fix: Use filesystem timestamps for entity sync instead of database operation time (#138) (#369)

Signed-off-by: Claude <noreply@anthropic.com>
Signed-off-by: phernandez <paul@basicmachines.co>

Author: phernandez

src/basic_memory/sync/sync_service.py

@@ -547,6 +547,11 @@ async def sync_markdown_file(self, path: str, new: bool = True) -> Tuple[Optiona
         file_content = await self._read_file_async(file_path)
         file_contains_frontmatter = has_frontmatter(file_content)
 
+        # Get file timestamps for tracking modification times
+        file_stats = self.file_service.file_stats(path)
+        created = datetime.fromtimestamp(file_stats.st_ctime).astimezone()
+        modified = datetime.fromtimestamp(file_stats.st_mtime).astimezone()
+
         # entity markdown will always contain front matter, so it can be used up create/update the entity
         entity_markdown = await self.entity_parser.parse_file(path)
 
@@ -585,8 +590,11 @@ async def sync_markdown_file(self, path: str, new: bool = True) -> Tuple[Optiona
         # after relation processing is complete
         final_checksum = await self._compute_checksum_async(path)
 
-        # set checksum
-        await self.entity_repository.update(entity.id, {"checksum": final_checksum})
+        # Update checksum and timestamps from file system
+        # This ensures temporal ordering in search and recent activity uses actual file modification times
+        await self.entity_repository.update(
+            entity.id, {"checksum": final_checksum, "created_at": created, "updated_at": modified}
+        )
 
         logger.debug(
             f"Markdown sync completed: path={path}, entity_id={entity.id}, "
@@ -659,13 +667,18 @@ async def sync_regular_file(self, path: str, new: bool = True) -> Tuple[Optional
                     # Re-raise if it's a different integrity error
                     raise
         else:
+            # Get file timestamps for updating modification time
+            file_stats = self.file_service.file_stats(path)
+            modified = datetime.fromtimestamp(file_stats.st_mtime).astimezone()
+
             entity = await self.entity_repository.get_by_file_path(path)
             if entity is None:  # pragma: no cover
                 logger.error(f"Entity not found for existing file, path={path}")
                 raise ValueError(f"Entity not found for existing file: {path}")
 
+            # Update checksum and modification time from file system
             updated = await self.entity_repository.update(
-                entity.id, {"file_path": path, "checksum": checksum}
+                entity.id, {"file_path": path, "checksum": checksum, "updated_at": modified}
             )
 
             if updated is None:  # pragma: no cover

tests/markdown/test_entity_parser_error_handling.py

@@ -1,7 +1,6 @@
 """Tests for entity parser error handling (issues #184 and #185)."""
 
 import pytest
-from pathlib import Path
 from textwrap import dedent
 
 from basic_memory.markdown.entity_parser import EntityParser
@@ -213,4 +212,4 @@ async def test_parse_valid_file_still_works(tmp_path):
     assert result is not None
     assert result.frontmatter.title == "Valid File"
     assert result.frontmatter.type == "knowledge"
-    assert result.frontmatter.tags == ["test", "valid"]
+    assert result.frontmatter.tags == ["test", "valid"]

tests/sync/test_sync_service.py

@@ -645,6 +645,90 @@ async def test_sync_preserves_timestamps(
     assert abs(entity_updated_epoch - file_stats.st_mtime) < tolerance  # Allow tolerance difference
 
 
+@pytest.mark.asyncio
+async def test_sync_updates_timestamps_on_file_modification(
+    sync_service: SyncService,
+    project_config: ProjectConfig,
+    entity_service: EntityService,
+):
+    """Test that sync updates entity timestamps when files are modified.
+
+    This test specifically validates that when an existing file is modified and re-synced,
+    the updated_at timestamp in the database reflects the file's actual modification time,
+    not the database operation time. This is critical for accurate temporal ordering in
+    search and recent_activity queries.
+    """
+    import time
+
+    project_dir = project_config.home
+
+    # Create initial file
+    initial_content = """
+---
+type: knowledge
+---
+# Test File
+Initial content for timestamp test
+"""
+    file_path = project_dir / "timestamp_test.md"
+    await create_test_file(file_path, initial_content)
+
+    # Initial sync
+    await sync_service.sync(project_config.home)
+
+    # Get initial entity and timestamps
+    entity_before = await entity_service.get_by_permalink("timestamp-test")
+    initial_updated_at = entity_before.updated_at
+
+    # Wait a bit to ensure filesystem timestamp changes
+    time.sleep(0.1)
+
+    # Modify the file content
+    modified_content = """
+---
+type: knowledge
+---
+# Test File
+Modified content for timestamp test
+
+## Observations
+- [test] This was modified
+"""
+    file_path.write_text(modified_content)
+
+    # Wait to ensure mtime is different
+    time.sleep(0.1)
+
+    # Get the file's modification time after our changes
+    file_stats_after_modification = file_path.stat()
+
+    # Re-sync the modified file
+    await sync_service.sync(project_config.home)
+
+    # Get entity after re-sync
+    entity_after = await entity_service.get_by_permalink("timestamp-test")
+
+    # Verify that updated_at changed
+    assert entity_after.updated_at != initial_updated_at, (
+        "updated_at should change when file is modified"
+    )
+
+    # Verify that updated_at matches the file's modification time, not db operation time
+    entity_updated_epoch = entity_after.updated_at.timestamp()
+    file_mtime = file_stats_after_modification.st_mtime
+
+    # Allow 2s difference on Windows due to filesystem timing precision
+    tolerance = 2 if os.name == "nt" else 1
+    assert abs(entity_updated_epoch - file_mtime) < tolerance, (
+        f"Entity updated_at ({entity_after.updated_at}) should match file mtime "
+        f"({datetime.fromtimestamp(file_mtime)}) within {tolerance}s tolerance"
+    )
+
+    # Verify the content was actually updated
+    assert len(entity_after.observations) == 1
+    assert entity_after.observations[0].content == "This was modified"
+
+
 @pytest.mark.asyncio
 async def test_file_move_updates_search_index(
     sync_service: SyncService,