Intelligently convert C/C++ comments to Rust

With this change, we can correctly parse C++ block comments.

```
/**
 * Does a thing
 *
 * More documentation. This test does something
 * useful.
 */
```

into

```
/// Does a thing
///
/// More documentation. This test does something
/// useful.
```

Fixes rust-lang#426.

Author: dylanmckay

src/ir/comment.rs

@@ -0,0 +1,92 @@
+/// The type of a comment.
+#[derive(Debug, PartialEq, Eq)]
+enum Kind {
+    /// A `///` comment, or something of the like.
+    /// All lines in a comment should start with the same symbol.
+    SingleLines,
+    /// A `/**` comment, where each other line can start with `*` and the
+    /// entire block ends with `*/`.
+    MultiLine,
+}
+/// Preprocesses a C/C++ comment so that it is a valid Rust comment.
+pub fn preprocess(comment: String) -> String {
+    match self::kind(&comment) {
+        Some(Kind::SingleLines) => preprocess_single_lines(&comment),
+        Some(Kind::MultiLine) => preprocess_multi_line(&comment),
+        None => comment.to_owned(),
+    }
+}
+
+/// Gets the kind of the doc comment, if it is one.
+fn kind(comment: &str) -> Option<Kind> {
+    if comment.starts_with("/*") {
+        Some(Kind::MultiLine)
+    } else if comment.starts_with("//") {
+        Some(Kind::SingleLines)
+    } else {
+        None
+    }
+}
+
+/// Preprocesses mulitple single line comments.
+///
+/// Handles lines starting with both `//` and `///`.
+fn preprocess_single_lines(comment: &str) -> String {
+    assert!(comment.starts_with("//"), "comment is not single line");
+
+    let lines: Vec<_> = comment.lines()
+                               .map(|l| l.trim_left_matches('/').trim())
+                               .map(|l| format!("/// {}", l).trim().to_owned())
+                               .collect();
+    lines.join("\n")
+}
+
+fn preprocess_multi_line(comment: &str) -> String {
+    let comment = comment.trim_left_matches('/')
+                         .trim_left_matches("*")
+                         .trim_left_matches("!")
+                         .trim_right_matches('/')
+                         .trim_right_matches('*')
+                         .trim();
+
+    // Strip any potential `*` characters preceding each line.
+    let mut lines: Vec<_> = comment.lines()
+        .map(|line| line.trim().trim_left_matches('*').trim())
+        .skip_while(|line| line.is_empty()) // Skip the first empty lines.
+        .map(|line| format!("/// {}", line).trim().to_owned())
+        .collect();
+
+    // Remove the trailing `*/`.
+    let last_idx = lines.len() - 1;
+    if lines[last_idx].is_empty() {
+        lines.remove(last_idx);
+    }
+
+    lines.join("\n")
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+
+    #[test]
+    fn picks_up_single_and_multi_line_doc_comments() {
+        assert_eq!(kind("/// hello"), Some(Kind::SingleLines));
+        assert_eq!(kind("/** world */"), Some(Kind::MultiLine));
+    }
+
+    #[test]
+    fn processes_single_lines_correctly() {
+        assert_eq!(preprocess("/// hello".to_owned()), "/// hello");
+        assert_eq!(preprocess("// hello".to_owned()), "/// hello");
+    }
+
+    #[test]
+    fn processes_multi_lines_correctly() {
+        assert_eq!(preprocess("/** hello \n * world \n * foo \n */".to_owned()),
+                   "/// hello\n/// world\n/// foo");
+
+        assert_eq!(preprocess("/**\nhello\n*world\n*foo\n*/".to_owned()),
+                   "/// hello\n/// world\n/// foo");
+    }
+}

src/ir/comp.rs

@@ -1,6 +1,7 @@
 //! Compound types (unions and structs) in our intermediate representation.
 
 use super::annotations::Annotations;
+use super::comment;
 use super::context::{BindgenContext, ItemId};
 use super::derive::{CanDeriveCopy, CanDeriveDebug, CanDeriveDefault};
 use super::dot::DotAttributes;
@@ -1101,7 +1102,7 @@ impl CompInfo {
                                                           Some(potential_id),
                                                           ctx);
 
-                    let comment = cur.raw_comment();
+                    let comment = cur.raw_comment().map(comment::preprocess);
                     let annotations = Annotations::new(&cur);
                     let name = cur.spelling();
                     let is_mutable = cursor.is_mutable_field();

src/ir/enum_ty.rs

@@ -1,5 +1,6 @@
 //! Intermediate representation for C/C++ enumerations.
 
+use super::comment;
 use super::context::{BindgenContext, ItemId};
 use super::item::Item;
 use super::ty::TypeKind;
@@ -113,7 +114,7 @@ impl Enum {
                                 })
                         });
 
-                    let comment = cursor.raw_comment();
+                    let comment = cursor.raw_comment().map(comment::preprocess);
                     variants.push(EnumVariant::new(name,
                                                    comment,
                                                    val,

src/ir/function.rs

@@ -1,5 +1,6 @@
 //! Intermediate representation for C/C++ functions and methods.
 
