Function get_match

etc/scripts/licenses/synclic.py:227–276

Return a tuple of (license key, True if exact match, match score, match text) e.g.: - top matched license key or None, - True if this an exact match, False if the match is ok, None if the match is weak, - match score or 0 or None - matched text or None

(text)

def get_match(text):
    """
    Return a tuple of (license key, True if exact match, match score, match text)
    e.g.:
        - top matched license key or None,
        - True if this an exact match, False if the match is ok, None if the match is weak,
        - match score or 0 or None
        - matched text or None
    """

    from licensedcode.cache import get_index

    idx = get_index()
    matches = list(idx.match(query_string=text, min_score=80))
    if not matches:
        return None, None, 0, None

    match = matches[0]
    matched_text = match.matched_text(whole_lines=False)
    query = match.query
    query_len = len(query.whole_query_run().tokens)
    rule = match.rule
    rule_licenses = rule.license_keys()
    key = rule_licenses[0]

    is_exact = (
        len(matches) == 1
        and rule.is_from_license
        and len(rule_licenses) == 1
        and match.matcher == "1-hash"
        and match.score() == 100
        and match.len() == query_len
    )

    if is_exact:
        return key, True, 100, matched_text

    is_ok = len(rule_licenses) == 1 and match.coverage() > 95 and match.score() > 95
    if is_ok:
        return key, False, match.score(), matched_text

    is_weak = len(rule_licenses) == 1 and match.coverage() > 90 and match.score() > 90
    if is_weak:
        return key, None, match.score(), matched_text

    if match.score() > 85:
        # junk match
        return key, -1, match.score(), matched_text
    else:
        return None, None, None, None


def get_response(url, headers, params):