feat(parser/html): emit diagnostics for parsing errors (#3846)

crates/biome_html_parser/src/syntax/mod.rs

@@ -1,14 +1,9 @@
 mod parse_error;
 
 use crate::parser::HtmlParser;
-use crate::syntax::parse_error::{expected_attribute, expected_child};
+use crate::syntax::parse_error::*;
 use crate::token_source::HtmlLexContext;
-use biome_html_syntax::HtmlSyntaxKind::{
-    HTML_ATTRIBUTE, HTML_ATTRIBUTE_INITIALIZER_CLAUSE, HTML_ATTRIBUTE_LIST, HTML_BOGUS_ELEMENT,
-    HTML_CLOSING_ELEMENT, HTML_CONTENT, HTML_DIRECTIVE, HTML_ELEMENT, HTML_ELEMENT_LIST,
-    HTML_LITERAL, HTML_NAME, HTML_OPENING_ELEMENT, HTML_ROOT, HTML_SELF_CLOSING_ELEMENT,
-    HTML_STRING, HTML_STRING_LITERAL, UNICODE_BOM,
-};
+use biome_html_syntax::HtmlSyntaxKind::*;
 use biome_html_syntax::{HtmlSyntaxKind, T};
 use biome_parser::parse_lists::ParseNodeList;
 use biome_parser::parse_recovery::{ParseRecoveryTokenSet, RecoveryResult};
@@ -59,28 +54,29 @@ fn parse_element(p: &mut HtmlParser) -> ParsedSyntax {
     let m = p.start();
 
     p.bump(T![<]);
-    // TODO: handle error
-    parse_literal(p).ok();
+    parse_literal(p).or_add_diagnostic(p, expected_element_name);
 
     AttributeList.parse_list(p);
 
     if p.at(T![/]) {
         p.bump(T![/]);
-        p.bump(T![>]);
+        p.expect(T![>]);
         Present(m.complete(p, HTML_SELF_CLOSING_ELEMENT))
     } else {
-        p.bump_with_context(T![>], HtmlLexContext::ElementList);
+        p.expect_with_context(T![>], HtmlLexContext::ElementList);
         let opening = m.complete(p, HTML_OPENING_ELEMENT);
         ElementList.parse_list(p);
-        // TODO: handle error
-        parse_closing_element(p).ok();
+        parse_closing_element(p).or_add_diagnostic(p, expected_closing_tag);
         let previous = opening.precede(p);
 
         Present(previous.complete(p, HTML_ELEMENT))
     }
 }
 
 fn parse_closing_element(p: &mut HtmlParser) -> ParsedSyntax {
+    if !p.at(T![<]) || !p.nth_at(1, T![/]) {
+        return Absent;
+    }
     let m = p.start();
     p.bump(T![<]);
     p.bump(T![/]);
@@ -112,7 +108,8 @@ impl ParseNodeList for ElementList {
     fn is_at_list_end(&self, p: &mut Self::Parser<'_>) -> bool {
         let at_l_angle0 = p.at(T![<]);
         let at_slash1 = p.nth_at(1, T![/]);
-        at_l_angle0 && at_slash1
+        let at_eof = p.at(EOF);
+        at_l_angle0 && at_slash1 || at_eof
     }
 
     fn recover(
@@ -142,7 +139,7 @@ impl ParseNodeList for AttributeList {
     }
 
     fn is_at_list_end(&self, p: &mut Self::Parser<'_>) -> bool {
-        p.at(T![>]) || p.at(T![/])
+        p.at(T![>]) || p.at(T![/]) || p.at(EOF)
     }
 
     fn recover(
@@ -163,8 +160,7 @@ fn parse_attribute(p: &mut HtmlParser) -> ParsedSyntax {
         return Absent;
     }
     let m = p.start();
-    // TODO: handle error
-    parse_literal(p).ok();
+    parse_literal(p).or_add_diagnostic(p, expected_attribute);
     if p.at(T![=]) {
         parse_attribute_initializer(p).ok();
         Present(m.complete(p, HTML_ATTRIBUTE))
@@ -201,7 +197,6 @@ fn parse_attribute_initializer(p: &mut HtmlParser) -> ParsedSyntax {
     }
     let m = p.start();
     p.bump(T![=]);
-    // TODO: handle error
-    parse_string_literal(p).ok();
+    parse_string_literal(p).or_add_diagnostic(p, expected_initializer);
     Present(m.complete(p, HTML_ATTRIBUTE_INITIALIZER_CLAUSE))
 }

crates/biome_html_parser/src/syntax/parse_error.rs

@@ -10,3 +10,32 @@ pub(crate) fn expected_attribute(p: &HtmlParser, range: TextRange) -> ParseDiagn
 pub(crate) fn expected_child(p: &HtmlParser, range: TextRange) -> ParseDiagnostic {
     expect_one_of(&["element", "text"], range).into_diagnostic(p)
 }
+
+/// The parser was expecting a value for an attribute initializer clause.
+///
+/// ```html
+/// <div id= />
+///         ^ expected initializer
+/// ```
+pub(crate) fn expected_initializer(p: &HtmlParser, range: TextRange) -> ParseDiagnostic {
+    expected_node("initializer", range, p).into_diagnostic(p)
+}
+
+/// The parser encountered a tag that does not have a corresponding closing tag.
+///
+/// ```html
+/// <div>foo
+/// ```
+pub(crate) fn expected_closing_tag(p: &HtmlParser, range: TextRange) -> ParseDiagnostic {
+    expected_node("closing tag", range, p).into_diagnostic(p)
+}
+
+/// The parser was encountered a tag that does not have a name.
+///
+/// ```html
+/// <>
+/// ^ expected element name
+/// ```
+pub(crate) fn expected_element_name(p: &HtmlParser, range: TextRange) -> ParseDiagnostic {
+    expected_node("element name", range, p).into_diagnostic(p)
+}

crates/biome_html_parser/tests/html_specs/error/attributes/missing-initializer.html

@@ -0,0 +1 @@
+<div class=></div>

crates/biome_html_parser/tests/html_specs/error/attributes/missing-initializer.html.snap

@@ -0,0 +1,99 @@
+---
+source: crates/biome_html_parser/tests/spec_test.rs
+expression: snapshot
+---
+## Input
+
+```html
+<div class=></div>
+
+```
+
+
+## AST
+
+```
+HtmlRoot {
+    bom_token: missing (optional),
+    directive: missing (optional),
+    html: HtmlElement {
+        opening_element: HtmlOpeningElement {
+            l_angle_token: L_ANGLE@0..1 "<" [] [],
+            name: HtmlName {
+                value_token: HTML_LITERAL@1..5 "div" [] [Whitespace(" ")],
+            },
+            attributes: HtmlAttributeList [
+                HtmlAttribute {
+                    name: HtmlName {
+                        value_token: HTML_LITERAL@5..10 "class" [] [],
+                    },
+                    initializer: HtmlAttributeInitializerClause {
+                        eq_token: EQ@10..11 "=" [] [],
+                        value: missing (required),
+                    },
+                },
+            ],
+            r_angle_token: R_ANGLE@11..12 ">" [] [],
+        },
+        children: HtmlElementList [],
+        closing_element: HtmlClosingElement {
+            l_angle_token: L_ANGLE@12..13 "<" [] [],
+            slash_token: SLASH@13..14 "/" [] [],
+            name: HtmlName {
+                value_token: HTML_LITERAL@14..17 "div" [] [],
+            },
+            r_angle_token: R_ANGLE@17..18 ">" [] [],
+        },
+    },
+    eof_token: EOF@18..19 "" [Newline("\n")] [],
+}
+```
+
+## CST
+
+```
+0: [email protected]
+  0: (empty)
+  1: (empty)
+  2: [email protected]
+    0: [email protected]
+      0: [email protected] "<" [] []
+      1: [email protected]
+        0: [email protected] "div" [] [Whitespace(" ")]
+      2: [email protected]
+        0: [email protected]
+          0: [email protected]
+            0: [email protected] "class" [] []
+          1: [email protected]
+            0: [email protected] "=" [] []
+            1: (empty)
+      3: [email protected] ">" [] []
+    1: [email protected]
+    2: [email protected]
+      0: [email protected] "<" [] []
+      1: [email protected] "/" [] []
+      2: [email protected]
+        0: [email protected] "div" [] []
+      3: [email protected] ">" [] []
+  3: [email protected] "" [Newline("\n")] []
+
+```
+
+## Diagnostics
+
+```
+missing-initializer.html:1:12 parse ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+  × Expected an initializer but instead found '>'.
+  
+  > 1 │ <div class=></div>
+      │            ^
+    2 │ 
+  
+  i Expected an initializer here.
+  
+  > 1 │ <div class=></div>
+      │            ^
+    2 │ 
+  
+```

crates/biome_html_parser/tests/html_specs/error/element/missing-close-tag.html

@@ -0,0 +1 @@
+<div>foo

crates/biome_html_parser/tests/html_specs/error/element/missing-close-tag.html.snap

@@ -0,0 +1,77 @@
+---
+source: crates/biome_html_parser/tests/spec_test.rs
+expression: snapshot
+---
+## Input
+
+```html
+<div>foo
+
+```
+
+
+## AST
+
+```
+HtmlRoot {
+    bom_token: missing (optional),
+    directive: missing (optional),
+    html: HtmlElement {
+        opening_element: HtmlOpeningElement {
+            l_angle_token: L_ANGLE@0..1 "<" [] [],
+            name: HtmlName {
+                value_token: HTML_LITERAL@1..4 "div" [] [],
+            },
+            attributes: HtmlAttributeList [],
+            r_angle_token: R_ANGLE@4..5 ">" [] [],
+        },
+        children: HtmlElementList [
+            HtmlContent {
+                value_token: HTML_LITERAL@5..9 "foo\n" [] [],
+            },
+        ],
+        closing_element: missing (required),
+    },
+    eof_token: EOF@9..9 "" [] [],
+}
+```
+
+## CST
+
+```
+0: [email protected]
+  0: (empty)
+  1: (empty)
+  2: [email protected]
+    0: [email protected]
+      0: [email protected] "<" [] []
+      1: [email protected]
+        0: [email protected] "div" [] []
+      2: [email protected]
+      3: [email protected] ">" [] []
+    1: [email protected]
+      0: [email protected]
+        0: [email protected] "foo\n" [] []
+    2: (empty)
+  3: [email protected] "" [] []
+
+```
+
+## Diagnostics
+
+```
+missing-close-tag.html:2:1 parse ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+  × Expected a closing tag but instead found the end of the file.
+  
+    1 │ <div>foo
+  > 2 │ 
+      │ 
+  
+  i Expected a closing tag here.
+  
+    1 │ <div>foo
+  > 2 │ 
+      │ 
+  
+```

crates/biome_html_parser/tests/html_specs/error/element/missing-element-name.html

@@ -0,0 +1 @@
+<></>

crates/biome_html_parser/tests/html_specs/error/element/missing-element-name.html.snap

@@ -0,0 +1,77 @@
+---
+source: crates/biome_html_parser/tests/spec_test.rs
+expression: snapshot
+---
+## Input
+
+```html
+<></>
+
+```
+
+
+## AST
+
+```
+HtmlRoot {
+    bom_token: missing (optional),
+    directive: missing (optional),
+    html: HtmlElement {
+        opening_element: HtmlOpeningElement {
+            l_angle_token: L_ANGLE@0..1 "<" [] [],
+            name: missing (required),
+            attributes: HtmlAttributeList [],
+            r_angle_token: R_ANGLE@1..2 ">" [] [],
+        },
+        children: HtmlElementList [],
+        closing_element: HtmlClosingElement {
+            l_angle_token: L_ANGLE@2..3 "<" [] [],
+            slash_token: SLASH@3..4 "/" [] [],
+            name: missing (required),
+            r_angle_token: R_ANGLE@4..5 ">" [] [],
+        },
+    },
+    eof_token: EOF@5..6 "" [Newline("\n")] [],
+}
+```
+
+## CST
+
+```
+0: [email protected]
+  0: (empty)
+  1: (empty)
+  2: [email protected]
+    0: [email protected]
+      0: [email protected] "<" [] []
+      1: (empty)
+      2: [email protected]
+      3: [email protected] ">" [] []
+    1: [email protected]
+    2: [email protected]
+      0: [email protected] "<" [] []
+      1: [email protected] "/" [] []
+      2: (empty)
+      3: [email protected] ">" [] []
+  3: [email protected] "" [Newline("\n")] []
+
+```
+
+## Diagnostics
+
+```
+missing-element-name.html:1:2 parse ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+
+  × Expected an element name but instead found '>'.
+  
+  > 1 │ <></>
+      │  ^
+    2 │ 
+  
+  i Expected an element name here.
+  
+  > 1 │ <></>
+      │  ^
+    2 │ 
+  
+```

crates/biome_html_parser/tests/html_specs/ok/attributes/multiline-attributes.html

@@ -0,0 +1,6 @@
+<div
+	class="foo"
+	role="button"
+>
+	foo
+</div>