(Gj UdZddlZddlZddlZddlZddlZddlZddlmZddl m Z m Z m Z m Z mZddlmZddlZddlZddlmZmZmZddlmZejeZdeezfd Zehd Z eee!d <ej"d ej#Z$ej%d Z&dedefdZ'ia(e ee ee ffe!d<da)e*e!d<iZ+e ee ee ffe!d<dZ,e*e!d<dZ-iZ.e ee ee ee fffe!d<iZ/e ee*fe!d<dZ0defdZ1de e*fdZ2de ee ee fffdZ3de ee ee ffddfdZ4gdZ5e5dZ6dZ7dZ8iZ9e e:e:fe!d <id!d"d#d"d$d"d%d"d&d"d'd"d(d"d)d"d*d"d+d"d,d-d.d/d0d1d2d1d3d/d4d5d6d5id7d1d8d9d:d5d;d<d=d>d?d>d@d>dAdBdCdDdEd"dFd"dGd"dHd"dId5dJdBdKd<dLd"idMdNdOdBdPd"dQdRdSd<dTdUdVd-dWd>dXd>dYdDdZd[d\d[d]d"d^d>d_dBd`dBdadBidbdNdcdNdddBdedNdfdNdgdBdhdBdidjdkdNdldNdmdNdndRdodNdpd<dqd<drd<dsdNdNdUdtZ;duZdxZ?dyZ@dzZAd{edefd|ZBdd~ede eeffdZCd{edefdZDd{edefdZEiddddddddddddddddddddddddQdddddddddd;iddddIddddddddddddddddddddddddddddZFe eefe!d< ddlGmHZIeID]&ZJeJKZLeLreLeFvr eJjMeFeL<'n #eN$rYnwxYwd{ede efdZOd{edefdZPd{edefdZQd{ededefdZRded{edeSddfdZT dded{edeSd~edeSf d„ZUd{edefdÄZVdd{ed~ede efdĄZWde fdƄZXdde deSdeSde eSfd˄ZYde ee fde:edfde eSfdτZZde ee fde eSfdЄZ[de ee fde eSfdфZ\de ee fde ee ffd҄Z]de ee ee ffdede ee fddfdքZ^ddede ee ee fffdلZ_ dd{ed~edede ee ee fffdڄZ` dded{ed~ede eSfdۄZadefd܄Zbde eeSffd݄Zcded{edeSddfdބZdded{ede eSfd߄Zeded{eddfdZfdeSde eSfdZgdede eSfdZhdedeSde eSfdZidede eSfdZjdedefdZkdededefdZldded{ed~ede eSfdZmdded{ed~ede efdZndded{ed~ede eSfdZodedefdZpdedefdZqdedefdZrdded{ed~ede eSfdZsdded{ed~ede eSfdZtdedefdZuded{ed~ede eSfdZvdddd5ddddddd Zwe eeSfe!d<iaxe eeSfe!d<daye*e!d<dZzdede eeSffdZ{ ddedede eSfdZ| dded{ed~edee eSeffdZ} dded{ed~edeSdzdede~dzdeSfdZ dded{ed~edeSdzdede~dzdeSfdZdedeSfdZde e ee fdeSfdZde ee fd eSdeSfd Zde ee fdeSfd Zd}dd de e ee fd ede e e ee fdeSfdZdS(zModel metadata, context lengths, and token estimation utilities. Pure utility functions with no AIAgent dependency. Used by ContextCompressor and run_agent.py for pre-flight context checks. N)Path)AnyDictListOptionalTuple)urlparse)atomic_json_writebase_url_host_matchesbase_url_hostname)OPENROUTER_MODELS_URLreturncdD];}tj|}|r#tj|r|cS>#&& JJJ 4>Ix-aix.aiz-aiz.aikimi-cnarcee-ai deep-seek gmi-cloud novita-ai minimax-cn nvidia-nim moonshot-cn qwen-portal xiaomi-mimo opencode-zen github-models google-gemini minimax-oauth tencent-cloudgoogle-ai-studiogoglmgminimxaizaizengrokkilokimimimonousqwenarceelocalzhipualiyunclaudecustomgeminigithubgooglenovitanvidiaollamaxiaomialibabaarceeaicopilotminimaxstepfuntencentdeepseekgmicloudkilocodemoonshotnemotronnovitaaiopencodetokenhub anthropic dashscope openrouter tencentmaas qwen-oauth copilot-acp kimi-coding opencode-go ollama-cloud openai-codexgithub-copilotkimi-coding-cntencent-tokenhub_PROVIDER_PREFIXESzE^(\d+\.?\d*b|latest|stable|q\d|fp?\d|instruct|chat|coder|vision|text)z 100.64.0.0/10modelc,d|vs|dr|S|dd\}}|}|tvr0t |r|S|S|S)uaStrip a recognised provider prefix from a model string. ``"local:my-model"`` → ``"my-model"`` ``"qwen3.5:27b"`` → ``"qwen3.5:27b"`` (unchanged — not a provider prefix) ``"qwen:0.5b"`` → ``"qwen:0.5b"`` (unchanged — Ollama model:tag) ``"deepseek:latest"``→ ``"deepseek:latest"``(unchanged — Ollama model:tag) :http) startswithsplitstriplowerre_OLLAMA_TAG_PATTERNmatch)rfprefixsuffix prefix_lowers r_strip_provider_prefixrtWs %5++F33 [[a((NFF<<>>''))L)))  $ $V\\^^ 4 4 L Lr_model_metadata_cache_model_metadata_cache_time_novita_metadata_cache_novita_metadata_cache_timei_endpoint_model_metadata_cache#_endpoint_model_metadata_cache_timei,c.ddlm}|dz dz S)z8Return path to the OpenRouter model metadata disk cache.rget_hermes_homecachezopenrouter_model_metadata.jsonhermes_constantsr}r|s r_get_model_metadata_cache_pathrts-000000 ?  w &)I IIrc t}|sdStj|jz }|dkrdS|S#t $rYdSwxYw)zBReturn disk-cache age in seconds, or None if freshness is unknown.Nr)rexiststimestatst_mtime Exception) cache_pathages r&_model_metadata_disk_cache_age_secondsrzs} 355   "" 4ikkJOO--66 774 tts"A3AA A+*A+cx t}|dd5}tj|}dddn #1swxYwYt |t siSd|DS#t$r'}t d|icYd}~Sd}~wwxYw)z3Load processed OpenRouter metadata cache from disk.rutf-8encodingNc^i|]*\}}t|tt||+S) isinstancedictstr.0keyvalues r z3_load_model_metadata_disk_cache..sD   U%&& HHe   rz7Failed to load OpenRouter model metadata disk cache: %s) ropenjsonloadrritemsrloggerdebug)rfdataes r_load_model_metadata_disk_cachers  355 __S7_ 3 3 q9Q<i@rg>@_LOCAL_CTX_PROBE_CACHEzclaude-fable-5@Bz claude-fablezclaude-opus-4-8zclaude-opus-4.8zclaude-opus-4-7zclaude-opus-4.7zclaude-opus-4-6zclaude-sonnet-4-6zclaude-opus-4.6zclaude-sonnet-4.6rAi@ gpt-5.5iz gpt-5.4-nano gpt-5.4-minigpt-5.4gpt-5.3-codex-sparkrz gpt-5.1-chatgpt-5zgpt-4.1izgpt-4rCizgemma-4rgemma4z gemma-4-31bzgemma-3igemmai zdeepseek-v4-prozdeepseek-v4-flashz deepseek-chatzdeepseek-reasonerrPllamaz qwen3.6-pluszqwen3-coder-plusz qwen3-coderir< minimax-m3rM zglm-5.2r1iz grok-composerz grok-buildzgrok-code-fastz grok-2-visionz grok-4-fastiz grok-4.20grok-4.3zgrok-4zgrok-3zgrok-2r7r9z hy3-previewrTtrinityelephantzQwen/Qwen3.5-397B-A17BzQwen/Qwen3.5-35B-A3Bzdeepseek-ai/DeepSeek-V3.2izmoonshotai/Kimi-K2.5zmoonshotai/Kimi-K2.6zmoonshotai/Kimi-K2-ThinkingzMiniMaxAI/MiniMax-M2.5zXiaomiMiMo/MiMo-V2-Flashz mimo-v2-proz mimo-v2.5-proz mimo-v2.5z mimo-v2-omni)z mimo-v2-flashz zai-org/GLM-5)z grok-3-minizgrok-4.20-multi-agentrc|pdsdSdD]"}|vr|dd#tfdtDS)a$Return True when an xAI Grok model accepts ``reasoning.effort``. Allowlist by substring (matches both bare ``grok-3-mini`` and aggregator-prefixed ``x-ai/grok-3-mini``). Conservative by design: if a future Grok model isn't listed, we send no effort dial rather than 400. F)/rjc3BK|]}|VdSN)rk)rrqnames r z1grok_supports_reasoning_effort..ns/SS6tv&&SSSSSSr)rmrnrsplitany_GROK_EFFORT_CAPABLE_PREFIXES)rfseprs @rgrok_supports_reasoning_effortr_s KR   & & ( (D u++ $;;;;sA&&r*D SSSS5RSSS S SSr) context_lengthcontext_window context_sizemax_context_lengthmax_position_embeddings max_model_lenmax_input_tokensmax_sequence_length max_seq_len n_ctx_trainn_ctxctx_size)max_completion_tokensmax_output_tokens max_tokens) localhostz 127.0.0.1z::1z0.0.0.0)z.docker.internalz.containers.internalz.lima.internalbase_urlcT|pddS)Nrr)rmrstriprs r_normalize_base_urlrs& N ! ! # # * *3 / //rrapi_keyc^t|pd}|siSdd|iS)Nr AuthorizationBearer )rrm)rtokens r _auth_headersrs@  2   $ $ & &E  .u.. //rc"t|dS)N openrouter.ai)r rs r_is_openrouter_base_urlrs ? ; ;;rc^t|}t|ot| Sr)rboolr)r normalizeds r_is_custom_endpointrs.$X..J    G$;J$G$G GGrzapi.openai.comopenaiz chatgpt.comapi.anthropic.comrXzapi.z.air5zopen.bigmodel.cnzapi.moonshot.air^zapi.moonshot.cnrcz api.kimi.comzapi.stepfun.airNzapi.stepfun.comz api.arcee.air=z api.minimaxzdashscope.aliyuncs.comrJzdashscope-intl.aliyuncs.comzportal.qwen.air\rrZz!generativelanguage.googleapis.comzinference-api.nousresearch.comr;zapi.deepseek.comzapi.githubcopilot.comrLz.githubcopilot.comzmodels.github.aizmodels.inference.ai.azure.comzapi.fireworks.ai fireworksz opencode.air_zapi.x.air4zintegrate.api.nvidia.comrGzapi.xiaomimimo.comrIzxiaomimimo.comzapi.gmi-serving.comr2 api.novita.airFztokenhub.tencentmaas.comrdz ollama.comr`_URL_TO_PROVIDER)list_providersct|}|sdStd|vr|nd|}|jp|j}t D] \}}||vr|cSdS)aInfer the models.dev provider name from a base URL. This allows context length resolution via models.dev for custom endpoints like DashScope (Alibaba), Z.AI, Kimi, etc. without requiring the user to explicitly set the provider name in config. N://zhttps://)rr netlocrnrrr)rrparsedhosturl_partproviders r_infer_provider_from_urlrs%X..J t EZ$7$7jj=T =T=T U UF =   7FK$5$5$7$7D.4466( t  OOO  4rct|d}dD]D}||r-|dt| d}nE|S)zBReturn the LM Studio server root for native ``/api/v1`` endpoints.r)z/api/v1z/api/v1N)rrendswithlen)rrootrrs r_lmstudio_server_rootrsu x ( ( / / 4 4D, ==  3v;;,'..s33D E  Krc$t|duSr)rrs r_is_known_provider_base_urlrs #H - -T 99rrc|dkS)uReturn True when the on-disk context cache must not short-circuit probing. LM Studio excludes caching because loaded context is transient — the user can reload the model with a different context_length at any time. lmstudior)rrs r_skip_persistent_context_cachers z !!rlengthcB|tkrt|||dSdS)ubPersist a locally probed context length only when it meets Hermes minimum. Sub-minimum live windows (e.g. vLLM ``--max-model-len 32768``) are still returned to callers so ``agent_init`` can fail with the existing minimum-context guidance — they must not be normalized into the on-disk cache as if they were valid operating limits. N)MINIMUM_CONTEXT_LENGTHsave_context_length)rfrrs r!