Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,4 +1,3 @@
|
|
| 1 |
-
import asyncio
|
| 2 |
import time
|
| 3 |
from typing import List, Dict
|
| 4 |
|
|
@@ -14,16 +13,16 @@ import coverage
|
|
| 14 |
import saved_searches
|
| 15 |
|
| 16 |
|
| 17 |
-
#
|
| 18 |
# Adapter Registry
|
| 19 |
-
#
|
| 20 |
|
| 21 |
ALL_ADAPTERS = get_all_adapters()
|
| 22 |
|
| 23 |
|
| 24 |
-
#
|
| 25 |
-
# Async Search Engine
|
| 26 |
-
#
|
| 27 |
|
| 28 |
async def _search_adapter(adapter, query: str) -> Dict:
|
| 29 |
start = time.time()
|
|
@@ -57,7 +56,7 @@ async def federated_search_async(
|
|
| 57 |
]
|
| 58 |
|
| 59 |
tasks = [_search_adapter(a, query) for a in adapters]
|
| 60 |
-
responses = await
|
| 61 |
|
| 62 |
flat_results = []
|
| 63 |
health = {}
|
|
@@ -73,9 +72,17 @@ async def federated_search_async(
|
|
| 73 |
return flat_results, health
|
| 74 |
|
| 75 |
|
| 76 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 77 |
# Rendering Helpers
|
| 78 |
-
#
|
| 79 |
|
| 80 |
def badge(result: Dict) -> str:
|
| 81 |
badges = []
|
|
@@ -113,11 +120,11 @@ def render_health(health: Dict) -> str:
|
|
| 113 |
return "\n".join(lines)
|
| 114 |
|
| 115 |
|
| 116 |
-
#
|
| 117 |
-
#
|
| 118 |
-
#
|
| 119 |
|
| 120 |
-
def run_search(
|
| 121 |
query: str,
|
| 122 |
hide_stubs: bool,
|
| 123 |
enable_extended: bool,
|
|
@@ -133,10 +140,7 @@ def run_search(
|
|
| 133 |
[a for a in enabled_extended_agencies if a in EXTENDED_AGENCIES]
|
| 134 |
)
|
| 135 |
|
| 136 |
-
|
| 137 |
-
results, health = loop.run_until_complete(
|
| 138 |
-
federated_search_async(query, agencies)
|
| 139 |
-
)
|
| 140 |
|
| 141 |
saved_searches.save(query, enable_extended)
|
| 142 |
|
|
@@ -150,28 +154,28 @@ def run_search(
|
|
| 150 |
return rendered, render_health(health), coverage.render_coverage()
|
| 151 |
|
| 152 |
|
| 153 |
-
#
|
| 154 |
# Gradio UI
|
| 155 |
-
#
|
| 156 |
|
| 157 |
-
with gr.Blocks(title="Federated FOIA Search") as demo:
|
| 158 |
gr.Markdown(
|
| 159 |
"""
|
| 160 |
# 🏛️ Federated FOIA Document Search
|
| 161 |
-
Search **public government FOIA reading rooms** across multiple agencies.
|
| 162 |
|
| 163 |
-
|
| 164 |
-
|
| 165 |
-
-
|
| 166 |
-
-
|
|
|
|
|
|
|
| 167 |
"""
|
| 168 |
)
|
| 169 |
|
| 170 |
-
|
| 171 |
-
|
| 172 |
-
|
| 173 |
-
|
| 174 |
-
)
|
| 175 |
|
| 176 |
with gr.Row():
|
| 177 |
hide_stubs = gr.Checkbox(
|
|
@@ -219,7 +223,7 @@ Search **public government FOIA reading rooms** across multiple agencies.
|
|
| 219 |
**Legal Notice**
|
| 220 |
|
| 221 |
This tool:
|
| 222 |
-
- Accesses **only publicly available FOIA
|
| 223 |
- Respects robots.txt and rate limits
|
| 224 |
- Does not bypass access controls
|
| 225 |
- Is intended for journalism, research, and public-interest use
|
|
@@ -227,9 +231,9 @@ This tool:
|
|
| 227 |
)
|
| 228 |
|
| 229 |
|
| 230 |
-
#
|
| 231 |
-
# HF
|
| 232 |
-
#
|
| 233 |
|
| 234 |
if __name__ == "__main__":
|
| 235 |
demo.launch()
|
|
|
|
|
|
|
| 1 |
import time
|
| 2 |
from typing import List, Dict
|
| 3 |
|
|
|
|
| 13 |
import saved_searches
|
| 14 |
|
| 15 |
|
| 16 |
+
# ======================================================
|
| 17 |
# Adapter Registry
|
| 18 |
+
# ======================================================
|
| 19 |
|
| 20 |
ALL_ADAPTERS = get_all_adapters()
|
| 21 |
|
| 22 |
|
| 23 |
+
# ======================================================
|
| 24 |
+
# Async Federated Search Engine
|
| 25 |
+
# ======================================================
|
| 26 |
|
| 27 |
async def _search_adapter(adapter, query: str) -> Dict:
|
| 28 |
start = time.time()
|
|
|
|
| 56 |
]
|
| 57 |
|
| 58 |
tasks = [_search_adapter(a, query) for a in adapters]
|
| 59 |
+
responses = await gradio_async_gather(tasks)
|
| 60 |
|
| 61 |
flat_results = []
|
| 62 |
health = {}
|
|
|
|
| 72 |
return flat_results, health
|
| 73 |
|
| 74 |
|
| 75 |
+
async def gradio_async_gather(tasks):
|
| 76 |
+
"""
|
| 77 |
+
HF-safe async gather wrapper.
|
| 78 |
+
"""
|
| 79 |
+
import asyncio
|
| 80 |
+
return await asyncio.gather(*tasks, return_exceptions=False)
|
| 81 |
+
|
| 82 |
+
|
| 83 |
+
# ======================================================
|
| 84 |
# Rendering Helpers
|
| 85 |
+
# ======================================================
|
| 86 |
|
| 87 |
def badge(result: Dict) -> str:
|
| 88 |
badges = []
|
|
|
|
| 120 |
return "\n".join(lines)
|
| 121 |
|
| 122 |
|
| 123 |
+
# ======================================================
|
| 124 |
+
# HF-SAFE ASYNC SEARCH HANDLER
|
| 125 |
+
# ======================================================
|
| 126 |
|
| 127 |
+
async def run_search(
|
| 128 |
query: str,
|
| 129 |
hide_stubs: bool,
|
| 130 |
enable_extended: bool,
|
|
|
|
| 140 |
[a for a in enabled_extended_agencies if a in EXTENDED_AGENCIES]
|
| 141 |
)
|
| 142 |
|
| 143 |
+
results, health = await federated_search_async(query, agencies)
|
|
|
|
|
|
|
|
|
|
| 144 |
|
| 145 |
saved_searches.save(query, enable_extended)
|
| 146 |
|
|
|
|
| 154 |
return rendered, render_health(health), coverage.render_coverage()
|
| 155 |
|
| 156 |
|
| 157 |
+
# ======================================================
|
| 158 |
# Gradio UI
|
| 159 |
+
# ======================================================
|
| 160 |
|
| 161 |
+
with gr.Blocks(title="Federated FOIA Document Search") as demo:
|
| 162 |
gr.Markdown(
|
| 163 |
"""
|
| 164 |
# 🏛️ Federated FOIA Document Search
|
|
|
|
| 165 |
|
| 166 |
+
Search **public government FOIA electronic reading rooms** across multiple agencies.
|
| 167 |
+
|
| 168 |
+
- **Default:** Safe, public-only sources
|
| 169 |
+
- **Extended mode:** Additional public releases (opt-in)
|
| 170 |
+
- No authentication
|
| 171 |
+
- No classified or restricted systems
|
| 172 |
"""
|
| 173 |
)
|
| 174 |
|
| 175 |
+
query = gr.Textbox(
|
| 176 |
+
label="Search term",
|
| 177 |
+
placeholder="e.g. UAP, radar incident, AATIP",
|
| 178 |
+
)
|
|
|
|
| 179 |
|
| 180 |
with gr.Row():
|
| 181 |
hide_stubs = gr.Checkbox(
|
|
|
|
| 223 |
**Legal Notice**
|
| 224 |
|
| 225 |
This tool:
|
| 226 |
+
- Accesses **only publicly available FOIA electronic reading rooms**
|
| 227 |
- Respects robots.txt and rate limits
|
| 228 |
- Does not bypass access controls
|
| 229 |
- Is intended for journalism, research, and public-interest use
|
|
|
|
| 231 |
)
|
| 232 |
|
| 233 |
|
| 234 |
+
# ======================================================
|
| 235 |
+
# HF ENTRY POINT
|
| 236 |
+
# ======================================================
|
| 237 |
|
| 238 |
if __name__ == "__main__":
|
| 239 |
demo.launch()
|