Merge #1834 from remote-tracking branch 'origin/1830-publicationImpro…

…vement' into updateTestsAfterEndOfWave3

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

@@ -324,4 +324,58 @@ do put_macro("subjectLabel")
   copy_field("$i.z","subject[].$last.label.$append")
   copy_field("$i.v","subject[].$last.label.$append")
   join_field("subject[].$last.label"," / ")
+end
+
+do put_macro("publication")
+  do list(path:"$[field]", "var":"$i")
+    add_field("publication[].$append.test","")
+    do list(path: "$i.c", "var":"$j")
+      replace_all("$j", "\\[|\\]|ca. |c ", "")
+      unless exists("publication[].$last.startDate")
+        if any_match("$j",".*?([01]\\d{3}|20\\d{2}).*")
+          paste("publication[].$last.startDate", "$j")
+        end
+      end
+      unless exists("publication[].$last.endDate")
+        if any_match("$j",".*-[ ]?([01]\\d{3}|20\\d{2})$")
+          paste("publication[].$last.endDate", "$j")
+        end
+      end
+    end
+    set_array("publication[].$last.type[]","PublicationEvent")
+    set_array("publication[].$last.location[]")
+    copy_field("$i.a", "publication[].$last.location[].$append")
+    set_array("publication[].$last.publishedBy[]")
+    copy_field("$i.b", "publication[].$last.publishedBy[].$append")
+    if exists("$i.6")
+      copy_field("$i.6","$i.linkageTest")
+      do list(path:"880??","var":"$880")
+        if in ("$i.linkageTest","$880.linkageTest")
+          if in ("$880.@script.id","alternateGraphicRepresentation[].*.script.id")   
+            do list(path:"alternateGraphicRepresentation[]","var":"$AGR")
+              if in ("$880.@script.id","$AGR.script.id")
+                unless exists("$AGR.record.publication[]")
+                  set_array("$AGR.record.publication[]")
+                end                
+                add_field ("$AGR.record.publication[].$append.dummi","")       
+                set_array("$AGR.record.publication[].$last.location[]")        
+                copy_field("$880.a", "$AGR.record.publication[].$last.location[].$append")
+                set_array("$AGR.record.publication[].$last.publishedBy[]")
+                copy_field("$880.b", "$AGR.record.publication[].$last.publishedBy[].$append")
+              end
+            end
+          else    
+            copy_field("$880.@script.id","alternateGraphicRepresentation[].$append.script.id")
+            copy_field("$880.@script.label","alternateGraphicRepresentation[].$last.script.label")
+            set_array("alternateGraphicRepresentation[].$last.publication[]")
+            add_field ("alternateGraphicRepresentation[].$last.publication[].$append.dummi","")       
+            set_array("alternateGraphicRepresentation[].$last.publication[].$last.location[]")        
+            copy_field("$880.a", "alternateGraphicRepresentation[].$last.publication[].$last.location[].$append")
+            set_array("alternateGraphicRepresentation[].$last.publication[].$last.publishedBy[]")
+            copy_field("$880.b", "alternateGraphicRepresentation[].$last.publication[].$last.publishedBy[].$append")
+          end
+        end
+      end
+    end
+  end
 end

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

@@ -221,86 +221,85 @@ end
 # 264 - Production, Publication, Distribution, Manufacture, and Copyright Notice (R) -  Subfield: $a (R), $b (R), $c  (R)
 # 008,18 for frequency
 # 515 - Numbering Peculiarities Note (R)
+# Prefer 264 over 260 since it can create duplicate info and 260 is discontinued in RDA.
 