_maybe_cache_local_context_lengthrs0'''E8V44444('rcachedcXt|||}|r|dkr||kr|tkr9td|||dtdt |||Std|||d|dt ||t ||||S|S)uReturn *cached* unless a live local probe reports a different limit. vLLM/Ollama operators can restart with a new ``--max-model-len`` / ``num_ctx`` without changing the model id. When the server is reachable, prefer its reported window over a stale disk entry; when the probe fails (offline tests, network blip), keep the cached value. Live probes below :data:`MINIMUM_CONTEXT_LENGTH` invalidate stale cache entries but are not persisted — startup should reject them, not bless a sub-64K window as config. rrukLive local probe for %s@%s reports %s (< minimum %s); invalidating stale cache — agent init should rejectrz@Reconciling stale local cache entry %s@%s: %s -> %s (live probe))_query_local_context_lengthrrinfo!_invalidate_cached_context_lengthr)rfrr rlive_ctxs r&_reconcile_local_cached_context_lengthrs"+5(GLLLHHqLLX%7%7 , , , KKHxH5K2O2O    .eX > > >O N 8]]xOO    *%:::)%8DDD Mrct|}|sdSd|vr|nd|} t|}|jpdn#t$rYdSwxYwtvrdSt fdt DrdSrdvrdS tj}|j s|j s|j rdSt|tj r |tvrdSn#t$rYnwxYwd}t#|dkr t%|d t%|d }}|d krdS|d krd |cxkrdkrnndS|dkr|dkrdS|dkrd|cxkrdkrnndSn#t$rYnwxYwdS)aReturn True if base_url points to a local machine. Recognises loopback (``localhost``, ``127.0.0.0/8``, ``::1``), container-internal DNS names (``host.docker.internal`` et al.), RFC-1918 private ranges (``10/8``, ``172.16/12``, ``192.168/16``), link-local, and Tailscale CGNAT (``100.64.0.0/10``). Tailscale CGNAT is included so remote-but-trusted Ollama boxes reached over a Tailscale mesh get the same timeout auto-bumps as localhost Ollama. Frzhttp://rTc3BK|]}|VdSr)r)rrrrs rrz$is_local_endpoint..Ps/ I IV4== I I I I I Ir.rrj d@)rr hostnamer _LOCAL_HOSTSr_CONTAINER_LOCAL_SUFFIXES ipaddress ip_address is_private is_loopback is_link_localr IPv4Address_TAILSCALE_CGNAT ValueErrorrlrint) rrurlraddrpartsfirstsecondrs @ris_local_endpointr/:s5%X..J u++**1G:1G1GC#$" uu |t I I I I/H I I IIIt 4t #D)) ? d. $2D 4 dI1 2 2 t?O7O7O4       JJsOOE 5zzQ aMM3uQx==6E{{t||f 2 2 2 2 2 2 2 2 2t||# t||f 3 3 3 3 3 3 3 3 3t    D  5sK: AA)C,#C CC 0E<=E< E<#E<< F F c ddl}t|}|}|dr |dd}t|}t |} |d|5} ||d}|jdkr dddd Sn#t$rYnwxYw ||d }|jdkr8 | } d | vr dddd Sn#t$rYnwxYwn#t$rYnwxYw ||d }|jdkr||d}|jdkrd|j vr ddddSn#t$rYnwxYw ||d}|jdkr&| } d| vr ddddSn#t$rYnwxYwdddn #1swxYwYn#t$rYnwxYwdS)zDetect which local server is running at base_url by probing known endpoints. Returns one of: "ollama", "lm-studio", "vllm", "llamacpp", or None. rNrg@timeoutheaders/api/v1/models lm-studioz /api/tagsmodelsrH /v1/props/propsdefault_generation_settingsllamacppz/versionversionvllm) httpxrrrrClientget status_coderrtext) rrr?r server_url lmstudio_urlr4clientrrs rdetect_local_server_typerGrsm LLL$X..JJ5!!%_ (22LG$$G* \\#w\ 7 7' 6 JJ,>>>??=C''& ' ' ' ' ' ' ' ' (     JJ*77788=C'' vvxx#t++#+#' ' ' ' ' ' ' '  ,$     JJ*77788=C'' j#8#8#899A=C'',IQV,S,S%9' ' ' ' ' ' ' ' :     JJ*66677=C''6688D D((%K' ' ' ' ' ' ' ' L    M' ' ' ' ' ' ' ' ' ' ' ' ' ' ' P       4sG>+G2-#B G>G2 B,)G2+B,,G20#D C;- G>:D ; DD DD  G2 DG2DG2AE;- G>:G2; FG2FG2 ;G G>G2 G# G2"G##G2& G>2G66G>9G6:G>> H  H rc#Kt|tr2|V|D]}t|Ed{VdSt|tr|D]}t|Ed{VdSdSr)rrvalues_iter_nested_dictslist)rnesteditems rrJrJs%0 llnn 2 2F)&11 1 1 1 1 1 1 1 1 2 2 E4 0 0 0D)$// / / / / / / / /00 0 0r逖minimummaximumc$ t|trdSt|tr(|dd}t |}n#t tf$rYdSwxYw||cxkr|krnn|SdS)Nrr)rrrrmreplacer) TypeErrorr()rrPrQresults r_coerce_reasonable_intrVs eT " " 4 eS ! ! 3KKMM))#r22EU z "tt&####G##### 4sA&A A&&A;:A;payloadkeys.cd|D}t|D]W}|D]@\}}t||vr)t |}||ccSAXdS)Nc6h|]}|Srrn)rrs r z%_extract_first_int..s * * *cciikk * * *r)rJrrrnrV)rWrXkeysetmappingrrcoerceds r_extract_first_intr`s * *T * * *F%g..!