Function merge_matches

src/licensedcode/match.py:869–1068

Return a list of merged LicenseMatch matches given a `matches` list of LicenseMatch. Merging is a "lossless" operation that combines two or more matches to the same rule and that are in sequence of increasing query and index positions in a single new match.

(matches, max_dist=None, trace=TRACE_MERGE)

def merge_matches(matches, max_dist=None, trace=TRACE_MERGE):
    """
    Return a list of merged LicenseMatch matches given a `matches` list of
    LicenseMatch. Merging is a "lossless" operation that combines two or more
    matches to the same rule and that are in sequence of increasing query and
    index positions in a single new match.
    """
    # shortcut for single matches
    if len(matches) < 2:
        return matches

    # only merge matches with the same rule: sort then group by rule for the
    # same rule, sort on start, longer high, longer match, matcher type
    sorter = lambda m: (m.rule.identifier, m.qspan.start, -m.hilen(), -m.len(), m.matcher_order)
    matches.sort(key=sorter)
    matches_by_rule = [
        (rid, list(rule_matches))
        for rid, rule_matches
        in groupby(matches, key=lambda m: m.rule.identifier)
    ]

    if trace:
        print('merge_matches: number of matches to process:', len(matches))

    if max_dist is None:
        max_dist = MAX_DIST

    merged = []
    merged_extend = merged.extend

    for rid, rule_matches in matches_by_rule:
        if trace:
            logger_debug('merge_matches: processing rule:', rid)

        rule_length = rule_matches[0].rule.length

        # FIXME this is likely too much as we are getting gaps that are often too big
        max_rule_side_dist = min((rule_length // 2) or 1, max_dist)

        # compare two matches in the sorted sequence: current and next
        i = 0
        while i < len(rule_matches) - 1:
            j = i + 1
            while j < len(rule_matches):
                current_match = rule_matches[i]
                next_match = rule_matches[j]

                if trace:
                    logger_debug('---> merge_matches: current:', current_match)
                    logger_debug('---> merge_matches: next:   ', next_match)

                # FIXME: also considers the match length!
                # stop if we exceed max dist
                # or distance over 1/2 of rule length
                if (current_match.qdistance_to(next_match) > max_rule_side_dist
                or current_match.idistance_to(next_match) > max_rule_side_dist):

                    if trace: