fix: validate date

metadata_tools/importxls/export_utils.py

@@ -3,6 +3,8 @@ import json
 import logging
 import urllib
 
+import re
+
 from datetime import datetime
 from json import dumps
 from socket import timeout
@@ -68,14 +70,26 @@ def process_possible_date(possible_date):
     if isinstance(possible_date, dt.date):
         return possible_date.strftime("%Y-%m-%d")
     elif isinstance(possible_date, int):
-        return ""
+        logging.warning("Date '{possible_date}' is of type Int. Casting to 'YYYY-MM-DD'")
+        d = datetime.fromordinal(datetime(1900, 1, 1).toordinal() + possible_date - 2)
+        return d.strftime("%Y-%m-%d")
     else:
         try:
-            d = datetime.strptime(possible_date.replace('/','.'), '%d.%m.%Y')
-            return d.strftime("%Y-%m-%d")
+          d = process_possible_date_str(possible_date)
+          return d
         except ValueError as e:
-            return ""
+            logging.error(f"Date '{possible_date}' is not a valid date of standard format ('YYYY-MM-DD' or 'DD/MM/YYYY')")
+            return None
 
+def process_possible_date_str(possible_date):
+    if possible_date == '':
+            return None
+    elif re.compile(r'^([0-9]{4})-(1[0-2]|0[1-9])-(3[01]|0[1-9]|[12][0-9])$').match(possible_date):
+        d = datetime.strptime(possible_date.replace('/','.'), '%d.%m.%Y')
+        return possible_date
+    else:
+        d = datetime.strptime(possible_date.replace('/','.'), '%d.%m.%Y')
+        return d.strftime("%Y-%m-%d")
 
 
 def get_names_from_string(full_name):