@vpxa/aikit 0.1.128 → 0.1.130

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@vpxa/aikit",
-  "version": "0.1.128",
+  "version": "0.1.130",
   "type": "module",
   "description": "Local-first AI developer toolkit — knowledge base, code analysis, context management, and developer tools for LLM agents",
   "license": "MIT",

package/packages/core/dist/index.js

@@ -1 +1 @@
-import{basename as e,extname as t,join as n,resolve as r}from"node:path";import{createHash as i}from"node:crypto";import{appendFileSync as a,closeSync as o,constants as s,existsSync as c,mkdirSync as l,openSync as ee,readFileSync as u,readdirSync as d,renameSync as f,statSync as te,unlinkSync as p,writeFileSync as m}from"node:fs";import{homedir as h}from"node:os";const g={ai:`.ai`,aiContext:`.ai/context`,aiCurated:`.ai/curated`,restorePoints:`.ai/restore-points`,data:`.aikit-data`,state:`.aikit-state`,logs:`.aikit-state/logs`,brainstorm:`.brainstorm`,handoffs:`.handoffs`},_={root:`.aikit-data`,registry:`registry.json`},v={markdown:{max:1500,min:100},code:{max:2e3,min:50},config:{max:3e3,min:50},default:{max:1500,min:100,overlap:200}},y={model:`mixedbread-ai/mxbai-embed-large-v1`,dimensions:512},b={backend:`sqlite-vec`,path:g.data,tableName:`knowledge`},x={maxFileSizeBytes:1e6,maxCuratedFileSizeBytes:5e4},S={maxResults:10,minScore:.25},C=/^[a-z][a-z0-9-]*$/,w=[`decisions`,`patterns`,`troubleshooting`,`conventions`,`architecture`],T={".ts":`code-typescript`,".tsx":`code-typescript`,".mts":`code-typescript`,".cts":`code-typescript`,".js":`code-javascript`,".jsx":`code-javascript`,".mjs":`code-javascript`,".cjs":`code-javascript`,".py":`code-python`,".json":`config-json`,".yaml":`config-yaml`,".yml":`config-yaml`,".toml":`config-toml`,".env":`config-env`,".md":`markdown`,".mdx":`markdown`},E=[/\.test\.[jt]sx?$/,/\.spec\.[jt]sx?$/,/(^|\/)__tests__\//,/(^|\/)test\//,/(^|\/)tests\//,/(^|\/)spec\//,/(^|\/)fixtures\//],ne=[/\.stack\.[jt]s$/,/(^|\/)stacks\//,/(^|\/)constructs\//,/cdk\.json$/];function re(n){let r=t(n).toLowerCase(),i=e(n).toLowerCase();return n.includes(`${g.aiContext}/`)?`produced-knowledge`:n.includes(`${g.aiCurated}/`)?`curated-knowledge`:E.some(e=>e.test(n))?`test-code`:ne.some(e=>e.test(n))?`cdk-stack`:r in T?T[r]:i.startsWith(`.env`)?`config-env`:[`.go`,`.rs`,`.java`,`.rb`,`.php`,`.sh`,`.ps1`,`.sql`,`.graphql`,`.proto`,`.css`,`.scss`,`.less`,`.html`,`.htm`,`.vue`,`.svelte`,`.astro`,`.hbs`,`.ejs`,`.svg`].includes(r)?`code-other`:`unknown`}const D={"code-typescript":`source`,"code-javascript":`source`,"code-python":`source`,"code-other":`source`,"cdk-stack":`source`,"test-code":`test`,markdown:`documentation`,documentation:`documentation`,"curated-knowledge":`documentation`,"produced-knowledge":`documentation`,"config-json":`config`,"config-yaml":`config`,"config-toml":`config`,"config-env":`config`,unknown:`source`};function ie(e){return D[e]??`source`}function O(e){return Object.entries(D).filter(([,t])=>t===e).map(([e])=>e)}var k=class extends Error{code;constructor(e,t,n){super(e,n===void 0?void 0:{cause:n}),this.code=t,this.name=`AikitError`}},A=class extends k{constructor(e,t){super(e,`EMBEDDING_ERROR`,t),this.name=`EmbeddingError`}},j=class extends k{constructor(e,t){super(e,`STORE_ERROR`,t),this.name=`StoreError`}},M=class extends k{constructor(e,t){super(e,`INDEX_ERROR`,t),this.name=`IndexError`}},N=class extends k{constructor(e,t){super(e,`CONFIG_ERROR`,t),this.name=`ConfigError`}};function P(){return process.env.AIKIT_GLOBAL_DATA_DIR??r(h(),_.root)}function F(t){let n=r(t);return`${e(n).toLowerCase().replace(/[^a-z0-9-]/g,`-`)||`workspace`}-${i(`sha256`).update(n).digest(`hex`).slice(0,8)}`}function I(){let e=r(P(),_.registry);if(!c(e))return{version:1,workspaces:{}};let t=u(e,`utf-8`);try{return JSON.parse(t)}catch{return{version:1,workspaces:{}}}}function L(e,t=5e3){let n=`${e}.lock`,r=Date.now()+t,i=10;for(;Date.now()<r;)try{let e=ee(n,s.O_CREAT|s.O_EXCL|s.O_WRONLY);return m(e,`${process.pid}\n`),o(e),n}catch(e){if(e.code!==`EEXIST`)throw e;try{let{mtimeMs:e}=te(n);if(Date.now()-e>3e4){p(n);continue}}catch{}let t=new SharedArrayBuffer(4);Atomics.wait(new Int32Array(t),0,0,i),i=Math.min(i*2,200)}throw Error(`Failed to acquire registry lock after ${t}ms`)}function R(e){try{p(e)}catch{}}function z(e){let t=P();l(t,{recursive:!0});let n=r(t,_.registry),i=L(n);try{let t=`${n}.tmp`;m(t,JSON.stringify(e,null,2),`utf-8`),f(t,n)}finally{R(i)}}function B(e){let t=I(),n=F(e),i=new Date().toISOString();return t.workspaces[n]?t.workspaces[n].lastAccessedAt=i:t.workspaces[n]={partition:n,workspacePath:r(e),registeredAt:i,lastAccessedAt:i},l(U(n),{recursive:!0}),z(t),t.workspaces[n]}function V(e){let t=I(),n=F(e);return t.workspaces[n]}function H(){let e=I();return Object.values(e.workspaces)}function U(e){return r(P(),e)}function W(){return c(r(P(),_.registry))}function G(e){return W()?r(U(B(e).partition),`state`):r(e,g.state)}const K={debug:0,info:1,warn:2,error:3},q=[];let J=process.env.AIKIT_LOG_LEVEL??