mirror of
https://github.com/diegosouzapw/OmniRoute.git
synced 2026-09-15 11:22:15 +03:00
65e81158ab
Landed with the design call resolved per the owner's pick — **option 1**: the synced store is now endpoint-agnostic (persistDiscoveredModels and managedModelImport no longer drop non-chat models at write time), and chat selectability moved to read time (auto-pool expansion in autoStrategy applies filterChatSelectableModels; the models-route projection already had its chatOnly filter). Your discovery test now passes end-to-end (3/3): /api/show capabilities persist per connection and image/embedding requests route through the advertising host. Reconciliation notes: conflicted areas merged onto the current tip (adobe discovery import, requestedModel preflight signature, resolvedProvider fast-path coexists with the synced-route override — explicit resolution wins); carried base-red drains (#10055 memoization, #11071 test variants) dropped as already-landed; the managed-model-import exclusion test was propagated to the new contract (image/video models persist; the read filter still hides them from chat pickers — pinned by a new assertion). Full battery: 205/206 focused (the one red is a confirmed periodic-timer timing flake on the loaded devbox — 20/20 isolated), autoCombo vitest 30/30, combo suites 46/46, gates + typecheck clean. Thank you @yourspraveen — the capability probe + routing design was right; it just needed the store contract opened up. Fixes #11087.
56 lines
2.3 KiB
TypeScript
56 lines
2.3 KiB
TypeScript
import test from "node:test";
|
|
import assert from "node:assert/strict";
|
|
|
|
// Regression guard for the `zai` provider catalog (Anthropic-direct transport).
|
|
//
|
|
// GLM-5.2 shipped in v3.8.26 across the glm/glm-cn/glmt providers (via
|
|
// GLM_SHARED_MODELS) but the curated `zai` provider subset was not updated and
|
|
// still advertised only the GLM-5 line. This test pins the refreshed catalog and,
|
|
// crucially, documents WHY the effort-tier aliases must NOT appear here.
|
|
//
|
|
// The `zai` provider uses the DefaultExecutor, which sends the requested model ID
|
|
// verbatim. The effort tiers `glm-5.2-high` / `glm-5.2-max` are OmniRoute aliases
|
|
// that only the GlmExecutor knows how to resolve (parseGlmEffortTier → base model
|
|
// "glm-5.2" + `effort` field + effort-2025-11-24 beta header). Listing them under
|
|
// `zai` would send unknown model IDs to z.ai's Anthropic endpoint, so they belong
|
|
// to the `glm` provider only.
|
|
|
|
const { getRegistryEntry } = await import("../../open-sse/config/providerRegistry.ts");
|
|
|
|
function modelIds(provider: string): string[] {
|
|
const entry = getRegistryEntry(provider);
|
|
assert.ok(entry, `provider "${provider}" should be registered`);
|
|
return (entry.models ?? []).map((m) => m.id);
|
|
}
|
|
|
|
test("zai provider advertises GLM-5.2 and GLM-4.7 base models", () => {
|
|
const ids = modelIds("zai");
|
|
for (const expected of ["glm-5.2", "glm-4.7", "glm-4.7-flash"]) {
|
|
assert.ok(ids.includes(expected), `zai should advertise ${expected}; got ${ids.join(", ")}`);
|
|
}
|
|
});
|
|
|
|
test("zai provider keeps the existing GLM-5 line", () => {
|
|
const ids = modelIds("zai");
|
|
for (const kept of ["glm-5.1", "glm-5", "glm-5-turbo"]) {
|
|
assert.ok(ids.includes(kept), `zai should still advertise ${kept}`);
|
|
}
|
|
});
|
|
|
|
test("zai provider does NOT advertise effort-tier aliases (DefaultExecutor cannot resolve them)", () => {
|
|
const ids = modelIds("zai");
|
|
for (const alias of ["glm-5.2-high", "glm-5.2-max"]) {
|
|
assert.ok(
|
|
!ids.includes(alias),
|
|
`zai must not list ${alias}: it is a GlmExecutor-only alias and would reach z.ai as an unknown model ID`
|
|
);
|
|
}
|
|
});
|
|
|
|
test("effort-tier aliases remain on the glm provider, where GlmExecutor resolves them", () => {
|
|
const ids = modelIds("glm");
|
|
for (const alias of ["glm-5.2", "glm-5.2-high", "glm-5.2-max"]) {
|
|
assert.ok(ids.includes(alias), `glm provider should expose ${alias}`);
|
|
}
|
|
});
|