@furongjun1999/dsh-memory 0.4.11 → 0.5.0

package/md_cg/test_census_prune.py

@@ -1,81 +1,81 @@
-# -*- coding: utf-8 -*-
-"""PR #15 census 归档区剪枝探针（脚本式，exit code 定成败）。
-
-用法：python -X utf8 -m md_cg.test_census_prune
-覆盖：
-- ARCHIVE_DIRS 五个目录整体剪枝（含 _protected_history/<id>/<ts>.md 两层下钻）
-- _protected_history_* 前缀目录剪枝
-- 普通子目录不误伤
-- 无 frontmatter 的散落 md 仍被跳过（既有语义保持）
-- 真实库（data/mdcg）上活跃数不再被墓碑/快照稀释
-"""
-import os
-import sys
-import tempfile
-
-sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
-
-from md_cg import census  # noqa: E402
-
-FM = "---\nid: {nid}\nlayer: knowledge\n---\n正文\n"
-passed = 0
-
-
-def check(name, cond, detail=""):
-    global passed
-    if cond:
-        passed += 1
-        print("  [PASS] " + name)
-    else:
-        print("  [FAIL] " + name + "  " + detail)
-
-
-def main():
-    root = tempfile.mkdtemp(prefix="census_prune_")
-    nodes = {
-        # (相对路径, 是否活跃)
-        ("act_root.md", True),
-        ("sub", "act_sub.md", True),          # 普通子目录不误伤
-        ("trash", "t1.md", False),
-        ("_protected_history", "n1", "20260101_000000.md", False),  # 两层下钻
-        ("_protected", "p1.md", False),
-        ("_index_log", "i1.md", False),
-        ("hippocampus", "h1.md", False),
-        ("_protected_history_extra", "x.md", False),  # 前缀分支
-        ("note.md", None),                    # 无 frontmatter 散落文档 → 跳过
-    }
-    for n in nodes:
-        parts = [p for p in n if p is not True and p is not False and p is not None]
-        active = n[-1] is True
-        rel = os.path.join(*parts) if parts else "note.md"
-        p = os.path.join(root, rel)
-        os.makedirs(os.path.dirname(p), exist_ok=True)
-        with open(p, "w", encoding="utf-8") as f:
-            f.write("" if n[-1] is None else FM.format(nid=os.path.splitext(parts[-1])[0]))
-    rows = census.load(root)
-    ids = sorted(r[0] for r in rows)
-    check("仅活跃节点入普查（2 个）", ids == ["act_root", "act_sub"], str(ids))
-    for bad in ("t1", "p1", "i1", "h1", "x", "20260101_000000"):
-        check(f"归档区 {bad} 被剪枝", bad not in ids, str(ids))
-
-    # 真实库对比：PR 版活跃数 vs 全遍历数（含归档区）
-    real = os.path.join(os.path.dirname(os.path.dirname(
-        os.path.abspath(__file__))), "data", "mdcg")
-    if os.path.isdir(real):
-        n_active = len(census.load(real))
-        n_all = sum(1 for dp, _ds, fs in os.walk(real) for f in fs
-                    if f.endswith(".md"))
-        print(f"\n  真实库 {real}")
-        print(f"  PR 版活跃节点 {n_active} / 全遍历 {n_all}"
-              f"（归档区剪掉 {n_all - n_active}）")
-        check("真实库剪枝后不再被墓碑稀释", n_active < n_all,
-              f"{n_active} vs {n_all}")
-
-    print(f"\ntest_census_prune: {passed} 断言通过")
-    return 0 if passed >= 7 else 1
-
-
-if __name__ == "__main__":
-    if hasattr(sys.stdout, "reconfigure"):
-        sys.stdout.reconfigure(encoding="utf-8")
-    sys.exit(main())
+# -*- coding: utf-8 -*-
+"""PR #15 census 归档区剪枝探针（脚本式，exit code 定成败）。
+
+用法：python -X utf8 -m md_cg.test_census_prune
+覆盖：
+- ARCHIVE_DIRS 五个目录整体剪枝（含 _protected_history/<id>/<ts>.md 两层下钻）
+- _protected_history_* 前缀目录剪枝
+- 普通子目录不误伤
+- 无 frontmatter 的散落 md 仍被跳过（既有语义保持）
+- 真实库（data/mdcg）上活跃数不再被墓碑/快照稀释
+"""
+import os
+import sys
+import tempfile
+
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+from md_cg import census  # noqa: E402
+
+FM = "---\nid: {nid}\nlayer: knowledge\n---\n正文\n"
+passed = 0
+
+
+def check(name, cond, detail=""):
+    global passed
+    if cond:
+        passed += 1
+        print("  [PASS] " + name)
+    else:
+        print("  [FAIL] " + name + "  " + detail)
+
+
+def main():
+    root = tempfile.mkdtemp(prefix="census_prune_")
+    nodes = {
+        # (相对路径, 是否活跃)
+        ("act_root.md", True),
+        ("sub", "act_sub.md", True),          # 普通子目录不误伤
+        ("trash", "t1.md", False),
+        ("_protected_history", "n1", "20260101_000000.md", False),  # 两层下钻
+        ("_protected", "p1.md", False),
+        ("_index_log", "i1.md", False),
+        ("hippocampus", "h1.md", False),
+        ("_protected_history_extra", "x.md", False),  # 前缀分支
+        ("note.md", None),                    # 无 frontmatter 散落文档 → 跳过
+    }
+    for n in nodes:
+        parts = [p for p in n if p is not True and p is not False and p is not None]
+        active = n[-1] is True
+        rel = os.path.join(*parts) if parts else "note.md"
+        p = os.path.join(root, rel)
+        os.makedirs(os.path.dirname(p), exist_ok=True)
+        with open(p, "w", encoding="utf-8") as f:
+            f.write("" if n[-1] is None else FM.format(nid=os.path.splitext(parts[-1])[0]))
+    rows = census.load(root)
+    ids = sorted(r[0] for r in rows)
+    check("仅活跃节点入普查（2 个）", ids == ["act_root", "act_sub"], str(ids))
+    for bad in ("t1", "p1", "i1", "h1", "x", "20260101_000000"):
+        check(f"归档区 {bad} 被剪枝", bad not in ids, str(ids))
+
+    # 真实库对比：PR 版活跃数 vs 全遍历数（含归档区）
+    real = os.path.join(os.path.dirname(os.path.dirname(
+        os.path.abspath(__file__))), "data", "mdcg")
+    if os.path.isdir(real):
+        n_active = len(census.load(real))
+        n_all = sum(1 for dp, _ds, fs in os.walk(real) for f in fs
+                    if f.endswith(".md"))
+        print(f"\n  真实库 {real}")
+        print(f"  PR 版活跃节点 {n_active} / 全遍历 {n_all}"
+              f"（归档区剪掉 {n_all - n_active}）")
+        check("真实库剪枝后不再被墓碑稀释", n_active < n_all,
+              f"{n_active} vs {n_all}")
+
+    print(f"\ntest_census_prune: {passed} 断言通过")
+    return 0 if passed >= 7 else 1
+
+
+if __name__ == "__main__":
+    if hasattr(sys.stdout, "reconfigure"):
+        sys.stdout.reconfigure(encoding="utf-8")
+    sys.exit(main())

package/md_cg/test_cond_compose_anchors.py

@@ -1,76 +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())
+# -*- 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())