Add FilterList.apply_to_buffer, apply_to_file, apply_to_blob

Author: jorio

pygit2/_libgit2/ffi.pyi

@@ -123,6 +123,10 @@ class GitBlameC:
     # incomplete
     pass
 
+class GitBlobC:
+    # incomplete
+    pass
+
 class GitMergeOptionsC:
     file_favor: int
     flags: int
@@ -268,6 +272,8 @@ def new(a: Literal['git_oid *']) -> GitOidC: ...
 @overload
 def new(a: Literal['git_blame **']) -> _Pointer[GitBlameC]: ...
 @overload
+def new(a: Literal['git_blob **']) -> _Pointer[GitBlobC]: ...
+@overload
 def new(a: Literal['git_clone_options *']) -> GitCloneOptionsC: ...
 @overload
 def new(a: Literal['git_merge_options *']) -> GitMergeOptionsC: ...

pygit2/decl/filter.h

@@ -26,6 +26,23 @@ int git_filter_list_contains(
 	git_filter_list *filters,
 	const char *name);
 
+int git_filter_list_apply_to_buffer(
+	git_buf *out,
+	git_filter_list *filters,
+	const char* in,
+	size_t in_len);
+
+int git_filter_list_apply_to_file(
+	git_buf *out,
+	git_filter_list *filters,
+	git_repository *repo,
+	const char *path);
+
+int git_filter_list_apply_to_blob(
+	git_buf *out,
+	git_filter_list *filters,
+	git_blob *blob);
+
 size_t git_filter_list_length(
 	const git_filter_list *fl);
 

pygit2/filter.py

@@ -29,12 +29,14 @@
 from collections.abc import Callable
 from typing import TYPE_CHECKING
 
-from ._pygit2 import FilterSource
+from ._pygit2 import Blob, FilterSource
+from .errors import check_error
 from .ffi import C, ffi
 from .utils import to_bytes
 
 if TYPE_CHECKING:
     from ._libgit2.ffi import GitFilterListC
+    from .repository import BaseRepository
 
 
 class Filter:
@@ -157,5 +159,49 @@ def __contains__(self, name: str) -> bool:
     def __len__(self) -> int:
         return C.git_filter_list_length(self._pointer)
 
+    def apply_to_buffer(self, data: bytes) -> bytes:
+        """
+        Apply a filter list to a data buffer.
+        Return the filtered contents.
+        """
+        buf = ffi.new('git_buf *')
+        err = C.git_filter_list_apply_to_buffer(buf, self._pointer, data, len(data))
+        check_error(err)
+        try:
+            return ffi.string(buf.ptr)
+        finally:
+            C.git_buf_dispose(buf)
+
+    def apply_to_file(self, repo: BaseRepository, path: str) -> bytes:
+        """
+        Apply a filter list to the contents of a file on disk.
+        Return the filtered contents.
+        """
+        buf = ffi.new('git_buf *')
+        c_path = to_bytes(path)
+        err = C.git_filter_list_apply_to_file(buf, self._pointer, repo._repo, c_path)
+        check_error(err)
+        try:
+            return ffi.string(buf.ptr)
+        finally:
+            C.git_buf_dispose(buf)
+
+    def apply_to_blob(self, blob: Blob) -> bytes:
+        """
+        Apply a filter list to a data buffer.
+        Return the filtered contents.
+        """
+        buf = ffi.new('git_buf *')
+
+        c_blob = ffi.new('git_blob **')
+        ffi.buffer(c_blob)[:] = blob._pointer[:]
+
+        err = C.git_filter_list_apply_to_blob(buf, self._pointer, c_blob[0])
+        check_error(err)
+        try:
+            return ffi.string(buf.ptr)
+        finally:
+            C.git_buf_dispose(buf)
+
     def __del__(self):
         C.git_filter_list_free(self._pointer)

test/test_filter.py

@@ -7,7 +7,7 @@
 
 import pygit2
 from pygit2 import Blob, Filter, FilterSource, Repository
-from pygit2.enums import BlobFilter
+from pygit2.enums import BlobFilter, FilterMode
 from pygit2.errors import Passthrough
 
 
@@ -146,22 +146,31 @@ def test_filterlist_none(testrepo: Repository) -> None:
     assert fl is None
 
 
-def test_filterlist_crlf(testrepo: Repository) -> None:
+def test_filterlist_apply_to_buffer_crlf_clean(testrepo: Repository) -> None:
     testrepo.config['core.autocrlf'] = True
 
-    fl = testrepo.load_filter_list('hello.txt')
+    fl = testrepo.load_filter_list('whatever.txt', mode=FilterMode.CLEAN)
     assert fl is not None
     assert len(fl) == 1
     assert 'crlf' in fl
-
+    assert 'bogus_filter_name' not in fl
     with pytest.raises(TypeError):
         1234 in fl  # type: ignore
 
+    filtered = fl.apply_to_buffer(b'hello\r\nworld\r\n')
+    assert filtered == b'hello\nworld\n'
 
-def test_filterlist_rot13(testrepo: Repository, rot13_filter: Filter) -> None:
-    fl = testrepo.load_filter_list('hello.txt')
+
+def test_filterlist_apply_to_buffer_crlf_smudge(testrepo: Repository) -> None:
+    testrepo.config['core.autocrlf'] = True
+
+    fl = testrepo.load_filter_list('whatever.txt', mode=FilterMode.SMUDGE)
     assert fl is not None
-    assert 'rot13' in fl
+    assert len(fl) == 1
+    assert 'crlf' in fl
+
+    filtered = fl.apply_to_buffer(b'hello\nworld\n')
+    assert filtered == b'hello\r\nworld\r\n'
 
 
 def test_filterlist_dangerous_unregister(testrepo: Repository) -> None:
@@ -182,3 +191,46 @@ def test_filterlist_dangerous_unregister(testrepo: Repository) -> None:
     del fl
     gc.collect()
     pygit2.filter_unregister('rot13')
+
+
+def test_filterlist_apply_to_file(
+    testrepo: Repository, rot13_filter: Filter
+) -> None:
+    fl = testrepo.load_filter_list('bye.txt')
+    assert fl is not None
+    assert len(fl) == 1
+    assert 'rot13' in fl
+
+    filtered = fl.apply_to_file(testrepo, 'bye.txt')
+    assert filtered == b'olr jbeyq\n'
+
+
+def test_filterlist_apply_to_blob(
+    testrepo: Repository, rot13_filter: Filter
+) -> None:
+    fl = testrepo.load_filter_list('whatever.txt')
+    assert fl is not None
+    assert len(fl) == 1
+    assert 'rot13' in fl
+
+    blob_oid = testrepo.create_blob(b'bye world\n')
+    blob = testrepo[blob_oid]
+    assert isinstance(blob, Blob)
+
+    filtered = fl.apply_to_blob(blob)
+    assert filtered == b'olr jbeyq\n'
+
+
+def test_filterlist_apply_to_buffer_multiple(
+    testrepo: Repository, rot13_filter: Filter
+) -> None:
+    testrepo.config['core.autocrlf'] = True
+
+    fl = testrepo.load_filter_list('whatever.txt')
+    assert fl is not None
+    assert len(fl) == 2
+    assert 'crlf' in fl
+    assert 'rot13' in fl
+
+    filtered = fl.apply_to_buffer(b'bye\r\nworld\r\n')
+    assert filtered == b'olr\njbeyq\n'