HumanSignal/label-studio · error · ValidationError

preannotated_fields

Error message

preannotated_fields

What it means

reformat_predictions converts legacy 'preannotated_fields' style data into task+predictions structures. While processing it collects validation errors about missing pre-annotation fields, and if any were found and raise_errors is set, it raises ValidationError({'preannotated_fields': validation_errors}). The dict key 'preannotated_fields' in the response identifies this error family.

Source

Thrown at label_studio/data_import/functions.py:281

                            'from_name': field,
                            'to_name': to_name,
                            'type': prediction_type,
                            'value': prediction_value,
                        }
                    ],
                    'score': 1.0,
                    'model_version': 'preannotated',
                }

                predictions.append(prediction)

        # Create new task structure
        new_task = {'data': task_data, 'predictions': predictions}
        new_tasks.append(new_task)

    # If there are validation errors, raise them
    if validation_errors and raise_errors:
        raise ValidationError({'preannotated_fields': validation_errors})

    return new_tasks


post_process_reimport = load_func(settings.POST_PROCESS_REIMPORT)


def _async_reimport_background_streaming(reimport, project, organization_id, user):
    """Streaming version of reimport that processes tasks in batches to reduce memory usage"""
    try:
        # Get batch size from settings or use default
        batch_size = settings.REIMPORT_BATCH_SIZE

        # Initialize counters
        total_task_count = 0
        total_annotation_count = 0
        total_prediction_count = 0
        all_found_formats = {}

View on GitHub (pinned to 0b49e9b539)

Solutions

1. Read validation_errors under the 'preannotated_fields' key; each entry names the missing/invalid field
2. Add the expected pre-annotated field keys to each task's data object
3. Update the import payload to the current annotations/predictions format instead of legacy preannotated_fields
4. Check the project/template configuration for which data keys it expects
5. Verify upstream data generation emits the required keys for every task

Example fix

// before: task data missing the pre-annotated field
{"data": {"text": "hello"}}
// after: include expected preannotated field
{"data": {"text": "hello", "preannotated_choices": ["positive"]}}

required = {'preannotated_choices'}  # keys the template/project expects
for t in tasks:
    missing = required - t.get('data', {}).keys()
    if missing:
        raise ValueError(f'task data missing preannotated fields: {missing}')

def has_preannotated_fields(task: dict, required: set) -> bool:
    data = task.get('data') if isinstance(task, dict) else None
    return isinstance(data, dict) and required.issubset(data.keys())

try:
    requests.post(import_url, headers=H, json=tasks).raise_for_status()
except requests.HTTPError as e:
    detail = e.response.json()
    if 'preannotated_fields' in detail:
        log.error('Pre-annotation field errors: %s', detail['preannotated_fields'])

When it happens

Trigger: Importing tasks whose data references pre-annotation fields (e.g. data keys configured as pre-annotated sources) that are missing or invalid in the item, during sync_import, async import background processing, or streaming re-import — with raise_errors enabled.

Common situations: Legacy pre-annotation workflows where the expected data keys were renamed or omitted; re-importing old exports into projects with different data key configuration; templates expecting {data}['annotations']-style fields that the payload lacks; automated pipelines producing tasks without the pre-annotated field keys.

Understand the failure class

Background: "Missing required field" and "field is required" errors: why libraries reject payloads that omit mandatory fields — this error's family across 20 libraries.

Related errors

• Prediction validation failed ({len(validation_errors)} error
• "url" must be 2048 characters or fewer
• Can't deserialize tasks due to {errors}
• "url" is not found in request data
• load_tasks: No data found in DATA or in FILES

AI-assisted analysis of HumanSignal/label-studio@0b49e9b539 (2026-08-29). Data as JSON: /api/errors/c3542155a1fc1dc1. Report an issue: GitHub.