基于Python编写一个PDF转换工具箱

import os
import re
import sys
from pyqt5.qtwidgets import qapplication, qmainwindow, qpushbutton, qvboxlayout, qwidget, qfiledialog, qlistwidget, \
    qmessagebox, qlineedit, qhboxlayout
from pyqt5.qtcore import qthread, pyqtsignal
from pypdf2 import pdfreader, pdfwriter, pdfmerger
from pdf2docx import converter
import fitz  # 用于pdf转jpg的处理

class customlistwidget(qlistwidget):
    def __init__(self, parent=none):
        super().__init__(parent)
        self.setacceptdrops(true)
        self.parentwindow = parent

    def dragenterevent(self, event):
        if any(url.tostring().lower().endswith('.pdf') for url in event.mimedata().urls()):
            event.acceptproposedaction()

    def dragmoveevent(self, event):
        if any(url.tostring().lower().endswith('.pdf') for url in event.mimedata().urls()):
            event.acceptproposedaction()

    def dropevent(self, event):
        pdf_files = [url.tolocalfile() for url in event.mimedata().urls() if url.tostring().lower().endswith('.pdf')]
        for f in pdf_files:
            self.parentwindow.addpdffile(f)

class worker(qthread):
    finished = pyqtsignal(str)
    error = pyqtsignal(str)

    def __init__(self, pdf_files, range_str=none, save_path=none, operation=none):
        super().__init__()
        self.pdf_files = pdf_files
        self.range_str = range_str
        self.save_path = save_path
        self.operation = operation

    def run(self):
        try:
            if self.operation == 'merge':
                self.merge_pdfs()
            elif self.operation == 'split':
                self.split_pdfs()
            elif self.operation == 'delete':
                self.delete_pages()
            elif self.operation == 'extract':
                self.extract_pages()
            elif self.operation == 'jpg':
                self.pdf_to_jpg()
            elif self.operation == 'word':
                self.pdf_to_word()
        except exception as e:
            self.error.emit(str(e))

    def merge_pdfs(self):
        merger = pdfmerger()
        for pdf in self.pdf_files:
            merger.append(pdf)
        merger.write(self.save_path)
        merger.close()
        self.finished.emit('pdf文件已成功合并。')

    def split_pdfs(self):
        ranges = self.parse_ranges(self.range_str)
        reader = pdfreader(self.pdf_files[0])
        os.makedirs(self.save_path, exist_ok=true)  # 确保目标文件夹存在
        file_index = 1  # 用于创建唯一的文件名
        for range_index, (start_page, end_page) in enumerate(ranges):
            # 对于每个范围，拆分出来的每个页面为一个单独的pdf文件
            for page_num in range(start_page, end_page + 1):
                writer = pdfwriter()
                writer.add_page(reader.pages[page_num])
                # 使用文件索引来确保每个文件的名称都是唯一的
                split_save_path = os.path.join(self.save_path, f'split_page_{file_index}.pdf')
                with open(split_save_path, 'wb') as f:
                    writer.write(f)
                file_index += 1
        self.finished.emit('pdf文件已成功拆分并保存。')

    def delete_pages(self):
        ranges = self.parse_ranges(self.range_str)
        reader = pdfreader(self.pdf_files[0])
        writer = pdfwriter()
        pages_to_delete = {page for start, end in ranges for page in range(start, end + 1)}
        for i in range(len(reader.pages)):
            if i not in pages_to_delete:
                writer.add_page(reader.pages[i])
        with open(self.save_path, 'wb') as f:
            writer.write(f)
        self.finished.emit('指定页面已从pdf中删除。')

    def extract_pages(self):
        ranges = self.parse_ranges(self.range_str)
        reader = pdfreader(self.pdf_files[0])
        os.makedirs(self.save_path, exist_ok=true)  # 在循环外提前确保目录存在
        for i, (start_page, end_page) in enumerate(ranges):
            writer = pdfwriter()
            for page_num in range(start_page, end_page + 1):
                writer.add_page(reader.pages[page_num])
            extract_save_path = os.path.join(self.save_path, f'extract_{i + 1}.pdf')
            with open(extract_save_path, 'wb') as f:
                writer.write(f)
        self.finished.emit('指定页面已从pdf中提取。')

