Merge branch 'caltechlibrary:main' into tests

Author: RohanBhattaraiNP

CITATION.cff

@@ -22,4 +22,4 @@ keywords:
   - metadata
   - software
   - InvenioRDM
-date-released: 2024-12-18
+

caltechdata_api/__init__.py

@@ -9,7 +9,8 @@
     caltechdata_unembargo,
     caltechdata_accept,
 )
-from .customize_schema import customize_schema
+from .customize_schema import customize_schema, validate_metadata
 from .get_metadata import get_metadata
 from .download_file import download_file, download_url
 from .utils import humanbytes
+from .md_to_json import parse_readme_to_json

caltechdata_api/cli.py

@@ -1,8 +1,7 @@
 import argparse
 import requests
 import s3fs
-from caltechdata_api import caltechdata_write, caltechdata_edit
-from md_to_json import parse_readme_to_json
+from caltechdata_api import caltechdata_write, caltechdata_edit, parse_readme_to_json
 import json
 import os
 from cryptography.fernet import Fernet
@@ -69,7 +68,7 @@ def get_or_set_token(production=True):
 
     key = load_or_generate_key()
 
-    # Use different token files for production and test environments
+    # Use different token files for production and test environment
     token_filename = "token.txt" if production else "token_test.txt"
     token_file = os.path.join(caltechdata_directory, token_filename)
 

caltechdata_api/customize_schema.py

@@ -480,18 +480,17 @@ def validate_metadata(json_record):
                     errors.append("Each 'subject' must have a 'subject' key.")
 
     # Check for 'dates'
-    if "dates" not in json_record:
-        errors.append("'dates' field is missing.")
-    elif not isinstance(json_record["dates"], list) or len(json_record["dates"]) == 0:
-        errors.append("'dates' should be a non-empty list.")
-    else:
-        for date in json_record["dates"]:
-            if (
-                not isinstance(date, dict)
-                or "date" not in date
-                or "dateType" not in date
-            ):
-                errors.append("Each 'date' must have 'date' and 'dateType'.")
+    if "dates" in json_record:
+        if not isinstance(json_record["dates"], list) or len(json_record["dates"]) == 0:
+            errors.append("'dates' should be a non-empty list.")
+        else:
+            for date in json_record["dates"]:
+                if (
+                    not isinstance(date, dict)
+                    or "date" not in date
+                    or "dateType" not in date
+                ):
+                    errors.append("Each 'date' must have 'date' and 'dateType'.")
 
     # Check for 'creators'
     if "creators" not in json_record:
@@ -601,10 +600,9 @@ def validate_metadata(json_record):
         errors.append("'publisher' should be a string.")
 
     # Check for 'publicationYear'
-    if "publicationYear" not in json_record:
-        errors.append("'publicationYear' field is missing.")
-    elif not isinstance(json_record["publicationYear"], str):
-        errors.append("'publicationYear' should be a string.")
+    if "publicationYear" in json_record:
+        if not isinstance(json_record["publicationYear"], str):
+            errors.append("'publicationYear' should be a string.")
 
     # Check for 'types'
     if "types" not in json_record: