"""垂直切片测试：幂等原子事务、IDOR fail-closed、CSRF、SSE 与回滚。""" import json import pytest from httpx import AsyncClient from sqlalchemy import select from app import models from tests.conftest import login MESSAGE_BODY = {"content": [{"type": "text", "text": "你好，魂镜"}], "start_run": True} async def _create_conversation(client: AsyncClient, headers: dict) -> str: resp = await client.post( "/v1/conversations", json={"title": "切片"}, headers=headers ) assert resp.status_code == 201, resp.text return resp.json()["id"] async def _post_message(client: AsyncClient, headers: dict, cid: str, key: str) -> object: return await client.post( f"/v1/conversations/{cid}/messages", json=MESSAGE_BODY, headers={**headers, "Idempotency-Key": key}, ) class TestMessagePipeline: async def test_create_and_post_message(self, client: AsyncClient) -> None: headers = await login(client, "alice@example.com") cid = await _create_conversation(client, headers) resp = await _post_message(client, headers, cid, "idem-1") assert resp.status_code == 201, resp.text body = resp.json() assert body["message"]["id"].startswith("message_") assert body["message"]["content"][0]["id"].startswith("block_") assert body["run"]["status"] == "completed" # 空状态闭环 async def test_idempotent_replay_returns_original(self, client: AsyncClient) -> None: headers = await login(client, "bob@example.com") cid = await _create_conversation(client, headers) first = await _post_message(client, headers, cid, "idem-replay") assert first.status_code == 201 second = await _post_message(client, headers, cid, "idem-replay") assert second.status_code == 200 # 重放返回原资源 assert second.json()["message"]["id"] == first.json()["message"]["id"] async def test_same_key_different_body_conflicts(self, client: AsyncClient) -> None: headers = await login(client, "carol@example.com") cid = await _create_conversation(client, headers) await _post_message(client, headers, cid, "idem-conflict") resp = await client.post( f"/v1/conversations/{cid}/messages", json={"content": [{"type": "text", "text": "不同内容"}], "start_run": True}, headers={**headers, "Idempotency-Key": "idem-conflict"}, ) assert resp.status_code == 409 async def test_missing_idempotency_key_rejected(self, client: AsyncClient) -> None: headers = await login(client, "dave@example.com") cid = await _create_conversation(client, headers) resp = await client.post( f"/v1/conversations/{cid}/messages", json=MESSAGE_BODY, headers=headers ) assert resp.status_code == 422 async def test_unknown_block_type_rejected(self, client: AsyncClient) -> None: headers = await login(client, "erin@example.com") cid = await _create_conversation(client, headers) resp = await client.post( f"/v1/conversations/{cid}/messages", json={"content": [{"type": "hologram", "file_id": "x"}]}, headers={**headers, "Idempotency-Key": "idem-bad"}, ) assert resp.status_code == 422 async def test_cross_domain_conversation_returns_404(self, client: AsyncClient) -> None: alice = await login(client, "ida@example.com") cid = await _create_conversation(client, alice) mallory = await login(client, "mallory@example.com") resp = await _post_message(client, mallory, cid, "idem-idor") assert resp.status_code == 404 # fail-closed，不泄露存在性 async def test_csrf_token_required_for_state_change(self, client: AsyncClient) -> None: await login(client, "frank@example.com") resp = await client.post("/v1/conversations", json={"title": "无 CSRF"}) assert resp.status_code == 403 async def test_unauthenticated_rejected(self, client: AsyncClient) -> None: resp = await client.post("/v1/conversations", json={"title": "匿名"}) assert resp.status_code == 401 class TestRunSse: async def test_sse_snapshot_and_terminal(self, client: AsyncClient) -> None: headers = await login(client, "grace@example.com") cid = await _create_conversation(client, headers) resp = await _post_message(client, headers, cid, "idem-sse") run_id = resp.json()["run"]["id"] sse = await client.get(f"/v1/runs/{run_id}/events") assert sse.status_code == 200 assert sse.headers["content-type"].startswith("text/event-stream") lines = [ln for ln in sse.text.splitlines() if ln.startswith("data: ")] events = [json.loads(ln.removeprefix("data: ")) for ln in lines] assert events[0]["type"] == "run.status.changed" assert events[0]["data"]["status"] == "completed" assert events[-1]["type"] == "message.completed" async def test_cross_domain_run_returns_404(self, client: AsyncClient) -> None: headers = await login(client, "heidi@example.com") cid = await _create_conversation(client, headers) resp = await _post_message(client, headers, cid, "idem-sse2") run_id = resp.json()["run"]["id"] await login(client, "eve@example.com") sse = await client.get(f"/v1/runs/{run_id}/events") assert sse.status_code == 404 class TestAtomicity: async def test_failure_leaves_no_orphans(self, client: AsyncClient, db) -> None: from app.conversations import service headers = await login(client, "ivan@example.com") cid = await _create_conversation(client, headers) class BoomRun: def __init__(self, *args, **kwargs): raise RuntimeError("注入失败：Run 写入阶段崩溃") original = service.Run service.Run = BoomRun # type: ignore[assignment] try: resp = await _post_message(client, headers, cid, "idem-atomic") assert resp.status_code == 500 finally: service.Run = original # type: ignore[assignment] async with db() as session: # 会话创建的合法审计/Outbox 保留；消息相关记录必须为零 assert (await session.scalars(select(models.Message))).all() == [] assert (await session.scalars(select(models.Run))).all() == [] audits = (await session.scalars(select(models.AuditEvent))).all() assert [a for a in audits if a.aggregate_type == "message"] == [] outbox = (await session.scalars(select(models.OutboxEvent))).all() assert [e for e in outbox if e.aggregate_type == "message"] == [] assert (await session.scalars(select(models.IdempotencyRecord))).all() == [] @pytest.mark.usefixtures("client") class TestLingtaiNormalization: async def test_markdown_input_expanded_by_engine(self, client: AsyncClient) -> None: headers = await login(client, "norm@example.com") cid = await _create_conversation(client, headers) resp = await client.post( f"/v1/conversations/{cid}/messages", json={ "content": [ {"type": "markdown", "text": "说明\n```mermaid\nA-->B\n```\n$$\nz\n$$"} ] }, headers={**headers, "Idempotency-Key": "idem-norm"}, ) assert resp.status_code == 201, resp.text blocks = resp.json()["message"]["content"] assert [b["type"] for b in blocks] == ["markdown", "mermaid", "math"] assert all(b["id"].startswith("block_") for b in blocks) assert resp.json()["message"].get("parser_version") == "hunjing-markdown-v1" async def test_python_cpp_golden_consistency(self) -> None: """绑定输出与 C++ doctest 期望字节一致（跨端一致性最小 golden）。""" from app.core import lingtai out = lingtai.transform_markdown("x\n```mermaid\nA-->B\n```\n$$\nz\n$$\n") assert out["parser_version"] == "hunjing-markdown-v1" assert [b["type"] for b in out["blocks"]] == ["markdown", "mermaid", "math"] # 与 engine/tests/test_transform.cpp 的确定性用例保持同一输入输出 out2 = lingtai.transform_markdown("x\n```mermaid\nA-->B\n```\n$$\nz\n$$\n") assert out == out2 async def test_engine_failure_fails_closed(self, client: AsyncClient, monkeypatch) -> None: from app.core import lingtai as lt def boom(text: str): raise lt.LingtaiUnavailable("灵台引擎调用失败：注入") monkeypatch.setattr(lt, "transform_markdown", boom) headers = await login(client, "fc@example.com") cid = await _create_conversation(client, headers) resp = await client.post( f"/v1/conversations/{cid}/messages", json={"content": [{"type": "markdown", "text": "hi"}]}, headers={**headers, "Idempotency-Key": "idem-fc"}, ) # fail closed：引擎不可用时显式失败，不静默降级直存 assert resp.status_code == 500 @pytest.mark.usefixtures("client") class TestContentBlocks: async def test_media_blocks_accepted(self, client: AsyncClient) -> None: headers = await login(client, "blocks@example.com") cid = await _create_conversation(client, headers) resp = await client.post( f"/v1/conversations/{cid}/messages", json={ "content": [ {"type": "image", "file_id": "file_1", "alt": "图"}, {"type": "math", "notation": "latex", "display": "inline", "source": "a+b"}, {"type": "table", "columns": ["a"], "rows": [["1"]]}, {"type": "citation", "source_id": "s1", "title": "t", "quoted_text": "q"}, {"type": "error", "message": "块错误"}, ] }, headers={**headers, "Idempotency-Key": "idem-blocks"}, ) assert resp.status_code == 201, resp.text types = [b["type"] for b in resp.json()["message"]["content"]] assert types == ["image", "math", "table", "citation", "error"] async def test_runtime_blocks_rejected_from_client(self, client: AsyncClient) -> None: """契约 3.5：tool_call/agent_status 等系统 block 不允许客户端伪造。""" headers = await login(client, "rtb@example.com") cid = await _create_conversation(client, headers) for block in [ {"type": "tool_call", "tool_name": "t", "call_id": "c1"}, {"type": "agent_status", "stage": "s"}, {"type": "approval_request", "approval_id": "a1", "action": "do"}, {"type": "artifact", "artifact_id": "ar1", "artifact_type": "file"}, ]: resp = await client.post( f"/v1/conversations/{cid}/messages", json={"content": [block]}, headers={**headers, "Idempotency-Key": f"idem-rtb-{block['type']}"}, ) assert resp.status_code == 422, (block["type"], resp.text) @pytest.mark.usefixtures("client") class TestSseReplay: async def test_events_replayed_in_order_with_sequence(self, client: AsyncClient) -> None: headers = await login(client, "sseq@example.com") cid = await _create_conversation(client, headers) resp = await _post_message(client, headers, cid, "idem-seq") run_id = resp.json()["run"]["id"] sse = await client.get(f"/v1/runs/{run_id}/events") lines = [ln for ln in sse.text.splitlines() if ln.startswith("data: ")] events = [json.loads(ln.removeprefix("data: ")) for ln in lines] sequences = [e["sequence"] for e in events] assert sequences == sorted(sequences) == list(range(len(events))) assert events[0]["type"] == "run.status.changed" assert events[-1]["type"] == "message.completed" # 用户消息两个 block：text 与（无 markdown 时）content.block.completed 数对应 assert sum(1 for e in events if e["type"] == "content.block.completed") >= 1 @pytest.mark.usefixtures("client") class TestPaginationAndBranches: async def test_message_cursor_pagination(self, client: AsyncClient) -> None: headers = await login(client, "page@example.com") cid = await _create_conversation(client, headers) for i in range(5): await client.post( f"/v1/conversations/{cid}/messages", json={"content": [{"type": "text", "text": f"m{i}"}], "start_run": False}, headers={**headers, "Idempotency-Key": f"idem-p{i}"}, ) page1 = await client.get(f"/v1/conversations/{cid}/messages?limit=2", headers=headers) assert page1.status_code == 200 body1 = page1.json() assert len(body1["data"]) == 2 and body1["has_more"] is True assert body1["next_cursor"] page2 = await client.get( f"/v1/conversations/{cid}/messages?limit=2&cursor={body1['next_cursor']}", headers=headers, ) body2 = page2.json() ids1 = {m["id"] for m in body1["data"]} ids2 = {m["id"] for m in body2["data"]} assert not (ids1 & ids2) last = await client.get( f"/v1/conversations/{cid}/messages?limit=100&cursor={body2['next_cursor']}", headers=headers, ) assert last.json()["has_more"] is False async def test_invalid_cursor_rejected(self, client: AsyncClient) -> None: headers = await login(client, "badc@example.com") cid = await _create_conversation(client, headers) resp = await client.get( f"/v1/conversations/{cid}/messages?cursor=%%%", headers=headers ) assert resp.status_code == 422 async def test_branches_and_current_path(self, client: AsyncClient) -> None: headers = await login(client, "branch@example.com") cid = await _create_conversation(client, headers) listing = await client.get(f"/v1/conversations/{cid}/branches", headers=headers) branches = listing.json()["data"] assert len(branches) == 1 and branches[0]["title"] == "主脉" # 切换到不存在的分支：422 fail-closed bad = await client.post( f"/v1/conversations/{cid}/current-branch", json={"branch_id": "branch_00000000-0000-7000-8000-000000000000"}, headers=headers, ) assert bad.status_code == 422 # 切换到合法分支 ok = await client.post( f"/v1/conversations/{cid}/current-branch", json={"branch_id": branches[0]["id"]}, headers=headers, ) assert ok.status_code == 200 assert ok.json()["current_branch_id"] == branches[0]["id"] async def test_cross_domain_branch_rejected(self, client: AsyncClient) -> None: alice = await login(client, "br-a@example.com") cid = await _create_conversation(client, alice) listing = await client.get(f"/v1/conversations/{cid}/branches", headers=alice) bid = listing.json()["data"][0]["id"] await login(client, "br-b@example.com") resp = await client.post( f"/v1/conversations/{cid}/current-branch", json={"branch_id": bid} ) assert resp.status_code in (401, 403) # 未带新会话 Cookie：401；带错 Cookie：拒绝 @pytest.mark.usefixtures("client") class TestRealmsAndCanonization: async def _make_message(self, client, headers, cid, key, text="hi") -> dict: resp = await client.post( f"/v1/conversations/{cid}/messages", json={"content": [{"type": "text", "text": text}], "start_run": False}, headers={**headers, "Idempotency-Key": key}, ) assert resp.status_code == 201, resp.text return resp.json()["message"] async def test_realm_switch_snapshots_into_new_messages(self, client: AsyncClient) -> None: headers = await login(client, "realm1@example.com") cid = await _create_conversation(client, headers) detail = await client.get(f"/v1/conversations/{cid}", headers=headers) assert detail.json()["realm"] == "reality" # 默认真界 sw = await client.post( f"/v1/conversations/{cid}/realm", json={"realm": "simulation"}, headers=headers ) assert sw.status_code == 200 and sw.json()["realm"] == "simulation" await self._make_message(client, headers, cid, "idem-sim1") await client.get(f"/v1/conversations/{cid}/messages", headers=headers) # （realm 字段在消息响应中暂未透出，以归真行为间接验证） # 回到真界 await client.post( f"/v1/conversations/{cid}/realm", json={"realm": "reality"}, headers=headers ) async def test_canonize_reality_message(self, client: AsyncClient) -> None: headers = await login(client, "can1@example.com") cid = await _create_conversation(client, headers) msg = await self._make_message(client, headers, cid, "idem-can1") resp = await client.post( f"/v1/messages/{msg['id']}/canonizations", json={"scope": "reality"}, headers=headers, ) assert resp.status_code == 201 assert resp.json()["canonical_scope"] == "reality" assert resp.json()["canonized_at"] is not None async def test_canonize_idempotent_replay(self, client: AsyncClient) -> None: headers = await login(client, "can2@example.com") cid = await _create_conversation(client, headers) msg = await self._make_message(client, headers, cid, "idem-can2") first = await client.post( f"/v1/messages/{msg['id']}/canonizations", json={"scope": "reality"}, headers=headers ) second = await client.post( f"/v1/messages/{msg['id']}/canonizations", json={"scope": "reality"}, headers=headers ) assert first.status_code == 201 and second.status_code == 200 assert second.json()["canonized_at"] == first.json()["canonized_at"] async def test_canonize_scope_mismatch_rejected(self, client: AsyncClient) -> None: """真界消息不能归真到境界/幻境作用域（ADR-020 作用域一致）。""" headers = await login(client, "can3@example.com") cid = await _create_conversation(client, headers) msg = await self._make_message(client, headers, cid, "idem-can3") resp = await client.post( f"/v1/messages/{msg['id']}/canonizations", json={"scope": "world"}, headers=headers ) assert resp.status_code == 422 async def test_simulation_message_canonizes_as_rehearsal(self, client: AsyncClient) -> None: """幻境消息归真仅标记试演记录（ADR-020）。""" headers = await login(client, "can4@example.com") cid = await _create_conversation(client, headers) await client.post( f"/v1/conversations/{cid}/realm", json={"realm": "simulation"}, headers=headers ) msg = await self._make_message(client, headers, cid, "idem-can4") resp = await client.post( f"/v1/messages/{msg['id']}/canonizations", json={"scope": "simulation"}, headers=headers ) assert resp.status_code == 201 async def test_canonize_cross_domain_404(self, client: AsyncClient) -> None: alice = await login(client, "can5@example.com") cid = await _create_conversation(client, alice) msg = await self._make_message(client, alice, cid, "idem-can5") await login(client, "can5-evil@example.com") resp = await client.post( f"/v1/messages/{msg['id']}/canonizations", json={"scope": "reality"} ) assert resp.status_code in (401, 403, 404) # 未携带新会话 Cookie/CSRF：拒绝 @pytest.mark.usefixtures("client") class TestRevisions: async def test_revision_keeps_original_intact(self, client: AsyncClient) -> None: headers = await login(client, "rev1@example.com") cid = await _create_conversation(client, headers) original = None resp = await client.post( f"/v1/conversations/{cid}/messages", json={"content": [{"type": "text", "text": "原始内容"}], "start_run": False}, headers={**headers, "Idempotency-Key": "idem-rev1"}, ) original = resp.json()["message"] rev = await client.post( f"/v1/messages/{original['id']}/revisions", json={"content": [{"type": "markdown", "text": "修订后\n```python\nprint(2)\n```"}]}, headers=headers, ) assert rev.status_code == 201, rev.text body = rev.json() assert body["parent_message_id"] == original["id"] assert body["variant_kind"] == "revision" # markdown 修订经引擎规范化为两块 assert [b["type"] for b in body["content"]] == ["markdown", "code"] # 原消息内容不变：列表中两条，原消息文本仍是"原始内容" listing = await client.get(f"/v1/conversations/{cid}/messages", headers=headers) by_id = {m["id"]: m for m in listing.json()["data"]} assert by_id[original["id"]]["content"][0]["text"] == "原始内容" async def test_revision_cross_domain_404(self, client: AsyncClient) -> None: alice = await login(client, "rev2@example.com") cid = await _create_conversation(client, alice) resp = await client.post( f"/v1/conversations/{cid}/messages", json={"content": [{"type": "text", "text": "x"}], "start_run": False}, headers={**alice, "Idempotency-Key": "idem-rev2"}, ) mid = resp.json()["message"]["id"] await login(client, "rev2-evil@example.com") rev = await client.post( f"/v1/messages/{mid}/revisions", json={"content": [{"type": "text", "text": "篡改"}]} ) assert rev.status_code in (401, 403, 404)