rustdoc: add doc_link_canonical feature

Author: lolbinarycat

compiler/rustc_ast_passes/src/feature_gate.rs

@@ -182,6 +182,7 @@ impl<'a> Visitor<'a> for PostExpansionVisitor<'a> {
 
                 gate_doc!(
                     "experimental" {
+                        html_link_canonical => doc_link_canonical
                         cfg => doc_cfg
                         cfg_hide => doc_cfg_hide
                         masked => doc_masked

compiler/rustc_feature/src/unstable.rs

@@ -479,6 +479,8 @@ declare_features! (
     (unstable, doc_cfg, "1.21.0", Some(43781)),
     /// Allows `#[doc(cfg_hide(...))]`.
     (unstable, doc_cfg_hide, "1.57.0", Some(43781)),
+    /// Allows `#![doc(html_link_canonical]`
+    (unstable, doc_link_canonical, "1.88.0", Some(143139)),
     /// Allows `#[doc(masked)]`.
     (unstable, doc_masked, "1.21.0", Some(44027)),
     /// Allows `dyn* Trait` objects.

compiler/rustc_passes/src/check_attr.rs

@@ -1329,6 +1329,7 @@ impl<'tcx> CheckAttrVisitor<'tcx> {
 
                         Some(
                             sym::html_favicon_url
+                            | sym::html_link_canonical
                             | sym::html_logo_url
                             | sym::html_playground_url
                             | sym::issue_tracker_base_url

compiler/rustc_span/src/symbol.rs

@@ -864,6 +864,7 @@ symbols! {
         doc_cfg,
         doc_cfg_hide,
         doc_keyword,
+        doc_link_canonical,
         doc_masked,
         doc_notable_trait,
         doc_primitive,
@@ -1134,6 +1135,7 @@ symbols! {
         homogeneous_aggregate,
         host,
         html_favicon_url,
+        html_link_canonical,
         html_logo_url,
         html_no_source,
         html_playground_url,

src/librustdoc/html/layout.rs

@@ -20,9 +20,13 @@ pub(crate) struct Layout {
     pub(crate) css_file_extension: Option<PathBuf>,
     /// If true, then scrape-examples.js will be included in the output HTML file
     pub(crate) scrape_examples_extension: bool,
+    /// if present, insert a rel="canonical" link with this prefix.
+    pub(crate) link_canonical: Option<String>,
 }
 
 pub(crate) struct Page<'a> {
+    /// url relative to documentation bundle root.
+    pub(crate) relative_url: Option<String>,
     pub(crate) title: &'a str,
     pub(crate) css_class: &'a str,
     pub(crate) root_path: &'a str,
@@ -47,7 +51,6 @@ struct PageLayout<'a> {
     static_root_path: String,
     page: &'a Page<'a>,
     layout: &'a Layout,
-
     files: &'static StaticFiles,
 
     themes: Vec<String>,

src/librustdoc/html/render/context.rs

@@ -49,6 +49,10 @@ pub(crate) struct Context<'tcx> {
     /// The current destination folder of where HTML artifacts should be placed.
     /// This changes as the context descends into the module hierarchy.
     pub(crate) dst: PathBuf,
+    /// Initial length of `dst`.
+    ///
+    /// Used to split `dst` into (doc bundle path, relative path)
+    dst_prefix_doc_bundle: usize,
     /// Tracks section IDs for `Deref` targets so they match in both the main
     /// body and the sidebar.
     pub(super) deref_id_map: RefCell<DefIdMap<String>>,
@@ -180,13 +184,18 @@ impl<'tcx> Context<'tcx> {
         self.id_map.borrow_mut().derive(id)
     }
 
+    pub(crate) fn dst_relative_to_doc_bundle_root(&self) -> &str {
+        str::from_utf8(&self.dst.as_os_str().as_encoded_bytes()[self.dst_prefix_doc_bundle..])
+            .expect("non-utf8 in name generated by rustdoc").trim_start_matches('/')
+    }
+
     /// String representation of how to get back to the root path of the 'doc/'
     /// folder in terms of a relative URL.
     pub(super) fn root_path(&self) -> String {
         "../".repeat(self.current.len())
     }
 
-    fn render_item(&mut self, it: &clean::Item, is_module: bool) -> String {
+    fn render_item(&mut self, it: &clean::Item, is_module: bool, relative_url: Option<String>) -> String {
         let mut render_redirect_pages = self.info.render_redirect_pages;
         // If the item is stripped but inlined, links won't point to the item so no need to generate
         // a file for it.
@@ -238,6 +247,7 @@ impl<'tcx> Context<'tcx> {
         if !render_redirect_pages {
             let content = print_item(self, it);
             let page = layout::Page {
+                relative_url,
                 css_class: tyname_s,
                 root_path: &self.root_path(),
                 static_root_path: self.shared.static_root_path.as_deref(),
@@ -511,6 +521,7 @@ impl<'tcx> FormatRenderer<'tcx> for Context<'tcx> {
             krate_version: krate_version.to_string(),
             css_file_extension: extension_css,
             scrape_examples_extension: !call_locations.is_empty(),
+            link_canonical: None,
         };
         let mut issue_tracker_base_url = None;
         let mut include_sources = !html_no_source;
@@ -537,6 +548,14 @@ impl<'tcx> FormatRenderer<'tcx> for Context<'tcx> {
                 (Some(sym::html_no_source), None) if attr.is_word() => {
                     include_sources = false;
                 }
+                (Some(sym::html_link_canonical), Some(s)) => {
+                    let mut s = s.to_string();
+                    // ensure trailing slash
+                    if !s.ends_with('/') {
+                        s.push('/');
+                    }
+                    layout.link_canonical = Some(s);
+                }
                 _ => {}
             }
         }
@@ -579,6 +598,7 @@ impl<'tcx> FormatRenderer<'tcx> for Context<'tcx> {
 
