@furongjun1999/dsh-memory 0.4.8 → 0.4.9

package/md_cg/test_comment_gate.py

@@ -0,0 +1,187 @@
+# -*- coding: utf-8 -*-
+"""comment_gate 验收（环二入口）。
+
+验收口径：
+1. 唯一真源：阈值取 refine.GATE_MIN_PASS_RATE（本模块不再定义一份）
+2. 候选判据机械：只有「代码节点且正文缺生效条件」入池（有注释的那条不入池）
+3. 工单：确定性抽样（同 seed 同工单）+ 带源码落点（code_ref 坐标 + 落点规则）
+4. 只读纪律：apply 只落留痕，**不改任何节点**（改前改后内容 sha 相同）
+5. 闸门：<0.90 不放行、>=0.90 放行、无批次不放行
+6. 分派：run(action=comment_gate) 走只读工单；comment_gate_verdict 走闸门
+"""
+from __future__ import annotations
+
+import os
+import sys
+import tempfile
+
+from . import codeindex, comment_gate, refindex, refine
+from .mdcos import MdCGSecure
+from .security import Principal
+
+_ok = 0
+_bad = []
+
+
+def _check(name, cond, detail=""):
+    global _ok
+    if cond:
+        _ok += 1
+        print("  ok   %s" % name)
+    else:
+        _bad.append("%s %s" % (name, detail))
+        print("  FAIL %s %s" % (name, detail))
+
+
+ANNOTATED = ('"""样本。"""' + chr(10) + chr(10) + chr(10) +
+             "# 生效条件：入参 x 为整数" + chr(10) +
+             "def with_cond(x):" + chr(10) + "    return x + 1" + chr(10))
+PLAIN = ('"""样本。"""' + chr(10) + chr(10) + chr(10) +
+         "def no_cond(x):" + chr(10) + "    return x + 2" + chr(10))
+
+
+def _mk_graph(root):
+    p = Principal(tenant="default", actor="t_test", role="designer",
+                  can_write=True, can_admin=True)
+    return MdCGSecure(root, principal=p)
+
+
+def main():
+    print("=" * 68)
+    print("md_cg comment_gate 验收（代码符号条件化注释抽样闸门）")
+    print("=" * 68)
+    tmp = tempfile.mkdtemp(prefix="mdcg_comment_gate_")
+    root = os.path.join(tmp, "root")
+    code = os.path.join(tmp, "code")
+    os.makedirs(code)
+    with open(os.path.join(code, "a.py"), "w", encoding="utf-8") as f:
+        f.write(ANNOTATED)
+    with open(os.path.join(code, "b.py"), "w", encoding="utf-8") as f:
+        f.write(PLAIN)
+    cg = _mk_graph(root)
+    items, errors, stats = refindex.index_dir(code, kind="code_ref",
+                                             ledger=refindex.Ledger(cg.root))
+    ids, _sens = refindex.add_items(cg, items, kind="code_ref", root=code)
+    cg.flush()
+    print("索引条目=%d 节点=%d 错误=%d" % (len(items), len(ids), len(errors)))
+
+    _check("阈值唯一真源=refine.GATE_MIN_PASS_RATE",
+           comment_gate.GATE_MIN_PASS_RATE == refine.GATE_MIN_PASS_RATE,
+           str(comment_gate.GATE_MIN_PASS_RATE))
+    by_name = {it["name"]: codeindex.node_id(it) for it in items}
+    cands, cmeta = comment_gate.candidates(cg)
+    # 注：模块级条目同样缺生效条件 → 也是合法候选；故按符号名判定而非计条数
+    _check("候选判据机械：缺生效条件者入池、已注释者不入池",
+           by_name["no_cond"] in cands and by_name["with_cond"] not in cands,
+           "cands=%s no=%s with=%s" % (len(cands), by_name["no_cond"],
+                                       by_name["with_cond"]))
+    _check("候选池状态如实上报（pool/candidates 与实取一致）",
+           cmeta.get("pool", 0) == len(items) and cmeta.get("candidates") == len(cands),
+           str(cmeta))
+    p1 = comment_gate.plan(cg, n=20, seed="cmt-1")
+    p2 = comment_gate.plan(cg, n=20, seed="cmt-1")
+    _check("同 seed 工单确定性（sample 与 sha 一致）",
+           p1["sample"] == p2["sample"] and p1["sample_sha"] == p2["sample_sha"],
+           str(p1["sample"]))
+    _check("工单只读（dry_run=True，不改节点）", p1["dry_run"] is True and p1["readonly"] is True)
+    it = next((x for x in p1["items"]
+               if (x.get("code_ref") or {}).get("name") == "no_cond"), {})
+    ref = (it.get("code_ref") or {})
+    _check("工单带源码落点坐标（path/lineno/end）",
+           ref.get("path") == "b.py" and ref.get("lineno") and ref.get("end"),
+           str(ref))
+    _check("工单带落点规则（两窗口，契约 §三.2）",
+           "leading" in str(it.get("landing_rule") or "")
+           and "body" in str(it.get("landing_rule") or ""),
+           str(it.get("landing_rule"))[:80])
+    _check("工单缺项口径=缺生效条件",
+           it.get("ccg_missing") == ["生效条件"], str(it.get("ccg_missing")))
+    _check("SPEC 四项核对清单与 0.90 闸门口径在案",
+           len(comment_gate.SPEC["review_items"]) == 4
+           and comment_gate.SPEC["gate"]["min_pass_rate"] == 0.90)
+
+    nid = by_name["no_cond"]
+    before = (cg.get(nid) or {}).get("content") or ""
+    r = comment_gate.apply(cg, n=20, seed="cmt-1", verdicts=["忠实"],
+                           actor="t_test", note="自测")
+    after = (cg.get(nid) or {}).get("content") or ""
+    _check("apply 落留痕且通过率=1.0", r["ok"] and r["gate"]["pass_rate"] == 1.0,
+           str(r.get("gate")))
+    _check("apply 之后放行（>=0.90）", r["gate"]["expand_allowed"] is True)
+    _check("apply 不改节点（内容逐字相同）", before == after and before != "")
+    _check("留痕文件已落盘", os.path.exists(os.path.join(root, comment_gate.LOG_NAME)))
+    g1 = comment_gate.gate(cg)
+    _check("gate 复算通过率=1.0 且放行", g1["pass_rate"] == 1.0 and g1["expand_allowed"] is True,
+           str(g1))
+    r2 = comment_gate.apply(cg, n=20, seed="cmt-1", verdicts=[False, False, False],
+                            actor="t_test", batch="low")
+    _check("低通过率（0/3）不放行且给原因",
+           r2["gate"]["expand_allowed"] is False and "rate" in str(r2["gate"]["reason"]),
+           str(r2["gate"]))
+    g2 = comment_gate.gate(cg, batch="low")
+    _check("gate(batch=low) 复算为不放行", g2["expand_allowed"] is False, str(g2))
+    g3 = comment_gate.gate(cg, batch="不存在")
+    _check("无批次 → no_batch 不放行", g3["expand_allowed"] is False
+           and g3["reason"] == "no_batch", str(g3))
+    run1 = comment_gate.run(cg, "comment_gate", n=5)
+    _check("run(comment_gate) 走只读工单", run1["dry_run"] is True
+           and run1["action"] == "comment_gate")
+    from . import mcp_server
+    d1 = mcp_server._maintain_call(cg, {"op": "maintain", "action": "comment_gate", "n": 5})
+    _check("经 maintain 分派可达（op 面接线）",
+           d1.get("action") == "comment_gate" and d1.get("dry_run") is True
+           and d1.get("readonly") is True, str(d1.get("action")))
+    d2 = mcp_server._maintain_call(cg, {"action": "comment_gate_gate"})
+    _check("经 maintain 分派可取闸门（comment_gate_gate）",
+           "expand_allowed" in d2, str(sorted(d2.keys()))[:90])
+    from .mdcos import MdCGSecure as _S
+    from .security import AccessDenied as _AD
+    cg_low = _S(root, principal=Principal(tenant="default", actor="t_low",
+                                          role="record", can_write=True,
+                                          can_admin=False))
+    denied = False
+    try:
+        comment_gate.apply(cg_low, n=5, seed="cmt-1", verdicts=["忠实"])
+    except _AD:
+        denied = True
+    except Exception:
+        denied = False
+    _check("apply 需管理权（闸门自持在模块内，分发面不漏挂）", denied)
+    d3 = mcp_server._maintain_call(cg, {"action": "comment_gate_verdict"})
+    _check("经 maintain 分派可取闸门（别名 comment_gate_verdict）",
+           "expand_allowed" in d3, str(sorted(d3.keys()))[:90])
+    run2 = comment_gate.run(cg, "comment_gate_verdict")
+    _check("run(comment_gate_verdict) 走闸门", "expand_allowed" in run2)
+    raised = False
+    try:
+        comment_gate.run(cg, "nope")
+    except ValueError:
+        raised = True
+    _check("未知 action fail-closed", raised)
+    # ---- 源级候选枚举（绕开重索引代价；不依赖认知图）----
+    sp1 = comment_gate.plan_sources(code, n=20, seed="src-1")
+    _check("源级枚举可见未注释符号（不依赖认知图）",
+           sp1["examined"] >= 1 and sp1["candidates"] >= 1 and sp1["sampled"] >= 1,
+           "examined=%s cands=%s sampled=%s" % (sp1["examined"], sp1["candidates"],
+                                                sp1["sampled"]))
+    src_names = [it["name"] for it in sp1["items"]]
+    _check("源级工单排除已注释符号（no_cond 在、with_cond 不在）",
+           "no_cond" in src_names and "with_cond" not in src_names, str(src_names))
+    _check("源级工单带源码坐标（path/lineno/end）",
+           all((it["code_ref"].get("path") and it["code_ref"].get("lineno")
+                and it["code_ref"].get("end")) for it in sp1["items"]))
+    _check("源级抽样确定性（同 seed 同工单）",
+           comment_gate.plan_sources(code, n=20, seed="src-1")["items"] == sp1["items"])
+    _check("源级工单只读且声明模式",
+           sp1["dry_run"] is True and sp1["mode"] == "source_level")
+    _check("源级枚举经 run 分派可达",
+           comment_gate.run(cg, "comment_gate_sources", n=5)["mode"] == "source_level")
+    print()
+    print("PASS %d / FAIL %d" % (_ok, len(_bad)))
+    for b in _bad:
+        print("  - " + b)
+    return 1 if _bad else 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

package/md_cg/test_cond_compose_anchors.py

@@ -0,0 +1,76 @@
+# -*- coding: utf-8 -*-
+"""cond_compose 行为锚点（在途 WIP 并集解）探针：脚本式，exit code 定成败。
+
+背景：2026-09-19 收口两线时，工作区残留一处 stash-pop 冲突（md_cg/cond_compose.py）：
+  ours(HEAD)=compose 的「生效条件」注释；theirs(stash)=新增 behavior_anchors()。
+两侧新增互不重叠 → 并集解（本分支）。该函数当前**全库无调用**，故本探针只把
+「冲突解正确 + 函数行为与文档一致」钉住，配线与否由使用者裁决。
+
+用法：python -X utf8 -m md_cg.test_cond_compose_anchors
+"""
+import os
+import sys
+
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+from md_cg import cond_compose as cc  # noqa: E402
+
+passed = 0
+failed = 0
+
+
+def check(name, cond, detail=""):
+    global passed, failed
+    if cond:
+        passed += 1
+        print("  [PASS] " + name)
