381 lines
17 KiB
Python
381 lines
17 KiB
Python
# coding=utf-8
|
||
import asyncio
|
||
import os
|
||
import sys
|
||
import time
|
||
import json
|
||
import cv2
|
||
from Apps.TeLaiDian.Kit import (
|
||
take_screenshot, get_image_content_md5, clean_station_name,
|
||
setup_logger, detect_price_click_point_cv, read_image, save_image
|
||
)
|
||
from Apps.TeLaiDian.ReadImageKit import ReadImageKit
|
||
from Apps.TeLaiDian.Service import TeLaiDianService
|
||
from Apps.TeLaiDian.Config.Setting import (
|
||
SCROLL_DISTANCE_RATIO, WAIT_AFTER_SCROLL, MAX_STATIONS_COUNT,
|
||
SAFE_EXCLUDE_RATIO, BOTTOM_SAFE_EXCLUDE_RATIO, WAIT_DETAIL_PAGE_LOAD,
|
||
WAIT_BACK_TO_LIST, DETAIL_SCROLL_DISTANCE_RATIO, FIRST_RUN_ONLY_ONE_STATION
|
||
)
|
||
from Core.BaseCrawler import BaseCrawler
|
||
import uiautomator2 as u2
|
||
|
||
# 项目根目录处理
|
||
project_root = os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
|
||
if project_root not in sys.path:
|
||
sys.path.append(project_root)
|
||
|
||
# 初始化日志
|
||
logger = setup_logger("TeLaiDianCrawler")
|
||
|
||
class TeLaiDianCrawler(BaseCrawler):
|
||
def __init__(self, service=None):
|
||
super().__init__(service or TeLaiDianService())
|
||
self.read_image_kit = ReadImageKit()
|
||
|
||
async def start(self):
|
||
"""
|
||
实现 BaseCrawler 的启动入口
|
||
注意：在 T4 等全流程脚本中，小程序由 Opener 提前打开，
|
||
此处直接开始爬取逻辑。
|
||
"""
|
||
d = u2.connect()
|
||
await self.crawl_list_logic(d)
|
||
|
||
async def open_app(self):
|
||
"""
|
||
打开特来电小程序
|
||
"""
|
||
from Apps.TeLaiDian import Opener
|
||
return await Opener.open_mini_program()
|
||
|
||
async def clear_ads(self, d, max_rounds=3):
|
||
"""
|
||
清理页面上的广告弹窗
|
||
"""
|
||
logger.info(f"开始清理广告弹窗，最多尝试 {max_rounds} 轮...")
|
||
for i in range(max_rounds):
|
||
ad_screen = take_screenshot(d, f"tld_ad_check_{int(time.time())}.jpg")
|
||
res = await self.read_image_kit.find_close_button_vlm(ad_screen)
|
||
|
||
if res.get("has_ad") and res.get("close_point"):
|
||
close_point = res.get("close_point")
|
||
w, h = d.window_size()
|
||
target_x = int(close_point[0] * w / 1000)
|
||
target_y = int(close_point[1] * h / 1000)
|
||
|
||
# 安全校验：绝对不能点击微信小程序的胶囊按钮区 (右上角)
|
||
# 胶囊按钮通常在 x > 80% 宽度 且 y < 100 像素(或 8% 高度) 的区域
|
||
if target_x > w * 0.75 and target_y < 150:
|
||
logger.warning(f"⚠️ 拒绝点击疑似微信胶囊按钮的区域: ({target_x}, {target_y})")
|
||
continue
|
||
|
||
logger.info(f"第 {i+1} 轮发现广告: {res.get('reason')}，点击关闭: ({target_x}, {target_y})")
|
||
d.click(target_x, target_y)
|
||
await asyncio.sleep(1.5)
|
||
else:
|
||
logger.info(f"第 {i+1} 轮未发现明显广告，清理结束。")
|
||
if os.path.exists(ad_screen): os.remove(ad_screen)
|
||
break
|
||
|
||
if os.path.exists(ad_screen): os.remove(ad_screen)
|
||
|
||
async def crawl_list_logic(self, d):
|
||
# [临时禁用] 广告清理逻辑误触严重，先解决主流程问题
|
||
# await self.clear_ads(d)
|
||
|
||
# [优化] 向下滚动以刷新/校准地理位置
|
||
# 使用更加显式的 swipe 方式：从屏幕 30% 划到 80%
|
||
w, h = d.window_size()
|
||
logger.info(f"执行显式下拉刷新操作: (x={w//2}, y1={int(h*0.3)} -> y2={int(h*0.8)})")
|
||
d.swipe(w // 2, int(h * 0.3), w // 2, int(h * 0.8), duration=0.5)
|
||
|
||
logger.info("等待 2.5 秒确保位置校准和列表刷新完成...")
|
||
await asyncio.sleep(2.5)
|
||
|
||
processed_count = 0
|
||
last_md5 = None
|
||
processed_station_names = set() # 用于列表级去重
|
||
|
||
while processed_count < MAX_STATIONS_COUNT:
|
||
# 1. 截图并分析
|
||
screenshot_path = take_screenshot(d, f"tld_list_{int(time.time())}.jpg")
|
||
|
||
# 检测是否滚动到底部
|
||
curr_md5 = get_image_content_md5(screenshot_path, top_ratio=SAFE_EXCLUDE_RATIO, bottom_ratio=BOTTOM_SAFE_EXCLUDE_RATIO)
|
||
if last_md5 == curr_md5:
|
||
logger.info("内容无变化，判定已到底部")
|
||
if os.path.exists(screenshot_path): os.remove(screenshot_path)
|
||
break
|
||
last_md5 = curr_md5
|
||
|
||
stations = await self.read_image_kit.analyze_station_list(screenshot_path)
|
||
if not stations:
|
||
# 检查是否意外退出了小程序
|
||
is_wrong_page = await self.check_wrong_page(d, screenshot_path)
|
||
if is_wrong_page:
|
||
logger.error("检测到已退出详情列表页（可能回到了搜索页），尝试重新进入...")
|
||
await self.open_app()
|
||
await asyncio.sleep(5)
|
||
continue
|
||
|
||
logger.info("本页未检测到场站，尝试滑动...")
|
||
d.swipe_ext("up", scale=SCROLL_DISTANCE_RATIO)
|
||
await asyncio.sleep(WAIT_AFTER_SCROLL)
|
||
continue
|
||
|
||
for station in stations:
|
||
if processed_count >= MAX_STATIONS_COUNT:
|
||
break
|
||
|
||
name = station.get("name")
|
||
point = station.get("point")
|
||
|
||
if not name or not point:
|
||
continue
|
||
|
||
if name in processed_station_names:
|
||
logger.info(f"跳过已处理场站: {name}")
|
||
continue
|
||
|
||
logger.info(f"处理场站: {name} (坐标: {point})")
|
||
|
||
# 点击进入详情
|
||
d.click(point[0], point[1])
|
||
logger.info(f"已点击场站 '{name}'，等待 {WAIT_DETAIL_PAGE_LOAD}s 加载详情页...")
|
||
await asyncio.sleep(WAIT_DETAIL_PAGE_LOAD)
|
||
|
||
# 截图验证是否进入详情页
|
||
detail_check_path = take_screenshot(d, f"tld_detail_check_{int(time.time())}.jpg")
|
||
logger.info(f"详情页快照已保存: {detail_check_path}")
|
||
|
||
# 简单验证：如果标题包含 "我的卡券"、"优惠券"、"新人福利" 等，说明点错了
|
||
is_wrong_page = await self.check_wrong_page(d, detail_check_path)
|
||
if is_wrong_page:
|
||
logger.warning(f"检测到进入了错误页面，尝试返回列表...")
|
||
# 尝试点击左上角的返回箭头，如果没有，则执行系统 back
|
||
d.click(40, 70) # 特来电通常左上角有返回箭头
|
||
await asyncio.sleep(1.0)
|
||
d.press("back")
|
||
await asyncio.sleep(WAIT_BACK_TO_LIST)
|
||
if os.path.exists(detail_check_path): os.remove(detail_check_path)
|
||
continue
|
||
|
||
# 爬取详情
|
||
await self.crawl_detail_logic(d, station)
|
||
processed_station_names.add(name)
|
||
|
||
# 返回列表
|
||
d.press("back")
|
||
await asyncio.sleep(WAIT_BACK_TO_LIST)
|
||
# 返回列表后也检查一下是否有新广告弹出
|
||
await self.clear_ads(d, max_rounds=1)
|
||
processed_count += 1
|
||
if FIRST_RUN_ONLY_ONE_STATION:
|
||
logger.info("已完成首个场站的全流程采集，根据配置退出爬取任务。")
|
||
if os.path.exists(screenshot_path):
|
||
os.remove(screenshot_path)
|
||
return
|
||
|
||
# 滑动到下一页
|
||
d.swipe_ext("up", scale=SCROLL_DISTANCE_RATIO)
|
||
await asyncio.sleep(WAIT_AFTER_SCROLL)
|
||
|
||
if os.path.exists(screenshot_path): os.remove(screenshot_path)
|
||
|
||
async def check_wrong_page(self, d, image_path):
|
||
"""
|
||
检查是否误触进入了错误的页面（如：我的卡券、活动页等）
|
||
"""
|
||
data = await self.read_image_kit.check_wrong_page_vlm(image_path)
|
||
is_detail = data.get("is_detail_page", True)
|
||
if not is_detail:
|
||
logger.warning(f"⚠️ 确认进入错误页面: {data.get('page_type')} ({data.get('reason')})")
|
||
return not is_detail
|
||
|
||
async def crawl_detail_logic(self, d, station_info):
|
||
"""
|
||
在详情页提取价格和状态信息
|
||
"""
|
||
# 1. 截取第一屏的基础信息图（名称、精确地址），并异步识别
|
||
first_screen_path = take_screenshot(d, f"tld_detail_basic_{int(time.time())}.jpg")
|
||
station_name = station_info.get("name")
|
||
address = station_info.get("address")
|
||
logger.info(f"已截取详情页首屏截图，启动异步基础信息识别任务: {station_name} | {address}")
|
||
|
||
async def analyze_basic_info_background(image_path, fallback_name, fallback_address):
|
||
try:
|
||
basic_info = await self.read_image_kit.analyze_detail_basic_info(image_path)
|
||
name2 = basic_info.get("name") or fallback_name
|
||
addr2 = basic_info.get("address") or fallback_address
|
||
logger.info(f"[异步] 详情页基础信息识别完成: {name2} | {addr2}")
|
||
except Exception as ex:
|
||
logger.error(f"[异步] 分析详情页基础信息失败: {ex}")
|
||
finally:
|
||
if os.path.exists(image_path):
|
||
try:
|
||
os.remove(image_path)
|
||
except:
|
||
pass
|
||
|
||
asyncio.create_task(analyze_basic_info_background(first_screen_path, station_name, address))
|
||
|
||
# 2. 一次向上滑动，确保当前底部文字完全顶到上沿之外
|
||
w, h = d.window_size()
|
||
logger.info("执行一次向上滑动，将当前底部文字完全推至屏幕上沿之外...")
|
||
d.swipe(w * 0.5, h * 0.75, w * 0.5, h * 0.25, 0.8)
|
||
await asyncio.sleep(1.5)
|
||
|
||
# 3. 使用模板 jgxx.jpg 匹配“价格信息 当前价”入口并点击
|
||
template_jgxx = os.path.join(os.path.dirname(__file__), "Template", "jgxx.jpg")
|
||
entrance_point = None
|
||
|
||
if os.path.exists(template_jgxx):
|
||
logger.info(f"使用模板匹配价格入口: {template_jgxx}")
|
||
try:
|
||
match_res = d.image.match(template_jgxx)
|
||
except Exception as e:
|
||
match_res = None
|
||
logger.error(f"模板匹配价格入口失败: {e}")
|
||
|
||
if match_res:
|
||
if hasattr(match_res, "point") and match_res.point:
|
||
entrance_point = match_res.point
|
||
elif isinstance(match_res, dict):
|
||
if "point" in match_res and match_res["point"]:
|
||
entrance_point = match_res["point"]
|
||
elif "x" in match_res and "y" in match_res:
|
||
entrance_point = (match_res["x"], match_res["y"])
|
||
elif isinstance(match_res, (list, tuple)) and len(match_res) >= 2:
|
||
entrance_point = (match_res[0], match_res[1])
|
||
|
||
if entrance_point:
|
||
logger.info(f"通过 jgxx.jpg 成功找到价格入口，点击坐标: {entrance_point}")
|
||
d.click(int(entrance_point[0]), int(entrance_point[1]))
|
||
await asyncio.sleep(WAIT_DETAIL_PAGE_LOAD)
|
||
else:
|
||
logger.warning("未能通过 jgxx.jpg 找到价格入口，使用坐标兜底点击页面中部偏下位置。")
|
||
d.click(w // 2, int(h * 0.45))
|
||
await asyncio.sleep(WAIT_DETAIL_PAGE_LOAD)
|
||
|
||
# 4. 进入分时电价页面后的处理
|
||
try:
|
||
logger.info("已点击进入价格详情，等待小程序自动滚动定位完成 (4秒)...")
|
||
await asyncio.sleep(4.0)
|
||
|
||
logger.info("执行两次大幅向下拉动，确保回到 00:00 时段顶部...")
|
||
for _ in range(2):
|
||
d.swipe_ext("down", scale=0.9)
|
||
await asyncio.sleep(1.0)
|
||
except Exception as e:
|
||
logger.error(f"处理分时电价页面初始状态失败: {e}")
|
||
|
||
# 5. 循环滑动抓取完整分时电价
|
||
all_prices = []
|
||
last_price_md5 = None
|
||
price_page_count = 0
|
||
max_price_pages = 4 # 增加到4页确保覆盖 00:00-24:00
|
||
screenshot_tasks = [] # 用于异步分析图片的任务列表
|
||
temp_screenshots = [] # 记录临时文件以便后续清理
|
||
|
||
logger.info("开始循环截图（UI操作优先，后台并行分析）...")
