David K. Bainbridge | 215e024 | 2017-09-05 23:18:24 -0700 | [diff] [blame] | 1 | // Copyright 2010 The Go Authors. All rights reserved. |
| 2 | // Use of this source code is governed by a BSD-style |
| 3 | // license that can be found in the LICENSE file. |
| 4 | |
| 5 | package html |
| 6 | |
| 7 | import ( |
| 8 | "errors" |
| 9 | "fmt" |
| 10 | "io" |
| 11 | "strings" |
| 12 | |
| 13 | a "golang.org/x/net/html/atom" |
| 14 | ) |
| 15 | |
| 16 | // A parser implements the HTML5 parsing algorithm: |
| 17 | // https://html.spec.whatwg.org/multipage/syntax.html#tree-construction |
| 18 | type parser struct { |
| 19 | // tokenizer provides the tokens for the parser. |
| 20 | tokenizer *Tokenizer |
| 21 | // tok is the most recently read token. |
| 22 | tok Token |
| 23 | // Self-closing tags like <hr/> are treated as start tags, except that |
| 24 | // hasSelfClosingToken is set while they are being processed. |
| 25 | hasSelfClosingToken bool |
| 26 | // doc is the document root element. |
| 27 | doc *Node |
| 28 | // The stack of open elements (section 12.2.3.2) and active formatting |
| 29 | // elements (section 12.2.3.3). |
| 30 | oe, afe nodeStack |
| 31 | // Element pointers (section 12.2.3.4). |
| 32 | head, form *Node |
| 33 | // Other parsing state flags (section 12.2.3.5). |
| 34 | scripting, framesetOK bool |
| 35 | // im is the current insertion mode. |
| 36 | im insertionMode |
| 37 | // originalIM is the insertion mode to go back to after completing a text |
| 38 | // or inTableText insertion mode. |
| 39 | originalIM insertionMode |
| 40 | // fosterParenting is whether new elements should be inserted according to |
| 41 | // the foster parenting rules (section 12.2.5.3). |
| 42 | fosterParenting bool |
| 43 | // quirks is whether the parser is operating in "quirks mode." |
| 44 | quirks bool |
| 45 | // fragment is whether the parser is parsing an HTML fragment. |
| 46 | fragment bool |
| 47 | // context is the context element when parsing an HTML fragment |
| 48 | // (section 12.4). |
| 49 | context *Node |
| 50 | } |
| 51 | |
| 52 | func (p *parser) top() *Node { |
| 53 | if n := p.oe.top(); n != nil { |
| 54 | return n |
| 55 | } |
| 56 | return p.doc |
| 57 | } |
| 58 | |
| 59 | // Stop tags for use in popUntil. These come from section 12.2.3.2. |
| 60 | var ( |
| 61 | defaultScopeStopTags = map[string][]a.Atom{ |
| 62 | "": {a.Applet, a.Caption, a.Html, a.Table, a.Td, a.Th, a.Marquee, a.Object, a.Template}, |
| 63 | "math": {a.AnnotationXml, a.Mi, a.Mn, a.Mo, a.Ms, a.Mtext}, |
| 64 | "svg": {a.Desc, a.ForeignObject, a.Title}, |
| 65 | } |
| 66 | ) |
| 67 | |
| 68 | type scope int |
| 69 | |
| 70 | const ( |
| 71 | defaultScope scope = iota |
| 72 | listItemScope |
| 73 | buttonScope |
| 74 | tableScope |
| 75 | tableRowScope |
| 76 | tableBodyScope |
| 77 | selectScope |
| 78 | ) |
| 79 | |
| 80 | // popUntil pops the stack of open elements at the highest element whose tag |
| 81 | // is in matchTags, provided there is no higher element in the scope's stop |
| 82 | // tags (as defined in section 12.2.3.2). It returns whether or not there was |
| 83 | // such an element. If there was not, popUntil leaves the stack unchanged. |
| 84 | // |
| 85 | // For example, the set of stop tags for table scope is: "html", "table". If |
| 86 | // the stack was: |
| 87 | // ["html", "body", "font", "table", "b", "i", "u"] |
| 88 | // then popUntil(tableScope, "font") would return false, but |
| 89 | // popUntil(tableScope, "i") would return true and the stack would become: |
| 90 | // ["html", "body", "font", "table", "b"] |
| 91 | // |
| 92 | // If an element's tag is in both the stop tags and matchTags, then the stack |
| 93 | // will be popped and the function returns true (provided, of course, there was |
| 94 | // no higher element in the stack that was also in the stop tags). For example, |
| 95 | // popUntil(tableScope, "table") returns true and leaves: |
| 96 | // ["html", "body", "font"] |
| 97 | func (p *parser) popUntil(s scope, matchTags ...a.Atom) bool { |
| 98 | if i := p.indexOfElementInScope(s, matchTags...); i != -1 { |
| 99 | p.oe = p.oe[:i] |
| 100 | return true |
| 101 | } |
| 102 | return false |
| 103 | } |
| 104 | |
| 105 | // indexOfElementInScope returns the index in p.oe of the highest element whose |
| 106 | // tag is in matchTags that is in scope. If no matching element is in scope, it |
| 107 | // returns -1. |
| 108 | func (p *parser) indexOfElementInScope(s scope, matchTags ...a.Atom) int { |
| 109 | for i := len(p.oe) - 1; i >= 0; i-- { |
| 110 | tagAtom := p.oe[i].DataAtom |
| 111 | if p.oe[i].Namespace == "" { |
| 112 | for _, t := range matchTags { |
| 113 | if t == tagAtom { |
| 114 | return i |
| 115 | } |
| 116 | } |
| 117 | switch s { |
| 118 | case defaultScope: |
| 119 | // No-op. |
| 120 | case listItemScope: |
| 121 | if tagAtom == a.Ol || tagAtom == a.Ul { |
| 122 | return -1 |
| 123 | } |
| 124 | case buttonScope: |
| 125 | if tagAtom == a.Button { |
| 126 | return -1 |
| 127 | } |
| 128 | case tableScope: |
| 129 | if tagAtom == a.Html || tagAtom == a.Table { |
| 130 | return -1 |
| 131 | } |
| 132 | case selectScope: |
| 133 | if tagAtom != a.Optgroup && tagAtom != a.Option { |
| 134 | return -1 |
| 135 | } |
| 136 | default: |
| 137 | panic("unreachable") |
| 138 | } |
| 139 | } |
| 140 | switch s { |
| 141 | case defaultScope, listItemScope, buttonScope: |
| 142 | for _, t := range defaultScopeStopTags[p.oe[i].Namespace] { |
| 143 | if t == tagAtom { |
| 144 | return -1 |
| 145 | } |
| 146 | } |
| 147 | } |
| 148 | } |
| 149 | return -1 |
| 150 | } |
| 151 | |
| 152 | // elementInScope is like popUntil, except that it doesn't modify the stack of |
| 153 | // open elements. |
| 154 | func (p *parser) elementInScope(s scope, matchTags ...a.Atom) bool { |
| 155 | return p.indexOfElementInScope(s, matchTags...) != -1 |
| 156 | } |
| 157 | |
| 158 | // clearStackToContext pops elements off the stack of open elements until a |
| 159 | // scope-defined element is found. |
| 160 | func (p *parser) clearStackToContext(s scope) { |
| 161 | for i := len(p.oe) - 1; i >= 0; i-- { |
| 162 | tagAtom := p.oe[i].DataAtom |
| 163 | switch s { |
| 164 | case tableScope: |
| 165 | if tagAtom == a.Html || tagAtom == a.Table { |
| 166 | p.oe = p.oe[:i+1] |
| 167 | return |
| 168 | } |
| 169 | case tableRowScope: |
| 170 | if tagAtom == a.Html || tagAtom == a.Tr { |
| 171 | p.oe = p.oe[:i+1] |
| 172 | return |
| 173 | } |
| 174 | case tableBodyScope: |
| 175 | if tagAtom == a.Html || tagAtom == a.Tbody || tagAtom == a.Tfoot || tagAtom == a.Thead { |
| 176 | p.oe = p.oe[:i+1] |
| 177 | return |
| 178 | } |
| 179 | default: |
| 180 | panic("unreachable") |
| 181 | } |
| 182 | } |
| 183 | } |
| 184 | |
| 185 | // generateImpliedEndTags pops nodes off the stack of open elements as long as |
| 186 | // the top node has a tag name of dd, dt, li, option, optgroup, p, rp, or rt. |
| 187 | // If exceptions are specified, nodes with that name will not be popped off. |
| 188 | func (p *parser) generateImpliedEndTags(exceptions ...string) { |
| 189 | var i int |
| 190 | loop: |
| 191 | for i = len(p.oe) - 1; i >= 0; i-- { |
| 192 | n := p.oe[i] |
| 193 | if n.Type == ElementNode { |
| 194 | switch n.DataAtom { |
| 195 | case a.Dd, a.Dt, a.Li, a.Option, a.Optgroup, a.P, a.Rp, a.Rt: |
| 196 | for _, except := range exceptions { |
| 197 | if n.Data == except { |
| 198 | break loop |
| 199 | } |
| 200 | } |
| 201 | continue |
| 202 | } |
| 203 | } |
| 204 | break |
| 205 | } |
| 206 | |
| 207 | p.oe = p.oe[:i+1] |
| 208 | } |
| 209 | |
| 210 | // addChild adds a child node n to the top element, and pushes n onto the stack |
| 211 | // of open elements if it is an element node. |
| 212 | func (p *parser) addChild(n *Node) { |
| 213 | if p.shouldFosterParent() { |
| 214 | p.fosterParent(n) |
| 215 | } else { |
| 216 | p.top().AppendChild(n) |
| 217 | } |
| 218 | |
| 219 | if n.Type == ElementNode { |
| 220 | p.oe = append(p.oe, n) |
| 221 | } |
| 222 | } |
| 223 | |
| 224 | // shouldFosterParent returns whether the next node to be added should be |
| 225 | // foster parented. |
| 226 | func (p *parser) shouldFosterParent() bool { |
| 227 | if p.fosterParenting { |
| 228 | switch p.top().DataAtom { |
| 229 | case a.Table, a.Tbody, a.Tfoot, a.Thead, a.Tr: |
| 230 | return true |
