txt_tm_importer 0.3.1 → 0.4.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 7f16238c6b6ad668720bc31949b5ea75e4cc9079
-  data.tar.gz: ea4f891cbf848221e16cb34ce76200b14934b727
+  metadata.gz: 8a437d5112648108747b10abbaf3b2253ba2bf78
+  data.tar.gz: 86f0997da5a86514c9c3b11ef1a9a251795f769e
 SHA512:
-  metadata.gz: 20b8130a49d629c8404b8ed8ac89e36d7084be82a7c509efc0f3e8d951745c8c90d30b843861d385e482671f62a04ff9de1c375903738c99445bd3d258d2b8ba
-  data.tar.gz: 03143efba5a48ebe19104d197a47f96b29ed8b2aa92d6c4c125d06f4e65f5a47e99cf7a3e48e72b76db485a6eadb0e5a65043101aab0e0f1dc6d8e8a935f93dc
+  metadata.gz: 975b00b591daa6f1561be0e7541dae0a92d1dbc485419c5981a594903ebfc61bb561936484d52499f776a2d9a01b1ceec2f28a595f4e01d4fec0aad0739b0a6b
+  data.tar.gz: c399dbc772ab1764988718efda2c3d6fb291b79dedadf4d468ec68411cbb66ac18293986da19323991e6b9b2ee5ca406bc8b250c52a72beeb12504c1b6a02015

data/lib/txt_tm_importer.rb

@@ -73,6 +73,7 @@ module TxtTmImporter
 
     def import_wordfast_file
       wordfast_lines.each_with_index do |line, index|
+        next if line.empty? || line.gsub(/\s+/, '').empty?
         line_array = line.split("\t")
         @doc[:source_language] = line_array[3].gsub(/%/, '').gsub(/\s/, '') if index.eql?(0)
         @doc[:target_language] = line_array[5].gsub(/%/, '').gsub(/\s/, '') if index.eql?(0)
@@ -96,7 +97,7 @@ module TxtTmImporter
         if line.include?('<Seg')
           write_tu if tu_tracker.eql?(1)
           tu_tracker = 0 if tu_tracker > 0
-          language = line.scan(/(?<=<Seg L=)\S+(?=>)/)[0] if !line.scan(/(?<=<Seg L=)\S+(?=>)/).empty?
+          language = line.scan(/(?<=<Seg L=)[^>]+(?=>)/)[0] if !line.scan(/(?<=<Seg L=)[^>]+(?=>)/).empty?
           if role_counter.eql?(0)
             write_seg(line.scan(/(?<=>).+/)[0], 'source', language)
             role_counter += 1
@@ -111,7 +112,11 @@ module TxtTmImporter
 
     def wordfast_stats
       lines = wordfast_lines
-      @doc[:tu][:counter] = lines.size - 1
+      lines.each_with_index do |line, index|
+        next if line.empty? || line.gsub(/\s+/, '').empty?
+        next if index.eql?(0)
+        @doc[:tu][:counter] += 1
+      end
       @doc[:seg][:counter] = @doc[:tu][:counter] * 2
       @doc[:source_language] = lines[0].split("\t")[3].gsub(/%/, '').gsub(/\s/, '')
       @doc[:target_language] = lines[0].split("\t")[5].gsub(/%/, '').gsub(/\s/, '')
@@ -124,8 +129,8 @@ module TxtTmImporter
       role_counter = 0
       @text.each_line do |line|
         if line.include?('<Seg L=')
-          @doc[:source_language] = line.scan(/(?<=<Seg L=)\S+(?=>)/)[0] if !line.scan(/(?<=<Seg L=)\S+(?=>)/).empty? && role_counter.eql?(0)
-          @doc[:target_language] = line.scan(/(?<=<Seg L=)\S+(?=>)/)[0] if !line.scan(/(?<=<Seg L=)\S+(?=>)/).empty? && role_counter.eql?(1)
+          @doc[:source_language] = line.scan(/(?<=<Seg L=)[^>]+(?=>)/)[0] if !line.scan(/(?<=<Seg L=)\S+(?=>)/).empty? && role_counter.eql?(0)
+          @doc[:target_language] = line.scan(/(?<=<Seg L=)[^>]+(?=>)/)[0] if !line.scan(/(?<=<Seg L=)\S+(?=>)/).empty? && role_counter.eql?(1)
           role_counter += 1 if role_counter.eql?(0)
         end
         @doc[:language_pairs] << [@doc[:source_language], @doc[:target_language]] if !@doc[:source_language].nil? && !@doc[:target_language].nil? && role_counter > 0

data/lib/txt_tm_importer/version.rb

@@ -1,3 +1,3 @@
 module TxtTmImporter
-  VERSION = "0.3.1"
+  VERSION = "0.4.0"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: txt_tm_importer
 version: !ruby/object:Gem::Version
-  version: 0.3.1
+  version: 0.4.0
 platform: ruby
 authors:
 - Kevin S. Dias