Strong AI. Elira2. Saving ChatGPT conversations
Hi all!
In this article, I'll show you how you can save conversations in ChatGPT along with your chat images!
Despite ChatGPT's existence for quite a long time, it still lacks the function of saving dialogs with images. Moreover, if there are images in the dialog, a save link is not created at all. That's why I decided to create a script that will allow me to save conversations directly through the ChatGPT web interface using Selenium.
Preparation
First you need to install the necessary dependencies. Create a file requirements.txt with content:
undetected-chromedriver
selenium
requests
Then install the dependencies using the command (in the console):pip install -r requirements.txt
Next, create a file credentials.txtin which we write down the login and password for automatic login to ChatGPT:
myemail@example.com
mypassword
Creating a script
Let's start by importing the necessary libraries and reading the credentials:
import os
import time
import requests
from undetected_chromedriver import Chrome, ChromeOptions
from selenium.webdriver.common.by import By
from selenium.webdriver.support.ui import WebDriverWait
from selenium.webdriver.support import expected_conditions as EC
# Чтение учетных данных из файла
with open('credentials.txt', 'r', encoding='utf-8') as f:
credentials = f.read().splitlines()
email = credentials[0]
password = credentials[1] To automate the browser (and bypass cloudflare) use Bible undetected_chromedriver.
We configure it and open the page for logging into ChatGPT:
# Настройка undetected_chromedriver
options = ChromeOptions()
options.add_argument('--no-sandbox')
options.add_argument('--disable-dev-shm-usage')
with Chrome(options=options) as driver:
driver.get('https://chat.openai.com/')
wait = WebDriverWait(driver, 10)
login_button = wait.until(EC.element_to_be_clickable((By.XPATH, '/html/body/div[1]/div[1]/div[2]/div[1]/div/div/button[1]/div/span')))
login_button.click()
# Ввод логина
email_field = wait.until(EC.presence_of_element_located((By.XPATH, '/html/body/div/div/main/section/div[2]/div[1]/input')))
email_field.send_keys(email)
next_button = driver.find_element(By.XPATH, '/html/body/div/div/main/section/div[2]/button')
next_button.click()
# Ввод пароля
password_field = wait.until(EC.presence_of_element_located((By.XPATH, '/html/body/div[1]/main/section/div/div/div/form/div[1]/div/div[2]/div/input')))
password_field.send_keys(password)
login_button = driver.find_element(By.XPATH, '/html/body/div[1]/main/section/div/div/div/form/div[2]/button')
login_button.click()
input("Выберите нужный диалог и нажмите Enter для продолжения...")We create folders for saving data and functions for saving text and downloading files:
# Создание папок для сохранения данных
if not os.path.exists('Диалог'):
os.makedirs('Диалог/Вложения')
# Функция для сохранения текста сообщения
def save_text(message, filename):
with open(filename, 'w', encoding='utf-8') as f:
f.write(message)
# Функция для скачивания файла по URL
def download_file(url, dest_path):
response = requests.get(url)
with open(dest_path, 'wb') as f:
f.write(response.content)Next, we create the main HTML file to save the dialogs. The HTML includes styles made in the colors of the ChatGPT web interface (for beauty):
# Создание основного HTML файла
html_content = """
<!DOCTYPE html>
<html lang="ru">
<head>
<meta charset="UTF-8">
<title>Диалог</title>
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/highlight.js/11.3.1/styles/default.min.css">
<script src="https://cdnjs.cloudflare.com/ajax/libs/highlight.js/11.3.1/highlight.min.js"></script>
<script>
hljs.highlightAll();
function copyCode(button) {
const code = button.parentElement.querySelector('code').innerText;
navigator.clipboard.writeText(code);
button.innerText="Скопировано";
setTimeout(() => { button.innerText="Копировать"; }, 2000);
}
function showModal(imageSrc) {
const modal = document.getElementById('imageModal');
const modalImg = document.getElementById('modalImg');
modal.style.display = 'block';
modalImg.src = imageSrc;
}
function closeModal() {
const modal = document.getElementById('imageModal');
modal.style.display = 'none';
}
</script>
<style>
body { font-family: Arial, sans-serif; line-height: 1.6; margin: 20px; background-color: #212121; color: #d1d5da; }
.user-message { color: #ececec !important; background-color: #2f2f2f !important; padding: 10px; border-radius: 8px; margin-bottom: 10px; }
.ai-message { color: #ececec !important; background-color: #212121; padding: 10px; border-radius: 8px; margin-bottom: 10px; }
.message-container { margin-bottom: 20px; }
.code-block { background-color: #000000 !important; border-left: 4px solid #00aaff; padding: 10px; position: relative; border-radius: 5px; }
.code-block code { display: block; background-color: #000000 !important; color: #f8f8f2; }
.copy-button { position: absolute; top: 10px; right: 10px; padding: 5px; background-color: #444654; border-radius: 8px; border: 1px solid #ccc; cursor: pointer; }
.image-container { width: 200px; display: inline-block; margin: 10px; }
.image-container img { width: 100%; cursor: pointer; border-radius: 5px; }
.image-modal { display: none; position: fixed; z-index: 1000; padding-top: 100px; left: 0; top: 0; width: 100%; height: 100%; overflow: auto; background-color: rgba(0,0,0,0.9); }
.image-modal-content { margin: auto; display: block; width: 80%; max-width: 700px; border-radius: 5px; }
.image-modal-content, .image-modal { animation-name: zoom; animation-duration: 0.6s; }
@keyframes zoom { from { transform: scale(0) } to { transform: scale(1) } }
.close { position: absolute; top: 50px; right: 35px; color: #f1f1f1; font-size: 40px; font-weight: bold; transition: 0.3s; }
.close:hover, .close:focus { color: #bbb; text-decoration: none; cursor: pointer; }
</style>
</head>
<body>
<div id="imageModal" class="image-modal">
<span class="close" onclick="closeModal()">×</span>
<img class="image-modal-content" id="modalImg">
</div>
"""
Now we begin collecting all messages and processing them. We will determine the “authorship” of the message and process the text, code and images separately:
# Инициализация счетчиков
total_messages = 0
user_messages = 0
ai_messages = 0
skipped_msg = 0
# Сбор всех сообщений
messages = driver.find_elements(By.XPATH, '//div[contains(@class, "group") and contains(@class, "flex") and not(contains(@class, "h-48"))]')
for idx, message in enumerate(messages):
if not message.text.strip():
skipped_msg += 1
continue # Пропускаем пустые сообщения
total_messages += 1
role = ""
# Проверка, является ли автор сообщения пользователем или ИИ
try:
author_role_element = message.find_element(By.XPATH, './/div[@data-message-author-role]')
author_role = author_role_element.get_attribute('data-message-author-role')
if author_role == 'user':
user_messages += 1
role = "Пользователь"
role_class = "user-message"
elif author_role == 'assistant':
ai_messages += 1
role = "ИИ"
role_class = "ai-message"
else:
continue # Пропускаем неизвестные роли
except:
continue # Пропускаем сообщения без атрибута data-message-author-role
# Заменяем роль на значок для ИИ
if role == "ИИ":
role_icon = "<img src="https://habr.com/ru/articles/GPT_icon.png" alt="ChatGPT Icon" width="20" height="20">"
role_html = f"<p>{role_icon}</p>"
else:
role_html = ""
html_content += f"<div class="message-container"><div class="{role_class}">{role_html}"
# Сохранение текста и кода
text_elements = message.find_elements(By.XPATH, './/div[contains(@class, "whitespace-pre-wrap")]')
if text_elements:
text_content = text_elements[0].text
save_text(text_content, f'Диалог/message_{idx}.txt')
code_elements = message.find_elements(By.XPATH, './/code[contains(@class, "!whitespace-pre hljs")]')
if code_elements:
for code_element in code_elements:
code_content = code_element.text
save_text(code_content, f'Диалог/code_{idx}.txt')
text_content = text_content.replace(code_content, f"<div class="code-block"><button class="copy-button" onclick='copyCode(this)'>Копировать</button><pre><code>{code_content}</code></pre></div>")
html_content += f"<p>{text_content}</p>"
print(f"Сообщение {idx - skipped_msg} сохранено (Текст: 1, Код: {len(code_elements)})")
# Сохранение изображений
img_elements = message.find_elements(By.TAG_NAME, 'img')
if img_elements:
for img_idx, img_element in enumerate(img_elements):
img_url = img_element.get_attribute('src')
img_path = f'Вложения/image_{idx}_{img_idx}.png'
download_file(img_url, f'Диалог/{img_path}')
html_content += f"<div class="image-container"><img src="https://habr.com/ru/articles/823428/{img_path}" onclick='showModal(\"{img_path}\")'></div>"
print(f"Изображения: {len(img_elements)}", end=", ")
else:
print(f"Изображения: 0", end=", ")
# Сохранение прикрепленных файлов
file_elements = message.find_elements(By.XPATH, './/a[contains(@href, "/attachment")]')
if file_elements:
for file_element in file_elements:
file_url = file_element.get_attribute('href')
file_name = file_element.text
file_path = f'Вложения/{file_name}'
download_file(file_url, f'Диалог/{file_path}')
html_content += f"<a href="https://habr.com/ru/articles/823428/{file_path}">{file_name}</a>"
print(f"Файлы: {len(file_elements)})")
else:
print(f"Файлы: 0)")
html_content += "</div></div>"Let's not forget to correctly close the HTML structure, save the main HTML file and display statistics:
# Закрытие HTML структуры
html_content += """
</body>
</html>
"""
# Сохранение основного HTML файла
with open('Диалог/диалог.html', 'w', encoding='utf-8') as f:
f.write(html_content)
# Вывод статистики
print(f"Всего сообщений: {total_messages}")
print(f"Сообщений пользователя: {user_messages}")
print(f"Сообщений ИИ: {ai_messages}")
print("Диалог сохранен.")Link to repo with source code:
https://github.com/Devastor/DevastorSaveDialogues_ChatGPT
You can also download the icon file there (to indicate the AI in the final dialogue)
What can (and should) be improved
Add the ability to save not only images, but also other attachment formats, for example, xls, csv tables (other attachments do not live long in the chat, unlike tables).
Fix “OSError: [WinError 6] Invalid descriptor”, which crashes after saving the dialogue (I didn’t succeed, although I didn’t try very hard, because I’m lazy and it doesn’t interfere with the script’s work, it should be quite simple).
Add saving all account dialogs in a loop (easy).
Add saving the name of the dialog (not as trivial as it seems – you need to somehow figure out how to catch the name of the container of the dialog that is selected).
Remove the extra “Copy” button from the upper right corner 🙂
Make logs disable (best written in args)
