da_fetch_mapping.fix

# Conversion rules for CBS2DigitalesArchiv
# @author: Felix Hemme, Luisa Kramer
# @version: 2022-09-05 v3.0.2

# Accept variable
add_field('target','{{target}}')
add_field('isil','{{isil}}')

# Add download method and default to selen
add_field('download_method','selen')

if all_match('target','owc-de-206')
  # For PPN from owc-de-206
  marc_map('001','identifier\.ppn')
elsif all_match('target','k10plus')
  # For PPN from K10plus
  marc_map('001','identifier\.ppn')
elsif all_match('target','ebooks')
  # For PPN from E-Book-Pool
  marc_map('001','identifier\.ppn')
  prepend('identifier\.ppn','EBP')
elsif all_match('target','nl-monographien')
  # For PPN from NL-Mono-Pool
  marc_map('001','identifier\.ppn')
  prepend('identifier\.ppn','NLM')
elsif all_match('target','nl-zeitschriften')
  # For PPN from NL-ZS-Pool
  marc_map('001','identifier\.ppn')
  prepend('identifier\.ppn','NLZ')
end

# Type "Article" (article from journal)
# 002@ Pos.2 = "s"
do marc_each()
  if marc_match('7737|||s','s')
    set_field('type','b')
  # No 773 $7 for target nl-zeitschriften
  # Check on LDR/07 bibliographic level
  elsif all_match('target','nl-zeitschriften')
    if marc_match('LDR/7','b')
      set_field('type','b')
    end
  end
end

# Type "Book Part" (article from monograph)
# 002@ Pos.2 = "s"
do marc_each()
  if marc_match('7737|||m','m')
    set_field('type','bp')
  # No 773 $7 for target ebooks and nl-monographien
  # Check on LDR/07 bibliographic level
  elsif all_match('target','ebooks')
    if marc_match('LDR/7','a')
      set_field('type','bp')
    end
  elsif all_match('target','nl-monographien')
    if marc_match('LDR/7','a')
      set_field('type','bp')
    end
  end
end

# Type "Book"
# 002@ Pos.2 = "a" if 013D $8 not one of
# [Konferenzschrift || Bericht || Forschungsbericht || Hochschulschrift]
do marc_each()
  if marc_match('LDR/7','m')
    unless marc_any_match('655[,7]a','Konferenzschrift')
      marc_map('LDR/7','type')
	end
	unless marc_any_match('655[,7]a','Bericht')
	  marc_map('LDR/7','type')
	end
	unless marc_any_match('655[,7]a','Forschungsbericht')
	  marc_map('LDR/7','type')
	end
	unless marc_any_match('655[,7]a','Hochschulschrift')
      marc_map('LDR/7','type')
	end
  end
end

# Type "Periodical Part"
# 002@ Pos.2 = "v"
do marc_each()
  if marc_match('LDR/19','c')
    marc_map('LDR/19','type')
  end
end

# Type "Conference Paper"
do marc_each()
  if marc_any_match('655[,7]a','Konferenzschrift')
    set_field('type','cp')
  end
end

# Type "Proceedings"
do marc_each()
  if marc_any_match('655[,7]a','Bericht')
    set_field('type','p')
  end
end

# Type "Research Report"
do marc_each()
  if marc_any_match('655[,7]a','Forschungsbericht')
    set_field('type','rr')
  end
end

# Type "Thesis"
do marc_each()
  if marc_any_match('655[,7]a','Hochschulschrift')
    set_field('type','t')
  end
end

# Date of publication
marc_map('008/07-10','date\.issued')

# ISSN
# if LDR 07 = m then map from 810/830 $x
# elsif 773 $7 = |||s then map from 773[08]x
# Multiple ISSN are separated from each other with "|"
if marc_any_match('LDR/7','m')
  if marc_has('810')
    marc_map('810x','relation\.issn', join:'|')
  elsif marc_has('830')
    marc_map('810x','relation\.issn', join:'|')
  end
elsif marc_match('7737|||s','s')
  marc_map('773[08]x','relation\.issn', join:'|')
end

# ISBN
# Multiple ISBN are separated from each other with "|"
#do marc each()
if marc_has('020')
  marc_map('020a','identifier\.isbn', join:'|')
elsif marc_has('773')
  marc_map('773[08]z','identifier\.isbn', join:'|')
end

# DOI, URN, Handle are contained in one combined column
# Multiple DOI, URN, Handle are separated from each other with "|"
# Known issue: Prefixes can't be created (2022-08-03)
do marc_each()
  if marc_match('024[7]2','doi') 
    marc_map('024[7]a','identifier\.pi.$append')
  elsif marc_match('024[7]2','hdl')
    marc_map('024[7]a','identifier\.pi.$append')
  elsif marc_match('024[7]2','urn')
    marc_map('024[7]a','identifier\.pi.$append')
  end
end
join_field('identifier\.pi','|')

# Citation/source for journal articles
do marc_each()
  marc_map('952d','econstor\.citation\.volume')
  #marc_map('952j','econstor\.citation\.year')
  marc_map('952e','econstor\.citation\.issue')
  marc_map('952h','econstor\.citation\.startpage')
  replace_all('econstor\.citation\.startpage','(^[\d]{1,})(-)([\d]{1,}$)','$1')
  marc_map('952h','econstor\.citation\.endpage')
  replace_all('econstor\.citation\.endpage','(^[\d]{1,})(-)([\d]{1,}$)','$3')
  marc_map('952i','econstor\.citation\.articlenumber')
end

# Languages
# Multiple languages are separated from each other with "|"
marc_map('041a','language\.iso', join:'|')

# If the number in 084 is a JEL code, it is mapped into column "subject.jel"
# Multiple JEL are separated from each other with "|"
if marc_match('0842','(?i)jelc')
  marc_map('084a','subject\.jel', join:'|')
end

# Check on relationship designators for field 100 and 700, distinction between author, editor, other
# 100 $4 = aut -> contributor.author
# 700 $4 = aut -> contributor.author
if marc_match('1004','aut')
  marc_map('100a','contributor\.author.$append')
end

do marc_each()
  if marc_match('7004','aut')
    marc_map('700a','contributor\.author.$append')
  end
end
join_field('contributor\.author','|')

# 700 $4 = edt -> contributor.editor
do marc_each()
  if marc_match('7004','edt')
    marc_map('700a','contributor\.editor.$append')
  end
end
join_field('contributor\.editor','|')

# All other relationship designators map to contributor.other
do marc_each()
  unless marc_match('7004','(edt|aut)')
    marc_map('700a','contributor\.other.$append')
  end
  unless marc_match('1004','aut')
    marc_map('100a','contributor\.other.$append')
  end
end
join_field('contributor\.other','|')

# Title
marc_map('245ab','title',join:' : ')

# Number/name of part of a work
marc_map('245np','part',join:' ; ')

# Variant title
marc_map('246[1]a','title\.alternative')

# Place of publication and publisher with delimiter " : "
# If independent publication
marc_map('264ab','publisher', join:' : ')
# If book article (type = book part)
if all_match('type','bp')
  marc_map('773[08]d','publisher')
  replace_all('publisher',',?\s?[0-9]{4}$','')
end

# Edition
# If independent publication
marc_map('250a','description\.version')
# If book article (type = book part)
if all_match('type','bp')
  marc_map('773[08]b','description\.version')
end

# Series name
marc_map('490a','relation\.ispartofseries',join:'. ')
#do marc_each()
#  marc_map('490a','relation\.ispartofseries.$append')
#end
join_field('relation\.ispartofseries','|')

# Abstract
do marc_each()
  marc_map('520a','description\.abstract')
end
join_field('description\.abstract','|')

#######################################
#773 cases
#######################################

# 773 BOOK cases

# Book name (Book Parts)
do marc_each()
if all_match('type','bp')
    marc_map('773t','relation\.ispartofbook',join:'|')
  end
end

# Book name (Books)
do marc_each()
  if marc_match('7737|||m','m')
    marc_map('773t','relation\.ispartofbook',join:'|')
  end
end