--//  JC3xx~~v--,U33G"#   4rc,t|tSr)r`_CONTEXT_LENGTH_KEYSrWs r_extract_context_lengthrd g'; < <.s.PPP*#uc#hhnn&&PPPrc3NK|]}tfd|DV dS)c3 K|]}|vV dSrr)raliasrs rrz-_extract_pricing...s(@@uu *@@@@@@rN)r)raliasesrs rrz#_extract_pricing..s@ccW3@@@@@@@@@ccccccr>Nr)rArfloatrJrrrI) rW novita_input novita_outputpricing alias_mapr^targetrrrs @r_extract_pricingrs;;899LKK :;;M=#<"$  # #E,$7$7&$@9$L M MGH   $$'m(<(2)F)F)J)JKbdh)i)i &(33 99Y33 E uh 6 6 6 "2B77I .sVJq$4G4Gj!!Z/rrr9r:r;r model_aliasz1Failed to fetch model metadata from %s/models: %s)rrryrArzr_ENDPOINT_MODEL_CACHE_TTLrrappendr/rGrrrrrrrr)rhrrrrrdrrSokrr)#rrrrr  cached_at candidates alternater4 last_errorrDrrWr~rfrrrinstcfgctxrralt_idexc candidater* is_llamacppbase_verify props_respprops gen_settingsrrs# rfetch_endpoint_model_metadatar=s%X..J 0<< /33J??7;;JJJ  49;;#:>W"W"WMJ5!!'ssO**3// & %Yj00)$$$8?G 3' 3 344RG&*J$$21 ' GDDD SS2:>> #<%%c**-==#355  ))+++"--//35$[[266AAE%eT22! $yy//B599T??H#! -3UYYvx5P5P,QE%)N % *>>"YYt__F!&#..A6Af>P>P*5&%@@@=B.z:BF)++3J? ]T^   JJJJJJ  44 s##i/2 |C"MeMgMghhhH  % % ' ' 'mmooG/1E VR00 ; ;!%.. 99T??)/681L1L(M!8!?!?!-.Bikk / ;LLL   JJJJJJ b H*V`aaa13":.6:ikk' 3 IsW1IL33 M =MM *D/V7C#U>=V7> V V7 V  )V77 WW  Wct||}||}|sht|dkr/tt |}n&|D]\}}||vs||vr|}n|r,|d}t|tr|SdS)zDResolve context length from an endpoint's live ``/models`` metadata.r rjrN) rrArnextiterrIrrr))rfrrendpoint_metadatamatchedrrrs r _resolve_endpoint_context_lengthrs 6hPPP##E**G   ! !Q & &4 1 8 8 : :;;< context_length cache from disk.rrNcontext_lengthsz'Failed to load context length cache: %s) rrryaml safe_loadrArrr)rrrrs r_load_context_cachers " $ $D ;;==  $ ) ) ) +Q>!$$*D + + + + + + + + + + + + + + +xx)2...  >BBB s@A;A A;AA;!A"A;; B,B'!B,'B,c|d|}t}|||krdS|||<t} |jddt |dd5}t jd|i|d dddn #1swxYwYt d ||d dS#t$r&}t d |Yd}~dSd}~wwxYw)zPersist a discovered context length for a model+provider combo. Cache key is ``model@base_url`` so the same model name served from different providers can have different limits. @NTparentsexist_okwrrrFdefault_flow_stylez%Cached context length %s -> %s tokensrz'Failed to save context length cache: %s) rrArparentmkdirrrdumprr rr)rfrrrr~rrrs rrrsp   X  C  ! !E yy~~E#J " $ $DC $666 $g . . . O! I(%0! N N N N O O O O O O O O O O O O O O O ;SV--PPPPP CCC >BBBBBBBBBCs<.C3B CBC B!!C C4C//C4cV|d|}t}||S)zBLook up a previously discovered context length for model+provider.r)rrA)rfrrr~s rget_cached_context_lengthr s0   X  C  ! !E 99S>>rc|d|}t}||vrdS||=t} |jddt |dd5}t jd|i|d ddddS#1swxYwYdS#t$r'}t d ||Yd}~dSd}~wwxYw) zCDrop a stale cache entry so it gets re-resolved on the next lookup.rNTrrrrrFrz6Failed to invalidate context length cache entry %s: %s) rrrrrrrrrr)rfrrr~rrrs rrrsV   X  C  ! !E % c " $ $DW $666 $g . . . O! I(%0! N N N N O O O O O O O O O O O O O O O O O O WWW MsTUVVVVVVVVVWs;.BB6 BBB B B CB<<Ccurrent_lengthc.tD] }||kr|cS dS)z@Return the next lower probe tier, or None if already at minimum.N)CONTEXT_PROBE_TIERS)rtiers rget_next_probe_tierr$s/# . KKK ! 4r error_msgc|}gd}|D]O}tj||}|r6t|d}d|cxkrdkrnK|cSPdS)a?Try to extract the actual context limit from an API error message. Many providers include the limit in their error text, e.g.: - "maximum context length is 32768 tokens" - "context_length_exceeded: 131072" - "Maximum context size 32768 exceeded" - "model's max context length is 65536" )z+max_model_len\s*(?:is\s*)?[:=(]?\s*(\d{4,})z2maximum model length\s*(?:is\s*)?[:=(]?\s*(\d{4,})zY(?:max(?:imum)?|limit)\s*(?:context\s*)?(?:length|size|window)?\s*(?:is|of|:)?\s*(\d{4,})z:context\s*(?:length|size|window)\s*(?:is|of|:)?\s*(\d{4,})z)(\d{4,})\s*(?:token)?\s*(?:context|limit)z">\s*(\d{4,})\s*(?:max|limit|token)z(\d{4,})\s*(?:max(?:imum)?)\brjrNrONrnresearchr)group)r error_lowerpatternspatternrplimits rparse_context_limit_from_errorr,s//##KH ';//   A''Eu**** ***** 4rcurrent_context_lengthc<t|}|dS||kr|SdS)uReturn a provider-reported lower context limit, if one is present. Context-overflow recovery must not invent a new model window size. Some providers only say that the input exceeds the context window without reporting the actual maximum. In that case callers should keep the configured context length and try compression only, rather than stepping down through guessed probe tiers (1M → 256K → 128K → ...). N)r)rr parsed_limits r&get_context_length_from_provider_errorrJs32)< window. Fix: reduce max_tokens (the output cap) for this call. Do NOT touch context_length — the window hasn't shrunk. Anthropic's API returns errors like: "max_tokens: 32768 > context_window: 200000 - input_tokens: 190000 = available_tokens: 10000" Returns the number of output tokens that would fit (e.g. 10000 above), or None if the error does not look like a max_tokens-too-large error. ravailable_tokensavailable tokens in the outputmaximum context length requested output tokensrange of max_tokens should beNz9range of max_tokens should be\s*\[\s*\d+\s*,\s*(\d+)\s*\]rj)zavailable_tokens[:\s]+(\d+)zavailable\s+tokens[:\s]+(\d+)z =\s*(\d+)\s*$zmaximum context length is (\d+)zK\((\d+)\s+of text input,\s*(\d+)\s+of tool input,\s*(\d+)\s+in the output\)z'maximum context length is (\d+)\s*tokenz!prompt contains (\d+)\s*characterz2prompt contains (?:at least )?(\d+)\s*input tokensr)rris_output_cap_error_m_range_caprrrptokens_m_ctx_m_parts _available _m_ctx_tok_m_chars_ctx _est_input _m_vllm_inputs r(parse_available_output_tokens_from_errorr^s"//##K  # U ; . S2D 2S ;& 4 $ 3  !K/ + ; & + { *!0 (;614 tyDH8>>!$$%% 199KH  ';//  Q((F{{ Y9; G GFyVH(a))Cq0A0A,B,BBSXYIZIZE[E[[ ?? E{SSJy={KKHh:##A&&''(..++,,q0Q6 J& ?? I={Mm))!,,--M4G4G4J4J0K0KK ??  4rc|}d|vpd|vpd|v}|sdSd|vp#d|vpd|vpd|vod |vpd |vod |vp d |vpd |vpd|v}|sdSd|vpd|vpd|vpd|vp d|vpd|vpd|v}| S)u5Return True if a 400 is about the OUTPUT cap (max_tokens) being too large. This is the broader sibling of :func:`parse_available_output_tokens_from_error`: that function only returns a number when it can extract the available output budget from a *known* provider phrasing. This one answers the cheaper yes/no question — "is this an output-cap error at all?" — across providers whose exact wording we may not yet parse a number from. Why this matters: an output-cap 400 is deterministic (every retry with the same ``max_tokens`` gets the identical rejection). If such an error is misclassified as a context-overflow it gets routed into the compression loop, the compressor re-issues the call with the same oversized ``max_tokens``, the provider rejects it identically, and the session death-loops until "cannot compress further" (issue #55546, DashScope/Qwen: "Range of max_tokens should be [1, 65536]"). Compression cannot help an output-cap error — the input already fits. The signal: the error talks about ``max_tokens`` (or its aliases) as a cap/range/limit, and does NOT talk about the INPUT/prompt/context window being too long. When both are present we defer to the context-overflow path (a real input overflow can also mention max_tokens). rrrFrrrrrrrz should bezless than or equalzmust bezprompt is too longzprompt too longzinput is too longz input tokenz prompt lengthzprompt containszreduce the lengthr[)rrmentions_output_paramoutput_cap_signalinput_overflow_signals rrrsX.//##K  # 2 + - 2 "k 1 !u (;6 $  , $  , $ { * 8(K7 $ ; & /;. $ + % $ ; . $  # u  + .  + . + - . K ' . k )  .  +  . + -% $$r candidate_id lookup_modelcb||krdSd|vr"|ddd|krdSdS)aReturn True if *candidate_id* (from server) matches *lookup_model* (configured). Supports two forms: - Exact match: "nvidia-nemotron-super-49b-v1" == "nvidia-nemotron-super-49b-v1" - Slug match: "nvidia/nvidia-nemotron-super-49b-v1" matches "nvidia-nemotron-super-49b-v1" (the part after the last "/" equals lookup_model) This covers LM Studio's native API which stores models as "publisher/slug" while users typically configure only the slug after the "local:" prefix. TrrjF)r)rr s r_model_id_matchesr sH|##t l|223::1=MMt 5rc@ddl}t|}|d}|dr |dd} t ||}n#t $rYdSwxYw|dkrdSt |} |d| 5}||d d |i } | j d kr ddddS| } | dd} d| vr| dD]s} d| vrm| } t| dkr4 t| dccdddS#t $rYowxYwt| di}|D]B\}}d|vr9t%|tt&frt|ccdddSC dddn #1swxYwYn#t $rYnwxYwdS)ayQuery an Ollama server for the model's context length. Returns the model's maximum context from GGUF metadata via ``/api/show``, or the explicit ``num_ctx`` from the Modelfile if set. Returns None if the server is unreachable or not Ollama. This is the value that should be passed as ``num_ctx`` in Ollama chat requests to override the default 2048. rNrrr1r rH@r2 /api/showrrr6 parametersrnum_ctx rr model_infor)r?rtrrrGrrr@postrBrrArlrmrr)r(rrr)rfrrr?rrD server_typer4rFresprparamsliner,rrrs rquery_ollama_num_ctxr,sLLL'..J%%J5!!