Add bindings for git_reference_normalize_name (rust-lang#620)

* Add bindings for `git_reference_normalize_name`

* fixup! Add bindings for `git_reference_normalize_name`

* Fix type

* GIT_REFNAME_MAX is not in the includes

Author: kim

libgit2-sys/lib.rs

@@ -1814,6 +1814,15 @@ git_enum! {
     }
 }
 
+git_enum! {
+    pub enum git_reference_format_t {
+        GIT_REFERENCE_FORMAT_NORMAL = 0,
+        GIT_REFERENCE_FORMAT_ALLOW_ONELEVEL = 1 << 0,
+        GIT_REFERENCE_FORMAT_REFSPEC_PATTERN = 1 << 1,
+        GIT_REFERENCE_FORMAT_REFSPEC_SHORTHAND = 1 << 2,
+    }
+}
+
 extern "C" {
     // threads
     pub fn git_libgit2_init() -> c_int;
@@ -2278,6 +2287,12 @@ extern "C" {
     ) -> c_int;
     pub fn git_reference_has_log(repo: *mut git_repository, name: *const c_char) -> c_int;
     pub fn git_reference_ensure_log(repo: *mut git_repository, name: *const c_char) -> c_int;
+    pub fn git_reference_normalize_name(
+        buffer_out: *mut c_char,
+        buffer_size: size_t,
+        name: *const c_char,
+        flags: u32,
+    ) -> c_int;
 
     // stash
     pub fn git_stash_save(

src/lib.rs

@@ -1403,6 +1403,40 @@ impl DiffFlags {
     is_bit_set!(exists, DiffFlags::EXISTS);
 }
 
+bitflags! {
+    /// Options for [`Reference::normalize_name`].
+    pub struct ReferenceFormat: u32 {
+        /// No particular normalization.
+        const NORMAL = raw::GIT_REFERENCE_FORMAT_NORMAL as u32;
+        /// Constrol whether one-level refname are accepted (i.e., refnames that
+        /// do not contain multiple `/`-separated components). Those are
+        /// expected to be written only using uppercase letters and underscore
+        /// (e.g. `HEAD`, `FETCH_HEAD`).
+        const ALLOW_ONELEVEL = raw::GIT_REFERENCE_FORMAT_ALLOW_ONELEVEL as u32;
+        /// Interpret the provided name as a reference pattern for a refspec (as
+        /// used with remote repositories). If this option is enabled, the name
+        /// is allowed to contain a single `*` in place of a full pathname
+        /// components (e.g., `foo/*/bar` but not `foo/bar*`).
+        const REFSPEC_PATTERN = raw::GIT_REFERENCE_FORMAT_REFSPEC_PATTERN as u32;
+        /// Interpret the name as part of a refspec in shorthand form so the
+        /// `ALLOW_ONELEVEL` naming rules aren't enforced and `master` becomes a
+        /// valid name.
+        const REFSPEC_SHORTHAND = raw::GIT_REFERENCE_FORMAT_REFSPEC_SHORTHAND as u32;
+    }
+}
+
+impl ReferenceFormat {
+    is_bit_set!(is_allow_onelevel, ReferenceFormat::ALLOW_ONELEVEL);
+    is_bit_set!(is_refspec_pattern, ReferenceFormat::REFSPEC_PATTERN);
+    is_bit_set!(is_refspec_shorthand, ReferenceFormat::REFSPEC_SHORTHAND);
+}
+
+impl Default for ReferenceFormat {
+    fn default() -> Self {
+        ReferenceFormat::NORMAL
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use super::ObjectType;

src/reference.rs

@@ -8,9 +8,14 @@ use std::str;
 use crate::object::CastOrPanic;
 use crate::util::{c_cmp_to_ordering, Binding};
 use crate::{
-    raw, Blob, Commit, Error, Object, ObjectType, Oid, ReferenceType, Repository, Tag, Tree,
+    raw, Blob, Commit, Error, Object, ObjectType, Oid, ReferenceFormat, ReferenceType, Repository,
+    Tag, Tree,
 };
 
+// Not in the public header files (yet?), but a hard limit used by libgit2
+// internally
+const GIT_REFNAME_MAX: usize = 1024;
+
 struct Refdb<'repo>(&'repo Repository);
 
 /// A structure to represent a git [reference][1].
@@ -34,12 +39,120 @@ pub struct ReferenceNames<'repo, 'references> {
 
 impl<'repo> Reference<'repo> {
     /// Ensure the reference name is well-formed.
+    ///
+    /// Validation is performed as if [`ReferenceFormat::ALLOW_ONELEVEL`]
+    /// was given to [`Reference::normalize_name`]. No normalization is
+    /// performed, however.
+    ///
+    /// ```rust
+    /// use git2::Reference;
+    ///
+    /// assert!(Reference::is_valid_name("HEAD"));
+    /// assert!(Reference::is_valid_name("refs/heads/master"));
+    ///
+    /// // But:
+    /// assert!(!Reference::is_valid_name("master"));
+    /// assert!(!Reference::is_valid_name("refs/heads/*"));
+    /// assert!(!Reference::is_valid_name("foo//bar"));
+    /// ```
+    ///
+    /// [`ReferenceFormat::ALLOW_ONELEVEL`]:
+    ///     struct.ReferenceFormat#associatedconstant.ALLOW_ONELEVEL
+    /// [`Reference::normalize_name`]: struct.Reference#method.normalize_name
     pub fn is_valid_name(refname: &str) -> bool {
         crate::init();
         let refname = CString::new(refname).unwrap();
         unsafe { raw::git_reference_is_valid_name(refname.as_ptr()) == 1 }
     }
 
