From c17243cdc36809fff78ce5cc56350796e365eade Mon Sep 17 00:00:00 2001
From: Fabian Steeg <steeg@hbz-nrw.de>
Date: Thu, 13 Jun 2024 18:56:02 +0200
Subject: [PATCH] Add workflow for updating `created` in Strapi (RPB-170)

Based on Allegro `f95_` values and Strapi `id`, merged via `rppdId`
---
 conf/rppd-strapi-to-strapi-created.flux | 47 +++++++++++++++++++++++++
 1 file changed, 47 insertions(+)
 create mode 100644 conf/rppd-strapi-to-strapi-created.flux

diff --git a/conf/rppd-strapi-to-strapi-created.flux b/conf/rppd-strapi-to-strapi-created.flux
new file mode 100644
index 0000000..56674de
--- /dev/null
+++ b/conf/rppd-strapi-to-strapi-created.flux
@@ -0,0 +1,47 @@
+// Create a shell script to update a specific field in Strapi (created),
+// using curl and the Strapi update API (can't send from here, need ID in path):
+// https://docs.strapi.io/dev-docs/api/rest#update-an-entry
+
+// Use export data from Allegro (for `f95_` values) and Strapi (for record `id`), merged via `rppdId`:
+// sbt "runMain rpb.ETL conf/rppd-to-strapi.flux IN_FILE=RPB-Export_HBZ_Bio.txt OUT_FILE=output-rppd-strapi.ndjson"
+// zgrep -a '"type":"api::person.person"' conf/strapi-export.tar.gz > conf/output/rppd-export.jsonl
+// sbt "runMain rpb.ETL conf/rppd-strapi-to-strapi-created.flux HOST=test-metadaten-nrw:1339 API_TOKEN=bb0..."
+// bash conf/output/rppd-strapi-update.sh
+
+default HOST = "localhost:1337"; // pass e.g. HOST=test-metadaten-nrw:1339
+default API_TOKEN = ""; // pass e.g. API_TOKEN=bb0...
+
+"conf/output/rppd-export.jsonl"
+| open-file
+| as-lines
+| decode-json
+| fix("
+unless all_match('data.rppdId', 'p.+')
+  reject()
+end
+move_field('data.rppdId', '_id')
+retain('id', '_id')
+")
+| stream-to-triples(redirect="true")
+| @X;
+
+"conf/output/output-rppd-strapi.ndjson"
+| open-file
+| as-lines
+| decode-json
+| fix("
+move_field('rppdId', '_id')
+move_field('f95_', 'data.created')
+retain('data.created', '_id')
+")
+| stream-to-triples(redirect="true")
+| @X;
+
+@X
+| wait-for-inputs("2")
+| sort-triples(by="subject")
+| collect-triples
+| encode-json(prettyPrinting="false")
+| match(pattern=".+?id...(.+)...data..(.+).+", replacement="curl -X PUT -H 'Content-Type: application/json' -d '{\"data\":$2}' -H 'Authorization: Bearer " + API_TOKEN + "' -w '\\\\n' http://" + HOST + "/api/persons/$1")
+| write(FLUX_DIR + "output/rppd-strapi-update.sh")
+;