feat(html): set up html parser and formatter benchmarks (#5155)

xtask/bench/Cargo.toml

@@ -16,6 +16,9 @@ biome_graphql_formatter = { workspace = true }
 biome_graphql_parser    = { workspace = true }
 biome_graphql_syntax    = { workspace = true }
 biome_grit_patterns     = { workspace = true }
+biome_html_formatter    = { workspace = true }
+biome_html_parser       = { workspace = true }
+biome_html_syntax       = { workspace = true }
 biome_js_analyze        = { workspace = true }
 biome_js_formatter      = { workspace = true }
 biome_js_parser         = { workspace = true }
@@ -91,5 +94,14 @@ name    = "graphql_formatter"
 harness = false
 name    = "gritql_search"
 
+# HTML benches
+[[bench]]
+harness = false
+name    = "html_parser"
+
+[[bench]]
+harness = false
+name    = "html_formatter"
+
 [lints]
 workspace = true

xtask/bench/benches/html_formatter.rs

@@ -0,0 +1,41 @@
+use std::collections::HashMap;
+use xtask_bench::{bench_formatter_group, TestCase};
+use xtask_bench::{criterion_group, criterion_main, Criterion};
+
+#[cfg(target_os = "windows")]
+#[global_allocator]
+static GLOBAL: mimalloc::MiMalloc = mimalloc::MiMalloc;
+
+#[cfg(all(
+    any(target_os = "macos", target_os = "linux"),
+    not(target_env = "musl"),
+))]
+#[global_allocator]
+static GLOBAL: tikv_jemallocator::Jemalloc = tikv_jemallocator::Jemalloc;
+
+#[cfg(all(target_env = "musl", target_os = "linux", target_arch = "aarch64"))]
+#[global_allocator]
+static GLOBAL: std::alloc::System = std::alloc::System;
+
+fn bench_html_formatter(criterion: &mut Criterion) {
+    let mut all_suites = HashMap::new();
+    all_suites.insert("html", include_str!("libs-html.txt"));
+    let mut libs = vec![];
+    libs.extend(all_suites.values().flat_map(|suite| suite.lines()));
+
+    let mut group = criterion.benchmark_group("html_formatter");
+    for lib in libs {
+        let test_case = TestCase::try_from(lib);
+
+        match test_case {
+            Ok(test_case) => {
+                bench_formatter_group(&mut group, test_case);
+            }
+            Err(e) => println!("{e:?}"),
+        }
+    }
+    group.finish();
+}
+
+criterion_group!(html_formatter, bench_html_formatter);
+criterion_main!(html_formatter);

xtask/bench/benches/html_parser.rs

@@ -0,0 +1,41 @@
+use std::collections::HashMap;
+use xtask_bench::{bench_parser_group, TestCase};
+use xtask_bench::{criterion_group, criterion_main, Criterion};
+
+#[cfg(target_os = "windows")]
+#[global_allocator]
+static GLOBAL: mimalloc::MiMalloc = mimalloc::MiMalloc;
+
+#[cfg(all(
+    any(target_os = "macos", target_os = "linux"),
+    not(target_env = "musl"),
+))]
+#[global_allocator]
+static GLOBAL: tikv_jemallocator::Jemalloc = tikv_jemallocator::Jemalloc;
+
+#[cfg(all(target_env = "musl", target_os = "linux", target_arch = "aarch64"))]
+#[global_allocator]
+static GLOBAL: std::alloc::System = std::alloc::System;
+
+fn bench_html_parser(criterion: &mut Criterion) {
+    let mut all_suites = HashMap::new();
+    all_suites.insert("html", include_str!("libs-html.txt"));
+    let mut libs = vec![];
+    libs.extend(all_suites.values().flat_map(|suite| suite.lines()));
+
+    let mut group = criterion.benchmark_group("html_parser");
+    for lib in libs {
+        let test_case = TestCase::try_from(lib);
+
+        match test_case {
+            Ok(test_case) => {
+                bench_parser_group(&mut group, test_case);
+            }
+            Err(e) => println!("{e:?}"),
+        }
+    }
+    group.finish();
+}
+
+criterion_group!(html_parser, bench_html_parser);
+criterion_main!(html_parser);

xtask/bench/benches/libs-html.txt

@@ -0,0 +1 @@
+https://raw.githubusercontent.com/dyc3/opentogethertube/e6b44c716289a8babb2f70624e452542e45c2a1b/client/index.html

xtask/bench/src/language.rs

@@ -10,6 +10,9 @@ use biome_css_syntax::{CssRoot, CssSyntaxNode};
 use biome_formatter::{FormatResult, Formatted, PrintResult, Printed};
 use biome_graphql_formatter::context::{GraphqlFormatContext, GraphqlFormatOptions};
 use biome_graphql_syntax::GraphqlSyntaxNode;
+use biome_html_formatter::context::HtmlFormatContext;
+use biome_html_formatter::HtmlFormatOptions;
+use biome_html_syntax::HtmlSyntaxNode;
 use biome_js_formatter::context::{JsFormatContext, JsFormatOptions};
 use biome_js_parser::JsParserOptions;
 use biome_js_syntax::{AnyJsRoot, JsFileSource, JsSyntaxNode};
@@ -25,6 +28,7 @@ pub enum Parse<'a> {
     Json(&'a str),
     Css(&'a str),
     Graphql(&'a str),
+    Html(&'a str),
 }
 
 impl Parse<'_> {
@@ -35,6 +39,7 @@ impl Parse<'_> {
                 "json" => Some(Parse::Json(case.code())),
                 "css" => Some(Parse::Css(case.code())),
                 "graphql" => Some(Parse::Graphql(case.code())),
+                "html" => Some(Parse::Html(case.code())),
                 _ => None,
             },
         }
@@ -57,6 +62,7 @@ impl Parse<'_> {
                     .allow_css_modules(),
             )),
             Parse::Graphql(code) => Parsed::Graphql(biome_graphql_parser::parse_graphql(code)),
+            Parse::Html(code) => Parsed::Html(biome_html_parser::parse_html(code)),
         }
     }
 
@@ -86,6 +92,9 @@ impl Parse<'_> {
             Parse::Graphql(code) => {
                 Parsed::Graphql(biome_graphql_parser::parse_graphql_with_cache(code, cache))
             }
+            Parse::Html(code) => {
+                Parsed::Html(biome_html_parser::parse_html_with_cache(code, cache))
+            }
         }
     }
 }
@@ -95,6 +104,7 @@ pub enum Parsed {
     Json(biome_json_parser::JsonParse),
     Css(biome_css_parser::CssParse),
     Graphql(biome_graphql_parser::GraphqlParse),
+    Html(biome_html_parser::HtmlParse),
 }
 
 impl Parsed {
@@ -106,6 +116,7 @@ impl Parsed {
             Parsed::Json(parse) => Some(FormatNode::Json(parse.syntax())),
             Parsed::Css(parse) => Some(FormatNode::Css(parse.syntax())),
             Parsed::Graphql(parse) => Some(FormatNode::Graphql(parse.syntax())),
+            Parsed::Html(parse) => Some(FormatNode::Html(parse.syntax())),
         }
     }
 
@@ -115,6 +126,7 @@ impl Parsed {
             Parsed::Json(_) => None,
             Parsed::Graphql(_) => None,
             Parsed::Css(parse) => Some(Analyze::Css(parse.tree())),
+            Parsed::Html(_) => None,
         }
     }
 
@@ -124,6 +136,7 @@ impl Parsed {
             Parsed::Json(parse) => parse.into_diagnostics(),
             Parsed::Css(parse) => parse.into_diagnostics(),
             Parsed::Graphql(parse) => parse.into_diagnostics(),
+            Parsed::Html(parse) => parse.into_diagnostics(),
         }
     }
 }
@@ -133,6 +146,7 @@ pub enum FormatNode {
     Json(JsonSyntaxNode),
     Css(CssSyntaxNode),
     Graphql(GraphqlSyntaxNode),
+    Html(HtmlSyntaxNode),
 }
 
 impl FormatNode {
@@ -152,6 +166,10 @@ impl FormatNode {
                 biome_graphql_formatter::format_node(GraphqlFormatOptions::default(), root)
                     .map(FormattedNode::Graphql)
             }
+            FormatNode::Html(root) => {
+                biome_html_formatter::format_node(HtmlFormatOptions::default(), root)
+                    .map(FormattedNode::Html)
+            }
         }
     }
 }
@@ -161,6 +179,7 @@ pub enum FormattedNode {
     Json(Formatted<JsonFormatContext>),
     Css(Formatted<CssFormatContext>),
     Graphql(Formatted<GraphqlFormatContext>),
+    Html(Formatted<HtmlFormatContext>),
 }
 
 impl FormattedNode {
@@ -170,6 +189,7 @@ impl FormattedNode {
             FormattedNode::Json(formatted) => formatted.print(),
             FormattedNode::Css(formatted) => formatted.print(),
             FormattedNode::Graphql(formatted) => formatted.print(),
+            FormattedNode::Html(formatted) => formatted.print(),
         }
     }
 }