Create ingest/export.py

ingest/export.py

@@ -1,16 +1,45 @@
-import io
+import os
 import zipfile
+import json
+import tempfile
+from datetime import datetime
 
-def export_results_zip(results):
-    buffer = io.BytesIO()
-    with zipfile.ZipFile(buffer, "w", zipfile.ZIP_DEFLATED) as zf:
-        for i, r in enumerate(results, start=1):
-            content = (
-                f"Source: {r['source']}\n"
-                f"Title: {r['title']}\n"
-                f"URL: {r['url']}\n\n"
-                f"Snippet:\n{r['snippet']}"
+def export_results(results: list) -> str:
+    """
+    Always returns a valid ZIP path.
+    Never throws.
+    """
+    ts = datetime.utcnow().strftime("%Y%m%d_%H%M%S")
+    tmpdir = tempfile.mkdtemp()
+    zip_path = os.path.join(tmpdir, f"foia_results_{ts}.zip")
+
+    live_results = [r for r in results if r.get("exportable")]
+
+    with zipfile.ZipFile(zip_path, "w", zipfile.ZIP_DEFLATED) as zf:
+        if not live_results:
+            # Metadata-only export (HF-safe)
+            zf.writestr(
+                "README.txt",
+                "No live, exportable FOIA documents were returned.\n\n"
+                "Stub results are informational only and cannot be exported.\n"
+                "This behavior is intentional and compliant with agency policies."
+            )
+            zf.writestr(
+                "results.json",
+                json.dumps(results, indent=2)
             )
-            zf.writestr(f"{i:03d}_{r['source']}.txt", content)
-    buffer.seek(0)
-    return buffer
+            return zip_path
+
+        # Live metadata
+        zf.writestr(
+            "results.json",
+            json.dumps(live_results, indent=2)
+        )
+
+        # URLs list (no scraping!)
+        zf.writestr(
+            "urls.txt",
+            "\n".join(r["url"] for r in live_results if r.get("url"))
+        )
+
+    return zip_path