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

114 lines
3.7 KiB
Python

import requests, bs4, urllib, sys, re, math, logging
from datetime import datetime
startTime = datetime.now()
def makeQuery(term) :
query = urllib.parse.quote_plus(term)
return query
def makeURL(num, query) :
url = "https://ad.search.naver.com/search.naver?where=ad&sm=svc_nrs&query=" + query + "&referenceId=&pagingIndex=" + str(num)
return url
def getHTML(url) :
resp = requests.get(url)
resp.raise_for_status()
resp.encoding='UTF-8'
html = resp.text
bs = bs4.BeautifulSoup(html, 'html.parser')
return bs
def resultCount(allHTML) :
res_count = allHTML.select('div.search_result div.inner span.num_result')[0].getText().strip()
res_count = re.sub(r"[0-9]+\-[0-9]+\ \/\s","",res_count)
res_count = re.sub(r"","",res_count)
return int(res_count)
def paging(resultCount) :
paging = math.ceil(int(resultCount)/25)
return paging
term = str(sys.argv[1])
url = makeURL(1,makeQuery(str(sys.argv[1])))
content = getHTML(url)
resultCount = resultCount(content)
paging = paging(resultCount)
if resultCount == 0 : print(term + "\t" + "NoResult")
else :
_list = content.select('div.ad_section ol.lst_type li.lst')
for i in range(len(_list)) :
title = _list[i].select('div.inner a.lnk_tit')[0].getText().strip()
subTitle = _list[i].select('div.inner a.sub_tit')
if len(subTitle) == 0 : SubT = ""
else : SubT = subTitle[0].getText().strip()
ad_thumb = _list[i].select('div.ad_thumb')
if len(ad_thumb) == 0 : thumb = ""
else : thumb = "Y"
url = _list[i].select('div.inner div.url_area a.url')[0].getText().strip()
menu = _list[i].select('div.inner ul.lst_link li.item a')
menulink = ''
if len(menu) != 0 :
for j in range(len(menu)) :
menulink = menulink + menu[j].getText().strip()
if j != len(menu) - 1 :
menulink = menulink + "||"
ev = _list[i].select('div.inner p.promotion')
if len(ev) == 0 :
event = ""
description = _list[i].select('div.inner p.ad_dsc_inner')[0].getText().strip()
else :
event = ev[0].getText().strip()
event = re.sub(r"\n|\t|\s\s"," ",event)
event = re.sub(r"\s\s+","|",event)
description = _list[i].select('div.inner p.ad_dsc_inner')[0].getText().strip()
description = re.sub(r"\n|\t|\s\s"," ",description)
description = re.sub(r"\s\s+"," ",description)
itemlst = ''
lst_price = _list[i].select('div.inner ul.lst_price li.item a')
for k in range(len(lst_price)) :
itemlst = itemlst + lst_price[k].select('div.txt span')[0].getText().strip() + " (" + lst_price[k].select('span.price')[0].getText().strip() + ")"
if k != (len(lst_price)-1) : itemlst = itemlst + "||"
adPeriod = _list[i].select('div.inner div.period_area')[0].getText().strip()
adPeriod = re.sub(r"광고집행기간|\n","",adPeriod)
iconStats = ''
iconArea = _list[i].select('div.inner div.url_area span.ico_area span.ico')
if len(iconArea) != 0 :
for l in range(len(iconArea)) :
if len(iconArea[l].select('span.ico_npay')) != 0 :
iconStats = iconStats + 'nPay'
if len(iconArea[l].select('span.ico_nlogin')) != 0 :
iconStats = iconStats + 'nLogin'
if len(iconArea[l].select('span.ico_talktalk')) != 0 :
iconStats = iconStats + 'nTalkTalk'
if len(iconArea[l].select('span.ico_nreserve')) != 0 :
iconStats = iconStats + 'nReserve'
if len(iconArea) > 1 :
if l < (len(iconArea)-1) :
iconStats = iconStats + ","
else : iconStats = iconStats + ''
print(
term + "\t" +
str(resultCount) + "\t" +
str(i+1) + "\t" +
title + "\t" +
SubT + "\t" +
thumb + "\t" +
url + "\t" +
iconStats + "\t" +
menulink + "\t" +
event + "\t" +
description + "\t" +
itemlst + "\t" +
adPeriod + "\t"
)
consumtime = datetime.now() - startTime
logging.warning(term + "\t" + str(consumtime))