damai/scripts/captcha_solver.py

# coding: utf-8
"""
验证码处理模块
支持：图像验证码、滑块验证码
依赖：Pillow (PIL)
"""

import io
import logging
from dataclasses import dataclass
from enum import Enum, auto
from typing import Optional, Tuple

from selenium.webdriver.common.by import By
from selenium.webdriver.remote.webelement import WebElement

logger = logging.getLogger("damai.captcha")


class CaptchaType(Enum):
    NONE = auto()
    IMAGE = auto()       # 图形验证码
    SLIDER = auto()      # 滑块验证码
    CLICK = auto()       # 点选验证码
    UNKNOWN = auto()


@dataclass
class CaptchaResult:
    success: bool
    captcha_type: CaptchaType
    message: str = ""


def detect_captcha(driver) -> CaptchaType:
    """检测当前页面是否存在验证码"""
    # 检测滑块
    slider_selectors = [
        ".captcha-slider",
        ".slide-verify",
        "[class*='slider']",
        ".nc-lang-cn",  # 阿里云盾滑块
    ]
    for sel in slider_selectors:
        try:
            driver.find_element(By.CSS_SELECTOR, sel)
            logger.info(f"检测到滑块验证码: {sel}")
            return CaptchaType.SLIDER
        except Exception:
            continue

    # 检测图形验证码
    image_selectors = [
        ".captcha-img",
        "img[src*='captcha']",
        "img[src*='verify']",
        "#captcha_img",
    ]
    for sel in image_selectors:
        try:
            driver.find_element(By.CSS_SELECTOR, sel)
            logger.info(f"检测到图形验证码: {sel}")
            return CaptchaType.IMAGE
        except Exception:
            continue

    # 检测弹窗中是否有验证码关键词
    try:
        page_source = driver.page_source.lower()
        if "captcha" in page_source or "验证码" in page_source:
            return CaptchaType.UNKNOWN
    except Exception:
        pass

    return CaptchaType.NONE


def solve_slider(driver, slider_element: Optional[WebElement] = None) -> CaptchaResult:
    """
    处理滑块验证码
    简单实现：拖动滑块从左到右
    更精确的方案需要图像处理来识别缺口位置
    """
    from selenium.webdriver.common.action_chains import ActionChains

    selectors = [
        ".captcha-slider .slide-btn",
        ".slide-verify-slider-mask-item",
        ".nc-lang-cn .btn_slide",
        "[class*='slider'] [class*='btn']",
    ]

    slider = slider_element
    if not slider:
        for sel in selectors:
            try:
                slider = driver.find_element(By.CSS_SELECTOR, sel)
                break
            except Exception:
                continue

    if not slider:
        return CaptchaResult(False, CaptchaType.SLIDER, "找不到滑块元素")

    try:
        # 获取滑块轨道宽度
        track = slider.find_element(By.XPATH, "..")
        track_width = track.size["width"]
        slider_width = slider.size["width"]
        distance = track_width - slider_width

        if distance <= 0:
            distance = 300  # fallback

        actions = ActionChains(driver)
        actions.click_and_hold(slider)
        actions.pause(0.1)

        # 模拟人类拖动：分段加速减速
        steps = 20
        for i in range(steps):
            progress = (i + 1) / steps
            # 加速-减速曲线
            if progress < 0.3:
                offset = distance * progress * 0.5
            elif progress < 0.7:
                offset = distance * progress
            else:
                offset = distance * progress * 0.95
            move = int(offset - (distance * (i / steps) if i > 0 else 0))
            actions.move_by_offset(move, 0)
            actions.pause(0.02)

        actions.release()
        actions.perform()

        logger.info("滑块拖动完成")
        return CaptchaResult(True, CaptchaType.SLIDER)

    except Exception as e:
        logger.error(f"滑块处理失败: {e}")
        return CaptchaResult(False, CaptchaType.SLIDER, str(e))


def handle_captcha(driver) -> CaptchaResult:
    """自动检测并处理验证码"""
    captcha_type = detect_captcha(driver)

    if captcha_type == CaptchaType.NONE:
        return CaptchaResult(True, CaptchaType.NONE)

    if captcha_type == CaptchaType.SLIDER:
        return solve_slider(driver)

    if captcha_type == CaptchaType.IMAGE:
        logger.warning("图形验证码需要人工处理或接入 OCR 服务")
        return CaptchaResult(False, CaptchaType.IMAGE, "需要人工处理")

    logger.warning(f"未知验证码类型，需要人工处理")
    return CaptchaResult(False, captcha_type, "未知类型")