]> git.proxmox.com Git - rustc.git/blame - src/vendor/html5ever/src/tree_builder/mod.rs
New upstream version 1.25.0+dfsg1
[rustc.git] / src / vendor / html5ever / src / tree_builder / mod.rs
CommitLineData
ea8adc8c
XL
1// Copyright 2014-2017 The html5ever Project Developers. See the
2// COPYRIGHT file at the top-level directory of this distribution.
3//
4// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
5// http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
6// <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
7// option. This file may not be copied, modified, or distributed
8// except according to those terms.
9
10#![allow(warnings)]
11
12//! The HTML5 tree builder.
13
14pub use interface::{QuirksMode, Quirks, LimitedQuirks, NoQuirks};
ff7c6d11 15pub use interface::{NodeOrText, AppendNode, AppendText, Attribute};
ea8adc8c
XL
16pub use interface::{TreeSink, Tracer, NextParserState, create_element, ElementFlags};
17
18use self::types::*;
ea8adc8c 19
ff7c6d11 20use {ExpandedName, QualName, LocalName, Namespace};
ea8adc8c
XL
21use tendril::StrTendril;
22
23use tokenizer;
ff7c6d11 24use tokenizer::{Doctype, StartTag, Tag, EndTag, TokenSink, TokenSinkResult};
ea8adc8c
XL
25use tokenizer::states as tok_state;
26
27use util::str::is_ascii_whitespace;
28
ff7c6d11
XL
29use std::{slice, fmt};
30use std::ascii::AsciiExt;
ea8adc8c
XL
31use std::borrow::Cow::Borrowed;
32use std::collections::VecDeque;
ff7c6d11
XL
33use std::default::Default;
34use std::iter::{Rev, Enumerate};
35use std::mem::replace;
36
37use tokenizer::states::{RawData, RawKind};
38use tree_builder::types::*;
39use tree_builder::tag_sets::*;
40use util::str::to_escaped_string;
41
42pub use self::PushFlag::*;
ea8adc8c
XL
43
44#[macro_use] mod tag_sets;
45
46mod data;
47mod types;
ea8adc8c 48
ff7c6d11 49include!(concat!(env!("OUT_DIR"), "/rules.rs"));
ea8adc8c
XL
50
51/// Tree builder options, with an impl for Default.
52#[derive(Copy, Clone)]
53pub struct TreeBuilderOpts {
54 /// Report all parse errors described in the spec, at some
55 /// performance penalty? Default: false
56 pub exact_errors: bool,
57
58 /// Is scripting enabled?
59 pub scripting_enabled: bool,
60
61 /// Is this an `iframe srcdoc` document?
62 pub iframe_srcdoc: bool,
63
64 /// Should we drop the DOCTYPE (if any) from the tree?
65 pub drop_doctype: bool,
66
67 /// Obsolete, ignored.
68 pub ignore_missing_rules: bool,
69
70 /// Initial TreeBuilder quirks mode. Default: NoQuirks
71 pub quirks_mode: QuirksMode,
72}
73
74impl Default for TreeBuilderOpts {
75 fn default() -> TreeBuilderOpts {
76 TreeBuilderOpts {
77 exact_errors: false,
78 scripting_enabled: true,
79 iframe_srcdoc: false,
80 drop_doctype: false,
81 ignore_missing_rules: false,
82 quirks_mode: NoQuirks,
83 }
84 }
85}
86
87/// The HTML tree builder.
88pub struct TreeBuilder<Handle, Sink> {
89 /// Options controlling the behavior of the tree builder.
90 opts: TreeBuilderOpts,
91
92 /// Consumer of tree modifications.
93 pub sink: Sink,
94
95 /// Insertion mode.
96 mode: InsertionMode,
97
98 /// Original insertion mode, used by Text and InTableText modes.
99 orig_mode: Option<InsertionMode>,
100
101 /// Stack of template insertion modes.
102 template_modes: Vec<InsertionMode>,
103
104 /// Pending table character tokens.
105 pending_table_text: Vec<(SplitStatus, StrTendril)>,
106
107 /// Quirks mode as set by the parser.
108 /// FIXME: can scripts etc. change this?
109 quirks_mode: QuirksMode,
110
111 /// The document node, which is created by the sink.
112 doc_handle: Handle,
113
114 /// Stack of open elements, most recently added at end.
115 open_elems: Vec<Handle>,
116
117 /// List of active formatting elements.
118 active_formatting: Vec<FormatEntry<Handle>>,
119
120 //§ the-element-pointers
121 /// Head element pointer.
122 head_elem: Option<Handle>,
123
124 /// Form element pointer.
125 form_elem: Option<Handle>,
126 //§ END
127
128 /// Frameset-ok flag.
129 frameset_ok: bool,
130
131 /// Ignore a following U+000A LINE FEED?
132 ignore_lf: bool,
133
134 /// Is foster parenting enabled?
135 foster_parenting: bool,
136
137 /// The context element for the fragment parsing algorithm.
138 context_elem: Option<Handle>,
139
140 /// Track current line
141 current_line: u64,
142
143 // WARNING: If you add new fields that contain Handles, you
144 // must add them to trace_handles() below to preserve memory
145 // safety!
146 //
147 // FIXME: Auto-generate the trace hooks like Servo does.
148}
149
150impl<Handle, Sink> TreeBuilder<Handle, Sink>
151 where Handle: Clone,
152 Sink: TreeSink<Handle=Handle>,
153{
154 /// Create a new tree builder which sends tree modifications to a particular `TreeSink`.
155 ///
156 /// The tree builder is also a `TokenSink`.
157 pub fn new(mut sink: Sink, opts: TreeBuilderOpts) -> TreeBuilder<Handle, Sink> {
158 let doc_handle = sink.get_document();
159 TreeBuilder {
160 opts: opts,
161 sink: sink,
162 mode: Initial,
163 orig_mode: None,
164 template_modes: vec!(),
165 pending_table_text: vec!(),
166 quirks_mode: opts.quirks_mode,
167 doc_handle: doc_handle,
168 open_elems: vec!(),
169 active_formatting: vec!(),
170 head_elem: None,
171 form_elem: None,
172 frameset_ok: true,
173 ignore_lf: false,
174 foster_parenting: false,
175 context_elem: None,
176 current_line: 1,
177 }
178 }
179
180 /// Create a new tree builder which sends tree modifications to a particular `TreeSink`.
181 /// This is for parsing fragments.
182 ///
183 /// The tree builder is also a `TokenSink`.
184 pub fn new_for_fragment(mut sink: Sink,
185 context_elem: Handle,
186 form_elem: Option<Handle>,
187 opts: TreeBuilderOpts) -> TreeBuilder<Handle, Sink> {
188 let doc_handle = sink.get_document();
189 let context_is_template =
190 sink.elem_name(&context_elem) == expanded_name!(html "template");
191 let mut tb = TreeBuilder {
192 opts: opts,
193 sink: sink,
194 mode: Initial,
195 orig_mode: None,
196 template_modes: if context_is_template { vec![InTemplate] } else { vec![] },
197 pending_table_text: vec!(),
198 quirks_mode: opts.quirks_mode,
199 doc_handle: doc_handle,
200 open_elems: vec!(),
201 active_formatting: vec!(),
202 head_elem: None,
203 form_elem: form_elem,
204 frameset_ok: true,
205 ignore_lf: false,
206 foster_parenting: false,
207 context_elem: Some(context_elem),
208 current_line: 1,
209 };
210
211 // https://html.spec.whatwg.org/multipage/syntax.html#parsing-html-fragments
212 // 5. Let root be a new html element with no attributes.
213 // 6. Append the element root to the Document node created above.
214 // 7. Set up the parser's stack of open elements so that it contains just the single element root.
215 tb.create_root(vec!());
216 // 10. Reset the parser's insertion mode appropriately.
217 tb.mode = tb.reset_insertion_mode();
218
219 tb
220 }
221
222 // https://html.spec.whatwg.org/multipage/syntax.html#concept-frag-parse-context
223 // Step 4. Set the state of the HTML parser's tokenization stage as follows:
224 pub fn tokenizer_state_for_context_elem(&self) -> tok_state::State {
225 let elem = self.context_elem.as_ref().expect("no context element");
226 let name = match self.sink.elem_name(elem) {
227 ExpandedName { ns: &ns!(html), local } => local,
228 _ => return tok_state::Data
229 };
230 match *name {
231 local_name!("title") | local_name!("textarea") => tok_state::RawData(tok_state::Rcdata),
232
233 local_name!("style") | local_name!("xmp") | local_name!("iframe")
234 | local_name!("noembed") | local_name!("noframes") => tok_state::RawData(tok_state::Rawtext),
235
236 local_name!("script") => tok_state::RawData(tok_state::ScriptData),
237
238 local_name!("noscript") => if self.opts.scripting_enabled {
239 tok_state::RawData(tok_state::Rawtext)
240 } else {
241 tok_state::Data
242 },
243
244 local_name!("plaintext") => tok_state::Plaintext,
245
246 _ => tok_state::Data
247 }
248 }
249
250 /// Call the `Tracer`'s `trace_handle` method on every `Handle` in the tree builder's
251 /// internal state. This is intended to support garbage-collected DOMs.
252 pub fn trace_handles(&self, tracer: &Tracer<Handle=Handle>) {
253 tracer.trace_handle(&self.doc_handle);
254 for e in &self.open_elems {
255 tracer.trace_handle(e);
256 }
257 for e in &self.active_formatting {
258 match e {
259 &Element(ref h, _) => tracer.trace_handle(h),
260 _ => (),
261 }
262 }
263 self.head_elem.as_ref().map(|h| tracer.trace_handle(h));
264 self.form_elem.as_ref().map(|h| tracer.trace_handle(h));
265 self.context_elem.as_ref().map(|h| tracer.trace_handle(h));
266 }
267
268 #[allow(dead_code)]
269 fn dump_state(&self, label: String) {
270 println!("dump_state on {}", label);
271 print!(" open_elems:");
272 for node in self.open_elems.iter() {
273 let name = self.sink.elem_name(node);
274 match *name.ns {
275 ns!(html) => print!(" {}", name.local),
276 _ => panic!(),
277 }
278 }
279 println!("");
280 print!(" active_formatting:");
281 for entry in self.active_formatting.iter() {
282 match entry {
283 &Marker => print!(" Marker"),
284 &Element(ref h, _) => {
285 let name = self.sink.elem_name(h);
286 match *name.ns {
287 ns!(html) => print!(" {}", name.local),
288 _ => panic!(),
289 }
290 }
291 }
292 }
293 println!("");
294 }
295
296 fn debug_step(&self, mode: InsertionMode, token: &Token) {
297 use util::str::to_escaped_string;
298 debug!("processing {} in insertion mode {:?}", to_escaped_string(token), mode);
299 }
300
301 fn process_to_completion(&mut self, mut token: Token) -> TokenSinkResult<Handle> {
302 // Queue of additional tokens yet to be processed.
303 // This stays empty in the common case where we don't split whitespace.
304 let mut more_tokens = VecDeque::new();
305
306 loop {
307 let should_have_acknowledged_self_closing_flag =
308 matches!(token, TagToken(Tag { self_closing: true, kind: StartTag, .. }));
309 let result = if self.is_foreign(&token) {
310 self.step_foreign(token)
311 } else {
312 let mode = self.mode;
313 self.step(mode, token)
314 };
315 match result {
316 Done => {
317 if should_have_acknowledged_self_closing_flag {
318 self.sink.parse_error(Borrowed("Unacknowledged self-closing tag"));
319 }
320 token = unwrap_or_return!(more_tokens.pop_front(), tokenizer::TokenSinkResult::Continue);
321 }
322 DoneAckSelfClosing => {
323 token = unwrap_or_return!(more_tokens.pop_front(), tokenizer::TokenSinkResult::Continue);
324 }
325 Reprocess(m, t) => {
326 self.mode = m;
327 token = t;
328 }
329 ReprocessForeign(t) => {
330 token = t;
331 }
332 SplitWhitespace(mut buf) => {
333 let p = buf.pop_front_char_run(is_ascii_whitespace);
334 let (first, is_ws) = unwrap_or_return!(p, tokenizer::TokenSinkResult::Continue);
335 let status = if is_ws { Whitespace } else { NotWhitespace };
336 token = CharacterTokens(status, first);
337
338 if buf.len32() > 0 {
339 more_tokens.push_back(CharacterTokens(NotSplit, buf));
340 }
341 }
342 Script(node) => {
343 assert!(more_tokens.is_empty());
344 return tokenizer::TokenSinkResult::Script(node);
345 }
346 ToPlaintext => {
347 assert!(more_tokens.is_empty());
348 return tokenizer::TokenSinkResult::Plaintext;
349 }
350 ToRawData(k) => {
351 assert!(more_tokens.is_empty());
352 return tokenizer::TokenSinkResult::RawData(k);
353 }
354 }
355 }
356 }
357
358 /// Are we parsing a HTML fragment?
359 pub fn is_fragment(&self) -> bool {
360 self.context_elem.is_some()
361 }
362
363 /// https://html.spec.whatwg.org/multipage/#appropriate-place-for-inserting-a-node
364 fn appropriate_place_for_insertion(&mut self,
365 override_target: Option<Handle>)
366 -> InsertionPoint<Handle> {
367 use self::tag_sets::*;
368
369 declare_tag_set!(foster_target = "table" "tbody" "tfoot" "thead" "tr");
370 let target = override_target.unwrap_or_else(|| self.current_node().clone());
371 if !(self.foster_parenting && self.elem_in(&target, foster_target)) {
372 if self.html_elem_named(&target, local_name!("template")) {
373 // No foster parenting (inside template).
374 let contents = self.sink.get_template_contents(&target);
375 return LastChild(contents);
376 } else {
377 // No foster parenting (the common case).
378 return LastChild(target);
379 }
380 }
381
382 // Foster parenting
383 let mut iter = self.open_elems.iter().rev().peekable();
384 while let Some(elem) = iter.next() {
385 if self.html_elem_named(&elem, local_name!("template")) {
386 let contents = self.sink.get_template_contents(&elem);
387 return LastChild(contents);
388 } else if self.html_elem_named(&elem, local_name!("table")) {
ff7c6d11
XL
389 return TableFosterParenting {
390 element: elem.clone(),
391 prev_element: (*iter.peek().unwrap()).clone(),
392 };
ea8adc8c
XL
393 }
394 }
395 let html_elem = self.html_elem();
396 LastChild(html_elem.clone())
397 }
398
399 fn insert_at(&mut self, insertion_point: InsertionPoint<Handle>, child: NodeOrText<Handle>) {
400 match insertion_point {
401 LastChild(parent) => self.sink.append(&parent, child),
ff7c6d11
XL
402 BeforeSibling(sibling) => self.sink.append_before_sibling(&sibling, child),
403 TableFosterParenting { element, prev_element } => self.sink.append_based_on_parent_node(
404 &element,
405 &prev_element,
406 child),
ea8adc8c
XL
407 }
408 }
409}
410
411impl<Handle, Sink> TokenSink
412 for TreeBuilder<Handle, Sink>
413 where Handle: Clone,
414 Sink: TreeSink<Handle=Handle>,
415{
416 type Handle = Handle;
417
418 fn process_token(&mut self, token: tokenizer::Token, line_number: u64) -> TokenSinkResult<Handle> {
419 if line_number != self.current_line {
420 self.sink.set_current_line(line_number);
421 }
422 let ignore_lf = replace(&mut self.ignore_lf, false);
423
424 // Handle `ParseError` and `DoctypeToken`; convert everything else to the local `Token` type.
