news_aggregator.ts 9.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286
  1. import TurndownService, { Options } from "turndown";
  2. import interval from "interval-promise";
  3. import { client, forumClient, FORUMS_DOMAIN } from "../client";
  4. import sha1 from "sha1";
  5. import * as path from "path";
  6. import * as fs from "fs";
  7. import { HTML2BBCode } from "html2bbcode";
  8. import { Dict } from "../util";
  9. import { IAggregator, NewsPostItem } from "./aggregators/aggregator";
  10. import { TextChannel, Message, Channel, ReactionCollector, MessageReaction, User, Collector, MessageEmbed } from "discord.js";
  11. import { getRepository, IsNull, Not } from "typeorm";
  12. import { KnownChannel } from "@shared/db/entity/KnownChannel";
  13. import { AggroNewsItem } from "@shared/db/entity/AggroNewsItem";
  14. import { v3beta1 } from "@google-cloud/translate";
  15. import { CommandSet } from "src/model/command";
  16. const { TranslationServiceClient } = v3beta1;
  17. const UPDATE_INTERVAL = process.env.NODE_ENV == "dev" ? 60 : 5;
  18. const MAX_PREVIEW_LENGTH = 300;
  19. const AGGREGATOR_MANAGER_CHANNEL = "aggregatorManager";
  20. const FORUMS_STAGING_ID = 54;
  21. const FORUMS_NEWS_ID = 49;
  22. @CommandSet
  23. export class NewsAggregator {
  24. tlClient = new TranslationServiceClient();
  25. aggregators: IAggregator[] = [];
  26. aggregateChannelID: string = null;
  27. bbCodeParser = new HTML2BBCode();
  28. turndown = new TurndownService();
  29. reactionCollectors: Dict<ReactionCollector> = {};
  30. verifyMessageIdToPost: Dict<AggroNewsItem> = {};
  31. constructor() {
  32. this.turndown.addRule("image", {
  33. filter: "img",
  34. replacement: () => ""
  35. });
  36. this.turndown.addRule("link", {
  37. filter: (node: HTMLElement, opts: Options) => node.nodeName === "A" && node.getAttribute("href") != null,
  38. replacement: (content: string, node: HTMLElement) => node.getAttribute("href")
  39. });
  40. }
  41. checkFeeds = async () => {
  42. console.log(`Aggregating feeds on ${new Date().toISOString()}`);
  43. let aggregatorJobs = [];
  44. for (let aggregator of this.aggregators) {
  45. aggregatorJobs.push(aggregator.aggregate());
  46. }
  47. let aggregatedItems = await Promise.all(aggregatorJobs);
  48. for (let itemSet of aggregatedItems) {
  49. for (let item of itemSet) {
  50. let itemObj = {
  51. ...item,
  52. cacheMessageId: null,
  53. postedMessageId: null
  54. } as NewsPostItem;
  55. itemObj.hash = sha1(itemObj.contents);
  56. await this.addNewsItem(itemObj);
  57. }
  58. }
  59. }
  60. clipText(text: string) {
  61. if (text.length <= MAX_PREVIEW_LENGTH)
  62. return text;
  63. return `${text.substring(0, MAX_PREVIEW_LENGTH)}...`;
  64. }
  65. async addNewsItem(item: NewsPostItem) {
  66. let repo = getRepository(AggroNewsItem);
  67. let ch = client.channels.resolve(this.aggregateChannelID);
  68. if (!(ch instanceof TextChannel))
  69. return;
  70. let isNew = true;
  71. let newsItem = await repo.findOne({
  72. where: { feedName: item.feedId, newsId: item.newsId }
  73. });
  74. if (newsItem) {
  75. if (process.env.IGNORE_CHANGED_NEWS === "TRUE") {
  76. newsItem.hash = item.hash;
  77. await repo.save(newsItem);
  78. return;
  79. }
  80. // No changes, skip
  81. if (newsItem.hash == item.hash)
  82. return;
  83. else
  84. await this.deleteCacheMessage(newsItem.editMessageId);
  85. isNew = false;
  86. } else {
  87. newsItem = repo.create({
  88. newsId: item.newsId,
  89. feedName: item.feedId,
  90. hash: item.hash
  91. });
  92. }
  93. if (item.needsTranslation)
  94. try {
  95. let request = {
  96. parent: this.tlClient.locationPath(process.env.GOOGLE_APP_ID, "global"),
  97. contents: [item.title, item.contents],
  98. mimeType: "text/html",
  99. sourceLanguageCode: "ja",
  100. targetLanguageCode: "en"
  101. };
  102. let [res] = await this.tlClient.translateText(request);
  103. item.title = res.translations[0].translatedText
  104. item.contents = res.translations[1].translatedText;
  105. } catch (err) {
  106. console.log(`Failed to translate because ${err}`);
  107. }
  108. item.contents = this.bbCodeParser.feed(item.contents).toString();
  109. if (!newsItem.forumsEditPostId) {
  110. let createResponse = await forumClient.createThread(FORUMS_STAGING_ID, item.title, item.contents);
  111. newsItem.forumsEditPostId = createResponse.thread.thread_id;
  112. } else {
  113. await forumClient.postReply(newsItem.forumsNewsPostId, item.contents);
  114. }
  115. let msg = await ch.send(new MessageEmbed({
