Merge branch 'js-field'

bin/validate

@@ -6,7 +6,7 @@ set -eu
 
 echo "Validating apps.json..."
 
-jsonlint-cli -s schema.json src/apps.json
+jsonlint-cli -tps schema.json src/apps.json > /tmp/apps.json && mv /tmp/apps.json src/apps.json
 
 echo "Validating regular expressions..."
 

schema.json

@@ -35,9 +35,9 @@
 						},
 						"required": true
 					},
-					"env": {
+					"js": {
-						"type": [ "string", "array" ],
+						"type": "object",
-						"items": {
+						"additionalProperties": {
 							"type": "string"
 						}
 					},

src/drivers/npm/driver.js

@@ -38,6 +38,8 @@ class Driver {
     this.wappalyzer.apps = json.apps;
     this.wappalyzer.categories = json.categories;
 
+    this.wappalyzer.parseJsPatterns();
+
     this.wappalyzer.driver.log = (message, source, type) => this.log(message, source, type);
     this.wappalyzer.driver.displayApps = (detected, meta, context) => this.displayApps(detected, meta, context);
   }
@@ -123,34 +125,15 @@ class Driver {
               .finally(() => {
                 this.timer('browser.wait end');
 
-                const headers = {};
+                const headers = this.getHeaders(browser);
-
+                const html    = this.getHtml(browser);
-                browser.resources['0'].response.headers._headers.forEach(header => {
+                const scripts = this.getScripts(browser);
-                  if ( !headers[header[0]] ){
+                const js      = this.getJs(browser);
-                    headers[header[0]] = [];
-                  }
-
-                  headers[header[0]].push(header[1]);
-                });
-
-                let html = '';
-
-                try {
-                  html = browser.html();
-                } catch ( e ) {
-                  this.wappalyzer.log(error.message, 'browser', 'error');
-                }
-
-                const vars = Object.getOwnPropertyNames(browser.window);
-                const scripts = Array.prototype.slice
-                  .apply(browser.document.scripts)
-                  .filter(s => s.src)
-                  .map(s => s.src);
 
                 this.wappalyzer.analyze(pageUrl.hostname, pageUrl.href, {
                   headers,
                   html,
-                  env: vars,
+                  js,
                   scripts
                 });
 
@@ -163,6 +146,70 @@ class Driver {
     });
   }
 
+  getHeaders(browser) {
+    const headers = {};
+
+    browser.resources['0'].response.headers._headers.forEach(header => {
+      if ( !headers[header[0]] ){
+        headers[header[0]] = [];
+      }
+
+      headers[header[0]].push(header[1]);
+    });
+
+    return headers;
+  }
+
+  getHtml(browser) {
+    let html = '';
+
+    try {
+      html = browser.html();
+    } catch ( e ) {
+      this.wappalyzer.log(error.message, 'browser', 'error');
+    }
+
+    return html;
+  }
+
+  getScripts(browser) {
+    const scripts = Array.prototype.slice
+      .apply(browser.document.scripts)
+      .filter(s => s.src)
+      .map(s => s.src);
+
+    return scripts;
+  }
+
+  getJs(browser) {
+    const patterns = this.wappalyzer.jsPatterns;
+    const js = {};
+
+    Object.keys(patterns).forEach(appName => {
+      js[appName] = {};
+
+      Object.keys(patterns[appName]).forEach(chain => {
+        js[appName][chain] = {};
+
+        patterns[appName][chain].forEach((pattern, index) => {
+          const properties = chain.split('.');
+
+          let value = properties.reduce((parent, property) => {
+            return parent && parent.hasOwnProperty(property) ? parent[property] : null;
+          }, browser.window);
+
+          value = typeof value === 'string' ? value : !!value;
+
+          if ( value ) {
+            js[appName][chain][index] = value;
+          }
+        });
+      });
+    });
+
+    return js;
+  }
+
   crawl(pageUrl, index = 1, depth = 1) {
     this.timer('crawl');
 

src/drivers/webextension/js/content.js

@@ -1,58 +1,58 @@
 /** global: browser */
 
 if ( typeof browser !== 'undefined' && typeof document.body !== 'undefined' ) {
-  var html = document.documentElement.outerHTML;
+  try {
+    var html = document.documentElement.outerHTML;
 
-  if ( html.length > 50000 ) {
+    if ( html.length > 50000 ) {
-    html = html.substring(0, 25000) + html.substring(html.length - 25000, html.length);
+      html = html.substring(0, 25000) + html.substring(html.length - 25000, html.length);
-  }
+    }
 
-  var scripts = Array.prototype.slice
+    const scripts = Array.prototype.slice
       .apply(document.scripts)
-      .filter(s => s.src)
+      .filter(script => script.src)
-      .map(s => s.src);
+      .map(script => script.src);
-
-  try {
-    browser.runtime.sendMessage({
-      id: 'analyze',
-      subject: { html },
-      source: 'content.js'
-    });
 
     browser.runtime.sendMessage({
       id: 'analyze',
-      subject: { scripts },
+      subject: { html, scripts },
       source: 'content.js'
     });
 
-    var container = document.createElement('wappalyzerData');
+    const script = document.createElement('script');
 
-    container.setAttribute('id',    'wappalyzerData');
+    script.onload = () => {
-    container.setAttribute('style', 'display: none');
+      addEventListener('message', event => {
+        if ( event.data.id !== 'js' ) {
+          return;
+        }
 
