gtin2atc 0.1.3 → 0.1.4

checksums.yaml

@@ -1,15 +1,7 @@
 ---
-!binary "U0hBMQ==":
-  metadata.gz: !binary |-
-    ZDRhOWFiNzFhNzM0YmE0N2NlYTA0NWI5MDEyOGRkNjY5YWZjYTU3Yg==
-  data.tar.gz: !binary |-
-    YmZlMDkxMWRiM2VkNTY0MDZiMjk0YWFkYjM0MGY3ZDFjMDEzMTBlYw==
+SHA1:
+  metadata.gz: 88ed6172e67a5d527bfa4c4f55fe1d49d4af214f
+  data.tar.gz: b7daf74015a79a3f797037eaa0cc8c45f84ec3b8
 SHA512:
-  metadata.gz: !binary |-
-    Yzk0OGY2MTU2NmY2ZmY0NmRiYTM2MGI4Mzc0NzdhNjFmM2RlMGJjNTVjZDhm
-    NmEzMDI1NjYxNzk0Njg2MDhhYTQ4ZDNkZDNiZTUyYzE0MTRiODdiZmYxODc5
-    NjJhMDk3MDY1ODRlMzBmY2YxNzUyMTM4ZDQwNzZjNzNjMDQ0MTY=
-  data.tar.gz: !binary |-
-    ZDk1NDkxZmY0NGY3OTZjYzY3MTYwMzFmMmFkMjQ3OTExZmRmZWE2YTcxZDQ0
-    MmUxNWE4OGRiOGRiZmIzMGEyNGQwYmYzMTA1NWQzYmEwMTBiODFkNTcxNmVj
-    MWE1YzAyNDg3OTgyYjJlMTg4ZmQ3YTI4ZjhhOWNhYTExNjllMGU=
+  metadata.gz: a98367cbe6b716cee86116d9e100b7ea0fd8c44a85e5b7bb633d292f0f04de78d907865c9eebddf9fb74609161dbe3e69464d8353287ffc2f19596e51318a3c7
+  data.tar.gz: a7d711d4fa0305d21540c6445669e09082734a99d0b30b9cfa4f05494ff242cadb365d2d89bc9d46d504a52959c5938e4675e0e435d7f33a43a4c749b180ca0c

data/History.txt

@@ -1,3 +1,7 @@
+=== 0.1.4 03.03.2015
+
+* Create file pharmacode_gtin_not_found.txt
+
 === 0.1.3 27.01.2015
 
 * Clearer filenames

data/lib/gtin2atc/builder.rb

@@ -4,6 +4,7 @@ require "gtin2atc/options"
 require "gtin2atc/downloader"
 require "gtin2atc/xml_definitions"
 require 'mechanize'
+require 'open-uri'
 
 module Gtin2atc
   class Builder
@@ -33,6 +34,18 @@ module Gtin2atc
       end
       ((10-(sum%10))%10).to_s
     end
+    def epha_atc_extractor
+      data = {}
+      body = open('https://download.epha.ch/cleaned/atc.csv').read.force_encoding('UTF-8')
+      Util.debug_msg "epha_atc_extractor is #{body.size} bytes long"
+      csv = CSV.new(body, { :headers => false, :col_sep => '|' } )
+      csv.to_a.each{
+        |line|
+        data[line[0]] = line[2] if line[2]
+      }
+      Util.debug_msg "epha_atc_extractor extracted #{data.size} items"
+      data
+    end
     def swissmedic_xls_extractor
       @swissmedic = SwissmedicDownloader.new
       filename = @swissmedic.download
@@ -111,20 +124,35 @@ module Gtin2atc
     def run(gtins_to_parse=[])
       Util.debug_msg("run #{gtins_to_parse}")
       Util.debug_msg("@use_swissindex true")
+      @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']
+        csvfile << ["gtin", "ATC", 'pharmacode', 'description', 'daily drug dose']
         @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])
-            csvfile << [gtin, item[:atc_code], item[:pharmacode], item[:description]]
+            gtins_to_parse.index(gtin.to_s) or
+            gtins_to_parse.index(item[:pharmacode])
+            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]
           end
         end
       end
       msg = "swissindex: Extracted #{gtins_to_parse.size} of #{@data_swissindex.size} items into #{output_name} for #{gtins_to_parse}"
       Util.debug_msg(msg)
+      missing_ids = []
+      gtins_to_parse.each{
+        |id|
+          next if emitted_ids.index(id)
+          missing_ids << id
+      }
+      File.open('pharmacode_gtin_not_found.txt', 'w+') { |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')
@@ -148,7 +176,6 @@ module Gtin2atc
       check_swissmedic
       compare
     end
-    # require 'pry';
     def check_bag
       matching_atc_codes = []
 
@@ -201,8 +228,6 @@ module Gtin2atc
         end
         total1 = not_in_swissindex + match_in_swissindex + longer_in_swissindex +  shorter_in_swissindex + different_atc_in_swissindex
         total2 = not_in_swissmedic + match_in_swissmedic + longer_in_swissmedic +  shorter_in_swissmedic + different_atc_in_swissmedic
-        # binding.pry if j != (total1 + matching_atc_codes)
-        # binding.pry if j != (total2 + matching_atc_codes)
         # Util.debug_msg "#{gtin}: j #{j} finished #{total1} #{total2} #{atc_code} matching_atc_codes #{matching_atc_codes}"
       }
       Util.info  "Result of verifing data from bag (SL):

data/lib/gtin2atc/version.rb

@@ -1,3 +1,3 @@
 module Gtin2atc
-  VERSION = "0.1.3"
+  VERSION = "0.1.4"
 end

data/spec/builder_spec.rb

@@ -45,6 +45,30 @@ describe Gtin2atc::Builder do
     Dir.chdir @savedDir if @savedDir and File.directory?(@savedDir)
   end
 
+  def check_csv(filename)
+    File.exists?(filename).should eq true
+    inhalt = IO.readlines(filename)
+    puts inhalt
+    /^\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")
+  end
+
+  context 'when 20273 41803 (Pharmacodes) is given' do
+    let(:cli) do
+      options = Gtin2atc::Options.new
+      options.parser.parse!('20273 41803'.split(' '))
+      Gtin2atc::Builder.new(options.opts)
+    end
+
+    it 'should produce a correct csv' do
+       # @res = buildr_capture(:stdout){ cli.run }
+      cli.run
+      check_csv(CSV_NAME)
+    end
+  end
+
   context 'when --log is given' do
     let(:cli) do
       options = Gtin2atc::Options.new
@@ -78,8 +102,8 @@ 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'
-      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
 
@@ -151,9 +175,10 @@ describe Gtin2atc::Builder do
       Gtin2atc::Builder.new(options.opts)
     end
 
-    it 'should produce a many report files' do
+    it 'should produce many report files' do
       @res = buildr_capture(:stdout){ cli.run }
       Dir.glob(Gtin2atc::WorkDir + '/compare_all_*.txt').size.should == 5
+      Dir.glob(Gtin2atc::WorkDir + '/pharmacode_gtin_not_found.txt').size.should == 1
       Dir.glob(Gtin2atc::WorkDir + '/compare_swissmedic_to_swisssindex_*.txt').size.should == 5
       Dir.glob(Gtin2atc::WorkDir + '/compare_bag_to_swissindex_*.txt').size.should == 5
       Dir.glob(Gtin2atc::WorkDir + '/compare_bag_to_swissmedic_*.txt').size.should == 5

data/spec/data/atc.csv

@@ -0,0 +1,18 @@
+A|ALIMENTÄRES SYSTEM UND STOFFWECHSEL||
+A01AA01|Natriumfluorid|1,1 mg O 0.5 mg Fluorid|
+A01AA03|Olaflur|1,1 mg O|
+A01AB02|Wasserstoffperoxid|60 mg O|
+A01AB33|Nystatin|500 000 IE O|
+A06AA01|Dickflüssiges Paraffin|15 g O|
+B03AA07|Eisen(II)sulfat|0,2 g O Fe2+|
+B03AE10|Verschiedene Kombinationen||
+D11AC03|Selen-haltige Verbindungen||
+G03GA01|Choriongonadotrophin|7,5 TSD E P zur Ovulationsauslösung|
+J01FA01|Erythromycin|1 g O; 2 g O Erythromycinethylsuccinat-Tabletten; 2 g P; 1 g O Kinder DDD für Erythromycinethylsuccinat-haltige Mittel|
+N02BE01|Paracetamol|3 g O,P,R; 0,75 g O Kinder DDD; 0,375 g R Säuglings DDD; 0,75 g R Kinder DDD|
+N05CP01|Baldrianwurzel|7 g O Droge; 6,5 ml O Tinktur|
+N07BC02|Methadon|25 mg O,P|
+S01EB02|Carbachol|0,4 ml|
+S01EB06|Neostigmin|40 mg Salbe; 0,4 ml|
+S01XA28|Hyaluronsäure|0,4 ml AT|
+S01XC20|Kombinationen|Standarddosis: 0,4 ml AT; 0,4 g AS|