Python clearSpace示例

示例#1

文件： OfficeWeiboMsg.py 项目： rabintang/weibo_crawler

 def get_retweet_info(self):
     tag = 'feed_list_forwardContent'
     pos = self.content.find(tag, self.pos, self.end)
     temp = pos  #被删除的情况时，用到这个position
     if pos != -1:  #被删除的微薄，tag也能找到，下面注意限定范围
         tag = 'feed_list_originNick'
         pos = self.content.find(tag, pos, self.end)
         if pos != -1:  #正常情况下
             self.pos = pos
             self.get_retweet_un()
             self.get_retweet_mc()
             self.get_weibopu()
             self.get_retweet_rc()
             self.get_retweet_cc()
             self.get_retweet_page()
             self.get_retweet_pt()
         else:  #被删除的情况；还有情况就是：只有文字，其他啥也没有
             tag = 'WB_deltxt'
             pos = self.content.find(tag, temp, self.end)
             bTag = '">'
             eTag = '<\/div>'
             pos1 = self.content.find(bTag, pos, self.end) + len(bTag)
             pos2 = self.content.find(eTag, pos1, self.end)
             self.weibomsg['rmc'] = self.content[pos1:pos2]
             self.weibomsg['rmc'] = self.eraseTag(
                 self.weibomsg['rmc']).replace('\\', '')
             self.weibomsg['rmc'] = utility.clearSpace(self.weibomsg['rmc'])
             self.pos = pos2
     else:
         pass  #转发信息都为空

示例#2

文件： CommonWeiboMsg.py 项目： Windy-Ground/weibo_crawler

	def get_retweet_info(self):
		tag = 'feed_list_forwardContent'
		pos =  self.content.find(tag, self.pos, self.end)
		temp = pos	#被删除的情况时，用到这个position
		if pos != -1:	#被删除的微薄，tag也能找到，下面注意限定范围
			tag = 'feed_list_originNick'
			pos =  self.content.find(tag, pos, self.end)
			if pos != -1:	#正常情况下
				self.pos = pos
				self.get_retweet_un()
				self.get_retweet_mc()
				self.get_weibopu()
				self.get_retweet_rc()
				self.get_retweet_cc()
				self.get_retweet_page()
				self.get_retweet_pt()
			else:	#被删除的情况；还有情况就是：只有文字，其他啥也没有
				tag = 'WB_deltxt'
				pos =  self.content.find(tag, temp, self.end)
				bTag = '">'
				eTag = '<\/div>'
				pos1 = self.content.find(bTag, pos, self.end)+len(bTag)
				pos2 = self.content.find(eTag, pos1, self.end)
				self.weibomsg['rmc'] = self.content[pos1:pos2]
				self.weibomsg['rmc'] = self.eraseTag(self.weibomsg['rmc']).replace('\\','')
				self.weibomsg['rmc'] = utility.clearSpace(self.weibomsg['rmc']);
				self.pos = pos2
		else:
			pass	#转发信息都为空

示例#3

文件： CommonWeiboMsg.py 项目： Windy-Ground/weibo_crawler

	def get_retweet_mc(self):
		bTag = 'node-type="feed_list_reason">'
		eTag = '<\/div>'
		pos1 = self.content.find(bTag, self.pos, self.end)+len(bTag)
		pos2 = self.content.find(eTag, pos1, self.end)
		self.weibomsg['rmc'] = self.content[pos1:pos2]
		self.pos = pos2
		self.weibomsg['rmc'] = self.eraseTag(self.weibomsg['rmc']).replace('\\','')
		self.weibomsg['rmc'] = utility.clearSpace(self.weibomsg['rmc']);

示例#4

文件： OfficeWeiboMsg.py 项目： rabintang/weibo_crawler

 def get_retweet_mc(self):
     bTag = 'node-type="feed_list_reason">'
     eTag = '<\/div>'
     pos1 = self.content.find(bTag, self.pos, self.end) + len(bTag)
     pos2 = self.content.find(eTag, pos1, self.end)
     self.weibomsg['rmc'] = self.content[pos1:pos2]
     self.pos = pos2
     self.weibomsg['rmc'] = self.eraseTag(self.weibomsg['rmc']).replace(
         '\\', '')
     self.weibomsg['rmc'] = utility.clearSpace(self.weibomsg['rmc'])

示例#5

文件： CommonWeiboMsg.py 项目： Windy-Ground/weibo_crawler

	def get_weibomc(self):
		tag = 'node-type="feed_list_content"'
		self.pos = self.content.find(tag, self.pos, self.end)+len(tag)
		bTag = '>'
		eTag = '<\/div>'
		pos1 = self.content.find(bTag, self.pos, self.end) + len(bTag)
		pos2 = self.content.find(eTag, pos1, self.end)
		self.weibomsg['mc'] = self.content[pos1:pos2]
		self.weibomsg['mc'] = self.eraseTag(self.weibomsg['mc']).replace('\\','')
		self.weibomsg['mc'] = utility.clearSpace(self.weibomsg['mc']);
		self.pos = pos2

示例#6

文件： OfficeWeiboMsg.py 项目： rabintang/weibo_crawler

 def get_weibomc(self):
     tag = 'node-type="feed_list_content"'
     self.pos = self.content.find(tag, self.pos, self.end) + len(tag)
     bTag = '>'
     eTag = '<\/div>'
     pos1 = self.content.find(bTag, self.pos, self.end) + len(bTag)
     pos2 = self.content.find(eTag, pos1, self.end)
     self.weibomsg['mc'] = self.content[pos1:pos2]
     self.weibomsg['mc'] = self.eraseTag(self.weibomsg['mc']).replace(
         '\\', '')
     self.weibomsg['mc'] = utility.clearSpace(self.weibomsg['mc'])
     self.pos = pos2

示例#7

 def get_retweet_cc(self):
     bTag = '评论'
     eTag = '<\/a>'
     pos1 = self.content.find(bTag, self.pos, self.end) + len(bTag)
     pos2 = self.content.find(eTag, pos1, self.end)
     if pos2 - pos1 == 0:
         self.weibomsg['rcc'] = '0'
         self.pos = pos2
     else:
         slug = self.content[pos1:pos2]
         slug = utility.clearSpace(slug)
         self.weibomsg['rcc'] = slug[1:-1]
         self.pos = pos2

示例#8

文件： CompanyWeiboMsg.py 项目： Windy-Ground/weibo_crawler

	def get_retweet_cc(self):
		bTag = '评论'
		eTag = '<\/a>'
		pos1 = self.content.find(bTag, self.pos, self.end)+len(bTag)
		pos2 = self.content.find(eTag, pos1, self.end)
		if pos2-pos1==0:
			self.weibomsg['rcc'] = '0'
			self.pos = pos2
		else:
			slug = self.content[pos1:pos2]
			slug = utility.clearSpace(slug);
			self.weibomsg['rcc'] = slug[1:-1];
			self.pos = pos2