1 | """data_lang/htm8.py
|
2 |
|
3 | TODO
|
4 |
|
5 | API:
|
6 | - Get rid of AttrValueLexer - this should be in the TagLexer
|
7 | - this also means that unquoted values can be more similar
|
8 | - We can use a single lexer mode for everything inside <>
|
9 | - the SPACE is the only difference
|
10 | - Deprecate tag_lexer.GetTagName() in favor of lx.CanonicalTagName() or
|
11 | _LiteralTagName()
|
12 | - UTF-8 check, like JSON8
|
13 | - re2c
|
14 | - port lexer, which will fix static typing issues
|
15 | - the abstraction needs to support submatch?
|
16 | - for finding the end of a tag, etc.?
|
17 |
|
18 | - LexError and ParseError need details
|
19 | - harmonize with data_lang/j8.py, which uses error.Decode(msg, ...,
|
20 | cur_line_num)
|
21 |
|
22 | - Copy all errors into doc/ref/chap-errors.md
|
23 | - This helps understand the language
|
24 |
|
25 | - Update doc/htm8.md
|
26 | - list of Algorithms:
|
27 | - lex just the top level
|
28 | - lex both levels
|
29 | - and match tags - this is the level for value.Htm8Frag?
|
30 | - convert to XML!
|
31 | - lazy selection by tag, or attr (id= and class=)
|
32 | - lazy selection by CSS selector expression
|
33 | - convert to DOMTree
|
34 | - sed-like replacement of DOM Tree or element
|
35 | - untrusted HTML filter, e.g. like StackOverflow / Reddit
|
36 | - this is Safe HTM8
|
37 | - should have a zero alloc way to support this, with good errors?
|
38 | - I think most of them silently strip data
|
39 | """
|
40 |
|
41 | import re
|
42 |
|
43 | from typing import Dict, List, Tuple, Optional, IO, Iterator, Any
|
44 |
|
45 | from _devbuild.gen.htm8_asdl import (h8_id, h8_id_t, h8_tag_id, h8_tag_id_t,
|
46 | h8_tag_id_str, attr_name_t, attr_value_t)
|
47 | from doctools.util import log
|
48 |
|
49 |
|
50 | class LexError(Exception):
|
51 | """
|
52 | Examples of lex errors:
|
53 |
|
54 | - h8_id.Invalid, like <> or &&
|
55 | - Unclosed <!-- <? <![CDATA[ <script> <style>
|
56 | """
|
57 |
|
58 | def __init__(self, s, start_pos):
|
59 | # type: (str, int) -> None
|
60 | self.s = s
|
61 | self.start_pos = start_pos
|
62 |
|
63 | def __str__(self):
|
64 | # type: () -> str
|
65 | return '(LexError %r)' % (self.s[self.start_pos:self.start_pos + 20])
|
66 |
|
67 |
|
68 | def _FindLineNum(s, error_pos):
|
69 | # type: (str, int) -> int
|
70 | current_pos = 0
|
71 | line_num = 1
|
72 | while True:
|
73 | newline_pos = s.find('\n', current_pos)
|
74 | #log('current = %d, N %d, line %d', current_pos, newline_pos, line_num)
|
75 |
|
76 | if newline_pos == -1: # this is the last line
|
77 | return line_num
|
78 | if newline_pos >= error_pos:
|
79 | return line_num
|
80 | line_num += 1
|
81 | current_pos = newline_pos + 1
|
82 |
|
83 |
|
84 | class ParseError(Exception):
|
85 | """
|
86 | Examples of parse errors
|
87 |
|
88 | - unbalanced tag structure
|
89 | - ul_table.py errors
|
90 | """
|
91 |
|
92 | def __init__(self, msg, s=None, start_pos=-1):
|
93 | # type: (str, Optional[str], int) -> None
|
94 | self.msg = msg
|
95 | self.s = s
|
96 | self.start_pos = start_pos
|
97 |
|
98 | def __str__(self):
|
99 | # type: () -> str
|
100 | if self.s is not None:
|
101 | assert self.start_pos != -1, self.start_pos
|
102 | snippet = (self.s[self.start_pos:self.start_pos + 20])
|
103 |
|
104 | line_num = _FindLineNum(self.s, self.start_pos)
|
105 | else:
|
106 | snippet = ''
|
107 | line_num = -1
|
108 | msg = 'line %d: %r %r' % (line_num, self.msg, snippet)
|
109 | return msg
|
110 |
|
111 |
|
112 | class Output(object):
|
113 | """Takes an underlying input buffer and an output file. Maintains a
|
114 | position in the input buffer.
|
115 |
|
116 | Print FROM the input or print new text to the output.
|
117 | """
|
118 |
|
119 | def __init__(self, s, f, left_pos=0, right_pos=-1):
|
120 | # type: (str, IO[str], int, int) -> None
|
121 | self.s = s
|
122 | self.f = f
|
123 | self.pos = left_pos
|
124 | self.right_pos = len(s) if right_pos == -1 else right_pos
|
125 |
|
126 | def SkipTo(self, pos):
|
127 | # type: (int) -> None
|
128 | """Skip to a position."""
|
129 | self.pos = pos
|
130 |
|
131 | def PrintUntil(self, pos):
|
132 | # type: (int) -> None
|
133 | """Print until a position."""
|
134 | piece = self.s[self.pos:pos]
|
135 | self.f.write(piece)
|
136 | self.pos = pos
|
137 |
|
138 | def PrintTheRest(self):
|
139 | # type: () -> None
|
140 | """Print until the end of the string."""
|
141 | self.PrintUntil(self.right_pos)
|
142 |
|
143 | def Print(self, s):
|
144 | # type: (str) -> None
|
145 | """Print text to the underlying buffer."""
|
146 | self.f.write(s)
|
147 |
|
148 |
|
149 | def MakeLexer(rules):
|
150 | return [(re.compile(pat, re.VERBOSE), i) for (pat, i) in rules]
|
151 |
|
152 |
|
153 | #
|
154 | # Eggex
|
155 | #
|
156 | # Tag = / ~['>']+ /
|
157 |
|
158 | # Is this valid? A single character?
|
159 | # Tag = / ~'>'* /
|
160 |
|
161 | # Maybe better: / [NOT '>']+/
|
162 | # capital letters not allowed there?
|
163 | #
|
164 | # But then this is confusing:
|
165 | # / [NOT ~digit]+/
|
166 | #
|
167 | # / [NOT digit] / is [^\d]
|
168 | # / ~digit / is \D
|
169 | #
|
170 | # Or maybe:
|
171 | #
|
172 | # / [~ digit]+ /
|
173 | # / [~ '>']+ /
|
174 | # / [NOT '>']+ /
|
175 |
|
176 | # End = / '</' Tag '>' /
|
177 | # StartEnd = / '<' Tag '/>' /
|
178 | # Start = / '<' Tag '>' /
|
179 | #
|
180 | # EntityRef = / '&' dot{* N} ';' /
|
181 |
|
182 | # Tag name, or attribute name
|
183 | # colon is used in XML
|
184 |
|
185 | # https://www.w3.org/TR/xml/#NT-Name
|
186 | # Hm there is a lot of unicode stuff. We are simplifying parsing
|
187 |
|
188 | _NAME = r'[a-zA-Z][a-zA-Z0-9:_\-]*' # must start with letter
|
189 |
|
190 | CHAR_LEX = [
|
191 | # Characters
|
192 | # https://www.w3.org/TR/xml/#sec-references
|
193 | (r'&\# [0-9]+ ;', h8_id.DecChar),
|
194 | (r'&\# x[0-9a-fA-F]+ ;', h8_id.HexChar),
|
195 | (r'& %s ;' % _NAME, h8_id.CharEntity),
|
196 | # Allow unquoted, and quoted
|
197 | (r'&', h8_id.BadAmpersand),
|
198 | ]
|
199 |
|
200 | HTM8_LEX = CHAR_LEX + [
|
201 | # TODO: CommentBegin, ProcessingBegin, CDataBegin could have an additional
|
202 | # action associated with them? The ending substring
|
203 | (r'<!--', h8_id.CommentBegin),
|
204 |
|
205 | # Processing instruction are used for the XML header:
|
206 | # <?xml version="1.0" encoding="UTF-8"?>
|
207 | # They are technically XML-only, but in HTML5, they are another kind of
|
208 | # comment:
|
209 | #
|
210 | # https://developer.mozilla.org/en-US/docs/Web/API/ProcessingInstruction
|
211 | #
|
212 | (r'<\?', h8_id.ProcessingBegin),
|
213 | # Not necessary in HTML5, but occurs in XML
|
214 | (r'<!\[CDATA\[', h8_id.CDataBegin), # <![CDATA[
|
215 |
|
216 | # Markup declarations
|
217 | # - In HTML5, there is only <!DOCTYPE html>
|
218 | # - XML has 4 more declarations: <!ELEMENT ...> ATTLIST ENTITY NOTATION
|
219 | # - these seem to be part of DTD
|
220 | # - it's useful to skip these, and be able to parse the rest of the document
|
221 | # - Note: < is allowed?
|
222 | (r'<! [^>\x00]+ >', h8_id.Decl),
|
223 |
|
224 | # Tags
|
225 | # Notes:
|
226 | # - We look for a valid tag name, but we don't validate attributes.
|
227 | # That's done in the tag lexer.
|
228 | # - We don't allow leading whitespace
|
229 | (r'</ (%s) >' % _NAME, h8_id.EndTag),
|
230 | # self-closing <br/> comes before StartTag
|
231 | # could/should these be collapsed into one rule?
|
232 | (r'< (%s) [^>\x00]* />' % _NAME, h8_id.StartEndTag), # end </a>
|
233 | (r'< (%s) [^>\x00]* >' % _NAME, h8_id.StartTag), # start <a>
|
234 |
|
235 | # HTML5 allows unescaped > in raw data, but < is not allowed.
|
236 | # https://stackoverflow.com/questions/10462348/right-angle-bracket-in-html
|
237 | #
|
238 | # - My early blog has THREE errors when disallowing >
|
239 | # - So do some .wwz files
|
240 | (r'[^&<>\x00]+', h8_id.RawData),
|
241 | (r'>', h8_id.BadGreaterThan),
|
242 | # < is an error
|
243 | (r'.', h8_id.Invalid),
|
244 | ]
|
245 |
|
246 | # Old notes:
|
247 | #
|
248 | # Non-greedy matches are regular and can be matched in linear time
|
249 | # with RE2.
|
250 | #
|
251 | # https://news.ycombinator.com/item?id=27099798
|
252 | #
|
253 |
|
254 | # This person tried to do it with a regex:
|
255 | #
|
256 | # https://skeptric.com/html-comment-regexp/index.html
|
257 |
|
258 | # . is any char except newline
|
259 | # https://re2c.org/manual/manual_c.html
|
260 |
|
261 | # Discarded options
|
262 | #(r'<!-- .*? -->', h8_id.Comment),
|
263 |
|
264 | # Hack from Claude: \s\S instead of re.DOTALL. I don't like this
|
265 | #(r'<!-- [\s\S]*? -->', h8_id.Comment),
|
266 | #(r'<!-- (?:.|[\n])*? -->', h8_id.Comment),
|
267 |
|
268 | HTM8_LEX_COMPILED = MakeLexer(HTM8_LEX)
|
269 |
|
270 |
|
271 | class Lexer(object):
|
272 |
|
273 | def __init__(self, s, left_pos=0, right_pos=-1, no_special_tags=False):
|
274 | # type: (str, int, int, bool) -> None
|
275 | self.s = s
|
276 | self.pos = left_pos
|
277 | self.right_pos = len(s) if right_pos == -1 else right_pos
|
278 | self.no_special_tags = no_special_tags
|
279 |
|
280 | # string -> compiled regex pattern object
|
281 | self.cache = {} # type: Dict[str, Any]
|
282 |
|
283 | # either </script> or </style> - we search until we see that
|
284 | self.search_state = None # type: Optional[str]
|
285 |
|
286 | # Position of tag name, if applicable
|
287 | # - Set after you get a StartTag, EndTag, or StartEndTag
|
288 | # - Unset on other tags
|
289 | self.tag_pos_left = -1
|
290 | self.tag_pos_right = -1
|
291 |
|
292 | def _Read(self):
|
293 | # type: () -> Tuple[h8_id_t, int]
|
294 | if self.pos == self.right_pos:
|
295 | return h8_id.EndOfStream, self.pos
|
296 |
|
297 | assert self.pos < self.right_pos, self.pos
|
298 |
|
299 | if self.search_state is not None and not self.no_special_tags:
|
300 | # TODO: case-insensitive search for </SCRIPT> <SCRipt> ?
|
301 | #
|
302 | # Another strategy: enter a mode where we find ONLY the end tag
|
303 | # regex, and any data that's not <, and then check the canonical
|
304 | # tag name for 'script' or 'style'.
|
305 | pos = self.s.find(self.search_state, self.pos)
|
306 | if pos == -1:
|
307 | # unterminated <script> or <style>
|
308 | raise LexError(self.s, self.pos)
|
309 | self.search_state = None
|
310 | # beginning
|
311 | return h8_id.HtmlCData, pos
|
312 |
|
313 | # Find the first match.
|
314 | # Note: frontend/match.py uses _LongestMatch(), which is different!
|
315 | # TODO: reconcile them. This lexer should be expressible in re2c.
|
316 |
|
317 | for pat, tok_id in HTM8_LEX_COMPILED:
|
318 | m = pat.match(self.s, self.pos)
|
319 | if m:
|
320 | if tok_id in (h8_id.StartTag, h8_id.EndTag, h8_id.StartEndTag):
|
321 | self.tag_pos_left = m.start(1)
|
322 | self.tag_pos_right = m.end(1)
|
323 | else:
|
324 | # Reset state
|
325 | self.tag_pos_left = -1
|
326 | self.tag_pos_right = -1
|
327 |
|
328 | if tok_id == h8_id.CommentBegin:
|
329 | pos = self.s.find('-->', self.pos)
|
330 | if pos == -1:
|
331 | # unterminated <!--
|
332 | raise LexError(self.s, self.pos)
|
333 | return h8_id.Comment, pos + 3 # -->
|
334 |
|
335 | if tok_id == h8_id.ProcessingBegin:
|
336 | pos = self.s.find('?>', self.pos)
|
337 | if pos == -1:
|
338 | # unterminated <?
|
339 | raise LexError(self.s, self.pos)
|
340 | return h8_id.Processing, pos + 2 # ?>
|
341 |
|
342 | if tok_id == h8_id.CDataBegin:
|
343 | pos = self.s.find(']]>', self.pos)
|
344 | if pos == -1:
|
345 | # unterminated <![CDATA[
|
346 | raise LexError(self.s, self.pos)
|
347 | return h8_id.CData, pos + 3 # ]]>
|
348 |
|
349 | if tok_id == h8_id.StartTag:
|
350 | # TODO: reduce allocations
|
351 | if (self.TagNameEquals('script') or
|
352 | self.TagNameEquals('style')):
|
353 | # <SCRipt a=b> -> </SCRipt>
|
354 | self.search_state = '</' + self._LiteralTagName() + '>'
|
355 |
|
356 | return tok_id, m.end()
|
357 | else:
|
358 | raise AssertionError('h8_id.Invalid rule should have matched')
|
359 |
|
360 | def TagNameEquals(self, expected):
|
361 | # type: (str) -> bool
|
362 | assert self.tag_pos_left != -1, self.tag_pos_left
|
363 | assert self.tag_pos_right != -1, self.tag_pos_right
|
364 |
|
365 | # TODO: In C++, this does not need an allocation. Can we test
|
366 | # directly?
|
367 | return expected == self.CanonicalTagName()
|
368 |
|
369 | def _LiteralTagName(self):
|
370 | # type: () -> str
|
371 | assert self.tag_pos_left != -1, self.tag_pos_left
|
372 | assert self.tag_pos_right != -1, self.tag_pos_right
|
373 |
|
374 | return self.s[self.tag_pos_left:self.tag_pos_right]
|
375 |
|
376 | def CanonicalTagName(self):
|
377 | # type: () -> str
|
378 | tag_name = self._LiteralTagName()
|
379 | # Most tags are already lower case, so avoid allocation with this conditional
|
380 | # TODO: this could go in the mycpp runtime?
|
381 | if tag_name.islower():
|
382 | return tag_name
|
383 | else:
|
384 | return tag_name.lower()
|
385 |
|
386 | def Read(self):
|
387 | # type: () -> Tuple[h8_id_t, int]
|
388 | tok_id, end_pos = self._Read()
|
389 | self.pos = end_pos # advance
|
390 | return tok_id, end_pos
|
391 |
|
392 | def LookAhead(self, regex):
|
393 | # type: (str) -> bool
|
394 | """
|
395 | Currently used for ul_table.py. But taking a dynamic regex string is
|
396 | not the right interface.
