systemy_dialogowe/utils/script.py

def remove_empty_lines(text):
    lines = text.split('\n')

    non_empty_lines = []
    for line in lines:
        if line.strip():
            non_empty_lines.append(line)

    return '\n'.join(non_empty_lines)


def join_lines(text):
    lines = text.split("\n")
    joined_lines = [lines[0]]

    for line in lines[1:]:
        if line.startswith("["):
            joined_lines.append(line)
        else:
            joined_lines[-1] += " " + line

    return "\n".join(joined_lines)


def process_text(text, nazwisko_user, nazwisko_system):
    lines = text.split('\n')
    result = []
    for line in lines:
        if nazwisko_system in line:
            result.append("system " + line.split(nazwisko_system, 1)[1].strip())
        elif nazwisko_user in line:
            result.append("user " + line.split(nazwisko_user, 1)[1].strip())
        else:
            result.append(line)
    return '\n'.join(result)


def read_tsv(file_path):
    with open(file_path, 'r', encoding='utf-8') as tsv_file:
        content = tsv_file.read()
    return content


def write_tsv(file_path, content):
    with open(file_path, 'w', encoding='utf-8') as tsv_file:
        tsv_file.write(content)


def replace_first_space_with_tab(text):
    lines = text.split("\n")
    new_lines = []
    for line in lines:
        first_space_index = line.find(" ")
        if first_space_index != -1:
            new_line = line[:first_space_index] + "\t" + line[first_space_index + 1:]
            new_lines.append(new_line)
        else:
            new_lines.append(line)
    return "\n".join(new_lines)


if __name__ == '__main__':
    # wypelnic przed uruchomieniem
    nazwisko_user = ""
    nazwisko_system = ""
    numer_user = ""
    numer_system = ""
    file_path = f"J:\PycharmProjects\systemy_dialogowe\data\dialog-{numer_system}-{numer_user}-01.tsv"

    text = read_tsv(file_path)
    text = remove_empty_lines(text)
    text = join_lines(text)
    text = process_text(text, nazwisko_user, nazwisko_system)
    text = replace_first_space_with_tab(text)
    write_tsv(file_path, text)