fix: follow redirects in httpx clients, better url joining, more verbose health check (#6)

* fix: follow redirects in httpx clients

* fix: handle trailing slashes more robustly

Author: hrodmn

src/stac_fastapi/collection_discovery/core.py

@@ -26,6 +26,14 @@
 logger = logging.getLogger(__name__)
 
 
+def _robust_urljoin(base: str, path: str) -> str:
+    """Join URL parts robustly, ensuring base is treated as a directory."""
+    # Ensure base ends with slash so it's treated as a directory
+    if not base.endswith("/"):
+        base += "/"
+    return urljoin(base, path)
+
+
 COLLECTION_SEARCH_CONFORMANCE_CLASSES = [
     STACConformanceClasses.CORE,
     STACConformanceClasses.COLLECTIONS,
@@ -40,6 +48,7 @@ class UpstreamApiStatus(BaseModel):
     """Status information for an upstream API."""
 
     healthy: bool
+    collection_search_conformance: list[str]
 
 
 class LifespanStatus(BaseModel):
@@ -112,7 +121,9 @@ def _get_search_state(
             logger.info("Continuing collection search with token pagination")
         else:
             search_state = {
-                "current": {api: f"{api}/collections?{param_str}" for api in apis},
+                "current": {
+                    api: _robust_urljoin(api, f"collections?{param_str}") for api in apis
+                },
                 "is_first_page": True,
             }
             logger.info(f"Starting new collection search across {len(apis)} APIs")
@@ -139,7 +150,9 @@ def _build_pagination_links(
             links.append(
                 {
                     "rel": "previous",
-                    "href": f"{request.base_url}collections?token={prev_token}",
+                    "href": _robust_urljoin(
+                        str(request.base_url), f"collections?token={prev_token}"
+                    ),
                 }
             )
 
@@ -152,7 +165,9 @@ def _build_pagination_links(
             links.append(
                 {
                     "rel": "next",
-                    "href": f"{request.base_url}collections?token={next_token}",
+                    "href": _robust_urljoin(
+                        str(request.base_url), f"collections?token={next_token}"
+                    ),
                 }
             )
 
@@ -164,7 +179,7 @@ async def _fetch_api_conformance(
         """Fetch conformance classes from a single API."""
         async with semaphore:
             try:
-                api_response = await client.get(f"{api}/conformance")
+                api_response = await client.get(api, follow_redirects=True)
                 if api_response.status_code == 200:
                     conformance_data = api_response.json()
                     return api, set(conformance_data.get("conformsTo", []))
@@ -507,12 +522,38 @@ async def health_check(
     async def check_api_health(client, api: str) -> tuple[str, UpstreamApiStatus]:
         async with semaphore:
             try:
-                # Check landing page health
-                api_response = await client.get(api)
-                return api, UpstreamApiStatus(healthy=api_response.status_code == 200)
+                # Check landing page health - follow redirects automatically
+                api_response = await client.get(api, follow_redirects=True)
+                # Consider 2xx and 3xx status codes as healthy
+                is_healthy = 200 <= api_response.status_code < 400
+
+                collection_search_conformance = []
+                if is_healthy:
+                    try:
+                        response_data = api_response.json()
+                        conforms_to = response_data.get("conformsTo", [])
+                        # Extract collection-search conformance classes with suffixes
+                        for cls in conforms_to:
+                            if "collection-search" in cls:
+                                # Extract from "collection-search" onwards
+                                collection_search_part = cls[
+                                    cls.find("collection-search") :
+                                ]
+                                collection_search_conformance.append(
+                                    collection_search_part
+                                )
+                    except Exception as e:
+                        logger.warning(f"Failed to parse conformance from {api}: {e}")
+
+                return api, UpstreamApiStatus(
+                    healthy=is_healthy,
+                    collection_search_conformance=collection_search_conformance,
+                )
 
             except Exception:
-                return api, UpstreamApiStatus(healthy=False)
+                return api, UpstreamApiStatus(
+                    healthy=False, collection_search_conformance=[]
+                )
 
     async with AsyncClient(timeout=HTTPX_TIMEOUT) as client:
         tasks = [check_api_health(client, api) for api in apis]

tests/unit/test_app.py

@@ -239,7 +239,7 @@ async def test_landing_page_with_single_custom_api(self, client):
     async def test_conformance_with_custom_apis(self, client):
         """Test conformance endpoint with custom APIs parameter."""
         # Mock upstream API conformance endpoints for custom APIs
-        respx.get("https://custom-api1.example.com/conformance").mock(
+        respx.get("https://custom-api1.example.com").mock(
             return_value=Response(
                 200,
                 json={
@@ -250,7 +250,7 @@ async def test_conformance_with_custom_apis(self, client):
                 },
             )
         )
-        respx.get("https://custom-api2.example.com/conformance").mock(
+        respx.get("https://custom-api2.example.com").mock(
             return_value=Response(
                 200,
                 json={

tests/unit/test_health.py

@@ -67,3 +67,23 @@ async def test_health_check_all_apis_unhealthy(self, mock_request):
 
         api2_result = result.upstream_apis["https://api2.example.com"]
         assert api2_result.healthy is False
+
+    @pytest.mark.asyncio
+    @respx.mock
+    async def test_health_check_redirects_healthy(self, mock_request):
+        """Test health check when APIs return redirects (should be considered healthy)."""
+        # Mock api1 to redirect once then return success
+        respx.get("https://api1.example.com").mock(return_value=Response(301))
+        # Mock api2 to redirect once then return success
+        respx.get("https://api2.example.com").mock(return_value=Response(302))
+
+        result = await health_check(mock_request)
+
+        assert result.status == "UP"
+        assert result.lifespan.status == "UP"
+
+        api1_result = result.upstream_apis["https://api1.example.com"]
+        assert api1_result.healthy is True
+
+        api2_result = result.upstream_apis["https://api2.example.com"]
+        assert api2_result.healthy is True