oddb2xml 2.5.0 → 2.5.1

data/lib/oddb2xml/extractor.rb

@@ -9,14 +9,15 @@ require 'oddb2xml/xml_definitions'
 module Oddb2xml
   module TxtExtractorMethods
     def initialize(str)
+      Oddb2xml.log("TxtExtractorMethods #{str} #{str.to_s.size} bytes")
       @io = StringIO.new(str)
     end
     def to_hash
       data = {}
       while line = @io.gets
         next unless line =~ /\d{13}/
-        ean = line.chomp.gsub("\"", '')
-        data[ean] = true
+        ean13 = line.chomp.gsub("\"", '')
+        data[ean13] = true
       end
       data
     end
@@ -24,6 +25,7 @@ module Oddb2xml
   class Extractor
     attr_accessor :xml
     def initialize(xml)
+      Oddb2xml.log("Extractor #{xml } xml #{xml.size} bytes")
       @xml = xml
     end
   end
@@ -37,6 +39,7 @@ module Oddb2xml
       result = PreparationsEntry.parse(@xml.sub(Strip_For_Sax_Machine, ''), :lazy => true)
       result.Preparations.Preparation.each do |seq|
         item = {}
+        item[:data_origin]  = 'bag_xml'
         item[:refdata]      = true
         item[:product_key]  = seq.ProductCommercial
         item[:desc_de]      = (desc = seq.DescriptionDe) ? desc : ''
@@ -81,8 +84,7 @@ module Oddb2xml
               puts "BagXmlExtractor: Setting missing GTIN  #{pac.GTIN} in SwissmedicNo8 #{pac.SwissmedicNo8}  BagDossierNo #{pac.BagDossierNo} PackId #{pac.PackId} #{item[:name_de]}."
             end
           end
-          ean = pac.GTIN.to_i
-          
+          ean13 = pac.GTIN.to_s
           # packages
           exf = {:price => '', :valid_date => '', :price_code => ''}
           if pac.Prices and pac.Prices.ExFactoryPrice
@@ -96,14 +98,14 @@ module Oddb2xml
             pub[:valid_date] =  pac.Prices.PublicPrice.ValidFromDate if pac.Prices.PublicPrice.ValidFromDate
             pub[:price_code] =  pac.Prices.PublicPrice.PriceTypeCode if pac.Prices.PublicPrice.PriceTypeCode
           end
-          item[:packages][ean] = {
-            :ean                 => ean,
+          item[:packages][ean13] = {
+            :ean13               => ean13,
             :swissmedic_category => (cat = pac.SwissmedicCategory) ? cat : '',
             :swissmedic_number8  => (num = pac.SwissmedicNo8)      ? num : '',
             :prices              => { :exf_price => exf, :pub_price => pub },
           }
           # related all limitations
-          item[:packages][ean][:limitations] = []
+          item[:packages][ean13][:limitations] = []
           limitations = Hash.new{|h,k| h[k] = [] }
           if seq.Limitations
             limitations[:seq] = seq.Limitations.Limitation.collect { |x| x }
@@ -132,11 +134,11 @@ module Oddb2xml
               id  = item[key].to_s
             when :pac
               key = :swissmedic_number8
-              id  = item[:packages][ean][key].to_s
+              id  = item[:packages][ean13][key].to_s
             end
-            if id.empty? && item[:packages][ean][ :swissmedic_number8]
+            if id.empty? && item[:packages][ean13][ :swissmedic_number8]
               key = :swissmedic_number8
-              id  = item[:packages][ean][key].to_s
+              id  = item[:packages][ean13][key].to_s
             end
             lims.each do |lim|
               limitation = {
@@ -160,13 +162,13 @@ module Oddb2xml
                 end
               end
               limitation[:del] = deleted
-              item[:packages][ean][:limitations] << limitation
+              item[:packages][ean13][:limitations] << limitation
             end if lims
           end
           # limitation points
           pts = pac.PointLimitations.PointLimitation.first # only first points
-          item[:packages][ean][:limitation_points] = pts ? pts.Points : ''
-          data[ean] = item
+          item[:packages][ean13][:limitation_points] = pts ? pts.Points : ''
+          data[ean13] = item
         end
       end
       data
@@ -183,24 +185,30 @@ module Oddb2xml
       result = SwissRegArticleEntry.parse(@xml.sub(Strip_For_Sax_Machine, ''), :lazy => true)
       items = result.ARTICLE.ITEM
       items.each do |pac|
+        ean13 = (gtin = pac.GTIN.to_s) ? gtin: '0'
+        if ean13.size != 13
+          puts "Refdata #{@type} ean13: Fixed incorrect length #{ean13.size} for #{ean13}"
+          ean13 = ean13[1..-1]
+        end
         item = {}
+        item[:data_origin]     = 'refdata'
         item[:refdata]         = true
         item[:_type]           = (typ  = pac.ATYPE.downcase.to_sym)  ? typ: ''
-        item[:ean]             = (gtin = pac.GTIN.to_i)   ? gtin: 0
-        item[:pharmacode]      = (phar = pac.PHAR.to_i)   ? phar: 0
+        item[:ean13]           = ean13
+        item[:pharmacode]      = (phar = pac.PHAR.to_s)   ? phar: '0'
         item[:last_change]     = (date = Time.parse(pac.DT).to_s)  ? date: ''  # Date and time of last data change
         item[:desc_de]         = (dscr = pac.NAME_DE)   ? dscr: ''
         item[:desc_fr]         = (dscr = pac.NAME_FR)   ? dscr: ''
         item[:atc_code]        = (code = pac.ATC)    ? code.to_s : ''
-				item[:company_name] = (nam = pac.AUTH_HOLDER_NAME) ? nam: ''
-				item[:company_ean]  = (gln = pac.AUTH_HOLDER_GLN)  ? gln: ''
+        item[:company_name] = (nam = pac.AUTH_HOLDER_NAME) ? nam: ''
+        item[:company_ean]  = (gln = pac.AUTH_HOLDER_GLN)  ? gln: ''
         unless item[:pharmacode]
           item[:pharmacode] = phar
           unless data[item[:pharmacode]] # pharmacode => GTINs
-            data[item[:ean]] = []
+            data[item[:ean13]] = []
           end
         end
-        data[item[:ean]] = item
+        data[item[:ean13]] = item
       end
       data
     end
@@ -210,6 +218,7 @@ module Oddb2xml
       @filename = File.join(Downloads, File.basename(filename))
       @filename = File.join(SpecData, File.basename(filename)) if defined?(RSpec) and not File.exists?(@filename)
       @type  = type
+      Oddb2xml.log("SwissmedicExtractor #{@filename} #{File.size(@filename)} bytes")
       return unless File.exists?(@filename)
       @sheet = RubyXL::Parser.parse(File.expand_path(@filename)).worksheets[0]
     end
@@ -280,8 +289,8 @@ module Oddb2xml
           unless no8.empty?
             next if no8.to_i == 0
             ean_base12 = "7680#{no8}"
-            data[no8.intern] = {
-              :ean                  => (ean_base12.ljust(12, '0') + Oddb2xml.calc_checksum(ean_base12)),
+            data[no8] = {
+              :ean13                => (ean_base12.ljust(12, '0') + Oddb2xml.calc_checksum(ean_base12)),
               :prodno               => prodno ? prodno : '',
               :ith_swissmedic       => row[ith] ? row[ith].value.to_s : '',
               :swissmedic_category  => row[cat].value.to_s,
@@ -296,6 +305,11 @@ module Oddb2xml
               :gen_production       => row[COLUMNS_JULY_2015.keys.index(:gen_production)].value.to_s,
               :insulin_category     => row[COLUMNS_JULY_2015.keys.index(:insulin_category)].value.to_s,
               :drug_index           => row[COLUMNS_JULY_2015.keys.index(:drug_index)].value.to_s,
+              :data_origin          => 'swissmedic_package',
+              :expiry_date          => row[COLUMNS_JULY_2015.keys.index(:expiry_date)].value.to_s,
+              :company_name         => row[COLUMNS_JULY_2015.keys.index(:company)].value.to_s,
+              :size                 => row[COLUMNS_JULY_2015.keys.index(:size)].value.to_s,
+              :unit                 => row[COLUMNS_JULY_2015.keys.index(:unit)].value.to_s,
             }
           end
         end
@@ -307,14 +321,14 @@ module Oddb2xml
     def cleanup_file
       begin
         File.unlink(@filename) if File.exists?(@filename)
-        rescue Errno::EACCES # Permission Denied on Windows      
+        rescue Errno::EACCES # Permission Denied on Windows
       end unless defined?(RSpec)
     end
 
   end
   class MigelExtractor < Extractor
     def initialize(bin)
-      io = StringIO.new(bin)
+      Oddb2xml.log("MigelExtractor #{io} #{File.size(io)} bytes")
       book = Spreadsheet.open(io, 'rb')
       @sheet = book.worksheet(0)
     end
@@ -322,19 +336,20 @@ module Oddb2xml
       data = {}
       @sheet.each_with_index do |row, i|
         next if i.zero?
-        phar = row[1].to_i
+        phar = row[1]
         next if phar == 0
-        ean = row[0].to_i
-        ean = phar unless ean.to_s.length == 13
+        ean13 = row[0]
+        ean13 = phar unless ean13.to_s.length == 13
         data[ean] = {
           :refdata         => true,
-          :ean             => ean,
-          :pharmacode      => phar.to_i,
+          :ean13           => ean13,
+          :pharmacode      => phar,
           :desc_de         => row[3],
           :desc_fr         => row[4],
           :quantity        => row[5], # quantity
           :company_name    => row[6],
-          :company_ean     => row[7].to_i,
+          :company_ean     => row[7],
+          :data_origin     => 'migel'
         }
       end
       data
@@ -350,7 +365,8 @@ module Oddb2xml
         lang = pac.lang.to_s
         next unless lang =~ /de|fr/
         item = {}
-        item[:refdata] = true,
+        item[:refdata] = true
+        item[:data_origin] = 'swissmedic_info'
         item[:name]  = (name = pac.title) ? name : ''
         item[:owner] = (ownr = pac.authHolder) ? ownr : ''
         item[:style] =  Nokogiri::HTML.fragment(pac.style).to_html(:encoding => 'UTF-8')
@@ -370,6 +386,7 @@ module Oddb2xml
 
   class EphaExtractor < Extractor
     def initialize(str)
+      Oddb2xml.log("EphaExtractor #{str.size} bytes")
       @io = StringIO.new(str)
     end
     def to_arry
@@ -384,6 +401,7 @@ module Oddb2xml
           row = CSV.parse_line(line.gsub('""','"'))
           action = {}
           next unless row.size > 8
+          action[:data_origin] = 'epha'
           action[:ixno]      = ixno
           action[:title]     = row[4]
           action[:atc1]      = row[0]
@@ -413,6 +431,7 @@ module Oddb2xml
           row = line.chomp.split("\t")
           next if row[0] =~ /^GLN/
           data << {
+            :data_origin   => 'medreg',
             :gln           => row[0].to_s.gsub(/[^0-9]/, ''), #=> GLN Betrieb
             :name_1        => row[1].to_s,                    #=> Betriebsname 1
             :name_2        => row[2].to_s,                    #=> Betriebsname 2
@@ -431,6 +450,7 @@ module Oddb2xml
           row = line.chomp.split("\t")
           next if row[0] =~ /^GLN/
           data << {
+            :data_origin   => 'medreg',
             :gln           => row[0].to_s.gsub(/[^0-9]/, ''), #=> GLN Person
             :last_name     => row[1].to_s,                    #=> Name
             :first_name    => row[2].to_s,                    #=> Vorname
@@ -449,8 +469,9 @@ module Oddb2xml
   end
   class ZurroseExtractor < Extractor
     # see http://dev.ywesee.com/Bbmb/TransferDat
-    def initialize(dat, extended = false)
+    def initialize(dat, extended = false, artikelstamm = false)
       @@extended = extended
+      @artikelstamm = artikelstamm
       FileUtils.makedirs(WorkDir)
       @@error_file ||= File.open(File.join(WorkDir, "duplicate_ean13_from_zur_rose.txt"), 'wb+:ISO-8859-14')
       @@items_without_ean13s ||= 0
@@ -470,37 +491,43 @@ module Oddb2xml
     def to_hash
       data = {}
       while line = @io.gets
-        line = Oddb2xml.patch_some_utf8(line)
+        line = Oddb2xml.patch_some_utf8(line).chomp
         next if line =~ /(ad us\.* vet)|(\(vet\))/i
         if @@extended
           next unless line =~ /(\d{13})(\d{1})$/
         else
           next unless line =~ /(7680\d{9})(\d{1})$/
         end
-        pharma_code = line[3..9].to_i
+        pharma_code = line[3..9]
         if $1.to_s == '0000000000000'
           @@items_without_ean13s += 1
-          ean13 = pharma_code # dummy ean13
+          ean13 = '999999' + pharma_code.to_s # dummy ean13
         else
-          ean13 = $1.to_i
+          ean13 = $1
         end
+        next if @artikelstamm && /(0{13})(\d{1})$/.match(line)
         if data[ean13]
-          @@error_file.puts "Duplicate ean13 #{ean13} in line \nact: #{line}\norg: #{data[ean13][:line]}"
+          @@error_file.puts "Duplicate ean13 #{ean13} in line \nact: #{line.chomp}\norg: #{data[ean13][:line]}"
           @@items_without_ean13s -= 1
           @@duplicated_ean13s += 1
           next
         end
 
+        pexf = sprintf("%.2f", line[60,6].gsub(/(\d{2})$/, '.\1').to_f)
+        ppub =  sprintf("%.2f", line[66,6].gsub(/(\d{2})$/, '.\1').to_f)
+        next if  @artikelstamm && /^113/.match(line) && /^7680/.match(ean13)
+        next if  @artikelstamm && /^113/.match(line) && ppub.eql?('0.0') && pexf.eql?('0.0')
         data[ean13] = {
-          :line   => line,
-          :ean   => ean13,
+          :data_origin   => 'zur_rose',
+          :line   => line.chomp,
+          :ean13 => ean13,
           :clag  => line[73],
           :vat   => line[96],
           :description => line[10..59].sub(/\s+$/, ''),
           :quantity => '',
           :pharmacode => pharma_code,
-          :price => sprintf("%.2f", line[60,6].gsub(/(\d{2})$/, '.\1').to_f),
-          :pub_price => sprintf("%.2f", line[66,6].gsub(/(\d{2})$/, '.\1').to_f),
+          :price => pexf,
+          :pub_price => ppub,
           :type => :nonpharma,
           :cmut => line[2],
         }

data/lib/oddb2xml/options.rb

@@ -1,85 +1,84 @@
 # encoding: utf-8
-require 'optparse'
+require 'trollop'
+require 'oddb2xml/version'
 
 module Oddb2xml
-  
-  class Options
-    attr_reader :parser, :opts
-    def Options.default_opts
-      {
-        :fi           => false,
-        :adr          => false,
-        :address      => false,
-        :nonpharma    => false,
-        :extended     => false,
-        :compress_ext => nil,
-        :format       => :xml,
-        :calc         => false,
-        :tag_suffix   => nil,
-        :debug        => false,
-        :ean14        => false,
-        :skip_download=> false,
-        :log          => false,
-        :percent      => nil,
-      }
-    end
-    def Options.help
-  <<EOS
-#$0 ver.#{Oddb2xml::VERSION}
-Usage:
-  oddb2xml [option]
-    produced files are found under data
-    -a,   --append       Additional target nonpharma
-    -c F, --compress=F   Compress format F. {tar.gz|zip}
-    -e    --extended     pharma, non-pharma plus prices and non-pharma from zurrose.
-                         Products without EAN-Code will also be listed.
-                         File oddb_calc.xml will also be generated
-    -f F, --format=F     File format F, default is xml. {xml|dat}
-                         If F is given, -o option is ignored.
-    -I x, --increment=x  Increment price by x percent. Forces -f dat -p zurrose.
-    -I x, --increment=x  create additional field price_resellerpub as
-                         price_extfactory incremented by x percent (rounded to the next 0.05 francs)
-                         in oddb_article.xml. In generated zurrose_transfer.dat PRPU is set to this price
-                         Forces -f dat -p zurrose.
-    -i,   --include      Include target option for ean14  for 'dat' format.
-                         'xml' format includes always ean14 records.
-    -o,   --option       Optional fachinfo output.
-    -p,   --price        Price source (transfer.dat) from ZurRose
-    -t S, --tag-suffix=S XML tag suffix S. Default is none. [A-z0-9]
-                         If S is given, it is also used as prefix of filename.
-    -x N, --context=N    context N {product|address}. product is default.
-    --calc               create only oddb_calc.xml with GTIN, name and galenic information
+  module Options
+    def self.parse(args =ARGV)
+      if args.is_a?(String)
+        args = args.split(' ') 
+      end
 
-                         For debugging purposes
-    --skip-download      skips downloading files it the file is already under downloads.
-                         Downloaded files are saved under downloads
-    --log                log important actions
-    -h,   --help         Show this help message.
+      @opts = Trollop::options(args) do
+        version "#$0 ver.#{Oddb2xml::VERSION}"
+        banner <<-EOS
+        #{File.expand_path($0)} version #{Oddb2xml::VERSION}
+        Usage:
+        oddb2xml [option]
+          produced files are found under data
 EOS
-    end
-    def initialize
-      @parser = OptionParser.new
-      @opts   = Options.default_opts
-      @parser.on('-a',   '--append')                       {|v| @opts[:nonpharma] = true }
-      @parser.on('-c v', '--compress v',   /^tar\.gz|zip$/){|v| @opts[:compress_ext] = v }
-      @parser.on('-e', '--extended')                       {|v| @opts[:extended] = true
-                                                              @opts[:nonpharma] = true
-                                                              @opts[:price] = :zurrose
-                                                              @opts[:calc] = true
-                                                            }
-      @parser.on('-f v', '--format v',     /^xml|dat$/)    {|v| @opts[:format] = v.intern }
-      @parser.on('--calc')                                 {|v| @opts[:calc] = true }
-      @parser.on('-o',   '--option')                       {|v| @opts[:fi] = true }
-      @parser.on('-I v', '--increment v',  /^[0-9]+$/)     {|v| @opts[:percent] = v ? v.to_i : 0
-                                                                @opts[:price] = :zurrose
-                                                           }
-      @parser.on('-i',   '--include')                      {|v| @opts[:ean14] = true }
-      @parser.on('-t v', '--tag-suffix v', /^[A-z0-9]*$/i) {|v| @opts[:tag_suffix] = v.upcase }
-      @parser.on('-x v', '--context v',    /^addr(ess)*$/i){|v| @opts[:address] = true }
-      @parser.on('-p', '--price')                          {|v| @opts[:price] = :zurrose }
-      @parser.on('--skip-download')                        {|v| @opts[:skip_download] = true }
-      @parser.on('--log')                                  {|v| @opts[:log] = true }
-      @parser.on_tail('-h', '--help') { puts Options.help; exit }
+        opt :append,       "Additional target nonpharma", :default => false
+        opt :artikelstamm, "Create Artikelstamm Version 3 and 5 for Elexis >= 3.1"
+        opt :compress_ext,     "format F. {tar.gz|zip}", :type => :string, :default => nil, :short => 'c'
+        opt :extended,     "pharma, non-pharma plus prices and non-pharma from zurrose.
+                            Products without EAN-Code will also be listed.
+                            File oddb_calc.xml will also be generated"
+        opt :format,        "File format F, default is xml. {xml|dat}
+                            If F is given, -o option is ignored.", :type => :string, :default => 'xml'
+        opt :include,       "Include target option for ean14  for 'dat' format.
+                            'xml' format includes always ean14 records.", :short => 'i'
+        opt :increment,     "Increment price by x percent. Forces -f dat -p zurrose.
+                            create additional field price_resellerpub as
+                            price_extfactory incremented by x percent (rounded to the next 0.05 francs)
+                            in oddb_article.xml. In generated zurrose_transfer.dat PRPU is set to this price
+                            Forces -f dat -p zurrose.", :type => :int, :default => nil, :short => 'I'
+        opt :fi,            "Optional fachinfo output.", :short => 'o'
+        opt :price,         "Price source (transfer.dat) from ZurRose", :default => nil
+        opt :tag_suffix,   "XML tag suffix S. Default is none. [A-z0-9]
+                            If S is given, it is also used as prefix of filename.", :type => :string, :short => 't'
+        opt :context,       "{product|address}. product is default.", :default => 'product', :type => :string, :short => 'x'
+        opt :calc,          "create only oddb_calc.xml with GTIN, name and galenic information"
+
+        opt :skip_download, "skips downloading files it the file is already under downloads.
+                            Downloaded files are saved under downloads"
+        opt :log,           "log important actions", :short => :none
+        opt :use_ra11zip,   "Use the ra11.zip (a zipped transfer.dat from Galexis)",
+                            :default => File.exist?('ra11.zip') ? 'ra11.zip' : nil, :type => :string
+      end
+          
+      @opts[:percent] = @opts[:increment]
+      if @opts[:increment] 
+        @opts[:nonpharma] = true
+        @opts[:price] = :zurrose
+      end
+      @opts[:ean14]  = @opts[:increment]
+      @opts.delete(:increment)
+      @opts[:nonpharma] = @opts[:append]
+      @opts.delete(:append)
+      if @opts[:extended] 
+        @opts[:nonpharma] = true
+        @opts[:price] = :zurrose
+        @opts[:calc] = true
+      end
+      if @opts[:artikelstamm]
+        @opts[:extended] = true
+        @opts[:price] = :zurrose
+      end      
+      @opts[:price] = :zurrose if @opts[:price].is_a?(TrueClass)
+      @opts[:price]  = @opts[:price].to_sym if @opts[:price]
+      @opts[:ean14] = @opts[:include]
+      @opts[:format] = @opts[:format].to_sym if @opts[:format]
+      @opts.delete(:include)
+      @opts.delete(:help)
+      @opts.delete(:version)
+
+      @opts[:address]  = false
+      @opts[:address]  = true  if /^addr(ess)*$/i.match(@opts[:context])
+      @opts.delete(:context)
+
+      @opts.delete(:price) unless @opts[:price]
+
+      @opts.each{|k,v| @opts.delete(k) if /_given$/.match(k.to_s)}
     end
   end
 end