fix: reject ambiguous output map merges

This commit is contained in:
lda
2026-07-29 08:25:01 +07:00 Verified
parent e920abe0f5
commit 1cee713850
2 changed files with 231 additions and 19 deletions
+57 -19
View File
@@ -489,8 +489,18 @@ class WorkflowDraftApi:
merge: bool = False, merge: bool = False,
) -> dict[str, Any]: ) -> dict[str, Any]:
if merge: if merge:
workspace = self._workspace_if_revision_matches(
workspace_id=workspace_id,
revision=revision,
)
if isinstance(workspace, dict):
return workspace
step = _draft_step(workspace.draft, step_id)
output_map = { output_map = {
**self._step_output_map(workspace_id=workspace_id, step_id=step_id), **_require_lossless_step_output_map_round_trip(
step.get("output", []),
step_id=step_id,
),
**output_map, **output_map,
} }
return await self.patch_draft_workspace( return await self.patch_draft_workspace(
@@ -514,12 +524,39 @@ class WorkflowDraftApi:
merge: bool = False, merge: bool = False,
) -> dict[str, Any]: ) -> dict[str, Any]:
output_bindings: list[dict[str, Any]] output_bindings: list[dict[str, Any]]
workspace: WorkflowDraftWorkspace | None = None
if merge: if merge:
workspace = self._draft_store().get_workspace(workspace_id) checked_workspace = self._workspace_if_revision_matches(
workspace_id=workspace_id,
revision=revision,
)
if isinstance(checked_workspace, dict):
return checked_workspace
workspace = checked_workspace
remaining = dict(output_map) remaining = dict(output_map)
output_bindings = [] output_bindings = []
output_payload = workspace.draft.get("output") output_payload = workspace.draft.get("output")
if isinstance(output_payload, list): if isinstance(output_payload, list):
ambiguous = next(
(
source
for source in output_map
if sum(
1
for binding in output_payload
if isinstance(binding, dict)
and binding.get("path") == source
)
> 1
),
None,
)
if ambiguous is not None:
raise ValueError(
f"workflow output source {ambiguous!r} has multiple "
"bindings and cannot be updated through a compatibility "
"map; replace the complete canonical binding list instead"
)
for binding in output_payload: for binding in output_payload:
if not isinstance(binding, dict): if not isinstance(binding, dict):
continue continue
@@ -545,7 +582,8 @@ class WorkflowDraftApi:
{"path": source, "target": target} {"path": source, "target": target}
for source, target in output_map.items() for source, target in output_map.items()
] ]
workspace = self._draft_store().get_workspace(workspace_id) if workspace is None:
workspace = self._draft_store().get_workspace(workspace_id)
output_schema = self._workflow_output_schema_for_bindings( output_schema = self._workflow_output_schema_for_bindings(
draft=workspace.draft, draft=workspace.draft,
output_bindings=output_bindings, output_bindings=output_bindings,
@@ -627,22 +665,6 @@ class WorkflowDraftApi:
projected = updated projected = updated
return projected if changed else output_schema return projected if changed else output_schema
def _step_input_maps(
self,
*,
workspace_id: str,
step_id: str,
) -> tuple[dict[str, str], dict[str, Any]]:
workspace = self._draft_store().get_workspace(workspace_id)
step = _draft_step(workspace.draft, step_id)
return _input_maps_from_payload(step.get("input", []))
def _step_output_map(self, *, workspace_id: str, step_id: str) -> dict[str, str]:
workspace = self._draft_store().get_workspace(workspace_id)
step = _draft_step(workspace.draft, step_id)
return _output_map_from_payload(step.get("output", []))
def _workflow_source_schema( def _workflow_source_schema(
draft: Mapping[str, Any], draft: Mapping[str, Any],
source_path: str, source_path: str,
@@ -766,6 +788,22 @@ def _output_map_from_payload(payload: Any) -> dict[str, str]:
return output_map return output_map
def _require_lossless_step_output_map_round_trip(
payload: object,
*,
step_id: str,
) -> dict[str, str]:
"""Return a compatibility map only when it reproduces the output list."""
output_map = _output_map_from_payload(payload)
rebuilt = _draft_output_bindings_payload(output_map)
if rebuilt != payload:
raise ValueError(
f"step {step_id!r} outputs cannot be safely merged through a "
"compatibility map; replace the complete canonical binding list instead"
)
return output_map
def _path_text(value: Any, *, expected_root: str | None = None) -> str: def _path_text(value: Any, *, expected_root: str | None = None) -> str:
"""Return compact dotted text for stored structural path JSON.""" """Return compact dotted text for stored structural path JSON."""
