PyQt6实例_批量下载pdf工具_整个项目代码_终结章

目录

前置：

代码：

视频：

---

前置：

1 本系列将以 “PyQt6实例_批量下载pdf工具”开头，放在 【PyQt6实例】 专栏
2 本系列涉及到的PyQt6知识点：
线程池：QThreadPool,QRunnable；
信号与槽：pyqtSignal,pyqtSlot；
界面：QTextEdit,QLabel,QLineText,QPushButton,QMainWindow,QWidget;
布局：QHBoxLayout,QVBoxLayout;
弹框：QFileDialog,QMessageBox。
3 本系列后续会在B站录制视频，到时会在文末贴出链接。本人还是建议先看博文，不懂的再看视频，这样效率高，节约时间。

代码：

import os
import sys,json
import traceback
from datetime import datetime
from pathlib import Path
import requests
from PyQt6.QtCore import (
QSize,Qt,QRunnable,QThreadPool,pyqtSignal,pyqtSlot,QObject
)
from PyQt6.QtWidgets import(
QApplication,
QMainWindow,
QLabel,
QLineEdit,
QPushButton,
QFileDialog,
QTextEdit,
QHBoxLayout,
QVBoxLayout,
QWidget,
QMessageBox
)

basedir = os.path.dirname(__file__)

class WorkerSignals(QObject):
    finished = pyqtSignal(int)
    error = pyqtSignal(tuple)
    result = pyqtSignal(tuple)
    pass

class Worker(QRunnable):
    def __init__(self,thread_num:int,task_data:dict):
        super().__init__()
        self.thread_num = thread_num
        self.task_data = task_data
        self.signals = WorkerSignals()
        self.is_stop = False
        pass
    @pyqtSlot()
    def run(self):
        try:
            '''
            每一个线程领一部分股票，执行下载任务
            1 下载完一个股票，记录这个股票
            2 当下载某个股票，下载一个pdf，记录pdfurl
            
            断点信息 temp_dict {ticker:[pdfurl,pdfurl,...]}
            股票列表  txt_list
            txt所在目录  txt_dir
            pdf存储位置  pdf_dir
            '''
            result = None
            if self.task_data['temp_dict'] is not None:
                temp_ticker_list = list(self.task_data['temp_dict'].keys())
                pass
            else:
                temp_ticker_list = []
            txt_list = self.task_data['txt_list']
            txt_dir = self.task_data['txt_dir'] + os.path.sep
            pdf_dir = self.task_data['pdf_dir'] + os.path.sep
            executed_ticker_list = []
            for txt_file in txt_list:
                ticker = txt_file[0:6]
                txt_file_path = txt_dir + txt_file
                one_ticker_executed_url_list = []
                with open(txt_file_path,'r',encoding='utf-8') as fr:
                    res_str = fr.read()
                url_list = res_str.split('\n')
                if ticker in temp_ticker_list:
                    one_ticker_executed_url_list = self.task_data['temp_dict'][ticker]
                    for one_url in one_ticker_executed_url_list:
                        url_list.remove(one_url)
                tar_pdf_dir = pdf_dir + ticker + os.path.sep
                if not os.path.exists(tar_pdf_dir):
                    os.mkdir(tar_pdf_dir)
                for one_url in url_list:
                    if self.is_stop:
                        # 强制停止退出
                        result = (self.thread_num,'stoped',ticker,one_ticker_executed_url_list,executed_ticker_list)
                        break
                        pass
                    tar_file_name00 = one_url.split('/')
                    tar_file_name = f"{tar_file_name00[-2]}_{tar_file_name00[-1]}.pdf"
                    try:
                        res = requests.get(one_url)
                        if res.status_code == 200:
                            with open(tar_pdf_dir + tar_file_name,'wb') as fw:
                                fw.write(res.content)
                            pass
                        pass
                    except:
                        pass
                    one_ticker_executed_url_list.append(one_url)
                    pass
                if self.is_stop:
                    if result is None:
                        result = (self.thread_num,'stoped',None,None,executed_ticker_list)
                    break
                executed_ticker_list.append(ticker)
            if not self.is_stop:
                result = (self.thread_num,'success',None,None,executed_ticker_list)
            pass
        except Exception:
            traceback.print_exc()
            exctype,value = sys.exc_info()[:2]
            self.signals.error.emit((self.thread_num,exctype,value,traceback.format_exc()))
            pass
        else:
            self.signals.result.emit(result)
            pass
        finally:
            self.signals.finished.emit(self.thread_num)

    def stop_run(self):
        self.is_stop = True
        pass
    pass


class MainWindow(QMainWindow):
    def __init__(self):
        super().__init__()
        self.setWindowTitle('PDF Downloader')
        self.setMinimumSize(QSize(600,400))

        label1 = QLabel('存储pdf网址txt文件所在目录:')
        self.txtdir_lineedit = QLineEdit()
        self.choicedir_btn = QPushButton('打开目录')
        self.choicedir_btn.clicked.connect(self.choicedir_btn_clicked)
        label2 = QLabel('设置下载后pdf存储目录：')
        self.savedir_lineedit = QLineEdit()
        self.savedir_btn = QPushButton('设置目录')
        self.savedir_btn.clicked.connect(self.savedir_btn_clicked)

        self.execute_btn = QPushButton('执行')
        self.execute_btn.clicked.connect(self.execute_btn_clicked)
        self.stop_btn = QPushButton('停止')
        self.stop_btn.clicked.connect(self.stop_btn_clicked)

        label3 = QLabel('运行日志：')
        label4 = QLabel('其他信息：')
        self.excutelog_textedit = QTextEdit()
        self.excutelog_textedit.setReadOnly(True)
        self.otherlog_textedit = QTextEdit()
        self.otherlog_textedit.setReadOnly(True)

        layout0 = QHBoxLayout()
        layout0.addWidget(label1)
        layout0.addWidget(self.txtdir_lineedit)
        layout0.addWidget(self.choicedir_btn)

        layout1 = QHBoxLayout()
        layout1.addWidget(label2)
        layout1.addWidget(self.savedir_lineedit)
        layout1.addWidget(self.savedir_btn)

        layout2 = QHBoxLayout()
        layout2.addWidget(self.execute_btn)
        layout2.addWidget(self.stop_btn)

        layout3 = QVBoxLayout()
        layout3.addWidget(label3)
        layout3.addWidget(self.excutelog_textedit)

        layout4 = QVBoxLayout()
        layout4.addWidget(label4)
        layout4.addWidget(self.otherlog_textedit)

        layout5 = QHBoxLayout()
        layout5.addLayout(layout3)
        layout5.addLayout(layout4)

        layout = QVBoxLayout()
        layout.addLayout(layout0)
        layout.addLayout(layout1)
        layout.addLayout(layout2)
        layout.addLayout(layout5)

        widget = QWidget()
        widget.setLayout(layout)

        self.setCentralWidget(widget)

        self.threadpool = QThreadPool()
        self.max_thread_count = self.threadpool.maxThreadCount()

