1
import os
2
import datetime
3
from bs4 import BeautifulSoup
4
import html2text
5

6
# --- 配置区 (根据你的 Astro 需求调整) ---
7
CONFIG = {
8
    "input_dir": "./html_files",      # 存放 Halo 导出的 HTML 文件
9
    "output_dir": "./posts",          # 转换后的 MD 文件存放处
10
    "default_category": "General",     # 默认分类
11
    "default_tags": ["Blog", "Imported"], # 默认标签
12
    "image_path": "./cover.webp",     # 默认封面路径
13
    "author": "LWR"                   # 作者名
14
}
15

16
def setup_markdown_converter():
17
    h2t = html2text.HTML2Text()
18
    h2t.body_width = 0                # 禁用自动换行，保持段落完整
19
    h2t.ignore_links = False
20
    return h2t
21

22
def convert_html_to_astro_md():
23
    if not os.path.exists(CONFIG["output_dir"]):
24
        os.makedirs(CONFIG["output_dir"])
25

26
    h2t = setup_markdown_converter()
27
    # 默认发布日期设为当前日期，你也可以手动修改
28
    today = datetime.datetime.now().strftime('%Y-%m-%d')
29

30
    print(f"🚀 开始转换任务...")
31

32
    for filename in os.listdir(CONFIG["input_dir"]):
33
        if filename.endswith(".html"):
34
            file_path = os.path.join(CONFIG["input_dir"], filename)
35

36
            with open(file_path, 'r', encoding='utf-8') as f:
37
                soup = BeautifulSoup(f.read(), 'html.parser')
38

39
                # 1. 提取标题
40
                title_tag = soup.find(['h1', 'h2'])
41
                title = title_tag.get_text().strip() if title_tag else filename.replace('.html', '')
42

43
                # 2. 提取描述 (取第一个段落前 50 字)
44
                first_p = soup.find('p')
45
                description = first_p.get_text().strip()[:50].replace('"', "'") + "..." if first_p else "暂无描述"
46

47
                # 3. 转换前清理：移除已提取的标题，避免正文重复
48
                if title_tag:
49
                    title_tag.decompose()
50

51
                # 4. 转换正文为 Markdown
52
                markdown_body = h2t.handle(str(soup))
53

54
                # 5. 组装 Astro 标准 Frontmatter
55
                frontmatter = (
56
                    f"---\n"
57
                    f"title: \"{title}\"\n"
58
                    f"published: {today}\n"
59
                    f"description: \"{description}\"\n"
60
                    f"image: \"{CONFIG['image_path']}\"\n"
61
                    f"tags: {CONFIG['default_tags']}\n"
62
                    f"category: \"{CONFIG['default_category']}\"\n"
63
                    f"author: \"{CONFIG['author']}\"\n"
64
                    f"draft: false\n"
65
                    f"---\n\n"
66
                    f"# {title}\n\n"
67
                    f"{markdown_body}"
68
                )
69

70
                new_filename = os.path.splitext(filename)[0] + '.md'
71
                with open(os.path.join(CONFIG["output_dir"], new_filename), 'w', encoding='utf-8') as f_out:
72
                    f_out.write(frontmatter)
73

74
                print(f"✅ 转换成功: {new_filename}")
75

76
    print(f"🎉 所有任务已完成！请检查 {CONFIG['output_dir']} 文件夹。")
77

78
if __name__ == "__main__":
79
    convert_html_to_astro_md()