argo-search 2.9.7 → 2.9.8

package/sub-skills/local-seek/tests/test_regression_20261003.py

@@ -1,169 +0,0 @@
-#!/usr/bin/env python3
-"""2026-10-03 审计轮回归测试：四类假阴性 + run_query 契约 + timeout 下传。
-
-守的四个实测缺陷（全部先在真机复现再修）：
-  1. run_query 目录不存在分支 print + 裸 return 1，违反 (text, rc) 契约，
-     MCP/include-local 进程内调用方解包直接 TypeError。
-  2. --since/--until 在 max 截断之后过滤：旧文件占满截断池，新命中被吃掉
-     （实测 20 个新文件只报出 8 个）。
-  3. build_patterns 中文分支不拆英文多词：「性能 asyncio tutorial」的英文段
-     整段化作固定短语，单词条目全部漏掉（eb0ed38 只修了纯英文分支）。
-  4. --scope doc 把 md/txt 从 exts 减掉，文档场景恰好漏掉最常搜的笔记。
-
-运行：
-  cd ~/.agents/skills/argo
-  python3 -m pytest sub-skills/local-seek/tests/test_regression_20261003.py -q
-"""
-
-from __future__ import annotations
-
-import io
-import json
-import os
-import sys
-import tempfile
-import unittest
-from contextlib import redirect_stdout
-from pathlib import Path
-from unittest import mock
-
-SCRIPTS = Path(__file__).resolve().parent.parent / "scripts"
-sys.path.insert(0, str(SCRIPTS))
-
-import seek as s  # noqa: E402
-
-
-def _run_main(argv):
-    """跑 seek.main()，捕获 stdout，返回 (exit_code, stdout)。"""
-    with mock.patch.object(sys, "argv", ["seek.py"] + argv):
-        buf = io.StringIO()
-        try:
-            with redirect_stdout(buf):
-                code = s.main()
-        except SystemExit as e:
-            code = e.code or 0
-    return code or 0, buf.getvalue()
-
-
-class TestBuildPatterns(unittest.TestCase):
-    """中英混排与纯英文分词（缺陷 3）。"""
-
-    def test_mixed_cjk_english_splits_words(self):
-        patterns, fixed = s.build_patterns("性能 asyncio tutorial")
-        self.assertIn("asyncio", patterns, "英文段必须按词拆开")
-        self.assertIn("tutorial", patterns)
-        self.assertIn("性能", patterns)
-        self.assertTrue(fixed)
-
-    def test_pure_english_multiword_unchanged(self):
-        patterns, _ = s.build_patterns("Python asyncio tutorial")
-        self.assertEqual(patterns, ["Python", "asyncio", "tutorial"])
-
-    def test_space_never_a_pattern(self):
-        patterns, _ = s.build_patterns("樯橹 灰飞烟灭")
-        self.assertNotIn("", patterns)
-        self.assertNotIn(" ", patterns)
-
-    def test_cjk_2gram_extension_kept(self):
-        patterns, _ = s.build_patterns("数据抓取")
-        self.assertIn("数据抓取", patterns)
-        self.assertIn("抓取", patterns)
-
-
-class TestSinceFilterBeforeTruncation(unittest.TestCase):
-    """--since 过滤必须发生在截断之前（缺陷 2，真实 rg 端到端）。"""
-
-    def test_new_files_not_swallowed_by_old(self):
-        if not s.tool_exists("rg"):
-            self.skipTest("本机无 rg")
-        with tempfile.TemporaryDirectory() as tmp:
-            ancient = 900000000  # 1998 年，稳落任何时间窗之外
-            for i in range(40):
-                p = Path(tmp) / f"old_{i:02d}.txt"
-                p.write_text("目标词命中 here\n")
-                os.utime(p, (ancient, ancient))
-            for i in range(20):
-                (Path(tmp) / f"new_{i:02d}.txt").write_text("目标词命中 here\n")
-            code, out = _run_main(
-                ["目标词", "--path", tmp, "--since", "1d", "--max", "30", "--json"])
-            self.assertEqual(code, 0)
-            self.assertEqual(json.loads(out)["count"], 20,