        # 断点续传
        self.open_init()
        self.pre_last_time_data = {}
        self.pre_last_time_executed_tickers = []
        self.thread_finished_count = 0
        self.waitting_close = False
        # 标记本次是否执行过
        self.this_time_start_yeah = False
        self.runner_list = []
        pass
    def open_init(self):
        '''
        params.txt   txt_dir;pdf_dir;str...
        temp_dict  temp.json
        executed.txt
        :return:
        '''
        self.create_empty_file()
        with open(os.path.join(basedir,'data','params.txt'),'r',encoding='utf-8') as fr:
            res_str = fr.read()
        res_str = res_str.strip()
        if len(res_str)>0:
            res_list = res_str.split(';')
            self.txtdir_lineedit.setText(res_list[0])
            self.savedir_lineedit.setText(res_list[1])
            self.otherlog_textedit.setText(res_list[2])
            with open(os.path.join(basedir,'data','temp.json'),'r',encoding='utf-8') as fr:
                self.last_time_data = json.load(fr)
            with open(os.path.join(basedir,'data','executed.txt'),'r',encoding='utf-8') as fr:
                tickers_str = fr.read()
            tickers_str = tickers_str.strip()
            if len(tickers_str)>0:
                self.last_time_executed_tickers = tickers_str.split('\n')
            else:
                self.last_time_executed_tickers = None
            pass
        else:
            self.last_time_data = None
            self.last_time_executed_tickers = None
        pass
    def create_empty_file(self):
        if not os.path.exists(os.path.join(basedir,'data')):
            os.mkdir(os.path.join(basedir,'data'))
        if not os.path.exists(os.path.join(basedir,'data','params.txt')):
            with open(os.path.join(basedir,'data','params.txt'),'w',encoding='utf-8') as fw:
                fw.write('')
        if not os.path.exists(os.path.join(basedir, 'data', 'temp.json')):
            with open(os.path.join(basedir, 'data', 'temp.json'), 'w', encoding='utf-8') as fw:
                json.dump({}, fw)
        if not os.path.exists(os.path.join(basedir, 'data', 'executed.txt')):
            with open(os.path.join(basedir, 'data', 'executed.txt'), 'w', encoding='utf-8') as fw:
                fw.write('')
        pass
    def choicedir_btn_clicked(self):
        dir_name = QFileDialog.getExistingDirectory(self,"打开存储pdf网址txt文件所在目录")
        if dir_name:
            self.txtdir_lineedit.setText(dir_name)
        pass
    def savedir_btn_clicked(self):
        dir_name = QFileDialog.getExistingDirectory(self, "指定下载后pdf存储的目录")
        if dir_name:
            self.savedir_lineedit.setText(dir_name)
        pass
    def execute_btn_clicked(self):
        txt_dir = self.txtdir_lineedit.text()
        if txt_dir is None or len(txt_dir.strip())<=0:
            self.information_dialog('请先选择txt所在目录')
            return
        txt_list = os.listdir(txt_dir)
        if len(txt_list)<=0:
            self.information_dialog('txt所在目录为空')
            return
        pdf_dir = self.savedir_lineedit.text()
        if pdf_dir is None or len(pdf_dir.strip())<=0:
            self.information_dialog('请设置pdf存储目录')
            return
        answer = QMessageBox.question(
            self,
            '确认启动？',
            f'如果确定启动，程序将把任务分成 {self.max_thread_count} 个线程执行。执行过程将占用设备资源。',
            QMessageBox.StandardButton.Yes | QMessageBox.StandardButton.No
        )
        if answer == QMessageBox.StandardButton.Yes:
            self.this_time_start_yeah = True
            self.thread_finished_count = 0
            self.txtdir_lineedit.setDisabled(True)
            self.savedir_lineedit.setDisabled(True)
            self.choicedir_btn.setDisabled(True)
            self.savedir_btn.setDisabled(True)
            self.execute_btn.setDisabled(True)
            if self.last_time_executed_tickers is not None:
                for one in self.last_time_executed_tickers:
                    res_txt_str = f'{one}.txt'
                    if res_txt_str in txt_list:
                        txt_list.remove(res_txt_str)
                pass
            # 分发任务
            interval = len(txt_list)//self.max_thread_count
            if interval == 0:
                self.max_thread_count = 1
                self.insert_executelog('需要执行的内容很少，只开启一个线程')
                pass
            for i in range(0,self.max_thread_count):
                if i == self.max_thread_count-1:
                    node_txt_list = txt_list[i*interval:]
                else:
                    node_txt_list = txt_list[i*interval:(i+1)*interval]
                task_data = {
                    'txt_dir':txt_dir,
                    'pdf_dir':pdf_dir,
                    'txt_list':node_txt_list,
                    'temp_dict':self.last_time_data
                }
                worker = Worker(i,task_data)
                worker.signals.result.connect(self.thread_result_fn)
                worker.signals.finished.connect(self.thread_finished_fn)
