parameters.append(ValueTerminal(';', 'parameter-separator')) |
| 2428 | value = value[1:] |
| 2429 | return mime_parameters |
| 2430 | |
| 2431 | def _find_mime_parameters(tokenlist, value): |
| 2432 | """Do our best to find the parameters in an invalid MIME header |
| 2433 | |
| 2434 | """ |
| 2435 | while value and value[0] != ';': |
| 2436 | if value[0] in PHRASE_ENDS: |
| 2437 | tokenlist.append(ValueTerminal(value[0], 'misplaced-special')) |
| 2438 | value = value[1:] |
| 2439 | else: |
| 2440 | token, value = get_phrase(value) |
| 2441 | tokenlist.append(token) |
| 2442 | if not value: |
| 2443 | return |
| 2444 | tokenlist.append(ValueTerminal(';', 'parameter-separator')) |
| 2445 | tokenlist.append(parse_mime_parameters(value[1:])) |
| 2446 | |
| 2447 | def parse_content_type_header(value): |
| 2448 | """ maintype "/" subtype *( ";" parameter ) |
| 2449 | |
| 2450 | The maintype and substype are tokens. Theoretically they could |
| 2451 | be checked against the official IANA list + x-token, but we |
| 2452 | don't do that. |
| 2453 | """ |
| 2454 | ctype = ContentType() |
| 2455 | recover = False |
| 2456 | if not value: |
| 2457 | ctype.defects.append(errors.HeaderMissingRequiredValue( |
| 2458 | "Missing content type specification")) |
| 2459 | return ctype |
| 2460 | try: |
| 2461 | token, value = get_token(value) |
| 2462 | except errors.HeaderParseError: |
| 2463 | ctype.defects.append(errors.InvalidHeaderDefect( |
| 2464 | "Expected content maintype but found {!r}".format(value))) |
| 2465 | _find_mime_parameters(ctype, value) |
| 2466 | return ctype |
| 2467 | ctype.append(token) |
Martin Panter | 46f5072 | 2016-05-26 05:35:26 +0000 | [diff] [blame] | 2468 | # XXX: If we really want to follow the formal grammar we should make |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 2469 | # mantype and subtype specialized TokenLists here. Probably not worth it. |
| 2470 | if not value or value[0] != '/': |
| 2471 | ctype.defects.append(errors.InvalidHeaderDefect( |
| 2472 | "Invalid content type")) |
| 2473 | if value: |
| 2474 | _find_mime_parameters(ctype, value) |
| 2475 | return ctype |
| 2476 | ctype.maintype = token.value.strip().lower() |
| 2477 | ctype.append(ValueTerminal('/', 'content-type-separator')) |
| 2478 | value = value[1:] |
| 2479 | try: |
| 2480 | token, value = get_token(value) |
| 2481 | except errors.HeaderParseError: |
| 2482 | ctype.defects.append(errors.InvalidHeaderDefect( |
| 2483 | "Expected content subtype but found {!r}".format(value))) |
| 2484 | _find_mime_parameters(ctype, value) |
| 2485 | return ctype |
| 2486 | ctype.append(token) |
| 2487 | ctype.subtype = token.value.strip().lower() |
| 2488 | if not value: |
| 2489 | return ctype |
| 2490 | if value[0] != ';': |
| 2491 | ctype.defects.append(errors.InvalidHeaderDefect( |
| 2492 | "Only parameters are valid after content type, but " |
| 2493 | "found {!r}".format(value))) |
| 2494 | # The RFC requires that a syntactically invalid content-type be treated |
| 2495 | # as text/plain. Perhaps we should postel this, but we should probably |
| 2496 | # only do that if we were checking the subtype value against IANA. |
| 2497 | del ctype.maintype, ctype.subtype |
| 2498 | _find_mime_parameters(ctype, value) |
| 2499 | return ctype |
| 2500 | ctype.append(ValueTerminal(';', 'parameter-separator')) |
| 2501 | ctype.append(parse_mime_parameters(value[1:])) |
| 2502 | return ctype |
| 2503 | |
| 2504 | def parse_content_disposition_header(value): |
| 2505 | """ disposition-type *( ";" parameter ) |
| 2506 | |
| 2507 | """ |
| 2508 | disp_header = ContentDisposition() |
| 2509 | if not value: |
| 2510 | disp_header.defects.append(errors.HeaderMissingRequiredValue( |
| 2511 | "Missing content disposition")) |
| 2512 | return disp_header |
| 2513 | try: |
| 2514 | token, value = get_token(value) |
| 2515 | except errors.HeaderParseError: |
Ezio Melotti | d577480 | 2014-08-04 17:16:49 +0300 | [diff] [blame] | 2516 | disp_header.defects.append(errors.InvalidHeaderDefect( |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 2517 | "Expected content disposition but found {!r}".format(value))) |
| 2518 | _find_mime_parameters(disp_header, value) |
| 2519 | return disp_header |
| 2520 | disp_header.append(token) |
| 2521 | disp_header.content_disposition = token.value.strip().lower() |
| 2522 | if not value: |
| 2523 | return disp_header |
| 2524 | if value[0] != ';': |
| 2525 | disp_header.defects.append(errors.InvalidHeaderDefect( |
| 2526 | "Only parameters are valid after content disposition, but " |
| 2527 | "found {!r}".format(value))) |
| 2528 | _find_mime_parameters(disp_header, value) |
| 2529 | return disp_header |
| 2530 | disp_header.append(ValueTerminal(';', 'parameter-separator')) |
| 2531 | disp_header.append(parse_mime_parameters(value[1:])) |
| 2532 | return disp_header |
| 2533 | |
| 2534 | def parse_content_transfer_encoding_header(value): |
| 2535 | """ mechanism |
| 2536 | |
| 2537 | """ |
| 2538 | # We should probably validate the values, since the list is fixed. |
| 2539 | cte_header = ContentTransferEncoding() |
| 2540 | if not value: |
| 2541 | cte_header.defects.append(errors.HeaderMissingRequiredValue( |
| 2542 | "Missing content transfer encoding")) |
| 2543 | return cte_header |
| 2544 | try: |
| 2545 | token, value = get_token(value) |
| 2546 | except errors.HeaderParseError: |
Ezio Melotti | d577480 | 2014-08-04 17:16:49 +0300 | [diff] [blame] | 2547 | cte_header.defects.append(errors.InvalidHeaderDefect( |
| 2548 | "Expected content transfer encoding but found {!r}".format(value))) |
R David Murray | 97f43c0 | 2012-06-24 05:03:27 -0400 | [diff] [blame] | 2549 | else: |
| 2550 | cte_header.append(token) |
| 2551 | cte_header.cte = token.value.strip().lower() |
| 2552 | if not value: |
| 2553 | return cte_header |
| 2554 | while value: |
| 2555 | cte_header.defects.append(errors.InvalidHeaderDefect( |
| 2556 | "Extra text after content transfer encoding")) |
| 2557 | if value[0] in PHRASE_ENDS: |
| 2558 | cte_header.append(ValueTerminal(value[0], 'misplaced-special')) |
| 2559 | value = value[1:] |
| 2560 | else: |
| 2561 | token, value = get_phrase(value) |
| 2562 | cte_header.append(token) |
| 2563 | return cte_header |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 2564 | |
| 2565 | |
| 2566 | # |
| 2567 | # Header folding |
| 2568 | # |
| 2569 | # Header folding is complex, with lots of rules and corner cases. The |
| 2570 | # following code does its best to obey the rules and handle the corner |
| 2571 | # cases, but you can be sure there are few bugs:) |
| 2572 | # |
| 2573 | # This folder generally canonicalizes as it goes, preferring the stringified |
| 2574 | # version of each token. The tokens contain information that supports the |
| 2575 | # folder, including which tokens can be encoded in which ways. |
| 2576 | # |
| 2577 | # Folded text is accumulated in a simple list of strings ('lines'), each |
| 2578 | # one of which should be less than policy.max_line_length ('maxlen'). |
| 2579 | # |
| 2580 | |
| 2581 | def _steal_trailing_WSP_if_exists(lines): |
| 2582 | wsp = '' |
| 2583 | if lines and lines[-1] and lines[-1][-1] in WSP: |
| 2584 | wsp = lines[-1][-1] |
| 2585 | lines[-1] = lines[-1][:-1] |
| 2586 | return wsp |
| 2587 | |
| 2588 | def _refold_parse_tree(parse_tree, *, policy): |
| 2589 | """Return string of contents of parse_tree folded according to RFC rules. |
| 2590 | |
| 2591 | """ |
| 2592 | # max_line_length 0/None means no limit, ie: infinitely long. |
| 2593 | maxlen = policy.max_line_length or float("+inf") |
| 2594 | encoding = 'utf-8' if policy.utf8 else 'us-ascii' |
| 2595 | lines = [''] |
| 2596 | last_ew = None |
| 2597 | wrap_as_ew_blocked = 0 |
| 2598 | want_encoding = False |
| 2599 | end_ew_not_allowed = Terminal('', 'wrap_as_ew_blocked') |
| 2600 | parts = list(parse_tree) |
| 2601 | while parts: |
| 2602 | part = parts.pop(0) |
| 2603 | if part is end_ew_not_allowed: |
| 2604 | wrap_as_ew_blocked -= 1 |
| 2605 | continue |
| 2606 | tstr = str(part) |
| 2607 | try: |
| 2608 | tstr.encode(encoding) |
| 2609 | charset = encoding |
| 2610 | except UnicodeEncodeError: |
| 2611 | if any(isinstance(x, errors.UndecodableBytesDefect) |
| 2612 | for x in part.all_defects): |
| 2613 | charset = 'unknown-8bit' |
| 2614 | else: |
| 2615 | # If policy.utf8 is false this should really be taken from a |
| 2616 | # 'charset' property on the policy. |
| 2617 | charset = 'utf-8' |
| 2618 | want_encoding = True |
| 2619 | if part.token_type == 'mime-parameters': |
| 2620 | # Mime parameter folding (using RFC2231) is extra special. |
| 2621 | _fold_mime_parameters(part, lines, maxlen, encoding) |
| 2622 | continue |
| 2623 | if want_encoding and not wrap_as_ew_blocked: |
| 2624 | if not part.as_ew_allowed: |
| 2625 | want_encoding = False |
| 2626 | last_ew = None |
| 2627 | if part.syntactic_break: |
Jens Troeger | 45b2f88 | 2019-05-14 11:07:39 +1000 | [diff] [blame^] | 2628 | encoded_part = part.fold(policy=policy)[:-len(policy.linesep)] |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 2629 | if policy.linesep not in encoded_part: |
| 2630 | # It fits on a single line |
| 2631 | if len(encoded_part) > maxlen - len(lines[-1]): |
| 2632 | # But not on this one, so start a new one. |
| 2633 | newline = _steal_trailing_WSP_if_exists(lines) |
| 2634 | # XXX what if encoded_part has no leading FWS? |
| 2635 | lines.append(newline) |
| 2636 | lines[-1] += encoded_part |
| 2637 | continue |
| 2638 | # Either this is not a major syntactic break, so we don't |
| 2639 | # want it on a line by itself even if it fits, or it |
| 2640 | # doesn't fit on a line by itself. Either way, fall through |
| 2641 | # to unpacking the subparts and wrapping them. |
| 2642 | if not hasattr(part, 'encode'): |
| 2643 | # It's not a Terminal, do each piece individually. |
| 2644 | parts = list(part) + parts |
| 2645 | else: |
| 2646 | # It's a terminal, wrap it as an encoded word, possibly |
| 2647 | # combining it with previously encoded words if allowed. |
| 2648 | last_ew = _fold_as_ew(tstr, lines, maxlen, last_ew, |
| 2649 | part.ew_combine_allowed, charset) |
| 2650 | want_encoding = False |
| 2651 | continue |
| 2652 | if len(tstr) <= maxlen - len(lines[-1]): |
| 2653 | lines[-1] += tstr |
| 2654 | continue |
| 2655 | # This part is too long to fit. The RFC wants us to break at |
| 2656 | # "major syntactic breaks", so unless we don't consider this |
| 2657 | # to be one, check if it will fit on the next line by itself. |
| 2658 | if (part.syntactic_break and |
| 2659 | len(tstr) + 1 <= maxlen): |
| 2660 | newline = _steal_trailing_WSP_if_exists(lines) |
| 2661 | if newline or part.startswith_fws(): |
| 2662 | lines.append(newline + tstr) |
| 2663 | continue |
| 2664 | if not hasattr(part, 'encode'): |
| 2665 | # It's not a terminal, try folding the subparts. |
| 2666 | newparts = list(part) |
| 2667 | if not part.as_ew_allowed: |
| 2668 | wrap_as_ew_blocked += 1 |
| 2669 | newparts.append(end_ew_not_allowed) |
| 2670 | parts = newparts + parts |
| 2671 | continue |
| 2672 | if part.as_ew_allowed and not wrap_as_ew_blocked: |
| 2673 | # It doesn't need CTE encoding, but encode it anyway so we can |
| 2674 | # wrap it. |
| 2675 | parts.insert(0, part) |
| 2676 | want_encoding = True |
| 2677 | continue |
| 2678 | # We can't figure out how to wrap, it, so give up. |
| 2679 | newline = _steal_trailing_WSP_if_exists(lines) |
| 2680 | if newline or part.startswith_fws(): |
| 2681 | lines.append(newline + tstr) |
| 2682 | else: |
| 2683 | # We can't fold it onto the next line either... |
| 2684 | lines[-1] += tstr |
| 2685 | return policy.linesep.join(lines) + policy.linesep |
| 2686 | |
| 2687 | def _fold_as_ew(to_encode, lines, maxlen, last_ew, ew_combine_allowed, charset): |
| 2688 | """Fold string to_encode into lines as encoded word, combining if allowed. |
| 2689 | Return the new value for last_ew, or None if ew_combine_allowed is False. |
| 2690 | |
| 2691 | If there is already an encoded word in the last line of lines (indicated by |
| 2692 | a non-None value for last_ew) and ew_combine_allowed is true, decode the |
| 2693 | existing ew, combine it with to_encode, and re-encode. Otherwise, encode |
| 2694 | to_encode. In either case, split to_encode as necessary so that the |
| 2695 | encoded segments fit within maxlen. |
| 2696 | |
| 2697 | """ |
| 2698 | if last_ew is not None and ew_combine_allowed: |
| 2699 | to_encode = str( |
| 2700 | get_unstructured(lines[-1][last_ew:] + to_encode)) |
| 2701 | lines[-1] = lines[-1][:last_ew] |
| 2702 | if to_encode[0] in WSP: |
| 2703 | # We're joining this to non-encoded text, so don't encode |
| 2704 | # the leading blank. |
| 2705 | leading_wsp = to_encode[0] |
| 2706 | to_encode = to_encode[1:] |
| 2707 | if (len(lines[-1]) == maxlen): |
| 2708 | lines.append(_steal_trailing_WSP_if_exists(lines)) |
| 2709 | lines[-1] += leading_wsp |
| 2710 | trailing_wsp = '' |
| 2711 | if to_encode[-1] in WSP: |
| 2712 | # Likewise for the trailing space. |
| 2713 | trailing_wsp = to_encode[-1] |
| 2714 | to_encode = to_encode[:-1] |
| 2715 | new_last_ew = len(lines[-1]) if last_ew is None else last_ew |
| 2716 | while to_encode: |
| 2717 | remaining_space = maxlen - len(lines[-1]) |
| 2718 | # The RFC2047 chrome takes up 7 characters plus the length |
| 2719 | # of the charset name. |
| 2720 | encode_as = 'utf-8' if charset == 'us-ascii' else charset |
| 2721 | text_space = remaining_space - len(encode_as) - 7 |
| 2722 | if text_space <= 0: |
| 2723 | lines.append(' ') |
| 2724 | # XXX We'll get an infinite loop here if maxlen is <= 7 |
| 2725 | continue |
| 2726 | first_part = to_encode[:text_space] |
| 2727 | ew = _ew.encode(first_part, charset=encode_as) |
| 2728 | excess = len(ew) - remaining_space |
| 2729 | if excess > 0: |
| 2730 | # encode always chooses the shortest encoding, so this |
| 2731 | # is guaranteed to fit at this point. |
| 2732 | first_part = first_part[:-excess] |
| 2733 | ew = _ew.encode(first_part) |
| 2734 | lines[-1] += ew |
| 2735 | to_encode = to_encode[len(first_part):] |
| 2736 | if to_encode: |
| 2737 | lines.append(' ') |
| 2738 | new_last_ew = len(lines[-1]) |
| 2739 | lines[-1] += trailing_wsp |
| 2740 | return new_last_ew if ew_combine_allowed else None |
| 2741 | |
| 2742 | def _fold_mime_parameters(part, lines, maxlen, encoding): |
| 2743 | """Fold TokenList 'part' into the 'lines' list as mime parameters. |
| 2744 | |
| 2745 | Using the decoded list of parameters and values, format them according to |
| 2746 | the RFC rules, including using RFC2231 encoding if the value cannot be |
Leo Arias | c3d9508 | 2018-02-03 18:36:10 -0600 | [diff] [blame] | 2747 | expressed in 'encoding' and/or the parameter+value is too long to fit |
| 2748 | within 'maxlen'. |
R. David Murray | 85d5c18 | 2017-12-03 18:51:41 -0500 | [diff] [blame] | 2749 | |
| 2750 | """ |
| 2751 | # Special case for RFC2231 encoding: start from decoded values and use |
| 2752 | # RFC2231 encoding iff needed. |
| 2753 | # |
| 2754 | # Note that the 1 and 2s being added to the length calculations are |
| 2755 | # accounting for the possibly-needed spaces and semicolons we'll be adding. |
| 2756 | # |
| 2757 | for name, value in part.params: |
| 2758 | # XXX What if this ';' puts us over maxlen the first time through the |
| 2759 | # loop? We should split the header value onto a newline in that case, |
| 2760 | # but to do that we need to recognize the need earlier or reparse the |
| 2761 | # header, so I'm going to ignore that bug for now. It'll only put us |
| 2762 | # one character over. |
| 2763 | if not lines[-1].rstrip().endswith(';'): |
| 2764 | lines[-1] += ';' |
| 2765 | charset = encoding |
| 2766 | error_handler = 'strict' |
| 2767 | try: |
| 2768 | value.encode(encoding) |
| 2769 | encoding_required = False |
| 2770 | except UnicodeEncodeError: |
| 2771 | encoding_required = True |
| 2772 | if utils._has_surrogates(value): |
| 2773 | charset = 'unknown-8bit' |
| 2774 | error_handler = 'surrogateescape' |
| 2775 | else: |
| 2776 | charset = 'utf-8' |
| 2777 | if encoding_required: |
| 2778 | encoded_value = urllib.parse.quote( |
| 2779 | value, safe='', errors=error_handler) |
| 2780 | tstr = "{}*={}''{}".format(name, charset, encoded_value) |
| 2781 | else: |
| 2782 | tstr = '{}={}'.format(name, quote_string(value)) |
| 2783 | if len(lines[-1]) + len(tstr) + 1 < maxlen: |
| 2784 | lines[-1] = lines[-1] + ' ' + tstr |
| 2785 | continue |
| 2786 | elif len(tstr) + 2 <= maxlen: |
| 2787 | lines.append(' ' + tstr) |
| 2788 | continue |
| 2789 | # We need multiple sections. We are allowed to mix encoded and |
| 2790 | # non-encoded sections, but we aren't going to. We'll encode them all. |
| 2791 | section = 0 |
| 2792 | extra_chrome = charset + "''" |
| 2793 | while value: |
| 2794 | chrome_len = len(name) + len(str(section)) + 3 + len(extra_chrome) |
| 2795 | if maxlen <= chrome_len + 3: |
| 2796 | # We need room for the leading blank, the trailing semicolon, |
| 2797 | # and at least one character of the value. If we don't |
| 2798 | # have that, we'd be stuck, so in that case fall back to |
| 2799 | # the RFC standard width. |
| 2800 | maxlen = 78 |
| 2801 | splitpoint = maxchars = maxlen - chrome_len - 2 |
| 2802 | while True: |
| 2803 | partial = value[:splitpoint] |
| 2804 | encoded_value = urllib.parse.quote( |
| 2805 | partial, safe='', errors=error_handler) |
| 2806 | if len(encoded_value) <= maxchars: |
| 2807 | break |
| 2808 | splitpoint -= 1 |
| 2809 | lines.append(" {}*{}*={}{}".format( |
| 2810 | name, section, extra_chrome, encoded_value)) |
| 2811 | extra_chrome = '' |
| 2812 | section += 1 |
| 2813 | value = value[splitpoint:] |
| 2814 | if value: |
| 2815 | lines[-1] += ';' |