+    /// Normalize reference name and check validity.
+    ///
+    /// This will normalize the reference name by collapsing runs of adjacent
+    /// slashes between name components into a single slash. It also validates
+    /// the name according to the following rules:
+    ///
+    /// 1. If [`ReferenceFormat::ALLOW_ONELEVEL`] is given, the name may
+    ///    contain only capital letters and underscores, and must begin and end
+    ///    with a letter. (e.g. "HEAD", "ORIG_HEAD").
+    /// 2. The flag [`ReferenceFormat::REFSPEC_SHORTHAND`] has an effect
+    ///    only when combined with [`ReferenceFormat::ALLOW_ONELEVEL`]. If
+    ///    it is given, "shorthand" branch names (i.e. those not prefixed by
+    ///    `refs/`, but consisting of a single word without `/` separators)
+    ///    become valid. For example, "master" would be accepted.
+    /// 3. If [`ReferenceFormat::REFSPEC_PATTERN`] is given, the name may
+    ///    contain a single `*` in place of a full pathname component (e.g.
+    ///    `foo/*/bar`, `foo/bar*`).
+    /// 4. Names prefixed with "refs/" can be almost anything. You must avoid
+    ///    the characters '~', '^', ':', '\\', '?', '[', and '*', and the
+    ///    sequences ".." and "@{" which have special meaning to revparse.
+    ///
+    /// If the reference passes validation, it is returned in normalized form,
+    /// otherwise an [`Error`] with [`ErrorCode::InvalidSpec`] is returned.
+    ///
+    /// ```rust
+    /// use git2::{Reference, ReferenceFormat};
+    ///
+    /// assert_eq!(
+    ///     Reference::normalize_name(
+    ///         "foo//bar",
+    ///         ReferenceFormat::NORMAL
+    ///     )
+    ///     .unwrap(),
+    ///     "foo/bar".to_owned()
+    /// );
+    ///
+    /// assert_eq!(
+    ///     Reference::normalize_name(
+    ///         "HEAD",
+    ///         ReferenceFormat::ALLOW_ONELEVEL
+    ///     )
+    ///     .unwrap(),
+    ///     "HEAD".to_owned()
+    /// );
+    ///
+    /// assert_eq!(
+    ///     Reference::normalize_name(
+    ///         "refs/heads/*",
+    ///         ReferenceFormat::REFSPEC_PATTERN
+    ///     )
+    ///     .unwrap(),
+    ///     "refs/heads/*".to_owned()
+    /// );
+    ///
+    /// assert_eq!(
+    ///     Reference::normalize_name(
+    ///         "master",
+    ///         ReferenceFormat::ALLOW_ONELEVEL | ReferenceFormat::REFSPEC_SHORTHAND
+    ///     )
+    ///     .unwrap(),
+    ///     "master".to_owned()
+    /// );
+    /// ```
+    ///
+    /// [`ReferenceFormat::ALLOW_ONELEVEL`]:
+    ///     struct.ReferenceFormat#associatedconstant.ALLOW_ONELEVEL
+    /// [`ReferenceFormat::REFSPEC_SHORTHAND`]:
+    ///     struct.ReferenceFormat#associatedconstant.REFSPEC_SHORTHAND
+    /// [`ReferenceFormat::REFSPEC_PATTERN`]:
+    ///     struct.ReferenceFormat#associatedconstant.REFSPEC_PATTERN
+    /// [`Error`]: struct.Error
+    /// [`ErrorCode::InvalidSpec`]: enum.ErrorCode#variant.InvalidSpec
+    pub fn normalize_name(refname: &str, flags: ReferenceFormat) -> Result<String, Error> {
+        crate::init();
+        let mut dst = [0u8; GIT_REFNAME_MAX];
+        let refname = CString::new(refname)?;
+        unsafe {
+            try_call!(raw::git_reference_normalize_name(
+                dst.as_mut_ptr() as *mut libc::c_char,
+                dst.len() as libc::size_t,
+                refname,
+                flags.bits()
+            ));
+            let s = &dst[..dst.iter().position(|&a| a == 0).unwrap()];
+            Ok(str::from_utf8(s).unwrap().to_owned())
+        }
+    }
+
     /// Get access to the underlying raw pointer.
     pub fn raw(&self) -> *mut raw::git_reference {
         self.raw