async def _booking(request):
valid = await validate_request(request)
if valid["errmsg"]:
return rest_result(request, {"status": 400, "errmsg": valid["errmsg"]})
cms_ids = valid["hotels"]
db = databases("hub")
start_time = valid["start_time"]
end_time = valid["end_time"]
days = valid["days"]
# 避免传入的开始时间早于当前日期导致无效查询
# TODO: 抽象成单独校验日期的方法
for index, hid in enumerate(cms_ids):
booking_url = await db["poi_items"].find_one(
{
"_id": ObjectId(hid),
"crawl_info.crawl_website": "bk_url"
},
{"crawl_info.$": "1"},
)
if not booking_url:
continue
await get_booking_prices(booking_url["crawl_info"][0]["crawl_url"],
hid, start_time, days)
calendar_one.delay(hid, start_time, end_time)
if index % 10 == 0:
_check_prices.apply_async(
kwargs={
"base_url": booking_url["crawl_info"][0]["crawl_url"],
"cms_id": hid,
"start_time": start_time,
"days": days,
},
countdown=settings.CHECK_PRICE_DELAY_TIME,
)
return rest_result(request, {"status": 200, "data": "ok"})
async def availability(request):
"""
查价任务统一发布接口\n
compare中为需要比价的第三方网站\n
目前支持ctrip 和 booking
"""
request.headers["Accept"] = "application/json"
valid = await validate_request(request)
logger.info(f"valid: {valid}")
if valid["errmsg"]:
return rest_result(request, {"status": 400, "errmsg": valid["errmsg"]})
compare = valid["compare"]
hotels = valid["hotels"]
start_time = valid["start_time"]
days = valid["days"]
only_compare = valid["only_compare"]
"""
获取第三方价格放在scrapy中执行，celery仅发布任务
优先放入celery中执行
"""
if "ctrip" in compare:
await ctrip(hotels, start_time, days)
if "booking" in compare:
await booking(hotels, start_time, days)
if not only_compare:
for uid, _hotels in hotels.items():
for hotel in uid.split(";"):
quoter_id, hotel_id = hotel.split("::")
if quoter_id == "cms":
calendar_one.delay(hotel_id,
start_time=start_time,
end_time=None,
days=days)
else:
get_provider_prices.delay(
hotel_id=hotel_id,
quoter_id=quoter_id,
start_time=start_time,
hotels=None,
days=days,
)
return rest_result(request, {"status": 200, "data": list(hotels.keys())})
async def booking_provider(request):
valid = await validate_request(request)
provider_compair_booking.delay(valid["start_time"], valid["days"],
valid["hotels"])
return rest_result(
request,
{
"status": 200,
"data": valid.get("errmsg") or valid["hotels"]
},
)
async def _weego(request):
valid = await validate_request(request)
logger.info(f"weego params : {valid}")
for hotel_id in valid.get("hotels", []):
calendar_one.delay(hotel_id, valid["start_time"], valid["end_time"])
return rest_result(
request,
{
"status": 200,
"data": valid.get("errmsg") or valid["hotels"]
},
)
async def _provider(request):
"""
hotel from statics providers data
"""
body = request.json
if not body:
return rest_result(request, {
"status": 400,
"errmsg": "request without params!"
})
start_time = body.get("start_time")
end_time = body.get("end_time")
days = body.get("days")
hotel_ids = body.get("hotel_ids", [])
quoter = body.get("quoter")
quoters = body.get("quoters", []) or body.get("hotels", [])
if quoter in settings.SUPPLIER_NAME_2_ID:
quoter = settings.SUPPLIER_NAME_2_ID[quoter]
if days and not end_time:
end_time = (datetime.strptime(start_time, "%Y-%m-%d") +
timedelta(days=days)).strftime("%Y-%m-%d")
for hotel_id in hotel_ids:
get_provider_prices.delay(
hotel_id=hotel_id,
quoter_id=quoter,
start_time=start_time,
end_time=end_time,
)
for hotel in quoters:
hotel_id = hotel["hotel_id"]
quoter_id = (hotel.get("quoter")
or settings.SUPPLIER_NAME_2_ID[hotel.get("provider")])
get_provider_prices.delay(
hotel_id=hotel_id,
quoter_id=quoter_id,
start_time=start_time,
end_time=end_time,
)
return rest_result(request, {"status": 200, "data": "ok"})
async def skyscanner(request):
valid = await validate_request(request)
if valid["errmsg"] and not request.json.get("provider"):
return rest_result(request, {"status": 400, "errmsg": valid["errmsg"]})
start_time = valid["start_time"]
days = valid["days"]
for hotel in valid["hotels"]:
get_skyscanner.delay(start_time, days, hotel_id=hotel)
db = databases("scripture")
for hotel_id, hotel_name in request.json.get("provider", {}).items():
sid = await db["statics.hotels.skyscanner"].find_one(
{"name": {
"$regex": hotel_name.lower(),
"$options": "i"
}},
{"sid": "1"},
)
if sid:
get_skyscanner.delay(start_time,
days,
sid=sid["sid"],
hotel_id=hotel_id)
return rest_result(request, {"status": 200, "data": "ok"})
async def ctrip_provider(request):
valid = await validate_request(request)
if valid["provider"]:
hotels = []
for _hotel in valid["hotels"]:
if isinstance(_hotel, str):
_hotel = {"hotel_id": _hotel, "provider": valid["provider"]}
elif isinstance(_hotel, dict):
_hotel["provider"] = valid["provider"]
hotels.append(_hotel)
valid["hotels"] = hotels
provider_compair_ctrip.delay(valid["start_time"], valid["days"],
valid["hotels"])
return rest_result(
request,
{
"status": 200,
"data": valid.get("errmsg") or list(valid["hotels"])
},
)
async def _crawl_upload(request):
ori_hotel_info = {}
opt = request.form.get('opt')
if not opt:
logger.warning(f'{request}无opt,非法请求')
return rest_result(request, {"status": 400, "errmsg": "非法请求"})
match_prob = request.form.get('prob')
if opt == 'form':
ori_hotel_name = request.form.get('name')
if not ori_hotel_name:
return rest_result(request, {"status": 400, "errmsg": "酒店名为必填项"})
ori_hotel_provider = request.form.get('provider')
if ori_hotel_provider and ori_hotel_provider not in settings.SUPPLIER_NAME_2_ID:
return rest_result(request, {"status": 400, "errmsg": "供应商有误"})
ori_hotel_code = request.form.get('code')
ori_hotel_address = request.form.get('address')
ori_hotel_country = request.form.get('country')
ori_hotel_info = {
"ori_hotel": [{
"ori_hotel_name": ori_hotel_name,
"ori_hotel_provider": ori_hotel_provider,
"ori_hotel_code": ori_hotel_code,
"ori_hotel_address": ori_hotel_address,
"ori_hotel_country": ori_hotel_country,
}]
}
logger.debug(f'ori_hotel_info:{ori_hotel_info}')
elif opt == 'excel':
excel = request.files.get('excel')
if not excel:
logger.info(f'{request}包含文件为空')
return rest_result(request, {"status": 400, "errmsg": "文件为空"})
excel_info = f'文件名: {excel.name},mime: {excel.type}'
logger.info(f'上传了一份文件:文件信息:{excel_info}')
if XLSXMIME != excel.type:
logger.info(f'{request}{excel_info} 上传了错误的文件类型')
return rest_result(request, {"status": 400, "errmsg": "错误的文件类型"})
book = xlrd.open_workbook(file_contents=excel.body)
_page = book.sheet_by_index(0)
ori_hotel_info["ori_hotel"] = []
for row_index in range(_page.nrows):
one_hotel_info = {}
for col_index, cel_value in enumerate(_page.row_values(row_index)):
if col_index == 0 and not cel_value:
break
if col_index == 1 and cel_value and cel_value not in settings.SUPPLIER_NAME_2_ID:
one_hotel_info = {}
break
if cel_value and col_index < 5:
one_hotel_info[hotel_info_key[col_index]] = cel_value
if one_hotel_info:
ori_hotel_info["ori_hotel"].append(one_hotel_info)
logger.debug(f'{request}上传了一份excel文件,response_dict:{ori_hotel_info}')
ori_hotel_info['match_prob'] = match_prob
logger.info(f'{request}开始了一次酒店匹配,原始酒店信息为:{ori_hotel_info}')
try:
result = requests.post(f'{settings.EXAMINER_API}/api/v1/match/',
json=ori_hotel_info)
if result.status_code == 200:
logger.debug(f'result.json(){result.json()}')
tr = ''
if not result.json():
return rest_result(request, {"status": 200, "结果": "匹配结束无相似酒店"})
for hotel in result.json():
_similar_hotel = hotel["similar_hotel"]
tr += f'<tr>' \
f'<td>原酒店名称:{hotel["ori_hotel_name"]}</td>' \
f'<td>' \
f'相似酒店名称:{_similar_hotel["hotel_name"]},' \
f'供应商:{_similar_hotel["hotel_provider"]},' \
f'酒店ID:{_similar_hotel["hotel_code"]},' \
f'地址:{_similar_hotel["hotel_addr"]},' \
f'国家:{_similar_hotel["hotel_country"]},' \
f'相似率:{_similar_hotel["hotel_prob"]}' \
f'</td>' \
f'</tr>'
table = f'<table border="1">{tr}</table>'
return html(table)
else:
logger.warning(f'酒店信息:{ori_hotel_info}对应查询异常')
return rest_result(request, {"status": 500, "errmsg": "查询异常"})
except Exception as exc:
logger.warning(f"examiner api 异常", exc_info=exc)
return rest_result(request, {
"status": 500,
"errmsg": "examiner lose effectiveness"
})
async def preparation_check(request):
body = request.json
check_preparation.delay(body["hotels"])
return rest_result(request, {"status": 200, "data": "ok"})
async def availability(request):
"""
查价任务统一查询数据接口\n
compare中为需要比价的第三方网站\n
目前支持ctrip 和 booking
return {
'status': 200,
'data': {
quoter_id1::hotel_id1;quoter_id2::hotel_id2:
[
{'checkin': {
'price': '', 'booking': '', 'ctrip', '', 'msg': '',
'type': -1/0/1
}
}
],
cms::cms_id
[
{'checkin': {
'price': '', 'booking': '', 'ctrip', '', 'msg': '',
'type': -1/0/1
}
}
],
}
}
"""
request.headers["Accept"] = "application/json"
valid = await validate_request(request)
if valid["errmsg"]:
return rest_result(request, {"status": 400, "errmsg": valid["errmsg"]})
results = []
if valid['is_ori_price']:
price_type = 3
elif valid['is_without_tax']:
price_type = 2
else:
price_type = 1
to_text = valid['to_text']
only_supplier = valid["only_supplier"]
start_time = datetime.strptime(valid["start_time"], "%Y-%m-%d")
end_time = datetime.strptime(valid["end_time"], "%Y-%m-%d")
logger.info(f"valid: {valid}")
_hotel_prices = {
datetime.strftime(start_time + timedelta(days=i), "%Y-%m-%d"): {
"type": -1,
"price": "",
"room": "",
"supplier": "",
"each_supplier": {},
"updated_at": "",
}
for i in range((end_time - start_time).days + 1)
}
for uid in valid["hotels"]:
hotel_prices = copy.deepcopy(_hotel_prices)
hotel_prices = await get_weego_prices(uid, hotel_prices, price_type)
for compare in valid["compare"]:
hotel_prices = await add_compare_prices(
compare, uid, hotel_prices, start_time, end_time
)
results.append({uid: hotel_prices})
results = add_msg(results, to_text, only_supplier)
return rest_result(request, {"status": 200, "data": results})
