Merge #1381 from branch '1340-adjustAdditionalElementsToConformToSche…

…ma' of https://github.com/hbz/lobid-resources
hbz · Aug 15, 2022 · 127804e · 127804e
2 parents a438bea + fde5ca6
commit 127804e
Show file tree

Hide file tree

Showing 84 changed files with 893 additions and 1,228 deletions.
diff --git a/src/main/resources/alma/alma.fix b/src/main/resources/alma/alma.fix
@@ -6,17 +6,17 @@ include ("./fix/maps.fix")
 copy_field("001","almaMmsId")
 
 #    <data name="id" source="@almaMmsId">
-#      <regexp match="(.*)" format="https://lobid.org/resources/${1}#!"/>
+#      <regexp match="(.*)" format="http://lobid.org/resources/${1}#!"/>
 #    </data>
 #  </rules>
 
-paste("id", "~https://lobid.org/resources/", "001", "~#!", join_char: "")
+paste("id", "~http://lobid.org/resources/", "001", "~#!", join_char: "")
 
 #    <data name="\@context" source="@almaMmsId">
 #      <constant value="https://lobid.org/resources/context.jsonld"/>
 #    </data>
 
-add_field("@context","https://lobid.org/resources/context.jsonld")
+add_field("@context","http://lobid.org/resources/context.jsonld")
 
 include ("./fix/titleRelatedFields.fix")
 include ("./fix/describedBy.fix")
@@ -29,38 +29,38 @@ include ("./fix/contribution.fix")
 
 vacuum()
 retain(
-	"@context",
-	"abstract[]",
-	"almaMmsId",
-	"alternativeTitle[]",
-	"describedBy",
-	"contribution[]",
-	"doi[]",
-	"edition[]",
-	"hasItem[]",
-	"hbzId",
-	"id",
-	"inCollection[]",
-	"isbn[]",
-	"issn[]",
-	"ismn[]",
-	"isPartOf[]",
-	"language[]",
-	"medium[]",
-	"natureOfContent[]",
-	"note[]",
-	"oclcNumber[]",
-	"otherTitleInformation[]",
-	"publication[]",
-	"related[]",
-	"responsibilityStatement[]",
-	"sameAs[]",
-	"subject[]",
-	"subjectAltLabel[]",
-	"tableOfContents[]",
-	"thesisInformation[]",
-	"title",
-	"titleOfSubSeries",
-	"type[]",
-	"zdbId"
-	)
+  "@context",
+  "abstract[]",
+  "almaMmsId",
+  "alternativeTitle[]",
+  "describedBy",
+  "contribution[]",
+  "doi[]",
+  "edition[]",
+  "hasItem[]",
+  "hbzId",
+  "id",
+  "inCollection[]",
+  "isbn[]",
+  "issn[]",
+  "ismn[]",
+  "isPartOf[]",
+  "language[]",
+  "medium[]",
+  "natureOfContent[]",
+  "note[]",
+  "oclcNumber[]",
+  "otherTitleInformation[]",
+  "publication[]",
+  "related[]",
+  "responsibilityStatement[]",
+  "sameAs[]",
+  "subject[]",
+  "subjectAltLabel[]",
+  "tableOfContents[]",
+  "thesisInformation[]",
+  "title",
+  "titleOfSubSeries",
+  "type[]",
+  "zdbId"
+  )
diff --git a/src/main/resources/alma/fix/contribution.fix b/src/main/resources/alma/fix/contribution.fix
@@ -99,7 +99,7 @@ do list(path:"100[01] ", "var":"$i")
       # type
       set_array("contribution[].$last.agent.type[]","Person")
       # role
-      copy_field("$j","contribution[].$last.agent.role.id")
+      copy_field("$j","contribution[].$last.role.id")
     end
   end
 end
@@ -128,7 +128,7 @@ do list(path:"700[01] ", "var":"$i")
       # type
       set_array("contribution[].$last.agent.type[]","Person")
       # role
-      copy_field("$j","contribution[].$last.agent.role.id")
+      copy_field("$j","contribution[].$last.role.id")
     end
   end
 end
@@ -201,7 +201,7 @@ do list(path:"110[012] ", "var":"$i")
       # type
       set_array("contribution[].$last.agent.type[]","CorporateBody")
       # role
-      copy_field("$j","contribution[].$last.agent.role.id")
+      copy_field("$j","contribution[].$last.role.id")
     end
   end
 end
@@ -231,7 +231,7 @@ do list(path:"710[012] ", "var":"$i")
       # type
       set_array("contribution[].$last.agent.type[]","CorporateBody")
       # role
-      copy_field("$j","contribution[].$last.agent.role.id")
+      copy_field("$j","contribution[].$last.role.id")
     end
   end
 end
@@ -240,6 +240,6 @@ end
 set_array("contribution[].*.type[]", "Contribution")
 replace_all("contribution[].*.agent.id","^\\(DE-588\\)(.*$)","https://d-nb.info/gnd/$1")
 replace_all("contribution[].*.agent.label","(?<!\\p{Upper})\\.$|[,]$","")
-prepend("contribution[].*.agent.role.id","http://id.loc.gov/vocabulary/relators/")
+prepend("contribution[].*.role.id","http://id.loc.gov/vocabulary/relators/")
 
 # Deleted the "conference"-Element TODO: specific conference Info needs to be added elsewhere.
diff --git a/src/main/resources/alma/fix/describedBy.fix b/src/main/resources/alma/fix/describedBy.fix
@@ -163,6 +163,7 @@ add_field("describedBy.license[].$last.label","Creative Commons-Lizenz CC0 1.0 U
 # 040 - Cataloging Source (NR) - Subfield: $a (NR), $c (NR), $d (R)
 
 copy_field("040  .a", "describedBy.sourceOrganization.id")
+replace_all("describedBy.sourceOrganization.id", " ", "")
 prepend("describedBy.sourceOrganization.id", "http://lobid.org/organisations/DE-")
 append("describedBy.sourceOrganization.id", "#!")
 if exists("describedBy.sourceOrganization.id")
@@ -180,6 +181,7 @@ end
 
 # TODO: It seems that there are a lot of organisations that are not in lobid, we should filter them out.
 copy_field("040  .c", "describedBy.provider.id")
+replace_all("describedBy.provider.id", " ", "")
 prepend("describedBy.provider.id", "http://lobid.org/organisations/DE-")
 append("describedBy.provider.id", "#!")
 if exists("describedBy.provider.id")
@@ -209,6 +211,7 @@ do list(path:"040  .d", "var":"$i")
 	end
 end
 replace_all("describedBy.modifiedBy.id", "(DE-)?(.*)", "$2")
+replace_all("describedBy.modifiedBy.id", " ", "")
 prepend("describedBy.modifiedBy.id", "http://lobid.org/organisations/DE-")
 append("describedBy.modifiedBy.id", "#!")
 if exists("describedBy.modifiedBy.id")

diff --git a/src/main/resources/alma/fix/fields.fix b/src/main/resources/alma/fix/fields.fix
@@ -804,14 +804,16 @@ end
 #
 
 # 960 ## no Information about repeatability
-
-do list(path:"960??", "var":"$i")
-  do list(path:"$i.?", "var": "$j")
-      copy_field("$j", "inCollection[].$append.id")
-      replace_all("inCollection[].$last.id", "^(.*)$", "http://lobid.org/collections/$1#!")
-      # TODO: Do we need a label? https://github.com/hbz/lobid-resources/issues/1305#issuecomment-912312471, also labels seem wrong.
-  end
-end
+# TODO: This needs further inspection if we need a collection fr all subfields: https://service-wiki.hbz-nrw.de/display/VDBE/960+-+Selektionskennzeichen+NZ
+# Values from r can be invalid.
+
+# do list(path:"960??", "var":"$i")
+#   do list(path:"$i.?", "var": "$j")
+#       copy_field("$j", "inCollection[].$append.id")
+#       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
 
 
 set_array("inCollection[].*.type[]","Collection")
diff --git a/src/main/resources/alma/fix/item.fix b/src/main/resources/alma/fix/item.fix
@@ -108,22 +108,22 @@ end
 #       </combine>
 #     </entity>
 
-# TODO: The morph only took up H52 without indices, but what about 1. inidice 8 ("H528 ") and 2. indice 1 ("H5281")
-do list(path:"H52  ", "var": "$i")
-  if any_match("$i.8", ".*(\\d{4})$")
-    copy_field("$i.8", "hasItem[].$append.id")
-    prepend("hasItem[].$last.id","https://lobid.org/item/")
-    add_field("hasItem[].$last.label", "lobid Bestandsressource")
-    set_array("hasItem[].$last.type[]", "Item", "H52")
-    copy_field("$i.h", "hasItem[].$last.callNumber")
-    paste("hasItem[].$last.currentLocation", "$i.b", "~/", "$i.c")    
-    copy_field("$i.8", "hasItem[].$last.heldBy.id")
-    replace_all("hasItem[].$last.heldBy.id",".*(\\d{4})$","$1")
-    lookup("hasItem[].$last.heldBy.id", "alma-institution-code-to-isil")
-    prepend("hasItem[].$last.heldBy.id", "http://lobid.org/organisations/")
-    append("hasItem[].$last.heldBy.id","#!")
-  end
-end
+# # TODO: The morph only took up H52 without indices, but what about 1. inidice 8 ("H528 ") and 2. indice 1 ("H5281")
+# do list(path:"H52  ", "var": "$i")
+#   if any_match("$i.8", ".*(\\d{4})$")
+#     copy_field("$i.8", "hasItem[].$append.id")
+#     prepend("hasItem[].$last.id","https://lobid.org/item/")
+#     add_field("hasItem[].$last.label", "lobid Bestandsressource")
+#     set_array("hasItem[].$last.type[]", "Item", "H52")
+#     copy_field("$i.h", "hasItem[].$last.callNumber")
+#     paste("hasItem[].$last.currentLocation", "$i.b", "~/", "$i.c")    
+#     copy_field("$i.8", "hasItem[].$last.heldBy.id")
+#     replace_all("hasItem[].$last.heldBy.id",".*(\\d{4})$","$1")
+#     lookup("hasItem[].$last.heldBy.id", "alma-institution-code-to-isil")
+#     prepend("hasItem[].$last.heldBy.id", "http://lobid.org/organisations/")
+#     append("hasItem[].$last.heldBy.id","#!")
+#   end
+# end
 
 # 
 #     <!-- ITM -->

diff --git a/src/test/resources/alma-fix/(CKB)2550000000015245.json b/src/test/resources/alma-fix/(CKB)2550000000015245.json
@@ -1,6 +1,6 @@
 {
   "almaMmsId" : "99370746459806441",
-  "@context" : "https://lobid.org/resources/context.jsonld",
+  "@context" : "http://lobid.org/resources/context.jsonld",
   "title" : "Equivalents of the axiom of choice",
   "alternativeTitle" : [ "Studies in Logic and the Foundations of Mathematics" ],
   "describedBy" : {
@@ -242,23 +242,23 @@
   "contribution" : [ {
     "agent" : {
       "label" : "Rubin, Herman",
-      "type" : [ "Person" ],
-      "role" : {
-        "id" : "http://id.loc.gov/vocabulary/relators/aut",
-        "label" : "Autor/in"
-      }
+      "type" : [ "Person" ]
+    },
+    "role" : {
+      "id" : "http://id.loc.gov/vocabulary/relators/aut",
+      "label" : "Autor/in"
     },
     "type" : [ "Contribution" ]
   }, {
     "agent" : {
       "label" : "Rubin, Jean",
-      "type" : [ "Person" ],
-      "role" : {
-        "id" : "http://id.loc.gov/vocabulary/relators/ctb",
-        "label" : "Beitragende/r"
-      }
+      "type" : [ "Person" ]
+    },
+    "role" : {
+      "id" : "http://id.loc.gov/vocabulary/relators/ctb",
+      "label" : "Beitragende/r"
     },
     "type" : [ "Contribution" ]
   } ],
-  "id" : "https://lobid.org/resources/99370746459806441#!"
+  "id" : "http://lobid.org/resources/99370746459806441#!"
 }
diff --git a/src/test/resources/alma-fix/(CKB)2940000000001369.json b/src/test/resources/alma-fix/(CKB)2940000000001369.json
@@ -1,6 +1,6 @@
 {
   "almaMmsId" : "99370699582506441",
-  "@context" : "https://lobid.org/resources/context.jsonld",
+  "@context" : "http://lobid.org/resources/context.jsonld",
   "title" : "BUW-Output",
   "alternativeTitle" : [ "BUW-Output", "BUW-Output", "Output" ],
   "otherTitleInformation" : [ "Forschungsmagazin = Research bulletin / Universität Wuppertal ; hrsg. im Auftrag des Rektorates" ],
@@ -180,13 +180,13 @@
       "gndIdentifier" : "(DE-588)64944-2",
       "id" : "https://d-nb.info/gnd/64944-2",
       "label" : "Bergische Universität Wuppertal",
-      "type" : [ "CorporateBody" ],
-      "role" : {
-        "id" : "http://id.loc.gov/vocabulary/relators/aut",
-        "label" : "Autor/in"
-      }
+      "type" : [ "CorporateBody" ]
+    },
+    "role" : {
+      "id" : "http://id.loc.gov/vocabulary/relators/aut",
+      "label" : "Autor/in"
     },
     "type" : [ "Contribution" ]
   } ],
-  "id" : "https://lobid.org/resources/99370699582506441#!"
+  "id" : "http://lobid.org/resources/99370699582506441#!"
 }
diff --git a/src/test/resources/alma-fix/(CKB)4100000010077918.json b/src/test/resources/alma-fix/(CKB)4100000010077918.json
@@ -1,6 +1,6 @@
 {
   "almaMmsId" : "99370763433806441",
-  "@context" : "https://lobid.org/resources/context.jsonld",
+  "@context" : "http://lobid.org/resources/context.jsonld",
   "title" : "Uncanny bodies",
   "otherTitleInformation" : [ "superhero comics and disability" ],
   "describedBy" : {
@@ -173,23 +173,23 @@
   "contribution" : [ {
     "agent" : {
       "label" : "Smith, Scott Thompson",
-      "type" : [ "Person" ],
-      "role" : {
-        "id" : "http://id.loc.gov/vocabulary/relators/ctb",
-        "label" : "Beitragende/r"
-      }
+      "type" : [ "Person" ]
+    },
+    "role" : {
+      "id" : "http://id.loc.gov/vocabulary/relators/ctb",
+      "label" : "Beitragende/r"
     },
     "type" : [ "Contribution" ]
   }, {
     "agent" : {
       "label" : "Alaniz, José",
-      "type" : [ "Person" ],
-      "role" : {
-        "id" : "http://id.loc.gov/vocabulary/relators/ctb",
-        "label" : "Beitragende/r"
-      }
+      "type" : [ "Person" ]
+    },
+    "role" : {
+      "id" : "http://id.loc.gov/vocabulary/relators/ctb",
+      "label" : "Beitragende/r"
     },
     "type" : [ "Contribution" ]
   } ],
-  "id" : "https://lobid.org/resources/99370763433806441#!"
+  "id" : "http://lobid.org/resources/99370763433806441#!"
 }
diff --git a/src/test/resources/alma-fix/(CKB)5280000000199164.json b/src/test/resources/alma-fix/(CKB)5280000000199164.json
@@ -1,6 +1,6 @@
 {
   "almaMmsId" : "99370682219806441",
-  "@context" : "https://lobid.org/resources/context.jsonld",
+  "@context" : "http://lobid.org/resources/context.jsonld",
   "title" : "Kirche weltweit",
   "alternativeTitle" : [ "Kirche weltweit", "Mitteilungsblatt des Leipziger Missionswerkes der Evangelisch-Lutherischen Landeskirchen Mecklenburgs, Sachsens, Thüringens" ],
   "otherTitleInformation" : [ "Mitteilungsblatt des Leipziger Missionswerkes der Evangelisch-Lutherischen Landeskirche Sachsens und der Evangelischen Kirche in Mitteldeutschland" ],
@@ -190,13 +190,13 @@
       "gndIdentifier" : "(DE-588)5500538-X",
       "id" : "https://d-nb.info/gnd/5500538-X",
       "label" : "Evangelisch-Lutherisches Missionswerk Leipzig",
-      "type" : [ "CorporateBody" ],
-      "role" : {
-        "id" : "http://id.loc.gov/vocabulary/relators/isb",
-        "label" : "Herausgeber/in"
-      }
+      "type" : [ "CorporateBody" ]
+    },
+    "role" : {
+      "id" : "http://id.loc.gov/vocabulary/relators/isb",
+      "label" : "Herausgeber/in"
     },
     "type" : [ "Contribution" ]
   } ],
-  "id" : "https://lobid.org/resources/99370682219806441#!"
+  "id" : "http://lobid.org/resources/99370682219806441#!"
 }
diff --git a/src/test/resources/alma-fix/(DE-6)1523217-49hbz_ulm.json b/src/test/resources/alma-fix/(DE-6)1523217-49hbz_ulm.json
@@ -1,6 +1,6 @@
 {
   "almaMmsId" : "991000688209706449",
-  "@context" : "https://lobid.org/resources/context.jsonld",
+  "@context" : "http://lobid.org/resources/context.jsonld",
   "titleOfSubSeries" : "Bd. 63",
   "title" : "Bayerische Vorgeschichtsblätter, Bd. 63",
   "otherTitleInformation" : [ "BVbl" ],
@@ -72,5 +72,5 @@
   } ],
   "type" : [ "BibliographicResource", "Book" ],
   "responsibilityStatement" : [ "hrsg. von der Kommission für Bayerische Landesgeschichte bei der Bayerischen Akademie der Wissenschaften in Verbindung mit der Archäologischen Staatssammlung und dem Bayerischen Landesamt für Denkmalpflege" ],
-  "id" : "https://lobid.org/resources/991000688209706449#!"
+  "id" : "http://lobid.org/resources/991000688209706449#!"
 }