@furongjun1999/dsh-memory 0.4.8 → 0.4.10

package/md_cg/insight.py

@@ -59,10 +59,12 @@ V_LABELS = {"v1": "可检索/被引用", "v2": "实践重复", "v3": "外部确
 
 # ---------------------------------------------------------------- 基础
 
+# 生效条件：cg 可提供 root 时返回 os.path.join(cg.root, INSIGHT_LOG)。
 def log_path(cg):
     return os.path.join(cg.root, INSIGHT_LOG)
 
 
+# 生效条件：向 log_path(cg) 追加 JSONL 的调用若抛任何异常都会被 except Exception 吞掉，无论成败最终都返回 rec。
 def _append(cg, rec):
     """留痕（best-effort）：审计不该反过来打断主流程。"""
     try:
@@ -72,6 +74,7 @@ def _append(cg, rec):
     return rec
 
 
+# 生效条件：getattr(cg,"index",None) 或其 "nodes" 为 None/假值时返回空列表，否则返回 nodes 中 (e or {}).get("tags") or [] 含 TAG_EVENT 的节点 id 列表。
 def _events(cg):
     """全部洞见事件节点 id（按 index 层标签粗筛，避免全量读盘）。"""
     nodes = (getattr(cg, "index", None) or {}).get("nodes") or {}
@@ -79,6 +82,7 @@ def _events(cg):
             if TAG_EVENT in ((e or {}).get("tags") or [])]
 
 
+# 生效条件：cg 具备 get，node_id 经 str(node_id or "").strip() 后非空（0/空串等假值回落空串即返回 None）、cg.get(nid) 返回真值节点、且该节点 frontmatter 的 tags（缺键或假值时按 [] 计）包含 TAG_EVENT 时返回该节点，否则返回 None。
 def _read_event(cg, node_id):
     """读洞见事件节点；非事件节点（标签不符）返回 None。"""
     nid = str(node_id or "").strip()
@@ -98,6 +102,7 @@ def _read_event(cg, node_id):
 
 # ---------------------------------------------------------------- 条件快照
 
+# 生效条件：conditions 为 None/假值时按空 dict 处理且 missing 登记 CONDITION_KEYS 中不在其中的键；retrievability 由 raw.get("retrievability",0.0) or 0.0 转 float（失败落 0.0）并截断到 [0,1]，pressure/tone 不在 PRESSURES/TONES 白名单（含空串回落 "medium"/"neutral"）时落 "medium"/"neutral"，cross_domain 为字符串时包成单元素列表再排序去重，continuity_turns 经 int(raw.get(...) or 0)（失败落 0）取 max(0,·)。
 def normalize(conditions=None):
     """C1–C8 归一化；**缺失项显式登记**而不是静默补默认值。"""
     raw = dict(conditions or {})
@@ -126,6 +131,7 @@ def normalize(conditions=None):
     return {"conditions": out, "missing": missing}
 
 
+# 生效条件：conditions 经 normalize 后，retrievability ≥ C1_WINDOW_MIN、cross_domain 非空、pressure == "low" 三者同时成立时 open 为 True 且 blocked_by 为空，否则 blocked_by 列出未通过的闸门键名。
 def window(conditions=None):
     """开窗判定：C1 ≥ 0.6 ∧ 跨域（≥1 个域）∧ 低压力。
 
@@ -145,6 +151,7 @@ def window(conditions=None):
             "note": "开窗仅表示「条件具备」；是否有效一律由 verify 的外部证据裁决"}
 
 
+# 生效条件：任意 statement（假值按 "" 计）与 ts（None/0 等假值时回落 time.time()）下恒返回 "ins_%d_%s"，其中时间数字为 int(ts or time.time())、哈希为 str(statement or "").encode("utf-8") 的 sha1 前 8 位。
 def event_id(statement, ts=None):
     return "ins_%d_%s" % (int(ts or time.time()),
                           hashlib.sha1(str(statement or "").encode("utf-8")).hexdigest()[:8])
@@ -152,6 +159,7 @@ def event_id(statement, ts=None):
 
 # ---------------------------------------------------------------- record
 
+# 生效条件：cg 具备 add/_append，statement 经 str(statement or "").strip() 为空时抛 ValueError；非空时以 str(node_id or event_id(statement)).strip() 为 nid（node_id 假值时回落到 event_id(statement)），_read_event 命中已有事件则返回 duplicate=True/existed_before=True 且 state 取该节点 frontmatter 的 insight_state，未命中则以 STATE_PENDING 写入新节点并返回 duplicate=False/existed_before=False，返回中的 window 来自 window(conditions)。
 def record(cg, statement=None, conditions=None, category="", source="",
            tags=None, importance=0.5, node_id=None, actor=None, **extra):
     """记录一条洞见事件（pending）＋ C1–C8 条件快照。
@@ -202,6 +210,7 @@ def record(cg, statement=None, conditions=None, category="", source="",
 
 # ---------------------------------------------------------------- verify
 
+# 生效条件：evidence 为 str/dict 时先包成单元素列表、为 None 或假值时按 [] 遍历；每个 dict 项 type 经 str(...).strip().lower() 属 V_TYPES 才保留否则回落 "v1"，ref 取 item.get("ref") 或 item.get("source") 或 ""、note 取 item.get("note") 或 ""，非 dict 项一律记为 {"type":"v1","ref":str(item),"note":""}；v_types 中仅 strip().lower() 后属 V_TYPES 的项追加空 ref/note 条目，最终返回 out。
 def _normalize_evidence(evidence=None, v_types=None):
     """证据归一化：接受 ``["ref", ...]`` / ``[{"type","ref","note"}]`` / ``v_types``。"""
     ev = evidence
@@ -223,6 +232,7 @@ def _normalize_evidence(evidence=None, v_types=None):
     return out
 
