Unrestricted Open-source alternative to AI video platforms — Free AI image & video generation studio with 600+ models (Flux, Midjourney, Kling, Sora, Veo). No content filters. Self-hosted, MIT licensed.
JavaScript ai-art-generatorai-image-generationai-video-generationfal-ai-alternativegenerative-aiimage-to-videoinvideo-alternativekling-ailipsynclitellm-alternativemidjourney-alternativemuapiopen-sourceopenrouter-alternativeseedance2sora-alternativetext-to-imagetext-to-videouncensoredwan-video
const test = require('node:test');
const assert = require('node:assert/strict');
const { createMcpClient, normalizeMcpUrl, toProgressEvent } = require('../electron/lib/wan2gpMcp');
// Minimal fake of WanGP's MCP server (Streamable HTTP, SSE-framed replies).
function fakeServer(tools) {
const calls = [];
const reply = (body, extraHeaders = {}) => ({
ok: true,
status: 200,
headers: new Map(Object.entries({ 'content-type': 'text/event-stream', ...extraHeaders })),
text: async () => `event: message\ndata: ${JSON.stringify(body)}\n\n`,
});
const fetchImpl = async (url, init = {}) => {
const u = String(url);
if (u.includes('/wangp_api/gallery/download/')) {
return { ok: true, status: 200, arrayBuffer: async () => new TextEncoder().encode('video-bytes').buffer };
}
if (u.includes('/wangp_api/gallery/upload/')) {
calls.push({ upload: u, bytes: init.body.length });
return { ok: true, status: 200, json: async () => ({ status: 'uploaded', media_id: 'visual:up1' }) };
}
const msg = JSON.parse(init.body);
calls.push(msg);
if (msg.method === 'initialize') return reply({ jsonrpc: '2.0', id: msg.id, result: { serverInfo: { name: 'WanGP', version: '1.30.0' } } }, { 'mcp-session-id': 's1' });
if (msg.method === 'notifications/initialized') return { ok: true, status: 202, headers: new Map(), text: async () => '' };
if (msg.method === 'tools/list') return reply({ jsonrpc: '2.0', id: msg.id, result: { tools: Object.keys(tools).map(name => ({ name })) } });
const handler = tools[msg.params.name];
const out = handler(msg.params.arguments);
const result = out instanceof Error
? { isError: true, content: [{ type: 'text', text: out.message }] }
: { structuredContent: out };
return reply({ jsonrpc: '2.0', id: msg.id, result });
};
return { fetchImpl, calls };
}
test('normalizeMcpUrl appends /mcp once', () => {
assert.equal(normalizeMcpUrl('http://gpu:7866'), 'http://gpu:7866/mcp');
assert.equal(normalizeMcpUrl('http://gpu:7866/mcp/'), 'http://gpu:7866/mcp');
assert.equal(normalizeMcpUrl(''), '');
});
test('toProgressEvent maps WanGP job events onto app progress', () => {
assert.deepEqual(
toProgressEvent({ kind: 'progress', data: { phase: 'inference', status: 'Denoising', progress: 36, current_step: 2, total_steps: 8 } }),
{ status: 'generating', progress: 0.36, step: 2, totalSteps: 8, message: 'Denoising' },
);
assert.equal(toProgressEvent({ kind: 'stream', data: { text: '25%|██' } }), null);
const status = (phase) => toProgressEvent({ kind: 'progress', data: { phase, progress: 10 } }).status;
assert.equal(status('encoding_text'), 'encoding');
assert.equal(status('inference_stage_2'), 'generating');
assert.equal(status('loading_model'), 'loading');
assert.equal(status('downloading_output'), 'saving');
assert.equal(status('something_new'), 'working');
});
test('probe initializes a session and requires WanGP generation tools', async () => {
const { fetchImpl, calls } = fakeServer({ wangp_generate: () => ({}), wangp_get_job: () => ({}) });