%_ .xIII tthtG$$G \\#w\ 7 7 &6;;*777vz>R;SSD3&& & & & & & & & &99;;DXXlB//FF"""LL..%%D D(( $ 2 2 4 4u::??%'*59~~ 5 5 & & & & & & & &$.%%% $%,33J(..00 & & U#s**z%#u/N/N*u::%%- & & & & & & & &( &) & & & & & & & & & & & & & & &.      4s A A)(A)H)H HBHE7(H* H7 FHFA#H' H4H6 HHH H H HHcdddl}t|}|r|sdS t||dkrdSn#t$rYdSwxYw|d}|dr |dd}t |} |d| 5}||d d |i }|j d kr ddddS| } dddn #1swxYwYn#t$rYdSwxYw| d} t| trtd| DrdS| rdS| d} t| tr+| D](} dt!| vrdS)dS)a Return True/False when Ollama ``/api/show`` reports vision support. Uses the ``capabilities`` field on Ollama 0.6.0+ and falls back to ``model_info.*.vision.block_count`` on older servers. Returns None when the server is unreachable, not Ollama, or the model is unknown. rNr rHrrr1r r2rrrr6 capabilitiesc3`K|])}t|dkV*dS)visionNr)rcaps rrz/query_ollama_supports_vision..s7<<s3xx~~8+<<<<< > >( J J4 K tt%%J5!!%_ G$$G \\#w\ 7 7 6;;*777vz>R;SSD3&&        99;;D                 tt 88N # #D$ <&?&?#&!4:cC<+H+H!4+.s88OOOOYB ,B ,B ,B ,B ,B ,B ,B ,Z"E"::????@@D3&&yy{{hh//g488N8O8OfSTSXSXYeSfSf,:cC<#@#@,#&s88OOEB ,B ,B ,B ,B ,B ,B ,B ,B ,B ,B ,B ,B ,B ,B ,B ,B ,B ,B ,B ,B ,B ,B ,F       4s'A99 BB S#B2SE9*S, S9 FSFA#S) S6DS SB(S= S C"S, S9S: SS  S S S SSc.|ddS)zNormalize version separators for matching. Nous uses dashes: claude-opus-4-6, claude-sonnet-4-5 OpenRouter uses dots: claude-opus-4.6, claude-sonnet-4.5 Normalize both to dashes for comparison. r-)rSr&s r_normalize_model_versionr5ws ==c " ""rcr|r|drdS |d}|dr |dd}|d}|dd}tj||d t }|jd krdS|}|d gD]O}|d |kr4|d} t| tr | dkr| cSPn2#t$r%} t d| Yd} ~ nd} ~ wwxYwdS)zQuery Anthropic's /v1/models endpoint for context length. Only works with regular ANTHROPIC_API_KEY (sk-ant-api*). OAuth tokens (sk-ant-oat*) from Claude Code return 401. z sk-ant-oatNrrr1z/v1/models?limit=1000z 2023-06-01)z x-api-keyzanthropic-versionrrr6rrrrz%Anthropic /v1/models query failed: %s) rkrrrrArrBrrr)rrr) rfrrrr*r4rrrrrs r_query_anthropic_context_lengthr7sz g((66tAs## ==   9D,,, !-  |C"E]E_E_```  s " "4yy{{&"%%  AuuT{{e##ee.//c3''C!GGJJJ   AAA  &>&>?+--      s " " LLZ    Iyy{{  5s;;; )34(>(>Fdhhx$$$BGF''$%%  xxhh'(( dC  'ZS%9%9 'cAgg#&F4::<<  .%+"*-' Ms$AB! B!! C+C C Cct|}|sdS|ret|}||vr||S|}|D]!\}}||kr|cS"|}t t ddD] \}}||vr|cSdS)zResolve a Codex OAuth model's real context window. Prefers a live probe of chatgpt.com/backend-api/codex/models (when we have a bearer token), then falls back to ``_CODEX_OAUTH_CONTEXT_FALLBACK``. Nc,t|dSNrrxs rz5_resolve_codex_oauth_context_length.. sS1YYrTrreverse)rtrmr@rnrsortedr8)rfr; model_barelive model_lowerr=rs r#_resolve_codex_oauth_context_lengthrNs(..4466J t1,??    # # &&((   ID#zz||{** +""$$K%++--3F3FPT c ;  JJJ  4rc8|rt|||}||dfSt}dtdtdtt fd}||vr||||}||dfSt |}|D]\}} d |vr| d d d n|} | |ks%t | |kr||| }||dfcS|} |D]\}} d |vr| d d d n|} | | ft | |ffD]g\} } | | rMt| t| ks| t| d vr||| }||dfccShd S) uMResolve Nous Portal model context length. Tries the live Nous inference endpoint first (authoritative), then falls back to OpenRouter metadata with suffix/version matching. Nous model IDs are bare after prefix-stripping (e.g. 'qwen3.6-plus', 'claude-opus-4-6') while OpenRouter uses prefixed IDs (e.g. 'qwen/qwen3.6-plus', 'anthropic/claude-opus-4.6'). Version normalization (dot↔dash) is applied to handle name drifts. Returns ``(context_length, source)`` where ``source`` is one of: - ``"portal"`` — live /v1/models response (authoritative) - ``"openrouter"`` — OpenRouter cache fallback (non-authoritative; callers must NOT persist this to the on-disk cache or a single portal blip will freeze the wrong value in forever) - ``""`` — could not resolve r Nportalor_idrrc|d}|dS|dkr-t|rtd||dS|S)Nrz{Rejecting OpenRouter metadata context=%s for %r (Kimi-family underreport, Nous path); falling through to hardcoded defaults)rAr$rr )rQrrs r _safe_ctxz/_resolve_nous_context_length.._safe_ctx5saii()) ;4 %<<5e<<< KK^U    4 rrZrrjz-:.)Nr) rrrrrr)r5rnrrlrkr)rfrr portal_ctxmetadatarTrrrQrbarerMrquerys r_resolve_nous_context_lengthrYsf4(5eXwWWW  !x' '#%%H  T hsm     ix// ? $ $)%006688J (()) u),u{{3""1%%5 ::<<5;;== ( (,DT,J,J,P,P,R,RV`,`,`)E5))CL((((++--K ((-- u),u{{3""1%%5"&**,, !r s - - Iu##E** -I#e**,, #e**0E0N0Niu--? ,,,,,,  - 8rconfig_context_lengthcustom_providersc|t|tr|dkr|S|pddkrT ddlm}ddlm}ddlm }|| dpi|} | dpi} t| d pd} t| d pd} | rf| rd| dkrL|| | } t| | d dpd| d dpd| Sn,#t$rtddYnwxYw|r.|r,|r* ddlm}||||}|r|Sn#t$rYnwxYwt#|}|rt%||sst'||}|`|dkr7|dkr1td|||dt+||n#|dkr?t-|r0td|||dt+||n|dkr?t/|r0td|||dt+||n|dkr?t1|r0td|||dt+||nTt3|dkrtd||n$t5|rt7|||| S|S|d!ks4|rTt9|d"r2t=|d#r" dd$lm }||S#tB$rYnwxYw|d%ks|r;t=|d&r+tE||pd'| }||rtG||||StI|rOtK|s?tE||| }||StK|stM||| }|#t%||stG||||St5|r=tO||| }|r)|dkr#t%||stQ||||Std(||tRd|}tUtV,d)d*D],\}}||vr#td+|d|||cS-tRS|d,ks|r*t9|d-krt[||pd.|}|r|S|}|r|d/vr|rt3|}|r|}|d0vr) dd1l.m/}||| }|r|Sn#t$rYnwxYw|dkr6ta||pd|pd2\}}|r|r|d3krtG||||S|dkr*tc||pd4}|r|rtG||||S|d5kr|rtE||| }||S|r7tM||| }|#t%||stG||||S|d6krlte}| |}|rG| d7}t|tr|dkr|dkrt-|s|S|rhdd8l3m4} | ||}|rTt/|rCtV d9}!|!r'||!kr!td:|||!d|!}|S|sgte}||vrU|| d7tR}|dkr,t-|rtd;||n|S|rLt5|r=tO||| }|r)|dkr#t%||stQ||||S|}tUtV,d<d*D] \}}||vr|cStRS)=a2Get the context length for a model. Resolution order: 0. Explicit config override (model.context_length or custom_providers per-model) 1. Persistent cache (previously discovered via probing). Nous URLs bypass the cache here so step 5b can always reconcile against the authoritative portal /v1/models response. 1b. AWS Bedrock static table (must precede custom-endpoint probe) 2. Active endpoint metadata (/models for explicit custom endpoints) 3. Local server query (for local endpoints) 4. Anthropic /v1/models API (API-key users only, not OAuth) 5. Provider-aware lookups (before generic OpenRouter cache): a. Copilot live /models API b. Nous: live /v1/models probe first (authoritative), then OR cache fallback with suffix/version normalisation. Only portal-derived values are persisted to disk. c. Codex OAuth /models probe d. GMI /models endpoint e. Ollama native /api/show probe (any base_url, provider-agnostic) f. models.dev registry lookup (with :cloud/-cloud suffix fallback) 6. OpenRouter live API metadata (Kimi-family 32k guard) 7. Local server query (before hardcoded defaults for local endpoints) 8. Hardcoded defaults (broad family patterns, longest-key-first) 9. Default fallback (256K)Nrrmoa) load_config)resolve_moa_preset)resolve_runtime_provider aggregatorrrf)r target_modelrr)rrrz/MoA aggregator context-length resolution failedT)exc_info)"get_custom_provider_context_length)rfrr[rarzaDropping stale Codex cache entry %s@%s -> %s (pre-fix value); re-resolving via live /models proberrSziDropping stale Kimi cache entry %s@%s -> %s (OpenRouter underreport); re-resolving via hardcoded defaultsrzjDropping stale MiniMax-M3 cache entry %s@%s -> %s (pre-catalog value); re-resolving via hardcoded defaultsrzhDropping stale Grok-4.3 cache entry %s@%s -> %s (pre-catalog value); re-resolving via hardcoded defaultsr;z@Bypassing persistent cache for %s@%s (Nous portal authoritative)r bedrockzbedrock-runtime.z amazonaws.com)get_bedrock_context_lengthrFrzhttps://api.novita.ai/openai/v1uCould not detect context length for model %r at %s — defaulting to %s tokens (probe-down). Set model.context_length in config.yaml to override.c,t|dSrCrDrEs rrGz*get_model_context_length..Csc!A$iirrHzUUsing hardcoded context length %s for model %r (custom endpoint, catalog match on %r)rXrzhttps://api.anthropic.com>rBrZ>rLr]rb)get_copilot_model_context)rrrP)r;r2rZr)lookup_models_dev_contextrz[Rejecting models.dev context=%s for %r (MiniMax-M3 underreport); using hardcoded default %szpRejecting OpenRouter metadata