# -*- coding: utf-8 -*- """系统比对:报告实际指标行 vs v5 提取器白名单""" import fitz, re, sys sys.path.insert(0, r'C:/Users/liaoxg.DESKTOP-DGLN79N/.workbuddy/skills/bun-report-parser/scripts') import extract_full_report_v5 as v5 doc = fitz.open(r'E:/cfc/docs/参考资料/560822024-张学红.pdf') # 收集全文行 all_lines = [] for page in doc: for l in page.get_text().split('\n'): l = l.strip() if l: all_lines.append(l) # 动态抓取「名称 + 数值(+状态) + 范围」三元组行(不看白名单) status_words = ['正常', '过多', '轻度产气', '过低', '不足', '缺乏', '偏高', '注意'] found = {} # 名称 -> (值, 状态) i = 0 cur = None while i < len(all_lines): line = all_lines[i] # 形态A: 名称 数值 状态 范围 (同一行) m = re.match(r'^(.+?)\s+(\d+)\s+(' + '|'.join(status_words) + r')\s+(\d+-\d+)$', line) if m: found[m.group(1)] = (m.group(2), m.group(3), m.group(4)) i += 1 continue # 形态B: 数值 状态 范围 单独行(名称在上一行,可能多行) m2 = re.match(r'^(\d+)\s+(' + '|'.join(status_words) + r')\s+(\d+-\d+)$', line) if m2: # 向上找名称:跳过页码/空行,取最近的非数值行 name_parts = [] j = i - 1 while j >= 0 and len(name_parts) < 4: prev = all_lines[j] if re.match(r'^\d+/169$', prev) or not prev: j -= 1 continue if re.match(r'^\d', prev) or prev in status_words: break name_parts.insert(0, prev) # 名称行通常以括号结尾或较短;遇到表头关键词停止 if prev.endswith(')') or prev.endswith(')') or len(prev) < 12: break j -= 1 name = ''.join(name_parts) if name: found[name] = (m2.group(1), m2.group(2), m2.group(3)) i += 1 continue i += 1 # 与白名单比对 whitelists = { 'barrier': set(v5.KNOWN_BARRIER), 'scfa': set(v5.KNOWN_SCFA), 'neuro': set(v5.KNOWN_NEUROTRANSMITTER), } all_known = set() for s in whitelists.values(): all_known |= s print('===== 动态抓到的指标行(白名单外的=新增/漏采)=====') extra = [] for name, (val, st, rng) in sorted(found.items()): # 名称归一化后查白名单 hit = any(v5._npar(name).startswith(v5._npar(k)) or v5._npar(k).startswith(v5._npar(name)) for k in all_known) flag = 'OK ' if hit else 'NEW' if not hit: extra.append((name, val, st, rng)) print(f' [{flag}] {name} = {val} {st} ({rng})') print() print('===== 白名单外指标(需要补采)=====') for name, val, st, rng in extra: print(f' {name}: {val} {st} 范围{rng}')