# Book PPN
if all_match('target','owc-de-206')
  if marc_any_match('7737|||m','m')
    marc_map('773w',tmp,split:1)
  end
end

do list(path:tmp,var:c)
  if all_match(c,'^\(DE-627\).*')
    replace_all(c,'^\([^\)]+\)(.*)$','$1')
    copy_field(c,'relation\.bookppn.$append')
  end
end
join_field('relation\.bookppn','|')
remove_field(tmp)


# 773 JOURNAL cases

# Journal name
do marc_each()
  if marc_match('7737|||s','s')
    marc_map('773t','relation\.ispartofjournal',join:'|')
  elsif any_match('type','^c$')
    marc_map('245ap','relation\.ispartofjournal',join:'. ')
  end
end

# Journal ZDB PPN and ZDB ID
do marc_each()
  #if marc_has('773w')
  if marc_any_match('7737|||s','s')
    marc_map('773w',tmp,split:1)
  end
end

do list(path:tmp,var:c)
  # ZDB PPN (only for target owc-de-206 and nl-zeitschriften)
  if all_match(c,'^\(DE-627\).*')
    if all_match('target','owc-de-206')
      replace_all(c,'^\([^\)]+\)(.*)$','$1')
      copy_field(c,'relation\.journalppn.$append')
    elsif all_match('target','nl-zeitschriften')
      replace_all(c,'^\([^\)]+\)(.*)$','$1')
      prepend('c','NLZ')
      copy_field(c,'relation\.journalppn.$append')
    end
  end
  # ZDB ID (for all targets)
  if all_match(c,'^\(DE-600\).*')
    replace_all(c,'^\([^\)]+\)(.*)$','$1')
    copy_field(c,'relation\.journalzdbid.$append')
  end
end
join_field('relation\.journalppn','|')
join_field('relation\.journalzdbid','|')
remove_field(tmp)

#######################################
# 8XX cases
#######################################

# 810
do marc_each()
  if marc_has('810w')
    marc_map('810w',tmp,split:1)
  end
end

do list(path:tmp,var:c)
  # PPN for series
  if all_match(c,'^\(DE-627\).*')
    replace_all(c,'^\([^\)]+\)(.*)$','$1')
    copy_field(c,'relation\.seriesppn.$append')
  end
  # ZDB ID for series
  if all_match(c,'^\(DE-600\).*')
    replace_all(c,'^\([^\)]+\)(.*)$','$1')
    copy_field(c,'relation\.serieszdbid.$append')
  end
end
join_field('relation\.seriesppn','|')
join_field('relation\.serieszdbid','|')
remove_field(tmp)

# 830
do marc_each()
  if marc_has('830w')
    marc_map('830w',tmp,split:1)
  end
end

do list(path:tmp,var:c)
  # PPN for series
  if all_match(c,'^\(DE-627\).*')
    replace_all(c,'^\([^\)]+\)(.*)$','$1')
    copy_field(c,'relation\.seriesppn.$append')
  end
  # ZDB ID for series
  if all_match(c,'^\(DE-600\).*')
    replace_all(c,'^\([^\)]+\)(.*)$','$1')
    copy_field(c,'relation\.serieszdbid.$append')
  end
end
join_field('relation\.seriesppn','|')
join_field('relation\.serieszdbid','|')
remove_field(tmp)


# Map callnumbers to Lax record PPNs
do marc_each()
  if marc_any_match('980d','ZSM 1')
    set_field('collection_handle','11159/536')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 3')
    set_field('collection_handle','11159/2270')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 4')
    set_field('collection_handle','11159/2926')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 6')
    set_field('collection_handle','11159/4684')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 7')
    set_field('collection_handle','11159/4739')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 8')
    set_field('collection_handle','11159/2551')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 9')
    set_field('collection_handle','11159/3309')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 10')
    set_field('collection_handle','11159/2308')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 11')
    set_field('collection_handle','11159/1678')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 12')
    set_field('collection_handle','11159/1103')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 13')
    set_field('collection_handle','11159/1663')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 14')
    set_field('collection_handle','11159/2275')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 15')
    set_field('collection_handle','11159/1045')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 16')
    set_field('collection_handle','11159/1047')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 17')
    set_field('collection_handle','11159/2290')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 18')
    set_field('collection_handle','11159/141')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 19')
    set_field('collection_handle','11159/3440')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 20')
    set_field('collection_handle','11159/2890')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 21')
    set_field('collection_handle','11159/2918')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 22')
    set_field('collection_handle','11159/5383')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 23')
    set_field('collection_handle','11159/12240')
  end
end
do marc_each()
  if marc_any_match('980d','ZSM 24')
    set_field('collection_handle','11159/654471')
  end
end

