fix(agent): recover usable trip photos and isolate malformed provider metadata

apps/agent/src/trip_images.py

@@ -14,9 +14,16 @@
 def fetch_card(title: str, client: httpx.Client) -> dict:
     summary = client.get("https://en.wikipedia.org/api/rest_v1/page/summary/" + quote(title, safe=""))
     summary.raise_for_status()
-    data = summary.json()
-    image = data.get("thumbnail", {}).get("source", "")
-    original = data.get("originalimage", {}).get("source", "")
+    def mapping(value):
+        if not isinstance(value, dict):
+            raise ValueError("The photo source returned invalid metadata.")
+        return value
+
+    data = mapping(summary.json())
+    original = mapping(data.get("originalimage") or {}).get("source", "")
+    image = mapping(data.get("thumbnail") or {}).get("source") or original
+    if not isinstance(image, str) or not isinstance(original, str):
+        raise ValueError("The photo source returned invalid image URLs.")
     if any(urlsplit(url).scheme != "https" or urlsplit(url).hostname not in IMAGE_HOSTS
            for url in [image, original]):
         raise ValueError("No supported destination image was available.")
@@ -33,10 +40,17 @@ def commons_filename(url):
         "iiprop": "extmetadata", "titles": "File:" + filename,
     })
     metadata.raise_for_status()
-    pages = metadata.json().get("query", {}).get("pages", {})
-    info = next(iter(pages.values()), {}).get("imageinfo", [{}])[0].get("extmetadata", {})
+    pages = mapping(mapping(mapping(metadata.json()).get("query")).get("pages"))
+    page = mapping(next(iter(pages.values()), None))
+    image_info = page.get("imageinfo")
+    if not isinstance(image_info, list) or not image_info:
+        raise ValueError("The photo source returned invalid image metadata.")
+    info = mapping(mapping(image_info[0]).get("extmetadata"))
     def plain(key):
-        return unescape(re.sub(r"<[^>]*>", "", info.get(key, {}).get("value", ""))).strip()[:500]
+        value = mapping(info.get(key) or {}).get("value", "")
+        if not isinstance(value, str):
+            raise ValueError("The photo source returned invalid attribution.")
+        return unescape(re.sub(r"<[^>]*>", "", value)).strip()[:500]
     license_name = plain("LicenseShortName")
     artist = plain("Artist")
     if not artist or not license_name or not (license_name.startswith("CC") or license_name == "Public domain"):

apps/agent/tests/test_trip_image_resilience_regression.py

@@ -0,0 +1,52 @@
+import httpx
+import pytest
+
+import src.trip_images as images
+
+
+IMAGE = 'https://upload.wikimedia.org/wikipedia/commons/a/ab/Test.jpg'
+METADATA = {'query': {'pages': {'1': {'imageinfo': [{'extmetadata': {
+    'Artist': {'value': 'A Photographer'}, 'LicenseShortName': {'value': 'CC BY-SA 4.0'},
+}}]}}}}
+
+
+def test_original_image_is_usable_when_summary_has_no_thumbnail():
+    def handle(request):
+        payload = {'originalimage': {'source': IMAGE}} if 'summary' in request.url.path else METADATA
+        return httpx.Response(200, json=payload)
+
+    with httpx.Client(transport=httpx.MockTransport(handle)) as client:
+        card = images.fetch_card('Coast', client)
+    assert card['status'] == 'available'
+    assert card['image_url'] == IMAGE
+    assert card['credit_url'].endswith('File:Test.jpg')
+
+
+@pytest.mark.parametrize(('summary', 'metadata'), [
+    ([], METADATA),
+    ({'thumbnail': {'source': IMAGE}, 'originalimage': {'source': IMAGE}}, {'query': []}),
+    ({'thumbnail': {'source': IMAGE}, 'originalimage': {'source': IMAGE}}, {'query': {'pages': []}}),
+    ({'thumbnail': {'source': IMAGE}, 'originalimage': {'source': IMAGE}}, {'query': {'pages': {'1': None}}}),
+    ({'thumbnail': {'source': IMAGE}, 'originalimage': {'source': IMAGE}}, {'query': {'pages': {'1': {'imageinfo': [{'extmetadata': None}]}}}}),
+])
+def test_malformed_stop_does_not_abort_other_destination_cards(monkeypatch, summary, metadata):
+    real_client = httpx.Client
+    def make_client(**kwargs):
+        current_title = None
+
+        def handle(request):
+            nonlocal current_title
+            if 'summary' in request.url.path:
+                current_title = request.url.path.rsplit('/', 1)[-1]
+                return httpx.Response(200, json=summary if current_title == 'Broken' else {
+                    'thumbnail': {'source': IMAGE}, 'originalimage': {'source': IMAGE},
+                })
+            return httpx.Response(200, json=metadata if current_title == 'Broken' else METADATA)
+
+        return real_client(transport=httpx.MockTransport(handle))
+
+    monkeypatch.setattr(images.httpx, 'Client', make_client)
+    result = images.get_trip_stop_images.invoke({'wikipedia_titles': ['Broken', 'Healthy']})
+    assert [stop['status'] for stop in result['stops']] == ['unavailable', 'available']
+    assert result['stops'][0]['place'] == 'Broken'
+    assert result['stops'][1]['image_url'] == IMAGE