         let mut cx = Context {
             current: Vec::new(),
+            dst_prefix_doc_bundle: dst.as_os_str().len(),
             dst,
             id_map: RefCell::new(id_map),
             deref_id_map: Default::default(),
@@ -626,6 +646,7 @@ impl<'tcx> FormatRenderer<'tcx> for Context<'tcx> {
             css_class: "mod sys",
             root_path: "../",
             static_root_path: shared.static_root_path.as_deref(),
+            relative_url: None,
             description: "List of all items in this crate",
             resource_suffix: &shared.resource_suffix,
             rust_logo: has_doc_flag(self.tcx(), LOCAL_CRATE.as_def_id(), sym::rust_logo),
@@ -787,7 +808,8 @@ impl<'tcx> FormatRenderer<'tcx> for Context<'tcx> {
         info!("Recursing into {}", self.dst.display());
 
         if !item.is_stripped() {
-            let buf = self.render_item(item, true);
+            let rel_path = format!("{}/index.html", self.dst_relative_to_doc_bundle_root());
+            let buf = self.render_item(item, true, Some(rel_path));
             // buf will be empty if the module is stripped and there is no redirect for it
             if !buf.is_empty() {
                 self.shared.ensure_dir(&self.dst)?;
@@ -842,12 +864,13 @@ impl<'tcx> FormatRenderer<'tcx> for Context<'tcx> {
             self.info.render_redirect_pages = item.is_stripped();
         }
 
-        let buf = self.render_item(&item, false);
+        let name = item.name.as_ref().unwrap();
+        let item_type = item.type_();
+        let file_name = print_item_path(item_type, name.as_str()).to_string();
+        let rel_path = format!("{}/{file_name}", self.dst_relative_to_doc_bundle_root());
+        let buf = self.render_item(&item, false, Some(rel_path));
         // buf will be empty if the item is stripped and there is no redirect for it
         if !buf.is_empty() {
-            let name = item.name.as_ref().unwrap();
-            let item_type = item.type_();
-            let file_name = print_item_path(item_type, name.as_str()).to_string();
             self.shared.ensure_dir(&self.dst)?;
             let joint_dst = self.dst.join(&file_name);
             self.shared.fs.write(joint_dst, buf)?;

src/librustdoc/html/render/write_shared.rs

@@ -428,6 +428,7 @@ impl CratesIndexPart {
             title: "Index of crates",
             css_class: "mod sys",
             root_path: "./",
+            relative_url: None,
             static_root_path: cx.shared.static_root_path.as_deref(),
             description: "List of crates",
             resource_suffix: &cx.shared.resource_suffix,

src/librustdoc/html/sources.rs

@@ -232,6 +232,7 @@ impl SourceCollector<'_, '_> {
             title: &title,
             css_class: "src",
             root_path: &root_path,
+            relative_url: None,
             static_root_path: shared.static_root_path.as_deref(),
             description: &desc,
             resource_suffix: &shared.resource_suffix,

src/librustdoc/html/templates/page.html

@@ -62,6 +62,9 @@
         <link rel="icon" type="image/svg+xml" {#+ #}
             href="{{static_root_path|safe}}{{files.rust_favicon_svg}}">
     {% endif %}
+    {% if layout.link_canonical.is_some() && page.relative_url.is_some() %}
+        <link rel="canonical" href="{{layout.link_canonical.as_ref().unwrap()|safe}}{{page.relative_url.as_ref().unwrap()|safe}}">
+    {% endif %}
     {{ layout.external_html.in_header|safe }}
 </head> {# #}
 <body class="rustdoc {{+page.css_class}}"> {# #}

src/librustdoc/lib.rs

@@ -711,6 +711,7 @@ fn opts() -> Vec<RustcOptGroup> {
             "removed, see issue #44136 <https://github.com/rust-lang/rust/issues/44136> for more information",
             "[rust]",
         ),
+
     ]
 }
 

tests/rustdoc/link-canonical.rs

@@ -0,0 +1,10 @@
+#![crate_name = "foo"]
+#![feature(doc_link_canonical)]
+#![doc(html_link_canonical = "https://foo.example/")]
+
+//@ has 'foo/index.html'
+//@ has - '//head/link[@rel="canonical"][@href="https://foo.example/foo/index.html"]' ''
+
+//@ has 'foo/struct.FooBaz.html'
+//@ has - '//head/link[@rel="canonical"][@href="https://foo.example/foo/struct.FooBaz.html"]' ''
+pub struct FooBaz;