don't load sklearn libraries unless needed

src/paperless/settings.py

@@ -4,7 +4,6 @@ import multiprocessing
 import os
 import re
 
-import dateparser
 from dotenv import load_dotenv
 
 from django.utils.translation import gettext_lazy as _
@@ -491,7 +490,11 @@ if PAPERLESS_TIKA_ENABLED:
 
 # List dates that should be ignored when trying to parse date from document text
 IGNORE_DATES = set()
-for s in os.getenv("PAPERLESS_IGNORE_DATES", "").split(","):
-    d = dateparser.parse(s)
-    if d:
-        IGNORE_DATES.add(d.date())
+
+if os.getenv("PAPERLESS_IGNORE_DATES", ""):
+    import dateparser
+
+    for s in os.getenv("PAPERLESS_IGNORE_DATES", "").split(","):
+        d = dateparser.parse(s)
+        if d:
+            IGNORE_DATES.add(d.date())