1 files changed, 149 insertions, 0 deletions
diff --git a/scratch.rb b/scratch.rb
new file mode 100755
index 0000000..2c3bcf6
--- /dev/null
+++ b/scratch.rb
@@ -0,0 +1,149 @@
+#!/usr/bin/env ruby
+# -*- coding: utf-8 -*-
+
+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$/ =~ page['title'] }.map{|page| page['title']}
+# 	print "Deleting #{titles.length} articles...\n"
+# 	if (titles.length > 0)
+# 		mw.delete_by_title(titles, { :reason => 'Spam' })
+# 	end
+# 	if data['query-continue'].nil?
+# 		continue = nil
+# 	else
+# 		continue = data['query-continue']['allimages']['aicontinue']
+# 	end
+# 	print "continue = #{continue.inspect}\n"
+# end
+
+
+@keywords=[
+	# brand names
+	/(Crimson|Pink|Purple|Green|Orange) Dye/i,
+	/Air Jordan/i,
+	/Andrew Ting/i,
+	/Beats by Dre/i,
+	/Buccaneers/i,
+	/Canada Goose/i,
+	/Club Penguin/i,
+	/Diablo 3/i,
+	/Doudoune/i,
+	/Gamma Blue/i,
+	/Garcinia/i,
+	/Jeffraham/i,
+	/Jordan Fusion/i,
+	/Jordan Retro/i,
+	/Kate Spade/i,
+	/Michael[ _]Kors/i,
+	/\b49ers\b/i,
+	/\bCisco 200-120\b/i,
+	/\bDr\.? Dre\b/i,
+	/\bGucci\b/i,
+	/\bretro 11 /i,
+	/\buggs?\b/i,
+	/officialnflprostore/i,
+	# script kiddie topics
+	/ on Hack Wi-Fi$/i,
+	/Cracked Steam/i,
+	/Psn code generator/i,
+	/Steam Key Generator/i,
+	/\bpc games? (free|crack)/i,
+	/crack pc/i,
+	# health topics
+	/Resistance Band/i,
+	/diabetes/i,
+	/elliptical (equipment|machines?)/i,
+	/fat burning/i,
+	/health care/i,
+	/heart (disease|attack)/i,
+	/more wellness/i,
+	/pilates/i,
+	/skin care/i,
+	/weight loss/i,
+	/(body|excess) weight/i,
+	/Arrhythmia/i,
+	/Cardiovascular/,
+	# sex topics
+	/\b(sex|adult) cam/i,
+	/\b(male|breast) enhancement\b/i,
+	/\bpenis\b/i,
+	# other topics 
+	/\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,
+	/\bmen.?s (fashion|casual wear|health)\b/i,
+	/\brap beats\b/i,
+	/\bsearch engine marketing\b/i,
+	/\bsocial media strategy\b/i,
+	/\bvigorous motivators\b/i,
+	# formats
+	/^(aid|assist) on (where|the place)/i,
+	/^A Background In/i,
+	/^An? ( (simple|informative))? analysis of /i,
+	/1st Impressions in/i,
+	/The (Selection|Choice|Decision) of the .* Is Your/i,
+	# unsorted
+	/\b(jerseys?|vegan|shit|marketing|finance|footwear|shoes|muscle|lesbian|islamist|bodybuilding|nfl|nba|ejaculation|wholesale|nike|Hypertension|sherbet|bankrupt|stairlifts?|outfits|surcharges?)\b/i,
+]
+
+def kw(title)
+	@keywords.each do |re|
+		if re =~ title
+			return true
+		end
+	end
+	return false
+end
+
+apcontinue = ''
+while not apcontinue.nil? do
+	print "Searching...\n"
+	data = mw.query(:list => :allpages, :aplimit => 200, :apcontinue => apcontinue)
+	titles = data['query']['allpages'].select{|page| kw(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?
+		apcontinue = nil
+	else
+		apcontinue = data['query-continue']['allpages']['apcontinue']
+	end
+	print "apcontinue = #{apcontinue.inspect}\n"
+end