class mainWindow(QtGui.QWidget):
def __init__(self, width=600, height=800):
super(mainWindow, self).__init__()
self.width = width
self.height = height
self.setGeometry(10, 10, self.width, self.height)
self.initial()
def initial(self):
self.MainHLayout = QtGui.QHBoxLayout()
self.WebView = WebView()
self.WebView.show()
self.WebView.load(QtCore.QUrl("http://127.0.0.1/tpcdebt/test.php"))
self.MainHLayout.addWidget(self.WebView)
self.WebView.loadFinished.connect(self.loadPageOK)
self.WebView.page().setLinkDelegationPolicy(
QtWebKit.QWebPage.DelegateAllLinks)
self.setLayout(self.MainHLayout)
self.show()
def loadPageOK(self, isFinished):
if isFinished:
print("load OK")
pass
class mainWindow(QtGui.QWidget):
def __init__(self, width=600, height=800):
super(mainWindow, self).__init__()
self.width = width
self.height = height
self.setGeometry(10, 10, self.width, self.height)
self.initial()
def initial(self):
self.MainHLayout = QtGui.QHBoxLayout()
self.WebView = WebView()
self.WebView.show()
self.WebView.load(QtCore.QUrl("http://127.0.0.1/tpcdebt/test.php"))
self.MainHLayout.addWidget(self.WebView)
self.WebView.loadFinished.connect(self.loadPageOK)
self.WebView.page().setLinkDelegationPolicy(QtWebKit.QWebPage.DelegateAllLinks)
self.setLayout(self.MainHLayout)
self.show()
def loadPageOK(self, isFinished):
if isFinished:
print("load OK")
pass
class Connection(QtCore.QObject):
closeConnection = QtCore.Signal()
def __init__(self, TcpSocket=None):
if TcpSocket == None:
raise ValueError
self.stop = False
self.webview = WebView()
self.webview.load("http://127.0.0.1")
self.TcpSocket = TcpSocket
def widget(self):
return self.webview
@QtCore.Slot()
def forceStop(self):
self.stop = True
self.closeConnection.emit(self.webview)
self.TcpSocket.abort()
self.TcpSocket.deleteLater()
# self.webview.deleteLater()
self.deleteLater()
self.thread().forceStop()
@QtCore.Slot()
def service(self):
print(self.thread().currentThreadId())
while not self.stop:
command = self.TcpSocket.read(4096)
if command["type"] == "loadUrl":
self.webview.load(cmd["url"])
data = QtCore.QByteArray()
data = "load finished"
self.TcpSocket.write(data)
self.TcpSocket.flush()
data.deleteLater()
elif cmd["type"] == "close":
self.forceStop()
else:
print("unknow command")
class Connection(QtCore.QObject) :
closeConnection = QtCore.Signal()
def __init__(self, TcpSocket = None) :
if TcpSocket == None :
raise ValueError
self.stop = False
self.webview = WebView()
self.webview.load("http://127.0.0.1")
self.TcpSocket = TcpSocket
def widget(self) :
return self.webview
@QtCore.Slot()
def forceStop(self) :
self.stop = True
self.closeConnection.emit(self.webview)
self.TcpSocket.abort()
self.TcpSocket.deleteLater()
# self.webview.deleteLater()
self.deleteLater()
self.thread().forceStop()
@QtCore.Slot()
def service(self) :
print(self.thread().currentThreadId())
while not self.stop :
command = self.TcpSocket.read(4096)
if command["type"] == "loadUrl" :
self.webview.load(cmd["url"])
data = QtCore.QByteArray()
data = "load finished"
self.TcpSocket.write(data)
self.TcpSocket.flush()
data.deleteLater()
elif cmd["type"] == "close" :
self.forceStop()
else :
print("unknow command")
class Crawler(QtGui.QWidget):
signalClickedToLoadUrl = QtCore.Signal(QtCore.QUrl)
def __init__(self, width = 800, height = 600):
super(Crawler, self).__init__()
self.setGeometry(10, 10, width, height)
self.setupUI()
def setupUI(self) :
self.mainHBoxLayout = QtGui.QHBoxLayout()
self.mainRVBoxLayout = QtGui.QVBoxLayout()
self.mainLVBoxLayout = QtGui.QVBoxLayout()
self.executeCrawler = QtGui.QPushButton("Get Content!!!")