-                             "20 个新文件全部在窗口内，一个都不能被旧文件挤掉")
-
-    def test_no_window_returns_all(self):
-        if not s.tool_exists("rg"):
-            self.skipTest("本机无 rg")
-        with tempfile.TemporaryDirectory() as tmp:
-            for i in range(35):
-                (Path(tmp) / f"f_{i:02d}.txt").write_text("目标词命中 here\n")
-            code, out = _run_main(
-                ["目标词", "--path", tmp, "--max", "30", "--json"])
-            self.assertEqual(json.loads(out)["count"], 30)
-
-
-class TestScopeDocIncludesMd(unittest.TestCase):
-    """--scope doc 必须搜到 md（缺陷 4）。"""
-
-    def test_md_hit_in_doc_scope(self):
-        if not s.tool_exists("rg"):
-            self.skipTest("本机无 rg")
-        with tempfile.TemporaryDirectory() as tmp:
-            (Path(tmp) / "notes.md").write_text("检索词在这里\n")
-            code, out = _run_main(
-                ["检索词", "--path", tmp, "--scope", "doc", "--json"])
-            self.assertEqual(code, 0, f"--scope doc 应命中 md，输出：{out}")
-            data = json.loads(out)
-            self.assertTrue(any("notes.md" in r["path"] for r in data["results"]))
-
-
-class TestRunQueryContract(unittest.TestCase):
-    """run_query 必须始终返回 (text, rc)（缺陷 1）。"""
-
-    def test_nonexistent_dir_returns_tuple(self):
-        result = s.run_query(["x", "--path", "/nonexistent-seek-test-xyz"])
-        self.assertIsInstance(result, tuple, "返回裸 int 会让进程内调用方解包崩溃")
-        text, rc = result
-        self.assertIsInstance(text, str)
-        self.assertEqual(rc, 1)
-        self.assertIn("目录不存在", text)
-
-    def test_cli_still_prints_message(self):
-        code, out = _run_main(["x", "--path", "/nonexistent-seek-test-xyz"])
-        self.assertEqual(code, 1)
-        self.assertIn("目录不存在", out)
-
-
-class TestTimeoutDownpass(unittest.TestCase):
-    """structural/git 路径必须把 time_budget 下传给 run（缺陷 5）。"""
-
-    def test_structural_passes_timeout(self):
-        captured = {}
-        def fake_run(cmd, timeout=None, **kw):
-            captured["timeout"] = timeout
-            proc = mock.Mock()
-            proc.returncode = 1  # 无命中
-            proc.stdout = ""
-            return proc
-        with mock.patch.object(s, "run", side_effect=fake_run):
-            s.structural_search("裸except", "/tmp", [], 5, timeout=5.5)
-        self.assertEqual(captured.get("timeout"), 5.5,
-                         "structural_search 必须把 timeout 传给 run")
-
-    def test_git_log_passes_timeout(self):
-        captured = {}
-        def fake_run(cmd, timeout=None, **kw):
-            captured["timeout"] = timeout
-            proc = mock.Mock()
-            proc.returncode = 1
-            proc.stdout = ""
-            return proc
-        with tempfile.TemporaryDirectory() as tmp:
-            f = Path(tmp) / "a.txt"
-            f.write_text("x\n")
-            with mock.patch.object(s, "run", side_effect=fake_run):
-                s.git_log(str(f), timeout=7.5)
-        self.assertEqual(captured.get("timeout"), 7.5)
-
-
-if __name__ == "__main__":
-    unittest.main()

package/sub-skills/local-seek/tests/test_regression_20261006.py

@@ -1,146 +0,0 @@
-#!/usr/bin/env python3
-"""2026-10-06 性能轮回归测试：噪声档判定去 syscall + read_lines 单遍读取。
-
-实测背景（cProfile，Documents/GPT 搜 "import" --count，8930 个文件）：
-    _is_noise_path 独占 1.52s / 1.91s（80%），内含 35727 次 posix.getcwd、
-    103266 次 posix.lstat、35720 次 realpath。根因是每行都做
-    `Path(fp).resolve().relative_to(Path(root).resolve())`——resolve() 对每个
-    路径段 lstat 且反复 getcwd；而 _apply_noise_floor 还把根 resolve 了两遍、
-    每行判了两遍。
-
-修法：纯字符串归一化（normpath + 前缀比较），根只算一次、每行只判一次，
-全程零 syscall。本文件锁定两件事：
-  1. 语义不变——相对/绝对根、根自己叫 tests/、repos/、日期归档、越界路径
-     的判定结果与原实现一致；
-  2. 不得回退到 syscall——_is_noise_path 再调 Path.resolve / realpath / lstat
-     即报红（这是本轮性能收益的守门用例，比计时断言稳定）。
-
-read_lines 原实现先 open 数行、再 open 读区间（同一文件开两遍，大文件 IO
-翻倍），改为单遍流式；本文件锁定区间/越界/空文件三种行为不变。
-
-运行：
-  cd ~/.agents/skills/argo
-  python3 -m pytest sub-skills/local-seek/tests/test_regression_20261006.py -q
-"""
-
-from __future__ import annotations
-
-import os
-import sys
-import tempfile
-import unittest
-from pathlib import Path
-from unittest import mock
-
-SCRIPTS = Path(__file__).resolve().parent.parent / "scripts"
-sys.path.insert(0, str(SCRIPTS))
-
-import seek as s  # noqa: E402
-
-
-class TestNoisePathSemantics(unittest.TestCase):
-    """去 syscall 之后，噪声档判定语义必须与原实现逐条一致。"""
-
-    def test_root_named_tests_is_not_noise(self):
-        # 搜索根自己叫 tests/ 时整棵树不得被判噪声（原实现的核心理由）
-        self.assertFalse(s._is_noise_path("/x/tests/a.py", "/x/tests"))
-
-    def test_src_under_root_is_clean(self):
-        self.assertFalse(s._is_noise_path("/x/src/app.py", "/x"))
-
-    def test_repos_under_root_is_noise(self):
-        self.assertTrue(s._is_noise_path("/x/repos/lib.py", "/x"))
-
-    def test_date_archive_dir_is_noise(self):
-        self.assertTrue(s._is_noise_path("/x/2026-09-26_吸纳/a.md", "/x"))
-
-    def test_relative_root(self):
-        # 默认根 "." —— rg 输出相对路径时的常见形态
-        self.assertTrue(s._is_noise_path("repos/lib.py", "."))
-        self.assertFalse(s._is_noise_path("src/app.py", "."))
-
-    def test_out_of_root_falls_back_to_full_path(self):
-        # 不在根之下（软链/越界）：退回整条路径判定
-        self.assertTrue(s._is_noise_path("/other/repos/lib.py", "/x"))
-        self.assertFalse(s._is_noise_path("/other/src/lib.py", "/x"))
-
-    def test_noise_floor_clean_first(self):
-        rows = [("/x/repos/a.py", 0, ""), ("/x/src/b.py", 0, "")]
-        out = s._apply_noise_floor(rows, "/x", 30)
-        self.assertEqual(out[0][0], "/x/src/b.py",
-                         "真源必须排在噪声档之前")
-
-    def test_noise_floor_backfills_when_clean_short(self):
-        # 真源不足时噪声档补满（不是排除，保证仍可达）
-        rows = [("/x/repos/a.py", 0, ""), ("/x/src/b.py", 0, "")]
-        out = s._apply_noise_floor(rows, "/x", 2)
-        self.assertEqual(len(out), 2)
-
-
-class TestNoisePathNoSyscall(unittest.TestCase):
-    """性能守门：判定路径不得再触发任何文件系统解析调用。"""
-
-    def test_does_not_call_path_resolve(self):
-        with mock.patch.object(Path, "resolve",
-                               side_effect=AssertionError("又调 Path.resolve() 了")):
-            rows = [(f"/root/repos/f{i}.py", 0, "") for i in range(100)]
-            s._apply_noise_floor(rows, "/root", 30)
-
-    def test_does_not_call_realpath_or_lstat(self):
-        with mock.patch.object(os.path, "realpath",
-                               side_effect=AssertionError("又调 realpath() 了")), \
-             mock.patch.object(os, "lstat",
-                               side_effect=AssertionError("又调 lstat() 了")):
-            rows = [(f"/root/repos/f{i}.py", 0, "") for i in range(100)]
-            s._apply_noise_floor(rows, "/root", 30)
-
-
-class TestReadLinesSinglePass(unittest.TestCase):
-    """read_lines 单遍读取：区间/夹取/越界/空文件行为与原实现一致。"""
-
-    def _write(self, tmp, text):
-        f = Path(tmp) / "a.txt"
-        f.write_text(text, encoding="utf-8")
-        return str(f)
-
-    def test_range_slice(self):
-        with tempfile.TemporaryDirectory() as tmp:
-            fp = self._write(tmp, "\n".join(f"L{i}" for i in range(1, 11)) + "\n")
-            text, rc = s.read_lines(fp, "3-5")
-            self.assertEqual(rc, 0)
-            self.assertIn("3: L3", text)
-            self.assertIn("5: L5", text)
-            self.assertNotIn("6: L6", text)
-            self.assertIn("共 10 行", text)
-
-    def test_clamped_to_total(self):
-        with tempfile.TemporaryDirectory() as tmp:
-            fp = self._write(tmp, "L1\nL2\n")
-            text, rc = s.read_lines(fp, "1-99")
-            self.assertEqual(rc, 0)
-            self.assertIn("第 1-2 行", text)
-
-    def test_start_past_end(self):
-        with tempfile.TemporaryDirectory() as tmp:
-            fp = self._write(tmp, "L1\n")
-            text, rc = s.read_lines(fp, "5-6")
-            self.assertEqual(rc, 1)
-            self.assertIn("只有 1 行", text)
-
-    def test_empty_file(self):
-        with tempfile.TemporaryDirectory() as tmp:
-            fp = self._write(tmp, "")
-            text, rc = s.read_lines(fp, "1-2")
-            self.assertEqual(rc, 1)
-            self.assertIn("只有 0 行", text)
-
-    def test_bad_spec(self):
-        with tempfile.TemporaryDirectory() as tmp:
-            fp = self._write(tmp, "L1\n")
-            text, rc = s.read_lines(fp, "abc")
-            self.assertEqual(rc, 1)
-            self.assertIn("N-M", text)
-
-
-if __name__ == "__main__":
-    unittest.main()

package/sub-skills/local-seek/tests/test_seek_degradation.py

@@ -1,111 +0,0 @@
-#!/usr/bin/env python3
-"""local-seek 降级链与 --count 排序回归测试（P0-2 / P0-4，2026-09-13）。
-
-守的两个缺陷：
-  1. 降级链曾把「工具缺失」伪装成「搜索结论」：scope=code 且缺 rg 时落
-     mdfind（Spotlight 不索引源码内容）→ 搜 argo 自身符号返回 24 条外部
-     陈旧副本、漏掉正主，还报「未找到匹配（mdfind）」。修复后：缺 rg 直落
-     grep，两者皆无才显式报错。
-  2. --count 曾按 rg 原始输出顺序（目录遍历序）截断，「哪些文件命中最多」
-     系统性答错。修复后：全量收集→按命中数降序→取前 N。
-
-运行：
-  cd ~/.agents/skills/argo
-  python3 -m pytest sub-skills/local-seek/tests/test_seek_degradation.py -q
-"""
-
-from __future__ import annotations
-
-import io
-import json
-import sys
-import tempfile
-import unittest
-from contextlib import redirect_stdout
-from pathlib import Path
-from unittest import mock
-
-SCRIPTS = Path(__file__).resolve().parent.parent / "scripts"
-sys.path.insert(0, str(SCRIPTS))
-
-import seek as s  # noqa: E402
-
-
-def _run_main(argv):
-    """跑 seek.main()，捕获 stdout，返回 (exit_code, stdout)。"""
-    with mock.patch.object(sys, "argv", ["seek.py"] + argv):
