def report_in(self, **kwargs):
if kwargs['flag'] == 'lookup':
res = mysql.get_data("update_reports",
l=1,
cn=["lookup"],
c={"link_id": "count"},
db=self.db_test)
mysql.update_data(t="update_reports",
set={
'lookup': res[0][0] + 1,
'last_time': kwargs['SpiderDate']
},
c={
'link_id': 'count',
'shop_id': kwargs['shop_id']
})
else:
self.report_item['stockid'] = kwargs['stockid']
self.report_item['link_id'] = kwargs['link_id']
self.report_item['shop_id'] = kwargs['shop_id']
self.report_item['price_tb'] = kwargs['price_tb']
# self.report_item['first_discount'] = kwargs['first_discount']
self.report_item['last_time'] = kwargs['SpiderDate']
self.report_item['attribute'] = kwargs['attribute']
self.report_item['flag'] = kwargs['flag']
self.report_item['description'] = kwargs['description']
mysql.insert_data(t="update_reports",
d=self.report_item,
db=self.db_test)
def save_in_sql(self, item, tableName):
if 'goodsCode' in item:
"""判断是否orderdetail"""
dict_select_condition = {
'orderNo': item['orderNo'],
'itemNo': item['itemNo']
}
result = mysql.get_data(t=tableName, c=dict_select_condition)
if result:
item.pop("goodsCode")
mysql.update_data(t=tableName,
set=item,
c=dict_select_condition)
# logger.info(time_now() + " " + concat(dict_select_condition, "|") + "|订单详情更新成功|")
else:
mysql.insert_data(t=tableName, d=item)
# logger.info(time_now() + " " + concat(dict_select_condition, "|") + "|新订单详情写入成功")
else:
dict_select_condition = {'orderNo': item['orderNo']}
result = mysql.get_data(t=tableName, c=dict_select_condition)
if result:
mysql.update_data(t=tableName,
set=item,
c=dict_select_condition)
# logger.info(time_now() + " " + concat(dict_select_condition, "|") + "|订单详情更新成功|")
else:
mysql.insert_data(t=tableName, d=item)
async def save(self):
"""
处理_parse()中获得的数据模型，并写入到数据库中
"""
async for i in self._parse():
if not i:
print("需要要切换淘宝账户")
return 0
res = mysql.get_data(db=test_server, t="tb_master",
c={'link_id': i['link_id']}, dict_result=True)
flag = ["update"]
narrative = []
if res:
if res[0]['price'] != i['price']:
flag.append("price")
narrative.append("更新销售价格:[{}]=>[{}]".format(res[0]['price'], i['price']))
if res[0]['promotionPrice'] != i['promotionPrice']:
flag.append("promotion")
narrative.append("更新优惠售价格:[{}]=>[{}]".format(res[0]['promotionPrice'], i['promotionPrice']))
if res[0]['sale_num'] != i['sale_num']:
flag.append("sale")
narrative.append("更新销量:[{}]=>[{}]".format(res[0]['sale_num'], i['sale_num']))
i['flag'] = "_".join(flag)
i['narrative'] = ";".join(narrative)
mysql.update_data(db=test_server, t='tb_master', set=i, c={"link_id": i['link_id']})
else:
i['flag'] = 'insert'
mysql.insert_data(db=test_server, t="tb_master", d=i)
def Verify():
l_orderNo = []
column_name = [
'orderNo', 'deliverFee', 'actualFee', 'couponPrice', 'fromStore',
'orderStatus'
]
condition = {'isVerify': '0', 'isDetaildown': '1'}
# kwargs = {'isVerify': '2', 'isDetaildown': '1'}
result = mysql.get_data(t="tb_order_spider", cn=column_name, c=condition)
if result:
for i in result:
total = 0
orderNo = i[0]
deliverFee = i[1]
actualFee = i[2]
couponPrice = i[3]
fromStore = i[4]
column_name = ['unitPrice', 'sellNum', 'unitBenefits']
condition = {'orderNo': orderNo}
result2 = mysql.get_data(t="tb_order_detail_spider",
cn=column_name,
c=condition)
for j in result2:
unitPrice = j[0]
sellNum = j[1]
unitBenefits = j[2]
total = total + unitPrice * sellNum - unitBenefits
a = round(total, 3) + deliverFee - actualFee - couponPrice
if int(a) != 0 and i[5] != '交易关闭':
list_tmp = []
list_tmp.append(str(round(total, 2)))
list_tmp.append(str(deliverFee))
list_tmp.append(str(actualFee))
list_tmp.append(str(couponPrice))
list_tmp.append(str(round(a, 2)))
list_tmp.append(store_trans(fromStore))
list_tmp.append(orderNo)
l_orderNo.append("|".join(list_tmp))
mysql.update_data(t="tb_order_spider",
set={'isVerify': 2},
c={'orderNo': orderNo})
else:
mysql.update_data(t="tb_order_spider",
set={'isVerify': 1},
c={'orderNo': orderNo})
# print('没有异常数据，验证完成！')
if l_orderNo:
s = "\n".join(l_orderNo)
# print(s)
mail("数据异常报告", s, ["*****@*****.**"])
taobao_check()
def main():
weather_list = datasource.read_data()
loop = asyncio.get_event_loop()
result_list = loop.run_until_complete(get_task(weather_list))
for id, result in result_list:
weather = util.safe_array_get(result.get('weather'), 0, {})
weather_main = weather.get('main')
weather_name = WeatherType(weather_main).japanese_name()
if weather_name is not None:
datasource.update_data(id, weather_name)
else:
slack.send_error_message(slack.error_message_format.format(id, result))
continue
def maintain(self, operation, **kwargs):
item = {
'stockid': kwargs['goodsCode'],
'link_id': kwargs['link_id'],
'shop_id': self.shop_id(kwargs['fromStore']),
'price_tb': kwargs['unitPrice'],
# 'first_discount': kwargs['unitBenefits'],
'currabrev': 'CNY',
'operator': '爬虫维护',
'SpiderDate': time_now(),
'attribute': kwargs['goodsAttribute'],
'flag': None,
'description': kwargs['tbName'],
'typeabbrev': "",
'price_erp': 0,
'last_time': time_now(),
'freight': "",
'ratio': 1,
'promotionprice': 0,
'sales': 0,
'rates': 0,
'Checker': "",
'package_number': 1,
'CheckDate': time_now(),
}
if operation == "更新":
item['flag'] = 'update'
item_set = {
'SpiderDate': time_now(),
'flag': 'update',
'price_tb': kwargs['unitPrice'],
'description': kwargs['tbName'],
'ratio': kwargs['ratio'],
'attribute': kwargs['goodsAttribute'],
# 'first_discount': kwargs['unitBenefits']
}
item_where = {
'stockid': kwargs['goodsCode'],
'link_id': kwargs['link_id'],
'shop_id': self.shop_id(kwargs['fromStore'])
}
mysql.update_data(t="prices_tb", set=item_set, c=item_where)
elif operation == "创建":
item['flag'] = 'create'
mysql.insert_data(t="prices_tb", d=item)
else:
item['flag'] = 'lookup'
self.report_in(**item)
async def run_link_spider(self):
sql = """
SELECT COUNT(a.id),fromStore FROM tb_order_detail_spider a
JOIN tb_order_spider b ON a.`orderNo`=b.`orderNo`
WHERE link_id="1" AND a.url IS NOT NULL
GROUP BY fromStore
ORDER BY COUNT(a.id) DESC
"""
time.sleep(2)
res = mysql.get_data(sql=sql)
if res:
b, p, f = await self.login(**STORE_INFO[res[0][1]])
await self.link_spider(p, f)
else:
mysql.update_data(t="tb_order_spider",
set={"isDetaildown": 0},
c={"isDetaildown": 2})
# if self.b:
# await self.b.close()
# self.b = None
await self.run_order_detail_spider()
def competitor_data():
if request.method == 'POST':
# print("adlsfjlsdjf")
result = request.form
res = result.to_dict()
stk_list = result.getlist("stockid[]")
res.pop("stockid[]")
attr_list = result.getlist("attribute[]")
res.pop("attribute[]")
price_list = result.getlist("price_tb[]")
res.pop("price_tb[]")
pgn_list = result.getlist("package_number[]")
res.pop("package_number[]")
for i in range(len(stk_list)):
item = res.copy()
item['stockid'] = stk_list[i]
if not item['stockid']:
continue
item['attribute'] = attr_list[i]
item['price_tb'] = price_list[i]
pgn = item.pop("package_number_t")
if int(pgn_list[i]) > 1:
item['package_number'] = pgn_list[i]
elif int(pgn) > 1:
item['package_number'] = pgn
else:
item['package_number'] = 1
c = {"stockid": item["stockid"], "link_id": item["link_id"]}
res_sql = mysql.get_data(c=c, t="prices_tb")
if res_sql:
mysql.update_data(set=item, c=c, t="prices_tb")
else:
mysql.insert_data(d=item, t="prices_tb")
# print(res)
# print(item)
return "添加成功!"
async def next_page(self, page_num=1):
"""执行翻页"""
temp = 0
while 1:
t = time_zone(["08:00", "18:00", "23:00"])
a = datetime.datetime.now()
if a < t[0]:
if not temp:
temp = 0
n_p_time = 600
elif t[0] < a < t[1]:
temp += 1
if temp == 1:
page_num = 1
n_p_time = NEXT_PAGE_TIME
elif a > t[2]:
n_p_time = 60
if not LINUX:
subprocess.call("shutdown /s")
exit("到点关机")
else:
n_p_time = 60
await self.page.bringToFront()
if self.orderno:
await self.page.focus("#bizOrderId")
await asyncio.sleep(1)
await self.page.keyboard.down("ShiftLeft")
await asyncio.sleep(1)
await self.page.keyboard.press("Home")
await asyncio.sleep(1)
await self.page.keyboard.down("ShiftLeft")
await asyncio.sleep(1)
await self.page.keyboard.press("Delete")
await asyncio.sleep(1)
orderno = input(time_now() + " | 输入订单号:")
await self.page.type("#bizOrderId", orderno)
await self.page.setRequestInterception(True)
self.page.on('request', self.intercept_request)
self.page.on('response', self.intercept_response)
net_check()
await self.page.click(".button-mod__primary___17-Uv")
await asyncio.sleep(10)
else:
while 1:
try:
await self.page.waitForSelector(
".pagination-options-go")
await self.page.focus(".pagination-options input")
# await self.page.click(".pagination-options input", clickCount=2)
await self.page.keyboard.press("Delete")
await self.page.keyboard.press("Delete")
await self.page.keyboard.press("Delete")
await self.page.keyboard.press("Backspace")
await self.page.keyboard.press("Backspace")
await self.page.keyboard.press("Backspace")
await self.page.setRequestInterception(True)
self.page.on('request', self.intercept_request)
self.page.on('response', self.intercept_response)
net_check()
await self.page.type(".pagination-options input",
str(page_num))
await self.page.keyboard.press("Enter")
self.page.waitForSelector(
".pagination-item.pagination-item-" +
str(page_num) + ".pagination-item-active",
timeout=10000)
except errors.TimeoutError:
logger.info('翻页超时，5秒后重新翻页')
sleep(5)
else:
break
# await self.page.waitForSelector(".pagination-item-" + str(page_num) + " a", timeout=30000)
# await self.page.click(".pagination-item-" + str(page_num) + " a")
while 1:
if self.complete == 1:
s = random.random()
if s > 0.5:
await self.link_spider()
await self.order_page()
logger.info(str(int(s * n_p_time)) + " 秒后开始下一页爬取")
sleep(int(s * n_p_time))
break
elif self.complete == 2:
page_num = 0
s = random.random()
if s > 0.9:
mysql.update_data(t="tb_order_spider",
set={"isDetaildown": 0},
c={
"isDetaildown": 2,
"fromStore": self.fromStore
})
sleep(int(s * n_p_time))
break
else:
# if i == 59:
# logger.info("超时")
# await self.page.screenshot({'path': './headless-test-result.png'})
await asyncio.sleep(3)
self.complete = 0
page_num += 1
async def parse_2(self, data):
verify = []
repeat_list = []
for i in data['skuOuterIdTable']['dataSource']:
self.item['stockid'] = i['skuOuterId']
logger.info(self.item['stockid'])
if not self.item['stockid']:
continue
else:
if self.item['stockid'] not in verify:
verify.append(self.item['stockid'])
else:
if self.item['stockid'] not in repeat_list:
repeat_list.append(self.item['stockid'])
skuId = str(i['skuId'])
temp_attr = self.prop.get(skuId)
self.item['attribute'] = temp_attr.replace("（",
"(").replace("）", ")")
if not self.item['attribute']:
self.item.pop('attribute')
self.item['price_tb'] = self.prices.get(skuId)
if self.promo_price:
self.item["promotionprice"] = self.promo_price.get(skuId)
else:
self.item["promotionprice"] = 0
condition = {
"stockid": self.item['stockid'],
"link_id": self.item['link_id'],
"shop_id": self.item['shop_id'],
}
res = mysql.get_data(t="prices_tb",
l=1,
cn=["price_tb"],
c=condition,
db=self.target_server)
if res:
if res[0][0] == 0:
self.item['ratio'] = 1
else:
self.item['ratio'] = round(
float(self.item['price_tb']) / float(res[0][0]), 2)
print(self.item)
mysql.update_data(t="prices_tb",
set=self.item,
c=condition,
db=self.target_server)
else:
insert_item = self.item.copy()
insert_item["currabrev"] = "CNY"
insert_item["price_erp"] = 0
insert_item["operator"] = self.operator
insert_item["last_time"] = time_now()
if self.operator == "爬虫维护":
insert_item["flag"] = "create"
else:
insert_item['flag'] = "add"
insert_item["ratio"] = 1
insert_item["package_number"] = 1
insert_item["Checker"] = ""
insert_item["CheckDate"] = "0000-00-00 00:00:00"
print(insert_item)
with open("reports/report_" + self.fromStore + "_insert.txt",
"a") as file:
file.write("物料编码：" + insert_item['stockid'] + " 与商品ID：" +
insert_item['link_id'] + " 为最新匹配，添加至ERP系统。\n" +
self.item_url + insert_item['link_id'] + "\n" +
self.item_erp_url + insert_item['link_id'] +
"\n\n")
mysql.insert_data(t="prices_tb",
d=insert_item,
db=self.target_server)
if repeat_list:
with open("reports/report_" + self.fromStore + "_repeat.txt",
"a") as file:
file.write("店铺：" + store_trans(self.fromStore) + ",商品id:" +
self.item['link_id'] + " 重复编码\n" + "重复编码:" +
",".join(repeat_list) + "\n" + self.item_url +
self.item['link_id'] + "\n\n")
if not verify:
with open("reports/report_" + self.fromStore + "_empty.txt",
"a") as file:
file.write("店铺：" + store_trans(self.fromStore) + ",商品id:" +
self.item['link_id'] + " 空编码\n" + self.item_url +
self.item['link_id'] + "\n\n")
sql = """
select id,stockid
from prices_tb
where link_id='%s'
and flag not in('del','XiaJia')
""" % (self.item['link_id'])
res_verify = mysql.get_data(sql=sql, db=self.target_server)
for rv in res_verify:
if rv[1] not in verify:
with open("reports/report_" + self.fromStore + "_delete.txt",
"a") as file:
file.write("物料编码：" + rv[1] + " 与 商品ID：" +
self.item['link_id'] + " 不匹配，已被爬虫从ERP系统中删除。\n" +
self.item_url + self.item['link_id'] + "\n" +
self.item_erp_url + self.item['link_id'] +
"\n\n")
mysql.update_data(t="prices_tb",
c={"id": rv[0]},
db=self.target_server,
set={
"flag": "del",
"operator": self.operator,
"last_time": time_now()
})
self.complete = 1
async def order_detail_spider(self, p, f):
sql1 = """
SELECT datailURL,a.orderNo FROM tb_order_spider a
JOIN taobaoorders b ON a.orderNo = b.OrderNo
WHERE isDetaildown=0 AND fromStore='%s' AND b.Flag = 8
ORDER BY createTime DESC;
""" % (f)
sql = """
SELECT datailURL,orderNo FROM tb_order_spider
WHERE isDetaildown=0 AND fromStore='%s'
ORDER BY createTime DESC
""" % (f)
results = mysql.get_data(sql=sql1, dict_result=True)
if not results:
results = mysql.get_data(sql=sql, dict_result=True)
if results:
for result in results:
order = {}
url = result['datailURL']
try:
net_check()
await p.goto(url)
except errors.TimeoutError:
continue
slider = await p.J('#nocaptcha')
if slider:
while True:
print("出现滑块验证码")
await asyncio.sleep(2)
await p.hover('#nc_1_n1z')
await p.mouse.down()
await p.mouse.move(
2000, 0, {'delay': random.randint(1000, 2000)})
await p.mouse.up()
try:
p.waitForSelector(".nc-lang-cnt a", timeout=10000)
await asyncio.sleep(2)
await p.click(".nc-lang-cnt a")
except errors.TimeoutError:
break
except errors.PageError:
break
try:
await p.waitForSelector('#detail-panel', timeout=30000)
except Exception as e:
continue
content = await p.content()
a = re.search("var data = JSON.parse\('(.*)'\);",
content).group(1)
b = a.replace('\\\\\\"', '')
data = b.replace('\\"', '"')
m = json.loads(data)
order['actualFee'] = m['mainOrder']['payInfo']['actualFee'][
'value']
order['orderStatus'] = status_format(
m['mainOrder']['statusInfo']['text'])
if order['orderStatus'] == '等待买家付款':
order['isDetaildown'] = 2
else:
order['isDetaildown'] = 1
coupon = 0
for k, v in m['mainOrder']['payInfo'].items():
if k == 'promotions':
promotions = m['mainOrder']['payInfo']['promotions']
for i in range(len(promotions)):
if 'prefix' and 'suffix' in promotions[i]:
coupon_temp = re.search(
"(\d+\.\d+)", promotions[i]['value'])
if coupon_temp:
coupon += float(coupon_temp.group(1))
order['couponPrice'] = round(coupon, 2)
for k, v in m.items():
if k == 'buyMessage':
order['buyerComments'] = v
orderNo = m['mainOrder']['id']
order_info = m['mainOrder']['orderInfo']['lines'][1]['content']
for i in range(len(order_info)):
if order_info[i]['value']['name'] == '支付宝交易号:':
try:
order['tradeNo'] = order_info[i]['value']['value']
except KeyError:
order['tradeNo'] = None
# elif order_info[i]['value']['name'] == '创建时间:':
# order['createTime'] = order_info[i]['value']['value']
# elif order_info[i]['value']['name'] == '发货时间:':
# order['shipTime'] = order_info[i]['value']['value']
elif order_info[i]['value']['name'] == '付款时间:':
order['payTime'] = order_info[i]['value']['value']
ship_info = m['tabs']
for i in range(len(ship_info)):
if ship_info[i]['id'] == 'logistics':
temp = ship_info[i]['content']
for k, v in temp.items():
if k == 'logisticsName':
order['shippingCompany'] = v
elif k == 'shipType':
order['shippingMethod'] = v
elif k == 'logisticsNum':
order['shippingNo'] = v
# elif k == 'logisticsUrl':
# order['shipUrl'] = "https" + v
elif k == 'address':
rec_info = v
order['receiverName'] = rec_info.split(
"，")[0].replace(" ", "")
order['receiverPhone'] = rec_info.split("，")[1]
order['receiverAddress'] = "".join(
rec_info.split("，")[2:])
sub_orders = m['mainOrder']['subOrders']
# mainOrder.subOrders[10].tradeStatus[0].content[0].value
line_no = 0
for i in range(len(sub_orders)):
if sub_orders[i]['tradeStatus'][0]['content'][0][
'value'] == '已取消':
continue
item = {}
temp = 0
itemNo = line_no
line_no += 1
if sub_orders[i]['promotionInfo']:
for j in sub_orders[i]['promotionInfo']:
for x in j['content']:
for k, v in x.items():
if k == 'value':
p_list = re.findall("-?\d+\.\d+", v)
if p_list:
temp += float(p_list.pop())
item['unitBenefits'] = temp
mysql.update_data(t="tb_order_detail_spider",
set=item,
c={
'orderNo': orderNo,
'itemNo': itemNo
})
mysql.update_data(t="tb_order_spider",
set=order,
c={'orderNo': orderNo})
Verify()
while True:
s = random.random()
if s > 0.9:
for i in range(int(s * n_o_time)):
await asyncio.sleep(1)
print(">", end="", flush=True)
print("")
break
else:
pass
await p.close()
await self.run_order_detail_spider()
async def link_spider(self, p, f):
test_server["db"] = "test"
while True:
sql = """
SELECT a.id,url,goodsCode,a.orderNo FROM tb_order_detail_spider a
JOIN tb_order_spider b ON a.`orderNo`=b.`orderNo`
WHERE link_id="1" AND b.`fromStore`='%s' AND a.url IS NOT NULL
ORDER BY b.createTime DESC
LIMIT 1
""" % (f)
url = "https://smf.taobao.com/promotionmonitor/orderPromotionQuery.htm?orderNo="
results = mysql.get_data(sql=sql, dict_result=True)
if not results:
break
orderno = results[0]['orderNo']
url += orderno
await p.goto(url)
content = await p.content()
data = re.findall(">(\{.*?\})<", content)
order = json.loads(data[0])
try:
sub_orders = order["data"]["subOrderViewDTOs"]
except KeyError:
continue
for so in sub_orders:
order_no = so["orderNoStr"]
link_id = so["itemId"]
sql = "select goodsCode from tb_order_detail_spider where url like '%%%s%%'" % (
order_no)
print(sql)
goodsCode = mysql.get_data(sql=sql, return_one=True)
del sql
sql = "update tb_order_detail_spider set link_id='%s' where url like '%%%s%%'" % (
link_id, order_no)
mysql.update_data(sql=sql)
del sql
sql = """
SELECT SpiderDate
FROM prices_tb
WHERE link_id='%s'
AND stockid='%s'
AND flag NOT IN ('del','XiaJia')
""" % (link_id, goodsCode)
res = mysql.get_data(sql=sql)
res_fix = mysql.get_data(db=test_server,
dict_result=True,
t='prices_tb_fix',
c={
"link_id": link_id,
"server": "production_server"
})
if res:
spider_date = res[0][0]
days = 1
if spider_date != '0000-00-00 00:00:00':
days = (datetime.datetime.now() - spider_date).days
if spider_date == '0000-00-00 00:00:00' or days > 14:
if not res_fix:
mysql.insert_data(db=test_server,
t="prices_tb_fix",
d={
"link_id": link_id,
"fromStore": f,
"flag": 1
})
elif res_fix[0]["isComplete"] != 0:
mysql.update_data(db=test_server,
t="prices_tb_fix",
set={
"isComplete": 0,
"flag": 1
},
c={
"link_id": link_id,
"server": "production_server"
})
else:
if not res_fix:
mysql.insert_data(db=test_server,
t="prices_tb_fix",
d={
"link_id": link_id,
"fromStore": f,
"flag": 0
})
elif res_fix[0]["isComplete"] != 0:
mysql.update_data(db=test_server,
t="prices_tb_fix",
set={
"flag": 0,
"isComplete": 0
},
c={
"link_id": link_id,
"server": "production_server"
})
sleep(5)
await p.close()
await self.run_link_spider()
def report_mail(self):
d = time_zone(["18:05", "18:05"])
d1, d2 = d[0], d[1]
d = (d1 - datetime.timedelta(days=1)).strftime("%Y-%m-%d %H:%M:%S")
sql = "SELECT shop_id,flag,COUNT(flag),lookup FROM update_reports " \
"WHERE last_time < '%s' AND last_time > '%s' " \
"GROUP BY Flag,shop_id" % (d1, d)
sql2 = "SELECT * FROM update_reports WHERE last_time < '%s' AND last_time > '%s' " % (
d1, d)
res = mysql.get_data(sql=sql, db=self.db_test)
con, c, cd = mysql.connection(self.db_test)
df = pd.read_sql(sql2, con)
con.close()
date = date_now_str()
df.to_csv("./reports/reports" + date + ".csv")
out_list = []
out_list.append("今日爬虫维护 开源店 价格 ：<br>")
for r in res:
# print(r)
if r[0] == '115443253':
if r[1] == 'create':
string = '创建了 ' + str(r[2]) + ' 条数据。<br>'
out_list.append(string)
elif r[1] == 'update':
string = '更新了 ' + str(r[2]) + ' 条数据。<br>'
out_list.append(string)
elif r[1] == 'lookup':
string = '查看了 ' + str(r[3]) + ' 条数据。<br>'
out_list.append(string)
out_list.append("今日爬虫维护 玉佳企业店 价格：<br>")
for r in res:
# print(r)
if r[0] == '197444037':
if r[1] == 'create':
string = '创建了 ' + str(r[2]) + ' 条数据。<br>'
out_list.append(string)
elif r[1] == 'update':
string = '更新了 ' + str(r[2]) + ' 条数据。<br>'
out_list.append(string)
elif r[1] == 'lookup':
string = '查看了 ' + str(r[3]) + ' 条数据。<br>'
out_list.append(string)
out_list.append("今日爬虫维护 赛宝电子店 价格：<br>")
for r in res:
# print(r)
if r[0] == '34933991':
if r[1] == 'create':
string = '创建了 ' + str(r[2]) + ' 条数据。<br>'
out_list.append(string)
elif r[1] == 'update':
string = '更新了 ' + str(r[2]) + ' 条数据。<br>'
out_list.append(string)
elif r[1] == 'lookup':
string = '查看了 ' + str(r[3]) + ' 条数据。<br>'
out_list.append(string)
out_list.append("今日爬虫维护 玉佳电子店 价格：<br>")
for r in res:
# print(r)
if r[0] == '68559944':
if r[1] == 'create':
string = '创建了 ' + str(r[2]) + ' 条数据。<br>'
out_list.append(string)
elif r[1] == 'update':
string = '更新了 ' + str(r[2]) + ' 条数据。<br>'
out_list.append(string)
elif r[1] == 'lookup':
string = '查看了 ' + str(r[3]) + ' 条数据。<br>'
out_list.append(string)
# print("".join(out_list))
mail_reports("爬虫更新erp价格报告", "".join(out_list), date,
*["*****@*****.**", '*****@*****.**']) #
dt = (d1 - datetime.timedelta(days=3)).strftime("%Y-%m-%d %H:%M:%S")
print(dt)
sql = "delete from update_reports where last_time<'%s'" % (dt)
mysql.delete_data(sql=sql, db=self.db_test)
mysql.update_data(t="update_reports",
set={"loopup": 0},
c={"link_id": "count"})
async def fix_data(self, link_id=None):
# page = await self.browser.newPage()
self.complete = 0
self.prices = {}
self.promo_price = {}
await asyncio.sleep(2)
await self.page.focus("input[name='queryItemId']")
await self.page.keyboard.down("ShiftLeft")
await self.page.keyboard.press("Home")
await self.page.keyboard.down("ShiftLeft")
await self.page.keyboard.press("Delete")
server_name = 'production_server'
self.sn = server_name
if not link_id:
if MODE == 1:
link_id = "585308692855"
elif MODE == 2:
while True:
link_id = input(time_now() + " | 输入link_id：")
isMatch = re.match("^\d{10,20}$", link_id)
if isMatch:
break
elif MODE == 3:
sql = """
SELECT link_id,updateTime,server,operator
FROM prices_tb_fix
WHERE fromStore='%s' and isComplete=0
ORDER BY flag LIMIT 1
""" % (self.fromStore)
res = mysql.get_data(sql=sql, db=self.db_test)
if res:
self.target_server = self.server[res[0][2]]
link_id = res[0][0]
updateTime = res[0][1]
server_name = res[0][2]
self.sn = server_name
self.operator = res[0][3]
else:
return 1
logger.info(link_id)
page = await self.browser.newPage()
await page.setViewport({'width': 1600, 'height': 900})
net_check()
await page.goto("https://item.taobao.com/item.htm?id=" + link_id,
timeout=0)
await asyncio.sleep(3)
error_page = await page.J(".error-notice-hd") # 判断宝贝是否正常在售
offline = await page.J("#J_detail_offline") # 判断宝贝是否正常在售
if error_page or offline:
logger.info("商品已下架")
mysql.update_data(t="prices_tb",
set={
"flag": "XiaJia",
"typeabbrev": self.fromStore
},
c={"link_id": link_id},
db=self.target_server)
# mysql.update_data(t="tb_order_detail_spider",
# set={"link_id": link_id + "xiajia"},
# c={"link_id": link_id},
# db=self.target_server)
mysql.update_data(db=self.db_test,
t="prices_tb_fix",
set={
"isComplete": "2",
"updateTime": time_now()
},
c={
"link_id": link_id,
"server": server_name
})
await page.close()
return
else:
while True:
content = await page.content()
# print(content)
doc = pq(content)
self.common['rates'] = doc.find("#J_RateCounter").text()
self.common['sales'] = doc.find("#J_SellCounter").text()
self.common['freight'] = doc.find("#J_WlServiceTitle").text()
mat1 = re.match("\d+", self.common['sales'])
mat2 = re.match("\d+", self.common['rates'])
if mat1 and mat2:
break
res = re.findall('";(.*?);".*?e":"(\d+\.\d+).*?d":"(\d+)"',
content) # 判断是否存在多属性
if res:
control = 1
benefit_price = 0
for r in res:
data_values = r[0].split(";")
prop = []
for data in data_values:
prop.append(
doc.find("li[data-value='" + data +
"'] span").text())
if control:
for data in data_values:
try:
await page.click('li[data-value="' + data +
'"]')
except errors.PageError:
pass
content_p = await page.content()
promo_price = re.findall(
'<em id="J_PromoPriceNum".*?>(\d+\.?\d*)</em>',
content_p) # 判断是否存在优惠
if len(promo_price) == 1:
benefit_price = float(r[1]) - float(promo_price[0])
control = 0
self.prices[r[2]] = r[1]
prop.reverse()
self.prop[r[2]] = "-".join(prop)
for r in res:
if benefit_price:
self.promo_price[r[2]] = round(
float(r[1]) - benefit_price, 2)
else:
promo_price = re.findall(
'<em id="J_PromoPriceNum".*?>(\d+.*\d*)</em>',
content) # 判断是否存在优惠
if promo_price:
self.promo_price[link_id] = promo_price[0]
else:
self.promo_price[link_id] = 0
# print(self.prices)
# print(self.promo_price)
await page.close()
await self.page.type("input[name='queryItemId']", link_id)
await self.page.setRequestInterception(True)
self.page.on('request', self.intercept_request)
self.page.on('response', self.intercept_response)
await asyncio.sleep(1)
net_check()
await self.page.click(".filter-footer button:first-child")
while True:
await asyncio.sleep(1)
if self.complete == 1:
res = mysql.get_data(db=self.db_test,
t="prices_tb_fix",
c={
"link_id": link_id,
"server": server_name
})
if res:
mysql.update_data(db=self.db_test,
t="prices_tb_fix",
set={
"isComplete": "1",
"updateTime": time_now()
},
c={
"link_id": link_id,
"server": server_name
})
break
elif self.complete == 2:
mysql.update_data(db=self.db_test,
t="prices_tb_fix",
set={"spe_link": "1"},
c={
"link_id": link_id,
"server": server_name
})
break
async def parse(self, data):
if data != "q":
for i in range(len(data)):
self.item = {}
self.item = self.common.copy()
self.item['stockid'] = re.search(
"编码:(.*)",
data[i]['itemDesc']['desc'][1]['text']).group(1).upper()
self.item['link_id'] = data[i]['itemId']
self.item['attribute'] = ""
self.item['flag'] = "update"
self.item['typeabbrev'] = self.fromStore
self.item['shop_id'] = self.shop_id(self.fromStore)
self.item['SpiderDate'] = time_now()
temp_des = data[i]['itemDesc']['desc'][0]['text']
self.item['description'] = temp_des.replace("（", "(").replace(
"）", ")")
self.item['price_tb'] = re.findall(
"(\d+.?\d*)", data[i]["managerPrice"]['currentPrice'])[0]
self.item['promotionprice'] = self.promo_price.get(
self.item['link_id'])
# print(self.promo_price)
sql = "select spe_link from prices_tb_fix where link_id='%s' and server='%s'" % (
self.item['link_id'], self.sn)
spe_link_id = mysql.get_data(db=self.db_test,
sql=sql,
return_one=True)
isMut = re.search("^MUT\D*", self.item['stockid'])
if isMut or spe_link_id:
await self.page.setRequestInterception(True)
self.page.on('request', self.intercept_request)
self.page.on('response', self.intercept_response)
await asyncio.sleep(1)
net_check()
await self.page.click(
".next-table-row td:nth-child(2) div.product-desc-hasImg span:nth-child(2) i"
)
await asyncio.sleep(1)
await self.page.keyboard.press('Escape')
else:
# print(self.item)
if self.item['promotionprice'] is None:
mail("price_tb_error",
self.fromStore + ":" + self.item['link_id'],
["*****@*****.**"])
logger.error("error:" + self.fromStore + " : " +
self.item['link_id'] + " and " +
mysql.concat(self.promo_price, "="))
self.complete = 2
break
condition = {
"stockid": self.item['stockid'],
"link_id": self.item['link_id'],
"shop_id": self.item['shop_id'],
}
res = mysql.get_data(t="prices_tb",
l=1,
cn=["id"],
c=condition,
db=self.target_server)
if res:
self.item['ratio'] = round(
float(self.item['price_tb']) / float(res[0][0]), 2)
print(self.item)
mysql.update_data(t="prices_tb",
set=self.item,
c=condition,
db=self.target_server)
else:
insert_item = self.item.copy()
insert_item["currabrev"] = "CNY"
insert_item["price_erp"] = 0
insert_item["operator"] = self.operator
insert_item["last_time"] = time_now()
if self.operator == "爬虫维护":
insert_item["flag"] = "create"
else:
insert_item['flag'] = "add"
insert_item["ratio"] = 1
insert_item["package_number"] = 1
insert_item["Checker"] = ""
insert_item["CheckDate"] = "0000-00-00 00:00:00"
print(insert_item)
with open(
"reports/report_" + self.fromStore +
"_insert.txt", "a") as file:
file.writelines("物料编码：" + insert_item['stockid'] +
" 与 商品ID：" +
insert_item['link_id'] +
" 为最新匹配，添加至ERP系统。\n" +
self.item_url +
insert_item['link_id'] + "\n" +
self.item_erp_url +
insert_item['link_id'] + "\n\n")
mysql.insert_data(t="prices_tb",
d=insert_item,
db=self.target_server)
result = mysql.get_data(
t="prices_tb",
cn=["*"],
c={"link_id": self.item['link_id']},
db=self.target_server,
dict_result=True)
if len(result) > 1:
for r in result:
if r['stockid'] != self.item['stockid'] and r[
'flag'] != "del":
with open(
"reports/report_" + self.fromStore +
"_delete.txt", "a") as file:
file.writelines("物料编码：" + r['stockid'] +
" 与 商品ID：" +
self.item['link_id'] +
" 不匹配，已被爬虫从ERP系统中删除。\n" +
self.item_url +
self.item['link_id'] +
"\n" + self.item_erp_url +
self.item['link_id'] +
"\n\n")
mysql.update_data(t="prices_tb",
c={"id": r['id']},
db=self.target_server,
set={"flag": "del"})
self.complete = 1
else:
pass
self.complete = 1
async def order_page(self, browser_in=None, page_in=None):
"""爬取订单详情"""
while 1:
result = mysql.get_data(t="tb_order_spider",
cn=["datailURL", "orderNo"],
c={
"isDetaildown": 0,
"fromStore": self.fromStore
},
o=["createTime"],
om="d")
if result:
logger.info("订单详情爬取")
for url in result:
start_time = datetime.datetime.now()
logger.info(store_trans(self.fromStore))
logger.info("开始订单 " + url[1] + " 详情爬取")
order = {}
await self._page_order_detail.bringToFront()
# if browser_in:
# page = await browser_in.newPage()
# else:
# page = page_in
page = self._page_order_detail
while 1:
try:
await page.goto(url[0])
except errors.PageError:
sleep(5)
except errors.TimeoutError:
sleep(5)
else:
break
try:
await page.waitForSelector('#detail-panel',
timeout=30000)
except errors.TimeoutError:
continue
content = await page.content()
a = re.search("var data = JSON.parse\('(.*)'\);",
content).group(1)
b = a.replace('\\\\\\"', '')
data = b.replace('\\"', '"')
m = json.loads(data)
order['actualFee'] = m['mainOrder']['payInfo'][
'actualFee']['value']
order['orderStatus'] = status_format(
m['mainOrder']['statusInfo']['text'])
if order['orderStatus'] == '等待买家付款':
order['isDetaildown'] = 2
else:
order['isDetaildown'] = 1
coupon = 0
for k, v in m['mainOrder']['payInfo'].items():
if k == 'promotions':
promotions = m['mainOrder']['payInfo'][
'promotions']
for i in range(len(promotions)):
if 'prefix' and 'suffix' in promotions[i]:
coupon_temp = re.search(
"(\d+\.\d+)", promotions[i]['value'])
if coupon_temp:
coupon += float(coupon_temp.group(1))
order['couponPrice'] = round(coupon, 2)
for k, v in m.items():
if k == 'buyMessage':
order['buyerComments'] = v
orderNo = m['mainOrder']['id']
order_info = m['mainOrder']['orderInfo']['lines'][1][
'content']
for i in range(len(order_info)):
if order_info[i]['value']['name'] == '支付宝交易号:':
try:
order['tradeNo'] = order_info[i]['value'][
'value']
except KeyError:
order['tradeNo'] = None
# elif order_info[i]['value']['name'] == '创建时间:':
# order['createTime'] = order_info[i]['value']['value']
# elif order_info[i]['value']['name'] == '发货时间:':
# order['shipTime'] = order_info[i]['value']['value']
elif order_info[i]['value']['name'] == '付款时间:':
order['payTime'] = order_info[i]['value']['value']
ship_info = m['tabs']
for i in range(len(ship_info)):
if ship_info[i]['id'] == 'logistics':
temp = ship_info[i]['content']
for k, v in temp.items():
if k == 'logisticsName':
order['shippingCompany'] = v
elif k == 'shipType':
order['shippingMethod'] = v
elif k == 'logisticsNum':
order['shippingNo'] = v
# elif k == 'logisticsUrl':
# order['shipUrl'] = "https" + v
elif k == 'address':
rec_info = v
order['receiverName'] = rec_info.split(
"，")[0].replace(" ", "")
order['receiverPhone'] = rec_info.split(
"，")[1]
order['receiverAddress'] = "".join(
rec_info.split("，")[2:])
sub_orders = m['mainOrder']['subOrders']
# print(len(sub_orders))
for i in range(len(sub_orders)):
item = {}
temp = 0
itemNo = i
if sub_orders[i]['promotionInfo']:
for j in sub_orders[i]['promotionInfo']:
for x in j['content']:
for k, v in x.items():
if k == 'value':
p_list = re.findall(
"-?\d+\.\d+", v)
if p_list:
temp += float(p_list.pop())
item['unitBenefits'] = temp
mysql.update_data(t="tb_order_detail_spider",
set=item,
c={
'orderNo': orderNo,
'itemNo': itemNo
})
logger.info("详细订单状态更新成功")
# print(item)
# print(order)
mysql.update_data(t="tb_order_spider",
set=order,
c={'orderNo': orderNo})
logger.info("订单状态更新成功")
# if browser_in:
# await page.close()
await self.page.bringToFront()
Verify()
end_time = datetime.datetime.now()
spend_time = end_time - start_time
logger.info(
str(spend_time.seconds) + " 秒完成订单 " + url[1] + " 详情爬取")
while True:
s = random.random()
if s > 0.3:
logger.info("休息 " + str(int(s * n_o_time)) +
" 秒完开始下一单详情爬取")
for i in range(int(s * n_o_time)):
await asyncio.sleep(1)
break
else:
logger.info("没有可以爬取的详情")
break
async def parse(self, mainOrders, pageNum):
"""解析爬取内容信息"""
t = time_zone(["08:00", "18:00", "23:59"])
a = datetime.datetime.now()
if a < t[0]:
eoc = EARLIEST_ORDER_CREATETIME
elif t[0] < a < t[1]:
eoc = 2
else:
eoc = 20
start_time = datetime.datetime.now()
logger.info("开始第 " + str(pageNum) + " 页订单爬取")
logger.info(store_trans(self.fromStore))
if pageNum == 1:
self._loop_start_time = datetime.datetime.now()
loop_control = 0
for i in range(len(mainOrders)):
order = {} # 用于存储订单详细信息
order['orderNo'] = mainOrders[i]["id"]
order['createTime'] = mainOrders[i]['orderInfo']['createTime']
order['buyerName'] = mainOrders[i]['buyer']['nick']
flag = mainOrders[i]['extra']['sellerFlag']
order['actualFee'] = mainOrders[i]['payInfo']['actualFee']
order['deliverFee'] = re.search(
"\(含快递:￥(\d+\.\d+)\)",
mainOrders[i]['payInfo']['postType']).group(1)
order['datailURL'] = "https:" + mainOrders[i]['statusInfo'][
'operations'][0]['url']
order['orderStatus'] = mainOrders[i]['statusInfo']['text']
order['fromStore'] = self.fromStore
order['updateTime'] = time_now()
if flag == 1:
data_url = self.base_url + mainOrders[i]['operations'][0][
'dataUrl']
order['sellerFlag'] = await self.get_flag_text(data_url)
try:
order['isPhoneOrder'] = mainOrders[i]['payInfo']['icons'][0][
'linkTitle']
except KeyError:
pass
items = mainOrders[i]['subOrders']
line_no = 0
for j in range(len(items)):
continue_code = 0
item = {} # 用于存储售出商品详细信息
item['orderNo'] = mainOrders[i]["id"]
item['itemNo'] = line_no
try:
item['goodsCode'] = items[j]['itemInfo']['extra'][0][
'value']
except KeyError:
item['goodsCode'] = 'error'
logger.error(time_now() + " 订单：" + item['orderNo'])
item['tbName'] = items[j]['itemInfo']['title'].strip() \
.replace("±", "±").replace("Φ", "Φ").replace("Ω", "Ω") \
.replace("—", "—").replace("°", "°").replace("×", "×") \
.replace("μ", "μ").replace(" ", "").replace("（", "(").replace("）", ")")
item['unitPrice'] = items[j]['priceInfo']['realTotal']
item['sellNum'] = items[j]['quantity']
item['orderStatus'] = order['orderStatus']
if self.orderno:
logger.info(item['orderStatus'])
item['refundStatus'] = None
item['isRefund'] = 0
item['goodsAttribute'] = ""
item['url'] = "https:" + items[j]['itemInfo']['itemUrl']
try:
goodsAttributes = items[j]['itemInfo']['skuText']
except KeyError:
pass
else:
temp = []
for k in range(len(goodsAttributes)):
try:
goodsAttributes[k]['name']
except KeyError:
n = len(temp)
temp[n - 1] += goodsAttributes[k]['value'].replace(
"&Omega", "Ω").replace("·", "·")
else:
temp.append(goodsAttributes[k]['value'].replace(
"&Omega", "Ω").replace("·", "·"))
temp_ga = "-".join(temp)
item['goodsAttribute'] = temp_ga.replace("（", "(").replace(
"）", ")")
try:
operations = items[j]['operations']
except KeyError:
pass
else:
for x in range(len(operations)):
t = operations[x]['style']
if t in ['t12', 't16'
] and operations[x]['text'] != "退运保险":
item['refundStatus'] = operations[x]['text']
item['isRefund'] = "1"
elif t == 't0' and operations[x]['text'] == '已取消':
continue_code = 1
delete_item = {
'orderNo': item['orderNo'],
'itemNo': item['itemNo'],
'goodsCode': item['goodsCode']
}
is_exist = mysql.get_data(
t="tb_order_detail_spider", l=1, c=delete_item)
if is_exist:
mysql.delete_data(t="tb_order_detail_spider",
c=delete_item)
sql = """
UPDATE tb_order_detail_spider
SET itemNo=itemNo-1
WHERE OrderNo='%s' and itemNo>'%s'
""" % (item['orderNo'], item['itemNo'])
mysql.update_data(sql=sql)
pass
if continue_code:
continue
else:
line_no += 1
self.save_in_sql(item=item, tableName='tb_order_detail_spider')
self.save_in_sql(item=order, tableName='tb_order_spider')
if self.orderno:
logger.info("定向爬取订单完成")
return
date = datetime.date.today()
date_limit = (
date - datetime.timedelta(eoc)).strftime("%Y-%m-%d %H:%M:%S")
if order['createTime'] < date_limit:
logger.info("完成本轮爬取，共翻 " + str(pageNum) + " 页。")
loop_control = 1
break
end_time = datetime.datetime.now()
spend_time = end_time - start_time
logger.info(
str(spend_time.seconds) + " 秒完成第 " + str(pageNum) + " 页订单爬取")
if loop_control:
self._loop_end_time = datetime.datetime.now()
loop_spend_time = round(
(self._loop_end_time - self._loop_start_time).seconds / 60, 0)
logger.info(str(loop_spend_time) + " 分钟完成本轮订单爬取")
self.complete = 2
else:
self.complete = 1
