Script/PaddleOCR_PDF.py

from pathlib import Path
from paddleocr import PPStructureV3
from pdf2image import convert_from_path
import os

# 输入 PDF 和输出路径
input_file = "C:/Users/zhens/Downloads/12.pdf"
output_path = Path("C:/Users/zhens/Downloads/output")
output_path.mkdir(parents=True, exist_ok=True)

# Step 1: 把 PDF 转为图片
pdf_pages = convert_from_path(input_file, dpi=300)
image_paths = []

for idx, page in enumerate(pdf_pages):
    img_path = output_path / f"page_{idx+1}.jpg"
    page.save(img_path, "JPEG")
    image_paths.append(str(img_path))

# Step 2: 使用 PPStructureV3 分析每页图像
pipeline = PPStructureV3()
markdown_list = []
markdown_images = []

for img in image_paths:
    output = pipeline.predict(img)
    for res in output:
        md_info = res.markdown
        markdown_list.append(md_info)
        markdown_images.append(md_info.get("markdown_images", {}))

# Step 3: 合并 Markdown 输出
markdown_texts = pipeline.concatenate_markdown_pages(markdown_list)
mkd_file_path = output_path / f"{Path(input_file).stem}.md"

with open(mkd_file_path, "w", encoding="utf-8") as f:
    f.write(markdown_texts)

# Step 4: 保存图像资源（markdown_images）
for item in markdown_images:
    if item:
        for path, image in item.items():
            file_path = output_path / path
            file_path.parent.mkdir(parents=True, exist_ok=True)
            image.save(file_path)