bot-sismos/src/utils/scraper-reports.ts

import { type AnyNode, load } from "cheerio";
import "dotenv/config";

import config from "../config";
import Scraper from "./scraper";

import type { IScraperReportsOptions } from "../interfaces/scaper-reports-options";
import type { IItem } from "../interfaces/item";

import LogLevels from "../enums/log-levels";
import Props from "../enums/props";

export default class ScraperReports {
  private readonly _name: string;
  private readonly _options: IScraperReportsOptions;
  private readonly _scraper: Scraper;

  constructor (name: string, options: IScraperReportsOptions) {
    this._name = name;
    this._options = options;
    this._scraper = new Scraper();
  }

  private getProperty (domNodeElement: AnyNode, selector: string, prop: Props = Props.TEXT): string {
    const $ = load(domNodeElement);
    let value: string | undefined = "";
    try {
      value = prop === Props.TEXT ? $(selector).text() : $(selector).attr(prop);
    } catch (err) {
      if (config.LOG_LEVEL === LogLevels.DEBUG) {
        console.debug(`${this._name} | Error raised\n`);
        console.debug(`From ${domNodeElement.type} can't get value using selector '${selector}'`);
        console.error(err.message);
      }
    }
    return value ?? "";
  }

  private getDate (node: AnyNode): string {
    const selector = this._options.dateSelector ?? "";
    return selector !== "" ? this.getProperty(node, selector).trim() : new Date(Date.now()).toLocaleDateString("es-CL");
  }

  private getLocation (node: AnyNode): string {
    const selector = this._options.locationSelector ?? "";
    return selector !== "" ? this.getProperty(node, selector).trim() : "";
  }

  private getDepth (node: AnyNode): string {
    const selector = this._options.depthSelector ?? "";
    return selector !== "" ? this.getProperty(node, selector).trim() : "";
  }

  private getMagnitude (node: AnyNode): number {
    const selector = this._options.magnitudeSelector ?? "";
    return selector !== "" ? Number(this.getProperty(node, selector).trim()) : 0;
  }

  private getLink (node: AnyNode): string {
    const selector = this._options.linkSelector ?? "";
    let link = selector !== "" ? this.getProperty(node, selector, Props.LINK).trim() : "";
    link = this._options.url.replace("/index.html", "") + link;
    return link;
  }

  private async getLocationImg (link: string): Promise<File | null> {
    return await this._scraper.scrapeFile(link);
  }

  public async getItems (): Promise<IItem[]> {
    const items: IItem[] = [];
    const startTime = Date.now();

    try {
      console.info("Starting scraping", this._options);
      const response = await this._scraper.scrape({ url: this._options.url });
      const html = response.data.data.html;

      const $ = load(html as string);
      const domElements = $(this._options.reportsSelector);

      if (config.LOG_LEVEL === LogLevels.DEBUG) {
        console.debug(`${this._name} | Items obtained: ${domElements.length} `);
      }

      for (let i = 1; i < domElements.length; i++) {
        const node = domElements[i];
        const link = this.getLink(node);
        items.push({
          date: this.getDate(node),
          location: this.getLocation(node),
          depth: this.getDepth(node),
          magnitude: this.getMagnitude(node),
          link,
          locationImg: await this.getLocationImg(link.replace("html", "jpeg"))
        });
      }
    } catch (err) {
      if (config.LOG_LEVEL === LogLevels.DEBUG) {
        console.debug(`${this._name} | Error\n`);
        console.error(err.message);
      }
    } finally {
      const endTime = Date.now();
      const duration = (endTime - startTime) / 1000;
      console.info(`${this._name} | Execution time: ${duration}s`);
    }

    return items;
  }
}