+    else:
+        failed += 1
+        print("  [FAIL] " + name + "  " + detail)
+
+
+def main():
+    # 1) 并集完整性：冲突两侧的东西都在，且文件里没有冲突标记
+    src = open(cc.__file__, encoding="utf-8").read()
+    check("并集：compose 与 behavior_anchors 同时在",
+          hasattr(cc, "compose") and hasattr(cc, "behavior_anchors"))
+    check("并集：无冲突标记", "<<<<<<<" not in src and ">>>>>>>" not in src)
+    check("并集：compose 的生效条件注释未丢", "调用 compose 时必须提供 rec" in src)
+
+    # 2) behavior_anchors 行为（按实现口径逐条断言）
+    rec = {"name": "f", "doc_head": "  计算 a 的摘要  ",
+           "guards": [{"early": True, "cond": "a"},
+                      {"early": False, "cond": "a"},
+                      {"early": True, "cond": "a and b"}],
+           "returns": ["a.strip()", "LIMIT + 1", "local_only"]}
+    doc, guards, rets = cc.behavior_anchors(rec, {"a"})
+    check("doc 取 doc_head 并去空白", doc == "计算 a 的摘要", repr(doc))
+    # explained_guards 的两道筛：early 必须为真；cond 内标识符必须全在 allowed∪BUILTINS∪self/cls
+    # （故 "a" 入选、"a and b" 因 b 不在 allowed 被剔、"a is None" 会因 is 不是 builtins 被剔）
+    check("guards 只留 early 且标识符合规者（走 explained_guards）",
+          [g.get("cond") for g in guards] == ["a"], str(guards))
+    check("returns 只留含 allowed 标识符者", rets == ["a.strip()"], str(rets))
+    _d2, _g2, rets2 = cc.behavior_anchors(rec, {"a", "LIMIT"})
+    check("allowed 含模块常量时其返回表达式入选（调用方负责传 入参∪常量）",
+          rets2 == ["a.strip()", "LIMIT + 1"], str(rets2))
+
+    # 3) 「只有形参清单、无任何行为锚点」→ 三元组全空（调用方据此可判 BLINDSPOT）
+    empty = cc.behavior_anchors({"name": "g", "guards": [], "returns": []}, {"x"})
+    check("无形为锚点时返回空三元组", empty == ("", [], []), str(empty))
+    check("缺失 doc_head/returns 键不抛异常",
+          cc.behavior_anchors({"name": "h"}, {"x"}) == ("", [], []))
+
+    # 4) 未回归：compose 仍能产出以「。」结尾的条件串
+    s = cc.compose({"name": "f", "required": ["a"], "optional": [], "externals": [],
+                    "guards": [], "returns": ["a.strip()"], "doc_head": "摘要"})
+    check("compose 未回归（非空且以「。」结尾）",
+          isinstance(s, str) and s.endswith("。") and len(s) > 4, s[:60])
+
+    print("\ntest_cond_compose_anchors: %d 通过 / %d 失败" % (passed, failed))
+    return 0 if failed == 0 else 1
+
+
+if __name__ == "__main__":
+    if hasattr(sys.stdout, "reconfigure"):
+        sys.stdout.reconfigure(encoding="utf-8")
+    sys.exit(main())

package/md_cg/test_condition_anchor.py

@@ -0,0 +1,82 @@
+# -*- coding: utf-8 -*-
+"""condition_anchor 验收：四类裁决 + 边界（self/cls、kwonly、默认值常量、不可判）。
+
+验收口径：本模块只判「锚点是否存在」，不判语义正确性；不可判必须 BLINDSPOT，不得默认通过。
+"""
+from __future__ import annotations
+
+from . import condition_anchor as ca
+
+_ok = 0
+_bad = []
+
+
+def _check(name, cond, detail=""):
+    global _ok
+    if cond:
+        _ok += 1
+        print("  ok   %s" % name)
+    else:
+        _bad.append("%s %s" % (name, detail))
+        print("  FAIL %s %s" % (name, detail))
+
+
+SRC_REQ = "def f(root, patterns=None):" + chr(10) + "    return os.walk(root)" + chr(10)
+SRC_KW = "def g(a, *, limit, mode=1):" + chr(10) + "    return scan(a, limit, mode)" + chr(10)
+SRC_CLS = "class C:" + chr(10) + "    def m(self, x):" + chr(10) + "        return x" + chr(10)
+SRC_BARE = "def h():" + chr(10) + "    return 1" + chr(10)
+SRC_CONST = "def k(x=1):" + chr(10) + "    return x + LIMIT" + chr(10)
+SRC_BROKEN = "def n(:" + chr(10) + "    pass" + chr(10)
+
+
+def main():
+    print("=" * 68)
+    print("md_cg condition_anchor 验收（候选生效条件的锚点机械判据）")
+    print("=" * 68)
+    r = ca.judge("root 为可遍历路径", SRC_REQ)
+    _check("A1 锚定必需参数 → ANCHORED 且 ok", r["verdict"] == "ANCHORED" and r["ok"] is True, str(r))
+    _check("A2 锚点清单含 root（不含 patterns）",
+           r["anchors"] == ["root"] and "patterns" in r["optional"], str(r["anchors"]))
+    _check("A3 默认值参数归入 optional", r["optional"] == ["patterns"], str(r["optional"]))
+    r = ca.judge("limit 为正整数", SRC_KW)
+    _check("A4 kwonly 无默认值算必需参数", r["verdict"] == "ANCHORED" and r["anchors"] == ["limit"], str(r))
+    r = ca.judge("x 为整数", SRC_CLS)
+    _check("A5 class 内方法：self 被跳过、锚定 x", r["verdict"] == "ANCHORED" and r["anchors"] == ["x"], str(r))
+    r = ca.judge("LIMIT 已定义", SRC_CONST)
+    _check("A6 体内外部引用（常量）可作锚点", r["verdict"] == "ANCHORED" and r["anchors"] == ["LIMIT"], str(r))
+    r = ca.judge("该函数总是正确", SRC_REQ)
+    _check("A7 无锚点 → WEAK 且 ok=False（不默认通过）",
+           r["verdict"] == "WEAK" and r["ok"] is False, str(r))
+    r = ca.judge("源文件存在于本地仓", SRC_REQ)
+    _check("A8 索引元条件话术 → REJECT_META", r["verdict"] == "REJECT_META" and r["meta_marks"], str(r))
+    r = ca.judge("x 为整数", SRC_BARE)
+    _check("A9 无参数且无引用 → BLINDSPOT（不猜测）", r["verdict"] == "BLINDSPOT", str(r))
+    r = ca.judge("x 为整数", SRC_BROKEN)
+    _check("A10 源码不可解析 → BLINDSPOT", r["verdict"] == "BLINDSPOT", str(r))
+    r = ca.judge("root 为可遍历路径", SRC_REQ)
+    r2 = ca.judge("root 为可遍历路径", SRC_REQ)
+    _check("A11 判定确定性（同输入同结果）", r == r2)
+    _check("A12 prefix/suffix 仅回显、不参与判定",
+           ca.judge("root 为可遍历路径", SRC_REQ, prefix="【1】")["verdict"] == "ANCHORED")
+    b = ca.judge_batch([("a", "root 为可遍历路径", SRC_REQ),
+                        ("b", "该函数总是正确", SRC_REQ),
+                        ("c", "源文件存在", SRC_REQ)])
+    _check("A13 批量统计正确（ANCHORED/WEAK/REJECT_META 各 1）",
+           b["stats"].get("ANCHORED") == 1 and b["stats"].get("WEAK") == 1
+           and b["stats"].get("REJECT_META") == 1 and b["ok"] == 1, str(b["stats"]))
+    _r = ca.judge("tree 为 AST 节点", "class NoInit:" + chr(10) + "    x = 1" + chr(10))
+    _check("A14 无 __init__ 的类不崩溃并给出裁决",
+           _r["verdict"] in ("ANCHORED", "WEAK", "BLINDSPOT"), str(_r))
+    _r = ca.judge("x 为整数", "class C:" + chr(10) + "    def __init__(self, x):" + chr(10) + "        self.x = x" + chr(10))
+    _check("A15 有 __init__ 的类按其形参锚定",
+           _r["verdict"] == "ANCHORED" and _r["anchors"] == ["x"], str(_r))
+    print()
+    print("PASS %d / FAIL %d" % (_ok, len(_bad)))
+    for x in _bad:
+        print("  - " + x)
+    return 1 if _bad else 0
+
+
+if __name__ == "__main__":
+    import sys
+    sys.exit(main())