fix(parser/html): fix incorrect parsing when text starts with a quote

biomejs · Sep 20, 2024 · b7a77c2 · b7a77c2
1 parent d137533
commit b7a77c2
Show file tree

Hide file tree

Showing 5 changed files with 132 additions and 24 deletions.
diff --git a/crates/biome_html_parser/src/syntax/mod.rs b/crates/biome_html_parser/src/syntax/mod.rs
@@ -144,7 +144,7 @@ fn parse_closing_element(p: &mut HtmlParser) -> ParsedSyntax {
         p.error(void_element_should_not_have_closing_tag(p, p.cur_range()).into_diagnostic(p));
     }
     let _name = parse_literal(p);
-    p.bump(T![>]);
+    p.bump_with_context(T![>], HtmlLexContext::OutsideTag);
     Present(m.complete(p, HTML_CLOSING_ELEMENT))
 }
 

diff --git a/crates/biome_html_parser/tests/html_specs/ok/element_list.html.snap b/crates/biome_html_parser/tests/html_specs/ok/element_list.html.snap
@@ -82,10 +82,7 @@ HtmlRoot {
                             },
                         },
                         HtmlContent {
-                            value_token: HTML_LITERAL@38..46 "some" [Newline("\n"), Whitespace("\t\t")] [Whitespace(" ")],
-                        },
-                        HtmlContent {
-                            value_token: HTML_LITERAL@46..50 "text" [] [],
+                            value_token: HTML_LITERAL@38..50 "some text" [Newline("\n"), Whitespace("\t\t")] [],
                         },
                         HtmlElement {
                             opening_element: HtmlOpeningElement {
@@ -419,11 +416,9 @@ HtmlRoot {
                 2: [email protected]
                   0: [email protected] "div" [] []
                 3: [email protected] ">" [] []
-            2: [email protected]
-              0: [email protected] "some" [Newline("\n"), Whitespace("\t\t")] [Whitespace(" ")]
-            3: [email protected]
-              0: [email protected] "text" [] []
-            4: [email protected]
+            2: [email protected]
+              0: [email protected] "some text" [Newline("\n"), Whitespace("\t\t")] []
+            3: [email protected]
               0: [email protected]
                 0: [email protected] "<" [Newline("\n"), Whitespace("\t\t")] []
                 1: [email protected]
@@ -437,7 +432,7 @@ HtmlRoot {
                 2: [email protected]
                   0: [email protected] "div" [] []
                 3: [email protected] ">" [] []
-            5: [email protected]
+            4: [email protected]
               0: [email protected]
                 0: [email protected] "<" [Newline("\n"), Whitespace("\t\t")] []
                 1: [email protected]
@@ -451,7 +446,7 @@ HtmlRoot {
                 2: [email protected]
                   0: [email protected] "div" [] []
                 3: [email protected] ">" [] []
-            6: [email protected]
+            5: [email protected]
               0: [email protected]
                 0: [email protected] "<" [Newline("\n"), Whitespace("\t\t")] []
                 1: [email protected]
@@ -465,7 +460,7 @@ HtmlRoot {
                 2: [email protected]
                   0: [email protected] "div" [] []
                 3: [email protected] ">" [] []
-            7: [email protected]
+            6: [email protected]
               0: [email protected]
                 0: [email protected] "<" [Newline("\n"), Whitespace("\t\t")] []
                 1: [email protected]

diff --git a/crates/biome_html_parser/tests/html_specs/ok/element_list2.html.snap b/crates/biome_html_parser/tests/html_specs/ok/element_list2.html.snap
@@ -48,10 +48,7 @@ HtmlRoot {
             },
         },
         HtmlContent {
-            value_token: HTML_LITERAL@21..27 "some" [Newline("\n")] [Whitespace(" ")],
-        },
-        HtmlContent {
-            value_token: HTML_LITERAL@27..31 "text" [] [],
+            value_token: HTML_LITERAL@21..31 "some text" [Newline("\n")] [],
         },
         HtmlElement {
             opening_element: HtmlOpeningElement {
@@ -140,11 +137,9 @@ HtmlRoot {
         2: [email protected]
           0: [email protected] "div" [] []
         3: [email protected] ">" [] []
-    3: [email protected]
-      0: [email protected] "some" [Newline("\n")] [Whitespace(" ")]
-    4: [email protected]
-      0: [email protected] "text" [] []
-    5: [email protected]
+    3: [email protected]
+      0: [email protected] "some text" [Newline("\n")] []
+    4: [email protected]
       0: [email protected]
         0: [email protected] "<" [Newline("\n")] []
         1: [email protected]
@@ -158,7 +153,7 @@ HtmlRoot {
         2: [email protected]
           0: [email protected] "div" [] []
         3: [email protected] ">" [] []
-    6: [email protected]
+    5: [email protected]
       0: [email protected]
         0: [email protected] "<" [Newline("\n")] []
         1: [email protected]
@@ -172,7 +167,7 @@ HtmlRoot {
         2: [email protected]
           0: [email protected] "div" [] []
         3: [email protected] ">" [] []
-    7: [email protected]
+    6: [email protected]
       0: [email protected]
         0: [email protected] "<" [Newline("\n")] []
         1: [email protected]

diff --git a/crates/biome_html_parser/tests/html_specs/ok/quote-in-child.html b/crates/biome_html_parser/tests/html_specs/ok/quote-in-child.html
@@ -0,0 +1 @@
+<p>"<span>Foo</span>" is the title.</p>
diff --git a/crates/biome_html_parser/tests/html_specs/ok/quote-in-child.html.snap b/crates/biome_html_parser/tests/html_specs/ok/quote-in-child.html.snap
@@ -0,0 +1,117 @@
+---
+source: crates/biome_html_parser/tests/spec_test.rs
+expression: snapshot
+---
+## Input
+
+```html
+<p>"<span>Foo</span>" is the title.</p>
+
+```
+
+
+## AST
+
+```
+HtmlRoot {
+    bom_token: missing (optional),
+    directive: missing (optional),
+    html: HtmlElementList [
+        HtmlElement {
+            opening_element: HtmlOpeningElement {
+                l_angle_token: L_ANGLE@0..1 "<" [] [],
+                name: HtmlName {
+                    value_token: HTML_LITERAL@1..2 "p" [] [],
+                },
+                attributes: HtmlAttributeList [],
+                r_angle_token: R_ANGLE@2..3 ">" [] [],
+            },
+            children: HtmlElementList [
+                HtmlContent {
+                    value_token: HTML_LITERAL@3..4 "\"" [] [],
+                },
+                HtmlElement {
+                    opening_element: HtmlOpeningElement {
+                        l_angle_token: L_ANGLE@4..5 "<" [] [],
+                        name: HtmlName {
+                            value_token: HTML_LITERAL@5..9 "span" [] [],
+                        },
+                        attributes: HtmlAttributeList [],
+                        r_angle_token: R_ANGLE@9..10 ">" [] [],
+                    },
+                    children: HtmlElementList [
+                        HtmlContent {
+                            value_token: HTML_LITERAL@10..13 "Foo" [] [],
+                        },
+                    ],
+                    closing_element: HtmlClosingElement {
+                        l_angle_token: L_ANGLE@13..14 "<" [] [],
+                        slash_token: SLASH@14..15 "/" [] [],
+                        name: HtmlName {
+                            value_token: HTML_LITERAL@15..19 "span" [] [],
+                        },
+                        r_angle_token: R_ANGLE@19..20 ">" [] [],
+                    },
+                },
+                HtmlContent {
+                    value_token: HTML_LITERAL@20..35 "\" is the title." [] [],
+                },
+            ],
+            closing_element: HtmlClosingElement {
+                l_angle_token: L_ANGLE@35..36 "<" [] [],
+                slash_token: SLASH@36..37 "/" [] [],
+                name: HtmlName {
+                    value_token: HTML_LITERAL@37..38 "p" [] [],
+                },
+                r_angle_token: R_ANGLE@38..39 ">" [] [],
+            },
+        },
+    ],
+    eof_token: EOF@39..40 "" [Newline("\n")] [],
+}
+```
+
+## CST
+
+```
+0: [email protected]
+  0: (empty)
+  1: (empty)
+  2: [email protected]
+    0: [email protected]
+      0: [email protected]
+        0: [email protected] "<" [] []
+        1: [email protected]
+          0: [email protected] "p" [] []
+        2: [email protected]
+        3: [email protected] ">" [] []
+      1: [email protected]
+        0: [email protected]
+          0: [email protected] "\"" [] []
+        1: [email protected]
+          0: [email protected]
+            0: [email protected] "<" [] []
+            1: [email protected]
+              0: [email protected] "span" [] []
+            2: [email protected]
+            3: [email protected] ">" [] []
+          1: [email protected]
+            0: [email protected]
+              0: [email protected] "Foo" [] []
+          2: [email protected]
+            0: [email protected] "<" [] []
+            1: [email protected] "/" [] []
+            2: [email protected]
+              0: [email protected] "span" [] []
+            3: [email protected] ">" [] []
+        2: [email protected]
+          0: [email protected] "\" is the title." [] []
+      2: [email protected]
+        0: [email protected] "<" [] []
+        1: [email protected] "/" [] []
+        2: [email protected]
+          0: [email protected] "p" [] []
+        3: [email protected] ">" [] []
+  3: [email protected] "" [Newline("\n")] []
+
+```