def get_shop_info_list(self):
#开启第一个页面使用fast_get_page()
time.sleep(1)
self.fast_get_page(url='http://piao.qunar.com/',
is_scroll_to_bottom=False)
self.until_scroll_to_center_send_text_by_css_selector(
css_selector="#searchValue", text=self.data_region)
time.sleep(3)
self.until_scroll_to_center_click_by_css_selector(
css_selector='#searchBtn')
self.until_click_no_next_page_by_css_selector(
nextpagesetup=NextPageCssSelectorSetup(
css_selector='#pager-container > div > a.next',
pre_pagefunc=PageFunc(func=self.driver.refresh),
main_pagefunc=PageFunc(func=self.get_shop_info)))
def get_shop_comment(self):
self.fast_new_page(url='http://www.baidu.com')
shop_collcetion = Mongodb(db=TravelDriver.db,
collection=TravelDriver.shop_collection,
host='10.1.17.15').get_collection()
shop_name_url_list = list()
for i in shop_collcetion.find(self.get_data_key()):
if i.get('shop_url'):
shop_name_url_list.append(
(i.get('shop_name'), i.get('shop_url')))
for i in range(len(shop_name_url_list)):
self.info_log(data='第%s个,%s' % (i + 1, shop_name_url_list[i][0]))
#第一次打开的时候进行验证 后面都不需要
while (True):
self.is_ready_by_proxy_ip()
self.switch_window_by_index(index=-1)
self.deal_with_failure_page()
self.fast_new_page(url=shop_name_url_list[i][1])
time.sleep(1)
self.switch_window_by_index(index=-1) # 页面选择
if '验证中心' in self.driver.title:
self.info_log(data='关闭验证页面!!!')
self.close_curr_page()
else:
break
self.until_click_no_next_page_by_css_selector(
nextpagesetup=NextPageCssSelectorSetup(
css_selector='#remark_page > a.page-next',
stop_css_selector='#remark_page > a.page-next.hidden',
main_pagefunc=PageFunc(func=self.from_page_get_data_list,
page=page_comment_1)))
self.close_curr_page()
def get_shop_comment(self):
shop_collcetion = Mongodb(db=TravelDriver.db,
collection=TravelDriver.shop_collection,
host='10.1.17.15').get_collection()
shop_name_url_list = list()
for i in shop_collcetion.find(self.get_data_key()):
if i.get('shop_comment_url'):
shop_name_url_list.append(
(i.get('shop_name'), i.get('shop_comment_url')))
for i in range(len(shop_name_url_list)):
self.info_log(data='第%s个,%s' % (i + 1, shop_name_url_list[i][0]))
while (True):
self.is_ready_by_proxy_ip()
self.switch_window_by_index(index=-1)
self.deal_with_failure_page()
self.fast_new_page(url=shop_name_url_list[i][1])
time.sleep(1)
self.switch_window_by_index(index=-1) # 页面选择
if '验证中心' in self.driver.title:
self.info_log(data='关闭验证页面!!!')
self.close_curr_page()
else:
break
self.until_click_no_next_page_by_partial_link_text(
nextpagesetup=NextPageLinkTextSetup(
link_text='下一页',
main_pagefunc=PageFunc(func=self.from_page_get_data_list,
page=page_comment_1)))
self.close_curr_page()
def get_comment_list(self):
self.fast_new_page(url="http://www.baidu.com")
shop_collcetion = Mongodb(db=TravelDriver.db,
collection=TravelDriver.shop_collection,
host='localhost').get_collection()
shop_name_url_list = list()
for i in shop_collcetion.find(self.get_data_key()):
if i.get('shop_url'):
shop_name_url_list.append(
(i.get('shop_name'), i.get('shop_url')))
for i in range(len(shop_name_url_list)):
self.info_log(data='第%s个,%s' % (i + 1, shop_name_url_list[i][0]))
self.shop_name = shop_name_url_list[i][0]
self.fast_new_page("http://www.baidu.com")
self.fast_new_page(url=shop_name_url_list[i][1])
self.until_click_no_next_page_by_css_selector(
nextpagesetup=NextPageCssSelectorSetup(
css_selector=
'#allCmtComment > div.paging.orangestyle > div > a.nextpage',
stop_css_selector=
'#allCmtComment > div.paging.orangestyle > div > a.nextpage.hidden',
main_pagefunc=PageFunc(func=self.from_page_get_data_list,
page=page_comment_1),
pause_time=2))
self.close_curr_page()
def get_shop_info_list(self):
self.fast_new_page(url='http://www.dianping.com/qiandaohu/food')
#这里加一个循环即可
shop_cooks = ['水果生鲜']
for shop_cook in shop_cooks:
# time.sleep(5)
self.shop_cook_style = shop_cook
for i in range(1, 10):
#self.fast_new_page(url='http://www.dianping.com/qiandaohu/ch10/g118r85473')
self.fast_click_first_item_same_page_by_partial_link_text(
link_text=shop_cook)
time.sleep(2)
try:
self.driver.find_element_by_css_selector(
css_selector='#bussi-nav > a:nth-child(' + str(i) +
')').click()
time.sleep(2)
self.shop_site = self.driver.find_element_by_xpath(
'//*[@id="bussi-nav"]/a[' + str(i) + ']').text
#self.shop_site = '千岛湖风景区'
self.until_click_no_next_page_by_partial_link_text(
NextPageLinkTextSetup(link_text='下一页',
is_proxy=False,
main_pagefunc=PageFunc(
self.from_page_get_data_list,
page=page_shop_1)))
#self.close_curr_page()
except Exception:
print("无标签")
def get_shop_info_list(self):
def get_shop_list(subtype):
shop_data_list = self.from_page_get_data_list(page=page_shop_1)
for shop_data in shop_data_list:
self.save_data_to_mongodb(
fieldlist=fl_shop1,
mongodb=Mongodb(db=TravelDriver.db,
collection=TravelDriver.shop_collection,
host='10.1.17.15'),
data=self.merge_dict(shop_data, subtype),
external_key_name=[FieldName.SUBTYPE_NAME])
self.fast_click_first_item_page_by_partial_link_text(link_text='美食')
time.sleep(2)
self.until_scroll_to_center_click_by_css_selector(
css_selector='#J_qs-btn')
time.sleep(2)
subtype_list = []
for i in self.until_presence_of_all_elements_located_by_css_selector(
css_selector='#classfy > a'):
if i.text and i.text != '更多':
subtype_list.append({
FieldName.SUBTYPE_NAME:
i.text,
FieldName.SUBTYPE_URL:
i.get_attribute('href')
})
for subtype in subtype_list:
self.fast_new_page(url=subtype.get(FieldName.SUBTYPE_URL))
self.until_click_no_next_page_by_partial_link_text(
NextPageLinkTextSetup(link_text='下一页',
is_proxy=False,
main_pagefunc=PageFunc(get_shop_list,
subtype=subtype)))
self.close_curr_page()
def get_shop_info_list(self):
self.fast_get_page(
'http://s.lvmama.com/hotel/U69C20180919O20180920?mdd=%E8%AF%B7%E8%BE%93%E5%85%A5%E7%9B%AE%E7%9A%84%E5%9C%B0#list',
is_max=False)
time.sleep(2)
city = self.get_city_from_region_CHN(self.data_region)
self.until_scroll_to_center_send_enter_by_css_selector(
css_selector='#js_destination', text=city)
self.until_scroll_to_center_send_enter_by_css_selector(
css_selector='#js_keyword', text=self.data_region)