| 231 | } |
| 232 | } |
| 233 | return false |
| 234 | } |
| 235 | |
| 236 | // fosterParent adds a child node according to the foster parenting rules. |
| 237 | // Section 12.2.5.3, "foster parenting". |
| 238 | func (p *parser) fosterParent(n *Node) { |
| 239 | var table, parent, prev *Node |
| 240 | var i int |
| 241 | for i = len(p.oe) - 1; i >= 0; i-- { |
| 242 | if p.oe[i].DataAtom == a.Table { |
| 243 | table = p.oe[i] |
| 244 | break |
| 245 | } |
| 246 | } |
| 247 | |
| 248 | if table == nil { |
| 249 | // The foster parent is the html element. |
| 250 | parent = p.oe[0] |
| 251 | } else { |
| 252 | parent = table.Parent |
| 253 | } |
| 254 | if parent == nil { |
| 255 | parent = p.oe[i-1] |
| 256 | } |
| 257 | |
| 258 | if table != nil { |
| 259 | prev = table.PrevSibling |
| 260 | } else { |
| 261 | prev = parent.LastChild |
| 262 | } |
| 263 | if prev != nil && prev.Type == TextNode && n.Type == TextNode { |
| 264 | prev.Data += n.Data |
| 265 | return |
| 266 | } |
| 267 | |
| 268 | parent.InsertBefore(n, table) |
| 269 | } |
| 270 | |
| 271 | // addText adds text to the preceding node if it is a text node, or else it |
| 272 | // calls addChild with a new text node. |
| 273 | func (p *parser) addText(text string) { |
| 274 | if text == "" { |
| 275 | return |
| 276 | } |
| 277 | |
| 278 | if p.shouldFosterParent() { |
| 279 | p.fosterParent(&Node{ |
| 280 | Type: TextNode, |
| 281 | Data: text, |
| 282 | }) |
| 283 | return |
| 284 | } |
| 285 | |
| 286 | t := p.top() |
| 287 | if n := t.LastChild; n != nil && n.Type == TextNode { |
| 288 | n.Data += text |
| 289 | return |
| 290 | } |
| 291 | p.addChild(&Node{ |
| 292 | Type: TextNode, |
| 293 | Data: text, |
| 294 | }) |
| 295 | } |
| 296 | |
| 297 | // addElement adds a child element based on the current token. |
| 298 | func (p *parser) addElement() { |
| 299 | p.addChild(&Node{ |
| 300 | Type: ElementNode, |
| 301 | DataAtom: p.tok.DataAtom, |
| 302 | Data: p.tok.Data, |
| 303 | Attr: p.tok.Attr, |
| 304 | }) |
| 305 | } |
| 306 | |
| 307 | // Section 12.2.3.3. |
| 308 | func (p *parser) addFormattingElement() { |
| 309 | tagAtom, attr := p.tok.DataAtom, p.tok.Attr |
| 310 | p.addElement() |
| 311 | |
| 312 | // Implement the Noah's Ark clause, but with three per family instead of two. |
| 313 | identicalElements := 0 |
| 314 | findIdenticalElements: |
| 315 | for i := len(p.afe) - 1; i >= 0; i-- { |
| 316 | n := p.afe[i] |
| 317 | if n.Type == scopeMarkerNode { |
| 318 | break |
| 319 | } |
| 320 | if n.Type != ElementNode { |
| 321 | continue |
| 322 | } |
| 323 | if n.Namespace != "" { |
| 324 | continue |
| 325 | } |
| 326 | if n.DataAtom != tagAtom { |
| 327 | continue |
| 328 | } |
| 329 | if len(n.Attr) != len(attr) { |
| 330 | continue |
| 331 | } |
| 332 | compareAttributes: |
| 333 | for _, t0 := range n.Attr { |
| 334 | for _, t1 := range attr { |
| 335 | if t0.Key == t1.Key && t0.Namespace == t1.Namespace && t0.Val == t1.Val { |
| 336 | // Found a match for this attribute, continue with the next attribute. |
| 337 | continue compareAttributes |
| 338 | } |
| 339 | } |
| 340 | // If we get here, there is no attribute that matches a. |
| 341 | // Therefore the element is not identical to the new one. |
| 342 | continue findIdenticalElements |
| 343 | } |
| 344 | |
| 345 | identicalElements++ |
| 346 | if identicalElements >= 3 { |
| 347 | p.afe.remove(n) |
| 348 | } |
| 349 | } |
| 350 | |
| 351 | p.afe = append(p.afe, p.top()) |
| 352 | } |
| 353 | |
| 354 | // Section 12.2.3.3. |
| 355 | func (p *parser) clearActiveFormattingElements() { |
| 356 | for { |
| 357 | n := p.afe.pop() |
| 358 | if len(p.afe) == 0 || n.Type == scopeMarkerNode { |
| 359 | return |
| 360 | } |
| 361 | } |
| 362 | } |
| 363 | |
| 364 | // Section 12.2.3.3. |
| 365 | func (p *parser) reconstructActiveFormattingElements() { |
| 366 | n := p.afe.top() |
| 367 | if n == nil { |
| 368 | return |
| 369 | } |
| 370 | if n.Type == scopeMarkerNode || p.oe.index(n) != -1 { |
| 371 | return |
| 372 | } |
| 373 | i := len(p.afe) - 1 |
| 374 | for n.Type != scopeMarkerNode && p.oe.index(n) == -1 { |
| 375 | if i == 0 { |
| 376 | i = -1 |
| 377 | break |
| 378 | } |
| 379 | i-- |
| 380 | n = p.afe[i] |
| 381 | } |
| 382 | for { |
| 383 | i++ |
| 384 | clone := p.afe[i].clone() |
| 385 | p.addChild(clone) |
| 386 | p.afe[i] = clone |
| 387 | if i == len(p.afe)-1 { |
| 388 | break |
| 389 | } |
| 390 | } |
| 391 | } |
| 392 | |
| 393 | // Section 12.2.4. |
| 394 | func (p *parser) acknowledgeSelfClosingTag() { |
| 395 | p.hasSelfClosingToken = false |
| 396 | } |
| 397 | |
| 398 | // An insertion mode (section 12.2.3.1) is the state transition function from |
| 399 | // a particular state in the HTML5 parser's state machine. It updates the |
| 400 | // parser's fields depending on parser.tok (where ErrorToken means EOF). |
| 401 | // It returns whether the token was consumed. |
| 402 | type insertionMode func(*parser) bool |
| 403 | |
| 404 | // setOriginalIM sets the insertion mode to return to after completing a text or |
| 405 | // inTableText insertion mode. |
| 406 | // Section 12.2.3.1, "using the rules for". |
| 407 | func (p *parser) setOriginalIM() { |
| 408 | if p.originalIM != nil { |
| 409 | panic("html: bad parser state: originalIM was set twice") |
| 410 | } |
| 411 | p.originalIM = p.im |
| 412 | } |
| 413 | |
| 414 | // Section 12.2.3.1, "reset the insertion mode". |
| 415 | func (p *parser) resetInsertionMode() { |
| 416 | for i := len(p.oe) - 1; i >= 0; i-- { |
| 417 | n := p.oe[i] |
| 418 | if i == 0 && p.context != nil { |
| 419 | n = p.context |
| 420 | } |
| 421 | |
| 422 | switch n.DataAtom { |
| 423 | case a.Select: |
| 424 | p.im = inSelectIM |
| 425 | case a.Td, a.Th: |
| 426 | p.im = inCellIM |
| 427 | case a.Tr: |
| 428 | p.im = inRowIM |
| 429 | case a.Tbody, a.Thead, a.Tfoot: |
| 430 | p.im = inTableBodyIM |
| 431 | case a.Caption: |
| 432 | p.im = inCaptionIM |
| 433 | case a.Colgroup: |
| 434 | p.im = inColumnGroupIM |
| 435 | case a.Table: |
| 436 | p.im = inTableIM |
| 437 | case a.Head: |
| 438 | p.im = inBodyIM |
| 439 | case a.Body: |
| 440 | p.im = inBodyIM |
| 441 | case a.Frameset: |
| 442 | p.im = inFramesetIM |
| 443 | case a.Html: |
| 444 | p.im = beforeHeadIM |
| 445 | default: |
| 446 | continue |
| 447 | } |
| 448 | return |
| 449 | } |
| 450 | p.im = inBodyIM |
| 451 | } |
| 452 | |
| 453 | const whitespace = " \t\r\n\f" |
| 454 | |
| 455 | // Section 12.2.5.4.1. |
| 456 | func initialIM(p *parser) bool { |
| 457 | switch p.tok.Type { |
| 458 | case TextToken: |
| 459 | p.tok.Data = strings.TrimLeft(p.tok.Data, whitespace) |
| 460 | if len(p.tok.Data) == 0 { |
| 461 | // It was all whitespace, so ignore it. |
| 462 | return true |
| 463 | } |
| 464 | case CommentToken: |
| 465 | p.doc.AppendChild(&Node{ |
| 466 | Type: CommentNode, |
| 467 | Data: p.tok.Data, |
| 468 | }) |
| 469 | return true |
| 470 | case DoctypeToken: |
| 471 | n, quirks := parseDoctype(p.tok.Data) |
| 472 | p.doc.AppendChild(n) |
| 473 | p.quirks = quirks |
| 474 | p.im = beforeHTMLIM |
| 475 | return true |
| 476 | } |
| 477 | p.quirks = true |
| 478 | p.im = beforeHTMLIM |
| 479 | return false |
| 480 | } |
| 481 | |
| 482 | // Section 12.2.5.4.2. |
| 483 | func beforeHTMLIM(p *parser) bool { |
| 484 | switch p.tok.Type { |
| 485 | case DoctypeToken: |
| 486 | // Ignore the token. |
| 487 | return true |
| 488 | case TextToken: |
| 489 | p.tok.Data = strings.TrimLeft(p.tok.Data, whitespace) |
| 490 | if len(p.tok.Data) == 0 { |
| 491 | // It was all whitespace, so ignore it. |
| 492 | return true |
| 493 | } |
| 494 | case StartTagToken: |
| 495 | if p.tok.DataAtom == a.Html { |
| 496 | p.addElement() |
| 497 | p.im = beforeHeadIM |
| 498 | return true |
| 499 | } |
| 500 | case EndTagToken: |
| 501 | switch p.tok.DataAtom { |
| 502 | case a.Head, a.Body, a.Html, a.Br: |
| 503 | p.parseImpliedToken(StartTagToken, a.Html, a.Html.String()) |
| 504 | return false |
| 505 | default: |
| 506 | // Ignore the token. |
| 507 | return true |
| 508 | } |
| 509 | case CommentToken: |
| 510 | p.doc.AppendChild(&Node{ |
| 511 | Type: CommentNode, |
| 512 | Data: p.tok.Data, |
| 513 | }) |
| 514 | return true |
| 515 | } |
| 516 | p.parseImpliedToken(StartTagToken, a.Html, a.Html.String()) |
| 517 | return false |
| 518 | } |
