ChatGPT-4V easy script for automated visual description

gistfile1.txt

# Start chromium with chromium --remote-debugging-port=9230 
# Login on your chromium to chatgpt and navigate to gpt4 (with visual feature)
from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver.chrome.options import Options


import time
# Set up the webdriver with options
options = Options()
#options.add_argument("--remote-debugging-port=9230")
options.add_experimental_option("debuggerAddress", "127.0.0.1:9230")
driver = webdriver.Chrome(options=options)

from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver.common.keys import Keys

def send_form_with_image(driver, image_path):
    # Navigate to the page URL (replace with your actual URL)
    print("Bot : Refreshing Page (25s)")
    driver.back()
    time.sleep(5)  # wait for the previous page to load
    driver.forward()
    time.sleep(20)
    # 1. Find the textarea and input a message.
    print("Bot : Writing prompt (1s)")
    textarea = driver.find_element(By.ID, "prompt-textarea")
    textarea.send_keys("Your text here!")
    time.sleep(1)

    print("Bot : Inserting Image (20s)")
    file_input = driver.find_element(By.CSS_SELECTOR, "input[type='file']")
    file_input.send_keys(image_path)
    time.sleep(10)

    # 3. Find the send button and click it to submit the form.
    print("Bot : Submitting Form (1s)")
    send_button = driver.find_element(By.CSS_SELECTOR, "button[data-testid='send-button']")
    send_button.click()

    print("Bot : Waiting for response (200s)")
    time.sleep(200)

    markdown_divs = driver.find_elements(By.CSS_SELECTOR, "div.markdown")

    latest_paragraphs = []

    for div in markdown_divs:
        # For each div, get all <p> elements
        p_elements = div.find_elements(By.TAG_NAME, "p")
        
        # Get the last <p> from the list of p_elements
        if p_elements:
            latest_paragraphs.append(p_elements[-1])

    latest_markdown_text = "ERRROR"  
    if latest_paragraphs:
        last_paragraph = latest_paragraphs[-1]
        # Do something with last_paragraph, for instance, extract its text
        latest_markdown_text = last_paragraph.text
    #latest_markdown_text = latest_paragraphs[-1].text

    print("######### MARKDOWN #########")
    print(latest_markdown_text) # Or save it to a file or other storage.


# Execute the function
send_form_with_image(driver, "/path/to/img")
# Close the driver after completing the task
driver.quit()