fix: restore stage-based Alpha check classification

.scratch/submission-limit-classification/issues/01-implementation.md

@@ -9,3 +9,9 @@ Status: ready-for-agent
 ## 完成记录
 
 已实现并完成本地验证：后端 471 项全量测试通过；最终调整后 23 项专项测试通过；Ruff 与 git diff --check 通过；前端构建通过；Alpha 管理和工作空间 3 项浏览器回归通过。迁移 0018 在隔离 SQLite 中验证 503 条记录的升级、降级、再升级和原始证据保留。尚未部署或对业务数据库执行迁移；Docker 后端部署启动时按既有流程执行 alembic upgrade head。
+
+## Comments
+
+用户后续确认恢复旧项目阶段语义：同步非空有效检查无 FAIL 为 PRE_CHECK，主动 /check 完成无 FAIL 为 PASS，PENDING/WARNING 不算失败；异常、空结果保留待定。同步刷新采用新的同步快照重新分类，不沿用旧检查阶段。MCP/HTTP 使用数据库派生状态；迁移 0019 使用晚于 synced_at 的已保存 checked 检查点识别历史主动检查，其余归同步阶段。保留 0018 历史迁移和所有原始 checks。
+
+阶段逻辑调整已完成：后端 499 项测试通过，含同步→主动检查→再同步、MCP PENDING/WARNING 返回和 503 条历史记录迁移；Ruff、git diff --check、前端构建通过。预检通过使用蓝色 Tag。未部署，未迁移实际业务数据库。

backend/app/alphas.py

@@ -7,7 +7,7 @@ from datetime import datetime
 from sqlalchemy import or_, select, update
 
 from .models import Alpha, Research, ResearchTag, SelfCorrelation, now
-from .platform_checks import split_checks, submission_limits
+from .platform_checks import check_result, split_checks, submission_limits
 from .research.provenance import source_alpha_ids
 
 METRIC_FIELDS = (
@@ -20,16 +20,17 @@ def failed_checks(checks):
     """Return failed Alpha check names, excluding submission limits and local correlation."""
     return [
         check.get("name") if isinstance(check.get("name"), str) else "未命名检查"
-        for check in split_checks(checks)[0] if isinstance(check, dict) and check.get("result") == "FAIL"
+        for check in split_checks(checks)[0] if isinstance(check, dict) and check_result(check) == "FAIL"
     ] if isinstance(checks, list) else []
 
 
-def snapshot_columns(settings, metrics, checks):
+def snapshot_columns(settings, metrics, checks, *, checked=False):
     """Derive list fields from a platform snapshot, preserving missing metrics as null.
 
     Submission limits are excluded. Only explicit Alpha FAIL results count.
-    Empty, malformed and unfinished Alpha checks are
-    pending; all known checks passing without PROD_CORRELATION is only a pre-check.
+    Sync snapshots with no failures are PRE_CHECK; a completed explicit /check
+    with no failures is PASS. WARNING/PENDING do not count as failures, matching
+    the legacy workflow. Empty, malformed or unknown results remain PENDING.
     No submission eligibility or activity eligibility is inferred here.
     """
     settings = settings if isinstance(settings, dict) else {}
@@ -40,10 +41,10 @@ def snapshot_columns(settings, metrics, checks):
     by_name = {check["name"]: check for check in valid if isinstance(check.get("name"), str)}
     if failures:
         check_type = "FAIL_1" if failures == 1 else "FAIL_2"
-    elif not checks or len(valid) != len(checks) or any(check.get("result") != "PASS" for check in valid):
+    elif not checks or len(valid) != len(checks) or any(check_result(check) not in ("PASS", "WARNING", "PENDING") for check in valid):
         check_type = "PENDING"
     else:
-        check_type = "PASS" if "PROD_CORRELATION" in by_name else "PRE_CHECK"
+        check_type = "PASS" if checked else "PRE_CHECK"
     # /check values are freshest; submitted snapshots also expose a scalar in IS.
     prod_correlation = number(by_name.get("PROD_CORRELATION", {}).get("value"))
     if prod_correlation is None:
@@ -66,12 +67,13 @@ def snapshot_columns(settings, metrics, checks):
     }
 
 
-def check_summary(checks):
+def check_summary(checks, *, check_type):
     """Separate cached Alpha findings from submission limits; infer no live eligibility."""
     return {
-        "check_type": snapshot_columns({}, {}, checks)["check_type"],
+        "check_type": check_type,
         "failed_checks": failed_checks(checks),
         "submission_limits": submission_limits(checks),
+        "meaning": "PRE_CHECK 为同步无失败项；PASS 为主动检查完成且无失败项。PENDING/WARNING 不算失败，不代表全部检查项 PASS 或当前可提交",
     }
 
 

backend/app/platform_checks.py

@@ -1,6 +1,12 @@
 """Classify platform evidence without discarding unknown checks or inferring eligibility."""
 
 
+def check_result(check):
+    """Normalize known upstream result casing without rewriting the raw evidence."""
+    value = check.get("result") if isinstance(check, dict) else None
+    return value.upper() if isinstance(value, str) else None
+
+
 def is_submission_limit(check):
     """Recognize only the confirmed account-limit check; unknown names remain Alpha checks."""
     return isinstance(check, dict) and check.get("name") == "REGULAR_SUBMISSION"
@@ -16,8 +22,8 @@ def split_checks(checks):
 def submission_limits(checks):
     """Summarize the observed limit, never the account's current allowance or reset time."""
     _, limits = split_checks(checks)
-    status = "blocked" if any(c.get("result") == "FAIL" for c in limits) else (
-        "not_blocked" if limits and all(c.get("result") == "PASS" for c in limits) else "unknown"
+    status = "blocked" if any(check_result(c) == "FAIL" for c in limits) else (
+        "not_blocked" if limits and all(check_result(c) == "PASS" for c in limits) else "unknown"
     )
     return {"status": status, "checks": limits,
             "meaning": "仅反映缓存观测时的提交限制，不代表当前额度或正式提交资格"}

backend/app/research_access/service.py

@@ -222,7 +222,7 @@ class ResearchAccess:
         return {"alpha_id": args.alpha_id, "snapshot": submission_fingerprint(context),
                 **context, "descriptions": {key: item["description"] for key, item in context["sections"].items()},
                 "can_check": alpha.status == "UNSUBMITTED" and await correlation_allows_check(self.db, args.alpha_id),
-                "checks": alpha.checks, "check_summary": check_summary(alpha.checks), "source": "local_cache", "production_submission": False,
+                "checks": alpha.checks, "check_summary": check_summary(alpha.checks, check_type=alpha.check_type), "source": "local_cache", "production_submission": False,
                 "job_id": job.id if job else None, "job_status": job.status if job else None,
                 "checked_at": job.checkpoint.get("checked_at") if job else None}
 

backend/app/submission.py

@@ -207,7 +207,7 @@ def router(runner, ai):
             return {
                 "snapshot": fingerprint(context),
                 "checks": alpha.checks,
-                "check_summary": check_summary(alpha.checks),
+                "check_summary": check_summary(alpha.checks, check_type=alpha.check_type),
                 "sections": context["sections"],
                 "descriptions": {key: item["description"] for key, item in context["sections"].items()},
                 "model": config.description_model,
@@ -355,7 +355,7 @@ async def run_check(runner, job_id, payload):
         alpha.checks = sanitize(checks)
         alpha.is_metrics = {**alpha.is_metrics, "checks": alpha.checks}
         alpha.raw = {**alpha.raw, "is": {**(alpha.raw.get("is") or {}), "checks": alpha.checks}}
-        for key, value in snapshot_columns(alpha.settings, alpha.is_metrics, alpha.checks).items():
+        for key, value in snapshot_columns(alpha.settings, alpha.is_metrics, alpha.checks, checked=True).items():
             setattr(alpha, key, value)
         db.add(JobItem(job_id=job_id, alpha_id=alpha_id))
         job.processed = 1

