Repository navigation
fix(write-to-file): clean partial state on missing-param and rooignore denial + integrate the #1066 split series (6/6 + FINAL) #1928
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Open
easonLiangWorldedtech
wants to merge
23
commits into
Zoo-Code-Org:main
Choose a base branch
from
easonLiangWorldedtech:p1066/u7-early-return-denial-cleanup
base: main
Could not load branches
Branch not found: {{ refName }}
Loading
Could not load tags
Nothing to show
Loading
Are you sure you want to change the base?
Some commits from the old base branch may be removed from the timeline,
and old review comments may become outdated.
Open
Changes from all commits
Commits
Show all changes
23 commits
Select commit
Hold shift + click to select a range
f4689a0
fix(core): early-return denial cleanup, rebased onto main as one commit
easonLiangWorldedtech 9155962
fix(tests): write the specs against the surface main actually declares
easonLiangWorldedtech 9ebacb5
test(core): move the 13 teardown specs out of Task.spec.ts
easonLiangWorldedtech a8d6e3b
test(core): gate the metadata retry and isolate the abort-path release
easonLiangWorldedtech 8d4d11d
fix(core): record a failed diff release and surface it at the abort
easonLiangWorldedtech c382cd9
fix(core): gate the late diff cleanup on what open published
easonLiangWorldedtech af8b816
fix(core): close the diff editor a late open() publishes after release
easonLiangWorldedtech feeaf07
fix(core): keep a placeholder whose content is not the one open() wrote
easonLiangWorldedtech 9b3ac07
test(core): cover the placeholder read failure that does not say the …
easonLiangWorldedtech 054af49
fix(core): check the placeholder before the discard saves into it
easonLiangWorldedtech f65ab53
test(core): pin the release against a diff command that resolves late
easonLiangWorldedtech 0533879
fix(editor): decline the placeholder save when the guard read proved …
easonLiangWorldedtech 08d66ef
fix(editor): re-check the placeholder immediately before unlinking it
easonLiangWorldedtech ae6d8c5
fix(editor): close the plain text tab a released session leaves behind
easonLiangWorldedtech ebad9d0
test(editor): pin disposal of the two temporary open listeners
easonLiangWorldedtech 62c0f10
test(editor): cover the dirty create whose placeholder still matches
easonLiangWorldedtech f287e55
test(tools): bind the repeated-release row to the provider's real cle…
easonLiangWorldedtech ec6d02a
test(editor): name every step of the released-session stand-down
easonLiangWorldedtech 935d773
test(editor): cover a guard read that rejects with no error object
easonLiangWorldedtech 4e2d548
test(editor): cover the endings of the re-read and unlink that had no…
easonLiangWorldedtech 427cd72
test(core): answer the mutation rows the digest printed at these coor…
easonLiangWorldedtech 045dccc
Merge branch 'main' into p1066/u7-early-return-denial-cleanup
easonLiangWorldedtech 0c330a8
Fix the partial-path mutation directive this branch added in WriteToF…
File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
292 changes: 292 additions & 0 deletions
292
src/core/assistant-message/__tests__/presentAssistantMessage-missing-native-args.spec.ts
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,292 @@ | ||
| // npx vitest src/core/assistant-message/__tests__/presentAssistantMessage-missing-native-args.spec.ts | ||
|
|
||
| import { describe, it, expect, beforeEach, vi } from "vitest" | ||
| import { presentAssistantMessage } from "../presentAssistantMessage" | ||
| import { DispatchState } from "../../task/dispatchState" | ||
| import { type RequestPolicySnapshot } from "../../prompts/tools/effective-tool-policy" | ||
| import { isValidToolName, validateToolUse } from "../../tools/validateToolUse" | ||
|
|
||