context=%s for %r (Kimi-family underreport); falling through to hardcoded defaultsc,t|dSrCrDrEs rrGz*get_model_context_length..ss1Q4yyr)5rr)rmrnhermes_cli.configr^hermes_cli.moa_configr_hermes_cli.runtime_providerr`rArget_model_context_lengthrrrrdrtrrr rr$r'r)rr/rr rkr agent.bedrock_adapterrf ImportErrorrrrrr"r rDEFAULT_FALLBACK_CONTEXTrJDEFAULT_CONTEXT_LENGTHSrr7hermes_cli.modelsrhrYrNragent.models_devri)"rfrrrZrr[r^r_r`presetagg agg_provider agg_modelrtrdcp_ctxr rfrr local_ctxrM default_modelreffective_providerinferredrhsource codex_ctxrVror_ctxricatalogs" rrnrn^s B(Z8Ms-S-S(XmpqXqXq$$ B%%''500 [ 5 5 5 5 5 5 @ @ @ @ @ @ L L L L L L'' (9(9%(@(@(FBNNF**\**0bCswwz228b99??AALCGGG,,23399;;I \ l.@.@.B.Be.K.K-- S\]]]/VVJ339rFF9b117R)   [ [ [ LLJUYL Z Z Z Z Z [ H    L L L L L L77!!1F        D  #5 ) )E E6xJJE*5(;;  >))f.?.? :8]] 2%BBBB5%>u%E%E :8]] 2%BBBB7""'Fu'M'M" :8]] 2%BBBB7""'DU'K'K" :8]] 2%BBBB*(33v== V8 %X..Ax 9 h ' ' 2 23E F F "(O < <   H H H H H H--e44 4    D 8-B8_-]-].uh6cBcmtuuu ? :#E8S999J8$$.,-H-R-R.,9%SZ[[[  %! !*844* ,)'JJJC5hII>'x===  ** %7xQXYYY %Q9(HMMV9%9UUU$$ KK.x$>>>>>''(:EBB  /u55 "155lCC"sW}}KKOUwNN "CJ  ')) H  e_(()9;STTF#image image_url input_imagerj_anthropic_content_blocksr _multimodal>rr)rrrArK)rrcountrpartptypestashedinners rrr* s E$.sD$9$9Ccggi   tG'4    DdD)) HHV$$E=== 6@d6K6KUcgg1222QUG'4    D$%% $((6*:*:g*E*E '4  W[[%?%? I&& eT " "   dD))dhhv.>.>BX.X.XQJE > !!rct|tstt|Si}|D]\}}|dkr |dkrt|t rg}|D]}t|trY|ddvr,||dddZ||p|||||<t|tr/|dr|dd ||<|||<|||< tt|S) zChar count for token estimation, excluding base64 image data. Base64 images are counted via `_count_image_tokens` instead; including their raw chars here would massively overestimate token usage. rrr>rrrz [stripped])rrr text_summaryr)rrrrrrKrAr)rshadowkvcleanedrs rrrD sv c4 3s88}}F 1 + + +  >>!T"" --D!$---88F++/TTT#NNDHHV4D4D|+\+\]]]]#NN40000t,,,,#q At$$ })=)= EE."55q q F1II s6{{  r) system_prompttoolsrrcd}|r|t|dzdzz }|r|t|z }|r%|tt|dzdzz }|S)uRough token estimate for a full chat-completions request. Includes the major payload buckets Hermes sends to providers: system prompt, conversation messages, and tool schemas. With 50+ tools enabled, schemas alone can add 20-30K tokens — a significant blind spot when only counting messages. Image content is counted at a flat per-image cost (see estimate_messages_tokens_rough). rrr)rrr)rrrtotals restimate_request_tokens_roughre su E/ #m$$q(Q..: /999 , #c%jj//A%!++ Lr)r)rNrO)F)rF)rr)rrNrN)__doc__r!rloggingrrrpathlibrtypingrrrrr urllib.parser rrutilsr r r rr getLogger__name__rrrr frozensetre__annotations__compile IGNORECASEro IPv4Networkr'rtrurvrrwrxrryrzrrrrrrrqrr,rtuplerrrrrbrgrr rrrrr providersr_list_providers_pp get_hostname_hostrrrrrrr)rrr/rGrJrVr`rdrhrrrrrrrrrrrrrrrr rr r"r$r'r)r r-r5r7r8r9r:r>r@rNrYrKrnrrrrrrrrrrs   33333333333333!!!!!! MMMMMMMMMM222222  8 $ $$**&/Y000&&IcN4!bjLM)9(99##&46tCc3h/0555$%E%%%46S$sCx.01666%&U&&&GIS$sDcN/B*C%C DIII8:#T#u*%5:::JJJJJ     c4S>.A)B$ S$sDcN/B*C S S S S S /q1   $-/UE\*/// A g A G AwAwAwAwAwAAwAA  f!A.w/A0F1A2F3A4w5AB6CADFEAAF VGAHwIAJ VKAN gOARvSAT fUAV6WAXvYAZ T[AjykAlmAnYoApqArsAv VwA|G}A~AAA@6AAB FCAN'OAPvQA^y_A` 6aAvVwAx&yAzf{A|T}A~7A@AABCAD fEAF fGAH fIAJ FKAAAN FOAV6WAZ[A^v_AbcAffgAhFiAj kAlFmAnFoAp"6qArfsAtuAv7wAxWyAz{A|F}AA~AAAA`!T#T$TTTT$ < 0#0#00000030S#X0000$Y?*$@ A*$B=C*$DE*$FG*$H(I*$JhK*$L5M*$NXO*$P 2Q*$R.S*$*$$sCx.***\ ;;;;;;  //  ""  /U"222&)h U #/   D sx}$CC:#:$::::"S"C"D"""" 5  5 5  5 5 5 5 5( "" "" " "  """"J555555p;;s;S;(3-;;;;|0c0000  #  S ZbcfZg     S#X eCHo (SV-    =T#s(^= =====DcN=x}====d38nc3hB,d3S#X#67,3,tTWY\T\~,bf,,,,666c4S>>Q9R6666vMMM MM #tCH~  MMMMf c] .;;;;; T#s(^    CsCcC3C4CCCC*SCHSM WS WC WD W W W W  chsm<c](xx xxxxv=%3=%4=%=%=%=%@Cst&444s4S4(SV-4444n///s/S/RZ[_R`////d::#::s:HUXM::::z ;S ;T ; ; ; ; )3 )4 ) ) ) ) ' ' ' ' ' 'scCQYZ]Q^D\\\s\S\ZbcfZg\\\\~#C#C####3#PXY\P]P!! # 11tCH~&.0DcN///),,,,!%0S0T#s(^0000h%' ! c]FGG GGG 8C=#  GGGGX(,$( O$O$ O$O$O$: O$  O$ Tk O$ O$O$O$O$h (,$(  :    Tk  :         3T$sCx.-A3c3333""T#s(^"S"S""""4c3hCH,0 4S>" Dc3h( )   s7L<<MM