fix: harden route parsing, column dedup, and column name access

- Normalize QU route output (strip/lowercase) before validation
- Fix column name deduplication to handle suffix collisions
(e.g., 'Score_A', 'Score_A', 'Score_A_2' no longer produces duplicates)
- Use defensive .get() for column name access in SQL result formatting

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

src/query_engine.py

@@ -162,6 +162,8 @@ def _parse_qu_result(raw: str, original_query: str) -> dict:
     search_query = parsed.get("search_query", original_query)
     display_query = parsed.get("display_query", original_query)
     route = parsed.get("route", "vector")
+    if isinstance(route, str):
+        route = route.strip().lower()
     if route not in ("sql", "vector", "both"):
         route = "vector"
     sql_query = parsed.get("sql_query") if route in ("sql", "both") else None

src/sql_ingest.py

@@ -499,15 +499,18 @@ def _ingest_tables(
 
             # Infer column types, collect samples and stats
             safe_headers = [_sanitize_column_name(h) for h in headers]
-            # Deduplicate: append _2, _3, etc. for collisions
-            seen: dict[str, int] = {}
+            # Deduplicate: ensure all column names are unique after sanitization.
+            # Checks every name against all previously assigned names,
+            # handling both same-name collisions and suffix collisions
+            # (e.g., 'Score_A', 'Score_A', 'Score_A_2' won't produce duplicates).
+            used: set[str] = set()
             for i, name in enumerate(safe_headers):
-                lower = name.lower()
-                if lower in seen:
-                    seen[lower] += 1
-                    safe_headers[i] = f"{name}_{seen[lower]}"
-                else:
-                    seen[lower] = 1
+                if name.lower() in used:
+                    suffix = 2
+                    while f"{name}_{suffix}".lower() in used:
+                        suffix += 1
+                    safe_headers[i] = f"{name}_{suffix}"
+                used.add(safe_headers[i].lower())
             col_types = []
             col_samples = []
             col_stats = []

src/sql_retriever.py

@@ -93,7 +93,7 @@ def format_sql_results_as_context(
         if table_desc:
             parts.append(f"Dataset description: {table_desc}\n")
         col_descs = {
-            c["name"]: c.get("description", "")
+            c.get("name", "unknown"): c.get("description", "")
             for c in table_info.get("columns", [])
         }
         desc_lines = [f"  {name}: {desc}" for name, desc in col_descs.items() if desc]