| 519 | |
| 520 | // Section 12.2.5.4.3. |
| 521 | func beforeHeadIM(p *parser) bool { |
| 522 | switch p.tok.Type { |
| 523 | case TextToken: |
| 524 | p.tok.Data = strings.TrimLeft(p.tok.Data, whitespace) |
| 525 | if len(p.tok.Data) == 0 { |
| 526 | // It was all whitespace, so ignore it. |
| 527 | return true |
| 528 | } |
| 529 | case StartTagToken: |
| 530 | switch p.tok.DataAtom { |
| 531 | case a.Head: |
| 532 | p.addElement() |
| 533 | p.head = p.top() |
| 534 | p.im = inHeadIM |
| 535 | return true |
| 536 | case a.Html: |
| 537 | return inBodyIM(p) |
| 538 | } |
| 539 | case EndTagToken: |
| 540 | switch p.tok.DataAtom { |
| 541 | case a.Head, a.Body, a.Html, a.Br: |
| 542 | p.parseImpliedToken(StartTagToken, a.Head, a.Head.String()) |
| 543 | return false |
| 544 | default: |
| 545 | // Ignore the token. |
| 546 | return true |
| 547 | } |
| 548 | case CommentToken: |
| 549 | p.addChild(&Node{ |
| 550 | Type: CommentNode, |
| 551 | Data: p.tok.Data, |
| 552 | }) |
| 553 | return true |
| 554 | case DoctypeToken: |
| 555 | // Ignore the token. |
| 556 | return true |
| 557 | } |
| 558 | |
| 559 | p.parseImpliedToken(StartTagToken, a.Head, a.Head.String()) |
| 560 | return false |
| 561 | } |
| 562 | |
| 563 | // Section 12.2.5.4.4. |
| 564 | func inHeadIM(p *parser) bool { |
| 565 | switch p.tok.Type { |
| 566 | case TextToken: |
| 567 | s := strings.TrimLeft(p.tok.Data, whitespace) |
| 568 | if len(s) < len(p.tok.Data) { |
| 569 | // Add the initial whitespace to the current node. |
| 570 | p.addText(p.tok.Data[:len(p.tok.Data)-len(s)]) |
| 571 | if s == "" { |
| 572 | return true |
| 573 | } |
| 574 | p.tok.Data = s |
| 575 | } |
| 576 | case StartTagToken: |
| 577 | switch p.tok.DataAtom { |
| 578 | case a.Html: |
| 579 | return inBodyIM(p) |
| 580 | case a.Base, a.Basefont, a.Bgsound, a.Command, a.Link, a.Meta: |
| 581 | p.addElement() |
| 582 | p.oe.pop() |
| 583 | p.acknowledgeSelfClosingTag() |
| 584 | return true |
| 585 | case a.Script, a.Title, a.Noscript, a.Noframes, a.Style: |
| 586 | p.addElement() |
| 587 | p.setOriginalIM() |
| 588 | p.im = textIM |
| 589 | return true |
| 590 | case a.Head: |
| 591 | // Ignore the token. |
| 592 | return true |
| 593 | } |
| 594 | case EndTagToken: |
| 595 | switch p.tok.DataAtom { |
| 596 | case a.Head: |
| 597 | n := p.oe.pop() |
| 598 | if n.DataAtom != a.Head { |
| 599 | panic("html: bad parser state: <head> element not found, in the in-head insertion mode") |
| 600 | } |
| 601 | p.im = afterHeadIM |
| 602 | return true |
| 603 | case a.Body, a.Html, a.Br: |
| 604 | p.parseImpliedToken(EndTagToken, a.Head, a.Head.String()) |
| 605 | return false |
| 606 | default: |
| 607 | // Ignore the token. |
| 608 | return true |
| 609 | } |
| 610 | case CommentToken: |
| 611 | p.addChild(&Node{ |
| 612 | Type: CommentNode, |
| 613 | Data: p.tok.Data, |
| 614 | }) |
| 615 | return true |
| 616 | case DoctypeToken: |
| 617 | // Ignore the token. |
| 618 | return true |
| 619 | } |
| 620 | |
| 621 | p.parseImpliedToken(EndTagToken, a.Head, a.Head.String()) |
| 622 | return false |
| 623 | } |
| 624 | |
| 625 | // Section 12.2.5.4.6. |
| 626 | func afterHeadIM(p *parser) bool { |
| 627 | switch p.tok.Type { |
| 628 | case TextToken: |
| 629 | s := strings.TrimLeft(p.tok.Data, whitespace) |
| 630 | if len(s) < len(p.tok.Data) { |
| 631 | // Add the initial whitespace to the current node. |
| 632 | p.addText(p.tok.Data[:len(p.tok.Data)-len(s)]) |
| 633 | if s == "" { |
| 634 | return true |
| 635 | } |
| 636 | p.tok.Data = s |
| 637 | } |
| 638 | case StartTagToken: |
| 639 | switch p.tok.DataAtom { |
| 640 | case a.Html: |
| 641 | return inBodyIM(p) |
| 642 | case a.Body: |
| 643 | p.addElement() |
| 644 | p.framesetOK = false |
| 645 | p.im = inBodyIM |
| 646 | return true |
| 647 | case a.Frameset: |
| 648 | p.addElement() |
| 649 | p.im = inFramesetIM |
| 650 | return true |
| 651 | case a.Base, a.Basefont, a.Bgsound, a.Link, a.Meta, a.Noframes, a.Script, a.Style, a.Title: |
| 652 | p.oe = append(p.oe, p.head) |
| 653 | defer p.oe.remove(p.head) |
| 654 | return inHeadIM(p) |
| 655 | case a.Head: |
| 656 | // Ignore the token. |
| 657 | return true |
| 658 | } |
| 659 | case EndTagToken: |
| 660 | switch p.tok.DataAtom { |
| 661 | case a.Body, a.Html, a.Br: |
| 662 | // Drop down to creating an implied <body> tag. |
| 663 | default: |
| 664 | // Ignore the token. |
| 665 | return true |
| 666 | } |
| 667 | case CommentToken: |
| 668 | p.addChild(&Node{ |
| 669 | Type: CommentNode, |
| 670 | Data: p.tok.Data, |
| 671 | }) |
| 672 | return true |
| 673 | case DoctypeToken: |
| 674 | // Ignore the token. |
| 675 | return true |
| 676 | } |
| 677 | |
| 678 | p.parseImpliedToken(StartTagToken, a.Body, a.Body.String()) |
| 679 | p.framesetOK = true |
| 680 | return false |
| 681 | } |
| 682 | |
| 683 | // copyAttributes copies attributes of src not found on dst to dst. |
| 684 | func copyAttributes(dst *Node, src Token) { |
| 685 | if len(src.Attr) == 0 { |
| 686 | return |
| 687 | } |
| 688 | attr := map[string]string{} |
| 689 | for _, t := range dst.Attr { |
| 690 | attr[t.Key] = t.Val |
| 691 | } |
| 692 | for _, t := range src.Attr { |
| 693 | if _, ok := attr[t.Key]; !ok { |
| 694 | dst.Attr = append(dst.Attr, t) |
| 695 | attr[t.Key] = t.Val |
| 696 | } |
| 697 | } |
| 698 | } |
| 699 | |
| 700 | // Section 12.2.5.4.7. |
| 701 | func inBodyIM(p *parser) bool { |
| 702 | switch p.tok.Type { |
| 703 | case TextToken: |
| 704 | d := p.tok.Data |
| 705 | switch n := p.oe.top(); n.DataAtom { |
| 706 | case a.Pre, a.Listing: |
| 707 | if n.FirstChild == nil { |
| 708 | // Ignore a newline at the start of a <pre> block. |
| 709 | if d != "" && d[0] == '\r' { |
| 710 | d = d[1:] |
| 711 | } |
| 712 | if d != "" && d[0] == '\n' { |
| 713 | d = d[1:] |
| 714 | } |
| 715 | } |
| 716 | } |
| 717 | d = strings.Replace(d, "\x00", "", -1) |
| 718 | if d == "" { |
| 719 | return true |
| 720 | } |
| 721 | p.reconstructActiveFormattingElements() |
| 722 | p.addText(d) |
| 723 | if p.framesetOK && strings.TrimLeft(d, whitespace) != "" { |
| 724 | // There were non-whitespace characters inserted. |
| 725 | p.framesetOK = false |
| 726 | } |
| 727 | case StartTagToken: |
| 728 | switch p.tok.DataAtom { |
| 729 | case a.Html: |
| 730 | copyAttributes(p.oe[0], p.tok) |
| 731 | case a.Base, a.Basefont, a.Bgsound, a.Command, a.Link, a.Meta, a.Noframes, a.Script, a.Style, a.Title: |
| 732 | return inHeadIM(p) |
| 733 | case a.Body: |
| 734 | if len(p.oe) >= 2 { |
| 735 | body := p.oe[1] |
| 736 | if body.Type == ElementNode && body.DataAtom == a.Body { |
| 737 | p.framesetOK = false |
| 738 | copyAttributes(body, p.tok) |
| 739 | } |
| 740 | } |
| 741 | case a.Frameset: |
| 742 | if !p.framesetOK || len(p.oe) < 2 || p.oe[1].DataAtom != a.Body { |
| 743 | // Ignore the token. |
| 744 | return true |
| 745 | } |
| 746 | body := p.oe[1] |
| 747 | if body.Parent != nil { |
| 748 | body.Parent.RemoveChild(body) |
| 749 | } |
| 750 | p.oe = p.oe[:1] |
| 751 | p.addElement() |
| 752 | p.im = inFramesetIM |
| 753 | return true |
| 754 | case a.Address, a.Article, a.Aside, a.Blockquote, a.Center, a.Details, a.Dir, a.Div, a.Dl, a.Fieldset, a.Figcaption, a.Figure, a.Footer, a.Header, a.Hgroup, a.Menu, a.Nav, a.Ol, a.P, a.Section, a.Summary, a.Ul: |
| 755 | p.popUntil(buttonScope, a.P) |
| 756 | p.addElement() |
| 757 | case a.H1, a.H2, a.H3, a.H4, a.H5, a.H6: |
| 758 | p.popUntil(buttonScope, a.P) |
| 759 | switch n := p.top(); n.DataAtom { |
| 760 | case a.H1, a.H2, a.H3, a.H4, a.H5, a.H6: |
| 761 | p.oe.pop() |
| 762 | } |
| 763 | p.addElement() |
| 764 | case a.Pre, a.Listing: |
| 765 | p.popUntil(buttonScope, a.P) |
| 766 | p.addElement() |
| 767 | // The newline, if any, will be dealt with by the TextToken case. |
| 768 | p.framesetOK = false |
| 769 | case a.Form: |
| 770 | if p.form == nil { |
| 771 | p.popUntil(buttonScope, a.P) |
| 772 | p.addElement() |
| 773 | p.form = p.top() |
| 774 | } |
| 775 | case a.Li: |
| 776 | p.framesetOK = false |
| 777 | for i := len(p.oe) - 1; i >= 0; i-- { |
| 778 | node := p.oe[i] |
| 779 | switch node.DataAtom { |
| 780 | case a.Li: |
| 781 | p.oe = p.oe[:i] |
| 782 | case a.Address, a.Div, a.P: |
| 783 | continue |
| 784 | default: |
| 785 | if !isSpecialElement(node) { |
| 786 | continue |
| 787 | } |
| 788 | } |
| 789 | break |
