AnonymousSpottedToad/securityPatternsRecovery / ground_truth /python /limit_request_rate__tumma72_mlx-manager.json
| { | |
| "role_mappings": [ | |
| { | |
| "patternRole": "Entity", | |
| "entity": "run_single", | |
| "file": "backend/mlx_manager/mlx_server/benchmark/runner.py", | |
| "lines": "102", | |
| "notes": "In-repo benchmark CLI client (BenchmarkRunner.run_single, driven by mlx_server/benchmark/cli.py) POSTs /v1/chat/completions to the MLX server (runner.py:139/158), whose requests pass the RateLimitMiddleware Enforcer. Caveat: it is a load/benchmark tool; other callers (frontend, OpenAI clients) are outside the repo.", | |
| "status": "present", | |
| "nodeId": null | |
| }, | |
| { | |
| "patternRole": "Enforcer", | |
| "entity": "dispatch", | |
| "file": "backend/mlx_manager/mlx_server/middleware/rate_limit.py", | |
| "lines": "110", | |
| "notes": "RateLimitMiddleware.dispatch intercepts every request, derives the client IP (_get_client_ip, rate_limit.py:86), calls bucket.consume() (rate_limit.py:121) and returns 429 with Retry-After or forwards via call_next. Registered by app_instance.add_middleware in create_app (mlx_server/main.py:248), only when rate_limit_rpm > 0 (default 0 = disabled; opt-in via config).", | |
| "status": "present", | |
| "nodeId": null | |
| }, | |
| { | |
| "patternRole": "Limiter", | |
| "entity": "consume", | |
| "file": "backend/mlx_manager/mlx_server/middleware/rate_limit.py", | |
| "lines": "39", | |
| "notes": "Called from dispatch (rate_limit.py:121). TokenBucket.consume refills tokens by elapsed time * rate (capped at capacity) and returns True/False to allow or block; it also updates the bucket state (history update step).", | |
| "status": "present", | |
| "nodeId": null | |
| }, | |
| { | |
| "patternRole": "HistoryProvider", | |
| "entity": "_get_bucket", | |
| "file": "backend/mlx_manager/mlx_server/middleware/rate_limit.py", | |
| "lines": "93", | |
| "notes": "Called from dispatch (rate_limit.py:119); returns the per-IP TokenBucket from the in-memory dict self._buckets (rate_limit.py:82), whose tokens/last_refill encode the request history and are updated in consume. Stale buckets pruned by _maybe_cleanup (rate_limit.py:99).", | |
| "status": "present", | |
| "nodeId": null | |
| }, | |
| { | |
| "patternRole": "PolicyProvider", | |
| "entity": "rate_limit_rpm", | |
| "file": "backend/mlx_manager/mlx_server/config.py", | |
| "lines": "169", | |
| "notes": "Policy is the settings field rate_limit_rpm (requests per minute per IP, default 0), passed to RateLimitMiddleware(rpm=...) at mlx_server/main.py:248 and turned into bucket rate/capacity in __init__ (rate_limit.py:78) and _get_bucket (rate_limit.py:96). No dedicated get_policy method exists, so the config field is cited.", | |
| "status": "present", | |
| "nodeId": null | |
| }, | |
| { | |
| "patternRole": "System", | |
| "entity": "create_chat_completion", | |
| "file": "backend/mlx_manager/mlx_server/api/v1/chat.py", | |
| "lines": "80", | |
| "notes": "Reached via call_next in dispatch (rate_limit.py:139) through v1_router (api/v1/__init__.py:16, included at mlx_server/main.py:254); the /v1/chat/completions handler performs inference. Other v1 routes (completions, embeddings, messages, ...) are protected equally.", | |
| "status": "present", | |
| "nodeId": null | |
| } | |
| ], | |
| "interaction_order": { | |
| "available": true, | |
| "language": "python", | |
| "repo": "tumma72_mlx-manager", | |
| "reason": "", | |
| "entryRole": "Enforcer", | |
| "entryRule": "behaviour", | |
| "interactions": [ | |
| { | |
| "rank": 1, | |
| "interaction": "Enforcer \u2192 HistoryProvider", | |
| "from": "Enforcer", | |
| "to": "HistoryProvider", | |
| "file": "backend/mlx_manager/mlx_server/middleware/rate_limit.py", | |
| "line": 119, | |
| "via": [], | |
| "mechanism": "call", | |
| "path": [ | |
| { | |
| "caller": "backend/mlx_manager/mlx_server/middleware/rate_limit.py:<module>.RateLimitMiddleware.dispatch", | |
| "file": "backend/mlx_manager/mlx_server/middleware/rate_limit.py", | |
| "line": 119, | |
| "code": "self._get_bucket(ip)", | |
| "callee": "backend/mlx_manager/mlx_server/middleware/rate_limit.py:<module>.RateLimitMiddleware._get_bucket", | |
| "mechanism": "call" | |
| } | |
| ], | |
| "parent": "" | |
| }, | |
| { | |
| "rank": 2, | |
| "interaction": "Enforcer \u2192 Limiter", | |
| "from": "Enforcer", | |
| "to": "Limiter", | |
| "file": "backend/mlx_manager/mlx_server/middleware/rate_limit.py", | |
| "line": 121, | |
| "via": [], | |
| "mechanism": "call", | |
| "path": [ | |
| { | |
| "caller": "backend/mlx_manager/mlx_server/middleware/rate_limit.py:<module>.RateLimitMiddleware.dispatch", | |
| "file": "backend/mlx_manager/mlx_server/middleware/rate_limit.py", | |
| "line": 121, | |
| "code": "bucket.consume()", | |
| "callee": "backend/mlx_manager/mlx_server/middleware/rate_limit.py:<module>.TokenBucket.consume", | |
| "mechanism": "call" | |
| } | |
| ], | |
| "parent": "" | |
| } | |
| ], | |
| "anchoredRoles": [ | |
| { | |
| "role": "Enforcer", | |
| "kind": "METHOD" | |
| }, | |
| { | |
| "role": "HistoryProvider", | |
| "kind": "METHOD" | |
| }, | |
| { | |
| "role": "Limiter", | |
| "kind": "METHOD" | |
| }, | |
| { | |
| "role": "PolicyProvider", | |
| "kind": "LOCAL" | |
| } | |
| ], | |
| "skippedRoles": [ | |
| "Entity", | |
| "System" | |
| ] | |
| } | |
| } |
Xet Storage Details
- Size:
- 5.36 kB
- Xet hash:
- ed0bbb09df148752154682fdc79baad02cf050c56a79b7d804f8e8531d373cc2
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.