tmx_importer 0.5.0 → 1.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 07447d45355fb271c832db8228367a0d14d04c5f
-  data.tar.gz: ab468fdf9295379cd036b351648f37eb3477468f
+  metadata.gz: bea8f8bb85029d0b336fdf89adecdcce179fc7df
+  data.tar.gz: c005d43b81b20c87a88b15f0814f63550655240c
 SHA512:
-  metadata.gz: 508a95ae93f7a43af68362b97ab429d5f746486ad2bc2216174d35490b1c9952e8059e5eb111074c05ba40055d1e255d34d23439caae5ebfcebfa6dcceb0116e
-  data.tar.gz: 5aa98d1fb3a4ed65554f149c9d8be21fc2a3fe2b9cc02034bb63718ec1a7ec68049da278c9793d036cf450aaf7099ce1799eccc01e554e724d69e52614d11daa
+  metadata.gz: 464e050de392148bd8348bc7ce23f5834f9f112cad6deedf971ba1f8a8095e2f9adac1d8ff670f9031546696e54296cac5213f9ac3e8faf90691dcf0fb373c71
+  data.tar.gz: 180830aae81b47dcff661e1b90435558226457c177ca1a10a691ccbd8d6d08f45f0e8705d1154524975dc2de607a4c6d0dc7e0829b1e7bb3aed1155291bed43e

data/README.md

@@ -23,8 +23,9 @@ gem 'tmx_importer'
 
 ```ruby
 # Get the high level stats of a TMX file
+# Including the encoding is optional. If not included the gem will attempt to detect the encoding.
 file_path = File.expand_path('../tmx_importer/spec/test_sample_files/test_tm(utf-8).tmx')
-TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-8').stats
+TmxImporter::Tmx.new(file_path: file_path).stats
 # => {:tu_count=>4, :seg_count=>8, :language_pairs=>[["de-DE", "en-US"]]}
 
 # Extract the segments of a TMX file

data/lib/tmx_importer.rb

@@ -7,17 +7,22 @@ require 'charlock_holmes'
 module TmxImporter
   class Tmx
     attr_reader :file_path, :encoding
-    def initialize(file_path:, encoding:)
+    def initialize(file_path:, **args)
       @file_path = file_path
-      @encoding = encoding.upcase
+      if args[:encoding].nil?
+        @encoding = CharlockHolmes::EncodingDetector.detect(File.read(@file_path)[0..100_000])[:encoding]
+      else
+        @encoding = args[:encoding].upcase
+      end
       @doc = {
         source_language: "",
         tu: { id: "", counter: 0, vals: [], lang: "", creation_date: "" },
         seg: { lang: "", counter: 0, vals: [], role: "" },
         language_pairs: []
       }
-      @text = CharlockHolmes::Converter.convert(File.read(open(@file_path)), encoding, 'UTF-8') if !@encoding.eql?('UTF-8')
+      raise "Encoding type could not be determined. Please set an encoding of UTF-8, UTF-16LE, or UTF-16BE" if @encoding.nil?
