diff --git a/archive_query_log/parsers/warc_snippets.py b/archive_query_log/parsers/warc_snippets.py index f8eea84..216ec3b 100644 --- a/archive_query_log/parsers/warc_snippets.py +++ b/archive_query_log/parsers/warc_snippets.py @@ -119,7 +119,6 @@ def _parse_warc_snippets( method="xml", pretty_print=False, with_tail=True, - with_comments=True, ) snippet_id_components = ( parser.id, diff --git a/archive_query_log/parsers/xml.py b/archive_query_log/parsers/xml.py index d7caa39..0421ed1 100644 --- a/archive_query_log/parsers/xml.py +++ b/archive_query_log/parsers/xml.py @@ -63,12 +63,11 @@ def xpaths_from_css_selector(css_selector: str) -> list[str]: return ["."] selectors = cssselect_parse(css_selector) return [ - _translator.selector_to_xpath( + _translator.selector_to_xpath( selector, prefix="", translate_pseudo_elements=True, - ).replace( - "/descendant-or-self::*/", "//") + ).replace("/descendant-or-self::*/", "//") for selector in selectors ]