做了一个樱花frp的自动签到脚本
- 内容介绍
- 文章标签
- 相关推荐
GitHub - RyanStarFox/SakuraFRP_Auto_AI_check: 自动签到樱花FRP，使用免费的glm4模型
自动签到樱花FRP，使用免费的glm4模型
之前想实现樱花frp签到的自动化，找了一个github上面的项目。但是最近樱花frp从九宫格验证码换成slider验证码了，刚刚又vibe coding了一个
需要的佬友可自取！
--【壹】--：
樱花的流量我没用完过
--【贰】--：
w我当时刚开始研究内网穿透的时候好像只有zerotier和frp，然后zerotier在大陆和校园网环境又不稳定，所以当时用了，一直延续到现在懒得动
现在发现cloudflared提供的内网穿透好方便
--【叁】--：
青龙可以用我这套，基于楼主的改写的，感谢楼主提供的源码！
步骤：
1. 安装依赖
在青龙面板的「依赖管理」中安装以下 Python 依赖：
playwright>=1.40.0
zhipuai>=2.0.0
pillow>=10.0.0
numpy>=1.24.0
captcha-recognizer>=1.0.2
重要： 安装完依赖后，需要在青龙面板的终端中执行以下命令安装 Playwright 浏览器：
# 进入青龙容器
docker exec -it qinglong bash
# 安装 Chromium 浏览器
playwright install chromium
# 如果提示缺少系统依赖，执行：
playwright install-deps chromium
2. 配置环境变量
在青龙面板的「环境变量」中添加以下配置：
| 变量名 | 说明 | 必填 | 示例 |
|---|---|---|---|
NATFRP_USERNAME |
NATFRP账号用户名 | your_username |
|
NATFRP_PASSWORD |
NATFRP账号密码 | your_password |
|
ZHIPU_API_KEY |
智谱AI的API Key | abcd1234... |
|
ZHIPU_MODEL_VISION |
视觉模型名称 | glm-4v-flash（默认） |
|
ZHIPU_MODEL_TEXT |
文本模型名称 | glm-4-flash（默认） |
获取智谱AI API Key：
- 访问 智谱AI开放平台
- 注册并登录账号
- 在控制台创建 API Key
- 免费模型无需付费即可使用
3. 添加定时任务
方式一：通过青龙面板界面添加
- 进入青龙面板 → 「定时任务」 → 「新建任务」
- 填写以下信息：
- 名称：
NATFRP自动签到 - 命令：
task ql_natfrp_checkin.py - 定时规则：
0 8 * * *（每天早上8点执行）
- 保存并启用任务
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
cron: 0 8 * * *
new Env('NATFRP自动签到');
"""
import sys
import io
import os
import time
import random
import re
import json
import base64
import traceback
from pathlib import Path
from datetime import datetime
from PIL import Image
from playwright.sync_api import sync_playwright
import numpy as np
# ========= 缓动函数（来自 PyAutoGUI，避免导入整个模块）=========
def easeInOutQuad(n):
"""缓动函数：先加速后减速"""
if n < 0.5:
return 2 * n * n
else:
n = n * 2 - 1
return -0.5 * (n * (n - 2) - 1)
def easeOutQuad(n):
"""缓动函数：快速启动，逐渐减速"""
return -n * (n - 2)
def easeInOutCubic(n):
"""缓动函数：更平滑的加速减速"""
if n < 0.5:
return 4 * n * n * n
else:
n = n * 2 - 2
return 0.5 * n * n * n + 1
# ========= 青龙面板环境变量配置 =========
# 在青龙面板中配置以下环境变量：
# NATFRP_USERNAME - NATFRP账号用户名
# NATFRP_PASSWORD - NATFRP账号密码
# ZHIPU_API_KEY - 智谱AI的API Key
# ZHIPU_MODEL_VISION - 视觉模型（可选，默认glm-4v-flash）
# ZHIPU_MODEL_TEXT - 文本模型（可选，默认glm-4-flash）
# ========= 配置 =========
domain = "www.natfrp.com"
target_url = f"https://{domain}/user/"
ALREADY_SIGNED_TEXT = "今天已经签到过啦"
# 青龙面板数据目录
QL_DATA_DIR = Path(os.getenv("QL_DATA_DIR", "/ql/data"))
STATE_FILE = QL_DATA_DIR / "scripts" / "natfrp_state.json"
# ========= AI服务类 =========
class AIService:
"""AI服务类，封装所有AI调用逻辑"""
def __init__(self):
"""初始化AI服务，从环境变量读取配置"""
self.api_key = os.getenv("ZHIPU_API_KEY", "")
self.model_vision = os.getenv("ZHIPU_MODEL_VISION", "glm-4v-flash")
self.model_text = os.getenv("ZHIPU_MODEL_TEXT", "glm-4-flash")
if not self.api_key:
raise ValueError("未找到ZHIPU_API_KEY环境变量，请在青龙面板中配置")
from zhipuai import ZhipuAI
self.client = ZhipuAI(api_key=self.api_key)
def safe_parse_json(self, text):
"""强力解析 AI 返回的 JSON 列表"""
try:
match = re.search(r'\[.*\]', text, re.DOTALL)
if match:
return json.loads(match.group())
return json.loads(text)
except Exception:
return None
def call_vision(self, image_bytes, prompt):
"""调用智谱多模态模型"""
base64_data = base64.b64encode(image_bytes).decode('utf-8')
try:
response = self.client.chat.completions.create(
model=self.model_vision,
messages=[{
"role": "user",
"content": [
{"type": "text", "text": prompt},
{"type": "image_url", "image_url": {"url": f"data:image/png;base64,{base64_data}"}}
]
}]
)
return response.choices[0].message.content.strip()
except Exception as e:
print(f"[ERROR] AI API 调用失败: {e}")
return ""
def identify_captcha_row(self, row_img_bytes, row_index):
"""分行识别逻辑"""
prompt = "这是验证码的一行图片，包含3个格子。请从左到右识别这3个格子的物体名称，只返回一个 JSON 字符串数组，例如：[\"猫\", \"狗\", \"汽车\"]。不要有任何解释文字。"
res = self.call_vision(row_img_bytes, prompt)
print(f"[AI] 第 {row_index} 行识别结果: {res}")
parsed = self.safe_parse_json(res)
if parsed and isinstance(parsed, list):
while len(parsed) < 3:
parsed.append("未知")
return parsed[:3]
return ["未知", "未知", "未知"]
def semantic_match(self, target, descriptions):
"""语义裁决逻辑"""
items_text = "\n".join([f"{i+1}. {d}" for i, d in enumerate(descriptions)])
prompt = f"题目是：找出图片中所有的【{target}】。\n当前 9 个格子的识别结果如下：\n{items_text}\n请根据描述，判断哪些序号（1-9）最符合题目要求？\n返回格式：只返回 JSON 数组，如 [1, 3, 5]。如果没有符合的，返回空数组 []。"
try:
response = self.client.chat.completions.create(
model=self.model_text,
messages=[{"role": "user", "content": prompt}]
)
content = response.choices[0].message.content.strip()
print(f"[AI] 语义裁决原始输出: {content}")
parsed = self.safe_parse_json(content)
return parsed if isinstance(parsed, list) else []
except Exception as e:
print(f"[ERROR] 语义匹配失败: {e}")
return []
# ========= 验证码识别函数 =========
def identify_gap_with_library(bg_img_bytes):
"""使用 captcha-recognizer 库识别滑块验证码缺口位置"""
try:
from captcha_recognizer.slider import Slider
bg_img = Image.open(io.BytesIO(bg_img_bytes))
bg_arr = np.array(bg_img)
box, confidence = Slider().identify(source=bg_arr, show=False)
if box and len(box) >= 4:
x1, y1, x2, y2 = box
gap_position = int(x1)
print(f"[DEBUG] captcha-recognizer 识别结果: 缺口位置={gap_position}px, 置信度={confidence:.2f}")
return gap_position
else:
print("[WARNING] captcha-recognizer 未识别到缺口")
return 0
except ImportError as e:
print(f"[WARNING] captcha-recognizer 库未安装: {e}")
return 0
except Exception as e:
print(f"[ERROR] captcha-recognizer 识别异常: {e}")
return 0
# ========= 验证码类型检测 =========
def detect_captcha_type(page):
"""检测验证码类型：九宫格或滑块"""
# 检查九宫格验证码
grid_visible = False
grid_selectors = [".geetest_table_box", ".geetest_grid"]
for selector in grid_selectors:
try:
if page.locator(selector).is_visible(timeout=2000):
grid_visible = True
break
except:
continue
# 检查滑块验证码
slider_visible = False
slider_selectors = [".geetest_slider", ".geetest_slider_button", ".geetest_canvas_bg"]
for selector in slider_selectors:
try:
if page.locator(selector).is_visible(timeout=2000):
slider_visible = True
break
except:
continue
if grid_visible:
print("[DEBUG] 检测到九宫格验证码")
return "grid"
elif slider_visible:
print("[DEBUG] 检测到滑块验证码")
return "slider"
else:
return "unknown"
# ========= 九宫格验证码处理 =========
def solve_geetest_multistep(page, ai_service):
"""使用AI服务处理九宫格验证码"""
print("[INFO] 开始处理九宫格验证码...")
img_container = page.locator(".geetest_table_box").first
if not img_container.is_visible(timeout=3000):
print("[DEBUG] 验证码容器不可见")
return False
# 识别题目
target_object = ""
tip_img = page.locator(".geetest_tip_img").first
if tip_img.is_visible(timeout=2000):
print("[DEBUG] 检测到图片提示，使用AI识别...")
try:
target_object = ai_service.call_vision(tip_img.screenshot(), "图中是什么物体？只回答物体名称，不要带标点。")
except Exception as e:
print(f"[ERROR] AI识别图片提示失败: {e}")
else:
tip_text_loc = page.locator(".geetest_tip_content").first
if tip_text_loc.is_visible(timeout=2000):
try:
target_object = tip_text_loc.inner_text()
except Exception as e:
print(f"[ERROR] 读取文本提示失败: {e}")
target_object = re.sub(r'[^\w]', '', target_object)
print(f">>> [Step 1] 识别题目为：【{target_object}】")
# 逐行抠图识别
all_descriptions = []
try:
grid_bytes = img_container.screenshot()
grid_img = Image.open(io.BytesIO(grid_bytes))
w, h = grid_img.size
row_h = h / 3
for i in range(3):
top = i * row_h
bottom = (i + 1) * row_h
row_crop = grid_img.crop((0, top, w, bottom))
buf = io.BytesIO()
row_crop.save(buf, format='PNG')
row_res = ai_service.identify_captcha_row(buf.getvalue(), i+1)
all_descriptions.extend(row_res)
except Exception as e:
print(f"[ERROR] 九宫格识别过程出错: {e}")
return False
# 语义匹配并模拟点击
try:
click_indices = ai_service.semantic_match(target_object, all_descriptions)
print(f">>> [Final] 最终决定点击序号: {click_indices}")
except Exception as e:
print(f"[ERROR] 语义匹配失败: {e}")
return False
if not click_indices:
print("[INFO] 未找到匹配项，刷新验证码...")
try:
refresh_btn = page.locator(".geetest_refresh").first
if refresh_btn.is_visible():
refresh_btn.click()
time.sleep(2)
except:
pass
return False
try:
box = img_container.bounding_box()
cell_w, cell_h = box['width']/3, box['height']/3
for idx in click_indices:
try:
val = int(idx)
if 1 <= val <= 9:
r, c = (val-1)//3, (val-1)%3
target_x = box['x'] + c*cell_w + cell_w/2
target_y = box['y'] + r*cell_h + cell_h/2
page.mouse.click(target_x, target_y)
time.sleep(random.uniform(0.3, 0.5))
except:
continue
except Exception as e:
print(f"[ERROR] 获取验证码容器位置失败: {e}")
return False
# 提交验证
for sel in [".geetest_commit", "text=确认", ".geetest_submit"]:
try:
btn = page.locator(sel).first
if btn.is_visible(timeout=2000):
btn.click()
return True
except:
continue
return False
# ========= 滑块验证码处理 =========
def solve_geetest_slider(page, ai_service):
"""使用AI服务处理滑块验证码"""
print("[INFO] 开始处理滑块验证码...")
# 查找滑块按钮
slider_button = None
slider_selectors = [".geetest_slider_button", ".geetest_slider_knob", ".geetest_btn"]
for selector in slider_selectors:
try:
btn = page.locator(selector).first
if btn.is_visible(timeout=1000):
slider_button = btn
print(f"[DEBUG] 找到滑块按钮: {selector}")
break
except:
continue
if not slider_button:
print("[ERROR] 未找到滑块按钮")
return False
button_box = slider_button.bounding_box()
if not button_box:
print("[ERROR] 无法获取滑块按钮位置")
return False
button_initial_x = button_box['x']
button_x = button_box['x'] + button_box['width'] / 2
button_y = button_box['y'] + button_box['height'] / 2
# 获取验证码图片
bg_img_bytes = None
bg_canvas = None
canvas_selectors = [".geetest_canvas_bg", "canvas.geetest_canvas_bg"]
for selector in canvas_selectors:
try:
canvas = page.locator(selector).first
if canvas.is_visible(timeout=1000):
bg_canvas = canvas
bg_img_bytes = canvas.screenshot()
print("[DEBUG] 成功获取背景图")
break
except:
continue
if not bg_img_bytes:
print("[ERROR] 无法获取验证码图片")
return False
# 使用 captcha-recognizer 识别缺口位置
gap_position = identify_gap_with_library(bg_img_bytes)
if gap_position <= 0:
print("[ERROR] 识别到的缺口位置无效")
return False
# 计算滑动距离
bg_canvas_box = None
if bg_canvas:
try:
bg_canvas_box = bg_canvas.bounding_box()
except:
pass
if bg_canvas_box:
bg_canvas_x = bg_canvas_box['x']
offset = button_initial_x - bg_canvas_x
drag_distance_base = gap_position + offset
human_error = random.uniform(-5.0, 5.0)
drag_distance = drag_distance_base + human_error
target_x = button_x + drag_distance
else:
drag_distance_base = gap_position
human_error = random.uniform(-5.0, 5.0)
drag_distance = drag_distance_base + human_error
target_x = button_x + drag_distance
print(f"[INFO] 滑动距离: {drag_distance:.1f}px, 目标位置: {target_x:.1f}px")
# 执行拖动
try:
page.mouse.move(button_x, button_y)
time.sleep(random.uniform(0.1, 0.2))
page.mouse.down()
time.sleep(random.uniform(0.1, 0.2))
# 使用缓动函数模拟人类拖动
steps = random.randint(20, 30)
easing_functions = [
easeInOutQuad,
easeOutQuad,
easeInOutCubic,
]
easing_func = random.choice(easing_functions)
for i in range(steps):
progress = easing_func(i / steps)
jitter_x = random.uniform(-1.5, 1.5)
jitter_y = random.uniform(-2, 2)
current_x = button_x + drag_distance * progress + jitter_x
current_y = button_y + jitter_y
page.mouse.move(current_x, current_y)
if i < steps * 0.3:
time.sleep(random.uniform(0.005, 0.015))
elif i > steps * 0.7:
time.sleep(random.uniform(0.02, 0.04))
else:
time.sleep(random.uniform(0.01, 0.025))
# 超调效果
if random.random() > 0.5:
overshoot = random.uniform(2, 5)
page.mouse.move(target_x + overshoot, button_y + random.uniform(-1, 1))
time.sleep(random.uniform(0.05, 0.1))
page.mouse.move(target_x, button_y)
time.sleep(random.uniform(0.15, 0.25))
page.mouse.up()
time.sleep(random.uniform(0.5, 1.0))
print("[DEBUG] 滑块拖动完成")
time.sleep(2)
# 检查验证结果
captcha_gone = True
try:
if page.locator(".geetest_slider").is_visible(timeout=1000):
captcha_gone = False
except:
pass
return captcha_gone
except Exception as e:
print(f"[ERROR] 拖动滑块失败: {e}")
return False
# ========= 主逻辑 =========
def find_signed_text_locator(page, timeout=3000):
"""查找已签到文本"""
try:
loc = page.get_by_text(ALREADY_SIGNED_TEXT).first
if loc.is_visible(timeout=timeout):
return loc
except:
pass
return None
def main():
"""主函数"""
print("=" * 60)
print("NATFRP 自动签到脚本 - 青龙面板版")
print("=" * 60)
# 从环境变量获取配置
username = os.getenv("NATFRP_USERNAME", "")
password = os.getenv("NATFRP_PASSWORD", "")
if not username or not password:
print("[ERROR] 未配置账号信息，请在青龙面板中设置环境变量：")
print(" - NATFRP_USERNAME: NATFRP账号用户名")
print(" - NATFRP_PASSWORD: NATFRP账号密码")
return
# 初始化AI服务
try:
ai_service = AIService()
print(f"[INFO] AI服务初始化成功")
print(f" - 视觉模型: {ai_service.model_vision}")
print(f" - 文本模型: {ai_service.model_text}")
except Exception as e:
print(f"[ERROR] AI服务初始化失败: {e}")
print("[提示] 请在青龙面板中配置 ZHIPU_API_KEY 环境变量")
return
with sync_playwright() as p:
browser = p.chromium.launch(headless=True, slow_mo=100)
# 尝试使用缓存的登录状态
storage_state = None
if STATE_FILE.exists():
try:
print(f"[INFO] 发现登录状态缓存: {STATE_FILE}")
storage_state = str(STATE_FILE)
except Exception as e:
print(f"[WARNING] 读取登录状态缓存失败: {e}")
context = browser.new_context(storage_state=storage_state)
page = context.new_page()
page.set_viewport_size({"width": 1280, "height": 900})
print(f"[INFO] 正在访问: {target_url}")
try:
page.goto(target_url, timeout=30000)
print(f"[DEBUG] 页面加载完成")
except Exception as e:
print(f"[ERROR] 页面访问失败: {e}")
browser.close()
return
# 登录判断
current_url = page.url
is_logged_in = True
try:
username_input_visible = page.locator("#username").is_visible(timeout=2000)
except:
username_input_visible = False
if "login" in current_url or username_input_visible:
is_logged_in = False
print("[INFO] 检测到需要登录")
try:
print("[INFO] 正在填写登录信息...")
page.fill("#username", username)
page.fill("#password", password)
page.click("#login")
print("[DEBUG] 等待登录完成...")
page.wait_for_selector("text=账号信息", timeout=10000)
# 保存登录状态
try:
context.storage_state(path=str(STATE_FILE))
print(f"[SUCCESS] 登录成功，状态已保存到: {STATE_FILE}")
except Exception as e:
print(f"[WARNING] 保存登录状态失败: {e}")
print("[SUCCESS] 登录成功")
is_logged_in = True
except Exception as e:
print(f"[ERROR] 登录失败: {e}")
browser.close()
return
else:
print("[INFO] 已登录状态（使用缓存）")
# 处理18岁弹窗
try:
btn_18 = page.get_by_text("是，我已满18岁")
if btn_18.is_visible(timeout=3000):
print("[DEBUG] 检测到18岁确认弹窗，正在点击...")
btn_18.click()
time.sleep(1)
except:
pass
# 签到
print("[DEBUG] 开始检查签到状态...")
signed_locator = find_signed_text_locator(page)
if signed_locator:
print("[SUCCESS] 今日已签到")
else:
print("[DEBUG] 未检测到已签到状态，查找签到按钮...")
sign_btn = page.get_by_text("点击这里签到")
try:
sign_btn_visible = sign_btn.is_visible(timeout=3000)
except:
sign_btn_visible = False
if sign_btn_visible:
print("[INFO] 点击签到按钮...")
sign_success = False
try:
sign_btn.click()
print("[DEBUG] 已点击签到按钮，等待验证码加载...")
# 等待15秒让验证码加载
print("[INFO] 等待15秒让验证码完全加载...")
for i in range(30):
time.sleep(0.5)
if (i + 1) % 10 == 0:
print(f"[DEBUG] 已等待 {(i+1)*0.5:.1f} 秒...")
signed_check = find_signed_text_locator(page, timeout=500)
if signed_check:
print(f"[SUCCESS] 签到完成（无需验证码）！")
sign_success = True
break
if not sign_success:
# 检查验证码
captcha_type = detect_captcha_type(page)
if captcha_type != "unknown":
print(f"[INFO] 检测到验证码类型: {captcha_type}")
max_attempts = 3
for attempt in range(1, max_attempts + 1):
print(f"[DEBUG] 第 {attempt}/{max_attempts} 次尝试处理验证码...")
try:
if captcha_type == "grid":
captcha_result = solve_geetest_multistep(page, ai_service)
elif captcha_type == "slider":
captcha_result = solve_geetest_slider(page, ai_service)
else:
captcha_result = False
if captcha_result:
print("[INFO] 验证码处理成功")
time.sleep(3)
# 检查是否签到成功
signed_check = find_signed_text_locator(page, timeout=2000)
if signed_check:
print("[SUCCESS] 签到完成！")
sign_success = True
break
else:
print("[WARNING] 验证码处理失败，重试...")
time.sleep(2)
# 重新检测验证码类型
captcha_type = detect_captcha_type(page)
if captcha_type == "unknown":
print("[DEBUG] 验证码已消失，检查签到状态...")
signed_check = find_signed_text_locator(page, timeout=2000)
if signed_check:
print("[SUCCESS] 签到完成！")
sign_success = True
break
except Exception as e:
print(f"[ERROR] 验证码处理异常: {e}")
traceback.print_exc()
else:
print("[WARNING] 未检测到验证码")
if not sign_success:
print("[ERROR] 签到失败：超时或验证码处理失败")
except Exception as e:
print(f"[ERROR] 签到过程出错: {e}")
traceback.print_exc()
else:
print("[ERROR] 未找到签到按钮")
print("[INFO] 脚本运行结束")
print("=" * 60)
browser.close()
if __name__ == "__main__":
main()
--【肆】--：
感谢分享，已经用上了
--【伍】--：
感谢大佬
--【陆】--：
用上了 感谢大佬！，大佬 这家也有签到拿会员的能看看吗 chmlfrp
--【柒】--：
大佬nb，感谢分享
--【捌】--：
附带成功的图片
image578×1716 147 KB
--【玖】--：
太强了佬
--【拾】--：
佬友牛逼！
--【拾壹】--：
感谢分享，好久没用了这个
--【拾贰】--：
很好用。不过有时候会失败，把失败日志给AI改了几轮，部署到了服务器上弄了个定时任务，又加了个邮件通知功能，每天手机微信就能自动收到签到通知。
64de4137d11ac3505a79a523527da39e.jpg610×1315 112 KB image.png614×1324 102 KB--【拾叁】--：
也可以试试vibe coding?我基本上纯vibe coding，是可以实现的
--【拾肆】--：
说实话我更期待chmlfrp的，樱花我不怎么用不用
--【拾伍】--：
佬友nb
GitHub - RyanStarFox/SakuraFRP_Auto_AI_check: 自动签到樱花FRP，使用免费的glm4模型
自动签到樱花FRP，使用免费的glm4模型
之前想实现樱花frp签到的自动化，找了一个github上面的项目。但是最近樱花frp从九宫格验证码换成slider验证码了，刚刚又vibe coding了一个
需要的佬友可自取！
--【壹】--：
樱花的流量我没用完过
--【贰】--：
w我当时刚开始研究内网穿透的时候好像只有zerotier和frp，然后zerotier在大陆和校园网环境又不稳定，所以当时用了，一直延续到现在懒得动
现在发现cloudflared提供的内网穿透好方便
--【叁】--：
青龙可以用我这套，基于楼主的改写的，感谢楼主提供的源码！
步骤：
1. 安装依赖
在青龙面板的「依赖管理」中安装以下 Python 依赖：
playwright>=1.40.0
zhipuai>=2.0.0
pillow>=10.0.0
numpy>=1.24.0
captcha-recognizer>=1.0.2
重要： 安装完依赖后，需要在青龙面板的终端中执行以下命令安装 Playwright 浏览器：
# 进入青龙容器
docker exec -it qinglong bash
# 安装 Chromium 浏览器
playwright install chromium
# 如果提示缺少系统依赖，执行：
playwright install-deps chromium
2. 配置环境变量
在青龙面板的「环境变量」中添加以下配置：
| 变量名 | 说明 | 必填 | 示例 |
|---|---|---|---|
NATFRP_USERNAME |
NATFRP账号用户名 | your_username |
|
NATFRP_PASSWORD |
NATFRP账号密码 | your_password |
|
ZHIPU_API_KEY |
智谱AI的API Key | abcd1234... |
|
ZHIPU_MODEL_VISION |
视觉模型名称 | glm-4v-flash（默认） |
|
ZHIPU_MODEL_TEXT |
文本模型名称 | glm-4-flash（默认） |
获取智谱AI API Key：
- 访问 智谱AI开放平台
- 注册并登录账号
- 在控制台创建 API Key
- 免费模型无需付费即可使用
3. 添加定时任务
方式一：通过青龙面板界面添加
- 进入青龙面板 → 「定时任务」 → 「新建任务」
- 填写以下信息：
- 名称：
NATFRP自动签到 - 命令：
task ql_natfrp_checkin.py - 定时规则：
0 8 * * *（每天早上8点执行）
- 保存并启用任务
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
cron: 0 8 * * *
new Env('NATFRP自动签到');
"""
import sys
import io
import os
import time
import random
import re
import json
import base64
import traceback
from pathlib import Path
from datetime import datetime
from PIL import Image
from playwright.sync_api import sync_playwright
import numpy as np
# ========= 缓动函数（来自 PyAutoGUI，避免导入整个模块）=========
def easeInOutQuad(n):
"""缓动函数：先加速后减速"""
if n < 0.5:
return 2 * n * n
else:
n = n * 2 - 1
return -0.5 * (n * (n - 2) - 1)
def easeOutQuad(n):
"""缓动函数：快速启动，逐渐减速"""
return -n * (n - 2)
def easeInOutCubic(n):
"""缓动函数：更平滑的加速减速"""
if n < 0.5:
return 4 * n * n * n
else:
n = n * 2 - 2
return 0.5 * n * n * n + 1
# ========= 青龙面板环境变量配置 =========
# 在青龙面板中配置以下环境变量：
# NATFRP_USERNAME - NATFRP账号用户名
# NATFRP_PASSWORD - NATFRP账号密码
# ZHIPU_API_KEY - 智谱AI的API Key
# ZHIPU_MODEL_VISION - 视觉模型（可选，默认glm-4v-flash）
# ZHIPU_MODEL_TEXT - 文本模型（可选，默认glm-4-flash）
# ========= 配置 =========
domain = "www.natfrp.com"
target_url = f"https://{domain}/user/"
ALREADY_SIGNED_TEXT = "今天已经签到过啦"
# 青龙面板数据目录
QL_DATA_DIR = Path(os.getenv("QL_DATA_DIR", "/ql/data"))
STATE_FILE = QL_DATA_DIR / "scripts" / "natfrp_state.json"
# ========= AI服务类 =========
class AIService:
"""AI服务类，封装所有AI调用逻辑"""
def __init__(self):
"""初始化AI服务，从环境变量读取配置"""
self.api_key = os.getenv("ZHIPU_API_KEY", "")
self.model_vision = os.getenv("ZHIPU_MODEL_VISION", "glm-4v-flash")
self.model_text = os.getenv("ZHIPU_MODEL_TEXT", "glm-4-flash")
if not self.api_key:
raise ValueError("未找到ZHIPU_API_KEY环境变量，请在青龙面板中配置")
from zhipuai import ZhipuAI
self.client = ZhipuAI(api_key=self.api_key)
def safe_parse_json(self, text):
"""强力解析 AI 返回的 JSON 列表"""
try:
match = re.search(r'\[.*\]', text, re.DOTALL)
if match:
return json.loads(match.group())
return json.loads(text)
except Exception:
return None
def call_vision(self, image_bytes, prompt):
"""调用智谱多模态模型"""
base64_data = base64.b64encode(image_bytes).decode('utf-8')
try:
response = self.client.chat.completions.create(
model=self.model_vision,
messages=[{
"role": "user",
"content": [
{"type": "text", "text": prompt},
{"type": "image_url", "image_url": {"url": f"data:image/png;base64,{base64_data}"}}
]
}]
)
return response.choices[0].message.content.strip()
except Exception as e:
print(f"[ERROR] AI API 调用失败: {e}")
return ""
def identify_captcha_row(self, row_img_bytes, row_index):
"""分行识别逻辑"""
prompt = "这是验证码的一行图片，包含3个格子。请从左到右识别这3个格子的物体名称，只返回一个 JSON 字符串数组，例如：[\"猫\", \"狗\", \"汽车\"]。不要有任何解释文字。"
res = self.call_vision(row_img_bytes, prompt)
print(f"[AI] 第 {row_index} 行识别结果: {res}")
parsed = self.safe_parse_json(res)
if parsed and isinstance(parsed, list):
while len(parsed) < 3:
parsed.append("未知")
return parsed[:3]
return ["未知", "未知", "未知"]
def semantic_match(self, target, descriptions):
"""语义裁决逻辑"""
items_text = "\n".join([f"{i+1}. {d}" for i, d in enumerate(descriptions)])
prompt = f"题目是：找出图片中所有的【{target}】。\n当前 9 个格子的识别结果如下：\n{items_text}\n请根据描述，判断哪些序号（1-9）最符合题目要求？\n返回格式：只返回 JSON 数组，如 [1, 3, 5]。如果没有符合的，返回空数组 []。"
try:
response = self.client.chat.completions.create(
model=self.model_text,
messages=[{"role": "user", "content": prompt}]
)
content = response.choices[0].message.content.strip()
print(f"[AI] 语义裁决原始输出: {content}")
parsed = self.safe_parse_json(content)
return parsed if isinstance(parsed, list) else []
except Exception as e:
print(f"[ERROR] 语义匹配失败: {e}")
return []
# ========= 验证码识别函数 =========
def identify_gap_with_library(bg_img_bytes):
"""使用 captcha-recognizer 库识别滑块验证码缺口位置"""
try:
from captcha_recognizer.slider import Slider
bg_img = Image.open(io.BytesIO(bg_img_bytes))
bg_arr = np.array(bg_img)
box, confidence = Slider().identify(source=bg_arr, show=False)
if box and len(box) >= 4:
x1, y1, x2, y2 = box
gap_position = int(x1)
print(f"[DEBUG] captcha-recognizer 识别结果: 缺口位置={gap_position}px, 置信度={confidence:.2f}")
return gap_position
else:
print("[WARNING] captcha-recognizer 未识别到缺口")
return 0
except ImportError as e:
print(f"[WARNING] captcha-recognizer 库未安装: {e}")
return 0
except Exception as e:
print(f"[ERROR] captcha-recognizer 识别异常: {e}")
return 0
# ========= 验证码类型检测 =========
def detect_captcha_type(page):
"""检测验证码类型：九宫格或滑块"""
# 检查九宫格验证码
grid_visible = False
grid_selectors = [".geetest_table_box", ".geetest_grid"]
for selector in grid_selectors:
try:
if page.locator(selector).is_visible(timeout=2000):
grid_visible = True
break
except:
continue
# 检查滑块验证码
slider_visible = False
slider_selectors = [".geetest_slider", ".geetest_slider_button", ".geetest_canvas_bg"]
for selector in slider_selectors:
try:
if page.locator(selector).is_visible(timeout=2000):
slider_visible = True
break
except:
continue
if grid_visible:
print("[DEBUG] 检测到九宫格验证码")
return "grid"
elif slider_visible:
print("[DEBUG] 检测到滑块验证码")
return "slider"
else:
return "unknown"
# ========= 九宫格验证码处理 =========
def solve_geetest_multistep(page, ai_service):
"""使用AI服务处理九宫格验证码"""
print("[INFO] 开始处理九宫格验证码...")
img_container = page.locator(".geetest_table_box").first
if not img_container.is_visible(timeout=3000):
print("[DEBUG] 验证码容器不可见")
return False
# 识别题目
target_object = ""
tip_img = page.locator(".geetest_tip_img").first
if tip_img.is_visible(timeout=2000):
print("[DEBUG] 检测到图片提示，使用AI识别...")
try:
target_object = ai_service.call_vision(tip_img.screenshot(), "图中是什么物体？只回答物体名称，不要带标点。")
except Exception as e:
print(f"[ERROR] AI识别图片提示失败: {e}")
else:
tip_text_loc = page.locator(".geetest_tip_content").first
if tip_text_loc.is_visible(timeout=2000):
try:
target_object = tip_text_loc.inner_text()
except Exception as e:
print(f"[ERROR] 读取文本提示失败: {e}")
target_object = re.sub(r'[^\w]', '', target_object)
print(f">>> [Step 1] 识别题目为：【{target_object}】")
# 逐行抠图识别
all_descriptions = []
try:
grid_bytes = img_container.screenshot()
grid_img = Image.open(io.BytesIO(grid_bytes))
w, h = grid_img.size
row_h = h / 3
for i in range(3):
top = i * row_h
bottom = (i + 1) * row_h
row_crop = grid_img.crop((0, top, w, bottom))
buf = io.BytesIO()
row_crop.save(buf, format='PNG')
row_res = ai_service.identify_captcha_row(buf.getvalue(), i+1)
all_descriptions.extend(row_res)
except Exception as e:
print(f"[ERROR] 九宫格识别过程出错: {e}")
return False
# 语义匹配并模拟点击
try:
click_indices = ai_service.semantic_match(target_object, all_descriptions)
print(f">>> [Final] 最终决定点击序号: {click_indices}")
except Exception as e:
print(f"[ERROR] 语义匹配失败: {e}")
return False
if not click_indices:
print("[INFO] 未找到匹配项，刷新验证码...")
try:
refresh_btn = page.locator(".geetest_refresh").first
if refresh_btn.is_visible():
refresh_btn.click()
time.sleep(2)
except:
pass
return False
try:
box = img_container.bounding_box()
cell_w, cell_h = box['width']/3, box['height']/3
for idx in click_indices:
try:
val = int(idx)
if 1 <= val <= 9:
r, c = (val-1)//3, (val-1)%3
target_x = box['x'] + c*cell_w + cell_w/2
target_y = box['y'] + r*cell_h + cell_h/2
page.mouse.click(target_x, target_y)
time.sleep(random.uniform(0.3, 0.5))
except:
continue
except Exception as e:
print(f"[ERROR] 获取验证码容器位置失败: {e}")
return False
# 提交验证
for sel in [".geetest_commit", "text=确认", ".geetest_submit"]:
try:
btn = page.locator(sel).first
if btn.is_visible(timeout=2000):
btn.click()
return True
except:
continue
return False
# ========= 滑块验证码处理 =========
def solve_geetest_slider(page, ai_service):
"""使用AI服务处理滑块验证码"""
print("[INFO] 开始处理滑块验证码...")
# 查找滑块按钮
slider_button = None
slider_selectors = [".geetest_slider_button", ".geetest_slider_knob", ".geetest_btn"]
for selector in slider_selectors:
try:
btn = page.locator(selector).first
if btn.is_visible(timeout=1000):
slider_button = btn
print(f"[DEBUG] 找到滑块按钮: {selector}")
break
except:
continue
if not slider_button:
print("[ERROR] 未找到滑块按钮")
return False
button_box = slider_button.bounding_box()
if not button_box:
print("[ERROR] 无法获取滑块按钮位置")
return False
button_initial_x = button_box['x']
button_x = button_box['x'] + button_box['width'] / 2
button_y = button_box['y'] + button_box['height'] / 2
# 获取验证码图片
bg_img_bytes = None
bg_canvas = None
canvas_selectors = [".geetest_canvas_bg", "canvas.geetest_canvas_bg"]
for selector in canvas_selectors:
try:
canvas = page.locator(selector).first
if canvas.is_visible(timeout=1000):
bg_canvas = canvas
bg_img_bytes = canvas.screenshot()
print("[DEBUG] 成功获取背景图")
break
except:
continue
if not bg_img_bytes:
print("[ERROR] 无法获取验证码图片")
return False
# 使用 captcha-recognizer 识别缺口位置
gap_position = identify_gap_with_library(bg_img_bytes)
if gap_position <= 0:
print("[ERROR] 识别到的缺口位置无效")
return False
# 计算滑动距离
bg_canvas_box = None
if bg_canvas:
try:
bg_canvas_box = bg_canvas.bounding_box()
except:
pass
if bg_canvas_box:
bg_canvas_x = bg_canvas_box['x']
offset = button_initial_x - bg_canvas_x
drag_distance_base = gap_position + offset
human_error = random.uniform(-5.0, 5.0)
drag_distance = drag_distance_base + human_error
target_x = button_x + drag_distance
else:
drag_distance_base = gap_position
human_error = random.uniform(-5.0, 5.0)
drag_distance = drag_distance_base + human_error
target_x = button_x + drag_distance
print(f"[INFO] 滑动距离: {drag_distance:.1f}px, 目标位置: {target_x:.1f}px")
# 执行拖动
try:
page.mouse.move(button_x, button_y)
time.sleep(random.uniform(0.1, 0.2))
page.mouse.down()
time.sleep(random.uniform(0.1, 0.2))
# 使用缓动函数模拟人类拖动
steps = random.randint(20, 30)
easing_functions = [
easeInOutQuad,
easeOutQuad,
easeInOutCubic,
]
easing_func = random.choice(easing_functions)
for i in range(steps):
progress = easing_func(i / steps)
jitter_x = random.uniform(-1.5, 1.5)
jitter_y = random.uniform(-2, 2)
current_x = button_x + drag_distance * progress + jitter_x
current_y = button_y + jitter_y
page.mouse.move(current_x, current_y)
if i < steps * 0.3:
time.sleep(random.uniform(0.005, 0.015))
elif i > steps * 0.7:
time.sleep(random.uniform(0.02, 0.04))
else:
time.sleep(random.uniform(0.01, 0.025))
# 超调效果
if random.random() > 0.5:
overshoot = random.uniform(2, 5)
page.mouse.move(target_x + overshoot, button_y + random.uniform(-1, 1))
time.sleep(random.uniform(0.05, 0.1))
page.mouse.move(target_x, button_y)
time.sleep(random.uniform(0.15, 0.25))
page.mouse.up()
time.sleep(random.uniform(0.5, 1.0))
print("[DEBUG] 滑块拖动完成")
time.sleep(2)
# 检查验证结果
captcha_gone = True
try:
if page.locator(".geetest_slider").is_visible(timeout=1000):
captcha_gone = False
except:
pass
return captcha_gone
except Exception as e:
print(f"[ERROR] 拖动滑块失败: {e}")
return False
# ========= 主逻辑 =========
def find_signed_text_locator(page, timeout=3000):
"""查找已签到文本"""
try:
loc = page.get_by_text(ALREADY_SIGNED_TEXT).first
if loc.is_visible(timeout=timeout):
return loc
except:
pass
return None
def main():
"""主函数"""
print("=" * 60)
print("NATFRP 自动签到脚本 - 青龙面板版")
print("=" * 60)
# 从环境变量获取配置
username = os.getenv("NATFRP_USERNAME", "")
password = os.getenv("NATFRP_PASSWORD", "")
if not username or not password:
print("[ERROR] 未配置账号信息，请在青龙面板中设置环境变量：")
print(" - NATFRP_USERNAME: NATFRP账号用户名")
print(" - NATFRP_PASSWORD: NATFRP账号密码")
return
# 初始化AI服务
try:
ai_service = AIService()
print(f"[INFO] AI服务初始化成功")
print(f" - 视觉模型: {ai_service.model_vision}")
print(f" - 文本模型: {ai_service.model_text}")
except Exception as e:
print(f"[ERROR] AI服务初始化失败: {e}")
print("[提示] 请在青龙面板中配置 ZHIPU_API_KEY 环境变量")
return
with sync_playwright() as p:
browser = p.chromium.launch(headless=True, slow_mo=100)
# 尝试使用缓存的登录状态
storage_state = None
if STATE_FILE.exists():
try:
print(f"[INFO] 发现登录状态缓存: {STATE_FILE}")
storage_state = str(STATE_FILE)
except Exception as e:
print(f"[WARNING] 读取登录状态缓存失败: {e}")
context = browser.new_context(storage_state=storage_state)
page = context.new_page()
page.set_viewport_size({"width": 1280, "height": 900})
print(f"[INFO] 正在访问: {target_url}")
try:
page.goto(target_url, timeout=30000)
print(f"[DEBUG] 页面加载完成")
except Exception as e:
print(f"[ERROR] 页面访问失败: {e}")
browser.close()
return
# 登录判断
current_url = page.url
is_logged_in = True
try:
username_input_visible = page.locator("#username").is_visible(timeout=2000)
except:
username_input_visible = False
if "login" in current_url or username_input_visible:
is_logged_in = False
print("[INFO] 检测到需要登录")
try:
print("[INFO] 正在填写登录信息...")
