minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

[CI][RFC] Replace black-jupyter with ruff-format (#37210)

Browse Source 
Co-authored-by: Alison Shao <a.shao@wustl.edu>
This commit is contained in:
Alex Nails
2026-09-02 19:46:08 -07:00
committed by GitHub
co-authored by Alison Shao
parent 2641e427be
commit 28262c20df
1411 changed files with 7766 additions and 8176 deletions
Download Patch File Download Diff File Expand all files Collapse all files
experimental/sgl-router/tests/e2e/chat_completions/test_two_router_convergence.py
+12 -12
View File
@@ -194,9 +194,9 @@ def _direct_warm(worker_url: str, model_id: str, prefix: str) -> None:
},
timeout=60.0,
)
assert (
r.status_code == 200
), f"direct warm to {worker_url} failed: HTTP {r.status_code} {r.text!r}"
assert r.status_code == 200, (
f"direct warm to {worker_url} failed: HTTP {r.status_code} {r.text!r}"
)
def _route_through(router_url: str, model_id: str, prompt: str) -> str:
@@ -212,9 +212,9 @@ def _route_through(router_url: str, model_id: str, prompt: str) -> str:
after = _success_counts_by_worker(router_url)
deltas = {w: after.get(w, 0) - before.get(w, 0) for w in set(after) | set(before)}
winners = [w for w, d in deltas.items() if d > 0]
assert (
len(winners) == 1
), f"expected exactly one worker delta on {router_url}, got {deltas}"
assert len(winners) == 1, (
f"expected exactly one worker delta on {router_url}, got {deltas}"
)
return winners[0]
@@ -309,15 +309,15 @@ def test_routers_route_by_prefix_content(
landed = _route_through(
router.base_url, spec["model"], PREFIX_X
)
assert (
landed == worker_x.url
), f"router {label}: PREFIX_X must route to {worker_x.url}; landed on {landed}"
assert landed == worker_x.url, (
f"router {label}: PREFIX_X must route to {worker_x.url}; landed on {landed}"
)
landed = _route_through(
router.base_url, spec["model"], PREFIX_Y
)
assert (
landed == worker_y.url
), f"router {label}: PREFIX_Y must route to {worker_y.url}; landed on {landed}"
assert landed == worker_y.url, (
f"router {label}: PREFIX_Y must route to {worker_y.url}; landed on {landed}"
)
except Exception:
_dump_logs(logs)
raise
experimental/sgl-router/tests/e2e/chat_completions/test_validation.py
+6 -6
View File
@@ -50,9 +50,9 @@ def test_chat_non_streaming_returns_assistant_message(
body = resp.json()
choice = body["choices"][0]
assert choice["message"]["role"] == "assistant"
assert choice["message"][
"content"
], f"empty assistant content: {choice!r}"
assert choice["message"]["content"], (
f"empty assistant content: {choice!r}"
)
assert choice.get("finish_reason"), choice
finally:
gpu_allocator.release(gpu)
@@ -91,8 +91,8 @@ def test_chat_streaming_emits_sse_chunks_with_done(
if line.startswith("data:"):
chunks.append(line.strip())
assert len(chunks) >= 2, f"expected >=2 SSE chunks, got: {chunks}"
assert any(
"[DONE]" in c for c in chunks
), f"no [DONE] terminator in stream: {chunks}"
assert any("[DONE]" in c for c in chunks), (
f"no [DONE] terminator in stream: {chunks}"
)
finally:
gpu_allocator.release(gpu)
experimental/sgl-router/tests/e2e/k8s_integration/test_discovery.py
+11 -9
View File
@@ -66,15 +66,17 @@ def test_router_discovers_multiple_workers(router_url):
# Scale down to 1 — router should still route after reconverging
_scale_fake_worker(1)
_poll_until(
lambda: httpx.post(
f"{router_url}/v1/chat/completions",
json={
"model": "tiny",
"messages": [{"role": "user", "content": "post-scale-down"}],
},
timeout=10.0,
).status_code
== 200,
lambda: (
httpx.post(
f"{router_url}/v1/chat/completions",
json={
"model": "tiny",
"messages": [{"role": "user", "content": "post-scale-down"}],
},
timeout=10.0,
).status_code
== 200
),
"router routes after scale-down to 1",
timeout=60,
interval=3,
experimental/sgl-router/tests/e2e/test_chat_smoke.py
+6 -6
View File
@@ -16,9 +16,9 @@ def test_models(router: str) -> None:
assert resp.status_code == 200, resp.text
data = resp.json()
ids = [m["id"] for m in data.get("data", [])]
assert any(
MODEL in mid for mid in ids
), f"Model {MODEL!r} not found in /v1/models response: {ids}"
assert any(MODEL in mid for mid in ids), (
f"Model {MODEL!r} not found in /v1/models response: {ids}"
)
def test_chat_non_streaming(router: str) -> None:
@@ -59,6 +59,6 @@ def test_chat_streaming(router: str) -> None:
chunks.append(line)
assert len(chunks) >= 2, f"Expected >=2 SSE chunks, got {len(chunks)}: {chunks}"
assert any(
"[DONE]" in c for c in chunks
), f"No [DONE] chunk found in SSE stream: {chunks}"
assert any("[DONE]" in c for c in chunks), (
f"No [DONE] chunk found in SSE stream: {chunks}"
)
experimental/sgl-router/tests/e2e/test_tokenize_smoke.py
+6 -6
View File
@@ -20,9 +20,9 @@ def test_tokenize_round_trip(router: str) -> None:
)
assert tok_resp.status_code == 200, tok_resp.text
tokens = tok_resp.json()["tokens"]
assert (
isinstance(tokens, list) and len(tokens) > 0
), f"Expected non-empty token list, got: {tokens}"
assert isinstance(tokens, list) and len(tokens) > 0, (
f"Expected non-empty token list, got: {tokens}"
)
# Detokenize
detok_resp = httpx.post(
@@ -32,6 +32,6 @@ def test_tokenize_round_trip(router: str) -> None:
)
assert detok_resp.status_code == 200, detok_resp.text
recovered = detok_resp.json()["text"]
assert (
TEXT in recovered or recovered in TEXT
), f"Round-trip mismatch: original={TEXT!r}, recovered={recovered!r}"
assert TEXT in recovered or recovered in TEXT, (
f"Round-trip mismatch: original={TEXT!r}, recovered={recovered!r}"
)
experimental/sgl-router/tests/scripts/generate_parity_fixtures.py
+1 -1
View File
@@ -85,7 +85,7 @@ def load_tokenizer_with_fallback(primary, fallback, slug):
raise
continue
raise RuntimeError(
f"No accessible tokenizer for slug={slug} " f"(tried: {primary}, {fallback})"
f"No accessible tokenizer for slug={slug} (tried: {primary}, {fallback})"
)