| 1 | #!/usr/bin/env python2
|
| 2 | """
|
| 3 | lazylex/html.py - Wrapper around HTM8
|
| 4 |
|
| 5 | See doc/lazylex.md for details.
|
| 6 |
|
| 7 | """
|
| 8 | from __future__ import print_function
|
| 9 |
|
| 10 | from _devbuild.gen.htm8_asdl import (h8_id, h8_id_t, h8_id_str)
|
| 11 | from data_lang import htm8
|
| 12 | from data_lang.htm8 import (Lexer, TagLexer, AttrValueLexer, LexError,
|
| 13 | ParseError, Output)
|
| 14 | from doctools.util import log
|
| 15 |
|
| 16 | try:
|
| 17 | from cStringIO import StringIO
|
| 18 | except ImportError:
|
| 19 | # for python3
|
| 20 | from io import StringIO # type: ignore
|
| 21 | import sys
|
| 22 |
|
| 23 | if sys.version_info.major == 2:
|
| 24 | from typing import List, Tuple, Iterator
|
| 25 |
|
| 26 |
|
| 27 | def _Tokens(s, left_pos, right_pos):
|
| 28 | # type: (str, int, int) -> Iterator[Tuple[h8_id_t, int]]
|
| 29 | """
|
| 30 | Args:
|
| 31 | s: string to parse
|
| 32 | left_pos, right_pos: Optional span boundaries.
|
| 33 | """
|
| 34 | lx = Lexer(s, left_pos, right_pos)
|
| 35 | while True:
|
| 36 | tok_id, pos = lx.Read()
|
| 37 | yield tok_id, pos
|
| 38 | if tok_id == h8_id.EndOfStream:
|
| 39 | break
|
| 40 |
|
| 41 |
|
| 42 | def ValidTokens(s, left_pos=0, right_pos=-1):
|
| 43 | # type: (str, int, int) -> Iterator[Tuple[h8_id_t, int]]
|
| 44 | """Wrapper around _Tokens to prevent callers from having to handle Invalid.
|
| 45 |
|
| 46 | I'm not combining the two functions because I might want to do a
|
| 47 | 'yield' transformation on Tokens()? Exceptions might complicate the
|
| 48 | issue?
|
| 49 | """
|
| 50 | pos = left_pos
|
| 51 | for tok_id, end_pos in _Tokens(s, left_pos, right_pos):
|
| 52 | if tok_id == h8_id.Invalid:
|
| 53 | raise LexError('ValidTokens() got invalid token', s, pos)
|
| 54 | yield tok_id, end_pos
|
| 55 | pos = end_pos
|
| 56 |
|
| 57 |
|
| 58 | def ReadUntilStartTag(it, tag_lexer, tag_name):
|
| 59 | # type: (Iterator[Tuple[h8_id_t, int]], TagLexer, str) -> Tuple[int, int]
|
| 60 | """Find the next <foo>, returning its (start, end) positions
|
| 61 |
|
| 62 | Raise ParseError if it's not found.
|
| 63 |
|
| 64 | tag_lexer is RESET.
|
| 65 | """
|
| 66 | pos = 0
|
| 67 | while True:
|
| 68 | try:
|
| 69 | tok_id, end_pos = next(it)
|
| 70 | except StopIteration:
|
| 71 | break
|
| 72 | tag_lexer.Reset(pos, end_pos)
|
| 73 | if tok_id == h8_id.StartTag and tag_lexer.GetTagName() == tag_name:
|
| 74 | return pos, end_pos
|
| 75 |
|
| 76 | pos = end_pos
|
| 77 |
|
| 78 | raise ParseError('No start tag %r' % tag_name)
|
| 79 |
|
| 80 |
|
| 81 | def ReadUntilEndTag(it, tag_lexer, tag_name):
|
| 82 | # type: (Iterator[Tuple[h8_id_t, int]], TagLexer, str) -> Tuple[int, int]
|
| 83 | """Find the next </foo>, returning its (start, end) position
|
| 84 |
|
| 85 | Raise ParseError if it's not found.
|
| 86 |
|
| 87 | tag_lexer is RESET.
|
| 88 | """
|
| 89 | pos = 0
|
| 90 | while True:
|
| 91 | try:
|
| 92 | tok_id, end_pos = next(it)
|
| 93 | except StopIteration:
|
| 94 | break
|
| 95 | tag_lexer.Reset(pos, end_pos)
|
| 96 | if tok_id == h8_id.EndTag and tag_lexer.GetTagName() == tag_name:
|
| 97 | return pos, end_pos
|
| 98 |
|
| 99 | pos = end_pos
|
| 100 |
|
| 101 | raise ParseError('No end tag %r' % tag_name)
|
| 102 |
|
| 103 |
|
| 104 | CHAR_ENTITY = {
|
| 105 | 'amp': '&',
|
| 106 | 'lt': '<',
|
| 107 | 'gt': '>',
|
| 108 | 'quot': '"',
|
| 109 | 'apos': "'",
|
| 110 | }
|
| 111 |
|
| 112 |
|
| 113 | def ToText(s, left_pos=0, right_pos=-1):
|
| 114 | # type: (str, int, int) -> str
|
| 115 | """Given HTML, return text by unquoting > and < etc.
|
| 116 |
|
| 117 | Used by:
|
| 118 | doctools/oils_doc.py: PygmentsPlugin
|
| 119 | doctools/help_gen.py: HelpIndexCards
|
| 120 |
|
| 121 | In the latter case, we cold process some tags, like:
|
| 122 |
|
| 123 | - Blue Link (not clickable, but still useful)
|
| 124 | - Red X
|
| 125 |
|
| 126 | That should be html.ToAnsi.
|
| 127 | """
|
| 128 | f = StringIO()
|
| 129 | out = Output(s, f, left_pos, right_pos)
|
| 130 |
|
| 131 | pos = left_pos
|
| 132 | for tok_id, end_pos in ValidTokens(s, left_pos, right_pos):
|
| 133 | if tok_id in (h8_id.RawData, h8_id.BadAmpersand, h8_id.BadGreaterThan,
|
| 134 | h8_id.BadLessThan):
|
| 135 | out.SkipTo(pos)
|
| 136 | out.PrintUntil(end_pos)
|
| 137 |
|
| 138 | elif tok_id == h8_id.CharEntity: # &
|
| 139 |
|
| 140 | entity = s[pos + 1:end_pos - 1]
|
| 141 |
|
| 142 | out.SkipTo(pos)
|
| 143 | out.Print(CHAR_ENTITY[entity])
|
| 144 | out.SkipTo(end_pos)
|
| 145 |
|
| 146 | # Not handling these yet
|
| 147 | elif tok_id == h8_id.HexChar:
|
| 148 | raise AssertionError('Hex Char %r' % s[pos:pos + 20])
|
| 149 |
|
| 150 | elif tok_id == h8_id.DecChar:
|
| 151 | raise AssertionError('Dec Char %r' % s[pos:pos + 20])
|
| 152 |
|
| 153 | else:
|
| 154 | # Skip everything else
|
| 155 | out.SkipTo(end_pos)
|
| 156 |
|
| 157 | pos = end_pos
|
| 158 |
|
| 159 | out.PrintTheRest()
|
| 160 | return f.getvalue()
|
| 161 |
|
| 162 |
|
| 163 | # https://developer.mozilla.org/en-US/docs/Glossary/Void_element
|
| 164 | VOID_ELEMENTS = [
|
| 165 | 'area',
|
| 166 | 'base',
|
| 167 | 'br',
|
| 168 | 'col',
|
| 169 | 'embed',
|
| 170 | 'hr',
|
| 171 | 'img',
|
| 172 | 'input',
|
| 173 | 'link',
|
| 174 | 'meta',
|
| 175 | 'param',
|
| 176 | 'source',
|
| 177 | 'track',
|
| 178 | 'wbr',
|
| 179 | ]
|
| 180 |
|
| 181 | LEX_ATTRS = 1 << 1
|
| 182 | LEX_QUOTED_VALUES = 1 << 2 # href="?x=42&y=99"
|
| 183 | NO_SPECIAL_TAGS = 1 << 3 # <script> <style>, VOID tags, etc.