425 let token = match token {
426 tokenizer::ParseError(e) => {
427 self.sink.parse_error(e);
428 return tokenizer::TokenSinkResult::Continue;
429 }
430
431 tokenizer::DoctypeToken(dt) => if self.mode == Initial {
432 let (err, quirk) = data::doctype_error_and_quirks(&dt, self.opts.iframe_srcdoc);
433 if err {
434 self.sink.parse_error(format_if!(
435 self.opts.exact_errors,
436 "Bad DOCTYPE",
437 "Bad DOCTYPE: {:?}", dt));
438 }
439 let Doctype { name, public_id, system_id, force_quirks: _ } = dt;
440 if !self.opts.drop_doctype {
441 self.sink.append_doctype_to_document(
442 name.unwrap_or(StrTendril::new()),
443 public_id.unwrap_or(StrTendril::new()),
444 system_id.unwrap_or(StrTendril::new())
445 );
446 }
447 self.set_quirks_mode(quirk);
448
449 self.mode = BeforeHtml;
450 return tokenizer::TokenSinkResult::Continue;
451 } else {
452 self.sink.parse_error(format_if!(
453 self.opts.exact_errors,
454 "DOCTYPE in body",
455 "DOCTYPE in insertion mode {:?}", self.mode));
456 return tokenizer::TokenSinkResult::Continue;
457 },
458
459 tokenizer::TagToken(x) => TagToken(x),
460 tokenizer::CommentToken(x) => CommentToken(x),
461 tokenizer::NullCharacterToken => NullCharacterToken,
462 tokenizer::EOFToken => EOFToken,
463
464 tokenizer::CharacterTokens(mut x) => {
465 if ignore_lf && x.starts_with("\n") {
466 x.pop_front(1);
467 }
468 if x.is_empty() {
469 return tokenizer::TokenSinkResult::Continue;
470 }
471 CharacterTokens(NotSplit, x)
472 }
473 };
474
475 self.process_to_completion(token)
476 }
477
478 fn end(&mut self) {
479 for elem in self.open_elems.drain(..).rev() {
480 self.sink.pop(&elem);
481 }
482 }
483
484 fn adjusted_current_node_present_but_not_in_html_namespace(&self) -> bool {
485 !self.open_elems.is_empty() &&
486 self.sink.elem_name(self.adjusted_current_node()).ns != &ns!(html)
487 }
488}
489
ff7c6d11
XL
490pub fn html_elem<Handle>(open_elems: &[Handle]) -> &Handle {
491 &open_elems[0]
492}
493
494pub struct ActiveFormattingIter<'a, Handle: 'a> {
495 iter: Rev<Enumerate<slice::Iter<'a, FormatEntry<Handle>>>>,
496}
497
498impl<'a, Handle> Iterator for ActiveFormattingIter<'a, Handle> {
499 type Item = (usize, &'a Handle, &'a Tag);
500 fn next(&mut self) -> Option<(usize, &'a Handle, &'a Tag)> {
501 match self.iter.next() {
502 None | Some((_, &Marker)) => None,
503 Some((i, &Element(ref h, ref t))) => Some((i, h, t)),
504 }
505 }
506}
507
508pub enum PushFlag {
509 Push,
510 NoPush,
511}
512
513enum Bookmark<Handle> {
514 Replace(Handle),
515 InsertAfter(Handle),
516}
517
518macro_rules! qualname {
519 ("", $local:tt) => {
520 QualName {
521 prefix: None,
522 ns: ns!(),
523 local: local_name!($local),
524 }
525 };
526 ($prefix: tt $ns:tt $local:tt) => {
527 QualName {
528 prefix: Some(namespace_prefix!($prefix)),
529 ns: ns!($ns),
530 local: local_name!($local),
531 }
532 }
533}
534
535#[doc(hidden)]
536impl<Handle, Sink> TreeBuilder<Handle, Sink>
537 where Handle: Clone,
538 Sink: TreeSink<Handle=Handle>,
539{
540 fn unexpected<T: fmt::Debug>(&mut self, _thing: &T) -> ProcessResult<Handle> {
541 self.sink.parse_error(format_if!(
542 self.opts.exact_errors,
543 "Unexpected token",
544 "Unexpected token {} in insertion mode {:?}", to_escaped_string(_thing), self.mode));
545 Done
546 }
547
548 fn assert_named(&mut self, node: &Handle, name: LocalName) {
549 assert!(self.html_elem_named(&node, name));
550 }
551
552 /// Iterate over the active formatting elements (with index in the list) from the end
553 /// to the last marker, or the beginning if there are no markers.
554 fn active_formatting_end_to_marker<'a>(&'a self) -> ActiveFormattingIter<'a, Handle> {
555 ActiveFormattingIter {
556 iter: self.active_formatting.iter().enumerate().rev(),
557 }
558 }
559
560 fn position_in_active_formatting(&self, element: &Handle) -> Option<usize> {
561 self.active_formatting
562 .iter()
563 .position(|n| {
564 match n {
565 &Marker => false,
566 &Element(ref handle, _) => self.sink.same_node(handle, element)
567 }
568 })
569 }
570
571 fn set_quirks_mode(&mut self, mode: QuirksMode) {
572 self.quirks_mode = mode;
573 self.sink.set_quirks_mode(mode);
574 }
575
576 fn stop_parsing(&mut self) -> ProcessResult<Handle> {
577 warn!("stop_parsing not implemented, full speed ahead!");
578 Done
579 }
580
581 //§ parsing-elements-that-contain-only-text
582 // Switch to `Text` insertion mode, save the old mode, and
583 // switch the tokenizer to a raw-data state.
584 // The latter only takes effect after the current / next
585 // `process_token` of a start tag returns!
586 fn to_raw_text_mode(&mut self, k: RawKind) -> ProcessResult<Handle> {
587 self.orig_mode = Some(self.mode);
588 self.mode = Text;
589 ToRawData(k)
590 }
591
592 // The generic raw text / RCDATA parsing algorithm.
593 fn parse_raw_data(&mut self, tag: Tag, k: RawKind) -> ProcessResult<Handle> {
594 self.insert_element_for(tag);
595 self.to_raw_text_mode(k)
596 }
597 //§ END
598
599 fn current_node(&self) -> &Handle {
600 self.open_elems.last().expect("no current element")
601 }
602
603 fn adjusted_current_node(&self) -> &Handle {
604 if self.open_elems.len() == 1 {
605 if let Some(ctx) = self.context_elem.as_ref() {
606 return ctx;
607 }
608 }
609 self.current_node()
610 }
611
612 fn current_node_in<TagSet>(&self, set: TagSet) -> bool
613 where TagSet: Fn(ExpandedName) -> bool
614 {
615 set(self.sink.elem_name(self.current_node()))
616 }
617
618 // Insert at the "appropriate place for inserting a node".
619 fn insert_appropriately(&mut self, child: NodeOrText<Handle>, override_target: Option<Handle>) {
620 let insertion_point = self.appropriate_place_for_insertion(override_target);
621 self.insert_at(insertion_point, child);
622 }
623
624 fn adoption_agency(&mut self, subject: LocalName) {
625 // 1.
626 if self.current_node_named(subject.clone()) {
627 if self.position_in_active_formatting(self.current_node()).is_none() {
628 self.pop();
629 return;
630 }
631 }
632
633 // 2. 3. 4.
634 for _ in 0..8 {
635 // 5.
636 let (fmt_elem_index, fmt_elem, fmt_elem_tag) = unwrap_or_return!(
637 // We clone the Handle and Tag so they don't cause an immutable borrow of self.
638 self.active_formatting_end_to_marker()
639 .filter(|&(_, _, tag)| tag.name == subject)
640 .next()
641 .map(|(i, h, t)| (i, h.clone(), t.clone())),
642
643 {
644 self.process_end_tag_in_body(Tag {
645 kind: EndTag,
646 name: subject,
647 self_closing: false,
648 attrs: vec!(),
649 });
650 }
651 );
652
653 let fmt_elem_stack_index = unwrap_or_return!(
654 self.open_elems.iter()
655 .rposition(|n| self.sink.same_node(n, &fmt_elem)),
656
657 {
658 self.sink.parse_error(Borrowed("Formatting element not open"));
659 self.active_formatting.remove(fmt_elem_index);
660 }
661 );
662
663 // 7.
664 if !self.in_scope(default_scope, |n| self.sink.same_node(&n, &fmt_elem)) {
665 self.sink.parse_error(Borrowed("Formatting element not in scope"));
666 return;
667 }
668
669 // 8.
670 if !self.sink.same_node(self.current_node(), &fmt_elem) {
671 self.sink.parse_error(Borrowed("Formatting element not current node"));
672 }
673
674 // 9.
675 let (furthest_block_index, furthest_block) = unwrap_or_return!(
676 self.open_elems.iter()
677 .enumerate()
678 .skip(fmt_elem_stack_index)
679 .filter(|&(_, open_element)| self.elem_in(open_element, special_tag))
680 .next()
681 .map(|(i, h)| (i, h.clone())),
682
683 // 10.
684 {
685 self.open_elems.truncate(fmt_elem_stack_index);
686 self.active_formatting.remove(fmt_elem_index);
687 }
688 );
689
690 // 11.
691 let common_ancestor = self.open_elems[fmt_elem_stack_index - 1].clone();
692
693 // 12.
694 let mut bookmark = Bookmark::Replace(fmt_elem.clone());
695
696 // 13.
697 let mut node;
698 let mut node_index = furthest_block_index;
699 let mut last_node = furthest_block.clone();
700
701 // 13.1.
702 let mut inner_counter = 0;
703 loop {
704 // 13.2.
705 inner_counter += 1;
706
707 // 13.3.
708 node_index -= 1;
709 node = self.open_elems[node_index].clone();
710
711 // 13.4.
712 if self.sink.same_node(&node, &fmt_elem) {
713 break;
714 }
715
716 // 13.5.
717 if inner_counter > 3 {
718 self.position_in_active_formatting(&node)
719 .map(|position| self.active_formatting.remove(position));
720 self.open_elems.remove(node_index);
721 continue;
722 }
723
724 let node_formatting_index = unwrap_or_else!(
725 self.position_in_active_formatting(&node),
726
727 // 13.6.
728 {
729 self.open_elems.remove(node_index);
730 continue;
731 }
732 );
733
734 // 13.7.
735 let tag = match self.active_formatting[node_formatting_index] {
736 Element(ref h, ref t) => {
737 assert!(self.sink.same_node(h, &node));
738 t.clone()
739 }
740 Marker => panic!("Found marker during adoption agency"),
741 };
742 // FIXME: Is there a way to avoid cloning the attributes twice here (once on their
743 // own, once as part of t.clone() above)?
744 let new_element = create_element(
745 &mut self.sink, QualName::new(None, ns!(html), tag.name.clone()),
746 tag.attrs.clone());
747 self.open_elems[node_index] = new_element.clone();
748 self.active_formatting[node_formatting_index] = Element(new_element.clone(), tag);
749 node = new_element;
750
751 // 13.8.
752 if self.sink.same_node(&last_node, &furthest_block) {
753 bookmark = Bookmark::InsertAfter(node.clone());
754 }
755
756 // 13.9.
757 self.sink.remove_from_parent(&last_node);
758 self.sink.append(&node, AppendNode(last_node.clone()));
759
760 // 13.10.
761 last_node = node.clone();
762
763 // 13.11.
764 }
765
766 // 14.
767 self.sink.remove_from_parent(&last_node);
768 self.insert_appropriately(AppendNode(last_node.clone()), Some(common_ancestor));
769
770 // 15.
771 // FIXME: Is there a way to avoid cloning the attributes twice here (once on their own,
772 // once as part of t.clone() above)?
773 let new_element = create_element(
774 &mut self.sink, QualName::new(None, ns!(html), fmt_elem_tag.name.clone()),
775 fmt_elem_tag.attrs.clone());
776 let new_entry = Element(new_element.clone(), fmt_elem_tag);
777
778 // 16.
779 self.sink.reparent_children(&furthest_block, &new_element);
780
781 // 17.
782 self.sink.append(&furthest_block, AppendNode(new_element.clone()));
783
784 // 18.
785 // FIXME: We could probably get rid of the position_in_active_formatting() calls here
786 // if we had a more clever Bookmark representation.
787 match bookmark {
788 Bookmark::Replace(to_replace) => {
789 let index = self.position_in_active_formatting(&to_replace)
790 .expect("bookmark not found in active formatting elements");
791 self.active_formatting[index] = new_entry;
792 }
793 Bookmark::InsertAfter(previous) => {
794 let index = self.position_in_active_formatting(&previous)
795 .expect("bookmark not found in active formatting elements") + 1;
796 self.active_formatting.insert(index, new_entry);
797 let old_index = self.position_in_active_formatting(&fmt_elem)
798 .expect("formatting element not found in active formatting elements");
799 self.active_formatting.remove(old_index);
800 }
801 }
802
803 // 19.
804 self.remove_from_stack(&fmt_elem);
805 let new_furthest_block_index = self.open_elems.iter()
806 .position(|n| self.sink.same_node(n, &furthest_block))
807 .expect("furthest block missing from open element stack");
808 self.open_elems.insert(new_furthest_block_index + 1, new_element);
809
810 // 20.
811 }
812 }
813
814 fn push(&mut self, elem: &Handle) {
815 self.open_elems.push(elem.clone());
816 }
817
818 fn pop(&mut self) -> Handle {
819 let elem = self.open_elems.pop().expect("no current element");
820 self.sink.pop(&elem);
821 elem
822 }
823
824 fn remove_from_stack(&mut self, elem: &Handle) {
825 let sink = &mut self.sink;
826 let position = self.open_elems
827 .iter()
828 .rposition(|x| sink.same_node(elem, &x));
829 if let Some(position) = position {
830 self.open_elems.remove(position);
831 sink.pop(elem);
832 }
833 }
834
835 fn is_marker_or_open(&self, entry: &FormatEntry<Handle>) -> bool {
836 match *entry {
837 Marker => true,
838 Element(ref node, _) => {
839 self.open_elems.iter()
840 .rev()
841 .any(|n| self.sink.same_node(&n, &node))
842 }
843 }
844 }
845
846 /// Reconstruct the active formatting elements.
847 fn reconstruct_formatting(&mut self) {
848 {
849 let last = unwrap_or_return!(self.active_formatting.last(), ());
850 if self.is_marker_or_open(last) {
851 return
852 }
853 }
854
855 let mut entry_index = self.active_formatting.len() - 1;
856 loop {
857 if entry_index == 0 {
858 break
859 }
860 entry_index -= 1;
861 if self.is_marker_or_open(&self.active_formatting[entry_index]) {
862 entry_index += 1;
863 break
864 }
865 }
866
867 loop {
868 let tag = match self.active_formatting[entry_index] {
869 Element(_, ref t) => t.clone(),
870 Marker => panic!("Found marker during formatting element reconstruction"),
871 };
872
873 // FIXME: Is there a way to avoid cloning the attributes twice here (once on their own,
874 // once as part of t.clone() above)?
875 let new_element = self.insert_element(Push, ns!(html), tag.name.clone(),
876 tag.attrs.clone());
877 self.active_formatting[entry_index] = Element(new_element, tag);
878 if entry_index == self.active_formatting.len() - 1 {
879 break
880 }
881 entry_index += 1;
882 }
883 }
884
885 /// Get the first element on the stack, which will be the <html> element.
886 fn html_elem(&self) -> &Handle {
887 &self.open_elems[0]
888 }
889
890 /// Get the second element on the stack, if it's a HTML body element.
891 fn body_elem(&self) -> Option<&Handle> {
892 if self.open_elems.len() <= 1 {
893 return None;
894 }
895
896 let node = &self.open_elems[1];
897 if self.html_elem_named(node, local_name!("body")) {
898 Some(node)
899 } else {
900 None
901 }
902 }
903
904 /// Signal an error depending on the state of the stack of open elements at
905 /// the end of the body.
906 fn check_body_end(&mut self) {
907 declare_tag_set!(body_end_ok =
908 "dd" "dt" "li" "optgroup" "option" "p" "rp" "rt" "tbody" "td" "tfoot" "th"
909 "thead" "tr" "body" "html");
910
911 for elem in self.open_elems.iter() {
912 let error;
913 {
914 let name = self.sink.elem_name(elem);
915 if body_end_ok(name) {
916 continue
917 }
918 error = format_if!(self.opts.exact_errors,
919 "Unexpected open tag at end of body",
920 "Unexpected open tag {:?} at end of body", name);
921 }
922 self.sink.parse_error(error);
923 // FIXME: Do we keep checking after finding one bad tag?
924 // The spec suggests not.
925 return;
926 }
927 }
928
929 fn in_scope<TagSet,Pred>(&self, scope: TagSet, pred: Pred) -> bool
930 where TagSet: Fn(ExpandedName) -> bool, Pred: Fn(Handle) -> bool
931 {
932 for node in self.open_elems.iter().rev() {
933 if pred(node.clone()) {
934 return true;
935 }
936 if scope(self.sink.elem_name(node)) {
937 return false;
938 }
939 }
940
941 // supposed to be impossible, because <html> is always in scope
942
943 false
944 }
945
946 fn elem_in<TagSet>(&self, elem: &Handle, set: TagSet) -> bool
947 where TagSet: Fn(ExpandedName) -> bool
948 {
949 set(self.sink.elem_name(elem))
950 }
951
952 fn html_elem_named(&self, elem: &Handle, name: LocalName) -> bool {
953 let expanded = self.sink.elem_name(elem);
954 *expanded.ns == ns!(html) && *expanded.local == name
955 }
956
957 fn in_html_elem_named(&self, name: LocalName) -> bool {
958 self.open_elems.iter().any(|elem| self.html_elem_named(elem, name.clone()))
959 }
960
961 fn current_node_named(&self, name: LocalName) -> bool {
962 self.html_elem_named(self.current_node(), name)
963 }
964
965 fn in_scope_named<TagSet>(&self, scope: TagSet, name: LocalName) -> bool
966 where TagSet: Fn(ExpandedName) -> bool
967 {
968 self.in_scope(scope, |elem| self.html_elem_named(&elem, name.clone()))
969 }
970
971 //§ closing-elements-that-have-implied-end-tags
972 fn generate_implied_end<TagSet>(&mut self, set: TagSet)
973 where TagSet: Fn(ExpandedName) -> bool
974 {
975 loop {
976 {
977 let elem = unwrap_or_return!(self.open_elems.last(), ());
978 let nsname = self.sink.elem_name(elem);
979 if !set(nsname) { return; }
980 }
981 self.pop();
982 }
983 }
984
985 fn generate_implied_end_except(&mut self, except: LocalName) {
986 self.generate_implied_end(|p| {
987 if *p.ns == ns!(html) && *p.local == except {
988 false
989 } else {
990 cursory_implied_end(p)
991 }
992 });
993 }
994 //§ END
995
996 // Pop elements until the current element is in the set.
997 fn pop_until_current<TagSet>(&mut self, pred: TagSet)
998 where TagSet: Fn(ExpandedName) -> bool
999 {
1000 loop {
1001 if self.current_node_in(|x| pred(x)) {
1002 break;
1003 }
1004 self.open_elems.pop();
1005 }
1006 }
1007
1008 // Pop elements until an element from the set has been popped. Returns the
1009 // number of elements popped.
1010 fn pop_until<P>(&mut self, pred: P) -> usize
1011 where P: Fn(ExpandedName) -> bool
1012 {
1013 let mut n = 0;
1014 loop {
1015 n += 1;
1016 match self.open_elems.pop() {
1017 None => break,
1018 Some(elem) => if pred(self.sink.elem_name(&elem)) { break; },
1019 }
1020 }
1021 n
1022 }
1023
1024 fn pop_until_named(&mut self, name: LocalName) -> usize {
1025 self.pop_until(|p| *p.ns == ns!(html) && *p.local == name)
1026 }
1027
1028 // Pop elements until one with the specified name has been popped.
1029 // Signal an error if it was not the first one.
1030 fn expect_to_close(&mut self, name: LocalName) {
1031 if self.pop_until_named(name.clone()) != 1 {
1032 self.sink.parse_error(format_if!(self.opts.exact_errors,
1033 "Unexpected open element",
1034 "Unexpected open element while closing {:?}", name));
1035 }
1036 }
1037
1038 fn close_p_element(&mut self) {
1039 declare_tag_set!(implied = [cursory_implied_end] - "p");
1040 self.generate_implied_end(implied);
1041 self.expect_to_close(local_name!("p"));
1042 }
1043
1044 fn close_p_element_in_button_scope(&mut self) {
1045 if self.in_scope_named(button_scope, local_name!("p")) {
1046 self.close_p_element();
1047 }
1048 }
1049
1050 // Check <input> tags for type=hidden
1051 fn is_type_hidden(&self, tag: &Tag) -> bool {
1052 match tag.attrs.iter().find(|&at| at.name.expanded() == expanded_name!("", "type")) {
1053 None => false,
1054 Some(at) => (&*at.value).eq_ignore_ascii_case("hidden"),
1055 }
1056 }
1057
1058 fn foster_parent_in_body(&mut self, token: Token) -> ProcessResult<Handle> {
1059 warn!("foster parenting not implemented");
1060 self.foster_parenting = true;
1061 let res = self.step(InBody, token);
1062 // FIXME: what if res is Reprocess?
1063 self.foster_parenting = false;
1064 res
1065 }
1066
1067 fn process_chars_in_table(&mut self, token: Token) -> ProcessResult<Handle> {
1068 declare_tag_set!(table_outer = "table" "tbody" "tfoot" "thead" "tr");
1069 if self.current_node_in(table_outer) {
1070 assert!(self.pending_table_text.is_empty());
1071 self.orig_mode = Some(self.mode);
1072 Reprocess(InTableText, token)
1073 } else {
1074 self.sink.parse_error(format_if!(self.opts.exact_errors,
1075 "Unexpected characters in table",
1076 "Unexpected characters {} in table", to_escaped_string(&token)));
1077 self.foster_parent_in_body(token)
1078 }
1079 }
1080
1081 // https://html.spec.whatwg.org/multipage/syntax.html#reset-the-insertion-mode-appropriately
1082 fn reset_insertion_mode(&mut self) -> InsertionMode {
1083 for (i, mut node) in self.open_elems.iter().enumerate().rev() {
1084 let last = i == 0usize;
1085 if let (true, Some(ctx)) = (last, self.context_elem.as_ref()) {
1086 node = ctx;
1087 }
1088 let name = match self.sink.elem_name(node) {
1089 ExpandedName { ns: &ns!(html), local } => local,
1090 _ => continue,
1091 };
1092 match *name {
1093 local_name!("select") => {
1094 for ancestor in self.open_elems[0..i].iter().rev() {
1095 if self.html_elem_named(ancestor, local_name!("template")) {
1096 return InSelect;
1097 } else if self.html_elem_named(ancestor, local_name!("table")) {
1098 return InSelectInTable;
1099 }
1100 }
1101 return InSelect;
1102 },
1103 local_name!("td") | local_name!("th") => if !last { return InCell; },
1104 local_name!("tr") => return InRow,
1105 local_name!("tbody") | local_name!("thead") | local_name!("tfoot") => return InTableBody,
1106 local_name!("caption") => return InCaption,
1107 local_name!("colgroup") => return InColumnGroup,
1108 local_name!("table") => return InTable,
1109 local_name!("template") => return *self.template_modes.last().unwrap(),
1110 local_name!("head") => if !last { return InHead },
1111 local_name!("body") => return InBody,
1112 local_name!("frameset") => return InFrameset,
1113 local_name!("html") => match self.head_elem {
1114 None => return BeforeHead,
1115 Some(_) => return AfterHead,
1116 },
1117
1118 _ => (),
1119 }
1120 }
1121 InBody
1122 }
1123
1124 fn close_the_cell(&mut self) {
1125 self.generate_implied_end(cursory_implied_end);
1126 if self.pop_until(td_th) != 1 {
1127 self.sink.parse_error(Borrowed("expected to close <td> or <th> with cell"));
1128 }
1129 self.clear_active_formatting_to_marker();
1130 }
1131
1132 fn append_text(&mut self, text: StrTendril) -> ProcessResult<Handle> {
1133 self.insert_appropriately(AppendText(text), None);
1134 Done
1135 }
1136
1137 fn append_comment(&mut self, text: StrTendril) -> ProcessResult<Handle> {
1138 let comment = self.sink.create_comment(text);
1139 self.insert_appropriately(AppendNode(comment), None);
1140 Done
1141 }
1142
1143 fn append_comment_to_doc(&mut self, text: StrTendril) -> ProcessResult<Handle> {
1144 let comment = self.sink.create_comment(text);
1145 self.sink.append(&self.doc_handle, AppendNode(comment));
1146 Done
1147 }
1148
1149 fn append_comment_to_html(&mut self, text: StrTendril) -> ProcessResult<Handle> {
1150 let target = html_elem(&self.open_elems);
1151 let comment = self.sink.create_comment(text);
1152 self.sink.append(target, AppendNode(comment));
1153 Done
1154 }
1155
1156 //§ creating-and-inserting-nodes
1157 fn create_root(&mut self, attrs: Vec<Attribute>) {
1158 let elem = create_element(
1159 &mut self.sink, QualName::new(None, ns!(html), local_name!("html")),
1160 attrs);
1161 self.push(&elem);
1162 self.sink.append(&self.doc_handle, AppendNode(elem));
1163 // FIXME: application cache selection algorithm
1164 }
1165
1166 // https://html.spec.whatwg.org/multipage/#create-an-element-for-the-token
1167 fn insert_element(&mut self, push: PushFlag, ns: Namespace, name: LocalName, attrs: Vec<Attribute>)
1168 -> Handle {
1169 declare_tag_set!(form_associatable =
1170 "button" "fieldset" "input" "object"
1171 "output" "select" "textarea" "img");
1172
1173 declare_tag_set!(listed = [form_associatable] - "img");
1174
1175 // Step 7.
