fix(html/parser): Fix parsing of cdata in foreign context (#4531)

crates/swc_html_parser/src/lexer/mod.rs

@@ -27,12 +27,12 @@ where
     state: State,
     return_state: State,
     errors: Vec<Error>,
-    in_foreign_node: bool,
     pub last_start_tag_token: Option<Token>,
     pending_tokens: Vec<TokenAndSpan>,
     cur_token: Option<Token>,
     character_reference_code: Option<Vec<(u8, u32)>>,
     temporary_buffer: Option<String>,
+    is_adjusted_current_node_is_element_in_html_namespace: Option<bool>,
     doctype_keyword: Option<String>,
     last_emitted_error_pos: Option<BytePos>,
 }
@@ -54,12 +54,12 @@ where
             state: State::Data,
             return_state: State::Data,
             errors: vec![],
-            in_foreign_node: false,
             last_start_tag_token: None,
             pending_tokens: vec![],
             cur_token: None,
             character_reference_code: None,
             temporary_buffer: None,
+            is_adjusted_current_node_is_element_in_html_namespace: None,
             doctype_keyword: None,
             last_emitted_error_pos: None,
         }
@@ -205,13 +205,17 @@ where
         self.input.reset_to(state.pos);
     }
 
-    fn set_input_state(&mut self, state: State) {
-        self.state = state;
-    }
-
     fn take_errors(&mut self) -> Vec<Error> {
         take(&mut self.errors)
     }
+
+    fn set_adjusted_current_node_to_html_namespace(&mut self, value: bool) {
+        self.is_adjusted_current_node_is_element_in_html_namespace = Some(value);
+    }
+
+    fn set_input_state(&mut self, state: State) {
+        self.state = state;
+    }
 }
 
 impl<I> Lexer<I>
@@ -3108,7 +3112,7 @@ where
                                         Some(a2 @ 'a' | a2 @ 'A') => {
                                             match self.consume_next_char() {
                                                 Some('[') => {
-                                                    if self.in_foreign_node {
+                                                    if let Some(false) = self.is_adjusted_current_node_is_element_in_html_namespace {
                                                         self.state = State::CdataSection;
                                                     } else {
                                                         self.emit_error(

crates/swc_html_parser/src/parser/input.rs

@@ -20,9 +20,11 @@ pub trait ParserInput {
 
     fn reset(&mut self, state: &Self::State);
 
+    fn take_errors(&mut self) -> Vec<Error>;
+
     fn set_input_state(&mut self, state: State);
 
-    fn take_errors(&mut self) -> Vec<Error>;
+    fn set_adjusted_current_node_to_html_namespace(&mut self, value: bool);
 }
 
 #[derive(Debug)]
@@ -120,4 +122,9 @@ where
     pub(super) fn set_input_state(&mut self, state: State) {
         self.input.set_input_state(state);
     }
+
+    pub(super) fn set_adjusted_current_node_to_html_namespace(&mut self, value: bool) {
+        self.input
+            .set_adjusted_current_node_to_html_namespace(value);
+    }
 }

crates/swc_html_parser/src/parser/mod.rs

@@ -278,17 +278,26 @@ where
         // to the current insertion mode in HTML content.
         let adjusted_current_node = self.get_adjusted_current_node();
 
+        let is_element_in_html_namespace = is_element_in_html_namespace(adjusted_current_node);
+        let is_mathml_text_integration_point =
+            is_mathml_text_integration_point(adjusted_current_node);
+        let is_mathml_annotation_xml = is_mathml_annotation_xml(adjusted_current_node);
+        let is_html_integration_point = is_html_integration_point(adjusted_current_node);
+
+        self.input
+            .set_adjusted_current_node_to_html_namespace(is_element_in_html_namespace);
+
         if self.open_elements_stack.items.is_empty()
-            || is_element_in_html_namespace(adjusted_current_node)
+            || is_element_in_html_namespace
-            || (is_mathml_text_integration_point(adjusted_current_node)
+            || (is_mathml_text_integration_point
                 && matches!(&token_and_info.token, Token::StartTag { tag_name, .. } if &*tag_name != "mglyph" &&  &*tag_name != "malignmark"))
-            || (is_mathml_text_integration_point(adjusted_current_node)
+            || (is_mathml_text_integration_point
                 && matches!(&token_and_info.token, Token::Character { .. }))
-            || (is_mathml_annotation_xml(adjusted_current_node)
+            || (is_mathml_annotation_xml
                 && matches!(&token_and_info.token, Token::StartTag { tag_name, .. } if &*tag_name == "svg"))
-            || (is_html_integration_point(adjusted_current_node)
+            || (is_html_integration_point
                 && matches!(&token_and_info.token, Token::StartTag { .. }))
-            || (is_html_integration_point(adjusted_current_node)
+            || (is_html_integration_point
                 && matches!(&token_and_info.token, Token::Character { .. }))
             || matches!(&token_and_info.token, Token::Eof)
         {