 set_array("publication[]")
-do list(path:"260[ 23] |264[ 23][ 1]", "var":"$i")
-  add_field("publication[].$append.test","")
-  copy_field("362??.a","publication[].$last.publicationHistory")
-  do list(path: "$i.c", "var":"$j")
-    replace_all("$j", "\\[|\\]|ca. |c ", "")
-    unless exists("publication[].$last.startDate")
-      if any_match("$j",".*?([01]\\d{3}|20\\d{2}).*")
-        paste("publication[].$last.startDate", "$j")
+if exists("264[ 23][ 1]")
+  call_macro("publication",field:"2643[ 1]") # 3 - Current/Latest
+  call_macro("publication",field:"2642[ 1]") # 2 - Intervening
+  call_macro("publication",field:"264 [ 1]") # # - Not applicable/No information provided/Earliest
+else
+  call_macro("publication",field:"2603[ 1]") # 3 - Current/Latest
+  call_macro("publication",field:"2602[ 1]") # 2 - Intervening
+  call_macro("publication",field:"260 [ 1]") # # - Not applicable/No information provided/Earliest
+end
+
+# Only add additional publication info to the first publication-object since it is the latest.
+
+if exists("publication[].$first")
+  copy_field("362??.a","publication[].$first.publicationHistory")
+    set_array("publication[].$first.frequency[]")
+    if any_match("leader","^.{6}(a[bis]|m[bis]).*$") # checks if continous ressource
+      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#$1")
       end
-    end
-    unless exists("publication[].$last.endDate")
-      if any_match("$j",".*-[ ]?([01]\\d{3}|20\\d{2})$")
-        paste("publication[].$last.endDate", "$j")
+    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#$1")
+          end
+        end
       end
     end
-  end
-  # TODO is there a way to distinguish PublicationEvent and SecondaryPublicationEvent?
-  set_array("publication[].$last.type[]","PublicationEvent")
-  set_array("publication[].$last.location[]")
-  copy_field("$i.a", "publication[].$last.location[].$append")
-  set_array("publication[].$last.publishedBy[]")
-  copy_field("$i.b", "publication[].$last.publishedBy[].$append")
-  set_array("publication[].$last.frequency[]")
-  if any_match("leader","^.{6}(a[bis]|m[bis]).*$") # checks if continous ressource
-    unless any_match("008","^.{18}[#\\| u].*$") # filters out not matching values and also the value unknown
-      copy_field("008","publication[].$last.frequency[].$append.id")
-      replace_all("publication[].$last.frequency[].$last.id", "^.{18}(.).*$", "http://marc21rdf.info/terms/continuingfre#$1")
+    set_array("publication[].$first.note[]")
+    copy_field("515??.a","publication[].$first.note[].$append")
+  do list(path:"500  ", "var":"$i")
+    if any_match("$i.a", "^.*saṃ. \\d{4}=(\\d{4}).*Chr.*")
+      remove_field("publication[].$first.startDate")
+      copy_field("$i.a","publication[].$first.startDate")
+      replace_all("publication[].$first.startDate","^.*saṃ. \\d{4}=(\\d{4}).*Chr.*","$1")
     end
-  elsif any_match("006","^s.*$")
-    do list(path: "006", "var":"$z")
-      if any_match("$z","^s.*$")
-        unless any_match("$z","^.[#\\| u].*$")
-          copy_field("$z","publication[].$last.frequency[].$append.id")
-          replace_all("publication[].$last.frequency[].$last.id", "^.(.).*$", "http://marc21rdf.info/terms/continuingfre#$1")
+  end
+  # Add fallbacks for missing publication dates and other publication info.
+  do list(path:"260[ 3][ 1]", "var":"$i")
+    do list(path: "$i.c", "var":"$j")
+      replace_all("$j", "\\[|\\]|ca. |c ", "")
+      unless exists("publication[].$first.startDate")
+        if any_match("$j",".*?([01]\\d{3}|20\\d{2}).*")
+          paste("publication[].$first.startDate", "$j")
         end
       end
-    end
-  end
-  set_array("publication[].$last.note[]")
-  copy_field("515??.a","publication[].$last.note[].$append")
-  if exists("$i.6")
-    copy_field("$i.6","$i.linkageTest")
-    do list(path:"880??","var":"$880")
-      if in ("$i.linkageTest","$880.linkageTest")
-        if in ("$880.@script.id","alternateGraphicRepresentation[].*.script.id")   
-          do list(path:"alternateGraphicRepresentation[]","var":"$AGR")
-            if in ("$880.@script.id","$AGR.script.id")
-              unless exists("$AGR.record.publication[]")
-                set_array("$AGR.record.publication[]")
-              end                
-              add_field ("$AGR.record.publication[].$append.dummi","")       
-              set_array("$AGR.record.publication[].$last.location[]")        
-              copy_field("$880.a", "$AGR.record.publication[].$last.location[].$append")
-              set_array("$AGR.record.publication[].$last.publishedBy[]")
-              copy_field("$880.b", "$AGR.record.publication[].$last.publishedBy[].$append")
-            end
-          end
-        else    
-          copy_field("$880.@script.id","alternateGraphicRepresentation[].$append.script.id")
-          copy_field("$880.@script.label","alternateGraphicRepresentation[].$last.script.label")
-          set_array("alternateGraphicRepresentation[].$last.publication[]")
-          add_field ("alternateGraphicRepresentation[].$last.publication[].$append.dummi","")       
-          set_array("alternateGraphicRepresentation[].$last.publication[].$last.location[]")        
-          copy_field("$880.a", "alternateGraphicRepresentation[].$last.publication[].$last.location[].$append")
-          set_array("alternateGraphicRepresentation[].$last.publication[].$last.publishedBy[]")
-          copy_field("$880.b", "alternateGraphicRepresentation[].$last.publication[].$last.publishedBy[].$append")
+      unless exists("publication[].$first.endDate")
+        if any_match("$j",".*-[ ]?([01]\\d{3}|20\\d{2})$")
+          paste("publication[].$last.endDate", "$j")
         end
       end
+      unless exists("publication[].$first.location[].1")
+        copy_field("$i.a", "publication[].$first.location[].$append")
+      end
+      unless exists("publication[].$first.publishedBy[].1")
+        copy_field("$i.b", "publication[].$first.publishedBy[].$append")
+      end
     end
   end
-end
-
-
-do list(path:"500  ", "var":"$i")
-  if any_match("$i.a", "^.*saṃ. \\d{4}=(\\d{4}).*Chr.*")
-    remove_field("publication[].$last.startDate")
-    copy_field("$i.a","publication[].$last.startDate")
-    replace_all("publication[].$last.startDate","^.*saṃ. \\d{4}=(\\d{4}).*Chr.*","$1")
+  unless exists("publication[].$first.startDate[]")
+    if any_match("008","^.{6}[brestikm](\\d{4}).*$")
+      copy_field("008","@008startDate")
+      replace_all("@008startDate","^.{7}(\\d{4}).*$","$1")
+      copy_field("@008startDate","publication[].$first.startDate")
+    end
+  end
+  unless exists("publication[].$first.endDate[]")
+    if any_match("008","^.{6}[km]\\d{4}(\\d{4}).*$")
+      copy_field("008","@008endDate")
+      replace_all("@008endDate","^.{11}(\\d{4}).*$","$1")
+      unless any_equal("@008endDate","9999")
+        copy_field("@008endDate","publication[].$first.endDate")
+      end
+    end
   end
 end
 

src/test/resources/alma-fix/990052965140206441.json

@@ -6,10 +6,10 @@
   "oclcNumber" : [ "183288969" ],
   "title" : "Zweckverband Naturpark Kottenforst-Ville",
   "publication" : [ {
-    "publicationHistory" : "1.1985 - 4.2001; damit Ersch. eingest.",
     "type" : [ "PublicationEvent" ],
     "location" : [ "Köln [i.e.] Pulheim" ],
-    "publishedBy" : [ "Rheinland-Verl." ]
+    "publishedBy" : [ "Rheinland-Verl." ],
+    "publicationHistory" : "1.1985 - 4.2001; damit Ersch. eingest."
   } ],
   "describedBy" : {
     "id" : "http://lobid.org/resources/990052965140206441",

src/test/resources/alma-fix/990053976760206441.json

@@ -9,25 +9,20 @@
   "hbzId" : "HT000312236",
   "title" : "Physik in der Schule",
   "publication" : [ {
-    "publicationHistory" : "2.1964,7 - 38.2000",
     "startDate" : "1964",
     "endDate" : "2000",
     "type" : [ "PublicationEvent" ],
     "location" : [ "Berlin" ],
     "publishedBy" : [ "Pädagogischer Zeitschriftenverl." ],
+    "publicationHistory" : "2.1964,7 - 38.2000",
     "frequency" : [ {
       "id" : "http://marc21rdf.info/terms/continuingfre#m",
       "label" : "monatlich"
     } ]
   }, {
-    "publicationHistory" : "2.1964,7 - 38.2000",
     "type" : [ "PublicationEvent" ],
     "location" : [ "Berlin" ],
-    "publishedBy" : [ "Verl. Volk & Wissen" ],
-    "frequency" : [ {
-      "id" : "http://marc21rdf.info/terms/continuingfre#m",
-      "label" : "monatlich"
-    } ]
+    "publishedBy" : [ "Verl. Volk & Wissen" ]
   } ],
   "shortTitle" : [ "Phys. Sch." ],
   "describedBy" : {

src/test/resources/alma-fix/990054215550206441.json

@@ -11,12 +11,12 @@
   "alternativeTitle" : [ "Bulletin annuel de statistiques des transports pour l'Europe et l'Amérique du Nord", "Ežegodnyj bjulleten' statistiki transporta dlja Evropy i Severnoj Ameriki", "Annual bulletin of transport statistics for Europe ( Hauptsacht. bis 44.1994 )", "Bulletin annuel de statistiques de transports européens ( Parallelsacht. bis 44.1994 )", "Ežegodnyj bjulleten' evropejskoj statistiki transporta ( Parallelsacht. bis 44.1994 )" ],
   "otherTitleInformation" : [ "= Bulletin annuel de statistiques des transports pour l'Europe et l'Amérique du Nord = Ežegodnyj bjulleten' statistiki transporta dlja Evropy i Severnoj Ameriki" ],
   "publication" : [ {
-    "publicationHistory" : "6.1954(1955) - 52.2004",
     "startDate" : "1955",
     "endDate" : "2004",
     "type" : [ "PublicationEvent" ],
     "location" : [ "Genève [u.a.]" ],
-    "publishedBy" : [ "UN" ]
+    "publishedBy" : [ "UN" ],
+    "publicationHistory" : "6.1954(1955) - 52.2004"
   } ],
   "describedBy" : {
     "id" : "http://lobid.org/resources/990054215550206441",

src/test/resources/alma-fix/990054301770206441.json

@@ -11,12 +11,12 @@
   "alternativeTitle" : [ "Reisen in Deutschland / 2", "Bayern ( Sachl. Benennung bis 25.1975 )" ],
   "otherTitleInformation" : [ "deutsches Handbuch für Fremdenverkehr" ],
   "publication" : [ {
-    "publicationHistory" : "19.[1966] - 34.1984",
     "startDate" : "1966",
     "endDate" : "1984",
     "type" : [ "PublicationEvent" ],
     "location" : [ "Darmstadt" ],
-    "publishedBy" : [ "Jaeger" ]
+    "publishedBy" : [ "Jaeger" ],
+    "publicationHistory" : "19.[1966] - 34.1984"
   } ],
   "describedBy" : {
     "id" : "http://lobid.org/resources/990054301770206441",

src/test/resources/alma-fix/990054345550206441.json

@@ -10,11 +10,11 @@
   "title" : "Eilendorfer Heimatblätter",
   "otherTitleInformation" : [ "Jahresschrift des Heimatvereins Eilendorf 1983 e.V" ],
   "publication" : [ {
-    "publicationHistory" : "1.1983(1984) -",
     "startDate" : "1984",
     "type" : [ "PublicationEvent" ],
     "location" : [ "Aachen-Eilendorf" ],
     "publishedBy" : [ "Heimatverein" ],
+    "publicationHistory" : "1.1983(1984) -",
     "frequency" : [ {
       "id" : "http://marc21rdf.info/terms/continuingfre#z",
       "label" : "unregelmäßig oder sonstige Erscheinungsfrequenz"

src/test/resources/alma-fix/990055981810206441.json

@@ -10,25 +10,20 @@
   "title" : "Bochumer Zeitpunkte",
   "otherTitleInformation" : [ "Beiträge zur Stadtgeschichte, Heimatkunde und Denkmalpflege" ],
   "publication" : [ {
-    "publicationHistory" : "1.1991 -",
     "startDate" : "1991",
     "type" : [ "PublicationEvent" ],
     "location" : [ "Bochum" ],
     "publishedBy" : [ "Kortum-Gesellschaft Bochum" ],
+    "publicationHistory" : "1.1991 -",
     "frequency" : [ {
       "id" : "http://marc21rdf.info/terms/continuingfre#z",
       "label" : "unregelmäßig oder sonstige Erscheinungsfrequenz"
     } ]
   }, {
-    "publicationHistory" : "1.1991 -",
     "startDate" : "1991",
     "type" : [ "PublicationEvent" ],
     "location" : [ "Bochum" ],
-    "publishedBy" : [ "Kracht" ],
-    "frequency" : [ {
-      "id" : "http://marc21rdf.info/terms/continuingfre#z",
-      "label" : "unregelmäßig oder sonstige Erscheinungsfrequenz"
-    } ]
+    "publishedBy" : [ "Kracht" ]
   } ],
   "describedBy" : {
     "id" : "http://lobid.org/resources/990055981810206441",

src/test/resources/alma-fix/990103770440206441.json

@@ -8,11 +8,11 @@
   "hbzId" : "HT012224491",
   "title" : "Wettersbacher Anzeiger",
   "publication" : [ {
-    "publicationHistory" : "Nachgewiesen 1979 -",
     "startDate" : "1972",
     "type" : [ "PublicationEvent" ],
     "location" : [ "Weil der Stadt" ],
     "publishedBy" : [ "Nussbaum" ],
+    "publicationHistory" : "Nachgewiesen 1979 -",
     "frequency" : [ {
       "id" : "http://marc21rdf.info/terms/continuingfre#w",
       "label" : "wöchentlich"

src/test/resources/alma-fix/990103899140206441.json

@@ -8,12 +8,12 @@
   "hbzId" : "HT012237361",
   "title" : "Statistische Berichte des Statistischen Landesamtes Rheinland-Pfalz, L. IV. 6: Finanzen und Steuern. Einheitswerte des Grundvermögens nach der Hauptfeststellung : Gemeindeergebnisse",
   "publication" : [ {
-    "publicationHistory" : "1964(1975); damit Ersch. eingest.",
     "startDate" : "1975",
     "endDate" : "1975",
     "type" : [ "PublicationEvent" ],
     "location" : [ "Bad Ems" ],
-    "publishedBy" : [ "Statist. Landesamt" ]
+    "publishedBy" : [ "Statist. Landesamt" ],
+    "publicationHistory" : "1964(1975); damit Ersch. eingest."
   } ],
   "describedBy" : {
     "id" : "http://lobid.org/resources/990103899140206441",

src/test/resources/alma-fix/990104908070206441.json

@@ -10,11 +10,11 @@
   "alternativeTitle" : [ "Appropriation acts ( Nebent. d. Mikrofiche-Ausg. )" ],
   "otherTitleInformation" : [ "passed at the session of ..." ],
   "publication" : [ {
-    "publicationHistory" : "Nachgewiesen 1911 -",
     "startDate" : "1911",
     "type" : [ "PublicationEvent" ],
     "location" : [ "Harrisburg, Pa." ],
-    "publishedBy" : [ "[Verlag nicht ermittelbar]" ]
+    "publishedBy" : [ "[Verlag nicht ermittelbar]" ],
+    "publicationHistory" : "Nachgewiesen 1911 -"
   }, {
     "type" : [ "SecondaryPublicationEvent" ],
     "location" : [ "Buffalo, NY" ],