chat-restaruacja/tsv_data_gen.py

import csv

operators_and_users = {
    # operator
    ("06", "Krzysztof Bojakowski"): [
        # users
        ("01", "Mikołaj Gawron"),
        ("05", "Patryk Osiński"),
        ("09", "Bartosz Wieczorek"),
        ("07", "Sergiusz Kański"),
        ("03", "Dawid Korzępa"),
    ],
    ("05", "Patryk Osiński"): [
        # ("02", "Anna Śmigiel"),
        # ("04", "Michał Kasprowicz"),
        ("06", "Krzysztof Bojakowski"),
        ("08", "Krystian Osiński"),
    ]
}

OP_ANON = "system"
USER_ANON = "user"

for operator, users in operators_and_users.items():
    op_id, op_name = operator
    for (user_id, user_name) in users:
        with open(f"./data/dialog-{op_id}-{user_id}-01.tsv", "w", newline='\n', encoding="utf-8") as csvfile:
            spamwriter = csv.writer(csvfile, delimiter='\t', quotechar='|', quoting=csv.QUOTE_MINIMAL)
            spamwriter.writerow(["kto", "treść"])
            with open(f"./data/raw/dialog-{op_id}-{user_id}-01.txt", "r", encoding="utf-8") as f:
                whos_typing = None
                text = ""
                while True:
                    line = f.readline().strip()
                    if '"' in line:
                        line = line.replace('"', "'") # Git preview can't display data with " in line
                    if not line: # eof
                        break

                    if user_name not in line and op_name not in line:
                        if text != "":
                            text += f"\\n{line}"
                        else:
                            text = line

                    if user_name in line:
                        if text != "" and whos_typing != USER_ANON:
                            spamwriter.writerow([whos_typing, f"{text}"])
                            text = ""
                        whos_typing = USER_ANON

                    if op_name in line:
                        if text != "" and whos_typing != OP_ANON:
                            spamwriter.writerow([whos_typing, f"{text}"])
                            text = ""
                        whos_typing = OP_ANON

                if text != "":
                    spamwriter.writerow([whos_typing, f"{text}"])