From b246fef5b1004328a35af629398655263adff5ff Mon Sep 17 00:00:00 2001
From: anton <smirnoff.anton@gmail.com>
Date: Wed, 2 Jul 2014 12:58:36 +0400
Subject: [PATCH 01/13] remove duplicated declarations

---
 share/apps.json | 2 --
 1 file changed, 2 deletions(-)

diff --git a/share/apps.json b/share/apps.json
index 75fc58a8e..07c5ed167 100644
--- a/share/apps.json
+++ b/share/apps.json
@@ -1092,8 +1092,6 @@
 			"cats": [ 36 ],
 			"env": [ "^google_ad_", "^__google_ad_", "^Goog_AdSense_" ],
 			"script": [ "googlesyndication\\.com/pagead/show_ads\\.js", "ad\\.ca\\.doubleclick\\.net", "2mdn\\.net" ],
-			"env": "^google_ad_",
-			"script": [ "googlesyndication\\.com/pagead/show_ads\\.js", "ad\\.ca\\.doubleclick.net" ]
 		},
 		"Google App Engine": {
 			"website": "code.google.com/appengine",

From 2a429eb6b9a5401b54c2998644c8edc1ccf2f323 Mon Sep 17 00:00:00 2001
From: anton <smirnoff.anton@gmail.com>
Date: Wed, 2 Jul 2014 12:58:36 +0400
Subject: [PATCH 02/13] remove duplicated declarations

---
 share/apps.json | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/share/apps.json b/share/apps.json
index 07c5ed167..464d11852 100644
--- a/share/apps.json
+++ b/share/apps.json
@@ -1091,7 +1091,7 @@
 			"website": "google.com/adsense",
 			"cats": [ 36 ],
 			"env": [ "^google_ad_", "^__google_ad_", "^Goog_AdSense_" ],
-			"script": [ "googlesyndication\\.com/pagead/show_ads\\.js", "ad\\.ca\\.doubleclick\\.net", "2mdn\\.net" ],
+			"script": [ "googlesyndication\\.com/pagead/show_ads\\.js", "ad\\.ca\\.doubleclick\\.net", "2mdn\\.net" ]
 		},
 		"Google App Engine": {
 			"website": "code.google.com/appengine",

From d13ad4151343a5b4037a0d7caaa90000f2ea1fce Mon Sep 17 00:00:00 2001
From: anton <smirnoff.anton@gmail.com>
Date: Wed, 2 Jul 2014 18:39:26 +0400
Subject: [PATCH 03/13] fix purecss.io declaration error

---
 share/apps.json | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/share/apps.json b/share/apps.json
index 464d11852..a00352c0e 100644
--- a/share/apps.json
+++ b/share/apps.json
@@ -2415,7 +2415,7 @@
 		"Pure CSS": {
 			"website": "purecss.io",
 			"cats": [18],
-			"html": "<link[^>]+(?:([\\d.])+/)?pure(?:-min)?\\.css\\;version=\\1"
+			"html": "<link[^>]+(?:([\\d.])+/)?pure(?:-min)?\\.css\\;version:\\1"
 		},
 		"Python": {
 			"website": "python.org",

From d9e80f994b867f79bde12b076c17a1aef7b46650 Mon Sep 17 00:00:00 2001
From: anton <smirnoff.anton@gmail.com>
Date: Wed, 2 Jul 2014 18:48:57 +0400
Subject: [PATCH 04/13] fix shopify detection

---
 share/apps.json | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/share/apps.json b/share/apps.json
index a00352c0e..e67dc8847 100644
--- a/share/apps.json
+++ b/share/apps.json
@@ -2630,7 +2630,7 @@
 		"Shopify": {
 			"website": "shopify.com",
 			"cats": [ 6 ],
-			"html": "<link[^>]+=cdn\\.shopify\\.com",
+			"html": "<link[^>]+=['\"]//cdn\\.shopify\\.com",
 			"env": "^Shopify$"
 		},
 		"Shopware": {

From 679513a7d6a3ab8852e202491b4fdbd3d7fc6716 Mon Sep 17 00:00:00 2001
From: anton <smirnoff.anton@gmail.com>
Date: Wed, 2 Jul 2014 19:15:14 +0400
Subject: [PATCH 05/13] fix weebly detection

---
 share/apps.json | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/share/apps.json b/share/apps.json
index e67dc8847..a72c59731 100644
--- a/share/apps.json
+++ b/share/apps.json
@@ -3192,7 +3192,7 @@
 		"Weebly": {
 			"website": "www.weebly.com",
 			"cats": [ 1 ],
-			"html": "<[^>]+class=\"weebly"
+			"script":"cdn\\d+\\.editmysite\\.com"
 		},
 		"WikkaWiki": {
 			"website": "wikkawiki.org",

From 8157c62a704a2b90778e038050f8634e34db5025 Mon Sep 17 00:00:00 2001
From: anton <smirnoff.anton@gmail.com>
Date: Wed, 2 Jul 2014 19:15:45 +0400
Subject: [PATCH 06/13] fix wix scripts detection

---
 share/apps.json | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/share/apps.json b/share/apps.json
index a72c59731..7c7b17057 100644
--- a/share/apps.json
+++ b/share/apps.json
@@ -3214,7 +3214,7 @@
 		"Wix": {
 			"website": "wix.com",
 			"cats": [ 1 ],
-			"script": "static\\.wix\\.com",
+			"script": "static\\.wixstatic\\.com",
 			"headers": { "X-Wix-Dispatcher-Cache-Hit": ".+", "Set-Cookie": "Domain=\\.wix\\.com" },
 			"env": "^wix(?:Events|Data|Errors)"
 		},

From fc6dadce68a42f5678dfbec169d386b2cee8956d Mon Sep 17 00:00:00 2001
From: anton <smirnoff.anton@gmail.com>
Date: Wed, 2 Jul 2014 19:57:09 +0400
Subject: [PATCH 07/13] add Webs.com cms detection

---
 share/apps.json | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/share/apps.json b/share/apps.json
index 7c7b17057..d155442af 100644
--- a/share/apps.json
+++ b/share/apps.json
@@ -3162,6 +3162,11 @@
 			"cats": [ 1 ],
 			"meta": { "generator": "WEB\\|Publisher" }
 		},
+		"Webs": {
+			"website": "webs.com",
+			"cats": [ 1 ],
+			"headers": {"Server":"Webs.com/?([\\d\\.]+)?\\;version:\\1"}
+		},
 		"Websale": {
 			"website": "websale.de",
 			"cats": [ 6 ],

From 029935cf39dee40e8a4ec45698d1c7d3091488a5 Mon Sep 17 00:00:00 2001
From: anton <smirnoff.anton@gmail.com>
Date: Wed, 2 Jul 2014 19:58:00 +0400
Subject: [PATCH 08/13] remove django-cms detection, it's undetectable can be
 verified on django-cms.org or  www.orange.ch

---
 share/apps.json | 2 --
 1 file changed, 2 deletions(-)

diff --git a/share/apps.json b/share/apps.json
index d155442af..adf6b27e3 100644
--- a/share/apps.json
+++ b/share/apps.json
@@ -743,8 +743,6 @@
 		"Django CMS": {
 			"website": "django-cms.org",
 			"cats": [ 1 ],
-			"script": "media/cms/js/csrf\\.js",
-			"headers": { "Set-Cookie": "django[^;]=" },
 			"implies": "Django"
 		},
 		"Dojo": {

From 0a4b87543b297f200b10b2b633b72b5e8f6a6596 Mon Sep 17 00:00:00 2001
From: anton <smirnoff.anton@gmail.com>
Date: Wed, 2 Jul 2014 20:05:39 +0400
Subject: [PATCH 09/13] fix dotnetnuke detection can be verified on
 http://dotnetnuke.com, http://www.schooldude.com, http://www.sportsdirect.com

---
 share/apps.json | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/share/apps.json b/share/apps.json
index adf6b27e3..b91539084 100644
--- a/share/apps.json
+++ b/share/apps.json
@@ -775,7 +775,8 @@
 			"website": "dotnetnuke.com",
 			"cats": [ 1 ],
 			"meta": { "generator": "DotNetNuke" },
-			"headers": { "X-Compressed-By": "DotNetNuke", "Set-Cookie": "DotNetNukeAnonymous=" },
+            "script":["/js/dnncore\\.js"],
+			"headers": { "DNNOutputCache":".+", "X-Compressed-By": "DotNetNuke", "Set-Cookie": "DotNetNukeAnonymous=" },
 			"html": "<!-- by DotNetNuke Corporation",
 			"env": "^DotNetNuke$",
 			"implies": "Microsoft ASP.NET"
@@ -1202,7 +1203,7 @@
 			"website": "hellobar.com",
 			"cats": [ 5 ],
 			"script": "(hellobar\\.com/hellobar\\.js|new HelloBar)",
-			"html": "/hellobar\\.js" 
+			"html": "/hellobar\\.js"
 		},
 		"Hiawatha": {
 			"website": "hiawatha-webserver.org",
@@ -2729,7 +2730,7 @@
 			"headers": { "solodev_session": ".*" },
 			"html": "<div class='dynamicDiv' id='dd\\.\\d\\.\\d'>",
 			"implies": "PHP"
-		},		
+		},
 		"SPDY": {
 			"website": "chromium.org/spdy",
 			"cats": [ 19 ],

From 66b908c660211309116601ee41d4f64e868c33cf Mon Sep 17 00:00:00 2001
From: anton <smirnoff.anton@gmail.com>
Date: Wed, 2 Jul 2014 20:08:11 +0400
Subject: [PATCH 10/13] fix livestreetcms detection can be verified on
 http://livestreetcms.com/

---
 share/apps.json | 1 +
 1 file changed, 1 insertion(+)

diff --git a/share/apps.json b/share/apps.json
index b91539084..e55f946bb 100644
--- a/share/apps.json
+++ b/share/apps.json
@@ -1662,6 +1662,7 @@
 		"LiveStreet CMS": {
 			"website": "livestreetcms.com",
 			"cats": [ 1 ],
+            "html":["var LIVESTREET_SECURITY_KEY"],
 			"headers": { "X-Powered-By": "LiveStreet CMS" }
 		},
 		"Lockerz Share": {

From 71b50628b45862260b810c6bafcd182521b228fa Mon Sep 17 00:00:00 2001
From: anton <smirnoff.anton@gmail.com>
Date: Wed, 2 Jul 2014 20:33:24 +0400
Subject: [PATCH 11/13] fix cs-cart detecting can be verified on
 http://demo.cs-cart.com/ , http://www.juliettesinteriors.co.uk/ or other from
 http://www.cs-cart.com/live-stores.html

---
 share/apps.json | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/share/apps.json b/share/apps.json
index e55f946bb..9d32d625e 100644
--- a/share/apps.json
+++ b/share/apps.json
@@ -628,7 +628,7 @@
 			"website": "www.cs-cart.com",
 			"cats": [ 6 ],
 			"env": "^fn_compare_strings$",
-			"html": "&nbsp;Powered by (?:<a href=[^>]+cs-cart\\.com|CS-Cart)",
+			"html": ["&nbsp;Powered by (?:<a href=[^>]+cs-cart\\.com|CS-Cart)", "(?:\\$|jQuery)\\.runCart\\('\\w'\\)"],
 			"implies": "PHP"
 		},
 		"CubeCart": {
@@ -1836,7 +1836,7 @@
 		"MODx": {
 			"website": "modxcms.com",
 			"cats": [ 1 ],
-			"html": "(?:<a[^>]+>Powered by MODx</a>|<(?:link|script)[^>]+assets/snippets/)",
+			"html": ["<a[^>]+>Powered by MODx</a>", "<(?:link|script)[^>]+assets/(?:templates|components|snippets)/\\;confidence:80"],
 			"env": "^MODX_MEDIA_PATH$",
 			"headers": { "X-Powered-By": "^MODx", "Set-Cookie": "SN4[a-f\\d]{12}" },
 			"implies": "PHP"

From 05ca2c718e5eb2c30630c5e3dbcdaa9aff77a7e8 Mon Sep 17 00:00:00 2001
From: anton <smirnoff.anton@gmail.com>
Date: Wed, 2 Jul 2014 20:38:46 +0400
Subject: [PATCH 12/13] add wordpress.com detection

---
 share/apps.json | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/share/apps.json b/share/apps.json
index 9d32d625e..814463821 100644
--- a/share/apps.json
+++ b/share/apps.json
@@ -3250,6 +3250,12 @@
 			"env": "^wp_username$",
 			"implies": "PHP"
 		},
+		"WordPress.Com": {
+			"website": "wordpress.com",
+			"cats": [ 1, 11 ],
+			"html": "<link[^>]+s\\d+\\.wp\\.com",
+			"implies": "WordPress"
+		},
 		"WordPress Super Cache": {
 			"website": "ocaoimh.ie/wp-super-cache/",
 			"cats": [ 23 ],

From 065384c34535123061aed3ff432deb4e5357403c Mon Sep 17 00:00:00 2001
From: anton <smirnoff.anton@gmail.com>
Date: Wed, 2 Jul 2014 20:52:24 +0400
Subject: [PATCH 13/13] add raw python driver with base functionality and tests

---
 drivers/python_raw/__init__.py   |   2 +
 drivers/python_raw/real_test.py  | 156 +++++++++++++++++++++++++++++++
 drivers/python_raw/tests.py      | 131 ++++++++++++++++++++++++++
 drivers/python_raw/wappalyzer.py | 135 ++++++++++++++++++++++++++
 links.sh                         |   2 +
 5 files changed, 426 insertions(+)
 create mode 100644 drivers/python_raw/__init__.py
 create mode 100644 drivers/python_raw/real_test.py
 create mode 100644 drivers/python_raw/tests.py
 create mode 100755 drivers/python_raw/wappalyzer.py

diff --git a/drivers/python_raw/__init__.py b/drivers/python_raw/__init__.py
new file mode 100644
index 000000000..df05229a7
--- /dev/null
+++ b/drivers/python_raw/__init__.py
@@ -0,0 +1,2 @@
+#! /usr/bin/python
+# -*- coding: utf-8 -*-
diff --git a/drivers/python_raw/real_test.py b/drivers/python_raw/real_test.py
new file mode 100644
index 000000000..6130f9f5f
--- /dev/null
+++ b/drivers/python_raw/real_test.py
@@ -0,0 +1,156 @@
+#! /usr/bin/python
+# -*- coding: utf-8 -*-
+from wappalyzer import Wappalyzer
+
+TESTS = [
+    {
+        'url': 'http://www.hardgraft.com',
+        'apps': ['jQuery', 'Shopify', 'Nginx']
+    },
+    {
+        'url': 'http://its.bplaced.net',
+        'apps': ['WordPress', 'jQuery', 'Apache']
+    },
+    {
+        'url': 'http://www.bodybuilding.com/',
+        'apps': ['jQuery', 'Optimizely', 'SiteCatalyst', 'Apache Tomcat']
+    },
+    {
+        'url': 'http://guidedhelp21.weebly.com/',
+        'apps': ['Weebly', 'Apache', 'Quantcast', 'Google Analytics', 'jQuery']
+    },
+    {
+        'url': 'http://www.bancadelparque.com/',
+        'apps': ['Wix', 'Twitter Bootstrap']
+    },
+    {
+        'url': 'http://joomla.ru/',
+        'apps': ['Joomla', 'jQuery', 'MooTools', 'Yandex.Metrika', 'LiteSpeed']
+    },
+    {
+        'url': 'http://demoshop21.e-stile.ru/',
+        'apps': ['SiteEdit', 'PHP']
+    },
+    {
+        'url': 'http://umbraco.com',
+        'apps': ['Umbraco', 'IIS', 'Microsoft ASP.NET']
+    },
+    {
+        'url': 'http://johnsciacca.webs.com/',
+        'apps': ['Webs', 'RequireJS', 'Site Meter', 'Modernizr']
+    },
+    {
+        'url': 'http://www.1c-bitrix.ru/',
+        'apps': ['1C-Bitrix', 'Yandex.Metrika']
+    },
+    {
+        'url': 'http://amirocms.com',
+        'apps': ['Amiro.CMS']
+    },
+    {
+        'url': 'http://dle-news.ru',
+        'apps': ['DataLife Engine', 'CloudFlare']
+    },
+    {
+        'url': 'http://dotnetnuke.com',
+        'apps': ['DotNetNuke', 'Microsoft ASP.NET']
+    },
+    {
+        'url': 'http://www.schooldude.com',
+        'apps': ['DotNetNuke', 'Microsoft ASP.NET']
+    },
+    {
+        'url': 'http://www.sportsdirect.com/',
+        'apps': ['DotNetNuke', 'Microsoft ASP.NET']
+    },
+    {
+        'url': 'http://drupal.org',
+        'apps': ['Drupal', 'Varnish']
+    },
+    {
+        'url': 'http://www.komodocms.com/',
+        'apps': ['Komodo CMS']
+    },
+    {
+        'url': 'http://livestreetcms.com/',
+        'apps': ['LiveStreet CMS']
+    },
+    {
+        'url': 'http://modxcms.com/',
+        'apps': ['MODx']
+    },
+    {
+        'url': 'http://modx.ru/',
+        'apps': ['MODx']
+    },
+    {
+        'url': 'http://revo.modx.ru/',
+        'apps': ['MODx']
+    },
+    {
+        'url': 'http://www.punchbrand.com',
+        'apps': ['CS Cart']
+    },
+    {
+        'url': 'http://demo.cs-cart.com/',
+        'apps': ['CS Cart']
+    },
+    {
+        'url': 'https://livedemo.installatron.com/1404307206magento/',
+        'apps': ['Magento']
+    },
+    {
+        'url': 'http://livedemo.installatron.com/1404300689prestashop/',
+        'apps': ['Prestashop']
+    },
+    {
+        'url': 'http://demo.opencart.com/',
+        'apps': ['OpenCart']
+    },
+    {
+        'url': 'https://livedemo.installatron.com/1404307206oscommerce/',
+        'apps': ['osCommerce']
+    },
+    {
+        'url': 'http://www.ubercartdemo.com/',
+        'apps': ['Ubercart']
+    },
+    {
+        'url': 'http://demostore.x-cart.com/',
+        'apps': ['X-Cart']
+    },
+    {
+        'url': 'https://livedemo.installatron.com/1404307206zencart/',
+        'apps': ['Zen Cart']
+    },
+    {
+        'url': 'http://oreonfray83.wordpress.com',
+        'apps': ['WordPress.Com']
+    },
+    {
+        'url': 'http://www.try-phpbb.com/30x/',
+        'apps': ['phpBB']
+    },
+]
+
+
+def test():
+    wappalyzer = Wappalyzer(datafile_path='../../share/apps.json')
+
+    for site in TESTS:
+        print 'testing %s ...' % site['url']
+        result = wappalyzer.analyze(site['url'])
+        for app in site['apps']:
+            found = result.pop(app, None)
+            if found:
+                print '\t%s\t- ok\tconfidence=%d' % (app, found.get_confidence())
+            else:
+                print '\t%s\t- NOT FOUND' % (app)
+                return
+        if result:
+            print '\tUNEXPECTED APPS:'
+            for app_name, app in result.iteritems():
+                print '\t\t%s\t- ok\tconfidence=%d' % (app_name, app.get_confidence())
+
+if __name__ == '__main__':
+    test()
\ No newline at end of file
diff --git a/drivers/python_raw/tests.py b/drivers/python_raw/tests.py
new file mode 100644
index 000000000..51b8120b8
--- /dev/null
+++ b/drivers/python_raw/tests.py
@@ -0,0 +1,131 @@
+import re
+import unittest
+import wappalyzer
+
+
+class FakeUrlopenResponse(object):
+    def __init__(self, url, html, headers):
+        self.url = url
+        self.html = html
