From 74c69d3377d2fc602f8f7d445262c404f49995cd Mon Sep 17 00:00:00 2001
From: Fabian Steeg <steeg@hbz-nrw.de>
Date: Mon, 30 Oct 2023 13:48:14 +0100
Subject: [PATCH] Tweak rppd-to-lobid transformation for lobid-gnd usage
 (RPB-102)

- Prefer GND IDs and namespace for `id` field
- Add `type` field with hard-coded values
- Set filenames and indexing settings
---
 conf/rppd-to-lobid.fix  | 24 +++++++++++++++---------
 conf/rppd-to-lobid.flux |  6 +++---
 2 files changed, 18 insertions(+), 12 deletions(-)

diff --git a/conf/rppd-to-lobid.fix b/conf/rppd-to-lobid.fix
index b2e3dbcb..0af63f78 100644
--- a/conf/rppd-to-lobid.fix
+++ b/conf/rppd-to-lobid.fix
@@ -4,10 +4,21 @@ nothing()
 #00 RPPD-ID 
 # Komentar Doku:  (Achtung: 00 BLANK)
 
-copy_field("f00_","rppdId")
 copy_field("f00_","id")
-prepend("id","https://lbz.rlp.de/rppd/")
+copy_field("f00_","rppdId")
+prepend("id","https://rppd.lobid.org/")
+
+# -------
+#82b	(GND-ID (R)) ->	gndIdentifier	
+# Kommentar Doku:  ohne das vorangestellte Präfix (DE-588)
 
+copy_field("f82b", "gndIdentifier")
+unless all_contain("f82b", "Keine GND-Ansetzung")
+  copy_field("f82b","id")
+  prepend("id","https://d-nb.info/gnd/")
+end
+
+set_array("type[]", "AuthorityResource", "Person", "DifferentiatedPerson")
 
 # #1na (Name, bevorzugte Form) -> preferredName
 # Name ist aber Kombination aus Geburtsdaten und Name e.g. "f1na": "Marquard, Udo / 1959-"
@@ -117,12 +128,6 @@ end
 replace_all("f1ny", "(\\d{4})(\\d{2})(\\d{2})", "$1-$2-$3")
 copy_field("f1ny", "describedBy.dateModified")
 
-# -------
-#82b	(GND-ID (R)) ->	gndIdentifier	
-# Kommentar Doku:  ohne das vorangestellte Präfix (DE-588)
-
-copy_field("f82b", "gndIdentifier")
-
 # -------
 #1z1	(1. biogr. Anmerkung) ->	biographicalOrHistoricalInformation
 # Kommentar Doku:  getrennt durch "; ". Keine Abkürzungen benutzen, da die Stichworte in Register 9 indexiert werden. Zitate aus Quellen in Anführungszeichen; bei sehr langen, biogr. Anmerkungen wird der Text auf mehrere Kategorien aufgeteilt: #1z2, #1z3, #1z4 ... #1z9. Bei eingespielten Biographien werden die Angaben zum Originalwerk am Ende angegeben:  --- [Daten übernommen aus: ....]
@@ -130,7 +135,8 @@ copy_field("f82b", "gndIdentifier")
 vacuum()
 retain(
   "rppdId", 
-  "id", 
+  "id",
+  "type[]",
   "preferredName",
   "variantName[]",
   "dateOfBirth[]",
diff --git a/conf/rppd-to-lobid.flux b/conf/rppd-to-lobid.flux
index e12c7d81..fce6039e 100644
--- a/conf/rppd-to-lobid.flux
+++ b/conf/rppd-to-lobid.flux
@@ -1,11 +1,11 @@
-default outfile = "conf/output/bulk/bulk-rppd-${i}.ndjson";
-"conf/output/test-output-rppd.json"
+default outfile = "conf/output/bulk/rppd/bulk-rppd-${i}.jsonl"; // lobid-gnd expects *.jsonl suffix
+"conf/output/output-rppd-strapi.ndjson"
 | open-file
 | as-lines
 | decode-json
 | fix(FLUX_DIR + "rppd-to-lobid.fix")
 | batch-reset(batchsize="1000")
 | encode-json(prettyPrinting="false")
-| json-to-elasticsearch-bulk(idkey="id", type="resource", index="resources-alma-fix-staging")
+| json-to-elasticsearch-bulk(idkey="id", type="authority", index="gnd-rppd-test")
 | write(outfile)
 ;