Python异步编程｜PySimpleGUI界面读取PDF转换Excel

import xlwt, pyperclip, asyncio, pdfplumber
import os, time, datetime as dt
import PySimpleGUI as sg
 
# 全局变量
table_head = '姓名,部门,应到,实到,出勤率,迟到次数,早退次数,加班(分钟)'
path, font = '', ('宋体',12)
date, data = [], []
DateFormat = '    .  . -    .  .  '
ErrMessage = '错误'
SortedType = ["出勤率排序","加班时长排序","迟到次数排序","早退次数排序"]
 
# 定义布局
layout = [
    [sg.Text("昆山分行考勤表",font=('',16)),
     sg.Text(pad=(132,10)),
     sg.Text("请选择考勤文件:",font=font),
     sg.Input(key="-FOLDER-", enable_events=True, readonly=True,font=font,size=18),
     sg.FolderBrowse(button_text='...', enable_events=True, initial_folder='./')
     ],
    [sg.Text("考勤日期：",font=font),
     sg.Text(DateFormat,key='-DATE-',font=font)
     ],
    [sg.Table(values='',
              headings=table_head.split(','),
              key='-TABLE-',
              auto_size_columns=False,
              justification='left',
              num_rows=10)],
    [sg.Button("输出Excel文件",size=(12,1),pad=(15,30)),
     sg.Button(SortedType[0], enable_events=True,size=(10,1),pad=(15,30)),
     sg.Button(SortedType[1], enable_events=True,size=(10,1),pad=(15,30)),
     sg.Button(SortedType[2], enable_events=True,size=(10,1),pad=(15,30)),
     sg.Button(SortedType[3], enable_events=True,size=(10,1),pad=(15,30)),
     sg.Button("退出",size=(10,1),pad=(15,30))],
    [sg.StatusBar('',key="-BAR-",font=font,size=92)]
]
 
# 读取pdf表格
async def read_table(file):
    dct = dict()
    with pdfplumber.open(file) as pdf:
        for page in pdf.pages:
            tables = page.extract_tables(table_settings = {})
            for table in tables:
                for lst in table:
                    tmp = lst[1:]
                    if not any(tmp): continue
                    tmp = [tmp[0]]+tmp[3:8]+[tmp[-1]]
                    tmp[0] = tmp[0].replace('\n','')
                    tmp[0] = tmp[0].split('/')
                    tmp[0] = tmp[0][-1]
                    if lst[0]=='时间':
                        dct[lst[0]] = tmp[0]
                    else:
                        dct[','.join([lst[0],tmp[0]])] = ','.join(tmp[1:])
    return dct
 
# 写入xls文件
def write_sheet():
    global data, date, table_head, ErrMessage
    if ErrMessage[:2] in ('错误','文件'): return
    myxl = xlwt.Workbook()
    style = xlwt.easyxf('align: wrap yes; align: horiz center; font: bold yes;') 
    sheet = myxl.add_sheet('考勤表')
    wcol = [20,40,60,30,30,40,40,40,60]
    for i,w in enumerate(wcol):
        sheet.col(i).width = w * 80
    sheet.write_merge(0,0,0,8,'出勤统计报表',style)
    style = xlwt.easyxf('borders:top thin; borders:bottom thin; borders:left thin; borders:right thin;') 
    sheet.write_merge(1,1,0,2,'考勤日期：'+date[0])
    for i,head in enumerate(['序号']+table_head.split(',')):
        sheet.write(2,i,head,style)
    for i,row in enumerate(data):
        for j,col in enumerate([str(i+1)]+row):
            sheet.write(3+i,j,col,style)
    for i,t in enumerate(SortedType):
        if t in ErrMessage:
            tmp = SortedType[i]
            break
    else: tmp = ""
    excel_file = f'昆山分行考勤表{date[0]}({tmp}{strDateTime()}).xls'
    ErrMessage = f'文件输出为：{excel_file}'
    try:
        myxl.save(excel_file)
    except:
        ErrMessage = '写入excel文件失败！'
    finally:
        pyperclip.copy('\\'.join((os.getcwd(),excel_file)))
        window['-BAR-'].update(ErrMessage)
 
# 获取当前时间
def strDateTime(diff=0):
    now = dt.datetime.now()
    time = now + dt.timedelta(days=diff)    
    return f'{time.year}{time.month:02}{time.day:02}{time.hour:02}{time.minute:02}{time.second:02}'
 
# 选择并处理文件
async def on_text_changed(event, values):
    global date, data, path, ErrMessage
    new_path = values["-FOLDER-"]
    window["-FOLDER-"].update(new_path.split('/')[-1])
    if path==new_path: return
    else: path = new_path
    pdfs = [f for f in os.listdir(path) if f.endswith('.pdf') and not f.startswith('PDFmerged')]
    nums = len(pdfs)
    if nums==0:
        ErrMessage = '错误：所选文件夹中没有PDF文件！'
        window['-BAR-'].update(ErrMessage)
        window['-DATE-'].update(DateFormat)
        window['-TABLE-'].update(values=[])
        return
    date, data, sheet = [], [], dict()
    tasks = []
    for pdf in pdfs:
        tasks.append(read_table('/'.join([path,pdf])))
    ErrMessage = f'文件读取中(共{nums}个PDF文件)......'
    window['-BAR-'].update(ErrMessage)
    window.refresh()
    results = await asyncio.gather(*tasks)
    for r in results:
        dt = r.get('时间',None)
        if dt: date.append(dt)
        sheet.update(r)
    if date:
        window['-DATE-'].update(date[-1])
    for k,v in sheet.items():
        if k in ('时间','姓名,所属组织','普通班个人出勤统计报表,'): continue
        data.append(','.join([k,v]).split(','))
        window['-TABLE-'].update(values=data)
    persons = len(data)
    departments = len(set([d[1] for d in data]))
    if 0:#len(set(date))!=1:
        data = []
        ErrMessage = f'错误：请检查所选文件存在多个时间段：{",".join(set(date))}'
    else:
        ErrMessage = f'考勤人数：{persons} / 部门数：{departments}'
    window['-BAR-'].update(ErrMessage)
 
# 表格排序
def on_table_sorted(event, data):
    global ErrMessage
    if not data: return
    slist = ['x[-4][:-1]', 'x[-1]', 'x[-3]', 'x[-2]']
    style = slist[SortedType.index(event)]
    data = sorted(data, key=lambda x: float(eval(style)), reverse=True)
    window['-TABLE-'].update(values=data)
    ErrMessage = f'已按{event}更新！'
    window['-BAR-'].update(ErrMessage)
 
# 创建窗口
window = sg.Window("考勤表汇总", layout, finalize=True)
 
# 事件循环
while True:
    event, values = window.read()
    if event == sg.WINDOW_CLOSED or event == "退出":
        break
    elif event == "-FOLDER-":
        asyncio.run(on_text_changed(event, values))
    elif event in SortedType:
        on_table_sorted(event, data)
    elif event == "输出Excel文件":
        write_sheet()
 
# 关闭窗口
window.close()