crates/swc_html_parser/tests/recovery/element/svg-1/input.html

@@ -0,0 +1,2 @@
+<svg><![CDATA[foo
+bar]]>

crates/swc_html_parser/tests/recovery/element/svg-1/output.json

@@ -0,0 +1,75 @@
+{
+  "type": "Document",
+  "span": {
+    "start": 0,
+    "end": 21,
+    "ctxt": 0
+  },
+  "mode": "no-quirks",
+  "children": [
+    {
+      "type": "Element",
+      "span": {
+        "start": 0,
+        "end": 21,
+        "ctxt": 0
+      },
+      "tagName": "html",
+      "namespace": "http://www.w3.org/1999/xhtml",
+      "attributes": [],
+      "children": [
+        {
+          "type": "Element",
+          "span": {
+            "start": 0,
+            "end": 5,
+            "ctxt": 0
+          },
+          "tagName": "head",
+          "namespace": "http://www.w3.org/1999/xhtml",
+          "attributes": [],
+          "children": [],
+          "content": null
+        },
+        {
+          "type": "Element",
+          "span": {
+            "start": 0,
+            "end": 21,
+            "ctxt": 0
+          },
+          "tagName": "body",
+          "namespace": "http://www.w3.org/1999/xhtml",
+          "attributes": [],
+          "children": [
+            {
+              "type": "Element",
+              "span": {
+                "start": 0,
+                "end": 21,
+                "ctxt": 0
+              },
+              "tagName": "svg",
+              "namespace": "http://www.w3.org/2000/svg",
+              "attributes": [],
+              "children": [
+                {
+                  "type": "Text",
+                  "span": {
+                    "start": 5,
+                    "end": 21,
+                    "ctxt": 0
+                  },
+                  "value": "foo\nbar"
+                }
+              ],
+              "content": null
+            }
+          ],
+          "content": null
+        }
+      ],
+      "content": null
+    }
+  ]
+}

crates/swc_html_parser/tests/recovery/element/svg-1/output.stderr

@@ -0,0 +1,6 @@
+
+  x Unexpected token
+   ,-[$DIR/tests/recovery/element/svg-1/input.html:1:1]
+ 1 | ,-> <svg><![CDATA[foo
+ 2 | `-> bar]]>
+   `----

crates/swc_html_parser/tests/recovery/element/svg-1/span.rust-debug

@@ -0,0 +1,42 @@
+
+  x Document
+   ,-[$DIR/tests/recovery/element/svg-1/input.html:1:1]
+ 1 | ,-> <svg><![CDATA[foo
+ 2 | `-> bar]]>
+   `----
+
+  x Child
+   ,-[$DIR/tests/recovery/element/svg-1/input.html:1:1]
+ 1 | ,-> <svg><![CDATA[foo
+ 2 | `-> bar]]>
+   `----
+
+  x Element
+   ,-[$DIR/tests/recovery/element/svg-1/input.html:1:1]
+ 1 | ,-> <svg><![CDATA[foo
+ 2 | `-> bar]]>
+   `----
+
+  x Child
+   ,-[$DIR/tests/recovery/element/svg-1/input.html:1:1]
+ 1 | <svg><![CDATA[foo
+   : ^^^^^
+   `----
+
+  x Element
+   ,-[$DIR/tests/recovery/element/svg-1/input.html:1:1]
+ 1 | <svg><![CDATA[foo
+   : ^^^^^
+   `----
+
+  x Child
+   ,-[$DIR/tests/recovery/element/svg-1/input.html:1:1]
+ 1 | ,-> <svg><![CDATA[foo
+ 2 | `-> bar]]>
+   `----
+
+  x Text
+   ,-[$DIR/tests/recovery/element/svg-1/input.html:1:1]
+ 1 | ,-> <svg><![CDATA[foo
+ 2 | `-> bar]]>
+   `----