 
+# 生效条件：_read_event(cg,node_id) 取不到事件节点、或 verdict 经 str(verdict or "").strip().lower() 后非空且不是 verified/falsified 时抛 ValueError；verdict 为 None 或空白时按 v3 → v2 → v1 条数 ≥ V1_MIN_EVIDENCE 的顺序定 verified，有证据但不达门槛或无证据则保持 STATE_PENDING 并附 reason；显式 verdict 直接采信，verified 分支重要度保底 IMPORTANCE_FLOOR 并置保护位，falsified 分支打 TAG_FALSIFIED。
 def verify(cg, node_id=None, evidence=None, v_types=None, verdict=None,
            actor=None, note=""):
     """用 V1/V2/V3 外部证据裁决洞见事件。
@@ -305,6 +315,7 @@ def verify(cg, node_id=None, evidence=None, v_types=None, verdict=None,
 
 # ---------------------------------------------------------------- list / report
 
+# 生效条件：state 为真值时只保留 insight_state 与之相等的项（state 为 None/空串等假值则不过滤），limit 为 0 等假值时返回按 created_at 排序后的全部列表，limit 为真值时返回 out[-int(limit):]——limit 为负时该切片等价于 out[|limit|:]，即去掉排序后前 |limit| 条而非取前 |limit| 条。
 def list_events(cg, state=None, limit=0):
     out = []
     for nid in _events(cg):
@@ -326,6 +337,7 @@ def list_events(cg, state=None, limit=0):
     return out[-int(limit):] if limit else out
 
 
+# 生效条件：now 为 None/0 等假值时取 time.time()，window_days 为真值时按 now - window_days*86400 过滤 created_at；已裁决数 verified+falsified < CER_MIN_SAMPLES 时直接返回 cer/se/two_se/significant 为 None、layer_state="insufficient" 的结果，否则按 p=verified/denom 算出 se 与 significant，significant 为假时 p<0.3 取 degraded、否则 watch，significant 为真时 p≥0.5 取 reliable、否则 watch。
 def report(cg, window_days=None, now=None):
     """CER 条件有效洞见率报告（样本不足不判定）。
 
@@ -369,20 +381,24 @@ def report(cg, window_days=None, now=None):
 
 # ---------------------------------------------------------------- outlook
 
+# 生效条件：entry 为假值或 (entry or {}).get("layer") 为假值时返回 "unknown"，否则返回该 layer 值的 str 形式。
 def _layer_of(entry):
     return str((entry or {}).get("layer") or "unknown")
 
 
+# 生效条件：values 为空/假值时返回 min/p50/p90/max 全为 None 的字典，否则对 sorted(values) 取 0.0/0.5/0.9/1.0 位置的值并各自 round(·,4)。
 def _quartiles(values):
     if not values:
         return {"min": None, "p50": None, "p90": None, "max": None}
     vs = sorted(values)
+# 生效条件：给定 f 与闭包列表 vs，取下标 i = min(len(vs)-1, max(0, int(round(f*(len(vs)-1)))))（f 越界被夹到两端）并返回 round(vs[i], 4)；vs 为空时 min 得到 -1 的取值行为源码未做校验。
     def _q(f):
         i = min(len(vs) - 1, max(0, int(round(f * (len(vs) - 1)))))
         return round(vs[i], 4)
     return {"min": _q(0.0), "p50": _q(0.5), "p90": _q(0.9), "max": _q(1.0)}
 
 
+# 生效条件：cg.index（取不到或假值时 nodes 为空、total=0 使 protected_rate 与 growth 回落 None）与 now（假值时回落 time.time()）决定 d1，float(recent_days) 参与 lo_recent 计算，window_days 透传 report 得 d2；suggestions 依序在 basis["unset"] 非零、hi_imp_unprotected 非空、no_neg 非零、pending>=3、layers["unresolved"] 非零、contextual>max(3, knowledge)、rep["layer_state"]=="insufficient" 时各自追加一条，high_importance_unprotected 取排序后前 sample_limit 项（sample_limit=0 时为空列表）。
 def outlook(cg, window_days=None, sample_limit=8, recent_days=7, now=None):
     """结构洞察：D1 结构判断 + D2 盲区与趋势建议（只读）。
 
@@ -482,4 +498,4 @@ def outlook(cg, window_days=None, sample_limit=8, recent_days=7, now=None):
             "d1": d1, "d2": d2, "suggestions": suggestions,
             "generated_at": now,
             "note": ("结构描述与差距定位，不判有效性；suggestions 按优先级给出，"
-                     "需调用方自行执行并核对")}
+                     "需调用方自行执行并核对")}

package/md_cg/lexicon/build_cedict_en_zh.py

@@ -50,6 +50,7 @@ CEDICT_URL = "https://www.mdbg.net/chinese/export/cedict/cedict_1_0_ts_utf-8_mdb
 LINE_RE = re.compile(r"^(\S+)\s+(\S+)\s+\[[^\]]+\]\s+/(.+)/\s*$")
 
