使用kaggle-cli执行kg download时遇'Response' object has no attribute 'soup'错误

云原生阿岚

2026-5-28

Fix for 'Response' object has no attribute 'soup' in kaggle-cli

I've run into this exact error with kaggle-cli before, and it usually boils down to either a compatibility issue with the latest version or a hiccup in how the tool parses Kaggle's response pages. Here are a few fixes you can try right away:

Try downloading a specific file first
Sometimes the bulk download logic in kaggle-cli breaks, but single-file downloads work. Test with one of the dataset files to rule out broader issues:
```
kg download -c instacart-market-basket-analysis -f orders.csv
```
If this works, you can iterate through all file names in the competition to download them one by one, or use a simple script to automate the batch download.
Re-authenticate your kaggle-cli session
Even if you logged in recently, your session might have expired or become corrupted. Re-run the authentication command to refresh it:
```
kg auth
```
Follow the prompts to log in again (you might need to enter your Kaggle username and password once more to re-establish a valid session).
Downgrade to a stable older version of kaggle-cli
The latest version might have introduced a bug with how it parses Kaggle's HTML responses. Roll back to a known-working version like 0.12.2:
```
pip uninstall -y kaggle-cli
pip install kaggle-cli==0.12.2
```
After downgrading, try your original bulk download command again.
Switch to the official Kaggle API (recommended)
Kaggle-cli is a third-party tool, and the official Kaggle API is more reliable and actively maintained. Here's how to make the switch:
1. Install the official API:
```
pip install kaggle
```
2. Go to your Kaggle account settings, download the kaggle.json API key file.
3. Place the file in ~/.kaggle/ (Linux/macOS) or C:\Users\<YourUsername>\.kaggle\ (Windows).
4. Run the download command to get all competition files in one zip:
```
kaggle competitions download -c instacart-market-basket-analysis
```
You can then extract the zip file to access all the dataset files.