const client = createMcpClient('http://gpu:7866', { fetchImpl });
assert.deepEqual(await client.probe(), { server: 'WanGP', version: '1.30.0' });
assert.deepEqual(calls.map(c => c.method), ['initialize', 'notifications/initialized', 'tools/list']);
const other = fakeServer({ some_tool: () => ({}) });
await assert.rejects(createMcpClient('http://x', { fetchImpl: other.fetchImpl }).probe(), /does not expose WanGP generation tools/);
});
test('modelAvailability distinguishes downloaded, missing and unknown models', async () => {
const { fetchImpl } = fakeServer({
wangp_get_model_availability: ({ model_type }) => model_type === 'nope'
? new Error('Error executing tool wangp_get_model_availability: Unknown model_type: nope')
: { model_type, status: model_type === 't2v_1.3B' ? 'available' : 'missing' },
});
const client = createMcpClient('http://gpu:7866', { fetchImpl });
assert.equal(await client.modelAvailability('t2v_1.3B'), 'available');
assert.equal(await client.modelAvailability('t2v_2_2'), 'missing');
assert.equal(await client.modelAvailability('nope'), 'unknown-model');
});
test('generate submits, polls with progress, and download fetches the gallery item', { timeout: 20000 }, async () => {
let polls = 0;
const { fetchImpl, calls } = fakeServer({
wangp_generate: () => ({ job_id: 'job1', done: false, events: [] }),
wangp_get_job: () => (++polls < 2
? { job_id: 'job1', done: false, events: [{ kind: 'progress', data: { phase: 'inference', status: 'Denoising', progress: 50, current_step: 4, total_steps: 8 } }] }
: { job_id: 'job1', done: true, result: { success: true, gallery_items: [{ media_id: 'visual:abc', media_type: 'video', filename: 'out.mp4' }] } }),
wangp_create_gallery_download: () => ({ download_url: '/wangp_api/gallery/download/tok', filename: 'out.mp4' }),
});
const client = createMcpClient('http://gpu:7866/mcp', { fetchImpl });
const progress = [];
let jobId = null;
const item = await client.generate({ model_type: 't2v_1.3B', prompt: 'x' }, { onProgress: p => progress.push(p), onJobId: id => { jobId = id; } });
assert.equal(jobId, 'job1');
assert.equal(item.media_id, 'visual:abc');
assert.deepEqual(progress.map(p => [p.status, p.step]), [['generating', 4]]);
const submit = calls.find(c => c.params?.name === 'wangp_generate');
assert.deepEqual(submit.params.arguments, { source: { model_type: 't2v_1.3B', prompt: 'x' }, wait: false, event_limit: 0 });
const file = await client.download(item.media_id);
assert.equal(file.filename, 'out.mp4');
assert.equal(file.bytes.toString(), 'video-bytes');
});
test('generate surfaces job errors', { timeout: 20000 }, async () => {
const { fetchImpl } = fakeServer({
wangp_generate: () => ({ job_id: 'job2', done: true, result: { success: false, errors: [{ message: 'CUDA out of memory' }] } }),
});
await assert.rejects(createMcpClient('http://gpu:7866', { fetchImpl }).generate({}), /generation failed:\nCUDA out of memory/);
});
test('uploadImage PUTs bytes to the one-time URL and returns the media id', async () => {
const { fetchImpl, calls } = fakeServer({
wangp_create_gallery_upload: ({ filename }) => ({ upload_url: '/wangp_api/gallery/upload/tok', filename }),
});
const mediaId = await createMcpClient('http://gpu:7866', { fetchImpl }).uploadImage({ name: 'start.png', bytes: Buffer.from('png'), mime: 'image/png' });
assert.equal(mediaId, 'visual:up1');
assert.deepEqual(calls.at(-1), { upload: 'http://gpu:7866/wangp_api/gallery/upload/tok', bytes: 3 });
});