Tease apart orthogonal concerns in markdown link rewriting

`hir` should know nothing about URLs, markdown and html. It should
only be able to:

* resolve stringy path from documentation
* generate canonical stringy path for a def

In contrast, link rewriting should not care about semantics of paths
and names resolution, and should be concern only with text mangling
bits.

3 files changed, 231 insertions, 15 deletions

diff --git a/crates/ide/Cargo.toml b/crates/ide/Cargo.toml
index e61c276df..a15f704ca 100644
--- a/crates/ide/Cargo.toml
+++ b/crates/ide/Cargo.toml
@@ -18,6 +18,7 @@ rustc-hash = "1.1.0"
 oorandom = "11.1.2"
 pulldown-cmark-to-cmark = "5.0.0"
 pulldown-cmark = {version = "0.7.2", default-features = false}
+url = "2.1.1"
 
 stdx = { path = "../stdx", version = "0.0.0" }
 syntax = { path = "../syntax", version = "0.0.0" }
diff --git a/crates/ide/src/hover.rs b/crates/ide/src/hover.rs
index 536411704..efec0184e 100644
--- a/crates/ide/src/hover.rs
+++ b/crates/ide/src/hover.rs
@@ -1756,6 +1756,60 @@ pub struct B<|>ar
     }
 
     #[test]
+    fn test_doc_links_enum_variant() {
+        check(
+            r#"
+enum E {
+    /// [E]
+    V<|> { field: i32 }
+}
+"#,
+            expect![[r#"
+                *V*
+
+                ```rust
+                test::E
+                ```
+
+                ```rust
+                V
+                ```
+
+                ---
+
+                [E](https://docs.rs/test/*/test/enum.E.html)
+            "#]],
+        );
+    }
+
+    #[test]
+    fn test_doc_links_field() {
+        check(
+            r#"
+struct S {
+    /// [`S`]
+    field<|>: i32
+}
+"#,
+            expect![[r#"
+                *field*
+
+                ```rust
+                test::S
+                ```
+
+                ```rust
+                field: i32
+                ```
+
+                ---
+
+                [`S`](https://docs.rs/test/*/test/struct.S.html)
+            "#]],
+        );
+    }
+
+    #[test]
     fn test_hover_macro_generated_struct_fn_doc_comment() {
         mark::check!(hover_macro_generated_struct_fn_doc_comment);
 
diff --git a/crates/ide/src/link_rewrite.rs b/crates/ide/src/link_rewrite.rs
index ff3200eef..acedea71b 100644
--- a/crates/ide/src/link_rewrite.rs
+++ b/crates/ide/src/link_rewrite.rs
@@ -2,11 +2,11 @@
 //!
 //! Most of the implementation can be found in [`hir::doc_links`].
 
+use hir::{Adt, Crate, HasAttrs, ModuleDef};
+use ide_db::{defs::Definition, RootDatabase};
 use pulldown_cmark::{CowStr, Event, Options, Parser, Tag};
 use pulldown_cmark_to_cmark::{cmark_with_options, Options as CmarkOptions};
-
-use hir::resolve_doc_link;
-use ide_db::{defs::Definition, RootDatabase};
+use url::Url;
 
 /// Rewrite documentation links in markdown to point to an online host (e.g. docs.rs)
 pub fn rewrite_links(db: &RootDatabase, markdown: &str, definition: &Definition) -> String {
@@ -26,19 +26,16 @@ pub fn rewrite_links(db: &RootDatabase, markdown: &str, definition: &Definition)
             // Two posibilities:
             // * path-based links: `../../module/struct.MyStruct.html`
             // * module-based links (AKA intra-doc links): `super::super::module::MyStruct`
-            let resolved = match definition {
-                Definition::ModuleDef(t) => resolve_doc_link(db, t, title, target),
-                Definition::Macro(t) => resolve_doc_link(db, t, title, target),
-                Definition::Field(t) => resolve_doc_link(db, t, title, target),
-                Definition::SelfType(t) => resolve_doc_link(db, t, title, target),
-                Definition::Local(t) => resolve_doc_link(db, t, title, target),
-                Definition::TypeParam(t) => resolve_doc_link(db, t, title, target),
-            };
-
-            match resolved {
-                Some((target, title)) => (target, title),
-                None => (target.to_string(), title.to_string()),
+            if let Some(rewritten) = rewrite_intra_doc_link(db, *definition, target, title) {
+                return rewritten;
+            }
+            if let Definition::ModuleDef(def) = *definition {
+                if let Some(target) = rewrite_url_link(db, def, target) {
+                    return (target, title.to_string());
+                }
             }
+
+            (target.to_string(), title.to_string())
         }
     });
     let mut out = String::new();
@@ -48,6 +45,64 @@ pub fn rewrite_links(db: &RootDatabase, markdown: &str, definition: &Definition)
     out
 }
 
