# Wikipedia dataset

Wikipedia dataset from HTML.

## Process and load the dataset

Process and load the dataset using streaming:

```python
from datasets import load_dataset

ds = load_dataset(
    "wikimedia/wikipedia",
    revision="script-html",
    language="ca",
    split="train",
    trust_remote_code=True,
    storage_options={"https": {"client_kwargs": {"headers": {"Authorization": f"Bearer {auth['access_token']}"}}}},
    streaming=True,
)
item = next(iter(ds))
item
```

Note that you need to authenticate first. See Authentication section below.

## Authentication

Wikimedia Enterprise API requires authentication credentials. See information here:
https://enterprise.wikimedia.com/docs/#getting-started

1. If you do not already have an account, signup: https://enterprise.wikimedia.com/signup/

2. Write your Wikimedia Enterprise username and password in the `.env` file:

    ```
    WME_USERNAME=<YOUR-USERNAME>
    WME_PASSWORD=<YOUR-PASWORD>
    ```

3. Get your authentication tokens:

    ```python
    import requests
    from dotenv import dotenv_values
    
    
    config = dotenv_values(".env")
    
    r = requests.post(
        "https://auth.enterprise.wikimedia.com/v1/login",
        headers={"Content-Type": "application/json"},
        json={
            "username": config["WME_USERNAME"],
            "password": config["WME_PASSWORD"],
        }
    )
    auth = r.json()
    ```
   
4. Now you can use your access token (`auth["access_token"]`) to authenticate:

    ```python
    storage_options={
        "https": {
            "client_kwargs": {
                "headers": {
                    "Authorization": f"Bearer {auth['access_token']}"
                }
            }
        }
    }
    ```