Handle query parameters when retrieving IIIF image information

tests/test_iiif_manifest_parser.py

@@ -106,23 +106,32 @@ def test_create_metadata_url(mock_worker):
     )
 
 
-def test_get_image(mock_worker, responses):
-    responses.add(
-        responses.GET, "http://imageurl/info.json", json={"@id": "http://imageurl/"}
-    )
+@pytest.mark.parametrize(
+    ("image_id", "expected_url"),
+    [
+        ("http://imageurl/", "http://imageurl/info.json"),
+        ("http://imageurl/a/b/c", "http://imageurl/a/b/c/info.json"),
+        # Path parameters should be left as they are
+        ("http://imageurl/a/b;k=v/c", "http://imageurl/a/b;k=v/c/info.json"),
+        # Query parameters are treated as part of the path to handle non-standard servers
+        ("http://imageurl/a/b/c?k=v", "http://imageurl/a/b/c?k=v/info.json"),
+    ],
+)
+def test_get_image(mock_worker, responses, image_id, expected_url):
+    responses.add(responses.GET, expected_url, json={"@id": "http://imageurl/"})
     mock_worker.api_client.add_response(
         "CreateIIIFInformation",
         body={"@id": "http://imageurl/"},
-        response={"id": "imageid", "url": "http://imageurl", "status": "checked"},
+        response={"id": "imageid", "url": image_id, "status": "checked"},
     )
 
     parser = ManifestParser(mock_worker, BytesIO())
-    assert parser.get_image("http://imageurl") == {
+    assert parser.get_image(image_id) == {
         "id": "imageid",
-        "url": "http://imageurl",
+        "url": image_id,
         "status": "checked",
     }
-    responses.assert_call_count("http://imageurl/info.json", 1)
+    responses.assert_call_count(expected_url, 1)
 
 
 def test_get_image_check_error(mock_worker, responses):

worker_file_import/iiif.py

@@ -255,10 +255,15 @@ class ManifestParser(IIIFParser):
         Get or create an Arkindex image from an IIIF image URL.
         Returns None if the image could not be created.
         """
+        parsed_url = urlparse(url)
 
         if not url.endswith("/"):
             url += "/"
-        info_url = urljoin(url, "info.json")
+
+        # When the image URL has a query parameter, concatenate strings instead of using `urljoin`
+        # because some servers do not do any URL rewriting and put the image identifier in a query parameter
+        info_url = url + "info.json" if parsed_url.query else urljoin(url, "info.json")
+
         try:
             payload = self.fetch_image_information(info_url)
         except Exception as e: