def pusanUrl(name, url):
subMenuUrls = clsSubMenuUrls()
subMenuUrls.name = name
subMenuUrls.url = url
print subMenuUrls.name.decode('utf-8') + ' : ' + subMenuUrls.url
#print >> exceptFile, subMenuUrls + ' : ' + subMenuUrls.url
detailProductPusanHtml = savefilegethtml.getHtml(subMenuUrls.url, 'class="container', '<!-- end .ot_tab_style1 -->', 'onlinetourSubPagePusan.txt')
for subMenu in detailProductPusanHtml:
#if subMenu.find('<li class="">') > -1 and subMenu.find('전체') < 0:
if subMenu.find('<li') > -1 and subMenu.find('<a') > -1 and (subMenu.find('전체') < 0 or subMenuUrls.url.find('D50') > -1 or subMenuUrls.url.find('D60') > -1 or subMenuUrls.url.find('D70') > -1):
detailRegionUrls = clsDetailRegionUrls()
detailRegionUrls.name = tourUtil.getRemovedHtmlTag(subMenu).strip()
detailRegionUrls.url = mainUrl + tourUtil.getTagAttr(subMenu, 'a', 'href')
subMenuUrls.detailRegionList.append(detailRegionUrls)
print detailRegionUrls.name.decode('utf-8') + ' : ' + detailRegionUrls.url
#print >> exceptFile, detailRegionUrls.name + ' : ' + detailRegionUrls.url
return subMenuUrls
try:
for menu in backpackMenuList:
#tit_position2 부산출발 검색조건..
print menu.url
productListHtml = savefilegethtml.getHtml(menu.url,
'<div id="sub_box2">',
'function btn(ckbtn){',
'productListHtml.txt')
for each_line in productListHtml:
if each_line.find('<h1 class="bic_h">') > -1:
productName = each_line.split('bic_h">')[2].split('<')[0]
#productNameSplit = productName.split(' ')
#period = productNameSplit[len(productNameSplit)-1].replace('일', '')
productNameSplit = re.findall(
'[\^0-9]+', tourUtil.getRemovedHtmlTag(each_line))
period = productNameSplit[len(productNameSplit) - 1]
#if each_line.find('<span class="goods_text">') > -1: # 여행 설명이 잘 적혀 있긴 하지만.. 내일투어에서는 국가 정보가 있는 걸 보조로 가져가야 할듯..
#tourRoute = each_line.split('px;">')[1].split('<')[0]
if each_line.find('<span class="travel_box">') > -1:
tourRoute = tourUtil.getRemovedHtmlTag(each_line)
if each_line.find("sview('") > -1:
productCode = each_line.split("sview('")[1].split("'")[0]
code2 = each_line.split("sview('")[1].split("'")[2]
detailUrl = 'http://www.naeiltour.co.kr/backpack/program_include_list.asp?good_cd=' + productCode + '&sel_ym=' + targetYear + targetMonth
print >> exceptFile, 'DetailUrl : ', detailUrl
listUrl = 'http://www.naeiltour.co.kr/backpack/program_include_list.asp?good_cd='
productDetailUrl = 'http://www.naeiltour.co.kr/backpack/show.asp?good_cd='
mainpageHtml = savefilegethtml.getHtml('http://www.onlinetour.co.kr/web/home', '<li id="n_pack">', '<!--}} ot_navi-->', 'onlinetourMainPage.txt')
# URL 쑤셔넣는 부분...
mainMenuList = list()
mainMenuUrls = clsMenuUrls()
subMenuUrls = clsSubMenuUrls()
detailRegionUrls = clsDetailRegionUrls()
chkFree = False
chkDomestic = False
for menuList in mainpageHtml:
try:
#print menuList
if menuList.find('<a href=') > -1 and menuList.find('<li>') < 0:
mainMenuUrls = clsMenuUrls()
mainMenuUrls.name = tourUtil.getRemovedHtmlTag(menuList).strip()
mainMenuUrls.url = tourUtil.getTagAttr(menuList, 'a', 'href')
mainMenuUrls.dmst_div = 'A'
if menuList.find('부산출발') > -1:
mainMenuUrls.departCity = 'PUS'
# 부산의 경우.. 세부 지역 URL이 바로 노출되어 있지 않아.. 강제로 쭈셔 넣어 준다..
mainMenuUrls.subMenuList.append(pusanUrl('동남아', 'http://www.onlinetour.co.kr/web/tour?region_cd=D10'))
mainMenuUrls.subMenuList.append(pusanUrl('일본', 'http://www.onlinetour.co.kr/web/tour?region_cd=D20'))
mainMenuUrls.subMenuList.append(pusanUrl('중국', 'http://www.onlinetour.co.kr/web/tour?region_cd=D30'))
mainMenuUrls.subMenuList.append(pusanUrl('괌/사이판', 'http://www.onlinetour.co.kr/web/tour?region_cd=D40'))
mainMenuUrls.subMenuList.append(pusanUrl('남태평양', 'http://www.onlinetour.co.kr/web/tour?region_cd=D50'))
mainMenuUrls.subMenuList.append(pusanUrl('유럽/특수', 'http://www.onlinetour.co.kr/web/tour?region_cd=D60'))
mainMenuUrls.subMenuList.append(pusanUrl('미주/특수', 'http://www.onlinetour.co.kr/web/tour?region_cd=D70'))
else:
mainMenuUrls.departCity = 'ICN'
#query = savefilegethtml.getMasterMergeQuery('vgtour', mastercode, '', '', productGroupCls.name, productName, tourType, region, productComment, '') # A : 해외(Abroad)
#print query
productCls = clsProduct()
#productListHtml = open('productListHtml.txt')
departConfirm = False
for product in productListHtml:
#print 'product : ' + product
if product.find('pro_date') > -1:
productCls = clsProduct()
departConfirm = False
#<td class="pro_date">07/07 (월) 16:15<br/><span>07/09 (수) 21:05</span></td>
#<td class="pro_date">07/07 (월) <br/><span></span></td>
#<td class="pro_date">07/28 (월) 09:10<br/><span>08/17 (<span style="color:red;margin-bottom:0;">일</span>) 05:50</span></td>
daySplit = tourUtil.getNumArray(tourUtil.getRemovedHtmlTag(product))
productCls.sDay = ''
productCls.sTime = ''
productCls.aDay = ''
productCls.aTime = ''
if len(daySplit) > 1:
productCls.sDay = '2014' + daySplit[0] + daySplit[1]
if len(daySplit) > 3:
productCls.sTime = daySplit[2] + daySplit[3]
if len(daySplit) > 5:
productCls.aDay = '2014' + daySplit[4] + daySplit[5]
if len(daySplit) > 7:
productCls.aTime = daySplit[6] + daySplit[7]
#productCls.sDay = '2014' + product.split('pro_date">')[1].split('(')[0].strip().replace('/', '')
#productCls.sTime = product.split('<br/>')[0].split(')')[1].strip().replace(':', '')
if period == '' and tourkind == 'F':
if product.find('<td class="FRIDAYSPACING" >'
) > -1 and product.find('.gif') > -1:
productCls.airCode = product[product.find('.gif') -
2:product.find('.gif')]
#<td class="FRIDAYSPACING" width="220"><B><FONT COLOR="RED">569,000원</FONT>2박3일 도미인 아사쿠사 호텔(12박 13일)</td>
#<td class="FRIDAYSPACING" width="220"><B><FONT COLOR="RED">369,000원</FONT> 2박3일<BR>신주쿠 워싱톤 호텔(더블룸)</td>
#<td class="FRIDAYSPACING" width="220"><B><FONT COLOR="RED">569,000원</FONT>(2박3일) 도미인 아사쿠사 호텔</td>
if product.find('idth="220">') > -1:
print product
print type(product)
splitText = product.split('박'.decode('utf-8'))
tmpText = re.findall('[\^0-9]+',
tourUtil.getRemovedHtmlTag(splitText[0]))
print 'Night : ', tmpText[len(tmpText) - 1]
tmpText = re.findall('[\^0-9]+',
tourUtil.getRemovedHtmlTag(splitText[1]))
print 'day : ', tmpText[0]
#if product.find('(') > -1:
#productCls.night = re.findall(r"\d", product.split('(')[1])[0]
#productCls.period = re.findall(r"\d", product.split('(')[1])[1]
#elif product.find('[') > -1:
#productCls.night = re.findall(r"\d", product.split('[')[1])[0]
#productCls.period = re.findall(r"\d", product.split('[')[1])[1]
#print productCls.toString()
query = savefilegethtml.getDetailMergeQueryTest1(
'naeiltour', 'productcode', productCls.code,
productCls.productname, '20' + productCls.dDay, '', 'ICN',
query = savefilegethtml.getDetailMergeQueryTest1('naeiltour', 'productcode', productCls.code, productCls.productname, '20' + productCls.dDay, '', 'ICN', '', productCls.airCode, productCls.status, productCls.url, productCls.price, '0', '0', '0', '', '')
#print query
#break
if period == '' and tourkind == 'F':
if product.find('<td class="FRIDAYSPACING" >') > -1 and product.find('.gif') > -1:
productCls.airCode = product[product.find('.gif') - 2:product.find('.gif')]
#<td class="FRIDAYSPACING" width="220"><B><FONT COLOR="RED">569,000원</FONT>2박3일 도미인 아사쿠사 호텔(12박 13일)</td>
#<td class="FRIDAYSPACING" width="220"><B><FONT COLOR="RED">369,000원</FONT> 2박3일<BR>신주쿠 워싱톤 호텔(더블룸)</td>
#<td class="FRIDAYSPACING" width="220"><B><FONT COLOR="RED">569,000원</FONT>(2박3일) 도미인 아사쿠사 호텔</td>
if product.find('idth="220">') > -1:
print product
print type(product)
splitText = product.split('박'.decode('utf-8'))
tmpText = re.findall('[\^0-9]+', tourUtil.getRemovedHtmlTag(splitText[0]))
print 'Night : ', tmpText[len(tmpText)-1]
tmpText = re.findall('[\^0-9]+', tourUtil.getRemovedHtmlTag(splitText[1]))
print 'day : ', tmpText[0]
#if product.find('(') > -1:
#productCls.night = re.findall(r"\d", product.split('(')[1])[0]
#productCls.period = re.findall(r"\d", product.split('(')[1])[1]
#elif product.find('[') > -1:
#productCls.night = re.findall(r"\d", product.split('[')[1])[0]
#productCls.period = re.findall(r"\d", product.split('[')[1])[1]
#print productCls.toString()
query = savefilegethtml.getDetailMergeQueryTest1('naeiltour', 'productcode', productCls.code, productCls.productname, '20' + productCls.dDay, '', 'ICN', '', productCls.airCode, productCls.status, productCls.url, productCls.price, '0', '0', '0', '', productCls.night)
#print 'Query : ' + query
#break
def searchProduct(filename, productcode, productName, period, targetUrl, listUrl, productDetailUrl, departCity, tourkind, dmst_div, country='', city='', comment=''):
detailHtml = savefilegethtml.getHtml(targetUrl, '', '', 'naeiltourDetailHtml.txt')
print >> filename, 'TargetUrl : ', targetUrl
departDayList = list()
for detail_each_line in detailHtml:
if detail_each_line.find("fn_goodDetail('") > -1:
departDayList.append(detail_each_line.split("fn_goodDetail('")[1].split("'")[0])
# 출발 가능 날짜에 항공사 찾아오는 부분
try:
con = cx_Oracle.connect("bigtour/[email protected]:1521/ora11g")
codeList = codes.getCityCode(productName, city, comment, country)
cityList = codeList[0]
nationList = codeList[1]
#print nationList
#print cityList
#print nationList
#print cityList
query = savefilegethtml.getMasterMergeQueryTest1('naeiltour', productcode, '', country, city, productName, tourkind, dmst_div, comment, '', nationList, cityList) # A : 해외(Abroad)
#query = savefilegethtml.getMasterMergeQuery('naeiltour', productcode, '', country, city, productName, tourkind, dmst_div, comment, '', nationList, cityList) # A : 해외(Abroad)
#print query
cursor = con.cursor()
cursor.execute(query)
con.commit()
productCls = clsProduct()
for dayInfo in departDayList:
try:
productListUrl = listUrl + productcode + '&sel_day=' + dayInfo
print 'ProductListUrl : ' + productListUrl
productListHtml = savefilegethtml.getHtml(productListUrl, '', '', 'naeiltourproductListHtml.txt')
print >> filename, 'ProductListUrl : ' + productListUrl
for product in productListHtml:
try:
if product.find("fn_price('") > -1:
productCls = clsProduct()
productSplit = product.split('fn_price')[1].split("'")
productCls.productCode = productSplit[1]
productCls.dDay = productSplit[3]
productCls.code = productSplit[5]
if tourkind == 'W' or tourkind == 'G':
productCls.airCode = product[product.find('.gif') - 2:product.find('.gif')]
else:
productCls.airCode = productSplit[7] # 한글 공항코드... but 우리는 영문2자리 공항코드가 필요하다...
