@furongjun1999/dsh-memory 0.7.1 → 0.7.3

package/md_cg/test_ghostref.py

@@ -0,0 +1,182 @@
+# -*- coding: utf-8 -*-
+"""test_ghostref · 幽灵引用检查器（A2）：短语层标记 + id 层降级告警（不拒收）。
+
+多主体世界模型对齐 v0.1 §2 L1「转引不得升级」的行为守卫。
+运行：python -m md_cg.test_ghostref
+"""
+from __future__ import annotations
+
+import os
+import tempfile
+import time
+
+from . import branches, ghostref, nodefile
+from .mdcos import MdCGOS
+from .writepipe import default_pipeline
+
+_ok = 0
+_fail = []
+
+
+def check(name, cond, detail=""):
+    global _ok
+    if cond:
+        _ok += 1
+        print("[ok] " + name)
+    else:
+        _fail.append(name)
+        print("[FAIL] %s  · %s" % (name, str(detail)[:240]))
+
+
+H = ("# 功能名：幽灵引用守卫\n# 生效条件：测试环境\n# 子功能：无\n"
+     "# 执行：无\n# 验证方式：test\n# 不适用条件：无\n")
+URF = nodefile.UNCERTAIN_REFS_FIELD
+
+
+def main():
+    tmp = tempfile.mkdtemp(prefix="mdcg_ghost_test_")
+    cg = MdCGOS(os.path.join(tmp, "root"))
+    pipe = default_pipeline()
+
+    def W(nid, body, **kw):
+        a = {"op": "write", "node_id": nid, "content": H + body + "\n",
+             "layer": "knowledge", "content_kind": "text",
+             "verification_basis": "test"}
+        a.update(kw)
+        return pipe.execute(cg, a)
+
+    def FM(nid):
+        return ((cg.get(nid) or {}).get("frontmatter")) or {}
+
+    # ---- 1. 短语层：无可解析出处 ⇒ 标记；**不拒收** ----
+    out1 = W("ghp_one", "可以允许你在帐篷里点那根你上次买的香薰蜡烛。")
+    check("T1a 幽灵回指仍落盘（不拒收）", out1.get("committed") is True,
+          {k: out1.get(k) for k in ("ok", "committed", "moved_to")})
+    _refs1 = FM("ghp_one").get(URF) or []
+    check("T1b 短语标记落 fm（上次买）", "上次买" in _refs1, _refs1)
+    check("T1c 出口带 ghost_refs 告警",
+          bool(out1.get("ghost_refs")) and "上次买" in out1["ghost_refs"],
+          out1.get("ghost_refs"))
+
+    # ---- 2. 有出处（句内含库内 id）⇒ 不标记 ----
+    W("base_node", "集市采购的凭证记录。")
+    out2 = W("ghp_two", "上次说的那件事，见 base_node 的记载。")
+    check("T2a 有出处的回指不落标记",
+          URF not in FM("ghp_two"), FM("ghp_two").get(URF))
+    check("T2b 有出处的回指建 reference 边",
+          any(e.get("target") == "base_node" for e in (FM("ghp_two").get("edges") or [])),
+          FM("ghp_two").get("edges"))
+    check("T2c 出口无 ghost_refs", not out2.get("ghost_refs"), out2.get("ghost_refs"))
+
+    # ---- 3. 无回指短语 ⇒ 不标记（对照）----
+    out3 = W("ghp_three", "集市购物，周末进行。")
+    check("T3a 无短语不落标记", URF not in FM("ghp_three"), FM("ghp_three").get(URF))
+    # 2026-10-05 复核实证：T3 在「写入被拒」时会假通过（无标记≠已落盘）——
+    # 补 committed 断言把「对照场景确实落盘」钉住。
+    check("T3b 对照场景确实落盘", out3.get("committed") is True,
+          out3.get("committed"))
+
+    # ---- 4. 幽灵 id（句内 id 形态但不在库）⇒ 无出处 ⇒ 标记 ----
+    W("ghp_four", "参照 ghost_absent 那条记载，明明讲过。")
+    _refs4 = FM("ghp_four").get(URF) or []
+    check("T4a 幽灵 id 句仍标记（明明讲过）", "明明讲过" in _refs4, _refs4)
+    check("T4b 幽灵 id 不建边", not (FM("ghp_four").get("edges") or []),
+          FM("ghp_four").get("edges"))
+
+    # ---- 5. id 层：目标晚于自身 ⇒ 降级告警（不拒收）----
+    W("late_target", "晚出现的凭证节点。")
+    out5 = W("late_src", "依据见 late_target 记载。",
+             valid_from=time.time() - 100000)
+    check("T5a 目标晚于自身仍落盘（不拒收）", out5.get("committed") is True,
+          out5.get("committed"))
+    _lr = out5.get("late_refs") or []
+    check("T5b 降级告警点名目标",
+          len(_lr) == 1 and _lr[0].get("target") == "late_target", _lr)
+
+    # ---- 6. 正常时序 ⇒ 无告警（对照）----
+    out6 = W("ok_src", "依据见 late_target 记载。",
+             valid_from=time.time() + 100)
+    check("T6a 正常时序无告警", not out6.get("late_refs"), out6.get("late_refs"))
+    check("T6b 对照场景确实落盘", out6.get("committed") is True,
+          out6.get("committed"))
+
+    # ---- 7. opt-out（ghostref=False）⇒ 不标记 ----
+    out7 = W("ghp_opt", "上次说的事。", ghostref=False)
+    check("T7a opt-out 仍落盘", out7.get("committed") is True, out7.get("committed"))
+    check("T7b opt-out 不落标记", URF not in FM("ghp_opt"),
+          FM("ghp_opt").get(URF))
+
+    # ---- 8. add 层：显式落键 / 覆写不继承 / 非法拒（直调，不经写链）----
+    cg.add("ghp_eight", H + "上次说的那件事。\n", layer="knowledge",
+           verification_basis="test", uncertain_refs=["上次说", "之前说"])
+    check("T8a 显式落键", FM("ghp_eight").get(URF) == ["上次说", "之前说"],
+          FM("ghp_eight").get(URF))
+    cg.add("ghp_eight", H + "改写后的干净正文。\n", layer="knowledge",
+           verification_basis="test")
+    check("T8b 覆写不继承（正文属性重算）", URF not in FM("ghp_eight"),
+          FM("ghp_eight").get(URF))
+    try:
+        cg.add("ghp_bad", H + "正文。\n", layer="knowledge",
+               verification_basis="test", uncertain_refs="上次的")
+        check("T8c 非法类型拒绝", False, "未抛 ValueError")
+    except ValueError:
+        check("T8c 非法类型拒绝", True)
+
+    # ---- 9. fork 副本携带标记 ----
+    r9 = branches.fork(cg, ["ghp_one"], branch_id="br_ghost")
+    _bid = (r9.get("forked") or [{}])[0].get("to")
+    check("T9a fork 成功", bool(r9.get("ok")) and bool(_bid), r9)
+    check("T9b 副本携带标记", (_bid and FM(_bid).get(URF) == FM("ghp_one").get(URF)),
+          (_bid, FM(_bid).get(URF) if _bid else None))
+
+    # ---- 11. rewrite 按新正文重算标记（复核 yellow-1 修）----
+    # 副本 ghp_one@br_ghost 有标记 ["上次买"]（T1/ fork 携带）：
+    # 改成干净正文 → 标记消失；改回幽灵正文 → 标记出现（双向，非「清空」）。
+    rr1 = branches.rewrite(cg, _bid, H + "改写为干净正文。\n")
+    check("T11a rewrite 成功", rr1.get("ok") is True, rr1)
+    check("T11b rewrite 干净正文 → 标记重算消失", URF not in FM(_bid),
+          FM(_bid).get(URF))
+    rr2 = branches.rewrite(cg, _bid, H + "之前说的那件事。\n")
+    check("T11c rewrite 幽灵正文 → 标记重算出现",
+          "之前说" in (FM(_bid).get(URF) or []), FM(_bid).get(URF))
+
+    # ---- 12. merge 回主支按分支正文重算（复核 yellow-2① 修）----
+    # 主支幽灵正文 → fork → rewrite 分支为干净 → merge 回主支 → 主支标记消失。
+    W("ghp_merge", "明明讲过的那件事。")
+    check("T12a 主支初始有标记", "明明讲过" in (FM("ghp_merge").get(URF) or []),
+          FM("ghp_merge").get(URF))
+    r12 = branches.fork(cg, ["ghp_merge"], branch_id="br_mg")
+    _bid12 = (r12.get("forked") or [{}])[0].get("to")
+    branches.rewrite(cg, _bid12, H + "干净正文（无回指）。\n")
+    rm = branches.merge(cg, "br_mg")
+    check("T12b merge 成功", rm.get("ok") is True, rm)
+    check("T12c 主支标记按分支正文重算（消失）", URF not in FM("ghp_merge"),
+          FM("ghp_merge").get(URF))
+    # 反方向（2026-10-05 自证补）：干净主支 → 分支改幽灵 → merge 回写 ⇒ 主支
+    # 标记应**出现**——「消失」方向抓「携带旧值」，「出现」方向抓「置空不重算」。
+    W("ghp_merge2", "干净正文示例。")
+    r13 = branches.fork(cg, ["ghp_merge2"], branch_id="br_mg2")
+    _bid13 = (r13.get("forked") or [{}])[0].get("to")
+    branches.rewrite(cg, _bid13, H + "上次说的那件事。\n")
+    branches.merge(cg, "br_mg2")
+    check("T12d 主支标记按分支正文重算（出现）",
+          "上次说" in (FM("ghp_merge2").get(URF) or []),
+          FM("ghp_merge2").get(URF))
+
+    # ---- 10. 纯函数级：短语去重保序 + late_targets 边界 ----
+    # 「上次说的那件」只命中「上次说」——「上次的」要求连续子串（说/的 之间
+    # 不跳字）；命中去重后跨句保序。
+    _pf = ghostref.find_ghost_phrases("上次说的事；上次说的那件；之前说的那批。")
+    check("T10a 短语去重保序", _pf == ["上次说", "之前说"], _pf)
+    check("T10b late_targets 无 own_from 直接空",
+          ghostref.late_targets(cg, ["late_target"], None) == [])
+    check("T10c late_targets 非数 own_from 空",
+          ghostref.late_targets(cg, ["late_target"], "昨天") == [])
+
+    print("\ntest_ghostref: %d 通过 / %d 失败" % (_ok, len(_fail)))
+    if _fail:
+        raise SystemExit(1)
+
+
+if __name__ == "__main__":
+    main()

