gtin2atc 0.1.0

data/lib/gtin2atc/downloader.rb

@@ -0,0 +1,252 @@
+# encoding: utf-8
+# Copied and only using BAG SwissIndex from oddb2xml/downloader.rb
+require 'mechanize'
+require 'zip'
+require 'savon'
+
+module Gtin2atc
+  module DownloadMethod
+    private
+    def download_as(file, option='r')
+      tempFile  = File.join(WorkDir,   File.basename(file))
+      file2save = File.join(Util.get_archive, File.basename(file))
+      Gtin2atc.log "download_as file #{file2save} via #{tempFile} from #{@url}"
+      data = nil
+      FileUtils.rm_f(tempFile, :verbose => false)
+      if Gtin2atc.skip_download(file)
+        io = File.open(file, option)
+        data = io.read
+      else
+        begin
+          response = @agent.get(@url)
+          response.save_as(file)
+          response = nil # win
+          io = File.open(file, option)
+          data = io.read
+        rescue Timeout::Error, Errno::ETIMEDOUT
+          retrievable? ? retry : raise
+        ensure
+          io.close if io and !io.closed? # win
+          Gtin2atc.download_finished(tempFile)
+        end
+      end
+      return data
+    end
+  end
+  class Downloader
+    attr_reader :type
+    def initialize(options={}, url=nil)
+      @options     = options
+      @url         = url
+      @retry_times = 3
+      HTTPI.log = false # disable httpi warning
+      Gtin2atc.log "Downloader from #{@url} for #{self.class}"
+      init
+    end
+    def init
+      @agent = Mechanize.new
+      @agent.user_agent = 'Mozilla/5.0 (X11; Linux x86_64; rv:16.0) Gecko/20100101 Firefox/16.0'
+      @agent.redirect_ok         = true
+      @agent.redirection_limit   = 5
+      @agent.follow_meta_refresh = true
+      if RUBY_PLATFORM =~ /mswin|mingw|bccwin|cygwin/i and
+         ENV['SSL_CERT_FILE'].nil?
+        cert_store = OpenSSL::X509::Store.new
+        cert_store.add_file(File.expand_path('../../../tools/cacert.pem', __FILE__))
+        @agent.cert_store = cert_store
+      end
+      Util.debug_msg "Downloader @agent ist #{@agent}"
+    end
+    protected
+    def retrievable?
+      if @retry_times > 0
+        sleep 5
+        @retry_times -= 1
+        true
+      else
+        false
+      end
+    end
+    def read_xml_from_zip(target, zipfile)
+      Gtin2atc.log "read_xml_from_zip target is #{target} zip: #{zipfile} #{File.exists?(zipfile)}"
+      entry = nil
+      Dir.glob(File.join(Util.get_archive, '*')).each { |name| if target.match(name) then entry = name; break end }
+      if entry
+        dest = "#{Util.get_archive}/#{File.basename(entry)}"
+        if File.exists?(dest)
+          Gtin2atc.log "read_xml_from_zip return content of #{dest} #{File.size(dest)} bytes "
+          return IO.read(dest)
+        else
+          Gtin2atc.log "read_xml_from_zip could not read #{dest}"
+        end
+      else
+        Gtin2atc.log "read_xml_from_zip could not find #{target.to_s}"
+      end
+      xml = ''
+      if RUBY_PLATFORM =~ /mswin|mingw|bccwin|cygwin/i
+        Zip::File.open(zipfile) do |zipFile|
+          zipFile.each do |entry|
+            if entry.name =~ target
+              Gtin2atc.log "read_xml_from_zip reading #{__LINE__}: #{entry.name}"
+              io = entry.get_input_stream
+              until io.eof?
+                bytes = io.read(1024)
+                xml << bytes
+                bytes = nil
+              end
+              io.close if io.respond_to?(:close)
+              dest = "#{Util.get_archive}/#{File.basename(entry.name)}"
+              File.open(dest, 'w+') { |f| f.write xml }
+              Gtin2atc.log "read_xml_from_zip saved as #{dest}"
+            end
+          end
+        end
+      else
+        Zip::File.foreach(zipfile) do |entry|
+          if entry.name =~ target
+            Gtin2atc.log "read_xml_from_zip #{__LINE__}: reading #{entry.name}"
+            dest = "#{Util.get_archive}/#{File.basename(entry.name)}"
+            entry.get_input_stream { |io| xml = io.read }
+            File.open(dest, 'w+') { |f| f.write xml }
+            Gtin2atc.log "read_xml_from_zip saved as #{dest}"
+          end
+        end
+      end
+      xml
+    end
+  end
+  class BagXmlDownloader < Downloader
+    def init
+      super
+      @url ||= 'http://bag.e-mediat.net/SL2007.Web.External/File.axd?file=XMLPublications.zip'
+    end
+    def origin
+      @url
+    end
+    def download
+      file = File.join(WorkDir, 'XMLPublications.zip')
+      Gtin2atc.log "BagXmlDownloader #{__LINE__}: #{file} from #{@url}"
+      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)
+        begin
+          response = @agent.get(@url)
+          response.save_as(file)
+          response = nil # win
+        rescue Timeout::Error, Errno::ETIMEDOUT
+          retrievable? ? retry : raise
+        ensure
+          Gtin2atc.download_finished(file)
+        end
+      end
+      content = read_xml_from_zip(/Preparations.xml/, File.join(Util.get_archive, File.basename(file)))
+      content
+    end
+  end
+  class SwissmedicDownloader < Downloader
+    def initialize(type=:orphan)
+      @type = :package
+      @xpath = "//div[@id='sprungmarke10_7']//a[@title='Excel-Version Zugelassene Verpackungen*']"
+      @url = "http://www.swissmedic.ch/arzneimittel/00156/00221/00222/00230/index.html?lang=de"
+      super({}, @url)
+    end
+    def origin
+      @url
+    end
+    def init
+      config = {
+        :log_level       => :info,
+        :log             => false, # $stdout
+        :raise_errors    => true,
+        :ssl_version     => :SSLv3,
+        :wsdl            => @url
+      }
+      @client = Savon::Client.new(config)
+    end
+    def download
+      file2save, dated = Gtin2atc::Util.get_latest_and_dated_name("swissmedic_package", '.xlsx')
+      if File.exists?(file2save) and diff_hours = ((Time.now-File.ctime(file2save)).to_i/3600) and diff_hours < 24
+        Util.debug_msg "Skip download of #{file2save} as only #{diff_hours} hours old"
+        return File.expand_path(file2save)
+      end
+      Util.debug_msg "Must download #{file2save} #{File.expand_path(file2save)}"
+      begin
+        @agent = Mechanize.new
+        page = @agent.get(@url)
+        if link_node = page.search(@xpath).first
+          link = Mechanize::Page::Link.new(link_node, @agent, page)
+          response = link.click
+          response.save_as(file2save)
+          response = nil # win
+        end
+        return File.expand_path(file2save)
+      rescue Timeout::Error, Errno::ETIMEDOUT
+        retrievable? ? retry : raise
+      ensure
+        Gtin2atc.download_finished(file2save, false)
+      end
+      return File.expand_path(file2save)
+    end
+  end
+
+  class SwissIndexDownloader < Downloader
+    def initialize(options={}, type=:pharma, lang='DE')
+      @type = (type == :pharma ? 'Pharma' : 'NonPharma')
+      @lang = lang
+      @url = "https://index.ws.e-mediat.net/Swissindex/#{@type}/ws_#{@type}_V101.asmx?WSDL"
+      super(options, @url)
+    end
+    def origin
+      @url
+    end
+    def init
+      config = {
+        :log_level       => :info,
+        :log             => false, # $stdout
+        :raise_errors    => true,
+        :ssl_version     => :SSLv3,
+        :wsdl            => @url
+      }
+      @client = Savon::Client.new(config)
+    end
+    def download
+      begin
+        file2save, dated = Gtin2atc::Util.get_latest_and_dated_name("swissindex_#{@type}_#{@lang}", '.xml')
+        if File.exists?(file2save) and diff_hours = ((Time.now-File.ctime(file2save)).to_i/3600) and diff_hours < 24
+          Util.debug_msg "Skip download of #{file2save} as only #{diff_hours} hours old"
+          return IO.read(file2save)
+        end
+        FileUtils.rm_f(file2save, :verbose => false)
+        soap = <<XML
+<?xml version="1.0" encoding="utf-8"?>
+<soap:Envelope xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema" xmlns:soap="http://schemas.xmlsoap.org/soap/envelope/">
+<soap:Body>
+  <lang xmlns="http://swissindex.e-mediat.net/Swissindex#{@type}_out_V101">#{@lang}</lang>
+</soap:Body>
+</soap:Envelope>
+XML
+        response = @client.call(:download_all, :xml => soap)
+        if response.success?
+          if xml = response.to_xml
+            response = nil # win
+            FileUtils.makedirs(WorkDir)
+            File.open(file2save, 'w+') { |file| file.write xml }
+            Util.debug_msg "Swissindex download successful"
+          else
+            # received broken data or internal error
+            raise StandardError
+          end
+        else
+          raise Timeout::Error
+        end
+      rescue HTTPI::SSLError
+        exit # catch me in Cli class
+      rescue Timeout::Error, Errno::ETIMEDOUT
+        retrievable? ? retry : raise
+      end
+      Util.debug_msg "Download of #{file2save} finished"
+      xml
+    end
+  end
+end

data/lib/gtin2atc/options.rb

@@ -0,0 +1,39 @@
+# encoding: utf-8
+require 'optparse'
+
+module Gtin2atc
+  
+  class Options
+    attr_reader :parser, :opts
+    def Options.default_opts
+      {
+        :log          => false,
+        :compare      => false,
+      }
+    end
+    def Options.help
+  <<EOS
+#$0 ver.#{Gtin2atc::VERSION}
+Usage:
+  gtin2atc [--compare] [--log] [file_with_gtin or gtin or pharmacode] [gtin..]
+    If file_with_gtin is given only the GTIN (or pharamacode) (one per line) is outputted.
+    If no file or gtin is given, alle GTIN will be processed.
+    --log               log important actions
+
+    --compare           download an compare GTIN/ATC_code from BAG, SwissIndex and RefData
+                        For each GTIN we will output a message if it can be only found in the
+                        * BAG http://bag.e-mediat.net/SL2007.Web.External/File.axd?file=XMLPublications.zip
+                        * SwissIndex e-mediat: http://swissindex.e-mediat.net/SwissindexPharma_out_V101
+                        * or if the ATC_Code does not not match
+    -h,   --help         Show this help message.
+EOS
+    end
+    def initialize
+      @parser = OptionParser.new
+      @opts   = Options.default_opts
+      @parser.on('--log')             {|v| @opts[:log] = true }
+      @parser.on('--compare')         {|v| @opts[:compare] = true }
+      @parser.on_tail('-h', '--help') { puts Options.help; exit }
+    end
+  end
+end

data/lib/gtin2atc/util.rb

@@ -0,0 +1,47 @@
+require 'date'
+module Gtin2atc
+  class Util
+    @@archive = File.expand_path(File.join(__FILE__, '../../..'))
+    @@today   = Date.today
+    @@logging = false
+    def Util.get_today
+      @@today
+    end
+    def Util.set_archive_dir(archiveDir)
+      @@archive = archiveDir
+    end
+    def Util.get_archive
+      @@archive
+    end
+    def Util.set_logging(default)
+      @@logging = default
+    end
+    def Util.info(msg)
+      puts msg
+      return unless @@logging
+      Util.init
+      @@checkLog.puts("#{Time.now}: #{msg}")
+    end
+    def Util.init
+      return unless @@logging
+      if not defined?(@@checkLog) or not @@checkLog
+        name = File.join(@@archive, 'log.log')
+        FileUtils.makedirs(@@archive)
+        @@checkLog = File.open(name, 'a+')
+      end
+    end
+    def Util.debug_msg(msg)
+      return unless @@logging
+      Util.init
+      if @@logging or defined?(MiniTest) then $stdout.puts Time.now.to_s + ': ' + msg; $stdout.flush; return end
+      @@checkLog.puts("#{Time.now}: #{msg}")
+      @@checkLog.flush
+    end
+    def Util.get_latest_and_dated_name(keyword, extension)
+      return File.expand_path(File.join(Util.get_archive, keyword + '-latest' + extension)),
+          File.expand_path(File.join(Util.get_archive, Util.get_today.strftime("#{keyword}-%Y.%m.%d" + extension)))
+    end
+  end
+  def Gtin2atc.download_finished(file, remove_file = true)
+  end
+end

data/lib/gtin2atc/version.rb

@@ -0,0 +1,3 @@
+module Gtin2atc
+  VERSION = "0.1.0"
+end

data/lib/gtin2atc/xml_definitions.rb

@@ -0,0 +1,250 @@
+# Definitions for the sax-machine parse to parse
+# Preparations.xml
+
+require 'sax-machine'
+
+class PriceElement
+  include SAXMachine
+  element :Price
+  element :ValidFromDate
+  element :DivisionDescription
+  element :PriceTypeCode
+  element :PriceTypeDescriptionDe
+  element :PriceTypeDescriptionFr
+  element :PriceTypeDescriptionIt
+  element :PriceChangeTypeDescriptionDe
+  element :PriceChangeTypeDescriptionFr
+  element :PriceChangeTypeDescriptionIt
+end
+
+class StatusElement
+  include SAXMachine
+  element :IntegrationDate
+  element :ValidFromDate
+  element :ValidThruDate
+  element :StatusTypeCodeSl
+  element :StatusTypeDescriptionSl
+  element :FlagApd
+end
+
+class PricesElement
+  include SAXMachine
+  element :ExFactoryPrice, :class => PriceElement
+  element :PublicPrice, :class => PriceElement
+end
+
+class LimitationElement
+  include SAXMachine
+  element :LimitationCode
+  element :LimitationType
+  element :LimitationValue
+  element :LimitationNiveau
+  element :DescriptionDe
+  element :DescriptionFr
+  element :DescriptionIt
+  element :ValidFromDate
+  element :ValidThruDate
+end
+
+class LimitationsElement
+  include SAXMachine
+  elements :Limitation, :class => LimitationElement
+end
+
+class PointLimitationElement
+  include SAXMachine
+  element :Points
+  element :Packs
+  element :ValidFromDate
+  element :ValidThruDate
+end
+
+class PointLimitationsElement
+  include SAXMachine
+  elements :PointLimitation, :class => PointLimitationElement
+end
+
+class PackContent
+  include SAXMachine
+  attribute :ProductKey
+  attribute :Pharmacode
+  attribute :PackId
+  element :DescriptionDe
+  element :DescriptionFr
+  element :DescriptionIt
+  element :SwissmedicCategory
+  element :SwissmedicNo8
+  element :FlagNarcosis
+  element :FlagModal
+  element :BagDossierNo
+  element :GTIN
+  element :Limitations, :class => LimitationsElement
+  element :PointLimitations, :class => PointLimitationsElement
+  element :Prices, :class => PricesElement
+end
+
+class PacksElement
+  include SAXMachine
+  elements :Pack, :class => PackContent
+end
+
+class ItCodeContent
+  include SAXMachine
+  attribute :Code
+  element :DescriptionDe
+  element :DescriptionFr
+  element :DescriptionIt
+  element :Limitations, :class => LimitationsElement
+end
+
+class ItCodeEntry
+  include SAXMachine
+  element :ItCode, :class => ItCodeContent
+end
+
+# handling attributes as suggested by https://github.com/pauldix/sax-machine/issues/30
+class ItCodesElement
+  include SAXMachine
+  elements :ItCode, :class => ItCodeContent
+end
+
+class SubstanceElement
+  include SAXMachine
+  element :DescriptionLa
+  element :Quantity
+  element :QuantityUnit
+end
+
+class SubstancesElement
+  include SAXMachine
+  elements :Substance, :class => SubstanceElement
+end
+
+class PreparationContent
+  include SAXMachine
+  attribute :ProductCommercial
+  element :NameFr
+  element :NameDe
+  element :NameIt
+  element :Status, :class => StatusElement
+  element :Dummy
+  element :DescriptionDe
+  element :DescriptionFr
+  element :DescriptionIt
+  element :AtcCode
+  element :SwissmedicNo5
+  element :FlagItLimitation
+  element :OrgGenCode
+  element :FlagSB20
+  element :CommentDe
+  element :CommentFr
+  element :CommentIt
+  element :VatInEXF
+  element :Limitations, :class => LimitationsElement
+  element :Substances, :class => SubstancesElement
+  element :Packs, :class => PacksElement
+  element :ItCodes, :class => ItCodesElement
+end
+
+class PreparationEntry
+  include SAXMachine
+  element :Preparation, :class => PreparationContent
+end
+
+class PreparationsContent
+  include SAXMachine
+  attribute :ReleaseDate
+  elements :Preparation, :class => PreparationContent
+end
+
+class PreparationsEntry
+  include SAXMachine
+  element :Preparations, :class => PreparationsContent
+end
+
+
+class CompElement
+  include SAXMachine
+  element :NAME
+  element :GLN
+end
+
+class ItemContent
+  include SAXMachine
+  attribute :DT
+  element :GTIN
+  element :PHAR
+  element :STATUS
+  element :SDATE
+  element :LANG
+  element :DSCR
+  element :ADDSCR
+  element :ATC
+  element :COMP, :class => CompElement
+end
+
+class PharmaContent
+  include SAXMachine
+  attribute :CREATION_DATETIME
+  elements :ITEM, :class => ItemContent
+end
+
+class PharmaEntry
+  include SAXMachine
+  element :CREATION_DATETIME
+  element :NONPHARMA, :as => :PHARMA, :class => PharmaContent
+  element :PHARMA, :class => PharmaContent
+end
+
+class ItemContent
+  include SAXMachine
+  attribute :DT
+  element :GTIN
+  element :PHAR
+  element :STATUS
+  element :STDATE
+  element :LANG
+  element :DSCR
+  element :ADDSCR
+  element :ATC
+  element :COMP, :class => CompElement
+end
+
+class PharmaContent
+  include SAXMachine
+  attribute :CREATION_DATETIME
+  elements :ITEM, :class => ItemContent
+end
+
+class PharmaEntry
+  include SAXMachine
+  element :CREATION_DATETIME
+  element :PHARMA, :class => PharmaContent
+end
+
+class MedicalInformationContent
+  include SAXMachine
+  attribute :type
+  attribute :version
+  attribute :lang
+  element :title
+  element :authHolder
+  element :style
+  element :content
+end
+
+class MedicalInformationEntry
+  include SAXMachine
+  element :medicalInformation, :class => MedicalInformationContent
+end
+
+class MedicalInformationsContent
+  include SAXMachine
+  elements :medicalInformation, :class => MedicalInformationContent
+end
+
+class MedicalInformationsEntry
+  include SAXMachine
+  element :medicalInformations, :class => MedicalInformationsContent
+end
+

data/lib/gtin2atc.rb

@@ -0,0 +1,9 @@
+require "gtin2atc/version"
+require "gtin2atc/util"
+
+module Gtin2atc
+  WorkDir       = Dir.pwd
+  def self.log(msg)
+    Util.debug_msg(msg)
+  end
+end

data/readme.textile

@@ -0,0 +1,19 @@
+h2. gtin2atc
+
+h3. usage
+
+There are three different useage:
+# gtin2atc --compare
+  Reads the GTIN/ATC-Code from BAG, Swissmedic, RefData and compares the from constency.
+  When adding --log it outputs a line for each difference/problem found.
+# gtin2atc gtin1 [gtinx]
+  Search for ATC-Code with GTIN(s) passed on command line. Outputs a file gtin2atc.csv
+# gtin2atc file_with_one_gtin_per_line
+  Reads file_with_one_gtin_per_line, then putputs a file gtin2atc.csv with info about each GTIN found in the input file
+
+h3. background
+
+We get the data from three sources
+* e-mediat: http://swissindex.e-mediat.net/SwissindexPharma_out_V101
+* BAG: Preparations.xml inside http://bag.e-mediat.net/SL2007.Web.External/File.axd?file=XMLPublications.zip
+* Swissmedic: http://www.swissmedic.ch/arzneimittel/00156/00221/00222/00230/index.html?lang=de