基於Python實現PDF區域文本提取工具
功能簡介
打開軟件後界面如下:
點擊打開文件按鈕打開之前的PDF文件後效果如下:
框選區域後,標題欄會自動顯示當前框選的區域提取到的文字,還可以左右按鈕切換:
實際我們需要提取文字的區域可能不止這一個,所以程序支持多區域框選:
完成區域框選後就可以點擊保存文件,將PDF每頁提取到的文本保存到一個csv文件中,當前選區的保存結果如下:
可以看到已經按框選順序依次保存瞭每一個區域的字符串。
如果選擇區域時發現提取結果不準確,可以撤銷後重新選擇:
保存圖片則會將PDF的每頁的整體保存為一張圖片,未選擇區域時,以頁碼為文件名保存圖片:
選擇區域時,會自動提取最後一個區域提取的文本作為當前頁的文件名:
開發代碼
當然這個項目由於本人是一次使用wxpython,功能非常簡約,現在將完整代碼開源出來期待各位大佬的改進。
源碼和已編譯工具下載地址:
https://codechina.csdn.net/as604049322/python_gui
完整代碼:
"""
小小明的代碼
CSDN主頁: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, "打開文件", (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, "保存圖片", (150, 0))
self.Bind(wx.EVT_BUTTON, self.save_img, b)
b = wx.Button(self, -1, "撤銷選區", (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="選擇一個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范圍:", 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"保存一個{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, "選擇圖片保存的文件夾:",
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"區域{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()
以上就是基於Python實現PDF區域文本提取工具的詳細內容,更多關於Python PDF文本提取的資料請關註WalkonNet其它相關文章!
推薦閱讀:
- python opencv鼠標畫矩形框之cv2.rectangle()函數
- 如何使用Python進行PDF圖片識別OCR
- Python 操作pdf pdfplumber讀取PDF寫入Exce
- Python提取PDF中的圖片的實現示例
- python用pdfplumber提取pdf表格數據並保存到excel文件中