Upload alpha_factory/cleanup.py

alpha_factory/cleanup.py

@@ -5,8 +5,11 @@ Run: uv run python -m alpha_factory.cleanup
 Also provides batch-level dedup to prevent theme repetition.
 """
 import duckdb
+import logging
 from pathlib import Path
 
+logger = logging.getLogger(__name__)
+
 DB_PATH = Path("factor_store/alphas.duckdb")
 
 # Fields that genuinely do NOT exist on BRAIN — any alpha using these is invalid.
@@ -19,15 +22,15 @@ FAKE_FIELDS = {
     "placeholder",
     "dummy_field",
     "test_field",
-    "my_field",
     "sample_field",
+    "my_field",
 }
 
 
 def cleanup_fake_alphas():
     """Remove alphas that used placeholder field names."""
     if not DB_PATH.exists():
-        print("No database found.")
+        logger.info("No database found.")
         return
 
     conn = duckdb.connect(str(DB_PATH))
@@ -57,15 +60,15 @@ def cleanup_fake_alphas():
     if to_delete:
         placeholders = ",".join(["?" for _ in to_delete])
         conn.execute(f"DELETE FROM alphas WHERE alpha_id IN ({placeholders})", to_delete)
-        print(f"Deleted {len(to_delete)} alphas with fake/placeholder fields:")
+        logger.info(f"Deleted {len(to_delete)} alphas with fake/placeholder fields")
         for aid in to_delete:
-            print(f"  - {aid}")
+            logger.info(f"  - {aid}")
     else:
-        print("No fake alphas found. Database is clean.")
+        logger.info("No fake alphas found. Database is clean.")
     
     # Show remaining
     count = conn.execute("SELECT COUNT(*) FROM alphas").fetchone()[0]
-    print(f"\nRemaining alphas in store: {count}")
+    logger.info(f"Remaining alphas in store: {count}")
     conn.close()
 
 
@@ -86,9 +89,9 @@ def cleanup_quoted_expressions():
             fixed += 1
     
     if fixed:
-        print(f"Fixed {fixed} expressions with quoted field names.")
+        logger.info(f"Fixed {fixed} expressions with quoted field names.")
     else:
-        print("No quoted field names found.")
+        logger.info("No quoted field names found.")
     conn.close()
 
 
@@ -130,17 +133,18 @@ def cleanup_orphans():
     if to_delete:
         placeholders = ",".join(["?" for _ in to_delete])
         conn.execute(f"DELETE FROM alphas WHERE alpha_id IN ({placeholders})", to_delete)
-        print(f"Deleted {len(to_delete)} alphas with unknown field references.")
+        logger.info(f"Deleted {len(to_delete)} alphas with unknown field references.")
     
     conn.close()
 
 
 if __name__ == "__main__":
-    print("=== Alpha Factory Cleanup v0.2.0 ===\n")
-    print("1. Removing alphas with fake/placeholder fields...")
+    logging.basicConfig(level=logging.INFO, format="%(levelname)s: %(message)s")
+    logger.info("=== Alpha Factory Cleanup v0.2.0 ===")
+    logger.info("1. Removing alphas with fake/placeholder fields...")
     cleanup_fake_alphas()
-    print("\n2. Fixing quoted field names in expressions...")
+    logger.info("2. Fixing quoted field names in expressions...")
     cleanup_quoted_expressions()
-    print("\n3. Removing alphas with unknown field references...")
+    logger.info("3. Removing alphas with unknown field references...")
     cleanup_orphans()
-    print("\nDone!")
+    logger.info("Done!")