self.urlBarLineEdit = QtGui.QLineEdit("http://www2.tpa.edu.tw/tpaedu/Home/login.asp")
self.myWebView = WebView()
self.myTabWidget = TabWidget()
#self.myWebView.settings().setDefaultTextEncoding("big5")
self.myWebView.show()
self.myTabWidget.addTab(self.myWebView, "my Web View")
self.mainLVBoxLayout.addWidget(self.urlBarLineEdit)
self.mainLVBoxLayout.addWidget(self.myTabWidget)
self.mainRVBoxLayout.addWidget(self.executeCrawler)
self.mainHBoxLayout.addLayout(self.mainLVBoxLayout)
self.mainHBoxLayout.addLayout(self.mainRVBoxLayout)
self.setLayout(self.mainHBoxLayout)
self.executeCrawler.clicked.connect(self.slotClickedToLoadUrl)
self.signalClickedToLoadUrl.connect(self.slotLoadUrl)
self.myWebView.loadFinished.connect(self.slotWebViewLoadFinished)
self.show()
@QtCore.Slot()
def slotClickedToLoadUrl(self) :
print(self.urlBarLineEdit.text())
self.signalClickedToLoadUrl.emit(QtCore.QUrl(self.urlBarLineEdit.text()))
@QtCore.Slot(QtCore.QUrl)
def slotLoadUrl(self, qUrl) :
self.myWebView.load(qUrl)
@QtCore.Slot(bool)
def slotWebViewLoadFinished(self, isFinished) :
if isFinished :
self.urlBarLineEdit.setText(self.myWebView.url().toString())
pageFrames = self.myWebView.page().mainFrame().childFrames()
print("base Url := " + self.myWebView.page().mainFrame().baseUrl().toString())
if self.myWebView.page().mainFrame().baseUrl().toString() == "http://www2.tpa.edu.tw/tpaedu/default.asp" :
print("it is tpa local net")
for frame in pageFrames :
print(frame.frameName())
#print(frame.toHtml())
elements = frame.findAllElements("td")
print(elements.count())
for index in range(elements.count()) :
"""print("elements" + str(index))
print(elements.at(index + 1).tagName())
print(elements.at(index + 1).firstChild().tagName())
print(elements.at(index + 1).firstChild().nextSibling().tagName())
print(elements.at(index + 1).firstChild().nextSibling().nextSibling().tagName())
print(elements.at(index + 1).firstChild().nextSibling().nextSibling().nextSibling().tagName())
print(str(type(elements.at(index + 1).firstChild().nextSibling().nextSibling().nextSibling().nextSibling())))"""
findAllElementsChild(elements.at(index + 1).firstChild())
if frame.frameName() == "logo" :
pass
elif frame.frameName() == "menu" :
# get menu for javascript robot action
pass
elif frame.frameName() == "main" :
# get main content to crawler data
pass
elif self.myWebView.page().mainFrame().baseUrl().toString() == "http://www2.tpa.edu.tw/tpaedu/Home/login.asp" :
pass
pass
class Crawler(QtGui.QWidget):
signalClickedToLoadUrl = QtCore.Signal(QtCore.QUrl)
def __init__(self, width=800, height=600):
super(Crawler, self).__init__()
self.setGeometry(10, 10, width, height)
self.setupUI()
def setupUI(self):
self.mainHBoxLayout = QtGui.QHBoxLayout()
self.mainRVBoxLayout = QtGui.QVBoxLayout()
self.mainLVBoxLayout = QtGui.QVBoxLayout()
self.executeCrawler = QtGui.QPushButton("Get Content!!!")
self.urlBarLineEdit = QtGui.QLineEdit(
"http://www2.tpa.edu.tw/tpaedu/Home/login.asp")
self.myWebView = WebView()
self.myTabWidget = TabWidget()
#self.myWebView.settings().setDefaultTextEncoding("big5")
self.myWebView.show()
self.myTabWidget.addTab(self.myWebView, "my Web View")
self.mainLVBoxLayout.addWidget(self.urlBarLineEdit)
self.mainLVBoxLayout.addWidget(self.myTabWidget)
self.mainRVBoxLayout.addWidget(self.executeCrawler)
self.mainHBoxLayout.addLayout(self.mainLVBoxLayout)
self.mainHBoxLayout.addLayout(self.mainRVBoxLayout)
self.setLayout(self.mainHBoxLayout)
self.executeCrawler.clicked.connect(self.slotClickedToLoadUrl)
self.signalClickedToLoadUrl.connect(self.slotLoadUrl)
self.myWebView.loadFinished.connect(self.slotWebViewLoadFinished)
self.show()
@QtCore.Slot()
def slotClickedToLoadUrl(self):
print(self.urlBarLineEdit.text())
self.signalClickedToLoadUrl.emit(
QtCore.QUrl(self.urlBarLineEdit.text()))
@QtCore.Slot(QtCore.QUrl)
def slotLoadUrl(self, qUrl):
self.myWebView.load(qUrl)
@QtCore.Slot(bool)
def slotWebViewLoadFinished(self, isFinished):
if isFinished:
self.urlBarLineEdit.setText(self.myWebView.url().toString())
pageFrames = self.myWebView.page().mainFrame().childFrames()
print("base Url := " +
self.myWebView.page().mainFrame().baseUrl().toString())
if self.myWebView.page().mainFrame().baseUrl().toString(
) == "http://www2.tpa.edu.tw/tpaedu/default.asp":
print("it is tpa local net")
for frame in pageFrames:
print(frame.frameName())
#print(frame.toHtml())
elements = frame.findAllElements("td")
print(elements.count())
for index in range(elements.count()):
"""print("elements" + str(index))
print(elements.at(index + 1).tagName())
print(elements.at(index + 1).firstChild().tagName())
print(elements.at(index + 1).firstChild().nextSibling().tagName())
print(elements.at(index + 1).firstChild().nextSibling().nextSibling().tagName())
print(elements.at(index + 1).firstChild().nextSibling().nextSibling().nextSibling().tagName())
print(str(type(elements.at(index + 1).firstChild().nextSibling().nextSibling().nextSibling().nextSibling())))"""
findAllElementsChild(
elements.at(index + 1).firstChild())
if frame.frameName() == "logo":
pass
elif frame.frameName() == "menu":
# get menu for javascript robot action
pass
elif frame.frameName() == "main":
# get main content to crawler data
pass
elif self.myWebView.page().mainFrame().baseUrl().toString(
) == "http://www2.tpa.edu.tw/tpaedu/Home/login.asp":
pass
pass