Coverage for /pythoncovmergedfiles/medio/medio/usr/local/lib/python3.8/site-packages/bleach/_vendor/html5lib/constants.py: 100%

1from __future__ import absolute_import, division, unicode_literals 

2 

3import string 

4 

5EOF = None 

6 

7E = { 

8 "null-character": 

9 "Null character in input stream, replaced with U+FFFD.", 

10 "invalid-codepoint": 

11 "Invalid codepoint in stream.", 

12 "incorrectly-placed-solidus": 

13 "Solidus (/) incorrectly placed in tag.", 

14 "incorrect-cr-newline-entity": 

15 "Incorrect CR newline entity, replaced with LF.", 

16 "illegal-windows-1252-entity": 

17 "Entity used with illegal number (windows-1252 reference).", 

18 "cant-convert-numeric-entity": 

19 "Numeric entity couldn't be converted to character " 

20 "(codepoint U+%(charAsInt)08x).", 

21 "illegal-codepoint-for-numeric-entity": 

22 "Numeric entity represents an illegal codepoint: " 

23 "U+%(charAsInt)08x.", 

24 "numeric-entity-without-semicolon": 

25 "Numeric entity didn't end with ';'.", 

26 "expected-numeric-entity-but-got-eof": 

27 "Numeric entity expected. Got end of file instead.", 

28 "expected-numeric-entity": 

29 "Numeric entity expected but none found.", 

30 "named-entity-without-semicolon": 

31 "Named entity didn't end with ';'.", 

32 "expected-named-entity": 

33 "Named entity expected. Got none.", 

34 "attributes-in-end-tag": 

35 "End tag contains unexpected attributes.", 

36 'self-closing-flag-on-end-tag': 

37 "End tag contains unexpected self-closing flag.", 

38 "expected-tag-name-but-got-right-bracket": 

39 "Expected tag name. Got '>' instead.", 

40 "expected-tag-name-but-got-question-mark": 

41 "Expected tag name. Got '?' instead. (HTML doesn't " 

42 "support processing instructions.)", 

43 "expected-tag-name": 

44 "Expected tag name. Got something else instead", 

45 "expected-closing-tag-but-got-right-bracket": 

46 "Expected closing tag. Got '>' instead. Ignoring '</>'.", 

47 "expected-closing-tag-but-got-eof": 

48 "Expected closing tag. Unexpected end of file.", 

49 "expected-closing-tag-but-got-char": 

50 "Expected closing tag. Unexpected character '%(data)s' found.", 

51 "eof-in-tag-name": 

52 "Unexpected end of file in the tag name.", 

53 "expected-attribute-name-but-got-eof": 

54 "Unexpected end of file. Expected attribute name instead.", 

55 "eof-in-attribute-name": 

56 "Unexpected end of file in attribute name.", 

57 "invalid-character-in-attribute-name": 

58 "Invalid character in attribute name", 

59 "duplicate-attribute": 

60 "Dropped duplicate attribute on tag.", 

61 "expected-end-of-tag-name-but-got-eof": 

62 "Unexpected end of file. Expected = or end of tag.", 

63 "expected-attribute-value-but-got-eof": 

64 "Unexpected end of file. Expected attribute value.", 

65 "expected-attribute-value-but-got-right-bracket": 

66 "Expected attribute value. Got '>' instead.", 

67 'equals-in-unquoted-attribute-value': 

68 "Unexpected = in unquoted attribute", 

69 'unexpected-character-in-unquoted-attribute-value': 

70 "Unexpected character in unquoted attribute", 

71 "invalid-character-after-attribute-name": 

72 "Unexpected character after attribute name.", 

73 "unexpected-character-after-attribute-value": 

74 "Unexpected character after attribute value.", 

75 "eof-in-attribute-value-double-quote": 

76 "Unexpected end of file in attribute value (\").", 

77 "eof-in-attribute-value-single-quote": 

78 "Unexpected end of file in attribute value (').", 

79 "eof-in-attribute-value-no-quotes": 

80 "Unexpected end of file in attribute value.", 

81 "unexpected-EOF-after-solidus-in-tag": 

82 "Unexpected end of file in tag. Expected >", 

83 "unexpected-character-after-solidus-in-tag": 

84 "Unexpected character after / in tag. Expected >", 

85 "expected-dashes-or-doctype": 

86 "Expected '--' or 'DOCTYPE'. Not found.", 

87 "unexpected-bang-after-double-dash-in-comment": 

88 "Unexpected ! after -- in comment", 

89 "unexpected-space-after-double-dash-in-comment": 

90 "Unexpected space after -- in comment", 

91 "incorrect-comment": 

92 "Incorrect comment.", 

93 "eof-in-comment": 

94 "Unexpected end of file in comment.", 

95 "eof-in-comment-end-dash": 

96 "Unexpected end of file in comment (-)", 

97 "unexpected-dash-after-double-dash-in-comment": 

98 "Unexpected '-' after '--' found in comment.", 

99 "eof-in-comment-double-dash": 

100 "Unexpected end of file in comment (--).", 

101 "eof-in-comment-end-space-state": 

102 "Unexpected end of file in comment.", 

103 "eof-in-comment-end-bang-state": 

104 "Unexpected end of file in comment.", 

105 "unexpected-char-in-comment": 

