algo.py

    def match_licenses(self):
        """Match license text against loaded licenses."""
        margin = min(20, len(self._known_licenses) / 20)
        found = 0
        test_ratio = 103
        while found < margin and test_ratio > 0:
            to_check = []
            test_ratio -= 3
            for digest in self._known_licenses.keys():
                license = self._known_licenses[digest]
                ratio = set_ratio(
                    self._cache,
                    (self.license_hash_lower, digest),
                    self.license_text_lower,
                    license[0].text.lower()
                )
                if ratio > test_ratio:
                    to_check.append(digest)
            found = len(to_check)

        lg.debug(
            "Prefiltered %d of %d licenses",
            len(to_check),
            len(self._known_licenses)
        )
        for digest in to_check:
            license = self._known_licenses[digest]
            ratio = partial_ratio(
                self._cache,
                (self.license_hash_lower, digest),
                clean_license.sub('', self.license_text_lower),
                clean_license.sub('', license[0].text.lower())
            )
            lg.debug(
                "Check %s for match against %s (%02d)",
                self.name,
                license[0].name,
                ratio
            )
            if ratio > 95:
                for license_copy in license:
                    name = license_copy.name
                    if name in _match_aliases:
                        self.licenses.add(_match_aliases[name])
                    else:
                        self.licenses.add(name)