time.sleep(1)
# self.until_send_enter_by_css_selector(css_selector='#js_keyword')
self.fast_click_page_by_css_selector('#btn_search')
# self.fast_click_page_by_css_selector('body > div.lv-ban > div.lv_s_all > div > div:nth-child(1) > div.lv_s_search > span')
#这里根据字段的不同重新进行编写
#千岛湖
# self.fast_click_page_by_css_selector('#search-params > div.search-nav-box.clearfix > p > a:nth-child(7)')
#千岛湖森林氧吧
# self.fast_click_page_by_css_selector('body > div.banWrap.pr > div.hotelSeach.fix.pa.yh.f14 > div.hotelSeachbtn.fl.pr.tc.f18')
# self.fast_click_page_by_css_selector('#_j_mfw_search_main > div.s-nav > div > div > a:nth-child(4)')
time.sleep(1)
#self.vertical_scroll_to() # 滚动到页面底部
self.until_click_no_next_page_by_partial_link_text(
nextpagesetup=NextPageLinkTextSetup(link_text="下一页",
main_pagefunc=PageFunc(
func=self.get_shop_info)))
def get_comment_info_list(self):
self.fast_new_page(url='http://www.baidu.com')
shop_collcetion = Mongodb(db=TravelDriver.db,
collection=TravelDriver.shop_collection,
host='localhost').get_collection()
shop_name_url_list = list()
for i in shop_collcetion.find(self.get_data_key()):
if i.get('shop_url'):
shop_name_url_list.append(
(i.get('shop_name'), i.get('shop_comment_url')))
for i in range(len(shop_name_url_list)):
self.fast_new_page(url="https://www.baidu.com")
# 可能会有反爬
self.info_log(data='第%s个,%s' % (i + 1, shop_name_url_list[i][0]))
self.fast_new_page(url=shop_name_url_list[i][1])
self.shop_name = shop_name_url_list[i][0]
time.sleep(5)
self.until_click_no_next_page_by_partial_link_text(
nextpagesetup=NextPageLinkTextSetup(
link_text='下一页',
main_pagefunc=PageFunc(func=self.from_page_get_data_list,
page=page_comment_1),
pause_time=2))
self.close_curr_page()
def get_shop_info(self):
try:
shop_data_list = self.from_page_get_data_list(page=page_shop_1)
nextpagesetup = NextPageCssSelectorSetup(css_selector='#review-list > div.review-list-container > div.review-list-main > div.reviews-wrapper > div.bottom-area.clearfix > div > a.NextPage',page=page_comment_1, pause_time=2, pre_pagefunc=PageFunc(func=self.more_comment), after_pagefunc=PageFunc(func=self.close_curr_page))
extra_pagefunc = PageFunc(func=self.get_newest_comment_data_by_css_selector, nextpagesetup=nextpagesetup, shop_name_css_selector='#poi-detail > div.container > div.base-info > div.main-detail.clearfix > div.main-detail-left > div.main-detail-left-top.clearfix > div.hotel-detail-info > div > h1', is_effective=False)
self.from_page_add_data_to_data_list(page=page_shop_2, pre_page=page_shop_1, data_list=shop_data_list, extra_pagefunc=extra_pagefunc)
except Exception as e:
self.error_log(e=str(e))
def get_shop_info_list(self):
self.driver.get('https://www.baidu.com')
self.fast_new_page(url='http://piao.ctrip.com/',
is_scroll_to_bottom=False)
self.driver.refresh()
self.until_scroll_to_center_send_text_by_css_selector(
css_selector="#mainInput", text=self.data_region)
time.sleep(3)
self.until_scroll_to_center_send_enter_by_css_selector(
css_selector="#mainInput")
self.until_click_no_next_page_by_css_selector(
nextpagesetup=NextPageCssSelectorSetup(
css_selector=
'#searchResultContainer > div.pkg_page.basefix > a.down',
stop_css_selector=
'#searchResultContainer > div.pkg_page.basefix > a.down.down_nocurrent',
pre_pagefunc=PageFunc(func=self.driver.refresh),
main_pagefunc=PageFunc(func=self.get_shop_info)))
def get_shop_info_list(self):
self.fast_click_first_item_page_by_partial_link_text(link_text='酒店')
time.sleep(2)
self.until_click_no_next_page_by_partial_link_text(
NextPageLinkTextSetup(link_text='下一页',
is_proxy=False,
main_pagefunc=PageFunc(
self.from_page_get_data_list,
page=page_shop_1)))
def get_shop_info(self):
try:
shop_data_list = self.from_page_get_data_list(page=page_shop_1)
nextpagesetup = NextPageCssSelectorSetup(
css_selector=
'#divCtripComment > div.c_page_box > div > a.c_down',
page=page_comment_1,
pause_time=2,
pre_pagefunc=PageFunc(func=self.change_index_sort))
extra_pagefunc = PageFunc(
func=self.get_newest_comment_data_by_css_selector,
nextpagesetup=nextpagesetup)
self.from_page_add_data_to_data_list(page=page_shop_2,
pre_page=page_shop_1,
data_list=shop_data_list,
extra_pagefunc=extra_pagefunc)
except Exception as e:
self.error_log(e=str(e))
def get_shop_info(self):
try:
shop_data_list = self.from_page_get_data_list(page=page_shop_1)
# root > div > div > div > div > div:nth-child(3) > div.main-bd > div > div:nth-child(5) > div.detail-left > div.content-wrapper.clearfix > ul.pkg_page > a.down
# root > div > div > div > div > div:nth-child(3) > div.main-bd > div > div:nth-child(5) > div.detail-left > div.content-wrapper.clearfix > ul.pkg_page > a.down.disabled.nocurrent
# nextpagesetup = NextPageCssSelectorSetup(css_selector='div.detail-left > div.content-wrapper.clearfix > ul.pkg_page > a.down', stop_css_selector='div.detail-left > div.content-wrapper.clearfix > ul.pkg_page > a.down.disabled.nocurrent', page=page_comment_1, pause_time=2)
extra_pagefunc = PageFunc(func=self.get_comment)
self.from_page_add_data_to_data_list(page=page_shop_2, pre_page=page_shop_1, data_list=shop_data_list,extra_pagefunc=extra_pagefunc)
except Exception as e:
self.error_log(e=str(e))
def get_shop_info(self):
try:
shop_data_list = self.from_page_get_data_list(page=page_shop_1)
# nextpagesetup = NextPageCssSelectorSetup(css_selector='div.detail-left > div.content-wrapper.clearfix > ul.pkg_page > a.down', stop_css_selector='div.detail-left > div.content-wrapper.clearfix > ul.pkg_page > a.down.disabled.nocurrent', page=page_comment_1, pause_time=2)
# extra_pagefunc = PageFunc(func=self.get_newest_comment_data_by_css_selector, nextpagesetup=nextpagesetup)
self.from_page_add_data_to_data_list(
page=page_shop_2,
pre_page=page_shop_1,
data_list=shop_data_list,
pre_pagefunc=PageFunc(func=self.shop_detail_page_unfold))
except Exception as e:
self.error_log(e=str(e))
def get_shop_info_list(self):
self.fast_get_page('http://www.mafengwo.cn/hotel/', is_max=True)
time.sleep(2)
self.until_send_text_by_css_selector(css_selector='#_j_search_input',
text=self.data_region)
self.until_send_enter_by_css_selector(css_selector='#_j_search_input')
time.sleep(1)
self.vertical_scroll_to() #滚动到页面底部,为了使得整个页面都加载完成
self.until_click_no_next_page_by_partial_link_text(
nextpagesetup=NextPageLinkTextSetup(link_text="后一页",
main_pagefunc=PageFunc(
func=self.get_shop_info)))
def get_shop_info_list(self):
self.fast_new_page(
url='http://www.qiandaohu.cc/qdhfg/jqjd/xcyjd/#wzqdh')
time.sleep(60)
self.until_click_no_next_page_by_partial_link_text(
NextPageLinkTextSetup(link_text='下一页',
is_proxy=False,
main_pagefunc=PageFunc(
self.from_page_get_data_list,
page=page_shop_1)))
def get_shop_info_list(self):
self.fast_get_page('http://www.mafengwo.cn/', is_max=False)
time.sleep(2)
self.until_send_text_by_css_selector(css_selector='#_j_index_search_input_all', text=self.data_region)
self.until_send_enter_by_css_selector(css_selector='#_j_index_search_input_all')
self.fast_click_first_item_same_page_by_partial_link_text('景点')
self.fast_click_page_by_css_selector('#_j_mfw_search_main > div.s-nav > div > div > a:nth-child(4)')
time.sleep(1)
self.vertical_scroll_to() # 滚动到页面底部
self.until_click_no_next_page_by_partial_link_text(
nextpagesetup=NextPageLinkTextSetup(link_text="下一页", main_pagefunc=PageFunc(func=self.get_shop_info)))
def get_shop_info(self):
try:
shop_data_list = self.from_page_get_data_list(page=page_shop_1)
nextpagesetup = NextPageCssSelectorSetup(
css_selector='div._j_commentlist > div.m-pagination > a.pi.pg-next'
,
stop_css_selector='div._j_commentlist > div.m-pagination > a.pi.pg-next.hidden',
page=page_comment_1, pause_time=2)
extra_pagefunc = PageFunc(func=self.get_newest_comment_data_by_css_selector, nextpagesetup=nextpagesetup)
self.from_page_add_data_to_data_list(page=page_shop_2, pre_page=page_shop_1, data_list=shop_data_list,extra_pagefunc=extra_pagefunc
)
except Exception as e:
self.error_log(e=str(e))
def get_shop_info_list(self):
self.driver.get('https://www.baidu.com')
self.fast_new_page('http://hotels.ctrip.com/',
is_scroll_to_bottom=False)
self.until_scroll_to_center_send_text_by_css_selector(
css_selector="#txtCity", text=self.data_region)
time.sleep(3)
self.until_scroll_to_center_send_enter_by_css_selector(
css_selector="#txtCity")
time.sleep(2)
self.fast_click_same_page_by_css_selector(
click_css_selector='#btnSearch')
self.until_click_no_next_page_by_css_selector(
nextpagesetup=NextPageCssSelectorSetup(
css_selector='#downHerf.c_down',
main_pagefunc=PageFunc(func=self.get_shop_info)))
def get_shop_info(self):
try:
print(1234567890)
shop_data_list = self.from_page_get_data_list(page=page_shop_1)
nextpagesetup = NextPageCssSelectorSetup(
css_selector='#remarksPage > a.page-next'
,
stop_css_selector='#remarksPage > a.page-next.hidden',
page=page_comment_1, pause_time=2)
extra_pagefunc = PageFunc(func=self.get_newest_comment_data_by_css_selector, nextpagesetup=nextpagesetup)
self.from_page_add_data_to_data_list(page=page_shop_2, pre_page=page_shop_1, data_list=shop_data_list,extra_pagefunc=extra_pagefunc
)
except Exception as e:
print(0000000000000000)
self.error_log(e=str(e))
def get_shop_good_middle_bad_comment(self):
# self.fast_new_page(url='http://www.baidu.com')
# shop_collcetion = Mongodb(db=TravelDriver.db, collection=TravelDriver.shop_collection,
# ).get_collection()
# # shop_name_url_list = ['http://www.dianping.com/shop/23242707/review_all','http://www.dianping.com/shop/114535359/review_all','http://www.dianping.com/shop/107667835/review_all']
# shop_name_url_list = list();
# for i in shop_collcetion.find(self.get_data_key()):
# # if i.get('shop_comment_url'):
# if(i.get('shop_comment_num') > 1000):
#
# shop_name_url_list.append((i.get('shop_name'),i.get('shop_comment_url')))
# print(len(shop_name_url_list))
# for i in range(len(shop_name_url_list)):
try:
# self.info_log(data='第%s个,%s' % (i + 1, shop_name_url_list[i][0]))
# time.sleep(10)
#self.fast_new_page(url="http://www.baidu.com", is_scroll_to_bottom=False)
self.fast_new_page(url="http://www.dianping.com",
is_scroll_to_bottom=False)
time.sleep(20)
self.fast_new_page(
url="http://www.dianping.com/shop/19139636/review_all",
is_scroll_to_bottom=False)
time.sleep(30)
# try:
# self.driver.find_element_by_link_text(link_text='默认排序').click();
# time.sleep(2)
# self.driver.find_element_by_link_text(link_text='最新点评').click();
# time.sleep(20)
# except Exception:
# print("无最新点评")
##随后进行点击标签和评阅
self.until_click_no_next_page_by_partial_link_text(
nextpagesetup=NextPageLinkTextSetup(
link_text='下一页',
main_pagefunc=PageFunc(func=self.from_page_get_data_list,
page=page_comment_1),
pause_time=2))
self.close_curr_page()
except Exception:
print("无元素")
self.close_curr_page()
def get_shop_info_list(self):
self.fast_get_page('http://piao.qunar.com/', is_max=False)
time.sleep(2)
# self.fast_click_page_by_css_selector('#js_nva_cgy > li.c_piao.js-searchnav.cur > a')
self.until_send_text_by_css_selector(css_selector='#searchValue',
text=self.data_region)
