Reduce usage of EtikettMaker #2106

src/main/java/org/lobid/resources/run/AlmaMarcXmlFix2lobidJsonEs.java

@@ -139,6 +139,9 @@ public void run() {
                 fixVariables.put("isil2opac_issn.tsv", "../../../../../../lookup-tables/data/opacLinks/isil2opac_issn.tsv");
                 fixVariables.put("isil2opac_zdbId.tsv", "../../../../../../lookup-tables/data/opacLinks/isil2opac_zdbId.tsv");
                 fixVariables.put("isil2opac_almaMmsId.tsv", "../../../../../../lookup-tables/data/opacLinks/isil2opac_almaMmsId.tsv");
+                fixVariables.put("marcRel.tsv", "./maps/marcRel.tsv");
+                fixVariables.put("collectionLabels.tsv", "./maps/collectionLabels.tsv");
+
 
 
                 XmlElementSplitter xmlElementSplitter = new XmlElementSplitter();

src/main/resources/alma/fix/describedBy.fix

@@ -107,16 +107,10 @@ add_array("describedBy.resultOf.object.modifiedBy[]")
 
 end
 
-call_macro("provenanceLinks",field: "describedBy.resultOf.object.sourceOrganization.id")
-copy_field("describedBy.resultOf.object.sourceOrganization.id","describedBy.resultOf.object.sourceOrganization.label")
-lookup("describedBy.resultOf.object.sourceOrganization.label","lobidOrgLabels",delete:"true")
-call_macro("provenanceLinks",field: "describedBy.resultOf.object.provider.id")
-copy_field("describedBy.resultOf.object.provider.id","describedBy.resultOf.object.provider.label")
-lookup("describedBy.resultOf.object.provider.label","lobidOrgLabels",delete:"true")
+call_macro("provenanceLinks",field: "describedBy.resultOf.object.sourceOrganization.id",label: "describedBy.resultOf.object.sourceOrganization.label")
+call_macro("provenanceLinks",field: "describedBy.resultOf.object.provider.id",label: "describedBy.resultOf.object.provider.label")
 do list(path:"describedBy.resultOf.object.modifiedBy[]","var":"$i")
-  call_macro("provenanceLinks",field: "$i.id")
-  copy_field("$i.id","$i.label")
+  call_macro("provenanceLinks",field: "$i.id",label:"$i.label")
 end
-lookup("describedBy.resultOf.object.modifiedBy[].*.label","lobidOrgLabels",delete:"true")
 
 uniq("describedBy.resultOf.object.modifiedBy[]")

src/main/resources/alma/fix/macros.fix

@@ -289,6 +289,13 @@ do put_macro("provenanceLinks")
     end
     prepend("$[field]", "http://lobid.org/organisations/")
     append("$[field]", "#!")
+    copy_field("$[field]","$[label]")
+    lookup("$[label]","lobidOrgLabels",delete:"true")
+    unless exists("$[label]")
+      if any_contain("$[field]","lobid")
+        add_field("$[label]","lobid Organisation")
+      end
+    end
   end
 
 
@@ -560,3 +567,16 @@ do put_macro("manufacture")
     end
   end
 end
+
+
+# lobid resources label
+do put_macro("lobidResourcesFallbackLabel")
+  do list(path:"$[field]","var":"$array")
+    unless exists("$array.label")
+      if any_contain("$array.id","lobid")
+        add_field("$array.label","lobid Ressource")
+      end
+    end
+  end
+end
+

src/main/resources/alma/fix/maps.fix

@@ -55,6 +55,13 @@ put_filemap("$[isil2opac_issn.tsv]","isil2opac_issn", sep_char:"\t")
 put_filemap("$[isil2opac_zdbId.tsv]","isil2opac_zdbId", sep_char:"\t")
 put_filemap("$[isil2opac_almaMmsId.tsv]","isil2opac_almaMmsId", sep_char:"\t")
 
+# marcRel
+put_filemap("$[marcRel.tsv]","marcRel", sep_char:"\t",key_column:"0",value_column:"1",expected_columns:"-1")
+
+# collection labels
+put_filemap("$[collectionLabels.tsv]","collectionLabels", sep_char:"\t",key_column:"0",value_column:"1",expected_columns:"-1")
+
+
 
 put_map("rswk-indicator",
   "p": "Person",
@@ -65,8 +72,22 @@ put_map("rswk-indicator",
   "s": "SubjectHeading"
 )
 
-
-
+put_map("marc-publication-frequency-label",
+  "http://marc21rdf.info/terms/continuingfre#d" : "täglich",
+  "http://marc21rdf.info/terms/continuingfre#i" : "dreimal wöchentlich",
+  "http://marc21rdf.info/terms/continuingfre#c" : "zweimal wöchentlich",
+  "http://marc21rdf.info/terms/continuingfre#w" : "wöchentlich",
+  "http://marc21rdf.info/terms/continuingfre#e" : "vierzehntägig",
+  "http://marc21rdf.info/terms/continuingfre#s" : "halbmonatlich",
+  "http://marc21rdf.info/terms/continuingfre#m" : "monatlich",
+  "http://marc21rdf.info/terms/continuingfre#b" : "alle zwei Monate",
+  "http://marc21rdf.info/terms/continuingfre#q" : "vierteljährlich",
+  "http://marc21rdf.info/terms/continuingfre#f" : "halbjährlich",
+  "http://marc21rdf.info/terms/continuingfre#a" : "jährlich",
+  "http://marc21rdf.info/terms/continuingfre#g" : "alle zwei Jahre",
+  "http://marc21rdf.info/terms/continuingfre#h" : "alle drei Jahre",
+  "http://marc21rdf.info/terms/continuingfre#z" : "unregelmäßig oder sonstige Erscheinungsfrequenz"
+)
 
 put_map("medium-id-to-label",
   "Audio-Dokument":	"http://purl.org/ontology/bibo/AudioDocument",

src/main/resources/alma/fix/relatedRessourcesAndLinks.fix

@@ -72,6 +72,7 @@ end
 replace_all("supplement[].*.id","^\\(DE-605\\)(.*)$","http://lobid.org/resources/$1#!")
 replace_all("supplement[].*.id","^\\(DE-600\\)(.*)$","http://lobid.org/resources/ZDB-$1#!")
 replace_all("supplement[].*.label","<<|>>","")
+call_macro("lobidResourcesFallbackLabel",field:"supplement[]")
 
 # isPartOf
 # it describes the relation between a published ressource and its superordinate series or collection.
@@ -201,10 +202,15 @@ replace_all("isPartOf[].*.hasSuperordinate[].*.id", "^\\(DE-605\\)(.*)$", "http:
 replace_all("isPartOf[].*.hasSuperordinate[].*.id", "^\\(DE-600\\)(.*)$", "http://lobid.org/resources/ZDB-$1#!")
 
 replace_all("isPartOf[].*.numbering", "^[©]|\\s?[,.:;/=]?$", "")
+do list(path:"isPartOf[]","var":"$i")
+  call_macro("lobidResourcesFallbackLabel",field:"$i.hasSuperordinate[]")
+end
+
 uniq("isPartOf[]")
 replace_all("containedIn[].*.id", "^\\(DE-605\\)(.*)$", "http://lobid.org/resources/$1#!")
 replace_all("containedIn[].*.id", "^\\(DE-600\\)(.*)$", "http://lobid.org/resources/ZDB-$1#!")
 replace_all("containedIn[].*.label","<<|>>","")
+call_macro("lobidResourcesFallbackLabel",field:"containedIn[]")
 
 uniq("containedIn[]")
 
@@ -231,6 +237,7 @@ end
 
 replace_all("primaryForm[].*.id", "^\\(DE-605\\)(.*)$", "http://lobid.org/resources/$1#!")
 replace_all("primaryForm[].*.id", "^\\(DE-600\\)(.*)$", "http://lobid.org/resources/ZDB-$1#!")
+call_macro("lobidResourcesFallbackLabel",field:"primaryForm[]")
 
