Consolidate methods for filtering filepaths by dates

harvester/harvest/alma.py

@@ -1,7 +1,6 @@
 """harvester.harvest.alma"""
 
 # ruff: noqa: TRY003, EM101
-import datetime
 import glob
 import logging
 import re
@@ -141,44 +140,33 @@ def _filter_filepaths_by_dates(self, filepaths: list[str]) -> list[str]:
         """Filter list of XML files by date.
 
         Given a list of XML files, the method will retrieve the date (YYYY-MM-DD)
-        from the filepath and determine whether the date is valid by checking
-        if the date falls within the specified 'from_date' and/or 'until_date'
-        arguments passed to the harvester.
+        from the filepath and check whether the date falls within the specified
+        'from_date' and/or 'until_date' arguments passed to the harvester.
+
+        Note: If 'from_date' and 'until_date' are not provided, the original list
+        of XML files will be returned. For full harvests, 'from_date' is required
+        and an error is thrown (by the full_harvest_get_source_records() method)
+        if not provided or cannot be derived.
 
         Example filepath: alma-2024-03-01-daily-extracted-records-to-index_19.xml
             - run_date=2024-03-01
         """
         filtered_filepaths = []
         for filepath in filepaths:
-            if _filepath_date := self._get_date_from_filepath(filepath):
-                filepath_date = convert_to_utc(date_parser(_filepath_date))
-            if self._validate_filepath_date(filepath_date):
-                filtered_filepaths.append(filepath)
-        return filtered_filepaths
-
-    def _validate_filepath_date(self, date: datetime.datetime) -> bool:
-        """Check if a given date is valid.
+            if filepath_date_string := self._get_date_from_filepath(filepath):
+                filepath_date = convert_to_utc(date_parser(filepath_date_string))
+
+                # include where filepath date meets harvester from/until date criteria
+                if (
+                    self.from_datetime_object is None
+                    or filepath_date >= self.from_datetime_object
+                ) and (
+                    self.until_datetime_object is None
+                    or filepath_date < self.until_datetime_object
+                ):
+                    filtered_filepaths.append(filepath)
 
-        Dates are valid if it meets at least one of the following condition:
-
-        * Date is (a) on or after MITAlmaHarvester.from_date AND (b) before
-            MITAlmaHarvester.until_date if 'from_date' and 'until_date' is specified.
-        * Date is on or after MITAlmaHarvester.from_date if only 'from_date' is specified.
-        * Date is before MITAlmaHarvester.until_date if only 'until_date' is specified.
-        """
-        if (self.from_datetime_object and self.until_datetime_object) and (
-            date >= self.from_datetime_object and date < self.until_datetime_object
-        ):
-            return True
-        if (self.from_datetime_object and self.until_datetime_object is None) and (
-            date >= self.from_datetime_object
-        ):
-            return True
-        if (self.until_datetime_object and self.from_datetime_object is None) and (
-            date < self.until_datetime_object
-        ):
-            return True
-        return False
+        return filtered_filepaths
 
     def _list_xml_files(self) -> list[str]:
         """Retrieve list of XML files from S3 or local filesystem."""