mirror of
https://github.com/diegosouzapw/OmniRoute.git
synced 2026-09-11 17:32:35 +03:00
* test(infra): retry recursive temp-dir removal instead of failing a shard on ENOTEMPTY (#11966) Two shards on release/v3.8.51 went red in one day with the same signature — "ENOTEMPTY, Directory not empty: /tmp/omniroute-<test>-XXXXXX" — from combo-same-provider-cascade (Unit Tests fast-path 4/4, on a PR that touches only .github/) and auth-policy-embeddings-webfetch-7785 (the 20k-test TIA step). Both pass alone and on re-run: the cleanup races something still writing into the directory (SQLite WAL/-shm checkpoint, a worker, the backup) and under a loaded hosted runner the window opens. 1154 test files do their own cleanup with fs.rmSync(dir, { recursive: true, force: true }); 57 already asked for retries. One-shot codemod (scripts/ad-hoc/codemod-rm-maxretries.mjs, kept for the record): every rm / rmSync / rmdirSync option object with `recursive: true` and no `maxRetries` gains `maxRetries: 5, retryDelay: 100` — Node itself then retries ENOTEMPTY/EBUSY/EPERM for up to ~0.5 s before giving up. 2243 call sites in 1292 files under tests/, the shared tests/_setup/isolateDataDir.ts exit hook included. Only the option object changes: no call site, assertion or import is touched. Validation: prettier and ESLint (with the frozen suppressions) clean on all 1292 files; a random 20-file sample runs green (quota-redis-store hangs identically on the untouched tree — it needs a Redis on localhost, an environment matter). The four unit shards on this PR are the full run. * fix(quality): let check-forgotten-sibling-tests read a 1,000-file diff The gate shells out to `git diff` through execFileSync with Node's default 1 MB maxBuffer; the 1,292-file codemod in this PR is the first diff large enough to overflow it, and the gate died with `spawnSync git ENOBUFS` before comparing anything. 64 MB is far above any real PR and costs nothing when unused.
184 lines
6.8 KiB
TypeScript
184 lines
6.8 KiB
TypeScript
import test from "node:test";
|
|
import assert from "node:assert/strict";
|
|
import fs from "node:fs";
|
|
import os from "node:os";
|
|
import path from "node:path";
|
|
|
|
// Split guard for the #3501 god-file decomposition (PR 2): the target-resolution
|
|
// stage of handleComboChat (wildcard expansion → weighted step groups → strategy
|
|
// ordering → stickiness/eval/compat/context filters → task-aware reorder →
|
|
// prompt-cache affinity → pre-screen) was extracted verbatim into
|
|
// resolveComboTargetPipeline. These tests pin the leaf's own contract: the shape it
|
|
// hands back to the attempt loop and pass-through ordering for the plain `priority`
|
|
// path. The strategy-specific branches stay covered end-to-end by the combo-*
|
|
// consumer suites through combo.ts.
|
|
|
|
const TEST_DATA_DIR = fs.mkdtempSync(path.join(os.tmpdir(), "omniroute-combo-target-resolution-"));
|
|
const ORIGINAL_DATA_DIR = process.env.DATA_DIR;
|
|
process.env.DATA_DIR = TEST_DATA_DIR;
|
|
|
|
const core = await import("../../src/lib/db/core.ts");
|
|
const { saveModelsDevCapabilities, clearModelsDevCapabilities } =
|
|
await import("../../src/lib/modelsDevSync.ts");
|
|
const { resolveComboTargetPipeline } =
|
|
await import("../../open-sse/services/combo/targetResolution.ts");
|
|
|
|
test.after(() => {
|
|
core.resetDbInstance();
|
|
if (ORIGINAL_DATA_DIR === undefined) {
|
|
delete process.env.DATA_DIR;
|
|
} else {
|
|
process.env.DATA_DIR = ORIGINAL_DATA_DIR;
|
|
}
|
|
fs.rmSync(TEST_DATA_DIR, { recursive: true, force: true, maxRetries: 5, retryDelay: 100 });
|
|
});
|
|
|
|
test.beforeEach(() => {
|
|
clearModelsDevCapabilities();
|
|
});
|
|
|
|
const noopLog = { info() {}, warn() {}, error() {}, debug() {} } as never;
|
|
|
|
function capabilityEntry(limitContext: number) {
|
|
return {
|
|
tool_call: true,
|
|
reasoning: false,
|
|
attachment: false,
|
|
structured_output: true,
|
|
temperature: true,
|
|
modalities_input: JSON.stringify(["text"]),
|
|
modalities_output: JSON.stringify(["text"]),
|
|
knowledge_cutoff: null,
|
|
release_date: null,
|
|
last_updated: null,
|
|
status: null,
|
|
family: null,
|
|
open_weights: false,
|
|
limit_context: limitContext,
|
|
limit_input: limitContext,
|
|
limit_output: 4096,
|
|
interleaved_field: null,
|
|
};
|
|
}
|
|
|
|
const deps = (overrides: Record<string, unknown> = {}): never =>
|
|
({
|
|
body: { messages: [{ role: "user", content: "hi" }] },
|
|
combo: { id: "c1", name: "c1", models: ["openai/gpt-4o", "anthropic/claude-3"], config: {} },
|
|
strategy: "priority",
|
|
config: {},
|
|
settings: null,
|
|
allCombos: null,
|
|
relayOptions: null,
|
|
signal: null,
|
|
apiKeyAllowedConnections: null,
|
|
log: noopLog,
|
|
resilienceSettings: { providerCooldown: { enabled: false } },
|
|
isModelAvailable: undefined,
|
|
handleSingleModelWithTimeout: async () => new Response("{}"),
|
|
buildAutoCandidates: async () => [],
|
|
...overrides,
|
|
}) as never;
|
|
|
|
test("exports resolveComboTargetPipeline", () => {
|
|
assert.equal(typeof resolveComboTargetPipeline, "function");
|
|
});
|
|
|
|
test("priority strategy resolves combo models into orderedTargets in declared order", async () => {
|
|
const result = await resolveComboTargetPipeline(deps());
|
|
assert.ok(!("earlyResponse" in result), "expected a resolved pipeline, not an early response");
|
|
if ("earlyResponse" in result) return;
|
|
assert.deepEqual(
|
|
result.orderedTargets.map((t) => t.modelStr),
|
|
["openai/gpt-4o", "anthropic/claude-3"]
|
|
);
|
|
});
|
|
|
|
test("returns the derived values the attempt loop consumes", async () => {
|
|
const result = await resolveComboTargetPipeline(deps());
|
|
assert.ok(!("earlyResponse" in result));
|
|
if ("earlyResponse" in result) return;
|
|
assert.equal(typeof result.stickyWeightedLimit, "number");
|
|
assert.equal(typeof result.getWeightedStepKeyForTarget, "function");
|
|
assert.ok(result.preScreenMap instanceof Map);
|
|
assert.equal(result.sticky.messageHash === null || typeof result.sticky.messageHash, "string");
|
|
// Non-weighted strategies have no weighted step resolution, so the mapper is a
|
|
// constant null — the sticky-weighted write-back in combo.ts is then skipped.
|
|
assert.equal(result.getWeightedStepKeyForTarget(result.orderedTargets[0]), null);
|
|
});
|
|
|
|
test("an empty combo yields an empty target pool (combo.ts turns it into a 404)", async () => {
|
|
const result = await resolveComboTargetPipeline(deps({ combo: { name: "empty", models: [] } }));
|
|
assert.ok(!("earlyResponse" in result));
|
|
if ("earlyResponse" in result) return;
|
|
assert.deepEqual(result.orderedTargets, []);
|
|
});
|
|
|
|
test("request exceeding every approximate context hint keeps the target pool", async () => {
|
|
saveModelsDevCapabilities({
|
|
"unit-target-resolution": {
|
|
tiny: capabilityEntry(8_000),
|
|
small: capabilityEntry(16_000),
|
|
},
|
|
});
|
|
|
|
const result = await resolveComboTargetPipeline(
|
|
deps({
|
|
combo: {
|
|
id: "c2",
|
|
name: "known-context-overflow",
|
|
models: ["unit-target-resolution/tiny", "unit-target-resolution/small"],
|
|
config: {},
|
|
},
|
|
body: { messages: [{ role: "user", content: "word ".repeat(200_000) }] },
|
|
})
|
|
);
|
|
|
|
assert.ok(!("earlyResponse" in result), "approximate context hints must not reject the pool");
|
|
if ("earlyResponse" in result) return;
|
|
assert.deepEqual(
|
|
result.orderedTargets.map((target) => target.modelStr),
|
|
["unit-target-resolution/tiny", "unit-target-resolution/small"]
|
|
);
|
|
});
|
|
|
|
// #8790: maxContextWindow rejects every target whose known context window
|
|
// exceeds the configured ceiling. When that empties the pool, the
|
|
// context-requirements guard (applyContinuityFilters → #8786's
|
|
// buildEmptyComboTargetsPayload) must surface a 404 context_requirements_exhausted
|
|
// early response instead of letting an empty orderedTargets[] fall through to the
|
|
// attempt loop.
|
|
test("maxContextWindow rejecting every target returns a 404 context_requirements_exhausted earlyResponse", async () => {
|
|
saveModelsDevCapabilities({
|
|
"unit-target-resolution-max": {
|
|
big1: capabilityEntry(500_000),
|
|
big2: capabilityEntry(1_000_000),
|
|
},
|
|
});
|
|
|
|
const result = await resolveComboTargetPipeline(
|
|
deps({
|
|
combo: {
|
|
id: "c3",
|
|
name: "max-context-window-exhausted",
|
|
models: ["unit-target-resolution-max/big1", "unit-target-resolution-max/big2"],
|
|
config: {},
|
|
},
|
|
config: {
|
|
contextRequirements: { maxContextWindow: 128_000, contextFilterMode: "strict" },
|
|
},
|
|
})
|
|
);
|
|
|
|
assert.ok("earlyResponse" in result, "expected a context-requirements-exhausted early response");
|
|
if (!("earlyResponse" in result)) return;
|
|
assert.equal(result.earlyResponse.status, 404);
|
|
const body = (await result.earlyResponse.json()) as {
|
|
error?: { code?: string };
|
|
diagnostics?: { terminalReason?: string; excluded?: unknown[] };
|
|
};
|
|
assert.equal(body.error?.code, "model_not_found");
|
|
assert.equal(body.diagnostics?.terminalReason, "context_requirements_exhausted");
|
|
assert.equal(body.diagnostics?.excluded?.length, 2);
|
|
});
|