Integrate Box with LlamaIndex for RAG
LlamaIndex readers that bring Box.com files, text, and Box AI extraction into RAG pipelines.
Why it matters
Empower your LLM applications by seamlessly integrating Box.com documents into LlamaIndex for advanced Retrieval Augmented Generation (RAG) and data querying.
Outcomes
What it gets done
Read files directly from Box.com
Extract text using Box's native text representation
Leverage Box AI for advanced text and structured data extraction (E+ customers)
Index Box content for efficient RAG queries
Install
Add it to your toolbox
Run in your project directory:
curl -fsSL https://spark.entire.vc/get/li-reader-readers-box | bash Overview
LlamaIndex: Box Readers
The Box Readers bring Box.com content into LlamaIndex through four readers -- plain file reading, text extraction, and two Box AI-powered readers for context prompting and structured-data extraction (E+ plan required) -- authenticated via CCG, JWT, OAuth 2.0, or a developer token. Use it when you need Box.com files, or AI-extracted content from them, loaded into LlamaIndex for RAG. The two Box AI readers require an E+ Box plan; the plain readers do not.
What it does
This integration brings Box.com's capabilities to LlamaIndex for RAG and other LLM applications. It provides four readers: Box Reader (implements the SimpleReader interface to read files from Box), Box Text Extraction (uses Box's text representation to extract text from a document), Box AI Prompt (uses Box AI to extract context from documents), and Box AI Extraction (uses Box AI to extract structured data from documents). Box AI features are limited to Box's Enterprise Plus (E+) customers.
When to use - and when NOT to
Use it when you need to read files from Box into LlamaIndex documents, optionally using Box AI to extract context or structured data from them for RAG pipelines. The Box AI readers specifically require an E+ Box plan, so those two readers are not usable on lower-tier Box accounts -- the plain Box Reader and Text Extraction reader have no such restriction.
Inputs and outputs
Install with:
pip install llama-index-readers-box
To work with any Box reader you first need a BoxClient, authenticated via one of four methods: Client Credential Grant (CCG), JSON Web Tokens (JWT), OAuth 2.0, or a developer token -- CCG and JWT are the two walked through in setup detail. For CCG, set BOX_CLIENT_ID, BOX_CLIENT_SECRET, BOX_ENTERPRISE_ID, and optionally BOX_USER_ID in a .env file, then:
from box_sdk_gen import CCGConfig, BoxCCGAuth, BoxClient
config = CCGConfig(
client_id="your_client_id",
client_secret="your_client_secret",
enterprise_id="your_enterprise_id",
user_id="your_ccg_user_id", # Optional
)
auth = BoxCCGAuth(config)
if config.user_id:
auth.with_user_subject(config.user_id)
client = BoxClient(auth)
reader = BoxReader(box_client=client)
The CCG snippet above builds a CCGConfig from the client ID, client secret, and enterprise ID, wraps it in a BoxCCGAuth, and -- if a user_id was supplied -- calls auth.with_user_subject(config.user_id) before constructing the BoxClient and passing it to BoxReader(box_client=client); that with_user_subject call is what switches the client from the default service account over to acting as a specific user.
For JWT, set BOX_ENTERPRISE_ID, BOX_USER_ID, and JWT_CONFIG_PATH (pointing to a .config.json from the Box Developer Console), and install the extra dependency box-sdk-gen[jwt]. By default both CCG and JWT use a service account, which may not have access to every file depending on how they are shared -- specifying a different user_id requires the application to be able to impersonate or generate tokens for that user.
Who it's for
Developers building RAG or other LLM applications on Box.com content, who need either straightforward file access or Box AI-powered context or structured-data extraction on an E+ plan.
Source README
LlamaIndex: Box Readers
This open-source integration brings the capabilities of Box.com to the LLama-Index, empowering developers building Retrieval Augmented Generation (RAG) and other LLM applications.
This README will guide you through installation, usage, and explore the functionalities of each reader.
Installation
pip install llama-index-readers-box
Available readers
We provide multiple readers, including:
- Box Reader - Implementation of the SimpleReader interface to read files from Box.
- Box Text Extraction - Uses Box text representation to extract text from document.
- Box AI Prompt - Uses Box AI to extract context from documents
- Box AI Extraction - Uses Box AI to extract structured data from documents
Authentication
Client credential gran (CCG)
Create a new application in the Box Developer Console and generate a new client ID and client secret.
Create a .env file with the following content:
### CCG settings
BOX_CLIENT_ID = YOUR_CLIENT_ID
BOX_CLIENT_SECRET = YOUR_CLIENT_SECRET
### Common Settings
BOX_ENTERPRISE_ID = YOUR_BOX_ENTERPRISE_ID
BOX_USER_ID = YOUR_BOX_USER_ID (optional)
By default the CCG client will use a service account associated with the application. Depending on how the files are shared, the service account may not have access to all the files.
If you want to use a different user, you can specify the user ID in the .env file. In this case make sure your application can impersonate and/or generate user tokens in the scope.
Checkout this guide for more information on how to setup the CCG: Box CCG Guide
JSON web tokens (JWT)
Create a new application in the Box Developer Console and generate a new .config.json file.
Create a .env file with the following content:
### Common settings
BOX_ENTERPRISE_ID = 877840855
BOX_USER_ID = 18622116055
### JWT Settings
JWT_CONFIG_PATH = /path/to/your/.config.json
By default the JWT client will use a service account associated with the application. Depending on how the files are shared, the service account may not have access to all the files.
If you want to use a different user, you can specify the user ID in the .env file. In this case make sure your application can impersonate and/or generate user tokens in the scope.
Checkout this guide for more information on how to setup the JWT: Box JWT Guide
pip install "box-sdk-gen[jwt]"
Box Client
To work with the box readers, you will need to provide a Box Client.
The Box Client can be created using either the Client Credential Grant (CCG), JSON Web Tokens (JWT), OAuth 2.0, and developer token.
Using CCG authentication
from box_sdk_gen import CCGConfig, BoxCCGAuth, BoxClient
config = CCGConfig(
client_id="your_client_id",
client_secret="your_client_secret",
enterprise_id="your_enterprise_id",
user_id="your_ccg_user_id", # Optional
)
auth = BoxCCGAuth(config)
if config.user_id:
auth.with_user_subject(config.user_id)
client = BoxClient(auth)
reader = BoxReader(box_client=client)
Using JWT authentication
from box_sdk_gen import JWTConfig, BoxJWTAuth, BoxClient
#### Using manual configuration
config = JWTConfig(
client_id="YOUR_BOX_CLIENT_ID",
client_secret="YOUR_BOX_CLIENT_SECRET",
jwt_key_id="YOUR_BOX_JWT_KEY_ID",
private_key="YOUR_BOX_PRIVATE_KEY",
private_key_passphrase="YOUR_BOX_PRIVATE_KEY_PASSPHRASE",
enterprise_id="YOUR_BOX_EN
FAQ
Common questions
Discussion
Questions & comments · 0
Sign In Sign in to leave a comment.