Ensure parsers initiated from DOMParser always complete. (#33056)

* Ensure parsers initiated from DOMParser always complete.

Signed-off-by: Josh Matthews <josh@joshmatthews.net>

* Add test for parseFromString with async parser.

Signed-off-by: Josh Matthews <josh@joshmatthews.net>

* Add expected failure.

Signed-off-by: Josh Matthews <josh@joshmatthews.net>

---------

Signed-off-by: Josh Matthews <josh@joshmatthews.net>

components/script/dom/servoparser/mod.rs

@@ -128,12 +128,6 @@ pub struct ServoParser {
     prefetch_input: DomRefCell<BufferQueue>,
 }
 
-#[derive(PartialEq)]
-enum LastChunkState {
-    Received,
-    NotReceived,
-}
-
 pub struct ElementAttribute {
     name: QualName,
     value: DOMString,
@@ -161,7 +155,6 @@ impl ServoParser {
             ServoParser::new(
                 document,
                 Tokenizer::AsyncHtml(self::async_html::Tokenizer::new(document, url, None)),
-                LastChunkState::NotReceived,
                 ParserKind::Normal,
             )
         } else {
@@ -173,14 +166,13 @@ impl ServoParser {
                     None,
                     ParsingAlgorithm::Normal,
                 )),
-                LastChunkState::NotReceived,
                 ParserKind::Normal,
             )
         };
 
         // Set as the document's current parser and initialize with `input`, if given.
         if let Some(input) = input {
-            parser.parse_string_chunk(String::from(input));
+            parser.parse_complete_string_chunk(String::from(input));
         } else {
             parser.document.set_current_parser(Some(&parser));
         }
@@ -239,10 +231,9 @@ impl ServoParser {
                 Some(fragment_context),
                 ParsingAlgorithm::Fragment,
             )),
-            LastChunkState::Received,
             ParserKind::Normal,
         );
-        parser.parse_string_chunk(String::from(input));
+        parser.parse_complete_string_chunk(String::from(input));
 
         // Step 14.
         let root_element = document.GetDocumentElement().expect("no document element");
@@ -260,7 +251,6 @@ impl ServoParser {
                 None,
                 ParsingAlgorithm::Normal,
             )),
-            LastChunkState::NotReceived,
             ParserKind::ScriptCreated,
         );
         *parser.bom_sniff.borrow_mut() = None;
@@ -271,13 +261,12 @@ impl ServoParser {
         let parser = ServoParser::new(
             document,
             Tokenizer::Xml(self::xml::Tokenizer::new(document, url)),
-            LastChunkState::NotReceived,
             ParserKind::Normal,
         );
 
         // Set as the document's current parser and initialize with `input`, if given.
         if let Some(input) = input {
-            parser.parse_string_chunk(String::from(input));
+            parser.parse_complete_string_chunk(String::from(input));
         } else {
             parser.document.set_current_parser(Some(&parser));
         }
@@ -421,12 +410,7 @@ impl ServoParser {
     }
 
     #[allow(crown::unrooted_must_root)]
-    fn new_inherited(
+    fn new_inherited(document: &Document, tokenizer: Tokenizer, kind: ParserKind) -> Self {
-        document: &Document,
-        tokenizer: Tokenizer,
-        last_chunk_state: LastChunkState,
-        kind: ParserKind,
-    ) -> Self {
         ServoParser {
             reflector: Reflector::new(),
             document: Dom::from_ref(document),
@@ -435,7 +419,7 @@ impl ServoParser {
             network_input: DomRefCell::new(BufferQueue::default()),
             script_input: DomRefCell::new(BufferQueue::default()),
             tokenizer: DomRefCell::new(tokenizer),
-            last_chunk_received: Cell::new(last_chunk_state == LastChunkState::Received),
+            last_chunk_received: Cell::new(false),
             suspended: Default::default(),
             script_nesting_level: Default::default(),
             aborted: Default::default(),
@@ -446,19 +430,9 @@ impl ServoParser {
     }
 
     #[allow(crown::unrooted_must_root)]
-    fn new(
+    fn new(document: &Document, tokenizer: Tokenizer, kind: ParserKind) -> DomRoot<Self> {
-        document: &Document,
-        tokenizer: Tokenizer,
-        last_chunk_state: LastChunkState,
-        kind: ParserKind,
-    ) -> DomRoot<Self> {
         reflect_dom_object(
-            Box::new(ServoParser::new_inherited(
+            Box::new(ServoParser::new_inherited(document, tokenizer, kind)),
-                document,
-                tokenizer,
-                last_chunk_state,
-                kind,
-            )),
             document.window(),
         )
     }
@@ -579,9 +553,10 @@ impl ServoParser {
         }
     }
 
-    fn parse_string_chunk(&self, input: String) {
+    fn parse_complete_string_chunk(&self, input: String) {
         self.document.set_current_parser(Some(self));
         self.push_string_input_chunk(input);
+        self.last_chunk_received.set(true);
         if !self.suspended.get() {
             self.parse_sync();
         }

tests/wpt/meta/xhr/responsexml-non-well-formed.htm.ini

@@ -1,4 +1,7 @@
 [responsexml-non-well-formed.htm]
+  [XMLHttpRequest: responseXML non well-formed tests]
+    expected: FAIL
+
   [XMLHttpRequest: responseXML non well-formed tests 1]
     expected: FAIL
 

tests/wpt/mozilla/meta/MANIFEST.json

@@ -12632,6 +12632,13 @@
       }
      ]
     ],
+    "DOMParser-parseFromString.html": [
+     "38715db5d923e4f153e2c4a2679f644fe1e6d14d",
+     [
+      null,
+      {}
+     ]
+    ],
     "DOMParser.html": [
      "f386a3e0191af2c70dcb05790ce7db15dd5ccbf1",
      [

tests/wpt/mozilla/meta/mozilla/DOMParser-parseFromString.html.ini

@@ -0,0 +1,2 @@
+[DOMParser-parseFromString.html]
+  prefs: ["dom.servoparser.async_html_tokenizer.enabled:true"]

tests/wpt/mozilla/tests/mozilla/DOMParser-parseFromString.html

@@ -0,0 +1,15 @@
+<!doctype html>
+<head>
+<title>Verify that parseFromString does not panic</title>
+<script src="/resources/testharness.js"></script>
+<script src="/resources/testharnessreport.js"></script>
+<iframe src="missing"></iframe>
+<script>
+  test(() => {
+    {
+      const parser = new DOMParser();
+      const doc = parser.parseFromString("", "text/html");
+      assert_equals(doc.URL, document.URL);
+    }
+  });
+</script>