Exclude all language-info that do not match the marc code #2062

hbz · Sep 13, 2024 · 2aba6dc · 2aba6dc
1 parent 1ff30a6
commit 2aba6dc
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 10 deletions.
diff --git a/src/main/resources/alma/fix/otherFields.fix b/src/main/resources/alma/fix/otherFields.fix
@@ -6,19 +6,23 @@ substring("@008-lang", "35", "3")
 copy_field("@008-lang", "@language.$append")
 copy_field("041[ 01] .[adj]", "@language.$append")
 
-uniq("@language")
+flatten("@language")
 
 set_array("language[]")
 do list(path:"@language", "var":"$i")
   unless any_match("$i","zxx|mul|sgn|und|.*[\\|\\#].*|\\s*")
-    copy_field("$i", "language[].$append.id")
-    copy_field("$i", "language[].$last.label")
+    if any_match("$i","[A-Za-z]{3}")
+      copy_field("$i", "language[].$append.id")
+      copy_field("$i", "language[].$last.label")
+    end
   end
 end
 
 lookup("language[].*.label","ISO639-2-to-GND")
 prepend("language[].*.id", "http://id.loc.gov/vocabulary/iso639-2/")
 
+uniq("language[]")
+
 # 300 - Physical Description (R)
 # We reuse the introx transformation here.
 

diff --git a/src/test/resources/alma-fix/990126426530206441.json b/src/test/resources/alma-fix/990126426530206441.json
@@ -80,10 +80,6 @@
     "label" : "hbz - Hochschulbibliothekszentrum des Landes Nordrhein-Westfalen, Netzwerkzone",
     "type" : [ "Collection" ]
   } ],
-  "language" : [ {
-    "id" : "http://id.loc.gov/vocabulary/iso639-2/deutsch",
-    "label" : "deutsch"
-  } ],
   "note" : [ "In: Energie Spektr.. - 11 (1996) Nr. 3 S. 18/21 : Abb.; 2 Lit." ],
   "bibliographicCitation" : "Energie Spektr.. - 11 (1996) Nr. 3 S. 18/21 : Abb.; 2 Lit.",
   "hasItem" : [ {

diff --git a/src/test/resources/alma-fix/990173607780206441.json b/src/test/resources/alma-fix/990173607780206441.json
@@ -79,9 +79,6 @@
   "language" : [ {
     "id" : "http://id.loc.gov/vocabulary/iso639-2/eng",
     "label" : "Englisch"
-  }, {
-    "id" : "http://id.loc.gov/vocabulary/iso639-2/engl.",
-    "label" : "engl."
   } ],
   "extent" : "192 S. : zahlr. Ill., teils farb.",
   "natureOfContent" : [ {