 # secondaryForm
 
@@ -250,6 +257,9 @@ do list(path: "77608", "var":"$i")
   end
 end
 
+call_macro("lobidResourcesFallbackLabel",field:"secondaryForm[]")
+
+
 
 # 856 - Electronic Location and Access (R) - Subfield: $u (R) $3 (NR)
 # 1. Indicator: 4 = HTTP
@@ -406,6 +416,7 @@ end
 replace_all("related[].*.id", "^\\(DE-605\\)(.*)$", "http://lobid.org/resources/$1#!")
 replace_all("related[].*.id", "^\\(DE-600\\)(.*)$", "http://lobid.org/resources/ZDB-$1#!")
 replace_all("related[].*.label","<<|>>","")
+call_macro("lobidResourcesFallbackLabel",field:"related[]")
 
 add_array("inCollection[]")
 
@@ -544,6 +555,9 @@ do list(path:"912  ", "var":"$i")
     replace_all("inCollection[].$last.id", "(ZDB-[0-9]{1,6}-[a-zA-Z|0-9\\-]*).*", "http://lobid.org/organisations/$1#!")
     copy_field("inCollection[].$last.id","$i.@label")
     lookup("$i.@label","lobidOrgLabels",delete:"true")
+    unless exists("$i.@label")
+      add_field("$i.@label","lobid Organisation ZDB Collection")
+    end
     move_field("$i.@label","inCollection[].$last.label")
   end
 end
@@ -557,7 +571,7 @@ do list(path:"962  ", "var":"$i")
   do list(path:"$i.e", "var": "$j")
     unless any_match("$j", "^ZDB.*")
       copy_field("$j", "inCollection[].$append.id")
-      replace_all("inCollection[].$last.id", "^(.*)$", "https://lobid.org/collections#$1")
+      replace_all("inCollection[].$last.id", "^(.*)$", "http://lobid.org/collections#$1")
       # TODO: Do we need a label?
     end
   end
@@ -571,14 +585,20 @@ end
 # do list(path:"960??", "var":"$i")
 #   do list(path:"$i.?", "var": "$j")
 #       copy_field("$j", "inCollection[].$append.id")
-#       replace_all("inCollection[].$last.id", "^(.*)$", "https://lobid.org/collections#$1")
+#       replace_all("inCollection[].$last.id", "^(.*)$", "http://lobid.org/collections#$1")
 #       # TODO: Do we need a label? https://github.com/hbz/lobid-resources/issues/1305#issuecomment-912312471, also labels seem wrong.
 #   end
 # end
 
 
 add_array("inCollection[].*.type[]","Collection")
 
+do list(path:"inCollection[]", "var":"$i")
+  unless exists("$i.label")
+    copy_field("$i.id","$i.label")
+    lookup("$i.label","collectionLabels",delete:"true")
+  end
+end
 
 # predecessor
 
@@ -603,7 +623,7 @@ end
 
 replace_all("predecessor[].*.id", "^\\(DE-605\\)(.*)$", "http://lobid.org/resources/$1#!")
 replace_all("predecessor[].*.id", "^\\(DE-600\\)(.*)$", "http://lobid.org/resources/ZDB-$1#!")
-
+call_macro("lobidResourcesFallbackLabel",field:"predecessor[]")
 
 replace_all("predecessor[].*.label","Vorg. ---> ","")
 

src/main/resources/alma/fix/titleRelatedFields.fix

@@ -216,13 +216,17 @@ if exists("publication[].$first")
       unless any_match("008","^.{18}[#\\| u].*$") # filters out not matching values and also the value unknown
         copy_field("008","publication[].$first.frequency[].$append.id")
         replace_all("publication[].$first.frequency[].$last.id", "^.{18}(.).*$", "http://marc21rdf.info/terms/continuingfre#$1")
+        copy_field("publication[].$first.frequency[].$last.id","publication[].$first.frequency[].$last.label")
+        lookup("publication[].$first.frequency[].$last.label","marc-publication-frequency-label")
       end
     elsif any_match("006","^s.*$")
       do list(path: "006", "var":"$z")
         if any_match("$z","^s.*$")
           unless any_match("$z","^.[#\\| u].*$")
             copy_field("$z","publication[].$first.frequency[].$append.id")
             replace_all("publication[].$first.frequency[].$last.id", "^.(.).*$", "http://marc21rdf.info/terms/continuingfre#$1")
+            copy_field("publication[].$first.frequency[].$last.id","publication[].$first.frequency[].$last.label")
+            lookup("publication[].$first.frequency[].$last.label","marc-publication-frequency-label")
           end
         end
       end

