clean up

7 files changed, 35 insertions, 116 deletions

diff --git a/.gitignore b/.gitignore
index 3cdd89b..1d3d6ac 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,3 +1,4 @@
-*cache
 *.bak
+*.txt
+*cache
 *~
diff --git a/autopatrol.rb b/autopatrol.rb
index cabc875..6928bf6 100755
--- a/autopatrol.rb
+++ b/autopatrol.rb
@@ -3,7 +3,6 @@
 
 load 'mwapi.rb'
 require 'yaml'
-require 'pp'
 
 mw = MWApi.new('https://wiki.parabolagnulinux.org/api.php')
 credentials = YAML.load_file('credentials.yml')
diff --git a/block-users.rb b/block-users.rb
index e9c2d84..ebf78df 100755
--- a/block-users.rb
+++ b/block-users.rb
@@ -3,7 +3,6 @@
 
 load 'mwapi.rb'
 require 'yaml'
-require 'pp'
 
 mw = MWApi.new('https://wiki.parabolagnulinux.org/api.php')
 credentials = YAML.load_file('credentials.yml')
diff --git a/delete-stdin.rb b/delete-stdin.rb
index 2f835c3..4300d6b 100755
--- a/delete-stdin.rb
+++ b/delete-stdin.rb
@@ -3,7 +3,6 @@
 
 load 'mwapi.rb'
 require 'yaml'
-require 'pp'
 
 mw = MWApi.new('https://wiki.parabolagnulinux.org/api.php')
 credentials = YAML.load_file('credentials.yml')
diff --git a/scratch.rb b/scratch.rb
index f52a562..b725a5a 100755
--- a/scratch.rb
+++ b/scratch.rb
@@ -3,70 +3,39 @@
 
 load 'mwapi.rb'
 require 'yaml'
-require 'pp'
 
 mw = MWApi.new('https://wiki.parabolagnulinux.org/api.php')
 credentials = YAML.load_file('credentials.yml')
 mw.login(credentials['username'], credentials['password'])
 
-# apcontinue = ''
-# while not apcontinue.nil? do
-# 	print "Searching...\n"
-# 	data = mw.query(:list => :allpages,:aplimit => 200, :apcontinue => apcontinue)
-# 	titles = data['query']['allpages'].select{|page| page['title'].split(' ').length > 9 }.map{|page| page['title']}
-# 	print "Deleting #{titles.length} articles...\n"
-# 	if (titles.length > 0)
-# 		mw.delete_by_title(titles, { :reason => 'Spam' })
-# 	end
-# 	if data['query-continue'].nil?
-# 		apcontinue = nil
-# 	else
-# 		apcontinue = data['query-continue']['allpages']['apcontinue']
-# 	end
-# 	print "apcontinue = #{apcontinue.inspect}\n"
-# end
-
-# continue = ''
-# while not continue.nil? do
-# 	print "Searching...\n"
-# 	data = mw.query(:list => :allimages,:ailimit => 200, :aicontinue => continue)
-# 	titles = data['query']['allimages'].select{|page| /^File:[A-Z].* [0-9]?[0-9][0-9][0-9]\.jpg$/ =~ page['title'] }.map{|page| page['title']}
-# 	print "Deleting #{titles.length} articles...\n"
-# 	if (titles.length > 0)
-# 		mw.delete_by_title(titles, { :reason => 'Spam' })
-# 	end
-# 	if data['query-continue'].nil?
-# 		continue = nil
-# 	else
-# 		continue = data['query-continue']['allimages']['aicontinue']
-# 	end
-# 	print "continue = #{continue.inspect}\n"
-# end
-
-
 @keywords=[
 	# brand names
 	/(Crimson|Pink|Purple|Green|Orange) Dye/i,
 	/Air Jordan/i,
 	/Andrew Ting/i,
 	/Beats by Dre/i,
+	/Braun 5270/i,
 	/Buccaneers/i,
 	/Canada Goose/i,
 	/Club Penguin/i,
 	/Diablo 3/i,
 	/Doudoune/i,
+	/Focus T25/i,
 	/Gamma Blue/i,
-	/Garcinia/i,
+	/Google Pagerank/i,
 	/Jeffraham/i,
 	/Jillian Michaels/i,
 	/Jordan Fusion/i,
 	/Jordan Retro/i,
 	/Kate Spade/i,
 	/Michael[ _]Kors/i,
+	/\b(NBA|NFL)\b/i,
+	/\b(world|globe|planet) cup\b/i,
 	/\b49ers\b/i,
 	/\bCisco 200-120\b/i,
 	/\bDr\.? Dre\b/i,
 	/\bGucci\b/i,
+	/\bNike\b/i
 	/\bretro 11 /i,
 	/\buggs?\b/i,
 	/officialnflprostore/i,
@@ -81,14 +50,21 @@ mw.login(credentials['username'], credentials['password'])
 	/(body|excess) weight/i,
 	/Arrhythmia/i,
 	/Cardiovascular/,
+	/Garcinia/i,
 	/P90X/,
 	/Resistance Band/i,
+	/\bbodybuilding\b/i,
+	/\bhypertension\b/i,
+	/\bmuscle\b/i,
+	/\bstairlifts?\b/i,
+	/\bvegan\b/i,
 	/diabetes/i,
 	/dr oz/i,
 	/elliptical (equipment|machines?)/i,
 	/fat (burning|loss)/i,
 	/health care/i,
 	/heart (disease|attack)/i,
+	/hemorrhoids/i,
 	/herbalife/i,
 	/more wellness/i,
 	/pilates/i,
@@ -96,29 +72,35 @@ mw.login(credentials['username'], credentials['password'])
 	/skin care/i,
 	/weight[ -](loss|reduction)/i,
 	# sex topics
-	/\b(sex|adult) cam/i,
 	/\b(male|breast) enhancement\b/i,
-	/\bpenis\b/i,
-	/viagra/i,
-	# other topics 
+	/\b(sex|adult) cam/i,
+	/\b(ejaculation|lesbian|penis|viagra)\b/i,
+	# consumer topics
+	/Possum (Removalist|Infestation)/i,
 	/\b(coffee|tea) extract\b/i,
-	/\b(good|quality) social media\b/i,
 	/\b(green|ginseng) (coffee|tea)\b/i,
-	/\b(world|globe|planet) cup\b/i,
 	/\bbaby shower\b/i,
-	/\bcash loan\b/i,
-	/\bclick here\b/i,
-	/\bcredit (check|repair)\b/i,
-	/\bcredit card\b/i,
 	/\bdiy l[ue]x[ue]ry\b/i,
 	/\bgreen pan\b/i,
 	/\bipage (web)?host/i,
+	/\bjerseys?\b/i,
 	/\bmen.?s (fashion|casual wear|health)\b/i,
 	/\brap beats\b/i,
 	/\breal estate\b/i,
 	/\bsearch engine (marketing|optimization)\b/i,
-	/\bsocial media strategy\b/i,
 	/\bvigorous motivators\b/i,
+	/apartment moves/i,
+	/auto insurance/i,
+	/furniture removal/i,
+	/goji berr(ies|y)/i,
+	/jewellery/i,
+	# internet topics
+	/Twitter follower/i,
+	/\b(good|quality) social media\b/i,
+	/\bclick here\b/i,
+	/\bsocial media strategy\b/i,
+	/affiliate advertising/i,
+	/pay[ -]per[ -]click/i
 	# formats
 	/^(aid|assist) on (where|the place)/i,
 	/^A Background In/i,
@@ -126,22 +108,10 @@ mw.login(credentials['username'], credentials['password'])
 	/1st Impressions in/i,
 	/The (Selection|Choice|Decision) of the .* Is Your/i,
 	# unsorted
-	/\b(chinchilla|jerseys?|vegan|shit|marketing|finance|footwear|shoes|muscle|lesbian|islamist|bodybuilding|nfl|nba|ejaculation|wholesale|nike|Hypertension|sherbet|bankrupt|stairlifts?|outfits|casinos?|surcharges?)\b/i,
+	/\b(chinchilla|shit|marketing|finance|footwear|shoes|islamist|wholesale|sherbet|bankrupt|outfits|casinos?|surcharges?)\b/i,
 	/(Plombier|Serrurier) paris/i,
-	/Possum (Removalist|Infestation)/i,
-	/Google Pagerank/i,
-	/Twitter follower/i,
-	/jewellery/i,
-	/auto insurance/i,
-	/apartment moves/i,
-	/goji berr(ies|y)/i,
-	/affiliate advertising/i,
-	/hemorrhoids/i,
-	/furniture removal/i,
-	/cash online/i,
-	/Braun 5270/i,
-	/Focus T25/i,
-	/pay[ -]per[ -]click/i
+	/\bcash (online|loan)/i,
+	/\bcredit (card|check|repair)\b/i,
 ]
 
 def kw(title)
diff --git a/user-whitelist.rb b/user-whitelist.rb
index f6891a9..a9d3060 100755
--- a/user-whitelist.rb
+++ b/user-whitelist.rb
@@ -3,7 +3,6 @@
 
 load 'mwapi.rb'
 require 'yaml'
-require 'pp'
 
 mw = MWApi.new('https://wiki.parabolagnulinux.org/api.php')
 credentials = YAML.load_file('credentials.yml')
diff --git a/user-whitelist.txt b/user-whitelist.txt
deleted file mode 100644
index 5636806..0000000
--- a/user-whitelist.txt
+++ /dev/null
@@ -1,48 +0,0 @@
-Airon90
-Alfplayer
-Aurelien
-Brendan
-Coadde
-Danilovboas
-Djs
-ElkyDori
-Em3rgent0rdr
-Emulatorman
-Encyclomundi
-Erkan Yilmaz
-Fauno
-Fowlslegs
-Gaming4JC
-GNUtoo
-Habiloid
-Habstinat
-Icarious
-Jaycee
-Jbranso
-Jbsys
-Jorginho
-Jpgg
-Jyaworski
-Kobe
-Leth
-Lukeshu
-Monphey
-Mtjm
-Nicoals
-Opticnerve
-Otyugh
-Parabolawiki
-Quiliro
-Rakyi
-Sayurlodeh
-Sergi
-Shackra
-Smile4ever
-Smv
-Svitostar
-Sysfu
-T67t67
-T67t67t67
-Xihh
-Xylon
-Zykotick9