sitemap.js 6.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214
  1. /**
  2. * 每天运行一次,生成站点地图
  3. */
  4. const fs = require('fs');
  5. const datefns = require('date-fns');
  6. const models = require('../../models');
  7. const categories = require('../../config/category');
  8. const { tags } = require('../../config/tag');
  9. const date = datefns.format(Date.now(), 'yyyy-MM-dd');
  10. const sitemapPath = __dirname + '/../../dist/sitemap.xml';
  11. const sitemapTempPath = __dirname + '/../../dist/sitemap.xml.temp';
  12. // 生成站点地图
  13. async function generateSitemap() {
  14. let commonXml = [
  15. '<?xml version="1.0" encoding="UTF-8"?>',
  16. // '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">',
  17. '<urlset',
  18. ' xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"',
  19. ' xsi:schemaLocation="http://www.sitemaps.org/schemas/sitemap/0.9 http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd http://www.w3.org/1999/xhtml http://www.w3.org/2002/08/xhtml/xhtml1-strict.xsd"',
  20. ' xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"',
  21. ' xmlns:xhtml="http://www.w3.org/1999/xhtml"',
  22. '>',
  23. ' <url>',
  24. ' <loc>https://art.pcoloring.com/en</loc>',
  25. ` <lastmod>${date}</lastmod>`,
  26. ' <changefreq>daily</changefreq>',
  27. ' <priority>1.0</priority>',
  28. ' </url>',
  29. ' <url>',
  30. ' <loc>https://art.pcoloring.com/en/gallery</loc>',
  31. ` <lastmod>${date}</lastmod>`,
  32. ' <changefreq>weekly</changefreq>',
  33. ' <priority>0.8</priority>',
  34. ' </url>',
  35. ' <url>',
  36. ' <loc>https://art.pcoloring.com/en/videos</loc>',
  37. ` <lastmod>${date}</lastmod>`,
  38. ' <changefreq>weekly</changefreq>',
  39. ' <priority>0.8</priority>',
  40. ' </url>',
  41. // ' <url>',
  42. // ' <loc>https://art.pcoloring.com/en/category</loc>',
  43. // ` <lastmod>${date}</lastmod>`,
  44. // ' <changefreq>daily</changefreq>',
  45. // ' <priority>0.8</priority>',
  46. // ' </url>',
  47. ' <url>',
  48. ' <loc>https://art.pcoloring.com/en/tag</loc>',
  49. ` <lastmod>${date}</lastmod>`,
  50. ' <changefreq>daily</changefreq>',
  51. ' <priority>0.8</priority>',
  52. ' </url>',
  53. ' <url>',
  54. ' <loc>https://art.pcoloring.com/en/albums</loc>',
  55. ` <lastmod>${date}</lastmod>`,
  56. ' <changefreq>weekly</changefreq>',
  57. ' <priority>0.8</priority>',
  58. ' </url>',
  59. ' <url>',
  60. ' <loc>https://art.pcoloring.com/en/special</loc>',
  61. ` <lastmod>${date}</lastmod>`,
  62. ' <changefreq>daily</changefreq>',
  63. ' <priority>0.8</priority>',
  64. ' </url>',
  65. ' <url>',
  66. ' <loc>https://art.pcoloring.com/en/artists</loc>',
  67. ` <lastmod>${date}</lastmod>`,
  68. ' <changefreq>monthly</changefreq>',
  69. ' <priority>0.8</priority>',
  70. ' </url>',
  71. ' <url>',
  72. ' <loc>https://art.pcoloring.com/en/info</loc>',
  73. ` <lastmod>${date}</lastmod>`,
  74. ' </url>',
  75. ];
  76. // categories
  77. // let categoriesXml = [];
  78. // categories.forEach(e => {
  79. // categoriesXml = categoriesXml.concat([
  80. // ' <url>',
  81. // ` <loc>https://art.pcoloring.com/en/category/${e.id}</loc>`,
  82. // ` <lastmod>${date}</lastmod>`,
  83. // ' </url>',
  84. // ]
  85. // );
  86. // });
  87. // console.log(categoriesXml);
  88. // tags
  89. let tagsXml = [];
  90. let newtags = tags.filter(e => e.count && e.count > 0);
  91. newtags.forEach(e => {
  92. tagsXml = tagsXml.concat([
  93. ' <url>',
  94. ` <loc>https://art.pcoloring.com/en/tag/${e.tag}</loc>`,
  95. ` <lastmod>${date}</lastmod>`,
  96. ' </url>',
  97. ]);
  98. });
  99. console.log(tagsXml);
  100. // albums
  101. let albums = await models.ArtAlbum
  102. .find({ pid: 'art', enabled: true })
  103. .sort({ order: 'asc' })
  104. .select('_id')
  105. .lean()
  106. .exec();
  107. let albumsXml = [];
  108. albums.forEach(e => {
  109. albumsXml = albumsXml.concat([
  110. ' <url>',
  111. ` <loc>https://art.pcoloring.com/en/coloring-page-album/${e._id}</loc>`,
  112. ` <lastmod>${date}</lastmod>`,
  113. ' </url>',
  114. ])
  115. });
  116. console.log(albumsXml);
  117. // designers
  118. let designers = await models.Art.aggregate([
  119. // 首先,过滤出 status = 9000 的文档
  120. { $match: { open: true, status: 9000 } },
  121. // 首先,根据 user 字段进行分组,并计算每个 user 出现的次数
  122. { $group: { _id: '$user', count: { $sum: 1 } } },
  123. // 然后,按照 count 字段进行降序排列
  124. { $sort: { count: -1 } },
  125. // 接着,与 users 集合进行连接,以获取用户的详细信息
  126. {
  127. $lookup: {
  128. from: 'users', // 要连接的集合名称
  129. localField: '_id', // 本地字段(即上一步分组后的 _id 字段)
  130. foreignField: '_id', // 要连接的集合中的字段
  131. as: 'userDetails' // 连接后结果存储在新字段 userDetails 中
  132. }
  133. },
  134. // 展开 userDetails 数组,以便将用户信息提升到顶层
  135. { $unwind: '$userDetails' },
  136. // 调整输出格式,只保留需要的字段
  137. { $project: { _id: 1, user: '$_id', count: 1, username: '$userDetails.username', name: '$userDetails.name' } },
  138. // 限制返回的记录数量
  139. { $limit: 40 }
  140. ]);
  141. let designersXml = [];
  142. designers.forEach(e => {
  143. designersXml = designersXml.concat([
  144. ' <url>',
  145. ` <loc>https://art.pcoloring.com/en/artist/${e._id}</loc>`,
  146. ` <lastmod>${date}</lastmod>`,
  147. ' </url>',
  148. ])
  149. });
  150. console.log(designersXml);
  151. // details
  152. let detailsXml = [];
  153. const cursor = await models.Art
  154. .find({ open: true, status: 9000 })
  155. .select('name title')
  156. .sort({ publishTime: -1 }).cursor(); // 使用cursor,避免一次查询太多记录内存受不了
  157. // 遍历游标中的每个文档
  158. for await (const doc of cursor) {
  159. let uriTitle = doc.name;
  160. if (doc.title) {
  161. try {
  162. let titleJson = JSON.parse(doc.title);
  163. if (titleJson.en) uriTitle = titleJson.en; // URL 里还是尽量用英文title
  164. } catch (e) {
  165. console.error(e.message);
  166. }
  167. }
  168. let utf8name = encodeURIComponent(uriTitle.replace(/[\s_]+/g, '-')).toLowerCase();
  169. detailsXml = detailsXml.concat([
  170. ' <url>',
  171. ` <loc>https://art.pcoloring.com/en/coloring-page/${utf8name}-${doc._id}</loc>`,
  172. ` <lastmod>${date}</lastmod>`,
  173. ' </url>',
  174. ]);
  175. console.log(`process detail ${doc._id} url done`);
  176. }
  177. let endXml = ['</urlset>'];
  178. let sitemapXml = [...commonXml, ...tagsXml, ...albumsXml, ...designersXml, ...detailsXml, ...endXml];
  179. let sitemapXmlStr = sitemapXml.join('\n');
  180. fs.writeFileSync(sitemapTempPath, sitemapXmlStr); // 先写入临时文件
  181. fs.renameSync(sitemapTempPath, sitemapPath); // 再重命名
  182. console.log('生成sitemap成功');
  183. }
  184. function run() {
  185. generateSitemap();
  186. }
  187. module.exports = { run }
  188. if (require.main == module) {
  189. run();
  190. }