use_database=True should only consider files in cache_files_urlpath (#…

…134)

* use_database=True should only consider files in cache_files_urlpath

* improve test

cacholote/clean.py

@@ -159,8 +159,9 @@ def known_files(self) -> dict[str, int]:
         known_files: dict[str, int] = {}
         with config.get().instantiated_sessionmaker() as session:
             for cache_entry in session.scalars(sa.select(database.CacheEntry)):
+                files = _get_files_from_cache_entry(cache_entry, key="file:size")
                 known_files.update(
-                    _get_files_from_cache_entry(cache_entry, key="file:size")
+                    {k: v for k, v in files.items() if k.startswith(self.urldir)}
                 )
         return known_files
 

tests/test_60_clean.py

@@ -388,3 +388,26 @@ def test_multiple(tmp_path: pathlib.Path) -> None:
         clean.clean_cache_files(0)
     assert not cached_newpath.exists()
     assert cached_oldpath.exists()
+
+
+@pytest.mark.parametrize("use_database", [True, False])
+def test_clean_multiple_urlpaths(tmp_path: pathlib.Path, use_database: bool) -> None:
+    # Create files
+    tmpfile1 = tmp_path / "file1.txt"
+    fsspec.filesystem("file").pipe_file(tmpfile1, ONE_BYTE)
+    tmpfile2 = tmp_path / "file2.txt"
+    fsspec.filesystem("file").pipe_file(tmpfile2, ONE_BYTE)
+
+    # Copy to cache
+    path1 = tmp_path / "cache_files" / "folder1"
+    with config.set(cache_files_urlpath=str(path1 / "today")):
+        cached_file1 = pathlib.Path(open_url(tmpfile1).path)
+    path2 = tmp_path / "cache_files" / "folder2"
+    with config.set(cache_files_urlpath=str(path2 / "today")):
+        cached_file2 = pathlib.Path(open_url(tmpfile2).path)
+
+    # Clean
+    with config.set(cache_files_urlpath=str(path1)):
+        clean.clean_cache_files(maxsize=0, use_database=use_database, depth=2)
+    assert not cached_file1.exists()
+    assert cached_file2.exists()