fix(parser/html): fix parsing html keyword in doctype

biomejs · Sep 4, 2024 · 7615b8a · 7615b8a
1 parent 30698e4
commit 7615b8a
Show file tree

Hide file tree

Showing 6 changed files with 61 additions and 5 deletions.
diff --git a/crates/biome_html_parser/src/lexer/mod.rs b/crates/biome_html_parser/src/lexer/mod.rs
@@ -2,8 +2,8 @@ mod tests;
 
 use crate::token_source::HtmlLexContext;
 use biome_html_syntax::HtmlSyntaxKind::{
-    COMMENT, DOCTYPE_KW, EOF, ERROR_TOKEN, HTML_LITERAL, HTML_STRING_LITERAL, NEWLINE, TOMBSTONE,
-    UNICODE_BOM, WHITESPACE,
+    COMMENT, DOCTYPE_KW, EOF, ERROR_TOKEN, HTML_KW, HTML_LITERAL, HTML_STRING_LITERAL, NEWLINE,
+    TOMBSTONE, UNICODE_BOM, WHITESPACE,
 };
 use biome_html_syntax::{HtmlSyntaxKind, TextLen, TextSize, T};
 use biome_parser::diagnostic::ParseDiagnostic;
@@ -149,6 +149,7 @@ impl<'src> HtmlLexer<'src> {
 
         match &buffer[..len] {
             b"doctype" | b"DOCTYPE" => DOCTYPE_KW,
+            b"html" | b"HTML" => HTML_KW,
             _ => HTML_LITERAL,
         }
     }

diff --git a/crates/biome_html_parser/src/lexer/tests.rs b/crates/biome_html_parser/src/lexer/tests.rs
@@ -157,7 +157,7 @@ fn doctype_with_quirk() {
         BANG: 1,
         DOCTYPE_KW: 7,
         WHITESPACE: 1,
-        HTML_LITERAL: 4,
+        HTML_KW: 4,
         R_ANGLE: 1,
     }
 }
@@ -170,7 +170,7 @@ fn doctype_with_quirk_and_system() {
         BANG: 1,
         DOCTYPE_KW: 7,
         WHITESPACE: 1,
-        HTML_LITERAL: 4,
+        HTML_KW: 4,
         WHITESPACE: 1,
         HTML_STRING_LITERAL: 44,
         R_ANGLE: 1,

diff --git a/crates/biome_html_parser/src/syntax/mod.rs b/crates/biome_html_parser/src/syntax/mod.rs
@@ -43,6 +43,10 @@ fn parse_doc_type(p: &mut HtmlParser) -> ParsedSyntax {
         p.eat(T![doctype]);
     }
 
+    if p.at(T![html]) {
+        p.eat(T![html]);
+    }
+
     p.eat(T![>]);
 
     Present(m.complete(p, HTML_DIRECTIVE))

diff --git a/crates/biome_html_parser/tests/html_specs/ok/ok2.html b/crates/biome_html_parser/tests/html_specs/ok/ok2.html
@@ -0,0 +1 @@
+<!doctype html>
diff --git a/crates/biome_html_parser/tests/html_specs/ok/ok2.html.snap b/crates/biome_html_parser/tests/html_specs/ok/ok2.html.snap
@@ -0,0 +1,50 @@
+---
+source: crates/biome_html_parser/tests/spec_test.rs
+expression: snapshot
+---
+## Input
+
+```css
+<!doctype html>
+
+```
+
+
+## AST
+
+```
+HtmlRoot {
+    bom_token: missing (optional),
+    directive: HtmlDirective {
+        l_angle_token: L_ANGLE@0..1 "<" [] [],
+        excl_token: BANG@1..2 "!" [] [],
+        doctype_token: DOCTYPE_KW@2..10 "doctype" [] [Whitespace(" ")],
+        html_token: HTML_KW@10..14 "html" [] [],
+        quirk_token: missing (optional),
+        public_id_token: missing (optional),
+        system_id_token: missing (optional),
+        r_angle_token: R_ANGLE@14..15 ">" [] [],
+    },
+    html: missing (optional),
+    eof_token: EOF@15..16 "" [Newline("\n")] [],
+}
+```
+
+## CST
+
+```
+0: [email protected]
+  0: (empty)
+  1: [email protected]
+    0: [email protected] "<" [] []
+    1: [email protected] "!" [] []
+    2: [email protected] "doctype" [] [Whitespace(" ")]
+    3: [email protected] "html" [] []
+    4: (empty)
+    5: (empty)
+    6: (empty)
+    7: [email protected] ">" [] []
+  2: (empty)
+  3: [email protected] "" [Newline("\n")] []
+
+```
diff --git a/xtask/codegen/html.ungram b/xtask/codegen/html.ungram
@@ -116,7 +116,7 @@ HtmlAttribute =
 
 
 // <a href="">
-//       ^^^^
+//        ^^^
 HtmlAttributeInitializerClause =
 	'='
 	value: HtmlString