Parser

This utility provides data parsing and deep validation using Pydantic.

Key features

• Defines data in pure Python classes, then parse, validate and extract only what you want
• Built-in envelopes to unwrap, extend, and validate popular event sources payloads
• Enforces type hints at runtime with user-friendly errors

Extra dependency

Warning

This will increase the compressed package size by >10MB due to the Pydantic dependency.

To reduce the impact on the package size at the expense of 30%-50% of its performance Pydantic can also be installed without binary files:

SKIP_CYTHON=1 pip install --no-binary pydantic aws-lambda-powertools[pydantic]

Install parser's extra dependencies using pip install aws-lambda-powertools[pydantic].

Defining models

You can define models to parse incoming events by inheriting from BaseModel.

from aws_lambda_powertools.utilities.parser import BaseModel
from typing import List, Optional

class OrderItem(BaseModel):
    id: int
    quantity: int
    description: str

class Order(BaseModel):
    id: int
    description: str
    items: List[OrderItem] # nesting models are supported
    optional_field: Optional[str] # this field may or may not be available when parsing

These are simply Python classes that inherit from BaseModel. Parser enforces type hints declared in your model at runtime.

Parsing events

You can parse inbound events using event_parser decorator, or the standalone parse function. Both are also able to parse either dictionary or JSON string as an input.

event_parser decorator

Use the decorator for fail fast scenarios where you want your Lambda function to raise an exception in the event of a malformed payload.

event_parser decorator will throw a ValidationError if your event cannot be parsed according to the model.

Note

This decorator will replace the event object with the parsed model if successful. This means you might be careful when nesting other decorators that expect event to be a dict.

from aws_lambda_powertools.utilities.parser import event_parser, BaseModel
from aws_lambda_powertools.utilities.typing import LambdaContext
from typing import List, Optional

import json

class OrderItem(BaseModel):
    id: int
    quantity: int
    description: str

class Order(BaseModel):
    id: int
    description: str
    items: List[OrderItem] # nesting models are supported
    optional_field: Optional[str] # this field may or may not be available when parsing


@event_parser(model=Order)
def handler(event: Order, context: LambdaContext):
    print(event.id)
    print(event.description)
    print(event.items)

    order_items = [item for item in event.items]
    ...

payload = {
    "id": 10876546789,
    "description": "My order",
    "items": [
        {
            "id": 1015938732,
            "quantity": 1,
            "description": "item xpto"
        }
    ]
}

handler(event=payload, context=LambdaContext())
handler(event=json.dumps(payload), context=LambdaContext()) # also works if event is a JSON string

parse function

Use this standalone function when you want more control over the data validation process, for example returning a 400 error for malformed payloads.

from aws_lambda_powertools.utilities.parser import parse, BaseModel, ValidationError
from typing import List, Optional

class OrderItem(BaseModel):
    id: int
    quantity: int
    description: str

class Order(BaseModel):
    id: int
    description: str
    items: List[OrderItem] # nesting models are supported
    optional_field: Optional[str] # this field may or may not be available when parsing


payload = {
    "id": 10876546789,
    "description": "My order",
    "items": [
        {
            # this will cause a validation error
            "id": [1015938732],
            "quantity": 1,
            "description": "item xpto"
        }
    ]
}

def my_function():
    try:
        parsed_payload: Order = parse(event=payload, model=Order)
        # payload dict is now parsed into our model
        return parsed_payload.items
    except ValidationError:
        return {
            "status_code": 400,
            "message": "Invalid order"
        }

Built-in models

Parser comes with the following built-in models:

Model name	Description
DynamoDBStreamModel	Lambda Event Source payload for Amazon DynamoDB Streams
EventBridgeModel	Lambda Event Source payload for Amazon EventBridge
SqsModel	Lambda Event Source payload for Amazon SQS
AlbModel	Lambda Event Source payload for Amazon Application Load Balancer
CloudwatchLogsModel	Lambda Event Source payload for Amazon CloudWatch Logs
S3Model	Lambda Event Source payload for Amazon S3
S3ObjectLambdaEvent	Lambda Event Source payload for Amazon S3 Object Lambda
KinesisDataStreamModel	Lambda Event Source payload for Amazon Kinesis Data Streams
SesModel	Lambda Event Source payload for Amazon Simple Email Service
SnsModel	Lambda Event Source payload for Amazon Simple Notification Service
APIGatewayProxyEventModel	Lambda Event Source payload for Amazon API Gateway
APIGatewayProxyEventV2Model	Lambda Event Source payload for Amazon API Gateway v2 payload

extending built-in models

You can extend them to include your own models, and yet have all other known fields parsed along the way.

