# pdf_parser.py —— 合同 PDF 解析（文本层 + OCR 降级） from pathlib import Path import pdfplumber from PIL import Image import pytesseract  MIN_TEXT_LEN = 30  # 单页少于这么多字符，判定为扫描件  def extract_text(pdf_path: str) -> str:     text_parts = []     with pdfplumber.open(pdf_path) as pdf:         for i, page in enumerate(pdf.pages):             txt = page.extract_text() or ""             if len(txt.strip()) < MIN_TEXT_LEN:                 txt = ocr_page(pdf_path, i)  # 降级 OCR             text_parts.append(txt)     return "\n".join(text_parts)  def ocr_page(pdf_path: str, page_no: int) -> str:     img = Image.open(pdf_path)     # 真实场景用 fitz 按页转图更稳，这里示意     return pytesseract.image_to_string(img, lang="chi_sim+eng")  if __name__ == "__main__":     out = extract_text("sample_contract.pdf")     Path("data/contract.txt").write_text(out, encoding="utf-8")     print(f"✅ 提取文本 {len(out)} 字")