From db0367aac3182c8a2e39b8832bb88771811201be Mon Sep 17 00:00:00 2001
From: Maximilian Linhoff <maximilian.linhoff@cta-observatory.org>
Date: Mon, 2 Nov 2026 18:17:24 +0100
Subject: [PATCH] Escape % in manifest filenames as required by RFC

---
 src/bagit/__init__.py |  9 ++++++---
 test.py               | 27 ++++++++++++++++++++++++---
 2 files changed, 30 insertions(+), 6 deletions(-)

diff --git a/src/bagit/__init__.py b/src/bagit/__init__.py
index 2b4adce..32d0cea 100755
--- a/src/bagit/__init__.py
+++ b/src/bagit/__init__.py
@@ -1423,11 +1423,9 @@ def generate_manifest_lines(filename, algorithms=DEFAULT_CHECKSUMS):
             for hasher in hashers.values():
                 hasher.update(block)
 
-    decoded_filename = _decode_filename(filename)
-
     # We'll generate a list of results in roughly manifest format but prefixed with the algorithm:
     results = [
-        (alg, hasher.hexdigest(), decoded_filename, total_bytes)
+        (alg, hasher.hexdigest(), filename, total_bytes)
         for alg, hasher in hashers.items()
     ]
 
@@ -1435,6 +1433,9 @@ def generate_manifest_lines(filename, algorithms=DEFAULT_CHECKSUMS):
 
 
 def _encode_filename(s):
+    # needs to be first to not escape the % added by escaping \n and \r
+    s = s.replace("%", "%25")
+
     s = s.replace("\r", "%0D")
     s = s.replace("\n", "%0A")
     return s
@@ -1443,6 +1444,8 @@ def _encode_filename(s):
 def _decode_filename(s):
     s = re.sub(r"%0D", "\r", s, flags=re.IGNORECASE)
     s = re.sub(r"%0A", "\n", s, flags=re.IGNORECASE)
+    # needs to be last to not potentially introduce matches for \n / \r
+    s = s.replace("%25", "%")
     return s
 
 
diff --git a/test.py b/test.py
index a0e7f31..1d7dc85 100644
--- a/test.py
+++ b/test.py
@@ -827,10 +827,31 @@ def test_missing_tagmanifest_valid(self):
         os.remove(j(self.tmpdir, "tagmanifest-md5.txt"))
         self.assertTrue(bag.is_valid())
 
-    def test_carriage_return_manifest(self):
-        with open(j(self.tmpdir, "newline\r"), "w") as whatever:
-            whatever.write("ugh")
+    def test_special_characters_in_filename(self):
+        names = [
+            "carriage-return\r",
+            "newline\n",
+            "percent%",
+            "escapes%0A%0D%25", # has all the escape sequences already in the name
+        ]
+
+        for name in names:
+            with open(j(self.tmpdir, name), "w") as f:
+                f.write("ugh")
+
         bag = bagit.make_bag(self.tmpdir)
+        bag_names = slurp_text_file(next(bag.manifest_files()))
+
+        escaped_names = [
+            "carriage-return%0D",
+            "newline%0A",
+            "percent%25",
+            "escapes%250A%250D%2525",
+        ]
+
+        for escaped_name in escaped_names:
+            self.assertIn(f"data/{escaped_name}", bag_names)
+
         self.assertTrue(bag.is_valid())
 
     def test_payload_permissions(self):