                worker.signals.error.connect(self.thread_error_fn)
                self.runner_list.append(worker)
                self.insert_otherlog(f'线程 {i} 启动。')
                self.threadpool.start(worker)
            pass
        else:
            return
        pass
    def thread_result_fn(self,res:tuple):
        # (thread_num,stoped,ticker,executed_url_list,excuted_ticker_list)
        # (thread_num,success,None,None,excuted_ticker_list)
        thread_num = res[0]
        status = res[1]
        if status == 'stoped':
            self.insert_otherlog(f'线程 {thread_num} 停止.')
            if res[2] is not None:
                self.pre_last_time_data[res[2]] = res[3]
            self.pre_last_time_executed_tickers.extend(res[4])
            pass
        else:
            self.insert_otherlog(f'线程 {thread_num} 正常结束。')
        pass
    def thread_finished_fn(self,res:int):
        self.thread_finished_count += 1
        res_str = f'线程 {res} 结束.'
        self.insert_otherlog(res_str)
        if self.thread_finished_count == self.max_thread_count:
            temp_str = '上次执行正常结束'
            if self.pre_last_time_data:
                temp_str = '上次被强制停止'
                with open(os.path.join(basedir,'data','temp.json'),'w',encoding='utf-8') as fw:
                    json.dump(self.pre_last_time_data,fw)
                pass
            if len(self.pre_last_time_executed_tickers)>0:
                temp_str = '上次被强制停止'
                tickers_str = '\n'.join(self.pre_last_time_executed_tickers)
                with open(os.path.join(basedir,'data','executed.txt'),'w',encoding='utf-8') as fw:
                    fw.write(tickers_str)
                pass
            if temp_str == '上次执行正常结束':
                if os.path.exists(os.path.join(basedir,'data','executed.txt')):
                    os.remove(os.path.join(basedir,'data','executed.txt'))
                if os.path.exists(os.path.join(basedir,'data','temp.json')):
                    os.remove(os.path.join(basedir,'data','temp.json'))
                pass
            pre_str = f"{self.txtdir_lineedit.text()};{self.savedir_lineedit.text()};{temp_str}"
            with open(os.path.join(basedir,'data','params.txt'),'w',encoding='utf-8') as fw:
                fw.write(pre_str)
            if self.waitting_close:
                self.close()
            else:
                self.txtdir_lineedit.setDisabled(False)
                self.savedir_lineedit.setDisabled(False)
                self.choicedir_btn.setDisabled(False)
                self.savedir_btn.setDisabled(False)
                self.execute_btn.setDisabled(False)
                self.information_dialog('所有工作线程停止完毕')
            pass
        pass
    def thread_error_fn(self,res:tuple):
        error_str = f"线程 {res[0]} 报错。报错类型：{res[1]}。值：{res[2]}。异常栈：{res[3]}"
        self.insert_executelog(error_str)
        pass
    def stop_btn_clicked(self):
        answer = QMessageBox.question(
            self,
            '确认停止？',
            '确定要停止任务么？',
            QMessageBox.StandardButton.Yes | QMessageBox.StandardButton.No
        )
        if answer == QMessageBox.StandardButton.Yes:
            self.force_stop()
            pass
        else:
            return
        pass
    def force_stop(self):
        for item in self.runner_list:
            item.stop_run()
        pass
    def insert_executelog(self,msg:str):
        pre_str = f"{datetime.now().strftime('%Y-%m-%d %H:%M:%S')} {msg}"
        self.excutelog_textedit.append(pre_str)
        pass
    def insert_otherlog(self,msg:str):
        pre_str = f"{datetime.now().strftime('%Y-%m-%d %H:%M:%S')} {msg}"
        self.otherlog_textedit.append(pre_str)
        pass
    def information_dialog(self,msg:str):
        QMessageBox.information(
            self,
            '提示',
            msg,
            QMessageBox.StandardButton.Ok
        )
        pass
    def closeEvent(self, a0):
        answer = QMessageBox.question(
            self,
            '确认退出？',
            '退出将中断操作，确定要退出么？',
            QMessageBox.StandardButton.Yes | QMessageBox.StandardButton.No
        )
        if answer == QMessageBox.StandardButton.Yes:
            if not self.this_time_start_yeah:
                a0.accept()
            else:
                if self.thread_finished_count == self.max_thread_count:
                    a0.accept()
                else:
                    QMessageBox.information(
                        self,
                        '提示',
                        '正在存储必要信息为下次断点续传做准备，需要等几秒后关闭',
                        QMessageBox.StandardButton.Ok
                    )
                    a0.ignore()
                    self.waitting_close = True
                    self.force_stop()
                    pass
            pass
        else:
            a0.ignore()
            pass
        pass
    pass

if __name__ == '__main__':
    app = QApplication([])
    window = MainWindow()
    window.show()
    app.exec()
    pass

视频：

https://www.bilibili.com/video/BV1p8ZAYAEz3/