基于Python实现PDF区域文本提取工具

目录

• 功能简介
• 开发代码

功能简介

打开软件后界面如下：

点击打开文件按钮打开之前的PDF文件后效果如下：

框选区域后，标题栏会自动显示当前框选的区域提取到的文字，还可以左右按钮切换：

实际我们需要提取文字的区域可能不止这一个，所以程序支持多区域框选：

完成区域框选后就可以点击保存文件，将PDF每页提取到的文本保存到一个csv文件中，当前选区的保存结果如下：

可以看到已经按框选顺序依次保存了每一个区域的字符串。

如果选择区域时发现提取结果不准确，可以撤销后重新选择：

保存图片则会将PDF的每页的整体保存为一张图片，未选择区域时，以页码为文件名保存图片：

选择区域时，会自动提取最后一个区域提取的文本作为当前页的文件名：

开发代码

当然这个项目由于本人是一次使用wxpython，功能非常简约，现在将完整代码开源出来期待各位大佬的改进。

源码和已编译工具下载地址：

https://codechina.csdn.net/as604049322/python_gui

完整代码：

"""
小小明的代码
CSDN主页：https://blog.csdn.net/as604049322
"""
__author__ = '小小明'
__time__ = '2021/11/24'

import csv

import wx
import os
import fitz

class MyCanvas(wx.Panel):
    def __init__(self, parent):
        wx.Panel.__init__(self, parent)
        self.parent = parent
        self.rects = []
        self.Bind(wx.EVT_LEFT_DOWN, self.OnLeftButtonEvent)
        self.Bind(wx.EVT_LEFT_UP, self.OnLeftButtonEvent)
        self.Bind(wx.EVT_MOTION, self.OnLeftButtonEvent)
        self.Bind(wx.EVT_PAINT, self.DoDrawing)
        b = wx.Button(self, -1, "打开文件", (0, 0))
        self.Bind(wx.EVT_BUTTON, self.OnButton, b)
        b = wx.Button(self, -1, "保存文件", (75, 0))
        self.Bind(wx.EVT_BUTTON, self.save_file, b)
        b = wx.Button(self, -1, "保存图片", (150, 0))
        self.Bind(wx.EVT_BUTTON, self.save_img, b)
        b = wx.Button(self, -1, "撤销选区", (225, 0))
        self.Bind(wx.EVT_BUTTON, self.back_select, b)

        b = wx.Button(self, -1, "《", (300, 0), size=(25, 25))
        self.Bind(wx.EVT_BUTTON, self.previous, b)
        b = wx.Button(self, -1, "》", (325, 0), size=(25, 25))
        self.Bind(wx.EVT_BUTTON, self.next, b)

        self.g1 = wx.Gauge(self, -1, 100, (0, 30), (-1, 100), wx.GA_VERTICAL)

    def previous(self, evt):
        if not hasattr(self, "pdfDoc"):
            return
        if self.i > 0:
            self.i -= 1
            self.change_pdf_page(self.i, False)
            self.DoDrawing(-1)
            if self.rects:
                self.parent.SetTitle(self.path + "|" + self.extract_pdf_text())

    def next(self, evt):
        if not hasattr(self, "pdfDoc"):
            return
        if self.i < self.pageCount - 1:
            self.i += 1
            self.change_pdf_page(self.i, False)
            self.DoDrawing(-1)
            if self.rects:
                self.parent.SetTitle(self.path + "|" + self.extract_pdf_text())

    def back_select(self, evt):
        if self.rects:
            self.rects.pop()
            self.DoDrawing(-1)

    def OnButton(self, evt):
        dlg = wx.FileDialog(
            self, message="选择一个PDF文件",
            defaultDir=os.getcwd(),
            defaultFile="",
            wildcard="PDF文件(*.pdf)|*.pdf",
            style=wx.FD_OPEN | wx.FD_CHANGE_DIR |
                  wx.FD_FILE_MUST_EXIST | wx.FD_PREVIEW
        )
        if dlg.ShowModal() == wx.ID_OK:
            self.rects = []
            path = dlg.GetPath()
            self.pdfDoc = fitz.open(path)
            self.i = 0
            self.pageCount = self.pdfDoc.pageCount
            self.change_pdf_page(self.i)
            self.path = os.path.basename(path)
            self.parent.SetTitle(self.path)
            self.DoDrawing(-1)
        dlg.Destroy()

    def change_pdf_page(self, i, move=True):
        page = self.pdfDoc[i]
        rect = page.rect
        print("pdf范围：", rect)
        mat = fitz.Matrix(1, 1)
        pix = page.get_pixmap(matrix=mat, alpha=False, clip=rect)
        pix.save("tmp.png")
        self.change_img("tmp.png", move)

    def save_FileDialog(self, format="csv"):
        dlg = wx.FileDialog(
            self, message=f"保存一个{format}文件", defaultDir=os.getcwd(),
            defaultFile="", wildcard=f"{format}文件(*.{format})|*.{format}", style=wx.FD_SAVE | wx.FD_OVERWRITE_PROMPT
        )
        path = None
        if dlg.ShowModal() == wx.ID_OK:
            path = dlg.GetPath()
        dlg.Destroy()
        return path

    def save_img(self, evt):
        if not hasattr(self, "pdfDoc"):
            return
        dlg = wx.DirDialog(self, "选择图片保存的文件夹:",
                           style=wx.DD_DEFAULT_STYLE
                           # | wx.DD_DIR_MUST_EXIST
                           # | wx.DD_CHANGE_DIR
                           )
        mat = fitz.Matrix(1, 1)
        if dlg.ShowModal() == wx.ID_OK:
            path = dlg.GetPath()
            for i in range(self.pdfDoc.pageCount):
                page = self.pdfDoc[i]
                clip = page.rect
                pix = page.get_pixmap(matrix=mat, alpha=False, clip=clip)
                if self.rects:
                    name = self.extract_pdf_text(page=page, rect=self.rects[-1])
                else:
                    name = f"p{i:0>3d}"
                pix.save(f"{path}/{name}.png")
                self.g1.SetValue((i + 1) * 100 // self.pdfDoc.pageCount)
        dlg.Destroy()
        os.system(f"explorer {path}")

    def save_file(self, evt):
        if not hasattr(self, "pdfDoc"):
            return
        path = self.save_FileDialog()
        if path is None:
            return
        data = []
        for i in range(self.pdfDoc.pageCount):
            page = self.pdfDoc[i]
            row = [self.extract_pdf_text(page, rect)
                   for i, rect in enumerate(self.rects)]
            data.append(row)
        with open(path, "w") as f:
            writer = csv.writer(f, lineterminator="\n")
            row = [f"区域{i}" for i in range(1, len(row) + 1)]
            writer.writerow(row)
            for row in data:
                writer.writerow(row)
        os.system(f"cmd /c start {path}")

    def extract_pdf_text(self, page=None, rect=None):
        if page is None:
            page = self.pdfDoc[self.i]
        if rect is None:
            rect = self.rects[-1]
        a, b, c, d = rect
        clip = fitz.Rect(a, b, a + c, b + d)
        text = page.get_text(clip=clip).strip()
        return text

    def change_img(self, img_path, move=True):
        self.bmp = wx.Bitmap(img_path)
        self.SetSize(self.bmp.GetSize())
        self.parent.SetSize(self.parent.GetBestSize())
        if move:
            self.parent.Center()

    def DoDrawing(self, evt):
        if not hasattr(self, "bmp"):
            return
        dc = wx.ClientDC(self)
        dc.DrawBitmap(self.bmp, 0, 0, True)
        dc.SetPen(wx.Pen('blue'))
        dc.SetBrush(wx.Brush('white', wx.BRUSHSTYLE_TRANSPARENT))
        dc.DrawRectangleList(self.rects)

    def OnLeftButtonEvent(self, event):
        if event.LeftDown():
            self.x, self.y = event.GetPosition()
            self.rects.append([self.x, self.y, 0, 0])
        elif event.Dragging():
            x, y = event.GetPosition()
            self.rects[-1][2] = x - self.x
            self.rects[-1][3] = y - self.y
            self.DoDrawing(-1)
        elif event.LeftUp():
            print(self.rects)
            if self.rects[-1][2] < 5 or self.rects[-1][3] < 5:
                self.rects.pop()
            else:
                self.parent.SetTitle(self.path + "|" + self.extract_pdf_text())

app = wx.App()
frm = wx.Frame(None)
pnl = MyCanvas(frm)
frm.Center()
frm.Show()
frm.SetTitle("PDF文本提取器")
app.MainLoop() 

以上就是基于Python实现PDF区域文本提取工具的详细内容，更多关于Python PDF文本提取的资料请关注我们其它相关文章！