im_onix 1.0.2

data/bin/onix_pp.rb

@@ -0,0 +1,216 @@
+#!/usr/bin/env ruby
+require 'im_onix'
+
+filename=ARGV[0]
+version=ARGV[1]
+encoding=ARGV[2]
+
+if filename
+  msg=ONIX::ONIXMessage.new
+  msg.parse(filename,encoding,version)
+
+  puts "Release: #{msg.release}"
+
+  if msg.sender
+    puts "Sender: #{msg.sender.name}"
+    if msg.sender.gln
+      puts "Sender's GLN: #{msg.sender.gln}"
+    end
+  end
+  if msg.sent_date_time
+    puts "Sent date: #{msg.sent_date_time}"
+  end
+
+  msg.products.each do |product|
+#  pp product
+    puts "---"
+    puts " EAN: #{product.ean}"
+    puts " Title: #{product.title}"
+    puts " Language: #{product.language_name_of_text} (#{product.language_code_of_text})"
+    if product.subtitle
+      puts " Subtitle: #{product.subtitle}"
+    end
+    if product.publication_date
+      puts " Publication date: #{product.publication_date}"
+    end
+    if product.embargo_date
+      puts " Embargo date: #{product.embargo_date}"
+    end
+    if product.preorder_embargo_date
+      puts " Preorder embargo date: #{product.preorder_embargo_date}"
+    end
+    if product.public_announcement_date
+      puts " Public announcement date: #{product.public_announcement_date}"
+    end
+    if product.pages
+      puts " Pages: #{product.pages}"
+    end
+    if product.publisher_collection_title
+      puts " Collection: #{product.publisher_collection_title}"
+    end
+
+    product.bisac_categories_codes.each do |bcc|
+      puts " BISAC: #{bcc}"
+    end
+
+    product.clil_categories_codes.each do |bcc|
+      puts " CLIL: #{bcc}"
+    end
+
+    puts " Description: #{product.raw_description}"
+
+    if product.frontcover_url
+      puts " Frontcover: #{product.frontcover_url} (#{product.frontcover_last_updated})"
+    end
+
+    if product.epub_sample_url
+      puts " Sample: #{product.epub_sample_url}"
+    end
+
+    if product.keywords.length > 0
+      puts " Keywords: #{product.keywords.join(', ')}"
+    end
+
+    if product.publisher_name
+      puts " Publisher: #{product.publisher_name}"
+    end
+
+#  puts " Publisher: #{product.publisher_name}"
+    if product.imprint_name
+      puts " Imprint: #{product.imprint_name}"
+    end
+
+    if product.distributor_name
+      puts " Distributor: #{product.distributor_name}"
+    end
+
+    if product.onix_outlets_values.length > 0
+      puts " Outlets: #{product.onix_outlets_values}"
+    end
+
+    product.contributors.each do |c|
+      puts " Contributor:"
+      puts "  Name: #{c.name}"
+      puts "  Role: #{c.role.human}"
+      if c.biography
+        puts "  Biography: #{c.raw_biography}"
+      end
+    end
+
+    if product.digital?
+      puts " -- Digital product"
+
+      if product.bundle?
+        puts " Multiple files bundle"
+        if product.file_description
+          puts " Description: #{product.raw_file_description}"
+        end
+
+        product.parts.each do |part|
+          puts "  -- Part"
+          if part.ean
+            puts "  EAN: #{part.ean}"
+          end
+          puts "  Format: #{part.file_format}"
+          if part.file_description
+            puts "  Description: #{part.raw_file_description}"
+          end
+          if part.protection_type
+            puts "  Protection: #{part.protection_type}"
+          end
+          if part.filesize
+            puts "  Filesize: #{part.filesize} bytes"
+          end
+        end
+
+      else
+        puts " Single file"
+        puts " Format: #{product.file_format}"
+        if product.file_description
+          puts " Description: #{product.raw_file_description}"
+        end
+        if product.protection_type
+          puts " Protection: #{product.protection_type}"
+        end
+        if product.filesize
+          puts " Filesize: #{product.filesize} bytes"
+        end
+      end
+
+      if product.print_product
+        puts " Paper EAN: #{product.print_product.ean}"
+      end
+    end
+
+#  pp product.supplies
+
+    current_price=product.current_price_amount_for('EUR', 'FR')
+    if current_price
+      puts " Current EUR price for FR: #{current_price/100.0} EUR"
+    end
+
+    if product.sold_separately?
+      puts " Available: #{product.available?}"
+      puts " Supplies:"
+
+      product.supplies_with_default_tax.each do |supply|
+#    if supply[:availability_date]
+#      puts " Availability date : #{supply[:availability_date]}"
+#    end
+        output="  "
+
+        if supply[:available]
+          output+="Available in "
+        else
+          output+="Not available in "
+        end
+
+        if ONIX::Territory.worldwide?(supply[:territory])
+          output+="WORLD"
+        else
+          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
+
+        end
+
+        puts output
+
+        puts "  Prices:"
+
+        supply[:prices].each do |price|
+          output="   "
+
+          output+="#{price[:amount].to_f/100.0} #{supply[:currency]}"
+
+          if supply[:including_tax]
+            output+=" tax included"
+          else
+            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
+          end
+          if price[:until_date]
+            output+=" until #{price[:until_date]}"
+            output+=" (#{price[:until_date].to_date})" if price[:until_date].class==Time
+          end
+          puts output
+        end
+      end
+    else
+      puts " Not sold separately"
+    end
+  end
+  puts "---"
+  puts "Number of products: #{msg.products.count}"
+
+else
+  puts "ONIX 3.0 pretty printer"
+  puts "Usage: onix_pp.rb onix.xml"
+end

data/bin/onix_split.rb

@@ -0,0 +1,72 @@
+#!/usr/bin/env ruby
+require 'im_onix'
+
+filename=ARGV[0]
+cnt=ARGV[1] || 1000
+cnt=cnt.to_i
+
+class OnixSplitter
+
+  def initialize(filename, output_name)
+    @filename=filename
+    @output_name=output_name
+  end
+
+  def write_part(file_count, part)
+    out_filename=@output_name+"."+file_count.to_s+".xml"
+    puts "Write file #{out_filename}"
+    fw=File.open(out_filename, 'w')
+    fw.write("<ONIXMessage release=\"3.0\">\n")
+    part.each do |p|
+      fw.write p
+    end
+    fw.write("</ONIXMessage>\n")
+    fw.close
+  end
+
+  def count
+    current_part_count=0
+    ONIX::Helper.each_xml_product(@filename) do |product_str|
+      current_part_count+=1
+    end
+    current_part_count
+  end
+
+  def split(max_parts)
+    file_count=0
+    current_part=[]
+    current_part_count=0
+    ONIX::Helper.each_xml_product(@filename) do |product_str|
+      tmp_msg=ONIX::ONIXMessage.new
+      tmp_msg.parse(product_str)
+
+      current_part[current_part_count] ||=""
+      current_part[current_part_count] += product_str + "\n"
+
+      if tmp_msg.products.first.sold_separately?
+        current_part_count+=1
+      end
+
+      if current_part_count > max_parts-1
+        write_part(file_count, current_part)
+        current_part=[]
+        current_part_count=0
+        file_count+=1
+      end
+    end
+
+    if current_part_count > 0
+      write_part(file_count, current_part)
+      current_part=[]
+      current_part_count=0
+    end
+
+    true
+  end
+
+end
+
+splitter=OnixSplitter.new(filename, "out/splitted")
+splitter.split(cnt)
+puts splitter.count
+

data/data/codelists/codelist-1.yml

@@ -0,0 +1,12 @@
+---
+:codelist:
+  '01': EarlyNotification
+  '02': AdvanceNotificationConfirmed
+  '03': NotificationConfirmedOnPublication
+  '04': UpdatePartial
+  '05': Delete
+  08: NoticeOfSale
+  09: NoticeOfAcquisition
+  '12': UpdateSupplydetailOnly
+  '13': UpdateMarketrepresentationOnly
+  '14': UpdateSupplydetailAndMarketrepresentation

data/data/codelists/codelist-10.yml

@@ -0,0 +1,51 @@
+---
+:codelist:
+  '000': EpublicationContentPackage
+  '001': Html
+  '002': Pdf
+  '003': Pdfmerchant
+  '004': AdobeEbookReader
+  '005': MicrosoftReaderLevel1level3
+  '006': MicrosoftReaderLevel5
+  '007': Netlibrary
+  008: Metatext
+  009: Mightywords
+  '010': EreaderAkaPalmReader
+  '011': Softbook
+  '012': Rocketbook
+  '013': GemstarReb1100
+  '014': GemstarReb1200
+  '015': FranklinEbookman
+  '016': Books24x7
+  '017': Digitalowl
+  018: Handheldmed
+  019: Wizeup
+  '020': Tk3
+  '021': Litraweb
+  '022': Mobipocket
+  '023': OpenEbook
+  '024': TownCompassDataviewer
+  '025': Txt
+  '026': Exebook
+  '027': SonyBbeb
+  028: VitalsourceBookshelf
+  029: Epub
+  '030': Myilibrary
+  '031': Kindle
+  '032': GoogleEdition
+  '033': Vook
+  '034': Dxreader
+  '035': Ebl
+  '036': Ebrary
+  '037': Isilo
+  038: Plucker
+  039: Vitalbook
+  '040': BookAppForIos
+  '041': AndroidApp
+  '042': OtherApp
+  '043': Xps
+  '044': Ibook
+  '045': Epib
+  '046': Scorm
+  098: MultipleFormats
+  099: Unspecified

data/data/codelists/codelist-100.yml

@@ -0,0 +1,8 @@
+---
+:codelist:
+  '01': BicDiscountGroupCode
+  '02': ProprietaryDiscountCode
+  '03': Boeksoort
+  '04': GermanTermsCode
+  '05': ProprietaryCommissionCode
+  '06': BicCommissionGroupCode

data/data/codelists/codelist-101.yml

@@ -0,0 +1,7 @@
+---
+:codelist:
+  '01': Proprietary
+  '02': Pnd
+  '04': Lccn
+  '16': Isni
+  '25': Gnd

data/data/codelists/codelist-102.yml

@@ -0,0 +1,5 @@
+---
+:codelist:
+  '01': Proprietary
+  '02': BicSalesOutletIdCode
+  '03': OnixSalesOutletIdCode

data/data/codelists/codelist-11.yml

@@ -0,0 +1,13 @@
+---
+:codelist:
+  '01': Html
+  '02': Pdf
+  '03': MicrosoftReader
+  '04': Rocketbook
+  '05': RichTextFormatRtf
+  '06': OpenEbookPublicationStructureOebpsFormatStandard
+  '07': Xml
+  08: Sgml
+  09: Exe
+  '10': Ascii
+  '11': MobipocketFormat

data/data/codelists/codelist-12.yml

@@ -0,0 +1,15 @@
+---
+:codelist:
+  '01': UkOpenMarketEdition
+  '02': AirportEdition
+  '03': Sonderausgabe
+  '04': PocketPaperback
+  '05': InternationalEditionUs
+  '06': LibraryAudioEdition
+  '07': UsOpenMarketEdition
+  08: LivreScolaireDéclaréParLéditeur
+  09: LivreScolaireNonSpécifié
+  '10': SupplementToNewspaper
+  '11': PrecioLibreTextbook
+  '12': NewsOutletEdition
+  '13': UsTextbook

data/data/codelists/codelist-121.yml

@@ -0,0 +1,162 @@
+---
+:codelist:
+  Afak: Afaka
+  Arab: Arabic
+  Armi: ImperialAramaic
+  Armn: Armenian
+  Avst: Avestan
+  Bali: Balinese
+  Bamu: Bamun
+  Bass: BassaVah
+  Batk: Batak
+  Beng: Bengali
+  Blis: Blissymbols
+  Bopo: Bopomofo
+  Brah: Brahmi
+  Brai: Braille
+  Bugi: Buginese
+  Buhd: Buhid
+  Cakm: Chakma
+  Cans: UnifiedCanadianAboriginalSyllabics
+  Cari: Carian
+  Cham: Cham
+  Cher: Cherokee
+  Cirt: Cirth
+  Copt: Coptic
+  Cprt: Cypriot
+  Cyrl: Cyrillic
+  Cyrs: CyrillicOldChurchSlavonicVariant
+  Deva: DevanagariNagari
+  Dsrt: DeseretMormon
+  Dupl: DuployanShorthandDuployanStenography
+  Egyd: EgyptianDemotic
+  Egyh: EgyptianHieratic
+  Egyp: EgyptianHieroglyphs
+  Elba: Elbasan
+  Ethi: EthiopicGeez
+  Geok: KhutsuriAsomtavruliAndKhutsuri
+  Geor: GeorgianMkhedruli
+  Glag: Glagolitic
+  Goth: Gothic
+  Gran: Grantha
+  Grek: Greek
+  Gujr: Gujarati
+  Guru: Gurmukhi
+  Hang: HangulHangŭlHangeul
+  Hani: HanHanziKanjiHanja
+  Hano: HanunooHanunóo
+  Hans: HanSimplifiedVariant
+  Hant: HanTraditionalVariant
+  Hebr: Hebrew
+  Hira: Hiragana
+  Hmng: PahawhHmong
+  Hrkt: JapaneseSyllabariesAliasForHiraganaKatakana
+  Hung: OldHungarian
+  Inds: IndusHarappan
+  Ital: OldItalicEtruscanOscanEtc
+  Java: Javanese
+  Jpan: JapaneseSyllabariesAliasForHanHiraganaKatakana
+  Jurc: Jurchen
+  Kali: KayahLi
+  Kana: Katakana
+  Khar: Kharoshthi
+  Khmr: Khmer
+  Khoj: Khojki
+  Knda: Kannada
+  Kore: KoreanAliasForHangulHan
+  Kpel: Kpelle
+  Kthi: Kaithi
+  Lana: LannaTaiTham
+  Laoo: Lao
+  Latf: LatinFrakturVariant
+  Latg: LatinGaelicVariant
+  Latn: Latin
+  Lepc: LepchaRóng
+  Limb: Limbu
+  Lina: LinearA
+  Linb: LinearB
+  Lisu: LisuFraser
+  Loma: Loma
+  Lyci: Lycian
+  Lydi: Lydian
+  Mand: MandaicMandaean
+  Mani: Manichaean
+  Maya: MayanHieroglyphs
+  Mend: Mende
+  Merc: MeroiticCursive
+  Mero: MeroiticHieroglyphs
+  Mlym: Malayalam
+  Mong: Mongolian
+  Moon: MoonMoonCodeMoonScriptMoonType
+  Mroo: MroMru
+  Mtei: MeiteiMayekMeitheiMeetei
+  Mymr: MyanmarBurmese
+  Narb: OldNorthArabianAncientNorthArabian
+  Nbat: Nabatean
+  Nkgb: NakhiGebaNakhi²ggŏ¹bawNaxiGeba
+  Nkoo: Nko
+  Nshu: Nüshu
+  Ogam: Ogham
+  Olck: OlChikiOlCemetOlSantali
+  Orkh: OldTurkicOrkhonRunic
+  Orya: Oriya
+  Osma: Osmanya
+  Palm: Palmyrene
+  Perm: OldPermic
+  Phag: Phagspa
+  Phli: InscriptionalPahlavi
+  Phlp: PsalterPahlavi
+  Phlv: BookPahlavi
+  Phnx: Phoenician
+  Plrd: MiaoPollard
+  Prti: InscriptionalParthian
+  Qaaa: ReservedForPrivateUseStart
+  Qabx: ReservedForPrivateUseEnd
+  Rjng: RejangRedjangKaganga
+  Roro: Rongorongo
+  Runr: Runic
+  Samr: Samaritan
+  Sara: Sarati
+  Sarb: OldSouthArabian
+  Saur: Saurashtra
+  Sgnw: Signwriting
+  Shaw: ShavianShaw
+  Shrd: SharadaŚāradā
+  Sind: KhudawadiSindhi
+  Sinh: Sinhala
+  Sora: SoraSompeng
+  Sund: Sundanese
+  Sylo: SylotiNagri
+  Syrc: Syriac
+  Syre: SyriacEstrangeloVariant
+  Syrj: SyriacWesternVariant
+  Syrn: SyriacEasternVariant
+  Tagb: Tagbanwa
+  Takr: TakriṬākrīṬāṅkrī
+  Tale: TaiLe
+  Talu: NewTaiLue
+  Taml: Tamil
+  Tang: Tangut
+  Tavt: TaiViet
+  Telu: Telugu
+  Teng: Tengwar
+  Tfng: TifinaghBerber
+  Tglg: TagalogBaybayinAlibata
+  Thaa: Thaana
+  Thai: Thai
+  Tibt: Tibetan
+  Tirh: Tiruta
+  Ugar: Ugaritic
+  Vaii: Vai
+  Visp: VisibleSpeech
+  Wara: WarangCitiVarangKshiti
+  Wole: Woleai
+  Xpeo: OldPersian
+  Xsux: CuneiformSumeroakkadian
+  Yiii: Yi
+  Zinh: CodeForInheritedScript
+  Zmth: MathematicalNotation
+  Zsym: Symbols
+  Zxxx: CodeForUnwrittenDocuments
+  Zyyy: CodeForUndeterminedScript
+  Zzzz: CodeForUncodedScript

data/data/codelists/codelist-13.yml

@@ -0,0 +1,10 @@
+---
+:codelist:
+  '01': Proprietary
+  '02': Issn
+  '03': GermanNationalBibliographySeriesId
+  '04': GermanBooksInPrintSeriesId
+  '05': ElectreSeriesId
+  '06': Doi
+  '15': Isbn13
+  '22': Urn

data/data/codelists/codelist-138.yml

@@ -0,0 +1,8 @@
+---
+:codelist:
+  SFS4900: FinnishStandardSfs4900
+  SFS5807: FinnishStandardSfs5807
+  SFS5755: FinnishStandardSfs5755
+  SFS3602: FinnishStandardSfs5824
+  ISO3602: Iso3602
+  ISO7098: Iso7098

data/data/codelists/codelist-139.yml

@@ -0,0 +1,102 @@
+---
+:codelist:
+  ACM: ACMoore
+  AAP: Aandp
+  ALB: Albertsons
+  AMZ: Amazon
+  ANR: AngusAndRobertson
+  ANB: Anobii
+  APC: Apple
+  ASD: Asda
+  AUD: Audible
+  BDL: BDalton
+  BNO: BarnesAndNoble
+  BBB: BedBathAndBeyond
+  BST: BestBuy
+  BIL: Bilbary
+  BJW: BjsWholesaleClub
+  BLK: Blackwells
+  BCA: BookClubAssociates
+  BSH: Bookish
+  BKP: Bookpeople
+  BKM: Booksamillion
+  BRD: Borders
+  BRB: BordersbooksEtc
+  BRT: BritishBookshops
+  CDL: CasaDelLibro
+  CHD: Christianbookcom
+  CST: Costco
+  CRB: CrateAndBarrel
+  CVS: CvsDrugStores
+  DSG: DicksSportingGoods
+  DYM: Dymocks
+  ELC: EarlyLearningCentre
+  ESN: Eason
+  EBC: EbooksCorp
+  ECH: Echristian
+  ECI: ElCorteInglés
+  ELE: Electre
+  ELB: Elibse
+  EMP: Empik
+  ENH: EnglishHeritage
+  FDB: Feedbooks
+  FNC: Fnac
+  FRY: FrysElectronics
+  GMS: Gamestop
+  GOO: GoogleBooks
+  GOS: BlinkboxFormerlyGospokenmobcast
+  HST: HastingsEntertainment
+  HMV: Hmv
+  HMD: HomeDepot
+  IND: Indigochapters
+  IMM: Immateriel.fr
+  JSM: JohnSmithAndSon
+  KMT: Kmart
+  KNB: Knfbblio
+  KNO: KnoInc
+  KBO: Kobo
+  KOO: Koorong
+  KRG: Kroger
+  LWE: Lowes
+  MKS: MarksAndSpencer
+  MAT: Matras
+  MMS: MediaMarktsaturn
+  MCR: Microcenter
+  MRR: Morrisons
+  MTC: Mothercare
+  MYB: Myboox
+  NTR: NationalTrust
+  NUM: Numilog
+  OFD: OfficeDepot
+  OFM: OfficeMax
+  OLF: Olf
+  PST: PastTimes
+  PTS: PetSmart
+  PTC: Petco
+  PLY: Playcom
+  PTB: PotteryBarn
+  RDB: Readbooks
+  RCL: Readcloud
+  RST: RestorationHardware
+  RET: Rethink
+  RTZ: RitzCamera
+  SFW: Safeway
+  SNS: Sainsburys
+  SLF: Selfridges
+  SNY: Sony
+  STP: Staples
+  TEA: TheEbookAlternative
+  TRG: Target
+  TES: Tesco
+  TSR: ToysRUs
+  TSO: TsoTheStationeryOffice
+  VRG: VirginMegastores
+  WHS: WHSmith
+  WTR: Waitrose
+  WLM: Walmart
+  WST: Waterstones
+  WHT: Whitcouls
+  WLS: WilliamsSonoma
+  WLW: Woolworths
+  ZVV: Zavvi
+  ZZZ: Other

data/data/codelists/codelist-14.yml

@@ -0,0 +1,6 @@
+---
+:codelist:
+  '00': Undefined
+  '01': SentenceCase
+  '02': TitleCase
+  '03': AllCapitals

data/data/codelists/codelist-140.yml

@@ -0,0 +1,9 @@
+---
+:codelist:
+  '01': WarningChokingHazardSmallPartsNotForChildrenUnder3Yrs
+  '02': WarningChokingHazardThisToyIsASmallBallNotForChildrenUnder3Yrs
+  '03': WarningChokingHazardToyContainsASmallBallNotForChildrenUnder3Yrs
+  '04': WarningChokingHazardChildrenUnder8YrsCanChokeOrSuffocateOnUninflatedOrBrokenBalloonsAdultSupervisionRequiredKeepUninflatedBalloonsFromChildrenDiscardBrokenBalloonsAtOnce
+  '05': WarningChokingHazardThisToyIsAMarbleNotForChildrenUnder3Yrs
+  '06': WarningChokingHazardToyContainsAMarbleNotForChildrenUnder3Yrs
+  '07': NoChokingHazardWarningNecessary