1 | #!/usr/bin/env python2
|
2 | """
|
3 | doctools/html_old.py - APIs that should be replaced by data_lang/htm8.py
|
4 |
|
5 | See doc/lazylex.md for details.
|
6 | """
|
7 | from __future__ import print_function
|
8 |
|
9 | import re
|
10 |
|
11 | from _devbuild.gen.htm8_asdl import (h8_id, h8_id_t, h8_tag_id, h8_tag_id_t,
|
12 | h8_tag_id_str)
|
13 | from data_lang.htm8 import (Lexer, LexError, ParseError, Output, _NAME_RE)
|
14 | from doctools.util import log
|
15 |
|
16 | _ = log
|
17 |
|
18 | try:
|
19 | from cStringIO import StringIO
|
20 | except ImportError:
|
21 | # for python3
|
22 | from io import StringIO # type: ignore
|
23 | import sys
|
24 |
|
25 | if sys.version_info.major == 2:
|
26 | from typing import List, Tuple, Iterator, Optional
|
27 |
|
28 |
|
29 | def _Tokens(s, left_pos, right_pos):
|
30 | # type: (str, int, int) -> Iterator[Tuple[h8_id_t, int]]
|
31 | """
|
32 | Args:
|
33 | s: string to parse
|
34 | left_pos, right_pos: Optional span boundaries.
|
35 | """
|
36 | lx = Lexer(s, left_pos, right_pos)
|
37 | while True:
|
38 | tok_id, pos = lx.Read()
|
39 | yield tok_id, pos
|
40 | if tok_id == h8_id.EndOfStream:
|
41 | break
|
42 |
|
43 |
|
44 | def ValidTokens(s, left_pos=0, right_pos=-1):
|
45 | # type: (str, int, int) -> Iterator[Tuple[h8_id_t, int]]
|
46 | """Wrapper around _Tokens to prevent callers from having to handle Invalid.
|
47 |
|
48 | I'm not combining the two functions because I might want to do a
|
49 | 'yield' transformation on Tokens()? Exceptions might complicate the
|
50 | issue?
|
51 | """
|
52 | pos = left_pos
|
53 | for tok_id, end_pos in _Tokens(s, left_pos, right_pos):
|
54 | if tok_id == h8_id.Invalid:
|
55 | raise LexError('ValidTokens() got invalid token', s, pos)
|
56 | yield tok_id, end_pos
|
57 | pos = end_pos
|
58 |
|
59 |
|
60 | def ReadUntilStartTag(it, tag_lexer, tag_name):
|
61 | # type: (Iterator[Tuple[h8_id_t, int]], TagLexer, str) -> Tuple[int, int]
|
62 | """Find the next <foo>, returning its (start, end) positions
|
63 |
|
64 | Raise ParseError if it's not found.
|
65 |
|
66 | tag_lexer is RESET.
|
67 | """
|
68 | pos = 0
|
69 | while True:
|
70 | try:
|
71 | tok_id, end_pos = next(it)
|
72 | except StopIteration:
|
73 | break
|
74 | tag_lexer.Reset(pos, end_pos)
|
75 | if tok_id == h8_id.StartTag and tag_lexer.GetTagName() == tag_name:
|
76 | return pos, end_pos
|
77 |
|
78 | pos = end_pos
|
79 |
|
80 | raise ParseError('No start tag %r' % tag_name)
|
81 |
|
82 |
|
83 | def ReadUntilEndTag(it, tag_lexer, tag_name):
|
84 | # type: (Iterator[Tuple[h8_id_t, int]], TagLexer, str) -> Tuple[int, int]
|
85 | """Find the next </foo>, returning its (start, end) position
|
86 |
|
87 | Raise ParseError if it's not found.
|
88 |
|
89 | tag_lexer is RESET.
|
90 | """
|
91 | pos = 0
|
92 | while True:
|
93 | try:
|
94 | tok_id, end_pos = next(it)
|
95 | except StopIteration:
|
96 | break
|
97 | tag_lexer.Reset(pos, end_pos)
|
98 | if tok_id == h8_id.EndTag and tag_lexer.GetTagName() == tag_name:
|
99 | return pos, end_pos
|
100 |
|
101 | pos = end_pos
|
102 |
|
103 | raise ParseError('No end tag %r' % tag_name)
|
104 |
|
105 |
|
106 | CHAR_ENTITY = {
|
107 | 'amp': '&',
|
108 | 'lt': '<',
|
109 | 'gt': '>',
|
110 | 'quot': '"',
|
111 | 'apos': "'",
|
112 | }
|
113 |
|
114 |
|
115 | def ToText(s, left_pos=0, right_pos=-1):
|
116 | # type: (str, int, int) -> str
|
117 | """Given HTML, return text by unquoting > and < etc.
|
118 |
|
119 | Used by:
|
120 | doctools/oils_doc.py: PygmentsPlugin
|
121 | doctools/help_gen.py: HelpIndexCards
|
122 |
|
123 | In the latter case, we cold process some tags, like:
|
124 |
|
125 | - Blue Link (not clickable, but still useful)
|
126 | - Red X
|
127 |
|
128 | That should be html.ToAnsi.
|
129 | """
|
130 | f = StringIO()
|
131 | out = Output(s, f, left_pos, right_pos)
|
132 |
|
133 | pos = left_pos
|
134 | for tok_id, end_pos in ValidTokens(s, left_pos, right_pos):
|
135 | if tok_id in (h8_id.RawData, h8_id.BadAmpersand, h8_id.BadGreaterThan,
|
136 | h8_id.BadLessThan):
|
137 | out.SkipTo(pos)
|
138 | out.PrintUntil(end_pos)
|
139 |
|
140 | elif tok_id == h8_id.CharEntity: # &
|
141 |
|
142 | entity = s[pos + 1:end_pos - 1]
|
143 |
|
144 | out.SkipTo(pos)
|
145 | out.Print(CHAR_ENTITY[entity])
|
146 | out.SkipTo(end_pos)
|
147 |
|
148 | # Not handling these yet
|
149 | elif tok_id == h8_id.HexChar:
|
150 | raise AssertionError('Hex Char %r' % s[pos:pos + 20])
|
151 |
|
152 | elif tok_id == h8_id.DecChar:
|
153 | raise AssertionError('Dec Char %r' % s[pos:pos + 20])
|
154 |
|
155 | else:
|
156 | # Skip everything else
|
157 | out.SkipTo(end_pos)
|
158 |
|
159 | pos = end_pos
|
160 |
|
161 | out.PrintTheRest()
|
162 | return f.getvalue()
|
163 |
|
164 |
|
165 | #
|
166 | # OLD TagLexer API - REMOVE THIS
|
167 | #
|
168 | # HTML 5 doesn't restrict tag names at all
|
169 | # https://html.spec.whatwg.org/#toc-syntax
|
170 | #
|
171 | # XML allows : - .
|
172 | # https://www.w3.org/TR/xml/#NT-NameChar
|
173 |
|
174 | # Namespaces for MathML, SVG
|
175 | # XLink, XML, XMLNS
|
176 | #
|
177 | # https://infra.spec.whatwg.org/#namespaces
|
178 | #
|
179 | # Allow - for td-attrs
|
180 |
|
181 | # Similar to _UNQUOTED_VALUE in data_lang/htm8.py
|
182 | _UNQUOTED_VALUE_OLD = r'''[^ \t\r\n<>&"'\x00]*'''
|
183 |
|
184 | _TAG_RE = re.compile(r'/? \s* (%s)' % _NAME_RE, re.VERBOSE)
|
185 |
|
186 | _TAG_LAST_RE = re.compile(r'\s* /? >', re.VERBOSE)
|
187 |
|
188 | # To match href="foo"
|
189 |
|
190 | # <button disabled> is standard usage
|
191 |
|
192 | # NOTE: This used to allow whitespace around =
|
193 | # <a foo = "bar"> makes sense in XML
|
194 | # But then you also have
|
195 | # <a foo= bar> - which is TWO attributes, in HTML5
|
196 | # So the space is problematic
|
197 |
|
198 | _ATTR_RE = re.compile(
|
199 | r'''
|
200 | \s+ # Leading whitespace is required
|
201 | (%s) # Attribute name
|
202 | (?: # Optional attribute value
|
203 | \s* = \s* # Spaces allowed around =
|
204 | (?:
|
205 | " ([^>"\x00]*) " # double quoted value
|
206 | | ' ([^>'\x00]*) ' # single quoted value
|
207 | | (%s) # Attribute value
|
208 | )
|
209 | )?
|
210 | ''' % (_NAME_RE, _UNQUOTED_VALUE_OLD), re.VERBOSE)
|
211 |
|
212 |
|
213 | class TagLexer(object):
|
214 | """
|
215 | Given a tag like <a href="..."> or <link type="..." />, the TagLexer
|
216 | provides a few operations:
|
217 |
|
218 | - What is the tag?
|
219 | - Iterate through the attributes, giving (name, value_start_pos, value_end_pos)
|
220 | """
|
221 |
|
222 | def __init__(self, s):
|
223 | # type: (str) -> None
|
224 | self.s = s
|
225 | self.start_pos = -1 # Invalid
|
226 | self.end_pos = -1
|
227 |
|
228 | def Reset(self, start_pos, end_pos):
|
229 | # type: (int, int) -> None
|
230 | """Reuse instances of this object."""
|
231 | assert start_pos >= 0, start_pos
|
232 | assert end_pos >= 0, end_pos
|
233 |
|
234 | self.start_pos = start_pos
|
235 | self.end_pos = end_pos
|
236 |
|
237 | def WholeTagString(self):
|
238 | # type: () -> str
|
239 | """Return the entire tag string, e.g. <a href='foo'>"""
|
240 | return self.s[self.start_pos:self.end_pos]
|
241 |
|
242 | def GetTagName(self):
|
243 | # type: () -> str
|
244 | # First event
|
245 | tok_id, start, end = next(self.Tokens())
|
246 | return self.s[start:end]
|
247 |
|
248 | def GetSpanForAttrValue(self, attr_name):
|
249 | # type: (str) -> Tuple[int, int]
|
250 | """
|
251 | Used by oils_doc.py, for href shortcuts
|
252 | """
|
253 | # Algorithm: search for QuotedValue or UnquotedValue after AttrName
|
254 | # TODO: Could also cache these
|
255 |
|
256 | events = self.Tokens()
|
257 | val = (-1, -1)
|
258 | try:
|
259 | while True:
|
260 | tok_id, start, end = next(events)
|
261 | if tok_id == h8_tag_id.AttrName:
|
262 | name = self.s[start:end]
|
263 | if name == attr_name:
|
264 | # The value should come next
|
265 | tok_id, start, end = next(events)
|
266 | assert tok_id in (
|
267 | h8_tag_id.QuotedValue, h8_tag_id.UnquotedValue,
|
268 | h8_tag_id.MissingValue), h8_tag_id_str(tok_id)
|
269 | val = start, end
|
270 | break
|
271 |
|
272 | except StopIteration:
|
273 | pass
|
274 | return val
|
275 |
|
276 | def GetAttrRaw(self, attr_name):
|
277 | # type: (str) -> Optional[str]
|
278 | """
|
279 | Return the value, which may be UNESCAPED.
|
280 | """
|
281 | start, end = self.GetSpanForAttrValue(attr_name)
|
282 | if start == -1:
|
283 | return None
|
284 | return self.s[start:end]
|
285 |
|
286 | def AllAttrsRawSlice(self):
|
287 | # type: () -> List[Tuple[str, int, int]]
|
288 | """
|
289 | Get a list of pairs [('class', 3, 5), ('href', 9, 12)]
|
290 | """
|
291 | slices = []
|
292 | events = self.Tokens()
|
293 | try:
|
294 | while True:
|
295 | tok_id, start, end = next(events)
|
296 | if tok_id == h8_tag_id.AttrName:
|
297 | name = self.s[start:end]
|
298 |
|
299 | # The value should come next
|
300 | tok_id, start, end = next(events)
|
301 | assert tok_id in (
|
302 | h8_tag_id.QuotedValue, h8_tag_id.UnquotedValue,
|
303 | h8_tag_id.MissingValue), h8_tag_id_str(tok_id)
|
304 | # Note: quoted values may have &
|
305 | # We would need ANOTHER lexer to unescape them, but we
|
306 | # don't need that for ul-table
|
307 | slices.append((name, start, end))
|
308 | except StopIteration:
|
309 | pass
|
310 | return slices
|
311 |
|
312 | def AllAttrsRaw(self):
|
313 | # type: () -> List[Tuple[str, str]]
|
314 | """
|
315 | Get a list of pairs [('class', 'foo'), ('href', '?foo=1&bar=2')]
|
316 |
|
317 | The quoted values may be escaped. We would need another lexer to
|
318 | unescape them.
|
319 | """
|
320 | slices = self.AllAttrsRawSlice()
|
321 | pairs = []
|
322 | for name, start, end in slices:
|
323 | pairs.append((name, self.s[start:end]))
|
324 | return pairs
|
325 |
|
326 | def Tokens(self):
|
327 | # type: () -> Iterator[Tuple[h8_tag_id_t, int, int]]
|
328 | """
|
329 | Yields a sequence of tokens: Tag (AttrName AttrValue?)*
|
330 |
|
331 | Where each Token is (Type, start_pos, end_pos)
|
332 |
|
333 | Note that start and end are NOT redundant! We skip over some unwanted
|
334 | characters.
|
335 | """
|
336 | m = _TAG_RE.match(self.s, self.start_pos + 1)
|
337 | if not m:
|
338 | raise RuntimeError("Couldn't find HTML tag in %r" %
|
339 | self.WholeTagString())
|
340 | yield h8_tag_id.TagName, m.start(1), m.end(1)
|
341 |
|
342 | pos = m.end(0)
|
343 | #log('POS %d', pos)
|
344 |
|
345 | while True:
|
346 | # don't search past the end
|
347 | m = _ATTR_RE.match(self.s, pos, self.end_pos)
|
348 | if not m:
|
349 | #log('BREAK pos %d', pos)
|
350 | break
|
351 | #log('AttrName %r', m.group(1))
|
352 |
|
353 | yield h8_tag_id.AttrName, m.start(1), m.end(1)
|
354 |
|
355 | #log('m.groups() %r', m.groups())
|
356 | if m.group(2) is not None:
|
357 | # double quoted
|
358 | yield h8_tag_id.QuotedValue, m.start(2), m.end(2)
|
359 | elif m.group(3) is not None:
|
360 | # single quoted - TODO: could have different token types
|
361 | yield h8_tag_id.QuotedValue, m.start(3), m.end(3)
|
362 | elif m.group(4) is not None:
|
363 | yield h8_tag_id.UnquotedValue, m.start(4), m.end(4)
|
364 | else:
|
365 | # <button disabled>
|
366 | end = m.end(0)
|
367 | yield h8_tag_id.MissingValue, end, end
|
368 |
|
369 | # Skip past the "
|
370 | pos = m.end(0)
|
371 |
|
372 | #log('TOK %r', self.s)
|
373 |
|
374 | m = _TAG_LAST_RE.match(self.s, pos)
|
375 | #log('_TAG_LAST_RE match %r', self.s[pos:])
|
376 | if not m:
|
377 | raise LexError('Extra data at end of tag', self.s, pos)
|