im_onix 1.1.1 → 1.2.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 291ef6c16051d0bf41321eab30c7b1217dd472242f9251d041c06fd48791726a
-  data.tar.gz: 749fb1ff1be0edd0d1711e8663f9d4a7c1091ba965da558a0b924a4937deefae
+  metadata.gz: 0f4feeceb8c2275507ee70aac9c2eb2cf8cc335d14ae60fb480b8493b539ecd9
+  data.tar.gz: ead4f5e2dfb4e4a5cbcd14c14a5a3f8efe0004aac253c9c26aab3a91500d689a
 SHA512:
-  metadata.gz: 0eb7dac0735de679d3be0b217b0c59c86304a759055d93d344c01291e31a745ad786eaa557f404748b7c3d99f5b7f246d5adb921843a47e4a89f2fe9f389af65
-  data.tar.gz: 01633b2daeb2a326869e47c0a512a160119e7c62055958a69dfb915b2fc3ff790114176a15c770bed389c6694c36eb7e7a843647c905fe6c094ccb5e806b648a
+  metadata.gz: 8a4c862b06d7613c47a9395737227d0f212f76a7780cba9832060a613e2fc0f0a1ee445e937d08e59921a788c93fc60ff38a3d571a1e8cbc4c46864b1b046215
+  data.tar.gz: f0c1496ae0a1b33d8b10a2e8403383e6918f51982bfe946eceac4111aac12ee543454a68d74211102f2193b73043c036425b3b3ac7890d279b0f995097115c4a

data/bin/html_codelist_to_yml.rb

@@ -4,9 +4,11 @@
 require 'im_onix'
 require 'nokogiri'
 require 'yaml'
+require 'unidecoder'
 
-# html_codelist_to_yml.rb html_codelist_dir
+# html_codelist_to_yml.rb html_codelist_dir dest_dir
 
+# generate YML data/codelists from editeur.org HTML codelists
 class HTMLCodelist
 
   private
@@ -26,19 +28,21 @@ class HTMLCodelist
 
   # from rails
   def self.rename(term)
-    term.gsub(/\(|\)|\,|\-|’|\/|“|”|‘|\.|\:|–|\||\+/, "").gsub(/\;/, " Or ").gsub(/\s+/, " ").split(" ").map { |t| t.capitalize }.join("")
+    result = term.to_ascii.gsub(/\(|\)|\,|'|’|\/|“|”|‘|\.|\:|–|\||\+/, "").gsub(/\-/," ").gsub(/\;/, " Or ").gsub(/\s+/, " ").split(" ").map { |t| t.capitalize }.join("")
+    if result.length > 63
+      puts "WARN: #{result} (#{term}) to long"
+    end
+    result
   end
 end
 
-files = `ls #{ARGV[1]}/*.htm`.split(/\n/)
-
-h = {}
+files = `ls #{ARGV[0]}/*.htm`.split(/\n/)
 
 files.sort.each do |file|
   codelist = file.gsub(/.*onix\-codelist\-(.*)\.htm/, '\1').to_i
   h = HTMLCodelist.parse_codelist(file)
 
-  File.open("data/codelists/codelist-#{codelist}.yml", 'w') do |fw|
+  File.open("#{ARGV[1]}/codelist-#{codelist}.yml", 'w') do |fw|
     fw.write({:codelist => h}.to_yaml)
   end
 end

data/bin/onix_bench.rb

@@ -1,15 +1,20 @@
 #!/usr/bin/env ruby
 require 'benchmark'
+require 'objspace'
 require 'im_onix'
 
-filename=ARGV[0]
+filename = ARGV[0]
 
 if filename
-  bench=Benchmark.measure do
-    msg=ONIX::ONIXMessage.new
+  GC.start
+  old_memsize = ObjectSpace.count_objects_size[:TOTAL]
+  msg = ONIX::ONIXMessage.new
+  bench = Benchmark.measure do
     msg.parse(filename)
   end
-  puts bench
+  GC.start
+  new_memsize = ObjectSpace.count_objects_size[:TOTAL]
+  puts "Parsed #{msg.products.length} products in #{bench.real.round(3)} seconds resulting in use of #{new_memsize - old_memsize} bytes of memory"
 else
   puts "ImOnix parse benchmark"
   puts "Usage: onix_bench.rb file.xml"

data/bin/onix_pp.rb

@@ -1,13 +1,13 @@
 #!/usr/bin/env ruby
 require 'im_onix'
 
