# Put your text in a file called input.txt in the same directory.

import subprocess
import csv
import string
import time

class MeCabNode:
    def __init__(self, term="", pos="", pos1="", pos2="", pos3="", rule="", conj="", dictform="", reading="", pronunc=""):
        self.term = term
        self.pos = pos
        self.pos1 = pos1
        self.pos2 = pos2
        self.pos3 = pos3
        self.rule = rule
        self.conj = conj
        self.dictform = dictform
        self.reading = reading
        self.pronunc = pronunc
    
    def toStringExtended(self):
        return "Term: " + self.term + "\nPos: " + self.pos + "\nPos1: " + self.pos1 + "\nPos2: " + self.pos2 + "\nPos3: " + self.pos3 + "\nRule: " + self.rule + "\nConjugation: " + self.conj + "\nDictionary Form: " + self.dictform + "\nReading: " + self.reading + "\nPronunciation: " + self.pronunc + "\n"
    
    def toString(self):
        return self.dictform

class MeCab:
    # Assumes that MeCab is installed and has been added to the system path.
    def getMatrix(self, mecabdir):
        subprocess.Popen('"'+mecabdir+'\\mecab.exe" input.txt -o mecab-raw-output.txt', shell=True)

        # Waits for MeCab to finish working.
        time.sleep(5)
        
        csvfile = open("mecab-raw-output.txt", encoding='utf-8')
        FILE = open("mecab-output-tidied.txt", "w", encoding='utf-8')
        
        for line in csvfile:
            if "\t" in line:
                line = line.replace("\t", ",")
                FILE.write(line)
        
        FILE.close()
        csvfile.close()
        
        csvfile = open("mecab-output-tidied.txt", encoding='utf-8')
        
        matrix = list(csv.reader(csvfile))
        
        FILE.close()
	
        return matrix

    def getNode(self, line):
        node = MeCabNode()
        if len(line) > 0:
            node.term = line[0]
            node.pos = line[1]
            node.pos1 = line[2]
            node.pos2 = line[3]
            node.pos3 = line[4]
            node.rule = line[5]
            node.conj = line[6]
        if len(line) >= 8:
            node.dictform = line[7]
        if len(line) >= 9:
            node.reading = line[8]
        if len(line) >= 10:
            node.pronunc = line[9]
        return node

    def isValidTerm(self, term):
        if term == "" or term == "*" or term == "。" or term == "、" or term == "（" or term == "）" or term == "の" or term == "は" or term == "が" or term == "と" or term == "で" or term == "を" or term == "に":
            return False
        else:
            return True

    def getNodeSet(self, mecabdir):
        matrix = self.getMatrix(mecabdir)
        
        nodeset = set([])
        
        for line in matrix:
            node = self.getNode(line)
            if node != None:
                if self.isValidTerm(node.dictform):
                    nodeset.add(node)
        
        return nodeset

mecab = MeCab()
nodeset = mecab.getNodeSet("C:\\Program Files (x86)\\MeCab\\bin")

FILE = open("output.txt", "w", encoding='utf-8')

for node in nodeset:
    FILE.write(node.toString() + "\n")

FILE.close()