# Everything with 8600 DA:oas
do marc_each()
  if marc_any_match('980h','DA:oas')
    set_field('collection_handle','11159/18')
  end
end

# Everything with 8600 DA:ras
do marc_each()
  if marc_any_match('980h','DA:ras')
    set_field('collection_handle','11159/17')
  end
end

# Everything with 8600 DA:cas
do marc_each()
  if marc_any_match('980h','DA:cas')
    set_field('collection_handle','11159/9034')
  end
end

# License statement from 540 is written into column "rights.license"
do marc_each()
    marc_map('540u','rights\.license')
end

# A hardcoded base license statement is written into column "rights"
# 2022-09-09: The field is populated by the ingest script, no
# need to have it written to the CSV file
# set_field('rights','https://zbw.eu/econis-archiv/termsofuse')

# Dependent on ISIL a value is written into 
# "collection_handle",
# the ISIL is extracted and written into "identifier.packageid",
# and sometimes a fixed value written into "type"
if all_match('isil','zdb-33')
  # If ISIL = zdb-33
  marc_map('912ab','identifier\.packageid.$append')
  join_field('identifier\.packageid','|')
  replace_all('identifier\.packageid','(ZDB-[\d]{1,4}-[A-Za-z]{1,})([\d]{4})','$1 \($2\)')
  set_field('collection_handle','11159/6527')
  set_field('type','wp')
elsif all_match('isil','zdb-1-eem')
  # If ISIL = zdb-1-eem
  marc_map('912ab','identifier\.packageid.$append')
  join_field('identifier\.packageid','|')
  replace_all('identifier\.packageid','(ZDB-[\d]{1,4}-[A-Za-z]{1,})([\d]{4})','$1 \($2\)')
  # set_field('identifier\.packageid','')
set_field('collection_handle','11159/8695')
elsif all_match('isil','zdb-30-pdt')
  marc_map('912ab','identifier\.packageid.$append')
  join_field('identifier\.packageid','|')
  replace_all('identifier\.packageid','(ZDB-[\d]{1,4}-[A-Za-z]{1,})([\d]{4})','$1 \($2\)')
  # set_field('identifier\.packageid','')
  set_field('collection_handle','11159/12451')
  set_field('type','t')
elsif all_match('isil','zdb-1-wba')
  # If ISIL = zdb-1-wba
  marc_map('912ab','identifier\.packageid.$append')
  join_field('identifier\.packageid','|')
  replace_all('identifier\.packageid','(ZDB-[\d]{1,4}-[A-Za-z]{1,})([\d]{4})','$1 \($2\)')
  # set_field('identifier\.packageid','')
  set_field('collection_handle','11159/632042')
elsif all_match('isil','zdb-1-bmen')
  # If ISIL = zdb-1-bmen
  marc_map('912ab','identifier\.packageid.$append')
  join_field('identifier\.packageid','|')
  replace_all('identifier\.packageid','(ZDB-[\d]{1,4}-[A-Za-z]{1,})([\d]{4})','$1 \($2\)')
  # set_field('identifier\.packageid','')
  set_field('collection_handle','11159/654928')
elsif any_match('isil','zdb-23-gbe')
  # If ISIL = zdb-23-gbe
  marc_map('912ab','identifier\.packageid.$append')
  join_field('identifier\.packageid','|')
  replace_all('identifier\.packageid','(ZDB-\d+-[A-Za-z]+)(\d{4}(?:-\d{4})?)','$1 ($2)')
  replace_all('identifier\.packageid', '\((\d{4})\)', '($1)')
  # set_field('identifier\.packageid','')
  set_field('collection_handle','11159/654929')
elsif all_match('isil','zbw-1-dgw')
  # If ISIL = zbw-1-dgw
  set_field('identifier\.packageid','ZBW-1-DGW')
  set_field('collection_handle','11159/654929')
elsif all_match('isil','zbw-1-dgl')
  # If ISIL = zbw-1-dgl
  set_field('identifier\.packageid','ZBW-1-DGL')
  set_field('collection_handle','11159/654929')
#elsif all_match('isil','???')
  # PLACEHOLDER
  # If ISIL = 
# set_field('identifier\.packageid','')
end

# URL
marc_map('856[4,0]u','url',join: '|')

# Extract DOI from 856, transform and write into "identifier.pi"
do marc_each()
  if marc_any_match('856[40]u','doi\.org.*')
    # Apply substring of doi only
    marc_replace_all('856[40]u','https?:\/\/(www\.)?doi\.org/','')
    marc_map('856[40]u','identifier\.pi.$append')
  end
end
join_field('identifier\.pi','|')

# Extract SSRN abstract ID from 856 and use it for value in "identifier"
# Extract SSRN abstract ID from 856 and use for value in "filepath"
if all_match('isil','zdb-33')
  do marc_each()
    if marc_any_match('856[40]m','X:ELVSSRN')
      if marc_any_match('856[40]u','ssrn.com/abstract=.*')
        marc_replace_all('856[40]u','https?:\/\/(www\.)?ssrn\.com\/abstract=','')
        marc_map('856[40]u','identifier')
        prepend('identifier','ELVSSRN:')
        marc_map('856[40]u','filepath')
        prepend('filepath','//zbw/metadaten/Work/SSRN/upload/data/pdfs/')
        append('filepath','.pdf')
      end
    end
  end  
end

# Extract Edward Elgar ISBN from 020a and use for value in "filepath"
if all_match('isil','zdb-1-eem')
        marc_map('020a','filepath')
        prepend('filepath','//zbw/metadaten/Work/ZDB-1-EWE_und_ZDB-1-EEM/upload/')
        append('filepath','.pdf') 
end

# Extract Emerald DOI from 024a and use for value in "filepath"
if all_match('isil','zdb-1-bmen')
  do marc_each()
    if marc_match('024[7]2','doi')
      marc_replace_all('024[7]a','10.1016\/|10.1108\/','')
      marc_replace_all('024[7]a', '\(|\)', '_')
      marc_replace_all('024[7]a', 'S', 's')
      marc_map('024[7]a','filepath')
      prepend('filepath','//zbw/metadaten/Work/ZDB-1-EPB_und_ZDB-1-BMEN/upload/')
      append('filepath','.pdf') 
    end
  end
end

# Extract De Gruyter (Standardpaket) DOI from 024a and use for value in "filepath"
if all_match('isil','zdb-23-gbe')
  do marc_each()
    if marc_match('024[7]2','doi')
      marc_replace_all('024[7]a', '\/', '_')
      marc_map('024[7]a','filepath')
      prepend('filepath','//zbw/metadaten/Work/ZBW-1-DGW_und_ZDB-23-GBE/upload/')
      append('filepath','.pdf') 
    end
  end
end

# Extract De Gruyter (Individualpaket bis 2023) DOI from 024a and use for value in "filepath"
if all_match('isil','zdb-1-dgw')
  do marc_each()
    if marc_match('024[7]2','doi')
      marc_replace_all('024[7]a', '\/', '_')
      marc_map('024[7]a','filepath')
      prepend('filepath','//zbw/metadaten/Work/ZBW-1-DGW_und_ZDB-23-GBE/upload/')
      append('filepath','.pdf') 
    end
  end
end

# Extract De Gruyter (Individualpaket ab 2024) DOI from 024a and use for value in "filepath"
if all_match('isil','zdb-1-dgl')
do marc_each()
set_field('filepath','identifier\.pi')
      prepend('filepath','//zbw/metadaten/Work/ZBW-1-DGW_und_ZDB-23-GBE/upload/')
      append('filepath','.pdf') 
end
end

# DDC and keywords were to be included in the schema, but subsequently it was decided,
# not to fill the fields. Therefore they are created empty as placeholders.
set_field('subject\.ddc','')
set_field('subject\.keyword','')

remove_field('_id')
remove_field('record')