Function detect_copyrights_from_lines

src/cluecode/copyrights.py:153–224

Yield Detection objects detected in a ``numbered_lines`` sequence of tuples of (line number, text). The flags ``include_copyrights``, ``include_holders`` and ``include_authors`` drive which actual detections are done and returned. For copyrights only: - If ``include_copyri

(
    numbered_lines,
    include_copyrights=True,
    include_holders=True,
    include_authors=True,
    include_copyright_years=True,
    include_copyright_allrights=False,
    deadline=sys.maxsize,
)

def detect_copyrights_from_lines(
    numbered_lines,
    include_copyrights=True,
    include_holders=True,
    include_authors=True,
    include_copyright_years=True,
    include_copyright_allrights=False,
    deadline=sys.maxsize,
):
    """
    Yield Detection objects detected in a ``numbered_lines`` sequence of
    tuples of (line number, text).

    The flags ``include_copyrights``, ``include_holders`` and
    ``include_authors`` drive which actual detections are done and returned.

    For copyrights only:
    - If ``include_copyright_years`` is True, include years and year ranges.
    - If ``include_copyright_allrights`` is True, include trailing
      "all rights reserved"-style mentions

    Run for up to ``deadline`` seconds and return results found so far.
    """
    if not numbered_lines:
        return

    include_copyright_years = include_copyrights and include_copyright_years
    include_copyright_allrights = include_copyrights and include_copyright_allrights

    global DETECTOR

    if not DETECTOR:
        DETECTOR = detector = CopyrightDetector()
    else:
        detector = DETECTOR

    candidate_lines_groups = list(collect_candidate_lines(numbered_lines))

    if TRACE or TRACE_TOK:
        candidate_lines_groups = candidate_lines_groups
        logger_debug(
            f'detect_copyrights_from_lines: ALL groups of candidate '
            f'lines collected: {len(candidate_lines_groups)}',
        )

    for candidate_lines in candidate_lines_groups:
        if TRACE or TRACE_DEEP:
            logger_debug(f'\n========================================================================')
            logger_debug(f'detect_copyrights_from_lines: processing candidate_lines group:')

            for can in candidate_lines:
                logger_debug(f'  {can}')

        detections = detector.detect(
            numbered_lines=candidate_lines,
            include_copyrights=include_copyrights,
            include_holders=include_holders,
            include_authors=include_authors,