Introduction to Jupyter Server

Jupyter Server extends Jupyter Notebook functionalities by offering additional backend services and enabling richer API experiences. It allows developers to create custom, interactive computing environments tailored to their workflows. In this article, we delve into some of the most useful APIs provided by Jupyter Server, complete with example code snippets.

Useful Jupyter Server APIs

1. Listing Available Kernels

Retrieving all available kernels:

  
  import requests

  response = requests.get('http://localhost:8888/api/kernelspecs')
  print(response.json())
  

2. Creating a New Kernel

Starting a new kernel instance:

  
  import requests

  data = {'name': 'python3'}
  response = requests.post('http://localhost:8888/api/kernels', json=data)
  print(response.json())
  

3. Interrupting a Kernel

Interrupting an existing kernel:

  
  import requests

  kernel_id = 'your_kernel_id'
  response = requests.post(f'http://localhost:8888/api/kernels/{kernel_id}/interrupt')
  print(response.status_code)
  

4. Executing Code

Running code within a specific kernel:

  
  import requests

  kernel_id = 'your_kernel_id'
  headers = {'Content-Type': 'application/json'}
  code = {'code': 'print("Hello, world!")'}
  response = requests.post(f'http://localhost:8888/api/kernels/{kernel_id}/execute', headers=headers, json=code)
  print(response.json())
  

5. Shutting Down a Kernel

Gracefully terminating an existing kernel:

  
  import requests

  kernel_id = 'your_kernel_id'
  response = requests.delete(f'http://localhost:8888/api/kernels/{kernel_id}')
  print(response.status_code)
  

Application Example: Creating a Custom Data Analysis Server

This example demonstrates how to utilize the introduced APIs to create a custom server for running data analysis jobs.

  
  from flask import Flask, request, jsonify
  import requests

  app = Flask(__name__)
  kernel_id = None

  @app.route('/start', methods=['POST'])
  def start_kernel():
      global kernel_id
      response = requests.post('http://localhost:8888/api/kernels', json={'name': 'python3'})
      kernel_id = response.json()['id']
      return jsonify(response.json())

  @app.route('/execute', methods=['POST'])
  def execute_code():
      global kernel_id
      code = request.json['code']
      headers = {'Content-Type': 'application/json'}
      response = requests.post(f'http://localhost:8888/api/kernels/{kernel_id}/execute', headers=headers, json={'code': code})
      return jsonify(response.json())

  @app.route('/stop', methods=['POST'])
  def stop_kernel():
      global kernel_id
      response = requests.delete(f'http://localhost:8888/api/kernels/{kernel_id}')
      kernel_id = None
      return jsonify({'status': response.status_code})
  
  if __name__ == '__main__':
      app.run(debug=True)
  

In this application, we use Flask to create endpoints for starting, executing code in, and stopping a Jupyter kernel.

Note: Always ensure that Jupyter Server is running before executing these endpoints.

Hash: f704eece60335757bb0b824f41783e8c4761953cd77aba0552e4a6d2e10d112f