# Helper utility to make thesaurus input file.
# Not a program for students.

import re

def main():
    # build phone book as list
    grewords = set()
    with open("grewords.txt") as file:
        for line in file:
            tokens = line.strip().split("\t")
            if len(tokens) != 2:
                continue
            word, defn = tokens
            grewords.add(word.lower())
    
    thesaurus = {}
    part_of_speech = {}
    with open("oo-mythes/th_en_US_new.dat") as file:
        line = file.readline()
        while line != "":
            line = file.readline()
            if re.match(pattern=r'^[a-zA-Z]+[|][0-9]+$', string=line):
                word, count = line.strip().split("|")
                if word in grewords:
                    synonyms = []
                    for i in range(int(count)):
                        for syn in file.readline().strip().split("|"):
                            if syn[0] == "(" and syn[-1] == ")":
                                part_of_speech[word] = syn[1:-1]
                            elif syn not in synonyms:
                                synonyms.append(syn)
                    thesaurus[word] = synonyms
    
    with open("thesaurus.txt", "w") as out:
        for word in sorted(thesaurus.keys()):
            print(word, file=out)
            print(part_of_speech[word], file=out)
            print("|".join(thesaurus[word]), file=out)
            print("", file=out)   # blank line

main()