 
+# 生效条件：无必需形参，被调用即创建模块常量 RAW_DIR、按 CEDICT_URL 发请求（UA 头 + timeout=120），把响应体写入 RAW_GZ，再以 gzip 解压并按 errors="replace" 解码写入 RAW_TXT，最后打印字节数与行数并返回 0；
 def download():
     os.makedirs(RAW_DIR, exist_ok=True)
     req = urllib.request.Request(
@@ -67,6 +68,7 @@ def download():
     return 0
 
 
+# 生效条件：无必需形参，迭代时逐行读模块常量 RAW_TXT，仅当 strip 后非空、不以 "#" 开头且 LINE_RE.match 命中时产出 (第 2 组, 第 3 组)，其余行跳过（生成器，无产出即为空）；
 def iter_entries():
     """CEDICT 行 → (简体, [英文释义段])。注释/空行跳过。"""
     with io.open(RAW_TXT, encoding="utf-8") as f:
@@ -80,6 +82,7 @@ def iter_entries():
             yield m.group(2), m.group(3)
 
 
+# 生效条件：min_len 取默认 3（调用方可传入覆盖，传入 0 则 len(sub)<min_len 恒不成立、不再做长度过滤），遍历 iter_entries() 的释义按 / 分段、剥 (…) 注、按 ;, 切子段并 strip+lower+去 to/a/an 前缀，子段须 fullmatch [a-z]+、不在 STOPWORDS、长度 ≥ min_len 才计入 rev[en]；末尾按 zh 集大小 1 写入 cand[en]、>1 写入 conflicts[en]，返回 (cand, conflicts, stat)；
 def reverse_map(min_len=3):
     """词级反查 → (候选 {en: zh}, 冲突 {en: [zh...]}, 统计 dict)。
 
@@ -122,6 +125,7 @@ def reverse_map(min_len=3):
     return cand, conflicts, stat
 
 
+# 生效条件：无必需形参，逐行读模块引用 b6.CORPUS567 并跳过空白行，每行 json.loads 后取 (c.get("zh") or "") 与 (c.get("zh_fields") or {}) 中所有 str 值拼接成一条，最终以换行连成整串返回（无有效行时返回空串）；
 def _corpus_text():
     """领域语料全文（zh 正文+五槽摘要）——冲突集频次消歧用。
 
@@ -143,6 +147,7 @@ def _corpus_text():
     return "\n".join(buf)
 
 
+# 生效条件：无必需形参，逐行读模块引用 b6.CORPUS567 并跳过空白行；摘要取 (c.get("zh_fields") or {}).get("summary") 且 str(...).strip() 后为空时回落 c.get("zh")，非空则 semantic_atoms 切分后空格连接入 sums；正文取 c.get("zh").strip() 非空则同样切分入 bodys；返回 ("\n".join(sums), "\n".join(bodys))（两级均可能为空串）；
 def _atoms_texts():
     """doc 侧同构消歧空间（2026-09-14 消歧口径升级）。
 
@@ -175,6 +180,7 @@ def _atoms_texts():
     return "\n".join(sums), "\n".join(bodys)
 
 
+# 生效条件：给定冲突英文词 en、候选中文集合 zhs、文本对 texts=(摘要原子串, 正文原子串)，仅对 2 ≤ len(z) ≤ 4 且按 sorted(zhs) 顺序的候选，依次用摘要 frag 计数、摘要全零时改用正文 frag 计数、前两级全零时改用单原子碎片计数求和（级别 0/1/2 依次），任一非零即入 scored；scored 为空返回 (None, -1)，否则按 (-计数, 级别, 长度, 字典序) 排序返回 scored[0] 的中文词与级别；
 def _disambiguate(en, zhs, texts):
     """单英文词冲突集三级消歧：摘要碎片串 → 正文碎片串 → 碎片字级和。
 
@@ -216,6 +222,7 @@ def _disambiguate(en, zhs, texts):
     return scored[0][3], scored[0][1]
 
 
+# 生效条件：min_len 取默认 3、domain 取默认 True（两者均可由调用方传入覆盖，domain 传假值则跳过消歧分支）；先 reverse_map(min_len) 取 cand/conf/stat，当 domain 为真且 conf 非空时以 _atoms_texts() 对每个冲突词调 _disambiguate 并计 disambiguated/body/char/dropped，命中则写回 cand 并 conf.pop(en)、未命中仅计 dropped；随后无条件把 set(cand) & set(EN_ZH) 的键从 cand 移除（手工表优先），再写出 OUT_JSON 并返回 0；
 def build(min_len=3, domain=True):
     cand, conf, stat = reverse_map(min_len)
     stat["disambiguated"] = 0
@@ -274,6 +281,7 @@ def build(min_len=3, domain=True):
     return 0
 
 
+# 生效条件：无必需形参，两次逐行读模块常量 QUESTIONS 并跳过空白行：第一次对 str(q.get("question") or "") 调 normalize_en_query，把 detail 中 action=="unknown_keep" 的 str(d.get("orig") or "").lower() 逐个计入 Counter；第二次累加各问句 normalize_en_query 返回的第 0 项长度得 total_tokens；打印去重词数/词次/原子序列长度与 top 30 后返回 0（无有效行时计数为零仍返回 0）；
 def analyze():
     """集成后 normalize 的最终残留取证（en_normalizer 已加载 cedict_en_zh.json，
     此处 unknown_keep 即全链路翻译后的真 OOV）。"""
@@ -304,6 +312,7 @@ def analyze():
     return 0
 
