perf(html): initialize ammonia once, faster anchorizer, reduce iter collections & vecs, make DocNodes smaller

Cargo.toml

@@ -62,6 +62,7 @@ tree-sitter-md = { version = "0.1.7", optional = true }
 tree-sitter-rust = { version = "0.20.4", optional = true }
 tree-sitter-html = { version = "0.20.0", optional = true }
 tree-sitter-bash = { version = "0.20.5", optional = true }
+itoa = "1.0.10"
 
 [dev-dependencies]
 anyhow = { version = "1.0.58" }

src/html/comrak_adapters.rs

@@ -232,11 +232,40 @@ pub struct ToCEntry {
   pub anchor: String,
 }
 
+#[derive(Default)]
+pub struct Anchorizer {
+  map: HashMap<String, i32>,
+  itoa_buffer: itoa::Buffer,
+}
+
+impl Anchorizer {
+  /// Returns a String that has been converted into an anchor using the GFM algorithm.
+  /// This replaces comrak's implementation to improve the performance.
+  /// @see https://docs.rs/comrak/latest/comrak/struct.Anchorizer.html#method.anchorize
+  pub fn anchorize(&mut self, s: &str) -> String {
+    let mut s = REJECTED_CHARS
+      .replace_all(&s.to_lowercase(), "")
+      .replace(' ', "-");
+
+    if let Some(count) = self.map.get_mut(&s) {
+      let a = self.itoa_buffer.format(*count);
+      s.push('-');
+      s.push_str(a);
+
+      *count += 1;
+    } else {
+      self.map.insert(s.clone(), 1);
+    }
+
+    s
+  }
+}
+
 #[derive(Clone)]
 pub struct HeadingToCAdapter {
   toc: Arc<Mutex<Vec<ToCEntry>>>,
-  anchorizer: Arc<Mutex<comrak::html::Anchorizer>>,
   offset: Arc<Mutex<u8>>,
+  anchorizer: Arc<Mutex<Anchorizer>>,
 }
 
 impl Default for HeadingToCAdapter {
@@ -249,18 +278,18 @@ impl Default for HeadingToCAdapter {
   }
 }
 
+lazy_static! {
+  static ref REJECTED_CHARS: regex::Regex =
+    regex::Regex::new(r"[^\p{L}\p{M}\p{N}\p{Pc} -]").unwrap();
+}
+
 impl HeadingToCAdapter {
-  pub fn anchorize(&self, content: String) -> String {
+  pub fn anchorize(&self, content: &str) -> String {
     let mut anchorizer = self.anchorizer.lock().unwrap();
-    anchorizer.anchorize(content.clone())
+    anchorizer.anchorize(content)
   }
 
-  pub fn add_entry(
-    &self,
-    level: u8,
-    content: String,
-    anchor: String,
-  ) -> String {
+  pub fn add_entry(&self, level: u8, content: &str, anchor: &str) {
     let mut toc = self.toc.lock().unwrap();
     let mut offset = self.offset.lock().unwrap();
 
@@ -269,12 +298,10 @@ impl HeadingToCAdapter {
     if toc.last().map_or(true, |toc| toc.content != content) {
       toc.push(ToCEntry {
         level,
-        content,
-        anchor: anchor.clone(),
+        content: content.to_owned(),
+        anchor: anchor.to_owned(),
       });
     }
-
-    anchor
   }
 
   pub fn render(self) -> Option<String> {
@@ -322,7 +349,7 @@ impl HeadingAdapter for HeadingToCAdapter {
     let mut anchorizer = self.anchorizer.lock().unwrap();
     let offset = self.offset.lock().unwrap();
 
-    let anchor = anchorizer.anchorize(heading.content.clone());
+    let anchor = anchorizer.anchorize(&heading.content);
     writeln!(output, r#"<h{} id="{anchor}">"#, heading.level)?;
 
     let mut toc = self.toc.lock().unwrap();

src/html/jsdoc.rs

@@ -26,6 +26,74 @@ lazy_static! {
     regex::Regex::new(r"(^\.{0,2}\/)|(^[A-Za-z]+:\S)").unwrap();
   static ref MODULE_LINK_RE: regex::Regex =
     regex::Regex::new(r"^\[(\S+)\](?:\.(\S+)|\s|)$").unwrap();
+
+  #[cfg(feature = "ammonia")]
+  static ref AMMONIA: ammonia::Builder<'static> = {
+    let mut ammonia_builder = ammonia::Builder::default();
+
+    ammonia_builder
+      .add_tags(["video", "button", "svg", "path", "rect"])
+      .add_generic_attributes(["id", "align"])
+      .add_tag_attributes("button", ["data-copy"])
+      .add_tag_attributes(
+        "svg",
+        [
+          "width",
+          "height",
+          "viewBox",
+          "fill",
+          "xmlns",
+          "stroke",
+          "stroke-width",
+          "stroke-linecap",
+          "stroke-linejoin",
+        ],
+      )
+      .add_tag_attributes(
+        "path",
+        [
+          "d",
+          "fill",
+          "fill-rule",
+          "clip-rule",
+          "stroke",
+          "stroke-width",
+          "stroke-linecap",
+          "stroke-linejoin",
+        ],
+      )
+      .add_tag_attributes("rect", ["x", "y", "width", "height", "fill"])
+      .add_tag_attributes("video", ["src", "controls"])
+      .add_allowed_classes("pre", ["highlight"])
+      .add_allowed_classes("button", ["context_button"])
+      .add_allowed_classes(
+        "div",
+        [
+          "alert",
+          "alert-note",
+          "alert-tip",
+          "alert-important",
+          "alert-warning",
+          "alert-caution",
+        ],
+      )
+      .link_rel(Some("nofollow"))
+      .url_relative(
+      ammonia::UrlRelative::Custom(Box::new(AmmoniaRelativeUrlEvaluator())));
+
+    #[cfg(feature = "syntect")]
+    ammonia_builder.add_tag_attributes("span", ["style"]);
+
+    #[cfg(feature = "tree-sitter")]
+    ammonia_builder.add_allowed_classes("span", super::tree_sitter::CLASSES);
+
+    ammonia_builder
+  };
+}
+
+thread_local! {
+  static CURRENT_FILE: RefCell<Option<Option<ShortPath>>> = RefCell::new(None);
+  static URL_REWRITER: RefCell<Option<Option<URLRewriter>>> = RefCell::new(None);
 }
 
 fn parse_links<'a>(md: &'a str, ctx: &RenderContext) -> Cow<'a, str> {
@@ -147,15 +215,23 @@ fn split_markdown_title(
 }
 
 #[cfg(feature = "ammonia")]
-struct AmmoniaRelativeUrlEvaluator {
-  current_file: Option<ShortPath>,
-  url_rewriter: URLRewriter,
-}
+struct AmmoniaRelativeUrlEvaluator();
 
 #[cfg(feature = "ammonia")]
 impl ammonia::UrlRelativeEvaluate for AmmoniaRelativeUrlEvaluator {
   fn evaluate<'a>(&self, url: &'a str) -> Option<Cow<'a, str>> {
-    Some((self.url_rewriter)(self.current_file.as_ref(), url).into())
+    URL_REWRITER.with(|url_rewriter| {
+      if let Some(url_rewriter) = url_rewriter.borrow().as_ref().unwrap() {
+        CURRENT_FILE.with(|current_file| {
+          Some(
+            url_rewriter(current_file.borrow().as_ref().unwrap().as_ref(), url)
+              .into(),
+          )
+        })
+      } else {
+        Some(Cow::Borrowed(url))
+      }
+    })
   }
 }
 
@@ -374,10 +450,13 @@ pub fn markdown_to_html(
   }
 
   let mut plugins = comrak::Plugins::default();
-  plugins.render.codefence_syntax_highlighter =
-    Some(&render_ctx.ctx.highlight_adapter);
-  if !render_options.no_toc {
-    plugins.render.heading_adapter = Some(&render_ctx.toc);
+
+  if !render_options.summary {
+    plugins.render.codefence_syntax_highlighter =
+      Some(&render_ctx.ctx.highlight_adapter);
+    if !render_options.no_toc {
+      plugins.render.heading_adapter = Some(&render_ctx.toc);
+    }
   }
 
   let md = parse_links(md, render_ctx);
@@ -400,7 +479,7 @@ pub fn markdown_to_html(
     "markdown"
   };
 