# self.until_send_enter_by_css_selector(css_selector='#searchValue')
self.fast_click_page_by_css_selector('#searchBtn')
# self.until_send_enter_by_css_selector(css_selector='#js-piao-ticket > div.qcbox > div.qunar-qcbox > input')
time.sleep(10)
self.until_click_no_next_page_by_css_selector(
nextpagesetup=NextPageCssSelectorSetup(
css_selector='#pager-container > div > a.next',
stop_css_selector='#pager-container > div > a.next.hidden',
main_pagefunc=PageFunc(func=self.from_page_get_data_list,
page=page_shop_1)))
def get_shop_info(self):
try:
shop_data_list = self.from_page_get_data_list(page=page_shop_1)
nextpagesetup = NextPageCssSelectorSetup(
css_selector=
'#allCmtComment > div.paging.orangestyle > div > a.nextpage',
stop_css_selector=
'#allCmtComment > div.paging.orangestyle > div > a.nextpage.hidden',
page=page_comment_1,
pause_time=2)
extra_pagefunc = PageFunc(
func=self.get_newest_comment_data_by_css_selector,
nextpagesetup=nextpagesetup)
self.from_page_add_data_to_data_list(page=page_shop_2,
pre_page=page_shop_1,
data_list=shop_data_list,
extra_pagefunc=extra_pagefunc)
except Exception as e:
self.error_log(e=str(e))
def get_shop_info_list(self):
self.driver.get('https://www.baidu.com')
self.fast_new_page(url='http://ticket.lvmama.com/',
is_scroll_to_bottom=False)
#self.driver.refresh()
self.until_scroll_to_center_send_text_by_css_selector(
css_selector=
"body > div.banWrap > div > div.lv_s_box > div.lv_s_all > div.lv_s_input_box > input",
text=self.data_region)
self.until_send_enter_by_css_selector(
css_selector=
'body > div.banWrap > div > div.lv_s_box > div.lv_s_all > div.lv_s_input_box > input'
)
time.sleep(5)
self.until_click_no_next_page_by_partial_link_text(
NextPageLinkTextSetup(link_text='下一页',
is_proxy=False,
main_pagefunc=PageFunc(
self.from_page_get_data_list,
page=page_shop_1)))
def get_shop_info_list(self):
self.fast_get_page(url='http://you.ctrip.com/')
self.until_scroll_to_center_send_text_by_css_selector(
css_selector="#gsSearch", text=self.data_region)
time.sleep(5)
self.fast_click_page_by_css_selector(click_css_selector='#SearchBtn')
time.sleep(10)
self.fast_click_page_by_css_selector(
click_css_selector=
'body > div.content.cf > div.main > div.search-content.cf > ul > li.current > a > span'
)
# #body > div.content.cf > div.main > div.search - content.cf > div > div.result > div.desNavigation.cf > p > span > a.left_arrow
# body > div.content.cf > div.main > div.search - content.cf > div > div.result > div.desNavigation.cf > p > span > em.right_arrow
self.until_click_no_next_page_by_partial_link_text(
NextPageLinkTextSetup(link_text='下一页',
is_proxy=False,
main_pagefunc=PageFunc(
self.from_page_get_data_list,
page=page_shop_1)))
def get_shop_info_list(self):
self.fast_get_page(
url=
'http://touch.piao.qunar.com/touch/list_%E5%8C%97%E4%BA%AC.html?isSearch=1&cityName=%E5%8C%97%E4%BA%AC'
)
time.sleep(5)
self.until_scroll_to_center_send_text_by_css_selector(
css_selector='#search-input-bind', text=self.data_region)
#睡得久一点 让整个页面都加载出来
time.sleep(4)
self.fast_click_page_by_css_selector(
click_css_selector='#search-form-submit')
time.sleep(8)
#shop_data_list = self.from_page_get_data_list(page=page_shop_1)
self.until_click_no_next_page_by_partial_link_text(
nextpagesetup=NextPageLinkTextSetup(
link_text='下一页',
pause_time=5,
main_pagefunc=PageFunc(func=self.from_page_get_data_list,
page=page_shop_1)))
def get_shop_info_list(self):
city = self.get_city_from_region_ENG(self.data_region)
self.fast_get_page('http://hotel.qunar.com/city' + city,
is_max=False,
is_scroll_to_bottom=False)
time.sleep(2)
self.until_send_text_by_css_selector(css_selector='#jxQ',
text=self.data_region)
self.until_send_enter_by_css_selector(css_selector='#jxQ')
self.vertical_scroll_to(offset=1000000000) # 滚动到页面底部
# self.until_ismore_by_send_key_arrow_down_judge_by_len(list_css_selector='#search-list > div',
# ele_css_selector='#loadingDiv > div')
self.vertical_scroll_to() # 滚动到页面底部
time.sleep(5)
self.until_click_no_next_page_by_partial_link_text(
nextpagesetup=NextPageLinkTextSetup(link_text="下一页",
main_pagefunc=PageFunc(
func=self.get_shop_info)))
def get_shop_info(self):
try:
shop_data_list = self.from_page_get_data_list(page=page_shop_1)
nextpagesetup = NextPageCssSelectorSetup(
css_selector=
'#comment_main > div > div.wrapper > div.b_ugcpager.clrfix.js-pager-container > div.js-pager > div > a.next > span',
stop_css_selector=
'#comment_main > div > div.wrapper > div.b_ugcpager.clrfix.js-pager-container > div.js-pager > div > a.next > span.hidden',
page=page_comment_1,
pause_time=2)
extra_pagefunc = PageFunc(
func=self.get_newest_comment_data_by_css_selector,
nextpagesetup=nextpagesetup)
self.from_page_add_data_to_data_list(page=page_shop_2,
pre_page=page_shop_1,
data_list=shop_data_list,
extra_pagefunc=extra_pagefunc)
except Exception as e:
self.error_log(e=str(e))
def get_shop_info_list(self):
self.fast_get_page(url='http://hotel.tuniu.com/', is_scroll_to_bottom=False)
city = self.get_city_from_region_CHN(self.data_region)
self.until_send_text_by_css_selector(css_selector='#txtCity',text=city)
time.sleep(5)
self.fast_click_page_by_css_selector('#search_hotel')
self.until_send_text_by_css_selector(css_selector='#keyWord', text=self.data_region)
time.sleep(5)
self.fast_click_page_by_css_selector('#search_hotel')
# self.fast_click_page_by_css_selector(
# '#topContainer > div > div > div.box.clearfix.topnav.common > a:nth-child(14)',is_scroll_to_bottom=True)
time.sleep(10)
self.until_click_no_next_page_by_css_selector(
nextpagesetup=NextPageCssSelectorSetup(css_selector='#main > div.pagination.clearfix > div > span.next',stop_css_selector='span.next > i.hidden',
main_pagefunc=PageFunc(func=self.get_shop_info)))
def get_shop_comment(self):
shop_collcetion = Mongodb(db=TravelDriver.db,
collection=TravelDriver.shop_collection,
host='10.1.17.15').get_collection()
shop_name_url_list = list()
for i in shop_collcetion.find(self.get_data_key()):
if i.get('shop_comment_url'):
shop_name_url_list.append(
(i.get('shop_name'), i.get('shop_comment_url')))
# for i in range(len(shop_name_url_list)):
# self.info_log(data='第%s个,%s'%(i+1, shop_name_url_list[i][0]))
# while (True):
# self.is_ready_by_proxy_ip()
# self.switch_window_by_index(index=-1)
# self.deal_with_failure_page()
# self.fast_new_page(url=shop_name_url_list[i][1])
# time.sleep(1)
# self.switch_window_by_index(index=-1) # 页面选择
# if '验证中心' in self.driver.title:
# self.info_log(data='关闭验证页面!!!')
# self.close_curr_page()
# else:
# break
# dianping = self.driver.find_element_by_css_selector(
# css_selector='#review-list > div.review-list-container > div.review-list-main > div.reviews-wrapper > div.reviews-tags > div.content')
# self.fast_click_same_page_by_css_selector(click_css_selector='#main-page > header > h2 > div:nth-child(2)')
self.fast_new_page(
url='http://www.dianping.com/shop/110833569/review_all')
span = '#review-list > div.review-list-container > div.review-list-main > div.reviews-wrapper > div.reviews-tags > div.content > span:nth-child(' + str(
1) + ')'
links = self.driver.find_elements_by_css_selector(
span)[0].get_attribute('text')
self.until_click_no_next_page_by_partial_link_text(
nextpagesetup=NextPageLinkTextSetup(
link_text='下一页',
main_pagefunc=PageFunc(func=self.from_page_get_data_list,
page=page_comment_1)))
self.close_curr_page()
