Create ingest/extended_live_adapter.py

ingest/extended_live_adapter.py

@@ -1,20 +1,20 @@
 import aiohttp
+import time
 from ingest.generic_public_foia import GenericFOIAAdapter
-
+from ingest.health import HealthStatus
 
 class ExtendedLiveAdapter(GenericFOIAAdapter):
-    """
-    Extended live adapter.
-    Activated only when Extended Features mode is enabled.
-    Uses conservative keyword-presence search on public FOIA pages.
-    """
-
     source_type = "live"
     extended_only = True
 
     async def _safe_search(self, search_url: str, query: str):
+        start = time.perf_counter()
         await self._rate_limit()
 
+        if not await self._guard(search_url):
+            self.last_health = HealthStatus(ok=False, latency_ms=0, error="robots.txt")
+            return []
+
         try:
             async with aiohttp.ClientSession() as session:
                 async with session.get(
@@ -23,20 +23,25 @@ class ExtendedLiveAdapter(GenericFOIAAdapter):
                     timeout=12,
                     headers={"User-Agent": "HF-FOIA-Search/1.0"}
                 ) as resp:
-                    if resp.status != 200:
-                        return []
                     text = await resp.text()
+                    latency = int((time.perf_counter() - start) * 1000)
 
             if query.lower() in text.lower():
+                self.last_health = HealthStatus(ok=True, latency_ms=latency)
                 return [{
                     "source": self.source_name,
                     "title": f"{self.source_name} public document mentioning '{query}'",
                     "url": search_url,
                     "snippet": "Public FOIA-released material (extended live mode).",
                     "live": True,
-                    "extended": True
+                    "extended": True,
+                    "health": self.last_health.__dict__
                 }]
-        except Exception:
+
+            self.last_health = HealthStatus(ok=True, latency_ms=latency)
             return []
 
-        return []
+        except Exception as e:
+            latency = int((time.perf_counter() - start) * 1000)
+            self.last_health = HealthStatus(ok=False, latency_ms=latency, error=str(e))
+            return []