null+****@clear*****
null+****@clear*****
Sat Jan 21 12:12:05 JST 2012
SHIMADA Koji 2012-01-21 12:12:05 +0900 (Sat, 21 Jan 2012)
New Revision: e26c69d5847563e549f28e1f1c19e4a3d92abe03
Merged c742d06: Merge pull request #21 from logaling/import-gene95
Log:
Refactor Gene95#convert
Modified files:
lib/logaling/external_glossaries/gene95.rb
Modified: lib/logaling/external_glossaries/gene95.rb (+11 -8)
===================================================================
--- lib/logaling/external_glossaries/gene95.rb 2012-01-21 12:11:23 +0900 (b3dea7c)
+++ lib/logaling/external_glossaries/gene95.rb 2012-01-21 12:12:05 +0900 (2bcd16d)
@@ -25,15 +25,18 @@ module Logaling
def convert
CSV.generate do |csv|
puts "downloading gene95 dictionary..."
- Zlib::GzipReader.open(open('http://www.namazu.org/~tsuchiya/sdic/data/gene95.tar.gz')) do |gz|
- contents = false
+ url = 'http://www.namazu.org/~tsuchiya/sdic/data/gene95.tar.gz'
+ Zlib::GzipReader.open(open(url)) do |gz|
puts "importing gene95 dictionary..."
- gz.readlines.map{|l| l.encode("UTF-8", "CP932", undef: :replace, replace: '').strip }.each_slice(2) do |source, target|
- if contents
- csv << [ source.split(" ").map(&:strip)[0], target ]
- else
- contents = true
- end
+
+ 2.times { gz.gets } # skip header
+
+ preprocessed_lines = gz.readlines.map do |line|
+ line.encode("UTF-8", "CP932", undef: :replace, replace: '').chomp
+ end
+
+ preprocessed_lines.each_slice(2) do |source, target|
+ csv << [source.sub(/( .*)/, ''), target]
end
end
end