Name	Name	Last commit message	Last commit date
parent directory ..
bin	bin
client_app	client_app
cypress	cypress
docs/images	docs/images
lib	lib
.gitignore	.gitignore
.npmignore	.npmignore
README.md	README.md
after_destroy_cleanup.py	after_destroy_cleanup.py
cdk.json	cdk.json
check_appsync.py	check_appsync.py
cypress.config.ts	cypress.config.ts
empty_buckets.py	empty_buckets.py
jest.config.js	jest.config.js
package-lock.json	package-lock.json
package.json	package.json
setup_client_app.py	setup_client_app.py
tsconfig.json	tsconfig.json

Document Explorer

Overview

The "Document Explorer" sample generative AI application demonstrates how to build end-to-end solutions leveraging AWS services and AWS Generative AI CDK Constructs.

It includes examples of key components needed in generative AI applications:

Data Ingestion Pipeline: Ingests documents, converts them to text, and stores them in a knowledge base for retrieval. This enables long context window approaches.
Document Summarization: Summarizes PDF documents leveraging Large Language Models like Anthropic Claude V2 via Amazon Bedrock.
Question Answering: Answers natural language questions by retrieving relevant documents from the knowledge base and leveraging Large Language Models like Anthropic Claude V2 via Amazon Bedrock.

By providing reusable constructs following AWS best practices, this app helps you quickly build custom generative AI apps on AWS. The constructs abstract complexity of orchestrating AWS services like Lambda, OpenSearch, Step Functions, Bedrock, etc.

Here is the architecture diagram of the sample application:

Folder Structure

This Document Explorer codebase is organized into folders containing the frontend and backend infrastructure code. The frontend client app is built with Streamlit and is located in the client_app folder. The backend code lives in bin/document_explorer.ts and uses the AWS CDK resources defined in the lib folder.

The key folders are:

samples/document_explorer
├── client_app                                   # Frontend using Python Streamlit
│   │
│   ├── Home.ts                                  # Sample app entry point
│   ├── assets/                                  # Static files
│   ├── common/                                  # Utility classes
│   ├── graphql/                                 # GraphQL statements and client
│   └── pages/                                   # Streamlit pages for document selection, summarization, and QA
│
├── bin
│   └── document_explorer.ts                     # Backend - CDK app
├── lib                                          # CDK Stacks
│   ├── networking-stack.ts                      # VPC resources
│   ├── persistence-stack.ts                     # S3 and OpenSearch resources
│   └── api-stack.ts                             # Cognito, AppSync, and generative-ai constructs
│
└── [email protected] # Local copy of generative-ai CDK constructs

Getting started

To deploy this Document Explorer, follow these steps to set up the required tools and configure your AWS environment:

Prerequisites

An AWS account. We recommend you deploy this solution in a new account.
AWS CLI: configure your credentials

aws configure --profile [your-profile] 
AWS Access Key ID [None]: xxxxxx
AWS Secret Access Key [None]:yyyyyyyyyy
Default region name [None]: us-east-1 
Default output format [None]: json

Node.js: v18.12.1
AWS CDK: 2.68.0
jq: jq-1.6

Deploy the solution

This project is built using the AWS Cloud Development Kit (CDK). See Getting Started With the AWS CDK for additional details and prerequisites.

Clone this repository.
```
git clone <this>
```
Enter the code sample backend directory.
```
cd samples/document_explorer
```
Install packages
```
npm install
```
Boostrap AWS CDK resources on the AWS account.
```
cdk bootstrap aws://ACCOUNT_ID/REGION
```
The persistence layer requires the existence of the AWSServiceRoleForAmazonElasticsearchService Service-Linked Role (SLR). The following command checks if the SLR exists and creates one if needed:
```
if ! aws iam get-role --role-name AWSServiceRoleForAmazonElasticsearchService > /dev/null 2>&1; then
aws iam create-service-linked-role --aws-service-name es.amazonaws.com 
fi
```
Enable Access to Amazon Bedrock Models

You must explicitly enable access to models before they can be used with the Amazon Bedrock service. Please follow these steps in the Amazon Bedrock User Guide to enable access to the models (at minimum, Anthropic::Claude):.

Deploy the sample in your account.
```
$ cdk deploy --all
```

The command above will deploy three stacks in your account. Some services require a certain amount of time to get provisioned, here are some values observed:

Networking stack: ~181s (~3 minutes)
Persistence stack: ~1137s (~19 minutes)
Api stack: ~531 seconds (~9 minutes)

Between each stack, to protect you against unintended changes that affect your security posture, the AWS CDK Toolkit prompts you to approve security-related changes before deploying them. You will need to answer yes at each step to get all the stacks deployed.

Note If your deployment fails with the message: MemorySize value failed to satisfy constraint: Member must have value less than or equal to 3008 then your AWS Account is likely either too new or unused for the region, and there are these choices to fix:

Contact AWS Support

to increase your AWS Lambda maximum memory size limits to 10Gb. This will not happen immediately, but is the recommended option.

Deploy the CDK samples with the optional cdk deploy --all --outputs-file apistack-outputs.json --context maximumLambdaMemorySize=3008 parameter. This method is available without waiting, but will limit functionality and cause unexpected failures when more memory than is available is required.

Configure client_app

cd client_app
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Still within the /client_app directory, create an .env file with the following content or mutate the .env-example. Replace the property values with the values retrieved from the stack outputs/console.

COGNITO_DOMAIN="<ApiStack.CognitoDomain>"
REGION="<ApiStack.Region>"
USER_POOL_ID="<ApiStack.UserPoolId>"
CLIENT_ID="<ApiStack.ClientId>"
CLIENT_SECRET="COGNITO_CLIENT_SECRET"
IDENTITY_POOL_ID="<ApiStack.IdentityPoolId>"
APP_URI="http://localhost:8501/"
AUTHENTICATED_ROLE_ARN="<ApiStack.AuthenticatedRoleArn>"
GRAPHQL_ENDPOINT = "<ApiStack.GraphQLEndpoint>"
S3_INPUT_BUCKET = "<PersistenceStack.InputsAssetsBucket>"
S3_PROCESSED_BUCKET = "<PersistenceStack.processedAssetsBucket>"

Note: The COGNITO_CLIENT_SECRET is a secret value that can be retrieved from the AWS Console. Go to the Amazon Cognito page in the AWS console, then select the created user pool. Under App integration, select App client settings. Then, select Show Details and copy the value of the App client secret.

Run client_app
```
streamlit run Home.py
```

Test

Create a user in the Cognito user pool. Go to the Amazon Cognito page in the AWS console, then select the created user pool. Under users, select Create user and fill in the form
Access the webapp (either locally or through the Amplify hosted domain) and sign in using the user credentials you just created
Upload sample PDF files to the input bucket. For example, download Amazon's Annual Letters to Shareholders from 1997-2022 from ir.aboutamazon.com. Then:

Step 01. Test document ingestion

Subscription (Optional - to track completion)

subscription UpdateIngestionJobStatus {
  updateIngestionJobStatus(ingestionjobid: "1997-2022") {
    files {
      name
      status
    }
  }
}

Mutation

mutation IngestDocuments {
  ingestDocuments(
    ingestioninput: {
      files: [
        {status: "", name: "1997 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "1998 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "1999 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2000 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2001 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2002 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2003 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2004 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2005 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2006 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2007 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2008 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2009 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2010 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2011 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2012 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2013 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2014 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2015 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2016 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2017 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2018 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2019 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2020 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2021 Amazon Shareholder Letter.pdf"}, 
        {status: "", name: "2022 Amazon Shareholder Letter.pdf"}, 
      ], 
      ingestionjobid: "1997-2022"}
    ) {
    __typename
  }
}

Step 02. Run summarization

Subscription

subscription UpdateSummaryJobStatus {
  updateSummaryJobStatus(summary_job_id: "2022_Amazon_Shareholder_Letter") {
    summary_job_id
    files {
      name
      status
      summary
    }
  }
}

Mutation

mutation GenerateSummary {
    generateSummary(summaryInput: {
        summary_job_id: "2022_Amazon_Shareholder_Letter",
        files: [{name: "2022 Amazon Shareholder Letter.txt"}],
    }) {
    __typename
  }
}

Step 03. Ask question

Subscription

subscription UpdateQAJobStatus {
  updateQAJobStatus(jobid: "11a94ffc-423a-4157-a9c2-892446f9a1fe") {
    question
    answer
    jobstatus
  }
}

Post Question

mutation PostQuestion {
    postQuestion(
        jobid: "11a94ffc-423a-4157-a9c2-892446f9a1fe"
        jobstatus: ""
        filename: "2022 Amazon Shareholder Letter.txt"
        question: "V2hvIGlzIEJlem9zPw=="
        max_docs: 1
        verbose: false
    ) {
    __typename
    }
}

Clean up

Do not forget to delete the stack to avoid unexpected charges.

