Add Repository.hashfile(path) which maps libgit2 git_repository_hashfile

pygit2/_pygit2.pyi

@@ -471,6 +471,7 @@ class Repository:
     def expand_id(self, hex: str) -> Oid: ...
     def free(self) -> None: ...
     def git_object_lookup_prefix(self, oid: _OidArg) -> Object: ...
+    def hashfile(self, path: str) -> Oid: ...
     def init_submodules(self, submodules: list[Submodule] = None, overwrite = False) -> None: ...
     def list_worktrees(self) -> list[str]: ...
     def listall_branches(self, flag: int = GIT_BRANCH_LOCAL) -> list[str]: ...

src/repository.c

@@ -2435,6 +2435,41 @@ Repository_listall_stashes(Repository *self, PyObject *args)
     }
 }
 
+PyDoc_STRVAR(Repository_hashfile__doc__,
+  "hashfile(path: str) -> Oid\n"
+  "\n"
+  "Calculate hash of file using repository filtering rules.\n"
+  "\n"
+  "If you simply want to calculate the hash of a file on disk with no filters, "
+  "you can just use the ``pygit2.hashfile`` API. 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.\n"
+  "\n"
+  "Note: if the repository has ``core.safecrlf`` set to fail and the filtering "
+  "triggers that failure, then this function will raise ``GitError``.");
+
+PyObject *
+Repository_hashfile(Repository *self, PyObject *args)
+{
+    git_oid oid;
+    PyBytesObject *py_path = NULL;
+    const char* path = NULL;
+    int err;
+
+    if (!PyArg_ParseTuple(args, "O&", PyUnicode_FSConverter, &py_path))
+        return NULL;
+
+    if (py_path != NULL)
+        path = PyBytes_AS_STRING(py_path);
+
+    err = git_repository_hashfile(&oid, self->repo, path, GIT_OBJ_BLOB, NULL);
+    Py_XDECREF(py_path);
+    if (err < 0)
+        return Error_set(err);
+
+    return git_oid_to_python(&oid);
+}
+
 PyMethodDef Repository_methods[] = {
     METHOD(Repository, create_blob, METH_VARARGS),
     METHOD(Repository, create_blob_fromworkdir, METH_O),
@@ -2490,6 +2525,7 @@ PyMethodDef Repository_methods[] = {
     METHOD(Repository, set_odb, METH_O),
     METHOD(Repository, set_refdb, METH_O),
     METHOD(Repository, listall_stashes, METH_NOARGS),
+    METHOD(Repository, hashfile, METH_VARARGS),
     {NULL}
 };
 

src/repository.h

@@ -69,4 +69,6 @@ PyObject* Repository_cherrypick(Repository *self, PyObject *py_oid);
 PyObject* Repository_apply(Repository *self, PyObject *py_diff, PyObject *kwds);
 PyObject* Repository_merge_analysis(Repository *self, PyObject *args);
 
+PyObject* Repository_hashfile(Repository *self, PyObject *args);
+
 #endif

test/test_repository.py

@@ -759,3 +759,54 @@ def test_is_shallow(testrepo):
         f.write('abcdef0123456789abcdef0123456789abcdef00\n')
 
     assert testrepo.is_shallow
+
+def test_repo_hashfile_same_hash(testrepo):
+    data = 'Some multi-\nline text\n'
+    with (Path(testrepo.workdir) / 'untracked.txt').open('w') as f:
+        f.write(data)
+
+    hashed_file_sha1 = testrepo.hashfile(str(Path(testrepo.workdir) / 'untracked.txt'))
+    hashed_data_sha1 = pygit2.hash(data)
+    assert hashed_file_sha1 == hashed_data_sha1
+
+def test_repo_hashfile_crlf_normalization(testrepo):
+    with (Path(testrepo.workdir) / '.gitattributes').open('w+') as f:
+        print('*.txt  eol=lf\n', file=f)
+
+    data = 'Some multi-\nline text\n'
+    with (Path(testrepo.workdir) / 'untracked_lf.txt').open('w') as f:
+        f.write(data)
+    with (Path(testrepo.workdir) / 'untracked_crlf.txt').open('w') as f:
+        f.write(data.replace('\n','\r\n'))
+
+    hashed_lf_sha1 = testrepo.hashfile(str(Path(testrepo.workdir) / 'untracked_lf.txt'))
+    hashed_crlf_sha1 = testrepo.hashfile(str(Path(testrepo.workdir) / 'untracked_crlf.txt'))
+    assert hashed_lf_sha1 == hashed_crlf_sha1
+
+def test_repo_hashfile_no_normalization(testrepo):
+    with (Path(testrepo.workdir) / '.gitattributes').open('w+') as f:
+        print('*.txt  -text\n', file=f)
+
+    data = 'Some multi-\nline text\n'
+    with (Path(testrepo.workdir) / 'untracked_lf.txt').open('w') as f:
+        f.write(data)
+    with (Path(testrepo.workdir) / 'untracked_crlf.txt').open('w') as f:
+        f.write(data.replace('\n','\r\n'))
+
+    hashed_lf_sha1 = testrepo.hashfile(str(Path(testrepo.workdir) / 'untracked_lf.txt'))
+    hashed_crlf_sha1 = testrepo.hashfile(str(Path(testrepo.workdir) / 'untracked_crlf.txt'))
+    assert hashed_lf_sha1 != hashed_crlf_sha1
+
+def test_repo_hashfile_crlf_normalization_error(testrepo):
+    testrepo.config['core.safecrlf'] = True
+    with (Path(testrepo.workdir) / '.gitattributes').open('w+') as f:
+        print('*.txt  eol=lf\n', file=f)
+    with (Path(testrepo.workdir) / 'untracked_crlf.txt').open('w') as f:
+        f.write('Some multi-\r\nline text\r\n')
+
+    with pytest.raises(pygit2.GitError) as exc:
+        testrepo.hashfile(str(Path(testrepo.workdir) / 'untracked_crlf.txt'))
+
+    assert "CRLF would be replaced by LF" in str(exc.value)
+
+