全网最新——免费PDF文字发票信息提取重命名软件

import tkinter as tk
from tkinter import filedialog, messagebox, ttk, Toplevel, StringVar, Checkbutton, Button
import logging
import pdfplumber
import re
import os
import subprocess
import xlwt
pdf_files_folder = None
 
# 配置日志记录
logging.basicConfig(filename='app.log',
                    filemode='w',
                    format='%(name)s - %(levelname)s - %(message)s',
                    level=logging.INFO)
 
reverse = False
#定义排序函数
def sorter(tree, column, data_type, reverse):
    l = [(tree.set(k, column), k) for k in tree.get_children('')]
    if data_type == 'num':
        try:
            l = [(float(x), k) for x, k in l]
        except ValueError as e:
            pass
    l.sort(reverse=reverse)
 
    for index, (val, k) in enumerate(l):
        tree.move(k, '', index)
 
def column_sorter(tree, column, data_type='str'):
    global reverse
    reverse = not reverse
    sorter(tree, column, data_type, reverse)
def read_pdf_content(file_path):
    logging.info('开始函数 read_pdf_content')
    with pdfplumber.open(file_path) as pdf:
        full_text = "".join(page.extract_text() for page in pdf.pages)
        invoice_number = re.findall(r'(?:发票号码)\s*[：:]*\s*([^\s]+)', full_text)
        name = re.findall(r'(?:名称|名\s{1}称|名\s{1}称\s{1}：)\s*[：:]*\s*([^\s]+)', full_text)
        date = re.findall(r'\d\s*\d\s*\d\s*\d\s*年\s*\d\s*\d?\s*月\s*\d\s*\d?\s*日', full_text)
        category = re.findall(r'\*([\u4e00-\u9fa5a-zA-Z]+)\*([\u4e00-\u9fa5a-zA-Z]+)', full_text)
    logging.info('结束函数 read_pdf_content')
    return {
            "text": full_text, 
            "invoice_number": invoice_number[0] if invoice_number else "", 
            "name": name[0] if name else "", 
            "date": date[0] if date else "",
            "category": category if category else []
           }
 
def get_pdf_files(pdf_dir):
    logging.info('Starting function get_pdf_files')
    pdf_files = []
    for root, dirs, files in os.walk(pdf_dir):
        for file in files:
            if file.endswith(".pdf"):
                # 在此处对文件路径进行处理，确保路径格式正确
                filepath = os.path.normpath(os.path.join(root, file))
                pdf_files.append(filepath)
    logging.info(f'Finished function get_pdf_files. Found {len(pdf_files)} pdf files.')
    return pdf_files
 
def rename_pdf_file(file_path, new_value):
    logging.info('Starting function rename_pdf_file')
    dir_path = os.path.dirname(file_path)
    new_file_name = f"{new_value}.pdf"
    new_file_path = os.path.join(dir_path, new_file_name)
    os.rename(file_path, new_file_path)
    logging.info(f'Finished function rename_pdf_file. File {file_path} renamed to {new_file_path}')
    return new_file_path
 
# Your remaining function definitions and script body here...
# Don't forget to add logging statements to the rest of your functions and to your script body.
 
def sort_amount(tree, prev_sort=None):
    if prev_sort is None:
        prev_sort = {"column": "Amount", "reverse": False}
 
    tree_items = tree.get_children()
    reverse = not prev_sort["reverse"] if prev_sort["column"] == "Amount" else False
 
    # 排序并更新序号
    sorted_items = sorted(tree_items, key=lambda item: float(tree.set(item, "Amount")), reverse=reverse)
    for index, item in enumerate(sorted_items, start=1):
        tree.move(item, '', index - 1)
        tree.item(item, values=(index, *tree.item(item, "values")[1:]))
 
    prev_sort["column"] = "Amount"
    prev_sort["reverse"] = reverse
    return prev_sort
 
def open_pdf(path):
    if os.name == 'nt': # For Windows
        os.startfile(path)
    else: # For MacOS and Linux
        opener = 'open' if os.name == 'posix' else 'xdg-open'
        subprocess.call([opener, path])
         
def display_results(values, total_amount, input_root):
    # 输入界面的根窗口销毁
    input_root.destroy()
 
    # 创建新的根窗口
    root = tk.Tk()
    root.title("发票金额统计")
    center_window(root, width=950, height=300)
 
    # 主框架设置
    main_frame = ttk.Frame(root, padding="10 10 10 10")
    main_frame.grid(column=0, row=0, sticky=(tk.W, tk.E, tk.N, tk.S))
    main_frame.columnconfigure(0, weight=1)
    main_frame.rowconfigure(0, weight=1)
 
    # 定义重命名选中文件的功能
    def rename_selected_files():
        from tkinter import Toplevel, StringVar, Checkbutton, Button
 
        selected_items = tree.selection()
        if not selected_items:
            messagebox.showerror("错误", "请先选择一个或多个PDF文件进行重命名")
            return
         
        # 创建对话框并注册变量
        dialog = Toplevel(root)
        dialog.title("选择需要的字段")
         
        include_amount = StringVar(value='no')
        include_category = StringVar(value='no')
        include_invoice_number = StringVar(value='no')
        include_name = StringVar(value='no')
 
        Checkbutton(dialog, text='金额', variable=include_amount, onvalue='yes', offvalue='no', anchor='w').pack(fill='x')
        Checkbutton(dialog, text='类别', variable=include_category, onvalue='yes', offvalue='no', anchor='w').pack(fill='x')
        Checkbutton(dialog, text='发票号码', variable=include_invoice_number, onvalue='yes', offvalue='no', anchor='w').pack(fill='x')
        Checkbutton(dialog, text='公司名称', variable=include_name, onvalue='yes', offvalue='no', anchor='w').pack(fill='x')
 
 
        def on_ok():
            # 对话框确定按钮的回调函数
            for item in selected_items:
                print(f"Processing item: {item}")
                item_values = tree.item(item, 'values')
                new_name_parts = []
                if include_amount.get() == 'yes':
                    new_name_parts.append(item_values[1])
                if include_category.get() == 'yes':
                    new_name_parts.append(item_values[5])  # Category
                if include_invoice_number.get() == 'yes':
                    new_name_parts.append(item_values[2])  # Invoice Number
                if include_name.get() == 'yes':
                    new_name_parts.append(item_values[3])  # Name
                 
                if not new_name_parts:  # 如果用户没有选择任何字段，显示错误信息
                    messagebox.showerror("错误", "至少需要选择一个字段来重命名文件。")
                    dialog.destroy()
                    return
 
                new_file_name = "_".join(new_name_parts) 
                current_file_path = item_values[6]  # 原始的pdf路径
                new_file_path = rename_pdf_file(current_file_path, new_file_name)
                tree.set(item, column="文件路径", value=new_file_path)
 
            messagebox.showinfo("完成", "选中的文件已重命名。")
            dialog.destroy()
 
        Button(dialog, text='确定', command=on_ok).pack()
 
        dialog.update_idletasks()
        width = dialog.winfo_reqwidth()
        height = dialog.winfo_reqheight()
        x = (root.winfo_screenwidth() // 2) - (width // 2)
        y = (root.winfo_screenheight() // 2) - (height // 2)
        dialog.geometry('+{}+{}'.format(x, y))
 
        dialog.transient(root)  
        dialog.grab_set()
        dialog.wait_window()  # 等待对话框关闭
     
    def export_to_xls():
        global pdf_files_folder
        if pdf_files_folder is None:
            messagebox.showerror("错误", "PDF文件目录未设置。")
            return
        xls_file_path = os.path.join(pdf_files_folder, '发票数据.xls')
        # 创建一个Workbook对象
        workbook = xlwt.Workbook()
        # 添加一个sheet
        sheet = workbook.add_sheet('发票数据')
        sheet.col(0).width = 256 * 5  # 序号列设为20个字符宽度
        sheet.col(1).width = 256 * 10  # 金额列
        sheet.col(2).width = 256 * 30  # 发票号码列
        sheet.col(3).width = 256 * 30  # 公司列
        sheet.col(4).width = 256 * 20  # 开票日期列
        sheet.col(5).width = 256 * 25  # 类别列
        sheet.col(6).width = 256 * 10  # 文件路径列
         
        style = xlwt.easyxf('align: vert centre, horiz centre')
        # 将Treeview列标题写入sheet的第一行
        for i, header in enumerate(headers):
            sheet.write(0, i, header, style)
 
        # 写入数据
        for i, item in enumerate(tree.get_children(), start=1):
            # 获取每一行的数据
            row_values = tree.item(item, 'values')
            for j, value in enumerate(row_values):
                # 假设你想将所有的单元格格式设置为文本，请确保这里的value是字符串
                sheet.write(i, j, str(value), style) # 写入数据，注意i是从1开始的因为0是标题行
 
        # 保存xls文件
        workbook.save(xls_file_path)
        messagebox.showinfo("完成", f"数据成功导出至 '{xls_file_path}' 文件。")
 
 
    # 定义复制总金额到剪贴板的功能
    def copy_total_amount_to_clipboard(total_amount):
        root.clipboard_clear()
        root.clipboard_append(f"{total_amount:.2f}")
 
    # 树视图设置
    tree = ttk.Treeview(main_frame, columns=("序号", "金额", "发票号码", "公司", "开票日期", "类别", "文件路径"), show="headings")
 
    tree.grid(column=0, row=0, pady=5, padx=5, sticky=(tk.N, tk.S, tk.E, tk.W))
    tree.bind('<Double-1>', lambda event: open_pdf(tree.item(tree.selection())['values'][6]))
 
 
    # 默认的列宽和标题设置
    columns = ("序号", "金额", "发票号码", "公司", "开票日期", "类别", "文件路径")
    headers = ("序号", "金额", "发票号码", "公司", "开票日期", "类别", "文件路径")
    column_widths = {"序号": 50, "金额": 100, "发票号码": 200, "公司": 180, "开票日期": 100, "类别": 100, "文件路径": 50}
    for col in tree['columns']:
        # 使用列的名字作为键来从字典中获取列的宽度
        tree.column(col, width=column_widths[col])
        tree.heading(col, text=col)
        tree.column(col, anchor='center') 
 
    # 插入数据到树视图
    for index, value in enumerate(values, start=1):
        tree.insert("", "end", values=(index, *value))
 
    # 右侧按钮容器
    button_frame = ttk.Frame(main_frame)
    button_frame.grid(column=1, row=0, padx=5, sticky=(tk.N, tk.S))
 
    button_frame.rowconfigure(0, weight=1)
    button_frame.rowconfigure(1, weight=1)
    button_frame.rowconfigure(2, weight=1)
    button_frame.rowconfigure(3, weight=1)
 
    # 创建按钮并添加到右侧按钮容器
    rename_button = ttk.Button(button_frame, text="重命名选中文件", command=rename_selected_files)
    rename_button.grid(column=0, row=0, pady=5, sticky=(tk.N, tk.S))
 
    copy_button = ttk.Button(button_frame, text="复制总金额", command=lambda: copy_total_amount_to_clipboard(total_amount))
    copy_button.grid(column=0, row=1, pady=5, sticky=(tk.N, tk.S))
     
    export_button = ttk.Button(button_frame, text="导出到XLS", command=export_to_xls)
    export_button.grid(column=0, row=2, pady=5, sticky=(tk.N, tk.S))  # 注意：row的索引是3，因为我们已经有三个按钮了
 
    exit_button = ttk.Button(button_frame, text="退出", command=root.destroy)
    exit_button.grid(column=0, row=3, pady=5, sticky=(tk.N, tk.S))
    # 累计金额标签
    total_amount_label = ttk.Label(main_frame, text=f"所有发票的累计金额: {total_amount:.2f}")
    total_amount_label.grid(column=0, row=1, pady=5, sticky=tk.W)
 
    # 启动事件循环前更新窗口
    root.update()
 
    # 启动事件循环
    root.mainloop()
 
    logging.info('Exited main loop')
    def copy_total_amount_to_clipboard():
        root.clipboard_clear()
        root.clipboard_append(f"{total_amount:.2f}")
 
    root.mainloop()
    logging.info('Exited main loop')
tree = None
root = None
def resize_treeview_columns(event):
    global tree, root
    new_width = event.width
    tree.column("Index", width=int(new_width*0.05), anchor=tk.CENTER)
    tree.column("Amount", width=int(new_width*0.05), anchor=tk.CENTER)
    tree.column("Invoice Number", width=int(new_width*0.15), anchor=tk.CENTER)
    tree.column("Name", width=int(new_width*0.15), anchor=tk.CENTER)
    tree.column("Date", width=int(new_width*0.1), anchor=tk.CENTER)
    tree.column("Category", width=int(new_width*0.15), anchor=tk.CENTER)
    tree.column("Renamed File", width=int(new_width*0.15), anchor=tk.CENTER)
    tree.update_idletasks()
 
 
    total_amount_label = ttk.Label(main_frame, text=f"所有发票的累计金额: {total_amount:.2f}")
    total_amount_label.grid(column=0, row=1, pady=5, sticky=tk.W)
     
 
    def copy_total_amount_to_clipboard():
        root.clipboard_clear()
        root.clipboard_append(total_amount_label["text"].split(": ")[1])
 
    copy_button = ttk.Button(main_frame, text="复制总金额", command=copy_total_amount_to_clipboard)
    copy_button.grid(column=1, row=1, pady=5, padx=5, sticky=tk.W)
 
    ttk.Button(main_frame, text="退出", command=root.destroy).grid(column=1, row=2, pady=10, sticky=tk.E)
 
    root.columnconfigure(0, weight=1)
    root.rowconfigure(0, weight=1)
    root.mainloop()
     
def center_window(root, width=420, height=100):
    screen_width = root.winfo_screenwidth()
    screen_height = root.winfo_screenheight()
 
    x = (screen_width - width) // 2
    y = (screen_height - height) // 2
 
    root.geometry(f"{width}x{height}+{x}+{y}")
 
     
def browse_folder(entry):
    folder = filedialog.askdirectory(title="请选择发票PDF文件夹路径")
    if folder:
        entry.delete(0, tk.END)
        entry.insert(0, folder)
 
def start_processing(entry, input_root):
    logging.info('Start processing')
    folder = entry.get()
    if not folder:
        messagebox.showerror("错误", "请先选择或输入一个文件夹路径")
        return
 
    global pdf_files_folder
    pdf_files = get_pdf_files(folder)
    if pdf_files:
        pdf_files_folder = os.path.dirname(pdf_files[0]) 
 
    values = []
    renamed_pdf_files = []
 
    for pdf_file in pdf_files:
        pdf_content = read_pdf_content(pdf_file)
        pdf_text = pdf_content["text"]
        pattern = r'[&#165;￥](\d+(?:\.\d{1,2})?)'
        result = re.findall(pattern, pdf_text)
         
        if result:
            max_value = max([float(i) for i in result])
            values.append((max_value, 
                           pdf_content["invoice_number"], 
                           pdf_content["name"], 
                           pdf_content["date"], 
                           pdf_content["category"],
                           pdf_file))  # 将原始pdf路径保存而不是重命名后的路径
 
 
    logging.info(f'Values: {values}')
    logging.info(f'Renamed pdf files: {renamed_pdf_files}')
     
    # Extract the first element of each tuple in values
    amounts = [value[0] for value in values]
     
    total_amount = sum(amounts)
 
    logging.info(f'Total amount: {total_amount}')
    logging.info('Calling display_results function')
     
    display_results(values, total_amount, input_root)
     
    logging.info('Finished processing')
 
def main():
    logging.info('Application start')
    root = tk.Tk()
    root.title("发票金额统计")
 
    center_window(root)
 
    main_frame = ttk.Frame(root, padding="10 10 10 10")
    main_frame.grid(column=0, row=0, sticky=(tk.W, tk.E, tk.N, tk.S))
 
    folder_entry = ttk.Entry(main_frame, width=40)
    folder_entry.grid(column=0, row=0, padx=5, pady=5, sticky=tk.W)
 
    browse_button = ttk.Button(main_frame, text="打开", command=lambda: browse_folder(folder_entry))
    browse_button.grid(column=1, row=0, padx=5, pady=5, sticky=tk.W)
 
    start_button = ttk.Button(main_frame, text="开始处理", command=lambda: start_processing(folder_entry, root))
    start_button.grid(column=1, row=1, padx=5, pady=5, sticky=tk.W)
 
    root.columnconfigure(0, weight=1)
    root.rowconfigure(0, weight=1)
    root.mainloop()
logging.info('Application finished')
if __name__ == "__main__":
    main()