Source code for langchain_community.document_loaders.college_confidential

from typing import List

from langchain_core.documents import Document

from langchain_community.document_loaders.web_base import WebBaseLoader


class CollegeConfidentialLoader(WebBaseLoader):
    """加载`College Confidential`网页。"""

    def load(self) -> List[Document]:
        """加载网页作为文档。"""
        soup = self.scrape()
        text = soup.select_one("main[class='skin-handler']").text
        metadata = {"source": self.web_path}
        return [Document(page_content=text, metadata=metadata)]