Fix wikimedia_commons.py

WordPress · Apr 28, 2023 · 0c7416f · 0c7416f
1 parent e9e5760
commit 0c7416f
Show file tree

Hide file tree

Showing 2 changed files with 6 additions and 10 deletions.
diff --git a/catalog/dags/providers/provider_api_scripts/wikimedia_commons.py b/catalog/dags/providers/provider_api_scripts/wikimedia_commons.py
@@ -311,8 +311,7 @@ def get_record_data(self, record):
         if license_info.url is None:
             return None
 
-        media_url = media_info.get("url")
-        if media_url is None:
+        if not (url := media_info.get("url")):
             return None
 
         creator, creator_url = self.extract_creator_info(media_info)
@@ -322,7 +321,7 @@ def get_record_data(self, record):
         meta_data = self.create_meta_data_dict(record)
 
         record_data = {
-            "media_url": media_url,
+            "url": url,
             "foreign_landing_url": media_info.get("descriptionshorturl"),
             "foreign_identifier": foreign_id,
             "license_info": license_info,
@@ -345,7 +344,6 @@ def get_record_data(self, record):
     @staticmethod
     def get_image_record_data(record_data, media_info):
         """Extend record_data with image-specific fields."""
-        record_data["url"] = record_data.pop("media_url")
         if record_data["filetype"] == "svg":
             record_data["category"] = "illustration"
 
@@ -357,8 +355,6 @@ def get_image_record_data(record_data, media_info):
 
     def get_audio_record_data(self, record_data, media_info):
         """Extend record_data with audio-specific fields."""
-        record_data["url"] = record_data.pop("media_url")
-
         duration = int(float(media_info.get("duration", 0)) * 1000)
         record_data["duration"] = duration
         record_data["category"] = self.extract_audio_category(record_data)

diff --git a/catalog/tests/dags/providers/provider_api_scripts/test_wikimedia_commons.py b/catalog/tests/dags/providers/provider_api_scripts/test_wikimedia_commons.py
@@ -421,7 +421,7 @@ def test_create_meta_data_tallies_zero_global_usage_count(wmc):
 
 def test_get_audio_record_data_parses_ogg_streams(wmc):
     file_metadata = _get_resource_json("audio_filedata_ogg.json")
-    original_data = {"media_url": "myurl.com", "meta_data": {}}
+    original_data = {"url": "myurl.com", "meta_data": {}}
     actual_parsed_data = wmc.get_audio_record_data(original_data, file_metadata)
 
     expected_parsed_data = {
@@ -435,7 +435,7 @@ def test_get_audio_record_data_parses_ogg_streams(wmc):
 
 def test_get_audio_record_data_parses_wav_audio_data(wmc):
     file_metadata = _get_resource_json("audio_filedata_wav.json")
-    original_data = {"media_url": "myurl.com", "meta_data": {}}
+    original_data = {"url": "myurl.com", "meta_data": {}}
     actual_parsed_data = wmc.get_audio_record_data(original_data, file_metadata)
 
     expected_parsed_data = {
@@ -449,7 +449,7 @@ def test_get_audio_record_data_parses_wav_audio_data(wmc):
 
 def test_get_audio_record_data_parses_wav_audio_data_missing_streams(wmc):
     file_metadata = _get_resource_json("audio_filedata_wav.json")
-    original_data = {"media_url": "myurl.com", "meta_data": {}}
+    original_data = {"url": "myurl.com", "meta_data": {}}
     # Remove any actual audio metadata
     file_metadata["metadata"] = (
         file_metadata["metadata"][:5] + file_metadata["metadata"][6:]
@@ -465,7 +465,7 @@ def test_get_audio_record_data_parses_wav_audio_data_missing_streams(wmc):
 
 def test_get_audio_record_data_parses_wav_invalid_bit_rate(wmc):
     file_metadata = _get_resource_json("audio_filedata_wav.json")
-    original_data = {"media_url": "myurl.com", "meta_data": {}}
+    original_data = {"url": "myurl.com", "meta_data": {}}
     # Set the bit rate higher than the int max
     file_metadata["metadata"][5]["value"][3]["value"][0]["value"][3][
         "value"