Unify the way html5ever and xml5ever block on script elements (servo#36284)

Companion PR for servo/html5ever#591

Testing: Covered by WPT
Part of servo#6414,
servo#24898, preparation for
servo/html5ever#590

---------

Signed-off-by: Simon Wülker <[email protected]>

Author: simonwuelker

Cargo.lock

@@ -71,7 +71,7 @@ gstreamer-video = "0.23"
 harfbuzz-sys = "0.6.1"
 headers = "0.4"
 hitrace = "0.1.4"
-html5ever = "0.30"
+html5ever = "0.31"
 http = "1.3"
 http-body-util = "0.1"
 hyper = "1.6"
@@ -92,7 +92,7 @@ log = "0.4"
 mach2 = "0.4"
 malloc_size_of = { package = "servo_malloc_size_of", path = "components/malloc_size_of" }
 malloc_size_of_derive = "0.1"
-markup5ever = "0.15"
+markup5ever = "0.16"
 memmap2 = "0.9.5"
 mime = "0.3.13"
 mime_guess = "2.0.5"
@@ -174,7 +174,7 @@ windows-sys = "0.59"
 wio = "0.2"
 wr_malloc_size_of = { git = "https://github.com/servo/webrender", branch = "0.66" }
 xi-unicode = "0.3.0"
-xml5ever = "0.21"
+xml5ever = "0.22"
 
 [profile.release]
 opt-level = 3

Cargo.toml

@@ -84,6 +84,7 @@ libc = { workspace = true }
 log = { workspace = true }
 malloc_size_of = { workspace = true }
 malloc_size_of_derive = { workspace = true }
+markup5ever = { workspace = true }
 media = { path = "../media" }
 metrics = { path = "../metrics" }
 mime = { workspace = true }

components/script/Cargo.toml

@@ -14,14 +14,14 @@ use crossbeam_channel::{Receiver, Sender, unbounded};
 use html5ever::buffer_queue::BufferQueue;
 use html5ever::tendril::fmt::UTF8;
 use html5ever::tendril::{SendTendril, StrTendril, Tendril};
-use html5ever::tokenizer::{Tokenizer as HtmlTokenizer, TokenizerOpts, TokenizerResult};
+use html5ever::tokenizer::{Tokenizer as HtmlTokenizer, TokenizerOpts};
 use html5ever::tree_builder::{
-    ElementFlags, NextParserState, NodeOrText as HtmlNodeOrText, QuirksMode, TreeBuilder,
-    TreeBuilderOpts, TreeSink,
+    ElementFlags, NodeOrText as HtmlNodeOrText, QuirksMode, TreeBuilder, TreeBuilderOpts, TreeSink,
 };
 use html5ever::{
     Attribute as HtmlAttribute, ExpandedName, QualName, local_name, namespace_url, ns,
 };
+use markup5ever::TokenizerResult;
 use servo_url::ServoUrl;
 use style::context::QuirksMode as ServoQuirksMode;
 
@@ -901,10 +901,6 @@ impl TreeSink for Sink {
         self.send_op(ParseOperation::MarkScriptAlreadyStarted { node: node.id });
     }
 
-    fn complete_script(&self, _: &Self::Handle) -> NextParserState {
-        panic!("complete_script should not be called here!");
-    }
-
     fn reparent_children(&self, parent: &Self::Handle, new_parent: &Self::Handle) {
         self.send_op(ParseOperation::ReparentChildren {
             parent: parent.id,

components/script/dom/servoparser/async_html.rs

@@ -10,9 +10,10 @@ use std::io;
 use html5ever::buffer_queue::BufferQueue;
 use html5ever::serialize::TraversalScope::IncludeNode;
 use html5ever::serialize::{AttrRef, Serialize, Serializer, TraversalScope};
-use html5ever::tokenizer::{Tokenizer as HtmlTokenizer, TokenizerOpts, TokenizerResult};
+use html5ever::tokenizer::{Tokenizer as HtmlTokenizer, TokenizerOpts};
 use html5ever::tree_builder::{TreeBuilder, TreeBuilderOpts};
 use html5ever::{QualName, local_name, namespace_url, ns};
+use markup5ever::TokenizerResult;
 use script_bindings::trace::CustomTraceable;
 use servo_url::ServoUrl;
 use xml5ever::LocalName;

components/script/dom/servoparser/html.rs

@@ -16,10 +16,10 @@ use encoding_rs::Encoding;
 use html5ever::buffer_queue::BufferQueue;
 use html5ever::tendril::fmt::UTF8;
 use html5ever::tendril::{ByteTendril, StrTendril, TendrilSink};
-use html5ever::tokenizer::TokenizerResult;
-use html5ever::tree_builder::{ElementFlags, NextParserState, NodeOrText, QuirksMode, TreeSink};
+use html5ever::tree_builder::{ElementFlags, NodeOrText, QuirksMode, TreeSink};
 use html5ever::{Attribute, ExpandedName, LocalName, QualName, local_name, namespace_url, ns};
 use hyper_serde::Serde;
+use markup5ever::TokenizerResult;
 use mime::{self, Mime};
 use net_traits::request::RequestId;
 use net_traits::{
@@ -1356,15 +1356,6 @@ impl TreeSink for Sink {
         }
     }
 
-    fn complete_script(&self, node: &Dom<Node>) -> NextParserState {
-        if let Some(script) = node.downcast() {
-            self.script.set(Some(script));
-            NextParserState::Suspend
-        } else {
-            NextParserState::Continue
-        }
-    }
-
     fn reparent_children(&self, node: &Dom<Node>, new_parent: &Dom<Node>) {
         while let Some(ref child) = node.GetFirstChild() {
             new_parent.AppendChild(child, CanGc::note()).unwrap();

components/script/dom/servoparser/mod.rs

@@ -10,10 +10,11 @@ use content_security_policy::Destination;
 use html5ever::buffer_queue::BufferQueue;
 use html5ever::tokenizer::states::RawKind;
 use html5ever::tokenizer::{
-    Tag, TagKind, Token, TokenSink, TokenSinkResult, Tokenizer as HtmlTokenizer, TokenizerResult,
+    Tag, TagKind, Token, TokenSink, TokenSinkResult, Tokenizer as HtmlTokenizer,
 };
 use html5ever::{Attribute, LocalName, local_name};
 use js::jsapi::JSTracer;
+use markup5ever::TokenizerResult;
 use net_traits::request::{
     CorsSettings, CredentialsMode, InsecureRequestsPolicy, ParserMetadata, Referrer,
 };

components/script/dom/servoparser/prefetch.rs

@@ -6,13 +6,14 @@
 
 use std::cell::Cell;
 
-use html5ever::tokenizer::TokenizerResult;
+use markup5ever::TokenizerResult;
 use script_bindings::trace::CustomTraceable;
 use servo_url::ServoUrl;
 use xml5ever::buffer_queue::BufferQueue;
 use xml5ever::tokenizer::XmlTokenizer;
 use xml5ever::tree_builder::XmlTreeBuilder;
 
+use crate::dom::bindings::inheritance::Castable;
 use crate::dom::bindings::root::{Dom, DomRoot};
 use crate::dom::document::Document;
 use crate::dom::htmlscriptelement::HTMLScriptElement;
@@ -43,10 +44,17 @@ impl Tokenizer {
     }
 
     pub(crate) fn feed(&self, input: &BufferQueue) -> TokenizerResult<DomRoot<HTMLScriptElement>> {
-        self.inner.run(input);
-        match self.inner.sink.sink.script.take() {
-            Some(script) => TokenizerResult::Script(script),
-            None => TokenizerResult::Done,
+        loop {
+            match self.inner.run(input) {
+                TokenizerResult::Done => return TokenizerResult::Done,
+                TokenizerResult::Script(handle) => {
+                    // Apparently the parser can sometimes create <script> elements without a namespace, resulting
+                    // in them not being HTMLScriptElements.
+                    if let Some(script) = handle.downcast::<HTMLScriptElement>() {
+                        return TokenizerResult::Script(DomRoot::from_ref(script));
+                    }
+                },
+            }
         }
     }