import axios from "axios"; import fs from "fs"; import path from "path"; import { timestampToDate, loopCall, keywordsInclude } from "./utils.js"; import config from "./config.js"; import { SQLiteMessageQueue } from "./sqlite.js"; class BYD { constructor() { this.info = []; console.log("比亚迪 爬虫启动..."); this.queue = new SQLiteMessageQueue(); this.start(); } async start() { try { await this.init(); } catch (err) { console.error("启动失败:", err); } } async init() { let announcements = this.queue.getAnnouncementsBySpider("比亚迪"); if (announcements.length > 0) { await this.increment(); } else { await this.fullFetch(); } } // 全量爬取 async fullFetch() { console.log("开始全量爬取..."); try { await loopCall(this.getInfo.bind(this), { time: config.fullFetchTime, pagenumber: 1, stopWhen: (pagenumber, result) => { return ( pagenumber >= result.pages || pagenumber >= config.pageNumberLimit ); }, readyForNext: (pagenumber, result) => { this.info.push(...result.info); return pagenumber + 1; }, complete: (result) => { this.info.push(...result.info); console.log(`爬取完成,共获取 ${this.info.length} 条有效数据`); try { if (this.info.length > 0) { this.queue.saveAnnouncements("比亚迪", this.info); // this.writeFile(this.info); this.queue.addMessage("比亚迪", this.info); } } catch (error) { console.error("数据库操作失败:", error); } }, }); } catch (error) { console.error("全量爬取失败:", error); } console.log("开始增量爬取..."); this.increment(); } // 增量爬取 async increment() { console.log("开始增量爬取模式,每5分钟检查一次新数据..."); try { await loopCall(this.getInfo.bind(this), { time: config.incrementFetchTime, // 5分钟间隔 pagenumber: 1, readyForNext: (pagenumber, result) => { try { let newInfo = this.queue.filterNewAnnouncements( "比亚迪", result.info ); // 存在新数据 if (newInfo.length > 0) { console.log(`发现 ${newInfo.length} 条新数据`); // this.info.push(...newInfo); this.queue.saveAnnouncements("比亚迪", newInfo); // this.writeFile(this.info); this.queue.addMessage("比亚迪", newInfo); // 全是新数据,继续下一页 if (newInfo.length === result.info.length) { return pagenumber + 1; } else { // 有部分重复数据,重新从第一页开始 return 1; } } else { console.log("没有发现新数据,继续监控..."); return 1; // 重新从第一页开始 } } catch (error) { console.error("数据库操作失败:", error); } }, }); } catch (error) { console.error("增量爬取失败:", error); } } async getInfo(pagenumber = 1) { let info = []; console.log(`正在获取第 ${pagenumber} 页数据...`); let result = await this.getList(pagenumber); if (result[0]) { // 出错, 记录错误日志 console.error("获取页面数据失败:", result[0]); return { pages: 0, info: [] }; } else { let total = result[1].data.total; let pages = Math.ceil(total / 10); let arr = result[1].data.records; for (let i = 0; i < arr.length; i++) { let item = arr[i]; let endTime = timestampToDate( new Date(item.signUpEndTime).getTime(), true ); // 命中关键词 if ( keywordsInclude(item.title) && endTime && +new Date(endTime) >= Date.now() ) { // console.log("处理项目:", item.sourcingId, item.title); info.push({ id: item.sourcingId, name: item.title, publishTime: timestampToDate( new Date(item.tenderNoticePublishTime).getTime(), true ), endTime: endTime, urls: `https://spcn.byd.com/#/tender-detail?sourcingId=${item.sourcingId}`, }); } } return { pages, info }; } } // 分页获取数据 getList(pagenumber) { return axios({ url: "https://spcn.byd.com/api/srm-sou-sp/supplier/supplier/getTenderAnnouncementInfo", data: { pageNo: pagenumber, pageSize: 10, }, method: "post", }) .then((res) => { let result = res.data; if (result.msg === "成功" && result.code === "000000") { return [null, result]; } else { return ["err", null]; } }) .catch((err) => { return [err, null]; }); } } new BYD();