sharepoint-transcribe/sp-transcribe-service/lambdas/submit.py

"""API Gateway handler: validate API key, create job, start Step Functions."""
import json
import uuid
import time
import os
import boto3
from shared import get_user, api_response, JOBS_TABLE, S3_BUCKET

sfn = boto3.client('stepfunctions')
ddb = boto3.resource('dynamodb')


def handler(event, context):
    # Parse body
    try:
        body = json.loads(event.get('body', '{}'))
    except Exception:
        return api_response(400, {'error': 'Invalid JSON body'})

    api_key = body.get('api_key') or event.get('headers', {}).get('x-api-key', '')
    curl_command = body.get('curl')
    cookies = body.get('cookies', '')

    if not api_key:
        return api_response(401, {'error': 'Missing api_key'})
    if not curl_command:
        return api_response(400, {'error': 'Missing curl'})

    # Validate user
    user = get_user(api_key)
    if not user:
        return api_response(401, {'error': 'Invalid API key'})

    # Create job
    job_id = str(uuid.uuid4())[:8]
    now = int(time.time())
    ttl = now + 7 * 86400  # 7 days

    table = ddb.Table(JOBS_TABLE)
    table.put_item(Item={
        'job_id': job_id,
        'api_key': api_key,
        'email': user.get('email', ''),
        'status': 'SUBMITTED',
        'created_at': now,
        'ttl': ttl,
    })

    # Store curl + cookies in S3 (avoid DynamoDB size limits)
    s3 = boto3.client('s3')
    s3.put_object(
        Bucket=S3_BUCKET,
        Key=f"jobs/{job_id}/input.json",
        Body=json.dumps({
            'curl': curl_command,
            'cookies': cookies,
            'language': body.get('language', 'auto'),
        }),
        ContentType='application/json',
    )

    # Start Step Functions
    state_machine_arn = os.environ.get('STATE_MACHINE', '')
    sfn.start_execution(
        stateMachineArn=state_machine_arn,
        name=f"job-{job_id}",
        input=json.dumps({
            'job_id': job_id,
            'email': user.get('email', ''),
            'user_name': user.get('name', ''),
        }),
    )

    return api_response(200, {
        'job_id': job_id,
        'message': 'Job submitted, you will receive an email when complete.',
    })