tools/hf_blog_post_id_list.py

from smolagents import Tool
from typing import Any, Optional

class SimpleTool(Tool):
    name = "hf_blog_post_id_list"
    description = "Get the list of Hugging Face blog post IDs."
    inputs = {}
    output_type = "array"

    def forward(self) -> List[str]:
        """
        Get the list of Hugging Face blog post IDs.

        Returns:
            List[str]: List of blog post IDs.
        """
        import requests
        from bs4 import BeautifulSoup
        import re
        from typing import List

        r = requests.get("https://huggingface.co/blog")
        soup = BeautifulSoup(r.text, "html.parser")
        # Find all divs with class "SVELTE_HYDRATER contents"
        blog_divs = soup.find_all('div', attrs={'class': 'SVELTE_HYDRATER contents', 'data-target': 'BlogThumbnail'})

        # Extract hrefs from nested <a> elements
        hrefs = []
        for div in blog_divs:
            a_tag = div.find('a', attrs={'class': re.compile('flex lg:col-span-1 hover:shadow-alternate')})
            if a_tag and a_tag.has_attr('href'):
                hrefs.append(a_tag['href'])

        return hrefs