[dateparser] Fix parsing very short weekday names

- improve the tokenization to handle suffixes like "th", "st", "nd", "rd" attached to numbers
- remove remove_multiple_occurrences this method
- fix tests to use the current year instead of 2025
- #1170

Author: adnan-awan

dateparser/languages/dictionary.py

@@ -201,14 +201,25 @@ def _split_by_known_words(self, string: str, keep_formatting: bool):
                 curr_split = (
                     [known] if self._should_capture(known, keep_formatting) else []
                 )
+
                 if unparsed and self._should_capture(unparsed, keep_formatting):
                     curr_split = (
                         self._split_by_numerals(unparsed, keep_formatting) + curr_split
                     )
+
                 if unknown:
                     string = unknown if string != unknown else ""
 
-            splitted.extend(curr_split)
+            for token in curr_split:
+                if (
+                    splitted
+                    and splitted[-1].isdigit()
+                    and token in {"st", "nd", "rd", "th"}
+                ):
+                    continue
+
+                splitted.append(token)
+
         return splitted
 
     def _split_by_numerals(self, string, keep_formatting):

dateparser/languages/locale.py

@@ -122,18 +122,6 @@ def weekdays(self):
         ]
         return weekdays
 
-    def remove_multiple_occurrences(self, date_str_tokens: list):
-        # first occurrence of day of the week will be considered
-        # followings occurrence(s) will be skipped and removed from the token list.
-        weekdays_counter = 0
-        for i, token in enumerate(date_str_tokens):
-            if token in self.weekdays:
-                weekdays_counter += 1
-
-                if weekdays_counter > 1:
-                    date_str_tokens.pop(i)
-                    continue
-
     def translate(self, date_string, keep_formatting=False, settings=None):
         """
         Translate the date string to its English equivalent.
@@ -169,8 +157,6 @@ def translate(self, date_string, keep_formatting=False, settings=None):
                     date_string_tokens[i] = dictionary[word] or fallback
         if "in" in date_string_tokens:
             date_string_tokens = self._clear_future_words(date_string_tokens)
-
-        self.remove_multiple_occurrences(date_string_tokens)
         return self._join(
             list(filter(bool, date_string_tokens)),
             separator="" if keep_formatting else " ",

tests/test_search.py

@@ -1098,9 +1098,10 @@ def test_search_dates_with_prepositions(self):
             add_detected_language=True,
             languages=["ru"],
         )
+        current_year = datetime.datetime.now().year
         expected = [
-            ("12 января", datetime.datetime(2025, 1, 12, 0, 0), "ru"),
-            ("30 апреля", datetime.datetime(2025, 4, 30, 0, 0), "ru"),
+            ("12 января", datetime.datetime(current_year, 1, 12, 0, 0), "ru"),
+            ("30 апреля", datetime.datetime(current_year, 4, 30, 0, 0), "ru"),
         ]
         assert result == expected