Add Repository.hashfile()

This exposes libgit2's `git_repository_hashfile()`.

`pygit2.hashfile()` already exists, but this lets you hash files using the repository's filters.

Author: jorio

pygit2/decl/repository.h

@@ -83,6 +83,7 @@ int git_repository_set_head_detached(
 	git_repository* repo,
 	const git_oid* commitish);
 
+int git_repository_hashfile(git_oid *out, git_repository *repo, const char *path, git_object_t type, const char *as_path);
 int git_repository_ident(const char **name, const char **email, const git_repository *repo);
 int git_repository_set_ident(git_repository *repo, const char *name, const char *email);
 int git_repository_index(git_index **out, git_repository *repo);

pygit2/decl/types.h

@@ -57,3 +57,14 @@ typedef enum {
 	GIT_SUBMODULE_IGNORE_DIRTY     = 3,
 	GIT_SUBMODULE_IGNORE_ALL       = 4,
 } git_submodule_ignore_t;
+
+typedef enum {
+	GIT_OBJECT_ANY       = ...,
+	GIT_OBJECT_INVALID   = ...,
+	GIT_OBJECT_COMMIT    = ...,
+	GIT_OBJECT_TREE      = ...,
+	GIT_OBJECT_BLOB      = ...,
+	GIT_OBJECT_TAG       = ...,
+	GIT_OBJECT_OFS_DELTA = ...,
+	GIT_OBJECT_REF_DELTA = ...,
+} git_object_t;

pygit2/repository.py

@@ -51,6 +51,7 @@
     MergeFavor,
     MergeFileFlag,
     MergeFlag,
+    ObjectType,
     RepositoryOpenFlag,
     RepositoryState,
 )
@@ -129,6 +130,58 @@ def pack_all_objects(pack_builder):
 
         return builder.written_objects_count
 
+    def hashfile(
+        self,
+        path: str,
+        object_type: ObjectType = ObjectType.BLOB,
+        as_path: typing.Optional[str] = None,
+    ):
+        """Calculate the hash of a file using repository filtering rules.
+
+        If you simply want to calculate the hash of a file on disk with no filters,
+        you can just use `pygit2.hashfile()`. However, if you want to hash a file
+        in the repository and you want to apply filtering rules (e.g. crlf filters)
+        before generating the SHA, then use this function.
+
+        Note: if the repository has `core.safecrlf` set to fail and the filtering
+        triggers that failure, then this function will raise an error and not
+        calculate the hash of the file.
+
+        Returns: Output value of calculated SHA (Oid)
+
+        Parameters:
+
+        path
+            Path to file on disk whose contents should be hashed. This may be
+            an absolute path or a relative path, in which case it will be treated
+            as a path within the working directory.
+
+        object_type
+            The object type to hash (e.g. enums.ObjectType.BLOB)
+
+        as_path
+            The path to use to look up filtering rules. If this is an empty string
+            then no filters will be applied when calculating the hash.
+            If this is `None` and the `path` parameter is a file within the
+            repository's working directory, then the `path` will be used.
+        """
+        c_path = to_bytes(path)
+
+        if as_path is None:
+            c_as_path = ffi.NULL
+        else:
+            c_as_path = to_bytes(as_path)
+
+        c_oid = ffi.new('git_oid *')
+
+        err = C.git_repository_hashfile(
+            c_oid, self._repo, c_path, int(object_type), c_as_path
+        )
+        check_error(err)
+
+        oid = Oid(raw=bytes(ffi.buffer(c_oid.id)[:]))
+        return oid
+
     def __iter__(self):
         return iter(self.odb)
 

test/test_repository.py

@@ -898,3 +898,65 @@ def test_is_shallow(testrepo):
         f.write('abcdef0123456789abcdef0123456789abcdef00\n')
 
     assert testrepo.is_shallow
+
+
+def test_repository_hashfile(testrepo):
+    original_hash = testrepo.index['hello.txt'].id
+
+    # Test simple use
+    h = testrepo.hashfile('hello.txt')
+    assert h == original_hash
+
+    # Test absolute path
+    h = testrepo.hashfile(str(Path(testrepo.workdir, 'hello.txt')))
+    assert h == original_hash
+
+    # Test missing path
+    with pytest.raises(KeyError):
+        testrepo.hashfile('missing-file')
+
+    # Test invalid object type
+    with pytest.raises(pygit2.GitError):
+        testrepo.hashfile('hello.txt', ObjectType.OFS_DELTA)
+
+
+def test_repository_hashfile_filter(testrepo):
+    original_hash = testrepo.index['hello.txt'].id
+
+    with open(Path(testrepo.workdir, 'hello.txt'), 'rb') as f:
+        original_text = f.read()
+
+    crlf_data = original_text.replace(b'\n', b'\r\n')
+    crlf_hash = utils.gen_blob_sha1(crlf_data)
+    assert crlf_hash != original_hash
+
+    # Write hellocrlf.txt as a copy of hello.txt with CRLF line endings
+    with open(Path(testrepo.workdir, 'hellocrlf.txt'), 'wb') as f:
+        f.write(crlf_data)
+
+    # Set up a CRLF filter
+    testrepo.config['core.autocrlf'] = True
+    with open(Path(testrepo.workdir, '.gitattributes'), 'wt') as f:
+        f.write('*.txt text\n*.bin binary\n\n')
+
+    # By default, hellocrlf.txt should have the same hash as the original,
+    # due to core.autocrlf=True
+    h = testrepo.hashfile('hellocrlf.txt')
+    assert h == original_hash
+
+    # Treat absolute path with filters
+    h = testrepo.hashfile(str(Path(testrepo.workdir, 'hellocrlf.txt')))
+    assert h == original_hash
+
+    # Bypass filters
+    h = testrepo.hashfile('hellocrlf.txt', as_path='')
+    assert h == crlf_hash
+
+    # Bypass filters via .gitattributes
+    h = testrepo.hashfile('hellocrlf.txt', as_path='foobar.bin')
+    assert h == crlf_hash
+
+    # If core.safecrlf=fail, hashing a non-CRLF file will fail
+    testrepo.config['core.safecrlf'] = 'fail'
+    with pytest.raises(pygit2.GitError):
+        h = testrepo.hashfile('hello.txt')