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

Co-authored-by: Alison Shao <a.shao@wustl.edu>

examples/frontend_language/usage/llava_video/srt_example_llava_v.py

@@ -124,7 +124,6 @@ def batch(video_dir, save_dir, cur_chunk, num_chunks, num_frames=16, batch_size=
 
 
 if __name__ == "__main__":
-
     url = "https://raw.githubusercontent.com/EvolvingLMMs-Lab/sglang/dev/onevision_local/assets/jobs.mp4"
 
     cache_dir = os.path.expanduser("~/.cache")

examples/frontend_language/usage/readme_examples.py

@@ -31,7 +31,7 @@ def tip_suggestion(s):
 
     forks = s.fork(2)
     for i, f in enumerate(forks):
-        f += f"Now, expand tip {i+1} into a paragraph:\n"
+        f += f"Now, expand tip {i + 1} into a paragraph:\n"
         f += sgl.gen(f"detailed_tip", max_tokens=256, stop="\n\n")
 
     s += "Tip 1:" + forks[0]["detailed_tip"] + "\n"

examples/profiler/nsys_profile_tools/gputrc2graph.py

@@ -86,7 +86,7 @@ class GPUTrace2Graph:
         # Update current_end for overlapping intervals
         for i in range(1, len(df)):
             if i % display_units == 0:
-                print(f"processing trace: {int(i/len(df) * 100)} %", end="\r")
+                print(f"processing trace: {int(i / len(df) * 100)} %", end="\r")
             if starts[i] <= current_end:
                 if ends[i] > current_end:
                     # Partial overlap
@@ -182,9 +182,9 @@ class GPUTrace2Graph:
 
     def is_valid_file(self, base_file):
         """asserts if base_file is non-existent or is empty"""
-        assert (
-            os.path.isfile(base_file) and os.path.getsize(base_file) > 0
-        ), f"{base_file} doesn't exist or is empty"
+        assert os.path.isfile(base_file) and os.path.getsize(base_file) > 0, (
+            f"{base_file} doesn't exist or is empty"
+        )
 
     def should_gen_file(self, new_file, base_file):
         """figure out if new file should be generated from base_file"""

examples/runtime/engine/fastapi_engine_inference.py

@@ -130,7 +130,7 @@ def send_requests(server_url, prompts, max_new_tokens, temperature):
     """Sends generation requests to the running server for a list of prompts."""
     # Iterate through prompts and send requests
     for i, prompt in enumerate(prompts):
-        print(f"\n[{i+1}/{len(prompts)}] Sending prompt: '{prompt}'")
+        print(f"\n[{i + 1}/{len(prompts)}] Sending prompt: '{prompt}'")
         payload = {
             "prompt": prompt,
             "max_new_tokens": max_new_tokens,

examples/runtime/engine/offline_batch_inference_qwen_1m.py

@@ -17,8 +17,7 @@ def load_prompt() -> str:
     # https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen2.5-1M/test-data/1m.txt
 
     with urlopen(
-        "https://qianwen-res.oss-cn-beijing.aliyuncs.com"
-        "/Qwen2.5-1M/test-data/64k.txt",
+        "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen2.5-1M/test-data/64k.txt",
         timeout=5,
     ) as response:
         prompt = response.read().decode("utf-8")
@@ -41,9 +40,7 @@ def process_requests(llm: sgl.Engine, prompts: list[str]) -> None:
     for output in outputs:
         prompt_token_ids = output["meta_info"]["prompt_tokens"]
         generated_text = output["text"]
-        print(
-            f"Prompt length: {prompt_token_ids}, " f"Generated text: {generated_text!r}"
-        )
+        print(f"Prompt length: {prompt_token_ids}, Generated text: {generated_text!r}")
 
 
 # Create an LLM.

examples/runtime/qwen3_vl_reranker.py

@@ -44,7 +44,7 @@ def rerank_text_only():
 
     print("Results (sorted by relevance):")
     for i, result in enumerate(results):
-        print(f"  {i+1}. Score: {result['score']:.4f} - {result['document'][:60]}...")
+        print(f"  {i + 1}. Score: {result['score']:.4f} - {result['document'][:60]}...")
     print()
 
 
@@ -99,7 +99,7 @@ def rerank_with_images():
 
     print("Results (sorted by relevance):")
     for i, result in enumerate(results):
-        print(f"  {i+1}. Index: {result['index']}, Score: {result['score']:.4f}")
+        print(f"  {i + 1}. Index: {result['index']}, Score: {result['score']:.4f}")
     print()
 
 
@@ -149,7 +149,7 @@ def rerank_multimodal_query():
 
     print("Results (sorted by relevance):")
     for i, result in enumerate(results):
-        print(f"  {i+1}. Index: {result['index']}, Score: {result['score']:.4f}")
+        print(f"  {i + 1}. Index: {result['index']}, Score: {result['score']:.4f}")
     print()
 
 

examples/usage/modelopt_quantize_and_export.py

@@ -213,7 +213,7 @@ def deploy_exported_model(
         outputs = llm.generate(prompts, sampling_params)
 
         for i, output in enumerate(outputs):
-            print(f"Prompt {i+1}: {prompts[i]}")
+            print(f"Prompt {i + 1}: {prompts[i]}")
             print(f"Output: {output['text']}")
             print()
 

examples/usage/reasoning_aware_compression/rac_collect_traces.py

@@ -335,7 +335,7 @@ def report(manifest: TraceManifest) -> None:
     print(f"  decode tokens   : {manifest.num_decode_tokens}")
     print(f"  total tokens    : {total}")
     print(f"  decode share    : {decode_share:.1%}")
-    print(f"  wall clock      : {manifest.elapsed_seconds/60:.1f} min")
+    print(f"  wall clock      : {manifest.elapsed_seconds / 60:.1f} min")
     if manifest.calibration_mode == "rac":
         print(
             "\nThe decode share is the activation mass that prompt-only "

examples/usage/reasoning_aware_compression/rac_serve_and_eval.py

@@ -209,7 +209,7 @@ def report(results: List[EvalResult]) -> None:
             f"{result.model_path:<{width}}  "
             f"{result.accuracy:>7.3f}  "
             f"{result.mean_completion_tokens:>16.0f}  "
-            f"{result.elapsed_seconds/60:>10.1f}m"
+            f"{result.elapsed_seconds / 60:>10.1f}m"
         )
 
     if len(results) > 1: