Sync with AliasIO (#1)

* Refactoring, allow multiple expressions per js field, remove env detection

* Add js field to NPM driver

* Port env to js

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

@@ -37,6 +37,8 @@ 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 => this.displayApps(detected);
   }
@@ -120,34 +122,15 @@ 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
                 });
 
@@ -160,6 +143,70 @@ 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

@@ -1,76 +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 },
+      subject: { html, scripts },
       source: 'content.js'
     });
 
-    browser.runtime.sendMessage({
+    const script = document.createElement('script');
-      id: 'analyze',
-      subject: { scripts },
-      source: 'content.js'
-    });
-
-    var container = document.createElement('wappalyzerData');
-
-    container.setAttribute('id',    'wappalyzerData');
-    container.setAttribute('style', 'display: none');
-
-    var script = document.createElement('script');
-
-    script.setAttribute('id', 'wappalyzerEnvDetection');
-    script.setAttribute('src', browser.extension.getURL('js/inject.js'));
 
-    container.addEventListener('wappalyzerEnvEvent', (event => {
+    script.onload = () => {
-      browser.runtime.sendMessage({
+      addEventListener('message', event => {
-        id: 'JS_ready',
+        if ( event.data.id !== 'js' ) {
-        subject: { },
-        source: 'content.js'
-      }, response => {
-        window.postMessage({patterns: response.patterns}, "*");
-      });
-      window.addEventListener('message', (event => {
-        if (event.data.js === undefined)
           return;
-        var js = event.data.js ;
+        }
+
         browser.runtime.sendMessage({
           id: 'analyze',
-          subject: { js },
+          subject: {
+            js: event.data.js
+          },
           source: 'content.js'
         });
-      }), true);
+      }, true);
-
-      var env = event.target.childNodes[0].nodeValue;
 
-      document.documentElement.removeChild(container);
+      ( chrome || browser ).runtime.sendMessage({
-      document.documentElement.removeChild(script);
+        id: 'init_js',
-
+        subject: {},
-      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

@@ -74,7 +74,8 @@ fetch('../apps.json')
   .then(json => {
     wappalyzer.apps = json.apps;
     wappalyzer.categories = json.categories;
-    wappalyzer.parseJS();
+
+    wappalyzer.parseJsPatterns();
 
     categoryOrder = Object.keys(wappalyzer.categories).sort((a, b) => wappalyzer.categories[a].priority - wappalyzer.categories[b].priority);
   })
@@ -193,9 +194,9 @@ browser.webRequest.onCompleted.addListener(request => {
         };
 
         break;
-      case 'JS_ready':
+      case 'init_js':
         response = {
-          patterns: wappalyzer.parsedJS
+          patterns: wappalyzer.jsPatterns
         };
 
         break;

src/drivers/webextension/js/inject.js

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

src/wappalyzer.js

@@ -18,7 +18,7 @@ class Wappalyzer {
     this.apps = {};
     this.categories = {};
     this.driver = {};
-    this.parsedJS = undefined;
+    this.jsPatterns = {};
 
     this.detected = {};
     this.hostnameCache = {};
@@ -85,7 +85,7 @@ class Wappalyzer {
 
     if ( data.js ) {
       Object.keys(data.js).forEach(appName => {
-        this.analyzeJS(apps[appName], data.js[appName]);
+        this.analyzeJs(apps[appName], data.js[appName]);
       });
     }
 
@@ -252,16 +252,14 @@ class Wappalyzer {
   }
 
   /**
-   * Parse JS patterns
+   * Parse JavaScript patterns
    */
-  parseJS() {
+  parseJsPatterns() {
-    if ( this.parsedJS === undefined){
+    Object.keys(this.apps).forEach(appName => {
-      this.parsedJS = {};
+      if ( this.apps[appName].js ) {
-      Object.keys(this.apps).forEach(appName => {
+        this.jsPatterns[appName] = this.parsePatterns(this.apps[appName].js);
-        if (this.apps[appName].js)
+      }
-          this.parsedJS[appName] = this.parsePatterns(this.apps[appName].js);
+    });
-      });
-    }
   }
 
   resolveExcludes(apps) {
@@ -508,16 +506,20 @@ class Wappalyzer {
   }
 
   /**
-   * Analyze JS variables
+   * Analyze JavaScript variables
    */
-  analyzeJS(app, js) {
+  analyzeJs(app, results) {
-    Object.keys(js).forEach(property => {
+
-      var content = js[property]["content"];
+    console.log(app, results);
-      delete js[property]["content"];
+
-      js[property].forEach(pattern => {
+    Object.keys(results).forEach(string => {
-          if ( pattern.regex.test(content) ) {
+      Object.keys(results[string]).forEach(index => {
-            this.addDetected(app, pattern, 'js', content, property);
+        const pattern = this.jsPatterns[app.name][string][index];
-          }
+        const value = results[string][index];
+
+        if ( pattern.regex.test(value) ) {
+          this.addDetected(app, pattern, 'js', value);
+        }
       });
     });
   }