PizzAI/finetune_dialogs_tool/main.py

# I know this code sucks and there are many repeating parts but it's just a simple temporary tool and I didn't bother to make it better
# This is not how something serious should be done

import os
import numpy as np
from order_items import possible_order_items
import json
from datetime import datetime

root= os.path.dirname(os.path.abspath(__file__))
vscode_available = any(os.access(os.path.join(path, "code"), os.X_OK) for path in os.environ["PATH"].split(os.pathsep))

def mkdir_if_not_exists_rel(path):
    if not os.path.exists(root + "/" + path):
        os.mkdir(root + "/" + path)

# create output dialogs folder if it does not exist
#if not os.path.exists(root + "/output_dialogs"):
#    os.mkdir(root + "/output_dialogs")

mkdir_if_not_exists_rel("output_dialogs")
mkdir_if_not_exists_rel("temp")


# if temp/dialog_input.py does not exist, do the processing instead of just opening it
if not os.path.exists(root + "/temp/dialog_input.py"):
    # read templates/dialog_input.txt
    with open(root + "/templates/dialog_input.txt", "r", encoding="utf-8") as f:
        dialog_input_template = f.read()


    # random phone number string. for example "123 456 789"
    phone_number = " ".join("".join(map(str, pack)) for pack in np.random.randint(0, 10, 9).reshape(-1, 3))

    payment_method = np.random.choice([
        "cash", 
        "card", 
        "Cash", 
        "Card", 
        "gotówka", 
        "karta", 
        "Gotówka", 
        "Karta",
        "gotówką",
        "kartą",
        "Gotówką",
        "Kartą",
    ])



    order_items = ""

    # randint high is exclusive so range 1 to 3 is actually 1 to 2 !!!

    # pizza
    pizza = possible_order_items["pizza"].copy()
    for _ in range(np.random.randint(1, 3)):
        item_idx = np.random.randint(0, len(pizza))

        order_items += f"    \"{np.random.randint(1, 4)}x Pizza {pizza[item_idx]}\",\n"

        del pizza[item_idx]

    # drink
    drink = possible_order_items["drink"].copy()
    for _ in range(np.random.randint(0, 3)):
        item_idx = np.random.randint(0, len(drink))

        order_items += f"    \"{np.random.randint(1, 4)}x {drink[item_idx]}\",\n"

        del drink[item_idx]

    # sauce
    sauce = possible_order_items["sauce"].copy()
    for _ in range(np.random.randint(0, 3)):
        item_idx = np.random.randint(0, len(sauce))

        order_items += f"    \"{np.random.randint(1, 4)}x {sauce[item_idx]}\",\n"

        del sauce[item_idx]

    order_items = order_items[:-2]



    dialog_input_template = dialog_input_template.replace("{{phone_number}}", phone_number)
    dialog_input_template = dialog_input_template.replace("{{payment_method}}", payment_method)
    dialog_input_template = dialog_input_template.replace("{{order_items}}", order_items)


    # save to temp/dialog_input.py
    with open(root + "/temp/dialog_input.py", "w", encoding="utf-8") as f:
        f.write(dialog_input_template)

while True:
    # open it in notepad and wait for user to close it
    # (or vscode if installed)
    if vscode_available:
        # open in new process of code. not the same window
        os.system("code " + root + "/temp/dialog_input.py --new-window --wait")
    else:
        os.system("notepad " + root + "/temp/dialog_input.py")





    # read and print
    dialog_input_f = open(root + "/temp/dialog_input.py", "r", encoding="utf-8")
    dialog_input = dialog_input_f.read()
    dialog_input_f.close()


    # execute part from the beginning to "## PYTHON END ##" as a python code here
    python_end_idx = dialog_input.find("## PYTHON END ##")
    phone_number, delivery_address, payment_method, order_items = None, None, None, None
    exec(dialog_input[:python_end_idx])
    # it will set variables: phone_number, payment_method and order_items

    # check if theres any empty variables
    not_ok = False
    for val in [phone_number, delivery_address, payment_method, order_items]:
        if val == "" or val is None:
            print("Some values are empty. Please fill them.")
            not_ok = True
            break
    if not_ok:
        continue

    # check the conversation

    dialog = []

    not_ok = False
    now_pizzeria = True
    for line in dialog_input[python_end_idx:].splitlines():
        if line.startswith("piz ="):
            if not now_pizzeria:
                print("Wrong order of pizzeria (piz) and customer (bot) messages.")
                not_ok = True
                break
            dialog.append(line[5:].strip())
            now_pizzeria = not now_pizzeria

        elif line.startswith("bot ="):
            if now_pizzeria:
                print("Wrong order of pizzeria (piz) and customer (bot) messages.")
                not_ok = True
                break
            dialog.append(line[5:].strip())
            now_pizzeria = not now_pizzeria

    if not_ok:
        continue

    # check if dialog starts on pizzeria side and ends on customer side (length is even)
    if len(dialog) % 2 != 0:
        print("Dialog must start on pizzeria side and end on customer side.")
        continue

    break




dialog[-1] += " CALLEND"

now_pizzeria = True
print("=============================================")
for line in dialog:
    print("🍕 -" if now_pizzeria else "🤖 -", line)
    now_pizzeria = not now_pizzeria
print("=============================================")



user_input = input("Everything seems okay. Press enter to save it as a new JSON file. Type anything to abort and exit.")
if user_input != "":
    print("Aborting and exiting.")
    exit()




# read templates/system_instructions.txt
with open(root + "/templates/system_instructions.txt", "r", encoding="utf-8") as f:
    system_instructions_template = f.read()

system_instructions_template = system_instructions_template.replace("{{phone_number}}", phone_number)
system_instructions_template = system_instructions_template.replace("{{delivery_location}}", delivery_address)
system_instructions_template = system_instructions_template.replace("{{payment_method}}", payment_method)

order_items_str = ""
for line in order_items:
    order_items_str += f"{line}\n"
order_items_str = order_items_str[:-1]

system_instructions_template = system_instructions_template.replace("{{order_items}}", order_items_str)


# use all of these to create chatgpt finetuning JSON file

training_dict = {
    "messages": []
}

# add system instructions
training_dict["messages"].append({
    "role": "system",
    "content": system_instructions_template
})

# add dialog
now_pizzeria = True
for line in dialog:
    training_dict["messages"].append({
        "role": "user" if now_pizzeria else "assistant",
        "content": line
    })
    now_pizzeria = not now_pizzeria


## find all json files consisting of number (0.json, 1.json .... 123.json)
#files = [f for f in os.listdir(root + "/output_dialogs") if f.endswith(".json") and f[:-5].isdigit()]
#
## get name for next file (number higher than the highest number in the folder)
#if len(files) == 0:
#    next_file = "0.json"
#else:
#    next_file = str(max([int(f[:-5]) for f in files]) + 1) + ".json"


next_file = datetime.now().strftime("%Y-%m-%d_%H-%M-%S") + ".json"

# save to output_dialogs/next_file
with open(root + "/output_dialogs/" + next_file, "w", encoding="utf-8") as f:
    json.dump(training_dict, f, indent=4, ensure_ascii=False)






# remove temp/dialog_input.py
os.remove(root + "/temp/dialog_input.py")


print(f"Saved as {next_file}")