ci: fix lint errors across project, update ruff.toml per-file ignores

scripts/check_input.py

@@ -1,11 +1,15 @@
+import os
+
 from minio import Minio
-import os, json
 
 mc = Minio(os.environ["MINIO_ENDPOINT"], access_key=os.environ["MINIO_ACCESS_KEY"], secret_key=os.environ["MINIO_SECRET_KEY"], secure=False)
 
 # Check the most recent extraction - what text did the model get?
 # Look at the normalized text for a known doc
-import asyncio, asyncpg
+import asyncio
+
+import asyncpg
+
 
 async def main():
     pool = await asyncpg.create_pool(
@@ -15,20 +19,20 @@ async def main():
         user=os.environ["POSTGRES_USER"],
         password=os.environ["POSTGRES_PASSWORD"],
     )
-    
+
     # Get a recently extracted doc
     row = await pool.fetchrow(
         "SELECT id, title, normalized_storage_ref, parse_quality_score "
         "FROM documents WHERE source_type = 'news_api' AND parse_quality_score > 0.8 "
         "ORDER BY updated_at DESC LIMIT 1"
     )
-    
+
     if row:
         print(f"Doc: {row['id']}")
         print(f"Title: {row['title']}")
         print(f"Quality: {row['parse_quality_score']}")
         print(f"Ref: {row['normalized_storage_ref']}")
-        
+
         ref = row["normalized_storage_ref"]
         parts = ref.replace("s3://", "").split("/", 1)
         if len(parts) == 2:
@@ -37,9 +41,9 @@ async def main():
             obj.close()
             obj.release_conn()
             print(f"Text length: {len(text)} chars")
-            print(f"First 500 chars:")
+            print("First 500 chars:")
             print(text[:500])
-    
+
     await pool.close()
 
 asyncio.run(main())