106 "Unexpected character in comment found.", 

107 "need-space-after-doctype": 

108 "No space after literal string 'DOCTYPE'.", 

109 "expected-doctype-name-but-got-right-bracket": 

110 "Unexpected > character. Expected DOCTYPE name.", 

111 "expected-doctype-name-but-got-eof": 

112 "Unexpected end of file. Expected DOCTYPE name.", 

113 "eof-in-doctype-name": 

114 "Unexpected end of file in DOCTYPE name.", 

115 "eof-in-doctype": 

116 "Unexpected end of file in DOCTYPE.", 

117 "expected-space-or-right-bracket-in-doctype": 

118 "Expected space or '>'. Got '%(data)s'", 

119 "unexpected-end-of-doctype": 

120 "Unexpected end of DOCTYPE.", 

121 "unexpected-char-in-doctype": 

122 "Unexpected character in DOCTYPE.", 

123 "eof-in-innerhtml": 

124 "XXX innerHTML EOF", 

125 "unexpected-doctype": 

126 "Unexpected DOCTYPE. Ignored.", 

127 "non-html-root": 

128 "html needs to be the first start tag.", 

129 "expected-doctype-but-got-eof": 

130 "Unexpected End of file. Expected DOCTYPE.", 

131 "unknown-doctype": 

132 "Erroneous DOCTYPE.", 

133 "expected-doctype-but-got-chars": 

134 "Unexpected non-space characters. Expected DOCTYPE.", 

135 "expected-doctype-but-got-start-tag": 

136 "Unexpected start tag (%(name)s). Expected DOCTYPE.", 

137 "expected-doctype-but-got-end-tag": 

138 "Unexpected end tag (%(name)s). Expected DOCTYPE.", 

139 "end-tag-after-implied-root": 

140 "Unexpected end tag (%(name)s) after the (implied) root element.", 

141 "expected-named-closing-tag-but-got-eof": 

142 "Unexpected end of file. Expected end tag (%(name)s).", 

143 "two-heads-are-not-better-than-one": 

144 "Unexpected start tag head in existing head. Ignored.", 

145 "unexpected-end-tag": 

146 "Unexpected end tag (%(name)s). Ignored.", 

147 "unexpected-start-tag-out-of-my-head": 

148 "Unexpected start tag (%(name)s) that can be in head. Moved.", 

149 "unexpected-start-tag": 

150 "Unexpected start tag (%(name)s).", 

151 "missing-end-tag": 

152 "Missing end tag (%(name)s).", 

153 "missing-end-tags": 

154 "Missing end tags (%(name)s).", 

155 "unexpected-start-tag-implies-end-tag": 

156 "Unexpected start tag (%(startName)s) " 

157 "implies end tag (%(endName)s).", 

158 "unexpected-start-tag-treated-as": 

159 "Unexpected start tag (%(originalName)s). Treated as %(newName)s.", 

160 "deprecated-tag": 

161 "Unexpected start tag %(name)s. Don't use it!", 

162 "unexpected-start-tag-ignored": 

163 "Unexpected start tag %(name)s. Ignored.", 

164 "expected-one-end-tag-but-got-another": 

165 "Unexpected end tag (%(gotName)s). " 

166 "Missing end tag (%(expectedName)s).", 

167 "end-tag-too-early": 

168 "End tag (%(name)s) seen too early. Expected other end tag.", 

169 "end-tag-too-early-named": 

170 "Unexpected end tag (%(gotName)s). Expected end tag (%(expectedName)s).", 

171 "end-tag-too-early-ignored": 

172 "End tag (%(name)s) seen too early. Ignored.", 

173 "adoption-agency-1.1": 

174 "End tag (%(name)s) violates step 1, " 

175 "paragraph 1 of the adoption agency algorithm.", 

176 "adoption-agency-1.2": 

177 "End tag (%(name)s) violates step 1, " 

178 "paragraph 2 of the adoption agency algorithm.", 

179 "adoption-agency-1.3": 

180 "End tag (%(name)s) violates step 1, " 

181 "paragraph 3 of the adoption agency algorithm.", 

182 "adoption-agency-4.4": 

183 "End tag (%(name)s) violates step 4, " 

184 "paragraph 4 of the adoption agency algorithm.", 

185 "unexpected-end-tag-treated-as": 

186 "Unexpected end tag (%(originalName)s). Treated as %(newName)s.", 

187 "no-end-tag": 

188 "This element (%(name)s) has no end tag.", 

189 "unexpected-implied-end-tag-in-table": 

190 "Unexpected implied end tag (%(name)s) in the table phase.", 

191 "unexpected-implied-end-tag-in-table-body": 

192 "Unexpected implied end tag (%(name)s) in the table body phase.", 

193 "unexpected-char-implies-table-voodoo": 

194 "Unexpected non-space characters in " 

195 "table context caused voodoo mode.", 

196 "unexpected-hidden-input-in-table": 

197 "Unexpected input with type hidden in table context.", 

198 "unexpected-form-in-table": 

199 "Unexpected form in table context.", 

200 "unexpected-start-tag-implies-table-voodoo": 

201 "Unexpected start tag (%(name)s) in " 

202 "table context caused voodoo mode.", 

203 "unexpected-end-tag-implies-table-voodoo": 

204 "Unexpected end tag (%(name)s) in " 

205 "table context caused voodoo mode.", 

206 "unexpected-cell-in-table-body": 

207 "Unexpected table cell start tag (%(name)s) " 

208 "in the table body phase.", 

209 "unexpected-cell-end-tag": 

210 "Got table cell end tag (%(name)s) " 

211 "while required end tags are missing.", 

212 "unexpected-end-tag-in-table-body": 

213 "Unexpected end tag (%(name)s) in the table body phase. Ignored.", 

214 "unexpected-implied-end-tag-in-table-row": 

215 "Unexpected implied end tag (%(name)s) in the table row phase.", 

216 "unexpected-end-tag-in-table-row": 

217 "Unexpected end tag (%(name)s) in the table row phase. Ignored.", 

218 "unexpected-select-in-select": 

219 "Unexpected select start tag in the select phase " 

220 "treated as select end tag.", 

221 "unexpected-input-in-select": 

222 "Unexpected input start tag in the select phase.", 

223 "unexpected-start-tag-in-select": 

224 "Unexpected start tag token (%(name)s in the select phase. " 

225 "Ignored.", 

226 "unexpected-end-tag-in-select": 

227 "Unexpected end tag (%(name)s) in the select phase. Ignored.", 

228 "unexpected-table-element-start-tag-in-select-in-table": 

229 "Unexpected table element start tag (%(name)s) in the select in table phase.", 

230 "unexpected-table-element-end-tag-in-select-in-table": 

231 "Unexpected table element end tag (%(name)s) in the select in table phase.", 

232 "unexpected-char-after-body": 

233 "Unexpected non-space characters in the after body phase.", 

234 "unexpected-start-tag-after-body": 

235 "Unexpected start tag token (%(name)s)" 

236 " in the after body phase.", 

237 "unexpected-end-tag-after-body": 

238 "Unexpected end tag token (%(name)s)" 

239 " in the after body phase.", 

240 "unexpected-char-in-frameset": 

241 "Unexpected characters in the frameset phase. Characters ignored.", 

242 "unexpected-start-tag-in-frameset": 

243 "Unexpected start tag token (%(name)s)" 

244 " in the frameset phase. Ignored.", 

245 "unexpected-frameset-in-frameset-innerhtml": 

246 "Unexpected end tag token (frameset) " 

247 "in the frameset phase (innerHTML).", 

248 "unexpected-end-tag-in-frameset": 

249 "Unexpected end tag token (%(name)s)" 

250 " in the frameset phase. Ignored.", 

251 "unexpected-char-after-frameset": 

252 "Unexpected non-space characters in the " 

253 "after frameset phase. Ignored.", 

254 "unexpected-start-tag-after-frameset": 

255 "Unexpected start tag (%(name)s)" 

256 " in the after frameset phase. Ignored.", 

257 "unexpected-end-tag-after-frameset": 

258 "Unexpected end tag (%(name)s)" 

259 " in the after frameset phase. Ignored.", 

260 "unexpected-end-tag-after-body-innerhtml": 

261 "Unexpected end tag after body(innerHtml)", 

262 "expected-eof-but-got-char": 

263 "Unexpected non-space characters. Expected end of file.", 

264 "expected-eof-but-got-start-tag": 

265 "Unexpected start tag (%(name)s)" 

266 ". Expected end of file.", 

267 "expected-eof-but-got-end-tag": 

268 "Unexpected end tag (%(name)s)" 

269 ". Expected end of file.", 

270 "eof-in-table": 

271 "Unexpected end of file. Expected table content.", 

272 "eof-in-select": 

273 "Unexpected end of file. Expected select content.", 

274 "eof-in-frameset": 

275 "Unexpected end of file. Expected frameset content.", 

276 "eof-in-script-in-script": 

277 "Unexpected end of file. Expected script content.", 

278 "eof-in-foreign-lands": 

279 "Unexpected end of file. Expected foreign content", 

280 "non-void-element-with-trailing-solidus": 

281 "Trailing solidus not allowed on element %(name)s", 

282 "unexpected-html-element-in-foreign-content": 

283 "Element %(name)s not allowed in a non-html context", 

284 "unexpected-end-tag-before-html": 

285 "Unexpected end tag (%(name)s) before html.", 

286 "unexpected-inhead-noscript-tag": 

287 "Element %(name)s not allowed in a inhead-noscript context", 

288 "eof-in-head-noscript": 

289 "Unexpected end of file. Expected inhead-noscript content", 

290 "char-in-head-noscript": 

291 "Unexpected non-space character. Expected inhead-noscript content", 

292 "XXX-undefined-error": 

293 "Undefined error (this sucks and should be fixed)", 

294} 

295 

296namespaces = { 

297 "html": "http://www.w3.org/1999/xhtml", 

298 "mathml": "http://www.w3.org/1998/Math/MathML", 

299 "svg": "http://www.w3.org/2000/svg", 

300 "xlink": "http://www.w3.org/1999/xlink", 

301 "xml": "http://www.w3.org/XML/1998/namespace", 

302 "xmlns": "http://www.w3.org/2000/xmlns/" 

303} 

304 

305scopingElements = frozenset([ 

306 (namespaces["html"], "applet"), 

307 (namespaces["html"], "caption"), 

308 (namespaces["html"], "html"), 

309 (namespaces["html"], "marquee"), 

310 (namespaces["html"], "object"), 

311 (namespaces["html"], "table"), 

312 (namespaces["html"], "td"), 

313 (namespaces["html"], "th"), 

314 (namespaces["mathml"], "mi"), 

315 (namespaces["mathml"], "mo"), 

316 (namespaces["mathml"], "mn"), 

317 (namespaces["mathml"], "ms"), 

318 (namespaces["mathml"], "mtext"), 

319 (namespaces["mathml"], "annotation-xml"), 

320 (namespaces["svg"], "foreignObject"), 

321 (namespaces["svg"], "desc"), 

322 (namespaces["svg"], "title"), 

323]) 

324 

325formattingElements = frozenset([ 

326 (namespaces["html"], "a"), 

327 (namespaces["html"], "b"), 

328 (namespaces["html"], "big"), 

329 (namespaces["html"], "code"), 

330 (namespaces["html"], "em"), 

331 (namespaces["html"], "font"), 

332 (namespaces["html"], "i"), 

333 (namespaces["html"], "nobr"), 

334 (namespaces["html"], "s"), 

335 (namespaces["html"], "small"), 

336 (namespaces["html"], "strike"), 

337 (namespaces["html"], "strong"), 

338 (namespaces["html"], "tt"), 

339 (namespaces["html"], "u") 

340]) 

341 

342specialElements = frozenset([ 

343 (namespaces["html"], "address"), 

344 (namespaces["html"], "applet"), 

345 (namespaces["html"], "area"), 

346 (namespaces["html"], "article"), 

347 (namespaces["html"], "aside"), 

348 (namespaces["html"], "base"), 

349 (namespaces["html"], "basefont"), 

350 (namespaces["html"], "bgsound"), 

351 (namespaces["html"], "blockquote"), 

352 (namespaces["html"], "body"), 

353 (namespaces["html"], "br"), 

354 (namespaces["html"], "button"), 

355 (namespaces["html"], "caption"), 

356 (namespaces["html"], "center"), 

357 (namespaces["html"], "col"), 

358 (namespaces["html"], "colgroup"), 

359 (namespaces["html"], "command"), 

360 (namespaces["html"], "dd"), 

361 (namespaces["html"], "details"), 

362 (namespaces["html"], "dir"), 

363 (namespaces["html"], "div"), 

364 (namespaces["html"], "dl"), 

365 (namespaces["html"], "dt"), 

366 (namespaces["html"], "embed"), 

367 (namespaces["html"], "fieldset"), 

368 (namespaces["html"], "figure"), 

369 (namespaces["html"], "footer"), 

370 (namespaces["html"], "form"), 

371 (namespaces["html"], "frame"), 

372 (namespaces["html"], "frameset"), 

373 (namespaces["html"], "h1"), 

374 (namespaces["html"], "h2"), 

375 (namespaces["html"], "h3"), 

376 (namespaces["html"], "h4"), 

377 (namespaces["html"], "h5"), 

378 (namespaces["html"], "h6"), 

379 (namespaces["html"], "head"), 

380 (namespaces["html"], "header"), 

381 (namespaces["html"], "hr"), 

382 (namespaces["html"], "html"), 

383 (namespaces["html"], "iframe"), 

384 # Note that image is commented out in the spec as "this isn't an 

385 # element that can end up on the stack, so it doesn't matter," 

386 (namespaces["html"], "image"), 

387 (namespaces["html"], "img"), 

388 (namespaces["html"], "input"), 

389 (namespaces["html"], "isindex"), 

390 (namespaces["html"], "li"), 

391 (namespaces["html"], "link"), 

392 (namespaces["html"], "listing"), 

393 (namespaces["html"], "marquee"), 

394 (namespaces["html"], "menu"), 

395 (namespaces["html"], "meta"), 

396 (namespaces["html"], "nav"), 

397 (namespaces["html"], "noembed"), 

398 (namespaces["html"], "noframes"), 

399 (namespaces["html"], "noscript"), 

400 (namespaces["html"], "object"), 

401 (namespaces["html"], "ol"), 

402 (namespaces["html"], "p"), 

403 (namespaces["html"], "param"), 

404 (namespaces["html"], "plaintext"), 

405 (namespaces["html"], "pre"), 

406 (namespaces["html"], "script"), 

407 (namespaces["html"], "section"), 

408 (namespaces["html"], "select"), 

409 (namespaces["html"], "style"), 

410 (namespaces["html"], "table"), 

411 (namespaces["html"], "tbody"), 

412 (namespaces["html"], "td"), 

413 (namespaces["html"], "textarea"), 

414 (namespaces["html"], "tfoot"), 

415 (namespaces["html"], "th"), 

416 (namespaces["html"], "thead"), 

417 (namespaces["html"], "title"), 

418 (namespaces["html"], "tr"), 

419 (namespaces["html"], "ul"), 

420 (namespaces["html"], "wbr"), 

421 (namespaces["html"], "xmp"), 

422 (namespaces["svg"], "foreignObject") 

423]) 

424 

425htmlIntegrationPointElements = frozenset([ 

426 (namespaces["mathml"], "annotation-xml"), 

427 (namespaces["svg"], "foreignObject"), 

428 (namespaces["svg"], "desc"), 

429 (namespaces["svg"], "title") 