-  let mut html = {
+  let html = {
     let arena = Arena::new();
     let root = comrak::parse_document(&arena, md, &options);
 
@@ -411,75 +490,24 @@ pub fn markdown_to_html(
 
   #[cfg(feature = "ammonia")]
   {
-    let mut ammonia_builder = ammonia::Builder::default();
-
-    ammonia_builder
-      .add_tags(["video", "button", "svg", "path", "rect"])
-      .add_generic_attributes(["id", "align"])
-      .add_tag_attributes("button", ["data-copy"])
-      .add_tag_attributes(
-        "svg",
-        [
-          "width",
-          "height",
-          "viewBox",
-          "fill",
-          "xmlns",
-          "stroke",
-          "stroke-width",
-          "stroke-linecap",
-          "stroke-linejoin",
-        ],
-      )
-      .add_tag_attributes(
-        "path",
-        [
-          "d",
-          "fill",
-          "fill-rule",
-          "clip-rule",
-          "stroke",
-          "stroke-width",
-          "stroke-linecap",
-          "stroke-linejoin",
-        ],
-      )
-      .add_tag_attributes("rect", ["x", "y", "width", "height", "fill"])
-      .add_tag_attributes("video", ["src", "controls"])
-      .add_allowed_classes("pre", ["highlight"])
-      .add_allowed_classes("button", ["context_button"])
-      .add_allowed_classes(
-        "div",
-        [
-          "alert",
-          "alert-note",
-          "alert-tip",
-          "alert-important",
-          "alert-warning",
-          "alert-caution",
-        ],
-      )
-      .link_rel(Some("nofollow"))
-      .url_relative(render_ctx.ctx.url_rewriter.as_ref().map_or(
-        ammonia::UrlRelative::PassThrough,
-        |url_rewriter| {
-          ammonia::UrlRelative::Custom(Box::new(AmmoniaRelativeUrlEvaluator {
-            current_file: render_ctx.get_current_resolve().get_file().cloned(),
-            url_rewriter: url_rewriter.clone(),
-          }))
-        },
-      ));
+    CURRENT_FILE
+      .set(Some(render_ctx.get_current_resolve().get_file().cloned()));
+    URL_REWRITER.set(Some(render_ctx.ctx.url_rewriter.clone()));
 
-    #[cfg(feature = "syntect")]
-    ammonia_builder.add_tag_attributes("span", ["style"]);
+    let html = Some(format!(
+      r#"<div class="{class_name}">{}</div>"#,
+      AMMONIA.clean(&html)
+    ));
 
-    #[cfg(feature = "tree-sitter")]
-    ammonia_builder.add_allowed_classes("span", super::tree_sitter::CLASSES);
+    CURRENT_FILE.set(None);
+    URL_REWRITER.set(None);
 
-    html = ammonia_builder.clean(&html).to_string();
+    html
+  }
+  #[cfg(not(feature = "ammonia"))]
+  {
+    Some(format!(r#"<div class="{class_name}">{html}</div>"#))
   }
-
-  Some(format!(r#"<div class="{class_name}">{html}</div>"#))
 }
 
 pub(crate) fn render_markdown_summary(
@@ -649,20 +677,17 @@ impl ModuleDocCtx {
 
       sections.extend(super::namespace::render_namespace(
         render_ctx,
-        partitions_by_kind
-          .into_iter()
-          .map(|(title, nodes)| {
-            (
-              SectionHeaderCtx {
-                title: title.clone(),
-                anchor: AnchorCtx { id: title },
-                href: None,
-                doc: None,
-              },
-              nodes,
-            )
-          })
-          .collect(),
+        partitions_by_kind.into_iter().map(|(title, nodes)| {
+          (
+            SectionHeaderCtx {
+              title: title.clone(),
+              anchor: AnchorCtx { id: title },
+              href: None,
+              doc: None,
+            },
+            nodes,
+          )
+        }),
       ));
     }