fix(formatter_htmlish): handle XHTML (#28703)

- bump `oxc-htmlish-parser` to support Xhtml dialect
- Use it in formatter

Cargo.lock

@@ -1699,9 +1699,9 @@ dependencies = [
 
 [[package]]
 name = "oxc-htmlish-parser"
-version = "0.0.1"
+version = "0.0.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "909fd32d29f076e5129a27244801bed7de5ce33a974310c710f8777dee016495"
+checksum = "d56b242831e1d43ca8b931fec32a68a263a7b1c1c7e7f74cfb7f94c4759f3ae9"
 dependencies = [
  "memchr",
  "oxc_allocator",

Cargo.toml

@@ -243,7 +243,7 @@ num-bigint = "0.5.0" # Big integers
 num-traits = "0.2.19" # Numeric traits
 oxc-css-parser = "0.0.18" # CSS/SCSS/Less parser (raffia 0.12.3 fork: adds `template_placeholder` typed placeholders for css-in-js + valid-syntax coverage fixes (see crates/oxc_formatter_css/AGENTS.md))
 oxc-graphql-parser = "0.0.9" # GraphQL parser (direct AST; apollo-parser 0.8.6 fork aligned with graphql-js 17 syntax)
-oxc-htmlish-parser = "0.0.1" # HTML / Vue / Angular parser (loose, span-faithful AST)
+oxc-htmlish-parser = "0.0.2" # HTML / Vue / Angular parser (loose, span-faithful AST)
 oxc-markdown-parser = "0.0.3" # CommonMark + GFM parser (span-faithful, style-preserving typed AST)
 oxc-toml = "0.15.1" # TOML parser and formatter
 oxc-yaml-parser = "0.0.7" # YAML 1.2 parser (comment-preserving typed AST, node naming from `yaml-unist-parser`)

apps/oxfmt/conformance/run_prettier.ts

@@ -43,6 +43,7 @@ const LANGUAGES: Record<string, Language> = {
   ".yaml": "yaml",
   ".md": "markdown",
   ".html": "html",
+  ".xhtml": "html",
   ".vue": "vue",
 };
 
@@ -74,7 +75,8 @@ const PARSERS: Record<string, (name: string) => string> = {
   graphql: withExt(".graphql"),
   yaml: withExt(".yaml"),
   markdown: withExt(".md"),
-  html: withExt(".html"),
+  // An HTML extension is kept, as oxfmt reads some differently (`.xhtml`)
+  html: (name) => (LANGUAGES[extname(name)] === "html" ? name : withExt(".html")(name)),
   // The whole file goes to Prettier, but its `<script>` is formatted by oxc.
   // A `.html` page under `vue` (e.g. `vue/html-vue/`) fails only on the doctype case for now:
   // Prettier lowercases `<!DOCTYPE html>` for a `.html` filepath only, and this runs it as `.vue`.

apps/oxfmt/conformance/snapshots/prettier.snap.md

@@ -670,7 +670,7 @@
 
 ## html
 
-221/228 (96.93%)
+222/228 (97.37%)
 
 ### Failed
 
@@ -685,11 +685,6 @@
 - externals/prettier/html/tags/tags2.html (💥💥✨✨, 97.17%)
   - crates/oxc_formatter_htmlish/DIVERGENCES.md#dialog-out-of-flow
 
-### Error
-
-- externals/prettier/misc/filename-matters/html-doctype/xhtml-doctype.xhtml: Syntax error: missing end tag of raw text element at 16..74
-  - ❓unclassified
-
 ## vue
 
 175/184 (95.11%)

crates/oxc_formatter_htmlish/src/options.rs

@@ -14,14 +14,15 @@ pub enum HtmlishVariant {
     /// Prettier's `parser: angular` equivalent.
     Angular,
     /// Prettier's `parser: html` on an XHTML file (`.xhtml`, `.xht`), which an XML parser may read:
-    /// read and printed as HTML, but the doctype keeps its case.
+    /// read and printed as HTML, but `/>` closes every element and the doctype keeps its case.
     Xhtml,
 }
 
 impl HtmlishVariant {
     pub(crate) fn to_dialect(self) -> Dialect {
         match self {
-            Self::Html | Self::Xhtml => Dialect::Html,
+            Self::Html => Dialect::Html,
+            Self::Xhtml => Dialect::Xhtml,
             Self::Vue => Dialect::Vue,
             Self::Angular => Dialect::Angular,
         }

crates/oxc_formatter_htmlish/src/rules.rs

@@ -1,5 +1,3 @@
-use oxc_htmlish_parser::Dialect;
-
 use crate::HtmlishVariant;
 
 /// What the variant changes in the shared rules, resolved once per document.
@@ -25,11 +23,11 @@ impl DialectRules {
     /// `sfc`: a Vue file read as a single-file component, not as HTML.
     /// `standalone`: a file of its own, not a part embedded in another language.
     pub fn new(variant: HtmlishVariant, sfc: bool, standalone: bool) -> Self {
-        let dialect = variant.to_dialect();
         Self {
-            names_ignore_case: dialect == Dialect::Html || (dialect == Dialect::Vue && !sfc),
-            keeps_element_gap_kind: dialect == Dialect::Vue,
-            has_blocks: dialect == Dialect::Angular,
+            names_ignore_case: matches!(variant, HtmlishVariant::Html | HtmlishVariant::Xhtml)
+                || (variant == HtmlishVariant::Vue && !sfc),
+            keeps_element_gap_kind: variant == HtmlishVariant::Vue,
+            has_blocks: variant == HtmlishVariant::Angular,
             lowercase_doctype: variant != HtmlishVariant::Xhtml,
             lowercase_known_names: standalone && variant == HtmlishVariant::Html,
         }

crates/oxc_formatter_htmlish/tests/fixtures/html/self-closing-in-xhtml.xhtml

@@ -0,0 +1,4 @@
+<script src="a"/>
+<div/>after
+<p><my-el/>in p</p>
+<textarea/>

crates/oxc_formatter_htmlish/tests/fixtures/html/self-closing-in-xhtml.xhtml.snap

@@ -0,0 +1,29 @@
+---
+source: crates/oxc_formatter_htmlish/tests/fixtures/mod.rs
+---
+==================== Input ====================
+<script src="a"/>
+<div/>after
+<p><my-el/>in p</p>
+<textarea/>
+
+==================== Output ====================
+------------------
+{ printWidth: 80 }
+------------------
+<script src="a" />
+<div />
+after
+<p><my-el />in p</p>
+<textarea />
+
+-------------------
+{ printWidth: 100 }
+-------------------
+<script src="a" />
+<div />
+after
+<p><my-el />in p</p>
+<textarea />
+
+===================== End =====================