src/main/resources/alma/maps/collectionLabels.tsv

@@ -0,0 +1,117 @@
+http://lobid.org/collections#asmi	American Society for Microbiology /EBooks
+http://lobid.org/collections#budri	Budrich academic / Open Access
+http://lobid.org/collections#caso	California Scholarship Online (University of California Press)
+http://lobid.org/collections#chiso	Chicago Scholarship Online (University of Chicago Press)
+http://lobid.org/collections#Cont	Content-Select
+http://lobid.org/collections#cuvillier	Cuvillier-E-Books
+http://lobid.org/collections#dawsonera	dawsonera
+http://lobid.org/collections#edso	Edinburgh University Press/ Edinburgh Scholarship Online
+http://lobid.org/collections#editlib	EdITLib-E-Books
+http://lobid.org/collections#elgar	Edward Elgar / Elgar-E-Books
+http://lobid.org/collections#lyell	The Geological Society / Lyell Collection
+http://lobid.org/collections#hade	Harri Deutsch
+http://lobid.org/collections#hirzel	Hirzel eLibrary
+http://lobid.org/collections#huguenots	Huguenots-E-Books
+http://lobid.org/collections#iorm	INFORMS
+http://lobid.org/collections#kenso	Kentucky Scholarship Online (The University Press of Kentucky)
+http://lobid.org/collections#learntechlib	LearnTechLib E-Books
+http://lobid.org/collections#Logos	Logos Verlag Berlin e-Book Bibliothek
+http://lobid.org/collections#mansi	Manchester Studies in Imperialism
+http://lobid.org/collections#misso	Mississippi Scholarship Online (University Press of Mississippi)
+http://lobid.org/collections#MPSO	MIT Press Scholarship Online
+http://lobid.org/collections#mpig	Max-Planck-Institut für Gesellschaftsforschung Open Access
+http://lobid.org/collections#NNg	Nielsen Norman Group
+http://lobid.org/collections#obp	OpenBook Publishers-E-Books
+http://lobid.org/collections#oso	Otto Schmidt online
+http://lobid.org/collections#pearson	Pearson-E-Books
+http://lobid.org/collections#philon	Philosophy Online-E-Books
+http://lobid.org/collections#luther	ProQuest / Luther-E-Books
+http://lobid.org/collections#smalib	The Smashing Library for Universities and Enterprises
+http://lobid.org/collections#minnso	University of Minnesota Press / Minnesota Scholarship Online
+http://lobid.org/collections#uncso	The University of North Carolina Press / North Carolina Scholarship Online
+http://lobid.org/collections#vkal	Verlag Karl Alber
+http://lobid.org/collections#vkv	Vittorio Klostermann / Klostermann-E-Books
+http://lobid.org/collections#vogel	Vogel-E-Books
+http://lobid.org/collections#wageningen	Wageningen Academic Publishers-E-Books
+http://lobid.org/collections#woodhead	Woodhead-E-Books
+http://lobid.org/collections#wtm	WTM-Verlag
+http://lobid.org/collections#NLZ	Nationallizenzen
+http://lobid.org/collections#beofamilien	Beck Online / Familienrecht
+http://lobid.org/collections#beosteuer	Beck Online / Steuerrecht
+http://lobid.org/collections#beozivil	Beck Online / Zivilrecht
+http://lobid.org/collections#BeltzLizenz2012	Beltz Lizenzzugehörigkeit 2012
+http://lobid.org/collections#BeltzLizenz2013	Beltz Lizenzzugehörigkeit 2013
