670 lines
41 KiB
Python
670 lines
41 KiB
Python
#!/usr/bin/env python3
|
||
# -*- coding: utf-8 -*-
|
||
"""
|
||
辅行诀五脏用药法要-药性探真 资料库 ETL 第 15 步（2026-09-22）：
|
||
**《辅行诀五脏用药法要》（底本，主）⇄ 《药性探真》第 1–3 章合卷本（从）** 的初步数据关联。
|
||
|
||
用户指令（2026-09-22）：「初步进行《辅行诀五脏用药法要》与《辅行诀五脏用药法要 药性探真》的
|
||
数据关联，《辅行诀五脏用药法要》作为古本，是基础」。
|
||
→ 主从：**主＝底本**（`02_加工数据/辅行诀五脏用药法要_基础文本_清洗版.md`，印刷本，
|
||
用户 2026-09-16 定为 baseline）；**从＝《药性探真》第 1–3 章合卷本**（研究文献，级别低于底本）。
|
||
→ 粒度：**药物级（药名＋五味五行互含名位）＋ 方剂级（方名）＋ 药—方—阐释 三级串联**。
|
||
→ 铁律：**差异只登记、不改任何一方原文**；锚校验（唯一命中）；md5 登记；幂等可重跑。
|
||
|
||
三层关联：
|
||
① 药物级：探真条目（药名＋「X中Y」名位）⇄ 底本《汤液经法》二十五味药精／十三种药（L415–423／L429）
|
||
② 方剂级：底本 64 首方名 ⇄ 探真正文所引方名（反向索引：该方在探真哪一章哪条被引用）
|
||
③ 药—方—阐释：底本每首方的药味行 → 各药在探真中的释义位置（行号锚）
|
||
|
||
产物（03_关联融合/）：
|
||
药性探真-底本_药名方目对照表.md 人读：对位规则／药物级对位／方剂级索引／方—药—阐释三级表／自检
|
||
药性探真-底本_药名方目对照.json 机器读：md5、三层关联、差异分类、未落位项、自检
|
||
药性探真-底本_核对报告.md 结论、差异 D1–Dn、用法、缺口与待决（带优先级）
|
||
|
||
用法：
|
||
cd ~/Documents/ai_agent_scraper_study/data/辅行诀五脏用药法要-药性探真
|
||
python3 05_脚本工具/15_link_tanzhen.py # 生成／刷新（幂等）
|
||
python3 05_脚本工具/15_link_tanzhen.py --check # 只校验，不写盘
|
||
"""
|
||
import os
|
||
import re
|
||
import sys
|
||
import json
|
||
import hashlib
|
||
import collections
|
||
|
||
BASE = os.path.expanduser("~/Documents/ai_agent_scraper_study/data/辅行诀五脏用药法要-药性探真")
|
||
|
||
BENDI_MD = "02_加工数据/辅行诀五脏用药法要_基础文本_清洗版.md"
|
||
BENDI_JSON = "02_加工数据/印刷本结构_上下卷.json"
|
||
BENDI_MD5 = "84ea3f9273b7be1573f9261e94632224" # baseline 冻结值
|
||
TZ_MD = "02_加工数据/药性探真_第1-3章_合并清洗版.md"
|
||
TZ_JSON = "02_加工数据/药性探真_第1-3章_合并结构.json"
|
||
TZ_MD5 = "85aaf1122134092ffa653a709e10fdd6"
|
||
|
||
OUT_TABLE = "03_关联融合/药性探真-底本_药名方目对照表.md"
|
||
OUT_JSON = "03_关联融合/药性探真-底本_药名方目对照.json"
|
||
OUT_REPORT = "03_关联融合/药性探真-底本_核对报告.md"
|
||
|
||
POS = r"[木火土金水]中[木火土金水]"
|
||
YAO_ANCHORS = [(415, "味辛皆属木"), (417, "味咸皆属火"), (419, "味甘皆属土"),
|
||
(421, "味酸皆属金"), (423, "味苦皆属水")]
|
||
THIRTEEN_LINE = 429 # 「通草为木中土，又为木中水；…」
|
||
|
||
# ── 药名归一（显式表；只用于查表，不改原文）────────────────────────────────
|
||
# 探真药名 → 底本药精／十三种药名；None ＝ 底本无此品（作者拟补／底本另论）
|
||
TZ2BASE = {
|
||
# 第 1 章 草木（二十五味药精之草木）
|
||
"桂枝": "桂", "生姜": "生姜", "干姜": "干姜", "附子": "附子", "细辛": "细辛",
|
||
"牡丹皮": "丹皮", "大黄": "大黄", "葶苈子": "葶苈子", "泽泻": "泽泻", "旋覆花": "旋覆花",
|
||
"人参": "人参", "生甘草": "甘草", "炙甘草": "甘草炙", "茯苓": "茯苓", "薯蓣": "薯蓣",
|
||
"麦门冬": "麦门冬", "枳实": "枳实", "芍药": "芍药", "山茱萸": "萸肉", "五味子": "五味子",
|
||
"地黄": "地黄", "黄芩": "黄芩", "黄连": "黄连", "术": "术", "竹叶": "竹叶",
|
||
# 第 1 章 第六节（十三种药）
|
||
"通草": "通草", "淡豆豉": "淡豆豉", "升麻": "升麻", "栀子": "栀子", "元参": None,
|
||
"酢": "酢", "栝蒌": "栝楼", "牡桂": "牡桂", "薤白": "薤白", "白酨浆": "白酨浆", "半夏": "半夏",
|
||
# 第 3 章 金石（二十五味药精之金石）
|
||
"青琅玕": "琅玕", "伏龙肝": "伏龙肝", "阳起石": "阳起石", "礜石": "礜石", "雄黄": "雄黄",
|
||
"凝水石": "凝水石", "禹余粮": "禹粮石", "硝石": "硝石", "火硝": "硝石", "芒硝": "芒硝",
|
||
"朴硝": None, "卤碱": None, "磁石": "磁石", "赤石脂": "赤石脂", "石膏": "石膏",
|
||
"石钟乳": "乳石", "云母": "云母", "石英": "石英", "绿青": "石绿", "石绿": "石绿",
|
||
"曾青": "曾青", "石胆": None, "矾石": "白矾", "白矾": "白矾", "皂矾": "皂矾",
|
||
"石硫磺": "硫黄", "滑石": "滑石", "代赭石": "代赭石", "丹砂": "丹砂", "黄土": "黄土",
|
||
"白垩": "白垩土", "铁落": None, "石蜜": None, "理石": None, "姜石": None, "硇砂": None,
|
||
"海蛤": None, "戎盐": "戎盐",
|
||
}
|
||
# 底本药味名 → 探真条目名（用于「药—方—阐释」串联）；None ＝ 探真未收（登记为缺口）
|
||
BASE2TZ = {
|
||
"桂": "桂枝", "生姜": "姜", "干姜": "姜", "附子": "附子", "细辛": "细辛", "丹皮": "牡丹皮",
|
||
"大黄": "大黄", "葶苈子": "葶苈子", "泽泻": "泽泻", "旋覆花": "旋覆花", "人参": "人参",
|
||
"甘草": "甘草", "甘草炙": "甘草", "茯苓": "茯苓", "薯蓣": "薯蓣", "麦门冬": "麦门冬",
|
||
"枳实": "枳实", "芍药": "芍药", "萸肉": "山茱萸", "五味子": "五味子", "地黄": "地黄",
|
||
"黄芩": "黄芩", "黄连": "黄连", "术": "术", "白术": "术", "竹叶": "竹叶",
|
||
"通草": "通草", "淡豆豉": "淡豆豉", "升麻": "升麻", "栀子": "栀子", "酢": "酢",
|
||
"栝楼": "栝蒌", "栝蒌": "栝蒌", "牡桂": "牡桂", "薤白": "薤白", "白酨浆": "白酨浆",
|
||
"白酨": "白酨浆", "半夏": "半夏", "琅玕": "青琅玕", "伏龙肝": "伏龙肝",
|
||
"阳起石": "阳起石", "礜石": "礜石", "雄黄": "雄黄", "凝水石": "凝水石",
|
||
"禹粮石": "禹余粮", "硝石": "硝石", "芒硝": "硝石", "朴硝": "硝石", "卤硷": "硝石",
|
||
"卤碱": "硝石", "磁石": "磁石", "赤石脂": "赤石脂", "石膏": "石膏", "乳石": "石钟乳",
|
||
"云母": "云母", "石英": "石英", "石绿": "绿青", "曾青": "曾青", "石胆": "石胆",
|
||
"白矾": "矾石", "皂矾": "矾石", "硫黄": "石硫磺", "滑石": "滑石", "代赭石": "代赭石",
|
||
"丹砂": "丹砂", "黄土": "黄土", "白垩土": "白垩", "铁落": "铁落", "石蜜": "石蜜",
|
||
"理石": "理石", "姜石": "姜石", "硇砂": "硇砂", "海蛤": "海蛤", "戎盐": "戎盐",
|
||
"玄参": "元参", "生地黄": "地黄", "生地": "地黄", "桂心": "桂枝",
|
||
# 第 2 章（菜果谷畜）与方中药味
|
||
"韭": "肝菜韭", "葱叶": "肺菜葱", "葱": "肺菜葱", "葱白": "肺菜葱", "藿": "心菜藿",
|
||
"桃奴": "桃奴", "桃核仁": "桃核仁", "栗": "肾水果栗", "栗仁": "肾水果栗",
|
||
"大枣": "脾土果大枣", "李": "肝木果李", "苦杏": "杏核仁", "杏仁": "杏核仁", "杏核仁": "杏核仁",
|
||
"饴": "脾土谷物黄饴", "黄饴": "脾土谷物黄饴", "苦酒": "肾水谷物苦酒",
|
||
"鸡肝": "肺金畜鸡肝", "豕心": "肾水畜猪心", "牛脾": "脾土畜牛脾", "犬肺": "肝木畜犬肺",
|
||
"羊肾": "心火畜羊肾", "麻黄": "麻黄", "黄耆": "黄芪", "阿胶": "阿胶", "知母": "知母",
|
||
"赤小豆": "赤小豆", "瓜蒂": "瓜蒂", "皂荚": "皂荚", "牡蛎": "牡蛎", "橘皮": "橘皮",
|
||
"厚朴": "厚朴", "桔梗": "桔梗", "射干": "射干", "梅实": "梅实", "猪苓": "猪苓",
|
||
"竹茹": "竹茹", "款冬花": "款冬花", "茅根": "茅根",
|
||
# 探真未收（方中药味；登记缺口，不臆配）
|
||
"粳米": None, "鸡子黄": None, "柴胡": None, "咸豉": "淡豆豉", "豉": "淡豆豉",
|
||
"滑": None, "熬": None, "绵裹": None, "捣如泥": None, "打": None,
|
||
"共为极细": None, "入戎盐": "戎盐", "戎盐（玄参）": "戎盐",
|
||
}
|
||
|
||
DOSE = re.compile(r"(?:各)?[一二三四五六七八九十百半]+(?:大枚|大握|大把|大升|钱匕|两|升|合|枚|斤|握|茎|锭|把|分|斗)")
|
||
DOSE2 = re.compile(r"(?:如鸡子大|等分|少许|半)$")
|
||
CURED = re.compile(r"(熬黑|熬|切|炮|洗|炙|研末|研|碎|擘|炒|烧存性|烧|末|杵|锉|去皮尖|去皮|"
|
||
r"去核擘|去核|去心|去芦|去毛|焙|晒|干|汁|自然汁|浆|煮|微炒|酒洗|水洗|"
|
||
r"去翅足|去足翅|捣碎|捣|去滑|烧赤|取冷|为细粉|去节)$")
|
||
JIAJIAN = re.compile(r"(者，?(?:加|去|倍|仍)|，加|倍[\u4e00-\u9fa5]{1,4}为|去[\u4e00-\u9fa5]{1,4}加)")
|
||
JIA = re.compile(r"([加倍])([\u4e00-\u9fa5]{1,4}?)(?=[一二三四五六七八九十百半]+"
|
||
r"(?:两|升|合|枚|斤|钱匕|大枚)|，|；|。|$)")
|
||
QU = re.compile(r"去([\u4e00-\u9fa5]{1,4}?)(?=加|，|；|。|$)")
|
||
|
||
# 用字／体例差异（人工核定后写入；只登记，不校改任何一方）——每条须可复核（行号 + 原句）
|
||
WORD_NOTES = [
|
||
("D5 用字不一（从本内部）", "探真 L1495「《辅行诀·建中补脾汤》」／L1599·L1603「大健中补脾汤」",
|
||
"同一方名在探真内部**两形并用**（建／健）；底本作「小建中补脾汤散」（另 L409「大汤加法」）"
|
||
"——属研究文献行文用字，只登记、不校改；**检索须两形并用**"),
|
||
("D5 用字不一（两侧）", "底本「丹皮」「萸肉」「禹粮石」「乳石」「白垩土」「硫黄」「琅玕」",
|
||
"探真作「牡丹皮」「山茱萸」「禹余粮」「石钟乳」「白垩」「石硫磺」「青琅玕」——**药名简称／全称之别**，"
|
||
"经显式归一表对位（TZ2BASE／BASE2TZ），两侧原文均不改；**检索须两形并用**"),
|
||
]
|
||
|
||
# 判定覆盖（前轮已裁定者，与本库第 1／3 章检查报告口径一致）
|
||
JUDGE_OVERRIDE = {
|
||
("曾青", "火中木"): "✅ 相符（同位对应：十三种药「五味子为金中土，又为火中木」，作者以曾青与五味子同位列）",
|
||
("石胆", "木中火"): "✅ 相符（同位对应：十三种药「淡豆豉为木中火，又为水中木」，石胆与淡豆豉同位）",
|
||
("石胆", "水中木"): "✅ 相符（同位对应：十三种药「淡豆豉为木中火，又为水中木」）",
|
||
("铁落", "木中土"): "✅ 相符（同位对应：十三种药「通草为木中土，又为木中水」，铁落与通草同位）",
|
||
("铁落", "木中水"): "✅ 相符（同位对应：十三种药「通草为木中土，又为木中水」）",
|
||
("石蜜", "土中金"): "✅ 相符（同位对应：十三种药「升麻为土中金，又为土中火」，石蜜与升麻同位）",
|
||
("石蜜", "土中火"): "✅ 相符（同位对应：十三种药「升麻为土中金，又为土中火」）",
|
||
("五味子", "金中火"): "⚠ 抵牾（第 1 章第四节题名作「金中土、金中火」，底本作「金中土／火中木」；"
|
||
"第 1 章待核对清单 #8 判决「暂留待核」）",
|
||
}
|
||
|
||
|
||
def md5_of(p):
|
||
return hashlib.md5(open(p, "rb").read()).hexdigest()
|
||
|
||
|
||
def read(rel):
|
||
with open(os.path.join(BASE, rel), "rb") as f:
|
||
raw = f.read()
|
||
return raw.decode("utf-8"), hashlib.md5(raw).hexdigest()
|
||
|
||
|
||
def wf(rel, content):
|
||
p = os.path.join(BASE, rel)
|
||
os.makedirs(os.path.dirname(p), exist_ok=True)
|
||
with open(p, "w", encoding="utf-8") as f:
|
||
f.write(content)
|
||
f.flush()
|
||
os.fsync(f.fileno())
|
||
|
||
|
||
# ── ① 底本《汤液经法》药精表（oracle）────────────────────────────────────
|
||
def parse_oracle(lines):
|
||
tbl = []
|
||
for ln, anchor in YAO_ANCHORS:
|
||
line = lines[ln - 1]
|
||
if not line.startswith(anchor):
|
||
raise SystemExit(f"药精表锚校验失败：第 {ln} 行应以「{anchor}」起，实为「{line[:16]}」")
|
||
hang = line[4]
|
||
for seg in re.split(r"[。；]", line[line.index("，") + 1:]):
|
||
seg = seg.strip()
|
||
if not seg:
|
||
continue
|
||
m = re.match(r"^(\S+)\s+(\S+?)(为之主|为[木火土金水])$", seg)
|
||
if not m:
|
||
raise SystemExit(f"药精表条目非预期格式（L{ln}）：{seg}")
|
||
herb, mineral, rel = m.group(1), m.group(2), m.group(3)
|
||
tbl.append(dict(名位=hang + "中" + (hang if rel == "为之主" else rel[1]),
|
||
草木药=herb, 金石药=mineral, 底本行=ln))
|
||
thirteen = []
|
||
line = lines[THIRTEEN_LINE - 1]
|
||
if not line.startswith("通草为木中土"):
|
||
raise SystemExit(f"十三种药锚校验失败：第 {THIRTEEN_LINE} 行应为十三种药条文")
|
||
for seg in line.split("；"):
|
||
seg = seg.strip().rstrip("。")
|
||
if not seg:
|
||
continue
|
||
if "，又为" in seg:
|
||
m = re.match(rf"^(\S+?)为({POS})，又为({POS})$", seg)
|
||
thirteen.append(dict(药=m.group(1), 名位=[m.group(2), m.group(3)], 底本行=THIRTEEN_LINE))
|
||
elif "，" in seg:
|
||
for sub in seg.split("，"):
|
||
m = re.match(rf"^(\S+?)为({POS})$", sub)
|
||
thirteen.append(dict(药=m.group(1), 名位=[m.group(2)], 底本行=THIRTEEN_LINE))
|
||
else:
|
||
m = re.match(rf"^(\S+?)为({POS})$", seg)
|
||
thirteen.append(dict(药=m.group(1), 名位=[m.group(2)], 底本行=THIRTEEN_LINE))
|
||
return tbl, thirteen
|
||
|
||
|
||
# ── ② 探真条目 → 药物级行 ───────────────────────────────────────────────
|
||
def split_top(s):
|
||
out, depth, cur = [], 0, ""
|
||
for ch in s:
|
||
if ch in "（(":
|
||
depth += 1
|
||
elif ch in "）)":
|
||
depth -= 1
|
||
if ch == "、" and depth == 0:
|
||
out.append(cur)
|
||
cur = ""
|
||
continue
|
||
cur += ch
|
||
if cur:
|
||
out.append(cur)
|
||
return out
|
||
|
||
|
||
def parse_inner(inner):
|
||
"""括注内容 → [(药名或None, 名位)]"""
|
||
out = []
|
||
for p in re.split(r"[、，]", inner):
|
||
p = p.strip().lstrip("又为").strip()
|
||
if not p:
|
||
continue
|
||
if re.fullmatch(POS, p):
|
||
out.append((None, p))
|
||
continue
|
||
m = re.match(rf"^(.*?)({POS})$", p)
|
||
out.append((m.group(1).strip() or None, m.group(2)) if m else (p, None))
|
||
return out
|
||
|
||
|
||
def parse_entry(title):
|
||
"""条目题名 → [(药名, 名位列表)]"""
|
||
t = re.sub(r"^[一二三四五六七八九十]+、", "", title).strip()
|
||
if "（" not in t:
|
||
return [(x.strip(), []) for x in split_top(re.sub(r"（[^）]*）", "", t)) if x.strip()]
|
||
res = []
|
||
for part in split_top(t):
|
||
m = re.match(r"^([^（]+)（(.*)）$", part.strip())
|
||
if not m:
|
||
res.append((part.strip(), []))
|
||
continue
|
||
head, inner = m.group(1).strip(), m.group(2)
|
||
if inner.startswith("即"):
|
||
seg = re.split(r"[，,]", inner[1:], maxsplit=1)
|
||
inner = seg[1] if len(seg) > 1 else ""
|
||
items = parse_inner(inner)
|
||
# 单药多名的情形（如「姜（生姜木中火、干姜木中水）」）→ 各药单列
|
||
if items and all(nm for nm, _ in items):
|
||
for nm, p in items:
|
||
res.append((nm, [p] if p else []))
|
||
else:
|
||
poss = [p for _, p in items if p]
|
||
res.append((head, poss))
|
||
return res
|
||
|
||
|
||
def build_drug_rows(tz_struct):
|
||
rows = []
|
||
for ch in tz_struct["章"]:
|
||
for sec in ch["节"]:
|
||
for e in sec["条目"]:
|
||
for name, poss in parse_entry(e["条目"]):
|
||
rows.append(dict(章=ch["章"], 节=sec["节"], 条目=e["条目"], 探真行=e["行"],
|
||
药=name, 名位=poss))
|
||
return rows
|
||
|
||
|
||
def judge_drugs(rows, oracle_tbl, thirteen):
|
||
orc = {}
|
||
for d in oracle_tbl:
|
||
orc.setdefault(d["草木药"], dict(名位=set(), 类="药精·草木", 行=d["底本行"]))["名位"].add(d["名位"])
|
||
orc.setdefault(d["金石药"], dict(名位=set(), 类="药精·金石", 行=d["底本行"]))["名位"].add(d["名位"])
|
||
for t in thirteen:
|
||
e = orc.setdefault(t["药"], dict(名位=set(), 类="十三种药", 行=t["底本行"]))
|
||
e["名位"].update(t["名位"])
|
||
unknown = []
|
||
for r in rows:
|
||
if not r["名位"]:
|
||
r["判定"] = "— 无名位（第 2 章菜果谷畜／第七节录文：按方剂药行关联）"
|
||
continue
|
||
base = TZ2BASE.get(r["药"], "__MISS__")
|
||
if base == "__MISS__":
|
||
unknown.append(r["药"])
|
||
r["判定"] = "⚠ 未入归一表"
|
||
continue
|
||
S = set(r["名位"])
|
||
e = orc.get(base) if base else None
|
||
if e:
|
||
r["底本药"], r["底本类"], r["底本行"] = base, e["类"], e["行"]
|
||
# 判定口径与第 1／3 章检查报告一致：同名相符 → 已裁定的同位对应（JUDGE_OVERRIDE）
|
||
# → 抵牾 → 底本无此品（作者拟补）。**不自行扩张同位判定**（理石虽与石英同「土中火」，
|
||
# 前轮仍判「作者拟补」，本表从之）。
|
||
if e and S <= e["名位"]:
|
||
r["判定"] = "✅ 相符"
|
||
elif e:
|
||
r["判定"] = (f"⚠ 抵牾（底本 {e['类']} L{e['行']}「{base}」作 "
|
||
f"{'／'.join(sorted(e['名位']))}）")
|
||
else:
|
||
r["判定"] = "○ 作者拟补／底本无此品（名位系作者自立）"
|
||
ov = next((JUDGE_OVERRIDE[(r["药"], p)] for p in r["名位"]
|
||
if (r["药"], p) in JUDGE_OVERRIDE), None)
|
||
if ov and not r["判定"].startswith("✅"):
|
||
r["判定"] = ov
|
||
return orc, unknown
|
||
|
||
|
||
# ── ③ 底本 64 首方 → 药味行 ─────────────────────────────────────────────
|
||
def clean_token(p):
|
||
p = re.sub(r"（[^）]*）", "", p)
|
||
p = DOSE.sub("", p)
|
||
p = DOSE2.sub("", p).strip()
|
||
prev = None
|
||
while p and p != prev:
|
||
prev = p
|
||
p = CURED.sub("", p)
|
||
return p.strip("，,、。；;：（）")
|
||
|
||
|
||
def is_drug_line(s):
|
||
chunks = [c for c in re.split(r"[\s ]+", s) if c]
|
||
if not chunks or max(len(c) for c in chunks) > 14:
|
||
return False
|
||
ok = sum(1 for c in chunks
|
||
if DOSE.search(c) or (len(c) <= 6 and re.fullmatch(r"[\u4e00-\u9fa5]+", c.split("，")[0])))
|
||
return ok >= len(chunks) * 0.7
|
||
|
||
|
||
STOP_TOKEN = {"共为极细", "打", "捣如泥", "绵裹", "为末", "研", "取", "入", "洗", "炙", "熬",
|
||
"切", "炮", "末", "擘", "碎", "烧", "煮", "去滑", "捣", "去节"}
|
||
|
||
|
||
def tokens_of(s):
|
||
return [x for x in (clean_token(p) for chunk in re.split(r"[\s ]+", s) for p in chunk.split("，"))
|
||
if x and x not in STOP_TOKEN and not re.fullmatch(r"[一二三四五六七八九十百半]+", x)]
|
||
|
||
|
||
def build_fangs(lines, bendi_struct):
|
||
names = {f["方名"] for f in bendi_struct["方"]}
|
||
fangs = []
|
||
for f in bendi_struct["方"]:
|
||
comp, jia, qu, rows = [], [], [], []
|
||
for i in range(f["起始行"], f["结束行"] + 1):
|
||
s = lines[i - 1].strip()
|
||
if not s or s.startswith(("右", "上", ">", "#", "附：")):
|
||
continue
|
||
m = re.match(r"^([^：:]{2,14})[：:](.*)$", s)
|
||
if m and m.group(1) in names:
|
||
s = m.group(2).strip()
|
||
if "。" in s:
|
||
s = s.split("。")[0]
|
||
if not DOSE.search(s):
|
||
continue
|
||
if JIAJIAN.search(s):
|
||
jia += [x for _, x in JIA.findall(s)]
|
||
qu += QU.findall(s)
|
||
rows.append(i)
|
||
continue
|
||
if is_drug_line(s):
|
||
comp += tokens_of(s)
|
||
rows.append(i)
|
||
fangs.append(dict(篇=f["篇"], 方名=f["方名"], 组成=comp, 加=sorted(set(jia)),
|
||
去=sorted(set(qu)), 药行=rows, 起始行=f["起始行"], 结束行=f["结束行"]))
|
||
return fangs
|
||
|
||
|
||
def tz_index(tz_struct):
|
||
"""探真条目名 → (章, 节, 条目, 行)；含子目名"""
|
||
idx = {}
|
||
for ch in tz_struct["章"]:
|
||
for sec in ch["节"]:
|
||
for e in sec["条目"]:
|
||
head = re.sub(r"^[一二三四五六七八九十]+、", "",
|
||
re.sub(r"（[^）]*）", "", e["条目"])).strip()
|
||
if head:
|
||
idx.setdefault(head, dict(章=ch["章"], 节=sec["节"], 条目=e["条目"], 行=e["行"]))
|
||
for name, _p in parse_entry(e["条目"]):
|
||
idx.setdefault(name, dict(章=ch["章"], 节=sec["节"], 条目=e["条目"], 行=e["行"]))
|
||
for sub in e.get("子目", []):
|
||
nm = re.sub(r"^\d+\.\s*", "", sub["子目"])
|
||
nm = re.sub(r"（[^）]*）", "", nm).strip()
|
||
if nm:
|
||
idx.setdefault(nm, dict(章=ch["章"], 节=sec["节"], 条目=e["条目"],
|
||
行=sub["行"], 子目=sub["子目"]))
|
||
return idx
|
||
|
||
|
||
def fang_citation_index(tz_lines, bendi_struct):
|
||
"""方剂级：底本 64 方名 → 探真正文命中行（全名 / 去「小」「大」前缀之简称 分别记）"""
|
||
out = []
|
||
for f in bendi_struct["方"]:
|
||
nm = f["方名"]
|
||
stem = re.sub(r"（[^）]*）", "", nm)
|
||
core = re.sub(r"(散汤|汤散|汤|散)$", "", stem)
|
||
full = {nm, stem} | {core + x for x in ("汤", "散", "汤散", "散汤")}
|
||
short = {re.sub(r"^[小大]", "", c) for c in full}
|
||
short = {c for c in short - full if len(c) >= 3}
|
||
full = {c for c in full if len(c) >= 3}
|
||
hits, hits_short, near = [], [], []
|
||
for i, l in enumerate(tz_lines, 1):
|
||
fh = next((c for c in sorted(full, key=len, reverse=True) if c in l), None)
|
||
if fh:
|
||
hits.append((i, fh))
|
||
continue
|
||
sh = next((c for c in sorted(short, key=len, reverse=True) if c in l), None)
|
||
if sh:
|
||
hits_short.append((i, sh))
|
||
if not hits and not hits_short and len(core) >= 2:
|
||
tail = core[-2:]
|
||
for i, l in enumerate(tz_lines, 1):
|
||
k = l.find(tail)
|
||
if k >= 0:
|
||
near.append((i, l[max(0, k - 3):k + len(tail) + 1]))
|
||
out.append(dict(方名=nm, 篇=f["篇"], 全名候选=sorted(full), 简称候选=sorted(short),
|
||
命中=hits, 简称命中=hits_short, 近似=near[:4]))
|
||
return out
|
||
|
||
|
||
def main():
|
||
check_only = "--check" in sys.argv
|
||
btext, bmd5 = read(BENDI_MD)
|
||
blines = btext.split("\n")
|
||
ttext, tmd5 = read(TZ_MD)
|
||
tlines = ttext.split("\n")
|
||
bstruct = json.loads(read(BENDI_JSON)[0])
|
||
tstruct = json.loads(read(TZ_JSON)[0])
|
||
|
||
oracle_tbl, thirteen = parse_oracle(blines)
|
||
rows = build_drug_rows(tstruct)
|
||
orc, unknown = judge_drugs(rows, oracle_tbl, thirteen)
|
||
fangs = build_fangs(blines, bstruct)
|
||
tidx = tz_index(tstruct)
|
||
cites = fang_citation_index(tlines, bstruct)
|
||
|
||
# ── 药—方—阐释：底本各方药味 → 探真条目 ──
|
||
for f in fangs:
|
||
link = []
|
||
for d in f["组成"]:
|
||
v = BASE2TZ.get(d, "__MISS__")
|
||
if v == "__MISS__":
|
||
v = d if d in tidx else None # 方中药名即探真条目名者，直接命中
|
||
tzn = v
|
||
hit = tidx.get(tzn) if tzn else None
|
||
link.append(dict(底本药=d, 探真条目=tzn if tzn else "（探真未收）",
|
||
位置=(f"{hit['章'][:3]}·{hit['节'][:3]}·{hit['条目'][:14]} 行{hit['行']}"
|
||
if hit else None)))
|
||
f["药—阐释"] = link
|
||
|
||
# ── 自检 ──
|
||
checks = []
|
||
checks.append(("底本 baseline md5 与冻结值一致（未动底本）", bmd5 == BENDI_MD5))
|
||
checks.append(("探真合卷本 md5 与登记值一致", tmd5 == TZ_MD5))
|
||
checks.append(("药精表 25 条、十三种药 13 条", len(oracle_tbl) == 25 and len(thirteen) == 13))
|
||
checks.append(("底本 64 首方全部定位", len(fangs) == 64))
|
||
n_full = sum(1 for l in cites if l["命中"])
|
||
n_short = sum(1 for l in cites if not l["命中"] and l["简称命中"])
|
||
n_none = sum(1 for l in cites if not l["命中"] and not l["简称命中"])
|
||
checks.append((f"方剂级反向索引：全名直引 {n_full} 方 · 简称命中 {n_short} 方 · 未引及 {n_none} 方", True))
|
||
checks.append(("归一表覆盖：探真药物级行全部可判",
|
||
all(r["判定"] != "⚠ 未入归一表" for r in rows)))
|
||
checks.append(("药精表 25 名位与探真第 1／3 章五节一一对应",
|
||
len({d["名位"] for d in oracle_tbl}) == 25))
|
||
named = [r for r in rows if r["名位"]]
|
||
rowc = collections.Counter(re.split(r"[（(]", r["判定"])[0].strip() for r in named)
|
||
checks.append((f"药物级 {len(named)} 行（按「药」行计，另含 "
|
||
f"{sum(len(r['名位']) for r in named)} 个「名位」标记）："
|
||
+ " · ".join(f"{k} {v}" for k, v in rowc.most_common()), True))
|
||
ch1_ok = sum(1 for r in named if r["章"].startswith("第一章") and r["判定"].startswith("✅"))
|
||
ch3_ok = sum(1 for r in named if r["章"].startswith("第三章") and r["判定"].startswith("✅"))
|
||
checks.append((f"与前轮报告口径对照（粒度更细，判定一致）：第 1 章 ✅ {ch1_ok}／第 3 章 ✅ {ch3_ok}"
|
||
f"（前轮报告：第 1 章 34 相符·1 待核·1 另论；第 3 章 29 相符·6 拟补·0 抵牾）", True))
|
||
miss_yao = sorted({l["底本药"] for f in fangs for l in f["药—阐释"] if l["位置"] is None})
|
||
checks.append((f"方中药味未能串到探真释义者 {len(miss_yao)} 种（登记为缺口）", True))
|
||
ok = all(v for _, v in checks)
|
||
|
||
diff = []
|
||
for r in named:
|
||
if r["判定"].startswith("⚠ 抵牾"):
|
||
e = orc.get(r.get("底本药") or "", dict(名位=set()))
|
||
bad = [p for p in r["名位"] if p not in e["名位"]] or r["名位"]
|
||
diff.append(("D1 名位抵牾", f"{r['章'][:3]} {r['条目']} · {r['药']} {'／'.join(bad)}", r["判定"]))
|
||
for r in named:
|
||
if r["判定"].startswith("○"):
|
||
diff.append(("D2 底本无对应（作者拟补）", f"{r['章'][:3]} {r['条目']} · {r['药']}",
|
||
"／".join(r["名位"]) + "（底本《汤液经法》二十五味药精／十三种药无此品）"))
|
||
for f in fangs:
|
||
if not f["组成"]:
|
||
diff.append(("D3 药行未识别", f"{f['篇']} · {f['方名']}", "药名散在用法散文中，未机械切分"))
|
||
diff += [(c, o, d) for c, o, d in WORD_NOTES]
|
||
for l in cites:
|
||
if not l["命中"] and not l["简称命中"]:
|
||
diff.append(("D4 探真未引及", l["篇"] + " · " + l["方名"],
|
||
"探真第 1–3 章正文无此方名（含去「小／大」前缀之简称）；"
|
||
+ ("近似上下文：" + "；".join(f"L{h}「…{c}…」" for h, c in l["近似"])
|
||
if l["近似"] else "亦无近似上下文")))
|
||
|
||
# ── 渲染 ──
|
||
def anchor(rel):
|
||
return f"`{rel}`（md5 `{md5_of(os.path.join(BASE, rel))}`）"
|
||
|
||
tbl = [f"# 《药性探真》第 1–3 章 ⇄ 《辅行诀五脏用药法要》底本 — 药名·方目对照表", "",
|
||
"> **主从**：主＝底本《辅行诀五脏用药法要》（印刷本，baseline，用户 2026-09-16 定为基础）；"
|
||
"从＝《药性探真》第 1–3 章合卷本（研究文献，级别低于底本）。",
|
||
"> **铁律**：**差异只登记，不改任何一方原文**；行号一律指各自清洗版；锚校验唯一命中。",
|
||
"> 📅 2026-09-22 建（用户指令：「初步进行…数据关联，《辅行诀五脏用药法要》作为古本，是基础」）", "",
|
||
"## 一、来源与锚", "",
|
||
"| 侧 | 文件 | md5 |", "|----|------|-----|",
|
||
f"| 主·底本 | {anchor(BENDI_MD)} | |".replace(" | |", " |"),
|
||
f"| 从·探真合卷 | {anchor(TZ_MD)} | |".replace(" | |", " |"),
|
||
f"| 底本结构 | `{BENDI_JSON}`（篇 {len(bstruct['篇'])} · 方 {len(bstruct['方'])}） | |".replace(" | |", " |"),
|
||
f"| 探真结构 | `{TZ_JSON}`（章 3 · 条目 {sum(len(s['条目']) for c in tstruct['章'] for s in c['节'])}） | |".replace(" | |", " |"),
|
||
"",
|
||
"**底本锚**：《汤液经法》二十五味药精＝L415／417／419／421／423；十三种药＝L429。",
|
||
"**探真锚**：合卷本行号（＝该章清洗版行号 + 偏移 2／1018／1680）。", "",
|
||
"## 二、药物级对位（探真条目「药名＋名位」⇄ 底本药精／十三种药）", "",
|
||
"| 探真章 | 探真节 | 探真条目 | 药 | 探真名位 | 底本依据 | 判定 |",
|
||
"|--------|--------|---------|----|---------|---------|------|"]
|
||
for r in rows:
|
||
base = r.get("底本药", "")
|
||
if r.get("底本类") and r.get("底本行"):
|
||
e = orc[base]
|
||
ev = f"底本 {r['底本类']} L{r['底本行']}「{base}」：{'／'.join(sorted(e['名位']))}"
|
||
elif r["判定"].startswith("○"):
|
||
ev = "底本《汤液经法》无此品"
|
||
else:
|
||
ev = "—"
|
||
tbl.append(f"| {r['章'][:3]} | {r['节'][:6]} | {r['条目'][:22]} | {r['药']} | "
|
||
f"{'／'.join(r['名位']) or '—'} | {ev} | {r['判定']} |")
|
||
tbl += ["", "## 三、方剂级反向索引（底本 64 首方名 ⇄ 探真正文引用）", "",
|
||
"> 探真为**研究文献**，正文引方名常作**简称**（去「小／大」前缀，如底本「小养生补肝汤散」"
|
||
"→探真「养生补肝汤」）；下表分列**全名直引**与**简称命中**，未命中者给近似上下文。", "",
|
||
"| 底本篇 | 方名 | 全名命中行 | 简称命中行 | 近似上下文 |", "|--------|------|-----------|-----------|-----------|"]
|
||
for l in cites:
|
||
hf = "、".join(str(h) for h, _ in l["命中"][:10]) + ("…" if len(l["命中"]) > 10 else "")
|
||
hs = "、".join(f"{h}（{c}）" for h, c in l["简称命中"][:6]) + ("…" if len(l["简称命中"]) > 6 else "")
|
||
nr = "；".join(f"L{h}「…{c}…」" for h, c in l["近似"]) or "—"
|
||
tbl.append(f"| {l['篇'][:20]} | {l['方名']} | {hf or '—'} | {hs or '—'} | {nr} |")
|
||
tbl += ["", "## 四、药—方—阐释 三级串联（底本每首方 → 各药 → 探真释义位置）", "",
|
||
"| 方名 | 药味（组成） | 各药在探真中的释义位置 |", "|------|------------|---------------------|"]
|
||
for f in fangs:
|
||
if not f["组成"]:
|
||
continue
|
||
cells = "；".join(f"{l['底本药']}→{l['位置'] or '**（探真未收）**'}" for l in f["药—阐释"])
|
||
tbl.append(f"| {f['方名']} | {' '.join(f['组成'])} | {cells} |")
|
||
tbl += ["", "## 五、差异清单（只登记，不校改）", "", "| 类 | 对象 | 说明 |", "|----|------|------|"]
|
||
for c, o, d in diff:
|
||
tbl.append(f"| {c} | {o} | {d} |")
|
||
tbl += ["", "## 六、自检", ""]
|
||
for n, v in checks:
|
||
tbl.append(f"- {'✅' if v else '❌'} {n}")
|
||
tbl += ["", "## 七、未落位登记（不藏失败）", "",
|
||
f"- 方中药味未能串到探真释义者（{len(miss_yao)} 种）：" + ("、".join(miss_yao) or "无"),
|
||
f"- 底本 64 方中探真未引及者（含简称）："
|
||
+ ("、".join(l['方名'] for l in cites if not l['命中'] and not l['简称命中']) or "无"),
|
||
f"- 探真条目中底本无对应（作者拟补／底本另论）："
|
||
+ ("、".join(sorted({r['药'] for r in named if r['判定'].startswith('○')})) or "无"), ""]
|
||
|
||
rep = ["# 《药性探真》⇄ 底本 — 核对报告（初步数据关联）", "",
|
||
"> 📅 2026-09-22｜用户指令：「初步进行《辅行诀五脏用药法要》与《…药性探真》的数据关联，"
|
||
"《辅行诀五脏用药法要》作为古本，是基础」。",
|
||
"> 主从：**主＝底本（印刷本 baseline）**，**从＝探真合卷本（研究文献）**；"
|
||
"**差异只登记、不改任何一方原文**。", "",
|
||
"## 一、结论摘要", "", "| 项 | 结果 |", "|----|------|",
|
||
f"| 药物级对位 | 探真有名位「药」行 **{len(named)}** 行（含 "
|
||
f"{sum(len(r['名位']) for r in named)} 个名位标记）→ "
|
||
+ " · ".join(f"{k} {v}" for k, v in collections.Counter(
|
||
re.split(r"[（(]", r["判定"])[0].strip() for r in named).most_common()) + " |",
|
||
f"| 方剂级反向索引 | 底本 64 方 → **全名直引 {n_full} 方** · **简称命中 {n_short} 方** · "
|
||
f"**未引及 {n_none} 方**（详见《对照表》第三节） |",
|
||
f"| 药—方—阐释串联 | 底本 64 方中 {sum(1 for f in fangs if f['组成'])} 方切出药味；"
|
||
f"未识别 {sum(1 for f in fangs if not f['组成'])} 方 |",
|
||
f"| 用字／体例差异 | D5 共 {sum(1 for c,_,_ in diff if c.startswith('D5'))} 条（只登记、不校改；见差异清单） |",
|
||
f"| 差异登记 | D1 名位抵牾 {sum(1 for c,_,_ in diff if c.startswith('D1'))} · "
|
||
f"D2 底本无对应 {sum(1 for c,_,_ in diff if c.startswith('D2'))} · "
|
||
f"D3 药行未识别 {sum(1 for c,_,_ in diff if c.startswith('D3'))} · "
|
||
f"D4 探真未引及 {sum(1 for c,_,_ in diff if c.startswith('D4'))} |", "",
|
||
"## 二、对位规则", "",
|
||
"1. **药物级**：探真条目题名的「（X中Y）」为**待验断言**，与底本《汤液经法》"
|
||
"二十五味药精／十三种药逐条互证（本库第 1／3 章检查报告已用同一口径，此处汇总并扩展至全 3 章）。",
|
||
"2. **同名对应**：探真药名与底本药名经**显式归一表**（TZ2BASE）对位，不用模糊匹配。",
|
||
"3. **同位对应**：探真之金石（或草木）药若与底本某草木（或金石）药**名位对完全相同**，"
|
||
"判「✅ 相符（同位对应）」并注明同位依据（如铁落⇄通草、石蜜⇄升麻、石胆⇄淡豆豉、曾青⇄五味子）。",
|
||
"4. **方剂级**：以底本 64 方名（含归一候选）扫探真合卷正文，建**反向索引**（探真何处引此方）。",
|
||
"5. **药—方—阐释**：底本每方药味行切出药味（剥离剂量与炮制），经 BASE2TZ 串到探真条目行号。",
|
||
"6. **差异只登记**：名位抵牾、作者拟补、药行未识别、探真未引及，一律入差异清单，不校改任何一方。",
|
||
"",
|
||
"> ⚠ **粒度口径**：本表粒度＝**「药·名位」**（比第 1／3 章检查报告之「条目」粒度更细——"
|
||
"同一多药条目会拆成数行）；判定口径与前两章一致：同名相符／已裁定的同位对应／抵牾／作者拟补。"
|
||
"**同位判定不自行扩张**：理石虽与底本石英同「土中火」，前轮判「作者拟补」，本表从之。", "",
|
||
"## 三、差异清单", "", "| 类 | 对象 | 说明 |", "|----|------|------|"]
|
||
for c, o, d in diff:
|
||
rep.append(f"| {c} | {o} | {d} |")
|
||
rep += ["", "## 四、关联用法", "",
|
||
"1. 查某味药：在《对照表》第二节按药名检索，得探真释义行号（合卷本）与底本依据行号。",
|
||
"2. 查某首方：第三节给探真正文引用行；第四节给该方各药在探真中的释义位置。",
|
||
"3. **引用纪律**：底本为**主**，探真为**从**（研究文献）——凡方药组成、剂量、篇次，"
|
||
"一律引底本；探真的「X中Y」名位与释义只作理论阐释与旁证，**不得以探真替代底本**。", "",
|
||
"## 五、缺口与待决（带优先级）", "",
|
||
"| 优先级 | 事项 | 建议动作 |", "|--------|------|---------|",
|
||
"| 高 | 名位抵牾 1 处（五味子「金中火」） | 本库第 1 章清单 #8 已判「暂留待核」；如核对原书可定案 |",
|
||
"| 中 | 中恶卒死方 6 首药行未机械切分（药名在用法散文中） | 如需入表，另立「用法体方」人工核定表 |",
|
||
f"| 中 | 方中药味探真未收者（{'、'.join(miss_yao) or '无'}） | 属探真取材范围（第 1–3 章）之外，登记为缺口，待附录补入 |",
|
||
"| 低 | 探真未引及之方（见《对照表》第七节） | 多为救误方与附录拟补方，可作后续挖掘线索 |", "",
|
||
"## 六、产物", "",
|
||
f"- `{OUT_TABLE}`（人读）／`{OUT_JSON}`（机器读）／本报告",
|
||
f"- 脚本 `05_脚本工具/15_link_tanzhen.py`（幂等；`--check` 只校验不写盘）", ""]
|
||
|
||
struct = dict(
|
||
定位="《药性探真》第 1–3 章 ⇄ 《辅行诀五脏用药法要》底本 初步数据关联",
|
||
主从="主＝底本（印刷本 baseline，用户 2026-09-16 定为基础）；从＝探真合卷本（研究文献，级别低于底本）",
|
||
粒度="药物级（药名＋五味五行互含名位）／方剂级（方名反向索引）／药—方—阐释 三级串联",
|
||
用户指令="2026-09-22「初步进行《辅行诀五脏用药法要》与《辅行诀五脏用药法要 药性探真》的数据关联，"
|
||
"《辅行诀五脏用药法要》作为古本，是基础」",
|
||
来源=dict(底本文件=BENDI_MD, 底本md5=bmd5, 底本结构=BENDI_JSON,
|
||
探真文件=TZ_MD, 探真md5=tmd5, 探真结构=TZ_JSON),
|
||
底本锚=dict(二十五味药精=[ln for ln, _ in YAO_ANCHORS], 十三种药=THIRTEEN_LINE),
|
||
药精表=oracle_tbl, 十三种药=thirteen,
|
||
药物级对位=[dict(章=r["章"], 节=r["节"], 条目=r["条目"], 探真行=r["探真行"], 药=r["药"],
|
||
名位=r["名位"], 底本药=r.get("底本药"), 底本类=r.get("底本类"),
|
||
底本行=r.get("底本行"), 判定=r["判定"]) for r in rows],
|
||
方剂级反向索引=[dict(篇=l["篇"], 方名=l["方名"],
|
||
全名命中行=[h for h, _ in l["命中"]],
|
||
简称命中行=[h for h, _ in l["简称命中"]],
|
||
近似上下文=[dict(行=h, 文=c) for h, c in l["近似"]]) for l in cites],
|
||
药方阐释=[dict(篇=f["篇"], 方名=f["方名"], 组成=f["组成"], 加=f["加"], 去=f["去"],
|
||
药行=f["药行"], 关联=f["药—阐释"]) for f in fangs],
|
||
差异清单=[dict(类=c, 对象=o, 说明=d) for c, o, d in diff],
|
||
未落位=dict(方中药味探真未收=miss_yao,
|
||
探真未引及之方=[l["方名"] for l in cites if not l["命中"] and not l["简称命中"]],
|
||
底本无对应之探真药=sorted({r["药"] for r in named if r["判定"].startswith("○")})),
|
||
自检=[dict(项=n, 结果="✅ 通过" if v else "❌ 未通过") for n, v in checks],
|
||
重跑="python3 05_脚本工具/15_link_tanzhen.py（幂等；--check 只校验）",
|
||
)
|
||
|
||
if not ok:
|
||
print("❌ 自检未通过，未写盘：")
|
||
for n, v in checks:
|
||
if not v:
|
||
print(" -", n)
|
||
return 1
|
||
if check_only:
|
||
print("--check：自检通过，未写盘。")
|
||
for n, _ in checks:
|
||
print(" ✅", n)
|
||
return 0
|
||
|
||
wf(OUT_TABLE, "\n".join(tbl) + "\n")
|
||
wf(OUT_REPORT, "\n".join(rep) + "\n")
|
||
wf(OUT_JSON, json.dumps(struct, ensure_ascii=False, indent=2) + "\n")
|
||
print("对照表:", OUT_TABLE, os.path.getsize(os.path.join(BASE, OUT_TABLE)), "bytes")
|
||
print("报告 :", OUT_REPORT, os.path.getsize(os.path.join(BASE, OUT_REPORT)), "bytes")
|
||
print("JSON :", OUT_JSON, os.path.getsize(os.path.join(BASE, OUT_JSON)), "bytes")
|
||
print(f"药物级 {len(rows)} 行（有名位 {len(named)}）｜方剂级 {len(cites)} 方｜方—药 {len(fangs)} 方")
|
||
for n, v in checks:
|
||
print(" ✅", n)
|
||
print(" ⓘ 方中药味探真未收:", "、".join(miss_yao) or "无")
|
||
print(f" ⓘ 方剂级：全名直引 {n_full} · 简称命中 {n_short} · 未引及 {n_none}")
|
||
print(" ⓘ 探真未引及之方:",
|
||
"、".join(l["方名"] for l in cites if not l["命中"] and not l["简称命中"]) or "无")
|
||
for rel in (OUT_TABLE, OUT_REPORT, OUT_JSON):
|
||
print(f" md5 {md5_of(os.path.join(BASE, rel))} {rel}")
|
||
return 0
|
||
|
||
|
||
if __name__ == "__main__":
|
||
sys.exit(main())
|