fix(stats): split Git records at line feeds only

With `core.quotePath=false`, Git emits U+0085, U+2028 and U+2029 directly
in filenames. `Commit.stats` and `Stats._list_from_string` used
`str.splitlines()`, treating those filename characters as record endings
and raising `ValueError` instead of returning statistics.

Split only at the line feeds used by Git's protocol. Keep empty commits
and the generated text's trailing newline working by ignoring empty
records. No filename encoding or public return types change.

The regression uses real Git output for all three Unicode separators
and covers root, later and empty commits. It failed before this change.
The complete commit/statistics modules passed: 44 tests and 8 subtests.
Ruff lint/format, mypy for both changed source files, and `git diff
--check` also passed.

Author: BHARTIYAYASH

git/objects/commit.py

@@ -405,6 +405,8 @@ def stats(self) -> Stats:
         def process_lines(lines: List[str]) -> str:
             text = ""
             for file_info, line in zip(lines, lines[len(lines) // 2 :]):
+                if not line:
+                    continue
                 change_type = file_info.split("\t")[0][-1]
                 (insertions, deletions, filename) = line.split("\t")
                 text += "%s\t%s\t%s\t%s\n" % (change_type, insertions, deletions, filename)
@@ -413,12 +415,12 @@ def process_lines(lines: List[str]) -> str:
         if not self.parents:
             lines = self.repo.git.diff_tree(
                 self.hexsha, "--", numstat=True, no_renames=True, root=True, raw=True
-            ).splitlines()[1:]
+            ).split("\n")[1:]
             text = process_lines(lines)
         else:
             lines = self.repo.git.diff(
                 self.parents[0].hexsha, self.hexsha, "--", numstat=True, no_renames=True, raw=True
-            ).splitlines()
+            ).split("\n")
             text = process_lines(lines)
         return Stats._list_from_string(self.repo, text)
 

git/util.py

@@ -1111,7 +1111,9 @@ def _list_from_string(cls, repo: "Repo", text: str) -> "Stats":
             "total": {"insertions": 0, "deletions": 0, "lines": 0, "files": 0},
             "files": {},
         }
-        for line in text.splitlines():
+        for line in text.split("\n"):
+            if not line:
+                continue
             (change_type, raw_insertions, raw_deletions, filename) = line.split("\t")
             insertions = raw_insertions != "-" and int(raw_insertions) or 0
             deletions = raw_deletions != "-" and int(raw_deletions) or 0

test/test_commit.py

@@ -200,6 +200,24 @@ def test_stats_preserve_filename_spaces(self, rw_dir):
         }
         assert removed_stats.total == {"insertions": 0, "deletions": 30, "lines": 30, "files": 5}
 
+    @with_rw_directory
+    def test_stats_preserve_unicode_line_separators(self, rw_dir):
+        with Repo.init(rw_dir) as repo:
+            repo.git.config("core.quotePath", "false")
+            actor = Actor("Test", "test@example.com")
+            paths = [f"before{separator}after.txt" for separator in ("\x85", "\u2028", "\u2029")]
+            for multiplier, change_type in ((1, "A"), (2, "M")):
+                for path in paths:
+                    data = b"line\n" * multiplier
+                    blob = repo.odb.store(IStream("blob", len(data), BytesIO(data)))
+                    repo.index.add([BaseIndexEntry((0o100644, blob.binsha, 0, path))])
+                commit = repo.index.commit("Update files", author=actor, committer=actor)
+                assert commit.stats.total == {"insertions": 3, "deletions": 0, "lines": 3, "files": 3}
+                assert commit.stats.files == {
+                    path: {"insertions": 1, "deletions": 0, "lines": 1, "change_type": change_type} for path in paths
+                }
+            assert repo.index.commit("No changes", author=actor, committer=actor).stats.files == {}
+
     def test_renames(self):
         commit = self.rorepo.commit("185d847ec7647fd2642a82d9205fb3d07ea71715")
         files = commit.stats.files