-filename=ARGV[0]
-version=ARGV[1]
-encoding=ARGV[2]
+filename = ARGV[0]
+version = ARGV[1]
+encoding = ARGV[2]
 
 if filename
-  msg=ONIX::ONIXMessage.new
-  msg.parse(filename,encoding,version)
+  msg = ONIX::ONIXMessage.new
+  msg.parse(filename, encoding, version)
 
   puts "Release: #{msg.release}"
 
@@ -18,7 +18,7 @@ if filename
     end
   end
   if msg.sent_date_time
-    puts "Sent date: #{msg.sent_date_time}"
+    puts "Sent date: #{msg.sent_date_time.human}"
   end
 
   msg.products.each do |product|
@@ -140,9 +140,9 @@ if filename
       end
     end
 
-    current_price=product.current_price_amount_for('EUR', 'FR')
+    current_price = product.current_price_amount_for('EUR', 'FR')
     if current_price
-      puts " Current EUR price for FR: #{current_price/100.0} EUR"
+      puts " Current EUR price for FR: #{current_price / 100.0} EUR"
     end
 
     if product.sold_separately?
@@ -150,26 +150,26 @@ if filename
       puts " Supplies:"
 
       product.supplies_with_default_tax.each do |supply|
-#    if supply[:availability_date]
-#      puts " Availability date : #{supply[:availability_date]}"
-#    end
-        output="  "
+        #    if supply[:availability_date]
+        #      puts " Availability date : #{supply[:availability_date]}"
+        #    end
+        output = "  "
 
         if supply[:available]
-          output+="Available in "
+          output += "Available in "
         else
-          output+="Not available in "
+          output += "Not available in "
         end
 
         if ONIX::Territory.worldwide?(supply[:territory])
-          output+="WORLD"
+          output += "WORLD"
         else
-          output+=supply[:territory].join(", ")
+          output += supply[:territory].join(", ")
         end
 
         if supply[:availability_date]
-          output+=" starting #{supply[:availability_date]}"
-          output+=" (#{supply[:availability_date].to_date})" if supply[:availability_date].class==Time
+          output += " starting #{supply[:availability_date]}"
+          output += " (#{supply[:availability_date].to_date})" if supply[:availability_date].class == Time
 
         end
 
@@ -178,23 +178,23 @@ if filename
         puts "  Prices:"
 
         supply[:prices].each do |price|
-          output="   "
+          output = "   "
 
-          output+="#{price[:amount].to_f/100.0} #{supply[:currency]}"
+          output += "#{price[:amount].to_f / 100.0} #{supply[:currency]}"
 
           if supply[:including_tax]
-            output+=" tax included"
+            output += " tax included"
           else
-            output+=" tax excluded"
+            output += " tax excluded"
           end
 
           if price[:from_date]
-            output+=" from #{price[:from_date]}"
-            output+=" (#{price[:from_date].to_date})" if price[:from_date].class==Time
+            output += " from #{price[:from_date]}"
+            output += " (#{price[:from_date].to_date})" if price[:from_date].class == Time
           end
           if price[:until_date]
-            output+=" until #{price[:until_date]}"
-            output+=" (#{price[:until_date].to_date})" if price[:until_date].class==Time
+            output += " until #{price[:until_date]}"
+            output += " (#{price[:until_date].to_date})" if price[:until_date].class == Time
           end
           puts output
         end

data/bin/onix_serialize.rb

@@ -1,27 +1,23 @@
 #!/usr/bin/env ruby
 require 'im_onix'
 require 'onix/serializer'
-filename=ARGV[0]
-version=ARGV[1]
+filename = ARGV[0]
+format = ARGV[1] || "dump"
+version = ARGV[2]
 
 if filename
-  msg=ONIX::ONIXMessage.new
+  msg = ONIX::ONIXMessage.new
   msg.parse(filename, nil, version)
-  if true
-    msg.products.each do |product|
-      ONIX::Serializer::Dump::Subset.serialize(STDOUT, "Product", product)
-    end
-  end
-
-  if false
+  case format
+  when "dump"
+    ONIX::Serializer::Dump.serialize(STDOUT, msg)
+  when "xml"
     builder = Nokogiri::XML::Builder.new(:encoding => "UTF-8") do |xml|
-      xml.ONIXMessage {
-        msg.products.each do |product|
-          ONIX::Serializer::Default::Subset.serialize(xml, "Product", product)
-        end
-      }
+      ONIX::Serializer::Default.serialize(xml, msg)
     end
     puts builder.to_xml
   end
-
+else
+  puts "ImOnix serializer"
+  puts "Usage: onix_serialize.rb file.xml [dump|xml] [3.0|2.1]"
 end

data/data/codelists/codelist-1.yml

@@ -5,8 +5,7 @@
   '03': NotificationConfirmedOnPublication
   '04': UpdatePartial
   '05': Delete
-  08: NoticeOfSale
-  09: NoticeOfAcquisition
-  '12': UpdateSupplydetailOnly
-  '13': UpdateMarketrepresentationOnly
-  '14': UpdateSupplydetailAndMarketrepresentation
+  '08': NoticeOfSale
+  '09': NoticeOfAcquisition
+  '88': TestUpdatePartial
+  '89': TestRecord

data/data/codelists/codelist-10.yml

@@ -3,13 +3,13 @@
   '000': EpublicationContentPackage
   '001': Html
   '002': Pdf
-  '003': Pdfmerchant
+  '003': PdfMerchant
   '004': AdobeEbookReader
   '005': MicrosoftReaderLevel1level3
   '006': MicrosoftReaderLevel5
   '007': Netlibrary
-  008: Metatext
-  009: Mightywords
+  '008': Metatext
+  '009': Mightywords
   '010': EreaderAkaPalmReader
   '011': Softbook
   '012': Rocketbook
@@ -18,8 +18,8 @@
   '015': FranklinEbookman
   '016': Books24x7
   '017': Digitalowl
-  018: Handheldmed
-  019: Wizeup
+  '018': Handheldmed
+  '019': Wizeup
   '020': Tk3
   '021': Litraweb
   '022': Mobipocket
@@ -28,8 +28,8 @@
   '025': Txt
   '026': Exebook
   '027': SonyBbeb
-  028: VitalsourceBookshelf
-  029: Epub
+  '028': VitalsourceBookshelf
+  '029': Epub
   '030': Myilibrary
   '031': Kindle
   '032': GoogleEdition
@@ -38,8 +38,8 @@
   '035': Ebl
   '036': Ebrary
   '037': Isilo
-  038: Plucker
-  039: Vitalbook
+  '038': Plucker
+  '039': Vitalbook
   '040': BookAppForIos
   '041': AndroidApp
   '042': OtherApp
@@ -47,5 +47,7 @@
   '044': Ibook
   '045': Epib
   '046': Scorm
-  098: MultipleFormats
-  099: Unspecified
+  '047': Ebp
+  '048': PagePerfect
+  '098': MultipleFormats
+  '099': Unspecified

data/data/codelists/codelist-100.yml

@@ -6,3 +6,4 @@
   '04': GermanTermsCode
   '05': ProprietaryCommissionCode
   '06': BicCommissionGroupCode
+  '07': IsniBasedDiscountGroupCode

data/data/codelists/codelist-102.yml

@@ -1,5 +1,6 @@
 ---
 :codelist:
   '01': Proprietary
-  '02': BicSalesOutletIdCode
-  '03': OnixSalesOutletIdCode
+  '03': OnixRetailSalesOutletIdCode
+  '04': RetailSalesOutletGln
+  '05': RetailSalesOutletSan

data/data/codelists/codelist-11.yml

@@ -7,7 +7,7 @@
   '05': RichTextFormatRtf
   '06': OpenEbookPublicationStructureOebpsFormatStandard
   '07': Xml
-  08: Sgml
-  09: Exe
+  '08': Sgml
+  '09': Exe
   '10': Ascii
   '11': MobipocketFormat

data/data/codelists/codelist-12.yml

@@ -3,13 +3,18 @@
   '01': UkOpenMarketEdition
   '02': AirportEdition
   '03': Sonderausgabe
-  '04': PocketPaperback
+  '04': PocketBook
   '05': InternationalEditionUs
   '06': LibraryAudioEdition
   '07': UsOpenMarketEdition
-  08: LivreScolaireDéclaréParLéditeur
-  09: LivreScolaireNonSpécifié
+  '08': LivreScolaireDeclareParLediteur
+  '09': LivreScolaireNonSpecifie
   '10': SupplementToNewspaper
   '11': PrecioLibreTextbook
   '12': NewsOutletEdition
   '13': UsTextbook
+  '14': EBookShort
+  '15': SuperpocketBook
+  '16': BeauLivre
+  '17': Podcast
+  '18': Periodical

data/data/codelists/codelist-121.yml

@@ -1,7 +1,11 @@
 ---
 :codelist:
+  Adlm: Adlam
   Afak: Afaka
+  Aghb: CaucasianAlbanian
+  Ahom: AhomTaiAhom
   Arab: Arabic
+  Aran: ArabicNastaliqVariant
   Armi: ImperialAramaic
   Armn: Armenian
   Avst: Avestan
@@ -9,7 +13,8 @@
   Bamu: Bamun
   Bass: BassaVah
   Batk: Batak
-  Beng: Bengali
+  Beng: BengaliBangla
+  Bhks: Bhaiksuki
   Blis: Blissymbols
   Bopo: Bopomofo
   Brah: Brahmi
@@ -27,6 +32,7 @@
   Cyrl: Cyrillic
   Cyrs: CyrillicOldChurchSlavonicVariant
   Deva: DevanagariNagari
+  Dogr: Dogra
   Dsrt: DeseretMormon
   Dupl: DuployanShorthandDuployanStenography
   Egyd: EgyptianDemotic
@@ -35,43 +41,52 @@
   Elba: Elbasan
   Ethi: EthiopicGeez
   Geok: KhutsuriAsomtavruliAndKhutsuri
-  Geor: GeorgianMkhedruli
+  Geor: GeorgianMkhedruliAndMtavruli
   Glag: Glagolitic
+  Gong: GunjalaGondi
+  Gonm: MasaramGondi
   Goth: Gothic
   Gran: Grantha
   Grek: Greek
   Gujr: Gujarati
   Guru: Gurmukhi
-  Hang: HangulHangŭlHangeul
+  Hanb: HanWithBopomofo
+  Hang: HangulHangulHangeul
   Hani: HanHanziKanjiHanja
-  Hano: HanunooHanunóo
+  Hano: HanunooHanunoo
   Hans: HanSimplifiedVariant
   Hant: HanTraditionalVariant
+  Hatr: Hatran
   Hebr: Hebrew
   Hira: Hiragana
+  Hluw: AnatolianHieroglyphsLuwianHieroglyphsHittiteHieroglyphs
   Hmng: PahawhHmong
   Hrkt: JapaneseSyllabariesAliasForHiraganaKatakana
-  Hung: OldHungarian
+  Hung: OldHungarianHungarianRunic
   Inds: IndusHarappan
   Ital: OldItalicEtruscanOscanEtc
+  Jamo: JamoAliasForJamoSubsetOfHangul
   Java: Javanese
-  Jpan: JapaneseSyllabariesAliasForHanHiraganaKatakana
+  Jpan: JapaneseAliasForHanHiraganaKatakana
   Jurc: Jurchen
   Kali: KayahLi
   Kana: Katakana
   Khar: Kharoshthi
   Khmr: Khmer
   Khoj: Khojki
+  Kitl: KhitanLargeScript
+  Kits: KhitanSmallScript
   Knda: Kannada
   Kore: KoreanAliasForHangulHan
   Kpel: Kpelle
   Kthi: Kaithi