|
| 184 | BALANCED_TAGS = 1 << 4 # are tags balanced?
|
| 185 |
|
| 186 |
|
| 187 | def ValidateOld(contents, flags, counters):
|
| 188 | # type: (str, int, Counters) -> None
|
| 189 |
|
| 190 | tag_lexer = TagLexer(contents)
|
| 191 | val_lexer = AttrValueLexer(contents)
|
| 192 |
|
| 193 | no_special_tags = bool(flags & NO_SPECIAL_TAGS)
|
| 194 | lx = Lexer(contents, no_special_tags=no_special_tags)
|
| 195 | tokens = []
|
| 196 | start_pos = 0
|
| 197 | tag_stack = []
|
| 198 | while True:
|
| 199 | tok_id, end_pos = lx.Read()
|
| 200 | #log('TOP %s %r', h8_id_str(tok_id), contents[start_pos:end_pos])
|
| 201 |
|
| 202 | if tok_id == h8_id.Invalid:
|
| 203 | raise LexError('ValidateOld() got invalid token', contents,
|
| 204 | start_pos)
|
| 205 | if tok_id == h8_id.EndOfStream:
|
| 206 | break
|
| 207 |
|
| 208 | tokens.append((tok_id, end_pos))
|
| 209 |
|
| 210 | if tok_id == h8_id.StartEndTag:
|
| 211 | counters.num_start_end_tags += 1
|
| 212 |
|
| 213 | tag_lexer.Reset(start_pos, end_pos)
|
| 214 | all_attrs = tag_lexer.AllAttrsRawSlice()
|
| 215 | counters.num_attrs += len(all_attrs)
|
| 216 | for name, val_start, val_end in all_attrs:
|
| 217 | val_lexer.Reset(val_start, val_end)
|
| 218 | counters.num_val_tokens += val_lexer.NumTokens()
|
| 219 |
|
| 220 | #counters.debug_attrs.extend(all_attrs)
|
| 221 |
|
| 222 | elif tok_id == h8_id.StartTag:
|
| 223 | counters.num_start_tags += 1
|
| 224 |
|
| 225 | tag_lexer.Reset(start_pos, end_pos)
|
| 226 | all_attrs = tag_lexer.AllAttrsRawSlice()
|
| 227 | counters.num_attrs += len(all_attrs)
|
| 228 | for name, val_start, val_end in all_attrs:
|
| 229 | val_lexer.Reset(val_start, val_end)
|
| 230 | counters.num_val_tokens += val_lexer.NumTokens()
|
| 231 |
|
| 232 | #counters.debug_attrs.extend(all_attrs)
|
| 233 |
|
| 234 | if flags & BALANCED_TAGS:
|
| 235 | tag_name = lx.CanonicalTagName()
|
| 236 | if flags & NO_SPECIAL_TAGS:
|
| 237 | tag_stack.append(tag_name)
|
| 238 | else:
|
| 239 | # e.g. <meta> is considered self-closing, like <meta/>
|
| 240 | if tag_name not in VOID_ELEMENTS:
|
| 241 | tag_stack.append(tag_name)
|
| 242 |
|
| 243 | counters.max_tag_stack = max(counters.max_tag_stack,
|
| 244 | len(tag_stack))
|
| 245 | elif tok_id == h8_id.EndTag:
|
| 246 | if flags & BALANCED_TAGS:
|
| 247 | try:
|
| 248 | expected = tag_stack.pop()
|
| 249 | except IndexError:
|
| 250 | raise ParseError('Tag stack empty',
|
| 251 | s=contents,
|
| 252 | start_pos=start_pos)
|
| 253 |
|
| 254 | actual = lx.CanonicalTagName()
|
| 255 | if expected != actual:
|
| 256 | raise ParseError(
|
| 257 | 'Got unexpected closing tag %r; opening tag was %r' %
|
| 258 | (contents[start_pos:end_pos], expected),
|
| 259 | s=contents,
|
| 260 | start_pos=start_pos)
|
| 261 |
|
| 262 | start_pos = end_pos
|
| 263 |
|
| 264 | if len(tag_stack) != 0:
|
| 265 | raise ParseError('Missing closing tags at end of doc: %s' %
|
| 266 | ' '.join(tag_stack),
|
| 267 | s=contents,
|
| 268 | start_pos=start_pos)
|
| 269 |
|
| 270 | counters.num_tokens += len(tokens)
|
| 271 |
|
| 272 |
|
| 273 | def Validate(contents, flags, counters):
|
| 274 | # type: (str, int, Counters) -> None
|
| 275 |
|
| 276 | attr_lx = htm8.AttrLexer(contents)
|
| 277 |
|
| 278 | no_special_tags = bool(flags & NO_SPECIAL_TAGS)
|
| 279 | lx = htm8.Lexer(contents, no_special_tags=no_special_tags)
|
| 280 | tokens = []
|
| 281 | start_pos = 0
|
| 282 | tag_stack = []
|
| 283 | while True:
|
| 284 | tok_id, end_pos = lx.Read()
|
| 285 | #log('TOP %s %r', h8_id_str(tok_id), contents[start_pos:end_pos])
|
| 286 |
|
| 287 | if tok_id == h8_id.Invalid:
|
| 288 | raise LexError('Validate() got invalid token', contents, start_pos)
|
| 289 | if tok_id == h8_id.EndOfStream:
|
| 290 | break
|
| 291 |
|
| 292 | tokens.append((tok_id, end_pos))
|
| 293 |
|
| 294 | if tok_id == h8_id.StartEndTag:
|
| 295 | counters.num_start_end_tags += 1
|
| 296 |
|
| 297 | attr_lx.Init(lx.TagNamePos(), end_pos)
|
| 298 | all_attrs = htm8.AllAttrsRaw(attr_lx)
|
| 299 | counters.num_attrs += len(all_attrs)
|
| 300 | # TODO: val_lexer.NumTokens() can be replaced with tokens_out
|
| 301 |
|
| 302 | elif tok_id == h8_id.StartTag:
|
| 303 | counters.num_start_tags += 1
|
| 304 |
|
| 305 | attr_lx.Init(lx.TagNamePos(), end_pos)
|
| 306 | all_attrs = htm8.AllAttrsRaw(attr_lx)
|
| 307 | counters.num_attrs += len(all_attrs)
|
| 308 |
|
| 309 | #counters.debug_attrs.extend(all_attrs)
|
| 310 |
|
| 311 | if flags & BALANCED_TAGS:
|
| 312 | tag_name = lx.CanonicalTagName()
|
| 313 | if flags & NO_SPECIAL_TAGS:
|
| 314 | tag_stack.append(tag_name)
|
| 315 | else:
|
| 316 | # e.g. <meta> is considered self-closing, like <meta/>
|
| 317 | if tag_name not in VOID_ELEMENTS:
|
| 318 | tag_stack.append(tag_name)
|
| 319 |
|
| 320 | counters.max_tag_stack = max(counters.max_tag_stack,
|
| 321 | len(tag_stack))
|
| 322 | elif tok_id == h8_id.EndTag:
|
| 323 | if flags & BALANCED_TAGS:
|
| 324 | try:
|
| 325 | expected = tag_stack.pop()
|
| 326 | except IndexError:
|
| 327 | raise ParseError('Tag stack empty',
|
| 328 | s=contents,
|
| 329 | start_pos=start_pos)
|
| 330 |
|
| 331 | actual = lx.CanonicalTagName()
|
| 332 | if expected != actual:
|
| 333 | raise ParseError(
|
| 334 | 'Got unexpected closing tag %r; opening tag was %r' %
|
| 335 | (contents[start_pos:end_pos], expected),
|
| 336 | s=contents,
|
| 337 | start_pos=start_pos)
|
| 338 |
|
| 339 | start_pos = end_pos
