Function refine_matches

src/licensedcode/match.py:2691–2832

Return a filtered list of kept LicenseMatch matches and a list of discardable matches given a `matches` list of LicenseMatch by removing matches that do not mee certain criteria as defined in multiple filters.

(
    matches,
    query=None,
    min_score=0,
    filter_false_positive=True,
    merge=True,
    trace_basic=TRACE,
    trace=TRACE_REFINE,
)

def refine_matches(
    matches,
    query=None,
    min_score=0,
    filter_false_positive=True,
    merge=True,
    trace_basic=TRACE,
    trace=TRACE_REFINE,
):
    """
    Return a filtered list of kept LicenseMatch matches and a list of
    discardable matches given a `matches` list of LicenseMatch by removing
    matches that do not mee certain criteria as defined in multiple filters.
    """

    if trace_basic:
        logger_debug()
        logger_debug(' #####refine_matches: STARTING matches#', len(matches))
    if trace:
        for m in matches:
            logger_debug(m)
    if not matches:
        return [], []

    all_discarded = []
    all_discarded_extend = all_discarded.extend

    if merge:
        matches = merge_matches(matches)

        if trace_basic:
            logger_debug('     ##### refine_matches: STARTING MERGED_matches#:', len(matches))

    def _log(_matches, _discarded, msg):
        if trace_basic:
            logger_debug('   #####refine_matches: KEPT', msg, '#', len(matches))

        if trace:
            for m in matches:
                logger_debug(m)

        if trace_basic:
            logger_debug('   #####refine_matches: DISCARDED NOT', msg, '#', len(_discarded))

        if trace:
            for m in matches:
                logger_debug(m)

    set_matched_lines(matches, query.line_by_pos)

    # FIXME: we should have only a single loop on all the matches at once!!
    # and not 10's of loops!!!

    matches, discarded = filter_matches_missing_required_phrases(matches)
    all_discarded_extend(discarded)
    _log(matches, discarded, 'HAS REQUIRED PHRASES')

    matches, discarded = filter_spurious_matches(matches)