| 790 | } |
| 791 | p.popUntil(buttonScope, a.P) |
| 792 | p.addElement() |
| 793 | case a.Dd, a.Dt: |
| 794 | p.framesetOK = false |
| 795 | for i := len(p.oe) - 1; i >= 0; i-- { |
| 796 | node := p.oe[i] |
| 797 | switch node.DataAtom { |
| 798 | case a.Dd, a.Dt: |
| 799 | p.oe = p.oe[:i] |
| 800 | case a.Address, a.Div, a.P: |
| 801 | continue |
| 802 | default: |
| 803 | if !isSpecialElement(node) { |
| 804 | continue |
| 805 | } |
| 806 | } |
| 807 | break |
| 808 | } |
| 809 | p.popUntil(buttonScope, a.P) |
| 810 | p.addElement() |
| 811 | case a.Plaintext: |
| 812 | p.popUntil(buttonScope, a.P) |
| 813 | p.addElement() |
| 814 | case a.Button: |
| 815 | p.popUntil(defaultScope, a.Button) |
| 816 | p.reconstructActiveFormattingElements() |
| 817 | p.addElement() |
| 818 | p.framesetOK = false |
| 819 | case a.A: |
| 820 | for i := len(p.afe) - 1; i >= 0 && p.afe[i].Type != scopeMarkerNode; i-- { |
| 821 | if n := p.afe[i]; n.Type == ElementNode && n.DataAtom == a.A { |
| 822 | p.inBodyEndTagFormatting(a.A) |
| 823 | p.oe.remove(n) |
| 824 | p.afe.remove(n) |
| 825 | break |
| 826 | } |
| 827 | } |
| 828 | p.reconstructActiveFormattingElements() |
| 829 | p.addFormattingElement() |
| 830 | case a.B, a.Big, a.Code, a.Em, a.Font, a.I, a.S, a.Small, a.Strike, a.Strong, a.Tt, a.U: |
| 831 | p.reconstructActiveFormattingElements() |
| 832 | p.addFormattingElement() |
| 833 | case a.Nobr: |
| 834 | p.reconstructActiveFormattingElements() |
| 835 | if p.elementInScope(defaultScope, a.Nobr) { |
| 836 | p.inBodyEndTagFormatting(a.Nobr) |
| 837 | p.reconstructActiveFormattingElements() |
| 838 | } |
| 839 | p.addFormattingElement() |
| 840 | case a.Applet, a.Marquee, a.Object: |
| 841 | p.reconstructActiveFormattingElements() |
| 842 | p.addElement() |
| 843 | p.afe = append(p.afe, &scopeMarker) |
| 844 | p.framesetOK = false |
| 845 | case a.Table: |
| 846 | if !p.quirks { |
| 847 | p.popUntil(buttonScope, a.P) |
| 848 | } |
| 849 | p.addElement() |
| 850 | p.framesetOK = false |
| 851 | p.im = inTableIM |
| 852 | return true |
| 853 | case a.Area, a.Br, a.Embed, a.Img, a.Input, a.Keygen, a.Wbr: |
| 854 | p.reconstructActiveFormattingElements() |
| 855 | p.addElement() |
| 856 | p.oe.pop() |
| 857 | p.acknowledgeSelfClosingTag() |
| 858 | if p.tok.DataAtom == a.Input { |
| 859 | for _, t := range p.tok.Attr { |
| 860 | if t.Key == "type" { |
| 861 | if strings.ToLower(t.Val) == "hidden" { |
| 862 | // Skip setting framesetOK = false |
| 863 | return true |
| 864 | } |
| 865 | } |
| 866 | } |
| 867 | } |
| 868 | p.framesetOK = false |
| 869 | case a.Param, a.Source, a.Track: |
| 870 | p.addElement() |
| 871 | p.oe.pop() |
| 872 | p.acknowledgeSelfClosingTag() |
| 873 | case a.Hr: |
| 874 | p.popUntil(buttonScope, a.P) |
| 875 | p.addElement() |
| 876 | p.oe.pop() |
| 877 | p.acknowledgeSelfClosingTag() |
| 878 | p.framesetOK = false |
| 879 | case a.Image: |
| 880 | p.tok.DataAtom = a.Img |
| 881 | p.tok.Data = a.Img.String() |
| 882 | return false |
| 883 | case a.Isindex: |
| 884 | if p.form != nil { |
| 885 | // Ignore the token. |
| 886 | return true |
| 887 | } |
| 888 | action := "" |
| 889 | prompt := "This is a searchable index. Enter search keywords: " |
| 890 | attr := []Attribute{{Key: "name", Val: "isindex"}} |
| 891 | for _, t := range p.tok.Attr { |
| 892 | switch t.Key { |
| 893 | case "action": |
| 894 | action = t.Val |
| 895 | case "name": |
| 896 | // Ignore the attribute. |
| 897 | case "prompt": |
| 898 | prompt = t.Val |
| 899 | default: |
| 900 | attr = append(attr, t) |
| 901 | } |
| 902 | } |
| 903 | p.acknowledgeSelfClosingTag() |
| 904 | p.popUntil(buttonScope, a.P) |
| 905 | p.parseImpliedToken(StartTagToken, a.Form, a.Form.String()) |
| 906 | if action != "" { |
| 907 | p.form.Attr = []Attribute{{Key: "action", Val: action}} |
| 908 | } |
| 909 | p.parseImpliedToken(StartTagToken, a.Hr, a.Hr.String()) |
| 910 | p.parseImpliedToken(StartTagToken, a.Label, a.Label.String()) |
| 911 | p.addText(prompt) |
| 912 | p.addChild(&Node{ |
| 913 | Type: ElementNode, |
| 914 | DataAtom: a.Input, |
| 915 | Data: a.Input.String(), |
| 916 | Attr: attr, |
| 917 | }) |
| 918 | p.oe.pop() |
| 919 | p.parseImpliedToken(EndTagToken, a.Label, a.Label.String()) |
| 920 | p.parseImpliedToken(StartTagToken, a.Hr, a.Hr.String()) |
| 921 | p.parseImpliedToken(EndTagToken, a.Form, a.Form.String()) |
| 922 | case a.Textarea: |
| 923 | p.addElement() |
| 924 | p.setOriginalIM() |
| 925 | p.framesetOK = false |
| 926 | p.im = textIM |
| 927 | case a.Xmp: |
| 928 | p.popUntil(buttonScope, a.P) |
| 929 | p.reconstructActiveFormattingElements() |
| 930 | p.framesetOK = false |
| 931 | p.addElement() |
| 932 | p.setOriginalIM() |
| 933 | p.im = textIM |
| 934 | case a.Iframe: |
| 935 | p.framesetOK = false |
| 936 | p.addElement() |
| 937 | p.setOriginalIM() |
| 938 | p.im = textIM |
| 939 | case a.Noembed, a.Noscript: |
| 940 | p.addElement() |
| 941 | p.setOriginalIM() |
| 942 | p.im = textIM |
| 943 | case a.Select: |
| 944 | p.reconstructActiveFormattingElements() |
| 945 | p.addElement() |
| 946 | p.framesetOK = false |
| 947 | p.im = inSelectIM |
| 948 | return true |
| 949 | case a.Optgroup, a.Option: |
| 950 | if p.top().DataAtom == a.Option { |
| 951 | p.oe.pop() |
| 952 | } |
| 953 | p.reconstructActiveFormattingElements() |
| 954 | p.addElement() |
| 955 | case a.Rp, a.Rt: |
| 956 | if p.elementInScope(defaultScope, a.Ruby) { |
| 957 | p.generateImpliedEndTags() |
| 958 | } |
| 959 | p.addElement() |
| 960 | case a.Math, a.Svg: |
| 961 | p.reconstructActiveFormattingElements() |
| 962 | if p.tok.DataAtom == a.Math { |
| 963 | adjustAttributeNames(p.tok.Attr, mathMLAttributeAdjustments) |
| 964 | } else { |
| 965 | adjustAttributeNames(p.tok.Attr, svgAttributeAdjustments) |
| 966 | } |
| 967 | adjustForeignAttributes(p.tok.Attr) |
| 968 | p.addElement() |
| 969 | p.top().Namespace = p.tok.Data |
| 970 | if p.hasSelfClosingToken { |
| 971 | p.oe.pop() |
| 972 | p.acknowledgeSelfClosingTag() |
| 973 | } |
| 974 | return true |
| 975 | case a.Caption, a.Col, a.Colgroup, a.Frame, a.Head, a.Tbody, a.Td, a.Tfoot, a.Th, a.Thead, a.Tr: |
| 976 | // Ignore the token. |
| 977 | default: |
| 978 | p.reconstructActiveFormattingElements() |
| 979 | p.addElement() |
| 980 | } |
| 981 | case EndTagToken: |
| 982 | switch p.tok.DataAtom { |
| 983 | case a.Body: |
| 984 | if p.elementInScope(defaultScope, a.Body) { |
| 985 | p.im = afterBodyIM |
| 986 | } |
| 987 | case a.Html: |
| 988 | if p.elementInScope(defaultScope, a.Body) { |
| 989 | p.parseImpliedToken(EndTagToken, a.Body, a.Body.String()) |
| 990 | return false |
| 991 | } |
| 992 | return true |
| 993 | case a.Address, a.Article, a.Aside, a.Blockquote, a.Button, a.Center, a.Details, a.Dir, a.Div, a.Dl, a.Fieldset, a.Figcaption, a.Figure, a.Footer, a.Header, a.Hgroup, a.Listing, a.Menu, a.Nav, a.Ol, a.Pre, a.Section, a.Summary, a.Ul: |
| 994 | p.popUntil(defaultScope, p.tok.DataAtom) |
| 995 | case a.Form: |
| 996 | node := p.form |
| 997 | p.form = nil |
| 998 | i := p.indexOfElementInScope(defaultScope, a.Form) |
| 999 | if node == nil || i == -1 || p.oe[i] != node { |
| 1000 | // Ignore the token. |
| 1001 | return true |
| 1002 | } |
| 1003 | p.generateImpliedEndTags() |
| 1004 | p.oe.remove(node) |
| 1005 | case a.P: |
| 1006 | if !p.elementInScope(buttonScope, a.P) { |
| 1007 | p.parseImpliedToken(StartTagToken, a.P, a.P.String()) |
| 1008 | } |
| 1009 | p.popUntil(buttonScope, a.P) |
| 1010 | case a.Li: |
| 1011 | p.popUntil(listItemScope, a.Li) |
| 1012 | case a.Dd, a.Dt: |
| 1013 | p.popUntil(defaultScope, p.tok.DataAtom) |
| 1014 | case a.H1, a.H2, a.H3, a.H4, a.H5, a.H6: |
| 1015 | p.popUntil(defaultScope, a.H1, a.H2, a.H3, a.H4, a.H5, a.H6) |
| 1016 | case a.A, a.B, a.Big, a.Code, a.Em, a.Font, a.I, a.Nobr, a.S, a.Small, a.Strike, a.Strong, a.Tt, a.U: |
| 1017 | p.inBodyEndTagFormatting(p.tok.DataAtom) |
| 1018 | case a.Applet, a.Marquee, a.Object: |
| 1019 | if p.popUntil(defaultScope, p.tok.DataAtom) { |
| 1020 | p.clearActiveFormattingElements() |
| 1021 | } |
| 1022 | case a.Br: |
| 1023 | p.tok.Type = StartTagToken |
| 1024 | return false |
| 1025 | default: |
| 1026 | p.inBodyEndTagOther(p.tok.DataAtom) |
| 1027 | } |
| 1028 | case CommentToken: |
| 1029 | p.addChild(&Node{ |
| 1030 | Type: CommentNode, |
| 1031 | Data: p.tok.Data, |
| 1032 | }) |
| 1033 | } |
| 1034 | |
| 1035 | return true |
| 1036 | } |
| 1037 | |
| 1038 | func (p *parser) inBodyEndTagFormatting(tagAtom a.Atom) { |
| 1039 | // This is the "adoption agency" algorithm, described at |
| 1040 | // https://html.spec.whatwg.org/multipage/syntax.html#adoptionAgency |
| 1041 | |
| 1042 | // TODO: this is a fairly literal line-by-line translation of that algorithm. |
| 1043 | // Once the code successfully parses the comprehensive test suite, we should |
| 1044 | // refactor this code to be more idiomatic. |
| 1045 | |
| 1046 | // Steps 1-4. The outer loop. |
| 1047 | for i := 0; i < 8; i++ { |
| 1048 | // Step 5. Find the formatting element. |
| 1049 | var formattingElement *Node |
| 1050 | for j := len(p.afe) - 1; j >= 0; j-- { |
| 1051 | if p.afe[j].Type == scopeMarkerNode { |
| 1052 | break |
| 1053 | } |
| 1054 | if p.afe[j].DataAtom == tagAtom { |
| 1055 | formattingElement = p.afe[j] |
| 1056 | break |
| 1057 | } |
| 1058 | } |
| 1059 | if formattingElement == nil { |
| 1060 | p.inBodyEndTagOther(tagAtom) |
| 1061 | return |
| 1062 | } |
| 1063 | feIndex := p.oe.index(formattingElement) |
| 1064 | if feIndex == -1 { |
| 1065 | p.afe.remove(formattingElement) |
| 1066 | return |
| 1067 | } |
| 1068 | if !p.elementInScope(defaultScope, tagAtom) { |
| 1069 | // Ignore the tag. |
| 1070 | return |
| 1071 | } |
| 1072 | |
| 1073 | // Steps 9-10. Find the furthest block. |
| 1074 | var furthestBlock *Node |
| 1075 | for _, e := range p.oe[feIndex:] { |
| 1076 | if isSpecialElement(e) { |
| 1077 | furthestBlock = e |
| 1078 | break |
| 1079 | } |
| 1080 | } |
| 1081 | if furthestBlock == nil { |
| 1082 | e := p.oe.pop() |
| 1083 | for e != formattingElement { |
| 1084 | e = p.oe.pop() |
| 1085 | } |
| 1086 | p.afe.remove(e) |
| 1087 | return |
| 1088 | } |
| 1089 | |
| 1090 | // Steps 11-12. Find the common ancestor and bookmark node. |
| 1091 | commonAncestor := p.oe[feIndex-1] |
| 1092 | bookmark := p.afe.index(formattingElement) |
| 1093 | |
| 1094 | // Step 13. The inner loop. Find the lastNode to reparent. |
| 1095 | lastNode := furthestBlock |
| 1096 | node := furthestBlock |
| 1097 | x := p.oe.index(node) |
| 1098 | // Steps 13.1-13.2 |
| 1099 | for j := 0; j < 3; j++ { |
| 1100 | // Step 13.3. |
| 1101 | x-- |
| 1102 | node = p.oe[x] |
| 1103 | // Step 13.4 - 13.5. |
| 1104 | if p.afe.index(node) == -1 { |
| 1105 | p.oe.remove(node) |
| 1106 | continue |
| 1107 | } |
| 1108 | // Step 13.6. |
| 1109 | if node == formattingElement { |
| 1110 | break |
| 1111 | } |
| 1112 | // Step 13.7. |
| 1113 | clone := node.clone() |
| 1114 | p.afe[p.afe.index(node)] = clone |
| 1115 | p.oe[p.oe.index(node)] = clone |
| 1116 | node = clone |
| 1117 | // Step 13.8. |
| 1118 | if lastNode == furthestBlock { |
| 1119 | bookmark = p.afe.index(node) + 1 |
| 1120 | } |
| 1121 | // Step 13.9. |
| 1122 | if lastNode.Parent != nil { |
| 1123 | lastNode.Parent.RemoveChild(lastNode) |
| 1124 | } |
| 1125 | node.AppendChild(lastNode) |
| 1126 | // Step 13.10. |
| 1127 | lastNode = node |
| 1128 | } |
| 1129 | |
| 1130 | // Step 14. Reparent lastNode to the common ancestor, |
| 1131 | // or for misnested table nodes, to the foster parent. |
| 1132 | if lastNode.Parent != nil { |
| 1133 | lastNode.Parent.RemoveChild(lastNode) |
| 1134 | } |
| 1135 | switch commonAncestor.DataAtom { |
| 1136 | case a.Table, a.Tbody, a.Tfoot, a.Thead, a.Tr: |
| 1137 | p.fosterParent(lastNode) |
| 1138 | default: |
| 1139 | commonAncestor.AppendChild(lastNode) |
| 1140 | } |
| 1141 | |
| 1142 | // Steps 15-17. Reparent nodes from the furthest block's children |
| 1143 | // to a clone of the formatting element. |
| 1144 | clone := formattingElement.clone() |
| 1145 | reparentChildren(clone, furthestBlock) |
| 1146 | furthestBlock.AppendChild(clone) |
| 1147 | |
| 1148 | // Step 18. Fix up the list of active formatting elements. |
| 1149 | if oldLoc := p.afe.index(formattingElement); oldLoc != -1 && oldLoc < bookmark { |
| 1150 | // Move the bookmark with the rest of the list. |
| 1151 | bookmark-- |
| 1152 | } |
| 1153 | p.afe.remove(formattingElement) |
| 1154 | p.afe.insert(bookmark, clone) |
| 1155 | |
| 1156 | // Step 19. Fix up the stack of open elements. |
| 1157 | p.oe.remove(formattingElement) |
| 1158 | p.oe.insert(p.oe.index(furthestBlock)+1, clone) |
| 1159 | } |
| 1160 | } |
| 1161 | |
| 1162 | // inBodyEndTagOther performs the "any other end tag" algorithm for inBodyIM. |
| 1163 | // "Any other end tag" handling from 12.2.5.5 The rules for parsing tokens in foreign content |
| 1164 | // https://html.spec.whatwg.org/multipage/syntax.html#parsing-main-inforeign |
| 1165 | func (p *parser) inBodyEndTagOther(tagAtom a.Atom) { |
| 1166 | for i := len(p.oe) - 1; i >= 0; i-- { |
| 1167 | if p.oe[i].DataAtom == tagAtom { |
| 1168 | p.oe = p.oe[:i] |
| 1169 | break |
| 1170 | } |
| 1171 | if isSpecialElement(p.oe[i]) { |
| 1172 | break |
| 1173 | } |
| 1174 | } |
| 1175 | } |
| 1176 | |
| 1177 | // Section 12.2.5.4.8. |
| 1178 | func textIM(p *parser) bool { |
| 1179 | switch p.tok.Type { |
| 1180 | case ErrorToken: |
| 1181 | p.oe.pop() |
| 1182 | case TextToken: |
| 1183 | d := p.tok.Data |
| 1184 | if n := p.oe.top(); n.DataAtom == a.Textarea && n.FirstChild == nil { |
| 1185 | // Ignore a newline at the start of a <textarea> block. |
| 1186 | if d != "" && d[0] == '\r' { |
| 1187 | d = d[1:] |
| 1188 | } |
| 1189 | if d != "" && d[0] == '\n' { |
| 1190 | d = d[1:] |
| 1191 | } |
| 1192 | } |
| 1193 | if d == "" { |
| 1194 | return true |
| 1195 | } |
| 1196 | p.addText(d) |
| 1197 | return true |
| 1198 | case EndTagToken: |
| 1199 | p.oe.pop() |
| 1200 | } |
| 1201 | p.im = p.originalIM |
| 1202 | p.originalIM = nil |
| 1203 | return p.tok.Type == EndTagToken |
| 1204 | } |
| 1205 | |
| 1206 | // Section 12.2.5.4.9. |
| 1207 | func inTableIM(p *parser) bool { |
| 1208 | switch p.tok.Type { |
| 1209 | case ErrorToken: |
| 1210 | // Stop parsing. |
| 1211 | return true |
| 1212 | case TextToken: |
| 1213 | p.tok.Data = strings.Replace(p.tok.Data, "\x00", "", -1) |
| 1214 | switch p.oe.top().DataAtom { |
| 1215 | case a.Table, a.Tbody, a.Tfoot, a.Thead, a.Tr: |
| 1216 | if strings.Trim(p.tok.Data, whitespace) == "" { |
| 1217 | p.addText(p.tok.Data) |
| 1218 | return true |
| 1219 | } |
| 1220 | } |
| 1221 | case StartTagToken: |
| 1222 | switch p.tok.DataAtom { |
| 1223 | case a.Caption: |
| 1224 | p.clearStackToContext(tableScope) |
| 1225 | p.afe = append(p.afe, &scopeMarker) |
| 1226 | p.addElement() |
| 1227 | p.im = inCaptionIM |
| 1228 | return true |
| 1229 | case a.Colgroup: |
| 1230 | p.clearStackToContext(tableScope) |
| 1231 | p.addElement() |
| 1232 | p.im = inColumnGroupIM |
| 1233 | return true |
| 1234 | case a.Col: |
| 1235 | p.parseImpliedToken(StartTagToken, a.Colgroup, a.Colgroup.String()) |
| 1236 | return false |
| 1237 | case a.Tbody, a.Tfoot, a.Thead: |
| 1238 | p.clearStackToContext(tableScope) |
| 1239 | p.addElement() |
| 1240 | p.im = inTableBodyIM |
| 1241 | return true |
| 1242 | case a.Td, a.Th, a.Tr: |
| 1243 | p.parseImpliedToken(StartTagToken, a.Tbody, a.Tbody.String()) |
| 1244 | return false |
| 1245 | case a.Table: |
| 1246 | if p.popUntil(tableScope, a.Table) { |
| 1247 | p.resetInsertionMode() |
| 1248 | return false |
| 1249 | } |
| 1250 | // Ignore the token. |
| 1251 | return true |
| 1252 | case a.Style, a.Script: |
| 1253 | return inHeadIM(p) |
| 1254 | case a.Input: |
| 1255 | for _, t := range p.tok.Attr { |
| 1256 | if t.Key == "type" && strings.ToLower(t.Val) == "hidden" { |
| 1257 | p.addElement() |
| 1258 | p.oe.pop() |
| 1259 | return true |
| 1260 | } |
| 1261 | } |
| 1262 | // Otherwise drop down to the default action. |
| 1263 | case a.Form: |
| 1264 | if p.form != nil { |
| 1265 | // Ignore the token. |
| 1266 | return true |
| 1267 | } |
| 1268 | p.addElement() |
| 1269 | p.form = p.oe.pop() |
| 1270 | case a.Select: |
| 1271 | p.reconstructActiveFormattingElements() |
| 1272 | switch p.top().DataAtom { |
| 1273 | case a.Table, a.Tbody, a.Tfoot, a.Thead, a.Tr: |
| 1274 | p.fosterParenting = true |
| 1275 | } |
| 1276 | p.addElement() |
| 1277 | p.fosterParenting = false |
| 1278 | p.framesetOK = false |
| 1279 | p.im = inSelectInTableIM |
| 1280 | return true |
| 1281 | } |
| 1282 | case EndTagToken: |
| 1283 | switch p.tok.DataAtom { |
| 1284 | case a.Table: |
| 1285 | if p.popUntil(tableScope, a.Table) { |
| 1286 | p.resetInsertionMode() |
| 1287 | return true |
| 1288 | } |
| 1289 | // Ignore the token. |
| 1290 | return true |
| 1291 | case a.Body, a.Caption, a.Col, a.Colgroup, a.Html, a.Tbody, a.Td, a.Tfoot, a.Th, a.Thead, a.Tr: |
| 1292 | // Ignore the token. |
| 1293 | return true |
| 1294 | } |
| 1295 | case CommentToken: |
| 1296 | p.addChild(&Node{ |
| 1297 | Type: CommentNode, |
| 1298 | Data: p.tok.Data, |
| 1299 | }) |
| 1300 | return true |
| 1301 | case DoctypeToken: |
| 1302 | // Ignore the token. |
| 1303 | return true |
| 1304 | } |
| 1305 | |
| 1306 | p.fosterParenting = true |
| 1307 | defer func() { p.fosterParenting = false }() |
| 1308 | |
| 1309 | return inBodyIM(p) |
| 1310 | } |
| 1311 | |
| 1312 | // Section 12.2.5.4.11. |
| 1313 | func inCaptionIM(p *parser) bool { |
| 1314 | switch p.tok.Type { |
| 1315 | case StartTagToken: |
| 1316 | switch p.tok.DataAtom { |
| 1317 | case a.Caption, a.Col, a.Colgroup, a.Tbody, a.Td, a.Tfoot, a.Thead, a.Tr: |
| 1318 | if p.popUntil(tableScope, a.Caption) { |
| 1319 | p.clearActiveFormattingElements() |
| 1320 | p.im = inTableIM |
| 1321 | return false |
| 1322 | } else { |
| 1323 | // Ignore the token. |
| 1324 | return true |
| 1325 | } |
| 1326 | case a.Select: |
| 1327 | p.reconstructActiveFormattingElements() |
| 1328 | p.addElement() |
| 1329 | p.framesetOK = false |
| 1330 | p.im = inSelectInTableIM |
| 1331 | return true |
| 1332 | } |
| 1333 | case EndTagToken: |
| 1334 | switch p.tok.DataAtom { |
| 1335 | case a.Caption: |
| 1336 | if p.popUntil(tableScope, a.Caption) { |
| 1337 | p.clearActiveFormattingElements() |
| 1338 | p.im = inTableIM |
| 1339 | } |
| 1340 | return true |
| 1341 | case a.Table: |
| 1342 | if p.popUntil(tableScope, a.Caption) { |
| 1343 | p.clearActiveFormattingElements() |
| 1344 | p.im = inTableIM |
| 1345 | return false |
| 1346 | } else { |
| 1347 | // Ignore the token. |
| 1348 | return true |
| 1349 | } |
| 1350 | case a.Body, a.Col, a.Colgroup, a.Html, a.Tbody, a.Td, a.Tfoot, a.Th, a.Thead, a.Tr: |
| 1351 | // Ignore the token. |
| 1352 | return true |
| 1353 | } |
| 1354 | } |
| 1355 | return inBodyIM(p) |
| 1356 | } |
| 1357 | |
| 1358 | // Section 12.2.5.4.12. |
| 1359 | func inColumnGroupIM(p *parser) bool { |
| 1360 | switch p.tok.Type { |
| 1361 | case TextToken: |
| 1362 | s := strings.TrimLeft(p.tok.Data, whitespace) |
| 1363 | if len(s) < len(p.tok.Data) { |
| 1364 | // Add the initial whitespace to the current node. |
| 1365 | p.addText(p.tok.Data[:len(p.tok.Data)-len(s)]) |
| 1366 | if s == "" { |
| 1367 | return true |
| 1368 | } |
| 1369 | p.tok.Data = s |
| 1370 | } |
| 1371 | case CommentToken: |
| 1372 | p.addChild(&Node{ |
| 1373 | Type: CommentNode, |
| 1374 | Data: p.tok.Data, |
| 1375 | }) |
| 1376 | return true |
| 1377 | case DoctypeToken: |
| 1378 | // Ignore the token. |
| 1379 | return true |
| 1380 | case StartTagToken: |
| 1381 | switch p.tok.DataAtom { |
| 1382 | case a.Html: |
| 1383 | return inBodyIM(p) |
| 1384 | case a.Col: |
| 1385 | p.addElement() |
| 1386 | p.oe.pop() |
| 1387 | p.acknowledgeSelfClosingTag() |
| 1388 | return true |
| 1389 | } |
| 1390 | case EndTagToken: |
| 1391 | switch p.tok.DataAtom { |
| 1392 | case a.Colgroup: |
| 1393 | if p.oe.top().DataAtom != a.Html { |
| 1394 | p.oe.pop() |
| 1395 | p.im = inTableIM |
| 1396 | } |
| 1397 | return true |
| 1398 | case a.Col: |
| 1399 | // Ignore the token. |
| 1400 | return true |
| 1401 | } |
| 1402 | } |
| 1403 | if p.oe.top().DataAtom != a.Html { |
| 1404 | p.oe.pop() |
| 1405 | p.im = inTableIM |
| 1406 | return false |
| 1407 | } |
| 1408 | return true |
| 1409 | } |
| 1410 | |
| 1411 | // Section 12.2.5.4.13. |
| 1412 | func inTableBodyIM(p *parser) bool { |
| 1413 | switch p.tok.Type { |
| 1414 | case StartTagToken: |
| 1415 | switch p.tok.DataAtom { |
| 1416 | case a.Tr: |
| 1417 | p.clearStackToContext(tableBodyScope) |
| 1418 | p.addElement() |
| 1419 | p.im = inRowIM |
| 1420 | return true |
| 1421 | case a.Td, a.Th: |
| 1422 | p.parseImpliedToken(StartTagToken, a.Tr, a.Tr.String()) |
| 1423 | return false |
| 1424 | case a.Caption, a.Col, a.Colgroup, a.Tbody, a.Tfoot, a.Thead: |
| 1425 | if p.popUntil(tableScope, a.Tbody, a.Thead, a.Tfoot) { |
| 1426 | p.im = inTableIM |
| 1427 | return false |
| 1428 | } |
| 1429 | // Ignore the token. |
| 1430 | return true |
| 1431 | } |
| 1432 | case EndTagToken: |
| 1433 | switch p.tok.DataAtom { |
| 1434 | case a.Tbody, a.Tfoot, a.Thead: |
| 1435 | if p.elementInScope(tableScope, p.tok.DataAtom) { |
| 1436 | p.clearStackToContext(tableBodyScope) |
| 1437 | p.oe.pop() |
| 1438 | p.im = inTableIM |
| 1439 | } |
| 1440 | return true |
| 1441 | case a.Table: |
| 1442 | if p.popUntil(tableScope, a.Tbody, a.Thead, a.Tfoot) { |
| 1443 | p.im = inTableIM |
| 1444 | return false |
| 1445 | } |
| 1446 | // Ignore the token. |
| 1447 | return true |
| 1448 | case a.Body, a.Caption, a.Col, a.Colgroup, a.Html, a.Td, a.Th, a.Tr: |
| 1449 | // Ignore the token. |
| 1450 | return true |
| 1451 | } |
| 1452 | case CommentToken: |
| 1453 | p.addChild(&Node{ |
| 1454 | Type: CommentNode, |
| 1455 | Data: p.tok.Data, |
| 1456 | }) |
| 1457 | return true |
| 1458 | } |
| 1459 | |
| 1460 | return inTableIM(p) |
| 1461 | } |
| 1462 | |
| 1463 | // Section 12.2.5.4.14. |
| 1464 | func inRowIM(p *parser) bool { |
| 1465 | switch p.tok.Type { |
| 1466 | case StartTagToken: |
| 1467 | switch p.tok.DataAtom { |
| 1468 | case a.Td, a.Th: |
| 1469 | p.clearStackToContext(tableRowScope) |
| 1470 | p.addElement() |
| 1471 | p.afe = append(p.afe, &scopeMarker) |
| 1472 | p.im = inCellIM |
| 1473 | return true |
| 1474 | case a.Caption, a.Col, a.Colgroup, a.Tbody, a.Tfoot, a.Thead, a.Tr: |
| 1475 | if p.popUntil(tableScope, a.Tr) { |
| 1476 | p.im = inTableBodyIM |
| 1477 | return false |
| 1478 | } |
| 1479 | // Ignore the token. |
| 1480 | return true |
| 1481 | } |
| 1482 | case EndTagToken: |
| 1483 | switch p.tok.DataAtom { |
| 1484 | case a.Tr: |
| 1485 | if p.popUntil(tableScope, a.Tr) { |
| 1486 | p.im = inTableBodyIM |
| 1487 | return true |
| 1488 | } |
| 1489 | // Ignore the token. |
| 1490 | return true |
| 1491 | case a.Table: |
| 1492 | if p.popUntil(tableScope, a.Tr) { |
| 1493 | p.im = inTableBodyIM |
| 1494 | return false |
| 1495 | } |
| 1496 | // Ignore the token. |
| 1497 | return true |
| 1498 | case a.Tbody, a.Tfoot, a.Thead: |
| 1499 | if p.elementInScope(tableScope, p.tok.DataAtom) { |
| 1500 | p.parseImpliedToken(EndTagToken, a.Tr, a.Tr.String()) |
| 1501 | return false |
| 1502 | } |
| 1503 | // Ignore the token. |
| 1504 | return true |
| 1505 | case a.Body, a.Caption, a.Col, a.Colgroup, a.Html, a.Td, a.Th: |
| 1506 | // Ignore the token. |
| 1507 | return true |
| 1508 | } |
| 1509 | } |
| 1510 | |
| 1511 | return inTableIM(p) |
| 1512 | } |
| 1513 | |
| 1514 | // Section 12.2.5.4.15. |
| 1515 | func inCellIM(p *parser) bool { |
| 1516 | switch p.tok.Type { |
| 1517 | case StartTagToken: |
| 1518 | switch p.tok.DataAtom { |
| 1519 | case a.Caption, a.Col, a.Colgroup, a.Tbody, a.Td, a.Tfoot, a.Th, a.Thead, a.Tr: |
| 1520 | if p.popUntil(tableScope, a.Td, a.Th) { |
| 1521 | // Close the cell and reprocess. |
| 1522 | p.clearActiveFormattingElements() |
| 1523 | p.im = inRowIM |
| 1524 | return false |
| 1525 | } |
| 1526 | // Ignore the token. |
| 1527 | return true |
| 1528 | case a.Select: |
| 1529 | p.reconstructActiveFormattingElements() |
| 1530 | p.addElement() |
| 1531 | p.framesetOK = false |
| 1532 | p.im = inSelectInTableIM |
| 1533 | return true |
| 1534 | } |
| 1535 | case EndTagToken: |
| 1536 | switch p.tok.DataAtom { |
| 1537 | case a.Td, a.Th: |
| 1538 | if !p.popUntil(tableScope, p.tok.DataAtom) { |
| 1539 | // Ignore the token. |
| 1540 | return true |
| 1541 | } |
| 1542 | p.clearActiveFormattingElements() |
| 1543 | p.im = inRowIM |
| 1544 | return true |
| 1545 | case a.Body, a.Caption, a.Col, a.Colgroup, a.Html: |
| 1546 | // Ignore the token. |
| 1547 | return true |
| 1548 | case a.Table, a.Tbody, a.Tfoot, a.Thead, a.Tr: |
| 1549 | if !p.elementInScope(tableScope, p.tok.DataAtom) { |
| 1550 | // Ignore the token. |
| 1551 | return true |
| 1552 | } |
| 1553 | // Close the cell and reprocess. |
| 1554 | p.popUntil(tableScope, a.Td, a.Th) |
