def __inner_weibo_parse(weibo_driver):
weibo_info = weibo_driver.find_element_by_class_name(
'WB_info').find_element_by_tag_name('a')
weibo_author = weibo_info.text[1:].encode('utf-8')
weibo_href = weibo_info.get_attribute('href')
comment = weibo_driver.find_element_by_class_name('WB_text').text.encode(
'utf-8')
info_list = weibo_driver.find_elements_by_tag_name('li')
fordward_num = info_list[-3].text[3:].encode('utf-8')
comment_num = info_list[-2].text[3:].encode('utf-8')
good_num = info_list[-1].text.encode('utf-8')
weibo = Weibo()
weibo.author_name = weibo_author
weibo.author_url = weibo_href
weibo.comment = comment
if fordward_num:
weibo.forward_num = fordward_num
if comment_num:
weibo.comment_num = comment_num
if good_num:
weibo.good_num = good_num
weibo.is_forward = False
return weibo
def __inner_weibo_parse(weibo_driver):
weibo_info = weibo_driver.find_element_by_class_name('WB_info').find_element_by_tag_name('a')
weibo_author = weibo_info.text[1:].encode('utf-8')
weibo_href = weibo_info.get_attribute('href')
comment = weibo_driver.find_element_by_class_name('WB_text').text.encode('utf-8')
info_list = weibo_driver.find_elements_by_tag_name('li')
fordward_num = info_list[-3].text[3:].encode('utf-8')
comment_num = info_list[-2].text[3:].encode('utf-8')
good_num = info_list[-1].text.encode('utf-8')
weibo = Weibo()
weibo.author_name = weibo_author
weibo.author_url = weibo_href
weibo.comment = comment
if fordward_num:
weibo.forward_num = fordward_num
if comment_num:
weibo.comment_num = comment_num
if good_num:
weibo.good_num = good_num
weibo.is_forward = False
return weibo
def weibo_parse_simple(weibo_driver):
fordward_num = weibo_driver.find_element_by_xpath(".//span[@node-type='forward_btn_text']").text[3:].encode('utf-8')
comment_num = weibo_driver.find_element_by_xpath(".//span[@node-type='comment_btn_text']").text[3:].encode('utf-8')
good_num = weibo_driver.find_element_by_xpath(".//span[@node-type='like_status']").text.encode('utf-8')
weibo_time = weibo_driver.find_element_by_xpath(".//a[@node-type='feed_list_item_date']").text.encode('utf-8')
weibo = Weibo()
weibo.time = time_parse(weibo_time)
weibo.weibo_driver = weibo_driver
if fordward_num:
weibo.forward_num = fordward_num
if comment_num:
weibo.comment_num = comment_num
if good_num:
weibo.good_num = good_num
return weibo
def weibo_parse_simple(weibo_driver):
fordward_num = weibo_driver.find_element_by_xpath(
".//span[@node-type='forward_btn_text']").text[3:].encode('utf-8')
comment_num = weibo_driver.find_element_by_xpath(
".//span[@node-type='comment_btn_text']").text[3:].encode('utf-8')
good_num = weibo_driver.find_element_by_xpath(
".//span[@node-type='like_status']").text.encode('utf-8')
weibo_time = weibo_driver.find_element_by_xpath(
".//a[@node-type='feed_list_item_date']").text.encode('utf-8')
weibo = Weibo()
weibo.time = time_parse(weibo_time)
weibo.weibo_driver = weibo_driver
if fordward_num:
weibo.forward_num = fordward_num
if comment_num:
weibo.comment_num = comment_num
if good_num:
weibo.good_num = good_num
return weibo
def weibo_parse(weibo_driver):
print('解析一条微博')
try:
#自己微博就不会有
weibo_info = weibo_driver.find_element_by_class_name(
'WB_info').find_element_by_tag_name('a')
weibo_author = weibo_info.text.encode('utf-8')
weibo_href = weibo_info.get_attribute('href')
except:
weibo_author = ''
weibo_href = ''
comment = weibo_driver.find_element_by_xpath(
".//div[@node-type='feed_list_content']").text.encode('utf-8')
comment_handled = __comment_text_filter(comment)
nums_driver = weibo_driver.find_element_by_class_name('WB_feed_handle')
fordward_num = nums_driver.find_element_by_xpath(
".//span[@node-type='forward_btn_text']").text[3:].encode('utf-8')
comment_num = nums_driver.find_element_by_xpath(
".//span[@node-type='comment_btn_text']").text[3:].encode('utf-8')
good_num = nums_driver.find_element_by_xpath(
".//span[@node-type='like_status']").text.encode('utf-8')
forward_weibo_list = weibo_driver.find_elements_by_xpath(
".//div[@node-type='feed_list_forwardContent']")
weibo_time = weibo_driver.find_element_by_xpath(
".//a[@node-type='feed_list_item_date']").text.encode('utf-8')
weibo = Weibo()
if forward_weibo_list:
try:
forward_weibo = __inner_weibo_parse(forward_weibo_list[0])
weibo.is_forward = True
except Exception as e:
print e
forward_weibo = None
else:
forward_weibo = None
weibo.author_name = weibo_author
weibo.author_url = weibo_href
weibo.comment = comment_handled
weibo.time = time_parse(weibo_time)
weibo.weibo_driver = weibo_driver
if fordward_num:
weibo.forward_num = int(fordward_num)
if comment_num:
weibo.comment_num = int(comment_num)
if good_num:
weibo.good_num = int(good_num)
weibo.fordward_weibo = forward_weibo
return weibo
def weibo_parse(weibo_driver):
print('解析一条微博')
try:
#自己微博就不会有
weibo_info = weibo_driver.find_element_by_class_name('WB_info').find_element_by_tag_name('a')
weibo_author = weibo_info.text.encode('utf-8')
weibo_href = weibo_info.get_attribute('href')
except:
weibo_author = ''
weibo_href = ''
comment = weibo_driver.find_element_by_xpath(".//div[@node-type='feed_list_content']").text.encode('utf-8')
comment_handled = __comment_text_filter(comment)
nums_driver = weibo_driver.find_element_by_class_name('WB_feed_handle')
fordward_num = nums_driver.find_element_by_xpath(".//span[@node-type='forward_btn_text']").text[3:].encode('utf-8')
comment_num = nums_driver.find_element_by_xpath(".//span[@node-type='comment_btn_text']").text[3:].encode('utf-8')
good_num = nums_driver.find_element_by_xpath(".//span[@node-type='like_status']").text.encode('utf-8')
forward_weibo_list = weibo_driver.find_elements_by_xpath(".//div[@node-type='feed_list_forwardContent']")
weibo_time = weibo_driver.find_element_by_xpath(".//a[@node-type='feed_list_item_date']").text.encode('utf-8')
weibo = Weibo()
if forward_weibo_list:
try:
forward_weibo = __inner_weibo_parse(forward_weibo_list[0])
weibo.is_forward = True
except Exception as e:
print e
forward_weibo = None
else:
forward_weibo = None
weibo.author_name = weibo_author
weibo.author_url = weibo_href
weibo.comment = comment_handled
weibo.time = time_parse(weibo_time)
weibo.weibo_driver = weibo_driver
if fordward_num:
weibo.forward_num = int(fordward_num)
if comment_num:
weibo.comment_num = int(comment_num)
if good_num:
weibo.good_num = int(good_num)
weibo.fordward_weibo = forward_weibo
return weibo