Timeout for label detection

copienator/commands/submit_batches.py

@@ -19,11 +19,7 @@ from copienator import (
 )
 
 
-def run(workspace: EvaluationWorkspace, *, client=None) -> ExitCode:
-    if client is None:
-        if not config.API_KEY:
-            raise CliError("GEMINI_API_KEY is not configured")
-        client = genai.Client(api_key=config.API_KEY)
+def run(workspace: EvaluationWorkspace, *, client=None, overwrite: bool = False) -> ExitCode:
     batches = (
         (
             "flash",
@@ -45,16 +41,32 @@ def run(workspace: EvaluationWorkspace, *, client=None) -> ExitCode:
     }
     if workspace.batch_jobs_file.is_file():
         previous = read_json(workspace.batch_jobs_file)
-        if isinstance(previous, dict) and isinstance(previous.get("jobs"), dict):
-            manifest["jobs"] = previous["jobs"]
+        if not isinstance(previous, dict) or not isinstance(previous.get("jobs"), dict):
+            raise CliError(f"Invalid batch manifest: {workspace.batch_jobs_file}")
+        manifest["jobs"] = previous["jobs"]
     started = 0
+    skipped = 0
     for tier, file_path, model_id, display_name in batches:
+        existing = manifest["jobs"].get(tier)
+        if tier in manifest["jobs"]:
+            if (not isinstance(existing, dict)
+                    or not isinstance(existing.get("name"), str)
+                    or not existing["name"].strip()):
+                raise CliError(f"Invalid batch job in {workspace.batch_jobs_file}: {tier}")
+            if not overwrite:
+                print(f"Skipping {tier}: already submitted as {existing['name']}.")
+                skipped += 1
+                continue
         if not file_path.is_file():
             print(f"Skipping {model_id}: {file_path.name} does not exist.")
             continue
         if file_path.stat().st_size == 0:
             print(f"Skipping {model_id}: {file_path.name} is empty.")
             continue
+        if client is None:
+            if not config.API_KEY:
+                raise CliError("GEMINI_API_KEY is not configured")
+            client = genai.Client(api_key=config.API_KEY)
         print(f"Uploading {file_path.name} for model {model_id}...")
         uploaded = client.files.upload(
             file=str(file_path),
@@ -77,19 +89,26 @@ def run(workspace: EvaluationWorkspace, *, client=None) -> ExitCode:
         }
         atomic_write_json(workspace.batch_jobs_file, manifest)
         print(f"Started batch job: {job.name}")
-    if not started:
+    if not started and not skipped:
         print("No non-empty batch request files were found.")
         return ExitCode.PARTIAL
     return ExitCode.SUCCESS
 
 
 def build_parser() -> argparse.ArgumentParser:
-    return evaluation_parser("Upload correction JSONL files and start Gemini batches")
+    parser = evaluation_parser("Upload correction JSONL files and start Gemini batches")
+    parser.add_argument(
+        "--overwrite", action="store_true",
+        help="Submit new jobs even for recorded tiers, replacing their saved job IDs",
+    )
+    return parser
 
 
 def main(argv: Sequence[str] | None = None) -> int:
     parser = build_parser()
-    return execute(parser, argv, lambda args: run(workspace_from_args(args)))
+    return execute(parser, argv, lambda args: run(
+        workspace_from_args(args), overwrite=args.overwrite,
+    ))
 
 
 if __name__ == "__main__":