|
| 1 | +# -*- coding: utf-8 -*- |
| 2 | + |
| 3 | +# Form implementation generated from reading ui file 'H:\Github\PythonScripts\PyQtDesigner\WebScrapper\WebScrapper.ui' |
| 4 | +# |
| 5 | +# Created by: PyQt5 UI code generator 5.10.1 |
| 6 | +# |
| 7 | +# WARNING! All changes made in this file will be lost! |
| 8 | + |
| 9 | +from PyQt5 import QtCore, QtGui, QtWidgets |
| 10 | +from PyQt5.QtWidgets import QSystemTrayIcon, QAction |
| 11 | +from PyQt5.QtGui import QIcon |
| 12 | +from datetime import datetime |
| 13 | +from requests_html import HTMLSession |
| 14 | +import requests.exceptions |
| 15 | +import os |
| 16 | +import sys |
| 17 | +import multiprocessing |
| 18 | + |
| 19 | +class Ui_MainWindow(object): |
| 20 | + def setupUi(self, MainWindow): |
| 21 | + MainWindow.setObjectName("MainWindow") |
| 22 | + MainWindow.resize(1366, 768) |
| 23 | + MainWindow.setWindowTitle('WebScrapper by [@Nishant Ghanate]') |
| 24 | + self.scriptDir = os.path.dirname(os.path.realpath(__file__)) |
| 25 | + MainWindow.setWindowIcon(QIcon(self.scriptDir + os.path.sep + 'icon.png')) |
| 26 | + |
| 27 | + # Gui.QSystemTrayIcon |
| 28 | + MainWindow.setAutoFillBackground(False) |
| 29 | + # MainWindow.setMaximumWidth(1366) |
| 30 | + # MainWindow.setMaximumHeight(768) |
| 31 | + MainWindow.setStyleSheet("QMainWindow#MainWindow\n" |
| 32 | +"{\n" |
| 33 | +" /*background-color:rgb(195,223,255);*/\n" |
| 34 | +" background-image: url(\'H:/Github/PythonScripts/PyQtDesigner/WebScrapper/bg.png\');\n" |
| 35 | +" background-position: center; /* Center the image */\n" |
| 36 | +" background-repeat: no-repeat;\n" |
| 37 | +" " |
| 38 | +" \n" |
| 39 | +" \n" |
| 40 | +"}\n" |
| 41 | +"\n" |
| 42 | +"\n" |
| 43 | +"QWidget#centralwidget\n" |
| 44 | +"{\n" |
| 45 | +"\n" |
| 46 | +"}\n" |
| 47 | +"\n" |
| 48 | +"QLabel\n" |
| 49 | +"{ \n" |
| 50 | +" color:rgb(255,255,255);\n" |
| 51 | +"}\n" |
| 52 | +"\n" |
| 53 | +"QRadioButton\n" |
| 54 | +"{\n" |
| 55 | +" color:rgb(255,255,255);\n" |
| 56 | +"}\n" |
| 57 | +"\n" |
| 58 | +"\n" |
| 59 | +"") |
| 60 | + # self.setWindowIcon(QIcon(self.scriptDir + os.path.sep + 'icon.png')) |
| 61 | + self.centralwidget = QtWidgets.QWidget(MainWindow) |
| 62 | + self.centralwidget.setStyleSheet("") |
| 63 | + self.centralwidget.setObjectName("centralwidget") |
| 64 | + self.listWidgetMain = QtWidgets.QListWidget(self.centralwidget) |
| 65 | + self.listWidgetMain.setGeometry(QtCore.QRect(30, 60, 481, 400)) |
| 66 | + font = QtGui.QFont() |
| 67 | + font.setPointSize(10) |
| 68 | + self.listWidgetMain.setFont(font) |
| 69 | + self.listWidgetMain.setObjectName("listWidgetMain") |
| 70 | + self.listWidgetLogs = QtWidgets.QListWidget(self.centralwidget) |
| 71 | + self.listWidgetLogs.setGeometry(QtCore.QRect(860, 60, 461, 541)) |
| 72 | + font = QtGui.QFont() |
| 73 | + font.setPointSize(10) |
| 74 | + self.listWidgetLogs.setFont(font) |
| 75 | + self.listWidgetLogs.setObjectName("listWidgetLogs") |
| 76 | + self.buttonGetInput = QtWidgets.QPushButton(self.centralwidget) |
| 77 | + self.buttonGetInput.setGeometry(QtCore.QRect(40, 630, 80, 30)) |
| 78 | + font = QtGui.QFont() |
| 79 | + font.setPointSize(10) |
| 80 | + self.buttonGetInput.setFont(font) |
| 81 | + self.buttonGetInput.setObjectName("buttonGetInput") |
| 82 | + self.textEditUrl = QtWidgets.QTextEdit(self.centralwidget) |
| 83 | + self.textEditUrl.setGeometry(QtCore.QRect(30, 490, 481, 31)) |
| 84 | + font = QtGui.QFont() |
| 85 | + font.setPointSize(12) |
| 86 | + self.textEditUrl.setFont(font) |
| 87 | + self.textEditUrl.setObjectName("textEditUrl") |
| 88 | + self.textEditSource = QtWidgets.QTextEdit(self.centralwidget) |
| 89 | + self.textEditSource.setGeometry(QtCore.QRect(30, 570, 481, 31)) |
| 90 | + font = QtGui.QFont() |
| 91 | + font.setPointSize(12) |
| 92 | + self.textEditSource.setFont(font) |
| 93 | + self.textEditSource.setObjectName("textEditSource") |
| 94 | + self.buttonClearLogs = QtWidgets.QPushButton(self.centralwidget) |
| 95 | + self.buttonClearLogs.setGeometry(QtCore.QRect(980, 630, 90, 30)) |
| 96 | + font = QtGui.QFont() |
| 97 | + font.setPointSize(10) |
| 98 | + self.buttonClearLogs.setFont(font) |
| 99 | + self.buttonClearLogs.setObjectName("buttonClearLogs") |
| 100 | + self.buttonSaveLogs = QtWidgets.QPushButton(self.centralwidget) |
| 101 | + self.buttonSaveLogs.setGeometry(QtCore.QRect(860, 630, 90, 30)) |
| 102 | + font = QtGui.QFont() |
| 103 | + font.setPointSize(10) |
| 104 | + self.buttonSaveLogs.setFont(font) |
| 105 | + self.buttonSaveLogs.setObjectName("buttonSaveLogs") |
| 106 | + self.radioButtonFirst = QtWidgets.QRadioButton(self.centralwidget) |
| 107 | + self.radioButtonFirst.setGeometry(QtCore.QRect(120, 540, 82, 17)) |
| 108 | + font = QtGui.QFont() |
| 109 | + font.setPointSize(10) |
| 110 | + self.radioButtonFirst.setFont(font) |
| 111 | + self.radioButtonFirst.setObjectName("radioButtonFirst") |
| 112 | + self.buttonStop = QtWidgets.QPushButton(self.centralwidget) |
| 113 | + self.buttonStop.setGeometry(QtCore.QRect(150, 630, 80, 30)) |
| 114 | + font = QtGui.QFont() |
| 115 | + font.setPointSize(10) |
| 116 | + self.buttonStop.setFont(font) |
| 117 | + self.buttonStop.setObjectName("buttonStop") |
| 118 | + self.radioButtonAll = QtWidgets.QRadioButton(self.centralwidget) |
| 119 | + self.radioButtonAll.setGeometry(QtCore.QRect(230, 540, 71, 20)) |
| 120 | + font = QtGui.QFont() |
| 121 | + font.setPointSize(10) |
| 122 | + self.radioButtonAll.setFont(font) |
| 123 | + self.radioButtonAll.setObjectName("radioButtonAll") |
| 124 | + self.labelMain = QtWidgets.QLabel(self.centralwidget) |
| 125 | + self.labelMain.setGeometry(QtCore.QRect(170, 20, 200, 31)) |
| 126 | + font = QtGui.QFont() |
| 127 | + font.setPointSize(12) |
| 128 | + self.labelMain.setFont(font) |
| 129 | + self.labelMain.setAutoFillBackground(False) |
| 130 | + self.labelMain.setFrameShape(QtWidgets.QFrame.StyledPanel) |
| 131 | + self.labelMain.setFrameShadow(QtWidgets.QFrame.Plain) |
| 132 | + self.labelMain.setAlignment(QtCore.Qt.AlignCenter) |
| 133 | + self.labelMain.setObjectName("labelMain") |
| 134 | + self.labelLog = QtWidgets.QLabel(self.centralwidget) |
| 135 | + self.labelLog.setGeometry(QtCore.QRect(1000, 20, 200, 31)) |
| 136 | + font = QtGui.QFont() |
| 137 | + font.setPointSize(12) |
| 138 | + self.labelLog.setFont(font) |
| 139 | + self.labelLog.setFrameShape(QtWidgets.QFrame.StyledPanel) |
| 140 | + self.labelLog.setFrameShadow(QtWidgets.QFrame.Sunken) |
| 141 | + self.labelLog.setAlignment(QtCore.Qt.AlignCenter) |
| 142 | + self.labelLog.setObjectName("labelLog") |
| 143 | + self.label = QtWidgets.QLabel(self.centralwidget) |
| 144 | + self.label.setGeometry(QtCore.QRect(30, 470, 91, 21)) |
| 145 | + font = QtGui.QFont() |
| 146 | + font.setPointSize(10) |
| 147 | + self.label.setFont(font) |
| 148 | + self.label.setObjectName("label") |
| 149 | + self.label_2 = QtWidgets.QLabel(self.centralwidget) |
| 150 | + self.label_2.setGeometry(QtCore.QRect(30, 540, 71, 31)) |
| 151 | + font = QtGui.QFont() |
| 152 | + font.setPointSize(10) |
| 153 | + self.label_2.setFont(font) |
| 154 | + self.label_2.setObjectName("label_2") |
| 155 | + self.buttonClear = QtWidgets.QPushButton(self.centralwidget) |
| 156 | + self.buttonClear.setGeometry(QtCore.QRect(260, 630, 80, 30)) |
| 157 | + font = QtGui.QFont() |
| 158 | + font.setPointSize(10) |
| 159 | + self.buttonClear.setFont(font) |
| 160 | + self.buttonClear.setObjectName("buttonClear") |
| 161 | + MainWindow.setCentralWidget(self.centralwidget) |
| 162 | + |
| 163 | + # self.setFixedSize(1366, 768) |
| 164 | + self.buttonGetInput.clicked.connect(self.getInput) |
| 165 | + self.buttonSaveLogs.clicked.connect(self.saveLogs) |
| 166 | + self.buttonClearLogs.clicked.connect(self.clearLogs) |
| 167 | + self.buttonClear.clicked.connect(self.clear) |
| 168 | + self.buttonStop.clicked.connect(self.stop) |
| 169 | + self.buttonStop.setEnabled(False) |
| 170 | + self.oldUrl = "_" |
| 171 | + |
| 172 | + |
| 173 | + self.retranslateUi(MainWindow) |
| 174 | + QtCore.QMetaObject.connectSlotsByName(MainWindow) |
| 175 | + |
| 176 | + def retranslateUi(self, MainWindow): |
| 177 | + _translate = QtCore.QCoreApplication.translate |
| 178 | + MainWindow.setWindowTitle(_translate("MainWindow", "MainWindow")) |
| 179 | + self.buttonGetInput.setText(_translate("MainWindow", "Get ")) |
| 180 | + self.buttonClearLogs.setText(_translate("MainWindow", "Clear logs")) |
| 181 | + self.buttonSaveLogs.setText(_translate("MainWindow", "Save Logs")) |
| 182 | + self.radioButtonFirst.setText(_translate("MainWindow", "Firs page")) |
| 183 | + self.buttonStop.setText(_translate("MainWindow", "Stop")) |
| 184 | + self.radioButtonAll.setText(_translate("MainWindow", "All pages")) |
| 185 | + self.labelMain.setText(_translate("MainWindow", "Command Logs")) |
| 186 | + self.labelLog.setText(_translate("MainWindow", "Source logs")) |
| 187 | + self.label.setText(_translate("MainWindow", "Website url :")) |
| 188 | + self.label_2.setText(_translate("MainWindow", "Xpath :")) |
| 189 | + self.buttonClear.setText(_translate("MainWindow", "Clear")) |
| 190 | + self.Stop = True |
| 191 | + |
| 192 | + #Acttivate UI fucntions |
| 193 | + # QtCore.pyqtSlot() |
| 194 | + |
| 195 | + # Returns Time in e.g 1 oct 2018 formart |
| 196 | + def getTimeStamp(self): |
| 197 | + return datetime.now().strftime("%A %d. %B %Y %I:%M:%S %p") |
| 198 | + |
| 199 | + def stop(self): |
| 200 | + self.buttonGetInput.setEnabled(True) |
| 201 | + self.Stop = False |
| 202 | + |
| 203 | + # Validate input url and throw Expection for invalid |
| 204 | + def validateUrl(self,url): |
| 205 | + try: |
| 206 | + # Create new session only if new url |
| 207 | + if self.oldUrl != url: |
| 208 | + self.oldUrl = url |
| 209 | + session = HTMLSession() |
| 210 | + self.r = session.get(url) |
| 211 | + return True |
| 212 | + # else True because if only xpathSrc changed and continue scrapping from given url |
| 213 | + else : |
| 214 | + return True |
| 215 | + # Throw expection and close the fucntion |
| 216 | + except requests.exceptions.RequestException as e: |
| 217 | + self.listWidgetLogs.addItem(str(e)) |
| 218 | + return False |
| 219 | + |
| 220 | + # This function will scrap it is also recurive if required |
| 221 | + def getSource(self): |
| 222 | + try: |
| 223 | + # returns list for single Xpath item |
| 224 | + src = self.r.html.xpath(self.xpathSrc) |
| 225 | + # proceed further if list is not empty |
| 226 | + if src and self.Stop : |
| 227 | + timeStamp = self.getTimeStamp() |
| 228 | + for s in src: |
| 229 | + # print(s) |
| 230 | + s = str(s) |
| 231 | + # List widget only supports string |
| 232 | + self.listWidgetLogs.addItem(s) |
| 233 | + # if all pages is checked |
| 234 | + if self.radioButtonAll.isChecked(): |
| 235 | + if self.r.html._next(): |
| 236 | + print(self.r.html._next()) |
| 237 | + url = self.validateUrl(self.r.html._next()) |
| 238 | + if url and self.r.status_code == 200: |
| 239 | + # self.listWidgetLogs.addItem('\n'+self.r.html._next()) |
| 240 | + self.getSource() |
| 241 | + except : |
| 242 | + self.listWidgetLogs.addItem('invalid xpath format') |
| 243 | + |
| 244 | + |
| 245 | + # Get button function connection |
| 246 | + def getInput(self): |
| 247 | + url = self.textEditUrl.toPlainText() |
| 248 | + self.xpathSrc = self.textEditSource.toPlainText() |
| 249 | + print(self.xpathSrc) |
| 250 | + if url and self.xpathSrc.strip() : |
| 251 | + source = self.validateUrl(url) |
| 252 | + if source: |
| 253 | + timestampDay = self.getTimeStamp() |
| 254 | + self.listWidgetLogs.addItem(timestampDay) |
| 255 | + self.listWidgetMain.addItem(self.xpathSrc) |
| 256 | + self.getSource() |
| 257 | + self.buttonStop.setEnabled(True) |
| 258 | + |
| 259 | + |
| 260 | + elif url is None or len(url) < 5: |
| 261 | + timestampDay = self.getTimeStamp() |
| 262 | + self.listWidgetLogs.addItem(timestampDay) |
| 263 | + self.listWidgetLogs.addItem('Url cannot be empty ') |
| 264 | + else : |
| 265 | + timestampDay = self.getTimeStamp() |
| 266 | + self.listWidgetLogs.addItem(timestampDay) |
| 267 | + self.listWidgetLogs.addItem('xpath input where is it ?') |
| 268 | + |
| 269 | + # Saves logs from Preserved logs |
| 270 | + def saveLogs(self): |
| 271 | + timestampDay = datetime.now().strftime("%A %d %B %Y %I %M %S%p") |
| 272 | + fileName = self.scriptDir + os.path.sep +'WebScrappingLogs '+ timestampDay +'.txt' |
| 273 | + file = open(fileName,'a+' , encoding='utf-8') |
| 274 | + |
| 275 | + file.writelines('Command Logs ' + timestampDay + '\n') |
| 276 | + for i in range(self.listWidgetMain.count()): |
| 277 | + file.writelines(self.listWidgetMain.item(i).text() + '\n') |
| 278 | + |
| 279 | + file.writelines('\nSource Logs') |
| 280 | + for i in range(self.listWidgetLogs.count()): |
| 281 | + file.writelines(self.listWidgetLogs.item(i).text() + '\n') |
| 282 | + # file.writelines(itemsTextList) |
| 283 | + file.close() |
| 284 | + self.listWidgetLogs.addItem('Logs saved Sucessfully') |
| 285 | + |
| 286 | + # clear preservedLogs |
| 287 | + def clearLogs(self): |
| 288 | + self.listWidgetLogs.clear() |
| 289 | + |
| 290 | + def clear(self): |
| 291 | + self.listWidgetMain.clear() |
| 292 | + |
| 293 | + |
| 294 | + |
| 295 | + |
| 296 | +if __name__ == "__main__": |
| 297 | + |
| 298 | + app = QtWidgets.QApplication(sys.argv) |
| 299 | + |
| 300 | + |
| 301 | + MainWindow = QtWidgets.QMainWindow() |
| 302 | + ui = Ui_MainWindow() |
| 303 | + ui.setupUi(MainWindow) |
| 304 | + MainWindow.setWindowTitle('WebScrapper by [@Nishant Ghanate]') |
| 305 | + MainWindow.show() |
| 306 | + |
| 307 | + sys.exit(app.exec_()) |
| 308 | + |
0 commit comments