package/md_cg/test_health_corrupt_utf8.py

@@ -0,0 +1,37 @@
+# -*- coding: utf-8 -*-
+"""health() skips a damaged UTF-8 node instead of crashing the health endpoint."""
+from __future__ import annotations
+
+import os
+import shutil
+import tempfile
+
+from md_cg.mdcos import MdCGOS
+
+
+def main() -> None:
+    root = tempfile.mkdtemp(prefix="health_utf8_guard_")
+    cg = MdCGOS(root, autoflush=0)
+    try:
+        cg.add("good", "可读的健康检查节点", tags=["domain:health"])
+        cg.add("bad", "待损坏节点", tags=["domain:health"])
+        cg.flush()
+        bad_path = os.path.join(root, cg.index["nodes"]["bad"]["path"].replace("/", os.sep))
+        with open(bad_path, "wb") as fh:
+            fh.write(bytes((0xff, 0xfe, 0xfd)))
+
+        assert cg.get("bad") is None
+        results, _meta = cg.search("可读", k=5, include_neg=True)
+        assert any(item[0]["id"] == "good" for item in results), results
+        health = cg.health()
+        assert health["total_nodes"] == 2, health
+        assert sum(v["total"] for v in health["ccg_by_layer"].values()) == 1, health
+        print("PASS health skips invalid UTF-8 node")
+    finally:
+        cg.close()
+        shutil.rmtree(root, ignore_errors=True)
+
+
+if __name__ == "__main__":
+    main()
+

package/md_cg/test_index_crossprocess_reload.py

@@ -262,30 +262,45 @@ def main():
         a4.close()
 
