I wrote a class that makes it easier to divide by specifying part of speech when using Mecab in python

When using mecab with python, I have to rewrite it in various ways when I want to specify the part of speech and write it freely, so I wrote the class myself to eliminate the inconvenience, so I will publish it.

import MeCab
import unicodedata
import re


class MecabParser():
    def __init__(self, word_classes=None, word_class_details=None):
        """
        Args:
            word_classes (list, optional):Part of speech specified in Japanese. Defaults to None.
            word_class_details (list, optional):Specifying the details of part of speech. Defaults to None.

See below for the part of speech defined by mecab
        https://taku910.github.io/mecab/posid.html
        """
        self._word_classes = word_classes
        self._word_class_details = word_class_details

    def _format_text(self, text):
        """
Formatting text before putting it in MeCab
        """
        text = re.sub(r'http(s)?://[\w/:%#\$&\?\(\)~\.=\+\-…]+', "", text)
        text = re.sub(r'[ -/:-@\[-~_]', "", text)  #Half-width symbol
        text = re.sub(r'[︰-＠]', "", text)  #Double-byte symbol
        text = re.sub(r'\d', "", text)  #Numbers
        text = re.sub('\n', " ", text)  #Newline character
        text = re.sub('\r', " ", text)  #Newline character

        return text

    def parse(self, text, is_base=False):
        text = self._format_text(text)

        #Character code conversion process. If not converted, the voiced sound mark and the semi-voiced sound mark will be separated.
        text = unicodedata.normalize('NFC', text)

        result = []
        tagger = MeCab.Tagger(
            '-d /usr/local/lib/mecab/dic/mecab-ipadic-neologd')
        #You can avoid surface read error by parsing once before parseToNode
        tagger.parse('')

        nodes = tagger.parseToNode(text)
        while nodes:
            wclass = nodes.feature.split(',')
            #If no part of speech is specified, all words are separated.
            if not self._word_classes:
                result.append(wclass[6] if is_base else nodes.surface)
                nodes = nodes.next
                continue

            #If no part of speech details are specified, all the part of speech is divided.
            if not self._word_class_details:
                if wclass[0] in self._word_classes:
                    result.append(wclass[6] if is_base else nodes.surface)
                    nodes = nodes.next
                    continue

            #Divide according to the detailed specification of part of speech
            if wclass[0] in self._word_classes and wclass[1] in self._word_class_details:
                result.append(wclass[6] if is_base else nodes.surface)
            nodes = nodes.next

        #Remove first and last whitespace strings
        if len(result) > 0:
            result.pop(0)
            result.pop(-1)

        return result

I save this in a file called parser.py and use it. The feeling of use is as follows, and it is relatively easy to divide words by specifying the part of speech.

>>> from parser import MecabParser
>>> mp =  MecabParser(word_classes=['noun'], word_class_details=['General','固有noun'])
>>> text = 'I'm hungry today, so I came to eat one of the best ramen in the neighborhood.'
>>> mp.parse(text)
['stomach', 'Neighborhood', 'Tenkaippin', 'ramen']

The dictionary uses mecab-ipadic-neologd, words with changed endings are corrected to their original form, and the text is formatted in advance, so I hope that the user can modify it to their liking. think.