一、安装所需要的python库
pip3 install fitz -i https://pypi.tuna.tsinghua.edu.cn/simple
pip3 install PyMuPDF -i https://pypi.tuna.tsinghua.edu.cn/simple
pip3 install paddleocr -i https://pypi.tuna.tsinghua.edu.cn/simple
pip3 install paddlepaddle -i https://mirror.baidu.com/pypi/simple
二、所需提取的内容
发票号码
开票日期
销售方名称
销售方纳税人识别号
金额
import fitz
from paddleocr import PaddleOCR
from openpyxl import load_workbook
pdf = fitz.open('test.pdf')
page = pdf.load_page(0) # 获取第0页
# 提高分辨率
pix = page.get_pixmap()
zoom_x = 8
zoom_y = 8
mat = fitz.Matrix(zoom_x, zoom_y)
pix = page.get_pixmap(matrix=mat, alpha=False)
pix.save('temp.png')
# for i in range(pdf.page_count):
# page = pdf.load_page(i) # 获取第i页
# pix = page.get_pixmap()
# zoom_x = 8
# zoom_y = 8
# mat = fitz.Matrix(zoom_x, zoom_y)
# pix = page.get_pixmap(matrix=mat, alpha=False)
# pix.save('temp.png')
ocr = PaddleOCR(use_angle_cls = True, use_gpu = False, lang='ch')
res = ocr.ocr('temp.png', cls=True)
invoice_dic = {
}
for r in res[0]:
text = r[1][0]
if text.startswith('发票号码:'):
invoice_dic.update({
'number': text[5:]})
print(f"number:{
text[5:]}")
if text.startswith('开票日期:'):
invoice_dic.update({
'date': text[5:]})
print(f"date:{
text[5:]}")
if text.startswith('名称:'):
if text.endswith('购买方名称'):
continue
invoice_dic.update({
'name': text[3:]})
print(f"name:{
text[3:]}")
if text.startswith('统一社会信用代码/纳税人识别号:'):
if text.endswith('购买方统一社会信用代码/纳税人识别号'):
continue
invoice_dic.update({
'id': text[16:]})
print(f"id:{
text[16:]}")
if text.startswith('(小写)¥'):
invoice_dic.update({
'price': text[6:]})
print(f"price:{
text[6:]}")
三、写入excel表格
wb = load_workbook('test.xlsx') # 打开一个文件
ws = wb.active
nrow = ws.max_row # 取得行数
append_row = nrow + 1
ws.cell(append_row, 1).value = invoice_dic['number']
ws.cell(append_row, 2).value = invoice_dic['date']
ws.cell(append_row, 3).value = invoice_dic['name']
ws.cell(append_row, 4).value = invoice_dic['id']
ws.cell(append_row, 5).value = invoice_dic['price']
wb.save('test.xlsx')
欢迎一键三连