    def pdf_to_jpg(self):
        for file in self.pdf_files:
            pdf = fitz.open(file)
            img_folder = os.path.join(self.save_path, os.path.splitext(os.path.basename(file))[0])
            os.makedirs(img_folder, exist_ok=true)
            for pg in range(pdf.page_count):
                page = pdf[pg]
                trans = fitz.matrix(2, 2)  # 设置转换矩阵为放大2倍
                pm = page.get_pixmap(matrix=trans, alpha=false)
                pic_name = f'page_{pg + 1}.jpg'
                pic_path = os.path.join(img_folder, pic_name)
                pm.save(pic_path)
        self.finished.emit('pdf文件已成功转换为图片。')

    def pdf_to_word(self):
        for file in self.pdf_files:
            docx_name = os.path.splitext(file)[0] + '.docx'
            cv = converter(file)
            cv.convert(docx_name, start=0, end=none)
            cv.close()
        self.finished.emit('pdf文件已成功转换为word文档。')

    def parse_ranges(self, ranges_str):
        ranges = []
        for part in re.split(',|，', ranges_str):
            if '-' in part:
                start_page, end_page = map(int, part.split('-'))
                ranges.append((start_page - 1, end_page - 1))
            else:
                page = int(part)
                ranges.append((page - 1, page - 1))
        return ranges

class pdfmergerapp(qmainwindow):
    def __init__(self):
        super().__init__()
        self.initui()
        self.pdf_files = []

    def initui(self):
        self.setwindowtitle('pdf 工具箱')
        self.setgeometry(100, 100, 800, 600)

        mainlayout = qvboxlayout()

        self.addbutton = qpushbutton('添加 pdf', self)
        self.addbutton.clicked.connect(self.addpdf)
        mainlayout.addwidget(self.addbutton)

        self.listwidget = customlistwidget(self)
        mainlayout.addwidget(self.listwidget)

        deletelayout = qhboxlayout()
        self.removebutton = qpushbutton('删除选定', self)
        self.removebutton.clicked.connect(self.removeselected)
        deletelayout.addwidget(self.removebutton)

        self.removeallbutton = qpushbutton('删除全部', self)
        self.removeallbutton.clicked.connect(self.removeall)
        deletelayout.addwidget(self.removeallbutton)
        mainlayout.addlayout(deletelayout)

        convertlayout = qhboxlayout()
        self.convertjpgbutton = qpushbutton('转换为图片', self)
        self.convertjpgbutton.clicked.connect(self.converttojpg)
        convertlayout.addwidget(self.convertjpgbutton)

        self.convertwordbutton = qpushbutton('转换为word', self)
        self.convertwordbutton.clicked.connect(self.converttoword)
        convertlayout.addwidget(self.convertwordbutton)
        mainlayout.addlayout(convertlayout)

        self.mergebutton = qpushbutton('合并 pdfs', self)
        self.mergebutton.clicked.connect(self.mergepdfs)
        mainlayout.addwidget(self.mergebutton)

        splitlayout = qhboxlayout()
        self.splitinput = qlineedit(self)
        self.splitinput.setplaceholdertext('输入拆分页码范围可输入多个范围，如1,3-4,8-15')
        splitlayout.addwidget(self.splitinput)
        self.splitbutton = qpushbutton('拆分页面', self)
        self.splitbutton.clicked.connect(self.splitpdf)
        splitlayout.addwidget(self.splitbutton)
        mainlayout.addlayout(splitlayout)

        deletepagelayout = qhboxlayout()
        self.deleteinput = qlineedit(self)
        self.deleteinput.setplaceholdertext('输入删除页码范围可输入多个范围，如1,3-4,8-15')
        deletepagelayout.addwidget(self.deleteinput)
        self.deletebutton = qpushbutton('删除页面', self)
        self.deletebutton.clicked.connect(self.deletepages)
        deletepagelayout.addwidget(self.deletebutton)
        mainlayout.addlayout(deletepagelayout)

        extractlayout = qhboxlayout()
        self.extractinput = qlineedit(self)
        self.extractinput.setplaceholdertext('输入提取页码范围可输入多个范围，如1,3-4,8-15')
        extractlayout.addwidget(self.extractinput)
        self.extractbutton = qpushbutton('提取页面', self)
        self.extractbutton.clicked.connect(self.extractpages)
        extractlayout.addwidget(self.extractbutton)
        mainlayout.addlayout(extractlayout)



        container = qwidget()
        container.setlayout(mainlayout)
        self.setcentralwidget(container)

    def addpdf(self):
        files, _ = qfiledialog.getopenfilenames(self, '打开文件', '', 'pdf files (*.pdf)')
        for file_path in files:
            self.addpdffile(file_path)

    def addpdffile(self, file_path):
        if file_path and file_path not in self.pdf_files:
            self.pdf_files.append(file_path)
            self.listwidget.additem(file_path)

    def removeselected(self):
        for item in self.listwidget.selecteditems():
            self.pdf_files.remove(item.text())
            self.listwidget.takeitem(self.listwidget.row(item))

    def removeall(self):
        self.pdf_files.clear()
        self.listwidget.clear()

    def mergepdfs(self):
        save_path, _ = qfiledialog.getsavefilename(self, '保存文件', '', 'pdf files (*.pdf)')
        if save_path:
            self.thread = worker(self.pdf_files, save_path=save_path, operation='merge')
            self.thread.finished.connect(self.onfinished)
            self.thread.error.connect(self.onerror)
            self.thread.start()

    def splitpdf(self):
        if len(self.pdf_files) != 1:
            qmessagebox.warning(self, "错误", "请只选择一个pdf文件进行拆分。")
            return

        range_str = self.splitinput.text().strip()
        folder_path = self.getfoldername()
        if range_str and folder_path:
            self.thread = worker(self.pdf_files, range_str=range_str, save_path=folder_path, operation='split')
            self.thread.finished.connect(self.onfinished)
            self.thread.error.connect(self.onerror)
            self.thread.start()

    def deletepages(self):
        if len(self.pdf_files) != 1:
            qmessagebox.warning(self, "错误", "请只选择一个pdf文件进行删除操作。")
            return

        range_str = self.deleteinput.text().strip()
        save_path = qfiledialog.getsavefilename(self, '保存文件', '', 'pdf files (*.pdf)')[0]
        if save_path and range_str:
            self.thread = worker(self.pdf_files, range_str=range_str, save_path=save_path, operation='delete')
            self.thread.finished.connect(self.onfinished)
            self.thread.error.connect(self.onerror)
            self.thread.start()

    def extractpages(self):
        if len(self.pdf_files) != 1:
            qmessagebox.warning(self, "错误", "请只选择一个pdf文件进行提取操作。")
            return

        range_str = self.extractinput.text().strip()
        save_path = qfiledialog.getsavefilename(self, '保存文件', '', 'pdf files (*.pdf)')[0]
        if save_path and range_str:
            self.thread = worker(self.pdf_files, range_str=range_str, save_path=save_path, operation='extract')
            self.thread.finished.connect(self.onfinished)
            self.thread.error.connect(self.onerror)
            self.thread.start()

    def converttojpg(self):
        save_path = qfiledialog.getexistingdirectory(self, "选择保存图片的位置")
        if save_path:
            self.thread = worker(self.pdf_files, save_path=save_path, operation='jpg')
            self.thread.finished.connect(self.onfinished)
            self.thread.error.connect(self.onerror)
            self.thread.start()

    def converttoword(self):
        save_path = qfiledialog.getexistingdirectory(self, "选择保存word的位置")
        if save_path:
            self.thread = worker(self.pdf_files, save_path=save_path, operation='word')
            self.thread.finished.connect(self.onfinished)
            self.thread.error.connect(self.onerror)
            self.thread.start()

    def getfoldername(self):
        folder_path = qfiledialog.getexistingdirectory(self, "选择保存拆分文件的位置")
        return folder_path

    def onfinished(self, message):
        qmessagebox.information(self, "操作完成", message)
        self.clear_pdf_list()
        self.clear_text_inputs()

    def onerror(self, error_message):
        qmessagebox.warning(self, "操作失败", error_message)

    def clear_pdf_list(self):
        self.pdf_files.clear()
        self.listwidget.clear()

    def clear_text_inputs(self):
        # 清除所有的qlineedit控件内容
        self.splitinput.clear()
        self.deleteinput.clear()
        self.extractinput.clear()
def main():
    app = qapplication(sys.argv)
    ex = pdfmergerapp()
    ex.show()
    sys.exit(app.exec_())

if __name__ == '__main__':
    main()