null+****@clear*****
null+****@clear*****
Mon Jan 23 13:05:29 JST 2012
Kouhei Sutou 2012-01-23 13:05:29 +0900 (Mon, 23 Jan 2012)
New Revision: f3bde7f32b630806fc48bb66f4507823d12e5c52
Log:
[external-glossary] share common codes.
Modified files:
lib/logaling.rb
lib/logaling/external_glossaries/debian_project.rb
lib/logaling/external_glossaries/gene95.rb
lib/logaling/external_glossaries/gnome_project.rb
lib/logaling/external_glossaries/postgresql_manual.rb
lib/logaling/external_glossary.rb
Modified: lib/logaling.rb (+3 -1)
===================================================================
--- lib/logaling.rb 2012-01-23 12:42:33 +0900 (91bf34f)
+++ lib/logaling.rb 2012-01-23 13:05:29 +0900 (1c1155a)
@@ -1,6 +1,7 @@
-# -*- encoding: utf-8 -*-
+# -*- coding: utf-8 -*-
#
# Copyright (C) 2011 Miho SUZUKI
+# Copyright (C) 2011 Kouhei Sutou <kou****@clear*****>
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
@@ -25,4 +26,5 @@ module Logaling
class GlossaryNotFound < LogalingError; end
class GlossaryDBNotFound < LogalingError; end
class ExternalGlossaryNotFound < LogalingError; end
+ class UnsupportedFormat < LogalingError; end
end
Modified: lib/logaling/external_glossaries/debian_project.rb (+8 -8)
===================================================================
--- lib/logaling/external_glossaries/debian_project.rb 2012-01-23 12:42:33 +0900 (48339da)
+++ lib/logaling/external_glossaries/debian_project.rb 2012-01-23 13:05:29 +0900 (9cf07cc)
@@ -1,4 +1,5 @@
# Copyright (C) 2011 Miho SUZUKI
+# Copyright (C) 2012 Kouhei Sutou <kou****@clear*****>
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
@@ -21,15 +22,14 @@ module Logaling
description 'Debian JP Project (http://www.debian.or.jp/community/translate/)'
source_language 'en'
target_language 'ja'
+ output_format 'csv'
- def convert
- buffer = ""
- CSV.generate(buffer) do |csv|
- doc = ::Nokogiri::HTML(open("http://www.debian.or.jp/community/translate/trans_table.html", "r"))
- doc.css('dl').each do |dl|
- dl.children.each_slice(3) do |dt, dd, _|
- csv << [dt.text, dd.text.gsub(/(^\/|\/$)/,'')]
- end
+ private
+ def convert_to_csv(csv)
+ doc = ::Nokogiri::HTML(open("http://www.debian.or.jp/community/translate/trans_table.html", "r"))
+ doc.css('dl').each do |dl|
+ dl.children.each_slice(3) do |dt, dd, _|
+ csv << [dt.text, dd.text.gsub(/(^\/|\/$)/,'')]
end
end
end
Modified: lib/logaling/external_glossaries/gene95.rb (+23 -22)
===================================================================
--- lib/logaling/external_glossaries/gene95.rb 2012-01-23 12:42:33 +0900 (c6617e8)
+++ lib/logaling/external_glossaries/gene95.rb 2012-01-23 13:05:29 +0900 (1667233)
@@ -1,4 +1,5 @@
-# Copyright (C) 2012 Koji SHIMADA
+# Copyright (C) 2012 Koji SHIMADA
+# Copyright (C) 2012 Kouhei Sutou <kou****@clear*****>
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
@@ -23,33 +24,33 @@ module Logaling
description 'GENE95 Dictionary (http://www.namazu.org/~tsuchiya/sdic/data/gene.html)'
source_language 'en'
target_language 'ja'
+ output_format 'csv'
- def convert
- CSV.generate do |csv|
- puts "downloading gene95 dictionary..."
- url = 'http://www.namazu.org/~tsuchiya/sdic/data/gene95.tar.gz'
- Zlib::GzipReader.open(open(url)) do |gz|
- puts "importing gene95 dictionary..."
+ private
+ def convert_to_csv(csv)
+ puts "downloading gene95 dictionary..."
+ url = 'http://www.namazu.org/~tsuchiya/sdic/data/gene95.tar.gz'
+ Zlib::GzipReader.open(open(url)) do |gz|
+ puts "importing gene95 dictionary..."
- Gem::Package::TarReader.new(gz) do |tar|
- tar.each do |entry|
- case entry.full_name
- when "gene.txt"
- lines = StringIO.new(entry.read).each_line
+ Gem::Package::TarReader.new(gz) do |tar|
+ tar.each do |entry|
+ case entry.full_name
+ when "gene.txt"
+ lines = StringIO.new(entry.read).each_line
- 2.times { lines.next } # skip header
+ 2.times { lines.next } # skip header
- preprocessed_lines = lines.map.map do |line|
- line.encode("UTF-8", "CP932",
- undef: :replace, replace: '').chomp
- end
+ preprocessed_lines = lines.map.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
- else
- # ignore
+ preprocessed_lines.each_slice(2) do |source, target|
+ csv << [source.sub(/( .*)/, ''), target]
end
+ else
+ # ignore
end
end
end
Modified: lib/logaling/external_glossaries/gnome_project.rb (+7 -7)
===================================================================
--- lib/logaling/external_glossaries/gnome_project.rb 2012-01-23 12:42:33 +0900 (3ac5d8e)
+++ lib/logaling/external_glossaries/gnome_project.rb 2012-01-23 13:05:29 +0900 (06d3837)
@@ -1,4 +1,5 @@
# Copyright (C) 2011 Miho SUZUKI
+# Copyright (C) 2012 Kouhei Sutou <kou****@clear*****>
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
@@ -21,14 +22,13 @@ module Logaling
description 'GNOME Translation Project Ja (http://live.gnome.org/TranslationProjectJa)'
source_language 'en'
target_language 'ja'
+ output_format 'csv'
- def convert
- buffer = ""
- CSV.generate(buffer) do |csv|
- doc = ::Nokogiri::HTML(open("http://www.gnome.gr.jp/l10n/trans-terms.html", "r"))
- doc.css('table tr')[1..-1].each do |tr|
- csv << [tr.children[0].text, tr.children[1].text]
- end
+ private
+ def convert_to_csv(csv)
+ doc = ::Nokogiri::HTML(open("http://www.gnome.gr.jp/l10n/trans-terms.html", "r"))
+ doc.css('table tr')[1..-1].each do |tr|
+ csv << [tr.children[0].text, tr.children[1].text]
end
end
end
Modified: lib/logaling/external_glossaries/postgresql_manual.rb (+8 -8)
===================================================================
--- lib/logaling/external_glossaries/postgresql_manual.rb 2012-01-23 12:42:33 +0900 (95238aa)
+++ lib/logaling/external_glossaries/postgresql_manual.rb 2012-01-23 13:05:29 +0900 (473a8dc)
@@ -1,4 +1,5 @@
# Copyright (C) 2011 Miho SUZUKI
+# Copyright (C) 2012 Kouhei Sutou <kou****@clear*****>
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
@@ -21,15 +22,14 @@ module Logaling
description 'PostgreSQL7.1 Manual (http://osb.sraoss.co.jp/PostgreSQL/Manual/)'
source_language 'en'
target_language 'ja'
+ output_format 'csv'
- def convert
- buffer = ""
- CSV.generate(buffer) do |csv|
- doc = ::Nokogiri::HTML(open("http://osb.sraoss.co.jp/PostgreSQL/Manual/word.html", "r:iso-2022-jp").read.encode("utf-8"))
- doc.css('table table tr')[2..-1].each do |tr|
- if tr.children[2]
- csv << [tr.children[2].text, tr.children[4].text]
- end
+ private
+ def convert_to_csv(csv)
+ doc = ::Nokogiri::HTML(open("http://osb.sraoss.co.jp/PostgreSQL/Manual/word.html", "r:iso-2022-jp").read.encode("utf-8"))
+ doc.css('table table tr')[2..-1].each do |tr|
+ if tr.children[2]
+ csv << [tr.children[2].text, tr.children[4].text]
end
end
end
Modified: lib/logaling/external_glossary.rb (+20 -2)
===================================================================
--- lib/logaling/external_glossary.rb 2012-01-23 12:42:33 +0900 (ab2b09c)
+++ lib/logaling/external_glossary.rb 2012-01-23 13:05:29 +0900 (0d9a0a7)
@@ -56,14 +56,32 @@ class Logaling::ExternalGlossary
def target_language val=nil
@target_language ||= val
end
+
+ def output_format(*args)
+ if args.empty?
+ @output_format ||= "csv"
+ else
+ @output_format = args.first
+ end
+ end
end
def import
- File.open(import_file_name, "w") {|f| f.write(self.convert) }
+ File.open(import_file_name, "w") do |output|
+ output_format = self.class.output_format
+ output_format = output_format.to_s if output_format.is_a?(Symbol)
+ case output_format
+ when "csv"
+ convert_to_csv(CSV.new(output))
+ else
+ raise UnsupportedFormat, "unsupported format: <#{output_format}>"
+ end
+ end
end
private
def import_file_name
- [self.class.name, self.class.source_language, self.class.target_language, 'csv'].join('.')
+ [self.class.name, self.class.source_language,
+ self.class.target_language, self.class.output_format].join('.')
end
end