|
||
|
||
try:
|
||
while price_page_count < max_price_pages:
|
||
price_screen_path = take_screenshot(d, f"tld_detail_price_{int(time.time())}_{price_page_count}.jpg")
|
||
|
||
# 校验页面是否发生滚动变化
|
||
curr_md5 = get_image_content_md5(price_screen_path, top_ratio=0.2, bottom_ratio=0.2)
|
||
if curr_md5 == last_price_md5:
|
||
logger.info("价格页面内容无变化，判定已触底")
|
||
if os.path.exists(price_screen_path): os.remove(price_screen_path)
|
||
break
|
||
last_price_md5 = curr_md5
|
||
temp_screenshots.append(price_screen_path)
|
||
|
||
logger.info(f"已截取价格详情页第 {price_page_count + 1} 页: {price_screen_path}，启动后台异步分析")
|
||
# 使用 asyncio.create_task 立即在后台开始执行分析
|
||
task = asyncio.create_task(self.read_image_kit.analyze_detail_price(price_screen_path))
|
||
screenshot_tasks.append(task)
|
||
|
||
# 向上滚动列表（手指向上划），看后面的时段
|
||
logger.info("向上滚动列表，准备截取下一屏价格...")
|
||
d.swipe_ext("up", scale=0.8)
|
||
await asyncio.sleep(1.0)
|
||
price_page_count += 1
|
||
|
||
# 等待所有后台分析任务完成
|
||
if screenshot_tasks:
|
||
logger.info(f"UI 操作已完成，等待 {len(screenshot_tasks)} 个后台分析任务结束...")
|
||
results = await asyncio.gather(*screenshot_tasks, return_exceptions=True)
|
||
|
||
for res in results:
|
||
if isinstance(res, Exception):
|
||
logger.error(f"后台分析任务出错: {res}")
|
||
continue
|
||
if res:
|
||
# 深度去重：根据时段 (start, end) 合并
|
||
for p in res:
|
||
is_duplicate = False
|
||
for existing in all_prices:
|
||
if p.get('start') == existing.get('start') and p.get('end') == existing.get('end'):
|
||
is_duplicate = True
|
||
# 字段补全逻辑
|
||
for key in ['price', 'plus_price', 'market_price', 'elec_price', 'service_price']:
|
||
if p.get(key) is not None and (existing.get(key) is None or existing.get(key) == 0):
|
||
existing[key] = p[key]
|
||
break
|
||
if not is_duplicate:
|
||
all_prices.append(p)
|
||
except Exception as e:
|
||
logger.error(f"抓取价格详情过程中发生异常: {e}")
|
||
finally:
|
||
# 无论是否异常，都要确保清理未完成的任务，避免 "never awaited" 警告
|
||
for task in screenshot_tasks:
|
||
if not task.done():
|
||
task.cancel()
|
||
|
||
# 清理所有临时截图
|
||
for path in temp_screenshots:
|
||
if os.path.exists(path):
|
||
try:
|
||
os.remove(path)
|
||
except:
|
||
pass
|
||
|
||
# 6. 保存数据
|
||
if all_prices:
|
||
station_name_clean = clean_station_name(station_name)
|
||
# 对价格按时间排序
|
||
try:
|
||
all_prices.sort(key=lambda x: x.get('start', '00:00'))
|
||
except:
|
||
pass
|
||
logger.info(f"✅ 场站 {station_name_clean} 共提取到 {len(all_prices)} 条价格信息，准备保存...")
|
||
await self.service.save_station_data(station_name_clean, address, all_prices)
|
||
else:
|
||
logger.warning(f"❌ 未能提取到任何价格信息，请检查页面识别逻辑")
|
||
# 清理价格页临时截图（详情页首屏截图已在异步任务中清理）
|
||
if os.path.exists(price_screen_path):
|
||
os.remove(price_screen_path)
|
||
|
||
async def crawl_list(self):
|
||
"""
|
||
实现 BaseCrawler 的抽象方法
|
||
"""
|
||
d = u2.connect()
|
||
await self.crawl_list_logic(d)
|
||
|
||
async def crawl_detail(self, station_info):
|
||
"""
|
||
实现 BaseCrawler 的抽象方法
|
||
"""
|
||
# 逻辑已在 crawl_list_logic 中通过 crawl_detail_logic 调用
|
||
pass
|
||
|
||
async def main(service=None):
|
||
crawler = TeLaiDianCrawler(service=service)
|
||
await crawler.start()
|
||
|
||
if __name__ == "__main__":
|
||
asyncio.run(main())
|