|
397 | """
|
398 | # Cache the regex compilation. This could also be LookAheadFor(THEAD)
|
399 | # or something.
|
400 | pat = self.cache.get(regex)
|
401 | if pat is None:
|
402 | pat = re.compile(regex)
|
403 | self.cache[regex] = pat
|
404 |
|
405 | m = pat.match(self.s, self.pos)
|
406 | return m is not None
|
407 |
|
408 |
|
409 | class AttrLexer(object):
|
410 | """
|
411 | We can also invert this
|
412 |
|
413 | Unquoted (List[h8_id] tok_ids, List[int] end_pos)
|
414 |
|
415 | It would be nice to have a special case for the singleton, since that is
|
416 | very common.
|
417 |
|
418 | Simple (int tag_name_start, int tag_name_end, int attr_value_tag,
|
419 | int value_start, int value_end)
|
420 | This would cover many cases
|
421 |
|
422 | The other option is to create many different events, and have AttrValueLexer
|
423 | But I think that is annoying and overly detailed.
|
424 |
|
425 | Operations:
|
426 | - GetAttrRaw('foo')
|
427 | - AllAttrsRaw()
|
428 | - AllAttrsRawSlice()
|
429 |
|
430 | class= query - well we can do this with Space tokens I think - we should
|
431 | have an optimization
|
432 | id= query - ditto, we should just have a predicate
|
433 |
|
434 | Zero allocs:
|
435 | tag query - TagNameEquals()
|
436 |
|
437 | So I guess we have to write a HasClass('foo') and IdEquals('bar') on top of
|
438 | this. Yes.
|
439 |
|
440 | tag_lx.Reset2(...) # we should pass it the tag_name_end position
|
441 | tag_lx.Read() -> bool # success or fail? Or Attr or Invalid
|
442 | .AttrNameEquals('foo') -> bool # id and class query
|
443 | .GetAttrName() -> str # for getting them all
|
444 | .GetRawValue() -> Tuple[h8_tag_id, start, end] # just beginning and end
|
445 | .GetValueTokens() -> Tuple[h8_tag_id, TokenList]
|
446 | .TokenList = Tuple[List[h8_id], List[int end_pos]
|
447 |
|
448 | You could also have
|
449 |
|
450 | tag_lx.GetValueTokenId() -> Tuple[h8_id, end_pos]
|
451 |
|
452 | And then you read it until it's " or ' or space ? We probably won't have
|
453 | that use case to start.
|
454 | """
|
455 |
|
456 | def __init__(self, s):
|
457 | # type: (str) -> None
|
458 | self.s = s
|
459 | self.tag_name_pos = -1 # Invalid
|
460 | self.tag_end_pos = -1
|
461 |
|
462 | def Init(self, tag_name_pos, end_pos):
|
463 | # type: (int, int) -> None
|
464 | """Initialize so we can read names and values.
|
465 |
|
466 | Example:
|
467 | 'x <a y>' # tag_name_pos=4, end_pos=6
|
468 | 'x <a>' # tag_name_pos=4, end_pos=4
|
469 |
|
470 | The Reset() method is used to reuse instances of the AttrLexer object.
|
471 | """
|
472 | assert tag_name_pos >= 0, tag_name_pos
|
473 | assert end_pos >= 0, end_pos
|
474 |
|
475 | self.tag_name_pos = tag_name_pos
|
476 | self.end_pos = end_pos
|
477 |
|
478 | def ReadName(self):
|
479 | # type: () -> Tuple[attr_name_t, int, int]
|
480 | """Reads the attribute name
|
481 |
|
482 | EOF case:
|
483 | <a>
|
484 | <a >
|
485 |
|
486 | Error case:
|
487 | <a !>
|
488 | <a foo=bar !>
|
489 | """
|
490 | pass
|
491 |
|
492 | def AttrNameEquals(self, s):
|
493 | # type: (str) -> bool
|
494 | """
|
495 | TODO: Must call this after ReadName() ?
|
496 | Because that can FAIL.
|
497 | """
|
498 | pass
|
499 |
|
500 | def ReadRawValue(self):
|
501 | # type: () -> Tuple[attr_value_t, int, int]
|
502 | """Read the attribute value.
|
503 |
|
504 | In general, it is escaped or "raw"
|
505 |
|
506 | Note: Assuming ReadName() returned a value, this should NOT fail.
|
507 | """
|
508 | # NOTE: if = is not found, set state
|
509 |
|
510 | pass
|
511 |
|
512 | def SkipValue(self):
|
513 | # type: () -> None
|
514 | # Just ignore it and return
|
515 | self.ReadRawValue()
|
516 |
|
517 | def ReadValueAndDecode(self):
|
518 | # type: () -> str
|
519 | """Read the attribute vlaue
|
520 | """
|
521 | # TODO: tokenize it
|
522 | pass
|
523 |
|
524 |
|
525 | # Tag names:
|
526 | # Match <a or </a
|
527 | # Match <h2, but not <2h
|
528 | #
|
529 | # HTML 5 doesn't restrict tag names at all
|
530 | # https://html.spec.whatwg.org/#toc-syntax
|
531 | #
|
532 | # XML allows : - .
|
533 | # https://www.w3.org/TR/xml/#NT-NameChar
|
534 |
|
535 | # Namespaces for MathML, SVG
|
536 | # XLink, XML, XMLNS
|
537 | #
|
538 | # https://infra.spec.whatwg.org/#namespaces
|
539 | #
|
540 | # Allow - for td-attrs
|
541 |
|
542 | # Be very lenient - just no whitespace or special HTML chars
|
543 | # I don't think this is more lenient than HTML5, though we should check.
|
544 | _UNQUOTED_VALUE = r'''[^ \t\r\n<>&"'\x00]*'''
|
545 |
|
546 | # TODO: we don't need to capture the tag name here? That's done at the top
|
547 | # level
|
548 | _TAG_RE = re.compile(r'/? \s* (%s)' % _NAME, re.VERBOSE)
|
549 |
|
550 | _TAG_LAST_RE = re.compile(r'\s* /? >', re.VERBOSE)
|
551 |
|
552 | # To match href="foo"
|
553 | # Note: in HTML5 and XML, single quoted attributes are also valid
|
554 |
|
555 | # <button disabled> is standard usage
|
556 |
|
557 | # NOTE: This used to allow whitespace around =
|
558 | # <a foo = "bar"> makes sense in XML
|
559 | # But then you also have
|
560 | # <a foo= bar> - which is TWO attributes, in HTML5
|
561 | # So the space is problematic
|
562 |
|
563 | _ATTR_RE = re.compile(
|
564 | r'''
|
565 | \s+ # Leading whitespace is required
|
566 | (%s) # Attribute name
|
567 | (?: # Optional attribute value
|
568 | \s* = \s* # Spaces allowed around =
|
569 | (?:
|
570 | " ([^>"\x00]*) " # double quoted value
|
571 | | ' ([^>'\x00]*) ' # single quoted value
|
572 | | (%s) # Attribute value
|
573 | )
|
574 | )?
|
575 | ''' % (_NAME, _UNQUOTED_VALUE), re.VERBOSE)
|
576 |
|
577 |
|
578 | class TagLexer(object):
|
579 | """
|
580 | Given a tag like <a href="..."> or <link type="..." />, the TagLexer
|
581 | provides a few operations:
|
582 |
|
583 | - What is the tag?
|
584 | - Iterate through the attributes, giving (name, value_start_pos, value_end_pos)
|
585 | """
|
586 |
|
587 | def __init__(self, s):
|
588 | # type: (str) -> None
|
589 | self.s = s
|
590 | self.start_pos = -1 # Invalid
|
591 | self.end_pos = -1
|
592 |
|
593 | def Reset(self, start_pos, end_pos):
|
594 | # type: (int, int) -> None
|
595 | """Reuse instances of this object."""
|
596 | assert start_pos >= 0, start_pos
|
597 | assert end_pos >= 0, end_pos
|
598 |
|
599 | self.start_pos = start_pos
|
600 | self.end_pos = end_pos
|
601 |
|
602 | def WholeTagString(self):
|
603 | # type: () -> str
|
604 | """Return the entire tag string, e.g. <a href='foo'>"""
|
605 | return self.s[self.start_pos:self.end_pos]
|
606 |
|
607 | def GetTagName(self):
|
608 | # type: () -> str
|
609 | # First event
|
610 | tok_id, start, end = next(self.Tokens())
|
611 | return self.s[start:end]
|
612 |
|
613 | def GetSpanForAttrValue(self, attr_name):
|
614 | # type: (str) -> Tuple[int, int]
|
615 | """
|
616 | Used by oils_doc.py, for href shortcuts
|
617 | """
|
618 | # Algorithm: search for QuotedValue or UnquotedValue after AttrName
|
619 | # TODO: Could also cache these
|
620 |
|
621 | events = self.Tokens()
|
622 | val = (-1, -1)
|
623 | try:
|
624 | while True:
|
625 | tok_id, start, end = next(events)
|
626 | if tok_id == h8_tag_id.AttrName:
|
627 | name = self.s[start:end]
|
628 | if name == attr_name:
|
629 | # The value should come next
|
630 | tok_id, start, end = next(events)
|
631 | assert tok_id in (
|
632 | h8_tag_id.QuotedValue, h8_tag_id.UnquotedValue,
|
633 | h8_tag_id.MissingValue), h8_tag_id_str(tok_id)
|
634 | val = start, end
|
635 | break
|
636 |
|
637 | except StopIteration:
|
638 | pass
|
639 | return val
|
640 |
|
641 | def GetAttrRaw(self, attr_name):
|
642 | # type: (str) -> Optional[str]
|
643 | """
|
644 | Return the value, which may be UNESCAPED.
|
645 | """
|
646 | start, end = self.GetSpanForAttrValue(attr_name)
|
647 | if start == -1:
|
648 | return None
|
649 | return self.s[start:end]
|
650 |
|
651 | def AllAttrsRawSlice(self):
|
652 | # type: () -> List[Tuple[str, int, int]]
|
653 | """
|
654 | Get a list of pairs [('class', 3, 5), ('href', 9, 12)]
|
655 | """
|
656 | slices = []
|
657 | events = self.Tokens()
|
658 | try:
|
659 | while True:
|
660 | tok_id, start, end = next(events)
|
661 | if tok_id == h8_tag_id.AttrName:
|
662 | name = self.s[start:end]
|
663 |
|
664 | # The value should come next
|
665 | tok_id, start, end = next(events)
|
666 | assert tok_id in (
|
667 | h8_tag_id.QuotedValue, h8_tag_id.UnquotedValue,
|
668 | h8_tag_id.MissingValue), h8_tag_id_str(tok_id)
|
669 | # Note: quoted values may have &
|
670 | # We would need ANOTHER lexer to unescape them, but we
|
671 | # don't need that for ul-table
|
672 | slices.append((name, start, end))
|
673 | except StopIteration:
|
674 | pass
|
675 | return slices
|
676 |
|
677 | def AllAttrsRaw(self):
|
678 | # type: () -> List[Tuple[str, str]]
|
679 | """
|
680 | Get a list of pairs [('class', 'foo'), ('href', '?foo=1&bar=2')]
|
681 |
|
682 | The quoted values may be escaped. We would need another lexer to
|
683 | unescape them.
|
684 | """
|
685 | slices = self.AllAttrsRawSlice()
|
686 | pairs = []
|
687 | for name, start, end in slices:
|
688 | pairs.append((name, self.s[start:end]))
|
689 | return pairs
|
690 |
|
691 | def Tokens(self):
|
692 | # type: () -> Iterator[Tuple[h8_tag_id_t, int, int]]
|
693 | """
|
694 | Yields a sequence of tokens: Tag (AttrName AttrValue?)*
|
695 |
|
696 | Where each Token is (Type, start_pos, end_pos)
|
697 |
|
698 | Note that start and end are NOT redundant! We skip over some unwanted
|
699 | characters.
|
700 | """
|
701 | m = _TAG_RE.match(self.s, self.start_pos + 1)
|
702 | if not m:
|
703 | raise RuntimeError("Couldn't find HTML tag in %r" %
|
704 | self.WholeTagString())
|
705 | yield h8_tag_id.TagName, m.start(1), m.end(1)
|
706 |
|
707 | pos = m.end(0)
|
708 | #log('POS %d', pos)
|
709 |
|
710 | while True:
|
711 | # don't search past the end
|
712 | m = _ATTR_RE.match(self.s, pos, self.end_pos)
|
713 | if not m:
|
714 | #log('BREAK pos %d', pos)
|
715 | break
|
716 | #log('AttrName %r', m.group(1))
|
717 |
|
718 | yield h8_tag_id.AttrName, m.start(1), m.end(1)
|
719 |
|
720 | #log('m.groups() %r', m.groups())
|
721 | if m.group(2) is not None:
|
722 | # double quoted
|
723 | yield h8_tag_id.QuotedValue, m.start(2), m.end(2)
|
724 | elif m.group(3) is not None:
|
725 | # single quoted - TODO: could have different token types
|
726 | yield h8_tag_id.QuotedValue, m.start(3), m.end(3)
|
727 | elif m.group(4) is not None:
|
728 | yield h8_tag_id.UnquotedValue, m.start(4), m.end(4)
|
729 | else:
|
730 | # <button disabled>
|
731 | end = m.end(0)
|
732 | yield h8_tag_id.MissingValue, end, end
|
733 |
|
734 | # Skip past the "
|
735 | pos = m.end(0)
|
736 |
|
737 | #log('TOK %r', self.s)
|
738 |
|
739 | m = _TAG_LAST_RE.match(self.s, pos)
|
740 | #log('_TAG_LAST_RE match %r', self.s[pos:])
|
741 | if not m:
|
742 | # Extra data at end of tag. TODO: add messages for all these.
|
743 | raise LexError(self.s, pos)
|
744 |
|
745 |
|
746 | # This is similar but not identical to
|
747 | # " ([^>"\x00]*) " # double quoted value
|
748 | # | ' ([^>'\x00]*) ' # single quoted value
|
749 | #
|
750 | # Note: for unquoted values, & isn't allowed, and thus & and c and
|
751 | # ™ are not allowed. We could relax that?
|
752 | ATTR_VALUE_LEX = CHAR_LEX + [
|
753 | (r'[^>&\x00]+', h8_id.RawData),
|
754 | (r'.', h8_id.Invalid),
|
755 | ]
|
756 |
|
757 | ATTR_VALUE_LEX_COMPILED = MakeLexer(ATTR_VALUE_LEX)
|
758 |
|
759 |
|
760 | class AttrValueLexer(object):
|
761 | """
|
762 | <a href="foo=99&bar">
|
763 | <a href='foo=99&bar'>
|
764 | <a href=unquoted>
|
765 | """
|
766 |
|
767 | def __init__(self, s):
|
768 | # type: (str) -> None
|
769 | self.s = s
|
770 | self.start_pos = -1 # Invalid
|
771 | self.end_pos = -1
|
772 |
|
773 | def Reset(self, start_pos, end_pos):
|
774 | # type: (int, int) -> None
|
775 | """Reuse instances of this object."""
|
776 | assert start_pos >= 0, start_pos
|
777 | assert end_pos >= 0, end_pos
|
778 |
|
779 | self.start_pos = start_pos
|
780 | self.end_pos = end_pos
|
781 |
|
782 | def NumTokens(self):
|
783 | # type: () -> int
|
784 | num_tokens = 0
|
785 | pos = self.start_pos
|
786 | for tok_id, end_pos in self.Tokens():
|
787 | if tok_id == h8_id.Invalid:
|
788 | raise LexError(self.s, pos)
|
789 | pos = end_pos
|
790 | #log('pos %d', pos)
|
791 | num_tokens += 1
|
792 | return num_tokens
|
793 |
|
794 | def Tokens(self):
|
795 | # type: () -> Iterator[Tuple[h8_id_t, int]]
|
796 | pos = self.start_pos
|
797 | while pos < self.end_pos:
|
798 | # Find the first match, like above.
|
799 | # Note: frontend/match.py uses _LongestMatch(), which is different!
|
800 | # TODO: reconcile them. This lexer should be expressible in re2c.
|
801 | for pat, tok_id in ATTR_VALUE_LEX_COMPILED:
|
802 | m = pat.match(self.s, pos)
|
803 | if m:
|
804 | if 0:
|
805 | tok_str = m.group(0)
|
806 | log('token = %r', tok_str)
|
807 |
|
808 | end_pos = m.end(0)
|
809 | yield tok_id, end_pos
|
810 | pos = end_pos
|
811 | break
|
812 | else:
|
813 | raise AssertionError('h8_id.Invalid rule should have matched')
|