-  Lana: LannaTaiTham
+  Lana: TaiThamLanna
   Laoo: Lao
   Latf: LatinFrakturVariant
   Latg: LatinGaelicVariant
   Latn: Latin
-  Lepc: LepchaRóng
+  Leke: Leke
+  Lepc: LepchaRong
   Limb: Limbu
   Lina: LinearA
   Linb: LinearB
@@ -79,36 +94,46 @@
   Loma: Loma
   Lyci: Lycian
   Lydi: Lydian
+  Mahj: Mahajani
+  Maka: Makasar
   Mand: MandaicMandaean
   Mani: Manichaean
+  Marc: Marchen
   Maya: MayanHieroglyphs
-  Mend: Mende
+  Medf: MedefaidrinOberiOkaimeOberiOkaime
+  Mend: MendeKikakui
   Merc: MeroiticCursive
   Mero: MeroiticHieroglyphs
   Mlym: Malayalam
+  Modi: ModiModi
   Mong: Mongolian
   Moon: MoonMoonCodeMoonScriptMoonType
   Mroo: MroMru
   Mtei: MeiteiMayekMeitheiMeetei
+  Mult: Multani
   Mymr: MyanmarBurmese
   Narb: OldNorthArabianAncientNorthArabian
   Nbat: Nabatean
-  Nkgb: NakhiGebaNakhi²ggŏ¹bawNaxiGeba
+  Newa: NewaNewarNewariNepalaLipi
+  Nkgb: NakhiGebaNaKhi2ggo1bawNaxiGeba
   Nkoo: Nko
-  Nshu: Nüshu
+  Nshu: Nushu
   Ogam: Ogham
   Olck: OlChikiOlCemetOlSantali
   Orkh: OldTurkicOrkhonRunic
-  Orya: Oriya
+  Orya: OriyaOdia
+  Osge: Osage
   Osma: Osmanya
   Palm: Palmyrene
+  Pauc: PauCinHau
   Perm: OldPermic
-  Phag: Phagspa
+  Phag: PhagsPa
   Phli: InscriptionalPahlavi
   Phlp: PsalterPahlavi
   Phlv: BookPahlavi
   Phnx: Phoenician
   Plrd: MiaoPollard
+  Piqd: KlingonKliPlqad
   Prti: InscriptionalParthian
   Qaaa: ReservedForPrivateUseStart
   Qabx: ReservedForPrivateUseEnd
@@ -121,10 +146,12 @@
   Saur: Saurashtra
   Sgnw: Signwriting
   Shaw: ShavianShaw
-  Shrd: SharadaŚāradā
+  Shrd: SharadaSarada
+  Sidd: SiddhamSiddhamSiddhamatrka
   Sind: KhudawadiSindhi
   Sinh: Sinhala
   Sora: SoraSompeng
+  Soyo: Soyombo
   Sund: Sundanese
   Sylo: SylotiNagri
   Syrc: Syriac
@@ -132,7 +159,7 @@
   Syrj: SyriacWesternVariant
   Syrn: SyriacEasternVariant
   Tagb: Tagbanwa
-  Takr: TakriṬākrīṬāṅkrī
+  Takr: TakriTakriTankri
   Tale: TaiLe
   Talu: NewTaiLue
   Taml: Tamil
@@ -145,17 +172,19 @@
   Thaa: Thaana
   Thai: Thai
   Tibt: Tibetan
-  Tirh: Tiruta
+  Tirh: Tirhuta
   Ugar: Ugaritic
   Vaii: Vai
   Visp: VisibleSpeech
   Wara: WarangCitiVarangKshiti
   Wole: Woleai
   Xpeo: OldPersian
-  Xsux: CuneiformSumeroakkadian
+  Xsux: CuneiformSumeroAkkadian
   Yiii: Yi
+  Zanb: ZanabazarSquareZanabazarinDorboljinUsegXewteeDorboljinBicigHorizontalSquareScript
   Zinh: CodeForInheritedScript
   Zmth: MathematicalNotation
+  Zsye: SymbolsEmojiVariant
   Zsym: Symbols
   Zxxx: CodeForUnwrittenDocuments
   Zyyy: CodeForUndeterminedScript