AWS Step Functions, Intrinsic Functions In-Depth | Serverless
awedis
Posted on March 7, 2023
AWS Step Functions enables you to orchestrate serverless workflows by integrating with different AWS services.
📋 Note: If you want to read more about AWS Step Functions before starting this article you can read AWS Step Functions In-Depth | Serverless
The main parts of this article:
- About Intrinsic functions
- Examples
1. what are Intrinsic functions?
Intrinsic functions allows you to run directly on your state machine some simple processing.
Well, let's take a look at how we used to manipulate our data in Step Functions. In order to do so, we needed a state and a Lambda function which would take the data as input, update it, and make all the necessary changes before passing the data to the next state. However, upon closer examination, it became clear that our architecture required excessive resources and involved too many steps.
The Amazon States Language provides several intrinsic functions, also known as intrinsics, that can help you perform basic data processing operations without using a Task state. These intrinsics are constructs that resemble functions in programming languages and can assist payload builders in processing the data that goes to and from the Resource field of a Task state.
Using intrinsic functions can help you build applications where you don't have to rely on Lambda functions, which can save you time and money, reduce complexity (by consolidating logic in one place), minimize throughput, and reduce the payload between different tasks.
Examples
Now let's build simple examples and try all the options that we have on Intrinsic functions, the following are the different operations that we will cover
Intrinsics for arrays:
States.Array
,States.ArrayPartition
,States.ArrayContains
,States.ArrayRange
,States.ArrayGetItem
,States.ArrayLength
,States.ArrayUnique
Intrinsics for data encoding and decoding:
States.Base64Encode
,States.Base64Decode
Intrinsic for hash calculation:
States.Hash
Intrinsics for JSON data manipulation:
States.JsonMerge
,States.StringToJson
,States.JsonToString
Intrinsics for Math operations:
States.MathRandom
,States.MathAdd
Intrinsic for String operation:
States.StringSplit
Intrinsic for unique identifier generation:
States.UUID
Intrinsic for generic operation:
States.Format
Before adding intrinsic functions, let's first build our Step Functions. To keep it simple, we will have only two states, and a successful return at the end. Furthermore, our Lambda functions for each state will be simple, only logging out the data and passing the value to the next state.
firstLambdaARN: &FIRST_ARN arn:aws:lambda:${env:region}:${env:accountId}:function:${self:service}-${env:stage}-firstState
secondLambdaARN: &SECOND_ARN arn:aws:lambda:${env:region}:${env:accountId}:function:${self:service}-${env:stage}-secondState
states:
IntrinsicExample:
name: IntrinsicExample
definition:
Comment: "Intrinsic Functions Example"
StartAt: firstState
States:
firstState:
Type: Task
Resource: *FIRST_ARN
Next: secondState
secondState:
Type: Task
Resource: *SECOND_ARN
Next: success
success:
Type: Succeed
"use strict";
module.exports.firstState = async (event) => {
console.log('firstState event =>', event);
return event;
};
module.exports.secondState = async (event) => {
console.log('secondState event =>', event);
return {
status: 'SUCCESS'
};
};
Everything seems ready, let's start adding our intrinsic functions one by one and start seeing the results.
📋 Note: I'm not going with description of each intrinsic function, if you want more documentation how each one is working you can visit the link Intrinsic functions
To use intrinsic functions you must specify .$ in the key value in your state machine definitions.
You can nest up to 10 intrinsic functions within a field in your workflows.
Intrinsics for arrays
In the first one we are going to pass the whole payload as it is to input
also we are going to get the length. Inside the state we can see that Parameters
are now added
-
States.ArrayLength
firstLambdaARN: &FIRST_ARN arn:aws:lambda:${env:region}:${env:accountId}:function:${self:service}-${env:stage}-firstState
secondLambdaARN: &SECOND_ARN arn:aws:lambda:${env:region}:${env:accountId}:function:${self:service}-${env:stage}-secondState
states:
IntrinsicExample:
name: IntrinsicExample
definition:
Comment: "Intrinsic Functions Example"
StartAt: firstState
States:
firstState:
Type: Task
Resource: *FIRST_ARN
Parameters:
input.$: "$"
length.$: "States.ArrayLength($.inputArray)"
Next: secondState
secondState:
Type: Task
Resource: *SECOND_ARN
Next: success
success:
Type: Succeed
Perfect! Now we need to execute our Step Function. The easiest way to do so is by creating a simple function that can be triggered from an API call. We can then use the AWS SDK to run our Step Function.
"use strict";
const { StepFunctions } = require('aws-sdk');
const stepFunctions = new StepFunctions();
module.exports.runIntrinsicFunction = async (event) => {
try {
const stepFunctionResult = stepFunctions.startExecution({
stateMachineArn: process.env.INTRINSIC_EXAMPLE_STEP_FUNCTION_ARN,
input: JSON.stringify({
inputArray: [
{
id: 1,
title: 'book',
},
{
id: 2,
title: 'pen',
},
{
id: 3,
title: 'watch',
},
{
id: 4,
title: 'laptop',
}
]
})
}).promise();
console.log('stepFunctionResult =>', stepFunctionResult);
return {
statusCode: 200,
body: JSON.stringify({
message: `This is test API`,
}, null, 2),
};
} catch (error) {
console.log(error);
}
};
Now we can see that our first state takes the following input. It's amazing to see how powerful this is and how it can help minimize our code logic. With this new functionality, we can simplify our code and make it more efficient. Let's now move on to the other examples.
-
States.Array
,States.ArrayPartition
Payload:
input: JSON.stringify({
inputInteger: 12345678,
inputForPartition: [1, 2, 3, 4, 5, 6, 7, 8]
})
Intrinsic:
Parameters:
buildId.$: "States.Array($.inputInteger)"
partition.$: "States.ArrayPartition($.inputForPartition, 2)"
Result:
{
"partition": [
[
1,
2
],
[
3,
4
],
[
5,
6
],
[
7,
8
]
],
"BuildId": [
12345678
],
}
-
States.ArrayContains
,States.ArrayRange
Payload:
input: JSON.stringify({
inputArray: [1,2,3,4,5,6,7,8,9],
lookingFor: 8
})
Intrinsic:
Parameters:
containsResult.$: "States.ArrayContains($.inputArray, $.lookingFor)"
rangeResult.$: "States.ArrayRange(1, 20, 3)"
Result:
{
"containsResult": true,
"rangeResult": [
1,
4,
7,
10,
13,
16,
19
]
}
-
States.ArrayGetItem
,States.ArrayUnique
Payload:
input: JSON.stringify({
inputArray: [1,2,3,3,3,3,3,3,4],
index: 5,
})
Intrinsic:
Parameters:
getItemResult.$: "States.ArrayGetItem($.inputArray, $.index)"
uniqueResult.$: "States.ArrayUnique($.inputArray)"
Result:
{
"getItemResult": 3,
"uniqueResult": [
1,
2,
3,
4
]
}
Intrinsics for data encoding and decoding
-
States.Base64Encode
,States.Base64Decode
Payload
input: JSON.stringify({
input: "Data to encode",
base64: "RGF0YSB0byBlbmNvZGU="
})
Intrinsic:
Parameters:
encodeResult.$: "States.Base64Encode($.input)"
decodeResult.$: "States.Base64Decode($.base64)"
Result:
{
"decodeResult": "Data to encode",
"encodeResult": "RGF0YSB0byBlbmNvZGU="
}
Intrinsic for hash calculation
-
States.Hash
The hashing algorithm: MD5, SHA-1, SHA-256, SHA-384, SHA-512
Payload:
input: JSON.stringify({
Data: "input data",
Algorithm: "SHA-1"
})
Intrinsic:
Parameters:
result.$: "States.Hash($.Data, $.Algorithm)"
Result:
{
"result": "aaff4a450a104cd177d28d18d74485e8cae074b7"
}
Intrinsics for JSON data manipulation
-
States.JsonMerge
,States.StringToJson
,States.JsonToString
Payload:
input: JSON.stringify({
json1: { "a": {"a1": 1, "a2": 2}, "b": 2, },
json2: { "a": {"a3": 1, "a4": 2}, "c": 3 },
escapedJsonString: "{\"foo\": \"bar\"}",
unescapedJson: {
foo: "bar"
}
})
Intrinsic:
Parameters:
output.$: "States.JsonMerge($.json1, $.json2, false)"
toJsonResult: "States.StringToJson($.escapedJsonString)"
toStringResult: "States.JsonToString($.unescapedJson)"
Result:
{
"output": {
"a": {
"a3": 1,
"a4": 2
},
"b": 2,
"c": 3
},
"toJsonResult": {
"foo": "bar"
},
"toStringResult": "{\"foo\":\"bar\"}"
}
Intrinsics for Math operations
-
States.MathRandom
,States.MathAdd
Payload:
input: JSON.stringify({
start: 1,
end: 999,
value: 100,
step: 2
})
Intrinsic:
Parameters:
randomResult.$: "States.MathRandom($.start, $.end)"
additionResult.$: "States.MathAdd($.value, $.step)"
Result:
{
"randomResult": 110,
"additionResult": 102
}
Intrinsic for String operation
States.StringSplit
Payload:
input: JSON.stringify({
inputStringOne: "1,2,3,4,5",
splitterOne: ",",
inputStringTwo: "This.is+a,test=string",
splitterTwo: ".+,="
})
Intrinsic:
Parameters:
resultOne.$: "States.StringSplit($.inputStringOne, $.splitterOne)"
resultTwo.$: "States.StringSplit($.inputStringTwo, $.splitterTwo)"
Result:
{
"resultOne": [
"1",
"2",
"3",
"4",
"5"
],
"resultTwo": [
"This",
"is",
"a",
"test",
"string"
]
}
Intrinsic for generic operation
States.UUID
Intrinsic:
Parameters:
uuid.$: "States.UUID()"
Result:
{
"uuid": "c23c5e29-e9df-4507-8152-c677792511a4"
}
Intrinsic for generic operation
States.Format
Payload:
input: JSON.stringify({
template: "Hello, this is {}."
})
Intrinsic:
Parameters:
formatResult.$: "States.Format($.template, $.name)"
Result:
{
"formatResult": "Hello, this is Awedis."
}
Now lets see how we can nest some intrinsic functions together.
First I want to merge two objects together and then escape it.
Payload:
input: JSON.stringify({
json1: { id: '001', title: 'book' },
json2: { id: '001', category: 'products', label: 'business' }
})
Intrinsic:
Parameters:
time.$: "$$.State.EnteredTime"
output.$: "States.JsonToString(States.JsonMerge($.json1, $.json2, false))"
Result:
{
"output": "{\"id\":\"001\",\"label\":\"business\",\"title\":\"book\",\"category\":\"products\"}",
"time": "2023-03-05T13:17:18.682Z"
}
Conclusion
Using intrinsic functions can make your work easier since software engineers use these functions almost daily. Intrinsic functions can help minimize and simplify your code, making it much cleaner.
This article is part of the "Messaging with Serverless" series that I have been writing for a while. If you are interested in reading about other serverless offerings from AWS, feel free to visit the links below.
Posted on March 7, 2023
Join Our Newsletter. No Spam, Only the good stuff.
Sign up to receive the latest update from our blog.