def parseMarket(): from lxml import etree from lxml.html import parse page = parse('http://www.sse.com.cn/sseportal/ps/zhs/home.html').getroot() result = etree.tostring(page) print result import io with io.open('test.xml','wb') as f: f.writelines(result) r = page.xpath('//table[@class="border_gray_right border_gray_bottom border_gray_left right_gk_bj"]'); #print len(r) j = 0 ; from stockmarket import stockmarket stockmarket = stockmarket() for a in r: tree= etree.ElementTree(a) #print etree.tostring(tree) datas = tree.xpath('//td[@align="right"]') #print len(datas) for data in datas: dataTree = etree.ElementTree(data); #print etree.tostring(dataTree) values = dataTree.xpath('//text()') #print values #temp = str(values[0]).strip() temp = values[0].strip() #print temp length = len(temp) index = length for i in range(length): #print temp[i] if (temp[i].isalnum() or temp[i] =='.'): pass else: index = i if (index == len(temp)): break; #print index temp = temp[0:index] #print temp+'****'+str(j) if (j == 0): stockmarket.ssgs = int(temp); elif (j == 1): stockmarket.sszq = int(temp); elif (j == 2): stockmarket.ssgp = int(temp); elif (j == 3): stockmarket.totalShares = float(temp); elif (j == 4): stockmarket.totalFloatingShares = float(temp); elif (j == 5): stockmarket.totalCap = float(temp); elif (j == 6): stockmarket.totalMarketCap = float(temp); elif (j == 7): stockmarket.marketPe = float(temp); j+= 1; print stockmarket
def parseMarket(): page = parse('http://www.sse.com.cn/sseportal/ps/zhs/home.html').getroot() result = etree.tostring(page) print result import io with io.open('test.xml', 'wb') as f: f.writelines(result) r = page.xpath( '//table[@class="border_gray_right border_gray_bottom border_gray_left right_gk_bj"]' ) #print len(r) j = 0 from stockmarket import stockmarket stockmarket = stockmarket() for a in r: tree = etree.ElementTree(a) #print etree.tostring(tree) datas = tree.xpath('//td[@align="right"]') #print len(datas) for data in datas: dataTree = etree.ElementTree(data) #print etree.tostring(dataTree) values = dataTree.xpath('//text()') #print values #temp = str(values[0]).strip() temp = values[0].strip() #print temp length = len(temp) index = length for i in range(length): #print temp[i] if (temp[i].isalnum() or temp[i] == '.'): pass else: index = i if (index == len(temp)): break #print index temp = temp[0:index] #print temp+'****'+str(j) if (j == 0): stockmarket.ssgs = int(temp) elif (j == 1): stockmarket.sszq = int(temp) elif (j == 2): stockmarket.ssgp = int(temp) elif (j == 3): stockmarket.totalShares = float(temp) elif (j == 4): stockmarket.totalFloatingShares = float(temp) elif (j == 5): stockmarket.totalCap = float(temp) elif (j == 6): stockmarket.totalMarketCap = float(temp) elif (j == 7): stockmarket.marketPe = float(temp) j += 1 print stockmarket
def parseSzMarket(): from lxml import etree from lxml.html import parse page = parse('http://www.szse.cn/').getroot() result = etree.tostring(page) print result import io with io.open('test.xml', 'wb') as f: f.writelines(result) r = page.xpath('//div[@class="agora"]') print len(r) j = 0 from stockmarket import stockmarket stockmarket = stockmarket() tree = etree.ElementTree(r[0]) print etree.tostring(tree) datas = tree.xpath('//span/text()') print datas stockmarket.totalCap = float(datas[3]) print stockmarket
def parseSzMarket(): from lxml import etree from lxml.html import parse page = parse('http://www.szse.cn/').getroot() result = etree.tostring(page) print result import io with io.open('test.xml','wb') as f: f.writelines(result) r = page.xpath('//div[@class="agora"]'); print len(r) j = 0 ; from stockmarket import stockmarket stockmarket = stockmarket() tree= etree.ElementTree(r[0]) print etree.tostring(tree) datas = tree.xpath('//span/text()') print datas stockmarket.totalCap = float(datas[3]); print stockmarket
def parseMarket(): from lxml import etree from lxml.html import parse page = parse( "http://www.google.com.hk/finance/stockscreener?gl=cn#c0=QuoteLast&min0=3&max0=4®ion=cn§or=AllSectors&exchange=SHA_A&sort=&sortOrder=" ).getroot() result = etree.tostring(page) print result import io with io.open("test.xml", "wb") as f: f.writelines(result) r = page.xpath('//table[@class="border_gray_right border_gray_bottom border_gray_left right_gk_bj"]') # print len(r) j = 0 from stockmarket import stockmarket stockmarket = stockmarket() for a in r: tree = etree.ElementTree(a) # print etree.tostring(tree) datas = tree.xpath('//td[@align="right"]') # print len(datas) for data in datas: dataTree = etree.ElementTree(data) # print etree.tostring(dataTree) values = dataTree.xpath("//text()") # print values # temp = str(values[0]).strip() temp = values[0].strip() # print temp length = len(temp) index = length for i in range(length): # print temp[i] if temp[i].isalnum() or temp[i] == ".": pass else: index = i if index == len(temp): break # print index temp = temp[0:index] # print temp+'****'+str(j) if j == 0: stockmarket.ssgs = int(temp) elif j == 1: stockmarket.sszq = int(temp) elif j == 2: stockmarket.ssgp = int(temp) elif j == 3: stockmarket.totalShares = float(temp) elif j == 4: stockmarket.totalFloatingShares = float(temp) elif j == 5: stockmarket.totalCap = float(temp) elif j == 6: stockmarket.totalMarketCap = float(temp) elif j == 7: stockmarket.marketPe = float(temp) j += 1 print stockmarket
def parseMarket(): from lxml import etree from lxml.html import parse page = parse( 'http://www.google.com.hk/finance/stockscreener?gl=cn#c0=QuoteLast&min0=3&max0=4®ion=cn§or=AllSectors&exchange=SHA_A&sort=&sortOrder=' ).getroot() result = etree.tostring(page) print result import io with io.open('test.xml', 'wb') as f: f.writelines(result) r = page.xpath( '//table[@class="border_gray_right border_gray_bottom border_gray_left right_gk_bj"]' ) #print len(r) j = 0 from stockmarket import stockmarket stockmarket = stockmarket() for a in r: tree = etree.ElementTree(a) #print etree.tostring(tree) datas = tree.xpath('//td[@align="right"]') #print len(datas) for data in datas: dataTree = etree.ElementTree(data) #print etree.tostring(dataTree) values = dataTree.xpath('//text()') #print values #temp = str(values[0]).strip() temp = values[0].strip() #print temp length = len(temp) index = length for i in range(length): #print temp[i] if (temp[i].isalnum() or temp[i] == '.'): pass else: index = i if (index == len(temp)): break #print index temp = temp[0:index] #print temp+'****'+str(j) if (j == 0): stockmarket.ssgs = int(temp) elif (j == 1): stockmarket.sszq = int(temp) elif (j == 2): stockmarket.ssgp = int(temp) elif (j == 3): stockmarket.totalShares = float(temp) elif (j == 4): stockmarket.totalFloatingShares = float(temp) elif (j == 5): stockmarket.totalCap = float(temp) elif (j == 6): stockmarket.totalMarketCap = float(temp) elif (j == 7): stockmarket.marketPe = float(temp) j += 1 print stockmarket