1176 let qname = QualName::new(None, ns, name);
1177 let elem = create_element(&mut self.sink, qname.clone(), attrs.clone());
1178
1179 let insertion_point = self.appropriate_place_for_insertion(None);
1180 let (node1, node2) = match insertion_point {
1181 LastChild(ref p) |
1182 BeforeSibling(ref p) => (p.clone(), None),
1183 TableFosterParenting { ref element, ref prev_element } => (element.clone(), Some(prev_element.clone())),
1184 };
1185
1186 // Step 12.
1187 if form_associatable(qname.expanded()) &&
1188 self.form_elem.is_some() &&
1189 !self.in_html_elem_named(local_name!("template")) &&
1190 !(listed(qname.expanded()) &&
1191 attrs.iter().any(|a| a.name.expanded() == expanded_name!("", "form"))) {
1192
1193 let form = self.form_elem.as_ref().unwrap().clone();
1194 let node2 = match node2 {
1195 Some(ref n) => Some(n),
1196 None => None,
1197 };
1198 self.sink.associate_with_form(&elem, &form, (&node1, node2));
1199 }
1200
1201 self.insert_at(insertion_point, AppendNode(elem.clone()));
1202
1203 match push {
1204 Push => self.push(&elem),
1205 NoPush => (),
1206 }
1207 // FIXME: Remove from the stack if we can't append?
1208 elem
1209 }
1210
1211 fn insert_element_for(&mut self, tag: Tag) -> Handle {
1212 self.insert_element(Push, ns!(html), tag.name, tag.attrs)
1213 }
1214
1215 fn insert_and_pop_element_for(&mut self, tag: Tag) -> Handle {
1216 self.insert_element(NoPush, ns!(html), tag.name, tag.attrs)
1217 }
1218
1219 fn insert_phantom(&mut self, name: LocalName) -> Handle {
1220 self.insert_element(Push, ns!(html), name, vec!())
1221 }
1222 //§ END
1223
1224 fn create_formatting_element_for(&mut self, tag: Tag) -> Handle {
1225 // FIXME: This really wants unit tests.
1226 let mut first_match = None;
1227 let mut matches = 0usize;
1228 for (i, _, old_tag) in self.active_formatting_end_to_marker() {
1229 if tag.equiv_modulo_attr_order(old_tag) {
1230 first_match = Some(i);
1231 matches += 1;
1232 }
1233 }
1234
1235 if matches >= 3 {
1236 self.active_formatting.remove(first_match.expect("matches with no index"));
1237 }
1238
1239 let elem = self.insert_element(Push, ns!(html), tag.name.clone(), tag.attrs.clone());
1240 self.active_formatting.push(Element(elem.clone(), tag));
1241 elem
1242 }
1243
1244 fn clear_active_formatting_to_marker(&mut self) {
1245 loop {
1246 match self.active_formatting.pop() {
1247 None | Some(Marker) => break,
1248 _ => (),
1249 }
1250 }
1251 }
1252
1253 fn process_end_tag_in_body(&mut self, tag: Tag) {
1254 // Look back for a matching open element.
1255 let mut match_idx = None;
1256 for (i, elem) in self.open_elems.iter().enumerate().rev() {
1257 if self.html_elem_named(elem, tag.name.clone()) {
1258 match_idx = Some(i);
1259 break;
1260 }
1261
1262 if self.elem_in(elem, special_tag) {
1263 self.sink.parse_error(Borrowed("Found special tag while closing generic tag"));
1264 return;
1265 }
1266 }
1267
1268 // Can't use unwrap_or_return!() due to rust-lang/rust#16617.
1269 let match_idx = match match_idx {
1270 None => {
1271 // I believe this is impossible, because the root
1272 // <html> element is in special_tag.
1273 self.unexpected(&tag);
1274 return;
1275 }
1276 Some(x) => x,
1277 };
1278
1279 self.generate_implied_end_except(tag.name.clone());
1280
1281 if match_idx != self.open_elems.len() - 1 {
1282 // mis-nested tags
1283 self.unexpected(&tag);
1284 }
1285 self.open_elems.truncate(match_idx);
1286 }
1287
1288 fn handle_misnested_a_tags(&mut self, tag: &Tag) {
1289 let node = unwrap_or_return!(
1290 self.active_formatting_end_to_marker()
1291 .filter(|&(_, n, _)| self.html_elem_named(n, local_name!("a")))
1292 .next()
1293 .map(|(_, n, _)| n.clone()),
1294
1295 ()
1296 );
1297
1298 self.unexpected(tag);
1299 self.adoption_agency(local_name!("a"));
1300 self.position_in_active_formatting(&node)
1301 .map(|index| self.active_formatting.remove(index));
1302 self.remove_from_stack(&node);
1303 }
1304
1305 //§ tree-construction
1306 fn is_foreign(&mut self, token: &Token) -> bool {
1307 if let EOFToken = *token {
1308 return false;
1309 }
1310
1311 if self.open_elems.len() == 0 {
1312 return false;
1313 }
1314
1315 let name = self.sink.elem_name(self.adjusted_current_node());
1316 if let ns!(html) = *name.ns {
1317 return false;
1318 }
1319
1320 if mathml_text_integration_point(name) {
1321 match *token {
1322 CharacterTokens(..) | NullCharacterToken => return false,
1323 TagToken(Tag { kind: StartTag, ref name, .. })
1324 if !matches!(*name, local_name!("mglyph") | local_name!("malignmark")) => return false,
1325 _ => (),
1326 }
1327 }
1328
1329 if svg_html_integration_point(name) {
1330 match *token {
1331 CharacterTokens(..) | NullCharacterToken => return false,
1332 TagToken(Tag { kind: StartTag, .. }) => return false,
1333 _ => (),
1334 }
1335 }
1336
1337 if let expanded_name!(mathml "annotation-xml") = name {
1338 match *token {
1339 TagToken(Tag { kind: StartTag, name: local_name!("svg"), .. }) => return false,
1340 CharacterTokens(..) | NullCharacterToken |
1341 TagToken(Tag { kind: StartTag, .. }) => {
1342 return !self.sink.is_mathml_annotation_xml_integration_point(
1343 self.adjusted_current_node())
1344 }
1345 _ => {}
1346 };
1347 }
1348
1349 true
1350 }
1351 //§ END
1352
1353 fn enter_foreign(&mut self, mut tag: Tag, ns: Namespace) -> ProcessResult<Handle> {
1354 match ns {
1355 ns!(mathml) => self.adjust_mathml_attributes(&mut tag),
1356 ns!(svg) => self.adjust_svg_attributes(&mut tag),
1357 _ => (),
1358 }
1359 self.adjust_foreign_attributes(&mut tag);
1360
1361 if tag.self_closing {
1362 self.insert_element(NoPush, ns, tag.name, tag.attrs);
1363 DoneAckSelfClosing
1364 } else {
1365 self.insert_element(Push, ns, tag.name, tag.attrs);
1366 Done
1367 }
1368 }
1369
1370 fn adjust_svg_tag_name(&mut self, tag: &mut Tag) {
1371 let Tag { ref mut name, .. } = *tag;
1372 match *name {
1373 local_name!("altglyph") => *name = local_name!("altGlyph"),
1374 local_name!("altglyphdef") => *name = local_name!("altGlyphDef"),
1375 local_name!("altglyphitem") => *name = local_name!("altGlyphItem"),
1376 local_name!("animatecolor") => *name = local_name!("animateColor"),
1377 local_name!("animatemotion") => *name = local_name!("animateMotion"),
1378 local_name!("animatetransform") => *name = local_name!("animateTransform"),
1379 local_name!("clippath") => *name = local_name!("clipPath"),
1380 local_name!("feblend") => *name = local_name!("feBlend"),
1381 local_name!("fecolormatrix") => *name = local_name!("feColorMatrix"),
1382 local_name!("fecomponenttransfer") => *name = local_name!("feComponentTransfer"),
1383 local_name!("fecomposite") => *name = local_name!("feComposite"),
1384 local_name!("feconvolvematrix") => *name = local_name!("feConvolveMatrix"),
1385 local_name!("fediffuselighting") => *name = local_name!("feDiffuseLighting"),
1386 local_name!("fedisplacementmap") => *name = local_name!("feDisplacementMap"),
1387 local_name!("fedistantlight") => *name = local_name!("feDistantLight"),
1388 local_name!("fedropshadow") => *name = local_name!("feDropShadow"),
1389 local_name!("feflood") => *name = local_name!("feFlood"),
1390 local_name!("fefunca") => *name = local_name!("feFuncA"),
1391 local_name!("fefuncb") => *name = local_name!("feFuncB"),
1392 local_name!("fefuncg") => *name = local_name!("feFuncG"),
1393 local_name!("fefuncr") => *name = local_name!("feFuncR"),
1394 local_name!("fegaussianblur") => *name = local_name!("feGaussianBlur"),
1395 local_name!("feimage") => *name = local_name!("feImage"),
1396 local_name!("femerge") => *name = local_name!("feMerge"),
1397 local_name!("femergenode") => *name = local_name!("feMergeNode"),
1398 local_name!("femorphology") => *name = local_name!("feMorphology"),
1399 local_name!("feoffset") => *name = local_name!("feOffset"),
1400 local_name!("fepointlight") => *name = local_name!("fePointLight"),
1401 local_name!("fespecularlighting") => *name = local_name!("feSpecularLighting"),
1402 local_name!("fespotlight") => *name = local_name!("feSpotLight"),
1403 local_name!("fetile") => *name = local_name!("feTile"),
1404 local_name!("feturbulence") => *name = local_name!("feTurbulence"),
1405 local_name!("foreignobject") => *name = local_name!("foreignObject"),
1406 local_name!("glyphref") => *name = local_name!("glyphRef"),
1407 local_name!("lineargradient") => *name = local_name!("linearGradient"),
1408 local_name!("radialgradient") => *name = local_name!("radialGradient"),
1409 local_name!("textpath") => *name = local_name!("textPath"),
1410 _ => (),
1411 }
1412 }
1413
1414 fn adjust_attributes<F>(&mut self, tag: &mut Tag, mut map: F)
1415 where F: FnMut(LocalName) -> Option<QualName>,
1416 {
1417 for &mut Attribute { ref mut name, .. } in &mut tag.attrs {
1418 if let Some(replacement) = map(name.local.clone()) {
1419 *name = replacement;
1420 }
1421 }
1422 }
1423
1424 fn adjust_svg_attributes(&mut self, tag: &mut Tag) {
1425 self.adjust_attributes(tag, |k| match k {
1426 local_name!("attributename") => Some(qualname!("", "attributeName")),
1427 local_name!("attributetype") => Some(qualname!("", "attributeType")),
1428 local_name!("basefrequency") => Some(qualname!("", "baseFrequency")),
1429 local_name!("baseprofile") => Some(qualname!("", "baseProfile")),
1430 local_name!("calcmode") => Some(qualname!("", "calcMode")),
1431 local_name!("clippathunits") => Some(qualname!("", "clipPathUnits")),
1432 local_name!("diffuseconstant") => Some(qualname!("", "diffuseConstant")),
1433 local_name!("edgemode") => Some(qualname!("", "edgeMode")),
1434 local_name!("filterunits") => Some(qualname!("", "filterUnits")),
1435 local_name!("glyphref") => Some(qualname!("", "glyphRef")),
1436 local_name!("gradienttransform") => Some(qualname!("", "gradientTransform")),
1437 local_name!("gradientunits") => Some(qualname!("", "gradientUnits")),
1438 local_name!("kernelmatrix") => Some(qualname!("", "kernelMatrix")),
1439 local_name!("kernelunitlength") => Some(qualname!("", "kernelUnitLength")),
1440 local_name!("keypoints") => Some(qualname!("", "keyPoints")),
1441 local_name!("keysplines") => Some(qualname!("", "keySplines")),
1442 local_name!("keytimes") => Some(qualname!("", "keyTimes")),
1443 local_name!("lengthadjust") => Some(qualname!("", "lengthAdjust")),
1444 local_name!("limitingconeangle") => Some(qualname!("", "limitingConeAngle")),
1445 local_name!("markerheight") => Some(qualname!("", "markerHeight")),
1446 local_name!("markerunits") => Some(qualname!("", "markerUnits")),
1447 local_name!("markerwidth") => Some(qualname!("", "markerWidth")),
1448 local_name!("maskcontentunits") => Some(qualname!("", "maskContentUnits")),
1449 local_name!("maskunits") => Some(qualname!("", "maskUnits")),
1450 local_name!("numoctaves") => Some(qualname!("", "numOctaves")),
1451 local_name!("pathlength") => Some(qualname!("", "pathLength")),
1452 local_name!("patterncontentunits") => Some(qualname!("", "patternContentUnits")),
1453 local_name!("patterntransform") => Some(qualname!("", "patternTransform")),
1454 local_name!("patternunits") => Some(qualname!("", "patternUnits")),
1455 local_name!("pointsatx") => Some(qualname!("", "pointsAtX")),
1456 local_name!("pointsaty") => Some(qualname!("", "pointsAtY")),
1457 local_name!("pointsatz") => Some(qualname!("", "pointsAtZ")),
1458 local_name!("preservealpha") => Some(qualname!("", "preserveAlpha")),
1459 local_name!("preserveaspectratio") => Some(qualname!("", "preserveAspectRatio")),
1460 local_name!("primitiveunits") => Some(qualname!("", "primitiveUnits")),
1461 local_name!("refx") => Some(qualname!("", "refX")),
1462 local_name!("refy") => Some(qualname!("", "refY")),
1463 local_name!("repeatcount") => Some(qualname!("", "repeatCount")),
1464 local_name!("repeatdur") => Some(qualname!("", "repeatDur")),
1465 local_name!("requiredextensions") => Some(qualname!("", "requiredExtensions")),
1466 local_name!("requiredfeatures") => Some(qualname!("", "requiredFeatures")),
1467 local_name!("specularconstant") => Some(qualname!("", "specularConstant")),
1468 local_name!("specularexponent") => Some(qualname!("", "specularExponent")),
1469 local_name!("spreadmethod") => Some(qualname!("", "spreadMethod")),
1470 local_name!("startoffset") => Some(qualname!("", "startOffset")),
1471 local_name!("stddeviation") => Some(qualname!("", "stdDeviation")),
1472 local_name!("stitchtiles") => Some(qualname!("", "stitchTiles")),
1473 local_name!("surfacescale") => Some(qualname!("", "surfaceScale")),
1474 local_name!("systemlanguage") => Some(qualname!("", "systemLanguage")),
1475 local_name!("tablevalues") => Some(qualname!("", "tableValues")),
1476 local_name!("targetx") => Some(qualname!("", "targetX")),
1477 local_name!("targety") => Some(qualname!("", "targetY")),
1478 local_name!("textlength") => Some(qualname!("", "textLength")),
1479 local_name!("viewbox") => Some(qualname!("", "viewBox")),
1480 local_name!("viewtarget") => Some(qualname!("", "viewTarget")),
1481 local_name!("xchannelselector") => Some(qualname!("", "xChannelSelector")),
1482 local_name!("ychannelselector") => Some(qualname!("", "yChannelSelector")),
1483 local_name!("zoomandpan") => Some(qualname!("", "zoomAndPan")),
1484 _ => None,
1485 });
1486 }
1487
1488 fn adjust_mathml_attributes(&mut self, tag: &mut Tag) {
1489 self.adjust_attributes(tag, |k| match k {
1490 local_name!("definitionurl") => Some(qualname!("", "definitionURL")),
1491 _ => None,
1492 });
1493 }
1494
1495 fn adjust_foreign_attributes(&mut self, tag: &mut Tag) {
1496 self.adjust_attributes(tag, |k| match k {
1497 local_name!("xlink:actuate") => Some(qualname!("xlink" xlink "actuate")),
