def __execute_sycm_product_actions(self, task):
raw_data = task.raw_data
cate_id = raw_data['cateId']
cate_name = raw_data['cateName']
account = raw_data['account']
date_str = raw_data['dateStr']
context = Context()
context.attach(Context.KEY_CURRENT_TASK, task)
context.attach(Context.KEY_IS_UPDATE_COOKIES, False)
referer = 'https://sycm.taobao.com/mc/mq/product_insight'
context.attach(Context.KEY_HEADERS, SpiderHttp.get_sycm_headers(referer))
requests_cookie_jar = self._cookie_service.load(account)
if not requests_cookie_jar:
raise CookieNotFoundException('cookie not found')
# cookies = spider_qt5_bootstrap(url=SpiderUrls.get_sycm_login_url(), account=account)
# requests_cookie_jar = cookies
# self._cookie_service.dump(cookies, account)
context.attach(Context.KEY_COOKIES, requests_cookie_jar)
context.attach(Context.KEY_IS_UPDATE_COOKIES, False)
context.attach(Context.KEY_CURRENT_TASK_ACCOUNT, account)
good = Good()
good.set_category_id(cate_id)
good.set_category_name(cate_name)
good.set_flag(str(int(GoodDataType.initial)))
# yesterday_date_str = yesterday().strftime("%Y-%m-%d")
# yesterday_date_str = day_before_yesterday().strftime("%Y-%m-%d")
yesterday_date_str = date_str
good.set_date(yesterday_date_str)
context.attach(Context.KEY_GOOD_DICT, good)
sycm_brands_url = SpiderUrls.get_sycm_product_get_brands_url(cate_id)
token = self._cookie_service.load(account=account, type_=PickleFileType.legality_token)
sycm_prod_hot_rank_url = SpiderUrls.get_sycm_product_prod_hot_rank(start_data=yesterday_date_str,
end_date=yesterday_date_str, cate_id=cate_id,
token=token)
sycm_brands_http_request = HttpRequest(url=sycm_brands_url, method=HttpMethod.GET)
sycm_prod_hot_rank_http_request = HttpRequest(url=sycm_prod_hot_rank_url, method=HttpMethod.GET)
context.attach(Context.KEY_SYCM_PRODUCT_GET_BRANDS_HTTP_REQUEST, sycm_brands_http_request)
context.attach(Context.KEY_SYCM_PRODUCT_PROD_HOT_RANK_HTTP_REQUEST, sycm_prod_hot_rank_http_request)
for action in self.get_sycm_product_actions():
action.execute(context=context)
return
def test(self):
requests_cookie_jar = self.__cookie_service.load(account=self.account)
context = Context()
context.attach(Context.KEY_IS_UPDATE_COOKIES, True)
context.attach(Context.KEY_HEADERS, SpiderHttp.get_taobao_headers('https://s.m.taobao.com/h5'))
context.attach(Context.KEY_COOKIES, requests_cookie_jar)
context.attach(Context.KEY_CURRENT_TASK_ACCOUNT, self.account)
proxy = self.__proxy_service.get_s_proxy(self.account['username'])
context.attach(Context.KEY_CURRENT_PROXY, proxy)
good = Good()
good.set_brand_name('Flyco/飞科')
good.set_model_name('FR5218')
context.attach(Context.KEY_GOOD_DICT, good)
action = TaobaoPresearchAction()
action.execute(context=context)
try:
# action = TaobaoBaichuanAction()
# action.execute(context=context)
action = TaobaoSaleListPageAction()
action.execute(context=context)
except CookieNeedUpdateException as e:
self.__cookie_service.dump(requests_cookie_jar, self.account)
def execute_direct_good_actions(self, date_str):
context = Context()
context.attach(Context.KEY_DIRECT_COLLECT_DATE, date_str)
context.attach(Context.KEY_CURRENT_TASK_TYPE, TaobaoTaskType.sycm_list)
action = TaskDirectCollectAction()
action.execute(context=context)
tasks = context.get(Context.KEY_CURRENT_TASKS, [])
for task in tasks:
raw_data = task.raw_data
good = Good(raw_data['goodResult'])
context.attach(Context.KEY_GOOD_DICT, good)
action = GoodDirectPersistAction()
action.execute(context=context)
def test_init_unhandle_tasks(self):
context = Context()
context.attach(Context.KEY_CURRENT_TASK_TYPE, TaobaoTaskType.sycm_init)
action = TaskCollectAction()
action.execute(context=context)
tasks = context.get(Context.KEY_CURRENT_TASKS, [])
for task in tasks:
id_ = task.id
raw_data = task.raw_data
date = raw_data['dateStr']
with write_session_scope() as session:
stream_unhandle_task_dao = get_stream_unhandle_task_dao(
session=session)
mod_dict = {'date': date}
stream_unhandle_task_dao.update(
mod_dict=mod_dict,
_filter=[CmmSysStreamUnhandleTask.id == id_])
context = Context()
context.attach(Context.KEY_CURRENT_TASK_TYPE, TaobaoTaskType.sycm_list)
action = TaskCollectAction()
action.execute(context=context)
tasks = context.get(Context.KEY_CURRENT_TASKS, [])
for task in tasks:
id_ = task.id
raw_data = task.raw_data
good = Good(raw_data['goodResult'])
date = good.get_date()
with write_session_scope() as session:
stream_unhandle_task_dao = get_stream_unhandle_task_dao(
session=session)
mod_dict = {'date': date}
stream_unhandle_task_dao.update(
mod_dict=mod_dict,
_filter=[CmmSysStreamUnhandleTask.id == id_])
context = Context()
context.attach(Context.KEY_CURRENT_TASK_TYPE,
TaobaoTaskType.taobao_list)
action = TaskCollectAction()
action.execute(context=context)
tasks = context.get(Context.KEY_CURRENT_TASKS, [])
for task in tasks:
id_ = task.id
raw_data = task.raw_data
good = Good(raw_data['goodResult'])
date = good.get_date()
with write_session_scope() as session:
stream_unhandle_task_dao = get_stream_unhandle_task_dao(
session=session)
mod_dict = {'date': date}
stream_unhandle_task_dao.update(
mod_dict=mod_dict,
_filter=[CmmSysStreamUnhandleTask.id == id_])
def __execute_taobao_integrate_list_actions(self, task, account, proxy):
raw_data = task.raw_data
# default_account = global_config.accounts[0]
# account = raw_data.get('account', default_account)
# account = raw_data.get('account', default_account)
good = Good(raw_data['goodResult'])
context = Context()
context.attach(Context.KEY_CURRENT_TASK, task)
context.attach(Context.KEY_CURRENT_TASK_ACCOUNT, account)
context.attach(Context.KEY_CURRENT_PROXY, proxy)
# query = 'Flyco/飞科 + FR5218'
# query = brand_name + '+' + model_name
# page = '1'
requests_cookie_jar = self._cookie_service.load(account)
if not requests_cookie_jar:
raise CookieExpiredException('integrate list need first login')
context.attach(Context.KEY_IS_UPDATE_COOKIES, True)
referer = 'https://s.m.taobao.com/h5'
context.attach(Context.KEY_HEADERS, SpiderHttp.get_taobao_headers(referer))
context.attach(Context.KEY_COOKIES, requests_cookie_jar)
context.attach(Context.KEY_CURRENT_SLEEP_SECS, 2)
context.attach(Context.KEY_GOOD_DICT, good)
actions = self.get_taobao_integrate_list_actions()
for action in actions:
try:
result = action.execute(context=context)
except CookieNeedUpdateException as e:
self._cookie_service.dump(requests_cookie_jar, account)
raise e
# except CookieExpiredException as e:
# raise e
# except InterruptException as e:
# raise e
if not result:
break
def __execut_taobao_detail_actions(self, task, proxy=None):
raw_data = task.raw_data
good_result = Good(raw_data['goodResult'])
model_name = good_result.get_model_name()
cate_id = good_result.get_category_id()
integrate_infos = raw_data['integrateInfos']
sale_infos = raw_data['saleInfos']
i = 0
j = 1
length = min(len(integrate_infos), len(sale_infos))
is_success = False
context = Context()
context.attach(Context.KEY_GOOD_DICT, good_result)
context.attach(Context.KEY_CURRENT_TASK, task)
context.attach(Context.KEY_CURRENT_PROXY, proxy)
for x in range(0, length):
is_need_retry = False
if i < len(sale_infos):
sale_info = sale_infos[i]
sale_item_id = sale_info['itemId']
sale_title = sale_info['title']
sale_cate_id = sale_info['category']
sale_price = sale_info['price']
# str(sale_title).upper()
# if str(sale_title).upper().find(str(model_name).upper()) != -1 and str(cate_id) == str(sale_cate_id):
if str(sale_title).upper().find(str(model_name).upper()) != -1 and Category.check_cate_id(cate_id,
sale_cate_id):
actions = self.get_taobao_detail_actions()
is_success = True
price_info = [{
'skuId': '-1',
'price': yuan_2_cent(sale_price)
}]
good_result.set_price_info(price_info=price_info)
good_result.set_flag(str(int(GoodDataType.success)))
for action in actions:
action.execute(context=context)
break
elif i < 5:
actions = self.get_taobao_http_detail_actions()
timestamps = int(datetime.now().timestamp() * 1000)
# sign = get_sign('414804c1e894540b7f18f703c74346cf', str(timestamps), '12574478',
# '{"itemNumId":"%s"' % (sale_item_id))
sale_detail_url = SpiderUrls.get_taobao_detail_url(timestamps, '', sale_item_id)
context = Context()
context.attach(Context.KEY_GOOD_DICT, good_result)
context.attach(Context.KEY_CURRENT_TASK, task)
context.attach(Context.KEY_COOKIES, RequestsCookieJar())
context.attach(Context.KEY_IS_UPDATE_COOKIES, True)
context.attach(Context.KEY_CURRENT_PROXY, proxy)
detail_m_url = SpiderUrls.get_detail_m_url(sale_info['userType'], sale_item_id)
detail_m_http_request = HttpRequest(detail_m_url, method=HttpMethod.GET)
context.attach(Context.KEY_DETAIL_M_HTTP_REQUEST, detail_m_http_request)
sale_http_request = HttpRequest(url=sale_detail_url, method=HttpMethod.GET)
context.attach(Context.KEY_TAOBAO_DETAIL_HTTP_REQUEST, sale_http_request)
context.attach(Context.KEY_HEADERS, SpiderHttp.get_taobao_headers(detail_m_url))
try:
for action in actions:
action.execute(context=context)
is_success = True
break
except RetryException as e:
logger.error(e)
time.sleep(5)
except InterruptException as e:
logger.exception(e)
time.sleep(10)
is_need_retry = True
# raise e
# if is_success:
# break
if j < len(integrate_infos):
integrate_info = integrate_infos[j]
integrate_item_id = integrate_info['itemId']
integrate_title = integrate_info['title']
integrate_cate_id = integrate_info['category']
integrate_price = integrate_info['price']
# if str(integrate_title).upper().find(
# str(model_name).upper()) != -1 and str(cate_id) == str(integrate_cate_id):
if str(integrate_title).upper().find(
str(model_name).upper()) != -1 and Category.check_cate_id(cate_id, integrate_cate_id):
actions = self.get_taobao_detail_actions()
is_success = True
price_info = [{
'skuId': '-2',
'price': yuan_2_cent(integrate_price)
}]
good_result.set_price_info(price_info=price_info)
good_result.set_flag(str(int(GoodDataType.success)))
for action in actions:
action.execute(context=context)
break
elif j < 6:
actions = self.get_taobao_http_detail_actions()
timestamps = int(datetime.now().timestamp() * 1000)
integrate_detail_url = SpiderUrls.get_taobao_detail_url(timestamps, '', integrate_item_id)
context = Context()
# referer = 'https://s.m.taobao.com/h5'
# context.attach(Context.KEY_HEADERS, SpiderHttp.get_taobao_headers(referer))
context.attach(Context.KEY_GOOD_DICT, good_result)
context.attach(Context.KEY_CURRENT_TASK, task)
context.attach(Context.KEY_COOKIES, RequestsCookieJar())
context.attach(Context.KEY_IS_UPDATE_COOKIES, True)
context.attach(Context.KEY_CURRENT_PROXY, proxy)
detail_m_url = SpiderUrls.get_detail_m_url(integrate_info['userType'], integrate_item_id)
detail_m_http_request = HttpRequest(detail_m_url, method=HttpMethod.GET)
context.attach(Context.KEY_DETAIL_M_HTTP_REQUEST, detail_m_http_request)
integrate_http_request = HttpRequest(url=integrate_detail_url, method=HttpMethod.GET)
context.attach(Context.KEY_TAOBAO_DETAIL_HTTP_REQUEST, integrate_http_request)
context.attach(Context.KEY_HEADERS, SpiderHttp.get_taobao_headers(detail_m_url))
try:
for action in actions:
action.execute(context=context)
is_success = True
break
except RetryException as e:
logger.exception(e)
time.sleep(5)
except InterruptException as e:
logger.exception(e)
time.sleep(10)
is_need_retry = True
# raise e
if not is_need_retry:
i += 1
j += 1
if not is_success:
actions = self.get_taobao_detail_actions()
good_result.set_flag(str(int(GoodDataType.not_found)))
for action in actions:
action.execute(context=context)
def do_execute(self, context):
account = context.get(Context.KEY_CURRENT_TASK_ACCOUNT)
# brands_result = context.get(Context.KEY_SYCM_PRODUCT_GET_BRANDS_RESULT)
hot_rank_result = context.get(
Context.KEY_SYCM_PRODUCT_PROD_HOT_RANK_RESULT)
hot_rank_result = hot_rank_result['data']
task = context.get(Context.KEY_CURRENT_TASK)
with write_session_scope() as session:
good = context.get(Context.KEY_GOOD_DICT)
stream_opt_data_dao = get_stream_opt_data_dao(session=session)
stream_unhandle_task_dao = get_stream_unhandle_task_dao(
session=session)
stream_handle_task_dao = get_stream_handle_task_dao(
session=session)
opt_data_entity = {
'raw_data': {
'hotRankResult': hot_rank_result,
'goodResult': good
},
'type': int(TaobaoPageType.sycm_list)
}
entity = stream_opt_data_dao.insert(**opt_data_entity)
unhandle_task_entities = []
for item in hot_rank_result:
c_good = Good(deepcopy(good))
c_good.set_brand_id(item['brandId'])
c_good.set_brand_name(item['brandName'])
c_good.set_model_id(item['modelId'])
c_good.set_model_name(item['modelName'])
c_good.set_sell_count(item['payItmCnt'])
unhandle_task_entities.append({
'raw_data': {
'hotRankInfo': item,
'goodResult': c_good,
'account': account
},
'type':
int(TaobaoTaskType.sycm_list),
'origin_id':
entity.id,
'date':
good['date']
})
stream_unhandle_task_dao.bulk_insert(unhandle_task_entities)
entity = stream_unhandle_task_dao.delete(
_filter=[CmmSysStreamUnhandleTask.id == task.id])
stream_handle_task_dao.insert(
**{
'type': task.type,
'raw_data': task.raw_data,
'origin_id': task.origin_id,
'date': good['date']
})
return True
def test_dict(self):
good = {'model': 'ccc'}
c_good = Good(good)
c_good.set_flag('1111')
print(good)
print(c_good)