Qt5 B站视频数据下载

import csv import sys, time, random, requests from lxml import etree import pandas as pd from run import Ui_Form # 继承run.py from PyQt5.QtWidgets import * from PyQt5 import QtCore, QtGui, QtWidgets from loguru import logger class MyPyQT_Form(QWidget, Ui_Form): def __init__(self): super(MyPyQT_Form, self).__init__() self.setupUi(self) def pushButton_click(self): self.textEdit_2.setText("下载中......") self.keyword = self.lineEdit.text() # 关键字 self.page = int(self.lineEdit_2.text()) # 页数 # result = pd.DataFrame() data_list = [] for i in range(1, self.page + 1): time.sleep(random.random() + 0.5) headers = { 'User-Agent': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_14_6) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/83.0.4103.116 Safari/537.36'} url = 'https://search.bilibili.com/all?keyword={0}&from_source=nav_suggest_new0&page={1}'.format( self.keyword, self.page) html = requests.get(url.format(i), headers=headers) bs = etree.HTML(html.text) items = bs.xpath('//li[@class = "video-item matrix"]') for item in items: title = item.xpath('div[@class = "info"]/div/a/@title')[0] click = item.xpath('div[@class = "info"]/div[3]/span[1]/text()')[0].strip('\n ').replace("万", "") danmu = item.xpath('div[@class = "info"]/div[3]/span[2]/text()')[0].strip('\n ') date = item.xpath('div[@class = "info"]/div[3]/span[3]/text()')[0].strip('\n ') up = item.xpath('div[@class = "info"]/div[3]/span[4]/a/text()')[0].strip('\n ') data_list.append([title, click, danmu, date, up]) data = {'标题': title, '播放量(万)': click, '弹幕': danmu, '日期': date, 'UP主': up} file_name = '{0}/{1}视频数据.csv'.format(self.download_path, self.keyword) logger.info('file_name: ', file_name) logger.info(title, click, danmu, date, up) with open(file_name, 'a+', encoding='utf_8_sig', newline='') as fp: fieldnames = ['标题', '播放量(万)', '弹幕', '日期', 'UP主'] writer = csv.DictWriter(fp, fieldnames=fieldnames) writer.writerow(data) self.textEdit_2.append("第{0}页下载完成".format(i)) self.textEdit_2.append("保存路径:{}".format(self.download_path)) self.textEdit_2.append("文件名称:{}".format('{0}共{1}个视频数据.csv'.format(self.keyword, self.page * 20))) # 创建Pandas DataFrame df = pd.DataFrame(data_list, columns=['标题', '播放量(万)', '弹幕', '日期', 'UP主']) # 将DataFrame保存为Excel文件 df.to_excel('{0}/{1}视频数据.xlsx'.format(self.download_path, self.keyword), index=False) def setBrowerPath(self): self.download_path = QFileDialog.getExistingDirectory(self) self.textEdit.setText(self.download_path) if __name__ == '__main__': app = QApplication(sys.argv) my_pyqt_form = MyPyQT_Form() my_pyqt_form.show() sys.exit(app.exec_())