+http://lobid.org/collections#BeltzLizenz2014	Beltz Lizenzzugehörigkeit 2014
+http://lobid.org/collections#BeltzLizenz2015	Beltz Lizenzzugehörigkeit 2015
+http://lobid.org/collections#BeltzLizenz2016	Beltz Lizenzzugehörigkeit 2016
+http://lobid.org/collections#BeltzLizenz2017	Beltz Lizenzzugehörigkeit 2017
+http://lobid.org/collections#BeltzLizenz2018	Beltz Lizenzzugehörigkeit 2018
+http://lobid.org/collections#BeltzLizenz2019	Beltz Lizenzzugehörigkeit 2019
+http://lobid.org/collections#BeltzLizenz2020	Beltz Lizenzzugehörigkeit 2020
+http://lobid.org/collections#BeltzLizenz2021	Beltz Lizenzzugehörigkeit 2021
+http://lobid.org/collections#bloomsbury
+http://lobid.org/collections#bloomsbury2013	Bloomsbury-E-Books
+http://lobid.org/collections#bloomsbury2014	Bloomsbury Lizenzzugehörigkeit 2014
+http://lobid.org/collections#bloomsbury2015	Bloomsbury Lizenzzugehörigkeit 2015
+http://lobid.org/collections#bloomsbury2016	Bloomsbury Lizenzzugehörigkeit 2016
+http://lobid.org/collections#bloomsbury2017	Bloomsbury Lizenzzugehörigkeit 2017
+http://lobid.org/collections#melanchthon	Brill Academic / Melanchthon-E-Books
+http://lobid.org/collections#Lizenz2001	de Gruyter-E-Books Lizenzzugehörigkeit 2001
+http://lobid.org/collections#Lizenz2002	de Gruyter-E-Books Lizenzzugehörigkeit 2002
+http://lobid.org/collections#Lizenz2003	de Gruyter-E-Books Lizenzzugehörigkeit 2003
+http://lobid.org/collections#Lizenz2004	de Gruyter-E-Books Lizenzzugehörigkeit 2004
+http://lobid.org/collections#Lizenz2005	de Gruyter-E-Books Lizenzzugehörigkeit 2005
+http://lobid.org/collections#Lizenz2006	de Gruyter-E-Books Lizenzzugehörigkeit 2006
+http://lobid.org/collections#Lizenz2007	de Gruyter-E-Books Lizenzzugehörigkeit 2007
+http://lobid.org/collections#Lizenz2008	de Gruyter-E-Books Lizenzzugehörigkeit 2008
+http://lobid.org/collections#Lizenz2009	de Gruyter-E-Books Lizenzzugehörigkeit 2009
+http://lobid.org/collections#Lizenz2010	de Gruyter-E-Books Lizenzzugehörigkeit 2010
+http://lobid.org/collections#Lizenz2011	de Gruyter-E-Books Lizenzzugehörigkeit 2011
+http://lobid.org/collections#Lizenz2012	de Gruyter-E-Books Lizenzzugehörigkeit 2012
+http://lobid.org/collections#Lizenz2013	de Gruyter-E-Books Lizenzzugehörigkeit 2013
+http://lobid.org/collections#Lizenz2014	de Gruyter-E-Books Lizenzzugehörigkeit 2014
+http://lobid.org/collections#Lizenz2015	de Gruyter-E-Books Lizenzzugehörigkeit 2015
+http://lobid.org/collections#Lizenz2016	de Gruyter-E-Books Lizenzzugehörigkeit 2016
+http://lobid.org/collections#Lizenz2017	de Gruyter-E-Books Lizenzzugehörigkeit 2017
+http://lobid.org/collections#Lizenz2018	de Gruyter-E-Books Lizenzzugehörigkeit 2018
+http://lobid.org/collections#Lizenz2019	de Gruyter-E-Books Lizenzzugehörigkeit 2019
+http://lobid.org/collections#Lizenz2020	de Gruyter-E-Books Lizenzzugehörigkeit 2020
+http://lobid.org/collections#herder	Herder-E-Books
+http://lobid.org/collections#juris	Juris-E-Books
+http://lobid.org/collections#KohlhammerLizenz2012	Kohlhammer Lizenzzugehörigkeit 2012
