Scrape and Analyze Bilibili’s “马保国” Videos with Python – A Complete Guide

The article presents a Python web‑scraping project targeting the Bilibili "马保国" channel, which hosts many meme videos.

The API endpoint used is:

https://api.bilibili.com/x/web-interface/web/channel/multiple/list?channel_id=3503796&sort_type=hot&page_size=30

The offset parameter is obtained from the JSON response of the previous request.

A concise scraping function is provided:

def get_data(url, headers):
    data_m = pd.DataFrame(columns=['id','name','view_count','like_count','duration','author_name','author_id','bvid'])
    html = requests.get(url, headers=headers).content
    data = json.loads(html.decode('utf-8'))
    offset = data['data']['offset']
    print(offset)
    for j in range(30):
        data_m = data_m.append({
            'id': data['data']['list'][j]['id'],
            'name': data['data']['list'][j]['name'],
            'view_count': data['data']['list'][j]['view_count'],
            'like_count': data['data']['list'][j]['like_count'],
            'duration': data['data']['list'][j]['duration'],
            'author_name': data['data']['list'][j]['author_name'],
            'author_id': data['data']['list'][j]['author_id'],
            'bvid': data['data']['list'][j]['bvid']
        }, ignore_index=True)
    return offset, data_m

Running this code quickly retrieves about 14,000 video records. After basic cleaning (e.g., converting view counts expressed in ten‑thousands), the data is visualized with a scatter plot of view count versus like count.

The analysis reveals that the most viewed videos have several million views, while the most liked video is a meme clip titled "武 林 高 手" from a popular Bilibili creator.

To explore audience comments, a word‑cloud is generated using the stylecloud library:

# Generate word cloud
stylecloud.gen_stylecloud(
    text=' '.join(text1),
    collocations=False,
    font_path=r'C:\Windows\Fonts\msyh.ttc',
    icon_name='fas fa-play-circle',
    size=653,
    output_name='马保国词云图.png'
)
Image(filename='马保国词云图.png')

The resulting word cloud highlights frequent phrases such as "耗子尾汁" and other meme terms.

Overall, the tutorial demonstrates how to obtain, process, and visualize large‑scale video data from Bilibili using Python, providing a practical example of data mining and visualization techniques.