"""PDF OCR处理后上传到服务器展示""" import sys, os, re, json, base64 sys.path.insert(0, os.path.expanduser("~/gov_crawler/venv/lib/python3.9/site-packages")) import pymupdf from rapidocr_onnxruntime import RapidOCR PDF_DIR = os.path.expanduser("~/Nutstore Files/Downloads_nuc") OUTPUT_DIR = "/tmp/cceup_ocr_preview" def ocr_pdf_to_html(pdf_path): """OCR PDF并生成预览HTML""" fname = os.path.basename(pdf_path).replace('.pdf', '') doc = pymupdf.open(pdf_path) ocr = RapidOCR() pages_html = [] total_ocr_text = [] for i, page in enumerate(doc): # 渲染页面为图片（200dpi） pix = page.get_pixmap(dpi=200) img_bytes = pix.tobytes("png") img_b64 = base64.b64encode(img_bytes).decode() # OCR识别 temp_path = f"/tmp/ocr_preview_{os.getpid()}_{i}.png" with open(temp_path, "wb") as f: f.write(img_bytes) result, _ = ocr(temp_path) os.remove(temp_path) # 收集OCR文本 ocr_lines = [] if result: for box, text, score in result: if score > 0.8: ocr_lines.append(f"[{score:.2f}] {text}") page_text = '\n'.join(ocr_lines) total_ocr_text.append(page_text) pages_html.append(f'''

第 {i+1} 页

第{i+1}页 
{"
".join(ocr_lines) if ocr_lines else "(无识别结果)"}
''') doc.close() # 提取联系方式 full_text = '\n'.join(total_ocr_text) phones = re.findall(r'1[3-9]\d{9}', full_text) landlines = re.findall(r'0[1-9]\d{1,2}-?\d{7,8}', full_text) names = re.findall( r'(?:联系人|联\s*系\s*人|项目负责人|法人|法人代表|法定代表人|代表|业主|经办人|' r'负责人|项目经理|采购人|招标人|姓名|姓\s*名)[：:\s]*([\u4e00-\u9fff·]{2,6})', full_text ) # 黑名单过滤 blacklist = {'手机','代表','单位注册地址','为外资','姓名及号码','姓名'} names = [n for n in names if n not in blacklist and len(n) >= 2] contacts_html = '' if names or phones or landlines: contacts_html = f'''

📞 提取到的联系方式

姓名: {"、".join(set(names)) or "未识别"}

手机: {"、".join(set(phones)) or "未识别"}

座机: {"、".join(set(landlines)) or "未识别"}

''' html = f''' 📄 {html_escape(fname)}
← 返回列表

📄 {html_escape(fname)}

共 {len(pages_html)} 页
{contacts_html} {"".join(pages_html)}
''' return html, fname def html_escape(s): return s.replace('&', '&').replace('<', '<').replace('>', '>').replace('"', '"') if __name__ == "__main__": os.makedirs(OUTPUT_DIR, exist_ok=True) # 选5个样本 all_pdfs = sorted([f for f in os.listdir(PDF_DIR) if f.endswith('.pdf')]) # 避开头尾，选中间几个不同类型的 samples = [all_pdfs[2], all_pdfs[10], all_pdfs[20], all_pdfs[35], all_pdfs[55]] index_items = [] for fname in samples: fpath = os.path.join(PDF_DIR, fname) print(f"处理: {fname}...") html, short_name = ocr_pdf_to_html(fpath) safe_name = fname.replace('.pdf', '').replace('/', '_') out_path = os.path.join(OUTPUT_DIR, f"{safe_name}.html") with open(out_path, 'w', encoding='utf-8') as f: f.write(html) print(f" → {out_path}") index_items.append((safe_name, fname)) # 生成索引页 index_html = ''' 📄 CCEUP PDF OCR 预览

📄 CCEUP PDF OCR 预览

5 个样本 PDF — 左侧为原始页面截图，右侧为 RapidOCR 识别结果

''' with open(os.path.join(OUTPUT_DIR, "index.html"), 'w', encoding='utf-8') as f: f.write(index_html) print(f"\n完成！共 {len(samples)} 个PDF") print(f"预览文件: {OUTPUT_DIR}/index.html")