Create app.py

app.py

@@ -1,90 +1,111 @@
 import gradio as gr
 import asyncio
 
-from ingest.registry import get_all_adapters
-from ingest.cluster import semantic_clusters
+from ingest.registry import get_live_adapters, get_stub_adapters
 from ingest.export import export_results_zip
-from ingest.health import check_health
 
-ADAPTERS = get_all_adapters()
+LIVE_ADAPTERS = get_live_adapters()
+STUB_ADAPTERS = get_stub_adapters()
 
-async def federated_search(query: str):
+async def federated_search(query: str, include_stubs: bool):
     results = []
-    for adapter in ADAPTERS.values():
+
+    for adapter in LIVE_ADAPTERS.values():
         try:
             res = await adapter.search(query)
             results.extend(res)
         except Exception:
             continue
+
+    if include_stubs:
+        for adapter in STUB_ADAPTERS.values():
+            try:
+                res = await adapter.search(query)
+                results.extend(res)
+            except Exception:
+                continue
+
     return results
 
-def run_search(query):
+def search_handler(query, include_stubs, confirm_extended):
+    if include_stubs and not confirm_extended:
+        return [], "⚠️ You must acknowledge the Extended Features warning."
+
     if not query.strip():
-        return [], []
+        return [], "⚠️ Enter a search term."
 
-    results = asyncio.run(federated_search(query))
+    results = asyncio.run(federated_search(query, include_stubs))
 
     rows = [
         [r["source"], r["title"], r["url"], r["snippet"]]
         for r in results
     ]
 
-    texts = [r["snippet"] for r in results]
-    clusters = semantic_clusters(texts) if texts else []
-
-    return rows, clusters
-
-def export_zip_handler(table):
-    results = []
-    for row in table:
-        results.append({
-            "source": row[0],
-            "title": row[1],
-            "url": row[2],
-            "snippet": row[3],
-        })
-    return export_results_zip(results)
+    return rows, ""
 
 with gr.Blocks() as app:
     gr.Markdown(
         "# **Federal FOIA Intelligence Search**\n"
         "### Public Electronic Reading Rooms Only\n\n"
-        "Live, robots-compliant search across U.S. government FOIA libraries."
+        "Live, robots-compliant search across U.S. government FOIA libraries.\n\n"
+        "**Default mode queries only agencies that explicitly permit automated access.**"
+    )
+
+    query = gr.Textbox(
+        label="Search term",
+        placeholder="e.g. UAP, procurement, surveillance"
+    )
+
+    include_stubs = gr.Checkbox(
+        label="Include Stub Results (Non-Live, Informational Only)",
+        value=False
     )
 
-    query = gr.Textbox(label="Search term", placeholder="e.g. UAP, procurement, surveillance")
+    with gr.Accordion("⚠️ Extended Features (Advanced – Read Carefully)", open=False):
+        gr.Markdown(
+            "**The following agencies do NOT permit automated querying or do not "
+            "provide public FOIA search endpoints:**\n\n"
+            "- NSA\n- NRO\n- SAP / Special Access Programs\n- TEN-CAP\n"
+            "- AATIP\n- Special Activities\n- DIA\n- NGA\n\n"
+            "If enabled, these sources return **clearly labeled stub results only**. "
+            "**No live requests are made.**"
+        )
+
+        confirm_extended = gr.Checkbox(
+            label="I understand and want to include stub-only results",
+            value=False
+        )
+
     search_btn = gr.Button("Search")
 
+    status = gr.Markdown("")
     results_table = gr.Dataframe(
-        headers=["Agency", "Title", "URL", "Snippet"],
-        interactive=False,
-        wrap=True
+        headers=["Source", "Title", "URL", "Snippet"],
+        wrap=True,
+        interactive=False
     )
 
-    cluster_output = gr.State()
-
     search_btn.click(
-        fn=run_search,
-        inputs=query,
-        outputs=[results_table, cluster_output]
+        fn=search_handler,
+        inputs=[query, include_stubs, confirm_extended],
+        outputs=[results_table, status]
     )
 
     gr.Markdown("### Export")
     export_btn = gr.Button("Export Results (ZIP)")
-    zip_file = gr.File(label="Download ZIP")
+    export_file = gr.File(label="Download ZIP")
 
     export_btn.click(
-        fn=export_zip_handler,
+        fn=lambda rows: export_results_zip([
+            {
+                "source": r[0],
+                "title": r[1],
+                "url": r[2],
+                "snippet": r[3]
+            } for r in rows
+        ]),
         inputs=results_table,
-        outputs=zip_file
+        outputs=export_file
     )
 
-    gr.Markdown("### Source Health")
-    for adapter in ADAPTERS.values():
-        health = check_health(adapter)
-        gr.Markdown(
-            f"- **{health['source']}**: {health['status']} "
-            f"({health['latency_ms']} ms)"
-        )
-
 app.launch()