gtin2atc 0.1.4 → 0.1.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 88ed6172e67a5d527bfa4c4f55fe1d49d4af214f
-  data.tar.gz: b7daf74015a79a3f797037eaa0cc8c45f84ec3b8
+  metadata.gz: b6d462072d516aec948e6cf2ec404d60d78634d0
+  data.tar.gz: 965cfb24672bc54f53ec3cd347104dda20d8d39b
 SHA512:
-  metadata.gz: a98367cbe6b716cee86116d9e100b7ea0fd8c44a85e5b7bb633d292f0f04de78d907865c9eebddf9fb74609161dbe3e69464d8353287ffc2f19596e51318a3c7
-  data.tar.gz: a7d711d4fa0305d21540c6445669e09082734a99d0b30b9cfa4f05494ff242cadb365d2d89bc9d46d504a52959c5938e4675e0e435d7f33a43a4c749b180ca0c
+  metadata.gz: 5a9490c9a9d5efa5498153c8020b9119cfba390d7d09fc59524decb26f497502402f01ab79f10fdb5ae1757fe65be9a4b839a430d0f21c4d4e53130d7ba2c6b1
+  data.tar.gz: d1acea60db039502005e629dbb48f611dd9f4c18c50ef58c3bbf7426cb843090cf227bb4d6e88ef3f51f1aef1529afc7428e7567491b02e25a0b95e66578fdee

data/History.txt

@@ -1,5 +1,12 @@
+=== 0.1.5 04.03.2015
+
+* Added optional second parameter to specify output file
+* Added output for dosis (name, unit, qty) for mono-preparations
+
 === 0.1.4 03.03.2015
 
+* Add selling units ot generated gtin2atc.csv file
+* Use ';' as column separator for generated CSV files
 * Create file pharmacode_gtin_not_found.txt
 
 === 0.1.3 27.01.2015

data/bin/gtin2atc

@@ -23,12 +23,12 @@ end
 
 opts = options.opts
 startTime = Time.now
-if ARGV.size == 1 and File.exists?(ARGV[0])
-  args = []
-  IO.readlines(ARGV[0]).each{ |x| args << x.chomp}
+if File.exists?(ARGV[0])
+  gtins2parse = []
+  IO.readlines(ARGV[0]).each{ |x| gtins2parse << x.chomp}
 else
-  args = ARGV.clone
+  gtins2parse = ARGV.clone
 end
-Gtin2atc::Builder.new(opts).run(args)
+Gtin2atc::Builder.new(opts).run(gtins2parse, ARGV[1])
 diff = (Time.now-startTime).to_i
 Gtin2atc::Util.debug_msg "#{File.basename(__FILE__)} done. Took #{diff} seconds"

data/lib/gtin2atc/builder.rb

@@ -14,11 +14,13 @@ module Gtin2atc
     AtcNotInSwissmedic    = 'atc not in swissmedic'
     AtcNotInBag           = 'atc not in bag'
     AtcDifferent          = 'atc differed'
+    CsvOutputOptions      =  { :col_sep => ';', :encoding => 'UTF-8'}
     def initialize(opts)
       Util.set_logging(opts[:log])
+      @output = opts[:output]
       @do_compare = opts[:compare]
       @gen_reports = opts[:compare] and opts[:full]
-      Util.debug_msg "Builder: opts are #{opts} @do_compare is #{@do_compare}"
+      Util.debug_msg "Builder: opts are #{opts} @do_compare is #{@do_compare} output #{@output}"
       @data_swissmedic = {}
       @data_bag = {}
       @data_swissindex = {}
@@ -93,6 +95,31 @@ module Gtin2atc
       Util.debug_msg "swissindex_xml_extractor extracted #{data.size} items"
       data
     end
+    def oddb_calc_xml_extractor
+      filename = 'oddb_calc.xml'
+      data = {}
+      unless File.exists?('oddb_calc.xml')
+        puts "Unable to open #{filename}"
+      else
+        xml = IO.read(filename)
+        Util.debug_msg "oddb_calc_xml_extractor xml is #{xml.size} bytes long"
+        result = ARTICLESEntry.parse(xml.sub(Strip_For_Sax_Machine, ''), :lazy => true)
+        result.ARTICLES.ARTICLE.each do |article|
+          item = {}
+          gtin = article.GTIN.to_i
+          item[:gtin]            = gtin
+          item[:PKG_SIZE]        = article.PKG_SIZE
+          item[:SELLING_UNITS]   = article.SELLING_UNITS
+          item[:MEASURE]         = article.MEASURE
+          if article.COMPOSITIONS.COMPONENT and article.COMPOSITIONS.COMPONENT.size == 1
+            item[:COMPOSITIONS]    = article.COMPOSITIONS
+          end
+          data[gtin]     = item
+        end
+        Util.debug_msg "oddb_calc_xml_extractor extracted #{data.size} items"
+      end
+      data
+    end
     def bag_xml_extractor
       data = {}
       @bag = BagDownloader.new
@@ -121,24 +148,37 @@ module Gtin2atc
       Util.debug_msg "bag_xml_extractor extracted #{data.size} items. Skipped #{@bag_entries_without_gtin} entries without gtin"
       data
     end
-    def run(gtins_to_parse=[])
+
+    def run(gtins_to_parse=[], output_name=nil)
       Util.debug_msg("run #{gtins_to_parse}")
       Util.debug_msg("@use_swissindex true")
+      @oddb_calc = oddb_calc_xml_extractor
       @data_epha_atc = epha_atc_extractor
       @data_swissindex = swissindex_xml_extractor
       emitted_ids = []
-      output_name =  File.join(Util.get_archive, @do_compare ? 'gtin2atc_swissindex.csv' : 'gtin2atc.csv')
-      CSV.open(output_name,'w+') do |csvfile|
-        csvfile << ["gtin", "ATC", 'pharmacode', 'description', 'daily drug dose']
+      if @do_compare
+        output_name =  File.join(Util.get_archive, 'gtin2atc_swissindex.csv')
+      else
+        output_name ||= 'gtin2atc.csv'
+        output_name =  File.join(Util.get_archive, output_name)
+      end
+      CSV.open(output_name,'w+', CsvOutputOptions) do |csvfile|
+        csvfile << ["gtin", "ATC", 'pharmacode', 'description', 'daily drug dose', 'selling units']
         @data_swissindex.sort.each do |gtin, item|
           if @do_compare or gtins_to_parse.size == 0 or
             gtins_to_parse.index(gtin.to_s) or
-            gtins_to_parse.index(item[:pharmacode])
+            gtins_to_parse.index(item[:pharmacode].to_s)
             atc = item[:atc_code]
             ddd = @data_epha_atc[atc]
-            emitted_ids << gtin if gtin
-            emitted_ids << item[:pharmacode] if item[:pharmacode]
-            csvfile << [gtin, atc, item[:pharmacode], item[:description], ddd]
+            selling_units = @oddb_calc[gtin] ? @oddb_calc[gtin][:SELLING_UNITS] : nil
+            emitted_ids << gtin.to_i if gtin
+            emitted_ids << item[:pharmacode].to_i if item[:pharmacode]
+            if @oddb_calc[gtin] and @oddb_calc[gtin][:COMPOSITIONS]
+              comp = @oddb_calc[gtin][:COMPOSITIONS].COMPONENT.first
+              csvfile << [gtin, atc, item[:pharmacode], item[:description], ddd, selling_units, comp.NAME, comp.QTY, comp.UNIT]
+            else
+              csvfile << [gtin, atc, item[:pharmacode], item[:description], ddd, selling_units]
+            end
           end
         end
       end
@@ -147,16 +187,16 @@ module Gtin2atc
       missing_ids = []
       gtins_to_parse.each{
         |id|
-          next if emitted_ids.index(id)
+          next if emitted_ids.index(id.to_i)
           missing_ids << id
       }
-      File.open('pharmacode_gtin_not_found.txt', 'w+') { |f| f.write missing_ids.join("\n") }
+      File.open('pharmacode_gtin_not_found.txt', 'w+', CsvOutputOptions) { |f| f.write missing_ids.join("\n") }
       msg = "swissindex: Could not find info for #{missing_ids.size} missing ids see file pharmacode_gtin_not_found.txt"
       Util.debug_msg(msg)
       return unless @do_compare
       @data_bag = bag_xml_extractor
       output_name =  File.join(Util.get_archive, 'gtin2atc_bag.csv')
-      CSV.open(output_name,'w+') do |csvfile|
+      CSV.open(output_name,'w+', CsvOutputOptions) do |csvfile|
         csvfile << ["gtin", "ATC", 'description']
         @data_bag.sort.each do |gtin, item|
           csvfile << [gtin, item[:atc_code], item[:description]]
@@ -165,7 +205,7 @@ module Gtin2atc
       Util.debug_msg "bag: Extracted #{gtins_to_parse.size} of #{@data_bag.size} items into #{output_name} for #{gtins_to_parse}"
       @data_swissmedic = swissmedic_xls_extractor
       output_name =  File.join(Util.get_archive, 'gtin2atc_swissmedic.csv')
-      CSV.open(output_name,'w+') do |csvfile|
+      CSV.open(output_name,'w+', CsvOutputOptions) do |csvfile|
         csvfile << ["gtin", "ATC", 'description']
         @data_swissmedic.sort.each do |gtin, item|
           csvfile << [gtin, item[:atc_code], item[:pharmacode], item[:description]]

data/lib/gtin2atc/downloader.rb

@@ -129,7 +129,7 @@ module Gtin2atc
       if File.exists?(file) and diff_hours = ((Time.now-File.ctime(file)).to_i/3600) and diff_hours < 24
         Util.debug_msg "Skip download of #{file} as only #{diff_hours} hours old"
       else
-        FileUtils.rm_f(file, :verbose => true)
+        FileUtils.rm_f(file, :verbose => false)
         begin
           response = @agent.get(@url)
           response.save_as(file)

data/lib/gtin2atc/options.rb

@@ -16,9 +16,13 @@ module Gtin2atc
   <<EOS
 #$0 ver.#{Gtin2atc::VERSION}
 Usage:
-  gtin2atc [--compare] [--log] [file_with_gtin or gtin or pharmacode] [gtin..]
+  gtin2atc [--compare] [--log] [file_with_gtin or gtin or pharmacode] [output_file]
     If file_with_gtin is given only the GTIN (or pharamacode) (one per line) is outputted.
+    If a second parameter output_file is given the name of the generated csv file.
     If no file or gtin is given, alle GTIN will be processed.
+    If a file oddb_calc.xml (produced via oddb2xml --calc) is found in the current directory,
+    its content is read to add the daily drug dose and dose information for each article.
+
     --log               log important actions
 
     --compare           download an compare GTIN/ATC_code from BAG, SwissIndex and RefData

data/lib/gtin2atc/version.rb

@@ -1,3 +1,3 @@
 module Gtin2atc
-  VERSION = "0.1.4"
+  VERSION = "0.1.5"
 end

data/lib/gtin2atc/xml_definitions.rb

@@ -248,3 +248,50 @@ class MedicalInformationsEntry
   element :medicalInformations, :class => MedicalInformationsContent
 end
 
+class COMPONENTContent
+  include SAXMachine
+  element :NAME
+  element :QTY
+  element :UNIT
+end
+
+class COMPONENTEntry
+  include SAXMachine
+  element :COMPONENT, :class => COMPONENTContent
+end
+
+class COMPONENTSContent
+  include SAXMachine
+  element :components, :class => COMPONENTContent
+end
+
+class COMPOSITIONSContent
+  include SAXMachine
+  attribute :ReleaseDate
+  elements :COMPONENT, :class => COMPONENTContent
+end
+
+
+class ARTICLEContent
+  include SAXMachine
+  element :GTIN
+  element :NAME
+  element :PKG_SIZE
+  element :MEASURE
+  element :SELLING_UNITS
+  element :GALENIC_FORM
+  element :GALENIC_GROUP
+  element :COMPOSITIONS, :class => COMPOSITIONSContent
+end
+
+class ARTICLESContent
+  include SAXMachine
+  attribute :ReleaseDate
+  elements :ARTICLE, :class => ARTICLEContent
+end
+
+class ARTICLESEntry
+  include SAXMachine
+  element :ARTICLES, :class => ARTICLESContent
+end
+

data/spec/builder_spec.rb

@@ -48,11 +48,10 @@ describe Gtin2atc::Builder do
   def check_csv(filename)
     File.exists?(filename).should eq true
     inhalt = IO.readlines(filename)
-    puts inhalt
-    /^\d{13},\w{4}/.should match inhalt[1]
+    /^\d{13};\w{4}/.should match inhalt[1]
     # Packungsgrösse, Dosierung, DDD, Route of Administration
-    /^gtin,ATC,pharmacode,description,daily drug dose/.should match inhalt.first
-    /^7680316440115,B03AA07,20244,FERRO-GRADUMET Depottabl,"0,2 g O Fe2\+"/.should match inhalt.join("\n")
+    /^gtin;ATC;pharmacode;description;daily drug dose/.should match inhalt.first
+    /^7680316440115;B03AA07;20244;FERRO-GRADUMET Depottabl;0,2 g O Fe2\+;/.should match inhalt.join("\n")
   end
 
   context 'when 20273 41803 (Pharmacodes) is given' do
@@ -102,8 +101,8 @@ describe Gtin2atc::Builder do
       check_csv(CSV_NAME)
       inhalt = IO.readlines(CSV_NAME)
       inhalt.size.should eq 2+1 # one header lines + two items
-      inhalt[1].chomp.should eq '7680147690482,N07BC02,41803,KETALGIN Inj Lös 10 mg/ml,"25 mg O,P"'
-      inhalt[2].chomp.should eq '7680353660163,B03AE10,20273,KENDURAL Depottabl,'
+      inhalt[1].chomp.should eq '7680147690482;N07BC02;41803;KETALGIN Inj Lös 10 mg/ml;25 mg O,P;'
+      inhalt[2].chomp.should eq '7680353660163;B03AE10;20273;KENDURAL Depottabl;;'
     end
   end
 
@@ -123,15 +122,15 @@ describe Gtin2atc::Builder do
       /7680353660163/.match(inhalt[1]).should == nil
       /7680147690482/.match(inhalt[2]).should == nil
       /7680353660163/.match(inhalt[2]).should_not == nil
-      /7680353660163,B03AE10,20273,KENDURAL Depottabl/.match(inhalt[2]).should_not == nil
+      /7680353660163;B03AE10;20273;KENDURAL Depottabl/.match(inhalt[2]).should_not == nil
     end
   end
 
   def check_csv(filename)
     File.exists?(filename).should eq true
     inhalt = IO.readlines(filename)
-    /^gtin,ATC/.match(inhalt.first).should_not == nil
-    /^\d{13},\w{4}/.match(inhalt[1]).should_not == nil
+    /^gtin;ATC/.match(inhalt.first).should_not == nil
+    /^\d{13};\w{4}/.match(inhalt[1]).should_not == nil
   end
 
   context 'when --compare is given' do
@@ -144,8 +143,8 @@ describe Gtin2atc::Builder do
     it 'should produce three correct csv' do
       @res = buildr_capture(:stdout){ cli.run }
       check_csv('gtin2atc_bag.csv')
-      check_csv('gtin2atc_swissindex.csv')
       check_csv('gtin2atc_swissmedic.csv')
+      check_csv('gtin2atc_swissindex.csv')
     end
 
     it 'should produce a good logging output' do
@@ -186,4 +185,34 @@ describe Gtin2atc::Builder do
 
   end
 
+  context 'when outputfile is given' do
+    let(:cli) do
+      options = Gtin2atc::Options.new
+      options.parser.parse!('7680316440115 tst.csv'.split(' '))
+      Gtin2atc::Builder.new(options.opts)
+    end
+
+    it 'should produce a correct tst.csv' do
+      tst_name = 'tst.csv'
+      buildr_capture(:stdout){ cli.run(["7680316440115"], tst_name) }
+      check_csv(tst_name)
+    end
+  end
+
+  context 'when producing csv' do
+    let(:cli) do
+      options = Gtin2atc::Options.new
+      Gtin2atc::Builder.new(options.opts)
+    end
+
+    it 'should contain dose for FERRO-GRADUMET Depottab' do
+      oddb_calc_xml = File.expand_path(File.join( __FILE__, '../data/oddb_calc.xml'))
+      FileUtils.cp(oddb_calc_xml, Gtin2atc::WorkDir, :verbose => false)
+      @res = buildr_capture(:stdout){ cli.run() }
+      check_csv(CSV_NAME)
+      inhalt = IO.readlines(CSV_NAME)
+      inhalt.index("7680316440115;B03AA07;20244;FERRO-GRADUMET Depottabl;0,2 g O Fe2+;30;Ferrum(ii);105.0;mg\n").should_not == nil
+    end
+  end
+
 end