pip install dddocr  # 安裝命令

pip install selenium

import time
import dddocr
from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver import ActionChains
from selenium.webdriver.support.ui import WebDriverWait
from webdriver_manager.chrome import ChromeDriverManager

???????# 初始化瀏覽器
def init_browser():
    options = webdriver.ChromeOptions()
    options.add_argument("--disable-blink-features=AutomationControlled")
    options.add_experimental_option("excludeSwitches", ["enable-automation"])
    driver = webdriver.Chrome(ChromeDriverManager().install(), options=options)
    driver.execute_script("Object.defineProperty(navigator, 'webdriver', {get: () => undefined})")
    return driver

def detect_gaps(bg_path, gap_path):
    ocr = dddocr.DdddOcr(show_ad=False)
    
    # 讀取背景圖和滑塊圖
    with open(bg_path, 'rb') as f:
        target_bytes = f.read()
    with open(gap_path, 'rb') as f:
        background_bytes = f.read()
    
    # 識別多個(gè)缺口位置
    results = ocr.slide_match(target_bytes, background_bytes, simple_target=True)
    
    # 過濾低置信度結(jié)果
    valid_gaps = [res for res in results if res['confidence'] > 0.7]
    
    # 按位置排序并選擇最可能的缺口
    if valid_gaps:
        # 選擇X坐標(biāo)最小的缺口（通常是最左側(cè)的正確缺口）
        target_gap = min(valid_gaps, key=lambda x: x['target'][0])
        return target_gap
    return None

def generate_move_track(distance):
    """生成擬人化移動軌跡"""
    track = []
    current = 0
    mid = distance * 0.8
    t = 0.2
    v = 0
    
    while current < distance:
        if current < mid:
            a = 2  # 加速階段
        else:
            a = -3  # 減速階段
            
        v0 = v
        v = v0 + a * t
        move = v0 * t + 0.5 * a * t * t
        current += move
        track.append(round(move))
    
    # 微調(diào)確保準(zhǔn)確到達(dá)
    overshoot = current - distance
    if overshoot > 0:
        track.append(-round(overshoot))
    
    return track

def execute_slide_verification(driver, gap_element, track):
    """執(zhí)行滑塊拖動操作"""
    action = ActionChains(driver)
    action.click_and_hold(gap_element).perform()
    
    # 分段移動模擬人工操作
    for move in track:
        action.move_by_offset(move, random.randint(-2, 2)).perform()
        time.sleep(random.uniform(0.01, 0.05))
    
    # 添加隨機(jī)抖動
    action.move_by_offset(random.randint(-3, 3), random.randint(-3, 3)).perform()
    time.sleep(0.1)
    action.release().perform()

def filter_valid_gaps(results, bg_width):
    """多維度過濾有效缺口"""
    valid_gaps = []
    
    for res in results:
        x, y = res['target'][0], res['target'][1]
        
        # 1. 置信度過濾
        if res['confidence'] < 0.65:
            continue
            
        # 2. 邊界過濾（排除邊緣10%區(qū)域）
        if x < bg_width * 0.1 or x > bg_width * 0.9:
            continue
            
        # 3. 缺口尺寸過濾
        gap_width = res['target'][2] - res['target'][0]
        if not (40 <= gap_width <= 70):  # 典型缺口寬度范圍
            continue
            
        # 4. 與其他缺口距離過濾
        if any(abs(x - gap['target'][0]) < 20 for gap in valid_gaps):
            continue
            
        valid_gaps.append(res)
    
    return valid_gaps

def select_most_likely_gap(gaps, previous_gaps=[]):
    """基于歷史記錄選擇最可能缺口"""
    if not gaps:
        return None
    
    # 優(yōu)先選擇水平位置最近的缺口
    if previous_gaps:
        last_x = previous_gaps[-1]['target'][0]
        return min(gaps, key=lambda x: abs(x['target'][0] - last_x))
    
    # 首次選擇最左側(cè)缺口（80%情況下正確）
    return min(gaps, key=lambda x: x['target'][0])

def solve_slide_captcha(driver, max_retry=3):
    """處理滑塊驗(yàn)證主函數(shù)"""
    retry_count = 0
    
    while retry_count < max_retry:
        try:
            # 1. 定位元素
            bg_element = WebDriverWait(driver, 10).until(
                EC.presence_of_element_located((By.CLASS_NAME, "captcha-bg"))
            )
            gap_element = WebDriverWait(driver, 10).until(
                EC.presence_of_element_located((By.CLASS_NAME, "captcha-gap"))
            )
            
            # 2. 保存驗(yàn)證圖片
            bg_location = bg_element.location
            bg_size = bg_element.size
            driver.save_screenshot("screenshot.png")
            
            # 3. 裁剪背景圖和滑塊圖
            crop_image("screenshot.png", "bg.png", 
                      bg_location['x'], bg_location['y'],
                      bg_size['width'], bg_size['height'])
            
            # 4. 缺口識別
            target_gap = detect_gaps("bg.png", "gap_template.png")
            if not target_gap:
                raise Exception("No valid gap detected")
            
            # 5. 計(jì)算移動距離（考慮縮放比例）
            scale_ratio = bg_size['width'] / target_gap['background'][0]
            move_distance = target_gap['target'][0] * scale_ratio - 25
            
            # 6. 生成移動軌跡
            track = generate_move_track(move_distance)
            
            # 7. 執(zhí)行滑塊操作
            execute_slide_verification(driver, gap_element, track)
            
            # 8. 驗(yàn)證結(jié)果檢查
            time.sleep(2)
            if "驗(yàn)證成功" in driver.page_source:
                return True
                
        except Exception as e:
            print(f"Attempt {retry_count+1} failed: {str(e)}")
            retry_count += 1
            # 點(diǎn)擊刷新按鈕
            driver.find_element(By.CLASS_NAME, "captcha-refresh").click()
            time.sleep(1)
    
    return False

def mask_browser_fingerprint(driver):
    # 修改WebGL供應(yīng)商
    driver.execute_script(
        "const getParameter = WebGLRenderingContext.getParameter;"
        "WebGLRenderingContext.prototype.getParameter = function(parameter) {"
        "  if (parameter === 37445) return 'Intel Inc.';"
        "  if (parameter === 37446) return 'Intel Iris OpenGL Engine';"
        "  return getParameter(parameter);"
        "};"
    )
    
    # 修改屏幕分辨率
    driver.execute_script(
        "Object.defineProperty(screen, 'width', {get: () => 1920});"
        "Object.defineProperty(screen, 'height', {get: () => 1080});"
    )
    
    # 修改時(shí)區(qū)
    driver.execute_script(
        "Object.defineProperty(Intl.DateTimeFormat.prototype, 'resolvedOptions', {"
        "  value: function() {"
        "    return { timeZone: 'Asia/Shanghai' };"
        "  }"
        "});"
    )

def human_like_mouse_movement(driver, element):
    """模擬人類鼠標(biāo)移動路徑"""
    action = ActionChains(driver)
    location = element.location_once_scrolled_into_view
    
    # 生成隨機(jī)起始點(diǎn)
    start_x = random.randint(100, 300)
    start_y = random.randint(300, 500)
    
    # 移動到元素上方隨機(jī)點(diǎn)
    action.move_by_offset(start_x, start_y).perform()
    time.sleep(random.uniform(0.2, 0.5))
    
    # 隨機(jī)曲線路徑
    points = generate_bezier_curve(
        start_x, start_y,
        location['x'] + 10, location['y'] + 10,
        points=30
    )
    
    for point in points:
        action.move_by_offset(point[0], point[1]).perform()
        time.sleep(random.uniform(0.01, 0.03))

def universal_captcha_solver(driver):
    """通用驗(yàn)證碼處理框架"""
    captcha_type = identify_captcha_type(driver)
    
    if captcha_type == "SLIDE":
        return solve_slide_captcha(driver)
    elif captcha_type == "TEXT":
        return solve_text_captcha(driver)
    elif captcha_type == "CLICK":
        return solve_click_captcha(driver)
    else:
        raise Exception("Unsupported captcha type")

# 圖像增強(qiáng)預(yù)處理
def enhance_image(image_path):
    img = cv2.imread(image_path)
    img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    img = cv2.GaussianBlur(img, (3, 3), 0)
    img = cv2.Canny(img, 100, 200)
    cv2.imwrite("enhanced.png", img)
    return "enhanced.png"

# 使用undetected_chromedriver
import undetected_chromedriver as uc

driver = uc.Chrome(version_main=105)
driver.get('https://target-website.com')

# 添加人類特征抖動
def humanize_track(track):
    # 隨機(jī)插入微停頓
    for _ in range(3):
        pos = random.randint(5, len(track)-5)
        track.insert(pos, 0)
        
    # 添加垂直抖動
    return [(x, random.randint(-1, 1)) for x in track]

def collect_training_data():
    """收集失敗案例用于模型改進(jìn)"""
    if not solve_slide_captcha(driver):
        save_failed_case(bg_image, gap_image, gap_position)
        # 定期使用新數(shù)據(jù)訓(xùn)練模型
        retrain_dddocr_model()

def identify_captcha_type(driver):
    """自動識別驗(yàn)證碼類型"""
    if driver.find_elements(By.CLASS_NAME, "slider-captcha"):
        return "SLIDE"
    elif driver.find_elements(By.ID, "captcha-image"):
        return "TEXT"
    elif driver.find_elements(By.CLASS_NAME, "point-captcha"):
        return "CLICK"
    else:
        return "UNKNOWN"