unprint 0.16.4-beta → 0.17.1

package/README.md

@@ -205,15 +205,24 @@ Extracts the CSS `url()` background from a style attribute. Alias for `query.sty
 * `unprint.request(url, body, [options], [method])`
 
 Options
-* `select`: Pre-query and initialize a specific element on the page
-* `selectAll`: Pre-query and initialize multiple specific element on the page
+* `select`: Pre-query and initialize a specific element on the page.
+* `selectAll`: Pre-query and initialize multiple specific element on the page.
 
 Use Playwright with Chromium (experimental)
 * `unprint.browserRequest(url, [options])`
-* `unprint.closeAllBrowsers()`
+* `unprint.closeAllBrowsers()`: Close reused browser instances.
 
 Additional options
-* `browser`: Options object passed to Playwright
+* `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.
+* `page`: Options object passed to Playwright's `goto`.
+
+This requires you to install the Chromium executable:
+* `sudo npx patchright install-deps`
+* `npx patchright install`
 
 Returns
 ```javascript

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "unprint",
-  "version": "0.16.4-beta",
+  "version": "0.17.1",
   "description": "Simplify common web scraping tasks while staying in control of the data.",
   "main": "src/app.js",
   "scripts": {},

package/src/app.js

@@ -1046,22 +1046,26 @@ function setProxy(instance, options, url) {
 const clients = new Map();
 
 /* eslint-enable no-param-reassign */
-async function getBrowserInstance(scope) {
+async function getBrowserInstance(scope, options) {
 	if (clients.has(scope)) {
 		return clients.get(scope);
 	}
 
 	const browser = await chromium.launch({
-		headless: false,
+		headless: true,
+		...options.browser,
 	});
 
 	const context = await browser.newContext({
 		userAgent: 'unprint',
+		...options.context,
 	});
 
 	const client = { context, browser };
 
-	clients.set(scope, client);
+	if (scope) {
+		clients.set(scope, client);
+	}
 
 	return client;
 }
@@ -1126,11 +1130,11 @@ async function browserRequest(url, customOptions = {}) {
 	};
 
 	return limiter.schedule(async () => {
-		const { context, browser } = await getBrowserInstance(options.scope);
+		const { context, browser } = await getBrowserInstance(options.scope, options);
 		const page = await context.newPage();
 
 		const res = await page.goto(url, {
-			...options.browser,
+			...options.page,
 		});
 
 		const status = res.status();
@@ -1169,7 +1173,11 @@ 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,

package/tests/browser.js

@@ -5,7 +5,10 @@ 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,
+			},
+			scope: null,
 			async control(_page) {
 				//
 			},
@@ -16,7 +19,10 @@ async function initTest() {
 			}, 1000);
 		}).then(async () => {
 			await unprint.browser('https://tools-httpstatus.pickup-services.com/200?sleep=2000', {
-				headless: false,
+				browser: {
+					headless: false,
+				},
+				scope: null,
 				async control(_page) {
 					//
 				},