基于PyQt5制作Excel文件数据去重小工具

from PyQt5.QtWidgets import *
from PyQt5.QtGui import *

from PyQt5.QtCore import *

from QCandyUi import CandyWindow

# w = CandyWindow.createWindow(EDataDel(), theme='blueGreen', title='Excel批数据去重器  公众号：[python 集中营]',
#                                  ico_path='数据去重.ico')
#     w.show()

import sys
import os

import pandas as pd
import openpyxl as pxl

class EDataDel(QWidget):
    def __init__(self):
        super(EDataDel, self).__init__()
        self.init_ui()

    def init_ui(self):
        self.brower = QTextBrowser()
        self.brower.setReadOnly(True)
        self.brower.setFont(QFont('微软雅黑', 8))
        self.brower.setPlaceholderText('处理进程展示区域...')
        self.brower.ensureCursorVisible()

        fORM = QFormLayout()
        self.file_paths = QLineEdit()
        self.file_paths.setReadOnly(True)

        self.file_paths_btn = QPushButton()
        self.file_paths_btn.setText('加载批文件')
        self.file_paths_btn.clicked.connect(self.file_paths_btn_click)

        self.colums_label = QLabel()
        self.colums_label.setText('自定义去重复列')

        self.colums_text = QLineEdit()
        self.colums_text.setPlaceholderText('列名1,列名2,列名3,...')


        form.addRow(self.file_paths, self.file_paths_btn)
        form.addRow(self.colums_label, self.colums_text)

        self.work = DataWork(self)
        self.work.trigger.connect(self.update_log)
        self.work.finished.connect(self.finished)

        vbox = QVBoxLayout()
        self.start_btn = QPushButton()
        self.start_btn.setText('开始执行')
        self.start_btn.clicked.connect(self.start_btn_click)

        vbox.addLayout(form)
        vbox.addWidget(self.start_btn)

        hbox = QHBoxLayout()
        hbox.addWidget(self.brower)
        hbox.addLayout(vbox)

        self.setLayout(hbox)

    def file_paths_btn_click(self):
        paths = QFileDialog.getOpenFileNames(self, '选择文件', os.getcwd(), 'Excel Files(*.xlsx)')
        files = paths[0]
        path_strs = ''
        for file in files:
            path_strs = path_strs + file + ';'
        self.file_paths.setText(path_strs)
        if self.file_paths.text().strip() != '':
            self.update_log('已经完成批文件路径加载！')
        else:
            self.update_log('没有选择任何文件！')

    def save_dir_btn_click(self):
        directory = QFileDialog.getExistingDirectory(self, '选择文件夹', os.getcwd())
        self.save_dir.setText(directory)

    def update_log(self, text):
        cursor = self.brower.textCursor()
        cursor.movePosition(QTextCursor.End)
        self.brower.append(text)
        self.brower.setTextCursor(cursor)
        self.brower.ensureCursorVisible()

    def start_btn_click(self):
        self.start_btn.setEnabled(False)
        self.work.start()

    def finished(self, finished):
        if finished is True:
            self.start_btn.setEnabled(True)

class DataWork(QThread):
    trigger = pyqtSignal(str)
    finished = pyqtSignal(bool)

    def __init__(self, parent=None):
        super(DataWork, self).__init__(parent)
        self.parent = parent
        self.working = True

    def __del__(self):
        self.working = False
        self.wait()

    def run(self):
        self.trigger.emit('启动批量处理子线程...')
        file_paths = self.parent.file_paths.text().strip()
        colums_text = self.parent.colums_text.text().strip()
        colums = []
        if ',' in colums_text:
            colums = colums_text.split(',')
        else:
            colums.append(colums_text)
        self.trigger.emit('获取配置项完成！')
        for file in file_paths.split(';'):
            if file.strip() != '':
                WEB_sheet = pxl.load_workbook(file)
                sheets = web_sheet.sheetnames
                print(file)
                new_file = file.split('.')[0] + '_已去重.' + file.split('.')[1]
                print(new_file)
                writer = pd.ExcelWriter(new_file)
                for sheet in sheets:
                    sheet_name = sheet.title()
                    print(sheet_name)
                    self.trigger.emit('准备处理工作表名称：' + str(sheet.title()))
                    data_frame = pd.read_excel(file, sheet_name=sheet_name)
                    print(data_frame)
                    repe = data_frame.duplicated(subset=colums)
                    repe = repe[repe]
                    print(data_frame.iloc[repe.index])
                    res = data_frame.drop_duplicates(subset=colums)
                    print(res)

                    self.trigger.emit(str(sheet.title()) + '：已清除')
                    res.to_excel(writer, sheet_name, index=False)
                writer.save()
            else:
                self.trigger.emit('当前文件路径为空，继续...')
        self.trigger.emit('数据处理完成...')
        self.finished.emit(True)

if __name__ == '__main__':
    app = QApplication(sys.argv)
    w = CandyWindow.createWindow(EDataDel(), theme='blueGreen', title='Excel批数据去重器  公众号：[Python 集中营]',
                                 ico_path='数据去重.ico')
    w.show()
    sys.exit(app.exec_())