license_identifier.py 文件源码

def get_top_candidates(self, source):
        # First, compute n-grams for all lines in the source file
        src_ng = n_grams.NGrams()
        src_ng.parse_text_list_items(
            source.lines,
            universe_ng=self.license_library.universe_n_grams)

        # Measure n-gram similarity relative to all licenses in the library
        similarities = OrderedDict()
        for license_name, lic in iteritems(self.license_library.licenses):
            similarity_score = lic.n_grams.measure_similarity(src_ng)
            similarities[license_name] = similarity_score

        # Filter out low-scoring licenses
        best_score = max(similarities.values())
        current_threshold = max(self.threshold,
                                best_score * self.keep_fraction_of_best)

        top_candidates = OrderedDict()
        for license_name, score in iteritems(similarities):
            if score >= current_threshold:
                top_candidates[license_name] = score

        return top_candidates