From 17ffef9b5366e056e0a6725e7bc9b0423687a470 Mon Sep 17 00:00:00 2001
From: alighazi288 <51366992+alighazi288@users.noreply.github.com>
Date: Mon, 20 Jan 2025 14:53:29 -0500
Subject: [PATCH] Fix file state tracking with st = None after unlink/rmdir

---
 src/borg/archive.py                | 16 +++++++---------
 src/borg/testsuite/archive_test.py |  3 ++-
 2 files changed, 9 insertions(+), 10 deletions(-)

diff --git a/src/borg/archive.py b/src/borg/archive.py
index ea5e1e75e8..7461e8ada1 100644
--- a/src/borg/archive.py
+++ b/src/borg/archive.py
@@ -719,9 +719,9 @@ def extract_helper(self, item, path, hlm, *, dry_run=False):
                 # In this case, we *want* to extract twice, because there is no other way.
                 pass
 
-    def compare_and_extract_chunks(self, item, fs_path, *, pi=None, sparse=False):
+    def compare_and_extract_chunks(self, item, fs_path, *, st, pi=None):
         """Compare file chunks and patch if needed. Returns True if patching succeeded."""
-        if not os.path.exists(fs_path):
+        if st is None:
             return False
         try:
             # First pass: Build fs chunks list
@@ -755,12 +755,8 @@ def compare_and_extract_chunks(self, item, fs_path, *, pi=None, sparse=False):
                     else:
                         chunk_data = next(chunk_data_iter)
 
-                        with backup_io("seek"):
-                            fs_file.seek(fs_file.tell())
-
                         with backup_io("write"):
-                            data = b"\0" * len(chunk_data) if sparse and not chunk_data.strip(b"\0") else chunk_data
-                            fs_file.write(data)
+                            fs_file.write(chunk_data)
                     if pi:
                         pi.show(increase=len(chunk_data), info=[remove_surrogates(item.path)])
 
@@ -863,12 +859,14 @@ def same_item(item, st):
                 return  # done! we already have fully extracted this file in a previous run.
             elif stat.S_ISDIR(st.st_mode):
                 os.rmdir(path)
+                st = None
             else:
                 os.unlink(path)
+                st = None
         except UnicodeEncodeError:
             raise self.IncompatibleFilesystemEncodingError(path, sys.getfilesystemencoding()) from None
         except OSError:
-            pass
+            st = None
 
         def make_parent(path):
             parent_dir = os.path.dirname(path)
@@ -882,7 +880,7 @@ def make_parent(path):
             with self.extract_helper(item, path, hlm) as hardlink_set:
                 if hardlink_set:
                     return
-                if self.compare_and_extract_chunks(item, path, pi=pi, sparse=sparse):
+                if self.compare_and_extract_chunks(item, path, st=st, pi=pi):
                     return
 
                 with backup_io("open"):
diff --git a/src/borg/testsuite/archive_test.py b/src/borg/testsuite/archive_test.py
index 93bc9bddc6..3aee648ab4 100644
--- a/src/borg/testsuite/archive_test.py
+++ b/src/borg/testsuite/archive_test.py
@@ -545,7 +545,8 @@ def test_compare_and_extract_chunks(setup_extractor, name, item_data, fs_data, e
     with open(target_path, "wb") as f:
         f.write(fs_data)
 
-    result = extractor.compare_and_extract_chunks(item, target_path)
+    st = os.stat(target_path)
+    result = extractor.compare_and_extract_chunks(item, target_path, st=st)
     assert result
 
     fetched_chunks = get_fetched_chunks()