`info`,Y=process.env.AIKIT_LOG_FILE_SINK===`true`||process.env.AIKIT_LOG_FILE_SINK!==`false`&&!process.env.VITEST&&process.env.NODE_ENV!==`test`;function ae(){return Y?process.env.VITEST||process.env.NODE_ENV===`test`?process.env.AIKIT_LOG_FILE_SINK===`true`:!0:!1}let X;function Z(){return X||=n(G(process.cwd()),`logs`),X}function oe(e){let t=e.toISOString().slice(0,10);return n(Z(),`${t}.jsonl`)}let Q=0;function se(){let e=Date.now();if(!(e-Q<36e5)){Q=e;try{let t=Z(),r=new Date(e-30*864e5).toISOString().slice(0,10);for(let e of d(t))if(e.endsWith(`.jsonl`)&&e.slice(0,10)<r)try{p(n(t,e))}catch{}}catch{}}}function ce(e,t){try{l(Z(),{recursive:!0}),a(oe(t),`${e}\n`),se()}catch{}}function le(e){J=e}function ue(){return J}function de(e){Y=e}function fe(){X=void 0}function pe(e){if(e instanceof Error){let t={error:e.message};return e.stack&&(t.stack=e.stack),e.cause!==void 0&&(t.cause=e.cause instanceof Error?e.cause.message:String(e.cause)),t}return{error:String(e)}}function $(e){return q.push(e),()=>{let t=q.indexOf(e);t>=0&&q.splice(t,1)}}function me(e){function t(t,n,r){if(K[t]<K[J])return;let i=new Date,a={ts:i.toISOString(),level:t,component:e,msg:n,...r},o=JSON.stringify(a);console.error(o);for(let i of q)try{i({level:t,component:e,message:n,data:r})}catch{}ae()&&(t===`warn`||t===`error`)&&ce(o,i)}return{debug:(e,n)=>t(`debug`,e,n),info:(e,n)=>t(`info`,e,n),warn:(e,n)=>t(`warn`,e,n),error:(e,n)=>t(`error`,e,n)}}const he=[`indexed`,`curated`,`produced`],ge=[`source`,`documentation`,`test`,`config`,`generated`],_e=[`auto`,`manual`,`smart`],ve=[`efficient`,`normal`,`full`],ye=[`documentation`,`code-typescript`,`code-javascript`,`code-python`,`code-other`,`config-json`,`config-yaml`,`config-toml`,`config-env`,`test-code`,`cdk-stack`,`markdown`,`curated-knowledge`,`produced-knowledge`,`unknown`];export{_ as AIKIT_GLOBAL_PATHS,g as AIKIT_PATHS,k as AikitError,C as CATEGORY_PATTERN,v as CHUNK_SIZES,ye as CONTENT_TYPES,N as ConfigError,w as DEFAULT_CATEGORIES,y as EMBEDDING_DEFAULTS,A as EmbeddingError,x as FILE_LIMITS,_e as INDEX_MODES,M as IndexError,he as KNOWLEDGE_ORIGINS,S as SEARCH_DEFAULTS,ge as SOURCE_TYPES,b as STORE_DEFAULTS,j as StoreError,ve as TOKEN_BUDGETS,$ as addLogListener,F as computePartitionKey,ie as contentTypeToSourceType,me as createLogger,re as detectContentType,P as getGlobalDataDir,ue as getLogLevel,U as getPartitionDir,W as isUserInstalled,H as listWorkspaces,I as loadRegistry,V as lookupWorkspace,B as registerWorkspace,fe as resetLogDir,G as resolveStateDir,z as saveRegistry,pe as serializeError,de as setFileSinkEnabled,le as setLogLevel,O as sourceTypeContentTypes};
+import{basename as e,extname as t,join as n,resolve as r}from"node:path";import{createHash as i}from"node:crypto";import{appendFileSync as a,closeSync as o,constants as s,existsSync as c,mkdirSync as l,openSync as ee,readFileSync as u,readdirSync as d,renameSync as f,statSync as te,unlinkSync as p,writeFileSync as m}from"node:fs";import{homedir as h}from"node:os";const g={ai:`.ai`,aiContext:`.ai/context`,aiCurated:`.ai/curated`,restorePoints:`.ai/restore-points`,data:`.aikit-data`,state:`.aikit-state`,logs:`.aikit-state/logs`,brainstorm:`.brainstorm`,handoffs:`.handoffs`},_={root:`.aikit-data`,registry:`registry.json`},v={markdown:{max:1500,min:100},code:{max:2e3,min:50},config:{max:3e3,min:50},default:{max:1500,min:100,overlap:200}},y={model:`mixedbread-ai/mxbai-embed-large-v1`,dimensions:512},b={backend:`sqlite-vec`,path:g.data,tableName:`knowledge`},x={maxFileSizeBytes:1e6,maxCuratedFileSizeBytes:5e4},S={maxResults:10,minScore:.25},C=/^[a-z][a-z0-9-]*$/,w=[`decisions`,`patterns`,`troubleshooting`,`conventions`,`architecture`],T={".ts":`code-typescript`,".tsx":`code-typescript`,".mts":`code-typescript`,".cts":`code-typescript`,".js":`code-javascript`,".jsx":`code-javascript`,".mjs":`code-javascript`,".cjs":`code-javascript`,".py":`code-python`,".json":`config-json`,".yaml":`config-yaml`,".yml":`config-yaml`,".toml":`config-toml`,".env":`config-env`,".md":`markdown`,".mdx":`markdown`},E=[/\.test\.[jt]sx?$/,/\.spec\.[jt]sx?$/,/(^|\/)__tests__\//,/(^|\/)test\//,/(^|\/)tests\//,/(^|\/)spec\//,/(^|\/)fixtures\//],ne=[/\.stack\.[jt]s$/,/(^|\/)stacks\//,/(^|\/)constructs\//,/cdk\.json$/];function re(n){let r=t(n).toLowerCase(),i=e(n).toLowerCase();return n.includes(`${g.aiContext}/`)?`produced-knowledge`:n.includes(`${g.aiCurated}/`)?`curated-knowledge`:E.some(e=>e.test(n))?`test-code`:ne.some(e=>e.test(n))?`cdk-stack`:r in T?T[r]:i.startsWith(`.env`)?`config-env`:[`.go`,`.rs`,`.java`,`.rb`,`.php`,`.sh`,`.ps1`,`.sql`,`.graphql`,`.proto`,`.css`,`.scss`,`.less`,`.html`,`.htm`,`.vue`,`.svelte`,`.astro`,`.hbs`,`.ejs`,`.svg`].includes(r)?`code-other`:`unknown`}const D={"code-typescript":`source`,"code-javascript":`source`,"code-python":`source`,"code-other":`source`,"cdk-stack":`source`,"test-code":`test`,markdown:`documentation`,documentation:`documentation`,"curated-knowledge":`documentation`,"produced-knowledge":`documentation`,"config-json":`config`,"config-yaml":`config`,"config-toml":`config`,"config-env":`config`,unknown:`source`};function ie(e){return D[e]??`source`}function O(e){return Object.entries(D).filter(([,t])=>t===e).map(([e])=>e)}var k=class extends Error{code;constructor(e,t,n){super(e,n===void 0?void 0:{cause:n}),this.code=t,this.name=`AikitError`}},A=class extends k{constructor(e,t){super(e,`EMBEDDING_ERROR`,t),this.name=`EmbeddingError`}},j=class extends k{constructor(e,t){super(e,`STORE_ERROR`,t),this.name=`StoreError`}},M=class extends k{constructor(e,t){super(e,`INDEX_ERROR`,t),this.name=`IndexError`}},N=class extends k{constructor(e,t){super(e,`CONFIG_ERROR`,t),this.name=`ConfigError`}};function P(){return process.env.AIKIT_GLOBAL_DATA_DIR??r(h(),_.root)}function F(t){let n=r(t);return`${e(n).toLowerCase().replace(/[^a-z0-9-]/g,`-`)||`workspace`}-${i(`sha256`).update(n).digest(`hex`).slice(0,8)}`}function I(){let e=r(P(),_.registry);if(!c(e))return{version:1,workspaces:{}};let t=u(e,`utf-8`);try{return JSON.parse(t)}catch{return{version:1,workspaces:{}}}}function L(e,t=5e3){let n=`${e}.lock`,r=Date.now()+t,i=10;for(;Date.now()<r;)try{let e=ee(n,s.O_CREAT|s.O_EXCL|s.O_WRONLY);return m(e,`${process.pid}\n`),o(e),n}catch(e){if(e.code!==`EEXIST`)throw e;try{let{mtimeMs:e}=te(n);if(Date.now()-e>3e4){p(n);continue}}catch{}let t=new SharedArrayBuffer(4);Atomics.wait(new Int32Array(t),0,0,i),i=Math.min(i*2,200)}throw Error(`Failed to acquire registry lock after ${t}ms`)}function R(e){try{p(e)}catch{}}function z(e){let t=P();l(t,{recursive:!0});let n=r(t,_.registry),i=L(n);try{let t=`${n}.tmp`;m(t,JSON.stringify(e,null,2),`utf-8`),f(t,n)}finally{R(i)}}function B(e){let t=I(),n=F(e),i=new Date().toISOString();return t.workspaces[n]?t.workspaces[n].lastAccessedAt=i:t.workspaces[n]={partition:n,workspacePath:r(e),registeredAt:i,lastAccessedAt:i},l(U(n),{recursive:!0}),z(t),t.workspaces[n]}function V(e){let t=I(),n=F(e);return t.workspaces[n]}function H(){let e=I();return Object.values(e.workspaces)}function U(e){return r(P(),e)}function W(){return c(r(P(),_.registry))}function G(e){return W()?r(U(B(e).partition),`state`):r(e,g.state)}const K={debug:0,info:1,warn:2,error:3},q=[];let J=process.env.AIKIT_LOG_LEVEL??`info`,Y=process.env.AIKIT_LOG_FILE_SINK===`true`||process.env.AIKIT_LOG_FILE_SINK!==`false`&&!process.env.VITEST&&process.env.NODE_ENV!==`test`;function ae(){return Y?process.env.VITEST||process.env.NODE_ENV===`test`?process.env.AIKIT_LOG_FILE_SINK===`true`:!0:!1}let X;function Z(){return X||=n(G(process.cwd()),`logs`),X}function oe(e){let t=e.toISOString().slice(0,10);return n(Z(),`${t}.jsonl`)}let Q=0;function se(){let e=Date.now();if(!(e-Q<36e5)){Q=e;try{let t=Z(),r=new Date(e-30*864e5).toISOString().slice(0,10);for(let e of d(t))if(e.endsWith(`.jsonl`)&&e.slice(0,10)<r)try{p(n(t,e))}catch{}}catch{}}}function ce(e,t){try{l(Z(),{recursive:!0}),a(oe(t),`${e}\n`),se()}catch{}}function le(e){J=e}function ue(){return J}function de(e){Y=e}function fe(){X=void 0}function pe(e){if(e instanceof Error){let t={error:e.message};return e.stack&&(t.stack=e.stack),e.cause!==void 0&&(t.cause=e.cause instanceof Error?e.cause.message:String(e.cause)),t}return{error:String(e)}}function $(e){return q.push(e),()=>{let t=q.indexOf(e);t>=0&&q.splice(t,1)}}function me(e){function t(t,n,r){if(K[t]<K[J])return;let i=new Date,a={ts:i.toISOString(),level:t,component:e,msg:n,...r},o=JSON.stringify(a);(t===`warn`||t===`error`)&&console.error(o);for(let i of q)try{i({level:t,component:e,message:n,data:r})}catch{}ae()&&(t===`warn`||t===`error`)&&ce(o,i)}return{debug:(e,n)=>t(`debug`,e,n),info:(e,n)=>t(`info`,e,n),warn:(e,n)=>t(`warn`,e,n),error:(e,n)=>t(`error`,e,n)}}const he=[`indexed`,`curated`,`produced`],ge=[`source`,`documentation`,`test`,`config`,`generated`],_e=[`auto`,`manual`,`smart`],ve=[`efficient`,`normal`,`full`],ye=[`documentation`,`code-typescript`,`code-javascript`,`code-python`,`code-other`,`config-json`,`config-yaml`,`config-toml`,`config-env`,`test-code`,`cdk-stack`,`markdown`,`curated-knowledge`,`produced-knowledge`,`unknown`];export{_ as AIKIT_GLOBAL_PATHS,g as AIKIT_PATHS,k as AikitError,C as CATEGORY_PATTERN,v as CHUNK_SIZES,ye as CONTENT_TYPES,N as ConfigError,w as DEFAULT_CATEGORIES,y as EMBEDDING_DEFAULTS,A as EmbeddingError,x as FILE_LIMITS,_e as INDEX_MODES,M as IndexError,he as KNOWLEDGE_ORIGINS,S as SEARCH_DEFAULTS,ge as SOURCE_TYPES,b as STORE_DEFAULTS,j as StoreError,ve as TOKEN_BUDGETS,$ as addLogListener,F as computePartitionKey,ie as contentTypeToSourceType,me as createLogger,re as detectContentType,P as getGlobalDataDir,ue as getLogLevel,U as getPartitionDir,W as isUserInstalled,H as listWorkspaces,I as loadRegistry,V as lookupWorkspace,B as registerWorkspace,fe as resetLogDir,G as resolveStateDir,z as saveRegistry,pe as serializeError,de as setFileSinkEnabled,le as setLogLevel,O as sourceTypeContentTypes};

package/packages/embeddings/dist/index.d.ts

@@ -24,6 +24,9 @@ interface IEmbedder {
 }
 //#endregion
 //#region packages/embeddings/src/embedder-proxy.d.ts
+type EmbedderProxyLogger = {
+  warn?: (message: string, data?: Record<string, unknown>) => void;
+};
 interface EmbedderProxyOptions {
   model?: string;
   dimensions?: number;
@@ -32,9 +35,15 @@ interface EmbedderProxyOptions {
   interOpNumThreads?: number;
   intraOpNumThreads?: number;
   idleTimeoutMs?: number;
+  logger?: EmbedderProxyLogger;
+  maxRetries?: number;
+  retryBaseDelayMs?: number;
 }
 declare class EmbedderProxy implements IEmbedder {
   private readonly options;
+  private readonly logger?;
+  private readonly maxRetries;
+  private readonly retryBaseDelayMs;
   private readonly workerPath;
   private readonly pendingRequests;
   private readonly childState;
@@ -53,13 +62,18 @@ declare class EmbedderProxy implements IEmbedder {
   embed(text: string): Promise<Float32Array>;
   embedQuery(text: string): Promise<Float32Array>;
   embedBatch(texts: string[], batchSize?: number): Promise<Float32Array[]>;
+  private sendBatchRequest;
   shutdown(): Promise<void>;
   private startWorker;
   private spawnChild;
   private handleChildMessage;
   private handleChildExit;
   private handleChildFailure;
+  private sendVectorRequestWithRetry;
   private sendVectorRequest;
+  private withWorkerExitRetry;
+  private isWorkerExitError;
+  private wait;
   private requireReadyChild;
   private buildInitConfig;
   private buildChildEnv;

package/packages/embeddings/dist/index.js

@@ -1 +1 @@
-import{fork as e}from"node:child_process";import{randomUUID as t}from"node:crypto";import{dirname as n,join as r}from"node:path";import{fileURLToPath as i}from"node:url";import{EMBEDDING_DEFAULTS as a}from"../../core/dist/index.js";import{rm as o}from"node:fs/promises";import{homedir as s}from"node:os";var c=class{options;workerPath=r(n(i(import.meta.url)),`embedder-worker.js`);pendingRequests=new Map;childState=new WeakMap;child=null;readyChild=null;pendingInit=null;pendingShutdown=null;initializePromise=null;shutdownPromise=null;currentDimensions;currentModelId;constructor(e={}){this.options=e,this.currentDimensions=e.dimensions??a.dimensions,this.currentModelId=e.model??a.model}get dimensions(){return this.currentDimensions}get modelId(){return this.currentModelId}async initialize(){if(!(this.readyChild&&this.child===this.readyChild)){if(this.initializePromise)return this.initializePromise;if(this.shutdownPromise){try{await this.shutdownPromise}catch{}if(this.readyChild&&this.child===this.readyChild)return}return this.initializePromise=this.startWorker().finally(()=>{this.initializePromise=null}),this.initializePromise}}async embed(e){return this.sendVectorRequest({type:`embed`,text:e})}async embedQuery(e){return this.sendVectorRequest({type:`embedQuery`,text:e})}async embedBatch(e,n){if(e.length===0)return[];await this.initialize();let r=this.requireReadyChild(),i=t(),a=new Promise((e,t)=>{this.pendingRequests.set(i,{child:r,resolve:t=>e(t),reject:t})});try{r.send({type:`embedBatch`,id:i,texts:e,batchSize:n})}catch(e){throw this.pendingRequests.delete(i),this.toError(e,`Failed to send embedBatch request to worker`)}return a}async shutdown(){if(this.shutdownPromise)return this.shutdownPromise;let e=this.child;if(!e)return;let t=this.requireChildState(e);t.shutdownRequested=!0,this.readyChild===e&&(this.readyChild=null),this.shutdownPromise=new Promise((t,n)=>{this.pendingShutdown={child:e,resolve:t,reject:n}}).finally(()=>{this.shutdownPromise=null});try{e.send({type:`shutdown`})}catch(t){let n=this.toError(t,`Failed to send shutdown request to worker`);throw this.clearChildReference(e),this.rejectPendingForChild(e,n),this.rejectLifecycleIfOwned(`shutdown`,e,n),n}return this.shutdownPromise}async startWorker(){this.child&&this.readyChild!