refactor: prepare for non puppeteer scrapers

Author: sarod

browser-extension/entrypoints/background/scraping/base-scraper.ts

@@ -1,9 +1,9 @@
 import { SocialNetworkName } from "@/entrypoints/shared/model/social-network-name";
-import { BaseScraper } from "./base-scraper";
+import { PuppeteerBaseScraper } from "./puppeteer/puppeteer-base-scraper";
 import { InstagramScraper } from "./instagram/instagram-scraper";
 import { YoutubeScraper } from "./youtube/youtube-scraper";
 
-export function createScraper(sn: SocialNetworkName): BaseScraper {
+export function createScraper(sn: SocialNetworkName): PuppeteerBaseScraper {
   switch (sn) {
     case "YOUTUBE":
       return new YoutubeScraper();

browser-extension/entrypoints/background/scraping/create-scraper.ts

@@ -1,5 +1,8 @@
-import { ElementHandle } from "puppeteer-core/lib/esm/puppeteer/puppeteer-core-browser.js";
-import { BaseScraper } from "../base-scraper";
+import {
+  ElementHandle,
+  Page,
+} from "puppeteer-core/lib/esm/puppeteer/puppeteer-core-browser.js";
+import { PuppeteerBaseScraper } from "../puppeteer/puppeteer-base-scraper";
 import {
   type Author,
   type Post,
@@ -10,7 +13,7 @@ import { currentIsoDate } from "../utils/current-iso-date";
 
 //TODO: gérer le scroll et le chargement des commentaires
 //TODO: gérer le scraping des réponses aux commentaires
-export class InstagramScraper extends BaseScraper {
+export class InstagramScraper extends PuppeteerBaseScraper {
   private INSTAGRAM_URL = "https://www.instagram.com/";
 
   extractPostId(url: string): string {
@@ -21,9 +24,7 @@ export class InstagramScraper extends BaseScraper {
     return parsed.postId;
   }
 
-  async scrapTab(tab: Browser.tabs.Tab): Promise<Post> {
-    const page = await this.getBrowserPageFromTab(tab);
-
+  async doScrapTab(tab: Browser.tabs.Tab, page: Page): Promise<Post> {
     // //main/div/div/div
     const cadre_publication = (await page.$("::-p-xpath(//main/div/div/div)"))!;
     const colonne_commentaires = (await cadre_publication.$(

browser-extension/entrypoints/background/scraping/instagram/instagram-scraper.ts

@@ -0,0 +1,42 @@
+import {
+  connect,
+  ExtensionTransport,
+  Page,
+  Browser as PuppeteerBrowser,
+} from "puppeteer-core/lib/esm/puppeteer/puppeteer-core-browser.js";
+import type { Post } from "../../../shared/model/post";
+import { Scraper } from "../scraper";
+
+export abstract class PuppeteerBaseScraper implements Scraper {
+  browser?: PuppeteerBrowser;
+  constructor() {}
+
+  protected async sleep(ms: number): Promise<void> {
+    return new Promise((resolve) => setTimeout(resolve, ms));
+  }
+
+  private async getBrowserPageFromTab(tab: Browser.tabs.Tab): Promise<Page> {
+    this.browser = await connect({
+      transport: await ExtensionTransport.connectTab(tab.id!),
+    });
+    const [page] = await this.browser.pages();
+    return page;
+  }
+
+  private async disconnectBrowser() {
+    if (this.browser) {
+      this.browser.disconnect();
+    }
+  }
+
+  public async scrapTab(tab: Browser.tabs.Tab): Promise<Post> {
+    const page = await this.getBrowserPageFromTab(tab);
+    try {
+      return this.doScrapTab(tab, page);
+    } finally {
+      this.disconnectBrowser();
+    }
+  }
+
+  abstract doScrapTab(tab: Browser.tabs.Tab, page: Page): Promise<Post>;
+}

…/background/scraping/utils/anchorHref.ts …kground/scraping/puppeteer/anchorHref.tsbrowser-extension/entrypoints/background/scraping/utils/anchorHref.ts renamed to browser-extension/entrypoints/background/scraping/puppeteer/anchorHref.ts browser-extension/entrypoints/background/scraping/utils/anchorHref.ts renamed to browser-extension/entrypoints/background/scraping/puppeteer/anchorHref.ts

@@ -13,11 +13,5 @@ export async function scrapTab(
     throw new Error("Url of tab is not scrapable!");
   }
   const scraper = createScraper(snUrl.socialNetwork);
-  try {
-    const socialNetworkPost = await scraper.scrapTab(tab);
-
-    return socialNetworkPost;
-  } finally {
-    await scraper.disconnectBrowser();
-  }
+  return await scraper.scrapTab(tab);
 }