1498 local_name!("xlink:arcrole") => Some(qualname!("xlink" xlink "arcrole")),
1499 local_name!("xlink:href") => Some(qualname!("xlink" xlink "href")),
1500 local_name!("xlink:role") => Some(qualname!("xlink" xlink "role")),
1501 local_name!("xlink:show") => Some(qualname!("xlink" xlink "show")),
1502 local_name!("xlink:title") => Some(qualname!("xlink" xlink "title")),
1503 local_name!("xlink:type") => Some(qualname!("xlink" xlink "type")),
1504 local_name!("xml:base") => Some(qualname!("xml" xml "base")),
1505 local_name!("xml:lang") => Some(qualname!("xml" xml "lang")),
1506 local_name!("xml:space") => Some(qualname!("xml" xml "space")),
1507 local_name!("xmlns") => Some(qualname!("" xmlns "xmlns")),
1508 local_name!("xmlns:xlink") => Some(qualname!("xmlns" xmlns "xlink")),
1509 _ => None,
1510 });
1511 }
1512
1513 fn foreign_start_tag(&mut self, mut tag: Tag) -> ProcessResult<Handle> {
1514 let current_ns = self.sink.elem_name(self.adjusted_current_node()).ns.clone();
1515 match current_ns {
1516 ns!(mathml) => self.adjust_mathml_attributes(&mut tag),
1517 ns!(svg) => {
1518 self.adjust_svg_tag_name(&mut tag);
1519 self.adjust_svg_attributes(&mut tag);
1520 }
1521 _ => (),
1522 }
1523 self.adjust_foreign_attributes(&mut tag);
1524 if tag.self_closing {
1525 // FIXME(#118): <script /> in SVG
1526 self.insert_element(NoPush, current_ns, tag.name, tag.attrs);
1527 DoneAckSelfClosing
1528 } else {
1529 self.insert_element(Push, current_ns, tag.name, tag.attrs);
1530 Done
1531 }
1532 }
1533
1534 fn unexpected_start_tag_in_foreign_content(&mut self, tag: Tag) -> ProcessResult<Handle> {
1535 self.unexpected(&tag);
1536 if self.is_fragment() {
1537 self.foreign_start_tag(tag)
1538 } else {
1539 self.pop();
1540 while !self.current_node_in(|n| {
1541 *n.ns == ns!(html) ||
1542 mathml_text_integration_point(n) ||
1543 svg_html_integration_point(n)
1544 }) {
1545 self.pop();
1546 }
1547 ReprocessForeign(TagToken(tag))
1548 }
1549 }
1550}
1551
ea8adc8c
XL
1552#[cfg(test)]
1553#[allow(non_snake_case)]
1554mod test {
1555 use markup5ever::interface::{QuirksMode, Quirks, LimitedQuirks, NoQuirks};
1556 use markup5ever::interface::{NodeOrText, AppendNode, AppendText};
1557 use markup5ever::interface::{TreeSink, Tracer, ElementFlags};
1558
1559 use super::types::*;
ea8adc8c
XL
1560
1561 use ExpandedName;
1562 use QualName;
1563 use tendril::StrTendril;
1564 use tendril::stream::{TendrilSink, Utf8LossyDecoder};
1565
1566 use tokenizer;
1567 use tokenizer::{Tokenizer, TokenizerOpts};
1568 use tokenizer::{Doctype, StartTag, Tag, TokenSink};
1569 use tokenizer::states as tok_state;
1570
1571 use util::str::is_ascii_whitespace;
1572
1573 use std::default::Default;
1574 use std::mem::replace;
1575 use std::borrow::Cow;
1576 use std::borrow::Cow::Borrowed;
1577 use std::collections::VecDeque;
1578
1579 use driver::*;
1580 use super::{TreeBuilderOpts, TreeBuilder};
1581 use markup5ever::Attribute;
1582 use rcdom::{Node, Handle, RcDom, NodeData};
1583
1584 pub struct LineCountingDOM {
1585 pub line_vec: Vec<(QualName, u64)>,
1586 pub current_line: u64,
1587 pub rcdom: RcDom,
1588 }
1589
1590 impl TreeSink for LineCountingDOM {
1591 type Output = Self;
1592
1593 fn finish(self) -> Self { self }
1594
1595 type Handle = Handle;
1596
1597 fn parse_error(&mut self, msg: Cow<'static, str>) {
1598 self.rcdom.parse_error(msg);
1599 }
1600
1601 fn get_document(&mut self) -> Handle {
1602 self.rcdom.get_document()
1603 }
1604
1605 fn get_template_contents(&mut self, target: &Handle) -> Handle {
1606 self.rcdom.get_template_contents(target)
1607 }
1608
1609 fn set_quirks_mode(&mut self, mode: QuirksMode) {
1610 self.rcdom.set_quirks_mode(mode)
1611 }
1612
1613 fn same_node(&self, x: &Handle, y: &Handle) -> bool {
1614 self.rcdom.same_node(x, y)
1615 }
1616
1617 fn elem_name<'a>(&'a self, target: &'a Handle) -> ExpandedName<'a> {
1618 self.rcdom.elem_name(target)
1619 }
1620
1621 fn create_element(&mut self, name: QualName, attrs: Vec<Attribute>, flags: ElementFlags)
1622 -> Handle {
1623 self.line_vec.push((name.clone(), self.current_line));
1624 self.rcdom.create_element(name, attrs, flags)
1625 }
1626
1627 fn create_comment(&mut self, text: StrTendril) -> Handle {
1628 self.rcdom.create_comment(text)
1629 }
1630
1631 fn create_pi(&mut self, target: StrTendril, content: StrTendril) -> Handle {
1632 self.rcdom.create_pi(target, content)
1633 }
1634
1635 fn has_parent_node(&self, node: &Handle) -> bool {
1636 self.rcdom.has_parent_node(node)
1637 }
1638
1639 fn append(&mut self, parent: &Handle, child: NodeOrText<Handle>) {
1640 self.rcdom.append(parent, child)
1641 }
1642
1643 fn append_before_sibling(&mut self,
1644 sibling: &Handle,
1645 child: NodeOrText<Handle>) {
1646 self.rcdom.append_before_sibling(sibling, child)
1647 }
1648
ff7c6d11
XL
1649 fn append_based_on_parent_node(
1650 &mut self,
1651 element: &Handle,
1652 prev_element: &Handle,
1653 child: NodeOrText<Handle>) {
1654 self.rcdom.append_based_on_parent_node(element, prev_element, child)
1655 }
1656
ea8adc8c
XL
1657 fn append_doctype_to_document(&mut self,
1658 name: StrTendril,
1659 public_id: StrTendril,
1660 system_id: StrTendril) {
1661 self.rcdom.append_doctype_to_document(name, public_id, system_id);
1662 }
1663
1664 fn add_attrs_if_missing(&mut self, target: &Handle, attrs: Vec<Attribute>) {
1665 self.rcdom.add_attrs_if_missing(target, attrs);
1666 }
1667
1668 fn remove_from_parent(&mut self, target: &Handle) {
1669 self.rcdom.remove_from_parent(target);
1670 }
1671
1672 fn reparent_children(&mut self, node: &Handle, new_parent: &Handle) {
1673 self.rcdom.reparent_children(node, new_parent);
1674 }
1675
1676 fn mark_script_already_started(&mut self, target: &Handle) {
1677 self.rcdom.mark_script_already_started(target);
1678 }
1679
1680 fn set_current_line(&mut self, line_number: u64) {
1681 self.current_line = line_number;
1682 }
1683 }
1684
1685 #[test]
1686 fn check_four_lines() {
1687 // Input
1688 let sink = LineCountingDOM {
1689 line_vec: vec!(),
1690 current_line: 1,
1691 rcdom: RcDom::default(),
1692 };
1693 let opts = ParseOpts::default();
1694 let mut resultTok = parse_document(sink, opts);
1695 resultTok.process(StrTendril::from("<a>\n"));
1696 resultTok.process(StrTendril::from("</a>\n"));
1697 resultTok.process(StrTendril::from("<b>\n"));
1698 resultTok.process(StrTendril::from("</b>"));
1699 // Actual Output
1700 let actual = resultTok.finish();
1701 // Expected Output
1702 let expected = vec![(QualName::new(None, ns!(html), local_name!("html")), 1),
1703 (QualName::new(None, ns!(html), local_name!("head")), 1),
1704 (QualName::new(None, ns!(html), local_name!("body")), 1),
1705 (QualName::new(None, ns!(html), local_name!("a")), 1),
1706 (QualName::new(None, ns!(html), local_name!("b")), 3)];
1707 // Assertion
1708 assert_eq!(actual.line_vec, expected);
1709 }
1710}