Use a CAPTCHA solver with Selenium

1. 1.

   Install the dependencies

   Use Python 3.10 or newer with Chrome. Install the selenium package through pip, then install the official solvethree Python SDK from its downloadable wheel. Use the direct-download installation instructions in the Python integration; don't substitute an assumed package-registry installation.

2. 2.

   Configure the target profile

   Set up the owned form's hostname in the solvethree console and make sure its profile is active for your workspace. Sending a hostname in a task doesn't create that profile. The sample requests the login action; change it if your application uses a different action.

3. 3.

   Set the environment variables

   Set TARGET_URL to your full HTTPS form URL, SOLVETHREE_PROXY to your HTTP or HTTPS proxy URL, and SOLVETHREE_API_KEY to your workspace API key. Keep the key in the Python process, not in page JavaScript.

This Chrome sample requires an IP-allowlisted proxy with no username or password. It rejects proxy URLs containing credentials and doesn't implement authenticated proxy setup.

The task uses the hostname extracted from TARGET_URL, the login action and one proxy. After solve returns a ready result, Chrome starts with that same proxy and the returned userAgent. Before navigation, the sample also sets the returned secChUa header.

# Install selenium and the official solvethree Python SDK first.
import os
from urllib.parse import urlsplit
from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver.support.ui import WebDriverWait
from solvethree import Solvethree

target = os.environ["TARGET_URL"]  # Your configured HTTPS form page.
proxy_url = os.environ["SOLVETHREE_PROXY"]
proxy = urlsplit(proxy_url)
if proxy.username or proxy.password:
    raise ValueError("This Chrome example requires an IP-allowlisted proxy without URL credentials")
client = Solvethree(os.environ["SOLVETHREE_API_KEY"])
result = client.solve({"site": urlsplit(target).hostname, "action": "login", "proxy": proxy_url})
solution = result["solution"]
options = webdriver.ChromeOptions()
options.add_argument("--proxy-server=" + proxy_url)
options.add_argument("--user-agent=" + solution["userAgent"])
driver = webdriver.Chrome(options=options)
try:
    driver.execute_cdp_cmd("Network.enable", {})
    driver.execute_cdp_cmd("Network.setExtraHTTPHeaders", {
        "headers": {"sec-ch-ua": solution["secChUa"]}
    })
    driver.get(target)
    # Adapt the field and button to your application's form contract.
    field = WebDriverWait(driver, 20).until(
        lambda browser: browser.find_element(By.NAME, "g-recaptcha-response")
    )
    driver.execute_script("""
        arguments[0].value = arguments[1];
        arguments[0].dispatchEvent(new Event('input', {bubbles:true}));
    """, field, solution["token"])
    driver.find_element(By.CSS_SELECTOR, 'button[type="submit"]').click()
    # Check your application's success response here.
finally:
    driver.quit()

Solvethree.solve creates the task once and polls until it's ready or raises. Only then does the browser navigate to the form. The script waits for the token field, assigns solution.token, dispatches an input event and clicks the submit button. Chrome is closed in the finally block.

• TARGET_URL is your application's form page. The task's site is its hostname, not the full URL; reCAPTCHA requests don't use a url or type field.
• login is the sample action. Replace it with the action used by the application and its configured profile.
• By.NAME with g-recaptcha-response locates the demo textarea. Use the field or submission parameter your application actually reads. Assigning the value and dispatching an input event is the demo's handoff, not a universal reCAPTCHA callback.
• button[type="submit"] identifies the demo submit button. Replace it with an unambiguous selector for your form and retain the application's required session, CSRF and callback behavior.

The sample's 20-second explicit wait checks only that the textarea exists. After clicking Submit, add a conditional wait for your application's actual outcome, such as its success state or validation error. Use an overall browser-flow deadline separate from the solver timeout; don't replace state checks with fixed sleeps or casually mix implicit and explicit waits.

The solver task fails before Chrome starts.

Inspect the terminal task's errorCode and error. Distinguish a failed task from an HTTP request error.

Try this: Correct the reported input, profile or connectivity problem before retrying. Stop polling a failed task; unchanged invalid inputs won't improve with another request.

Selenium times out finding the textarea.

Confirm that Chrome reached the intended form and that the page actually contains the demo's named field.

Try this: Replace the demo locator with your application's token handoff and wait for the state that makes it available. A longer timeout won't fix the wrong selector.

The token is ready, but the application rejects submission.

Inspect the outgoing application request first. Check that it contains solution.token—not the task ID or an old value—and that the action, proxy, browser identity and session match the intended flow.

Try this: Correct the submitted parameter or application requirements before requesting another token. A ready task supplies a solution; the application's response determines whether submission was accepted.