From f35c7ff3e721c445e66f61b74b046ba506024cb4 Mon Sep 17 00:00:00 2001
From: Elbert Alias <77259+AliasIO@users.noreply.github.com>
Date: Fri, 23 Feb 2018 14:34:45 +1100
Subject: [PATCH] Add robots.txt queuing

---
 src/drivers/npm/package.json           |  2 +-
 src/drivers/webextension/js/driver.js  | 56 ++++++++++++++------------
 src/drivers/webextension/js/iframe.js  |  2 +-
 src/drivers/webextension/manifest.json |  2 +-
 src/wappalyzer.js                      |  9 ++---
 5 files changed, 37 insertions(+), 34 deletions(-)

diff --git a/src/drivers/npm/package.json b/src/drivers/npm/package.json
index f6b5cf37c..473839445 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": "5.4.7",
+  "version": "5.4.8",
   "author": "Elbert Alias",
   "license": "GPL-3.0",
   "repository": {
diff --git a/src/drivers/webextension/js/driver.js b/src/drivers/webextension/js/driver.js
index 198d67074..c7d550537 100644
--- a/src/drivers/webextension/js/driver.js
+++ b/src/drivers/webextension/js/driver.js
@@ -11,6 +11,7 @@ var tabCache = {};
 var headersCache = {};
 var categoryOrder = [];
 var options = {};
+var robotsTxtQueue = {};
 
 browser.tabs.onRemoved.addListener(tabId => {
   tabCache[tabId] = null;
@@ -287,7 +288,13 @@ wappalyzer.driver.displayApps = (detected, meta, context) => {
  * Fetch and cache robots.txt for host
  */
 wappalyzer.driver.getRobotsTxt = (host, secure = false) => {
-  return new Promise((resolve, reject) => {
+  if ( robotsTxtQueue.hasOwnProperty(host) ) {
+    wappalyzer.log('robotTxt fetch already in queue');
+
+    return robotsTxtQueue[host];
+  }
+
+  robotsTxtQueue[host] = new Promise((resolve, reject) => {
     getOption('tracking', true)
       .then(tracking => {
         if ( !tracking ) {
@@ -299,34 +306,31 @@ wappalyzer.driver.getRobotsTxt = (host, secure = false) => {
             robotsTxtCache = robotsTxtCache || {};
 
             if ( host in robotsTxtCache ) {
-              resolve(robotsTxtCache[host]);
-            } else {
-              const url = 'http' + ( secure ? 's' : '' ) + '://' + host + '/robots.txt';
-
-              fetch('http' + ( secure ? 's' : '' ) + '://' + host + '/robots.txt')
-                .then(response => {
-                  if ( !response.ok ) {
-                    if ( response.status === 404 ) {
-                      return '';
-                    } else {
-                      throw 'GET ' + response.url + ' was not ok';
-                    }
-                  }
-
-                  return response.text();
-                })
-                .then(robotsTxt => {
-                  robotsTxtCache[host] = wappalyzer.parseRobotsTxt(robotsTxt);
-
-                  setOption('robotsTxtCache', robotsTxtCache);
-
-                  resolve(robotsTxtCache[host]);
-                })
-                .catch(reject);
+              return resolve(robotsTxtCache[host]);
             }
+
+            const timeout = setTimeout(() => resolve([]), 3000);
+
+            fetch('http' + ( secure ? 's' : '' ) + '://' + host + '/robots.txt', { redirect: 'follow' })
+              .then(response => {
+                clearTimeout(timeout);
+
+                return response.ok ? response.text() : '';
+              })
+              .then(robotsTxt => {
+                robotsTxtCache[host] = wappalyzer.parseRobotsTxt(robotsTxt);
+
+                setOption('robotsTxtCache', robotsTxtCache);
+
+                resolve(robotsTxtCache[host]);
+              })
+              .catch(err => resolve([]));
           });
       });
-  });
+  })
+  .finally(() => delete robotsTxtQueue[host]);
+
+  return robotsTxtQueue[host];
 };
 
 /**
diff --git a/src/drivers/webextension/js/iframe.js b/src/drivers/webextension/js/iframe.js
index 2370a3eda..883ebea71 100644
--- a/src/drivers/webextension/js/iframe.js
+++ b/src/drivers/webextension/js/iframe.js
@@ -890,7 +890,7 @@ var exports = {};
 	var _pageTags;
 	var INIT_MS_BW_SEARCHES = 2000;
 	var PAGE_TAG_RE = new RegExp('gpt|oascentral');
-	var POST_MSG_ID = '1511804838-25881-9878-26947-14879';
+	var POST_MSG_ID = '1519242200-10756-12873-1462-13403';
 	var AD_SERVER_RE = new RegExp('^(google_ads_iframe|oas_frame|atwAdFrame)');
 
 	function getPageTags(doc) {
diff --git a/src/drivers/webextension/manifest.json b/src/drivers/webextension/manifest.json
index de6e15489..6accf3520 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": "5.4.7",
+	"version": "5.4.8",
 	"default_locale": "en",
 	"manifest_version": 2,
 	"icons": {
diff --git a/src/wappalyzer.js b/src/wappalyzer.js
index d2937d212..d95f90154 100644
--- a/src/wappalyzer.js
+++ b/src/wappalyzer.js
@@ -19,7 +19,6 @@ class Wappalyzer {
     this.categories = {};
     this.driver = {};
     this.jsPatterns = {};
-
     this.detected = {};
     this.hostnameCache = {};
     this.adCache = [];
@@ -139,12 +138,12 @@ class Wappalyzer {
 
       this.driver.getRobotsTxt(parsed.host, parsed.protocol === 'https:')
         .then(robotsTxt => {
-          if (robotsTxt.some(disallowedPath => parsed.pathname.indexOf(disallowedPath) === 0)) {
+          if ( robotsTxt.some(disallowedPath => parsed.pathname.indexOf(disallowedPath) === 0) ) {
             return reject();
-          } else {
-            return resolve();
           }
-        });
+
+          return resolve();
+        }, () => resolve());
     });
   };