aws-sdk-python-usage

Use client.get_paginator() instead of manually looping with NextToken. Call paginate() to iterate pages, or use .search() with a JMESPath expression to extract and flatten specific fields across all pages.

Clients map directly to AWS APIs and return plain dicts, covering every service. Resources provide an object-oriented interface for select services like S3 and DynamoDB, with automatic type marshalling that is especially useful for DynamoDB.

Catch typed exceptions via client.exceptions, such as s3.exceptions.NoSuchKey, rather than matching error code strings. Import the generic ClientError from botocore.exceptions, not boto3, and only catch exceptions when you have an actionable response.

The low-level client returns raw types like {"N": "42"}. Use boto3.resource("dynamodb") or resource.meta.client to auto-marshal values into native Python types like strings, Decimals, and sets.

Call s3.generate_presigned_url() with the operation name such as get_object or put_object, a Params dict with Bucket and Key, and ExpiresIn in seconds. For browser form uploads, use generate_presigned_post with conditions.

Pass a botocore.config.Config object to the client with retries settings (total_max_attempts and mode like standard or adaptive), connect_timeout, read_timeout, and max_pool_connections. Merge multiple configs with Config.merge().