+fn rewrite_intra_doc_link(
+    db: &RootDatabase,
+    def: Definition,
+    target: &str,
+    title: &str,
+) -> Option<(String, String)> {
+    let link = if target.is_empty() { title } else { target };
+    let (link, ns) = parse_link(link);
+    let resolved = match def {
+        Definition::ModuleDef(def) => match def {
+            ModuleDef::Module(it) => it.resolve_doc_path(db, link, ns),
+            ModuleDef::Function(it) => it.resolve_doc_path(db, link, ns),
+            ModuleDef::Adt(it) => it.resolve_doc_path(db, link, ns),
+            ModuleDef::EnumVariant(it) => it.resolve_doc_path(db, link, ns),
+            ModuleDef::Const(it) => it.resolve_doc_path(db, link, ns),
+            ModuleDef::Static(it) => it.resolve_doc_path(db, link, ns),
+            ModuleDef::Trait(it) => it.resolve_doc_path(db, link, ns),
+            ModuleDef::TypeAlias(it) => it.resolve_doc_path(db, link, ns),
+            ModuleDef::BuiltinType(_) => return None,
+        },
+        Definition::Macro(it) => it.resolve_doc_path(db, link, ns),
+        Definition::Field(it) => it.resolve_doc_path(db, link, ns),
+        Definition::SelfType(_) | Definition::Local(_) | Definition::TypeParam(_) => return None,
+    }?;
+    let krate = resolved.module(db)?.krate();
+    let canonical_path = resolved.canonical_path(db)?;
+    let new_target = get_doc_url(db, &krate)?
+        .join(&format!("{}/", krate.display_name(db)?))
+        .ok()?
+        .join(&canonical_path.replace("::", "/"))
+        .ok()?
+        .join(&get_symbol_filename(db, &resolved)?)
+        .ok()?
+        .into_string();
+    let new_title = strip_prefixes_suffixes(title);
+    Some((new_target, new_title.to_string()))
+}
+
+/// Try to resolve path to local documentation via path-based links (i.e. `../gateway/struct.Shard.html`).
+fn rewrite_url_link(db: &RootDatabase, def: ModuleDef, target: &str) -> Option<String> {
+    if !(target.contains("#") || target.contains(".html")) {
+        return None;
+    }
+
+    let module = def.module(db)?;
+    let krate = module.krate();
+    let canonical_path = def.canonical_path(db)?;
+    let base = format!("{}/{}", krate.display_name(db)?, canonical_path.replace("::", "/"));
+
+    get_doc_url(db, &krate)
+        .and_then(|url| url.join(&base).ok())
+        .and_then(|url| {
+            get_symbol_filename(db, &def).as_deref().map(|f| url.join(f).ok()).flatten()
+        })
+        .and_then(|url| url.join(target).ok())
+        .map(|url| url.into_string())
+}
+
 // Rewrites a markdown document, resolving links using `callback` and additionally striping prefixes/suffixes on link titles.
 fn map_links<'e>(
     events: impl Iterator<Item = Event<'e>>,
@@ -79,3 +134,109 @@ fn map_links<'e>(
         _ => evt,
     })
 }
+
+fn parse_link(s: &str) -> (&str, Option<hir::Namespace>) {
+    let path = strip_prefixes_suffixes(s);
+    let ns = ns_from_intra_spec(s);
+    (path, ns)
+}
+
+/// Strip prefixes, suffixes, and inline code marks from the given string.
+fn strip_prefixes_suffixes(mut s: &str) -> &str {
+    s = s.trim_matches('`');
+
+    [
+        (TYPES.0.iter(), TYPES.1.iter()),
+        (VALUES.0.iter(), VALUES.1.iter()),
+        (MACROS.0.iter(), MACROS.1.iter()),
+    ]
+    .iter()
+    .for_each(|(prefixes, suffixes)| {
+        prefixes.clone().for_each(|prefix| s = s.trim_start_matches(*prefix));
+        suffixes.clone().for_each(|suffix| s = s.trim_end_matches(*suffix));
+    });
+    s.trim_start_matches("@").trim()
+}
+
+static TYPES: ([&str; 7], [&str; 0]) =
+    (["type", "struct", "enum", "mod", "trait", "union", "module"], []);
+static VALUES: ([&str; 8], [&str; 1]) =
+    (["value", "function", "fn", "method", "const", "static", "mod", "module"], ["()"]);
+static MACROS: ([&str; 1], [&str; 1]) = (["macro"], ["!"]);
+
+/// Extract the specified namespace from an intra-doc-link if one exists.
+///
+/// # Examples
+///
+/// * `struct MyStruct` -> `Namespace::Types`
+/// * `panic!` -> `Namespace::Macros`
+/// * `fn@from_intra_spec` -> `Namespace::Values`
+fn ns_from_intra_spec(s: &str) -> Option<hir::Namespace> {
+    [
+        (hir::Namespace::Types, (TYPES.0.iter(), TYPES.1.iter())),
+        (hir::Namespace::Values, (VALUES.0.iter(), VALUES.1.iter())),
+        (hir::Namespace::Macros, (MACROS.0.iter(), MACROS.1.iter())),
+    ]
+    .iter()
+    .filter(|(_ns, (prefixes, suffixes))| {
+        prefixes
+            .clone()
+            .map(|prefix| {
+                s.starts_with(*prefix)
+                    && s.chars()
+                        .nth(prefix.len() + 1)
+                        .map(|c| c == '@' || c == ' ')
+                        .unwrap_or(false)
+            })
+            .any(|cond| cond)
+            || suffixes
+                .clone()
+                .map(|suffix| {
+                    s.starts_with(*suffix)
+                        && s.chars()
+                            .nth(suffix.len() + 1)
+                            .map(|c| c == '@' || c == ' ')
+                            .unwrap_or(false)
+                })
+                .any(|cond| cond)
+    })
+    .map(|(ns, (_, _))| *ns)
+    .next()
+}
+
+fn get_doc_url(db: &RootDatabase, krate: &Crate) -> Option<Url> {
+    krate
+        .get_html_root_url(db)
+        .or_else(|| {
+            // Fallback to docs.rs. This uses `display_name` and can never be
+            // correct, but that's what fallbacks are about.
+            //
+            // FIXME: clicking on the link should just open the file in the editor,
+            // instead of falling back to external urls.
+            Some(format!("https://docs.rs/{}/*/", krate.display_name(db)?))
+        })
+        .and_then(|s| Url::parse(&s).ok())
+}
+
+/// Get the filename and extension generated for a symbol by rustdoc.
+///
+/// Example: `struct.Shard.html`
+fn get_symbol_filename(db: &RootDatabase, definition: &ModuleDef) -> Option<String> {
+    Some(match definition {
+        ModuleDef::Adt(adt) => match adt {
+            Adt::Struct(s) => format!("struct.{}.html", s.name(db)),
+            Adt::Enum(e) => format!("enum.{}.html", e.name(db)),
+            Adt::Union(u) => format!("union.{}.html", u.name(db)),
+        },
+        ModuleDef::Module(_) => "index.html".to_string(),
+        ModuleDef::Trait(t) => format!("trait.{}.html", t.name(db)),
+        ModuleDef::TypeAlias(t) => format!("type.{}.html", t.name(db)),
+        ModuleDef::BuiltinType(t) => format!("primitive.{}.html", t),
+        ModuleDef::Function(f) => format!("fn.{}.html", f.name(db)),
+        ModuleDef::EnumVariant(ev) => {
+            format!("enum.{}.html#variant.{}", ev.parent_enum(db).name(db), ev.name(db))
+        }
+        ModuleDef::Const(c) => format!("const.{}.html", c.name(db)?),
+        ModuleDef::Static(s) => format!("static.{}.html", s.name(db)?),
+    })
+}