From 3d119f01e2d8c51d1e310a2b1bb39fa798f08359 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Kevin=20R=C3=B6bert?= <dev@xn--rb-fka.it>
Date: Thu, 24 Aug 2017 00:16:07 +0200
Subject: [PATCH] Add new rules and rewrite old rules

+ New rules for:
+ youtube
+ facebook
+ imdb
+ adsensecustomsearchads
+ new global rules

Rewrite old rules regex after the = from "[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?"
to "[^&]*(\\?|&(amp;)?)?", because it is stronger.
---
 data/data.json | 111 ++++++++++++++++++++++++++++++++++---------------
 1 file changed, 78 insertions(+), 33 deletions(-)

diff --git a/data/data.json b/data/data.json
index 6f618b4..da78a76 100644
--- a/data/data.json
+++ b/data/data.json
@@ -4,17 +4,16 @@
             "urlPattern": "(https:\\/\\/||http:\\/\\/).*(\\.amazon\\.)\\w{2,}\\/.*",
             "completeProvider": false,
             "rules": [
-                "pf_rd_[a-zA-Z]=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "qid=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "sr=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "srs=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                ".*(amazon-adsystem\\.com)\\/.*",
+                "pf_rd_[a-zA-Z]=[^&]*(\\?|&(amp;)?)?",
+                "qid=[^&]*(\\?|&(amp;)?)?",
+                "sr=[^&]*(\\?|&(amp;)?)?",
+                "srs=[^&]*(\\?|&(amp;)?)?",
                 ".*(adsensecustomsearchads\\.com)\\/.*",
-                "pd_rd_[a-zA-Z]*=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "__mk_[a-zA-Z]{1,3}_[a-zA-Z]{1,3}=[a-zA-Z0-9\\-\\.\\_\\%]*[\\?|&]?",
-                "url=[a-zA-Z0-9\\-\\.\\_\\%]*[\\?|&]?",
-                "spIA=[a-zA-Z0-9\\-\\.\\_\\%]*[\\?|&]?",
-                "rh=[a-zA-Z0-9\\-\\.\\_\\%]*[\\?|&]?"
+                "pd_rd_[a-zA-Z]*=[^&]*(\\?|&(amp;)?)?",
+                "__mk_[a-zA-Z]{1,3}_[a-zA-Z]{1,3}=[^&]*(\\?|&(amp;)?)?",
+                "url=[^&]*(\\?|&(amp;)?)?",
+                "spIA=[^&]*(\\?|&(amp;)?)?",
+                "rh=[^&]*(\\?|&(amp;)?)?"
             ],
             "exceptions": [
                 ".*(amazon\\.)\\w{2,}(\/gp\/).*\\/redirector.html\\/.*"
@@ -24,27 +23,27 @@
             "urlPattern": "(https:\\/\\/||http:\\/\\/).*(\\.google\\.)\\w{2,}\\/.*",
             "completeProvider": false,
             "rules": [
-                "ved=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "bi[a-zA-Z]*=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "gfe_[a-zA-Z]*=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "ei=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "source=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "gs_[a-zA-Z]*=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "site=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "&\\.[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "oq=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "esrc=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "uact=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "cd=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "cad=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "gws_[a-zA-Z]*=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "im[a-zA-Z]*=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "atyp=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "vet=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "zx=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "_u=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "je=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?",
