summary

# coding:utf-8
import re

class Parser(object):
    def __init__(self, text):
        self.text = text
        self.sentences = self.split_sentences()

    def split_sentences(self):
        """
            Split the text into sentences
        """
        abbreviations = r"(Sr|Sra|Dr|Dra|Ex|Vs|Versus)\."
        sentences = re.sub(abbreviations, '', self.text)
        sentences = re.sub(r'!|\?|\.|\n', '<stop>', sentences)
        sentences = sentences.strip()
        sentences = sentences.split('<stop>')
        return list(filter(lambda s: s != '', sentences))

    def remove_punct(self, sentence):
        """
            Remove punctuations from text
        """
        return re.sub(r"\?|\.{1,}|!|,|;|-|:|\(|\)|'|\"|\*", '', sentence)

    def is_sentence(self, sentence):
        """
            Check if is a sentence
        """
        return True if ' ' in sentence else False

    def intersec_score(self, sentenceX, sentenceY):
        """
            Get score of intersection
        """
        if self.is_sentence(sentenceX) and self.is_sentence(sentenceY):
            result = 0
            sentenceX, sentenceY = self.remove_punct(sentenceX).split(), self.remove_punct(sentenceY).split()
            wordsInX = {(w, sentenceX.count(w)) for w in sentenceX}
            wordsInY = {(w, sentenceY.count(w)) for w in sentenceY}
            for k,v in wordsInX:
                result += v
            for k,v in wordsInY:
                result += v
            return result // 2
        else:
            return False

    def calc_score(self):
        """
            I suppose that method calc text score and return a summary?!
        """
        high_result = 0
        winner = {'score': 0, 'sentence': ''}
        sentences = []
        for i in range(0, len(self.sentences)):
            for j in range(0, len(self.sentences)):
                if self.sentences[i] != self.sentences[j]:
                    score = self.intersec_score(self.sentences[i], self.sentences[j])
                    if score > high_result:
                        high_result = score
                        sentences.append({'score': score, 'sentence': '{}. {}.'.format(self.sentences[i], self.sentences[j])})
        return self.get_average(sentences)

    def get_average(self, data):
        """
            Idnk how i did it
        """
        length = len(data)
        total = 0
        minimum = 1000000
        result = None
        total = sum([d['score'] for d in data])
        avg = round(total/length)
        for d in data:
            m = abs(d['score'] - avg)
            if m < minimum:
                minimum = m
                result = d['sentence']
        return result