First make sure to remove all data from the Amazon Simple Storage Service (Amazon S3) Buckets. Then:

    $ cdk destroy --all

Then in the AWS Console delete the S3 buckets.

Deployment Options

This application implements search capabilities using Amazon OpenSearch Serverless by default. However, you can configure it to use Amazon OpenSearch Service instead. Review the developer guide for help choosing between Serverless and Service.

The key differences when using OpenSearch Service are:

Networking Stack

Set openSearchServiceType to 'es' instead of 'aoss'

Persistence Stack

Set openSearchServiceType to 'es'
Pass OpenSearchServiceProps instead of OpenSearchServerlessProps
Configure domain properties like instance types, number of nodes, storage, etc.
Create an OpenSearch domain instead of a collection

API Stack

Reference the OpenSearch domain instead of the collection

Here is an example bin/document_explorer.ts configured for OpenSearch Service:

// ...

//-----------------------------------------------------------------------------
// Networking Layer
//-----------------------------------------------------------------------------
const network = new NetworkingStack(app, 'NetworkingStack', {
  env: env,
  openSearchServiceType: 'es',
});
cdk.Tags.of(network).add("stacl", "network");

//-----------------------------------------------------------------------------
// Persistence Layer
//-----------------------------------------------------------------------------
const persistence = new PersistenceStack(app, 'PersistenceStack', {
  env: env,
  vpc: network.vpc,
  securityGroups: network.securityGroups,
  openSearchServiceType: 'es',
  openSearchProps: {
    masterNodes: 3,
    dataNodes: 3,
    masterNodeInstanceType: 'm6g.large.search',
    dataNodeInstanceType: 'm6g.large.search',
    availabilityZoneCount: 3,
    volumeSize: 20,
  } as OpenSearchServiceProps,
  removalPolicy: cdk.RemovalPolicy.DESTROY  
});
cdk.Tags.of(persistence).add("stack", "persistence");

//-----------------------------------------------------------------------------
// API Layer
//-----------------------------------------------------------------------------
const api = new ApiStack(app, 'ApiStack', {
  env: env,
  description: '(uksb-1tupboc43) API Layer stack',
  existingOpensearchDomain: persistence.opensearchDomain,
  existingVpc: network.vpc,
  existingSecurityGroup: network.securityGroups[0],
  existingInputAssetsBucketObj: persistence.inputAssetsBucket,
  existingProcessedAssetsBucketObj: persistence.processedAssetsBucket,
  openSearchIndexName: 'doc-explorer',
  cacheNodeType: 'cache.r6g.xlarge',
  engine: 'redis',
  numCacheNodes: 1,
  removalPolicy: cdk.RemovalPolicy.DESTROY,
  clientUrl: 'http://localhost:8501/'
});
cdk.Tags.of(api).add("stack", "api");

// ...

Content Security Legal Disclaimer

The sample code; software libraries; command line tools; proofs of concept; templates; or other related technology (including any of the foregoing that are provided by our personnel) is provided to you as AWS Content under the AWS Customer Agreement, or the relevant written agreement between you and AWS (whichever applies). You should not use this AWS Content in your production accounts, or on production or other critical data. You are responsible for testing, securing, and optimizing the AWS Content, such as sample code, as appropriate for production grade use based on your specific quality control practices and standards. Deploying AWS Content may incur AWS charges for creating or using AWS chargeable resources, such as running Amazon EC2 instances or using Amazon S3 storage.

Operational Metrics Collection

This solution collects anonymous operational metrics to help AWS improve the quality and features of the solution. Data collection is subject to the AWS Privacy Policy (https://aws.amazon.com/privacy/). To opt out of this feature, simply remove the tag(s) starting with “uksb-” or “SO” from the description(s) in any CloudFormation templates or CDK TemplateOptions.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

document_explorer

document_explorer

README.md

Document Explorer

Overview

Folder Structure

Getting started

Prerequisites

Deploy the solution

Test

Step 01. Test document ingestion

Step 02. Run summarization

Step 03. Ask question

Clean up

Deployment Options

Networking Stack

Persistence Stack

API Stack

Content Security Legal Disclaimer

Operational Metrics Collection

Files

document_explorer

Directory actions

More options

Directory actions

More options

Latest commit

History

document_explorer

Folders and files

parent directory

README.md

Document Explorer

Overview

Folder Structure

Getting started

Prerequisites

Deploy the solution

Test

Step 01. Test document ingestion

Step 02. Run summarization

Step 03. Ask question

Clean up

Deployment Options

Networking Stack

Persistence Stack

API Stack

Content Security Legal Disclaimer

Operational Metrics Collection