Spaces:
Sleeping
Sleeping
Update ingest/fbi_vault_live.py
Browse files- ingest/fbi_vault_live.py +31 -15
ingest/fbi_vault_live.py
CHANGED
|
@@ -1,20 +1,36 @@
|
|
| 1 |
-
import
|
| 2 |
from ingest.generic_public_foia import GenericFOIAAdapter
|
| 3 |
|
|
|
|
| 4 |
class FBIAdapter(GenericFOIAAdapter):
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
base_url = "https://vault.fbi.gov/search"
|
| 9 |
|
| 10 |
async def search(self, query: str):
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import aiohttp
|
| 2 |
from ingest.generic_public_foia import GenericFOIAAdapter
|
| 3 |
|
| 4 |
+
|
| 5 |
class FBIAdapter(GenericFOIAAdapter):
|
| 6 |
+
source_name = "FBI Vault"
|
| 7 |
+
base_url = "https://vault.fbi.gov"
|
| 8 |
+
source_type = "live"
|
|
|
|
| 9 |
|
| 10 |
async def search(self, query: str):
|
| 11 |
+
await self._rate_limit()
|
| 12 |
+
try:
|
| 13 |
+
async with aiohttp.ClientSession() as session:
|
| 14 |
+
async with session.get(
|
| 15 |
+
f"{self.base_url}/search",
|
| 16 |
+
params={"q": query},
|
| 17 |
+
timeout=12
|
| 18 |
+
) as resp:
|
| 19 |
+
if resp.status != 200:
|
| 20 |
+
return []
|
| 21 |
+
|
| 22 |
+
text = await resp.text()
|
| 23 |
+
|
| 24 |
+
if "reading-room" in text.lower():
|
| 25 |
+
return [{
|
| 26 |
+
"source": self.source_name,
|
| 27 |
+
"title": f"FBI Vault document mentioning '{query}'",
|
| 28 |
+
"url": f"{self.base_url}/search?q={query}",
|
| 29 |
+
"snippet": "Public FBI FOIA document from The Vault.",
|
| 30 |
+
"live": True,
|
| 31 |
+
"extended": False
|
| 32 |
+
}]
|
| 33 |
+
except Exception:
|
| 34 |
+
return []
|
| 35 |
+
|
| 36 |
+
return []
|