from POS.POSTagger import POSTagger as PPOSTagger

def cleanup_string(text):
    elim_char_string = "qwertyuiopasdfghjklzxcvbnmQWERTYUIOPASDFGHJKLZXCVBNM12345678910\/\n\t\r.,:-<>!@#$%^&*+=?..\"\')([]«»"
    text = text.translate(str.maketrans(elim_char_string, ' ' * len(elim_char_string)))
    text = text.replace("  ",'')
    text = text.rstrip()

    return text


def process_text_with_perpos_tagger(input_string):
    global tagger,perpos_tagger
    try:
        input_string.strip()
        print("input_string")
        print(input_string)
        elim_tag_list = ['N','Ne','P','Pe','RES','RESe','DET','DETe','CONJ','CONJe','CL','POSTP','PRO','PROe','NUM','NUMe','PUNC ']

        tokenized_data = perpos_tagger.parse(input_string.split())
        print(tokenized_data)
        reduced_list = []
        i = 0
        j = 0
        while(i<len(tokenized_data)-j):
            d = tokenized_data[i][1]
            if d not in elim_tag_list:
                reduced_list.append(tokenized_data[i][0])
                j+=1
            i+=1


        if len(reduced_list)>0:
            reduced_string = " ".join(reduced_list)
            reduced_string = cleanup_string(reduced_string)
            print("reduced string")
            print(reduced_string)
            return reduced_string
        else:
            return None
    except:
        return None