流式传输批次结果：在验证码解决方案到达时对其进行处理

一批 500 个验证码任务的完成情况不均匀 - 有些在 8 秒内解决，有些需要 45 秒。在处理结果之前等待每个任务完成会浪费第一个和最后一个解决方案之间的时间。流式处理让您的下游管道在每个结果到达时就使用它。

流式传输与批处理

Python：用于流结果的异步生成器

使用 asyncio 和 aiohttp，每个解决方案都会通过异步生成器立即产生：

import asyncio
import aiohttp
import time

API_KEY = "YOUR_API_KEY"
SUBMIT_URL = "https://ocr.captchaai.com/in.php"
RESULT_URL = "https://ocr.captchaai.com/res.php"


async def submit_task(session, task_data):
    """Submit a single CAPTCHA task."""
    params = {
        "key": API_KEY,
        "method": task_data.get("method", "userrecaptcha"),
        "json": 1,
    }
    if params["method"] == "userrecaptcha":
        params["googlekey"] = task_data["sitekey"]
        params["pageurl"] = task_data["pageurl"]
    elif params["method"] == "turnstile":
        params["sitekey"] = task_data["sitekey"]
        params["pageurl"] = task_data["pageurl"]

    async with session.post(SUBMIT_URL, data=params) as resp:
        result = await resp.json(content_type=None)
        if result.get("status") != 1:
            return None, result.get("request", "unknown")
        return result["request"], None


async def poll_task(session, task_id, timeout=300):
    """Poll until solved or timeout."""
    start = time.monotonic()
    while time.monotonic() - start < timeout:
        await asyncio.sleep(5)
        params = {"key": API_KEY, "action": "get", "id": task_id, "json": 1}
        async with session.get(RESULT_URL, params=params) as resp:
            result = await resp.json(content_type=None)

        if result.get("request") == "CAPCHA_NOT_READY":
            continue
        if result.get("status") == 1:
            return result["request"], None
        return None, result.get("request", "unknown")

    return None, "TIMEOUT"


async def solve_one(session, index, task_data, semaphore):
    """Solve a single task within concurrency limits."""
    async with semaphore:
        start = time.monotonic()
        task_id, error = await submit_task(session, task_data)
        if error:
            return {"index": index, "status": "failed", "error": error, "time": 0}

        token, error = await poll_task(session, task_id)
        elapsed = time.monotonic() - start

        if token:
            return {"index": index, "status": "solved", "token": token, "time": round(elapsed, 1)}
        return {"index": index, "status": "failed", "error": error, "time": round(elapsed, 1)}


async def stream_results(tasks, max_concurrent=20):
    """
    Async generator that yields each result as it completes.
    Results arrive in completion order, not submission order.
    """
    semaphore = asyncio.Semaphore(max_concurrent)

    async with aiohttp.ClientSession() as session:
        pending = set()
        for i, task in enumerate(tasks):
            coro = solve_one(session, i, task, semaphore)
            pending.add(asyncio.ensure_future(coro))

        while pending:
            done, pending = await asyncio.wait(pending, return_when=asyncio.FIRST_COMPLETED)
            for future in done:
                yield future.result()


async def main():
    tasks = [
        {"sitekey": "SITE_KEY", "pageurl": f"https://example.com/page{i}"}
        for i in range(50)
    ]

    solved = 0
    failed = 0

    async for result in stream_results(tasks, max_concurrent=15):
        # Process each result immediately
        if result["status"] == "solved":
            solved += 1
            print(f"  [{solved + failed}/{len(tasks)}] Task {result['index']} SOLVED in {result['time']}s")

            # Use token immediately — don't wait for batch
            # await submit_form(result["token"])
            # await save_to_database(result)
        else:
            failed += 1
            print(f"  [{solved + failed}/{len(tasks)}] Task {result['index']} FAILED: {result['error']}")

    print(f"\nDone: {solved} solved, {failed} failed")


asyncio.run(main())

安装依赖项：

pip install aiohttp

JavaScript：EventEmitter 流模式

Node.js 使用事件驱动的方法——在解析时发出每个结果：

const { EventEmitter } = require("events");

const API_KEY = "YOUR_API_KEY";
const SUBMIT_URL = "https://ocr.captchaai.com/in.php";
const RESULT_URL = "https://ocr.captchaai.com/res.php";

class CaptchaStream extends EventEmitter {
  constructor(maxConcurrent = 15) {
    super();
    this.maxConcurrent = maxConcurrent;
    this.active = 0;
    this.queue = [];
    this.total = 0;
    this.completed = 0;
  }

  async submitAndPoll(index, taskData) {
    const params = new URLSearchParams({
      key: API_KEY,
      method: taskData.method || "userrecaptcha",
      googlekey: taskData.sitekey,
      pageurl: taskData.pageurl,
      json: "1",
    });

    const start = Date.now();
    const submitResp = await (await fetch(SUBMIT_URL, { method: "POST", body: params })).json();

    if (submitResp.status !== 1) {
      return { index, status: "failed", error: submitResp.request, time: 0 };
    }

    const taskId = submitResp.request;
    for (let i = 0; i < 60; i++) {
      await new Promise((r) => setTimeout(r, 5000));
      const url = `${RESULT_URL}?key=${API_KEY}&action=get&id=${taskId}&json=1`;
      const poll = await (await fetch(url)).json();

      if (poll.request === "CAPCHA_NOT_READY") continue;
      const elapsed = ((Date.now() - start) / 1000).toFixed(1);
      if (poll.status === 1) return { index, status: "solved", token: poll.request, time: elapsed };
      return { index, status: "failed", error: poll.request, time: elapsed };
    }
    return { index, status: "failed", error: "TIMEOUT", time: ((Date.now() - start) / 1000).toFixed(1) };
  }

  async processNext() {
    if (this.queue.length === 0 || this.active >= this.maxConcurrent) return;

    const { index, taskData } = this.queue.shift();
    this.active++;

    try {
      const result = await this.submitAndPoll(index, taskData);
      this.emit("result", result);
    } catch (err) {
      this.emit("result", { index, status: "failed", error: err.message });
    } finally {
      this.active--;
      this.completed++;

      if (this.completed === this.total) {
        this.emit("done");
      } else {
        this.processNext();
      }
    }
  }

  start(tasks) {
    this.total = tasks.length;
    this.queue = tasks.map((taskData, index) => ({ index, taskData }));

    // Launch initial batch
    const initial = Math.min(this.maxConcurrent, tasks.length);
    for (let i = 0; i < initial; i++) {
      this.processNext();
    }
    return this;
  }
}

// Usage
const tasks = Array.from({ length: 50 }, (_, i) => ({
  sitekey: "SITE_KEY",
  pageurl: `https://example.com/page${i}`,
}));

const stream = new CaptchaStream(15);
let solved = 0, failed = 0;

stream.on("result", (result) => {
  if (result.status === "solved") {
    solved++;
    console.log(`[${solved + failed}/${tasks.length}] Task ${result.index} SOLVED (${result.time}s)`);
    // Use token immediately
    // submitForm(result.token);
  } else {
    failed++;
    console.log(`[${solved + failed}/${tasks.length}] Task ${result.index} FAILED: ${result.error}`);
  }
});

stream.on("done", () => {
  console.log(`\nComplete: ${solved} solved, ${failed} failed`);
});

stream.start(tasks);

何时使用流式处理与全部收集

故障排除

常问问题

与批处理相比，流式处理是否会增加 API 调用？

不会——无论哪种方式都会发生相同数量的提交和轮询调用。流仅在应用程序处理每个结果时发生变化，而不是进行了多少次 API 调用。

流式传输时如何保持任务顺序？

每个结果都带有其原始的 index。如果顺序对于下游处理很重要，则缓冲区会产生排序结构并刷新连续运行（如 TCP 数据包重组）。

我可以将流式传输与检查点结合起来吗？

是的。在每个结果到达时将其附加到检查点文件。恢复时，加载检查点，过滤掉已完成的索引，并仅重新处理剩余的任务。

相关文章

• Kafka Captchaai 流式验证码处理
• 队列批量验证码优先处理

下一步

在验证码解决方案到达后立即对其进行处理 –”获取您的 CaptchaAI API 密钥并构建流媒体管道。

相关指南：

• CSV 驱动的批量验证码解决
• 基于队列的批量验证码处理
• 并行验证码求解