       raise "Encoding type not supported. Please choose an encoding of UTF-8, UTF-16LE, or UTF-16BE" unless @encoding.eql?('UTF-8') || @encoding.eql?('UTF-16LE') || @encoding.eql?('UTF-16BE')
+      @text = CharlockHolmes::Converter.convert(File.read(open(@file_path)), @encoding, 'UTF-8') if !@encoding.eql?('UTF-8')
     end
 
     def stats

data/lib/tmx_importer/version.rb

@@ -1,3 +1,3 @@
 module TmxImporter
-  VERSION = "0.5.0"
+  VERSION = "1.1.0"
 end

data/spec/tmx_importer_spec.rb

@@ -5,36 +5,36 @@ describe TmxImporter do
     expect(TmxImporter::VERSION).not_to be nil
   end
 
-  it 'raises an error if the encoding is not supported' do
-    file_path = File.expand_path('../tmx_importer/spec/test_sample_files/test_tm(utf-8).tmx')
-    -> { expect(TmxImporter::Tmx.new(file_path: file_path, encoding: 'ISO-8859-9').stats).to raise_error }
-  end
-
   it 'raises an error if the wrong encoding is specified in the file' do
     file_path = File.expand_path('../tmx_importer/spec/test_sample_files/bad_encoding.tmx')
-    -> { expect(TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-8').stats).to raise_error }
+    -> { expect(TmxImporter::Tmx.new(file_path: file_path, encoding: 'ISO-8859-9').stats).to raise_error }
   end
 
   it 'raises an error if the file contains bad markup' do
     file_path = File.expand_path('../tmx_importer/spec/test_sample_files/bad_markup(utf-8).tmx')
-    -> { expect(TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-8').stats).to raise_error }
+    -> { expect(TmxImporter::Tmx.new(file_path: file_path).stats).to raise_error }
   end
 
   it 'raises an error if the file contains bad markup 2' do
     file_path = File.expand_path('../tmx_importer/spec/test_sample_files/bad_markup(utf-16).tmx')
-    -> { expect(TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-16le').stats).to raise_error }
+    -> { expect(TmxImporter::Tmx.new(file_path: file_path).stats).to raise_error }
+  end
+
+  it 'raises an error if the encoding can not be determined' do
+    file_path = File.expand_path('../tmx_importer/spec/test_sample_files/test_tm(utf-16LE).tmx')
+    -> { expect(TmxImporter::Tmx.new(file_path: file_path).import).to raise_error }
   end
 
   describe '#stats' do
     it 'reports the stats of a UTF-8 TMX file' do
       file_path = File.expand_path('../tmx_importer/spec/test_sample_files/test_tm(utf-8).tmx')
-      tmx = TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-8')
+      tmx = TmxImporter::Tmx.new(file_path: file_path)
       expect(tmx.stats).to eq({:tu_count=>4, :seg_count=>8, :language_pairs=>[["de-DE", "en-US"]]})
     end
 
     it 'reports the stats of a UTF-8 TMX file 2' do
       file_path = File.expand_path('../tmx_importer/spec/test_sample_files/test_tm_2(utf-8).tmx')
-      tmx = TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-8')
+      tmx = TmxImporter::Tmx.new(file_path: file_path)
       expect(tmx.stats).to eq({:tu_count=>4, :seg_count=>8, :language_pairs=>[["de", "en"]]})
     end
 
@@ -46,31 +46,31 @@ describe TmxImporter do
 
     it 'reports the stats of a UTF-16LE BOM TMX file' do
       file_path = File.expand_path('../tmx_importer/spec/test_sample_files/test_tm(utf-16LE BOM).tmx')
-      tmx = TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-16le')
+      tmx = TmxImporter::Tmx.new(file_path: file_path)
       expect(tmx.stats).to eq({:tu_count=>4, :seg_count=>8, :language_pairs=>[["de-DE", "en-US"]]})
     end
 
     it 'reports the stats of a multiple language pair TMX file' do
       file_path = File.expand_path('../tmx_importer/spec/test_sample_files/multiple_language_pairs.tmx')
-      tmx = TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-8')
+      tmx = TmxImporter::Tmx.new(file_path: file_path)
       expect(tmx.stats).to eq({:tu_count=>4, :seg_count=>10, :language_pairs=>[["de-DE", "en-US"], ["de-DE", "it"], ["de-DE", "fr"]]})
     end
 
     it 'reports the stats of a srclang equals *all* TMX file' do
       file_path = File.expand_path('../tmx_importer/spec/test_sample_files/srclang_all.tmx')
-      tmx = TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-8')