+        self.headers = headers
+
+    def read(self):
+        return self.html
+
+    def info(self):
+        _cls = self
+
+        class _Info:
+            @property
+            def dict(self):
+                return _cls.headers
+
+        return _Info()
+
+
+class WappalyzerCustomTestCase(unittest.TestCase):
+    def setUp(self):
+        self.wappalyzer = wappalyzer.Wappalyzer({'categories':[],'apps':[]})
+
+    def get_wappalyzer(self, categories, apps):
+        return wappalyzer.Wappalyzer({'categories': categories, 'apps': apps})
+
+    def test_parse_simple(self):
+        parsed = self.wappalyzer.parse_patterns('control/userimage\\.html')
+        self.assertEqual(1, len(parsed))
+        self.assertTrue(hasattr(parsed[0].regex, 'search'))
+
+    def test_parse_confidence_version(self):
+        parsed = self.wappalyzer.parse_patterns('control/userimage\\.html\\;version:1\\;confidence:80')
+        self.assertEqual(1, len(parsed))
+        self.assertEqual('1', getattr(parsed[0], 'version'))
+        self.assertEqual(80, getattr(parsed[0], 'confidence'))
+
+    def _construct_response(self, url=None, headers=None, html=None):
+        return FakeUrlopenResponse(
+            url=url or '',
+            headers=headers or {},
+            html=html or ''
+        )
+
+    def test_by_url(self):
+        wappalyzer = self.get_wappalyzer(
+            {},
+            {'test1': {'url': 'mysite\d.com'}, 'test2': {'url': 'hissite\d.com'},
+             'test3': {'url': ['my', 'his']}})
+        resp = self._construct_response(url='http://mysite2.com')
+
+        result = wappalyzer.analyze(response=resp)
+
+        self.assertIn('test1', result)
+        self.assertIn('test3', result)
+
+    def test_by_html_with_confidence(self):
+        wappalyzer = self.get_wappalyzer(
+            {},
+            {'test1': {'html': 'body\d\\;confidence:70'}, 'test2': {'html': 'body\w'}})
+        resp = self._construct_response(html='body123')
+
+        result = wappalyzer.analyze(response=resp)
+
+        self.assertIn('test1', result)
+        self.assertEqual(70, result['test1'].get_confidence())
+
+    def test_by_headers(self):
+        wappalyzer = self.get_wappalyzer({},
+            {
+                'test1': {
+                    "headers": {"Server": "debut\\/?([\\d\\.]+)?\\;version:\\1"},
+                }
+            })
+        resp = self._construct_response(headers={"Server": 'debut'})
+        result = wappalyzer.analyze(response=resp)
+        self.assertIn('test1', result)
+
+        resp = self._construct_response(headers={"Server": 'debut/12'})
+        result = wappalyzer.analyze(response=resp)
+        self.assertIn('test1', result)
+
+    def test_by_meta(self):
+        wappalyzer = self.get_wappalyzer({},
+            {
+                'test1': {
+                    "meta": {"generator": "uCore PHP Framework"},
+                },
+                'test2': {
+                    "meta": {"generator2": "0"},
+                }
+            })
+        resp = self._construct_response(html="<html><meta name='generator'  content='uCore PHP Framework'>")
+        result = wappalyzer.analyze(response=resp)
+        self.assertIn('test1', result)
+        self.assertNotIn('test2', result)
+
+        resp = self._construct_response(html="<html><meta property='generator'  content='uCore PHP Framework'>")
+        result = wappalyzer.analyze(response=resp)
+        self.assertIn('test1', result)
+        self.assertNotIn('test2', result)
+
+        resp = self._construct_response(html="<html><meta content='uCore PHP Framework' name='generator'>")
+        result = wappalyzer.analyze(response=resp)
+        self.assertIn('test1', result)
+        self.assertNotIn('test2', result)
+
+    def test_by_scripts(self):
+        wappalyzer = self.get_wappalyzer(
+            {},
+            {'jquery': {
+            "script": ["jquery(?:\\-|\\.)([\\d.]*\\d)[^/]*\\.js\\;version:\\1",
+                       "/([\\d.]+)/jquery(\\.min)?\\.js\\;version:\\1", "jquery.*\\.js"],
+            }})
+        resp = self._construct_response(
+            html='<html><script src="//ajax.googleapis.com/ajax/libs/jquery/1.8.2/jquery.min.js"></script>')
+
+        result = wappalyzer.analyze(response=resp)
+
+        self.assertIn('jquery', result)
+
+
+
+if __name__ == '__main__':
+    unittest.main()
diff --git a/drivers/python_raw/wappalyzer.py b/drivers/python_raw/wappalyzer.py
new file mode 100755
index 000000000..73df758f2
--- /dev/null
+++ b/drivers/python_raw/wappalyzer.py
@@ -0,0 +1,135 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+import os
+import re
+import sys
+import urllib
+from urlparse import urlparse
+
+try:
+    import json
+except ImportError:
+    import simplejson as json
+
+
+class Application(object):
+    def __init__(self, app):
+        self.app = app
+        self.confidence = {}
+        self.detected = False
+
+    def set_detected(self, pattern, type, value, key=None):
