Hardcoding CaptchaAI API calls throughout your scraping code creates tight coupling. If you swap providers, add a mock for testing, or want to route between services based on price, you rewrite everything. An abstract interface defines the contract once — implementations handle the details.
Why Abstract the Solver?
| Scenario | Without abstraction | With abstraction |
|---|---|---|
| Switch providers | Modify every file that calls the API | Swap one implementation class |
| Run tests | Mock HTTP calls or hit the live API | Inject a mock solver |
| A/B test providers | Duplicate all solving code | Create two implementations, route via config |
| Add logging/metrics | Modify every solve call | Wrap the interface with a decorator |
Python: Protocol-Based Interface
Python's Protocol defines structural typing — any class with matching methods satisfies the interface without explicit inheritance.
import requests
import time
from typing import Protocol, runtime_checkable
from dataclasses import dataclass
SUBMIT_URL = "https://ocr.captchaai.com/in.php"
RESULT_URL = "https://ocr.captchaai.com/res.php"
@dataclass
class SolveResult:
"""Standardised result from any solver."""
token: str
solve_time_ms: int
provider: str
cost_usd: float | None = None
@runtime_checkable
class CaptchaSolver(Protocol):
"""Contract that any CAPTCHA solving provider must satisfy."""
def solve_recaptcha_v2(self, sitekey: str, pageurl: str) -> SolveResult: ...
def solve_recaptcha_v3(self, sitekey: str, pageurl: str,
action: str, min_score: float) -> SolveResult: ...
def solve_turnstile(self, sitekey: str, pageurl: str) -> SolveResult: ...
def solve_hcaptcha(self, sitekey: str, pageurl: str) -> SolveResult: ...
def solve_image(self, base64_image: str) -> SolveResult: ...
def get_balance(self) -> float: ...
# --- CaptchaAI implementation ---
class CaptchaAISolver:
"""CaptchaAI implementation of the solver interface."""
def __init__(self, api_key: str):
self._api_key = api_key
def _submit_and_poll(self, params: dict, timeout: int = 180) -> tuple[str, int]:
params["key"] = self._api_key
params["json"] = 1
start = time.monotonic()
resp = requests.post(SUBMIT_URL, data=params, timeout=30).json()
if resp.get("status") != 1:
raise RuntimeError(f"Submit failed: {resp.get('request')}")
task_id = resp["request"]
while time.monotonic() - start < timeout:
time.sleep(5)
poll = requests.get(RESULT_URL, params={
"key": self._api_key, "action": "get",
"id": task_id, "json": 1,
}, timeout=15).json()
if poll.get("request") == "CAPCHA_NOT_READY":
continue
if poll.get("status") == 1:
elapsed_ms = int((time.monotonic() - start) * 1000)
return poll["request"], elapsed_ms
raise RuntimeError(f"Solve failed: {poll.get('request')}")
raise RuntimeError("Timeout")
def _result(self, token: str, solve_time_ms: int) -> SolveResult:
return SolveResult(token=token, solve_time_ms=solve_time_ms, provider="captchaai")
def solve_recaptcha_v2(self, sitekey: str, pageurl: str) -> SolveResult:
token, ms = self._submit_and_poll({
"method": "userrecaptcha", "googlekey": sitekey, "pageurl": pageurl,
})
return self._result(token, ms)
def solve_recaptcha_v3(self, sitekey: str, pageurl: str,
action: str = "verify", min_score: float = 0.3) -> SolveResult:
token, ms = self._submit_and_poll({
"method": "userrecaptcha", "version": "v3",
"googlekey": sitekey, "pageurl": pageurl,
"action": action, "min_score": str(min_score),
})
return self._result(token, ms)
def solve_turnstile(self, sitekey: str, pageurl: str) -> SolveResult:
token, ms = self._submit_and_poll({
"method": "turnstile", "sitekey": sitekey, "pageurl": pageurl,
})
return self._result(token, ms)
def solve_hcaptcha(self, sitekey: str, pageurl: str) -> SolveResult:
token, ms = self._submit_and_poll({
"method": "hcaptcha", "sitekey": sitekey, "pageurl": pageurl,
})
return self._result(token, ms)
def solve_image(self, base64_image: str) -> SolveResult:
token, ms = self._submit_and_poll({"method": "base64", "body": base64_image})
return self._result(token, ms)
def get_balance(self) -> float:
resp = requests.get(RESULT_URL, params={
"key": self._api_key, "action": "getbalance", "json": 1,
}, timeout=15).json()
return float(resp.get("request", 0))
# --- Mock implementation for testing ---
class MockSolver:
"""Test double that returns predictable tokens."""
def __init__(self, token: str = "MOCK_TOKEN_123"):
self._token = token
self.calls: list[dict] = []
def _record(self, method: str, **kwargs) -> SolveResult:
self.calls.append({"method": method, **kwargs})
return SolveResult(token=self._token, solve_time_ms=0, provider="mock")
def solve_recaptcha_v2(self, sitekey: str, pageurl: str) -> SolveResult:
return self._record("recaptcha_v2", sitekey=sitekey, pageurl=pageurl)
def solve_recaptcha_v3(self, sitekey: str, pageurl: str,
action: str = "verify", min_score: float = 0.3) -> SolveResult:
return self._record("recaptcha_v3", sitekey=sitekey, pageurl=pageurl)
def solve_turnstile(self, sitekey: str, pageurl: str) -> SolveResult:
return self._record("turnstile", sitekey=sitekey, pageurl=pageurl)
def solve_hcaptcha(self, sitekey: str, pageurl: str) -> SolveResult:
return self._record("hcaptcha", sitekey=sitekey, pageurl=pageurl)
def solve_image(self, base64_image: str) -> SolveResult:
return self._record("image")
def get_balance(self) -> float:
return 999.99
# --- Consumer code stays clean ---
def scrape_page(solver: CaptchaSolver, url: str, sitekey: str) -> str:
"""Works with any solver implementation."""
assert isinstance(solver, CaptchaSolver) # runtime check
result = solver.solve_recaptcha_v2(sitekey=sitekey, pageurl=url)
print(f"Solved by {result.provider} in {result.solve_time_ms}ms")
return result.token
# Production
solver = CaptchaAISolver("YOUR_API_KEY")
token = scrape_page(solver, "https://example.com", "SITEKEY")
# Testing
mock = MockSolver()
token = scrape_page(mock, "https://example.com", "SITEKEY")
assert mock.calls[0]["method"] == "recaptcha_v2"
JavaScript: Class-Based Interface
const SUBMIT_URL = "https://ocr.captchaai.com/in.php";
const RESULT_URL = "https://ocr.captchaai.com/res.php";
// Abstract interface (throws if not implemented)
class CaptchaSolver {
async solveRecaptchaV2(sitekey, pageurl) { throw new Error("Not implemented"); }
async solveRecaptchaV3(sitekey, pageurl, action, minScore) { throw new Error("Not implemented"); }
async solveTurnstile(sitekey, pageurl) { throw new Error("Not implemented"); }
async solveHCaptcha(sitekey, pageurl) { throw new Error("Not implemented"); }
async solveImage(base64Image) { throw new Error("Not implemented"); }
async getBalance() { throw new Error("Not implemented"); }
}
class CaptchaAISolver extends CaptchaSolver {
#apiKey;
constructor(apiKey) {
super();
this.#apiKey = apiKey;
}
async #submitAndPoll(params) {
const start = Date.now();
const body = new URLSearchParams({ key: this.#apiKey, json: "1", ...params });
const resp = await (await fetch(SUBMIT_URL, { method: "POST", body })).json();
if (resp.status !== 1) throw new Error(`Submit: ${resp.request}`);
const taskId = resp.request;
for (let i = 0; i < 60; i++) {
await new Promise((r) => setTimeout(r, 5000));
const url = `${RESULT_URL}?key=${this.#apiKey}&action=get&id=${taskId}&json=1`;
const poll = await (await fetch(url)).json();
if (poll.request === "CAPCHA_NOT_READY") continue;
if (poll.status === 1) {
return { token: poll.request, solveTimeMs: Date.now() - start, provider: "captchaai" };
}
throw new Error(`Solve: ${poll.request}`);
}
throw new Error("Timeout");
}
async solveRecaptchaV2(sitekey, pageurl) {
return this.#submitAndPoll({ method: "userrecaptcha", googlekey: sitekey, pageurl });
}
async solveTurnstile(sitekey, pageurl) {
return this.#submitAndPoll({ method: "turnstile", sitekey, pageurl });
}
async solveHCaptcha(sitekey, pageurl) {
return this.#submitAndPoll({ method: "hcaptcha", sitekey, pageurl });
}
async solveImage(base64Image) {
return this.#submitAndPoll({ method: "base64", body: base64Image });
}
async getBalance() {
const url = `${RESULT_URL}?key=${this.#apiKey}&action=getbalance&json=1`;
const resp = await (await fetch(url)).json();
return parseFloat(resp.request);
}
}
class MockSolver extends CaptchaSolver {
constructor(token = "MOCK_TOKEN") {
super();
this.token = token;
this.calls = [];
}
async solveRecaptchaV2(sitekey, pageurl) {
this.calls.push({ method: "recaptcha_v2", sitekey, pageurl });
return { token: this.token, solveTimeMs: 0, provider: "mock" };
}
async solveTurnstile(sitekey, pageurl) {
this.calls.push({ method: "turnstile", sitekey, pageurl });
return { token: this.token, solveTimeMs: 0, provider: "mock" };
}
async solveHCaptcha(sitekey, pageurl) {
this.calls.push({ method: "hcaptcha", sitekey, pageurl });
return { token: this.token, solveTimeMs: 0, provider: "mock" };
}
async solveImage(base64Image) {
this.calls.push({ method: "image" });
return { token: this.token, solveTimeMs: 0, provider: "mock" };
}
async getBalance() { return 999.99; }
}
// Consumer code — provider-agnostic
async function scrapePage(solver, url, sitekey) {
const result = await solver.solveRecaptchaV2(sitekey, url);
console.log(`Solved by ${result.provider} in ${result.solveTimeMs}ms`);
return result.token;
}
Decorator Pattern: Adding Cross-Cutting Concerns
Wrap the abstract interface to add behaviour without modifying implementations:
class LoggingSolver:
"""Decorator that logs every solve call."""
def __init__(self, inner: CaptchaSolver):
self._inner = inner
def solve_recaptcha_v2(self, sitekey: str, pageurl: str) -> SolveResult:
print(f"[LOG] solve_recaptcha_v2 sitekey={sitekey[:10]}...")
result = self._inner.solve_recaptcha_v2(sitekey, pageurl)
print(f"[LOG] Solved in {result.solve_time_ms}ms")
return result
# Delegate remaining methods via __getattr__
def __getattr__(self, name):
return getattr(self._inner, name)
# Stack decorators
solver = LoggingSolver(CaptchaAISolver("YOUR_API_KEY"))
Troubleshooting
| Issue | Cause | Fix |
|---|---|---|
isinstance check fails with Protocol |
Missing @runtime_checkable |
Add @runtime_checkable decorator to Protocol class |
| Mock doesn't satisfy interface | Missing method | Implement all Protocol methods in the mock |
__getattr__ delegates wrong method |
Decorator missing explicit method | Implement all critical methods explicitly |
| Type checker warns about missing methods | Partial implementation | Add all methods defined in the Protocol/base class |
| SolveResult fields inconsistent | Providers return different data | Normalize in each implementation's methods |
FAQ
Should I use ABC or Protocol in Python?
Use Protocol for structural typing — classes satisfy the interface if they have matching methods, without inheriting from anything. Use ABC when you want to enforce inheritance and shared base functionality. Protocol is more flexible for third-party code you don't control.
How do I handle provider-specific features?
Keep the abstract interface generic. Expose provider-specific options via keyword arguments or a params dict. The consumer code stays clean — only provider-aware code passes the extra params.
Can I combine this with the factory pattern?
Yes. The factory selects which implementation to create based on configuration or detection. The abstract interface ensures all implementations share the same contract. The factory returns CaptchaSolver, and consumer code never knows the concrete type.
Next Steps
Build provider-agnostic CAPTCHA solving — get your CaptchaAI API key and implement the abstract interface.
Related guides:
Discussions (0)
Join the conversation
Sign in to share your opinion.
Sign InNo comments yet.