From cf1b7eb8e815decbd285e7228c5d71f825479793 Mon Sep 17 00:00:00 2001
From: Bardi Harborow <bardi@bardiharborow.com>
Date: Sat, 2 Sep 2017 13:44:11 +0000
Subject: [PATCH] [Wolfram.com] Update rulesets.

---
 src/chrome/content/rules/Wolfram.com.xml      | 38 +++++---
 src/chrome/content/rules/Wolfram_Alpha.xml    | 97 ++++++++++++++-----
 src/chrome/content/rules/Wolfram_CDN.com.xml  | 17 +++-
 src/chrome/content/rules/Wolframcloud.com.xml | 38 +++++++-
 utils/ruleset-whitelist.csv                   |  4 +-
 5 files changed, 140 insertions(+), 54 deletions(-)

diff --git a/src/chrome/content/rules/Wolfram.com.xml b/src/chrome/content/rules/Wolfram.com.xml
index 87a0c214ba8e..d8c6af5c2abd 100644
--- a/src/chrome/content/rules/Wolfram.com.xml
+++ b/src/chrome/content/rules/Wolfram.com.xml
@@ -1,32 +1,38 @@
 <!--
-	Other Wolfram Research rulesets:
+	For other Wolfram Research coverage:
 
 		- Wolfram_Alpha.xml
 		- Wolfram_CDN.com.xml
+		- Wolframcloud.com.xml
 
+	Problematic hosts in *wolfram.com:
 
-	Nonfunctional subdomains:
+		- challenges ᵐ
+		- datarepository ᵐ
+		- mvid ᵐ
+		- open.exploration ᵐ
+		- open.sandbox ᵐ
+		- streamingmedia (timeout)
 
-		- support	(times out)
+	ᵐ Mismatched
 
 -->
-<ruleset name="Wolfram.com (partial) ">
+<ruleset name="Wolfram (partial)">
 
 	<target host="wolfram.com" />
-	<target host="*.wolfram.com" />
+	<target host="www.wolfram.com" />
+	<target host="account.wolfram.com" />
+	<target host="amoeba.wolfram.com" />
+	<target host="billing.wolfram.com" />
+	<target host="calendar.wolfram.com" />
+	<target host="community.wolfram.com" />
+	<target host="education.wolfram.com" />
+	<target host="reference.wolfram.com" />
+	<target host="user.wolfram.com" />
 
-
-	<!--securecookie host="^\.wolfram\.com$" name="^WR_SID$" /-->
 	<securecookie host="^user\.wolfram\.com$" name=".+" />
 
-
-	<rule from="^http://(?:www\.)?wolfram\.com/"
-		to="https://www.wolfram.com/" />
-
-	<rule from="^http://user\.wolfram\.com/"
-		to="https://user.wolfram.com/" />
-
-	<rule from="^http://reference\.wolfram\.com/"
-		to="https://reference.wolfram.com/" />
+	<rule from="^http:"
+		to="https:" />
 
 </ruleset>
diff --git a/src/chrome/content/rules/Wolfram_Alpha.xml b/src/chrome/content/rules/Wolfram_Alpha.xml
index c2baea422ea8..4b837cde66a6 100644
--- a/src/chrome/content/rules/Wolfram_Alpha.xml
+++ b/src/chrome/content/rules/Wolfram_Alpha.xml
@@ -1,34 +1,79 @@
 <!--
-Invalid certificate:
-	api-cn.wolframalpha.com
-	api-maps.wolframalpha.com
-	api-tw.wolframalpha.com
-	dev1.cvs.wolframalpha.com
-	dev2.cvs.wolframalpha.com
-	reg.cvs.wolframalpha.com
-	sit.cvs.wolframalpha.com
-	cvsrx.wolframalpha.com
-	cvsrx.devel.wolframalpha.com
-	www-cn.wolframalpha.com
-	www-tw.wolframalpha.com
-	volunteer.wolframalpha.com
-
-Refused:
-	community.wolframalpha.com
-
-Timeout:
-	blog.wolframalpha.com
-	nba.devel.wolframalpha.com
-	homeworkday.wolframalpha.com
-	www2.wolframalpha.com
+	For other Wolfram Research coverage:
+
+		- Wolfram.com.xml
+		- Wolfram_CDN.com.xml
+		- Wolframcloud.com.xml
+
+	Refused hosts:
+
+		- api-cn
+		- api-maps 
+		- api-tw
+		- blog
+		- citi
+		- community
+		- www[0-9][0-9]
+
+	SHA-1 certificate hosts:
+
+		- cvsrx
+		- nielsen
+		- volunteer
+		- www-cn
+		- www-tw
+
+	Timed out hosts:
+
+		- ^
+		- homeworkday
+
+	Mismatched hosts:
+
+		- cvsrx.devel
+
 -->
 <ruleset name="Wolfram Alpha">
+
+	<target host="wolframalpha.com" />
 	<target host="*.wolframalpha.com" />
+
+	<test url="http://api.wolframalpha.com/" />
+	<test url="http://api-cn.wolframalpha.com/" />
+	<test url="http://api-maps.wolframalpha.com/" />
+	<test url="http://api-tw.wolframalpha.com/" />
+	<test url="http://developer.wolframalpha.com/" />
+	<test url="http://m.wolframalpha.com/" />
+	<test url="http://preview.wolframalpha.com/" />
+	<test url="http://products.wolframalpha.com/" />
+	<test url="http://volunteer.wolframalpha.com/" />
+	<test url="http://wc.wolframalpha.com/" />
+	<test url="http://www.wolframalpha.com/" />
+	<test url="http://www1.wolframalpha.com/" />
 	<test url="http://www10.wolframalpha.com/" />
 	<test url="http://www20.wolframalpha.com/" />
+	<test url="http://www3.wolframalpha.com/" />
+	<test url="http://www30.wolframalpha.com/" />
+	<test url="http://www4b.wolframalpha.com/" />
+	<test url="http://www4c.wolframalpha.com/" />
+	<test url="http://www4d.wolframalpha.com/" />
+	<test url="http://www4f.wolframalpha.com/" />
+	<test url="http://www5a.wolframalpha.com/" />
+	<test url="http://www5b.wolframalpha.com/" />
+
+	<rule from="^http://wolframalpha\.com/"
+		to="https://www.wolframalpha.com/" />
+
+	<rule from="^http://api-(cn|maps|tw)\.wolframalpha\.com/"
+		to="https://www.wolframalpha.com/" />
+
+	<rule from="^http://volunteer\.wolframalpha\.com/"
+		to="https://www.wolframalpha.com/" />
+
+	<rule from="^http://www[0-9]?[0-9a-z]?\.wolframalpha\.com/"
+		to="https://www.wolframalpha.com/" />
+
+	<rule from="^http://(api|developer|m|preview|products|wc)\.wolframalpha\.com/"
+		to="https://$1.wolframalpha.com/" />
 
-	<rule from="^http://api-(cn|maps|tw)\.wolframalpha\.com/" to="https://www.wolframalpha.com/" />
-	<rule from="^http://volunteer\.wolframalpha\.com/" to="https://www.wolframalpha.com/" />
-	<rule from="^http://(www[0-9][0-9]\.)wolframalpha\.com/" to="https://www.wolframalpha.com/" />
-	<rule from="^http:" to="https:" />
 </ruleset>
diff --git a/src/chrome/content/rules/Wolfram_CDN.com.xml b/src/chrome/content/rules/Wolfram_CDN.com.xml
index c4958241ba14..4cff2fafc4fe 100644
--- a/src/chrome/content/rules/Wolfram_CDN.com.xml
+++ b/src/chrome/content/rules/Wolfram_CDN.com.xml
@@ -1,12 +1,21 @@
 <!--
-	For other Wolfram Research coverage, see Wolfram.com.xml.
+	For other Wolfram Research coverage:
+	
+		- Wolfram.com.xml
+		- Wolfram_Alpha.xml
+		- Wolframcloud.com.xml
+
+	Problematic hosts in *wolframcdn.com:
+
+		- ^ (https timeout)
 
