1
const GhostContentAPI = require('@tryghost/content-api');
2
const TurndownService = require('turndown');
3
const fs = require('fs');
4
const path = require('path');
5
const axios = require('axios');
6
const cheerio = require('cheerio');
7
const argv = require('minimist')(process.argv.slice(2));
8

9
// 1. 获取命令行参数
10
const { url: ghostUrl, key: ghostKey, tag: targetTag } = argv;
11

12
if (!ghostUrl || !ghostKey) {
13
  console.error('\n❌ 缺少参数！用法: node export.js --url=https://yourblog.com --key=YOUR_KEY [--tag=VPS]');
14
  process.exit(1);
15
}
16

17
const api = new GhostContentAPI({ url: ghostUrl, key: ghostKey, version: "v5.0" });
18
const turndownService = new TurndownService();
19

20
// 2. 建立目录结构
21
const baseDir = path.join(__dirname, 'ghost_export_final');
22
const resourceDir = path.join(baseDir, 'resource');
23
if (!fs.existsSync(baseDir)) fs.mkdirSync(baseDir);
24
if (!fs.existsSync(resourceDir)) fs.mkdirSync(resourceDir);
25

26
// 3. 增强型文件下载函数 (支持重试与流写入)
27
async function downloadMedia(fileUrl, saveFolder) {
28
  if (!fileUrl || !fileUrl.startsWith('http')) return fileUrl;
29

30
  try {
31
    const urlObj = new URL(fileUrl);
32
    const fileName = path.basename(urlObj.pathname);
33
    const localPath = path.join(saveFolder, fileName);
34
    const relativePath = `./resource/${fileName}`;
35

36
    if (fs.existsSync(localPath)) return relativePath;
37

38
    const response = await axios({ url: fileUrl, method: 'GET', responseType: 'stream', timeout: 30000 });
39
    const writer = fs.createWriteStream(localPath);
40
    response.data.pipe(writer);
41

42
    return new Promise((resolve, reject) => {
43
      writer.on('finish', () => resolve(relativePath));
44
      writer.on('error', reject);
45
    });
46
  } catch (e) {
47
    console.warn(`  ⚠️ 下载失败: ${fileUrl} (跳过)`);
48
    return fileUrl;
49
  }
50
}
51

52
async function startExport() {
53
  try {
54
    console.log(`📡 正在连接 Ghost: ${ghostUrl}`);
55
    const posts = await api.posts.browse({
56
      limit: 'all',
57
      include: 'tags',
58
      status: 'all',
59
      filter: targetTag ? `tag:${targetTag}` : undefined
60
    });
61

62
    console.log(`✅ 找到 ${posts.length} 篇文章，开始解析媒体资源...\n`);
63

64
    for (const post of posts) {
65
      process.stdout.write(`📦 处理中: ${post.title} ... `);
66

67
      const $ = cheerio.load(post.html || '');
68

69
      // 处理封面图
70
      let localFeatureImage = "";
71
      if (post.feature_image) {
72
        localFeatureImage = await downloadMedia(post.feature_image, resourceDir);
73
      }
74

75
      // 处理正文内的 <img>, <video>, <audio> 以及 <source>
76
      const mediaTags = $('img, video, audio, source');
77
      for (let i = 0; i < mediaTags.length; i++) {
78
        const el = $(mediaTags[i]);
79
        const src = el.attr('src');
80
        if (src) {
81
          const localRelPath = await downloadMedia(src, resourceDir);
82
          el.attr('src', localRelPath);
83
        }
84
      }
85

86
      // 转换为 Markdown
87
      const updatedHtml = $.html();
88
      const markdownBody = turndownService.turndown(updatedHtml);
89

90
      // 构建 Front-matter
91
      const rawDate = post.published_at ? post.published_at.split('T')[0] : '';
92
      const rawTags = post.tags ? post.tags.map(t => `  - ${t.name}`).join('\n') : '';
93

94
      const frontMatter = [
95
        '---',
96
        `title: ${post.title}`,
97
        `description: "${post.custom_excerpt || ""}"`,
98
        `category: Blog`,
99
        `tags:`,
100
        `${rawTags}`,
101
        `image: "${localFeatureImage}"`,
102
        `published: ${rawDate}`,
103
        `draft: ${post.status === 'draft'}`,
104
        `slug: "${post.slug}"`,
105
        '---',
106
        '\n'
107
      ].join('\n');
108

109
      fs.writeFileSync(path.join(baseDir, `${post.slug}.md`), frontMatter + markdownBody, 'utf8');
110
      console.log('完成');
111
    }
112

113
    console.log(`\n🎉 导出成功！所有文件位于: ${baseDir}`);
114
  } catch (err) {
115
    console.error(`\n❌ 运行失败: ${err.message}`);
116
  }
117
}
118

119
startExport();