|
| 340 |
|
| 341 | if len(tag_stack) != 0:
|
| 342 | raise ParseError('Missing closing tags at end of doc: %s' %
|
| 343 | ' '.join(tag_stack),
|
| 344 | s=contents,
|
| 345 | start_pos=start_pos)
|
| 346 |
|
| 347 | counters.num_tokens += len(tokens)
|
| 348 |
|
| 349 |
|
| 350 | def ToXml(htm8_str):
|
| 351 | # type: (str) -> str
|
| 352 |
|
| 353 | # TODO:
|
| 354 | # 1. Lex it
|
| 355 | # 2. < & > must be escaped
|
| 356 | # a. in raw data
|
| 357 | # b. in quoted strings
|
| 358 | # 3. <script> turned into CDATA
|
| 359 | # 4. void tags turned into self-closing tags
|
| 360 | # 5. case-sensitive tag matching - not sure about this
|
| 361 |
|
| 362 | tag_lexer = TagLexer(htm8_str)
|
| 363 | val_lexer = AttrValueLexer(htm8_str)
|
| 364 |
|
| 365 | f = StringIO()
|
| 366 | out = Output(htm8_str, f)
|
| 367 |
|
| 368 | lx = Lexer(htm8_str)
|
| 369 |
|
| 370 | pos = 0
|
| 371 | while True:
|
| 372 | tok_id, end_pos = lx.Read()
|
| 373 |
|
| 374 | if tok_id == h8_id.Invalid:
|
| 375 | raise LexError('ToXml() got invalid token', htm8_str, pos)
|
| 376 | if tok_id == h8_id.EndOfStream:
|
| 377 | break
|
| 378 |
|
| 379 | if tok_id in (h8_id.RawData, h8_id.CharEntity, h8_id.HexChar,
|
| 380 | h8_id.DecChar):
|
| 381 | out.PrintUntil(end_pos)
|
| 382 | elif tok_id in (h8_id.StartTag, h8_id.StartEndTag):
|
| 383 | tag_lexer.Reset(pos, end_pos)
|
| 384 | # TODO: reduce allocations here
|
| 385 | all_attrs = tag_lexer.AllAttrsRawSlice()
|
| 386 | for name, val_start, val_end in all_attrs:
|
| 387 | val_lexer.Reset(val_start, val_end)
|
| 388 | # TODO: get the kind of string
|
| 389 | #
|
| 390 | # Quoted: we need to replace & with & and < with <
|
| 391 | # note > is not allowed
|
| 392 | # Unquoted: right now, we can just surround with double quotes
|
| 393 | # because we don't allow any bad chars
|
| 394 | # Empty : add "", so empty= becomes =""
|
| 395 | # Missing : add ="", so missing becomes missing=""
|
| 396 |
|
| 397 | tag_name = lx.CanonicalTagName()
|
| 398 | if tok_id == h8_id.StartTag and tag_name in VOID_ELEMENTS:
|
| 399 | # TODO: instead of closing >, print />
|
| 400 | pass
|
| 401 |
|
| 402 | elif tok_id == h8_id.BadAmpersand:
|
| 403 | #out.SkipTo(pos)
|
| 404 | out.Print('&')
|
| 405 | out.SkipTo(end_pos)
|
| 406 |
|
| 407 | elif tok_id == h8_id.BadGreaterThan:
|
| 408 | #out.SkipTo(pos)
|
| 409 | out.Print('>')
|
| 410 | out.SkipTo(end_pos)
|
| 411 | else:
|
| 412 | out.PrintUntil(end_pos)
|
| 413 |
|
| 414 | pos = end_pos
|
| 415 |
|
| 416 | out.PrintTheRest()
|
| 417 | return f.getvalue()
|
| 418 |
|
| 419 |
|
| 420 | class Counters(object):
|
| 421 |
|
| 422 | def __init__(self):
|
| 423 | # type: () -> None
|
| 424 | self.num_tokens = 0
|
| 425 | self.num_start_tags = 0
|
| 426 | self.num_start_end_tags = 0
|
| 427 | self.num_attrs = 0
|
| 428 | self.max_tag_stack = 0
|
| 429 | self.num_val_tokens = 0
|
| 430 |
|
| 431 | #self.debug_attrs = []
|
| 432 |
|
| 433 |
|
| 434 | def main(argv):
|
| 435 | # type: (List[str]) -> int
|
| 436 | action = argv[1]
|
| 437 |
|
| 438 | if action == 'tokens':
|
| 439 | contents = sys.stdin.read()
|
| 440 |
|
| 441 | lx = Lexer(contents)
|
| 442 | start_pos = 0
|
| 443 | while True:
|
| 444 | tok_id, end_pos = lx.Read()
|
| 445 | if tok_id == h8_id.Invalid:
|
| 446 | raise LexError('Invalid token', contents, start_pos)
|
| 447 | if tok_id == h8_id.EndOfStream:
|
| 448 | break
|
| 449 |
|
| 450 | frag = contents[start_pos:end_pos]
|
| 451 | log('%d %s %r', end_pos, h8_id_str(tok_id), frag)
|
| 452 | start_pos = end_pos
|
| 453 |
|
| 454 | return 0
|
| 455 |
|
| 456 | elif action in ('lex-htm8', 'parse-htm8', 'parse-xml'):
|
| 457 |
|
| 458 | errors = []
|
| 459 | counters = Counters()
|
| 460 |
|
| 461 | flags = LEX_ATTRS | LEX_QUOTED_VALUES
|
| 462 | if action.startswith('parse-'):
|
| 463 | flags |= BALANCED_TAGS
|
| 464 | if action == 'parse-xml':
|
| 465 | flags |= NO_SPECIAL_TAGS
|
| 466 |
|
| 467 | i = 0
|
| 468 | for line in sys.stdin:
|
| 469 | filename = line.strip()
|
| 470 | with open(filename) as f:
|
| 471 | contents = f.read()
|
| 472 |
|
| 473 | try:
|
| 474 | Validate(contents, flags, counters)
|
| 475 | except LexError as e:
|
| 476 | log('Lex error in %r: %s', filename, e)
|
| 477 | errors.append((filename, e))
|
| 478 | except ParseError as e:
|
| 479 | log('Parse error in %r: %s', filename, e)
|
| 480 | errors.append((filename, e))
|
| 481 | i += 1
|
| 482 |
|
| 483 | log('')
|
| 484 | log('%10d tokens', counters.num_tokens)
|
| 485 | log('%10d start/end tags', counters.num_start_end_tags)
|
| 486 | log('%10d start tags', counters.num_start_tags)
|
| 487 | log('%10d attrs', counters.num_attrs)
|
| 488 | log('%10d max tag stack depth', counters.max_tag_stack)
|
| 489 | log('%10d attr val tokens', counters.num_val_tokens)
|
| 490 | log('%10d errors', len(errors))
|
| 491 | if len(errors):
|
| 492 | return 1
|
| 493 | return 0
|
| 494 |
|
| 495 | elif action == 'todo':
|
| 496 | # Other algorithms:
|
| 497 | #
|
| 498 | # - select first subtree with given ID
|
| 499 | # - this requires understanding the void tags I suppose
|
| 500 | # - select all subtrees that have a class
|
| 501 | # - materialize DOM
|
| 502 |
|
| 503 | # Safe-HTM8? This is a filter
|
| 504 | return 0
|
| 505 |
|
| 506 | else:
|
| 507 | raise RuntimeError('Invalid action %r' % action)
|
| 508 |
|
| 509 |
|
| 510 | if __name__ == '__main__':
|
| 511 | sys.exit(main(sys.argv))
|