如何从外部托管的Jupyter Notebook导出数据？

DevOps 小满

2026-5-20

Alright, let's tackle how you can get your data out of that remote Jupyter Notebook setup—since you can't SSH into the Linux server directly but have Python access to the file system, there are several practical ways to do this:

1. Save to standard file formats and use platform file downloads

Most hosted Jupyter environments come with a built-in file browser or download option. First, export your data to a widely supported format like CSV, JSON, or Parquet:

For a Pandas DataFrame:

import pandas as pd
# Replace `your_dataframe` with your actual data object
your_dataframe.to_csv('exported_data.csv', index=False)

For Python dictionaries/lists:

import json
# Replace `your_data` with your actual data structure
with open('exported_data.json', 'w') as f:
    json.dump(your_data, f, indent=2)

After running this, confirm the file exists by executing !ls (this runs a Linux command directly in the Notebook). Then look for a file management panel in your Jupyter interface—you should be able to select the saved file and download it to your local Windows machine.

2. Print and copy-paste (for small datasets)

If your dataset isn't too large, you can convert it to a human-readable string, print it, and manually copy the output to your local machine:

For a DataFrame:

print(your_dataframe.to_csv(index=False))

For structured data:

import json
print(json.dumps(your_data, indent=2))

Just highlight the printed text, copy it, and paste it into a local text editor (like Notepad or VS Code), then save it with the appropriate file extension (.csv, .json). This is quick but not ideal for large datasets—it'll slow down your Notebook and be tedious to copy.

3. Send data to a local HTTP server (if outbound requests are allowed)

If the remote server can make external HTTP requests, you can set up a simple local server to receive the data:

First, open a terminal on your local Windows machine and run:

python -m http.server 8000 --bind 0.0.0.0

Then, in your remote Notebook, use requests to send the data to your local machine (replace your-local-ip with your machine's public IP, or local network IP if the server is on the same LAN):

import requests
import pandas as pd

# Convert DataFrame to JSON
data_json = your_dataframe.to_json(orient='records')
response = requests.post('http://your-local-ip:8000/receive_data', json=data_json)

# Alternatively, send raw CSV text
csv_text = your_dataframe.to_csv(index=False)
response = requests.post('http://your-local-ip:8000/save_csv', data=csv_text)

You'll need to handle the incoming request on your local server (you can write a simple custom script instead of using the default http.server if you want automatic saving). Note: Some hosted platforms block outbound requests, so test first with requests.get('https://httpbin.org/get') to confirm connectivity.

4. Use cloud storage as a middleman (if you can install packages)

If you can install third-party libraries, upload your data to a cloud storage bucket (like AWS S3, Azure Blob Storage, or Google Cloud Storage) then download it locally:

Example with AWS S3 (install boto3 first if needed: !pip install boto3):

import boto3
from io import StringIO
import pandas as pd

# Initialize the S3 client (use environment variables or your platform's secret manager to store credentials—never hardcode them!)
s3 = boto3.client('s3')

# Save DataFrame to an in-memory buffer
csv_buffer = StringIO()
your_dataframe.to_csv(csv_buffer, index=False)

# Upload to your S3 bucket
s3.put_object(Bucket='your-bucket-name', Key='exported_data.csv', Body=csv_buffer.getvalue())

Once uploaded, you can use your cloud provider's CLI or desktop app to download the file to your Windows machine. This works great for large datasets and is secure if you manage your credentials properly.

内容的提问来源于stack exchange，提问作者Phil-ZXX