zy123/zbparse
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

1.6 insert_del_pagemark.py修改

Browse Source
This commit is contained in:
zy123 2025-01-06 10:34:30 +08:00
parent 65a7ddf1b8
commit 04816cdedf
2 changed files with 9 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
flask_app/general/insert_del_pagemark.py
View File

@ -5,7 +5,7 @@ from PyPDF2 import PdfReader, PdfWriter
from docx import Document from docx import Document
from reportlab.pdfgen import canvas from reportlab.pdfgen import canvas
from reportlab.lib.units import cm from reportlab.lib.units import cm
import copy
#复制input_pdf_path的内容到invalid_added.pdf 这一步可能报错！绘制新的一页基本不会报错 #复制input_pdf_path的内容到invalid_added.pdf 这一步可能报错！绘制新的一页基本不会报错
def insert_mark(input_pdf_path): def insert_mark(input_pdf_path):
@ -48,8 +48,10 @@ def insert_mark(input_pdf_path):
# 将内存中的PDF读入PyPDF2 # 将内存中的PDF读入PyPDF2
packet.seek(0) packet.seek(0)
new_pdf = PdfReader(packet) new_pdf = PdfReader(packet)
blank_page = new_pdf.pages[0] # blank_page = new_pdf.pages[0]
blank_page = copy.deepcopy(new_pdf.pages[0])
packet.truncate(0)
packet.seek(0)
# 将带有文本的空白页添加到写入器 # 将带有文本的空白页添加到写入器
pdf_writer.add_page(blank_page) pdf_writer.add_page(blank_page)
@ -92,10 +94,10 @@ def delete_mark(docx_path):
return new_file_path return new_file_path
if __name__ == '__main__': if __name__ == '__main__':
# input=r'D:\flask_project\flask_app\static\output\output1\90b073b5-b8fb-4b11-9962-10de7c3f3854\invalid_added.docx' # input=r'C:\Users\Administrator\Desktop\fsdownload\0bb9cf31-280c-4d96-bc21-0871ee7fd6df\tmp\ztbfile.pdf'
# input=r'C:\Users\Administrator\Desktop\new招标文件\output5\广水市公安局音视频监控系统设备采购项目_procurement.pdf' # input=r'C:\Users\Administrator\Desktop\new招标文件\output5\广水市公安局音视频监控系统设备采购项目_procurement.pdf'
# output=insert_mark(input) # output=insert_mark(input)
doc_path = r'D:\flask_project\flask_app\static\output\output1\90b073b5-b8fb-4b11-9962-10de7c3f3854\invalid_added.docx' doc_path = r'C:\Users\Administrator\Desktop\fsdownload\0bb9cf31-280c-4d96-bc21-0871ee7fd6df\tmp\invalid_added.docx'
res=delete_mark(doc_path) res=delete_mark(doc_path)
if res: if res:
print(res) print(res)

4
flask_app/货物标/截取pdf货物标版.py
View File

@ -612,10 +612,10 @@ if __name__ == "__main__":
logger = get_global_logger("123") logger = get_global_logger("123")
# input_path = r"C:\Users\Administrator\Desktop\new招标文件\货物标" # input_path = r"C:\Users\Administrator\Desktop\new招标文件\货物标"
# pdf_path = r"C:\Users\Administrator\Desktop\招标文件-采购类\2024-贵州-贵州医科大学附属医院导视系统零星制作安装项目.pdf" # pdf_path = r"C:\Users\Administrator\Desktop\招标文件-采购类\2024-贵州-贵州医科大学附属医院导视系统零星制作安装项目.pdf"
pdf_path=r"D:\flask_project\flask_app\static\output\output1\a43b4a55-06c1-465d-afb3-5844695ad9d8\ztbfile.pdf" pdf_path=r"C:\Users\Administrator\Desktop\fsdownload\0bb9cf31-280c-4d96-bc21-0871ee7fd6df\ztbfile.pdf"
# input_path = r"C:\Users\Administrator\Desktop\货物标\zbfiles\2-招标文件（广水市教育局封闭管理）.pdf" # input_path = r"C:\Users\Administrator\Desktop\货物标\zbfiles\2-招标文件（广水市教育局封闭管理）.pdf"
# pdf_path=r"C:\Users\Administrator\Desktop\文件解析问题\文件解析问题\1414cb9c-7bf4-401c-8761-2acde151b9c2\ztbfile.pdf" # pdf_path=r"C:\Users\Administrator\Desktop\文件解析问题\文件解析问题\1414cb9c-7bf4-401c-8761-2acde151b9c2\ztbfile.pdf"
output_folder = r"D:\flask_project\flask_app\static\output\output1\a43b4a55-06c1-465d-afb3-5844695ad9d8\tmp" output_folder = r"C:\Users\Administrator\Desktop\fsdownload\0bb9cf31-280c-4d96-bc21-0871ee7fd6df\tmp"
# output_folder = r"C:\Users\Administrator\Desktop\new招标文件\output2" # output_folder = r"C:\Users\Administrator\Desktop\new招标文件\output2"
selection = 5 # 例如：1 - 公告, 2 - 评标办法, 3 - 资格审查后缀有qualification1或qualification2（与评标办法一致） 4.投标人须知前附表part1 投标人须知正文part2 5-采购需求 selection = 5 # 例如：1 - 公告, 2 - 评标办法, 3 - 资格审查后缀有qualification1或qualification2（与评标办法一致） 4.投标人须知前附表part1 投标人须知正文part2 5-采购需求
generated_files = truncate_pdf_main_goods(pdf_path, output_folder, selection,logger) generated_files = truncate_pdf_main_goods(pdf_path, output_folder, selection,logger)