From 3c6dcf6afbda987b5bbaa0ab8b4107be4bf4363e Mon Sep 17 00:00:00 2001
From: Elbert Alias <77259+AliasIO@users.noreply.github.com>
Date: Thu, 27 Feb 2020 15:38:10 +1100
Subject: [PATCH] Fix NPM/Puppeteer error handling

---
 src/drivers/npm/browsers/puppeteer.js  | 198 +++++++++++++------------
 src/drivers/npm/package.json           |   2 +-
 src/drivers/webextension/manifest.json |   2 +-
 3 files changed, 103 insertions(+), 99 deletions(-)

diff --git a/src/drivers/npm/browsers/puppeteer.js b/src/drivers/npm/browsers/puppeteer.js
index 5b422e833..efc70061d 100644
--- a/src/drivers/npm/browsers/puppeteer.js
+++ b/src/drivers/npm/browsers/puppeteer.js
@@ -57,117 +57,121 @@ class PuppeteerBrowser extends Browser {
     super(options);
   }
 
-  visit(url) {
-    return new Promise(async (resolve, reject) => {
-      let done = false;
-      let browser;
-
-      try {
-        browser = await puppeteer.launch(chromium ? {
-          args: [...chromium.args, '--ignore-certificate-errors'],
-          defaultViewport: chromium.defaultViewport,
-          executablePath: await chromium.executablePath,
-          headless: chromium.headless,
-        } : {
-          args: ['--no-sandbox', '--headless', '--disable-gpu', '--ignore-certificate-errors', '--disable-dev-shm-usage'],
-          executablePath: CHROME_BIN,
-        });
+  async visit(url) {
+    let done = false;
+    let browser;
 
-        browser.on('disconnected', () => {
-          if (!done) {
-            reject(new Error('browser: disconnected'));
-          }
-        });
+    try {
+      await new Promise(async (resolve, _reject) => {
+        try {
+          browser = await puppeteer.launch(chromium ? {
+            args: [...chromium.args, '--ignore-certificate-errors'],
+            defaultViewport: chromium.defaultViewport,
+            executablePath: await chromium.executablePath,
+            headless: chromium.headless,
+          } : {
+            args: ['--no-sandbox', '--headless', '--disable-gpu', '--ignore-certificate-errors', '--disable-dev-shm-usage'],
+            executablePath: CHROME_BIN,
+          });
+
+          browser.on('disconnected', () => {
+            if (!done) {
+              _reject(new Error('browser: disconnected'));
+            }
+          });
 
-        const page = await browser.newPage();
+          const page = await browser.newPage();
 
-        page.setDefaultTimeout(this.options.maxWait * 2);
+          page.setDefaultTimeout(this.options.maxWait * 2);
 
-        page.on('error', error => reject(new Error(`page error: ${error.message || error}`)));
+          page.on('error', error => _reject(new Error(`page error: ${error.message || error}`)));
 
-        page.on('response', (response) => {
-          try {
-            if (response.status() === 301 || response.status() === 302) {
-              return;
-            }
+          page.on('response', (response) => {
+            try {
+              if (response.status() === 301 || response.status() === 302) {
+                return;
+              }
 
-            if (!this.statusCode) {
-              this.statusCode = response.status();
+              if (!this.statusCode) {
+                this.statusCode = response.status();
 
-              this.headers = {};
+                this.headers = {};
 
-              const headers = response.headers();
+                const headers = response.headers();
 
-              Object.keys(headers).forEach((key) => {
-                this.headers[key] = Array.isArray(headers[key]) ? headers[key] : [headers[key]];
-              });
+                Object.keys(headers).forEach((key) => {
+                  this.headers[key] = Array.isArray(headers[key]) ? headers[key] : [headers[key]];
+                });
 
-              this.contentType = headers['content-type'] || null;
+                this.contentType = headers['content-type'] || null;
+              }
+            } catch (error) {
+              _reject(new Error(`page error: ${error.message || error}`));
             }
-          } catch (error) {
-            reject(new Error(`page error: ${error.message || error}`));
-          }
-        });
+          });
+
+          page.on('console', ({ _type, _text, _location }) => this.log(`${_text} (${_location.url}: ${_location.lineNumber})`, _type));
+
+          await page.setUserAgent(this.options.userAgent);
+
+          await Promise.race([
+            page.goto(url, { waitUntil: 'domcontentloaded' }),
+            new Promise(_resolve => setTimeout(() => {
+              this.log('Timeout', 'error');
+
+              _resolve();
+            }, this.options.maxWait)),
+          ]);
+
+          // eslint-disable-next-line no-undef
+          const links = await page.evaluateHandle(() => Array.from(document.getElementsByTagName('a')).map(({
+            hash, hostname, href, pathname, protocol, rel,
+          }) => ({
+            hash,
+            hostname,
+            href,
+            pathname,
+            protocol,
+            rel,
+          })));
 
-        page.on('console', ({ _type, _text, _location }) => this.log(`${_text} (${_location.url}: ${_location.lineNumber})`, _type));
-
-        await page.setUserAgent(this.options.userAgent);
-
-        await Promise.race([
-          page.goto(url, { waitUntil: 'domcontentloaded' }),
-          new Promise(_resolve => setTimeout(() => {
-            this.log('Timeout', 'error');
-
-            _resolve();
-          }, this.options.maxWait)),
-        ]);
-
-        // eslint-disable-next-line no-undef
-        const links = await page.evaluateHandle(() => Array.from(document.getElementsByTagName('a')).map(({
-          hash, hostname, href, pathname, protocol, rel,
-        }) => ({
-          hash,
-          hostname,
-          href,
-          pathname,
-          protocol,
-          rel,
-        })));
-
-        this.links = await links.jsonValue();
-
-        // eslint-disable-next-line no-undef
-        const scripts = await page.evaluateHandle(() => Array.from(document.getElementsByTagName('script')).map(({
-          src,
-        }) => src));
-
-        this.scripts = (await scripts.jsonValue()).filter(script => script);
-
-        this.js = await page.evaluate(getJs);
-
-        this.cookies = (await page.cookies()).map(({
-          name, value, domain, path,
-        }) => ({
-          name, value, domain, path,
-        }));
-
-        this.html = await page.content();
-
-        resolve();
-      } catch (error) {
-        reject(new Error(`visit error: ${error.message || error}`));
-      } finally {
-        done = true;
-
-        if (browser) {
-          try {
-            await browser.close();
-          } catch (error) {
-            this.log(error.message || error.toString(), 'error');
-          }
+          this.links = await links.jsonValue();
+
+          // eslint-disable-next-line no-undef
+          const scripts = await page.evaluateHandle(() => Array.from(document.getElementsByTagName('script')).map(({
+            src,
+          }) => src));
+
+          this.scripts = (await scripts.jsonValue()).filter(script => script);
+
+          this.js = await page.evaluate(getJs);
+
+          this.cookies = (await page.cookies()).map(({
+            name, value, domain, path,
+          }) => ({
+            name, value, domain, path,
+          }));
+
+          this.html = await page.content();
+
+          resolve();
+        } catch (error) {
+          _reject(new Error(`visit error: ${error.message || error}`));
+        }
+      });
+    } catch (error) {
+      done = true;
+
+      if (browser) {
+        try {
+          await browser.close();
+        } catch (_error) {
+          this.log(_error.message || _error.toString(), 'error');
         }
       }
-    });
+
+      throw new Error(error.message || error.toString());
+    }
   }
 }
 
diff --git a/src/drivers/npm/package.json b/src/drivers/npm/package.json
index 21613c8ea..64a29d62c 100644
--- a/src/drivers/npm/package.json
+++ b/src/drivers/npm/package.json
@@ -2,7 +2,7 @@
   "name": "wappalyzer",
   "description": "Uncovers the technologies used on websites",
   "homepage": "https://github.com/AliasIO/Wappalyzer",
-  "version": "5.9.21",
+  "version": "5.9.22",
   "author": "Elbert Alias",
   "license": "GPL-3.0",
   "repository": {
diff --git a/src/drivers/webextension/manifest.json b/src/drivers/webextension/manifest.json
index bc7fe4856..be890ceca 100644
--- a/src/drivers/webextension/manifest.json
+++ b/src/drivers/webextension/manifest.json
@@ -4,7 +4,7 @@
   "author": "Elbert Alias",
   "homepage_url": "https://www.wappalyzer.com",
   "description": "Identify web technologies",
-  "version": "5.9.20",
+  "version": "5.9.22",
   "default_locale": "en",
   "manifest_version": 2,
   "icons": {