-                "[a-zA-Z\\_]+id=[a-zA-Z0-9\\-\\.\\_]*[\\?|&]?"
+                "ved=[^&]*(\\?|&(amp;)?)?",
+                "bi[a-zA-Z]*=[^&]*(\\?|&(amp;)?)?",
+                "gfe_[a-zA-Z]*=[^&]*(\\?|&(amp;)?)?",
+                "ei=[^&]*(\\?|&(amp;)?)?",
+                "source=[^&]*(\\?|&(amp;)?)?",
+                "gs_[a-zA-Z]*=[^&]*(\\?|&(amp;)?)?",
+                "site=[^&]*(\\?|&(amp;)?)?",
+                "&\\.[^&]*(\\?|&(amp;)?)?",
+                "oq=[^&]*(\\?|&(amp;)?)?",
+                "esrc=[^&]*(\\?|&(amp;)?)?",
+                "uact=[^&]*(\\?|&(amp;)?)?",
+                "cd=[^&]*(\\?|&(amp;)?)?",
+                "cad=[^&]*(\\?|&(amp;)?)?",
+                "gws_[a-zA-Z]*=[^&]*(\\?|&(amp;)?)?",
+                "im[a-zA-Z]*=[^&]*(\\?|&(amp;)?)?",
+                "atyp=[^&]*(\\?|&(amp;)?)?",
+                "vet=[^&]*(\\?|&(amp;)?)?",
+                "zx=[^&]*(\\?|&(amp;)?)?",
+                "_u=[^&]*(\\?|&(amp;)?)?",
+                "je=[^&]*(\\?|&(amp;)?)?",
+                "[a-zA-Z\\_]+id=[^&]*(\\?|&(amp;)?)?"
             ],
             "exceptions": []
         },
@@ -60,11 +59,26 @@
             "rules": [],
             "exceptions": []
         },
-        "utm": {
+        "globalRules": {
             "urlPattern": ".*",
             "completeProvider": false,
             "rules": [
-                "utm_[a-zA-Z]*=.*[\\?|&]?"
+                "utm_[a-zA-Z]*=[^&]*(\\?|&(amp;)?)?",
+                "ga_source=[^&]*(\\?|&(amp;)?)?",
+                "ga_medium=[^&]*(\\?|&(amp;)?)?",
+                "ga_term=[^&]*(\\?|&(amp;)?)?",
+                "ga_content=[^&]*(\\?|&(amp;)?)?",
+                "ga_campaign=[^&]*(\\?|&(amp;)?)?",
+                "ga_place=[^&]*(\\?|&(amp;)?)?",
+                "yclid=[^&]*(\\?|&(amp;)?)?",
+                "_openstat=[^&]*(\\?|&(amp;)?)?",
+                "fb_action_ids=[^&]*(\\?|&(amp;)?)?",
+                "fb_action_types=[^&]*(\\?|&(amp;)?)?",
+                "fb_ref=[^&]*(\\?|&(amp;)?)?",
+                "fb_source=[^&]*(\\?|&(amp;)?)?",
+                "action_object_map=[^&]*(\\?|&(amp;)?)?",
+                "action_type_map=[^&]*(\\?|&(amp;)?)?",
+                "action_ref_map=[^&]*(\\?|&(amp;)?)?"
             ],
             "exceptions": []
         },
@@ -91,6 +105,37 @@
             "completeProvider": true,
             "rules": [],
             "exceptions": []
+        },
+        "adsensecustomsearchads": {
+            "urlPattern": ".*(adsensecustomsearchads).*",
+            "completeProvider": true,
+            "rules": [],
+            "exceptions": []
+        },
+        "youtube": {
+            "urlPattern": "(https:\\/\\/||http:\\/\\/).*(\\.youtube\\.)\\w{2,}\\/.*",
+            "completeProvider": false,
+            "rules": [
+                "feature=[^&]*(\\?|&(amp;)?)?"
+            ],
+            "exceptions": []
+        },
+        "facebook": {
+            "urlPattern": "(https:\\/\\/||http:\\/\\/).*(\\.facebook\\.)\\w{2,}\\/.*",
+            "completeProvider": false,
+            "rules": [
+                "[a-zA-Z]*ref=[^&]*(\\?|&(amp;)?)?",
+                "hc_location=[^&]*(\\?|&(amp;)?)?"
+            ],
+            "exceptions": []
+        },
+        "imdb": {
+            "urlPattern": "(https:\\/\\/||http:\\/\\/).*(\\.imdb\\.com)\\/.*",
+            "completeProvider": false,
+            "rules": [
+                "ref_=[^&]*(\\?|&(amp;)?)?"
+            ],
+            "exceptions": []
         }
     }  
 }
\ No newline at end of file