insurance-spider/chery.js

import axios from "axios";
import fs from "fs";
import path from "path";
import {
  timestampToDate,
  loopCall,
  keywordsInclude,
  // addToMessageQueue,
} from "./utils.js";
import config from "./config.js";
import { SQLiteMessageQueue } from "./sqlite.js";
// import { messageQueue } from "./msgManager.js";
// import cheerio from "cheerio";

class Chery {
  constructor() {
    this.jsonMap = [
      {
        name: "奇瑞采购公告",
        // filepath: path.resolve("chery_cg.json"),
        info: [],
        options: {
          name: "采购公告",
          url: "https://ebd.mychery.com/cms/api/dynamicData/queryContentPage",
          categoryId: "5035",
          siteId: "747",
        },
      },
      {
        name: "奇瑞寻源预告",
        // filepath: path.resolve("chery_xy.json"),
        info: [],
        options: {
          name: "寻源预告",
          url: "https://ebd.mychery.com/cms/api/dynamicData/queryContentPage",
          categoryId: "965901485789413376",
          siteId: "747",
        },
      },
      {
        name: "奇瑞变更公告",
        // filepath: path.resolve("chery_bg.json"),
        info: [],
        options: {
          name: "变更公告",
          url: "https://ebd.mychery.com/cms/api/dynamicData/queryContentPage",
          categoryId: "5032",
          siteId: "747",
        },
      },
    ];
    console.log("奇瑞 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() {
    for (let item of this.jsonMap) {
      let announcements = this.queue.getAnnouncementsBySpider(item.name);
      if (announcements.length > 0) {
        this.loopFetchIncrement(item);
      } else {
        this.loopFetchFull(item);
      }
      // if (fs.existsSync(item.filepath)) {
      //   let data = fs.readFileSync(item.filepath, "utf-8");
      //   item.info = data ? JSON.parse(data) : [];
      //   if (item.info.length > 0) {
      //     // await this.increment(item);
      //     console.log(`${item.name} 历史文件存在，开始增量爬取`);
      //     this.loopFetchIncrement(item);
      //   } else {
      //     this.loopFetchFull(item);
      //   }
      // } else {
      //   console.log(`${item.name}历史文件不存在，开始全量爬取`);
      //   this.loopFetchFull(item);
      // }
    }
  }
  // 全量爬取
  loopFetchFull(props) {
    try {
      loopCall(this.getInfo.bind(this), {
        time: config.fullFetchTime,
        pagenumber: 1,
        additional: props.options,
        stopWhen: (pagenumber, result) => {
          return (
            pagenumber >= result.pages || pagenumber >= config.pageNumberLimit
          );
        },
        readyForNext: (pagenumber, result) => {
          props.info.push(...result.info);
          return pagenumber + 1;
        },
        complete: (result) => {
          props.info.push(...result.info);
          console.log(`爬取完成，共获取 ${props.info.length} 条有效数据`);
          try {
            this.queue.saveAnnouncements(props.name, props.info);
            // this.writeFile(props);
            this.queue.addMessage(props.name, props.info);
          } catch (error) {
            console.error("数据库操作失败:", error);
          }
          this.loopFetchIncrement(props);
        },
      });
    } catch (error) {
      console.error(`奇瑞${props.options.name}全量爬取失败:`, error);
    }
  }
  loopFetchIncrement(props) {
    try {
      loopCall(this.getInfo.bind(this), {
        time: config.incrementFetchTime, // 5分钟间隔
        pagenumber: 1,
        additional: props.options,
        readyForNext: (pagenumber, result) => {
          try {
            let newInfo = this.queue.filterNewAnnouncements(
              props.name,
              result.info
            );
            // 存在新数据
            if (newInfo.length > 0) {
              console.log(`发现 ${newInfo.length} 条新数据`);
              // props.info.push(...newInfo);
              this.queue.saveAnnouncements(props.name, newInfo);
              // this.writeFile(props);
              this.queue.addMessage(props.name, newInfo);
              // 全是新数据，继续下一页
              if (newInfo.length === result.info.length) {
                return pagenumber + 1;
              } else {
                // 有部分重复数据，重新从第一页开始
                return 1;
              }
            } else {
              console.log("没有发现新数据，继续监控...");
              return 1; // 重新从第一页开始
            }
          } catch (error) {
            console.error("数据库操作失败:", error);
          }
        },
      });
    } catch (error) {
      console.error(`奇瑞${props.options.name}增量爬取失败:`, error);
    }
  }
  async getInfo(pagenumber = 1, config) {
    let info = [];
    console.log(`${config.name}--获取第 ${pagenumber} 页数据...`);
    let result = await this.getList(pagenumber, config);
    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: 30, info: [] };
    } else {
      let pages = 30;
      let arr = result[1].res.rows;

      for (let i = 0; i < arr.length; i++) {
        let item = arr[i];
        let endTime, publishTime;
        if (config.categoryId === "965901485789413376") {
          publishTime = item.publishDate.replace("T", " ").split(".")[0];
          endTime = this.extractDeadlineTime(item.text);
        } else {
          endTime = item.signUpEndTime.replace("T", " ").split(".")[0];
          publishTime = item.signUpBeginTime.replace("T", " ").split(".")[0];
        }
        // 命中关键词
        if (
          endTime &&
          keywordsInclude(item.title) &&
          +new Date(endTime) >= Date.now()
        ) {
          // console.log("处理项目:", item.id, item.projectName);
          info.push({
            id: item.url,
            name: item.title,
            publishTime: publishTime,
            endTime: endTime,
            urls: `https://ebd.mychery.com/cms` + item.url,
          });
        }
      }
      return { pages, info };
    }
  }
  // 分页获取数据
  getList(pagenumber, config) {
    return axios({
      url: config.url,
      data: {
        dto: {
          bidType: "",
          categoryId: config.categoryId,
          city: "",
          county: "",
          province: "",
          purchaseMode: "",
          secondCompanyId: "",
          siteId: config.siteId,
        },
        pageNo: pagenumber,
        pageSize: "10",
      },
      method: "post",
    })
      .then((res) => {
        let result = res.data;
        if (result.code === 0) {
          return [null, result];
        } else {
          return ["err", null];
        }
      })
      .catch((err) => {
        return [err, null];
      });
  }

  // writeFile(props) {
  //   fs.writeFileSync(props.filepath, JSON.stringify(props.info), "utf-8");
  // }

  extractDeadlineTime(html) {
    // 匹配"预告报名截止时间："后面的时间格式
    const regex = /预告报名截止时间：(\d{4}-\d{2}-\d{2}\s+\d{2}:\d{2}:\d{2})/;
    const match = html.match(regex);

    if (match) {
      return match[1];
    }

    return null;
  }
}

new Chery();