mirror of
https://github.com/diegosouzapw/OmniRoute.git
synced 2026-08-02 13:22:11 +03:00
* chore(release): open v3.8.37 development cycle * chore(ci): harden release flow — ratchet decoupling, fast-path drift gates, build-scope guard, heap default (#5054) Implements improvements 1-4 from the v3.8.36 release benchmark (_tasks/release-bench/v3.8.36/PLANO-MELHORIA.md): 1. Quality Ratchet decoupled from flaky coverage (ci.yml): the shard→coverage→ratchet chain meant a single flaky Coverage Shard SKIPPED the whole Quality Ratchet on the release PR (v3.8.36 #4854), so cycle drift only surfaced post-merge in #5029. The job now runs on !cancelled(); coverage download is continue-on-error and the ratchet runs --allow-missing, so the DETERMINISTIC gates (eslint/complexity/cognitive/duplication/ codeql) stay blocking even when coverage is unavailable. 2. Fast-path drift gates (quality.yml PR→release): added check:complexity, check:cognitive-complexity, and a new lightweight check:pack-policy (pack-artifact unexpected-files check WITHOUT a build, via --policy-only) so drift + stray-tarball-file regressions are caught/rebaselined PER-PR instead of cascading onto the release PR. 3. Build heap default 4096→8192 MB (build-next-isolated.mjs): the clean graph peaks ~3.9 GB and brushed the old 4 GB ceiling; 8 GB gives headroom. Comment notes heap is NOT the fix for a poisoned scope (run check:build-scope instead). 4. check:build-scope gate (new): fails if .ts/.tsx/.js/.jsx files in the tsconfig scope exceed a threshold — catches worktrees/cruft leaking into the build scope (the v3.8.36 OOM root cause: 355,215 vs 4,547 files) BEFORE it detonates next build. Wired into the fast-path. * fix(auth): only trust forwarding headers from loopback TCP peers (#4689) Integrated into release/v3.8.37 — loopback-gated forwarding headers (IP spoofing fix). Cherry-picked onto current release tip; ipUtils.test.ts 9/9 green. * fix(codex): treat OAuth 401 as unrecoverable refresh failure (#4686) Integrated into release/v3.8.37 — codex OAuth 401 treated as unrecoverable refresh. Cherry-picked onto release tip; token-refresh-service.test.ts 38/38 green. * fix(translator): preserve reasoning_effort for non-Copilot Responses clients (#4688) Integrated into release/v3.8.37 — preserve reasoning_effort for non-Copilot Responses clients. Cherry-picked onto release tip; tests 47/47 green. * fix(translator): coerce tool descriptions to strings in OpenAI normalization (#4675) Integrated into release/v3.8.37 — coerce tool descriptions to strings in OpenAI normalization. Cherry-picked onto release tip; tests 3/3 green. * feat(sse): x-omniroute-strip-reasoning header to drop reasoning_content (#4678) Integrated into release/v3.8.37 — x-omniroute-strip-reasoning header. Cherry-picked onto release tip (resolved chatCore.ts/headers.ts adjacency conflict, kept resolveCompressionHeader + isStripReasoningRequested); tests 8/8 green. * fix(combo): flatten Anthropic tool messages + tool history to prevent upstream 503 (#4648) Integrated into release/v3.8.37 — flattenToolHistory helper (combo anti-503). Cherry-picked onto release tip; tests 9/9 green. * feat(headroom): proxy lifecycle management + dashboard UI (Docker sidecar supported) (#4649) Integrated into release/v3.8.37 — headroom proxy lifecycle (status/start/stop, local-only + spawn-capable per Rules #15/#17). Cherry-picked onto release tip; lifecycle 7/7 + route-guard 43/43 + check:cycles green. * feat(cli): multi-model support for Factory Droid CLI (#4682) Integrated into release/v3.8.37 — Factory Droid multi-model support. Cherry-picked onto release tip (kept readJsoncConfig + droidCustomModels imports); droid-custom-models 11/11 green. * fix(providers): require Default Model in compatible-provider API-key setup (#4641) Integrated into release/v3.8.37 — require Default Model in compatible-provider API-key setup. Cherry-picked fix + test-move onto release tip (kept release providerSpecificData + QuotaScrapingFields; fixed moved-test import path; baseline rebaseline unneeded, 865<866); UI test 2/2 green. * fix(dashboard): stop double-masking already-masked API key in list (E2E 3/9 regression) (#4671) Integrated into release/v3.8.37 — render server-masked key verbatim (drop redundant maskKey call). Note: release's maskKey already guards '****' (since v3.8.34), so this is a safe simplification; added a contract test pinning the **** passthrough invariant (2/2 green, would fail against the pre-guard maskKey = the historical double-mask bug). * chore(quality): rebaseline file-size for rc17 PR batch drift Own growth from the merged rc17 PRs (#4678/#4686/#4688) at existing chokepoints — cohesive, not extractable: - open-sse/handlers/responseSanitizer.ts 1103->1122 (SanitizeOpenAIResponseOptions + stripReasoning, #4678) - open-sse/services/tokenRefresh.ts 2070->2090 (codex 401 unrecoverable-refresh guard, #4686) - tests/unit/token-refresh-service.test.ts 1322->1353 (401 regression case, #4686) - tests/unit/translator-openai-responses-req.test.ts 1047->1050 (reasoning_effort assertion, #4688) * docs(env): document HEADROOM_URL in .env.example + ENVIRONMENT.md The headroom proxy lifecycle (#4649) reads HEADROOM_URL (src/lib/headroom/detect.ts, default http://localhost:8787) but it was missing from the env contract, tripping check:env-doc-sync. Adds the var to both .env.example (commented, has a default) and the Proxy Health table in ENVIRONMENT.md. * fix(sse): stream writer mock abort() returns a Promise (#4788) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; tests green. * fix(cli): fall back to default data dir when DATA_DIR is not writable (#4767) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; tests green. * fix(oauth): verify Cursor installation on Linux before auto-import (#4770) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; tests green. * fix(sse): track Ollama streaming usage from raw NDJSON chunks (#4754) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; tests green. * fix(sse): strip enumDescriptions from antigravity tool schema (#4740) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; tests green. * fix(sse): include low-level cause details in formatProviderError (#4741) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; tests green. * fix(translator): strip x-anthropic-billing-header in claude-to-openai (#4728) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; tests green. * fix(sse): gate Kiro image attachments behind a Claude-capability check (#4763) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; tests green. * fix(sse): read Antigravity usage from the response.usageMetadata envelope (#4785) Integrated into release/v3.8.37 — Antigravity response.usageMetadata envelope. Cherry-picked onto release tip (resolved test-tail adjacency with #4754 Ollama block); usage-extractor 23/23 green. * fix(api): fall back to existing access token for any OAuth provider on refresh failure (#4786) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; tests green. * fix(cli): verify launchd registration + skip self-SIGTERM in macOS autostart (#4765) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; tests green. * fix(executors): anthropic-compatible-* gateways get Bearer alongside x-api-key (#4729) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; tests green. * fix(sse): json_schema fallback for OpenAI-compatible providers (#4766) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; tests green. * fix(sse): use workos auth token shape for cline (#4787) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; tests green. * feat(sse): parse Gemini CLI 429 retryDelay from structured RetryInfo (#4738) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; tests green. * fix(sse): finalize tool_calls finish_reason on early stream end in OpenAI Responses translator (#4764) Integrated into release/v3.8.37 — computeFinishReason finalizes tool_calls on early stream end (Responses translator). Cherry-picked onto release tip; responses-translation-fixes 29/29 green. * test(sse): golden-lock provider.ts translate-path across all providers (#4734) Integrated into release/v3.8.37 — golden-lock for provider.ts translate-path. Cherry-picked onto release tip; snapshot regenerated against the current provider set (UPDATE_GOLDEN=1, 167 entries); golden test 3/3 deterministic. * chore(quality): rebaseline file-size for rc17 leva2 PR batch drift Own growth from the merged leva2 PRs (cohesive, not extractable): - src/lib/usage/providerLimits.ts 950->955 (#4786) - open-sse/executors/default.ts NEW frozen @828 (#4729 + #4766 + #4787 header branches) - open-sse/translator/request/openai-to-kiro.ts 807->814 (#4763) - open-sse/translator/response/openai-responses.ts 923->937 (#4764) - tests/unit/executor-default-base.test.ts 1339->1440 (#4766) - tests/unit/translator-openai-to-kiro.test.ts 918->980 (#4763) * fix(dashboard): align Engine Combos editor engines with API schema (#4955) (#5062) The named-combos pipeline dropdown offered four engines (headroom, session-dedup, ccr, llmlingua) that stackedPipelineStepSchema rejects, so selecting one made PUT /api/context/combos/[id] return HTTP 400 while saveCombo swallowed the non-OK response (if (!res.ok) return). Editing the default 'Standard Savings' combo and changing an engine reproduced the 400. - Add canonical STACKED_PIPELINE_ENGINE_INTENSITIES next to the schema as the single source of truth; the client dropdown imports it so it can never drift from the discriminated union the API validates against. - Surface save errors and empty-name/empty-pipeline validation in the editor instead of failing silently. - Add a parity unit test asserting the UI engine map equals the schema union and that every (engine, intensity) the UI emits is accepted. * fix(sse): filter nameless hosted tools when converting Responses API to Chat format (#4789) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(dashboard): keep desktop sidebar visible via explicit CSS class (#4812) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): strip enumDescriptions from Antigravity tool schemas (#4813) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(dashboard): resolve passthrough model aliases by providerId in ModelSelectModal (#4815) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(oauth): allow per-connection refresh lead-time override via providerSpecificData.refreshLeadMs (#4818) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): strip X-Stainless-* headers and normalize SDK User-Agent for OpenAI-compatible endpoints (#4820) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): strip Gemini built-in tools when functionDeclarations present in Antigravity envelope (#4821) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(api): surface a Docker-localhost hint on provider-node validation connection errors (#4822) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): resolve bare model names to connection defaultModel before upstream calls (#4825) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(build): trace-include sql.js sql-wasm.wasm in standalone bundle (#4839) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): strip Composer <|final|> sentinel markers leaking after Composer reasoning (#4842) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(config): sync full SiliconFlow model list into registry (#4844) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): close reasoning before message content in Responses stream (#4848) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): reject unsupported Kiro [1m] context suffix (#4816) Integrated into release/v3.8.37 — cherry-picked onto release tip; test-tail conflict with #4763 resolved (kept both image + [1m] test blocks); CHANGELOG re-merged; 29/29 green. * fix(db): validate HuggingFace tokens via whoami-v2 auth probe (#4819) Integrated into release/v3.8.37 — defining commit re-homed onto the god-file-split validation module (validateHuggingFaceProvider in validation/openaiFormat.ts + map wiring); 115/115 green. * fix(sse): make anthropic-version default-guard case-insensitive (#4823) Integrated into release/v3.8.37 — conflict with #4729 Bearer-fallback resolved (kept both Bearer fallback + case-insensitive anthropic-version guard); 48/48 green. * fix(sse): sanitize Kiro tool schemas to avoid 400 "Improperly formed request" (#4847) Integrated into release/v3.8.37 — conflict in kiro-to-openai.ts resolved (kept release fallbackToolCallId + adopted #1375 toolNameMap remap); 7/7 green. * feat(sse): add GPT-4 to the GitHub Copilot provider (#4798) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * feat(sse): add GPT-4o mini to GitHub Copilot provider (#4797) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * feat(api): add MiniMax-M3 pricing row (#4814) Integrated into release/v3.8.37 — pricing row re-homed onto god-file-split pricing/regional.ts (pricing.ts is now a barrel); 4/4 green. * fix(cli): save runtime deps with --save-exact so a sibling install can't prune them (#4841) Integrated into release/v3.8.37 — trayRuntime conflict resolved (kept release SYSTRAY_SPEC + added --save-exact); 2/2 green. * fix(sse): preserve required fields in antigravity tool schemas (#4843) Integrated into release/v3.8.37 — conflict resolved (kept #4740/#4813 enumDescriptions strip + typed normalizeSchemaTypes, added required-preservation helpers; test-tail merged keeping both enumDescriptions + required tests); 7/7 green. * chore(quality): rebaseline file-size for rc17b leva3 PR batch drift * fix(sse): strip reasoning blobs from agentic context to prevent O(n^2) token growth (#4849) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): unwrap Qoder HTTP 200 SSE error envelope so fallback can trigger (#4850) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): strip temperature for Claude models with extended thinking (#4853) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): emit valid concatenable kiro tool_calls.arguments deltas (#4855) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * feat(sse): add toggleable tool-source diagnostics (#4856) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): redact api key from the AUTH debug log in the chat handler (#4858) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): forward AI SDK image parts in Responses translator (#4859) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): resolve custom combos by id and case-insensitive name (#4446) (#4869) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): exclude WS bridge controller-closed error from provider breaker (#4602) (#4870) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * feat(providers): add xAI Grok inbound translators and thinking patcher (#4910) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * feat(embeddings): add dimensions override field to embedding combos (#4913) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * feat(oauth): Codex bulk-import endpoint — POST /api/oauth/codex/import (#4914) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(antigravity): retry transient upstream failures (#4941) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): surface malformed HTTP-200 upstream responses (#4942) Integrated into release/v3.8.37 — cherry-picked defining commit onto release tip; CHANGELOG re-merged; tests green. * fix(sse): normalize Codex custom tools (apply_patch) to { input: string } schema (#4862) Integrated into release/v3.8.37 — conflict in request/openai-responses.ts resolved (kept #4789 nameless-tool skip + added #1007 custom-tool {input:string} normalization); 48/48 green incl. #4789/#4859 regression. * fix(sse): dense, deterministic output ordering in Responses API response.completed (#4906) Integrated into release/v3.8.37 — manual integration with #4862 in response/openai-responses.ts (custom-tool funcItem + dense recordCompletedItem). Fixed a latent #4848 interaction: the close-reasoning-before-message guard force-closed <think>-tag reasoning prematurely, which dense output (#4906) then snapshotted as a partial buffer ("plan" vs "planning") — scoped the guard to native reasoning_content (!inThinking) in BOTH transformer + translator paths. Full Responses suite 203/203 green incl. #4848/#4862 regression. * feat(sse): auto-promote successful combo model to position #1 (#4852) Integrated into release/v3.8.37 — dropped the stale file-size-baseline.json hunk (re-derived against the rc17b rebaseline); code+test applied clean; 13/13 green. * feat(providers): add Pioneer AI (Fastino Labs) provider (#4909) Integrated into release/v3.8.37 — providers.ts apikey block re-homed onto god-file-split src/shared/constants/providers/apikey/frontier-labs.ts (inline APIKEY_PROVIDERS no longer exists); registry/pioneer + providers/index.ts applied clean; 6/6 green. * add DGrid AI gateway provider (#4931) Integrated into release/v3.8.37 — rebased the contributor's commit onto the release tip; providers.ts god-file-split conflict resolved by relocating the dgrid APIKEY_PROVIDERS entry into apikey/gateways.ts; CHANGELOG added. 7/7 green. Thanks @dgridOP! * chore(quality): rebaseline file-size for rc17b leva4 PR batch drift * docs(routing): sync combo strategy docs for Fusion (17 strategies) (#5067) Fusion (16th strategy, panel fan-out + judge synthesis) and headroom shipped but the strategy-count docs were stale (14/15) and omitted both. Update every combo-strategy reference to the canonical 17, add fusion + headroom to all strategy tables, and add a dedicated Fusion section to AUTO-COMBO.md documenting judgeModel / fusionTuning config + an example. - CLAUDE.md, README.md, FEATURES.md, RESILIENCE_GUIDE.md, ARCHITECTURE.md, OPEN_SSE_ARCHITECTURE.md, OMNIROUTE_VS_ALTERNATIVES.md, docs/README.md, request-pipeline.mmd: 14/15 -> 17, list fusion + headroom - docs/routing/AUTO-COMBO.md: strategy table + new Fusion strategy section - docs/openapi.yaml: add reset-window, headroom, fusion to the strategy enum * fix(oauth): classify /api/oauth/cursor/auto-import as local-only (route-guard) (#5070) The Cursor auto-import route runs execFile("which", ["cursor"]) to verify a local Cursor install before importing credentials — a child-process spawn. The check:route-guard-membership gate (Hard Rules #15/#17) flagged it as an unclassified spawn-capable route: reachable past the loopback gate, an RCE-via-tunnel surface (a leaked JWT over a tunnel could trigger the spawn). Classify the specific path in LOCAL_ONLY_API_PREFIXES so loopback enforcement runs unconditionally before any auth check. Scoped to the exact path — the rest of /api/oauth/ (browser redirect/callback flows) stays remote-reachable. TDD: added a failing-then-passing assertion in route-guard-local-prefix.test.ts (classification + an over-broadening guard proving sibling OAuth paths stay remote). check:route-guard-membership now reports 0 new gaps. * chore(release): v3.8.37 — 2026-06-26 --------- Co-authored-by: dgridOP <dgrid_op@outlook.com>
549 lines
16 KiB
TypeScript
549 lines
16 KiB
TypeScript
import { trackPendingRequest } from "@/lib/usageDb";
|
|
import { STREAM_IDLE_TIMEOUT_MS } from "../config/constants.ts";
|
|
import { FORMATS } from "../translator/formats.ts";
|
|
import { PENDING_REQUEST_CLEARED_MARKER } from "./stream.ts";
|
|
|
|
// Stream handler with disconnect detection - shared for all providers
|
|
|
|
const DISCONNECT_ABORT_DELAY_MS = 2_000;
|
|
|
|
// Default budget for the pipeWithDisconnect raw-upstream stall watchdog.
|
|
// Inherits STREAM_IDLE_TIMEOUT_MS so a single env knob still governs the
|
|
// max time we tolerate silence from upstream. Reasoning models (Claude
|
|
// thinking, Kiro EventStream binary frames) emit zero post-transform
|
|
// output for long stretches while raw bytes keep arriving — measuring
|
|
// stall on the transform output false-positives on those streams, so
|
|
// the watchdog must track upstream byte activity instead. Ported from
|
|
// decolua/9router#1243.
|
|
const DEFAULT_STREAM_STALL_TIMEOUT_MS = STREAM_IDLE_TIMEOUT_MS;
|
|
|
|
type StreamDisconnectEvent = {
|
|
reason: string;
|
|
duration: number;
|
|
};
|
|
|
|
type StreamErrorEvent = {
|
|
error: unknown;
|
|
message: string;
|
|
statusCode: number;
|
|
duration: number;
|
|
};
|
|
|
|
type StreamControllerOptions = {
|
|
onDisconnect?: (event: StreamDisconnectEvent) => void;
|
|
onError?: (event: StreamErrorEvent) => boolean | void;
|
|
provider?: string;
|
|
model?: string;
|
|
connectionId?: string | null;
|
|
clientResponseFormat?: string | null;
|
|
};
|
|
|
|
type StreamController = ReturnType<typeof createStreamController>;
|
|
|
|
type StreamErrorStatusKind = "rate_limit" | "authentication" | "permission" | "client" | "server";
|
|
|
|
type StreamErrorStatusMapping = {
|
|
responses: {
|
|
type: string;
|
|
code: string;
|
|
};
|
|
claude: {
|
|
type: string;
|
|
};
|
|
};
|
|
|
|
function isResponsesClientFormat(clientResponseFormat?: string | null): boolean {
|
|
return (
|
|
clientResponseFormat === FORMATS.OPENAI_RESPONSES ||
|
|
clientResponseFormat === FORMATS.OPENAI_RESPONSE
|
|
);
|
|
}
|
|
|
|
function getStreamErrorStatusKind(statusCode: number): StreamErrorStatusKind {
|
|
if (statusCode === 429) return "rate_limit";
|
|
if (statusCode === 401) return "authentication";
|
|
if (statusCode === 403) return "permission";
|
|
if (statusCode >= 400 && statusCode < 500) return "client";
|
|
return "server";
|
|
}
|
|
|
|
function getStreamErrorStatusMapping(statusCode: number): StreamErrorStatusMapping {
|
|
switch (getStreamErrorStatusKind(statusCode)) {
|
|
case "rate_limit":
|
|
return {
|
|
responses: { type: "rate_limit_error", code: "rate_limit_exceeded" },
|
|
claude: { type: "rate_limit_error" },
|
|
};
|
|
case "authentication":
|
|
return {
|
|
responses: { type: "authentication_error", code: "invalid_authentication" },
|
|
claude: { type: "authentication_error" },
|
|
};
|
|
case "permission":
|
|
return {
|
|
responses: { type: "authentication_error", code: "permission_denied" },
|
|
claude: { type: "permission_error" },
|
|
};
|
|
case "client":
|
|
return {
|
|
responses: { type: "invalid_request_error", code: "bad_request" },
|
|
claude: { type: "invalid_request_error" },
|
|
};
|
|
case "server":
|
|
return {
|
|
responses: { type: "server_error", code: "server_error" },
|
|
claude: { type: "api_error" },
|
|
};
|
|
default:
|
|
return {
|
|
responses: { type: "server_error", code: "server_error" },
|
|
claude: { type: "api_error" },
|
|
};
|
|
}
|
|
}
|
|
|
|
function encodeSseEvent(
|
|
data: unknown,
|
|
{
|
|
event,
|
|
includeDone = false,
|
|
}: {
|
|
event?: string;
|
|
includeDone?: boolean;
|
|
} = {}
|
|
) {
|
|
if (event && /[\r\n]/.test(event)) {
|
|
throw new Error("SSE event names must not contain newlines");
|
|
}
|
|
|
|
const encoder = new TextEncoder();
|
|
const prefix = event ? `event: ${event}\n` : "";
|
|
const chunks = [encoder.encode(`${prefix}data: ${JSON.stringify(data)}\n\n`)];
|
|
if (includeDone) {
|
|
chunks.push(encoder.encode("data: [DONE]\n\n"));
|
|
}
|
|
return chunks;
|
|
}
|
|
|
|
// Get HH:MM:SS timestamp
|
|
function getTimeString() {
|
|
return new Date().toLocaleTimeString("en-US", {
|
|
hour12: false,
|
|
hour: "2-digit",
|
|
minute: "2-digit",
|
|
second: "2-digit",
|
|
});
|
|
}
|
|
|
|
function isPendingRequestClearedError(error: unknown): boolean {
|
|
return (
|
|
!!error &&
|
|
typeof error === "object" &&
|
|
(error as Record<string, unknown>)[PENDING_REQUEST_CLEARED_MARKER] === true
|
|
);
|
|
}
|
|
|
|
function getErrorMessage(error: unknown): string {
|
|
if (error instanceof Error && error.message) return error.message;
|
|
if (typeof error === "string" && error.trim().length > 0) return error;
|
|
return "Upstream stream error";
|
|
}
|
|
|
|
function getErrorStatusCode(error: unknown): number {
|
|
if (error && typeof error === "object" && "statusCode" in error) {
|
|
const statusCode = Number((error as { statusCode?: unknown }).statusCode);
|
|
if (Number.isFinite(statusCode) && statusCode >= 400 && statusCode <= 599) {
|
|
return statusCode;
|
|
}
|
|
}
|
|
return 502;
|
|
}
|
|
|
|
/**
|
|
* Create stream controller with abort and disconnect detection
|
|
* @param {object} options
|
|
* @param {function} options.onDisconnect - Callback when client disconnects
|
|
* @param {object} options.log - Logger instance
|
|
* @param {string} options.provider - Provider name
|
|
* @param {string} options.model - Model name
|
|
*/
|
|
/** @param {StreamControllerOptions} options */
|
|
export function createStreamController({
|
|
onDisconnect,
|
|
onError,
|
|
provider,
|
|
model,
|
|
connectionId,
|
|
clientResponseFormat,
|
|
}: StreamControllerOptions = {}) {
|
|
const abortController = new AbortController();
|
|
const startTime = Date.now();
|
|
let disconnected = false;
|
|
let abortTimeout: ReturnType<typeof setTimeout> | null = null;
|
|
let pendingRequestCleared = false;
|
|
|
|
const logStream = (status) => {
|
|
const duration = Date.now() - startTime;
|
|
const p = provider?.toUpperCase() || "UNKNOWN";
|
|
console.log(
|
|
`[${getTimeString()}] 🌊 [STREAM] ${p} | ${model || "unknown"} | ${duration}ms | ${status}`
|
|
);
|
|
};
|
|
|
|
const clearPendingRequest = (error?: unknown) => {
|
|
if (pendingRequestCleared) return;
|
|
if (
|
|
error &&
|
|
typeof error === "object" &&
|
|
(error as Record<string, unknown>)[PENDING_REQUEST_CLEARED_MARKER] === true
|
|
) {
|
|
pendingRequestCleared = true;
|
|
return;
|
|
}
|
|
|
|
pendingRequestCleared = true;
|
|
if (!model && !provider && !connectionId) return;
|
|
try {
|
|
trackPendingRequest(model || "", provider || "", connectionId ?? null, false);
|
|
} catch {}
|
|
};
|
|
|
|
return {
|
|
signal: abortController.signal,
|
|
startTime,
|
|
|
|
isConnected: () => !disconnected,
|
|
|
|
// Call when client disconnects
|
|
handleDisconnect: (reason = "client_closed") => {
|
|
if (disconnected) return;
|
|
disconnected = true;
|
|
|
|
logStream(`disconnect: ${reason}`);
|
|
|
|
// Decrement pending request counter — the TransformStream flush() won't
|
|
// fire when the client aborts mid-stream, so we must clean up here.
|
|
clearPendingRequest();
|
|
|
|
// Delay abort to allow cleanup
|
|
abortTimeout = setTimeout(() => {
|
|
abortController.abort();
|
|
}, DISCONNECT_ABORT_DELAY_MS);
|
|
|
|
onDisconnect?.({ reason, duration: Date.now() - startTime });
|
|
},
|
|
|
|
// Call when stream completes normally
|
|
handleComplete: () => {
|
|
if (disconnected) return;
|
|
disconnected = true;
|
|
|
|
logStream("complete");
|
|
|
|
if (abortTimeout) {
|
|
clearTimeout(abortTimeout);
|
|
abortTimeout = null;
|
|
}
|
|
},
|
|
|
|
// Call on error
|
|
handleError: (error: unknown) => {
|
|
if (abortTimeout) {
|
|
clearTimeout(abortTimeout);
|
|
abortTimeout = null;
|
|
}
|
|
|
|
const alreadyCleared = isPendingRequestClearedError(error);
|
|
let handled = false;
|
|
if (!alreadyCleared) {
|
|
try {
|
|
handled =
|
|
onError?.({
|
|
error,
|
|
message: getErrorMessage(error),
|
|
statusCode: getErrorStatusCode(error),
|
|
duration: Date.now() - startTime,
|
|
}) === true;
|
|
} catch {}
|
|
}
|
|
|
|
if (!handled) {
|
|
clearPendingRequest(error);
|
|
} else {
|
|
pendingRequestCleared = true;
|
|
}
|
|
|
|
if (error instanceof Error && error.name === "AbortError") {
|
|
logStream("aborted");
|
|
return;
|
|
}
|
|
|
|
if (error instanceof Error) {
|
|
logStream(`error: ${error.message}`);
|
|
return;
|
|
}
|
|
logStream("error: unknown");
|
|
},
|
|
|
|
abort: () => abortController.abort(),
|
|
clientResponseFormat,
|
|
};
|
|
}
|
|
|
|
function buildStreamErrorChunks(
|
|
errorMsg: string,
|
|
statusCode: number,
|
|
clientResponseFormat?: string | null
|
|
) {
|
|
const statusMapping = getStreamErrorStatusMapping(statusCode);
|
|
|
|
if (isResponsesClientFormat(clientResponseFormat)) {
|
|
const errorEvent = {
|
|
type: "response.failed",
|
|
response: {
|
|
id: null,
|
|
status: "failed",
|
|
error: {
|
|
message: errorMsg,
|
|
type: statusMapping.responses.type,
|
|
code: statusMapping.responses.code,
|
|
},
|
|
},
|
|
};
|
|
|
|
return encodeSseEvent(errorEvent, { event: "response.failed" });
|
|
}
|
|
|
|
if (clientResponseFormat === FORMATS.CLAUDE) {
|
|
const errorEvent = {
|
|
type: "error",
|
|
error: {
|
|
type: statusMapping.claude.type,
|
|
message: errorMsg,
|
|
},
|
|
};
|
|
|
|
return encodeSseEvent(errorEvent, { event: "error" });
|
|
}
|
|
|
|
const errorEvent = {
|
|
object: "chat.completion.chunk",
|
|
choices: [
|
|
{
|
|
index: 0,
|
|
delta: {},
|
|
finish_reason: "error",
|
|
},
|
|
],
|
|
error: {
|
|
message: errorMsg,
|
|
type: statusMapping.responses.type,
|
|
code: statusMapping.responses.code,
|
|
},
|
|
};
|
|
|
|
return encodeSseEvent(errorEvent, { includeDone: true });
|
|
}
|
|
|
|
/**
|
|
* Minimal `writable` half used by `pipeWithDisconnect`. The real writable is
|
|
* driven entirely by the upstream-piped readable, so the writer only needs an
|
|
* `abort()` hook for `createDisconnectAwareStream`'s `cancel()` path.
|
|
*
|
|
* `abort()` returns `Promise<void>` to match the native
|
|
* `WritableStreamDefaultWriter.abort()` contract — `cancel()` (and any caller
|
|
* that awaits the writer) gets a real thenable instead of `undefined`, which
|
|
* keeps abort/error handling clean. Ported from decolua/9router@6b624af4.
|
|
*/
|
|
export function createNoopAbortWritable(): {
|
|
getWriter: () => { abort: () => Promise<void> };
|
|
} {
|
|
return { getWriter: () => ({ abort: () => Promise.resolve() }) };
|
|
}
|
|
|
|
/**
|
|
* Create transform stream with disconnect detection
|
|
* Wraps existing transform stream and adds abort capability
|
|
*/
|
|
export function createDisconnectAwareStream(transformStream, streamController) {
|
|
const reader = transformStream.readable.getReader();
|
|
const writer = transformStream.writable.getWriter();
|
|
|
|
return new ReadableStream(
|
|
{
|
|
async pull(controller) {
|
|
if (!streamController.isConnected()) {
|
|
controller.close();
|
|
return;
|
|
}
|
|
|
|
try {
|
|
const { done, value } = await reader.read();
|
|
if (done) {
|
|
streamController.handleComplete();
|
|
controller.close();
|
|
return;
|
|
}
|
|
controller.enqueue(value);
|
|
} catch (error) {
|
|
streamController.handleError(error);
|
|
|
|
// T35: Encapsulate mid-stream errors as SSE events instead of abruptly aborting
|
|
// This prevents TransferEncodingError on the client side
|
|
const errorMsg = getErrorMessage(error);
|
|
const statusCode = getErrorStatusCode(error);
|
|
|
|
for (const chunk of buildStreamErrorChunks(
|
|
errorMsg,
|
|
statusCode,
|
|
streamController.clientResponseFormat
|
|
)) {
|
|
controller.enqueue(chunk);
|
|
}
|
|
|
|
controller.close();
|
|
}
|
|
},
|
|
|
|
cancel(reason) {
|
|
streamController.handleDisconnect(reason || "cancelled");
|
|
reader.cancel();
|
|
setTimeout(() => {
|
|
writer.abort();
|
|
}, DISCONNECT_ABORT_DELAY_MS).unref?.();
|
|
},
|
|
},
|
|
{ highWaterMark: 16384 }
|
|
);
|
|
}
|
|
|
|
/**
|
|
* Pipe provider response through transform with disconnect detection.
|
|
*
|
|
* Stall watchdog tracks raw upstream byte activity, not transform output.
|
|
* Reasoning models (Claude thinking via Kiro, etc.) can produce zero SSE
|
|
* output for long stretches while partial EventStream frames keep arriving;
|
|
* measuring stall on the transform output caused false stalls. Any upstream
|
|
* chunk resets the timer. If no bytes arrive for `stallTimeoutMs`, the
|
|
* stream surfaces a "stream stall timeout" error and aborts.
|
|
*
|
|
* Ported from decolua/9router#1243 by @zakirkun.
|
|
*
|
|
* @param providerResponse - Response from provider
|
|
* @param transformStream - Transform stream for SSE
|
|
* @param streamController - Stream controller from createStreamController
|
|
* @param opts.stallTimeoutMs - Override the stall budget (defaults to
|
|
* STREAM_IDLE_TIMEOUT_MS / DEFAULT_STREAM_STALL_TIMEOUT_MS). `0` disables
|
|
* the watchdog.