  116. title: item.title,
  117. url: item.link,
  118. color: item.embedColor,
  119. timestamp: new Date(),
  120. description: `${(isNew ? "**[NEW]**" : "**[EDIT]**")}\n[**Edit on forums**](${FORUMS_DOMAIN}/index.php?threads/.${newsItem.forumsEditPostId}/)`,
  121. author: {
  122. name: item.author
  123. },
  124. footer: {
  125. text: "NoctBot News Aggregator"
  126. }
  127. })) as Message;
  128. newsItem.editMessageId = msg.id;
  129. await msg.react("✅");
  130. await msg.react("❌");
  131. let collector = msg.createReactionCollector(this.isVerifyReaction, { maxEmojis: 1 });
  132. collector.on("collect", this.collectReaction)
  133. this.reactionCollectors[msg.id] = collector;
  134. this.verifyMessageIdToPost[msg.id] = newsItem;
  135. await repo.save(newsItem);
  136. }
  137. isVerifyReaction(reaction: MessageReaction, user: User) {
  138. return (reaction.emoji.name == "✅" || reaction.emoji.name == "❌") && !user.bot && user.id != client.user.id;
  139. }
  140. collectReaction = async (reaction: MessageReaction, collector: Collector<string, MessageReaction>) => {
  141. let repo = getRepository(AggroNewsItem);
  142. let m = reaction.message;
  143. collector.stop();
  144. delete this.reactionCollectors[m.id];
  145. let post = this.verifyMessageIdToPost[m.id];
  146. if (reaction.emoji.name == "✅") {
  147. let res = await forumClient.getThread(post.forumsEditPostId);
  148. let forumPost = await forumClient.getPost(res.thread.first_post_id);
  149. if (!post.forumsNewsPostId) {
  150. let newThread = await forumClient.createThread(FORUMS_NEWS_ID, res.thread.title, forumPost.message);
  151. post.forumsNewsPostId = newThread.thread.thread_id;
  152. } else {
  153. let curThread = await forumClient.editThread(post.forumsNewsPostId, {
  154. title: res.thread.title
  155. });
  156. await forumClient.editPost(curThread.thread.first_post_id, {
  157. message: forumPost.message
  158. });
  159. }
  160. }
  161. await forumClient.deleteThread(post.forumsEditPostId);
  162. await repo.update({ newsId: post.newsId, feedName: post.feedName }, { editMessageId: null, forumsEditPostId: null, forumsNewsPostId: post.forumsNewsPostId });
  163. await reaction.message.delete();
  164. delete this.verifyMessageIdToPost[m.id];
  165. };
  166. async deleteCacheMessage(messageId: string) {
  167. let ch = client.channels.resolve(this.aggregateChannelID);
  168. if (!(ch instanceof TextChannel))
  169. return;
  170. let msg = await this.tryFetchMessage(ch, messageId);
  171. if (msg)
  172. await msg.delete();
  173. }
  174. async tryFetchMessage(channel: Channel, messageId: string) {
  175. try {
  176. if (!(channel instanceof TextChannel))
  177. return null;
  178. return await channel.messages.fetch(messageId);
  179. } catch (error) {
  180. return null;
  181. }
  182. }
  183. initAggregators() {
  184. let aggregatorsPath = path.join(path.dirname(module.filename), "aggregators");
  185. let files = fs.readdirSync(aggregatorsPath);
  186. for (let file of files) {
  187. let ext = path.extname(file);
  188. let name = path.basename(file);
  189. if (name == "aggregator.js")
  190. continue;
  191. if (ext != ".js")
  192. continue;
  193. let obj = require(path.resolve(aggregatorsPath, file)).default as IAggregator;
  194. if (obj)
  195. this.aggregators.push(obj);
  196. if (obj.init)
  197. obj.init();
  198. }
  199. }
  200. async initPendingReactors() {
  201. let verifyChannel = client.channels.resolve(this.aggregateChannelID);
  202. let repo = getRepository(AggroNewsItem);
  203. let pendingVerifyMessages = await repo.find({
  204. where: { editMessageId: Not(IsNull()) }
  205. });
  206. for (let msg of pendingVerifyMessages) {
  207. let m = await this.tryFetchMessage(verifyChannel, msg.editMessageId);
  208. if (!m) {
  209. await repo.update({ feedName: msg.feedName, newsId: msg.newsId }, { editMessageId: null });
  210. continue;
  211. }
  212. let collector = m.createReactionCollector(this.isVerifyReaction, { maxEmojis: 1 });
  213. collector.on("collect", this.collectReaction);
  214. this.reactionCollectors[m.id] = collector;
  215. this.verifyMessageIdToPost[m.id] = msg;
  216. }
  217. }
  218. async onStart() {
  219. let repo = getRepository(KnownChannel);
  220. let ch = await repo.findOne({
  221. where: { channelType: AGGREGATOR_MANAGER_CHANNEL }
  222. });
  223. if (!ch)
  224. return;
  225. this.aggregateChannelID = ch.channelId;
  226. await this.initPendingReactors();
  227. this.initAggregators();
  228. interval(this.checkFeeds, UPDATE_INTERVAL * 60 * 1000);
  229. }
  230. }