def preprocess_text(text):

    lower_text = text.lower()
    cleaned_text = re.sub("\\(.*?\\)", "", lower_text)

    tokens = stemmer.analyze(cleaned_text)

    lemmas = []

    for token in tokens:

        if "analysis" in token:
            if len(token["analysis"]):
                if "lex" in token["analysis"][0]:
                    lemmas.append(token["analysis"][0]["lex"])
            else:
                if "text" not in token:
                    continue

                # english
                if re.match("[a-z]", token["text"].lower()):
                    if not len(lemmas) or lemmas[-1] != "_ENG_":
                        lemmas.append("_ENG_")
                else:
                    if re.sub("\s", "", token["text"]):
                        if not len(lemmas) or lemmas[-1] != "_NAME_":
                            lemmas.append("_NAME_")
        else:
            if "text" not in token:
                continue

            if token["text"].isdigit():
                if not len(lemmas) or lemmas[-1] != "_DIGITS_":
                    lemmas.append("_DIGITS_")
            else:
                clean_text = re.sub(" ", "", token["text"])
                if clean_text in [".", ",", "?", "!", ":", "(", ")"]:
                    lemmas.append(clean_text)


    return " ".join(lemmas)