backend/migrations/versions/0019_check_stages.py

@@ -0,0 +1,91 @@
+"""Restore stage-based check classification, preserving raw platform snapshots."""
+
+from datetime import datetime, timezone
+
+import sqlalchemy as sa
+from alembic import op
+
+revision = "0019"
+down_revision = "0018"
+branch_labels = None
+depends_on = None
+
+
+def timestamp(value):
+    """Read historical timestamps; missing or invalid evidence cannot prove a /check."""
+    try:
+        value = datetime.fromisoformat(value) if isinstance(value, str) else value
+        return value.replace(tzinfo=timezone.utc) if value.tzinfo is None else value
+    except (ValueError, TypeError, AttributeError):
+        return None
+
+
+def classify(checks, checked, legacy):
+    """Frozen migration rules; legacy means the pre-0019 correlation-presence rule."""
+    checks = checks if isinstance(checks, list) else []
+    checks = [c for c in checks if not (isinstance(c, dict) and c.get("name") == "REGULAR_SUBMISSION")]
+    valid = [c for c in checks if isinstance(c, dict)]
+    results = [c.get("result") for c in valid]
+    if not legacy:
+        results = [v.upper() if isinstance(v, str) else None for v in results]
+    failures = sum(v == "FAIL" for v in results)
+    if failures:
+        return "FAIL_1" if failures == 1 else "FAIL_2"
+    allowed = ("PASS",) if legacy else ("PASS", "PENDING", "WARNING")
+    if not checks or len(valid) != len(checks) or any(v not in allowed for v in results):
+        return "PENDING"
+    passed = any(c.get("name") == "PROD_CORRELATION" for c in valid) if legacy else checked
+    return "PASS" if passed else "PRE_CHECK"
+
+
+def reclassify(legacy=False):
+    """Recompute in batches; only a check checkpoint newer than the sync proves its stage.
+
+    A prior PASS is not evidence because the old rule inferred it from a check name.
+    A subsequent sync replaces the snapshot and is classified as pre-check again.
+    """
+    alphas = sa.table("alphas", sa.column("id", sa.String()), sa.column("checks", sa.JSON()),
+                      sa.column("synced_at", sa.DateTime(timezone=True)), sa.column("check_type", sa.String()))
+    jobs = sa.table("sync_jobs", sa.column("kind", sa.String()), sa.column("payload", sa.JSON()),
+                    sa.column("checkpoint", sa.JSON()))
+    connection = op.get_bind()
+    last_id = None
+    while True:
+        query = sa.select(alphas.c.id, alphas.c.checks, alphas.c.synced_at).order_by(alphas.c.id).limit(500)
+        if last_id is not None:
+            query = query.where(alphas.c.id > last_id)
+        rows = connection.execute(query).mappings().all()
+        if not rows:
+            break
+        checked_at = {}
+        if not legacy:
+            observations = connection.execute(sa.select(jobs.c.checkpoint).where(
+                jobs.c.kind == "submission_check",
+                jobs.c.payload["alpha_ids"][0].as_string().in_([r["id"] for r in rows]),
+            )).scalars()
+            for checkpoint in observations:
+                if not isinstance(checkpoint, dict) or checkpoint.get("phase") != "checked":
+                    continue
+                alpha_id = checkpoint.get("alpha_id")
+                observed = timestamp(checkpoint.get("checked_at"))
+                if isinstance(alpha_id, str) and observed and (
+                    alpha_id not in checked_at or observed > checked_at[alpha_id]
+                ):
+                    checked_at[alpha_id] = observed
+        updates = []
+        for row in rows:
+            synced = timestamp(row["synced_at"])
+            observed = checked_at.get(row["id"])
+            updates.append({"snapshot_id": row["id"], "classification": classify(
+                row["checks"], bool(synced and observed and observed > synced), legacy)})
+        connection.execute(alphas.update().where(alphas.c.id == sa.bindparam("snapshot_id"))
+                           .values(check_type=sa.bindparam("classification")), updates)
+        last_id = rows[-1]["id"]
+
+
+def upgrade():
+    reclassify()
+
+
+def downgrade():
+    reclassify(legacy=True)

backend/tests/test_alpha_list_fields.py

@@ -37,10 +37,10 @@ def checks(failures):
         (None, "PENDING"),
         ([None], "PENDING"),
         ([{}], "PENDING"),
-        ([{"name": "LOW_SHARPE", "result": "WARNING"}], "PENDING"),
+        ([{"name": "LOW_SHARPE", "result": "WARNING"}], "PRE_CHECK"),
         ([{"name": "LOW_SHARPE", "result": "PASS"}], "PRE_CHECK"),
-        ([{"name": "PROD_CORRELATION", "result": "PENDING"}], "PENDING"),
-        (checks(0), "PASS"),
+        ([{"name": "PROD_CORRELATION", "result": "PENDING"}], "PRE_CHECK"),
+        (checks(0), "PRE_CHECK"),
         (checks(1), "FAIL_1"),
         (checks(2), "FAIL_2"),
         (checks(3), "FAIL_2"),
@@ -58,7 +58,7 @@ async def test_checks_filter_before_pagination_and_share_export_scope(app, logge
     for check_type, expected in [
         ("FAIL_1", ["failed1"]),
         ("FAIL_2", ["failed2", "failed3"]),
-        ("PASS", ["failed0"]),
+        ("PRE_CHECK", ["failed0"]),
         ("PENDING", ["unknown"]),
     ]:
         response = await logged_in.get(
@@ -218,7 +218,7 @@ def test_migration_backfills_multiple_batches_and_preserves_research(tmp_path, m
             rows = db.execute(sa.select(alphas).order_by(alphas.c.id)).mappings().all()
             assert len(rows) == 503
             for i, row in enumerate(rows):
-                expected = snapshot_columns(row["settings"], row["is_metrics"], checks(i % 4))
+                expected = snapshot_columns(row["settings"], row["is_metrics"], checks(i % 4), checked=True)
                 assert {key: row[key] for key in expected} == expected
             record = db.execute(sa.select(research)).mappings().one()
             assert record["tags"] == ["PPAC"] and record["note"] == "keep" and record["version"] == 7

backend/tests/test_check_stage_migration.py

@@ -0,0 +1,64 @@
+"""Historical stage recovery requires a persisted /check newer than the latest sync."""
+
+from datetime import datetime, timezone
+from pathlib import Path
+
+import sqlalchemy as sa
+from alembic import command
+from alembic.config import Config
+from cryptography.fernet import Fernet
+
+
+def test_stage_backfill_preserves_evidence_and_uses_checkpoints(tmp_path, monkeypatch):
+    path = tmp_path / "stages.db"
+    monkeypatch.setenv("DATABASE_URL", f"sqlite+aiosqlite:///{path}")
+    monkeypatch.setenv("ADMIN_PASSWORD", "migration-test-only")
+    monkeypatch.setenv("ENCRYPTION_KEY", Fernet.generate_key().decode())
+    monkeypatch.setenv("WQ_EMAIL", "")
+    monkeypatch.setenv("WQ_PASSWORD", "")
+    root = Path(__file__).resolve().parents[1]
+    config = Config(str(root / "alembic.ini"))
+    config.set_main_option("script_location", str(root / "migrations"))
+    command.upgrade(config, "0018")
+    engine = sa.create_engine(f"sqlite:///{path}")
+    alphas = sa.Table("alphas", sa.MetaData(), autoload_with=engine)
+    jobs = sa.Table("sync_jobs", sa.MetaData(), autoload_with=engine)
+    synced = datetime(2026, 9, 12, 0, tzinfo=timezone.utc)
+    pending = [{"name": "PROD_CORRELATION", "result": "PENDING"},
+               {"name": "REGULAR_SUBMISSION", "result": "FAIL"}]
+    passed = [{"name": "PROD_CORRELATION", "result": "PASS"}]
+    patterns = [
+        (pending, "PENDING", "PRE_CHECK", None),
+        (pending, "PENDING", "PASS", {"phase": "checked", "checked_at": "2026-09-12T01:00:00+00:00"}),
+        (pending, "PENDING", "PRE_CHECK", {"phase": "checked", "checked_at": "2026-09-11T23:00:00+00:00"}),
+        (pending, "PENDING", "PRE_CHECK", {"phase": "check"}),
+        (passed, "PASS", "PRE_CHECK", None),
+        ([{"name": "LOW_SHARPE", "result": "FAIL"}], "FAIL_1", "FAIL_1", None),
+        ([{}], "PENDING", "PENDING", {"phase": "checked", "checked_at": "2026-09-12T01:00:00Z"}),
+    ]
+    with engine.begin() as db:
+        db.execute(alphas.insert(), [
+            {"id": f"stage{i:04}", "hidden": False, "settings": {}, "os_metrics": {},
+             "is_metrics": {"checks": patterns[i % 7][0]}, "checks": patterns[i % 7][0],
+             "check_type": patterns[i % 7][1], "synced_at": synced,
+             "raw": {"is": {"checks": patterns[i % 7][0]}}}
+            for i in range(503)
+        ])
+        db.execute(jobs.insert(), [
+            {"id": f"job{i}", "kind": "submission_check", "status": "completed",
+             "payload": {"alpha_ids": [f"stage{i:04}"]},
+             "checkpoint": {**patterns[i % 7][3], "alpha_id": f"stage{i:04}"},
+             "processed": 1, "failed": 0, "total": 1, "cancel_requested": False,
+             "created_at": synced, "updated_at": synced}
+            for i in range(503) if patterns[i % 7][3]
+        ])
+    for target, expected_index in [("0019", 2), ("0018", 1), ("0019", 2)]:
+        (command.upgrade if target == "0019" else command.downgrade)(config, target)
+        with engine.connect() as db:
+            rows = db.execute(sa.select(alphas).order_by(alphas.c.id)).mappings().all()
+            assert len(rows) == 503
+            for i, row in enumerate(rows):
+                assert row["check_type"] == patterns[i % 7][expected_index]
+                assert row["checks"] == row["raw"]["is"]["checks"] == row["is_metrics"]["checks"] == patterns[i % 7][0]
+    command.check(config)
+    engine.dispose()

backend/tests/test_check_stages.py

@@ -0,0 +1,56 @@
+"""The same checks have different meanings at sync and explicit /check stages."""
+
+import pytest
+
+from app.alphas import snapshot_columns
+
+
+@pytest.mark.parametrize("checks", [
+    [{"name": "LOW_SHARPE", "result": "PASS"}],
+    [{"name": "PROD_CORRELATION", "result": "PASS"}],
+    [{"name": "PROD_CORRELATION", "result": "PENDING"}],
+    [{"name": "MATCHES_THEMES", "result": "WARNING"}],
+])
+def test_stage_not_correlation_presence_decides_pass(checks):
+    assert snapshot_columns({}, {}, checks)["check_type"] == "PRE_CHECK"
+    assert snapshot_columns({}, {}, checks, checked=True)["check_type"] == "PASS"
+
+
+@pytest.mark.parametrize("checked", [False, True])
+@pytest.mark.parametrize("checks,expected", [
+    ([], "PENDING"),
+    ([None], "PENDING"),
+    ([{}], "PENDING"),
+    ([{"name": "UNKNOWN", "result": "OTHER"}], "PENDING"),
+    ([{"name": "REGULAR_SUBMISSION", "result": "FAIL"}], "PENDING"),
+    ([{"name": "LOW_SHARPE", "result": "fail"}, {"name": "REGULAR_SUBMISSION", "result": "FAIL"}], "FAIL_1"),
+    ([{"name": "LOW_SHARPE", "result": "FAIL"}, {"name": "LOW_FITNESS", "result": "Fail"}], "FAIL_2"),
+])
+def test_stage_preserves_failures_and_missing_evidence(checked, checks, expected):
+    assert snapshot_columns({}, {}, checks, checked=checked)["check_type"] == expected
+
+
+async def test_sync_check_and_resync_use_distinct_stages(app, logged_in, monkeypatch):
+    from app.alphas import upsert_alpha
+    from tests import test_submission
+    from tests.conftest import alpha
+
+    checks = [{"name": "LOW_SHARPE", "result": "PASS"},
+              {"name": "PROD_CORRELATION", "result": "PENDING"},
+              {"name": "MATCHES_THEMES", "result": "WARNING"},
+              {"name": "REGULAR_SUBMISSION", "result": "FAIL"}]
+    monkeypatch.setattr(test_submission, "CHECKS", checks)
+    await test_submission.setup(app, alpha(**{"is": {"checks": checks}}))
+    endpoint = "/api/v1/alphas/alpha1/submission"
+    assert (await logged_in.get(endpoint)).json()["check_summary"]["check_type"] == "PRE_CHECK"
+    response = await test_submission.enqueue(logged_in)
+    assert response.status_code == 202
+    await app.state.runner.execute(response.json()["id"])
+    state = (await logged_in.get(endpoint)).json()
+    assert state["job"]["status"] == "completed"
+    assert state["check_summary"]["check_type"] == "PASS"
+    assert state["check_summary"]["submission_limits"]["status"] == "blocked"
+    assert (await logged_in.get("/api/v1/alphas/alpha1")).json()["check_type"] == "PASS"
+    async with app.state.sessions.begin() as db:
+        await upsert_alpha(db, alpha(**{"is": {"checks": checks}}))
+    assert (await logged_in.get(endpoint)).json()["check_summary"]["check_type"] == "PRE_CHECK"

backend/tests/test_mcp_submission.py

@@ -11,7 +11,7 @@ from tests.test_submission import FIELDS, Description, setup
 
 
 @pytest.mark.parametrize("kind", ["REGULAR", "SUPER"])
-@pytest.mark.parametrize("result", ["PASS", "FAIL"])
+@pytest.mark.parametrize("result", ["PASS", "FAIL", "PENDING", "WARNING"])
 @pytest.mark.parametrize("limited", [False, True])
 async def test_mcp_check_never_submits(mcp_app, kind, result, limited, monkeypatch):
     from tests import test_submission
@@ -38,7 +38,7 @@ async def test_mcp_check_never_submits(mcp_app, kind, result, limited, monkeypat
     assert job["status"] == "completed", job
     data = await invoke(mcp_app, principal, "get_submission_check", {"alpha_id": "alpha1"})
     assert data["checks"] == checks and data["checked_at"]
-    assert data["check_summary"]["check_type"] == ("PASS" if result == "PASS" else "FAIL_1")
+    assert data["check_summary"]["check_type"] == ("FAIL_1" if result == "FAIL" else "PASS")
     assert data["check_summary"]["submission_limits"]["status"] == ("blocked" if limited else "unknown")
     assert data["production_submission"] is False
     assert platform.patches == [{s: {"description": args["descriptions"][s]} for s in sections}]

backend/tests/test_submission_limit_migration.py

@@ -47,5 +47,6 @@ def test_limit_migration_is_reversible_and_preserves_snapshots(tmp_path, monkeyp
             for i, row in enumerate(rows):
                 assert row["check_type"] == patterns[i % 5][position]
                 assert row["checks"] == row["is_metrics"]["checks"] == row["raw"]["is"]["checks"] == patterns[i % 5][0]
+    command.upgrade(config, "head")
     command.check(config)
     engine.dispose()

backend/tests/test_submission_limits.py

@@ -8,13 +8,13 @@ LIMIT = {"name": "REGULAR_SUBMISSION", "result": "FAIL", "value": 4, "limit": 4}
 
 
 @pytest.mark.parametrize("checks,expected", [
-    ([{"name": "PROD_CORRELATION", "result": "PASS"}], "PASS"),
+    ([{"name": "PROD_CORRELATION", "result": "PASS"}], "PRE_CHECK"),
     ([{"name": "LOW_SHARPE", "result": "PASS"}], "PRE_CHECK"),
     ([{"name": "LOW_SHARPE", "result": "FAIL"}], "FAIL_1"),
     ([], "PENDING"),
     ([None], "PENDING"),
     ([{"name": "UNKNOWN_CHECK", "result": "FAIL"}], "FAIL_1"),
-    ([{"name": "PROD_CORRELATION", "result": "PENDING"}], "PENDING"),
+    ([{"name": "PROD_CORRELATION", "result": "PENDING"}], "PRE_CHECK"),
 ])
 def test_limit_does_not_change_alpha_verdict(checks, expected):
     original = [*checks, LIMIT]
@@ -28,7 +28,7 @@ def test_all_limit_states_are_separate(result):
     from app.alphas import check_summary
 
     checks = [{"name": "PROD_CORRELATION", "result": "PASS"}, {**LIMIT, "result": result}]
-    summary = check_summary(checks)
+    summary = check_summary(checks, check_type="PASS")
     assert summary["check_type"] == "PASS"
     assert summary["submission_limits"]["status"] == ("not_blocked" if result == "PASS" else "unknown")
 

frontend/src/pages/AlphaPage.tsx

@@ -523,7 +523,9 @@ export function AlphaPage({
               ? "red"
               : row!.check_type === "PASS"
                 ? "green"
-                : "grey"
+                : row!.check_type === "PRE_CHECK"
+                  ? "blue"
+                  : "grey"
           }
         >
           {checkLabels[row!.check_type] || "待检查"}