if isinstance(value, str): if isinstance(value, str):
+174
View File
@@ -1404,6 +1404,84 @@ async def test_step_input_map_merge_checks_revision_before_lossless_preflight(
assert after == before assert after == before
@pytest.mark.asyncio
async def test_step_output_map_merge_rejects_canonical_source_fan_out(
tmp_path: Path,
) -> None:
artifact_store = FileWorkflowArtifactStore(tmp_path / "output_merge_fan_out")
api, _service, _authoring = _draft_api(artifact_store)
draft = _echo_draft()
draft["state_schema"] = {
"fields": {
"echoed": {"type": "string"},
"other": {"type": "string"},
}
}
draft["steps"]["echo"]["output"] = [
{"source": "echoed", "target": "state.echoed"},
{"source": "echoed", "target": "state.other"},
]
await api.create_draft_workspace(workspace_id="echo_ws", draft=draft)
before = await api.get_draft_workspace(
workspace_id="echo_ws",
include_draft=True,
)
with pytest.raises(ValueError, match="complete canonical binding list"):
await api.set_step_output_map(
workspace_id="echo_ws",
revision=1,
step_id="echo",
output_map={"extra": "state.extra"},
merge=True,
)
after = await api.get_draft_workspace(
workspace_id="echo_ws",
include_draft=True,
)
assert after == before
@pytest.mark.asyncio
async def test_step_output_map_merge_checks_revision_before_lossless_preflight(
tmp_path: Path,
) -> None:
artifact_store = FileWorkflowArtifactStore(tmp_path / "output_merge_stale")
api, _service, _authoring = _draft_api(artifact_store)
draft = _echo_draft()
draft["steps"]["echo"]["output"] = [
{"source": "echoed", "target": "state.echoed"},
{"source": "echoed", "target": "state.other"},
]
await api.create_draft_workspace(workspace_id="echo_ws", draft=draft)
await api.set_draft_name(
workspace_id="echo_ws",
revision=1,
name="echo_v2",
)
before = await api.get_draft_workspace(
workspace_id="echo_ws",
include_draft=True,
)
result = await api.set_step_output_map(
workspace_id="echo_ws",
revision=1,
step_id="echo",
output_map={"extra": "state.extra"},
merge=True,
)
after = await api.get_draft_workspace(
workspace_id="echo_ws",
include_draft=True,
)
assert result["status"] == "conflict"
assert result["diagnostics"][0]["code"] == "revision_conflict"
assert after == before
@pytest.mark.asyncio @pytest.mark.asyncio
async def test_validate_draft_workspace_refreshes_status(tmp_path: Path) -> None: async def test_validate_draft_workspace_refreshes_status(tmp_path: Path) -> None:
artifact_store = FileWorkflowArtifactStore(tmp_path / "drafts_validate_workspace") artifact_store = FileWorkflowArtifactStore(tmp_path / "drafts_validate_workspace")
@@ -4938,6 +5016,102 @@ async def test_set_workflow_output_map_merges_top_level_output(tmp_path: Path) -
] ]
@pytest.mark.asyncio
async def test_workflow_output_map_merge_rejects_requested_fan_out_source(
tmp_path: Path,
) -> None:
artifact_store = FileWorkflowArtifactStore(tmp_path / "workflow_merge_fan_out")
api, _service, _authoring = _draft_api(artifact_store, register_echo=True)
draft = {
**_echo_draft(),
"output_schema": {
"type": "object",
"properties": {
"first": {"type": "string"},
"second": {"type": "string"},
"kind": {"type": "string"},
},
},
"output": [
{"path": "state.echoed", "target": "first"},
{"path": "state.echoed", "target": "second"},
{"value": "markdown", "target": "kind"},
],
}
await api.create_draft_workspace(workspace_id="report", draft=draft)
before = await api.get_draft_workspace(
workspace_id="report",
include_draft=True,
)
with pytest.raises(ValueError, match="multiple bindings"):
await api.set_workflow_output_map(
workspace_id="report",
revision=1,
output_map={"state.echoed": "renamed"},
merge=True,
)
after = await api.get_draft_workspace(
workspace_id="report",
include_draft=True,
)
assert after == before
@pytest.mark.asyncio
async def test_workflow_output_map_merge_preserves_unrequested_fan_out_and_literals(
tmp_path: Path,
) -> None:
artifact_store = FileWorkflowArtifactStore(
tmp_path / "workflow_merge_unrequested_fan_out"
)
api, _service, _authoring = _draft_api(artifact_store, register_echo=True)
draft = {
**_echo_draft(),
"state_schema": {
"fields": {
"echoed": {"type": "string"},
"other": {"type": "string"},
}
},
"output_schema": {
"type": "object",
"properties": {
"first": {"type": "string"},
"second": {"type": "string"},
"kind": {"type": "string"},
"other": {"type": "string"},
},
},
"output": [
{"path": "state.echoed", "target": "first"},
{"path": "state.echoed", "target": "second"},
{"value": "markdown", "target": "kind"},
],
}
await api.create_draft_workspace(workspace_id="report", draft=draft)
result = await api.set_workflow_output_map(
workspace_id="report",
revision=1,
output_map={"state.other": "other"},
merge=True,
)
fetched = await api.get_draft_workspace(
workspace_id="report",
include_draft=True,
)
assert result["revision"] == 2
assert fetched["draft"]["output"] == [
{"path": "state.echoed", "target": "first"},
{"path": "state.echoed", "target": "second"},
{"value": "markdown", "target": "kind"},
{"path": "state.other", "target": "other"},
]
@pytest.mark.asyncio @pytest.mark.asyncio
async def test_set_workflow_output_map_projects_missing_output_schema( async def test_set_workflow_output_map_projects_missing_output_schema(
tmp_path: Path, tmp_path: Path,