synonym_file_parsers.py

"""
This module defines the parsers for the synonym file.

An abstract class is defined for the parser. This will allow us to
write parsers for other synonym input formats easily if needed in the
future.
"""

from abc import ABC, abstractmethod


class AbstractSynonymParser(ABC):

    """
    An abstract base class for which defines a synonym file parser.
    """

    @abstractmethod
    def __init__(self, synonym_file):
        """
        In a subclass method should instantiate the parser and use the
        parse method (see below) to read the file and generate the
        synonyms_dict class variable
        This method may throw an IOError.
        """
        pass

    @abstractmethod
    def parse(self, file_name):
        """
        In a subclass this method should read the file given as the
        argument file_name and generates a dict of synonyms.
        This method may throw an IOError.
        """
        pass

    @abstractmethod
    def get_synonyms(self):
        """
        In a subclass this method should return the synonyms
        dictionary generated by the parser.

        @return a dictionary of strings which maps a synonym to the
        first synonym in the list of synonyms as defined in the synonym file
        """
        pass


class SpaceSeparatedSynonymFileParser(AbstractSynonymParser):

    """
    A class which instantiates a parser for the space separated
    synonym file format.
    """

    def __init__(self, synonym_file):
        """
        initializes a space seperated synonym_file parser
        """
        self.synonym_dict = {}
        self.parse(synonym_file)

    def parse(self, file_name):
        """
        Reads the file given as the argument file_name and generates a
        dict of synonyms.
        This method may throw a IOError Exception.
        @return None
        """
        with open(file_name) as file_handler:
            for line in file_handler.readlines():
                words = line.split()
                synonym = words[0].lower()
                for word in words:
                    self.synonym_dict[word.lower()] = synonym
        return

    def get_synonyms(self):
        """
        This method returns the list of synonyms dict generated by the parser.

        @return a dictionary of strings which maps a synonym to the
        first synonym in the list of synonyms as defined in the synonym file
        """
        return self.synonym_dict