Function remove

archivebox/main.py:701–777

Remove the specified URLs from the archive

(filter_str: Optional[str]=None,
           filter_patterns: Optional[List[str]]=None,
           filter_type: str='exact',
           snapshots: Optional[QuerySet]=None,
           after: Optional[float]=None,
           before: Optional[float]=None,
           yes: bool=False,
           delete: bool=False,
           out_dir: Path=OUTPUT_DIR)

@enforce_types
def remove(filter_str: Optional[str]=None,
           filter_patterns: Optional[List[str]]=None,
           filter_type: str='exact',
           snapshots: Optional[QuerySet]=None,
           after: Optional[float]=None,
           before: Optional[float]=None,
           yes: bool=False,
           delete: bool=False,
           out_dir: Path=OUTPUT_DIR) -> List[Link]:
    """Remove the specified URLs from the archive"""
    
    check_data_folder(out_dir=out_dir)

    if snapshots is None:
        if filter_str and filter_patterns:
            stderr(
                '[X] You should pass either a pattern as an argument, '
                'or pass a list of patterns via stdin, but not both.\n',
                color='red',
            )
            raise SystemExit(2)
        elif not (filter_str or filter_patterns):
            stderr(
                '[X] You should pass either a pattern as an argument, '
                'or pass a list of patterns via stdin.',
                color='red',
            )
            stderr()
            hint(('To remove all urls you can run:',
                'archivebox remove --filter-type=regex ".*"'))
            stderr()
            raise SystemExit(2)
        elif filter_str:
            filter_patterns = [ptn.strip() for ptn in filter_str.split('\n')]

    list_kwargs = {
        "filter_patterns": filter_patterns,
        "filter_type": filter_type,
        "after": after,
        "before": before,
    }
    if snapshots:
        list_kwargs["snapshots"] = snapshots

    log_list_started(filter_patterns, filter_type)
    timer = TimedProgress(360, prefix='      ')
    try:
        snapshots = list_links(**list_kwargs)
    finally:
        timer.end()


    if not snapshots.exists():
        log_removal_finished(0, 0)
        raise SystemExit(1)


    log_links = [link.as_link() for link in snapshots]