Updates

record-migration/migration/dec20-to-feb21.py

+#!/usr/bin/env python3
+
+import json
+import os
+import sys
+
+
+def map_identifiers(identifiers):
+    ret = []
+    for k in identifiers:
+        val = {
+            "identifier": identifiers[k],
+            "scheme": k,
+        }
+        ret.append(val)
+    return ret
+
+
+def map_affiliation(affiliation):
+    return {
+        "name": affiliation["name"],
+        "identifiers": map_identifiers(affiliation.get("identifiers", {})),
+    }
+
+
+def map_creatibutor(creatibutor, default_role="editor"):
+    return {
+        "role": creatibutor.get("role", default_role),
+        "affiliations": [map_affiliation(aff) for aff in creatibutor["affiliations"]],
+        "person_or_org": {
+            "type": creatibutor["type"],
+            "name": creatibutor.get("name"),
+            "given_name": creatibutor.get("given_name"),
+            "family_name": creatibutor.get("family_name"),
+            "identifiers": map_identifiers(creatibutor.get("identifiers", [])),
+        },
+    }
+
+
+def map_right(right):
+    return {
+        "link": right.get("uri"),
+        "id": right.get("identifier"),
+        "title": right.get("rights"),
+    }
+
+
+def map_location(location):
+    ret = location.copy()
+    ret["identifiers"] = map_identifiers(location.get("identifiers", []))
+    return ret
+
+
+# - - - - - - - - - - - #
+# start doing the stuff #
+# - - - - - - - - - - - #
+
+directory = sys.argv[1] if len(sys.argv) > 1 else "outputs"
+for f in os.listdir(directory):
+    ff = os.path.join(directory, f)
+    recid, ext = os.path.splitext(f)
+    if ext.lower() != ".json":
+        continue
+
+    with open(ff, "r") as data_file:
+        data = json.load(data_file)
+
+    new = data.copy()
+    for k in ["id", "pid", "conceptid", "conceptpid"]:
+        new.pop(k, None)
+
+    new["metadata"]["contributors"] = [
+        map_creatibutor(c) for c in data["metadata"].get("contributors", [])
+    ]
+    new["metadata"]["creators"] = [
+        map_creatibutor(c) for c in data["metadata"]["creators"]
+    ]
+    new["metadata"]["rights"] = [
+        map_right(right) for right in data["metadata"].get("rights", [])
+    ]
+    new["metadata"]["locations"] = [
+        map_location(location) for location in data["metadata"].get("locations", [])
+    ]
+
+    # TODO: check why this is the case
+    new["metadata"].pop("references", None)
+
+    # because subjects aren't supported right now (TODO: check)
+    new["metadata"].pop("subjects", None)
+
+    # because the identifiers may have used weird schemes in both cases
+    new["metadata"].pop("locations", None)
+    new["metadata"].pop("funding", None)
+
+    new["access"] = {"record": "public", "files": "public", "owned_by": [{"user": "1"}]}
+
+    ff = os.path.join(directory, recid + ".migrated")
+    with open(ff, "w") as data_file:
+        json.dump(new, data_file)