pyqt4制作简单爬虫

最新推荐文章于 2023-02-01 23:32:58 发布

scarymonster

最新推荐文章于 2023-02-01 23:32:58 发布

阅读量1.6k

点赞数

分类专栏： python 文章标签：爬虫

本文链接：https://blog.csdn.net/u013505084/article/details/18671943

版权

python 专栏收录该内容

2 篇文章 0 订阅

订阅专栏

# -*- coding: utf-8 -*-
import urllib
import re
import sys
from urllib import request
from PyQt4 import QtGui,QtCore

class 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_())