Create ingest/utils.py

ingest/utils.py

@@ -0,0 +1,25 @@
+def highlight_terms(text: str, query: str) -> str:
+    if not text or not query:
+        return text
+    return text.replace(query, f"**{query}**")
+
+
+def classify_foia_exemptions(text: str):
+    # Explainable, rule-based, non-inferential
+    exemptions = []
+    lowered = text.lower()
+    if "national security" in lowered:
+        exemptions.append("b(1)")
+    if "internal personnel" in lowered:
+        exemptions.append("b(2)")
+    if "privacy" in lowered:
+        exemptions.append("b(6)")
+    return exemptions or ["none"]
+
+
+def format_citation(record: dict) -> str:
+    return (
+        f"{record.get('source')}. "
+        f"\"{record.get('title')}\". "
+        f"Retrieved from {record.get('url')}."
+    )