==this.child&&(this.child=null);let e=this.spawnChild();this.child=e,this.readyChild=null;let t=new Promise((t,n)=>{this.pendingInit={child:e,resolve:t,reject:n}});try{e.send({type:`init`,config:this.buildInitConfig()})}catch(t){let n=this.toError(t,`Failed to send init request to worker`);throw this.pendingInit=null,this.clearChildReference(e),n}return t}spawnChild(){let t=e(this.workerPath,[],{env:this.buildChildEnv()});return this.childState.set(t,{idleExitNotified:!1,shutdownRequested:!1,terminated:!1}),t.on(`message`,e=>{this.handleChildMessage(t,e)}),t.once(`error`,e=>{this.handleChildFailure(t,this.toError(e,`Embedder worker failed`))}),t.once(`exit`,(e,n)=>{this.handleChildExit(t,e,n)}),t}handleChildMessage(e,t){switch(t.type){case`ready`:{this.currentDimensions=t.dimensions,this.currentModelId=t.modelId,this.child===e&&(this.readyChild=e);let n=this.pendingInit;n?.child===e&&(this.pendingInit=null,n.resolve());return}case`result`:{let n=this.pendingRequests.get(t.id);if(!n||n.child!==e)return;this.pendingRequests.delete(t.id),n.resolve(new Float32Array(t.data));return}case`batchResult`:{let n=this.pendingRequests.get(t.id);if(!n||n.child!==e)return;this.pendingRequests.delete(t.id),n.resolve(t.data.map(e=>new Float32Array(e)));return}case`error`:{let n=Error(t.message);if(t.id===`init`){this.clearChildReference(e),this.rejectLifecycleIfOwned(`init`,e,n);return}if(t.id===`shutdown`){this.rejectLifecycleIfOwned(`shutdown`,e,n);return}let r=this.pendingRequests.get(t.id);if(!r||r.child!==e)return;this.pendingRequests.delete(t.id),r.reject(n);return}case`idle-exit`:{let t=this.requireChildState(e);t.idleExitNotified=!0,this.clearChildReference(e);return}}}handleChildExit(e,t,n){let r=this.requireChildState(e);if(r.terminated)return;r.terminated=!0,this.clearChildReference(e);let i=r.shutdownRequested||r.idleExitNotified,a=Error(i?`Embedder worker exited before completing request`:`Embedder worker exited unexpectedly (code ${t??`null`}${n?`, signal ${n}`:``})`);this.rejectLifecycleIfOwned(`init`,e,a),i?this.resolveLifecycleIfOwned(`shutdown`,e):this.rejectLifecycleIfOwned(`shutdown`,e,a),this.rejectPendingForChild(e,a)}handleChildFailure(e,t){let n=this.requireChildState(e);n.terminated||(n.terminated=!0,this.clearChildReference(e),this.rejectLifecycleIfOwned(`init`,e,t),this.rejectLifecycleIfOwned(`shutdown`,e,t),this.rejectPendingForChild(e,t))}async sendVectorRequest(e){await this.initialize();let n=this.requireReadyChild(),r=t(),i=new Promise((e,t)=>{this.pendingRequests.set(r,{child:n,resolve:t=>e(t),reject:t})});try{n.send({...e,id:r})}catch(t){throw this.pendingRequests.delete(r),this.toError(t,`Failed to send ${e.type} request to worker`)}return i}requireReadyChild(){if(!this.child||this.readyChild!==this.child)throw Error(`Embedder worker is not initialized`);return this.child}buildInitConfig(){return{model:this.options.model,dimensions:this.options.dimensions,nativeDim:this.options.nativeDim,queryPrefix:this.options.queryPrefix,interOpNumThreads:this.options.interOpNumThreads,intraOpNumThreads:this.options.intraOpNumThreads}}buildChildEnv(){return this.options.idleTimeoutMs===void 0?process.env:{...process.env,AIKIT_EMBED_IDLE_MS:String(this.options.idleTimeoutMs)}}requireChildState(e){let t=this.childState.get(e);if(!t)throw Error(`Embedder worker state not found`);return t}clearChildReference(e){this.child===e&&(this.child=null),this.readyChild===e&&(this.readyChild=null)}rejectPendingForChild(e,t){for(let[n,r]of this.pendingRequests)r.child===e&&(this.pendingRequests.delete(n),r.reject(t))}resolveLifecycleIfOwned(e,t){let n=this.pendingShutdown;!n||n.child!==t||(this.pendingShutdown=null,n.resolve())}rejectLifecycleIfOwned(e,t,n){let r=e===`init`?this.pendingInit:this.pendingShutdown;!r||r.child!==t||(e===`init`?this.pendingInit=null:this.pendingShutdown=null,r.reject(n))}toError(e,t){return e instanceof Error?e:Error(`${t}: ${String(e)}`)}};let l=null;async function u(){return l||(l=await import(`@huggingface/transformers`),l.env.cacheDir=r(s(),`.cache`,`huggingface`,`transformers-js`)),l}var d=class{pipe=null;shutdownPromise=null;dimensions;modelId;nativeDim;queryPrefix;threadConfig;constructor(e){if(this.modelId=e?.model??a.model,this.nativeDim=e?.nativeDim??1024,this.dimensions=e?.dimensions??a.dimensions,this.dimensions>this.nativeDim)throw Error(`Configured dimensions (${this.dimensions}) exceeds model native output (${this.nativeDim}). Matryoshka truncation cannot upscale — dimensions must be <= nativeDim.`);this.queryPrefix=e?.queryPrefix??this.detectQueryPrefix(this.modelId),this.threadConfig={interOp:e?.interOpNumThreads??1,intraOp:e?.intraOpNumThreads??4}}getPipelineOptions(e){let t=e.backends.onnx;t.wasm||={};let n=t.wasm;return n.numThreads=this.threadConfig.intraOp,{dtype:`q8`,session_options:{interOpNumThreads:this.threadConfig.interOp,intraOpNumThreads:this.threadConfig.intraOp}}}truncateAndRenorm(e){if(this.dimensions>=this.nativeDim)return e;let t=e.subarray(0,this.dimensions),n=0;for(let e=0;e<t.length;e++)n+=t[e]*t[e];if(n=Math.sqrt(n),n===0)return new Float32Array(this.dimensions);let r=new Float32Array(this.dimensions);for(let e=0;e<this.dimensions;e++)r[e]=t[e]/n;return r}detectQueryPrefix(e){let t=e.toLowerCase();return t.includes(`bge`)||t.includes(`mxbai-embed`)?`Represent this sentence for searching relevant passages: `:t.includes(`/e5-`)||t.includes(`multilingual-e5`)?`query: `:``}async initialize(){if(this.pipe)return;this.shutdownPromise=null;let{pipeline:e,env:t}=await u();try{this.pipe=await e(`feature-extraction`,this.modelId,this.getPipelineOptions(t))}catch(n){let i=n.message?.toLowerCase()??``;if(this.isCorruptionError(i)){let n=r(t.cacheDir??r(s(),`.cache`,`huggingface`,`transformers-js`),this.modelId);console.error(`[aikit:auto-heal] Detected corrupted model cache for "${this.modelId}". Clearing cache at ${n} and retrying download...`);try{await o(n,{recursive:!0,force:!0})}catch{}try{this.pipe=await e(`feature-extraction`,this.modelId,this.getPipelineOptions(t)),console.error(`[aikit:auto-heal] Model "${this.modelId}" re-downloaded successfully.`);return}catch(e){throw Error(`Failed to initialize embedding model "${this.modelId}" after auto-heal: ${e.message}`)}}throw Error(`Failed to initialize embedding model "${this.modelId}": ${n.message}`)}}isCorruptionError(e){return[`protobuf`,`invalid model`,`invalid onnx`,`unexpected end`,`unexpected token`,`failed to load`,`checksum`,`corrupt`,`could not load`,`onnx`,`malformed`].some(t=>e.includes(t))}async shutdown(){return this.shutdownPromise||=this._doShutdown(),this.shutdownPromise}async _doShutdown(){let e=this.pipe;if(e)try{let t=e;typeof t.dispose==`function`?await t.dispose():typeof t.model?.dispose==`function`&&await t.model.dispose()}catch{}finally{this.pipe=null}}async embed(e){this.pipe||await this.initialize();let t=await this.pipe?.(e,{pooling:`mean`,normalize:!0});if(!t?.data)throw Error(`Embedding pipeline returned no output`);try{let e=new Float32Array(t.data);return this.truncateAndRenorm(e)}finally{t.dispose?.()}}async embedQuery(e){return this.embed(this.queryPrefix+e)}async embedBatch(e,t=64){if(e.length===0)return[];this.pipe||await this.initialize();let n=[];for(let r=0;r<e.length;r+=t){let i=e.slice(r,r+t),a=await this.pipe?.(i,{pooling:`mean`,normalize:!0});if(!a?.data)throw Error(`Embedding pipeline returned no output`);try{if(i.length===1){let e=new Float32Array(a.data);n.push(this.truncateAndRenorm(e))}else for(let e=0;e<i.length;e++){let t=e*this.nativeDim,r=a.data.slice(t,t+this.nativeDim);n.push(this.truncateAndRenorm(new Float32Array(r)))}}finally{a.dispose?.()}}return n}};export{c as EmbedderProxy,d as OnnxEmbedder};
+import{fork as e}from"node:child_process";import{randomUUID as t}from"node:crypto";import{dirname as n,join as r}from"node:path";import{fileURLToPath as i}from"node:url";import{EMBEDDING_DEFAULTS as a}from"../../core/dist/index.js";import{rm as o}from"node:fs/promises";import{homedir as s}from"node:os";var c=class{options;logger;maxRetries;retryBaseDelayMs;workerPath=r(n(i(import.meta.url)),`embedder-worker.js`);pendingRequests=new Map;childState=new WeakMap;child=null;readyChild=null;pendingInit=null;pendingShutdown=null;initializePromise=null;shutdownPromise=null;currentDimensions;currentModelId;constructor(e={}){this.options=e,this.logger=e.logger,this.maxRetries=Math.max(0,e.maxRetries??3),this.retryBaseDelayMs=Math.max(0,e.retryBaseDelayMs??500),this.currentDimensions=e.dimensions??a.dimensions,this.currentModelId=e.model??a.model}get dimensions(){return this.currentDimensions}get modelId(){return this.currentModelId}async initialize(){if(!(this.readyChild&&this.child===this.readyChild)){if(this.initializePromise)return this.initializePromise;if(this.shutdownPromise){try{await this.shutdownPromise}catch{}if(this.readyChild&&this.child===this.readyChild)return}return this.initializePromise=this.startWorker().finally(()=>{this.initializePromise=null}),this.initializePromise}}async embed(e){return this.sendVectorRequestWithRetry({type:`embed`,text:e})}async embedQuery(e){return this.sendVectorRequestWithRetry({type:`embedQuery`,text:e})}async embedBatch(e,t){return e.length===0?[]:this.withWorkerExitRetry(`embedBatch`,()=>this.sendBatchRequest(e,t))}async sendBatchRequest(e,n){await this.initialize();let r=this.requireReadyChild(),i=t(),a=new Promise((e,t)=>{this.pendingRequests.set(i,{child:r,resolve:t=>e(t),reject:t})});try{r.send({type:`embedBatch`,id:i,texts:e,batchSize:n})}catch(e){throw this.pendingRequests.delete(i),this.toError(e,`Failed to send embedBatch request to worker`)}return a}async shutdown(){if(this.shutdownPromise)return this.shutdownPromise;let e=this.child;if(!e)return;let t=this.requireChildState(e);t.shutdownRequested=!0,this.readyChild===e&&(this.readyChild=null),this.shutdownPromise=new Promise((t,n)=>{this.pendingShutdown={child:e,resolve:t,reject:n}}).finally(()=>{this.shutdownPromise=null});try{e.send({type:`shutdown`})}catch(t){let n=this.toError(t,`Failed to send shutdown request to worker`);throw this.clearChildReference(e),this.rejectPendingForChild(e,n),this.rejectLifecycleIfOwned(`shutdown`,e,n),n}return this.shutdownPromise}async startWorker(){this.child&&this.readyChild!==this.child&&(this.child=null);let e=this.spawnChild();this.child=e,this.readyChild=null;let t=new Promise((t,n)=>{this.pendingInit={child:e,resolve:t,reject:n}});try{e.send({type:`init`,config:this.buildInitConfig()})}catch(t){let n=this.toError(t,`Failed to send init request to worker`);throw this.pendingInit=null,this.clearChildReference(e),n}return t}spawnChild(){let t=e(this.workerPath,[],{env:this.buildChildEnv()});return this.childState.set(t,{idleExitNotified:!1,shutdownRequested:!1,terminated:!1}),t.on(`message`,e=>{this.handleChildMessage(t,e)}),t.once(`error`,e=>{this.handleChildFailure(t,this.toError(e,`Embedder worker failed`))}),t.once(`exit`,(e,n)=>{this.handleChildExit(t,e,n)}),t}handleChildMessage(e,t){switch(t.type){case`ready`:{this.currentDimensions=t.dimensions,this.currentModelId=t.modelId,this.child===e&&(this.readyChild=e);let n=this.pendingInit;n?.child===e&&(this.pendingInit=null,n.resolve());return}case`result`:{let n=this.pendingRequests.get(t.id);if(!n||n.child!==e)return;this.pendingRequests.delete(t.id),n.resolve(new Float32Array(t.data));return}case`batchResult`:{let n=this.pendingRequests.get(t.id);if(!n||n.child!==e)return;this.pendingRequests.delete(t.id),n.resolve(t.data.map(e=>new Float32Array(e)));return}case`error`:{let n=Error(t.message);if(t.id===`init`){this.clearChildReference(e),this.rejectLifecycleIfOwned(`init`,e,n);return}if(t.id===`shutdown`){this.rejectLifecycleIfOwned(`shutdown`,e,n);return}let r=this.pendingRequests.get(t.id);if(!r||r.child!==e)return;this.pendingRequests.delete(t.id),r.reject(n);return}case`idle-exit`:{let t=this.requireChildState(e);t.idleExitNotified=!0,this.clearChildReference(e);return}}}handleChildExit(e,t,n){let r=this.requireChildState(e);if(r.terminated)return;r.terminated=!0,this.clearChildReference(e);let i=r.shutdownRequested||r.idleExitNotified,a=Error(i?`Embedder worker exited before completing request`:`Embedder worker exited unexpectedly (code ${t??`null`}${n?`, signal ${n}`:``})`);this.rejectLifecycleIfOwned(`init`,e,a),i?this.resolveLifecycleIfOwned(`shutdown`,e):this.rejectLifecycleIfOwned(`shutdown`,e,a),this.rejectPendingForChild(e,a)}handleChildFailure(e,t){let n=this.requireChildState(e);n.terminated||(n.terminated=!0,this.clearChildReference(e),this.rejectLifecycleIfOwned(`init`,e,t),this.rejectLifecycleIfOwned(`shutdown`,e,t),this.rejectPendingForChild(e,t))}async sendVectorRequestWithRetry(e){return this.withWorkerExitRetry(e.type,()=>this.sendVectorRequest(e))}async sendVectorRequest(e){await this.initialize();let n=this.requireReadyChild(),r=t(),i=new Promise((e,t)=>{this.pendingRequests.set(r,{child:n,resolve:t=>e(t),reject:t})});try{n.send({...e,id:r})}catch(t){throw this.pendingRequests.delete(r),this.toError(t,`Failed to send ${e.type} request to worker`)}return i}async withWorkerExitRetry(e,t){let n=null;for(let r=0;r<=this.maxRetries;r++)try{return await t()}catch(t){let i=this.toError(t,`Failed to process ${e} request`);if(!this.isWorkerExitError(i))throw i;if(n??=i,r===this.maxRetries)throw n;let a=r+1,o=this.retryBaseDelayMs*2**r;this.logger?.warn?.(`Embedder retry ${a}/${this.maxRetries} after ${o}ms`,{requestType:e,delayMs:o,error:i.message}),this.child=null,this.readyChild=null,await this.wait(o)}throw n??Error(`Failed to process ${e} request`)}isWorkerExitError(e){return/embedder worker exited/i.test(e.message)}wait(e){return new Promise(t=>{setTimeout(t,e)})}requireReadyChild(){if(!this.child||this.readyChild!