python_apps/crwlers/exceptionCrawler.py
2023-11-03 14:49:12 +09:00

57 lines
1.7 KiB
Python

import requests, bs4, urllib, sys, re, math, logging, os
from datetime import datetime
import subprocess
ScriptLocation = os.path.dirname(os.path.abspath(__file__))
sys.path.append(ScriptLocation + "/lib")
import naver
def nonSerpCrawl(naverTerm):
na = naver.NaverGet(naverTerm).getSummaryResult()
res = ""
for key,value in na.items():
res = res + str(value) + "\t"
return res
imarketTerm = str(sys.argv[1])
naverTerm = str(sys.argv[2])
mode = str(sys.argv[3])
res = imarketTerm + "\t" + mode + "\t\t\t\t\t\t\t\t\t\t"
srclocation = "https://www.imarket.co.kr/display/malls.do"
params = {}
if mode == "기획전":
srclocation = "https://www.imarket.co.kr/plant/PlantMaster.do"
params['_method'] = 'Initial'
params['plantId'] = imarketTerm
params['utm_campaign'] = 'theme'
if mode =="카테고리":
params['_method'] = '1Depth'
params['sc.dispNo'] = imarketTerm
params['sc.shopNo'] = '0000100000'
params['sc.viewType'] = 'list'
params['sc.row'] = 20
params['utm_campaign'] = 'category'
if mode =="메인" :
params['_method'] = 'welcome'
params['utm_campaign'] = 'brand'
if mode =="의료몰":
srclocation = "https://www.imarket.co.kr/medical"
params['utm_campaign'] = 'medical'
if mode =="호텔":
srclocation = "https://www.imarket.co.kr/hotel"
params['utm_campaign'] = 'hotel'
if mode =="디지털":
srclocation = "https://www.imarket.co.kr/digital"
params['utm_campaign'] = 'digital'
params['BIZ_CD'] = '1010187'
params['utm_source'] = 'naverPowerlink'
params['utm_medium'] = 'prdcpc'
params['utm_content'] = 'undefined'
params['utm_term'] = naverTerm
resp = requests.get(srclocation, params=params)
res = res + resp.url + "\t" + nonSerpCrawl(naverTerm)
print(res)