430]) 

431 

432mathmlTextIntegrationPointElements = frozenset([ 

433 (namespaces["mathml"], "mi"), 

434 (namespaces["mathml"], "mo"), 

435 (namespaces["mathml"], "mn"), 

436 (namespaces["mathml"], "ms"), 

437 (namespaces["mathml"], "mtext") 

438]) 

439 

440adjustSVGAttributes = { 

441 "attributename": "attributeName", 

442 "attributetype": "attributeType", 

443 "basefrequency": "baseFrequency", 

444 "baseprofile": "baseProfile", 

445 "calcmode": "calcMode", 

446 "clippathunits": "clipPathUnits", 

447 "contentscripttype": "contentScriptType", 

448 "contentstyletype": "contentStyleType", 

449 "diffuseconstant": "diffuseConstant", 

450 "edgemode": "edgeMode", 

451 "externalresourcesrequired": "externalResourcesRequired", 

452 "filterres": "filterRes", 

453 "filterunits": "filterUnits", 

454 "glyphref": "glyphRef", 

455 "gradienttransform": "gradientTransform", 

456 "gradientunits": "gradientUnits", 

457 "kernelmatrix": "kernelMatrix", 

458 "kernelunitlength": "kernelUnitLength", 

459 "keypoints": "keyPoints", 

460 "keysplines": "keySplines", 

461 "keytimes": "keyTimes", 

462 "lengthadjust": "lengthAdjust", 

463 "limitingconeangle": "limitingConeAngle", 

464 "markerheight": "markerHeight", 

465 "markerunits": "markerUnits", 

466 "markerwidth": "markerWidth", 

467 "maskcontentunits": "maskContentUnits", 

468 "maskunits": "maskUnits", 

469 "numoctaves": "numOctaves", 

470 "pathlength": "pathLength", 

471 "patterncontentunits": "patternContentUnits", 

472 "patterntransform": "patternTransform", 

473 "patternunits": "patternUnits", 

474 "pointsatx": "pointsAtX", 

475 "pointsaty": "pointsAtY", 

476 "pointsatz": "pointsAtZ", 

477 "preservealpha": "preserveAlpha", 

478 "preserveaspectratio": "preserveAspectRatio", 

479 "primitiveunits": "primitiveUnits", 

480 "refx": "refX", 

481 "refy": "refY", 

482 "repeatcount": "repeatCount", 

483 "repeatdur": "repeatDur", 

484 "requiredextensions": "requiredExtensions", 

485 "requiredfeatures": "requiredFeatures", 

486 "specularconstant": "specularConstant", 

487 "specularexponent": "specularExponent", 

488 "spreadmethod": "spreadMethod", 

489 "startoffset": "startOffset", 

490 "stddeviation": "stdDeviation", 

491 "stitchtiles": "stitchTiles", 

492 "surfacescale": "surfaceScale", 

493 "systemlanguage": "systemLanguage", 

494 "tablevalues": "tableValues", 

495 "targetx": "targetX", 

496 "targety": "targetY", 

497 "textlength": "textLength", 

498 "viewbox": "viewBox", 

499 "viewtarget": "viewTarget", 

500 "xchannelselector": "xChannelSelector", 

501 "ychannelselector": "yChannelSelector", 

502 "zoomandpan": "zoomAndPan" 

503} 

504 

505adjustMathMLAttributes = {"definitionurl": "definitionURL"} 

506 

507adjustForeignAttributes = { 

508 "xlink:actuate": ("xlink", "actuate", namespaces["xlink"]), 

509 "xlink:arcrole": ("xlink", "arcrole", namespaces["xlink"]), 

510 "xlink:href": ("xlink", "href", namespaces["xlink"]), 

511 "xlink:role": ("xlink", "role", namespaces["xlink"]), 

512 "xlink:show": ("xlink", "show", namespaces["xlink"]), 

513 "xlink:title": ("xlink", "title", namespaces["xlink"]), 

514 "xlink:type": ("xlink", "type", namespaces["xlink"]), 

515 "xml:base": ("xml", "base", namespaces["xml"]), 

516 "xml:lang": ("xml", "lang", namespaces["xml"]), 

517 "xml:space": ("xml", "space", namespaces["xml"]), 

518 "xmlns": (None, "xmlns", namespaces["xmlns"]), 

519 "xmlns:xlink": ("xmlns", "xlink", namespaces["xmlns"]) 

520} 

521 

522unadjustForeignAttributes = {(ns, local): qname for qname, (prefix, local, ns) in 

523 adjustForeignAttributes.items()} 

524 

525spaceCharacters = frozenset([ 

526 "\t", 

527 "\n", 

528 "\u000C", 

529 " ", 

530 "\r" 

531]) 

532 

533tableInsertModeElements = frozenset([ 

534 "table", 

535 "tbody", 

536 "tfoot", 

537 "thead", 

538 "tr" 

539]) 

540 

541asciiLowercase = frozenset(string.ascii_lowercase) 

542asciiUppercase = frozenset(string.ascii_uppercase) 

543asciiLetters = frozenset(string.ascii_letters) 

544digits = frozenset(string.digits) 

545hexDigits = frozenset(string.hexdigits) 

