Copds 1939 cleaner

cacholote/clean.py

@@ -96,16 +96,16 @@ def __init__(self) -> None:
         self.logger = config.get().logger
         self.fs, self.dirname = utils.get_cache_files_fs_dirname()
 
-        urldir = self.fs.unstrip_protocol(self.dirname)
+        self.urldir = self.fs.unstrip_protocol(self.dirname)
 
         self.logger.info("getting disk usage")
         self.file_sizes: dict[str, int] = collections.defaultdict(int)
         for path, size in self.fs.du(self.dirname, total=False).items():
             # Group dirs
             urlpath = self.fs.unstrip_protocol(path)
-            basename, *_ = urlpath.replace(urldir, "", 1).strip("/").split("/")
+            basename, *_ = urlpath.replace(self.urldir, "", 1).strip("/").split("/")
             if basename:
-                self.file_sizes[posixpath.join(urldir, basename)] += size
+                self.file_sizes[posixpath.join(self.urldir, basename)] += size
 
         self.disk_usage = sum(self.file_sizes.values())
         self.log_disk_usage()
@@ -254,16 +254,16 @@ def delete_cache_files(
                 sa.select(database.CacheEntry).filter(*filters).order_by(*sorters)
             ):
                 files = _get_files_from_cache_entry(cache_entry)
-                if files:
+                if any(file.startswith(self.urldir) for file in files):
                     n_entries_to_delete += 1
                     session.delete(cache_entry)
 
-                for file, file_type in files.items():
-                    self.pop_file_size(file)
-                    if file_type == "application/vnd+zarr":
-                        dirs_to_delete.append(file)
-                    else:
-                        files_to_delete.append(file)
+                    for file, file_type in files.items():
+                        self.pop_file_size(file)
+                        if file_type == "application/vnd+zarr":
+                            dirs_to_delete.append(file)
+                        else:
+                            files_to_delete.append(file)
 
                 if self.stop_cleaning(maxsize):
                     break

tests/test_60_clean.py

@@ -349,3 +349,22 @@ def test_expire_cache_entries_created_at() -> None:
     assert clean.expire_cache_entries(after=toc) == 0
     assert clean.expire_cache_entries(before=toc) == 1
     assert clean.expire_cache_entries(after=tic) == 1
+
+
+def test_multiple(tmp_path: pathlib.Path) -> None:
+    oldpath = tmp_path / "old.txt"
+    oldpath.write_bytes(ONE_BYTE)
+    with config.set(cache_files_urlpath=str(tmp_path / "old")):
+        cached_oldpath = pathlib.Path(open_url(oldpath).path)
+    assert cached_oldpath.exists()
+
+    newpath = tmp_path / "new.txt"
+    newpath.write_bytes(ONE_BYTE)
+    with config.set(cache_files_urlpath=str(tmp_path / "new")):
+        cached_newpath = pathlib.Path(open_url(newpath).path)
+    assert cached_newpath.exists()
+
+    with config.set(cache_files_urlpath=str(tmp_path / "new")):
+        clean.clean_cache_files(0)
+    assert not cached_newpath.exists()
+    assert cached_oldpath.exists()