Obtained from: https://github.com/adbar/htmldate/commit/b32422541cbc1f610838784f010a2e8cfc36fa05 --- htmldate/extractors.py.orig 2024-11-28 17:24:58 UTC +++ htmldate/extractors.py @@ -12,7 +12,6 @@ from dateparser import DateDataParser # type: ignore # coverage for date parsing from dateparser import DateDataParser # type: ignore # third-party, slow -from dateparser_data.settings import default_parsers from dateutil.parser import parse as dateutil_parse @@ -34,9 +33,8 @@ EXTERNAL_PARSER = DateDataParser( settings={ "NORMALIZE": True, # False may be faster "PARSERS": [ - p - for p in default_parsers - if p not in ("no-spaces-time", "relative-time", "timestamp") + "custom-formats", + "absolute-time", ], "PREFER_DATES_FROM": "past", "PREFER_LOCALE_DATE_ORDER": True, --- pyproject.toml.orig 2024-11-27 17:50:43 UTC +++ pyproject.toml @@ -53,7 +53,7 @@ dependencies = [ "dateparser >= 1.1.2", # 1.1.3+ slower # see tests on Github Actions "lxml == 4.9.2 ; platform_system == 'Darwin' and python_version <= '3.8'", - "lxml >= 5.3.0, < 6 ; platform_system != 'Darwin' or python_version > '3.8'", + "lxml >= 5.3.0 ; platform_system != 'Darwin' or python_version > '3.8'", "python-dateutil >= 2.9.0.post0", "urllib3 >= 1.26, < 3", ]