insurance-spider/yiqi.js

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

class YiQi {
  constructor() {
    // this.filepath = path.resolve("yiqi.json");
    this.info = [];
    console.log("一汽 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("一汽");
    if (announcements.length > 0) {
      await this.increment();
    } else {
      await this.fullFetch();
    }
    // if (fs.existsSync(this.filepath)) {
    //   let data = fs.readFileSync(this.filepath, "utf-8");
    //   this.info = data ? JSON.parse(data) : [];
    //   if (this.info.length > 0) {
    //     await this.increment();
    //   } else {
    //     await this.fullFetch();
    //   }
    // } else {
    //   console.log("历史文件不存在，开始全量爬取");
    //   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 {
            this.queue.saveAnnouncements("一汽", this.info);
            // this.writeFile(this.info);
            this.queue.addMessage("一汽", 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(
              "一汽",
              result.info
            );
            // let newInfo = result.info.filter(
            //   (item) => !this.info.some((info) => info.id === item.id)
            // );
            // 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("一汽", newInfo);
              // this.writeFile(this.info);
              this.queue.addMessage("一汽", 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.getHtml(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: 30, info: [] };
    } else {
      let pages = 30;
      let html = result[1];
      const $ = cheerio.load(html);
      let noticeEl = $(".zl-list-main .zl-col-6");
      noticeEl.each((index, element) => {
        let id = $(element).find(".zl-desc-item:contains('项目编号')").text();
        let name = $(element).find(".title").text();
        let publishTime = $(element)
          .find(".zl-desc-item:contains('发布时间')")
          .text();
        let endTime = $(element).find(".daojishi").attr("data-time");
        // 获取生产链接的参数
        let funcStr = $(element).find(".jump").attr("onclick");

        let funcArgs = parseToGgDetailsParams(funcStr);
        // 公告未过期 && 命中关键词
        if (endTime && keywordsInclude(name)) {
          let noticeUrl = getYiqiNoticeUrl(...funcArgs);
          info.push({
            id: id.replace("项目编号：", ""),
            name: name.trim(),
            publishTime: publishTime.replace("发布时间：", "").trim(),
            endTime: timestampToDate(Number(endTime)),
            urls: noticeUrl,
          });
        }
      });

      return { pages, info };
    }
  }
  // 分页获取数据
  getHtml(pagenumber) {
    return axios({
      url: "https://etp.faw.cn/gg/allJYTypeGGList?hangYeType=-1&xmLeiXing=&ggStartTimeEnd=&gongGaoType=5&isNew=1",
      data: {
        searchType: "",
        searchText: "",
        currentPage: pagenumber,
      },
      headers: {
        "Content-Type": "application/x-www-form-urlencoded",
      },
      method: "post",
    })
      .then((res) => {
        let result = res.data;
        return [null, result];
      })
      .catch((err) => {
        return [err, null];
      });
  }

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

new YiQi();