From 4c95e4289f2b89376b3cee55afddca795985d4cd Mon Sep 17 00:00:00 2001
From: Jannis Mohlin Tsiroyannis <jannis.mohlintsiroyannis@kb.se>
Date: Tue, 21 Apr 2020 14:30:49 +0200
Subject: [PATCH] Add cleaning script for empty "hasPart" entities containg
 only types.

---
 .../scripts/cleanups/2020/04/lxl-2963.groovy  | 34 +++++++++++++++++++
 1 file changed, 34 insertions(+)
 create mode 100644 whelktool/scripts/cleanups/2020/04/lxl-2963.groovy

diff --git a/whelktool/scripts/cleanups/2020/04/lxl-2963.groovy b/whelktool/scripts/cleanups/2020/04/lxl-2963.groovy
new file mode 100644
index 0000000000..367eaf08ba
--- /dev/null
+++ b/whelktool/scripts/cleanups/2020/04/lxl-2963.groovy
@@ -0,0 +1,34 @@
+PrintWriter scheduledForUpdating = getReportWriter("scheduled-updates")
+PrintWriter failedUpdating = getReportWriter("failed-updates")
+
+String where = "collection = 'bib' and data#>'{@graph,2,hasPart}' is not null"
+
+selectBySqlWhere(where) { data ->
+    def (record, mainEntity, work) = data.graph
+
+    boolean changed = false
+
+    Object parts = work.hasPart
+    if ( parts == null || ! (parts instanceof List) )
+        return
+
+    Iterator it = parts.iterator()
+    while (it.hasNext()) {
+        Map part = (Map) it.next()
+
+        if (part.get("@type") != null && part.size() == 1) {
+            changed = true
+            it.remove()
+        }
+    }
+
+    if (parts.isEmpty())
+        work.remove("hasPart")
+
+    if (changed) {
+        scheduledForUpdating.println("${data.doc.getURI()}")
+        data.scheduleSave(onError: { e ->
+            failedUpdating.println("Failed to update ${data.doc.shortId} due to: $e")
+        })
+    }
+}