| 1555 | p.clearActiveFormattingElements() |
| 1556 | p.im = inRowIM |
| 1557 | return false |
| 1558 | } |
| 1559 | } |
| 1560 | return inBodyIM(p) |
| 1561 | } |
| 1562 | |
| 1563 | // Section 12.2.5.4.16. |
| 1564 | func inSelectIM(p *parser) bool { |
| 1565 | switch p.tok.Type { |
| 1566 | case ErrorToken: |
| 1567 | // Stop parsing. |
| 1568 | return true |
| 1569 | case TextToken: |
| 1570 | p.addText(strings.Replace(p.tok.Data, "\x00", "", -1)) |
| 1571 | case StartTagToken: |
| 1572 | switch p.tok.DataAtom { |
| 1573 | case a.Html: |
| 1574 | return inBodyIM(p) |
| 1575 | case a.Option: |
| 1576 | if p.top().DataAtom == a.Option { |
| 1577 | p.oe.pop() |
| 1578 | } |
| 1579 | p.addElement() |
| 1580 | case a.Optgroup: |
| 1581 | if p.top().DataAtom == a.Option { |
| 1582 | p.oe.pop() |
| 1583 | } |
| 1584 | if p.top().DataAtom == a.Optgroup { |
| 1585 | p.oe.pop() |
| 1586 | } |
| 1587 | p.addElement() |
| 1588 | case a.Select: |
| 1589 | p.tok.Type = EndTagToken |
| 1590 | return false |
| 1591 | case a.Input, a.Keygen, a.Textarea: |
| 1592 | if p.elementInScope(selectScope, a.Select) { |
| 1593 | p.parseImpliedToken(EndTagToken, a.Select, a.Select.String()) |
| 1594 | return false |
| 1595 | } |
| 1596 | // In order to properly ignore <textarea>, we need to change the tokenizer mode. |
| 1597 | p.tokenizer.NextIsNotRawText() |
| 1598 | // Ignore the token. |
| 1599 | return true |
| 1600 | case a.Script: |
| 1601 | return inHeadIM(p) |
| 1602 | } |
| 1603 | case EndTagToken: |
| 1604 | switch p.tok.DataAtom { |
| 1605 | case a.Option: |
| 1606 | if p.top().DataAtom == a.Option { |
| 1607 | p.oe.pop() |
| 1608 | } |
| 1609 | case a.Optgroup: |
| 1610 | i := len(p.oe) - 1 |
| 1611 | if p.oe[i].DataAtom == a.Option { |
| 1612 | i-- |
| 1613 | } |
| 1614 | if p.oe[i].DataAtom == a.Optgroup { |
| 1615 | p.oe = p.oe[:i] |
| 1616 | } |
| 1617 | case a.Select: |
| 1618 | if p.popUntil(selectScope, a.Select) { |
| 1619 | p.resetInsertionMode() |
| 1620 | } |
| 1621 | } |
| 1622 | case CommentToken: |
| 1623 | p.addChild(&Node{ |
| 1624 | Type: CommentNode, |
| 1625 | Data: p.tok.Data, |
| 1626 | }) |
| 1627 | case DoctypeToken: |
| 1628 | // Ignore the token. |
| 1629 | return true |
| 1630 | } |
| 1631 | |
| 1632 | return true |
| 1633 | } |
| 1634 | |
| 1635 | // Section 12.2.5.4.17. |
| 1636 | func inSelectInTableIM(p *parser) bool { |
| 1637 | switch p.tok.Type { |
| 1638 | case StartTagToken, EndTagToken: |
| 1639 | switch p.tok.DataAtom { |
| 1640 | case a.Caption, a.Table, a.Tbody, a.Tfoot, a.Thead, a.Tr, a.Td, a.Th: |
| 1641 | if p.tok.Type == StartTagToken || p.elementInScope(tableScope, p.tok.DataAtom) { |
| 1642 | p.parseImpliedToken(EndTagToken, a.Select, a.Select.String()) |
| 1643 | return false |
| 1644 | } else { |
| 1645 | // Ignore the token. |
| 1646 | return true |
| 1647 | } |
| 1648 | } |
| 1649 | } |
| 1650 | return inSelectIM(p) |
| 1651 | } |
| 1652 | |
| 1653 | // Section 12.2.5.4.18. |
| 1654 | func afterBodyIM(p *parser) bool { |
| 1655 | switch p.tok.Type { |
| 1656 | case ErrorToken: |
| 1657 | // Stop parsing. |
| 1658 | return true |
| 1659 | case TextToken: |
| 1660 | s := strings.TrimLeft(p.tok.Data, whitespace) |
| 1661 | if len(s) == 0 { |
| 1662 | // It was all whitespace. |
| 1663 | return inBodyIM(p) |
| 1664 | } |
| 1665 | case StartTagToken: |
| 1666 | if p.tok.DataAtom == a.Html { |
| 1667 | return inBodyIM(p) |
| 1668 | } |
| 1669 | case EndTagToken: |
| 1670 | if p.tok.DataAtom == a.Html { |
| 1671 | if !p.fragment { |
| 1672 | p.im = afterAfterBodyIM |
| 1673 | } |
| 1674 | return true |
| 1675 | } |
| 1676 | case CommentToken: |
| 1677 | // The comment is attached to the <html> element. |
| 1678 | if len(p.oe) < 1 || p.oe[0].DataAtom != a.Html { |
| 1679 | panic("html: bad parser state: <html> element not found, in the after-body insertion mode") |
| 1680 | } |
| 1681 | p.oe[0].AppendChild(&Node{ |
| 1682 | Type: CommentNode, |
| 1683 | Data: p.tok.Data, |
| 1684 | }) |
| 1685 | return true |
| 1686 | } |
| 1687 | p.im = inBodyIM |
| 1688 | return false |
| 1689 | } |
| 1690 | |
| 1691 | // Section 12.2.5.4.19. |
| 1692 | func inFramesetIM(p *parser) bool { |
| 1693 | switch p.tok.Type { |
| 1694 | case CommentToken: |
| 1695 | p.addChild(&Node{ |
| 1696 | Type: CommentNode, |
| 1697 | Data: p.tok.Data, |
| 1698 | }) |
| 1699 | case TextToken: |
| 1700 | // Ignore all text but whitespace. |
| 1701 | s := strings.Map(func(c rune) rune { |
| 1702 | switch c { |
| 1703 | case ' ', '\t', '\n', '\f', '\r': |
| 1704 | return c |
| 1705 | } |
| 1706 | return -1 |
| 1707 | }, p.tok.Data) |
| 1708 | if s != "" { |
| 1709 | p.addText(s) |
| 1710 | } |
| 1711 | case StartTagToken: |
| 1712 | switch p.tok.DataAtom { |
| 1713 | case a.Html: |
| 1714 | return inBodyIM(p) |
| 1715 | case a.Frameset: |
| 1716 | p.addElement() |
| 1717 | case a.Frame: |
| 1718 | p.addElement() |
| 1719 | p.oe.pop() |
| 1720 | p.acknowledgeSelfClosingTag() |
| 1721 | case a.Noframes: |
| 1722 | return inHeadIM(p) |
| 1723 | } |
| 1724 | case EndTagToken: |
| 1725 | switch p.tok.DataAtom { |
| 1726 | case a.Frameset: |
| 1727 | if p.oe.top().DataAtom != a.Html { |
| 1728 | p.oe.pop() |
| 1729 | if p.oe.top().DataAtom != a.Frameset { |
| 1730 | p.im = afterFramesetIM |
| 1731 | return true |
| 1732 | } |
| 1733 | } |
| 1734 | } |
| 1735 | default: |
| 1736 | // Ignore the token. |
| 1737 | } |
| 1738 | return true |
| 1739 | } |
| 1740 | |
| 1741 | // Section 12.2.5.4.20. |
| 1742 | func afterFramesetIM(p *parser) bool { |
| 1743 | switch p.tok.Type { |
| 1744 | case CommentToken: |
| 1745 | p.addChild(&Node{ |
| 1746 | Type: CommentNode, |
| 1747 | Data: p.tok.Data, |
| 1748 | }) |
| 1749 | case TextToken: |
| 1750 | // Ignore all text but whitespace. |
| 1751 | s := strings.Map(func(c rune) rune { |
| 1752 | switch c { |
| 1753 | case ' ', '\t', '\n', '\f', '\r': |
| 1754 | return c |
| 1755 | } |
| 1756 | return -1 |
| 1757 | }, p.tok.Data) |
| 1758 | if s != "" { |
| 1759 | p.addText(s) |
| 1760 | } |
| 1761 | case StartTagToken: |
| 1762 | switch p.tok.DataAtom { |
| 1763 | case a.Html: |
| 1764 | return inBodyIM(p) |
| 1765 | case a.Noframes: |
| 1766 | return inHeadIM(p) |
| 1767 | } |
| 1768 | case EndTagToken: |
| 1769 | switch p.tok.DataAtom { |
| 1770 | case a.Html: |
| 1771 | p.im = afterAfterFramesetIM |
| 1772 | return true |
| 1773 | } |
| 1774 | default: |
| 1775 | // Ignore the token. |
| 1776 | } |
| 1777 | return true |
| 1778 | } |
| 1779 | |
| 1780 | // Section 12.2.5.4.21. |
| 1781 | func afterAfterBodyIM(p *parser) bool { |
| 1782 | switch p.tok.Type { |
| 1783 | case ErrorToken: |
| 1784 | // Stop parsing. |
| 1785 | return true |
| 1786 | case TextToken: |
| 1787 | s := strings.TrimLeft(p.tok.Data, whitespace) |
| 1788 | if len(s) == 0 { |
| 1789 | // It was all whitespace. |
| 1790 | return inBodyIM(p) |
| 1791 | } |
| 1792 | case StartTagToken: |
| 1793 | if p.tok.DataAtom == a.Html { |
| 1794 | return inBodyIM(p) |
| 1795 | } |
| 1796 | case CommentToken: |
| 1797 | p.doc.AppendChild(&Node{ |
| 1798 | Type: CommentNode, |
| 1799 | Data: p.tok.Data, |
| 1800 | }) |
| 1801 | return true |
| 1802 | case DoctypeToken: |
| 1803 | return inBodyIM(p) |
| 1804 | } |
| 1805 | p.im = inBodyIM |
| 1806 | return false |
| 1807 | } |
| 1808 | |
| 1809 | // Section 12.2.5.4.22. |
| 1810 | func afterAfterFramesetIM(p *parser) bool { |
| 1811 | switch p.tok.Type { |
| 1812 | case CommentToken: |
| 1813 | p.doc.AppendChild(&Node{ |
| 1814 | Type: CommentNode, |
| 1815 | Data: p.tok.Data, |
| 1816 | }) |
| 1817 | case TextToken: |
| 1818 | // Ignore all text but whitespace. |
| 1819 | s := strings.Map(func(c rune) rune { |
| 1820 | switch c { |
| 1821 | case ' ', '\t', '\n', '\f', '\r': |
| 1822 | return c |
| 1823 | } |
| 1824 | return -1 |
| 1825 | }, p.tok.Data) |
| 1826 | if s != "" { |
| 1827 | p.tok.Data = s |
| 1828 | return inBodyIM(p) |
| 1829 | } |
| 1830 | case StartTagToken: |
| 1831 | switch p.tok.DataAtom { |
| 1832 | case a.Html: |
| 1833 | return inBodyIM(p) |
| 1834 | case a.Noframes: |
| 1835 | return inHeadIM(p) |
| 1836 | } |
| 1837 | case DoctypeToken: |
| 1838 | return inBodyIM(p) |
| 1839 | default: |
| 1840 | // Ignore the token. |
| 1841 | } |
| 1842 | return true |
| 1843 | } |
| 1844 | |
| 1845 | const whitespaceOrNUL = whitespace + "\x00" |
| 1846 | |
| 1847 | // Section 12.2.5.5. |
| 1848 | func parseForeignContent(p *parser) bool { |
| 1849 | switch p.tok.Type { |
| 1850 | case TextToken: |
| 1851 | if p.framesetOK { |
| 1852 | p.framesetOK = strings.TrimLeft(p.tok.Data, whitespaceOrNUL) == "" |
| 1853 | } |
| 1854 | p.tok.Data = strings.Replace(p.tok.Data, "\x00", "\ufffd", -1) |
| 1855 | p.addText(p.tok.Data) |
| 1856 | case CommentToken: |
| 1857 | p.addChild(&Node{ |
| 1858 | Type: CommentNode, |
| 1859 | Data: p.tok.Data, |
| 1860 | }) |
| 1861 | case StartTagToken: |
| 1862 | b := breakout[p.tok.Data] |
| 1863 | if p.tok.DataAtom == a.Font { |
| 1864 | loop: |
| 1865 | for _, attr := range p.tok.Attr { |
| 1866 | switch attr.Key { |
| 1867 | case "color", "face", "size": |
| 1868 | b = true |
| 1869 | break loop |
| 1870 | } |
| 1871 | } |
| 1872 | } |
| 1873 | if b { |
| 1874 | for i := len(p.oe) - 1; i >= 0; i-- { |
| 1875 | n := p.oe[i] |
| 1876 | if n.Namespace == "" || htmlIntegrationPoint(n) || mathMLTextIntegrationPoint(n) { |
| 1877 | p.oe = p.oe[:i+1] |
| 1878 | break |
| 1879 | } |
| 1880 | } |
| 1881 | return false |
| 1882 | } |
| 1883 | switch p.top().Namespace { |
| 1884 | case "math": |
| 1885 | adjustAttributeNames(p.tok.Attr, mathMLAttributeAdjustments) |
| 1886 | case "svg": |
| 1887 | // Adjust SVG tag names. The tokenizer lower-cases tag names, but |
| 1888 | // SVG wants e.g. "foreignObject" with a capital second "O". |
| 1889 | if x := svgTagNameAdjustments[p.tok.Data]; x != "" { |
| 1890 | p.tok.DataAtom = a.Lookup([]byte(x)) |
| 1891 | p.tok.Data = x |
| 1892 | } |
| 1893 | adjustAttributeNames(p.tok.Attr, svgAttributeAdjustments) |
| 1894 | default: |
| 1895 | panic("html: bad parser state: unexpected namespace") |
| 1896 | } |
| 1897 | adjustForeignAttributes(p.tok.Attr) |
| 1898 | namespace := p.top().Namespace |
| 1899 | p.addElement() |
| 1900 | p.top().Namespace = namespace |
| 1901 | if namespace != "" { |
| 1902 | // Don't let the tokenizer go into raw text mode in foreign content |
| 1903 | // (e.g. in an SVG <title> tag). |
| 1904 | p.tokenizer.NextIsNotRawText() |
| 1905 | } |
| 1906 | if p.hasSelfClosingToken { |
| 1907 | p.oe.pop() |
| 1908 | p.acknowledgeSelfClosingTag() |
| 1909 | } |
| 1910 | case EndTagToken: |
| 1911 | for i := len(p.oe) - 1; i >= 0; i-- { |
| 1912 | if p.oe[i].Namespace == "" { |
| 1913 | return p.im(p) |
| 1914 | } |
| 1915 | if strings.EqualFold(p.oe[i].Data, p.tok.Data) { |
| 1916 | p.oe = p.oe[:i] |
| 1917 | break |
| 1918 | } |
| 1919 | } |
| 1920 | return true |
| 1921 | default: |
| 1922 | // Ignore the token. |
| 1923 | } |
| 1924 | return true |
| 1925 | } |
| 1926 | |
| 1927 | // Section 12.2.5. |
| 1928 | func (p *parser) inForeignContent() bool { |
| 1929 | if len(p.oe) == 0 { |
| 1930 | return false |
| 1931 | } |
| 1932 | n := p.oe[len(p.oe)-1] |
| 1933 | if n.Namespace == "" { |
| 1934 | return false |
| 1935 | } |
| 1936 | if mathMLTextIntegrationPoint(n) { |
| 1937 | if p.tok.Type == StartTagToken && p.tok.DataAtom != a.Mglyph && p.tok.DataAtom != a.Malignmark { |
| 1938 | return false |
| 1939 | } |
| 1940 | if p.tok.Type == TextToken { |
| 1941 | return false |
| 1942 | } |
| 1943 | } |
| 1944 | if n.Namespace == "math" && n.DataAtom == a.AnnotationXml && p.tok.Type == StartTagToken && p.tok.DataAtom == a.Svg { |
| 1945 | return false |
| 1946 | } |
| 1947 | if htmlIntegrationPoint(n) && (p.tok.Type == StartTagToken || p.tok.Type == TextToken) { |
| 1948 | return false |
| 1949 | } |
| 1950 | if p.tok.Type == ErrorToken { |
| 1951 | return false |
| 1952 | } |
| 1953 | return true |
| 1954 | } |
| 1955 | |
| 1956 | // parseImpliedToken parses a token as though it had appeared in the parser's |
| 1957 | // input. |
| 1958 | func (p *parser) parseImpliedToken(t TokenType, dataAtom a.Atom, data string) { |
| 1959 | realToken, selfClosing := p.tok, p.hasSelfClosingToken |
| 1960 | p.tok = Token{ |
| 1961 | Type: t, |
| 1962 | DataAtom: dataAtom, |
| 1963 | Data: data, |
| 1964 | } |
| 1965 | p.hasSelfClosingToken = false |
| 1966 | p.parseCurrentToken() |
| 1967 | p.tok, p.hasSelfClosingToken = realToken, selfClosing |
| 1968 | } |
| 1969 | |
| 1970 | // parseCurrentToken runs the current token through the parsing routines |
| 1971 | // until it is consumed. |
| 1972 | func (p *parser) parseCurrentToken() { |
| 1973 | if p.tok.Type == SelfClosingTagToken { |
| 1974 | p.hasSelfClosingToken = true |
| 1975 | p.tok.Type = StartTagToken |
| 1976 | } |
| 1977 | |
| 1978 | consumed := false |
| 1979 | for !consumed { |
| 1980 | if p.inForeignContent() { |
| 1981 | consumed = parseForeignContent(p) |
| 1982 | } else { |
| 1983 | consumed = p.im(p) |
| 1984 | } |
| 1985 | } |
| 1986 | |
| 1987 | if p.hasSelfClosingToken { |
| 1988 | // This is a parse error, but ignore it. |
| 1989 | p.hasSelfClosingToken = false |
| 1990 | } |
| 1991 | } |
| 1992 | |
| 1993 | func (p *parser) parse() error { |
| 1994 | // Iterate until EOF. Any other error will cause an early return. |
| 1995 | var err error |
| 1996 | for err != io.EOF { |
| 1997 | // CDATA sections are allowed only in foreign content. |
| 1998 | n := p.oe.top() |
| 1999 | p.tokenizer.AllowCDATA(n != nil && n.Namespace != "") |
| 2000 | // Read and parse the next token. |
| 2001 | p.tokenizer.Next() |
| 2002 | p.tok = p.tokenizer.Token() |
| 2003 | if p.tok.Type == ErrorToken { |
| 2004 | err = p.tokenizer.Err() |
| 2005 | if err != nil && err != io.EOF { |
| 2006 | return err |
| 2007 | } |
| 2008 | } |
| 2009 | p.parseCurrentToken() |
| 2010 | } |
| 2011 | return nil |
| 2012 | } |
| 2013 | |
| 2014 | // Parse returns the parse tree for the HTML from the given Reader. |
| 2015 | // The input is assumed to be UTF-8 encoded. |
| 2016 | func Parse(r io.Reader) (*Node, error) { |
| 2017 | p := &parser{ |
| 2018 | tokenizer: NewTokenizer(r), |
| 2019 | doc: &Node{ |
| 2020 | Type: DocumentNode, |
| 2021 | }, |
| 2022 | scripting: true, |
| 2023 | framesetOK: true, |
| 2024 | im: initialIM, |
| 2025 | } |
| 2026 | err := p.parse() |
| 2027 | if err != nil { |
| 2028 | return nil, err |
| 2029 | } |
| 2030 | return p.doc, nil |
| 2031 | } |
| 2032 | |
| 2033 | // ParseFragment parses a fragment of HTML and returns the nodes that were |
| 2034 | // found. If the fragment is the InnerHTML for an existing element, pass that |
| 2035 | // element in context. |
| 2036 | func ParseFragment(r io.Reader, context *Node) ([]*Node, error) { |
| 2037 | contextTag := "" |
| 2038 | if context != nil { |
| 2039 | if context.Type != ElementNode { |
| 2040 | return nil, errors.New("html: ParseFragment of non-element Node") |
| 2041 | } |
| 2042 | // The next check isn't just context.DataAtom.String() == context.Data because |
| 2043 | // it is valid to pass an element whose tag isn't a known atom. For example, |
| 2044 | // DataAtom == 0 and Data = "tagfromthefuture" is perfectly consistent. |
| 2045 | if context.DataAtom != a.Lookup([]byte(context.Data)) { |
| 2046 | return nil, fmt.Errorf("html: inconsistent Node: DataAtom=%q, Data=%q", context.DataAtom, context.Data) |
| 2047 | } |
| 2048 | contextTag = context.DataAtom.String() |
| 2049 | } |
| 2050 | p := &parser{ |
| 2051 | tokenizer: NewTokenizerFragment(r, contextTag), |
| 2052 | doc: &Node{ |
| 2053 | Type: DocumentNode, |
| 2054 | }, |
| 2055 | scripting: true, |
| 2056 | fragment: true, |
| 2057 | context: context, |
| 2058 | } |
| 2059 | |
| 2060 | root := &Node{ |
| 2061 | Type: ElementNode, |
| 2062 | DataAtom: a.Html, |
| 2063 | Data: a.Html.String(), |
| 2064 | } |
| 2065 | p.doc.AppendChild(root) |
| 2066 | p.oe = nodeStack{root} |
| 2067 | p.resetInsertionMode() |
| 2068 | |
| 2069 | for n := context; n != nil; n = n.Parent { |
| 2070 | if n.Type == ElementNode && n.DataAtom == a.Form { |
| 2071 | p.form = n |
| 2072 | break |
| 2073 | } |
| 2074 | } |
| 2075 | |
| 2076 | err := p.parse() |
| 2077 | if err != nil { |
| 2078 | return nil, err |
| 2079 | } |
| 2080 | |
| 2081 | parent := p.doc |
| 2082 | if context != nil { |
| 2083 | parent = root |
| 2084 | } |
| 2085 | |
| 2086 | var result []*Node |
| 2087 | for c := parent.FirstChild; c != nil; { |
| 2088 | next := c.NextSibling |
| 2089 | parent.RemoveChild(c) |
| 2090 | result = append(result, c) |
| 2091 | c = next |
| 2092 | } |
| 2093 | return result, nil |
| 2094 | } |