unprint 0.17.0 → 0.17.2

package/README.md

@@ -210,10 +210,11 @@ Options
 
 Use Playwright with Chromium (experimental)
 * `unprint.browserRequest(url, [options])`
-* `unprint.closeAllBrowsers()`
+* `unprint.closeAllBrowsers()`: Close reused browser instances.
 
 Additional options
-* `scope`: Browser instance to (re)use, default `main`.
+* `control`: Async function to interface with Playwright page passed as argument
+* `scope`: Browser instance to (re)use, set to `null` to force new scope every request, default `main`.
 * `browser`: Options object passed to Playwright's `launch`, requires new scope.
 * `browser.headless`: Headless mode, set to `false` to launch visible browser, default `true`.
 * `context`: Options object passed to Playwright's `newContext`, requires new scope.
@@ -238,6 +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": "0.17.0",
+  "version": "0.17.2",
   "description": "Simplify common web scraping tasks while staying in control of the data.",
   "main": "src/app.js",
   "scripts": {},

package/src/app.js

@@ -1063,7 +1063,9 @@ async function getBrowserInstance(scope, options) {
 
 	const client = { context, browser };
 
-	clients.set(scope, client);
+	if (scope) {
+		clients.set(scope, client);
+	}
 
 	return client;
 }
@@ -1072,7 +1074,7 @@ 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,
@@ -1080,6 +1082,7 @@ 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') {
@@ -1162,8 +1165,10 @@ 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);
@@ -1171,14 +1176,22 @@ async function browserRequest(url, customOptions = {}) {
 		const data = await page.content();
 
 		await page.close();
-		// await browser.close();
+
+		if (options.scope === null) {
+			// this browser won't be reused
+			await browser.close();
+		}
 
 		return curateResponse({
 			data,
 			status,
 			statusText,
 			headers,
-		}, options, { url, customOptions });
+		}, options, {
+			url,
+			customOptions,
+			control,
+		});
 	});
 }
 

package/tests/browser.js

@@ -5,7 +5,9 @@ const unprint = require('../src/app');
 async function initTest() {
 	await Promise.all([
 		unprint.browser('https://tools-httpstatus.pickup-services.com/200?sleep=5000', {
-			headless: false,
+			browser: {
+				headless: false,
+			},
 			async control(_page) {
 				//
 			},
@@ -16,7 +18,9 @@ async function initTest() {
 			}, 1000);
 		}).then(async () => {
 			await unprint.browser('https://tools-httpstatus.pickup-services.com/200?sleep=2000', {
-				headless: false,
+				browser: {
+					headless: false,
+				},
 				async control(_page) {
 					//
 				},
@@ -28,13 +32,14 @@ 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();
 }