🔥 (operations) Remove cleanup_archive

Author: simonwoerpel

docs/reference/operations.md

@@ -88,8 +88,6 @@ Operations for storing data.
 
 ::: memorious.operations.store.lakehouse
 
-::: memorious.operations.store.cleanup_archive
-
 ---
 
 ## Debug

memorious/operations/store.py

@@ -378,36 +378,6 @@ def lakehouse(context: Context, data: dict[str, Any]) -> None:
         context.emit(data=data)
 
 
-@register("cleanup_archive")
-def cleanup_archive(context: Context, data: dict[str, Any]) -> None:
-    """Remove a blob from the archive.
-
-    Deletes a file from the archive after processing is complete.
-    Useful for cleaning up temporary files.
-
-    Args:
-        context: The crawler context.
-        data: Must contain content_hash of file to delete.
-
-    Example:
-        ```yaml
-        pipeline:
-          cleanup:
-            method: cleanup_archive
-        ```
-    """
-    content_hash = data.get("content_hash")
-    if content_hash is None:
-        context.emit_warning("No content hash in data.")
-        return
-    file_info = context.archive.get(content_hash)
-    if file_info:
-        try:
-            context.archive.delete(file_info)
-        except NotImplementedError:
-            context.log.warning("File deletion not supported by storage backend")
-
-
 @register("store")
 def store(context: Context, data: dict[str, Any]) -> None:
     """Store with configurable backend and incremental marking.

tests/test_operations.py

@@ -8,7 +8,6 @@
 from memorious.operations.parse import parse
 from memorious.operations.store import (
     _compute_file_path,
-    cleanup_archive,
     directory,
     lakehouse,
 )
@@ -284,16 +283,6 @@ def test_directory_same_content(context, mocker, httpbin_url):
     assert len(json_files) == 1, "Should only have one content file"
 
 
-def test_cleanup_archive(context, httpbin_url):
-    url = f"{httpbin_url}/user-agent"
-    result = context.http.get(url, headers={"User-Agent": "Memorious Test"})
-    data = result.serialize()
-    assert context.archive.get(data["content_hash"]) is not None
-    # cleanup_archive may not actually delete due to NotImplementedError in ftm_lakehouse
-    cleanup_archive(context, data)
-    # NOTE: File may still exist if storage backend doesn't support deletion
-
-
 def test_lakehouse_default(context, mocker, httpbin_url):
     """Test lakehouse store with default archive."""
     from ftm_lakehouse import get_lakehouse