feat(scraping): change lookup impl, add randomize sleep (#110)

Co-authored-by: Jef LeCompte <jeffreylec@gmail.com>

.env-example

@@ -11,7 +11,8 @@ PHONE_CARRIER="tmobile"
 PLAY_SOUND="notification.mp3"
 PUSHOVER_TOKEN="123pushover-token456"
 PUSHOVER_USER="123pushover-user-key"
-RATE_LIMIT_TIMEOUT="5000"
+PAGE_SLEEP_MIN="5000"
+PAGE_SLEEP_MAX="10000"
 SHOW_ONLY_BRANDS="evga"
 SLACK_CHANNEL="SlackChannelName"
 SLACK_TOKEN="slack-token"

README.md

@@ -75,7 +75,8 @@ Here is a list of variables that you can use to customize your newly copied `.en
 | `PLAY_SOUND` | Play this sound notification if a card is found | E.g.: `path/to/notification.wav`, relative path accepted, valid formats: wav, mp3, flac, [free sounds available](https://notificationsounds.com/) |
 | `PUSHOVER_TOKEN` | Pushover access token | Generate at https://pushover.net/apps/build |
 | `PUSHOVER_USERNAME` | Pushover username |
-| `RATE_LIMIT_TIMEOUT` | Rate limit timeout for each full store cycle | Default: `5000` |
+| `PAGE_SLEEP_MIN` | Minimum sleep time between queries of the same store | Default: `5000` |
+| `PAGE_SLEEP_MAX` | Maximum sleep time between queries of the same store | Default: `10000` |
 | `SHOW_ONLY_BRANDS` | Filter to show specified brands | Comma separated, E.g.: `evga,zotac` |
 | `SLACK_CHANNEL` | Slack channel for posting | E.g., `update`, no need for `#` |
 | `SLACK_TOKEN` | Slack API token |

package.json

@@ -22,7 +22,6 @@
   },
   "homepage": "https://github.com/jef/nvidia-snatcher#readme",
   "dependencies": {
-    "async": "^3.2.0",
     "dotenv": "^8.2.0",
     "messaging-api-telegram": "^1.0.0",
     "nodemailer": "^6.4.11",

src/config.ts

@@ -6,7 +6,8 @@ config({path: resolve(__dirname, '../.env')});
 const browser = {
 	isHeadless: process.env.HEADLESS ? process.env.HEADLESS === 'true' : true,
 	open: process.env.OPEN_BROWSER === 'true',
-	rateLimitTimeout: process.env.RATE_LIMIT_TIMEOUT ? Number(process.env.RATE_LIMIT_TIMEOUT) : 5000
+	minSleep: Number(process.env.PAGE_SLEEP_MIN ?? 5000),
+	maxSleep: Number(process.env.PAGE_SLEEP_MAX ?? 10000)
 };
 
 const logLevel = process.env.LOG_LEVEL ?? 'info';
@@ -53,7 +54,7 @@ const page = {
 	capture: process.env.SCREENSHOT === 'true',
 	width: 1920,
 	height: 1080,
-	navigationTimeout: Number(process.env.PAGE_TIMEOUT) ?? 30000,
+	navigationTimeout: Number(process.env.PAGE_TIMEOUT ?? 30000),
 	userAgent: 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/78.0.3904.108 Safari/537.36'
 };
 

src/index.ts

@@ -5,11 +5,28 @@ import {Config} from './config';
 import {Store, Stores} from './store/model';
 import {Logger} from './logger';
 import {lookup} from './store';
-import async from 'async';
+import {Browser} from 'puppeteer';
 
 puppeteer.use(stealthPlugin());
 puppeteer.use(adblockerPlugin({blockTrackers: true}));
 
+function getSleepTime() {
+	return Config.browser.minSleep + (Math.random() * (Config.browser.maxSleep - Config.browser.minSleep));
+}
+
+async function tryLookupAndLoop(browser: Browser, store: Store) {
+	Logger.debug(`[${store.name}] Starting lookup...`);
+	try {
+		await lookup(browser, store);
+	} catch (error) {
+		Logger.error(error);
+	}
+
+	const sleepTime = getSleepTime();
+	Logger.debug(`[${store.name}] Lookup done, next one in ${sleepTime} ms`);
+	setTimeout(tryLookupAndLoop, sleepTime, browser, store);
+}
+
 /**
  * Starts the bot.
  */
@@ -22,32 +39,10 @@ async function main() {
 		}
 	});
 
-	const q = async.queue<Store>(async (store: Store, cb) => {
-		setTimeout(async () => {
-			try {
-				Logger.debug(`↗ scraping initialized - ${store.name}`);
-				await lookup(browser, store);
-			} catch (error) {
-				// Ignoring errors; more than likely due to rate limits
-				Logger.error(error);
-			} finally {
-				cb();
-				q.push(store);
-			}
-		}, Config.browser.rateLimitTimeout);
-	}, Stores.length);
-
 	for (const store of Stores) {
 		Logger.debug(store.links);
-		q.push(store);
-		if (Stores.length === 1) {
-			q.push(store);
-		} // Keep from completely draining
+		setTimeout(tryLookupAndLoop, getSleepTime(), browser, store);
 	}
-
-	await q.drain();
-
-	await browser.close();
 }
 
 /**