productCls.price = productSplit[9].replace(',', '')
#print productSplit[11]
productCls.status = codes.getStatus('naeiltour', productSplit[11]) # 공백 : 예약가능, 03 : 마감임박, 05 : 마감
#if tourkind == 'W':
#productCls.city = productSplit[13]
productCls.url = productDetailUrl + productcode + '&sel_day=' + productCls.dDay
productCls.productname = productName
productCls.dTime = ''
productCls.aDay = ''
productCls.aTime = ''
if period != '' and tourkind == 'F':
if product.find('<td width="134">') > -1:
productCls.period = period
#print productCls.toString()
productCls.airCode = product[product.find('.gif') - 2:product.find('.gif')]
query = savefilegethtml.getDetailMergeQueryTest1('naeiltour', productcode, productCls.code, productCls.productname, '20' + productCls.dDay, '', productCls.period, departCity, '', productCls.airCode, productCls.status, productCls.url, productCls.price, '0', '0', '0', '', '')
#query = savefilegethtml.getDetailMergeQuery('naeiltour', productcode, productCls.code, productCls.productname, '20' + productCls.dDay, '', productCls.period, departCity, '', productCls.airCode, productCls.status, productCls.url, productCls.price, '0', '0', '0', '', '')
#print query
cursor = con.cursor()
cursor.execute(query)
con.commit()
#break
if period == '' and tourkind == 'F':
if product.find('<td class="FRIDAYSPACING" >') > -1 and product.find('.gif') > -1:
productCls.airCode = product[product.find('.gif') - 2:product.find('.gif')]
if product.find('idth="220">') > -1:
"""
# 날짜 가져오는 부분... 종류가 너무 많아서 좀 수정
if product.find('(') > -1:
productCls.night = re.findall(r"\d", product.split('(')[1])[0]
productCls.period = re.findall(r"\d", product.split('(')[1])[1]
elif product.find('[') > -1:
productCls.night = re.findall(r"\d", product.split('[')[1])[0]
productCls.period = re.findall(r"\d", product.split('[')[1])[1]
"""
splitText = product.decode('cp949').split(u'박')
if len(splitText) > 1:
tmpText = re.findall(u'[\^0-9]+', tourUtil.getRemovedHtmlTag(splitText[0]))
productCls.night = tmpText[len(tmpText)-1].encode('utf-8')
tmpText = re.findall(u'[\^0-9]+', tourUtil.getRemovedHtmlTag(splitText[1]))
productCls.period = tmpText[0].encode('utf-8')
else:
productCls.night = '0'
productCls.period = '0'
#############################################################################################
#print productCls.toString()
query = savefilegethtml.getDetailMergeQueryTest1('naeiltour', productcode, productCls.code, productCls.productname, '20' + productCls.dDay, '', productCls.period, departCity, '', productCls.airCode, productCls.status, productCls.url, productCls.price, '0', '0', '0', '', productCls.night)
#print 'Query : ' + query
cursor = con.cursor()
cursor.execute(query)
con.commit()
if period == '' and tourkind == 'W':
if product.find('valign="middle"') > -1:
"""
if product.find('(') > -1:
productCls.night = re.findall(r"\d", product.split('(')[1])[0]
productCls.period = re.findall(r"\d", product.split('(')[1])[1]
elif product.find('[') > -1:
productCls.night = re.findall(r"\d", product.split('[')[1])[0]
productCls.period = re.findall(r"\d", product.split('[')[1])[1]
"""
splitText = product.decode('cp949').split(u'박')
if len(splitText) > 1:
tmpText = re.findall(u'[\^0-9]+', tourUtil.getRemovedHtmlTag(splitText[0]))
productCls.night = tmpText[len(tmpText)-1].encode('utf-8')
tmpText = re.findall(u'[\^0-9]+', tourUtil.getRemovedHtmlTag(splitText[1]))
productCls.period = tmpText[0].encode('utf-8')
else:
productCls.night = '0'
productCls.period = '0'
#############################################################################################
#print productCls.toString()
query = savefilegethtml.getDetailMergeQueryTest1('naeiltour', productcode, productCls.code, productCls.productname, '20' + productCls.dDay, '', productCls.period, departCity, '', productCls.airCode, productCls.status, productCls.url, productCls.price, '0', '0', '0', '', productCls.night)
#print 'Query : ' + query
cursor = con.cursor()
cursor.execute(query)
con.commit()
#break
if period == '' and tourkind == 'G':
if product.find('valign="middle"') > -1:
"""
if product.find('(') > -1:
productCls.night = re.findall(r"\d", product.split('(')[1])[0]
productCls.period = re.findall(r"\d", product.split('(')[1])[1]
elif product.find('[') > -1:
productCls.night = re.findall(r"\d", product.split('[')[1])[0]
productCls.period = re.findall(r"\d", product.split('[')[1])[1]
"""
splitText = product.decode('cp949').split(u'박')
if len(splitText) > 1:
tmpText = re.findall(u'[\^0-9]+', tourUtil.getRemovedHtmlTag(splitText[0]))
productCls.night = tmpText[len(tmpText)-1].encode('utf-8')
tmpText = re.findall(u'[\^0-9]+', tourUtil.getRemovedHtmlTag(splitText[1]))
productCls.period = tmpText[0].encode('utf-8')
else:
productCls.night = '0'
productCls.period = '0'
#############################################################################################
#print productCls.toString()
query = savefilegethtml.getDetailMergeQueryTest1('naeiltour', productcode, productCls.code, productCls.productname, '20' + productCls.dDay, '', productCls.period, departCity, '', productCls.airCode, productCls.status, productCls.url, productCls.price, '0', '0', '0', '', productCls.night)
#print 'Query : ' + query
cursor = con.cursor()
cursor.execute(query)
con.commit()
#break
if period == '' and tourkind == 'D':
if product.find('<td class="FRIDAYSPACING" >') > -1 and product.find('.gif') > -1:
productCls.airCode = product[product.find('.gif') - 2:product.find('.gif')]
if product.find('idth="220">') > -1:
"""
if product.find('(') > -1:
productCls.night = re.findall(r"\d", product.split('[')[1])[0]
productCls.period = re.findall(r"\d", product.split('')[1])[1]
else:
productCls.night = re.findall(r"\d", product.split('COLOR=#FF7A73>')[1])[0]
productCls.period = re.findall(r"\d", product.split('COLOR=#FF7A73>')[1])[1]
"""
splitText = product.decode('cp949').split(u'박')
if len(splitText) > 1:
tmpText = re.findall(u'[\^0-9]+', tourUtil.getRemovedHtmlTag(splitText[0]))
productCls.night = tmpText[len(tmpText)-1].encode('utf-8')
tmpText = re.findall(u'[\^0-9]+', tourUtil.getRemovedHtmlTag(splitText[1]))
productCls.period = tmpText[0].encode('utf-8')
else:
productCls.night = '0'
productCls.period = '0'
#############################################################################################
if product.find('COLOR=BLUE>') > -1:
departCity = 'PUS'
else:
departCity = 'ICN'
#print productCls.toString()
query = savefilegethtml.getDetailMergeQueryTest1('naeiltour', productcode, productCls.code, productCls.productname, '20' + productCls.dDay, '', productCls.period, departCity, '', productCls.airCode, productCls.status, productCls.url, productCls.price, '0', '0', '0', '', productCls.night)
#print 'Query : ' + query
cursor = con.cursor()
cursor.execute(query)
con.commit()
#break
except cx_Oracle.DatabaseError as err1:
print >> filename, err1
pass
except:
print >> filename, "Depth3 Error:", sys.exc_info()[0]
pass
#break
except:
print >> filename, "Depth2 Error:", sys.exc_info()[0]
pass
except:
print >> filename, "Depth1 Error:", sys.exc_info()[0]
pass
finally:
con.close()
productNameHtml = savefilegethtml.htmlToList(
productNameHtml, 'xxx.txt')
for pdName in productNameHtml:
if pdName.find('height="110" alt="') > 0:
productNameList.append(
pdName.split(
'alt="')[1].split('"')[0].replace(
"'",
"").strip().decode('utf-8'))
# description을.. 다른놈으로 가져가야 할듯.. route로..
#if pdName.find('<p class="desc">') > 0:
#productCommentList.append(pdName.split('desc">')[1].split('<')[0].replace("'", "").strip().decode('utf-8'))
if pdName.find('<p class="route">') > 0:
productCommentList.append(
tourUtil.getRemovedHtmlTag(
pdName).strip().replace(
"'", "").decode('utf-8'))
#today = today.replace(month = today.month + 1)
codeIdx = 0
for pcode in codeList:
detailProduct = pcode.split('s')[1]
detailProductUrl = ''
if not (
package.menuCode == 'A03'
or package.menuCode == 'A06'
): # 출발일정 눌렀을때 List가 펼쳐지는 경우랑, 페이지가 이동하는 경우 나눔..
detailProductUrl = ''
#if package.menuCode == 'A01':
if period == '' and tourkind == 'F':
if product.find('<td class="FRIDAYSPACING" >'
) > -1 and product.find('.gif') > -1:
productCls.airCode = product[product.find('.gif') -
2:product.find('.gif')]
#<td class="FRIDAYSPACING" width="220"><B><FONT COLOR="RED">569,000원</FONT>2박3일 도미인 아사쿠사 호텔(12박 13일)</td>
#<td class="FRIDAYSPACING" width="220"><B><FONT COLOR="RED">369,000원</FONT> 2박3일<BR>신주쿠 워싱톤 호텔(더블룸)</td>
#<td class="FRIDAYSPACING" width="220"><B><FONT COLOR="RED">569,000원</FONT>(2박3일) 도미인 아사쿠사 호텔</td>
if product.find('idth="220">') > -1:
print product
print type(product)
splitText = product.split('박'.decode('utf-8'))
tmpText = re.findall('[\^0-9]+',
tourUtil.getRemovedHtmlTag(splitText[0]))
print 'Night : ', tmpText[len(tmpText) - 1]
tmpText = re.findall('[\^0-9]+',
tourUtil.getRemovedHtmlTag(splitText[1]))
print 'day : ', tmpText[0]
#if product.find('(') > -1:
#productCls.night = re.findall(r"\d", product.split('(')[1])[0]
#productCls.period = re.findall(r"\d", product.split('(')[1])[1]
#elif product.find('[') > -1:
#productCls.night = re.findall(r"\d", product.split('[')[1])[0]
#productCls.period = re.findall(r"\d", product.split('[')[1])[1]
#print productCls.toString()
query = savefilegethtml.getDetailMergeQueryTest1(
'naeiltour', 'productcode', productCls.code,
productCls.productname, '20' + productCls.dDay, '', 'ICN',
def insertData(productCls, detailUrl, regionUrl, tourAgency, kind, dmst_div):
print 'Product Url : ', productCls.url
print >> exceptFile, 'Product Url : ', productCls.url
# 2014. 7. 23. 카테고리의 국가는 넣지 않기로 함...
#codeList = codes.getCityCode(productCls.name.decode('utf-8'), detailUrl.name.decode('utf-8'), regionUrl.name.decode('utf-8'))
codeList = codes.getCityCode(productCls.name.decode('utf-8'))
cityList = codeList[0]
nationList = codeList[1]
continentList = codeList[2]
siteList = codeList[3] # 2014. 8. 3. site 추가
if len(cityList) == 0 and len(nationList) == 0 and len(continentList) == 0:
codeList = codes.getCityCode(detailUrl.name.decode('utf-8'))
cityList = codeList[0]
nationList = codeList[1]
continentList = codeList[2]
siteList = codeList[3] # 2014. 8. 3. site 추가
# Master 상품 입력
query = tourQuery.getMasterMergeQuery(tourAgency, productCls.code, productCls.name.decode('utf-8'), menu.kind, dmst_div, '', '')
#print query
cursor = con.cursor()
cursor.execute(query)
con.commit()
# Region Data 삭제
codes.insertRegionData(tourAgency, productCls.code, cityList, nationList, continentList, siteList)
detailProductHtml = savefilegethtml.getHtml(productCls.url, '', '', 'tour2000DetailHtml'+targetMonth+'.txt')
pl10Idx = 0
for detailProduct in detailProductHtml:
try:
if detailProduct.find('<span class="text_pink">') > -1 and detailProduct.find('<a href=') < 0:
detailCls = clsProductDetail()
numArray = tourUtil.getNumArray(detailProduct)
if len(numArray) > 7:
detailCls.dDay = targetYear + numArray[0] + numArray[1]
detailCls.dTime = numArray[2] + numArray[3]
detailCls.aDay = targetYear + numArray[4] + numArray[5]
detailCls.aTime = numArray[6] + numArray[7]
elif len(numArray) == 4:
detailCls.dDay = targetYear + numArray[0] + numArray[1]
detailCls.dTime = ''
detailCls.aDay = targetYear + numArray[2] + numArray[3]
detailCls.aTime = ''
elif detailProduct.find('onError') > -1:
detailCls.airCode = detailProduct[detailProduct.find('.gif') - 4:detailProduct.find('.gif') - 2]
elif detailProduct.find('text_redB') > -1:
numArray = tourUtil.getNumArray(tourUtil.getRemovedHtmlTag(detailProduct))
for num in numArray:
detailCls.price += num
elif detailProduct.find('</a></td>') > -1:
if detailProduct.find('text_pink') > -1:
detailCls.status = codes.getStatus('tour2000', '예약가능')
elif detailProduct.find('text_blau') > -1:
detailCls.status = codes.getStatus('tour2000', '출발가능')
elif detailProduct.find('text_green') > -1:
detailCls.status = codes.getStatus('tour2000', '대기예약')
elif detailProduct.find('text_grayLightSmall') > -1:
detailCls.status = codes.getStatus('tour2000', '예약마감')
detailCls.remainSeat = tourUtil.getRemovedHtmlTag(detailProduct).replace("'", "").strip()
elif detailProduct.find('<p class="pl10">') > -1:
if pl10Idx == 0:
pl10Idx = 1
detailCls.productName = tourUtil.getRemovedHtmlTag(detailProduct).replace("'", "").strip()
detailCls.url = mainUrl + tourUtil.getTagAttr(detailProduct, 'a', 'href')
detailCls.productSeq = detailProduct.split('ev_ym=')[1].split('&')[0] + detailProduct.split('ev_seq=')[1].split('&')[0]
else:
pl10Idx = 0
if detailCls.productName.find('부산출발') > -1:
departCity = 'PUS'
else:
departCity = 'ICN'
query = tourQuery.getDetailMergeQuery(tourAgency, productCls.code, detailCls.productSeq, detailCls.productName.decode('utf-8'), detailCls.dDay+detailCls.dTime, detailCls.aDay+detailCls.aTime, productCls.period, departCity, '', detailCls.airCode, detailCls.status, detailCls.url, detailCls.price, '0', '0', '0', '', productCls.night)
#print >> exceptFile, query
#print query
cursor = con.cursor()
cursor.execute(query)
con.commit()
#break
except:
print >> exceptFile, 'detail parcing Error : ', sys.exc_info()[0]
pass
print >> exceptFile, 'List URL : ', defaultproductListUrl
productList = urllib2.urlopen(defaultproductListUrl).read()
codeList = re.findall(r"goodFocus\w*", productList)
productNameList = list()
productCommentList = list()
productNameHtml = productList[productList.find('travel_top_section'):productList.find('frmGD')]
productNameHtml = savefilegethtml.htmlToList(productNameHtml, 'xxx.txt')
for pdName in productNameHtml:
if pdName.find('height="110" alt="') > 0:
productNameList.append(pdName.split('alt="')[1].split('"')[0].replace("'", "").strip().decode('utf-8'))
# description을.. 다른놈으로 가져가야 할듯.. route로..
#if pdName.find('<p class="desc">') > 0:
#productCommentList.append(pdName.split('desc">')[1].split('<')[0].replace("'", "").strip().decode('utf-8'))
if pdName.find('<p class="route">') > 0:
productCommentList.append(tourUtil.getRemovedHtmlTag(pdName).strip().replace("'", "").decode('utf-8'))
#today = today.replace(month = today.month + 1)
codeIdx = 0
for pcode in codeList:
detailProduct = pcode.split('s')[1]
detailProductUrl = ''
if not (package.menuCode == 'A03' or package.menuCode == 'A06'): # 출발일정 눌렀을때 List가 펼쳐지는 경우랑, 페이지가 이동하는 경우 나눔..
detailProductUrl = ''
#if package.menuCode == 'A01':
detailProductUrl = 'http://www.ybtour.co.kr/Goods/' + urlMap[package.menuCode] + '/inc_evList_ajax.asp?goodCD=' + detailProduct + '&startDT=' + targetYear + targetMonth
#detailProductUrl = 'http://www.ybtour.co.kr/Goods/overseas/inc_evList_ajax.asp?goodCD=150201119&startDT=201408'
exceptFileName = 'tour2000Exception' + scrappingStartTime + '.txt'
exceptFile = open(exceptFileName, 'w')
print >> exceptFile, "Start : %s" % time.ctime()
mainUrl = 'http://www.tour2000.co.kr'
mainHtml = savefilegethtml.getHtml('http://www.tour2000.co.kr/index.asp', '<div class="navi_wholeMenu_box">', '<!-- navi_wholeMenu_wrapper// -->', 'tour2000mainHtml.txt')
startMainUrl = False
menuList = list()
MenuUrlCls = clsMenuUrls()
for each_line in mainHtml:
if each_line.find('text_pinkB14') > -1:
MenuUrlCls = clsMenuUrls()
MenuUrlCls.kind = codes.getTourKind(tourAgency, tourUtil.getRemovedHtmlTag(each_line).strip())
startMainUrl = True
# 해외여행(패키지), 허니문, 골프, 국내(제주) 제외하고는 일단 패스
if MenuUrlCls.kind == 'A' or MenuUrlCls.kind == 'F' or MenuUrlCls.kind == 'H' or MenuUrlCls.kind == 'No':
continue
if startMainUrl and each_line.find('<li>') > -1:
SubMenuCls = clsSubMenu()
SubMenuCls.name = tourUtil.getRemovedHtmlTag(each_line).strip()
SubMenuCls.url = mainUrl + tourUtil.getTagAttr(each_line, 'a', 'href')
MenuUrlCls.subMenuList.append(SubMenuCls)
if startMainUrl and each_line.find('</div>') > -1:
startMainUrl = False
menuList.append(MenuUrlCls)
def insertData(productCls, detailUrl, regionUrl, tourAgency, kind, dmst_div):
print 'Product Url : ', productCls.url
print >> exceptFile, 'Product Url : ', productCls.url
# 2014. 7. 23. 카테고리의 국가는 넣지 않기로 함...
#codeList = codes.getCityCode(productCls.name.decode('utf-8'), detailUrl.name.decode('utf-8'), regionUrl.name.decode('utf-8'))
codeList = codes.getCityCode(productCls.name.decode('utf-8'))
cityList = codeList[0]
nationList = codeList[1]
continentList = codeList[2]
siteList = codeList[3] # 2014. 8. 3. site 추가
if len(cityList) == 0 and len(nationList) == 0 and len(continentList) == 0:
codeList = codes.getCityCode(detailUrl.name.decode('utf-8'))
cityList = codeList[0]
nationList = codeList[1]
continentList = codeList[2]
siteList = codeList[3] # 2014. 8. 3. site 추가
# Master 상품 입력
query = tourQuery.getMasterMergeQuery(tourAgency, productCls.code,
productCls.name.decode('utf-8'),
menu.kind, dmst_div, '', '')
#print query
cursor = con.cursor()
cursor.execute(query)
con.commit()
# Region Data 삭제
codes.insertRegionData(tourAgency, productCls.code, cityList, nationList,
continentList, siteList)
detailProductHtml = savefilegethtml.getHtml(
productCls.url, '', '', 'tour2000DetailHtml' + targetMonth + '.txt')
pl10Idx = 0
for detailProduct in detailProductHtml:
try:
if detailProduct.find(
'<span class="text_pink">') > -1 and detailProduct.find(
'<a href=') < 0:
detailCls = clsProductDetail()
numArray = tourUtil.getNumArray(detailProduct)
if len(numArray) > 7:
detailCls.dDay = targetYear + numArray[0] + numArray[1]
detailCls.dTime = numArray[2] + numArray[3]
detailCls.aDay = targetYear + numArray[4] + numArray[5]
detailCls.aTime = numArray[6] + numArray[7]
elif len(numArray) == 4:
detailCls.dDay = targetYear + numArray[0] + numArray[1]
detailCls.dTime = ''
detailCls.aDay = targetYear + numArray[2] + numArray[3]
detailCls.aTime = ''
elif detailProduct.find('onError') > -1:
detailCls.airCode = detailProduct[detailProduct.find('.gif') -
4:detailProduct.find('.gif'
) - 2]
elif detailProduct.find('text_redB') > -1:
numArray = tourUtil.getNumArray(
tourUtil.getRemovedHtmlTag(detailProduct))
for num in numArray:
detailCls.price += num
elif detailProduct.find('</a></td>') > -1:
if detailProduct.find('text_pink') > -1:
detailCls.status = codes.getStatus('tour2000', '예약가능')
elif detailProduct.find('text_blau') > -1:
detailCls.status = codes.getStatus('tour2000', '출발가능')
elif detailProduct.find('text_green') > -1:
detailCls.status = codes.getStatus('tour2000', '대기예약')
elif detailProduct.find('text_grayLightSmall') > -1:
detailCls.status = codes.getStatus('tour2000', '예약마감')
detailCls.remainSeat = tourUtil.getRemovedHtmlTag(
detailProduct).replace("'", "").strip()
elif detailProduct.find('<p class="pl10">') > -1:
if pl10Idx == 0:
pl10Idx = 1
detailCls.productName = tourUtil.getRemovedHtmlTag(
detailProduct).replace("'", "").strip()
detailCls.url = mainUrl + tourUtil.getTagAttr(
detailProduct, 'a', 'href')
detailCls.productSeq = detailProduct.split(
'ev_ym=')[1].split('&')[0] + detailProduct.split(
'ev_seq=')[1].split('&')[0]
else:
pl10Idx = 0
if detailCls.productName.find('부산출발') > -1:
departCity = 'PUS'
else:
departCity = 'ICN'
query = tourQuery.getDetailMergeQuery(
tourAgency, productCls.code, detailCls.productSeq,
detailCls.productName.decode('utf-8'),
detailCls.dDay + detailCls.dTime,
detailCls.aDay + detailCls.aTime, productCls.period,
departCity, '', detailCls.airCode, detailCls.status,
detailCls.url, detailCls.price, '0', '0', '0', '',
productCls.night)
#print >> exceptFile, query
#print query
cursor = con.cursor()
cursor.execute(query)
con.commit()
#break
except:
print >> exceptFile, 'detail parcing Error : ', sys.exc_info()[0]
pass
mainUrl = 'http://www.tour2000.co.kr'
mainHtml = savefilegethtml.getHtml('http://www.tour2000.co.kr/index.asp',
'<div class="navi_wholeMenu_box">',
'<!-- navi_wholeMenu_wrapper// -->',
'tour2000mainHtml.txt')
startMainUrl = False
menuList = list()
MenuUrlCls = clsMenuUrls()
for each_line in mainHtml:
if each_line.find('text_pinkB14') > -1:
MenuUrlCls = clsMenuUrls()
MenuUrlCls.kind = codes.getTourKind(
tourAgency,
tourUtil.getRemovedHtmlTag(each_line).strip())
startMainUrl = True
# 해외여행(패키지), 허니문, 골프, 국내(제주) 제외하고는 일단 패스
if MenuUrlCls.kind == 'A' or MenuUrlCls.kind == 'F' or MenuUrlCls.kind == 'H' or MenuUrlCls.kind == 'No':
continue
if startMainUrl and each_line.find('<li>') > -1:
SubMenuCls = clsSubMenu()
SubMenuCls.name = tourUtil.getRemovedHtmlTag(each_line).strip()
SubMenuCls.url = mainUrl + tourUtil.getTagAttr(each_line, 'a', 'href')
MenuUrlCls.subMenuList.append(SubMenuCls)
if startMainUrl and each_line.find('</div>') > -1:
startMainUrl = False
menuList.append(MenuUrlCls)
def searchProduct(filename,
productcode,
productName,
period,
targetUrl,
listUrl,
productDetailUrl,
departCity,
tourkind,
dmst_div,
country='',
city='',
comment=''):
detailHtml = savefilegethtml.getHtml(targetUrl, '', '',
'naeiltourDetailHtml.txt')
print >> filename, 'TargetUrl : ', targetUrl
departDayList = list()
for detail_each_line in detailHtml:
if detail_each_line.find("fn_goodDetail('") > -1:
departDayList.append(
detail_each_line.split("fn_goodDetail('")[1].split("'")[0])
# 출발 가능 날짜에 항공사 찾아오는 부분
try:
con = cx_Oracle.connect(
"bigtour/[email protected]:1521/ora11g")
codeList = codes.getCityCode(productName, city, comment, country)
cityList = codeList[0]
nationList = codeList[1]
#print nationList
#print cityList
#print nationList
#print cityList
query = savefilegethtml.getMasterMergeQueryTest1(
'naeiltour', productcode, '', country, city, productName, tourkind,
dmst_div, comment, '', nationList, cityList) # A : 해외(Abroad)
#query = savefilegethtml.getMasterMergeQuery('naeiltour', productcode, '', country, city, productName, tourkind, dmst_div, comment, '', nationList, cityList) # A : 해외(Abroad)
#print query
cursor = con.cursor()
cursor.execute(query)
con.commit()
productCls = clsProduct()
for dayInfo in departDayList:
try:
productListUrl = listUrl + productcode + '&sel_day=' + dayInfo
print 'ProductListUrl : ' + productListUrl
productListHtml = savefilegethtml.getHtml(
productListUrl, '', '', 'naeiltourproductListHtml.txt')
print >> filename, 'ProductListUrl : ' + productListUrl
for product in productListHtml:
try:
if product.find("fn_price('") > -1:
productCls = clsProduct()
productSplit = product.split('fn_price')[1].split(
"'")
productCls.productCode = productSplit[1]
productCls.dDay = productSplit[3]
productCls.code = productSplit[5]
if tourkind == 'W' or tourkind == 'G':
productCls.airCode = product[
product.find('.gif') -
2:product.find('.gif')]
else:
productCls.airCode = productSplit[
7] # 한글 공항코드... but 우리는 영문2자리 공항코드가 필요하다...
productCls.price = productSplit[9].replace(',', '')
#print productSplit[11]
productCls.status = codes.getStatus(
'naeiltour', productSplit[11]
) # 공백 : 예약가능, 03 : 마감임박, 05 : 마감
#if tourkind == 'W':
#productCls.city = productSplit[13]
productCls.url = productDetailUrl + productcode + '&sel_day=' + productCls.dDay
productCls.productname = productName
productCls.dTime = ''
productCls.aDay = ''
productCls.aTime = ''
if period != '' and tourkind == 'F':
if product.find('<td width="134">') > -1:
productCls.period = period
#print productCls.toString()
productCls.airCode = product[
product.find('.gif') -
2:product.find('.gif')]
query = savefilegethtml.getDetailMergeQueryTest1(
'naeiltour', productcode, productCls.code,
productCls.productname,
'20' + productCls.dDay, '',
productCls.period, departCity, '',
productCls.airCode, productCls.status,
productCls.url, productCls.price, '0', '0',
'0', '', '')
#query = savefilegethtml.getDetailMergeQuery('naeiltour', productcode, productCls.code, productCls.productname, '20' + productCls.dDay, '', productCls.period, departCity, '', productCls.airCode, productCls.status, productCls.url, productCls.price, '0', '0', '0', '', '')
#print query
cursor = con.cursor()
cursor.execute(query)
con.commit()
#break
if period == '' and tourkind == 'F':
if product.find('<td class="FRIDAYSPACING" >'
) > -1 and product.find(
'.gif') > -1:
productCls.airCode = product[
product.find('.gif') -
2:product.find('.gif')]
if product.find('idth="220">') > -1:
"""
# 날짜 가져오는 부분... 종류가 너무 많아서 좀 수정
if product.find('(') > -1:
productCls.night = re.findall(r"\d", product.split('(')[1])[0]
productCls.period = re.findall(r"\d", product.split('(')[1])[1]
elif product.find('[') > -1:
productCls.night = re.findall(r"\d", product.split('[')[1])[0]
productCls.period = re.findall(r"\d", product.split('[')[1])[1]
"""
splitText = product.decode('cp949').split(u'박')
if len(splitText) > 1:
tmpText = re.findall(
u'[\^0-9]+',
tourUtil.getRemovedHtmlTag(
splitText[0]))
productCls.night = tmpText[
len(tmpText) - 1].encode('utf-8')
tmpText = re.findall(
u'[\^0-9]+',
tourUtil.getRemovedHtmlTag(
splitText[1]))
productCls.period = tmpText[0].encode(
'utf-8')
else:
productCls.night = '0'
productCls.period = '0'
#############################################################################################
#print productCls.toString()
query = savefilegethtml.getDetailMergeQueryTest1(
'naeiltour', productcode, productCls.code,
productCls.productname,
'20' + productCls.dDay, '',
productCls.period, departCity, '',
productCls.airCode, productCls.status,
productCls.url, productCls.price, '0', '0',
'0', '', productCls.night)
#print 'Query : ' + query
cursor = con.cursor()
cursor.execute(query)
con.commit()
if period == '' and tourkind == 'W':
if product.find('valign="middle"') > -1:
"""
if product.find('(') > -1:
productCls.night = re.findall(r"\d", product.split('(')[1])[0]
productCls.period = re.findall(r"\d", product.split('(')[1])[1]
elif product.find('[') > -1:
productCls.night = re.findall(r"\d", product.split('[')[1])[0]
productCls.period = re.findall(r"\d", product.split('[')[1])[1]
"""
splitText = product.decode('cp949').split(u'박')
if len(splitText) > 1:
tmpText = re.findall(
u'[\^0-9]+',
tourUtil.getRemovedHtmlTag(
splitText[0]))
productCls.night = tmpText[
len(tmpText) - 1].encode('utf-8')
tmpText = re.findall(
u'[\^0-9]+',
tourUtil.getRemovedHtmlTag(
splitText[1]))
productCls.period = tmpText[0].encode(
'utf-8')
else:
productCls.night = '0'
productCls.period = '0'
#############################################################################################
#print productCls.toString()
query = savefilegethtml.getDetailMergeQueryTest1(
'naeiltour', productcode, productCls.code,
productCls.productname,
'20' + productCls.dDay, '',
productCls.period, departCity, '',
productCls.airCode, productCls.status,
productCls.url, productCls.price, '0', '0',
'0', '', productCls.night)
#print 'Query : ' + query
cursor = con.cursor()
cursor.execute(query)
con.commit()
#break
if period == '' and tourkind == 'G':
if product.find('valign="middle"') > -1:
"""
if product.find('(') > -1:
productCls.night = re.findall(r"\d", product.split('(')[1])[0]
productCls.period = re.findall(r"\d", product.split('(')[1])[1]
elif product.find('[') > -1:
productCls.night = re.findall(r"\d", product.split('[')[1])[0]
productCls.period = re.findall(r"\d", product.split('[')[1])[1]
"""
splitText = product.decode('cp949').split(u'박')
if len(splitText) > 1:
tmpText = re.findall(
u'[\^0-9]+',
tourUtil.getRemovedHtmlTag(
splitText[0]))
productCls.night = tmpText[
len(tmpText) - 1].encode('utf-8')
tmpText = re.findall(
u'[\^0-9]+',
tourUtil.getRemovedHtmlTag(
splitText[1]))
productCls.period = tmpText[0].encode(
'utf-8')
else:
productCls.night = '0'
productCls.period = '0'
#############################################################################################
#print productCls.toString()
query = savefilegethtml.getDetailMergeQueryTest1(
'naeiltour', productcode, productCls.code,
productCls.productname,
'20' + productCls.dDay, '',
productCls.period, departCity, '',
productCls.airCode, productCls.status,
productCls.url, productCls.price, '0', '0',
'0', '', productCls.night)
#print 'Query : ' + query
cursor = con.cursor()
cursor.execute(query)
con.commit()
#break
if period == '' and tourkind == 'D':
if product.find('<td class="FRIDAYSPACING" >'
) > -1 and product.find(
'.gif') > -1:
productCls.airCode = product[
product.find('.gif') -
2:product.find('.gif')]
if product.find('idth="220">') > -1:
"""
if product.find('(') > -1:
productCls.night = re.findall(r"\d", product.split('[')[1])[0]
productCls.period = re.findall(r"\d", product.split('')[1])[1]
else:
productCls.night = re.findall(r"\d", product.split('COLOR=#FF7A73>')[1])[0]
productCls.period = re.findall(r"\d", product.split('COLOR=#FF7A73>')[1])[1]
"""
splitText = product.decode('cp949').split(u'박')
if len(splitText) > 1:
tmpText = re.findall(
u'[\^0-9]+',
tourUtil.getRemovedHtmlTag(
splitText[0]))
productCls.night = tmpText[
len(tmpText) - 1].encode('utf-8')
tmpText = re.findall(
u'[\^0-9]+',
tourUtil.getRemovedHtmlTag(
splitText[1]))
productCls.period = tmpText[0].encode(
'utf-8')
else:
productCls.night = '0'
productCls.period = '0'
#############################################################################################
if product.find('COLOR=BLUE>') > -1:
departCity = 'PUS'
else:
departCity = 'ICN'
#print productCls.toString()
query = savefilegethtml.getDetailMergeQueryTest1(
'naeiltour', productcode, productCls.code,
productCls.productname,
'20' + productCls.dDay, '',
productCls.period, departCity, '',
productCls.airCode, productCls.status,
productCls.url, productCls.price, '0', '0',
'0', '', productCls.night)
#print 'Query : ' + query
cursor = con.cursor()
cursor.execute(query)
con.commit()
#break
except cx_Oracle.DatabaseError as err1:
print >> filename, err1
pass
except:
print >> filename, "Depth3 Error:", sys.exc_info()[0]
pass
#break
except:
print >> filename, "Depth2 Error:", sys.exc_info()[0]
pass
except:
print >> filename, "Depth1 Error:", sys.exc_info()[0]
pass
finally:
con.close()
except:
print >> exceptFile, "backpack :", sys.exc_info()[0]
pass
try:
for menu in backpackMenuList:
#tit_position2 부산출발 검색조건..
print menu.url
productListHtml = savefilegethtml.getHtml(menu.url, '<div id="sub_box2">', 'function btn(ckbtn){', 'productListHtml.txt')
for each_line in productListHtml:
if each_line.find('<h1 class="bic_h">') > -1:
productName = each_line.split('bic_h">')[2].split('<')[0]
#productNameSplit = productName.split(' ')
#period = productNameSplit[len(productNameSplit)-1].replace('일', '')
productNameSplit = re.findall('[\^0-9]+', tourUtil.getRemovedHtmlTag(each_line))
period = productNameSplit[len(productNameSplit)-1]
#if each_line.find('<span class="goods_text">') > -1: # 여행 설명이 잘 적혀 있긴 하지만.. 내일투어에서는 국가 정보가 있는 걸 보조로 가져가야 할듯..
#tourRoute = each_line.split('px;">')[1].split('<')[0]
if each_line.find('<span class="travel_box">') > -1:
tourRoute = tourUtil.getRemovedHtmlTag(each_line)
if each_line.find("sview('") > -1:
productCode = each_line.split("sview('")[1].split("'")[0]
code2 = each_line.split("sview('")[1].split("'")[2]
detailUrl = 'http://www.naeiltour.co.kr/backpack/program_include_list.asp?good_cd='+ productCode + '&sel_ym=' + targetYear + targetMonth
print >> exceptFile, 'DetailUrl : ', detailUrl
listUrl = 'http://www.naeiltour.co.kr/backpack/program_include_list.asp?good_cd='
productDetailUrl = 'http://www.naeiltour.co.kr/backpack/show.asp?good_cd='
for each_line in mainpageHtml:
#print main
if each_line.find('<!--') > -1 :
startComment = True
elif each_line.find('-->') > -1:
startComment = False
if firstOversea and each_line.find('해외패키지') > -1:
clsMain = mainCls()
clsMain.name = codes.getTourKind('lottetour', 'package')
firstOversea = False
elif not startComment and each_line.find('<li') > -1 and each_line.find('<a href=') > -1:
clsSubMenu = subMenuCls()
clsSubMenu.url = mainUrl + tourUtil.getTagAttr(each_line, 'a', 'href')
if each_line.find('title') > -1:
clsSubMenu.name = tourUtil.getRemovedHtmlTag(each_line).strip()
clsMain.subMenuList.append(clsSubMenu)
else:
subMenu = True
elif not startComment and subMenu and each_line.find('title=') > -1:
clsSubMenu.name = each_line.split('>')[1].split('<')[0]
clsMain.subMenuList.append(clsSubMenu)
subMenu = False
elif each_line.find('sub_depth0') > -1:
if len(clsMain.subMenuList) > 0:
clsMain.subMenuList.pop()
elif each_line.find('class="fit"') > -1 or each_line.find('class="honeymoon _open"') > -1 or each_line.find('class="golf"') > -1 or each_line.find('class="fit"') > -1 or each_line.find('class="cruise line"') > -1 or each_line.find('class="air line"') > -1:
mainList.append(clsMain)
clsMain = mainCls()
clsMain.name = codes.getTourKind('lottetour', tourUtil.getTagAttr(each_line, 'li', 'class'))
#최종 상품들 잡아넣자..
try:
productCls = clsProduct()
#productListHtml = open('productListHtml.txt')
departConfirm = False
for product in productListHtml:
#print 'product : ' + product
if product.find('pro_date') > -1:
productCls = clsProduct()
departConfirm = False
#productCls.sDay = targetYear + product.split('pro_date">')[1].split('(')[0].strip().replace('/', '')
#productCls.sTime = product.split('<br/>')[0].split(')')[1].strip().replace(':', '')
#productCls.aDay = targetYear + product.split('<span>')[1].split('(')[0].strip().replace('/', '')
#productCls.aTime = product.split('<span>')[1].split(')')[1].split('<')[0].strip().replace(':', '')
daySplit = tourUtil.getNumArray(
tourUtil.getRemovedHtmlTag(
product))
productCls.sDay = ''
productCls.sTime = ''
productCls.aDay = ''
productCls.aTime = ''
if len(daySplit) > 1:
productCls.sDay = targetYear + daySplit[
0] + daySplit[1]
if len(daySplit) > 3:
productCls.sTime = daySplit[
2] + daySplit[3]
if len(daySplit) > 5:
productCls.aDay = targetYear + daySplit[
4] + daySplit[5]
if len(daySplit) > 7: