def task( self ):
"""
自动抓取微信群消息
"""
app.logger.info( 'executing %s' % ( __name__ ) )
try:
app.logger.info( "正在建立Appium会话..." )
self.driver = webdriver.Remote( self.appium_server, self.desired_capability ) # Appium会话配置
self.wait=WebDriverWait(self.driver, self.timeout)
app.logger.info( "正在点击微信首页消息列表第一条,进入群消息对话框..." )
self.wait.until( EC.presence_of_element_located( ( By.ID,'com.tencent.mm:id/as4' ) ) ).click()
# 获取群聊名称
group_name = self.wait.until( EC.presence_of_element_located( ( By.ID,'com.tencent.mm:id/hm' ) ) ).get_attribute( 'text' )
if not group_name or not re.compile( self.group_tag_pattern ).findall( group_name ):
app.logger.info( "当前会话不是群聊，即将关闭..." )
return
group_name = re.sub( self.group_tag_pattern, "", group_name )
app.logger.info( "当前群聊名称为：" + group_name )
current_simulator_window_size = get_simulator_window_size( self.driver )
app.logger.info( "当前屏幕尺寸为:{0}".format( current_simulator_window_size ) )
time.sleep( 1 ) # 强制等待1秒
# 消息发布时间
send_time = get_current_date()
app.logger.info( '消息发布时间:%s' %( send_time ) )
# 不断向下滑屏，获取当前屏消息
while True:
self.get_current_page_msgs( group_name=group_name, send_time=send_time )
try:
self.driver.find_element_by_id( 'com.tencent.mm:id/abj' )
app.logger.info( "发现新消息提示框,正在继续向下滑屏..." )
driver_swipe_down( current_simulator_window_size, self.driver )
except NoSuchElementException:
app.logger.info( "没有找到新消息提示框，只向下滑屏3次..." )
for i in range(0, 3):
driver_swipe_down( current_simulator_window_size, self.driver )
self.get_current_page_msgs( group_name=group_name, send_time=send_time )
break
# 一次过滤，first_filter_keys表， 并保存到数据库（messages表），调用公共方法
app.logger.info( "正在根据关键词过滤原始消息记录表..." )
MessagesService.filter_by_keys()
# 二次过滤，second_filter_qqnumber表、second_filter_groupnumber表， 并保存到数据库（messages表），调用公共方法
app.logger.info( "根据给定的QQ/微信群号码或者QQ/微信号过滤消息记录表" )
MessagesService.filter_by_numbers()
except:
app.logger.info( traceback.format_exc() )
finally:
app.logger.info( "正在关闭当前Appium会话..." ) # 关闭当前会话
self.driver.quit()
app.logger.info( "finished %s" % ( __name__ ) )
def task( self ):
"""
执行任务
"""
app.logger.info( "launching %s ..." % ( __name__ ) )
# 一次过滤，first_filter_keys表， 并保存到数据库（messages表），调用公共方法
app.logger.info( "正在根据关键词过滤原始消息记录表..." )
MessagesService.filter_by_keys()
# 二次过滤，second_filter_qqnumber表、second_filter_groupnumber表， 并保存到数据库（messages表），调用公共方法
app.logger.info( "根据给定的QQ群号码或者QQ号过滤消息记录表" )
MessagesService.filter_by_numbers()
app.logger.info( "finished %s" % ( __name__ ) )
def task( self ):
"""
导出QQ群消息记录并写入数据库
"""
app.logger.info( "launching %s ..." % ( __name__ ) )
app.logger.info( "当前屏幕宽为：" + str( self.screen_width ) + "," + "当前屏幕高为：" + str( self.screen_height ) )
for qq_account in self.qq_account_list: # 依次使用不同的账号登录QQ，导出QQ群消息记录
qq_number = qq_account["qq_number"]
qq_password = qq_account["qq_password"]
# 打开qq
app.logger.info( "正在打开QQ..." )
qq = Application( backend='uia' ).start( self.qq_path )
# qq.QQ.print_control_identifiers()
# 输入账号
app.logger.info( "正在输入账号..." )
qq.QQ.Pane16.child_window(title="QQ号码", control_type="ComboBox").click_input()
qq.QQ.Pane16.child_window(title="QQ号码", control_type="ComboBox").type_keys( qq_number )
# 输入密码
app.logger.info( "正在输入密码..." )
qq.QQ.Pane16.child_window(title="密码", control_type="Pane").click_input()
qq.QQ.Pane16.child_window(title="密码", control_type="Pane").type_keys( qq_password )
# 点击登录按钮
app.logger.info( "正在登录..." )
qq.QQ.child_window(title="登 录", control_type="Button").click_input()
app.logger.info( '强制等待10秒，若需要滑块验证，请迅速手动完成' )
time.sleep( 10 ) # 强制等待10秒，若需要滑块验证，请迅速手动完成
# 打开QQ消息管理器
app.logger.info( "正在连接新的QQ进程..." )
qq = None
qq = Application( backend='uia' ).connect( path=self.qq_path )
# 点击第三个tab选项卡，即群聊
app.logger.info( "正在打开群聊选项卡..." )
qq.QQ.TabControl.TabItem3.click_input()
# 单击"我的QQ群"
# 默认所有QQ账号的“我的QQ群”栏处于关闭状态
app.logger.info( "正在打开我的QQ群..." )
qq.QQ.child_window(title="我的QQ群", control_type="ListItem").click_input()
# 利用pyautogui获取鼠标当前的坐标，然后在第一个QQ群的位置，点击鼠标右键
x, y = pyautogui.position()
app.logger.info( "鼠标当前的坐标 X:" + str( x ).rjust( 4 ) + " Y:" + str( y ).rjust( 4 ) )
# 将鼠标向下移动35个像素
y = y + 35
app.logger.info( "鼠标当前的坐标 X:" + str( x ).rjust( 4 ) + " Y:" + str( y ).rjust( 4 ) )
pyautogui.moveTo( x, y )
# 点击鼠标右键
pyautogui.rightClick()
# 将鼠标向右移动40个像素,向下移动90个像素
x = x + 40
y = y + 90
app.logger.info( "鼠标当前的坐标 X:" + str( x ).rjust( 4 ) + " Y:" + str( y ).rjust( 4 ) )
pyautogui.moveTo( x, y )
# 点击“查看消息记录”按钮，打开消息管理器
app.logger.info( "正在打开消息管理器..." )
pyautogui.click()
app.logger.info( "等待消息加载完毕" )
time.sleep( 3 ) # 强制等待3秒
# 导出全部消息记录
app.logger.info( "正在导出消息记录..." )
x = self.export_x_coor
y = self.export_y_coor
app.logger.info( "鼠标当前的坐标 X:" + str( x ).rjust( 4 ) + " Y:" + str( y ).rjust( 4 ) )
pyautogui.moveTo( x, y )
pyautogui.click(x, y)
# 点击导出按钮
app.logger.info( "正在点击导出按钮..." )
x = x + 88
y = y + 49
app.logger.info( "鼠标当前的坐标 X:" + str( x ).rjust( 4 ) + " Y:" + str( y ).rjust( 4 ) )
pyautogui.moveTo( x, y )
pyautogui.click( x, y )
# 选择消息文件保存类型
app.logger.info( "正在选择文件保存类型..." )
x = self.msg_file_x_coor
y = self.msg_file_y_coor
app.logger.info( "鼠标当前的坐标 X:" + str( x ).rjust( 4 ) + " Y:" + str( y ).rjust( 4 ) )
pyautogui.moveTo( x, y )
pyautogui.click( x, y )
y = y + 39
app.logger.info( "鼠标当前的坐标 X:" + str( x ).rjust( 4 ) + " Y:" + str( y ).rjust( 4 ) )
pyautogui.moveTo( x, y )
pyautogui.click( x, y )
# 向上移动鼠标，防止失去焦点
y = y - 39
app.logger.info( "鼠标当前的坐标 X:" + str( x ).rjust( 4 ) + " Y:" + str( y ).rjust( 4 ) )
pyautogui.moveTo( x, y )
# 发送快捷键"Alt+S",保存消息记录文件
app.logger.info( "正在保存..." )
pyautogui.keyDown('altleft')
pyautogui.keyDown('s')
pyautogui.keyUp('altleft')
pyautogui.keyUp('s')
# 如果文件已经存在，则覆盖掉
pyautogui.keyDown('altleft')
pyautogui.keyDown('y')
pyautogui.keyUp('altleft')
pyautogui.keyUp('y')
app.logger.info( '等待消息保存完成' )
time.sleep( 10 ) # 强制等待10秒
# 关闭消息管理器
app.logger.info( "正在关闭消息管理器..." )
pyautogui.keyDown( 'altleft' )
pyautogui.keyDown( 'f4' )
pyautogui.keyUp( 'altleft' )
pyautogui.keyUp( 'f4' )
# qq.QQ.ScrollBar.print_control_identifiers()
# 关闭QQ
app.logger.info( "正在关闭QQ..." )
qq.QQ.child_window(title="我的QQ群", control_type="ListItem").click_input() # 将已经打开的“我的QQ群”栏收起
pyautogui.keyDown( 'altleft' )
pyautogui.keyDown( 'f4' )
pyautogui.keyUp( 'altleft' )
pyautogui.keyUp( 'f4' )
msgs = open( self.msg_path + r"\全部消息记录.txt","rb" ).read().decode("utf8")
# 对消息记录进行数据格式化，并增量保存到数据库(original_messages表)
app.logger.info( "正在将消息写入数据库..." )
MessagesService.save_export_group_msgs( msgs=msgs )
# 一次过滤，first_filter_keys表， 并保存到数据库（messages表），调用公共方法
app.logger.info( "正在根据关键词过滤原始消息记录表..." )
MessagesService.filter_by_keys()
# 二次过滤，second_filter_qqnumber表、second_filter_groupnumber表， 并保存到数据库（messages表），调用公共方法
app.logger.info( "根据给定的QQ群号码或者QQ号过滤消息记录表" )
MessagesService.filter_by_numbers()
app.logger.info( "finished %s" % ( __name__ ) )
def patch_delete_qq_numbers(number_id):
"""
patch /filter/qq_numbers/number_id
更新一个QQ号码
delete /filter/qq_numbers/number_id
删除一个QQ号码
"""
# 获取所有请求参数
req_dict = request.values
# 封装响应字段
resp_data = {"code": 0, "count": 0, "msg": "操作成功~", "data": []}
if not number_id:
resp_data["msg"] = "缺少参数"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 404)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
try:
number_id = int(number_id)
except:
resp_data["msg"] = "参数错误"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 406)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
number_model = SecondFilterQqnumber.query.filter_by(id=number_id).first()
if not number_model:
resp_data["msg"] = "资源不存在"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 404)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
if "PATCH" == request.method:
qq_number = req_dict.get("qq_number", None)
if not qq_number:
req_dict = request.get_json()
qq_number = req_dict.get("qq_number", None)
if not qq_number:
resp_data["msg"] = "缺少参数"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 404)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
number_model.qq_number = qq_number
db.session.add(number_model)
db.session.commit()
resp_data["data"].append({
"id": number_model.id,
"qq_number": number_model.qq_number
})
resp_data["count"] = 1
response = make_response(json.dumps(resp_data), 200)
response.headers["Content-Type"] = "application/json;charset=utf-8"
# 更新消息记录表
MessagesService.update_messages()
return response
if "DELETE" == request.method:
db.session.delete(number_model)
db.session.commit()
response = make_response(json.dumps(resp_data), 200)
response.headers["Content-Type"] = "application/json;charset=utf-8"
# 更新消息记录表
MessagesService.update_messages()
return response
def get_post_qq_numbers():
"""
get /message/filter/qq_numbers
获取过滤QQ号码列表
post /message/filter/qq_numbers
新增一个QQ号码
"""
# 获取所有请求参数
req_dict = request.values
# 封装响应字段
resp_data = {"code": 0, "count": 0, "msg": "操作成功~", "data": []}
if "GET" == request.method:
# 实现分页加载
page = req_dict.get("page", 1)
try:
page = int(page)
except:
resp_data["msg"] = "参数错误"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 406)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
if page < 1:
page = 1
page_size = req_dict.get("limit", 10)
try:
page_size = int(page_size)
except:
resp_data["msg"] = "参数错误"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 406)
response.headers["Content-Type"] = "application/json;charset=utf-8"
offset = (page - 1) * page_size
number_model_list = SecondFilterQqnumber.query.order_by(
SecondFilterQqnumber.id.desc()).offset(offset).limit(
page_size).all()
number_data_list = []
if number_model_list:
for model in number_model_list:
number_data_list.append({
"id": model.id,
"qq_number": model.qq_number
})
resp_data["data"] = number_data_list
resp_data["count"] = db.session.query(
func.count(SecondFilterQqnumber.id)).scalar()
response = make_response(json.dumps(resp_data), 200)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
if "POST" == request.method:
qq_number = req_dict.get("qq_number", None)
if not qq_number:
req_dict = request.get_json()
qq_number = req_dict.get("qq_number", None)
if not qq_number:
resp_data["msg"] = "参数错误"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 404)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
number_model = SecondFilterQqnumber.query.filter_by(
qq_number=qq_number).first()
if number_model:
resp_data["msg"] = "QQ号码已经存在"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 406)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
else:
number_model = SecondFilterQqnumber()
number_model.qq_number = qq_number
db.session.add(number_model)
db.session.commit()
resp_data["data"].append({
"id": number_model.id,
"qq_number": number_model.qq_number
})
resp_data["count"] = 1
response = make_response(json.dumps(resp_data), 200)
response.headers["Content-Type"] = "application/json;charset=utf-8"
# 更新消息记录表
MessagesService.update_messages()
return response
def patch_delete_keys(key_id):
"""
patch /message/filter/keys/key_id
更新一个关键词
delete /message/filter/keys/key_id
删除一个关键词
"""
# 获取所有请求参数
req_dict = request.values
# 封装响应字段
resp_data = {"code": 0, "count": 0, "msg": "操作成功~", "data": []}
if not key_id:
resp_data["msg"] = "缺少参数"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 404)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
try:
key_id = int(key_id)
except:
resp_data["msg"] = "参数错误"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 406)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
key_model = FirstFilterKey.query.filter_by(id=key_id).first()
if not key_model:
resp_data["msg"] = "资源不存在"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 404)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
if "PATCH" == request.method:
key_name = req_dict.get("key_name", None)
if not key_name:
req_dict = request.get_json()
key_name = req_dict.get("key_name", None)
if not key_name:
resp_data["msg"] = "缺少参数"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 404)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
key_model.key_name = key_name
db.session.add(key_model)
db.session.commit()
resp_data["data"].append({
"id": key_model.id,
"key_name": key_model.key_name
})
resp_data["count"] = 1
response = make_response(json.dumps(resp_data), 200)
response.headers["Content-Type"] = "application/json;charset=utf-8"
# 更新消息记录表
MessagesService.update_messages()
return response
if "DELETE" == request.method:
db.session.delete(key_model)
db.session.commit()
response = make_response(json.dumps(resp_data), 200)
response.headers["Content-Type"] = "application/json;charset=utf-8"
# 更新消息记录表
MessagesService.update_messages()
return response
def get_post_keys():
"""
get /message/filter/keys
获取过滤关键词列表
post /message/filter/keys
新增一个关键词
"""
# 获取所有请求参数
req_dict = request.values
# 封装响应字段
resp_data = {"code": 0, "count": 0, "msg": "操作成功~", "data": []}
if "GET" == request.method:
# 实现分页加载
page = req_dict.get("page", 1)
try:
page = int(page)
except:
resp_data["msg"] = "参数错误"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 406)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
if page < 1:
page = 1
page_size = req_dict.get("limit", 10)
try:
page_size = int(page_size)
except:
resp_data["msg"] = "参数错误"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 406)
response.headers["Content-Type"] = "application/json;charset=utf-8"
offset = (page - 1) * page_size
key_model_list = FirstFilterKey.query.order_by(
FirstFilterKey.id.desc()).offset(offset).limit(page_size).all()
key_data_list = []
if key_model_list:
for model in key_model_list:
key_data_list.append({
"id": model.id,
"key_name": model.key_name
})
resp_data["data"] = key_data_list
resp_data["count"] = db.session.query(func.count(
FirstFilterKey.id)).scalar()
response = make_response(json.dumps(resp_data), 200)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
if "POST" == request.method:
req_dict = request.get_json()
key_name = req_dict["key_name"] if "key_name" in req_dict else None
if not key_name:
resp_data["msg"] = "参数错误"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 404)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
key_model = FirstFilterKey.query.filter_by(key_name=key_name).first()
if key_model:
resp_data["msg"] = "关键字已经存在"
resp_data["code"] = -1
response = make_response(json.dumps(resp_data), 406)
response.headers["Content-Type"] = "application/json;charset=utf-8"
return response
else:
key_model = FirstFilterKey()
key_model.key_name = key_name
db.session.add(key_model)
db.session.commit()
resp_data["data"].append({
"id": key_model.id,
"key_name": key_model.key_name
})
resp_data["count"] = 1
response = make_response(json.dumps(resp_data), 200)
response.headers["Content-Type"] = "application/json;charset=utf-8"
# 更新消息记录表
MessagesService.update_messages()
return response
def get_current_page_msgs( self, group_name=None, group_number=None, send_time=None ):
"""
提取当前屏/页消息
"""
app.logger.info( '提取当前屏消息...' )
# 当前获取到的消息块列表
message_list = self.wait.until( EC.presence_of_all_elements_located(
( By.ID, 'com.tencent.mm:id/y' )
) )
app.logger.info( '当前页共有%d条消息' %( len( message_list ) ) )
# 遍历消息块列表，处理每条消息
app.logger.info( '遍历消息...' )
for message in message_list:
try:
# 点击用户头像，进入用户详情页，获取用户信息（如昵称等）
message.find_element_by_id( 'com.tencent.mm:id/kf' ).click()
except NoSuchElementException:
app.logger.info( '当前消息信息不完全, 跳过当前消息' )
continue
# 用户昵称
sender_name = self.wait.until( EC.presence_of_element_located(
( By.ID, 'com.tencent.mm:id/qj' )
) ).get_attribute( 'text' )
app.logger.info( '当前消息用户昵称为:%s' %( sender_name ) )
app.logger.info( '正在点击返回按钮,返回消息对话框界面...' )
self.driver.find_element_by_id( 'com.tencent.mm:id/hs' ).click()
self.wait.until( EC.presence_of_element_located( ( By.ID,'com.tencent.mm:id/hm' ) ) )
# 消息内容
content = ''
try:
content += message.find_element_by_id( 'com.tencent.mm:id/af2' ).get_attribute( 'text' )
content += message.find_element_by_id( 'com.tencent.mm:id/af5' ).get_attribute( 'text' )
app.logger.info( '匹配到超链接消息:%s' %( content ) )
except NoSuchElementException:
try:
content_box = message.find_element_by_id( 'com.tencent.mm:id/kh' ) # 文字消息box
except NoSuchElementException:
app.logger.info( '当前消息信息不完全, 跳过当前消息' )
continue
try:
app.logger.info( '长按"消息正文"2秒钟' )
TouchAction( self.driver ).long_press( # 长按"消息正文"2秒钟
x=content_box.location.get('x'), y=content_box.location.get('y'), duration=2000
).wait( 200 ).release().perform()
app.logger.info( '复制当前文字消息' )
self.driver.find_element_by_android_uiautomator( # 复制当前文字消息
'new UiSelector().text("复制")'
).click()
app.logger.info( '获取剪贴板的文字消息' )
content = get_clipboard_text()
app.logger.info( '匹配到文字消息:%s' %( content ) )
except:
app.logger.info( traceback.format_exc() )
app.logger.info( '当前消息信息不完全, 跳过当前消息' )
continue
# 对消息记录进行数据格式化，并增量保存到数据库(original_messages表)
app.logger.info( "正在将消息写入数据库..." )
MessagesService.save_wechat_group_msg( sender_name=sender_name, content=content, group_name=group_name, group_number=group_number, send_time=send_time )