Tip

For Mypy users, we only allow type override for fields where payload is injected e.g. detail, body, etc.

from aws_lambda_powertools.utilities.parser import parse, BaseModel
from aws_lambda_powertools.utilities.parser.models import EventBridgeModel

from typing import List, Optional

class OrderItem(BaseModel):
    id: int
    quantity: int
    description: str

class Order(BaseModel):
    id: int
    description: str
    items: List[OrderItem]

class OrderEventModel(EventBridgeModel):
    detail: Order

payload = {
    "version": "0",
    "id": "6a7e8feb-b491-4cf7-a9f1-bf3703467718",
    "detail-type": "OrderPurchased",
    "source": "OrderService",
    "account": "111122223333",
    "time": "2020-10-22T18:43:48Z",
    "region": "us-west-1",
    "resources": ["some_additional"],
    "detail": {
        "id": 10876546789,
        "description": "My order",
        "items": [
            {
                "id": 1015938732,
                "quantity": 1,
                "description": "item xpto"
            }
        ]
    }
}

ret = parse(model=OrderEventModel, event=payload)

assert ret.source == "OrderService"
assert ret.detail.description == "My order"
assert ret.detail_type == "OrderPurchased" # we rename it to snake_case since detail-type is an invalid name

for order_item in ret.detail.items:
    ...

What's going on here, you might ask:

1. We imported our built-in model EventBridgeModel from the parser utility
2. Defined how our Order should look like
3. Defined how part of our EventBridge event should look like by overriding detail key within our OrderEventModel
4. Parser parsed the original event against OrderEventModel

Envelopes

When trying to parse your payloads wrapped in a known structure, you might encounter the following situations:

• Your actual payload is wrapped around a known structure, for example Lambda Event Sources like EventBridge
• You're only interested in a portion of the payload, for example parsing the detail of custom events in EventBridge, or body of SQS records

You can either solve these situations by creating a model of these known structures, parsing them, then extracting and parsing a key where your payload is.

This can become difficult quite quickly. Parser makes this problem easier through a feature named Envelope.

Envelopes can be used via envelope parameter available in both parse function and event_parser decorator.

Here's an example of parsing a model found in an event coming from EventBridge, where all you want is what's inside the detail key.

from aws_lambda_powertools.utilities.parser import event_parser, parse, BaseModel, envelopes
from aws_lambda_powertools.utilities.typing import LambdaContext

class UserModel(BaseModel):
    username: str
    password1: str
    password2: str

payload = {
    "version": "0",
    "id": "6a7e8feb-b491-4cf7-a9f1-bf3703467718",
    "detail-type": "CustomerSignedUp",
    "source": "CustomerService",
    "account": "111122223333",
    "time": "2020-10-22T18:43:48Z",
    "region": "us-west-1",
    "resources": ["some_additional_"],
    "detail": {
        "username": "universe",
        "password1": "myp@ssword",
        "password2": "repeat password"
    }
}

ret = parse(model=UserModel, envelope=envelopes.EventBridgeEnvelope, event=payload)

# Parsed model only contains our actual model, not the entire EventBridge + Payload parsed
assert ret.password1 == ret.password2

# Same behaviour but using our decorator
@event_parser(model=UserModel, envelope=envelopes.EventBridgeEnvelope)
def handler(event: UserModel, context: LambdaContext):
    assert event.password1 == event.password2

What's going on here, you might ask:

1. We imported built-in envelopes from the parser utility
2. Used envelopes.EventBridgeEnvelope as the envelope for our UserModel model
3. Parser parsed the original event against the EventBridge model
4. Parser then parsed the detail key using UserModel

Built-in envelopes

Parser comes with the following built-in envelopes, where Model in the return section is your given model.

