https://github.com/RapidAI/RapidOCRPDF

# 基于rapidocr_onnxruntimepip 
# 基于rapidocr_openvinopip

from rapidocr_pdf import PDFExtracter
pdf_extracter = PDFExtracter()
pdf_path = 'tests/test_files/direct_and_image.pdf'
texts = pdf_extracter(pdf_path)
print(texts)

$ rapidocr_pdf -h
usage: rapidocr_pdf [-h] [-path FILE_PATH]
options:
-h, --help            show this help message and exit
-path FILE_PATH, --file_path FILE_PATH 
                       File path, PDF or images
$ rapidocr_pdf -path tests/test_files/direct_and_image.pdf

[
    ['0', '人之初，性本善。性相近，習(xí)相遠。', '0.8969868'],
        ['1', 'Men at their birth, are naturally good.', '0.8969868'],
]