myTools/ExtractWikipediaSection.py

from smolagents import Tool
import requests
import re
from markdownify import markdownify as md

class ExtractWikipediaSection(Tool):
    name = "extract_wikipedia_section"
    description = "Extracts a specific section from a Wikipedia page in Markdown format."

    inputs = {
        "url": {
            "type": "string",
            "description": "URL of the Wikipedia page"
        },
        "section": {
            "type": "string",
            "description": "Title of the section to extract"
        },
    }

    output_type = "string"

    def forward(self, url: str, section: str) -> str:
        headers = {
            "User-Agent": "Mozilla/5.0 (compatible; WebScraper/1.0; +https://example.com/bot)"
        }

        try:
            response = requests.get(url, headers=headers, timeout=10)
            response.raise_for_status()
        except Exception as e:
            raise RuntimeError(f"Failed to fetch page: {e}")

        markdown = md(response.text, heading_style="ATX")

        # RegEx pour détecter la section markdown
        pattern = rf"^##+\s*{re.escape(section)}\s*$(.*?)^##+"
        match = re.search(pattern, markdown, re.DOTALL | re.MULTILINE)
        if match:
            return match.group(1).strip()
        else:
            return f"❌ Section '{section}' not found on page."