jam-cloud/ruby/lib/jam_ruby/models/max_mind_isp.rb

57 lines
1.7 KiB
Ruby

module JamRuby
class MaxMindIsp < ActiveRecord::Base
self.table_name = 'max_mind_isp'
def self.import_from_max_mind(file)
# File Geo-142
# Format:
# "beginIp","endIp","countryCode","ISP"
MaxMindIsp.transaction do
MaxMindIsp.delete_all
File.open(file, 'r:ISO-8859-1') do |io|
io.gets # eat the copyright line. gah, why do they have that in their file??
MaxMindIsp.pg_copy_from io, :map => { 'beginIp' => 'ip_bottom', 'endIp' => 'ip_top', 'countryCode' => 'country', 'ISP' => 'isp'}, :columns => [:beginIp, :endIp, :countryCode, :ISP] do |row|
row[0] = ip_address_to_int(strip_quotes(row[0]))
row[1] = ip_address_to_int(strip_quotes(row[1]))
row[2] = row[2]
row[3] = row[3..-1].join(',') # this is because the parser just cuts on any ',' and ignores double quotes. essentially postgres-copy isn't a great csv parser -- or I need to configure it better
while row.length > 4
row.delete_at(4)
end
end
end
end
end
# Make an IP address fit in a signed int. Just divide it by 2, as the least significant part
# just can't possibly matter. We can verify this if needed. My guess is the entire bottom octet is
# actually irrelevant
def self.ip_address_to_int(ip)
ip.split('.').inject(0) {|total,value| (total << 8 ) + value.to_i} / 2
end
private
def self.strip_quotes str
return nil if str.nil?
if str.chr == '"'
str = str[1..-1]
end
if str.rindex('"') == str.length - 1
str = str.chop
end
return str
end
def self.escape str
str.gsub(/\"/, '""')
end
end
end