import fitz
p = r'C:\Users\12637\Desktop\hermes\焦煤\基础知识\焦煤交易手册.pdf'
doc = fitz.open(p)
print('pages', len(doc))
for i in range(min(5, len(doc))):
    text = doc.load_page(i).get_text('text')
    print(f'---PAGE {i+1}---')
    print(text[:4000])