+        self.detected = True
+        self.confidence[type + ' ' + (key + ' ' if key else '') + pattern.str] = pattern.confidence
+
+        # todo: detect version
+
+    def get_confidence(self):
+        total = sum(v for v in self.confidence.itervalues())
+        return min(100, total)
+
+
+class Wappalyzer(object):
+    def __init__(self, data=None, datafile_path=None):
+        data = data or self.load_data(datafile_path)
+        self.categories = data['categories']
+        self.apps = data['apps']
+
+    def load_data(self, datafile_path=NotImplementedError):
+        if not datafile_path:
+            file_dir = os.path.dirname(__file__)
+            datafile_path = os.path.join(file_dir, 'apps.json')
+        with open(datafile_path) as f:
+            data = json.load(f)
+        return data
+
+    def analyze(self, url=None, response=None):
+        if not response and not url:
+            raise ValueError
+
+        if not response:
+            response = urllib.urlopen(url)
+
+        url = response.url.split('#')[0]
+        html = response.read()
+        data = {
+            'url': url,
+            'html': html,
+            'script': re.findall(r'<script[^>]+src=(?:"|\')([^"\']+)', html, re.I | re.M),
+            'meta': dict((n.lower(), v) for n, v in
+                         re.findall('<meta\s+(?:name|property)=["\']([^"\']+)["\'].+?content=["\']([^"\']+)["\']', html,
+                                    re.I | re.M) +
+                         [(m2, m1) for m1, m2 in
+                          re.findall('<meta\s+content=["\']([^"\']+)["\'].+?(?:name|property)=["\']([^"\']+)["\']',
+                                     html,
+                                     re.I | re.M)]),
+            'headers': dict((n.lower(), v) for n, v in response.info().dict.iteritems()),
+            'env': None
+        }
+        detected_apps = {}
+
+        for app_name, app in self.apps.iteritems():
+            application = Application(app)
+            for detection_type, patterns in app.iteritems():
+                try:
+                    if detection_type in ['url', 'html']:
+                        for pattern in self.parse_patterns(patterns):
+                            if pattern.regex.search(data[detection_type]):
+                                application.set_detected(pattern, detection_type, data[detection_type])
+                    elif detection_type in ['meta', 'headers']:
+                        for hm_name, hm_pattern in patterns.iteritems():
+                            for pattern in self.parse_patterns(hm_pattern):
+                                value = data[detection_type].get(hm_name.lower())
+                                if value and pattern.regex.search(value):
+                                    application.set_detected(pattern, detection_type, value, hm_name)
+                    elif detection_type in ['script']:
+                        for script in data[detection_type]:
+                            for pattern in self.parse_patterns(patterns):
+                                if pattern.regex.search(script):
+                                    application.set_detected(pattern, detection_type, script)
+                    elif detection_type in ['website', 'excludes', 'cats', 'implies', 'env']:
+                        pass
+                    else:
+                        raise NotImplementedError
+                except:
+                    print 'error while detecting by %s application %s' % (detection_type, app)
+
+            if application.detected:
+                detected_apps[app_name] = application
+
+        return detected_apps
+
+    class Pattern:
+        def __init__(self, str):
+            self.str = str
+            self.regex = re.compile(str, re.I)
+            self.confidence = 100
+
+
+    def parse_patterns(self, patterns):
+        if isinstance(patterns, basestring):
+            patterns = [patterns]
+        elif not isinstance(patterns, list):
+            raise ValueError
+
+        parsed = []
+        for pattern in patterns:
+            parts = pattern.split('\\;')
+            result = Wappalyzer.Pattern(parts[0])
+            for part in parts[1:]:
+                name, value = part.split(':', 1)
+                if name in ['confidence']:
+                    value = float(value)
+                setattr(result, name, value)
+            parsed.append(result)
+
+        return parsed
+
+
+if __name__ == '__main__':
+    try:
+        w = Wappalyzer(sys.argv[1])
+        print w.analyze()
+    except IndexError:
+        print ('Usage: python %s <url>' % sys.argv[0])
diff --git a/links.sh b/links.sh
index 2c7d0b73c..fd1758dd8 100755
--- a/links.sh
+++ b/links.sh
@@ -25,3 +25,5 @@ ln -f share/js/wappalyzer.js      drivers/php/js
 
 ln -f share/apps.json             drivers/python
 ln -f share/js/wappalyzer.js      drivers/python/js
+
+ln -f share/apps.json             drivers/python_raw
\ No newline at end of file