Update transformation to current lobid fix RPB-225

hbz · Feb 28, 2025 · bb13c82 · bb13c82
1 parent deedcd5
commit bb13c82
Show file tree

Hide file tree

Showing 6 changed files with 53 additions and 20 deletions.
diff --git a/conf/hebisMarc2lobid-transformation/fix/describedBy.fix b/conf/hebisMarc2lobid-transformation/fix/describedBy.fix
@@ -41,9 +41,9 @@ nothing() # currently no transformation for describedBy is needed.
 #   substring("@initialCataloguingDate","0","6")
 # end
 # 
-# if any_match("@initialCataloguingDate","^[0-4]\\d*") # Complete dates after 2000
+# if any_match("@initialCataloguingDate","^[0-4]\\d(0[1-9]|1[012])(0[1-9]|[12][0-9]|3[01])") # Assume dates from 2000-01-01 to  2049-12-31 ( e.g. matching 491231)
 #   prepend("@initialCataloguingDate","20")
-# elsif any_match("@initialCataloguingDate","\\d*") # Complete dates before 2000
+# elsif any_match("@initialCataloguingDate","\\d{2}(0[1-9]|1[012])(0[1-9]|[12][0-9]|3[01])") # Assume dates from 1900-01-01 to  1999-12-31 ( e.g. matching 991231)
 #   prepend("@initialCataloguingDate","19")
 # else
 #   copy_field("MNG  .b","@initialCataloguingDate")
@@ -67,16 +67,18 @@ nothing() # currently no transformation for describedBy is needed.
 # replace_all("describedBy.resultOf.object.dateModified","^(\\d{4})(\\d{2})(\\d{2})$","$1-$2-$3")
 # replace_all("describedBy.resultOf.object.dateCreated","^(\\d{4})$","$1-01-01")
 # replace_all("describedBy.resultOf.object.dateModified","^(\\d{4})$","$1-01-01")
+# call_macro("leapYearChecker",date:"describedBy.resultOf.object.dateCreated")
+# call_macro("leapYearChecker",date:"describedBy.resultOf.object.dateModified")
 # 
 # add_array("describedBy.resultOf.object.type[]", "DataFeedItem")
 # 
 # copy_field("almaMmsId","describedBy.resultOf.object.label")
 # prepend("describedBy.resultOf.object.label","hbz-Ressource ")
 # append("describedBy.resultOf.object.label"," im Exportformat MARC21 XML")
 # 
-# add_field("describedBy.resultOf.object.inDataset.id", "http://sru.hebis.de/sru/DB=2.1?version=1.1")
+# add_field("describedBy.resultOf.object.inDataset.id", "https://datahub.io/dataset/hbz_unioncatalog")
 # 
-# add_field("describedBy.resultOf.object.inDataset.label", "Hebis SRU")
+# add_field("describedBy.resultOf.object.inDataset.label", "hbz_unioncatalog")
 # 
 # add_array("describedBy.license[]")
 # add_field("describedBy.license[].$append.id","http://creativecommons.org/publicdomain/zero/1.0" )
@@ -115,4 +117,3 @@ nothing() # currently no transformation for describedBy is needed.
 # end
 # 
 # uniq("describedBy.resultOf.object.modifiedBy[]")
-# 
diff --git a/conf/hebisMarc2lobid-transformation/fix/macros.fix b/conf/hebisMarc2lobid-transformation/fix/macros.fix
@@ -585,3 +585,22 @@ do put_macro("lobidResourcesFallbackLabel")
   end
 end
 
+# validate leap years
+do put_macro("leapYearChecker")
+  if any_match("$[date]","....-02-29")
+    unless any_match("$[date]","(((18|19|20)(04|08|[2468][048]|[13579][26]))|2000)-02-29")
+      replace_all("$[date]","(....-02)-29","$1-28")
+    end
+  end
+end
+
+
+# DE Sol1 Holding Tester excludes ZDB Records
+
+do put_macro("deSol1BridgeTester")
+  if any_match("$[holdingId]",".*7830$")
+    unless exists("zdbId")
+      add_field("$i.deSol1Bridge","true")
+    end
+  end
+end
diff --git a/conf/hebisMarc2lobid-transformation/fix/otherFields.fix b/conf/hebisMarc2lobid-transformation/fix/otherFields.fix
@@ -112,7 +112,9 @@ replace_all("extent", "  ", " ")
 # 500 - General Note (R) Subfield: $a (NR)
 add_array("note[]")
 do list(path:"500  ", "var": "$i")
-  copy_field("$i.a", "note[].$append")
+  unless any_contain("$i.a","In:")
+    copy_field("$i.a", "note[].$append")
+  end
 end
 uniq("note[]")
 
@@ -130,6 +132,9 @@ do list(path:"520[ 23] ", "var": "$i")
   copy_field("$i.[ab]", "abstract[].$append")
 end
 
+replace_all("abstract[].*","^<!\\[CDATA\\[(.*)\\]\\]>$","$1")
+replace_all("abstract[].*","<[\\/]?.{1,2}>","")
+
 # 502 - Dissertation Note (R) Subfield: $a (R)
 add_array("thesisInformation[]")
 do list(path:"502  ", "var": "$i")

diff --git a/conf/hebisMarc2lobid-transformation/fix/relatedRessourcesAndLinks.fix b/conf/hebisMarc2lobid-transformation/fix/relatedRessourcesAndLinks.fix
@@ -86,7 +86,7 @@ unless any_match("leader", "^.{7}[ad].*")
   do list(path: "773??", "var": "$i")
     unless any_equal("$i.9","LOCAL")
       do list(path: "$i.w", "var": "$j")
-        add_array("isPartOf[].$append.type[]", "IsPartOfRelation")
+        add_hash("isPartOf[].$append")
         add_array("isPartOf[].$last.hasSuperordinate[]")
         add_hash( "isPartOf[].$last.hasSuperordinate[].$append")
         if all_match("$j", "^\\((?:DE-600|DE-605)\\)(.*)$")
@@ -108,7 +108,7 @@ end
 # 490 with 1. Indicator 1 has an identical entry in 830. So only 490 with 1. Indicator 0
 
 do list(path: "4900?", "var": "$i")
-  add_array("isPartOf[].$append.type[]", "IsPartOfRelation")
+  add_hash("isPartOf[].$append")
   add_array("isPartOf[].$last.hasSuperordinate[]")
   add_hash( "isPartOf[].$last.hasSuperordinate[].$append")
   add_array("isPartOf[].$last.hasSuperordinate[].$last.label")
@@ -130,18 +130,19 @@ end
 # Element can be repeatable with local entries they have subfield $M.
 
 do list(path: "830??", "var": "$i")
-  add_array("isPartOf[].$append.type[]", "IsPartOfRelation")
+  add_hash("isPartOf[].$append")
   add_array("isPartOf[].$last.hasSuperordinate[]")
   add_hash( "isPartOf[].$last.hasSuperordinate[].$append")
   if all_match("$i.w", "^\\((?:DE-600|DE-605)\\)(.*)$")
     copy_field("$i.w", "isPartOf[].$last.hasSuperordinate[].$last.id")
   end
-  add_array("isPartOf[].$last.hasSuperordinate[].$last.label")
   do list(path:"$i.a", "var":"$j")
     copy_field("$j", "isPartOf[].$last.hasSuperordinate[].$last.label.$append")
   end
   join_field("isPartOf[].$last.hasSuperordinate[].$last.label", " / ")
-  copy_field("$i.v", "isPartOf[].$last.numbering")
+  unless is_empty("isPartOf[].$last.hasSuperordinate[].1")
+    copy_field("$i.v", "isPartOf[].$last.numbering")
+  end
 end
 
 do list(path: "4901?", "var": "$j")
@@ -164,7 +165,7 @@ if any_match("leader", "^.{7}[ad].*")
     do list(path: "773??", "var": "$i")
       unless any_equal("$i.9","LOCAL")
         do list(path: "$i.w", "var": "$j")
-          add_array("isPartOf[].$append.type[]", "IsPartOfRelation")
+          add_hash("isPartOf[].$append")
           add_array("isPartOf[].$last.hasSuperordinate[]")
           add_hash( "isPartOf[].$last.hasSuperordinate[].$append")
           if all_match("$j", "^\\((?:DE-600|DE-605)\\)(.*)$")
@@ -192,21 +193,26 @@ if any_match("leader", "^.{7}[ad].*")
   end
 end
 
-do list(path: "isPartOf[].*.hasSuperordinate[]", "var": "$i")
-  unless exists("$i.label")
-    copy_field("@title", "$i.label")
+do list(path: "isPartOf[]","var":"$i")
+  unless is_empty("$i.hasSuperordinate[].1")
+    do list(path:"$i.hasSuperordinate[]", "var": "$j") ## This is the fallback for isPartOf[].*.hasSuperordinate[].*.label
+      unless exists("$j.label")
+        copy_field("@title", "$j.label")
+      end
+    end
+    add_array("$i.type[]", "IsPartOfRelation")
   end
 end
 
 replace_all("isPartOf[].*.hasSuperordinate[].*.id", "^\\(DE-605\\)(.*)$", "http://lobid.org/resources/$1#!")
 replace_all("isPartOf[].*.hasSuperordinate[].*.id", "^\\(DE-600\\)(.*)$", "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","<<|>>","")

diff --git a/conf/hebisMarc2lobid-transformation/fix/titleRelatedFields.fix b/conf/hebisMarc2lobid-transformation/fix/titleRelatedFields.fix
@@ -326,6 +326,8 @@ do list(path: "publication[]", "var": "$i")
   replace_all("$i.location[].*", "^\\[(.*)\\]$", "$1")
   replace_all("$i.location[].*", "\\s?[,:;]$", "")
   replace_all("$i.publishedBy[].*", "^[©]|\\s?[,:;/=]?$", "")
+  call_macro("leapYearChecker",date:"$i.startDate")
+  call_macro("leapYearChecker",date:"$i.endDate")
   uniq("$i.location[]")
 end
 

diff --git a/conf/output/test-hebis-to-lobid-output-9.json b/conf/output/test-hebis-to-lobid-output-9.json
@@ -21,11 +21,11 @@
     "label" : "OCLC Ressource"
   } ],
   "isPartOf" : [ {
-    "type" : [ "IsPartOfRelation" ],
     "hasSuperordinate" : [ {
       "label" : "Beiträge zur Geschichte des Gau-Algesheimer Raumes"
     } ],
-    "numbering" : "42"
+    "numbering" : "42",
+    "type" : [ "IsPartOfRelation" ]
   } ],
   "language" : [ {
     "id" : "http://id.loc.gov/vocabulary/iso639-2/ger",