test: 修复 13 个腐化用例,套件恢复全绿且顺序无关
按「测试腐化(a)/ 源码缺陷(b)/ 测试污染(c)」逐项定性,仅改 tests/: - 外机绝对路径(4): 迁移文件路径改为相对仓库根解析(沿用 test_regressions._read 约定),断言内容保持不变。 - cutoff 用例(4+1): mock 未生效的真因是 _agent_provider 被换成同步 lambda,await 抛 TypeError 被生产代码吞掉后 IndexError;改为 async mock 并按 run_specialists/load_match_header/_agent_provider 的真实契约 打补丁。degraded 用例再加 _upsert_prediction 顶层 kwargs(model)采集。 - 陈旧断言(3): agent 键按现契约断言中文映射;h2h mock 改 async; Match.stats 按设计为 lazy="select",从 MATCH_RELATIONS 移出并单独 固化该设计决定。 - P0-3 守卫(1): seg 越界扫到下游 stats 管线导致误报,改为按缩进收口; 合法形状含经 raw 派生变量中转的写法,并补元测试确保守卫仍能抓到回归。 - 交叉污染(6): test_multi_agent_cutoff 用 patch.object 精确还原,消除 裸赋值泄漏的同步 mock;现已验证顺序无关。
This commit is contained in:
shangfangjian
+149
-19
@@ -14,8 +14,10 @@ from pathlib import Path
|
||||
|
||||
SRC = Path(__file__).resolve().parent.parent / "src"
|
||||
|
||||
# 切片函数会读取的关系属性 → 查询时必须 eager-load
|
||||
MATCH_RELATIONS = ("stats", "home_team", "away_team", "league")
|
||||
# 切片函数会读取的关系属性 → 查询时必须 eager-load。
|
||||
# Match.stats 刻意排除: 它按设计用 lazy="select",由 selectinload(Match.stats)
|
||||
# 显式预加载(见 test_stats_relationship_is_lazy_select_by_design)。
|
||||
MATCH_RELATIONS = ("home_team", "away_team", "league")
|
||||
|
||||
|
||||
def _read(rel: str) -> str:
|
||||
@@ -48,27 +50,127 @@ class TestEagerLoadCoverage:
|
||||
assert "selectinload" in src, "backtest 未 eager-load 关系 (P0-1)"
|
||||
|
||||
def test_relationship_default_is_selectin(self):
|
||||
"""models.py 中 Match 的高频关系应声明 lazy='selectin' 作为兜底。"""
|
||||
"""models.py 中 Match 的高频关系应声明 lazy='selectin' 作为兜底。
|
||||
|
||||
这里只要求「高频一起读取」的关系(set MATCH_RELATIONS)声明 selectin。
|
||||
Match.stats 刻意用 lazy="select" —— 它只在 stats 管线里按需取,不在
|
||||
每个切片都读,而且它的预加载由 selectinload(Match.stats) 显式表达
|
||||
(见 test_context_builder_getters_eager_load)。
|
||||
"""
|
||||
src = _read("db/models.py")
|
||||
# 找到 Match 类定义段
|
||||
m = re.search(r"class Match\(Base\):.*?(?=\nclass )", src, re.S)
|
||||
assert m, "Match 类未找到"
|
||||
body = m.group(0)
|
||||
for rel in MATCH_RELATIONS:
|
||||
# 关系声明可能跨多行(stats/home_team/away_team 都是),因此按
|
||||
# 「从 `rel: Mapped` 到下一个 `xxx: Mapped` 之前」整段匹配。
|
||||
# 关系声明可能跨多行(home_team/away_team 都是),因此按
|
||||
# 「从 `rel: Mapped` 到下一个单行注解声明之前」整段匹配。
|
||||
m_rel = re.search(
|
||||
rf"^\s*{rel}: Mapped.*?(?=^\s*\w+: Mapped|\Z)", body, re.M | re.S
|
||||
rf"^[ \t]*{rel}: Mapped.*?(?=^[ \t]*\w+:[^\n]*Mapped|\Z)",
|
||||
body, re.M | re.S,
|
||||
)
|
||||
assert m_rel, f"Match.{rel} 未找到"
|
||||
assert 'lazy="selectin"' in m_rel.group(0), (
|
||||
f"Match.{rel} 未声明 lazy='selectin' —— 兜底缺失 (P0-2)"
|
||||
)
|
||||
|
||||
def test_stats_relationship_is_lazy_select_by_design(self):
|
||||
"""Match.stats 刻意保持 lazy="select"(不是回归)。
|
||||
|
||||
它是唯一需要显式 selectinload 才预加载的关系 —— 若哪天有人把它也
|
||||
改成 selectin,上面的 test_context_builder_getters_eager_load 和
|
||||
bzzoiro stats 管线仍应工作,但本用例会提醒复核该设计决定。
|
||||
"""
|
||||
src = _read("db/models.py")
|
||||
body = re.search(r"class Match\(Base\):.*?(?=\nclass )", src, re.S).group(0)
|
||||
m_rel = re.search(
|
||||
r"^[ \t]*stats: Mapped.*?(?=^[ \t]*\w+:[^\n]*Mapped|\Z)",
|
||||
body, re.M | re.S,
|
||||
)
|
||||
assert m_rel, "Match.stats 未找到"
|
||||
assert 'lazy="select"' in m_rel.group(0), (
|
||||
"Match.stats 预期为 lazy='select'(按需加载),实际声明已变 —— 请复核设计"
|
||||
)
|
||||
|
||||
|
||||
class TestBzzoiroLineage:
|
||||
"""P0-3: source_event_id 必须取配对的 raw,不能是循环残留变量。"""
|
||||
|
||||
# 消费循环的起始行匹配模式(见 bzzoiro.py 顶部 events 管线的内层循环)
|
||||
_LOOP_PATTERN = "for nm, raw in normalized_matches"
|
||||
|
||||
# source_event_id 的合法赋值形状(两种,都必须取配对的 raw):
|
||||
# 1) 构造新比赛: `source_event_id=_to_int_or_none(raw.get("id")),`
|
||||
# 2) 回填已有比赛: `eid = _to_int_or_none(raw.get("id"))` →
|
||||
# `existing_match.source_event_id = eid`
|
||||
# 非法形状(即 P0-3 回归): 直接用未配对的变量给 ORM 对象赋值。
|
||||
_ASSIGN_DIRECT = re.compile(
|
||||
r"source_event_id\s*=\s*(?:[A-Za-z_][\w.]*\s*\(\s*)?raw(?:\.get\(|\s*\[)"
|
||||
)
|
||||
# 赋值给未配对的局部变量: `source_event_id = <变量>`
|
||||
_ASSIGN_VIA_VAR = re.compile(
|
||||
r"source_event_id\s*=\s*([A-Za-z_]\w*)\s*$"
|
||||
)
|
||||
|
||||
def _consume_loop_body(self, src: str) -> str:
|
||||
"""截取 `for nm, raw in normalized_matches` 循环体,不含循环之后的下游代码。
|
||||
|
||||
原实现是 `seg = "\\n".join(lines[start:])`,一直取到文件末尾,于是把
|
||||
无关的下游 stats 管线(bzzoiro.py 的 `_backfill_stats`)也扫了进来 ——
|
||||
那里合法地在 ORM 对象上访问 `m.source_event_id`,导致误报 P0-3。
|
||||
这里按缩进边界正确收口:循环体内每行要么是空行/注释,要么缩进严格
|
||||
大于 `for` 行。
|
||||
"""
|
||||
lines = src.splitlines()
|
||||
start = next(
|
||||
(i for i, ln in enumerate(lines) if self._LOOP_PATTERN in ln), None
|
||||
)
|
||||
assert start is not None, f"未找到消费循环: {self._LOOP_PATTERN}"
|
||||
|
||||
for_indent = len(lines[start]) - len(lines[start].lstrip())
|
||||
kept: list[str] = [lines[start]]
|
||||
for ln in lines[start + 1:]:
|
||||
stripped = ln.strip()
|
||||
# 顺序要保持: 空行与注释行缩进为 0,不能拿它们做边界判断
|
||||
if not stripped or stripped.startswith("#"):
|
||||
continue
|
||||
indent = len(ln) - len(ln.lstrip())
|
||||
if indent <= for_indent:
|
||||
break # 循环结束,后续属下游代码
|
||||
kept.append(ln)
|
||||
# 右侧剥离注释:避免 `# ... raw.get(...)` 这类注释误命中赋值正则
|
||||
return "\n".join(ln.split("#", 1)[0] for ln in kept)
|
||||
|
||||
def _bad_assignments(self, seg: str) -> list[str]:
|
||||
"""返回循环体内未取配对 raw 的 source_event_id 赋值行。"""
|
||||
# 先收集"来自配对 raw"的局部变量: `eid = _to_int_or_none(raw.get("id"))`
|
||||
# (不受行序影响,所以必须先建好,再判定中转赋值)
|
||||
raw_vars: set[str] = set()
|
||||
for ln in seg.splitlines():
|
||||
m = re.match(
|
||||
r"\s*([A-Za-z_]\w*)\s*=\s*.*raw(?:\.get\(|\s*\[)", ln
|
||||
)
|
||||
if m:
|
||||
raw_vars.add(m.group(1))
|
||||
|
||||
bad: list[str] = []
|
||||
for ln in seg.splitlines():
|
||||
stripped = ln.strip()
|
||||
if "source_event_id" not in stripped:
|
||||
continue
|
||||
# 读取判断/比较(`if x.source_event_id is None:`)不算赋值
|
||||
if re.search(r"source_event_id\s*(?:is|==|!=)", stripped):
|
||||
continue
|
||||
if re.search(r"source_event_id\s*\.\s*\w+\s*\(", stripped):
|
||||
continue # 方法调用,不是赋值
|
||||
if self._ASSIGN_DIRECT.search(stripped):
|
||||
continue # 直接取配对 raw
|
||||
m_var = self._ASSIGN_VIA_VAR.search(stripped)
|
||||
if m_var and m_var.group(1) in raw_vars:
|
||||
continue # 经由已确认来自 raw 的局部变量中转
|
||||
bad.append(stripped)
|
||||
return bad
|
||||
|
||||
def test_normalized_matches_carries_raw(self):
|
||||
src = _read("data/bzzoiro.py")
|
||||
# 规范化结果必须与原始 event 成对保存
|
||||
@@ -81,19 +183,47 @@ class TestBzzoiroLineage:
|
||||
)
|
||||
|
||||
def test_no_orphan_raw_use(self):
|
||||
"""source_event_id 所在行必须在解包循环内(用缩进 + 上下文粗判)。"""
|
||||
"""循环体内每个 source_event_id 赋值都必须取自配对的 raw(P0-3)。
|
||||
|
||||
用正则而不是 `raw.get(` 子串匹配:合法写法含「回填已有比赛」那条
|
||||
(`eid = raw.get("id")` 之后 `existing_match.source_event_id = eid`),
|
||||
它不是 `raw.get(` 同一行,但同样正确。非法写法(回归)是直接
|
||||
`existing_match.source_event_id = orphan_var`。
|
||||
"""
|
||||
src = _read("data/bzzoiro.py")
|
||||
lines = src.splitlines()
|
||||
# 找到 "for nm, raw in normalized_matches" 所在行号
|
||||
start = next(
|
||||
(i for i, ln in enumerate(lines) if "for nm, raw in normalized_matches" in ln),
|
||||
None,
|
||||
seg = self._consume_loop_body(src)
|
||||
bad = self._bad_assignments(seg)
|
||||
assert len(bad) == 0, (
|
||||
f"source_event_id 未使用配对的 raw (P0-3),问题行: {bad}"
|
||||
)
|
||||
assert start is not None
|
||||
# 该循环之后、下一个同/更低缩进的顶层语句之前的范围
|
||||
seg = "\n".join(lines[start:])
|
||||
uses = [ln for ln in seg.splitlines() if "source_event_id" in ln]
|
||||
assert uses, "未找到 source_event_id 赋值"
|
||||
assert all("raw.get(" in ln for ln in uses), (
|
||||
"source_event_id 未使用配对的 raw (P0-3)"
|
||||
|
||||
def test_loop_body_scope_excludes_downstream_stats_pipeline(self):
|
||||
"""作用域守卫: 截取段不能扫到循环之后的下游 stats 管线。
|
||||
|
||||
下游 `_backfill_stats` 里合法地在 ORM 对象上访问 `m.source_event_id`
|
||||
(与配对 raw 无关)。若 seg 越界,test_no_orphan_raw_use 会误报。
|
||||
"""
|
||||
src = _read("data/bzzoiro.py")
|
||||
seg = self._consume_loop_body(src)
|
||||
assert "m.source_event_id" not in seg, (
|
||||
"循环体截取越界,扫到了下游 stats 管线 —— 会误报 P0-3"
|
||||
)
|
||||
# 但配对使用必须仍在作用域内
|
||||
assert "raw.get(" in seg, "循环体内应保留 `raw.get(...)` 的配对用法"
|
||||
|
||||
def test_guard_detects_orphan_variable_regression(self):
|
||||
"""守卫有效性: 若 source_event_id 改成取循环外残留变量,必须被判失败。
|
||||
|
||||
回归保护的"元测试"——确保上面的正则在真实缺陷面前确实会红,
|
||||
而不是恒真的空断言。
|
||||
"""
|
||||
orphan = """
|
||||
for nm, raw in normalized_matches:
|
||||
m = Match(
|
||||
league_id=1,
|
||||
source_event_id=_to_int_or_none(orphan.get("id")),
|
||||
)
|
||||
"""
|
||||
seg = self._consume_loop_body(orphan)
|
||||
bad = self._bad_assignments(seg)
|
||||
assert bad, "守卫失效: 未配对的 orphan 变量未被识别为 P0-3 回归"
|
||||
|
||||
Reference in New Issue
Block a user