-        buf = io.StringIO()
-        try:
-            with redirect_stdout(buf):
-                code = s.main()
-        except SystemExit as e:
-            code = e.code or 0
-    return code or 0, buf.getvalue()
-
-
-class TestDegradationChain(unittest.TestCase):
-    """scope=code 缺 rg 时的降级（全部 mock，不依赖本机装了什么）。"""
-
-    def _argv(self, tmp: str) -> list[str]:
-        return ["def needle", "--scope", "code", "--path", tmp, "--json"]
-
-    def test_rg_missing_falls_to_grep(self):
-        with tempfile.TemporaryDirectory() as tmp:
-            (Path(tmp) / "a.py").write_text("def needle(x):\n    return x\n")
-            with mock.patch.object(s, "tool_exists",
-                                   lambda name: name in ("grep", "mdfind")), \
-                 mock.patch.object(s, "resolve_grep", lambda: "/usr/bin/grep"), \
-                 mock.patch.object(s, "grep_search",
-                                   lambda *a, **k: ([(str(Path(tmp) / "a.py"), 1, "def needle(x)")], None)):
-                code, out = _run_main(self._argv(tmp))
-        self.assertEqual(code, 0)
-        data = json.loads(out)
-        self.assertEqual(data["engine"], "grep",
-                         "缺 rg 时必须落 grep，不能落 mdfind")
-
-    def test_rg_and_grep_missing_reports_tools_not_nomatch(self):
-        with tempfile.TemporaryDirectory() as tmp:
-            with mock.patch.object(s, "tool_exists", lambda name: False), \
-                 mock.patch.object(s, "resolve_grep", lambda: None):
-                code, out = _run_main(self._argv(tmp))
-        self.assertEqual(code, 1)
-        self.assertIn("rg", out)
-        self.assertIn("grep", out)
-        self.assertNotIn("未找到匹配", out,
-                         "工具缺失不得伪装成搜索结论")
-
-    def test_real_rg_still_used(self):
-        """rg 在位时行为不变（防把好路径改坏）。"""
-        with tempfile.TemporaryDirectory() as tmp:
-            (Path(tmp) / "a.py").write_text("def needle(x):\n    return x\n")
-            code, out = _run_main(self._argv(tmp))
-            self.assertEqual(code, 0)
-            self.assertEqual(json.loads(out)["engine"], "rg")
-
-
-class TestCountOrdering(unittest.TestCase):
-    """--count 按命中数降序（真实 rg 端到端）。"""
-
-    def test_most_hits_first(self):
-        if not s.tool_exists("rg"):
-            self.skipTest("本机无 rg")
-        with tempfile.TemporaryDirectory() as tmp:
-            # 遍历序（字典序）在前面的文件命中少、后面的命中多：
-            # 旧实现按遍历序截断会答错。
-            (Path(tmp) / "aaa.txt").write_text("needle\n")          # 1 次
-            (Path(tmp) / "zzz.txt").write_text("needle\nneedle\nneedle\n")  # 3 次
-            code, out = _run_main(
-                ["needle", "--scope", "code", "--path", tmp, "--count",
-                 "--json"])
-            self.assertEqual(code, 0)
-            data = json.loads(out)
-            results = data.get("results") or []
-            files = [r.get("path") if isinstance(r, dict) else r[0]
-                     for r in results]
-            self.assertTrue(files, "无结果")
-            self.assertIn("zzz.txt", files[0],
-                          f"命中最多的文件必须排第一，实际 {files}")
-
-
-if __name__ == "__main__":
-    unittest.main()