 
+# 生效条件：无必需形参，mode = sys.argv[1] if len(sys.argv) > 1 else "analyze"（仅按参数个数回落，sys.argv[1] 为空串时 mode 即空串、不回落默认）；mode=="download" 返回 download()、"build" 返回 build()、"analyze" 返回 analyze()，其余取值打印模块 doc 并返回 2；
 def main():
     mode = sys.argv[1] if len(sys.argv) > 1 else "analyze"
     if mode == "download":

package/md_cg/lexicon/build_standard_en.py

@@ -1,168 +1,171 @@
-# -*- coding: utf-8 -*-
-"""build_standard_en.py · 英文标准库构建器（语素表 + 词条 → standard_en.json）
-
-词条语素带显式英文注 (zh_morph, en_gloss)，规避单字歧义（月=month/moon、生=person/living）。
-构建即验证：标准英文必须等于语素英文的 snake_case 直拼。
-"""
-import collections
-import json, io, os, re
-
-MORPHEMES = {}
-def M(zh, en):
-    MORPHEMES[zh] = en
-
-# 动物
-for zh, en in [("牛","cow"),("猪","pig"),("羊","sheep"),("鸡","chicken"),("鸭","duck"),("马","horse"),("狗","dog"),("猫","cat"),("鱼","fish"),("鸟","bird")]:
-    M(zh, en)
-# 食
-for zh, en in [("肉","meat"),("奶","milk"),("蛋","egg"),("米","rice"),("菜","vegetable"),("果","fruit"),("油","oil"),("盐","salt"),("糖","sugar"),("茶","tea"),("酒","wine"),("水","water")]:
-    M(zh, en)
-# 时间
-for zh, en in [("年","year"),("月","month"),("日","day"),("天","day"),("周","week"),("期","period"),("时","hour"),("分","minute"),("秒","second"),("春","spring"),("秋","autumn"),("冬","winter"),("夏","summer"),("今","this"),("明","next"),("去","past")]:
-    M(zh, en)
-# 数
-for zh, en in [("一","one"),("二","two"),("三","three"),("四","four"),("五","five"),("六","six"),("七","seven"),("八","eight"),("九","nine"),("十","ten"),("百","hundred"),("千","thousand")]:
-    M(zh, en)
-# 亲
-for zh, en in [("父","father"),("母","mother"),("兄","elder_brother"),("弟","younger_brother"),("姐","elder_sister"),("妹","younger_sister"),("儿","son"),("女","daughter")]:
-    M(zh, en)
-# 身
-for zh, en in [("头","head"),("手","hand"),("脚","foot"),("眼","eye"),("口","mouth"),("耳","ear"),("鼻","nose"),("心","heart"),("血","blood"),("骨","bone"),("皮","skin"),("脏","organ")]:
-    M(zh, en)
-# 自然
-for zh, en in [("山","mountain"),("水","water"),("火","fire"),("土","soil"),("木","wood"),("金","metal"),("光","light"),("电","electric"),("风","wind"),("雨","rain"),("雪","snow"),("星","star"),("海","sea"),("河","river"),("石","stone"),("云","cloud"),("月","moon")]:
-    M(zh, en)
-# 器
-for zh, en in [("车","vehicle"),("船","boat"),("机","machine"),("器","device"),("刀","knife"),("笔","pen"),("书","book"),("门","door"),("窗","window"),("桥","bridge"),("路","road"),("池","cell")]:
-    M(zh, en)
-# 动（基形）
-for zh, en in [("走","walk"),("来","come"),("看","see"),("听","listen"),("吃","eat"),("喝","drink"),("说","speak"),("读","read"),("写","write"),("买","buy"),("卖","sell"),("学","study"),("教","teach"),("做","make"),("跑","run"),("飞","fly"),("睡","sleep"),("想","think"),("爱","love"),("用","use"),("开","open"),("关","close")]:
-    M(zh, en)
-# 形
-for zh, en in [("大","big"),("小","small"),("高","high"),("低","low"),("新","new"),("旧","old"),("好","good"),("坏","bad"),("长","long"),("短","short"),("快","fast"),("慢","slow"),("热","hot"),("冷","cold"),("多","many"),("少","few"),("重","heavy"),("轻","light"),("深","deep")]:
-    M(zh, en)
-# 抽象
-for zh, en in [("力","force"),("气","gas"),("道","way"),("理","principle"),("文","text"),("化","transform"),("话","speech"),("影","shadow"),("音","sound"),("名","name"),("国","country"),("家","home"),("城","city"),("王","king"),("民","people"),("军","army"),("医","medicine"),("药","drug"),("病","disease"),("课","lesson"),("题","question"),("工","work"),("师","master"),("律","law"),("图","graph"),("谱","map"),("数","number"),("物","thing"),("地","earth"),("人","person"),("脑","brain"),("视","view")]:
-    M(zh, en)
-# 多字语素（化石词 / 复合单位）
-for zh, en in [("星期","week"),("经济","economy"),("逻辑","logic"),("沙发","sofa"),("咖啡","coffee"),("巧克力","chocolate"),("知识","knowledge"),("工作","work"),("纪律","discipline"),("记忆","memory"),("认知","cognition"),("月光","moon_light"),("十二","twelve"),("兄弟","brothers"),("姐妹","sisters"),("汽","gas")]:
-    M(zh, en)
-
-# 词条：(zh, [(语素, 英文注)...], 标准英文, legacy 别名, class)
-E = []
-DEFAULT_CLS = "compositional"
-
-
-def A(zh, glosses, std, legacy, cls=DEFAULT_CLS):
-    E.append((zh, glosses, std, legacy, cls))
-
-# 食物（语素直拼）
-A("牛肉", [("牛","cow"),("肉","meat")], "cow_meat", ["beef"])
-A("猪肉", [("猪","pig"),("肉","meat")], "pig_meat", ["pork"])
-A("羊肉", [("羊","sheep"),("肉","meat")], "sheep_meat", ["mutton"])
-A("鸡肉", [("鸡","chicken"),("肉","meat")], "chicken_meat", ["chicken"])
-A("鸭肉", [("鸭","duck"),("肉","meat")], "duck_meat", [])
-A("牛奶", [("牛","cow"),("奶","milk")], "cow_milk", ["milk"])
-A("鸡蛋", [("鸡","chicken"),("蛋","egg")], "chicken_egg", ["egg"])
-A("牛肉面", [("牛","cow"),("肉","meat"),("面","noodle")], "cow_meat_noodle", [])
-# 时间
-A("月份", [("月","month")], "month", [], "compositional")          # 份=名物化后缀，剔
-A("一月", [("一","one"),("月","month")], "one_month", ["January"])
-A("二月", [("二","two"),("月","month")], "two_month", ["February"])
-A("三月", [("三","three"),("月","month")], "three_month", ["March"])
-A("十二月", [("十二","twelve"),("月","month")], "twelve_month", ["December"])
-A("星期一", [("星期","week"),("一","one")], "week_one", ["Monday"])
-A("星期日", [("星期","week"),("日","sun")], "week_sun", ["Sunday"])
-A("今天", [("今","this"),("天","day")], "this_day", ["today"])
-A("明天", [("明","next"),("天","day")], "next_day", ["tomorrow"])
-A("今年", [("今","this"),("年","year")], "this_year", [])
-A("明年", [("明","next"),("年","year")], "next_year", [])
-A("去年", [("去","past"),("年","year")], "past_year", ["last year"])
-A("春天", [("春","spring")], "spring", ["spring"])                  # 天=季节后缀，剔
-A("冬天", [("冬","winter")], "winter", ["winter"])
-# 家庭
-A("父亲", [("父","father")], "father", ["father"])                  # 亲=亲缘标记，剔
-A("母亲", [("母","mother")], "mother", ["mother"])
-A("兄弟", [("兄弟","brothers")], "brothers", ["brothers"], "compositional")  # 复数是语义本身
-A("姐妹", [("姐妹","sisters")], "sisters", ["sisters"], "compositional")
-A("儿子", [("儿","son")], "son", ["son"])
-A("女儿", [("女","daughter")], "daughter", ["daughter"])
-# 身体
-A("眼睛", [("眼","eye")], "eye", ["eye"])                          # 睛=瞳，并入 eye
-A("心脏", [("心","heart"),("脏","organ")], "heart_organ", ["heart"])
-# 自然
-A("火山", [("火","fire"),("山","mountain")], "fire_mountain", ["volcano"])
-A("海水", [("海","sea"),("水","water")], "sea_water", ["seawater"])
-A("月光", [("月","moon"),("光","light")], "moon_light", ["moonlight"])
-A("雪山", [("雪","snow"),("山","mountain")], "snow_mountain", ["snow mountain"])
-# 科技
-A("电脑", [("电","electric"),("脑","brain")], "electric_brain", ["computer"])
-A("电话", [("电","electric"),("话","speech")], "electric_speech", ["telephone"])
-A("电视", [("电","electric"),("视","view")], "electric_view", ["television"])
-A("电影", [("电","electric"),("影","shadow")], "electric_shadow", ["movie", "film"])
-A("电池", [("电","electric"),("池","cell")], "electric_cell", ["battery"])
-A("火车", [("火","fire"),("车","vehicle")], "fire_vehicle", ["train"])
-A("汽车", [("汽","gas"),("车","vehicle")], "gas_vehicle", ["car"])
-A("飞机", [("飞","fly"),("机","machine")], "fly_machine", ["airplane"])
-A("手机", [("手","hand"),("机","machine")], "hand_machine", ["mobile phone"])
-# 学科
-A("数学", [("数","number"),("学","study")], "number_study", ["mathematics"])
-A("化学", [("化","transform"),("学","study")], "transform_study", ["chemistry"])
-A("物理学", [("物","thing"),("理","principle"),("学","study")], "thing_principle_study", ["physics"])
-A("地理学", [("地","earth"),("理","principle"),("学","study")], "earth_principle_study", ["geography"])
-A("生物学", [("生","living"),("物","thing"),("学","study")], "living_thing_study", ["biology"])
-A("热力学", [("热","heat"),("力","force"),("学","study")], "heat_force_study", ["thermodynamics"])
-# 人物
-A("医生", [("医","medicine"),("生","person")], "medicine_person", ["doctor"])
-A("学生", [("学","study"),("生","person")], "study_person", ["student"])
-A("律师", [("律","law"),("师","master")], "law_master", ["lawyer"])
-A("工人", [("工","work"),("人","person")], "work_person", ["worker"])
-# 领域词（认知图相关）
-A("知识图谱", [("知识","knowledge"),("图","graph"),("谱","map")], "knowledge_graph_map", ["knowledge graph"])
-A("认知图", [("认知","cognition"),("图","graph")], "cognition_graph", [])
-A("工作纪律", [("工作","work"),("纪律","discipline")], "work_discipline", [])
-A("记忆库", [("记忆","memory"),("库","store")], "memory_store", [])
-# 化石词（音译/古典，稳定单语素，保留既有英文）
-A("经济", [("经济","economy")], "economy", ["economy"], "opaque")
-A("逻辑", [("逻辑","logic")], "logic", ["logic"], "opaque")
-A("沙发", [("沙发","sofa")], "sofa", ["sofa"], "opaque")
-A("咖啡", [("咖啡","coffee")], "coffee", ["coffee"], "opaque")
-A("巧克力", [("巧克力","chocolate")], "chocolate", ["chocolate"], "opaque")
-
-
-def main():
-    problems = []
-    entries = []
-    for zh, glosses, std, legacy, cls in E:
-        join_zh = "".join(z for z, _ in glosses)
-        if not zh.startswith(join_zh):
-            problems.append("%s: 语素拼合 %s != 词条 %s" % (zh, join_zh, zh))
-            continue
-        join_en = "_".join(e for _, e in glosses)
-        if join_en != std:
-            problems.append("%s: 标准英文 %s != 语素拼接 %s" % (zh, std, join_en))
-            continue
-        if re.search(r"[A-Z]", std):
-            problems.append("%s: 标准英文含大写（违 snake_case）" % zh)
-            continue
-        entries.append({"zh": zh, "glosses": [{"zh": z, "en": e} for z, e in glosses],
-                        "standard_en": std, "legacy_en": legacy, "class": cls})
-    out = {
-        "meta": {"name": "lingshu-standard-en", "version": "0.1.0", "rules": "RULES.md",
-                 "note": "英文标准 = 中文语素序直拼（snake_case）；屈折归零；虚词不入名；音译化石词保留为 opaque；四问皆否不入库"},
-        "morphemes": MORPHEMES,
-        "entries": entries,
-    }
-    path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "standard_en.json")
-    with io.open(path, "w", encoding="utf-8", newline="\n") as f:
-        json.dump(out, f, ensure_ascii=False, indent=1)
-    cls_cnt = collections.Counter(e["class"] for e in entries)
-    print("entries=%d morphemes=%d classes=%s problems=%d" % (
-        len(entries), len(MORPHEMES), dict(cls_cnt), len(problems)))
-    for p in problems:
-        print("  !", p)
-    return 1 if problems else 0
-
-
-if __name__ == "__main__":
-    raise SystemExit(main())
+# -*- coding: utf-8 -*-
+"""build_standard_en.py · 英文标准库构建器（语素表 + 词条 → standard_en.json）
+
+词条语素带显式英文注 (zh_morph, en_gloss)，规避单字歧义（月=month/moon、生=person/living）。
+构建即验证：标准英文必须等于语素英文的 snake_case 直拼。
+"""
+import collections
+import json, io, os, re
+
+MORPHEMES = {}
+# 生效条件：以形参 zh、en 调用时无条件执行 MORPHEMES[zh] = en（同键覆盖），无返回值、无前置校验。
+def M(zh, en):
+    MORPHEMES[zh] = en
+
+# 动物
+for zh, en in [("牛","cow"),("猪","pig"),("羊","sheep"),("鸡","chicken"),("鸭","duck"),("马","horse"),("狗","dog"),("猫","cat"),("鱼","fish"),("鸟","bird")]:
+    M(zh, en)
+# 食
+for zh, en in [("肉","meat"),("奶","milk"),("蛋","egg"),("米","rice"),("菜","vegetable"),("果","fruit"),("油","oil"),("盐","salt"),("糖","sugar"),("茶","tea"),("酒","wine"),("水","water")]:
+    M(zh, en)
+# 时间
+for zh, en in [("年","year"),("月","month"),("日","day"),("天","day"),("周","week"),("期","period"),("时","hour"),("分","minute"),("秒","second"),("春","spring"),("秋","autumn"),("冬","winter"),("夏","summer"),("今","this"),("明","next"),("去","past")]:
+    M(zh, en)
+# 数
+for zh, en in [("一","one"),("二","two"),("三","three"),("四","four"),("五","five"),("六","six"),("七","seven"),("八","eight"),("九","nine"),("十","ten"),("百","hundred"),("千","thousand")]:
+    M(zh, en)
+# 亲
+for zh, en in [("父","father"),("母","mother"),("兄","elder_brother"),("弟","younger_brother"),("姐","elder_sister"),("妹","younger_sister"),("儿","son"),("女","daughter")]:
+    M(zh, en)
+# 身
+for zh, en in [("头","head"),("手","hand"),("脚","foot"),("眼","eye"),("口","mouth"),("耳","ear"),("鼻","nose"),("心","heart"),("血","blood"),("骨","bone"),("皮","skin"),("脏","organ")]:
+    M(zh, en)
+# 自然
+for zh, en in [("山","mountain"),("水","water"),("火","fire"),("土","soil"),("木","wood"),("金","metal"),("光","light"),("电","electric"),("风","wind"),("雨","rain"),("雪","snow"),("星","star"),("海","sea"),("河","river"),("石","stone"),("云","cloud"),("月","moon")]:
+    M(zh, en)
+# 器
+for zh, en in [("车","vehicle"),("船","boat"),("机","machine"),("器","device"),("刀","knife"),("笔","pen"),("书","book"),("门","door"),("窗","window"),("桥","bridge"),("路","road"),("池","cell")]:
+    M(zh, en)
+# 动（基形）
+for zh, en in [("走","walk"),("来","come"),("看","see"),("听","listen"),("吃","eat"),("喝","drink"),("说","speak"),("读","read"),("写","write"),("买","buy"),("卖","sell"),("学","study"),("教","teach"),("做","make"),("跑","run"),("飞","fly"),("睡","sleep"),("想","think"),("爱","love"),("用","use"),("开","open"),("关","close")]:
+    M(zh, en)
+# 形
+for zh, en in [("大","big"),("小","small"),("高","high"),("低","low"),("新","new"),("旧","old"),("好","good"),("坏","bad"),("长","long"),("短","short"),("快","fast"),("慢","slow"),("热","hot"),("冷","cold"),("多","many"),("少","few"),("重","heavy"),("轻","light"),("深","deep")]:
+    M(zh, en)
+# 抽象
+for zh, en in [("力","force"),("气","gas"),("道","way"),("理","principle"),("文","text"),("化","transform"),("话","speech"),("影","shadow"),("音","sound"),("名","name"),("国","country"),("家","home"),("城","city"),("王","king"),("民","people"),("军","army"),("医","medicine"),("药","drug"),("病","disease"),("课","lesson"),("题","question"),("工","work"),("师","master"),("律","law"),("图","graph"),("谱","map"),("数","number"),("物","thing"),("地","earth"),("人","person"),("脑","brain"),("视","view")]:
+    M(zh, en)
+# 多字语素（化石词 / 复合单位）
+for zh, en in [("星期","week"),("经济","economy"),("逻辑","logic"),("沙发","sofa"),("咖啡","coffee"),("巧克力","chocolate"),("知识","knowledge"),("工作","work"),("纪律","discipline"),("记忆","memory"),("认知","cognition"),("月光","moon_light"),("十二","twelve"),("兄弟","brothers"),("姐妹","sisters"),("汽","gas")]:
+    M(zh, en)
+
+# 词条：(zh, [(语素, 英文注)...], 标准英文, legacy 别名, class)
+E = []
+DEFAULT_CLS = "compositional"
+
+
+# 生效条件：以形参 zh、glosses、std、legacy 调用（cls 省略时取 DEFAULT_CLS）时无条件向 E 追加五元组 (zh, glosses, std, legacy, cls)，无返回、无校验。
+def A(zh, glosses, std, legacy, cls=DEFAULT_CLS):
+    E.append((zh, glosses, std, legacy, cls))
+
+# 食物（语素直拼）
+A("牛肉", [("牛","cow"),("肉","meat")], "cow_meat", ["beef"])
+A("猪肉", [("猪","pig"),("肉","meat")], "pig_meat", ["pork"])
+A("羊肉", [("羊","sheep"),("肉","meat")], "sheep_meat", ["mutton"])
+A("鸡肉", [("鸡","chicken"),("肉","meat")], "chicken_meat", ["chicken"])
+A("鸭肉", [("鸭","duck"),("肉","meat")], "duck_meat", [])
+A("牛奶", [("牛","cow"),("奶","milk")], "cow_milk", ["milk"])
+A("鸡蛋", [("鸡","chicken"),("蛋","egg")], "chicken_egg", ["egg"])
+A("牛肉面", [("牛","cow"),("肉","meat"),("面","noodle")], "cow_meat_noodle", [])
+# 时间
+A("月份", [("月","month")], "month", [], "compositional")          # 份=名物化后缀，剔
+A("一月", [("一","one"),("月","month")], "one_month", ["January"])
+A("二月", [("二","two"),("月","month")], "two_month", ["February"])
+A("三月", [("三","three"),("月","month")], "three_month", ["March"])
+A("十二月", [("十二","twelve"),("月","month")], "twelve_month", ["December"])
+A("星期一", [("星期","week"),("一","one")], "week_one", ["Monday"])
+A("星期日", [("星期","week"),("日","sun")], "week_sun", ["Sunday"])
+A("今天", [("今","this"),("天","day")], "this_day", ["today"])
+A("明天", [("明","next"),("天","day")], "next_day", ["tomorrow"])
+A("今年", [("今","this"),("年","year")], "this_year", [])
+A("明年", [("明","next"),("年","year")], "next_year", [])
+A("去年", [("去","past"),("年","year")], "past_year", ["last year"])
+A("春天", [("春","spring")], "spring", ["spring"])                  # 天=季节后缀，剔
+A("冬天", [("冬","winter")], "winter", ["winter"])
+# 家庭
+A("父亲", [("父","father")], "father", ["father"])                  # 亲=亲缘标记，剔
+A("母亲", [("母","mother")], "mother", ["mother"])
+A("兄弟", [("兄弟","brothers")], "brothers", ["brothers"], "compositional")  # 复数是语义本身
+A("姐妹", [("姐妹","sisters")], "sisters", ["sisters"], "compositional")
+A("儿子", [("儿","son")], "son", ["son"])
+A("女儿", [("女","daughter")], "daughter", ["daughter"])
+# 身体
+A("眼睛", [("眼","eye")], "eye", ["eye"])                          # 睛=瞳，并入 eye
+A("心脏", [("心","heart"),("脏","organ")], "heart_organ", ["heart"])
+# 自然
+A("火山", [("火","fire"),("山","mountain")], "fire_mountain", ["volcano"])
+A("海水", [("海","sea"),("水","water")], "sea_water", ["seawater"])
+A("月光", [("月","moon"),("光","light")], "moon_light", ["moonlight"])
+A("雪山", [("雪","snow"),("山","mountain")], "snow_mountain", ["snow mountain"])
+# 科技
+A("电脑", [("电","electric"),("脑","brain")], "electric_brain", ["computer"])
+A("电话", [("电","electric"),("话","speech")], "electric_speech", ["telephone"])
+A("电视", [("电","electric"),("视","view")], "electric_view", ["television"])
+A("电影", [("电","electric"),("影","shadow")], "electric_shadow", ["movie", "film"])
+A("电池", [("电","electric"),("池","cell")], "electric_cell", ["battery"])
+A("火车", [("火","fire"),("车","vehicle")], "fire_vehicle", ["train"])
+A("汽车", [("汽","gas"),("车","vehicle")], "gas_vehicle", ["car"])
+A("飞机", [("飞","fly"),("机","machine")], "fly_machine", ["airplane"])
+A("手机", [("手","hand"),("机","machine")], "hand_machine", ["mobile phone"])
+# 学科
+A("数学", [("数","number"),("学","study")], "number_study", ["mathematics"])
+A("化学", [("化","transform"),("学","study")], "transform_study", ["chemistry"])
+A("物理学", [("物","thing"),("理","principle"),("学","study")], "thing_principle_study", ["physics"])
+A("地理学", [("地","earth"),("理","principle"),("学","study")], "earth_principle_study", ["geography"])
+A("生物学", [("生","living"),("物","thing"),("学","study")], "living_thing_study", ["biology"])
+A("热力学", [("热","heat"),("力","force"),("学","study")], "heat_force_study", ["thermodynamics"])
+# 人物
+A("医生", [("医","medicine"),("生","person")], "medicine_person", ["doctor"])
+A("学生", [("学","study"),("生","person")], "study_person", ["student"])
+A("律师", [("律","law"),("师","master")], "law_master", ["lawyer"])
+A("工人", [("工","work"),("人","person")], "work_person", ["worker"])
+# 领域词（认知图相关）
+A("知识图谱", [("知识","knowledge"),("图","graph"),("谱","map")], "knowledge_graph_map", ["knowledge graph"])
+A("认知图", [("认知","cognition"),("图","graph")], "cognition_graph", [])
+A("工作纪律", [("工作","work"),("纪律","discipline")], "work_discipline", [])
+A("记忆库", [("记忆","memory"),("库","store")], "memory_store", [])
+# 化石词（音译/古典，稳定单语素，保留既有英文）
+A("经济", [("经济","economy")], "economy", ["economy"], "opaque")
+A("逻辑", [("逻辑","logic")], "logic", ["logic"], "opaque")
+A("沙发", [("沙发","sofa")], "sofa", ["sofa"], "opaque")
+A("咖啡", [("咖啡","coffee")], "coffee", ["coffee"], "opaque")
+A("巧克力", [("巧克力","chocolate")], "chocolate", ["chocolate"], "opaque")
+
+
+# 生效条件：无参调用时遍历模块级 E 的每条 (zh, glosses, std, legacy, cls)，凡 zh 不以 glosses 语素 zh 顺序拼接结果开头、或 "_".join(glosses 英文) != std、或 re.search(r"[A-Z]", std) 命中者追加 problems 并 continue，其余条目进 entries；随后将含 MORPHEMES 与 entries 的 out 写入 __file__ 同目录下 standard_en.json，打印统计并逐条打印 problems，problems 非空返回 1，否则返回 0。
+def main():
+    problems = []
+    entries = []
+    for zh, glosses, std, legacy, cls in E:
+        join_zh = "".join(z for z, _ in glosses)
+        if not zh.startswith(join_zh):
+            problems.append("%s: 语素拼合 %s != 词条 %s" % (zh, join_zh, zh))
+            continue
+        join_en = "_".join(e for _, e in glosses)
+        if join_en != std:
+            problems.append("%s: 标准英文 %s != 语素拼接 %s" % (zh, std, join_en))
+            continue
+        if re.search(r"[A-Z]", std):
+            problems.append("%s: 标准英文含大写（违 snake_case）" % zh)
+            continue
+        entries.append({"zh": zh, "glosses": [{"zh": z, "en": e} for z, e in glosses],
+                        "standard_en": std, "legacy_en": legacy, "class": cls})
+    out = {
+        "meta": {"name": "lingshu-standard-en", "version": "0.1.0", "rules": "RULES.md",
+                 "note": "英文标准 = 中文语素序直拼（snake_case）；屈折归零；虚词不入名；音译化石词保留为 opaque；四问皆否不入库"},
+        "morphemes": MORPHEMES,
+        "entries": entries,
+    }
+    path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "standard_en.json")
+    with io.open(path, "w", encoding="utf-8", newline="\n") as f:
+        json.dump(out, f, ensure_ascii=False, indent=1)
+    cls_cnt = collections.Counter(e["class"] for e in entries)
+    print("entries=%d morphemes=%d classes=%s problems=%d" % (
+        len(entries), len(MORPHEMES), dict(cls_cnt), len(problems)))
+    for p in problems:
+        print("  !", p)
+    return 1 if problems else 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())