API Tutorials

Bash 脚本 + cURL + CaptchaAI:Shell CAPTCHA 自动化

裸 Linux 环境没装 Python 也没装 Node.js,怎么处理验证码?光靠 Bash 和 cURL 就够——CaptchaAI API 就是几个 POST/GET 请求,curl 直接调用。本文用纯 Bash 识别 reCAPTCHA v2/v3、Turnstile 和图片验证码,接进 crontab、CI、Docker。


为什么用 Bash + cURL 处理验证码

优势

优势 说明
零依赖 Bash、cURL 自带
足够轻 无需安装
对 cron 友好 标准 crontab 即可
CI/CD 天然适配 Docker、GitHub Actions 都能跑
能接管道 配合 jqgrepawk 使用

阿里云、腾讯云上的运维脚本本来就是纯 Shell,没必要为一个 Turnstile 装 Python。

环境准备

  1. Bash 4.0 及以上
  2. cURL(Linux/macOS 默认自带)
  3. jqapt install jqbrew install jq;国内服务器可先切 TUNA 镜像源
  4. CaptchaAI API Key点此获取

核心函数:提交任务与轮询结果

两个函数打底:submit_task 提交给 in.phppoll_result 轮询 res.php 直到拿到 token 或超时。后面各求解函数都是加参数封装。

提交任务函数

#!/bin/bash

CAPTCHAAI_URL="https://ocr.captchaai.com"

submit_task() {
    local api_key="$1"
    shift
    local params=("$@")

    local response
    response=$(curl -s -X POST "${CAPTCHAAI_URL}/in.php" \
        -d "key=${api_key}" \
        -d "json=1" \
        "${params[@]}")

    local status
    status=$(echo "$response" | jq -r '.status')
    local request
    request=$(echo "$response" | jq -r '.request')

    if [ "$status" != "1" ]; then
        echo "ERROR: Submit failed: $request" >&2
        return 1
    fi

    echo "$request"
}

轮询结果函数

poll_result() {
    local api_key="$1"
    local task_id="$2"
    local max_wait="${3:-300}"
    local interval="${4:-5}"

    local elapsed=0

    while [ "$elapsed" -lt "$max_wait" ]; do
        sleep "$interval"
        elapsed=$((elapsed + interval))

        local response
        response=$(curl -s "${CAPTCHAAI_URL}/res.php?key=${api_key}&action=get&id=${task_id}&json=1")

        local status
        status=$(echo "$response" | jq -r '.status')
        local request
        request=$(echo "$response" | jq -r '.request')

        if [ "$request" = "CAPCHA_NOT_READY" ]; then
            echo "Waiting... (${elapsed}s/${max_wait}s)" >&2
            continue
        fi

        if [ "$status" != "1" ]; then
            echo "ERROR: Solve failed: $request" >&2
            return 1
        fi

        echo "$request"
        return 0
    done

    echo "ERROR: Timeout after ${max_wait}s" >&2
    return 1
}

识别各类型验证码

同一套提交/轮询函数,参数不同:

reCAPTCHA v2

solve_recaptcha_v2() {
    local api_key="$1"
    local site_url="$2"
    local sitekey="$3"

    echo "Submitting reCAPTCHA v2..." >&2
    local task_id
    task_id=$(submit_task "$api_key" \
        -d "method=userrecaptcha" \
        -d "googlekey=${sitekey}" \
        -d "pageurl=${site_url}")

    if [ $? -ne 0 ]; then return 1; fi
    echo "Task ID: $task_id" >&2

    echo "Polling for solution..." >&2
    local token
    token=$(poll_result "$api_key" "$task_id")

    if [ $? -ne 0 ]; then return 1; fi
    echo "$token"
}

# Usage
API_KEY="YOUR_API_KEY"
TOKEN=$(solve_recaptcha_v2 "$API_KEY" \
    "https://staging.example.com/qa-login" \
    "6Le-wvkSAAAAAPBMRTvw0Q4Muexq9bi0DJwx_mJ-")

echo "Token: ${TOKEN:0:50}..."

Cloudflare Turnstile

参数名不同(key 而非 googlekey),其余同 v2:

solve_turnstile() {
    local api_key="$1"
    local site_url="$2"
    local sitekey="$3"

    local task_id
    task_id=$(submit_task "$api_key" \
        -d "method=turnstile" \
        -d "key=${sitekey}" \
        -d "pageurl=${site_url}")

    if [ $? -ne 0 ]; then return 1; fi

    poll_result "$api_key" "$task_id"
}

# Usage
TOKEN=$(solve_turnstile "$API_KEY" \
    "https://example.com/form" \
    "0x4AAAAAAAB5...")

reCAPTCHA v3(带 action 参数)

v3 没界面,只返回分数,多带 version=v3action

solve_recaptcha_v3() {
    local api_key="$1"
    local site_url="$2"
    local sitekey="$3"
    local action="${4:-verify}"

    local task_id
    task_id=$(submit_task "$api_key" \
        -d "method=userrecaptcha" \
        -d "googlekey=${sitekey}" \
        -d "pageurl=${site_url}" \
        -d "version=v3" \
        -d "action=${action}" \

    if [ $? -ne 0 ]; then return 1; fi

    poll_result "$api_key" "$task_id"
}

图片验证码

base64 方法,编码后作为 body 参数发送:

solve_image_captcha() {
    local api_key="$1"
    local image_path="$2"

    if [ ! -f "$image_path" ]; then
        echo "ERROR: File not found: $image_path" >&2
        return 1
    fi

    local base64_data
    base64_data=$(base64 -w 0 "$image_path" 2>/dev/null || base64 "$image_path")

    local task_id
    task_id=$(submit_task "$api_key" \
        -d "method=base64" \
        --data-urlencode "body=${base64_data}")

    if [ $? -ne 0 ]; then return 1; fi

    poll_result "$api_key" "$task_id"
}

# From URL
solve_image_from_url() {
    local api_key="$1"
    local image_url="$2"
    local tmp_file
    tmp_file=$(mktemp /tmp/captcha_XXXXXX.png)

    curl -s -o "$tmp_file" "$image_url"
    local result
    result=$(solve_image_captcha "$api_key" "$tmp_file")
    rm -f "$tmp_file"

    echo "$result"
}

# Usage
TEXT=$(solve_image_captcha "$API_KEY" "captcha.png")
echo "CAPTCHA text: $TEXT"

打包成一个可复用的脚本库

整理进一个文件,其他脚本 source 即可调用:

#!/bin/bash
# CaptchaAI Solver Library
# Source this file: source ./captchaai.sh

CAPTCHAAI_URL="https://ocr.captchaai.com"
CAPTCHAAI_POLL_INTERVAL=5
CAPTCHAAI_MAX_WAIT=300

captchaai_submit() {
    local api_key="$1"; shift
    local response
    response=$(curl -s -X POST "${CAPTCHAAI_URL}/in.php" \
        -d "key=${api_key}" -d "json=1" "$@")
    local status=$(echo "$response" | jq -r '.status')
    local request=$(echo "$response" | jq -r '.request')
    [ "$status" = "1" ] && echo "$request" || { echo "Submit: $request" >&2; return 1; }
}

captchaai_poll() {
    local api_key="$1" task_id="$2" elapsed=0
    while [ "$elapsed" -lt "$CAPTCHAAI_MAX_WAIT" ]; do
        sleep "$CAPTCHAAI_POLL_INTERVAL"
        elapsed=$((elapsed + CAPTCHAAI_POLL_INTERVAL))
        local resp=$(curl -s "${CAPTCHAAI_URL}/res.php?key=${api_key}&action=get&id=${task_id}&json=1")
        local req=$(echo "$resp" | jq -r '.request')
        local st=$(echo "$resp" | jq -r '.status')
        [ "$req" = "CAPCHA_NOT_READY" ] && continue
        [ "$st" = "1" ] && { echo "$req"; return 0; }
        echo "Solve: $req" >&2; return 1
    done
    echo "Timeout" >&2; return 1
}

captchaai_balance() {
    local api_key="$1"
    curl -s "${CAPTCHAAI_URL}/res.php?key=${api_key}&action=getbalance&json=1" | jq -r '.request'
}

captchaai_recaptcha_v2() {
    local key="$1" url="$2" sk="$3"
    local tid=$(captchaai_submit "$key" -d "method=userrecaptcha" -d "googlekey=$sk" -d "pageurl=$url") || return 1
    captchaai_poll "$key" "$tid"
}

captchaai_turnstile() {
    local key="$1" url="$2" sk="$3"
    local tid=$(captchaai_submit "$key" -d "method=turnstile" -d "sitekey=$sk" -d "pageurl=$url") || return 1
    captchaai_poll "$key" "$tid"
}

captchaai_image() {
    local key="$1" path="$2"
    local b64=$(base64 -w 0 "$path" 2>/dev/null || base64 "$path")
    local tid=$(captchaai_submit "$key" -d "method=base64" --data-urlencode "body=$b64") || return 1
    captchaai_poll "$key" "$tid"
}

在其他脚本里引用这个库

#!/bin/bash
source ./captchaai.sh

API_KEY="YOUR_API_KEY"

# Check balance
echo "Balance: $(captchaai_balance "$API_KEY")"

# Solve reCAPTCHA v2
TOKEN=$(captchaai_recaptcha_v2 "$API_KEY" \
    "https://staging.example.com/qa-login" \
    "6Le-wvkSAAAAAPBMRTvw0Q4Muexq9bi0DJwx_mJ-")

echo "Token: ${TOKEN:0:50}..."

拿到 token 之后怎么提交表单

submit_form_with_token() {
    local url="$1"
    local token="$2"
    shift 2

    curl -s -X POST "$url" \
        -d "g-recaptcha-response=${token}" \
        "$@"
}

# Usage: solve then submit
TOKEN=$(captchaai_recaptcha_v2 "$API_KEY" \
    "https://staging.example.com/qa-login" "SITEKEY")

RESPONSE=$(submit_form_with_token "https://staging.example.com/qa-login" \
    "$TOKEN" \
    -d "[email protected]" \
    -d "password=password")

echo "Response: $RESPONSE"

进阶用法:并发与重试

多站点并行识别,用后台任务提速

用后台任务(& + wait)并发发起,跑完统一收集:

#!/bin/bash
source ./captchaai.sh

API_KEY="YOUR_API_KEY"
RESULTS_DIR=$(mktemp -d)

# Define tasks
declare -A TASKS
TASKS["site-a"]="https://site-a.com|SITEKEY_A"
TASKS["site-b"]="https://site-b.com|SITEKEY_B"
TASKS["site-c"]="https://site-c.com|SITEKEY_C"

# Launch parallel solves
pids=()
for name in "${!TASKS[@]}"; do
    IFS='|' read -r url sitekey <<< "${TASKS[$name]}"
    (
        token=$(captchaai_recaptcha_v2 "$API_KEY" "$url" "$sitekey" 2>/dev/null)
        if [ $? -eq 0 ]; then
            echo "$token" > "${RESULTS_DIR}/${name}.token"
        else
            echo "FAILED" > "${RESULTS_DIR}/${name}.token"
        fi
    ) &
    pids+=($!)
done

# Wait for all
for pid in "${pids[@]}"; do
    wait "$pid"
done

# Collect results
echo "=== Results ==="
for name in "${!TASKS[@]}"; do
    token=$(cat "${RESULTS_DIR}/${name}.token")
    if [ "$token" = "FAILED" ]; then
        echo "$name: FAILED"
    else
        echo "$name: ${token:0:50}..."
    fi
done

rm -rf "$RESULTS_DIR"

失败了怎么重试:指数退避

ERROR_NO_SLOT_AVAILABLE 值得重试,其他错误重试也没用:

solve_with_retry() {
    local api_key="$1"
    local solve_cmd="$2"
    shift 2
    local max_retries="${1:-3}"

    local retryable_errors=("ERROR_NO_SLOT_AVAILABLE" "ERROR_CAPTCHA_UNSOLVABLE")
    local attempt=0

    while [ "$attempt" -le "$max_retries" ]; do
        if [ "$attempt" -gt 0 ]; then
            local delay=$((2 ** attempt + RANDOM % 3))
            echo "Retry $attempt/$max_retries after ${delay}s..." >&2
            sleep "$delay"
        fi

        local result
        result=$($solve_cmd "$api_key" "${@:2}")

        if [ $? -eq 0 ]; then
            echo "$result"
            return 0
        fi

        # Check if error is retryable
        local is_retryable=0
        for err in "${retryable_errors[@]}"; do
            if echo "$result" | grep -q "$err"; then
                is_retryable=1
                break
            fi
        done

        if [ "$is_retryable" -eq 0 ]; then
            echo "$result"
            return 1
        fi

        attempt=$((attempt + 1))
    done

    echo "Max retries exceeded" >&2
    return 1
}

接入自动化环境:crontab 与 Docker

crontab 定时任务

加一行即可,国内主机先确认时区是不是 Asia/Shanghai

# Edit crontab: crontab -e
# Run daily at 8 AM
0 8 * * * /path/to/captcha-automation.sh >> /var/log/captcha.log 2>&1

cron 脚本示例

#!/bin/bash
source /path/to/captchaai.sh

API_KEY="YOUR_API_KEY"
LOG_FILE="/var/log/captcha-$(date +%Y%m%d).log"

log() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] $*" >> "$LOG_FILE"; }

# Check balance first
BALANCE=$(captchaai_balance "$API_KEY")
log "Balance: $BALANCE"

if (( $(echo "$BALANCE < 1.0" | bc -l) )); then
    log "WARNING: Low balance!"
    exit 1
fi

# Solve and process
TOKEN=$(captchaai_recaptcha_v2 "$API_KEY" \
    "https://portal.example.com" "SITEKEY")

if [ $? -eq 0 ]; then
    log "Solved successfully"
    # Submit form, download data, etc.
    curl -s "https://portal.example.com/data" \
        -d "g-recaptcha-response=$TOKEN" \
        -o "/data/export-$(date +%Y%m%d).csv"
    log "Data exported"
else
    log "ERROR: Failed to solve CAPTCHA"
    exit 1
fi

打进 Docker 镜像

Alpine 加三个包,体积能控制在 10MB 左右:

FROM alpine:3.19

RUN apk add --no-cache bash curl jq

COPY captchaai.sh /usr/local/lib/captchaai.sh
COPY automation.sh /app/automation.sh

RUN chmod +x /app/automation.sh

CMD ["/app/automation.sh"]

常见报错与排查

错误 原因 处理方式
ERROR_WRONG_USER_KEY Key 无效 到控制台核对
ERROR_ZERO_BALANCE 余额不足 前往控制台充值
curl: (60) SSL certificate 缺 CA 证书 --cacert /path/to/ca-bundle.crt,测试用 -k
jq: command not found 没装 jq apt install jqbrew install jq
base64: invalid option -- 'w' macOS BSD 版 base64 改用 base64 file
响应为空 网络问题 -v 查看日志

常见问题

不装 jq,还能解析 JSON 吗?

能用 grep/sed 抠字段,但顺序一变易翻车,建议装 jq

国内云服务器跑 crontab 要注意什么?

先用 timedatectl 确认时区,输出记得重定向到日志——cron 没有终端,出错不落盘就没法排查。

curl 超时或返回空响应,怎么排查?

-v 看连接超时还是握手失败,再单测 action=getbalance 排除问题。

能直接搬进 GitHub Actions 或 GitLab CI 吗?

可以,captchaai.sh 是纯 Bash 函数,runner 装好 bash/curl/jq 即可 source 进 job。

API Key 除了写死在脚本里,还有更安全的存法吗?

用环境变量 export CAPTCHAAI_KEY="...",别把 Key 提交进版本控制,CI 优先用平台自带 Secrets。


相关阅读


从命令行识别验证码——获取你的 API Key,用 Bash 接进你现有的流程。

该文章已禁用评论。