546 

547asciiUpper2Lower = {ord(c): ord(c.lower()) for c in string.ascii_uppercase} 

548 

549# Heading elements need to be ordered 

550headingElements = ( 

551 "h1", 

552 "h2", 

553 "h3", 

554 "h4", 

555 "h5", 

556 "h6" 

557) 

558 

559voidElements = frozenset([ 

560 "base", 

561 "command", 

562 "event-source", 

563 "link", 

564 "meta", 

565 "hr", 

566 "br", 

567 "img", 

568 "embed", 

569 "param", 

570 "area", 

571 "col", 

572 "input", 

573 "source", 

574 "track" 

575]) 

576 

577cdataElements = frozenset(['title', 'textarea']) 

578 

579rcdataElements = frozenset([ 

580 'style', 

581 'script', 

582 'xmp', 

583 'iframe', 

584 'noembed', 

585 'noframes', 

586 'noscript' 

587]) 

588 

589booleanAttributes = { 

590 "": frozenset(["irrelevant", "itemscope"]), 

591 "style": frozenset(["scoped"]), 

592 "img": frozenset(["ismap"]), 

593 "audio": frozenset(["autoplay", "controls"]), 

594 "video": frozenset(["autoplay", "controls"]), 

595 "script": frozenset(["defer", "async"]), 

596 "details": frozenset(["open"]), 

597 "datagrid": frozenset(["multiple", "disabled"]), 

598 "command": frozenset(["hidden", "disabled", "checked", "default"]), 

599 "hr": frozenset(["noshade"]), 

600 "menu": frozenset(["autosubmit"]), 

601 "fieldset": frozenset(["disabled", "readonly"]), 

602 "option": frozenset(["disabled", "readonly", "selected"]), 

603 "optgroup": frozenset(["disabled", "readonly"]), 

604 "button": frozenset(["disabled", "autofocus"]), 

605 "input": frozenset(["disabled", "readonly", "required", "autofocus", "checked", "ismap"]), 

606 "select": frozenset(["disabled", "readonly", "autofocus", "multiple"]), 

607 "output": frozenset(["disabled", "readonly"]), 

608 "iframe": frozenset(["seamless"]), 

609} 

610 

611# entitiesWindows1252 has to be _ordered_ and needs to have an index. It 

612# therefore can't be a frozenset. 

613entitiesWindows1252 = ( 

614 8364, # 0x80 0x20AC EURO SIGN 

615 65533, # 0x81 UNDEFINED 

616 8218, # 0x82 0x201A SINGLE LOW-9 QUOTATION MARK 

617 402, # 0x83 0x0192 LATIN SMALL LETTER F WITH HOOK 

618 8222, # 0x84 0x201E DOUBLE LOW-9 QUOTATION MARK 

619 8230, # 0x85 0x2026 HORIZONTAL ELLIPSIS 

620 8224, # 0x86 0x2020 DAGGER 

621 8225, # 0x87 0x2021 DOUBLE DAGGER 

622 710, # 0x88 0x02C6 MODIFIER LETTER CIRCUMFLEX ACCENT 

623 8240, # 0x89 0x2030 PER MILLE SIGN 

624 352, # 0x8A 0x0160 LATIN CAPITAL LETTER S WITH CARON 

625 8249, # 0x8B 0x2039 SINGLE LEFT-POINTING ANGLE QUOTATION MARK 

626 338, # 0x8C 0x0152 LATIN CAPITAL LIGATURE OE 

627 65533, # 0x8D UNDEFINED 

628 381, # 0x8E 0x017D LATIN CAPITAL LETTER Z WITH CARON 

629 65533, # 0x8F UNDEFINED 

630 65533, # 0x90 UNDEFINED 

631 8216, # 0x91 0x2018 LEFT SINGLE QUOTATION MARK 

632 8217, # 0x92 0x2019 RIGHT SINGLE QUOTATION MARK 

633 8220, # 0x93 0x201C LEFT DOUBLE QUOTATION MARK 

634 8221, # 0x94 0x201D RIGHT DOUBLE QUOTATION MARK 

635 8226, # 0x95 0x2022 BULLET 

636 8211, # 0x96 0x2013 EN DASH 

637 8212, # 0x97 0x2014 EM DASH 

638 732, # 0x98 0x02DC SMALL TILDE 

639 8482, # 0x99 0x2122 TRADE MARK SIGN 

640 353, # 0x9A 0x0161 LATIN SMALL LETTER S WITH CARON 

641 8250, # 0x9B 0x203A SINGLE RIGHT-POINTING ANGLE QUOTATION MARK 

642 339, # 0x9C 0x0153 LATIN SMALL LIGATURE OE 

643 65533, # 0x9D UNDEFINED 

644 382, # 0x9E 0x017E LATIN SMALL LETTER Z WITH CARON 

645 376 # 0x9F 0x0178 LATIN CAPITAL LETTER Y WITH DIAERESIS 

646) 

647 

648xmlEntities = frozenset(['lt;', 'gt;', 'amp;', 'apos;', 'quot;']) 