|
|
*/
|
|
export function pipeWithDisconnect(
|
|
providerResponse: Response,
|
|
transformStream: TransformStream<Uint8Array, Uint8Array>,
|
|
streamController: StreamController,
|
|
opts: { stallTimeoutMs?: number } = {}
|
|
) {
|
|
const stallTimeoutMs = opts.stallTimeoutMs ?? DEFAULT_STREAM_STALL_TIMEOUT_MS;
|
|
|
|
// Watchdog disabled — preserve legacy behavior verbatim.
|
|
if (!stallTimeoutMs || stallTimeoutMs <= 0) {
|
|
const transformedBody = providerResponse.body.pipeThrough(transformStream);
|
|
return createDisconnectAwareStream(
|
|
{ readable: transformedBody, writable: createNoopAbortWritable() },
|
|
streamController
|
|
);
|
|
}
|
|
|
|
let stallTimer: ReturnType<typeof setTimeout> | null = null;
|
|
// Captured on the upstream tap's `start`, used by the watchdog to error the
|
|
// pipeline so the downstream reader unblocks and emits a clean SSE error
|
|
// event. Without this, aborting the AbortController alone does not unblock
|
|
// a `reader.read()` already suspended on the transform pipe — the request
|
|
// would hang until the upstream finally closed the socket.
|
|
let upstreamTapController: TransformStreamDefaultController<Uint8Array> | null = null;
|
|
// Set when the watchdog fires so the downstream pull() catch (which sees
|
|
// the same error propagated through the pipeline) does not call
|
|
// handleError a second time — pending-cleanup is idempotent but onError
|
|
// callbacks should fire once per error.
|
|
let stallFired = false;
|
|
|
|
const clearStall = () => {
|
|
if (stallTimer) {
|
|
clearTimeout(stallTimer);
|
|
stallTimer = null;
|
|
}
|
|
};
|
|
const armStall = () => {
|
|
clearStall();
|
|
stallTimer = setTimeout(() => {
|
|
stallTimer = null;
|
|
stallFired = true;
|
|
const stallError = new Error("stream stall timeout");
|
|
// Notify the controller (onError callback + pending-request cleanup).
|
|
try {
|
|
streamController.handleError?.(stallError);
|
|
} catch {}
|
|
// Error the pipeline so the downstream reader unblocks. createDisconnect-
|
|
// AwareStream's catch block translates this into buildStreamErrorChunks
|
|
// (sanitized SSE error event with finish_reason:"error", per the format).
|
|
try {
|
|
upstreamTapController?.error(stallError);
|
|
} catch {}
|
|
// Abort the underlying fetch so upstream releases the connection.
|
|
try {
|
|
streamController.abort?.();
|
|
} catch {}
|
|
}, stallTimeoutMs);
|
|
};
|
|
|
|
// Wrap controller so every termination path clears the stall timer.
|
|
// Without this, abort/complete/error/disconnect paths leave the timer armed
|
|
// and a stale abort could fire after the request has already ended.
|
|
const wrappedController: StreamController = {
|
|
...streamController,
|
|
handleComplete: () => {
|
|
clearStall();
|
|
streamController.handleComplete();
|
|
},
|
|
handleError: (e: unknown) => {
|
|
clearStall();
|
|
// Watchdog already fired its own handleError — the inner pull() catch
|
|
// sees the same error propagated through the pipeline; suppress the
|
|
// duplicate to keep onError callbacks single-fire.
|
|
if (stallFired) return;
|
|
streamController.handleError(e);
|
|
},
|
|
handleDisconnect: (reason?: string) => {
|
|
clearStall();
|
|
streamController.handleDisconnect(reason);
|
|
},
|
|
abort: () => {
|
|
clearStall();
|
|
streamController.abort();
|
|
},
|
|
};
|
|
|
|
// Inert tap that resets the stall timer on every raw upstream byte chunk.
|
|
// Sits between the provider body and the SSE transform so reasoning models
|
|
// that buffer many raw bytes into a single emitted event do not look
|
|
// stalled to the watchdog.
|
|
const upstreamTap = new TransformStream<Uint8Array, Uint8Array>({
|
|
start(controller) {
|
|
upstreamTapController = controller;
|
|
armStall();
|
|
},
|
|
transform(chunk, controller) {
|
|
armStall();
|
|
controller.enqueue(chunk);
|
|
},
|
|
flush() {
|
|
clearStall();
|
|
},
|
|
});
|
|
|
|
const transformedBody = providerResponse.body.pipeThrough(upstreamTap).pipeThrough(transformStream);
|
|
return createDisconnectAwareStream(
|
|
{ readable: transformedBody, writable: createNoopAbortWritable() },
|
|
wrappedController
|
|
);
|
|
}
|