         # ---------- ⑥ 性能粗测：无跨进程写入时逐读开销 ----------
-        print("\n【⑥】性能粗测（无跨进程写入，接线态 vs 重载桩空转态）")
-        N = 300
-        a2c = MdCGSecure(root3, principal=_principal())
-        for _ in range(5):
-            a2c.get("own_0")
-        t0 = time.perf_counter()
-        for _ in range(N):
-            a2c.get("own_0")
-        t_on = (time.perf_counter() - t0) / N * 1e6
-        orig = a2c._maybe_reload_index
-        a2c._maybe_reload_index = lambda: False
-        for _ in range(5):
-            a2c.get("own_0")
-        t0 = time.perf_counter()
-        for _ in range(N):
-            a2c.get("own_0")
-        t_off = (time.perf_counter() - t0) / N * 1e6
-        a2c._maybe_reload_index = orig
-        a2c.close()
-        print("       接线态 %.1f µs/读 · 桩空转态 %.1f µs/读 · 增幅 %.1f%%"
-              % (t_on, t_off, (t_on - t_off) / max(t_off, 1e-9) * 100))
-        check("性能上界：接线态 ≤ 桩空转态×3 + 30µs（stat 微秒级；宽松防抖）",
-              t_on <= t_off * 3 + 30.0,
-              "on=%.1f off=%.1f" % (t_on, t_off))
+        # 并行豁免 + 3 轮中位数（2026-10-03 外部报告核验新增）：
+        #   · 并行全量下 CPU 争抢使本断言假红（报告机器实测 on 242.0µs / off 35.6µs，
+        #     单独跑 3 次 41~42µs 全 PASS——接线态是 syscall 密集面，争抢长尾远重于
+        #     空转桩态）⇒ 并行时**明示豁免**（run_tests 经 MDCG_RUN_TESTS_PARALLEL
+        #     传信号；串行（--jobs 1，含 CI）照测——语义断言不受本豁免影响）；
+        #   · 串行下采样改 3 轮取中位（抗偶发长尾），阈值口径不变（仍防真回归）。
+        _parallel = os.environ.get("MDCG_RUN_TESTS_PARALLEL") == "1"
+        if _parallel:
+            print("\n【⑥】性能粗测（无跨进程写入，接线态 vs 重载桩空转态）")
+            print("  [SKIP] 性能上界断言（并行豁免）：CPU 争抢下本断言假红（外部报告"
+                  "实测 on 242.0µs vs off 35.6µs；单跑 41~42µs）——串行（--jobs 1，含 CI）照测")
+        else:
+            print("\n【⑥】性能粗测（无跨进程写入，接线态 vs 重载桩空转态；3 轮取中位）")
+            N = 300
+            a2c = MdCGSecure(root3, principal=_principal())
+
+            def _probe():
+                for _ in range(5):
+                    a2c.get("own_0")
+                t0 = time.perf_counter()
+                for _ in range(N):
+                    a2c.get("own_0")
+                return (time.perf_counter() - t0) / N * 1e6
+
+            on_runs = [_probe() for _ in range(3)]
+            orig = a2c._maybe_reload_index
+            a2c._maybe_reload_index = lambda: False
+            off_runs = [_probe() for _ in range(3)]
+            a2c._maybe_reload_index = orig
+            a2c.close()
+            t_on = sorted(on_runs)[1]
+            t_off = sorted(off_runs)[1]
+            print("       接线态 %.1f µs/读（3 轮 %s）· 桩空转态 %.1f µs/读（3 轮 %s）· 增幅 %.1f%%"
+                  % (t_on, "/".join("%.1f" % v for v in on_runs),
+                     t_off, "/".join("%.1f" % v for v in off_runs),
+                     (t_on - t_off) / max(t_off, 1e-9) * 100))
+            check("性能上界：接线态（3 轮中位）≤ 桩空转态（3 轮中位）×3 + 30µs（中位抗长尾）",
+                  t_on <= t_off * 3 + 30.0,
+                  "on=%.1f off=%.1f" % (t_on, t_off))
     finally:
         for r_ in roots:
             shutil.rmtree(r_, ignore_errors=True)

package/md_cg/test_interop.py

@@ -62,7 +62,7 @@ def main():
         check("3a 合规 verdict 通过门禁", isinstance(v, dict) and v["passed"] == 27)
         for name, bad in [
             ("3b 绝对路径拦截",
-             {**v, "details": [{"log": r"D:\program\secret\path.txt"}]}),
+             {**v, "details": [{"log": r"D:\secret\path.txt"}]}),
             ("3c prompt 字段拦截",
              {**v, "prompt": "你是一个…"}),
             ("3d API key 拦截",

package/md_cg/test_issue52_scan_condition_first.py

@@ -11,7 +11,7 @@
   (c) 截断完全静默（返回体无任何标记）——「读不全」与「读不到」不可区分；
   (d) 被排除集合随索引重排漂移（运行期 dict 追加 vs 重建后 sorted 写盘）。
 三接口 timeline/anchors/consistency 共用 `_scan` 同病；MCP 面 `_stg_call` 不透传
-max_scan（永远默认口径）；DSH 侧 hooks.ts:423 auto-recall 直连 timeline 默认口径。
+max_scan（永远默认口径）；DSH 侧 hooks.ts:432 auto-recall 直连 timeline 默认口径。
 
 本批修法（设计者裁定 2026-10-03，原文）：「遇到更多的检索节点，应该要建立索引，
 明确检索条件，和不适用条件，而不是扩大节点数」⇒ 只落「条件先行 + 截断可观测」