+http://lobid.org/collections#KohlhammerLizenz2013	Kohlhammer Lizenzzugehörigkeit 2013
+http://lobid.org/collections#KohlhammerLizenz2014	Kohlhammer Lizenzzugehörigkeit 2014
+http://lobid.org/collections#KohlhammerLizenz2015	Kohlhammer Lizenzzugehörigkeit 2015
+http://lobid.org/collections#KohlhammerLizenz2016	Kohlhammer Lizenzzugehörigkeit 2016
+http://lobid.org/collections#KohlhammerLizenz2017	Kohlhammer Lizenzzugehörigkeit 2017
+http://lobid.org/collections#KohlhammerLizenz2018	Kohlhammer Lizenzzugehörigkeit 2018
+http://lobid.org/collections#KohlhammerLizenz2019	Kohlhammer Lizenzzugehörigkeit 2019
+http://lobid.org/collections#KohlhammerLizenz2020	Kohlhammer Lizenzzugehörigkeit 2020
+http://lobid.org/collections#logos	Logos Verlag Berlin
+http://lobid.org/collections#MohrSiebeckLizenz2013-2015	MohrSiebeck Lizenzzugehörigkeit 2013-2015
+http://lobid.org/collections#MohrSiebeckLizenz2013	MohrSiebeck Lizenzzugehörigkeit 2013
+http://lobid.org/collections#MohrSiebeckLizenz2014	MohrSiebeck Lizenzzugehörigkeit 2014
+http://lobid.org/collections#MohrSiebeckLizenz2015	MohrSiebeck Lizenzzugehörigkeit 2015
+http://lobid.org/collections#MohrSiebeckLizenz2016	MohrSiebeck Lizenzzugehörigkeit 2016
+http://lobid.org/collections#MohrSiebeckLizenz2017	MohrSiebeck Lizenzzugehörigkeit 2017
+http://lobid.org/collections#MohrSiebeckLizenz2018	MohrSiebeck Lizenzzugehörigkeit 2018
+http://lobid.org/collections#MohrSiebeckLizenz2019	MohrSiebeck Lizenzzugehörigkeit 2019
+http://lobid.org/collections#synthesis lectures	Morgan & Claypool / Synthesis Lectures-Reihen
+http://lobid.org/collections#thieref	Thieme eRef
+http://lobid.org/collections#taylor francis	Taylor & Francis E-Books
+http://lobid.org/collections#V&RELibraryLizenz2014	V&R eLibrary Lizenzzugehörigkeit 2014
+http://lobid.org/collections#V&RELibraryLizenz2015	V&R eLibrary Lizenzzugehörigkeit 2015
+http://lobid.org/collections#V&RELibraryLizenz2016	V&R eLibrary Lizenzzugehörigkeit 2016
+http://lobid.org/collections#V&RELibraryLizenz2017	V&R eLibrary Lizenzzugehörigkeit 2017
+http://lobid.org/collections#V&RELibraryLizenz2018	V&R eLibrary Lizenzzugehörigkeit 2018
+http://lobid.org/collections#V&RELibraryLizenz2019	V&R eLibrary Lizenzzugehörigkeit 2019
+http://lobid.org/collections#wbv	W. Bertelsmann Verlag E-Books
+http://lobid.org/collections#WallsteinLizenz2018	Wallstein Lizenzzugehörigkeit 2018
+http://lobid.org/collections#WallsteinLizenz2019	Wallstein Lizenzzugehörigkeit 2019
+http://lobid.org/collections#WBGLizenz2016	WBG Lizenzzugehörigkeit 2016
+http://lobid.org/collections#WBGLizenz2017	WBG Lizenzzugehörigkeit 2017
+http://lobid.org/collections#WBGLizenz2018	WBG Lizenzzugehörigkeit 2018
+http://lobid.org/collections#WBGLizenz2019	WBG Lizenzzugehörigkeit 2019
+http://lobid.org/collections#WBGLizenz2020	WBG Lizenzzugehörigkeit 2020
+http://lobid.org/collections#TTP-MCE	Trans Tech Publikations: Materials Science and Engineering
+http://lobid.org/collections#palgraveoa	Palgrave Open Access E-Books