can1357
daf2680298
Two defects flagged by review on the final head, both in the new drain/settle machinery: - drainInFlightDispatches looped on Promise.all unconditionally. Exec handlers have no cancellation contract (the bridge calls tool.execute with no signal), so a hung or long-running tool held the terminal error hostage after Ctrl+C. The drain now returns once the signal aborts; late results were discarded after agent_end regardless. - A host todoSync callback that throws (session persistence on disk failure) skipped both the paired result and toolcall_end, stranding the live card and leaving the resolved block to be stripped from rebuilt transcripts. The callback is now caught and the call settles as a failure carrying the thrown message. Both regression tests fail without their fix (the first by hanging the stream).
952 lines
31 KiB
TypeScript
952 lines
31 KiB
TypeScript
import { describe, expect, it } from "bun:test";
|
|
import { create, fromBinary, toBinary } from "@bufbuild/protobuf";
|
|
import {
|
|
type BlockState,
|
|
processInteractionUpdate,
|
|
type ToolCallState,
|
|
type UsageState,
|
|
} from "@oh-my-pi/pi-ai/providers/cursor";
|
|
import type { AssistantMessage, CursorTodoSnapshot, ToolResultMessage } from "@oh-my-pi/pi-ai/types";
|
|
import { kCursorExecResolved } from "@oh-my-pi/pi-ai/utils/block-symbols";
|
|
import { AssistantMessageEventStream } from "@oh-my-pi/pi-ai/utils/event-stream";
|
|
import {
|
|
AgentServerMessageSchema,
|
|
InteractionUpdateSchema,
|
|
McpArgsSchema,
|
|
McpToolCallSchema,
|
|
ReadTodosArgsSchema,
|
|
ReadTodosResultSchema,
|
|
ReadTodosSuccessSchema,
|
|
ReadTodosToolCallSchema,
|
|
type TodoItem,
|
|
TodoItemSchema,
|
|
type ToolCall,
|
|
ToolCallCompletedUpdateSchema,
|
|
ToolCallSchema,
|
|
ToolCallStartedUpdateSchema,
|
|
UpdateTodosArgsSchema,
|
|
UpdateTodosErrorSchema,
|
|
UpdateTodosResultSchema,
|
|
UpdateTodosSuccessSchema,
|
|
UpdateTodosToolCallSchema,
|
|
} from "@oh-my-pi/pi-catalog/discovery/cursor-gen/agent_pb";
|
|
|
|
/** One `todoSync` invocation, recorded verbatim. */
|
|
interface SyncCall {
|
|
snapshot: CursorTodoSnapshot | null;
|
|
toolCallId: string;
|
|
error: string | null;
|
|
}
|
|
|
|
interface Harness {
|
|
output: AssistantMessage;
|
|
stream: AssistantMessageEventStream;
|
|
state: BlockState;
|
|
usageState: UsageState;
|
|
/** Every settle, including refusals and server errors. */
|
|
syncCalls: SyncCall[];
|
|
/** Mirrored snapshots only — the subset that carries a list. */
|
|
snapshots: CursorTodoSnapshot[];
|
|
/** Call ids handed to `todoSync`, in order. */
|
|
syncCallIds: string[];
|
|
/** Results persisted for server-resolved blocks. */
|
|
toolResults: ToolResultMessage[];
|
|
}
|
|
|
|
function newHarness(): Harness {
|
|
const output: AssistantMessage = {
|
|
role: "assistant",
|
|
content: [],
|
|
api: "cursor-agent",
|
|
provider: "cursor",
|
|
model: "cursor-composer-2.5",
|
|
usage: {
|
|
input: 0,
|
|
output: 0,
|
|
cacheRead: 0,
|
|
cacheWrite: 0,
|
|
totalTokens: 0,
|
|
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0, total: 0 },
|
|
},
|
|
stopReason: "stop",
|
|
timestamp: 0,
|
|
};
|
|
const stream = new AssistantMessageEventStream();
|
|
const syncCalls: SyncCall[] = [];
|
|
const snapshots: CursorTodoSnapshot[] = [];
|
|
const syncCallIds: string[] = [];
|
|
const toolResults: ToolResultMessage[] = [];
|
|
let textBlock: BlockState["currentTextBlock"] = null;
|
|
let thinkingBlock: BlockState["currentThinkingBlock"] = null;
|
|
let toolCall: ToolCallState | null = null;
|
|
const state: BlockState = {
|
|
get currentTextBlock() {
|
|
return textBlock;
|
|
},
|
|
get currentThinkingBlock() {
|
|
return thinkingBlock;
|
|
},
|
|
get currentToolCall() {
|
|
return toolCall;
|
|
},
|
|
resolvedMcpToolCallIds: new Set(),
|
|
firstTokenTime: undefined,
|
|
setTextBlock: b => {
|
|
textBlock = b;
|
|
},
|
|
setThinkingBlock: b => {
|
|
thinkingBlock = b;
|
|
},
|
|
setToolCall: t => {
|
|
toolCall = t;
|
|
},
|
|
setFirstTokenTime: () => {},
|
|
onTodoSnapshot: (snapshot, toolCallId, error) => {
|
|
syncCalls.push({ snapshot, toolCallId, error });
|
|
if (snapshot) snapshots.push(snapshot);
|
|
syncCallIds.push(toolCallId);
|
|
// Stands in for the host's phase-grouped result: the provider must
|
|
// persist it verbatim rather than synthesizing its own.
|
|
return {
|
|
role: "toolResult",
|
|
toolCallId,
|
|
toolName: "todo",
|
|
content: [{ type: "text", text: error ?? "host result" }],
|
|
isError: error !== null,
|
|
timestamp: 0,
|
|
};
|
|
},
|
|
onToolResult: result => {
|
|
toolResults.push(result);
|
|
return result;
|
|
},
|
|
};
|
|
return {
|
|
output,
|
|
stream,
|
|
state,
|
|
usageState: { sawTokenDelta: false },
|
|
syncCalls,
|
|
snapshots,
|
|
syncCallIds,
|
|
toolResults,
|
|
};
|
|
}
|
|
|
|
function start(h: Harness, toolCall: unknown, callId = "call-1"): void {
|
|
processInteractionUpdate(
|
|
{ message: { case: "toolCallStarted", value: { callId, toolCall } } },
|
|
h.output,
|
|
h.stream,
|
|
h.state,
|
|
h.usageState,
|
|
);
|
|
}
|
|
|
|
function complete(h: Harness, toolCall: unknown): void {
|
|
processInteractionUpdate(
|
|
{ message: { case: "toolCallCompleted", value: { toolCall } } },
|
|
h.output,
|
|
h.stream,
|
|
h.state,
|
|
h.usageState,
|
|
);
|
|
}
|
|
|
|
function todoBlocks(h: Harness): ToolCallState[] {
|
|
return h.output.content.filter((c): c is ToolCallState => c.type === "toolCall" && c.name === "todo");
|
|
}
|
|
|
|
describe("cursor native todo bridge", () => {
|
|
it("never marks a native todo block runnable by the shared tool loop", () => {
|
|
// `ExecServerMessage` has no todo case: Cursor settles these server-side.
|
|
// An unresolved block would make `agent-loop` execute a tool that has no
|
|
// local counterpart and drive a spurious continuation turn.
|
|
const h = newHarness();
|
|
start(h, { updateTodosToolCall: { args: { todos: [{ content: "a", status: 1 }] } } });
|
|
start(h, { readTodosToolCall: { args: {} } }, "call-2");
|
|
|
|
const blocks = todoBlocks(h);
|
|
expect(blocks).toHaveLength(2);
|
|
for (const block of blocks) {
|
|
expect(block[kCursorExecResolved]).toBe(true);
|
|
}
|
|
});
|
|
|
|
it("takes the server success snapshot as truth over the requested args", () => {
|
|
const h = newHarness();
|
|
const args = { merge: true, todos: [{ content: "requested", status: 1 }] };
|
|
start(h, { updateTodosToolCall: { args } });
|
|
complete(h, {
|
|
updateTodosToolCall: {
|
|
args,
|
|
result: {
|
|
result: {
|
|
case: "success",
|
|
value: {
|
|
wasMerge: true,
|
|
todos: [
|
|
{ content: "pre-existing", status: 3 },
|
|
{ content: "requested", status: 2 },
|
|
],
|
|
},
|
|
},
|
|
},
|
|
},
|
|
});
|
|
|
|
expect(h.snapshots).toEqual([
|
|
{
|
|
todos: [
|
|
{ content: "pre-existing", status: "completed" },
|
|
{ content: "requested", status: "in_progress" },
|
|
],
|
|
merged: true,
|
|
},
|
|
]);
|
|
expect(todoBlocks(h)[0].arguments).toEqual({
|
|
todos: [
|
|
{ content: "pre-existing", status: "completed" },
|
|
{ content: "requested", status: "in_progress" },
|
|
],
|
|
merged: true,
|
|
});
|
|
});
|
|
|
|
it("maps TODO_STATUS_CANCELLED to abandoned instead of resurrecting the task", () => {
|
|
const h = newHarness();
|
|
start(h, { updateTodosToolCall: { args: { todos: [] } } });
|
|
complete(h, {
|
|
updateTodosToolCall: {
|
|
args: { todos: [] },
|
|
result: { result: { case: "success", value: { todos: [{ content: "dropped", status: 4 }] } } },
|
|
},
|
|
});
|
|
|
|
expect(h.snapshots[0].todos).toEqual([{ content: "dropped", status: "abandoned" }]);
|
|
});
|
|
|
|
it("leaves local state untouched when the server reports an error", () => {
|
|
const h = newHarness();
|
|
const args = { todos: [{ content: "requested", status: 1 }] };
|
|
start(h, { updateTodosToolCall: { args } });
|
|
const before = todoBlocks(h)[0].arguments;
|
|
complete(h, {
|
|
updateTodosToolCall: { args, result: { result: { case: "error", value: { error: "quota exceeded" } } } },
|
|
});
|
|
|
|
expect(h.snapshots).toEqual([]);
|
|
expect(todoBlocks(h)[0].arguments).toEqual(before);
|
|
});
|
|
|
|
it("leaves local state untouched when the completion carries no result", () => {
|
|
const h = newHarness();
|
|
const args = { todos: [{ content: "requested", status: 1 }] };
|
|
start(h, { updateTodosToolCall: { args } });
|
|
complete(h, { updateTodosToolCall: { args } });
|
|
|
|
expect(h.snapshots).toEqual([]);
|
|
});
|
|
|
|
it("refreshes local state from a read_todos snapshot", () => {
|
|
const h = newHarness();
|
|
start(h, { readTodosToolCall: { args: {} } });
|
|
complete(h, {
|
|
readTodosToolCall: {
|
|
args: {},
|
|
result: {
|
|
result: { case: "success", value: { todos: [{ content: "remote", status: 2 }], totalCount: 1 } },
|
|
},
|
|
},
|
|
});
|
|
|
|
expect(h.snapshots).toEqual([{ todos: [{ content: "remote", status: "in_progress" }], merged: false }]);
|
|
});
|
|
|
|
it("refuses a status-filtered read_todos result, which is a subset and not the list", () => {
|
|
// `ReadTodosArgs.status_filter` narrows the response; mirroring it would
|
|
// delete every task the filter excluded.
|
|
const h = newHarness();
|
|
const args = { statusFilter: [2] };
|
|
start(h, { readTodosToolCall: { args } });
|
|
complete(h, {
|
|
readTodosToolCall: {
|
|
args,
|
|
result: { result: { case: "success", value: { todos: [{ content: "only in progress", status: 2 }] } } },
|
|
},
|
|
});
|
|
|
|
expect(h.snapshots).toEqual([]);
|
|
});
|
|
|
|
it("refuses an id-filtered read_todos result", () => {
|
|
const h = newHarness();
|
|
const args = { idFilter: ["task-1"] };
|
|
start(h, { readTodosToolCall: { args } });
|
|
complete(h, {
|
|
readTodosToolCall: {
|
|
args,
|
|
result: { result: { case: "success", value: { todos: [{ content: "one task", status: 1 }] } } },
|
|
},
|
|
});
|
|
|
|
expect(h.snapshots).toEqual([]);
|
|
});
|
|
|
|
it("refuses a read_todos result truncated below the server's own total_count", () => {
|
|
const h = newHarness();
|
|
start(h, { readTodosToolCall: { args: {} } });
|
|
complete(h, {
|
|
readTodosToolCall: {
|
|
args: {},
|
|
result: {
|
|
result: {
|
|
case: "success",
|
|
value: { todos: [{ content: "first of three", status: 2 }], totalCount: 3 },
|
|
},
|
|
},
|
|
},
|
|
});
|
|
|
|
expect(h.snapshots).toEqual([]);
|
|
});
|
|
|
|
it("refuses an empty read_todos whose total_count is zero or unset", () => {
|
|
// proto3 defaults an unset `total_count` to 0, so todos=[] + totalCount=0
|
|
// is indistinguishable from a genuinely empty list. Mirroring it would
|
|
// wipe every local task; update_todos remains the clear path.
|
|
const h = newHarness();
|
|
start(h, { readTodosToolCall: { args: {} } });
|
|
complete(h, {
|
|
readTodosToolCall: {
|
|
args: {},
|
|
result: {
|
|
result: {
|
|
case: "success",
|
|
value: { todos: [], totalCount: 0 },
|
|
},
|
|
},
|
|
},
|
|
});
|
|
expect(h.snapshots).toEqual([]);
|
|
|
|
// Positive control: the same empty snapshot via update_todos DOES sync —
|
|
// clearing the list is an authoritative write, not an ambiguous read.
|
|
const cleared = newHarness();
|
|
start(cleared, { updateTodosToolCall: { args: { todos: [] } } });
|
|
complete(cleared, {
|
|
updateTodosToolCall: {
|
|
args: { todos: [] },
|
|
result: {
|
|
result: {
|
|
case: "success",
|
|
value: { todos: [], totalCount: 0 },
|
|
},
|
|
},
|
|
},
|
|
});
|
|
expect(cleared.snapshots).toEqual([{ todos: [], merged: false }]);
|
|
});
|
|
|
|
it("accepts a read_todos result whose row count matches total_count", () => {
|
|
const h = newHarness();
|
|
start(h, { readTodosToolCall: { args: {} } });
|
|
complete(h, {
|
|
readTodosToolCall: {
|
|
args: {},
|
|
result: {
|
|
result: {
|
|
case: "success",
|
|
value: {
|
|
todos: [
|
|
{ content: "one", status: 3 },
|
|
{ content: "two", status: 2 },
|
|
],
|
|
totalCount: 2,
|
|
},
|
|
},
|
|
},
|
|
},
|
|
});
|
|
|
|
expect(h.snapshots[0].todos).toEqual([
|
|
{ content: "one", status: "completed" },
|
|
{ content: "two", status: "in_progress" },
|
|
]);
|
|
});
|
|
});
|
|
|
|
/**
|
|
* The fixtures above hand-shape `toolCall` with a flattened
|
|
* `updateTodosToolCall` property. A decoded `agent.v1.ToolCall` does NOT look
|
|
* like that: `tool` is a protobuf oneof, so the variant only ever arrives as
|
|
* `tool: { case, value }`. These tests drive the bridge with messages that
|
|
* round-trip through the actual wire encoding, which is the only shape
|
|
* production ever sees.
|
|
*/
|
|
describe("cursor native todo bridge (wire-encoded protobuf)", () => {
|
|
function wireUpdate(kind: "toolCallStarted" | "toolCallCompleted", toolCall?: ToolCall): unknown {
|
|
// `toolCall` is optional on the wire: omitting it exercises a completion
|
|
// frame that carries no result at all.
|
|
const value =
|
|
kind === "toolCallStarted"
|
|
? create(ToolCallStartedUpdateSchema, { callId: "call-1", toolCall })
|
|
: create(ToolCallCompletedUpdateSchema, { callId: "call-1", toolCall });
|
|
const server = create(AgentServerMessageSchema, {
|
|
message: {
|
|
case: "interactionUpdate",
|
|
value: create(InteractionUpdateSchema, { message: { case: kind, value } } as never),
|
|
},
|
|
});
|
|
// handleServerMessage forwards `message.value` to processInteractionUpdate.
|
|
return fromBinary(AgentServerMessageSchema, toBinary(AgentServerMessageSchema, server)).message.value;
|
|
}
|
|
|
|
function items(rows: [string, string, number][]) {
|
|
return rows.map(([id, content, status]) => create(TodoItemSchema, { id, content, status }));
|
|
}
|
|
|
|
/** Rows carrying `TodoItem.dependencies` — the ids each row waits on. */
|
|
function depItems(rows: [string, string, number, string[]][]) {
|
|
return rows.map(([id, content, status, dependencies]) =>
|
|
create(TodoItemSchema, { id, content, status, dependencies }),
|
|
);
|
|
}
|
|
|
|
function successResult(todos: TodoItem[], totalCount: number, wasMerge = false) {
|
|
return create(UpdateTodosResultSchema, {
|
|
result: {
|
|
case: "success",
|
|
value: create(UpdateTodosSuccessSchema, { todos, totalCount, wasMerge }),
|
|
},
|
|
});
|
|
}
|
|
|
|
// `read_todos` settles on its own result message: no `was_merge`, and a
|
|
// `total_count` that reports the full size even when rows are withheld.
|
|
function readSuccessResult(todos: TodoItem[], totalCount: number) {
|
|
return create(ReadTodosResultSchema, {
|
|
result: { case: "success", value: create(ReadTodosSuccessSchema, { todos, totalCount }) },
|
|
});
|
|
}
|
|
|
|
function updateCall(todos: TodoItem[], totalCount: number, wasMerge = false): ToolCall {
|
|
return create(ToolCallSchema, {
|
|
tool: {
|
|
case: "updateTodosToolCall",
|
|
value: create(UpdateTodosToolCallSchema, {
|
|
args: create(UpdateTodosArgsSchema, { todos: [], merge: wasMerge }),
|
|
result: successResult(todos, totalCount, wasMerge),
|
|
}),
|
|
},
|
|
});
|
|
}
|
|
|
|
function readCall(
|
|
todos: TodoItem[],
|
|
totalCount: number,
|
|
args: { statusFilter?: number[]; idFilter?: string[] } = {},
|
|
): ToolCall {
|
|
return create(ToolCallSchema, {
|
|
tool: {
|
|
case: "readTodosToolCall",
|
|
value: create(ReadTodosToolCallSchema, {
|
|
args: create(ReadTodosArgsSchema, args),
|
|
result: readSuccessResult(todos, totalCount),
|
|
}),
|
|
},
|
|
});
|
|
}
|
|
|
|
/** A completed `update_todos` the server rejected outright. */
|
|
function errorCall(error: string): ToolCall {
|
|
return create(ToolCallSchema, {
|
|
tool: {
|
|
case: "updateTodosToolCall",
|
|
value: create(UpdateTodosToolCallSchema, {
|
|
args: create(UpdateTodosArgsSchema, { todos: [], merge: false }),
|
|
result: create(UpdateTodosResultSchema, {
|
|
result: { case: "error", value: create(UpdateTodosErrorSchema, { error }) },
|
|
}),
|
|
}),
|
|
},
|
|
});
|
|
}
|
|
|
|
function drive(toolCall: ToolCall): Harness {
|
|
const h = newHarness();
|
|
processInteractionUpdate(
|
|
wireUpdate("toolCallStarted", toolCall) as never,
|
|
h.output,
|
|
h.stream,
|
|
h.state,
|
|
h.usageState,
|
|
);
|
|
processInteractionUpdate(
|
|
wireUpdate("toolCallCompleted", toolCall) as never,
|
|
h.output,
|
|
h.stream,
|
|
h.state,
|
|
h.usageState,
|
|
);
|
|
return h;
|
|
}
|
|
|
|
it("synthesizes a todo block from a wire-decoded update_todos oneof", () => {
|
|
const h = drive(updateCall(items([["1", "done task", 3]]), 1));
|
|
|
|
expect(todoBlocks(h)).toHaveLength(1);
|
|
expect(todoBlocks(h)[0][kCursorExecResolved]).toBe(true);
|
|
});
|
|
|
|
it("mirrors the server snapshot from a wire-decoded update_todos oneof", () => {
|
|
const h = drive(
|
|
updateCall(
|
|
items([
|
|
["1", "done task", 3],
|
|
["2", "active task", 2],
|
|
]),
|
|
2,
|
|
true,
|
|
),
|
|
);
|
|
|
|
expect(h.snapshots).toEqual([
|
|
{
|
|
todos: [
|
|
{ content: "done task", status: "completed" },
|
|
{ content: "active task", status: "in_progress" },
|
|
],
|
|
merged: true,
|
|
},
|
|
]);
|
|
});
|
|
|
|
it("mirrors a complete wire-decoded read_todos oneof", () => {
|
|
const h = drive(readCall(items([["1", "only task", 2]]), 1));
|
|
|
|
expect(h.snapshots).toEqual([{ todos: [{ content: "only task", status: "in_progress" }], merged: false }]);
|
|
});
|
|
|
|
it("refuses a wire-decoded read_todos truncated below total_count", () => {
|
|
const h = drive(readCall(items([["1", "only task", 2]]), 5));
|
|
|
|
expect(todoBlocks(h)).toHaveLength(1);
|
|
expect(h.snapshots).toEqual([]);
|
|
});
|
|
|
|
it("refuses an update_todos snapshot truncated below total_count", () => {
|
|
// A partial or size-limited merge response is as incomplete as a filtered
|
|
// read: mirroring it would delete every task the server omitted. The
|
|
// refusal is not read-only.
|
|
//
|
|
// Positive control: the same row with an honest count does sync.
|
|
expect(drive(updateCall(items([["1", "only task", 2]]), 1)).snapshots).toHaveLength(1);
|
|
|
|
const h = drive(updateCall(items([["1", "only task", 2]]), 5));
|
|
|
|
expect(todoBlocks(h)).toHaveLength(1);
|
|
expect(h.snapshots).toEqual([]);
|
|
});
|
|
|
|
it("still mirrors an empty update_todos, the authoritative clear path", () => {
|
|
// Unlike an empty read (ambiguous: proto3 defaults `total_count` to 0),
|
|
// an empty update is an explicit "the list is now nothing" and must still
|
|
// clear local state — the count guard above must not swallow it.
|
|
const h = drive(updateCall([], 0));
|
|
|
|
expect(h.snapshots).toEqual([{ todos: [], merged: false }]);
|
|
});
|
|
|
|
it("refuses an empty update_todos whose total_count is nonzero", () => {
|
|
// The most destructive shape: an empty partial/size-limited merge response
|
|
// that still reports rows. Accepting it as an authoritative clear would
|
|
// delete every local task at once. Only a matching zero count is a clear.
|
|
const h = drive(updateCall([], 3));
|
|
|
|
expect(todoBlocks(h)).toHaveLength(1);
|
|
expect(h.snapshots).toEqual([]);
|
|
expect(h.syncCalls).toEqual([{ snapshot: null, toolCallId: todoBlocks(h)[0].id, error: null }]);
|
|
});
|
|
|
|
it("settles a completion frame that carries no toolCall at all", () => {
|
|
// `ToolCallCompletedUpdate.tool_call` is optional. The started frame has
|
|
// already marked the block `kCursorExecResolved`, so `agent-loop.ts` emits
|
|
// no placeholder result for it — staying silent here would leave the call
|
|
// unpaired and `buildSessionContext` would strip the whole interaction
|
|
// from every rebuilt transcript.
|
|
const h = newHarness();
|
|
const toolCall = updateCall(items([["1", "step one", 2]]), 1);
|
|
processInteractionUpdate(
|
|
wireUpdate("toolCallStarted", toolCall) as never,
|
|
h.output,
|
|
h.stream,
|
|
h.state,
|
|
h.usageState,
|
|
);
|
|
processInteractionUpdate(wireUpdate("toolCallCompleted") as never, h.output, h.stream, h.state, h.usageState);
|
|
|
|
const callId = todoBlocks(h)[0].id;
|
|
expect(todoBlocks(h)).toHaveLength(1);
|
|
// Nothing to mirror: no snapshot and no server error.
|
|
expect(h.snapshots).toEqual([]);
|
|
expect(h.syncCalls).toEqual([{ snapshot: null, toolCallId: callId, error: null }]);
|
|
// But it IS paired, so the block survives a rebuild.
|
|
expect(h.toolResults.map(r => r.toolCallId)).toEqual([callId]);
|
|
expect(h.toolResults[0]).toMatchObject({ role: "toolResult", toolName: "todo", isError: false });
|
|
});
|
|
|
|
it("refuses a snapshot carrying a row with empty content", () => {
|
|
// `content` is a proto3 string: missing or default arrives as `""`. The
|
|
// local list is keyed by content and `resolveTaskOrError` rejects a falsy
|
|
// one before lookup, so the row would be unreachable to every
|
|
// task-targeted `done`/`drop`/`rm`.
|
|
//
|
|
// Positive control: the same pair with real content syncs.
|
|
expect(
|
|
drive(
|
|
updateCall(
|
|
items([
|
|
["1", "real task", 1],
|
|
["2", "other task", 1],
|
|
]),
|
|
2,
|
|
),
|
|
).snapshots,
|
|
).toHaveLength(1);
|
|
|
|
const h = drive(
|
|
updateCall(
|
|
items([
|
|
["1", "real task", 1],
|
|
["2", "", 1],
|
|
]),
|
|
2,
|
|
),
|
|
);
|
|
|
|
expect(todoBlocks(h)).toHaveLength(1);
|
|
expect(h.snapshots).toEqual([]);
|
|
// Still settles as a benign no-op, like every other refusal.
|
|
expect(h.syncCalls).toEqual([{ snapshot: null, toolCallId: todoBlocks(h)[0].id, error: null }]);
|
|
});
|
|
|
|
it("refuses a wire-decoded empty read_todos with total_count 0", () => {
|
|
const h = drive(readCall([], 0));
|
|
expect(todoBlocks(h)).toHaveLength(1);
|
|
expect(h.snapshots).toEqual([]);
|
|
// Still settles as a benign non-mirror, not a wipe and not a failure.
|
|
expect(h.syncCalls).toEqual([{ snapshot: null, toolCallId: todoBlocks(h)[0].id, error: null }]);
|
|
});
|
|
|
|
it("refuses a snapshot whose rows collide on content", () => {
|
|
// The wire model identifies rows by `id` and can represent two rows with
|
|
// the same `content`, so the bridge must survive one. The local list is
|
|
// keyed by content alone, so importing the pair would make every later
|
|
// task-targeted `done`/`drop` resolve to the first row and strand the
|
|
// second.
|
|
//
|
|
// Positive control: the same two rows with distinct content do sync, so
|
|
// the refusal below is attributable to the collision.
|
|
const distinct = drive(
|
|
updateCall(
|
|
items([
|
|
["1", "task a", 1],
|
|
["2", "task b", 1],
|
|
]),
|
|
2,
|
|
),
|
|
);
|
|
expect(distinct.snapshots).toHaveLength(1);
|
|
|
|
const h = drive(
|
|
updateCall(
|
|
items([
|
|
["1", "same task", 1],
|
|
["2", "same task", 3],
|
|
]),
|
|
2,
|
|
),
|
|
);
|
|
const callId = todoBlocks(h)[0].id;
|
|
|
|
expect(todoBlocks(h)).toHaveLength(1);
|
|
// Nothing mutable reaches the host: no snapshot at all, so it cannot
|
|
// mirror a list the local model is unable to key.
|
|
expect(h.snapshots).toEqual([]);
|
|
expect(h.syncCalls).toEqual([{ snapshot: null, toolCallId: callId, error: null }]);
|
|
// Still settles: the call happened, only the mirror was declined. Without
|
|
// a completion the card animates forever, and without a paired result the
|
|
// block is stripped on rebuild. It reads as a benign no-op, not a failure.
|
|
expect(h.toolResults.map(r => r.toolCallId)).toEqual([callId]);
|
|
expect(h.toolResults[0]).toMatchObject({ role: "toolResult", toolName: "todo", isError: false });
|
|
});
|
|
|
|
it("settles a collided snapshot as a no-op when no host handler is registered", () => {
|
|
// Without a host the provider builds the result itself: a declined mirror
|
|
// must read as "not mirrored", never as an empty list, or a rebuilt
|
|
// transcript would claim the server wiped every task. The server may have
|
|
// accepted the update — claiming "No todo changes" would be false about that.
|
|
const h = newHarness();
|
|
h.state.onTodoSnapshot = undefined;
|
|
const toolCall = updateCall(
|
|
items([
|
|
["1", "same task", 1],
|
|
["2", "same task", 3],
|
|
]),
|
|
2,
|
|
);
|
|
for (const kind of ["toolCallStarted", "toolCallCompleted"] as const) {
|
|
processInteractionUpdate(wireUpdate(kind, toolCall) as never, h.output, h.stream, h.state, h.usageState);
|
|
}
|
|
|
|
expect(h.toolResults[0]).toMatchObject({
|
|
toolCallId: todoBlocks(h)[0].id,
|
|
isError: false,
|
|
content: [{ type: "text", text: "Todo snapshot not mirrored" }],
|
|
});
|
|
});
|
|
|
|
it("refuses a snapshot whose rows carry unresolved dependencies", () => {
|
|
// `TodoItem.dependencies` is a graph the local model cannot store: rows are
|
|
// keyed by content and have no id, so an edge cannot be replayed or
|
|
// re-evaluated when the blocker finishes. Importing the row anyway files it
|
|
// as plain `pending`, and `nextActionableTask` then offers work the server
|
|
// says is not ready.
|
|
//
|
|
// Positive control: the same two rows without the edge do sync, so the
|
|
// refusal is attributable to the dependency and not to a decode failure.
|
|
const rows: [string, string, number][] = [
|
|
["1", "blocker task", 1],
|
|
["2", "dependent task", 1],
|
|
];
|
|
expect(drive(updateCall(items(rows), 2)).snapshots).toHaveLength(1);
|
|
|
|
const h = drive(
|
|
updateCall(
|
|
depItems([
|
|
["1", "blocker task", 1, []],
|
|
["2", "dependent task", 1, ["1"]],
|
|
]),
|
|
2,
|
|
),
|
|
);
|
|
const callId = todoBlocks(h)[0].id;
|
|
|
|
expect(todoBlocks(h)).toHaveLength(1);
|
|
expect(h.snapshots).toEqual([]);
|
|
// Settles as a benign no-op under the streamed id, like every other refusal.
|
|
expect(h.syncCalls).toEqual([{ snapshot: null, toolCallId: callId, error: null }]);
|
|
expect(h.toolResults[0]).toMatchObject({ toolCallId: callId, isError: false });
|
|
});
|
|
|
|
it("still mirrors when every dependency is already finished", () => {
|
|
// A dependency on a completed row constrains nothing, so refusing it would
|
|
// strand late-session snapshots — by then most edges point at done work.
|
|
const h = drive(
|
|
updateCall(
|
|
depItems([
|
|
["1", "blocker task", 3, []],
|
|
["2", "dependent task", 1, ["1"]],
|
|
]),
|
|
2,
|
|
),
|
|
);
|
|
|
|
expect(h.snapshots).toEqual([
|
|
{
|
|
todos: [
|
|
{ content: "blocker task", status: "completed" },
|
|
{ content: "dependent task", status: "pending" },
|
|
],
|
|
merged: false,
|
|
},
|
|
]);
|
|
});
|
|
|
|
it("refuses a wire-decoded read_todos narrowed by status_filter", () => {
|
|
const rows: [string, string, number][] = [["1", "only task", 3]];
|
|
// Positive control: the identical response without the filter does sync,
|
|
// so the refusal below is attributable to the filter and not to the
|
|
// bridge failing to decode the oneof at all.
|
|
expect(drive(readCall(items(rows), 1)).snapshots).toHaveLength(1);
|
|
|
|
const h = drive(readCall(items(rows), 1, { statusFilter: [3] }));
|
|
|
|
expect(todoBlocks(h)).toHaveLength(1);
|
|
expect(h.snapshots).toEqual([]);
|
|
});
|
|
|
|
it("leaves local state untouched when the wire result carries an error", () => {
|
|
const h = drive(errorCall("boom"));
|
|
|
|
expect(todoBlocks(h)).toHaveLength(1);
|
|
expect(h.snapshots).toEqual([]);
|
|
});
|
|
|
|
it("syncs under the streamed call id so the visible block can resolve", () => {
|
|
// The interactive transcript files the block under the streamed `callId`
|
|
// and only clears it when `tool_execution_end.toolCallId` matches. A
|
|
// freshly generated id leaves that card pending and animating forever.
|
|
const h = drive(updateCall(items([["1", "task", 2]]), 1));
|
|
|
|
expect(h.syncCallIds).toEqual([todoBlocks(h)[0].id]);
|
|
});
|
|
|
|
it("persists a paired result so the block survives a transcript rebuild", () => {
|
|
// `buildSessionContext` strips any `toolCall` with no matching
|
|
// `toolResult`, so an unpaired resolved block vanishes on reload.
|
|
const h = drive(updateCall(items([["1", "task", 3]]), 1));
|
|
|
|
expect(h.toolResults.map(r => r.toolCallId)).toEqual([todoBlocks(h)[0].id]);
|
|
expect(h.toolResults[0]).toMatchObject({ role: "toolResult", toolName: "todo", isError: false });
|
|
});
|
|
|
|
it("still pairs a result when the snapshot is refused", () => {
|
|
// The call happened and the block is rendered; only local state was left
|
|
// alone. Without a result the block would be stripped on rebuild.
|
|
const h = drive(readCall(items([["1", "task", 2]]), 5));
|
|
|
|
expect(h.snapshots).toEqual([]);
|
|
expect(h.toolResults.map(r => r.toolCallId)).toEqual([todoBlocks(h)[0].id]);
|
|
});
|
|
|
|
it("settles a refused read as a successful no-op under the streamed call id", () => {
|
|
// The card leaves `pendingTools` only on a matching completion, so a
|
|
// refusal that stayed silent would animate forever. Nothing changed
|
|
// locally, so it settles as a success.
|
|
const h = drive(readCall(items([["1", "task", 2]]), 5));
|
|
const callId = todoBlocks(h)[0].id;
|
|
|
|
expect(h.syncCalls).toEqual([{ snapshot: null, toolCallId: callId, error: null }]);
|
|
expect(h.toolResults[0]).toMatchObject({ toolCallId: callId, isError: false });
|
|
});
|
|
|
|
it("settles a server error as a failure carrying the server's text", () => {
|
|
// Collapsing an `UpdateTodosError` into the benign no-op would replay the
|
|
// failure as a success and hide it from the rebuilt transcript.
|
|
const h = drive(errorCall("boom"));
|
|
const callId = todoBlocks(h)[0].id;
|
|
|
|
expect(h.syncCalls).toEqual([{ snapshot: null, toolCallId: callId, error: "boom" }]);
|
|
expect(h.toolResults[0]).toMatchObject({
|
|
toolCallId: callId,
|
|
isError: true,
|
|
content: [{ type: "text", text: "boom" }],
|
|
});
|
|
});
|
|
|
|
it("persists the host's result verbatim instead of its own summary", () => {
|
|
// Only the host knows the phase grouping the todo renderer replays from,
|
|
// so a provider-synthesized summary would replay the list as `0 tasks`.
|
|
const h = drive(updateCall(items([["1", "task", 3]]), 1));
|
|
|
|
expect(h.toolResults[0]).toMatchObject({ content: [{ type: "text", text: "host result" }] });
|
|
});
|
|
|
|
it("falls back to a summary-only result when no host handler is registered", () => {
|
|
// A host with no todo state registers no handler at all; the block still
|
|
// needs a paired result or the rebuild strips it.
|
|
const h = newHarness();
|
|
h.state.onTodoSnapshot = undefined;
|
|
const toolCall = updateCall(items([["1", "task", 3]]), 1);
|
|
for (const kind of ["toolCallStarted", "toolCallCompleted"] as const) {
|
|
processInteractionUpdate(wireUpdate(kind, toolCall) as never, h.output, h.stream, h.state, h.usageState);
|
|
}
|
|
|
|
expect(h.syncCalls).toEqual([]);
|
|
expect(h.toolResults[0]).toMatchObject({
|
|
toolCallId: todoBlocks(h)[0].id,
|
|
isError: false,
|
|
content: [{ type: "text", text: "1/1 tasks completed" }],
|
|
});
|
|
});
|
|
|
|
it("falls back to the server's error text when no host handler is registered", () => {
|
|
const h = newHarness();
|
|
h.state.onTodoSnapshot = undefined;
|
|
const toolCall = errorCall("boom");
|
|
for (const kind of ["toolCallStarted", "toolCallCompleted"] as const) {
|
|
processInteractionUpdate(wireUpdate(kind, toolCall) as never, h.output, h.stream, h.state, h.usageState);
|
|
}
|
|
|
|
expect(h.toolResults[0]).toMatchObject({
|
|
isError: true,
|
|
content: [{ type: "text", text: "boom" }],
|
|
});
|
|
});
|
|
|
|
it("settles the call as a failure when the host sync callback throws", () => {
|
|
// The host callback persists to the session branch and can throw
|
|
// synchronously (e.g. a disk failure). The exception must not skip the
|
|
// paired result and `toolcall_end`: the block is already marked
|
|
// resolved, so left unpaired it is stripped from every rebuilt
|
|
// transcript and the live card never resolves.
|
|
const h = newHarness();
|
|
h.state.onTodoSnapshot = () => {
|
|
throw new Error("session persistence failed");
|
|
};
|
|
const toolCall = updateCall(items([["1", "task", 3]]), 1);
|
|
for (const kind of ["toolCallStarted", "toolCallCompleted"] as const) {
|
|
processInteractionUpdate(wireUpdate(kind, toolCall) as never, h.output, h.stream, h.state, h.usageState);
|
|
}
|
|
|
|
expect(h.state.currentToolCall).toBeNull();
|
|
expect(h.toolResults).toHaveLength(1);
|
|
expect(h.toolResults[0]).toMatchObject({
|
|
toolCallId: todoBlocks(h)[0].id,
|
|
isError: true,
|
|
content: [{ type: "text", text: "session persistence failed" }],
|
|
});
|
|
});
|
|
|
|
it("recognizes an MCP call through the wire-encoded oneof, start and completion", () => {
|
|
// Same wire-shape trap the native todo calls fell into: `ToolCall.tool` is
|
|
// a protobuf oneof, so a decoded message exposes the variant as
|
|
// `{ case, value }` and never as a flat `mcpToolCall` property. Reading
|
|
// the flat one produced `undefined` on every real message while
|
|
// hand-shaped fixtures kept passing. Encode/decode for real here, and
|
|
// cover BOTH reads: the started frame that creates the block and the
|
|
// completed frame that merges the decoded argument map into it.
|
|
const h = newHarness();
|
|
const mcpCall = (args: Record<string, Uint8Array>) =>
|
|
create(ToolCallSchema, {
|
|
tool: {
|
|
case: "mcpToolCall",
|
|
value: create(McpToolCallSchema, {
|
|
args: create(McpArgsSchema, {
|
|
// `name` and `toolName` differ on purpose: the block must be
|
|
// named the same way `decodeMcpCall` names the paired result,
|
|
// which prefers `toolName`.
|
|
name: "fixture_report",
|
|
toolName: "mcp__fixture_report",
|
|
toolCallId: "call-mcp-wire",
|
|
providerIdentifier: "pi-agent",
|
|
args,
|
|
}),
|
|
}),
|
|
},
|
|
});
|
|
processInteractionUpdate(
|
|
wireUpdate("toolCallStarted", mcpCall({})) as never,
|
|
h.output,
|
|
h.stream,
|
|
h.state,
|
|
h.usageState,
|
|
);
|
|
processInteractionUpdate(
|
|
wireUpdate("toolCallCompleted", mcpCall({ query: new TextEncoder().encode('"weather"') })) as never,
|
|
h.output,
|
|
h.stream,
|
|
h.state,
|
|
h.usageState,
|
|
);
|
|
|
|
const blocks = h.output.content.filter((block): block is ToolCallState => block.type === "toolCall");
|
|
expect(blocks).toHaveLength(1);
|
|
expect(blocks[0]).toMatchObject({ id: "call-mcp-wire", name: "mcp__fixture_report" });
|
|
// The completion read is what merges the decoded args onto the block.
|
|
expect(blocks[0].arguments).toMatchObject({ query: "weather" });
|
|
});
|
|
});
|