| // Tool policy is frozen at request build and the presenter refuses to run without the | ||
| // snapshot; these tests only exercise the presenter's bypass exits, so the empty policy is | ||
| // enough (validateToolUse itself is mocked). | ||
| const baseSnapshot: RequestPolicySnapshot = { | ||
| disabledTools: [], | ||
| customModes: [], | ||
| } | ||
|
|
||
| const mockTeardown = vi.hoisted(() => vi.fn()) | ||
| const mockWriteHandle = vi.hoisted(() => vi.fn()) | ||
|
|
||
| vi.mock("../../task/Task") | ||
| vi.mock("../../tools/validateToolUse", () => ({ | ||
| validateToolUse: vi.fn(), | ||
| isValidToolName: vi.fn(() => true), | ||
| })) | ||
| vi.mock("../../tools/WriteToFileTool", () => ({ | ||
| writeToFileTool: { handle: mockWriteHandle, teardownAbandonedStream: mockTeardown }, | ||
| })) | ||
| vi.mock("@roo-code/telemetry", () => ({ | ||
| TelemetryService: { | ||
| instance: { | ||
| captureToolUsage: vi.fn(), | ||
| captureConsecutiveMistakeError: vi.fn(), | ||
| captureException: vi.fn(), | ||
| }, | ||
| }, | ||
| })) | ||
|
|
||
| type ToolResultBlock = { type: string; tool_use_id: string; content: string; is_error: boolean } | ||
|
|
||
| describe("presentAssistantMessage - finalized block without nativeArgs", () => { | ||
| // The presenter reads only a subset of Task. A Record keeps the double honest without | ||
| // an `any`; the single cast at the call site is the documented boundary. | ||
| let mockTask: Record<string, unknown> | ||
| let userMessageContent: ToolResultBlock[] | ||
|
|
||
| beforeEach(() => { | ||
| mockTeardown.mockReset() | ||
| mockWriteHandle.mockReset() | ||
| userMessageContent = [] | ||
| mockTask = { | ||
| taskId: "test-task-id", | ||
| instanceId: "test-instance", | ||
| abort: false, | ||
| // main renamed the dispatcher fields (src/core/task/dispatchState.ts): the presenter reads | ||
| // dispatchState and pendingWork, not the old lock/flag pair. | ||
| dispatchState: DispatchState.IDLE, | ||
| pendingWork: null, | ||
| currentStreamingContentIndex: 0, | ||
| assistantMessageContent: [ | ||
| { | ||
| type: "tool_use", | ||
| name: "write_to_file", | ||
| params: {}, | ||
| partial: false, | ||
| // Streaming JSON never parsed: Task completes the block with no nativeArgs. | ||
| id: "toolu_1", | ||
| nativeArgs: undefined, | ||
| }, | ||
| ], | ||
| userMessageContent, | ||
| didCompleteReadingStream: false, | ||
| // The presenter keeps per-tool-call context in this map (added on main since this | ||
| // spec was written); the bypass exits below run after its first use. | ||
| activeToolContexts: new Map(), | ||
| didRejectTool: false, | ||
| didAlreadyUseTool: false, | ||
| consecutiveMistakeCount: 0, | ||
| consecutiveMistakeLimit: 3, | ||
| apiConfiguration: { apiProvider: "test-provider" }, | ||
| clineMessages: [], | ||
| getTaskMode: vi.fn().mockResolvedValue("code"), | ||
| api: { getModel: () => ({ id: "test-model", info: {} }) }, | ||
| recordToolUsage: vi.fn(), | ||
| recordToolError: vi.fn(), | ||
| toolRepetitionDetector: { check: vi.fn().mockReturnValue({ allowExecution: true }) }, | ||
| providerRef: { | ||
| deref: () => ({ getState: vi.fn().mockResolvedValue({ mode: "code", customModes: [] }) }), | ||
| }, | ||
| say: vi.fn().mockResolvedValue(undefined), | ||
| ask: vi.fn().mockResolvedValue({ response: "yesButtonClicked" }), | ||
| pushToolResultToUserContent: vi.fn((toolResult: ToolResultBlock) => { | ||
| userMessageContent.push(toolResult) | ||
| return true | ||
| }), | ||
| } | ||
| }) | ||
|
|
||
| it("tears the write_to_file stream state down instead of dispatching the tool", async () => { | ||
| await presentAssistantMessage( | ||
| mockTask as unknown as Parameters<typeof presentAssistantMessage>[0], | ||
| baseSnapshot, | ||
| ) | ||
|
|
||
| // The malformed call must not be executed, and exactly one error tool_result is | ||
| // emitted for the provider. | ||
| expect(isValidToolName).toHaveBeenCalled() | ||
| expect(mockWriteHandle).not.toHaveBeenCalled() | ||
| expect(mockTask.pushToolResultToUserContent).toHaveBeenCalledTimes(1) | ||
| expect(userMessageContent).toEqual([ | ||
| expect.objectContaining({ | ||
| type: "tool_result", | ||
| tool_use_id: "toolu_1", | ||
| is_error: true, | ||
| content: expect.stringContaining("missing nativeArgs"), | ||
| }), | ||
| ]) | ||
| // The guard bypasses handle(), so the per-task stream state has to be released here: | ||
| // otherwise the entry, its TaskAborted listener and any streamed diff view leak into | ||
| // the next API request of the same task. | ||
| expect(mockTeardown).toHaveBeenCalledTimes(1) | ||
| expect(mockTeardown).toHaveBeenCalledWith(mockTask) | ||
| }) | ||
|
|
||
| it("tears the stream state down when tool validation rejects the call", async () => { | ||
| // A mode file restriction (or any validateToolUse failure) lands here. The block | ||
| // streamed partial deltas - handlePartial ran - and this exit bypasses handle(), so | ||
| // without the teardown the per-task stream state, its TaskAborted listener and any | ||
| // streamed diff view leak into the next API request. | ||
| mockTask.assistantMessageContent = [ | ||
| { | ||
| type: "tool_use", | ||
| name: "write_to_file", | ||
| params: { path: "restricted.ts", content: "partial model output" }, | ||
| partial: false, | ||
| id: "toolu_2", | ||
| nativeArgs: { path: "restricted.ts", content: "partial model output" }, | ||
| }, | ||
| ] | ||
| vi.mocked(validateToolUse).mockImplementationOnce(() => { | ||
| throw new Error("write_to_file is not allowed to write restricted.ts in this mode") | ||
| }) | ||
|
|
||
| await presentAssistantMessage( | ||
| mockTask as unknown as Parameters<typeof presentAssistantMessage>[0], | ||
| baseSnapshot, | ||
| ) | ||
|
|
||
| expect(mockWriteHandle).not.toHaveBeenCalled() | ||
| expect(userMessageContent).toEqual([ | ||
| expect.objectContaining({ | ||
| type: "tool_result", | ||
| tool_use_id: "toolu_2", | ||
| is_error: true, | ||
| content: expect.stringContaining("restricted.ts"), | ||
| }), | ||
| ]) | ||
| expect(mockTeardown).toHaveBeenCalledTimes(1) | ||
| expect(mockTeardown).toHaveBeenCalledWith(mockTask) | ||
| }) | ||
|
|
||
| it("tears the stream state down when the tool repetition limit stops the call", async () => { | ||
| mockTask.assistantMessageContent = [ | ||
| { | ||
| type: "tool_use", | ||
| name: "write_to_file", | ||
| params: { path: "same.ts", content: "same content" }, | ||
| partial: false, | ||
| id: "toolu_3", | ||
| nativeArgs: { path: "same.ts", content: "same content" }, | ||
| }, | ||
| ] | ||
| // The detector's result is a tagged union now (allow / soft_block / hard_block): only the | ||
| // hard block asks the user and stops the call, so that is the exit under test. | ||
| vi.mocked(mockTask.toolRepetitionDetector as unknown as { check: unknown }).check = vi.fn().mockReturnValue({ | ||
| action: "hard_block", | ||
| askUser: { messageKey: "tool_repetition", messageDetail: "write_to_file" }, | ||
| }) | ||
|
|
||
| await presentAssistantMessage( | ||
| mockTask as unknown as Parameters<typeof presentAssistantMessage>[0], | ||
| baseSnapshot, | ||
| ) | ||
|
|
||
| expect(mockWriteHandle).not.toHaveBeenCalled() | ||
| // The repetition exit reports through the local pushToolResult, which wraps the | ||
| // message in the tool-error envelope rather than the provider is_error flag. | ||
| expect(userMessageContent).toEqual([ | ||
| expect.objectContaining({ | ||
| type: "tool_result", | ||
| tool_use_id: "toolu_3", | ||
| content: expect.stringContaining("repetition limit reached"), | ||
| }), | ||
| ]) | ||
| expect(mockTeardown).toHaveBeenCalledTimes(1) | ||
| expect(mockTeardown).toHaveBeenCalledWith(mockTask) | ||
| }) | ||
|
|
||
| it("leaves the stream state alone at the missing-nativeArgs exit for a tool that opened no diff view", async () => { | ||
| // The release at this bypass is scoped to the one tool whose streamed diff view a skipped | ||
| // handle() would strand. Applying it to every block name would ask other tools to release | ||
| // state they never opened, so the guard has to read the name rather than the shape alone. | ||
| mockTask.assistantMessageContent = [ | ||
| { | ||
| type: "tool_use", | ||
| name: "execute_command", | ||
| params: { cmd: "ls" }, | ||
| partial: false, | ||
| id: "toolu_4", | ||
| nativeArgs: undefined, | ||
| }, | ||
| ] | ||
|
|
||
| await presentAssistantMessage( | ||
| mockTask as unknown as Parameters<typeof presentAssistantMessage>[0], | ||
| baseSnapshot, | ||
| ) | ||
|
|
||
| expect(mockWriteHandle).not.toHaveBeenCalled() | ||
| expect(userMessageContent).toEqual([ | ||
| expect.objectContaining({ type: "tool_result", tool_use_id: "toolu_4", is_error: true }), | ||
| ]) | ||
| expect(mockTeardown).not.toHaveBeenCalled() | ||
| }) | ||
|
|
||
| it("leaves the stream state alone at the validation exit for a tool that opened no diff view", async () => { | ||
| // A refusal lands here whatever the tool. Only write_to_file has a streamed diff view at | ||
| // this point, so the release stays scoped to it. | ||
| mockTask.assistantMessageContent = [ | ||
| { | ||
| type: "tool_use", | ||
| name: "execute_command", | ||
| params: { cmd: "ls" }, | ||
| partial: false, | ||
| id: "toolu_5", | ||
| nativeArgs: { cmd: "ls" }, | ||
| }, | ||
| ] | ||
| vi.mocked(validateToolUse).mockImplementationOnce(() => { | ||
| throw new Error("execute_command is not allowed in this mode") | ||
| }) | ||
|
|
||
| await presentAssistantMessage( | ||
| mockTask as unknown as Parameters<typeof presentAssistantMessage>[0], | ||
| baseSnapshot, | ||
| ) | ||
|
|
||
| expect(mockWriteHandle).not.toHaveBeenCalled() | ||
| expect(userMessageContent).toEqual([ | ||
| expect.objectContaining({ | ||
| type: "tool_result", | ||
| tool_use_id: "toolu_5", | ||
| is_error: true, | ||
| content: expect.stringContaining("not allowed"), | ||
| }), | ||
| ]) | ||
| expect(mockTeardown).not.toHaveBeenCalled() | ||
| }) | ||
|
|
||
| it("leaves the stream state alone at the repetition exit for a tool that opened no diff view", async () => { | ||
| // The repetition stop bypasses handle() for any tool, and the release is still only owed by | ||
| // the one that streams a diff view. | ||
| mockTask.assistantMessageContent = [ | ||
| { | ||
| type: "tool_use", | ||
| name: "execute_command", | ||
| params: { cmd: "ls" }, | ||
| partial: false, | ||
| id: "toolu_6", | ||
| nativeArgs: { cmd: "ls" }, | ||
| }, | ||
| ] | ||
| vi.mocked(mockTask.toolRepetitionDetector as unknown as { check: unknown }).check = vi.fn().mockReturnValue({ | ||
| action: "hard_block", | ||
| askUser: { messageKey: "tool_repetition", messageDetail: "execute_command" }, | ||
| }) | ||
|
|
||
| await presentAssistantMessage( | ||
| mockTask as unknown as Parameters<typeof presentAssistantMessage>[0], | ||
| baseSnapshot, | ||
| ) | ||
|
|
||
| expect(mockWriteHandle).not.toHaveBeenCalled() | ||
| expect(userMessageContent).toEqual([ | ||
| expect.objectContaining({ | ||
| type: "tool_result", | ||
| tool_use_id: "toolu_6", | ||
| content: expect.stringContaining("repetition limit reached"), | ||
| }), | ||
| ]) | ||
| expect(mockTeardown).not.toHaveBeenCalled() | ||
| }) | ||
| }) |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Uh oh!
There was an error while loading. Please reload this page.