스크래퍼 업데이트: 단축 url 처리, 시설 정보 추출
parent
03a80afc71
commit
1516f2807c
|
|
@ -1,13 +1,14 @@
|
|||
import asyncio
|
||||
import json
|
||||
import re
|
||||
|
||||
import aiohttp
|
||||
import json
|
||||
import asyncio
|
||||
import bs4
|
||||
|
||||
PLACE_PATTERN = r"/place/(\d+)"
|
||||
GRAPHQL_URL = "https://pcmap-api.place.naver.com/graphql"
|
||||
NAVER_COOKIES = "NAC=mQ7mBownbQf4A; NNB=TQPII6AKDBFGQ; PLACE_LANGUAGE=ko; NACT=1; nid_inf=1431570813; NID_AUT=k2T7FraXOdIMRCHzEZIFtHQup+I7b87M5fd7+p65AXZTdGB/gelRmW8s/Q4oDxm8; tooltipDisplayed=true; SRT30=1762660151; NID_SES=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; _naver_usersession_=Kkgzim/64JicPJzgkIIvqQ==; page_uid=jesTPsqVWUZssE4qJeossssssD0-011300; SRT5=1762662010; BUC=z5Fu3sAYtFwpbRDrrDFYdn4AgK5hNkOqX-DdaLU7VJM="
|
||||
NAVER_COOKIES="NAC=mQ7mBownbQf4A; NNB=TQPII6AKDBFGQ; PLACE_LANGUAGE=ko; NACT=1; nid_inf=1431570813; NID_AUT=k2T7FraXOdIMRCHzEZIFtHQup+I7b87M5fd7+p65AXZTdGB/gelRmW8s/Q4oDxm8; tooltipDisplayed=true; SRT30=1762660151; NID_SES=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; _naver_usersession_=Kkgzim/64JicPJzgkIIvqQ==; page_uid=jesTPsqVWUZssE4qJeossssssD0-011300; SRT5=1762662010; BUC=z5Fu3sAYtFwpbRDrrDFYdn4AgK5hNkOqX-DdaLU7VJM="
|
||||
|
||||
OVERVIEW_QUERY = """
|
||||
OVERVIEW_QUERY = '''
|
||||
query getAccommodation($id: String!, $deviceType: String) {
|
||||
business: placeDetail(input: {id: $id, isNx: true, deviceType: $deviceType}) {
|
||||
base {
|
||||
|
|
@ -25,81 +26,94 @@ query getAccommodation($id: String!, $deviceType: String) {
|
|||
images { images { origin url } }
|
||||
cpImages(source: [ugcImage]) { images { origin url } }
|
||||
}
|
||||
}"""
|
||||
}'''
|
||||
|
||||
REQUEST_HEADERS = {
|
||||
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/129.0.0.0 Safari/537.36",
|
||||
"Referer": "https://map.naver.com/",
|
||||
"Origin": "https://map.naver.com",
|
||||
"Content-Type": "application/json",
|
||||
"Cookie": NAVER_COOKIES,
|
||||
"Cookie": NAVER_COOKIES
|
||||
}
|
||||
|
||||
|
||||
class GraphQLException(Exception):
|
||||
pass
|
||||
|
||||
|
||||
class nvMapScraper:
|
||||
url: str = None
|
||||
scrap_type: str = None
|
||||
rawdata: dict = None
|
||||
image_link_list: list[str] = None
|
||||
base_info: dict = None
|
||||
class nvMapScraper():
|
||||
url : str = None
|
||||
scrap_type : str = None
|
||||
rawdata : dict = None
|
||||
image_link_list : list[str] = None
|
||||
base_info : dict = None
|
||||
|
||||
|
||||
def __init__(self, url):
|
||||
self.url = url
|
||||
|
||||
def parse_url(self):
|
||||
place_pattern = r"/place/(\d+)"
|
||||
async def parse_url(self):
|
||||
if 'place' not in self.url:
|
||||
if 'naver.me' in self.url:
|
||||
async with aiohttp.ClientSession() as session:
|
||||
async with session.get(self.url) as response:
|
||||
self.url = str(response.url)
|
||||
else:
|
||||
raise GraphQLException("this shorten url not have place id")
|
||||
try:
|
||||
place_id = re.search(place_pattern, self.url)[1]
|
||||
except:
|
||||
raise GraphQLException()
|
||||
|
||||
place_id = re.search(PLACE_PATTERN, self.url)[1]
|
||||
except Exception as E:
|
||||
raise GraphQLException("Cannot find place id")
|
||||
|
||||
return place_id
|
||||
|
||||
|
||||
async def scrap(self):
|
||||
try:
|
||||
place_id = self.parse_url()
|
||||
place_id = await self.parse_url()
|
||||
data = await self.call_get_accomodation(place_id)
|
||||
self.rawdata = data
|
||||
self.image_link_list = [
|
||||
nv_image["origin"]
|
||||
for nv_image in data["data"]["business"]["images"]["images"]
|
||||
]
|
||||
self.base_info = data["data"]["business"]["base"]
|
||||
fac_data = await self.get_facility_string(place_id)
|
||||
self.rawdata['facilities'] = fac_data
|
||||
self.image_link_list = [nv_image['origin'] for nv_image in data['data']['business']['images']['images']]
|
||||
self.base_info = data['data']['business']['base']
|
||||
self.facility_info = fac_data
|
||||
self.scrap_type = "GraphQL"
|
||||
|
||||
except GraphQLException:
|
||||
except GraphQLException as G:
|
||||
print (G)
|
||||
print("fallback")
|
||||
self.scrap_type = "Playwright"
|
||||
pass # 나중에 pw 이용한 crawling으로 fallback 추가
|
||||
pass # 나중에 pw 이용한 crawling으로 fallback 추가
|
||||
|
||||
return
|
||||
|
||||
async def call_get_accomodation(self, place_id):
|
||||
payload = {
|
||||
"operationName": "getAccommodation",
|
||||
"variables": {"id": place_id, "deviceType": "pc"},
|
||||
"operationName" : "getAccommodation",
|
||||
"variables": { "id": place_id, "deviceType": "pc" },
|
||||
"query": OVERVIEW_QUERY,
|
||||
}
|
||||
json_payload = json.dumps(payload)
|
||||
|
||||
|
||||
async with aiohttp.ClientSession() as session:
|
||||
async with session.post(
|
||||
GRAPHQL_URL, data=json_payload, headers=REQUEST_HEADERS
|
||||
) as response:
|
||||
response.encoding = "utf-8"
|
||||
if response.status == 200: # 요청 성공
|
||||
return await response.json() # await 주의
|
||||
else: # 요청 실패
|
||||
print("실패 상태 코드:", response.status)
|
||||
async with session.post(GRAPHQL_URL, data=json_payload, headers=REQUEST_HEADERS) as response:
|
||||
response.encoding = 'utf-8'
|
||||
if response.status == 200: # 요청 성공
|
||||
return await response.json() # await 주의
|
||||
else: # 요청 실패
|
||||
print('실패 상태 코드:', response.status)
|
||||
print(response.text)
|
||||
raise Exception()
|
||||
|
||||
|
||||
url = "https://map.naver.com/p/search/%EC%8A%A4%ED%85%8C%EC%9D%B4%EB%A8%B8%EB%AD%84/place/1133638931?c=14.70,0,0,0,dh&placePath=/photo?businessCategory=pension&fromPanelNum=2&locale=ko&searchText=%EC%8A%A4%ED%85%8C%EC%9D%B4%EB%A8%B8%EB%AD%84&svcName=map_pcv5×tamp=202512191123&fromPanelNum=2&locale=ko&searchText=%EC%8A%A4%ED%85%8C%EC%9D%B4%EB%A8%B8%EB%AD%84&svcName=map_pcv5×tamp=202512191007&from=map&entry=bmp&filterType=%EC%97%85%EC%B2%B4&businessCategory=pension"
|
||||
|
||||
async def get_facility_string(self, place_id):
|
||||
url = f"https://pcmap.place.naver.com/accommodation/{place_id}/home"
|
||||
async with aiohttp.ClientSession() as session:
|
||||
async with session.get(url, headers=REQUEST_HEADERS) as response:
|
||||
soup = bs4.BeautifulSoup(await response.read(), 'html.parser')
|
||||
c_elem = soup.find('span', 'place_blind', string='편의')
|
||||
facilities = c_elem.parent.parent.find('div').string
|
||||
return facilities
|
||||
|
||||
url = "https://naver.me/IgJGCCic"
|
||||
scraper = nvMapScraper(url)
|
||||
asyncio.run(scraper.scrap())
|
||||
print(scraper.image_link_list)
|
||||
print(len(scraper.image_link_list))
|
||||
print(scraper.base_info)
|
||||
print(len(scraper.image_link_list))
|
||||
Loading…
Reference in New Issue