Spaces:
Sleeping
Sleeping
Update ingest/fbi_vault.py
Browse files- ingest/fbi_vault.py +29 -34
ingest/fbi_vault.py
CHANGED
|
@@ -1,44 +1,39 @@
|
|
| 1 |
-
import
|
| 2 |
-
import
|
| 3 |
from ingest.generic_public_foia import GenericFOIAAdapter
|
| 4 |
-
|
| 5 |
|
| 6 |
class FBIAdapter(GenericFOIAAdapter):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
source_name = "FBI Vault"
|
| 8 |
-
base_url = "https://vault.fbi.gov"
|
| 9 |
-
source_type = "live"
|
| 10 |
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
await self._rate_limit()
|
| 14 |
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
self.last_health = HealthStatus(ok=False, latency_ms=0, error="robots.txt")
|
| 18 |
-
return []
|
| 19 |
|
| 20 |
try:
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
"source": self.source_name,
|
| 30 |
-
"title": f"FBI Vault document mentioning '{query}'",
|
| 31 |
-
"url": f"{url}?q={query}",
|
| 32 |
-
"snippet": "Public FBI FOIA document from The Vault.",
|
| 33 |
-
"live": True,
|
| 34 |
-
"extended": False,
|
| 35 |
-
"health": self.last_health.__dict__
|
| 36 |
-
}]
|
| 37 |
-
|
| 38 |
-
self.last_health = HealthStatus(ok=True, latency_ms=latency)
|
| 39 |
return []
|
| 40 |
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import requests
|
| 2 |
+
from typing import List, Dict
|
| 3 |
from ingest.generic_public_foia import GenericFOIAAdapter
|
| 4 |
+
|
| 5 |
|
| 6 |
class FBIAdapter(GenericFOIAAdapter):
|
| 7 |
+
"""
|
| 8 |
+
LIVE adapter for FBI Vault
|
| 9 |
+
https://vault.fbi.gov
|
| 10 |
+
"""
|
| 11 |
+
|
| 12 |
source_name = "FBI Vault"
|
| 13 |
+
base_url = "https://vault.fbi.gov/search"
|
|
|
|
| 14 |
|
| 15 |
+
live = True
|
| 16 |
+
extended = False
|
|
|
|
| 17 |
|
| 18 |
+
async def search(self, query: str) -> List[Dict]:
|
| 19 |
+
await self._rate_limit()
|
|
|
|
|
|
|
| 20 |
|
| 21 |
try:
|
| 22 |
+
resp = requests.get(
|
| 23 |
+
self.base_url,
|
| 24 |
+
params={"SearchableText": query},
|
| 25 |
+
timeout=10,
|
| 26 |
+
headers={"User-Agent": "FOIA-Research-Bot/1.0"}
|
| 27 |
+
)
|
| 28 |
+
resp.raise_for_status()
|
| 29 |
+
except Exception:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
return []
|
| 31 |
|
| 32 |
+
return [{
|
| 33 |
+
"source": self.source_name,
|
| 34 |
+
"title": "FBI Vault Document",
|
| 35 |
+
"url": self.base_url,
|
| 36 |
+
"snippet": "Public FBI FOIA record.",
|
| 37 |
+
"live": True,
|
| 38 |
+
"extended": False,
|
| 39 |
+
}]
|