base_summarizer.py

from operator import attrgetter
from collections import namedtuple
from utils import ItemsCount

SentenceInfo = namedtuple("SentenceInfo", ("sentence", "order", "rating",))

class BaseSummarizer(object):
    
    def __call__(self, document, sentences_count):
        raise NotImplementedError("This method should be overriden in subclass")

    @staticmethod
    def normalize_word(word):
        return word.lower()

    @staticmethod
    def _get_best_sentences(sentences, count, rating, *args, **kwargs):
        rate = rating
        if isinstance(rating, dict):
            assert not args and not kwargs
            rate = lambda s: rating[s]

        infos = (SentenceInfo(s, o, rate(s, *args, **kwargs))
            for o, s in enumerate(sentences))

        # sorting sentences by rating in descending order
        infos = sorted(infos, key=attrgetter("rating"), reverse=True)
        # getting `count` first best rated sentences
        if not isinstance(count, ItemsCount):
            count = ItemsCount(count)
        infos = count(infos)
        # sorting sentences by their order in document
        infos = sorted(infos, key=attrgetter("order"))

        return tuple(i.sentence for i in infos)