Merge pull request #1853 from jcole-crowdstrike/fix-branch-name-regex-to-handle-nbsp

Byron · web-flow · commit d40e7d03c69b · 2024-03-11T14:13:45.000+01:00
Update regex pattern to handle unicode whitespaces.
diff --git a/git/remote.py b/git/remote.py
@@ -325,7 +325,7 @@ class FetchInfo(IterableObj):
         ERROR,
     ) = [1 << x for x in range(8)]
 
-    _re_fetch_result = re.compile(r"^\s*(.) (\[[\w\s\.$@]+\]|[\w\.$@]+)\s+(.+) -> ([^\s]+)(    \(.*\)?$)?")
+    _re_fetch_result = re.compile(r"^ *(.) (\[[\w \.$@]+\]|[\w\.$@]+) +(.+) -> ([^ ]+)(    \(.*\)?$)?")
 
     _flag_map: Dict[flagKeyLiteral, int] = {
         "!": ERROR,
@@ -895,7 +895,7 @@ def _get_fetch_info_from_stderr(
             None,
             progress_handler,
             finalizer=None,
-            decode_streams=False,
+            decode_streams=True,
             kill_after_timeout=kill_after_timeout,
         )
 
@@ -1072,7 +1072,7 @@ def fetch(
             Git.check_unsafe_options(options=list(kwargs.keys()), unsafe_options=self.unsafe_git_fetch_options)
 
         proc = self.repo.git.fetch(
-            "--", self, *args, as_process=True, with_stdout=False, universal_newlines=True, v=verbose, **kwargs
+            "--", self, *args, as_process=True, with_stdout=False, universal_newlines=False, v=verbose, **kwargs
         )
         res = self._get_fetch_info_from_stderr(proc, progress, kill_after_timeout=kill_after_timeout)
         if hasattr(self.repo.odb, "update_cache"):
@@ -1126,7 +1126,7 @@ def pull(
             Git.check_unsafe_options(options=list(kwargs.keys()), unsafe_options=self.unsafe_git_pull_options)
 
         proc = self.repo.git.pull(
-            "--", self, refspec, with_stdout=False, as_process=True, universal_newlines=True, v=True, **kwargs
+            "--", self, refspec, with_stdout=False, as_process=True, universal_newlines=False, v=True, **kwargs
         )
         res = self._get_fetch_info_from_stderr(proc, progress, kill_after_timeout=kill_after_timeout)
         if hasattr(self.repo.odb, "update_cache"):
diff --git a/git/util.py b/git/util.py
@@ -611,20 +611,6 @@ def _parse_progress_line(self, line: AnyStr) -> None:
             self.error_lines.append(self._cur_line)
             return
 
-        # Find escape characters and cut them away - regex will not work with
-        # them as they are non-ASCII. As git might expect a tty, it will send them.
-        last_valid_index = None
-        for i, c in enumerate(reversed(line_str)):
-            if ord(c) < 32:
-                # its a slice index
-                last_valid_index = -i - 1
-            # END character was non-ASCII
-        # END for each character in line
-        if last_valid_index is not None:
-            line_str = line_str[:last_valid_index]
-        # END cut away invalid part
-        line_str = line_str.rstrip()
-
         cur_count, max_count = None, None
         match = self.re_op_relative.match(line_str)
         if match is None:
diff --git a/test/test_remote.py b/test/test_remote.py
@@ -1002,6 +1002,22 @@ def test_push_unsafe_options_allowed(self, rw_repo):
                 assert tmp_file.exists()
                 tmp_file.unlink()
 
+    @with_rw_and_rw_remote_repo("0.1.6")
+    def test_fetch_unsafe_branch_name(self, rw_repo, remote_repo):
+        # Create branch with a name containing a NBSP
+        bad_branch_name = f"branch_with_{chr(160)}_nbsp"
+        Head.create(remote_repo, bad_branch_name)
+
+        # Fetch and get branches
+        remote = rw_repo.remote("origin")
+        branches = remote.fetch()
+
+        # Test for truncated branch name in branches
+        assert f"origin/{bad_branch_name}" in [b.name for b in branches]
+
+        # Cleanup branch
+        Head.delete(remote_repo, bad_branch_name)
+
 
 class TestTimeouts(TestBase):
     @with_rw_repo("HEAD", bare=False)
