mirror of
https://github.com/langgenius/dify.git
synced 2026-06-07 16:32:01 +08:00
0bfbd2061e
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com> Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com> Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
84 lines
3.1 KiB
TypeScript
84 lines
3.1 KiB
TypeScript
// service/base is the dependency we're mocking in this test; the
|
|
// no-restricted-imports rule targets production imports, not test
|
|
// instrumentation — mirrors sibling service specs (annotation.spec.ts etc.).
|
|
// eslint-disable-next-line no-restricted-imports
|
|
import { post } from './base'
|
|
import { generateWorkflow } from './debug'
|
|
|
|
// Stub the shared `post` wrapper so tests verify only what `generateWorkflow`
|
|
// composes on top of it — URL, body, and the typed response surface.
|
|
vi.mock('./base', () => ({
|
|
post: vi.fn(),
|
|
get: vi.fn(),
|
|
ssePost: vi.fn(),
|
|
}))
|
|
|
|
describe('debug service — generateWorkflow', () => {
|
|
beforeEach(() => {
|
|
vi.clearAllMocks()
|
|
})
|
|
|
|
// The new endpoint lives at /workflow-generate; the controller mirrors
|
|
// /rule-generate so the body must flow through unchanged.
|
|
it('should POST to /workflow-generate with the body verbatim', () => {
|
|
const body = {
|
|
mode: 'workflow' as const,
|
|
instruction: 'Summarize a URL',
|
|
ideal_output: 'A 3-sentence summary.',
|
|
model_config: { provider: 'openai', name: 'gpt-4o', mode: 'chat', completion_params: {} },
|
|
}
|
|
|
|
generateWorkflow(body)
|
|
|
|
expect(post).toHaveBeenCalledWith('/workflow-generate', { body })
|
|
})
|
|
|
|
// The optional fields must still POST cleanly — `ideal_output` defaulting
|
|
// server-side requires the helper to forward the body as-is, not augment it.
|
|
it('should pass the body through even when ideal_output is omitted', () => {
|
|
const body = {
|
|
mode: 'advanced-chat' as const,
|
|
instruction: 'Friendly support bot',
|
|
model_config: { provider: 'openai', name: 'gpt-4o', mode: 'chat' },
|
|
}
|
|
|
|
generateWorkflow(body)
|
|
|
|
expect(post).toHaveBeenCalledWith('/workflow-generate', { body })
|
|
})
|
|
|
|
// When the caller threads a ``getAbortController`` callback (the modal's
|
|
// pattern for cancelling the in-flight request on close / double-click /
|
|
// 60 s timeout), it must reach ``post()`` as the third argument so the
|
|
// shared fetch wrapper wires it into the AbortController plumbing.
|
|
// Without this the modal cannot abort the request and a close-while-
|
|
// loading leaks the request beyond its UI surface.
|
|
it('should forward getAbortController to post when provided', () => {
|
|
const body = {
|
|
mode: 'workflow' as const,
|
|
instruction: 'Long-running generation',
|
|
model_config: { provider: 'openai', name: 'gpt-4o', mode: 'chat' },
|
|
}
|
|
const getAbortController = vi.fn()
|
|
|
|
generateWorkflow(body, { getAbortController })
|
|
|
|
expect(post).toHaveBeenCalledWith('/workflow-generate', { body }, { getAbortController })
|
|
})
|
|
|
|
// No options → no third argument. Keeps the call site clean and lets the
|
|
// shared wrapper apply its own defaults without a phantom empty object.
|
|
it('should NOT pass a third argument when no options are provided', () => {
|
|
const body = {
|
|
mode: 'workflow' as const,
|
|
instruction: 'Plain call',
|
|
model_config: { provider: 'openai', name: 'gpt-4o', mode: 'chat' },
|
|
}
|
|
|
|
generateWorkflow(body)
|
|
|
|
expect(post).toHaveBeenCalledWith('/workflow-generate', { body })
|
|
expect(vi.mocked(post).mock.calls[0]).toHaveLength(2)
|
|
})
|
|
})
|