-    var script = document.createElement('script');
+        browser.runtime.sendMessage({
+          id: 'analyze',
+          subject: {
+            js: event.data.js
+          },
+          source: 'content.js'
+        });
+      }, true);
 
-    script.setAttribute('id', 'wappalyzerEnvDetection');
+      ( chrome || browser ).runtime.sendMessage({
-    script.setAttribute('src', browser.extension.getURL('js/inject.js'));
+        id: 'init_js',
-
+        subject: {},
-    container.addEventListener('wappalyzerEvent', (event => {
-      var env = event.target.childNodes[0].nodeValue;
-
-      document.documentElement.removeChild(container);
-      document.documentElement.removeChild(script);
-
-      env = env.split(' ').slice(0, 500);
-
-      browser.runtime.sendMessage({
-        id: 'analyze',
-        subject: { env },
         source: 'content.js'
+      }, response => {
+        postMessage({
+          id: 'patterns',
+          patterns: response.patterns
+        }, '*');
       });
-    }), true);
+    };
+
+    script.setAttribute('id', 'wappalyzer');
+    script.setAttribute('src', browser.extension.getURL('js/inject.js'));
 
-    document.documentElement.appendChild(container);
+    document.body.appendChild(script);
-    document.documentElement.appendChild(script);
+  } catch (e) {
-  } catch(e) {
     log(e);
   }
 }

src/drivers/webextension/js/driver.js

@@ -75,6 +75,8 @@ fetch('../apps.json')
     wappalyzer.apps = json.apps;
     wappalyzer.categories = json.categories;
 
+    wappalyzer.parseJsPatterns();
+
     categoryOrder = Object.keys(wappalyzer.categories).sort((a, b) => wappalyzer.categories[a].priority - wappalyzer.categories[b].priority);
   })
   .catch(error => {
@@ -191,6 +193,12 @@ browser.webRequest.onCompleted.addListener(request => {
           categories: wappalyzer.categories
         };
 
+        break;
+      case 'init_js':
+        response = {
+          patterns: wappalyzer.jsPatterns
+        };
+
         break;
       default:
     }

src/drivers/webextension/js/inject.js

@@ -1,16 +1,43 @@
 (function() {
 	try {
-		var i, environmentVars = '', e = document.createEvent('Events');
+    addEventListener('message', (event => {
+      if ( event.data.id !== 'patterns' ) {
+        return;
+      }
 
-		e.initEvent('wappalyzerEvent', true, false);
+      const patterns = event.data.patterns || {};
 
-		for ( i in window ) {
+      const js = {};
-			environmentVars += i + ' ';
-		}
 
-		document.getElementById('wappalyzerData').appendChild(document.createComment(environmentVars));
+      Object.keys(patterns).forEach(appName => {
-		document.getElementById('wappalyzerData').dispatchEvent(e);
+        js[appName] = {};
-	} catch(e) {
+
-		// Fail quietly
+        Object.keys(patterns[appName]).forEach(chain => {
- 	}
+          js[appName][chain] = {};
+
+          patterns[appName][chain].forEach((pattern, index) => {
+            const value = detectJs(chain);
+
+            if ( value ) {
+              js[appName][chain][index] = value;
+            }
+          });
+        });
+      });
+
+      postMessage({ id: 'js', js }, '*');
+    }), false);
+  } catch(e) {
+    // Fail quietly
+  }
 }());
+
+function detectJs(chain) {
+  const properties = chain.split('.');
+
+  const value = properties.reduce((parent, property) => {
+    return parent && parent.hasOwnProperty(property) ? parent[property] : null;
+  }, window);
+
+  return typeof value === 'string' ? value : !!value;
+}

src/wappalyzer.js

@@ -18,6 +18,7 @@ class Wappalyzer {
     this.apps = {};
     this.categories = {};
     this.driver = {};
+    this.jsPatterns = {};
 
     this.detected = {};
     this.hostnameCache = {};
@@ -87,6 +88,12 @@ class Wappalyzer {
       }
     })
 
+    if ( data.js ) {
+      Object.keys(data.js).forEach(appName => {
+        this.analyzeJs(apps[appName], data.js[appName]);
+      });
+    }
+
     Object.keys(apps).forEach(appName => {
       var app = apps[appName];
 
@@ -249,6 +256,17 @@ class Wappalyzer {
     return parsed;
   }
 
+  /**
+   * Parse JavaScript patterns
+   */
+  parseJsPatterns() {
+    Object.keys(this.apps).forEach(appName => {
+      if ( this.apps[appName].js ) {
+        this.jsPatterns[appName] = this.parsePatterns(this.apps[appName].js);
+      }
+    });
+  }
+
   resolveExcludes(apps) {
     var excludes = [];
 
@@ -487,6 +505,25 @@ class Wappalyzer {
     }
   }
 
+  /**
+   * Analyze JavaScript variables
+   */
+  analyzeJs(app, results) {
+
+    console.log(app, results);
+
+    Object.keys(results).forEach(string => {
+      Object.keys(results[string]).forEach(index => {
+        const pattern = this.jsPatterns[app.name][string][index];
+        const value = results[string][index];
+
+        if ( pattern.regex.test(value) ) {
+          this.addDetected(app, pattern, 'js', value);
+        }
+      });
+    });
+  }
+
   /**
    * Analyze robots.txt
    */