Sha256: 0f6bf1e69ab27a1b693971b1a920342d95090a9f0a8a1f5540dea8dcb548b294
Contents?: true
Size: 1.22 KB
Versions: 12
Compression:
Stored size: 1.22 KB
Contents
module Gluttonberg
module Content
require 'despamilator/filter'
module DespamilatorFilter
class NumbersAndWords < Despamilator::Filter
def parse subject
text = tidy_text(subject)
[
/\w\d+/,
/\d+\w/,
/\d+($|\b)/
].each do |regexp|
matches = text.scan(regexp)
next if matches.empty?
matches.each do |to_remove|
to_remove = to_remove.to_s
text.sub!(to_remove, '') unless to_remove.empty?
subject.register_match!({:score => 0.1, :filter => self})
end
end
end
def name
'Numbers next to words'
end
def description
'Detects unusual number/word combinations'
end
private
def tidy_text subject
text = subject.text.without_uris
text.downcase!
# strip out "good numbers"
text.gsub!(/h[1-6]/, '')
text.gsub!(/(^|\b)\d+($|\b)/, '')
text.gsub!(/(^|\b)\d+(,|\.)\d+($|\b)/, '')
text.gsub!(/(^|\b)\d+(st|nd|rd|th)($|\b)/, '')
text
end
end
end
end #Content
end #Gluttonberg
Version data entries
12 entries across 12 versions & 1 rubygems