From f9a92572f79710d6085f2522c5547afcb24b7334 Mon Sep 17 00:00:00 2001
From: David McKee <david.mckee@dxw.com>
Date: Fri, 13 Dec 2024 15:15:36 +0000
Subject: [PATCH] Use corb2 to set identifiers for all primary documents that
 aren't failures

---
 .gitignore                   |  2 ++
 corb2/README.md              |  7 +++++++
 corb2/corb                   |  1 +
 corb2/get-uris.xqy           |  4 ++++
 corb2/migrate-ncn.properties |  9 +++++++++
 corb2/migrate-ncn.xqy        | 34 ++++++++++++++++++++++++++++++++++
 6 files changed, 57 insertions(+)
 create mode 100644 corb2/README.md
 create mode 100755 corb2/corb
 create mode 100644 corb2/get-uris.xqy
 create mode 100644 corb2/migrate-ncn.properties
 create mode 100644 corb2/migrate-ncn.xqy

diff --git a/.gitignore b/.gitignore
index 2036290..30499b3 100644
--- a/.gitignore
+++ b/.gitignore
@@ -5,3 +5,5 @@ gradle-*.properties
 !gradle-development.properties
 __pycache__
 node_modules/
+corb2/*.jar
+corb2/*.log
diff --git a/corb2/README.md b/corb2/README.md
new file mode 100644
index 0000000..df6e42a
--- /dev/null
+++ b/corb2/README.md
@@ -0,0 +1,7 @@
+`brew install temurin` to get a working Java (I had to install-uninstall-install)
+
+Download corb2 from https://github.com/marklogic-community/corb2/releases
+Download xcc from https://repo1.maven.org/maven2/com/marklogic/marklogic-xcc/11.1.0/ or https://developer.marklogic.com/products/xcc-2/ (I used maven)
+(both into this directory)
+
+`corb migrate-ncn` will run the code, `migrate-ncn.log` will show what it writes to the identifiers
diff --git a/corb2/corb b/corb2/corb
new file mode 100755
index 0000000..387f6f2
--- /dev/null
+++ b/corb2/corb
@@ -0,0 +1 @@
+java -server -cp .:marklogic-xcc-11.1.0.jar:marklogic-corb-2.5.6.jar -DOPTIONS-FILE=$1.properties com.marklogic.developer.corb.Manager xcc://admin:admin@localhost:8011
diff --git a/corb2/get-uris.xqy b/corb2/get-uris.xqy
new file mode 100644
index 0000000..df6c30d
--- /dev/null
+++ b/corb2/get-uris.xqy
@@ -0,0 +1,4 @@
+let $uris := cts:uris("",(),cts:collection-query(
+   ("http://marklogic.com/collections/dls/latest-version")
+))
+return (count($uris), $uris)
diff --git a/corb2/migrate-ncn.properties b/corb2/migrate-ncn.properties
new file mode 100644
index 0000000..e0162ae
--- /dev/null
+++ b/corb2/migrate-ncn.properties
@@ -0,0 +1,9 @@
+MODULES-DATABASE=caselaw-modules
+INSTALL = 1
+
+URIS-MODULE=get-uris.xqy
+PROCESS-MODULE=migrate-ncn.xqy
+
+PROCESS-TASK=com.marklogic.developer.corb.ExportBatchToFileTask
+EXPORT-FILE-NAME=migrate-ncn.log
+THREAD-COUNT=10
diff --git a/corb2/migrate-ncn.xqy b/corb2/migrate-ncn.xqy
new file mode 100644
index 0000000..35db7d7
--- /dev/null
+++ b/corb2/migrate-ncn.xqy
@@ -0,0 +1,34 @@
+declare namespace uk = "https://caselaw.nationalarchives.gov.uk/akn";
+
+import module namespace sem = "http://marklogic.com/semantics"
+      at "/MarkLogic/semantics.xqy";
+
+(: This is intended to be a migration script that runs once;
+   it should not be run on a database which already has identifiers :)
+
+declare variable $URI external;
+
+let $cite := fn:doc($URI)//uk:cite/text()
+let $slug := fn:replace(
+                fn:replace($URI, "\.xml$", "")
+                , "^/", "")
+let $log := ("")
+let $uuid := "id-"||sem:uuid-string()
+let $log := ($log, "Processing", $URI, $cite, $uuid)
+let $node :=
+<identifiers><identifier>
+<namespace>ukncn</namespace>
+<uuid>{$uuid}</uuid>
+<value>{$cite}</value>
+<url_slug>{$slug}</url_slug>
+</identifier></identifiers>
+let $log := ($log, xdmp:quote($node))
+
+let $log := ($log, if
+            (fn:starts-with($URI, "/failures/") or fn:starts-with($URI, "/collisions/"))
+          then
+            "ignored as failure/collision"
+          else
+            "set property" || xdmp:document-set-property($URI, $node))
+
+return string-join($log, "&#10;")