feat(etl): add museum data from museum authority list

packages/etl/src/index.js

@@ -145,6 +145,7 @@ async function processTeiFiles(inputPath, outputPath, options = {}) {
         delete result.provenanceFound;
         delete result.provenanceObserved;
         delete result.provenanceLost;
+        delete result.repository.museum;
         delete result.bibliographyEdition;
         delete result.bibliographyDiscussion;
 

packages/etl/src/metadata.js

@@ -1,4 +1,5 @@
 import xml2js from "xml2js";
+import museums from "../../../data/processed/museums.json" assert { type: "json" };
 
 /**
  * Extracts metadata from an XML string.
@@ -312,11 +313,28 @@ function getMsIdentifier(xml) {
     country: msIdentifier.country?.trim(),
     region: msIdentifier.region?.trim(),
     settlement: msIdentifier.settlement?.trim(),
-    repository: msIdentifier.repository,
+    repository: getRepository(msIdentifier),
     idno: msIdentifier.idno,
   };
 }
 
+function getRepository(msIdentifier) {
+  const ref = msIdentifier.repository?.ref;
+
+  if (!ref) return msIdentifier.repository;
+
+  const museum = museums.find((m) => m.uri === ref);
+
+  if (!museum) return msIdentifier.repository;
+
+  return {
+    _: museum.name,
+    role: museum.type,
+    ref: museum.uri,
+    museum,
+  };
+}
+
 function getTextLang(xml) {
   const textLang =
     xml.TEI.teiHeader.fileDesc.sourceDesc.msDesc.msContents.textLang;