+use super::comment;
 use super::context::{BindgenContext, ItemId};
 use super::dot::DotAttributes;
 use super::item::Item;
@@ -405,7 +406,7 @@ impl ClangSubItemParser for Function {
             mangled_name = None;
         }
 
-        let comment = cursor.raw_comment();
+        let comment = cursor.raw_comment().map(comment::preprocess);
 
         let function = Self::new(name, mangled_name, sig, comment);
         Ok(ParseResult::New(function, Some(cursor)))

src/ir/item.rs

@@ -2,6 +2,7 @@
 
 use super::super::codegen::CONSTIFIED_ENUM_MODULE_REPR_NAME;
 use super::annotations::Annotations;
+use super::comment;
 use super::context::{BindgenContext, ItemId, PartialType};
 use super::derive::{CanDeriveCopy, CanDeriveDebug, CanDeriveDefault};
 use super::dot::DotAttributes;
@@ -1001,7 +1002,7 @@ impl ClangItemParser for Item {
             return Err(ParseError::Continue);
         }
 
-        let comment = cursor.raw_comment();
+        let comment = cursor.raw_comment().map(comment::preprocess);
         let annotations = Annotations::new(&cursor);
 
         let current_module = ctx.current_module();
@@ -1207,7 +1208,8 @@ impl ClangItemParser for Item {
         };
 
         let comment = decl.raw_comment()
-            .or_else(|| location.raw_comment());
+            .or_else(|| location.raw_comment())
+            .map(comment::preprocess);
         let annotations = Annotations::new(&decl)
             .or_else(|| Annotations::new(&location));
 

src/ir/mod.rs

@@ -5,6 +5,7 @@
 
 pub mod annotations;
 pub mod comp;
+pub mod comment;
 pub mod context;
 pub mod derive;
 pub mod dot;

tests/expectations/tests/accessors.rs

@@ -8,11 +8,11 @@
 #[derive(Debug, Default, Copy)]
 pub struct SomeAccessors {
     pub mNoAccessor: ::std::os::raw::c_int,
-    /** <div rustbindgen accessor></div> */
+    /// <div rustbindgen accessor></div>
     pub mBothAccessors: ::std::os::raw::c_int,
-    /** <div rustbindgen accessor="unsafe"></div> */
+    /// <div rustbindgen accessor="unsafe"></div>
     pub mUnsafeAccessors: ::std::os::raw::c_int,
-    /** <div rustbindgen accessor="immutable"></div> */
+    /// <div rustbindgen accessor="immutable"></div>
     pub mImmutableAccessor: ::std::os::raw::c_int,
 }
 #[test]
@@ -68,7 +68,7 @@ impl SomeAccessors {
         &self.mImmutableAccessor
     }
 }
-/** <div rustbindgen accessor></div> */
+/// <div rustbindgen accessor></div>
 #[repr(C)]
 #[derive(Debug, Default, Copy)]
 pub struct AllAccessors {
@@ -114,7 +114,7 @@ impl AllAccessors {
         &mut self.mAlsoBothAccessors
     }
 }
-/** <div rustbindgen accessor="unsafe"></div> */
+/// <div rustbindgen accessor="unsafe"></div>
 #[repr(C)]
 #[derive(Debug, Default, Copy)]
 pub struct AllUnsafeAccessors {
@@ -162,16 +162,16 @@ impl AllUnsafeAccessors {
         &mut self.mAlsoBothAccessors
     }
 }
-/** <div rustbindgen accessor></div> */
+/// <div rustbindgen accessor></div>
 #[repr(C)]
 #[derive(Debug, Default, Copy)]
 pub struct ContradictAccessors {
     pub mBothAccessors: ::std::os::raw::c_int,
-    /** <div rustbindgen accessor="false"></div> */
+    /// <div rustbindgen accessor="false"></div>
     pub mNoAccessors: ::std::os::raw::c_int,
-    /** <div rustbindgen accessor="unsafe"></div> */
+    /// <div rustbindgen accessor="unsafe"></div>
     pub mUnsafeAccessors: ::std::os::raw::c_int,
-    /** <div rustbindgen accessor="immutable"></div> */
+    /// <div rustbindgen accessor="immutable"></div>
     pub mImmutableAccessor: ::std::os::raw::c_int,
 }
 #[test]
@@ -229,7 +229,7 @@ impl ContradictAccessors {
         &self.mImmutableAccessor
     }
 }
-/** <div rustbindgen accessor replaces="Replaced"></div> */
+/// <div rustbindgen accessor replaces="Replaced"></div>
 #[repr(C)]
 #[derive(Debug, Default, Copy)]
 pub struct Replaced {
@@ -258,7 +258,7 @@ impl Replaced {
         &mut self.mAccessor
     }
 }
-/** <div rustbindgen accessor></div> */
+/// <div rustbindgen accessor></div>
 #[repr(C)]
 #[derive(Debug, Default, Copy)]
 pub struct Wrapper {

tests/expectations/tests/annotation_hide.rs

@@ -4,9 +4,7 @@
 #![allow(dead_code, non_snake_case, non_camel_case_types, non_upper_case_globals)]
 
 
-/**
- * <div rustbindgen opaque></div>
- */
+/// <div rustbindgen opaque></div>
 #[repr(C)]
 #[derive(Debug, Default, Copy)]
 pub struct D {

tests/expectations/tests/class_use_as.rs

@@ -4,9 +4,7 @@
 #![allow(dead_code, non_snake_case, non_camel_case_types, non_upper_case_globals)]
 
 
-/**
- * <div rustbindgen="true" replaces="whatever"></div>
- */
+/// <div rustbindgen="true" replaces="whatever"></div>
 #[repr(C)]
 #[derive(Debug, Default, Copy)]
 pub struct whatever {