pyqt4制作简单爬虫

来源:互联网 发布:网络教育的意义 编辑:程序博客网 时间:2024/05/17 13:05
# -*- coding: utf-8 -*-import urllibimport reimport sysfrom urllib import requestfrom PyQt4 import QtGui,QtCoreclass InputDialog(QtGui.QWidget):    def __init__(self,parent=None):        QtGui.QWidget.__init__(self)        self.setGeometry(250,250,300,80)        self.setWindowTitle('网页爬虫 VER1.0')        self.button = QtGui.QPushButton('开始',self)        self.button.setFocusPolicy(QtCore.Qt.NoFocus)        self.button.move(20,20)        self.connect(self.button, QtCore.SIGNAL('clicked()'), self.Inputurl)        self.setFocus()                self.label = QtGui.QLineEdit(self)        self.label.move(130,22)    def Inputurl(self):        url,ok = QtGui.QInputDialog.getText(self,'version 1.0','输入网址')                if ok:            self.url = url            mydata = self.gethtml()    def gethtml(self):        text = urllib.request.urlopen(self.url)        data = str(text.read())        self.data = data        self.getdata()    def getdata(self):        reg = r'src="(.*?\.jpg)"'        tar = re.compile(reg)        imglist = re.findall(tar,self.data)        print(imglist)        x = 1        for imgurl in imglist:            urllib.request.urlretrieve(imgurl,'%s.jpg' %x)            print("第%s张图片下载完成" %x)            x += 1        print('下载完成')app = QtGui.QApplication(sys.argv)icon = InputDialog()icon.show()sys.exit(app.exec_())

0 0