added override to assign media types by media use

WorkbenchConfig.py

@@ -286,6 +286,7 @@ def get_default_config(self):
             "csv_value_templates_rand_length": 5,
             "allow_csv_value_templates_if_field_empty": [],
             "remind_user_to_run_check": False,
+            "media_type_by_media_use": False,
         }
 
     # Tests validity and existence of configuration file path.

i7Import/i7ImportUtilities.py

@@ -56,6 +56,7 @@ def __init__(self, config_location):
         "start": 0,
         "rows": 100000,
         "secure_ssl_only": True,
+        "pids": False,
     }
 
     def get_config(self):
@@ -181,6 +182,12 @@ def get_default_metadata_solr_request(self):
                 fedora_collections.append(f'{fedora_prefix}"{collection}"')
             fq_string = "&fq=" + " or ".join(fedora_collections)
             query = f"{query}{fq_string}"
+        if self.config["pids"]:
+            pids_to_use = []
+            for candidate in self.config["pids"]:
+                pids_to_use.append(f"PID:{candidate}")
+            fq_string = "&fq=" + " or ".join(pids_to_use)
+            query = f"{query}{fq_string}"
 
         # Get the populated CSV from Solr, with the object namespace and field list filters applied.
         return query

tests/assets/set_media_type_test/multi_types_config.yml

@@ -16,3 +16,6 @@ media_types_override:
  - sometextmedia: ['txt']
 
 secure_ssl_only: false
+
+media_type_by_media_use:
+  - https://projects.iq.harvard.edu/fits: fits_technical_metadata

workbench_utils.py

@@ -79,6 +79,14 @@ def set_media_type(config, filepath, file_fieldname, csv_row):
     """
     if "media_type" in config:
         return config["media_type"]
+    if config["media_type_by_media_use"] and len(config["media_type_by_media_use"]) > 0:
+        additional_files = get_additional_files_config(config)
+        media_url = additional_files.get(file_fieldname)
+        if file_fieldname in additional_files:
+            for entry in config["media_type_by_media_use"]:
+                for key, value in entry.items():
+                    if key == media_url:
+                        return value
 
     # Determine if the incomtimg filepath matches a registered eEmbed media type.
     oembed_media_type = get_oembed_url_media_type(config, filepath)