 -->
 <ruleset name="Wolfram CDN.com">
 
 	<target host="www.wolframcdn.com" />
+	<target host="dl.wolframcdn.com" />
 
+	<rule from="^http:"
+		to="https:" />
 
-	<rule from="^http:" to="https:" />
-
-</ruleset>
\ No newline at end of file
+</ruleset>
diff --git a/src/chrome/content/rules/Wolframcloud.com.xml b/src/chrome/content/rules/Wolframcloud.com.xml
index 44b307cce4ca..a6a671cb650c 100644
--- a/src/chrome/content/rules/Wolframcloud.com.xml
+++ b/src/chrome/content/rules/Wolframcloud.com.xml
@@ -1,8 +1,36 @@
+<!--
+
+	For other Wolfram Research coverage:
+
+		- Wolfram.com.xml
+		- Wolfram_Alpha.xml
+		- Wolfram_CDN.com.xml
+
+	Problematic hosts in *wolframcloud.com:
+
+		- ^ (no https)
+
+-->
 <ruleset name="Wolframcloud.com">
-  <target host="wolframcloud.com" />
-  <target host="www.wolframcloud.com" />
-  <target host="programming.wolframcloud.com" />
 
-  <rule from="^http://(?:www\.)?wolframcloud\.com/" to="https://www.wolframcloud.com/" />
-  <rule from="^http://programming\.wolframcloud\.com/" to="https://programming.wolframcloud.com/" />
+	<target host="wolframcloud.com" />
+	<target host="www.wolframcloud.com" />
+	<target host="billing.wolframcloud.com" />
+	<target host="channelbroker.wolframcloud.com" />
+	<target host="datadrop.wolframcloud.com" />
+	<target host="datarepository.wolframcloud.com" />
+	<target host="develop.wolframcloud.com" />
+	<target host="lab.wolframcloud.com" />
+	<target host="mathematica.wolframcloud.com" />
+	<target host="one.wolframcloud.com" />
+	<target host="open.wolframcloud.com" />
+	<target host="programming.wolframcloud.com" />
+	<target host="reference.wolframcloud.com" />
+	
+	<rule from="^http://wolframcloud\.com/"
+		to="https://www.wolframcloud.com/" />
+	
+	<rule from="^http:"
+		to="https:" />
+
 </ruleset>
diff --git a/utils/ruleset-whitelist.csv b/utils/ruleset-whitelist.csv
index 9da986c875ac..88bbbb55e8ff 100644
--- a/utils/ruleset-whitelist.csv
+++ b/utils/ruleset-whitelist.csv
@@ -3628,9 +3628,7 @@ b859ff3f78e0cc95b3eaf81ec38153ffd631b75974902b50e181c2b9a73b08f5,1,1,WinSuperSit
 0812bcf9185f25ab18b3a7f034f332203cdd4f3de385ab2b5e2737e83e9d969c,1,1,Wishabi.xml
 83ef353b9aab96e56973c3d8150c1f1e3b6cff4f68cad34cd754f3e2de999108,1,1,WiTopia.net.xml
 107b0e53285b3cd5b876c03108d666cba1dfac53af5a62e0190c990cb08c4cc8,1,1,Wizard-Internet-Services.xml
-b9f8cc3d837e320d48965b779ff0977fb14329226b2cff5ccf86f8b6b583f698,1,1,Wolfram_Alpha.xml
-e1440a89fa9e0be3d51832b2c2d6f5f63d7edcbc30ef786a77bca5c8d9c7e4bb,0,1,Wolframcloud.com.xml
-9cfc632cb53d9f01e6d82e0f36008181b0a4010ba453d154c20629bf9964832f,1,1,Wolfram.com.xml
+a1b7325cf1b8968daca8a883ab2a7d04a6902283ff93d0a7889bcd6e5420451c,1,0,Wolfram_Alpha.xml
 e237009069a1f7e25b8d4d83fb77aa2f273405abbeafa0f3d9f70110cc08c309,1,1,Womens_Health_Specialists.xml
 f7999de568e74fdd703ef7c2aaa057ce2dc1a418d112ea5cc64aa6d29e57915f,1,1,Wonderlandmovies.xml
 072d4bc8ac964f1db9fc4b3d1113dbe48d854b975c89839f3d5395226fee2b33,1,1,Woobox.com.xml