unprint 0.17.1 → 0.17.3

package/README.md

@@ -239,6 +239,7 @@ Returns
 	ok,					// (boolean)	status code >= 200 and < 300
 	response,			// (object)		the original axios response object, alias 'res'
 	res,				// (object)		alias for 'response'
+	control,			//				return value from browser control function
 }
 ```
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "unprint",
-  "version": "0.17.1",
+  "version": "0.17.3",
   "description": "Simplify common web scraping tasks while staying in control of the data.",
   "main": "src/app.js",
   "scripts": {},

package/src/app.js

@@ -1048,25 +1048,34 @@ const clients = new Map();
 /* eslint-enable no-param-reassign */
 async function getBrowserInstance(scope, options) {
 	if (clients.has(scope)) {
-		return clients.get(scope);
+		const client = clients.get(scope);
+
+		await client.launchers;
+
+		return client;
 	}
 
-	const browser = await chromium.launch({
+	// if we await here, we create a race condition, and a second call to getBrowserInstance would launch another browser that will overwrite the first one
+	const browserLauncher = chromium.launch({
 		headless: true,
 		...options.browser,
 	});
 
-	const context = await browser.newContext({
+	const contextLauncher = browserLauncher.then((browser) => browser.newContext({
 		userAgent: 'unprint',
 		...options.context,
-	});
+	}));
 
-	const client = { context, browser };
+	const launchers = Promise.all([browserLauncher, contextLauncher]);
+	const client = { launchers };
 
 	if (scope) {
 		clients.set(scope, client);
 	}
 
+	client.browser = await browserLauncher;
+	client.context = await contextLauncher;
+
 	return client;
 }
 
@@ -1074,7 +1083,7 @@ async function closeAllBrowsers() {
 	await Promise.all(Array.from(clients.values()).map(async (client) => client.browser.close()));
 }
 
-function curateResponse(res, options, { url, customOptions }) {
+function curateResponse(res, options, { url, control, customOptions }) {
 	const base = {
 		ok: true,
 		status: res.status,
@@ -1082,6 +1091,7 @@ function curateResponse(res, options, { url, customOptions }) {
 		headers: res.headers,
 		response: res,
 		res,
+		control,
 	};
 
 	if (['application/json', 'application/javascript'].some((type) => res.headers['content-type']?.includes(type)) && typeof res.data === 'object') {
@@ -1164,8 +1174,10 @@ async function browserRequest(url, customOptions = {}) {
 
 		await page.waitForLoadState();
 
+		let control = null;
+
 		if (customOptions.control) {
-			await customOptions.control(page, { context, browser });
+			control = await customOptions.control(page, { context, browser });
 		}
 
 		events.emit('controlSuccess', feedbackBase);
@@ -1184,7 +1196,11 @@ async function browserRequest(url, customOptions = {}) {
 			status,
 			statusText,
 			headers,
-		}, options, { url, customOptions });
+		}, options, {
+			url,
+			customOptions,
+			control,
+		});
 	});
 }
 

package/tests/browser.js

@@ -3,12 +3,30 @@
 const unprint = require('../src/app');
 
 async function initTest() {
+	// concurrency
+	await Promise.all([
+		unprint.browser('https://tools-httpstatus.pickup-services.com/200?sleep=500', {
+			browser: {
+				headless: false,
+			},
+		}),
+		unprint.browser('https://tools-httpstatus.pickup-services.com/200?sleep=500', {
+			browser: {
+				headless: false,
+			},
+		}),
+		unprint.browser('https://tools-httpstatus.pickup-services.com/200?sleep=500', {
+			browser: {
+				headless: false,
+			},
+		}),
+	]);
+
 	await Promise.all([
 		unprint.browser('https://tools-httpstatus.pickup-services.com/200?sleep=5000', {
 			browser: {
 				headless: false,
 			},
-			scope: null,
 			async control(_page) {
 				//
 			},
@@ -22,7 +40,6 @@ async function initTest() {
 				browser: {
 					headless: false,
 				},
-				scope: null,
 				async control(_page) {
 					//
 				},
@@ -34,13 +51,14 @@ async function initTest() {
 	// await unprint.browser('https://www.scrapingcourse.com/', {
 		headless: false,
 		async control(_page) {
-			//
+			return 'test';
 		},
 	});
 
 	const cards = res.context.query.contents('h2');
 
 	console.log('CARD TITLES', cards);
+	console.log('CONTROL OUT', res.control);
 
 	await unprint.closeAllBrowsers();
 }