649 

650entities = { 

651 "AElig": "\xc6", 

652 "AElig;": "\xc6", 

653 "AMP": "&", 

654 "AMP;": "&", 

655 "Aacute": "\xc1", 

656 "Aacute;": "\xc1", 

657 "Abreve;": "\u0102", 

658 "Acirc": "\xc2", 

659 "Acirc;": "\xc2", 

660 "Acy;": "\u0410", 

661 "Afr;": "\U0001d504", 

662 "Agrave": "\xc0", 

663 "Agrave;": "\xc0", 

664 "Alpha;": "\u0391", 

665 "Amacr;": "\u0100", 

666 "And;": "\u2a53", 

667 "Aogon;": "\u0104", 

668 "Aopf;": "\U0001d538", 

669 "ApplyFunction;": "\u2061", 

670 "Aring": "\xc5", 

671 "Aring;": "\xc5", 

672 "Ascr;": "\U0001d49c", 

673 "Assign;": "\u2254", 

674 "Atilde": "\xc3", 

675 "Atilde;": "\xc3", 

676 "Auml": "\xc4", 

677 "Auml;": "\xc4", 

678 "Backslash;": "\u2216", 

679 "Barv;": "\u2ae7", 

680 "Barwed;": "\u2306", 

681 "Bcy;": "\u0411", 

682 "Because;": "\u2235", 

683 "Bernoullis;": "\u212c", 

684 "Beta;": "\u0392", 

685 "Bfr;": "\U0001d505", 

686 "Bopf;": "\U0001d539", 

687 "Breve;": "\u02d8", 

688 "Bscr;": "\u212c", 

689 "Bumpeq;": "\u224e", 

690 "CHcy;": "\u0427", 

691 "COPY": "\xa9", 

692 "COPY;": "\xa9", 

693 "Cacute;": "\u0106", 

694 "Cap;": "\u22d2", 

695 "CapitalDifferentialD;": "\u2145", 

696 "Cayleys;": "\u212d", 

697 "Ccaron;": "\u010c", 

698 "Ccedil": "\xc7", 

699 "Ccedil;": "\xc7", 

700 "Ccirc;": "\u0108", 

701 "Cconint;": "\u2230", 

702 "Cdot;": "\u010a", 

703 "Cedilla;": "\xb8", 

704 "CenterDot;": "\xb7", 

705 "Cfr;": "\u212d", 

706 "Chi;": "\u03a7", 

707 "CircleDot;": "\u2299", 

708 "CircleMinus;": "\u2296", 

709 "CirclePlus;": "\u2295", 

710 "CircleTimes;": "\u2297", 

711 "ClockwiseContourIntegral;": "\u2232", 

712 "CloseCurlyDoubleQuote;": "\u201d", 

713 "CloseCurlyQuote;": "\u2019", 

714 "Colon;": "\u2237", 

715 "Colone;": "\u2a74", 

716 "Congruent;": "\u2261", 

717 "Conint;": "\u222f", 

718 "ContourIntegral;": "\u222e", 

719 "Copf;": "\u2102", 

720 "Coproduct;": "\u2210", 

721 "CounterClockwiseContourIntegral;": "\u2233", 

722 "Cross;": "\u2a2f", 

723 "Cscr;": "\U0001d49e", 

724 "Cup;": "\u22d3", 

725 "CupCap;": "\u224d", 

726 "DD;": "\u2145", 

727 "DDotrahd;": "\u2911", 

728 "DJcy;": "\u0402", 

729 "DScy;": "\u0405", 

730 "DZcy;": "\u040f", 

731 "Dagger;": "\u2021", 

732 "Darr;": "\u21a1", 

733 "Dashv;": "\u2ae4", 

734 "Dcaron;": "\u010e", 

735 "Dcy;": "\u0414", 

736 "Del;": "\u2207", 

737 "Delta;": "\u0394", 

738 "Dfr;": "\U0001d507", 

739 "DiacriticalAcute;": "\xb4", 

740 "DiacriticalDot;": "\u02d9", 

741 "DiacriticalDoubleAcute;": "\u02dd", 

742 "DiacriticalGrave;": "`", 

743 "DiacriticalTilde;": "\u02dc", 

744 "Diamond;": "\u22c4", 

745 "DifferentialD;": "\u2146", 

746 "Dopf;": "\U0001d53b", 

747 "Dot;": "\xa8", 

748 "DotDot;": "\u20dc", 

749 "DotEqual;": "\u2250", 

750 "DoubleContourIntegral;": "\u222f", 

751 "DoubleDot;": "\xa8", 

752 "DoubleDownArrow;": "\u21d3", 

753 "DoubleLeftArrow;": "\u21d0", 

754 "DoubleLeftRightArrow;": "\u21d4", 

755 "DoubleLeftTee;": "\u2ae4", 

756 "DoubleLongLeftArrow;": "\u27f8", 

757 "DoubleLongLeftRightArrow;": "\u27fa", 

758 "DoubleLongRightArrow;": "\u27f9", 

759 "DoubleRightArrow;": "\u21d2", 

760 "DoubleRightTee;": "\u22a8", 

761 "DoubleUpArrow;": "\u21d1", 

762 "DoubleUpDownArrow;": "\u21d5", 

763 "DoubleVerticalBar;": "\u2225", 

764 "DownArrow;": "\u2193", 

765 "DownArrowBar;": "\u2913", 

766 "DownArrowUpArrow;": "\u21f5", 

767 "DownBreve;": "\u0311", 

768 "DownLeftRightVector;": "\u2950", 

769 "DownLeftTeeVector;": "\u295e", 

770 "DownLeftVector;": "\u21bd", 

771 "DownLeftVectorBar;": "\u2956", 

772 "DownRightTeeVector;": "\u295f", 

773 "DownRightVector;": "\u21c1", 

774 "DownRightVectorBar;": "\u2957", 

775 "DownTee;": "\u22a4", 

776 "DownTeeArrow;": "\u21a7", 

777 "Downarrow;": "\u21d3", 

778 "Dscr;": "\U0001d49f", 

779 "Dstrok;": "\u0110", 

780 "ENG;": "\u014a", 

781 "ETH": "\xd0", 

782 "ETH;": "\xd0", 

783 "Eacute": "\xc9", 

784 "Eacute;": "\xc9", 

785 "Ecaron;": "\u011a", 

786 "Ecirc": "\xca", 

787 "Ecirc;": "\xca", 

788 "Ecy;": "\u042d", 

789 "Edot;": "\u0116", 

790 "Efr;": "\U0001d508", 

791 "Egrave": "\xc8", 

792 "Egrave;": "\xc8", 

793 "Element;": "\u2208", 

794 "Emacr;": "\u0112", 

795 "EmptySmallSquare;": "\u25fb", 

796 "EmptyVerySmallSquare;": "\u25ab", 

797 "Eogon;": "\u0118", 

798 "Eopf;": "\U0001d53c", 

799 "Epsilon;": "\u0395", 

800 "Equal;": "\u2a75", 

801 "EqualTilde;": "\u2242", 

802 "Equilibrium;": "\u21cc", 

803 "Escr;": "\u2130", 

804 "Esim;": "\u2a73", 

805 "Eta;": "\u0397", 

806 "Euml": "\xcb", 

807 "Euml;": "\xcb", 

808 "Exists;": "\u2203", 

809 "ExponentialE;": "\u2147", 

810 "Fcy;": "\u0424", 

811 "Ffr;": "\U0001d509", 

812 "FilledSmallSquare;": "\u25fc", 

813 "FilledVerySmallSquare;": "\u25aa", 

814 "Fopf;": "\U0001d53d", 

815 "ForAll;": "\u2200", 

816 "Fouriertrf;": "\u2131", 

817 "Fscr;": "\u2131", 

818 "GJcy;": "\u0403", 

819 "GT": ">", 

820 "GT;": ">", 

821 "Gamma;": "\u0393", 

822 "Gammad;": "\u03dc", 

823 "Gbreve;": "\u011e", 

824 "Gcedil;": "\u0122", 

825 "Gcirc;": "\u011c", 

826 "Gcy;": "\u0413", 

827 "Gdot;": "\u0120", 

828 "Gfr;": "\U0001d50a", 

829 "Gg;": "\u22d9", 

830 "Gopf;": "\U0001d53e", 

831 "GreaterEqual;": "\u2265", 

832 "GreaterEqualLess;": "\u22db", 

833 "GreaterFullEqual;": "\u2267", 

834 "GreaterGreater;": "\u2aa2", 

835 "GreaterLess;": "\u2277", 

836 "GreaterSlantEqual;": "\u2a7e", 

837 "GreaterTilde;": "\u2273", 

838 "Gscr;": "\U0001d4a2", 

839 "Gt;": "\u226b", 

840 "HARDcy;": "\u042a", 

841 "Hacek;": "\u02c7", 

842 "Hat;": "^", 

843 "Hcirc;": "\u0124", 

844 "Hfr;": "\u210c", 

845 "HilbertSpace;": "\u210b", 

846 "Hopf;": "\u210d", 

847 "HorizontalLine;": "\u2500", 

848 "Hscr;": "\u210b", 

849 "Hstrok;": "\u0126", 

850 "HumpDownHump;": "\u224e", 

851 "HumpEqual;": "\u224f", 

852 "IEcy;": "\u0415", 

853 "IJlig;": "\u0132", 

854 "IOcy;": "\u0401", 

855 "Iacute": "\xcd", 

856 "Iacute;": "\xcd", 

857 "Icirc": "\xce", 

858 "Icirc;": "\xce", 

859 "Icy;": "\u0418", 

860 "Idot;": "\u0130", 

861 "Ifr;": "\u2111", 

862 "Igrave": "\xcc", 

863 "Igrave;": "\xcc", 

864 "Im;": "\u2111", 

865 "Imacr;": "\u012a", 

866 "ImaginaryI;": "\u2148", 

867 "Implies;": "\u21d2", 

868 "Int;": "\u222c", 

869 "Integral;": "\u222b", 

870 "Intersection;": "\u22c2", 

871 "InvisibleComma;": "\u2063", 

872 "InvisibleTimes;": "\u2062", 

873 "Iogon;": "\u012e", 

874 "Iopf;": "\U0001d540", 

875 "Iota;": "\u0399", 

876 "Iscr;": "\u2110", 

877 "Itilde;": "\u0128", 

878 "Iukcy;": "\u0406", 

879 "Iuml": "\xcf", 

880 "Iuml;": "\xcf", 

881 "Jcirc;": "\u0134", 

882 "Jcy;": "\u0419", 

883 "Jfr;": "\U0001d50d", 

884 "Jopf;": "\U0001d541",