Add trailing slash to Jamendo thumbnail URLs (#1038)

* Add trailing slash to Jamendo thumbnail URLs

* Apply URL fix to all thumbnail URLs

openverse_catalog/dags/providers/provider_api_scripts/jamendo.py

@@ -90,7 +90,24 @@ def _remove_trackid(self, thumbnail_url: str | None) -> str | None:
         """
         if thumbnail_url is None:
             return None
-        return self._remove_param_from_url(thumbnail_url, "trackid")
+        return self._remove_param_from_url(
+            self._add_trailing_slash(thumbnail_url), "trackid"
+        )
+
+    @staticmethod
+    def _add_trailing_slash(url: str | None) -> str | None:
+        """
+        Jamendo image URLs are missing a trailing slash, which when viewed normally in
+        the browser get redirected to the correct URL. Example:
+        - https://usercontent.jamendo.com?type=album&id=100007&width=300 (before)
+        - https://usercontent.jamendo.com/?type=album&id=100007&width=300 (after)
+
+        Due to the way photon processes thumbnails, we need to add this trailing slash
+        to the url prior to the query params if it does not have one.
+        """
+        if url and "/?" not in url:
+            url = url.replace("?", "/?")
+        return url
 
     def _get_audio_url(self, data):
         """
@@ -178,7 +195,7 @@ def get_record_data(self, data):
         if duration:
             duration = int(duration) * 1000
         title = data.get("name")
-        thumbnail = data.get("image")
+        thumbnail = self._add_trailing_slash(data.get("image"))
         genres = data.get("musicinfo", {}).get("tags", {}).get("genres")
         creator, creator_url = self._get_creator_data(data)
         metadata = self._get_metadata(data)

tests/dags/providers/provider_api_scripts/test_jamendo.py

@@ -92,9 +92,9 @@ def test_get_record_data():
         "raw_tags": ["instrumental", "speed_medium"],
         "set_foreign_id": "119",
         "set_position": 6,
-        "set_thumbnail": "https://usercontent.jamendo.com?type=album&id=119&width=200",
+        "set_thumbnail": "https://usercontent.jamendo.com/?type=album&id=119&width=200",
         "set_url": "https://www.jamendo.com/album/119/opera-i",
-        "thumbnail_url": "https://usercontent.jamendo.com?type=album&id=119&width=200&trackid=732",
+        "thumbnail_url": "https://usercontent.jamendo.com/?type=album&id=119&width=200&trackid=732",
         "title": "Thoughtful",
     }
     assert actual == expected
@@ -152,3 +152,23 @@ def test_get_tags():
     expected_tags = ["vocal", "male", "speed_medium", "engage"]
     actual_tags = jamendo._get_tags(item_data)
     assert expected_tags == actual_tags
+
+
+@pytest.mark.parametrize(
+    "url, expected",
+    [
+        (None, None),
+        ("", ""),
+        (
+            "https://usercontent.jamendo.com?type=album&id=100007&width=300",
+            "https://usercontent.jamendo.com/?type=album&id=100007&width=300",
+        ),
+        (
+            "https://usercontent.jamendo.com/some-other-page/subpage?type=album&id=100007&width=300",
+            "https://usercontent.jamendo.com/some-other-page/subpage/?type=album&id=100007&width=300",
+        ),
+    ],
+)
+def test_add_trailing_slash(url, expected):
+    actual = jamendo._add_trailing_slash(url)
+    assert actual == expected