Envelope name	Behaviour	Return
DynamoDBStreamEnvelope	1. Parses data using DynamoDBStreamModel. 2. Parses records in NewImage and OldImage keys using your model. 3. Returns a list with a dictionary containing NewImage and OldImage keys	List[Dict[str, Optional[Model]]]
EventBridgeEnvelope	1. Parses data using EventBridgeModel. 2. Parses detail key using your model and returns it.	Model
SqsEnvelope	1. Parses data using SqsModel. 2. Parses records in body key using your model and return them in a list.	List[Model]
CloudWatchLogsEnvelope	1. Parses data using CloudwatchLogsModel which will base64 decode and decompress it. 2. Parses records in message key using your model and return them in a list.	List[Model]
KinesisDataStreamEnvelope	1. Parses data using KinesisDataStreamModel which will base64 decode it. 2. Parses records in in Records key using your model and returns them in a list.	List[Model]
SnsEnvelope	1. Parses data using SnsModel. 2. Parses records in body key using your model and return them in a list.	List[Model]
SnsSqsEnvelope	1. Parses data using SqsModel. 2. Parses SNS records in body key using SnsNotificationModel. 3. Parses data in Message key using your model and return them in a list.	List[Model]
ApiGatewayEnvelope	1. Parses data using APIGatewayProxyEventModel. 2. Parses body key using your model and returns it.	Model
ApiGatewayV2Envelope	1. Parses data using APIGatewayProxyEventV2Model. 2. Parses body key using your model and returns it.	Model

Bringing your own envelope

You can create your own Envelope model and logic by inheriting from BaseEnvelope, and implementing the parse method.

Here's a snippet of how the EventBridge envelope we demonstrated previously is implemented.

EventBridge ModelEventBridge Envelope

from datetime import datetime
from typing import Any, Dict, List

from aws_lambda_powertools.utilities.parser import BaseModel, Field


class EventBridgeModel(BaseModel):
    version: str
    id: str  # noqa: A003,VNE003
    source: str
    account: str
    time: datetime
    region: str
    resources: List[str]
    detail_type: str = Field(None, alias="detail-type")
    detail: Dict[str, Any]

from aws_lambda_powertools.utilities.parser import BaseEnvelope, models
from aws_lambda_powertools.utilities.parser.models import EventBridgeModel

from typing import Any, Dict, Optional, TypeVar

Model = TypeVar("Model", bound=BaseModel)

class EventBridgeEnvelope(BaseEnvelope):

    def parse(self, data: Optional[Union[Dict[str, Any], Any]], model: Model) -> Optional[Model]:
        """Parses data found with model provided

        Parameters
        ----------
        data : Dict
            Lambda event to be parsed
        model : Model
            Data model provided to parse after extracting data using envelope

        Returns
        -------
        Any
            Parsed detail payload with model provided
        """
        parsed_envelope = EventBridgeModel.parse_obj(data)
        return self._parse(data=parsed_envelope.detail, model=model)

What's going on here, you might ask:

1. We defined an envelope named EventBridgeEnvelope inheriting from BaseEnvelope
2. Implemented the parse abstract method taking data and model as parameters
3. Then, we parsed the incoming data with our envelope to confirm it matches EventBridge's structure defined in EventBridgeModel
4. Lastly, we call _parse from BaseEnvelope to parse the data in our envelope (.detail) using the customer model

Data model validation

Warning

This is radically different from the Validator utility which validates events against JSON Schema.

You can use parser's validator for deep inspection of object values and complex relationships.

There are two types of class method decorators you can use:

• validator - Useful to quickly validate an individual field and its value
• root_validator - Useful to validate the entire model's data

Keep the following in mind regardless of which decorator you end up using it:

• You must raise either ValueError, TypeError, or AssertionError when value is not compliant
• You must return the value(s) itself if compliant

validating fields

Quick validation to verify whether the field message has the value of hello world.

from aws_lambda_powertools.utilities.parser import parse, BaseModel, validator

class HelloWorldModel(BaseModel):
    message: str

    @validator('message')
    def is_hello_world(cls, v):
        if v != "hello world":
            raise ValueError("Message must be hello world!")
        return v

parse(model=HelloWorldModel, event={"message": "hello universe"})

If you run as-is, you should expect the following error with the message we provided in our exception:

message
  Message must be hello world! (type=value_error)

Alternatively, you can pass '*' as an argument for the decorator so that you can validate every value available.

from aws_lambda_powertools.utilities.parser import parse, BaseModel, validator

class HelloWorldModel(BaseModel):
    message: str
    sender: str

    @validator('*')
    def has_whitespace(cls, v):
        if ' ' not in v:
            raise ValueError("Must have whitespace...")

        return v

parse(model=HelloWorldModel, event={"message": "hello universe", "sender": "universe"})

validating entire model

root_validator can help when you have a complex validation mechanism. For example finding whether data has been omitted, comparing field values, etc.

from aws_lambda_powertools.utilities.parser import parse, BaseModel, root_validator

class UserModel(BaseModel):
    username: str
    password1: str
    password2: str

    @root_validator
    def check_passwords_match(cls, values):
        pw1, pw2 = values.get('password1'), values.get('password2')
        if pw1 is not None and pw2 is not None and pw1 != pw2:
            raise ValueError('passwords do not match')
        return values

payload = {
    "username": "universe",
    "password1": "myp@ssword",
    "password2": "repeat password"
}

parse(model=UserModel, event=payload)

Info

You can read more about validating list items, reusing validators, validating raw inputs, and a lot more in Pydantic's documentation.

Advanced use cases

Tip: Looking to auto-generate models from JSON, YAML, JSON Schemas, OpenApi, etc?

Use Koudai Aono's data model code generation tool for Pydantic

There are number of advanced use cases well documented in Pydantic's doc such as creating immutable models, declaring fields with dynamic values) e.g. UUID, and helper functions to parse models from files, str, etc.

Two possible unknown use cases are Models and exception' serialization. Models have methods to export them as dict, JSON, JSON Schema, and Validation exceptions can be exported as JSON.

from aws_lambda_powertools.utilities import Logger
from aws_lambda_powertools.utilities.parser import parse, BaseModel, ValidationError, validator

logger = Logger(service="user")

class UserModel(BaseModel):
    username: str
    password1: str
    password2: str

payload = {
    "username": "universe",
    "password1": "myp@ssword",
    "password2": "repeat password"
}

def my_function():
    try:
        return parse(model=UserModel, event=payload)
    except ValidationError as e:
        logger.exception(e.json())
        return {
            "status_code": 400,
            "message": "Invalid username"
        }

User: UserModel = my_function()
user_dict = User.dict()
user_json = User.json()
user_json_schema_as_dict = User.schema()
user_json_schema_as_json = User.schema_json(indent=2)

These can be quite useful when manipulating models that later need to be serialized as inputs for services like DynamoDB, EventBridge, etc.

FAQ

When should I use parser vs data_classes utility?

Use data classes utility when you're after autocomplete, self-documented attributes and helpers to extract data from common event sources.

Parser is best suited for those looking for a trade-off between defining their models for deep validation, parsing and autocomplete for an additional dependency to be brought in.

How do I import X from Pydantic?

We export most common classes, exceptions, and utilities from Pydantic as part of parser e.g. from aws_lambda_powertools.utilities.parser import BaseModel.

If what's your trying to use isn't available as part of the high level import system, use the following escape hatch mechanism:

from aws_lambda_powertools.utilities.parser.pydantic import <what you'd like to import'>

What is the cold start impact in bringing this additional dependency?

No significant cold start impact. It does increase the final uncompressed package by 71M, when you bring the additional dependency that parser requires.

Artillery load test sample against a hello world sample using Tracer, Metrics, and Logger with and without parser.

No parser

Info

Uncompressed package size: 55M, p99: 180.3ms

Summary report @ 14:36:07(+0200) 2020-10-23
Scenarios launched:  10
Scenarios completed: 10
Requests completed:  2000
Mean response/sec: 114.81
Response time (msec):
    min: 54.9
    max: 1684.9
    median: 68
    p95: 109.1
    p99: 180.3
Scenario counts:
    0: 10 (100%)
Codes:
    200: 2000

With parser

Info

Uncompressed package size: 128M, p99: 193.1ms

Summary report @ 14:29:23(+0200) 2020-10-23
Scenarios launched:  10
Scenarios completed: 10
Requests completed:  2000
Mean response/sec: 111.67
Response time (msec):
    min: 54.3
    max: 1887.2
    median: 66.1
    p95: 113.3
    p99: 193.1
Scenario counts:
    0: 10 (100%)
Codes:
    200: 2000

---

Last update: 2022-04-12