Jina Proxy Server

A lightweight proxy server for making authenticated requests to the Jina AI API. This server helps handle CORS issues and securely manages API keys when making requests to Jina's services.

Features

• Proxies requests to Jina AI API
• Handles CORS
• Secure API key management
• JSON and text response handling
• Docker support

Prerequisites

• Node.js 18 or higher
• Docker (optional)
• Jina AI API key

Setup

1. Clone the repository
2. Create a .env file in the root directory with:

   JINA_API_KEY=your_api_key_here
   PORT=3000 # optional, defaults to 3000
   

3. Install dependencies:

   npm install
   

Running the Server

Local Development

npm start

Using Docker

Build and run with Docker:

npm run docker:start

Or build and run separately:

npm run docker:build
npm run docker:run

API Usage

DeepSearch Chat Endpoint

POST /deepsearch/chat

Handle complex queries with Jina's DeepSearch API supporting streaming responses and adjustable reasoning effort. The server automatically adds your Jina API key for authentication.

Example Request:

curl -X POST http://localhost:3000/deepsearch/chat \
  -H "Content-Type: application/json" \
  -d '{
    "messages": [
      {"role": "user", "content": "What's the latest blog post from Jina AI?"}
    ],
    "stream": true,
    "reasoning_effort": "medium"
  }'

Parameters:

• messages: Array of message objects with role (user/assistant) and content
• model: DeepSearch model version (default: jina-deepsearch-v1)
• stream: Enable Server-Sent Events (SSE) streaming (default: true)
• reasoning_effort: Processing intensity level (low/medium/high)

Response:

• Streaming (text/event-stream): Real-time SSE events with data chunks
• Non-streaming (application/json): Standard JSON response with complete answer

Read Endpoint

GET /read?url=<encoded-url>

Makes a proxied request to the Jina AI API. The Jina AI API can be used for various document processing tasks:

Common Use Cases:

1. Document Reading & Analysis

   • Extract text from PDFs, Word documents, and images
   • Parse structured data from documents
   • Convert documents to different formats

2. Web Scraping & Processing

   • Extract content from web pages
   • Process HTML content
   • Parse news articles and blog posts

3. Content Extraction

   • Extract tables from documents
   • Pull specific sections or paragraphs
   • Identify headers and metadata

4. Document Intelligence

   • Analyze document layout
   • Extract key-value pairs
   • Identify document types

Example Request:

# Extract content from a PDF
curl "http://localhost:3000/read?url=https://example.com/document.pdf"

# Process a webpage
curl "http://localhost:3000/read?url=https://news-article.com/story"

# Extract tables from a document
curl "http://localhost:3000/read?url=https://example.com/report.xlsx"

Response formats vary based on the document type and processing requested. The server handles both JSON and text responses appropriately.

Error Handling

The server provides appropriate error responses for:

• Missing URL parameters
• API authentication issues
• Invalid responses
• Server errors

Security

• API keys are stored securely in environment variables
• CORS is enabled for cross-origin requests
• Sensitive information is not exposed to clients

License

MIT