==this.child)throw Error(`Embedder worker is not initialized`);return this.child}buildInitConfig(){return{model:this.options.model,dimensions:this.options.dimensions,nativeDim:this.options.nativeDim,queryPrefix:this.options.queryPrefix,interOpNumThreads:this.options.interOpNumThreads,intraOpNumThreads:this.options.intraOpNumThreads}}buildChildEnv(){return this.options.idleTimeoutMs===void 0?process.env:{...process.env,AIKIT_EMBED_IDLE_MS:String(this.options.idleTimeoutMs)}}requireChildState(e){let t=this.childState.get(e);if(!t)throw Error(`Embedder worker state not found`);return t}clearChildReference(e){this.child===e&&(this.child=null),this.readyChild===e&&(this.readyChild=null)}rejectPendingForChild(e,t){for(let[n,r]of this.pendingRequests)r.child===e&&(this.pendingRequests.delete(n),r.reject(t))}resolveLifecycleIfOwned(e,t){let n=this.pendingShutdown;!n||n.child!==t||(this.pendingShutdown=null,n.resolve())}rejectLifecycleIfOwned(e,t,n){let r=e===`init`?this.pendingInit:this.pendingShutdown;!r||r.child!==t||(e===`init`?this.pendingInit=null:this.pendingShutdown=null,r.reject(n))}toError(e,t){return e instanceof Error?e:Error(`${t}: ${String(e)}`)}};let l=null;async function u(){return l||(l=await import(`@huggingface/transformers`),l.env.cacheDir=r(s(),`.cache`,`huggingface`,`transformers-js`)),l}var d=class{pipe=null;shutdownPromise=null;dimensions;modelId;nativeDim;queryPrefix;threadConfig;constructor(e){if(this.modelId=e?.model??a.model,this.nativeDim=e?.nativeDim??1024,this.dimensions=e?.dimensions??a.dimensions,this.dimensions>this.nativeDim)throw Error(`Configured dimensions (${this.dimensions}) exceeds model native output (${this.nativeDim}). Matryoshka truncation cannot upscale — dimensions must be <= nativeDim.`);this.queryPrefix=e?.queryPrefix??this.detectQueryPrefix(this.modelId),this.threadConfig={interOp:e?.interOpNumThreads??1,intraOp:e?.intraOpNumThreads??4}}getPipelineOptions(e){let t=e.backends.onnx;t.wasm||={};let n=t.wasm;return n.numThreads=this.threadConfig.intraOp,{dtype:`q8`,session_options:{interOpNumThreads:this.threadConfig.interOp,intraOpNumThreads:this.threadConfig.intraOp}}}truncateAndRenorm(e){if(this.dimensions>=this.nativeDim)return e;let t=e.subarray(0,this.dimensions),n=0;for(let e=0;e<t.length;e++)n+=t[e]*t[e];if(n=Math.sqrt(n),n===0)return new Float32Array(this.dimensions);let r=new Float32Array(this.dimensions);for(let e=0;e<this.dimensions;e++)r[e]=t[e]/n;return r}detectQueryPrefix(e){let t=e.toLowerCase();return t.includes(`bge`)||t.includes(`mxbai-embed`)?`Represent this sentence for searching relevant passages: `:t.includes(`/e5-`)||t.includes(`multilingual-e5`)?`query: `:``}async initialize(){if(this.pipe)return;this.shutdownPromise=null;let{pipeline:e,env:t}=await u();try{this.pipe=await e(`feature-extraction`,this.modelId,this.getPipelineOptions(t))}catch(n){let i=n.message?.toLowerCase()??``;if(this.isCorruptionError(i)){let n=r(t.cacheDir??r(s(),`.cache`,`huggingface`,`transformers-js`),this.modelId);console.error(`[aikit:auto-heal] Detected corrupted model cache for "${this.modelId}". Clearing cache at ${n} and retrying download...`);try{await o(n,{recursive:!0,force:!0})}catch{}try{this.pipe=await e(`feature-extraction`,this.modelId,this.getPipelineOptions(t)),console.error(`[aikit:auto-heal] Model "${this.modelId}" re-downloaded successfully.`);return}catch(e){throw Error(`Failed to initialize embedding model "${this.modelId}" after auto-heal: ${e.message}`)}}throw Error(`Failed to initialize embedding model "${this.modelId}": ${n.message}`)}}isCorruptionError(e){return[`protobuf`,`invalid model`,`invalid onnx`,`unexpected end`,`unexpected token`,`failed to load`,`checksum`,`corrupt`,`could not load`,`onnx`,`malformed`].some(t=>e.includes(t))}async shutdown(){return this.shutdownPromise||=this._doShutdown(),this.shutdownPromise}async _doShutdown(){let e=this.pipe;if(e)try{let t=e;typeof t.dispose==`function`?await t.dispose():typeof t.model?.dispose==`function`&&await t.model.dispose()}catch{}finally{this.pipe=null}}async embed(e){this.pipe||await this.initialize();let t=await this.pipe?.(e,{pooling:`mean`,normalize:!0});if(!t?.data)throw Error(`Embedding pipeline returned no output`);try{let e=new Float32Array(t.data);return this.truncateAndRenorm(e)}finally{t.dispose?.()}}async embedQuery(e){return this.embed(this.queryPrefix+e)}async embedBatch(e,t=64){if(e.length===0)return[];this.pipe||await this.initialize();let n=[];for(let r=0;r<e.length;r+=t){let i=e.slice(r,r+t),a=await this.pipe?.(i,{pooling:`mean`,normalize:!0});if(!a?.data)throw Error(`Embedding pipeline returned no output`);try{if(i.length===1){let e=new Float32Array(a.data);n.push(this.truncateAndRenorm(e))}else for(let e=0;e<i.length;e++){let t=e*this.nativeDim,r=a.data.slice(t,t+this.nativeDim);n.push(this.truncateAndRenorm(new Float32Array(r)))}}finally{a.dispose?.()}}return n}};export{c as EmbedderProxy,d as OnnxEmbedder};