python-anticaptcha

Client library for solving captchas with Anticaptcha.com support. The library requires Python >= 3.9.

The library is cyclically and automatically tested for proper operation. We are constantly making the best efforts for its effective operation.

In case of any problems with integration - read the documentation, create an issue, use Gitter or contact privately.

Getting Started

Install as standard Python package using:

pip install python-anticaptcha

For async support (FastAPI, aiohttp, Starlette, etc.):

pip install python-anticaptcha[async]

Usage

To use this library you need Anticaptcha.com API key.

You can pass the key explicitly or set the ANTICAPTCHA_API_KEY environment variable:

# Explicit key
client = AnticaptchaClient("my-api-key")

# Or set ANTICAPTCHA_API_KEY environment variable
client = AnticaptchaClient()

The client can be used as a context manager to ensure the underlying session is closed:

with AnticaptchaClient(api_key) as client:
    job = client.create_task(task)
    job.join()

Async Usage

For async frameworks, use AsyncAnticaptchaClient — the API mirrors the sync client but all methods are awaitable:

from python_anticaptcha import AsyncAnticaptchaClient, NoCaptchaTaskProxylessTask

api_key = '174faff8fbc769e94a5862391ecfd010'
site_key = '6Le-wvkSAAAAAPBMRTvw0Q4Muexq9bi0DJwx_mJ-'
url = 'https://www.google.com/recaptcha/api2/demo'

async with AsyncAnticaptchaClient(api_key) as client:
    task = NoCaptchaTaskProxylessTask(url, site_key)
    job = await client.create_task(task)
    await job.join()
    print(job.get_solution_response())

The full integration example is available in file examples/async_recaptcha_request.py.

Sync Usage

Solve recaptcha

Example snippet for Recaptcha:

from python_anticaptcha import AnticaptchaClient, NoCaptchaTaskProxylessTask

api_key = '174faff8fbc769e94a5862391ecfd010'
site_key = '6Le-wvkSAAAAAPBMRTvw0Q4Muexq9bi0DJwx_mJ-'  # grab from site
url = 'https://www.google.com/recaptcha/api2/demo'

client = AnticaptchaClient(api_key)
task = NoCaptchaTaskProxylessTask(url, site_key)
job = client.create_task(task)
job.join()
print(job.get_solution_response())

The full integration example is available in file examples/sync_recaptcha_request.py.

If you process the same page many times, to increase reliability you can specify whether the captcha is visible or not. This parameter is not required, as the system detects invisible sitekeys automatically. To provide that, pass is_invisible parameter to NoCaptchaTaskProxylessTask or NoCaptchaTask eg.:

from python_anticaptcha import AnticaptchaClient, NoCaptchaTaskProxylessTask

api_key = '174faff8fbc769e94a5862391ecfd010'
site_key = '6Lc-0DYUAAAAAOPM3RGobCfKjIE5STmzvZfHbbNx'  # grab from site
url = 'https://losangeles.craigslist.org/lac/kid/d/housekeeper-sitting-pet-care/6720136191.html'

client = AnticaptchaClient(api_key)
task = NoCaptchaTaskProxylessTask(url, site_key, is_invisible=True)
job = client.create_task(task)
job.join()
print(job.get_solution_response())

Solve text captcha

Example snippet for text captcha:

from python_anticaptcha import AnticaptchaClient, ImageToTextTask

api_key = '174faff8fbc769e94a5862391ecfd010'
client = AnticaptchaClient(api_key)

# From a file path:
task = ImageToTextTask('examples/captcha_ms.jpeg')

# Or from raw bytes:
# task = ImageToTextTask(open('examples/captcha_ms.jpeg', 'rb').read())

# Or from a file object:
# task = ImageToTextTask(open('examples/captcha_ms.jpeg', 'rb'))

job = client.create_task(task)
job.join()
print(job.get_captcha_text())

Solve funcaptcha

Example snippet for funcaptcha:

from python_anticaptcha import AnticaptchaClient, FunCaptchaTask, Proxy
UA = 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_10_1) AppleWebKit/537.36 ' \
     '(KHTML, like Gecko) Chrome/39.0.2171.95 Safari/537.36'

api_key = '174faff8fbc769e94a5862391ecfd010'
site_key = 'DE0B0BB7-1EE4-4D70-1853-31B835D4506B'  # grab from site
url = 'https://www.google.com/recaptcha/api2/demo'
proxy = Proxy.parse_url("socks5://login:password@123.123.123.123:1080")

client = AnticaptchaClient(api_key)
task = FunCaptchaTask(url, site_key, user_agent=UA, **proxy.to_kwargs())
job = client.create_task(task)
job.join()
print(job.get_token_response())

Monitor solve progress

You can pass an on_check callback to job.join() to monitor progress:

def progress(elapsed, status):
    print(f"Elapsed: {elapsed}s, status: {status}")

job = client.create_task(task)
job.join(on_check=progress)

Report incorrect image

Example snippet for reporting an incorrect image task:

from python_anticaptcha import AnticaptchaClient, ImageToTextTask

api_key = '174faff8fbc769e94a5862391ecfd010'
captcha_fp = open('examples/captcha_ms.jpeg', 'rb')
client = AnticaptchaClient(api_key)
task = ImageToTextTask(captcha_fp)
job = client.create_task(task)
job.join()
print(job.get_captcha_text())
job.report_incorrect_image()

Setup proxy

The library is not responsible for managing the proxy server. However, we point to the possibility of simply launching such a server by:

pip install mitmproxy
mitmweb -p 9190 -b 0.0.0.0 --ignore '.' --socks

Then in your application use something like:

proxy = Proxy.parse_url("socks5://123.123.123.123:9190")

We recommend entering IP-based access control for incoming addresses to proxy. IP address required by Anticaptcha.com is:

69.65.41.21
209.212.146.168

Error handling

In the event of an application error, the AnticaptchaException exception is thrown. To handle the exception, do the following:

from python_anticaptcha import AnticaptchaException, ImageToTextTask

try:
    # any actions
except AnticaptchaException as e:
    if e.error_code == 'ERROR_ZERO_BALANCE':
        notify_about_no_funds(e.error_id, e.error_code, e.error_description)
    else:
        raise

Note: The legacy misspelled AnticatpchaException alias is still available for backward compatibility.

Versioning

We use SemVer for versioning. For the versions available, see the tags on this repository.

Authors

• Adam Dobrawy - Initial work - ad-m

See also the list of contributors who participated in this project.

License

This project is licensed under the MIT License - see the LICENSE.md file for details