page.fill("#username", username)
page.fill("#password", password)
page.click("#login")
print("[DEBUG] 等待登录完成...")
page.wait_for_selector("text=账号信息", timeout=10000)
# 保存登录状态
try:
context.storage_state(path=str(STATE_FILE))
print(f"[SUCCESS] 登录成功，状态已保存到: {STATE_FILE}")
except Exception as e:
print(f"[WARNING] 保存登录状态失败: {e}")
print("[SUCCESS] 登录成功")
is_logged_in = True
except Exception as e:
print(f"[ERROR] 登录失败: {e}")
browser.close()
return
else:
print("[INFO] 已登录状态（使用缓存）")
# 处理18岁弹窗
try:
btn_18 = page.get_by_text("是，我已满18岁")
if btn_18.is_visible(timeout=3000):
print("[DEBUG] 检测到18岁确认弹窗，正在点击...")
btn_18.click()
time.sleep(1)
except:
pass
# 签到
print("[DEBUG] 开始检查签到状态...")
signed_locator = find_signed_text_locator(page)
if signed_locator:
print("[SUCCESS] 今日已签到")
else:
print("[DEBUG] 未检测到已签到状态，查找签到按钮...")
sign_btn = page.get_by_text("点击这里签到")
try:
sign_btn_visible = sign_btn.is_visible(timeout=3000)
except:
sign_btn_visible = False
if sign_btn_visible:
print("[INFO] 点击签到按钮...")
sign_success = False
try:
sign_btn.click()
print("[DEBUG] 已点击签到按钮，等待验证码加载...")
# 等待15秒让验证码加载
print("[INFO] 等待15秒让验证码完全加载...")
for i in range(30):
time.sleep(0.5)
if (i + 1) % 10 == 0:
print(f"[DEBUG] 已等待 {(i+1)*0.5:.1f} 秒...")
signed_check = find_signed_text_locator(page, timeout=500)
if signed_check:
print(f"[SUCCESS] 签到完成（无需验证码）！")
sign_success = True
break
if not sign_success:
# 检查验证码
captcha_type = detect_captcha_type(page)
if captcha_type != "unknown":
print(f"[INFO] 检测到验证码类型: {captcha_type}")
max_attempts = 3
for attempt in range(1, max_attempts + 1):
print(f"[DEBUG] 第 {attempt}/{max_attempts} 次尝试处理验证码...")
try:
if captcha_type == "grid":
captcha_result = solve_geetest_multistep(page, ai_service)
elif captcha_type == "slider":
captcha_result = solve_geetest_slider(page, ai_service)
else:
captcha_result = False
if captcha_result:
print("[INFO] 验证码处理成功")
time.sleep(3)
# 检查是否签到成功
signed_check = find_signed_text_locator(page, timeout=2000)
if signed_check:
print("[SUCCESS] 签到完成！")
sign_success = True
break
else:
print("[WARNING] 验证码处理失败，重试...")
time.sleep(2)
# 重新检测验证码类型
captcha_type = detect_captcha_type(page)
if captcha_type == "unknown":
print("[DEBUG] 验证码已消失，检查签到状态...")
signed_check = find_signed_text_locator(page, timeout=2000)
if signed_check:
print("[SUCCESS] 签到完成！")
sign_success = True
break
except Exception as e:
print(f"[ERROR] 验证码处理异常: {e}")
traceback.print_exc()
else:
print("[WARNING] 未检测到验证码")
if not sign_success:
print("[ERROR] 签到失败：超时或验证码处理失败")
except Exception as e:
print(f"[ERROR] 签到过程出错: {e}")
traceback.print_exc()
else:
print("[ERROR] 未找到签到按钮")
print("[INFO] 脚本运行结束")
print("=" * 60)
browser.close()
if __name__ == "__main__":
main()
--【肆】--：
感谢分享，已经用上了
--【伍】--：
感谢大佬
--【陆】--：
用上了 感谢大佬！，大佬 这家也有签到拿会员的能看看吗 chmlfrp
--【柒】--：
大佬nb，感谢分享
--【捌】--：
附带成功的图片
image578×1716 147 KB
--【玖】--：
太强了佬
--【拾】--：
佬友牛逼！
--【拾壹】--：
感谢分享，好久没用了这个
--【拾贰】--：
很好用。不过有时候会失败，把失败日志给AI改了几轮，部署到了服务器上弄了个定时任务，又加了个邮件通知功能，每天手机微信就能自动收到签到通知。
64de4137d11ac3505a79a523527da39e.jpg610×1315 112 KB image.png614×1324 102 KB--【拾叁】--：
也可以试试vibe coding?我基本上纯vibe coding，是可以实现的
--【拾肆】--：
说实话我更期待chmlfrp的，樱花我不怎么用不用
--【拾伍】--：
佬友nb