+      tmx = TmxImporter::Tmx.new(file_path: file_path)
       expect(tmx.stats).to eq({:tu_count=>4, :seg_count=>10, :language_pairs=>[["de-DE", "en-US"], ["de-DE", "it"], ["de-DE", "fr"]]})
     end
 
     it 'reports the stats of a TMX file with out of order segments' do
       file_path = File.expand_path('../tmx_importer/spec/test_sample_files/out_of_order_segments.tmx')
-      tmx = TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-8')
+      tmx = TmxImporter::Tmx.new(file_path: file_path)
       expect(tmx.stats).to eq({:tu_count=>4, :seg_count=>8, :language_pairs=>[["de-DE", "en-US"]]})
     end
 
     it 'imports a TMX file with UTF-16 LE BOM encoding' do
       file_path = File.expand_path('../tmx_importer/spec/test_sample_files/strange_encoding.tmx')
-      tmx = TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-16LE')
+      tmx = TmxImporter::Tmx.new(file_path: file_path)
       expect(tmx.stats).to eq({:tu_count=>1, :seg_count=>2, :language_pairs=>[["tr", "en"]]})
     end
   end
@@ -78,13 +78,13 @@ describe TmxImporter do
   describe '#import' do
     it 'imports a UTF-8 TMX file' do
       file_path = File.expand_path('../tmx_importer/spec/test_sample_files/test_tm(utf-8).tmx')
-      tmx = TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-8').import
+      tmx = TmxImporter::Tmx.new(file_path: file_path).import
       expect(tmx[1][2][3]).to eq("de-DE")
     end
 
     it 'imports a UTF-8 TMX file 2' do
       file_path = File.expand_path('../tmx_importer/spec/test_sample_files/test_tm_2(utf-8).tmx')
-      tmx = TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-8')
+      tmx = TmxImporter::Tmx.new(file_path: file_path)
       expect(tmx.import[1][2][4]).to eq("Rückenlehneneinstellung")
     end
 
@@ -96,37 +96,37 @@ describe TmxImporter do
 
     it 'imports a UTF-16LE BOM TMX file' do
       file_path = File.expand_path('../tmx_importer/spec/test_sample_files/test_tm(utf-16LE BOM).tmx')
-      tmx = TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-16le')
+      tmx = TmxImporter::Tmx.new(file_path: file_path)
       expect(tmx.import[1][2][3]).to eq("de-DE")
     end
 
     it 'imports a multiple language pair TMX file' do
       file_path = File.expand_path('../tmx_importer/spec/test_sample_files/multiple_language_pairs.tmx')
-      tmx = TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-8')
+      tmx = TmxImporter::Tmx.new(file_path: file_path)
       expect(tmx.import[1][2][3]).to eq("it")
     end
 
     it 'imports a srclang equals *all* TMX file' do
       file_path = File.expand_path('../tmx_importer/spec/test_sample_files/srclang_all.tmx')
-      tmx = TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-8')
+      tmx = TmxImporter::Tmx.new(file_path: file_path)
       expect(tmx.import[1][2][3]).to eq("it")
     end
 
     it 'imports a TMX file with out of order segments' do
       file_path = File.expand_path('../tmx_importer/spec/test_sample_files/out_of_order_segments.tmx')
-      tmx = TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-8')
+      tmx = TmxImporter::Tmx.new(file_path: file_path)
       expect(tmx.import[1][2][3]).to eq("en-US")
     end
 
     it 'imports a TMX file with out of order segments' do
       file_path = File.expand_path('../tmx_importer/spec/test_sample_files/out_of_order_segments.tmx')
-      tmx = TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-8')
+      tmx = TmxImporter::Tmx.new(file_path: file_path)
       expect(tmx.import[1][2][1]).to eq("target")
     end
 
     it 'imports a TMX file with UTF-16 LE BOM encoding' do
       file_path = File.expand_path('../tmx_importer/spec/test_sample_files/strange_encoding.tmx')
-      tmx = TmxImporter::Tmx.new(file_path: file_path, encoding: 'utf-16LE')
+      tmx = TmxImporter::Tmx.new(file_path: file_path)
       expect(tmx.import[1][1][3]).to eq("en")
     end
   end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: tmx_importer
 version: !ruby/object:Gem::Version
-  version: 0.5.0
+  version: 1.1.0
 platform: ruby
 authors:
 - Kevin S. Dias