refactorws PR to obey standard code approach

src/auto_archiver/core/orchestrator.py

@@ -274,14 +274,6 @@ Here's how that would look: \n\nsteps:\n  extractors:\n  - [your_extractor_name_
             default=False,
         )
 
-        parser.add_argument(
-            "--metadata",
-            dest="requested_metadata",
-            help="An array of specific metadata fields to select from the collected content.",
-            default=[],
-            nargs="?",
-        )
-
     def add_individual_module_args(
         self, modules: list[LazyBaseModule] = None, parser: argparse.ArgumentParser = None
     ) -> None:

src/auto_archiver/modules/metadata_enricher/__manifest__.py

@@ -3,6 +3,13 @@
     "type": ["enricher"],
     "requires_setup": True,
     "dependencies": {"python": ["loguru"], "bin": ["exiftool"]},
+    "configs": {
+        "look_for_keys": {
+            "default": [],
+            "help": "list of lowercased metadata keys that will be included in the enriched metadata. Special keys: 'author', 'datetimes', 'location' to include related metadata fields. The default empty list `[]` means all metadata will be included.",
+            "type": "list",
+        },
+    },
     "description": """
     Extracts metadata information from files using ExifTool.
 

src/auto_archiver/modules/metadata_enricher/metadata_enricher.py

@@ -11,15 +11,14 @@ class MetadataEnricher(Enricher):
     Extracts metadata information from files using exiftool.
     """
 
-    def enrich(self, to_enrich: Metadata, md_grocery_list=["author", "datetimes", "location"]) -> None:
+    def enrich(self, to_enrich: Metadata) -> None:
         logger.debug("Extracting EXIF metadata")
 
         for i, m in enumerate(to_enrich.media):
             if len(md := self.get_metadata(m.filename)):
-                # feature flag has this currently turned on
+                if self.look_for_keys != []:
-                specified_md = self.select_metadata(md, md_grocery_list)
+                    md = self.select_metadata(md, self.look_for_keys)
-                to_enrich.media[i].set("metadata", specified_md)
+                to_enrich.media[i].set("metadata", md)
-                # to_enrich.media[i].set("metadata", md)
 
     def get_metadata(self, filename: str) -> dict:
         try:
@@ -38,7 +37,7 @@ class MetadataEnricher(Enricher):
             logger.error(f"Error occurred: {e}: {traceback.format_exc()}")
         return {}
 
-    def select_metadata(self, all_md, md_grocery_list):
+    def select_metadata(self, all_md, requested_metadata_keys):
         """
         coordinates the selection of metadata from the general exiftool output to the user-specified grocery list
         """
@@ -51,19 +50,19 @@ class MetadataEnricher(Enricher):
         for md_key in all_md.keys():
             md_key_lower = md_key.lower()
             # checking for special baskets within the grocery list of requested metadata
-            if ("author" in md_grocery_list) and any(
+            if ("author" in requested_metadata_keys) and any(
                 term in md_key_lower and len(all_md[md_key]) for term in author_key_terms
             ):
                 specified_md[md_key] = all_md[md_key]
-            if ("datetime" in md_grocery_list) and any(
+            if ("datetime" in requested_metadata_keys) and any(
                 term in md_key_lower and len(all_md[md_key]) for term in datetime_key_terms
             ):
                 specified_md[md_key] = all_md[md_key]
-            if ("location" in md_grocery_list) and any(
+            if ("location" in requested_metadata_keys) and any(
                 term in md_key_lower and len(all_md[md_key]) for term in location_key_terms
             ):
                 specified_md[md_key] = all_md[md_key]
             # if the metadata value is requested directly
-            if md_key_lower in md_grocery_list or md_key in md_grocery_list and len(all_md[md_key]):
+            if md_key_lower in requested_metadata_keys or md_key in requested_metadata_keys and len(all_md[md_key]):
                 specified_md[md_key] = all_md[md_key]
         return specified_md

tests/enrichers/test_metadata_enricher.py

@@ -49,7 +49,7 @@ def test_enrich_sets_metadata(enricher, mocker):
     metadata.media = [media1, media2]
     enricher.get_metadata = lambda f: {"key": "value"} if f == "img1.jpg" else {}
 
-    enricher.enrich(metadata, ["key"])
+    enricher.enrich(metadata)
 
     media1.set.assert_called_once_with("metadata", {"key": "value"})
     media2.set.assert_not_called()
@@ -62,6 +62,7 @@ def test_enrich_no_metadata_selection(enricher, mocker):
     metadata = mocker.Mock()
     metadata.media = [media1, media2]
     enricher.get_metadata = lambda f: {"key": "value"} if f == "img1.jpg" else {}
+    enricher.look_for_keys = ["no-key"]
     enricher.enrich(metadata)
     media1.set.assert_called_once_with("metadata", {})
     media2.set.assert_not_called()
@@ -106,7 +107,8 @@ def test_metadata_pickle_megapixel(enricher, unpickle, mocker):
     mock_run.return_value = unpickle("metadata_enricher_exif.pickle")
     metadata = unpickle("metadata_enricher_ytshort_input.pickle")
 
-    enricher.enrich(metadata, ["megapixels"])
+    enricher.look_for_keys = ["megapixels"]
+    enricher.enrich(metadata)
     actual_media = metadata.media
 
     assert actual_media[0].properties.get("metadata") == {"Megapixels": "0.922"}
@@ -116,9 +118,9 @@ def test_metadata_specify_datetime_and_metapixels(enricher, unpickle, mocker):
     mock_run = mocker.patch("subprocess.run")
     mock_run.return_value = unpickle("metadata_enricher_exif.pickle")
     metadata = unpickle("metadata_enricher_ytshort_input.pickle")
-    # expected_md = {"Metapixels":"0.922", "File Inode Change Date/Time":"2025:02:18 19:42:50+00:00"}
 
-    enricher.enrich(metadata, ["datetime", "megapixels"])
+    enricher.look_for_keys = ["datetime", "megapixels"]
+    enricher.enrich(metadata)
     actual_media = metadata.media
 
     assert actual_media[0].properties.get("metadata") == {