rustdoc: Add the concept of 'sections'

Author: brson

src/rustdoc/doc.rs

@@ -11,6 +11,15 @@ enum page {
     itempage(itemtag)
 }
 
+#[doc = "
+Most rustdocs can be parsed into 'sections' according to their markdown
+headers
+"]
+type section = {
+    header: str,
+    body: str
+};
+
 // FIXME: We currently give topmod the name of the crate.  There would
 // probably be fewer special cases if the crate had its own name and
 // topmod's name was the empty string.
@@ -36,6 +45,7 @@ type itemdoc = {
     path: [str],
     brief: option<str>,
     desc: option<str>,
+    sections: [section],
     // Indicates that this node is a reexport of a different item
     reexport: bool
 };
@@ -99,6 +109,7 @@ type methoddoc = {
     name: str,
     brief: option<str>,
     desc: option<str>,
+    sections: [section],
     args: [argdoc],
     return: retdoc,
     failure: option<str>,
@@ -400,4 +411,8 @@ impl util<A:item> for A {
     fn desc() -> option<str> {
         self.item().desc
     }
+
+    fn sections() -> [section] {
+        self.item().sections
+    }
 }

src/rustdoc/extract.rs

@@ -44,6 +44,7 @@ fn mk_itemdoc(id: ast::node_id, name: ast::ident) -> doc::itemdoc {
         path: [],
         brief: none,
         desc: none,
+        sections: [],
         reexport: false
     }
 }
@@ -250,6 +251,7 @@ fn ifacedoc_from_iface(
                 name: method.ident,
                 brief: none,
                 desc: none,
+                sections: [],
                 args: argdocs_from_args(method.decl.inputs),
                 return: {
                     desc: none
@@ -292,6 +294,7 @@ fn impldoc_from_impl(
                 name: method.ident,
                 brief: none,
                 desc: none,
+                sections: [],
                 args: argdocs_from_args(method.decl.inputs),
                 return: {
                     desc: none

src/rustdoc/rustdoc.rc

@@ -40,3 +40,4 @@ mod sort_item_type_pass;
 mod reexport_pass;
 mod par;
 mod page_pass;
+mod sectionalize_pass;

src/rustdoc/rustdoc.rs

@@ -148,8 +148,9 @@ fn run(config: config::config) {
             // prune_undoc_items_pass::mk_pass(),
             prune_hidden_pass::mk_pass(),
             desc_to_brief_pass::mk_pass(),
-            trim_pass::mk_pass(),
             unindent_pass::mk_pass(),
+            sectionalize_pass::mk_pass(),
+            trim_pass::mk_pass(),
             sort_item_name_pass::mk_pass(),
             sort_item_type_pass::mk_pass(),
             markdown_index_pass::mk_pass(config),

src/rustdoc/sectionalize_pass.rs

@@ -0,0 +1,237 @@
+#[doc = "Breaks rustdocs into sections according to their headers"];
+
+export mk_pass;
+
+fn mk_pass() -> pass {
+    {
+        name: "sectionalize",
+        f: run
+    }
+}
+
+fn run(_srv: astsrv::srv, doc: doc::doc) -> doc::doc {
+    let fold = fold::fold({
+        fold_item: fold_item,
+        fold_iface: fold_iface,
+        fold_impl: fold_impl
+        with *fold::default_any_fold(())
+    });
+    fold.fold_doc(fold, doc)
+}
+
+fn fold_item(fold: fold::fold<()>, doc: doc::itemdoc) -> doc::itemdoc {
+    let doc = fold::default_seq_fold_item(fold, doc);
+    let (desc, sections) = sectionalize(doc.desc);
+
+    {
+        desc: desc,
+        sections: sections
+        with doc
+    }
+}
+
+fn fold_iface(fold: fold::fold<()>, doc: doc::ifacedoc) -> doc::ifacedoc {
+    let doc = fold::default_seq_fold_iface(fold, doc);
+
+    {
+        methods: par::anymap(doc.methods) {|method|
+            let (desc, sections) = sectionalize(method.desc);
+
+            {
+                desc: desc,
+                sections: sections
+                with method
+            }
+        }
+        with doc
+    }
+}
+
+fn fold_impl(fold: fold::fold<()>, doc: doc::impldoc) -> doc::impldoc {
+    let doc = fold::default_seq_fold_impl(fold, doc);
+
+    {
+        methods: par::anymap(doc.methods) {|method|
+            let (desc, sections) = sectionalize(method.desc);
+
+            {
+                desc: desc,
+                sections: sections
+                with method
+            }
+        }
+        with doc
+    }
+}
+
+fn sectionalize(desc: option<str>) -> (option<str>, [doc::section]) {
+
+    #[doc = "
+
+    Take a description of the form
+
+        General text
+
+        # Section header
+
+        Section text
+
+        # Section header
+
+        Section text
+
+    and remove each header and accompanying text into section records.
+
+    "];
+
+    if option::is_none(desc) {
+        ret (none, []);
+    }
+
+    let lines = str::lines(option::get(desc));
+
+    let new_desc = none;
+    let current_section = none;
+    let sections = [];
+
+    for line in lines {
+        alt parse_header(line) {
+          some(header) {
+            if option::is_some(current_section) {
+                sections += [option::get(current_section)];
+            }
+            current_section = some({
+                header: header,
+                body: ""
+            });
+          }
+          none {
+            alt current_section {
+              some(section) {
+                current_section = some({
+                    body: section.body + "\n" + line
+                    with section
+                });
+              }
+              none {
+                alt new_desc {
+                  some(desc) {
+                    new_desc = some(desc + "\n" + line);
+                  }
+                  none {
+                    new_desc = some(line);
+                  }
+                }
+              }
+            }
+          }
+        }
+    }
+
+    if option::is_some(current_section) {
+        sections += [option::get(current_section)];
+    }
+
+    (new_desc, sections)
+}
+
+fn parse_header(line: str) -> option<str> {
+    if str::starts_with(line, "# ") {
+        some(str::slice(line, 2u, str::len(line)))
+    } else {
+        none
+    }
+}
+
+#[test]
+fn should_create_section_headers() {
+    let doc = test::mk_doc(
+        "#[doc = \"\
+         # Header\n\
+         Body\"]\
+         mod a { }");
+    assert str::contains(
+        doc.cratemod().mods()[0].item.sections[0].header,
+        "Header");
+}
+
+#[test]
+fn should_create_section_bodies() {
+    let doc = test::mk_doc(
+        "#[doc = \"\
+         # Header\n\
+         Body\"]\
+         mod a { }");
+    assert str::contains(
+        doc.cratemod().mods()[0].item.sections[0].body,
+        "Body");
+}
+
+#[test]
+fn should_not_create_sections_from_indented_headers() {
+    let doc = test::mk_doc(
+        "#[doc = \"\n\
+         Text\n             # Header\n\
+         Body\"]\
+         mod a { }");
+    assert vec::is_empty(doc.cratemod().mods()[0].item.sections);
+}
+
+#[test]
+fn should_remove_section_text_from_main_desc() {
+    let doc = test::mk_doc(
+        "#[doc = \"\
+         Description\n\n\
+         # Header\n\
+         Body\"]\
+         mod a { }");
+    assert !str::contains(
+        option::get(doc.cratemod().mods()[0].desc()),
+        "Header");
+    assert !str::contains(
+        option::get(doc.cratemod().mods()[0].desc()),
+        "Body");
+}
+
+#[test]
+fn should_eliminate_desc_if_it_is_just_whitespace() {
+    let doc = test::mk_doc(
+        "#[doc = \"\
+         # Header\n\
+         Body\"]\
+         mod a { }");
+    assert doc.cratemod().mods()[0].desc() == none;
+}
+
+#[test]
+fn should_sectionalize_iface_methods() {
+    let doc = test::mk_doc(
+        "iface i {
+         #[doc = \"\
+         # Header\n\
+         Body\"]\
+         fn a(); }");
+    assert doc.cratemod().ifaces()[0].methods[0].sections.len() == 1u;
+}
+
+#[test]
+fn should_sectionalize_impl_methods() {
+    let doc = test::mk_doc(
+        "impl i for bool {
+         #[doc = \"\
+         # Header\n\
+         Body\"]\
+         fn a() { } }");
+    assert doc.cratemod().impls()[0].methods[0].sections.len() == 1u;
+}
+
+#[cfg(test)]
+mod test {
+    fn mk_doc(source: str) -> doc::doc {
+        astsrv::from_str(source) {|srv|
+            let doc = extract::from_srv(srv, "");
+            let doc = attr_pass::mk_pass().f(srv, doc);
+            run(srv, doc)
+        }
+    }
+}