@kanonak-protocol/sdk 3.50.0 → 3.51.0

package/dist/search/SearchIndex.d.ts

@@ -9,10 +9,18 @@ export interface SearchHit {
 export interface BuildOptions {
     /**
      * Directory to persist/reuse the embedding cache. When set, an unchanged
-     * catalog reuses cached vectors instead of re-embedding. Defaults to
-     * `~/.kanonak/search` when `cache` is true.
+     * catalog reuses cached vectors instead of re-embedding.
      */
     cacheDir?: string;
+    /** Progress callback during embedding: (resources embedded so far, total). */
+    onProgress?: (done: number, total: number) => void;
+    /**
+     * Only index subjects passing this predicate (e.g. workspace-local
+     * resources, excluding imported vocabulary). Embed-text is still resolved
+     * against the FULL catalog, so type labels and the rdfs.Resource floor
+     * still apply. Default: index everything.
+     */
+    include?: (subject: SubjectKanonak) => boolean;
 }
 export interface BuildResult {
     /** Number of resources indexed. */

package/dist/search/index.js

@@ -1,6 +1,6 @@
-import{a as d}from"../chunk-2WQR7OHH.js";import"../chunk-3NEUHF7E.js";import"../chunk-DC774EQL.js";import"../chunk-NIGFQYVA.js";import"../chunk-H2LEQLVD.js";import{d as c,g as x}from"../chunk-N224PFBI.js";import{a as h,f as j}from"../chunk-Q4DKP5DO.js";import{a as k}from"../chunk-FQHALFRR.js";import"../chunk-64TVBQFR.js";import{c as v}from"../chunk-W6T7MOKY.js";import"../chunk-FUUTGGJS.js";import{homedir as F}from"os";import{join as y}from"path";import{existsSync as E,mkdirSync as B,readFileSync as C,writeFileSync as O}from"fs";import{createHash as P}from"crypto";var p="Xenova/all-MiniLM-L6-v2",f=null;async function S(){return f||(f=(async()=>{let n;try{n=await import("@huggingface/transformers")}catch(i){throw new Error(`Semantic search requires the embedding runtime, which is an optional dependency.
+import{a as p}from"../chunk-2WQR7OHH.js";import"../chunk-3NEUHF7E.js";import"../chunk-DC774EQL.js";import"../chunk-NIGFQYVA.js";import"../chunk-H2LEQLVD.js";import{d as m,g as x}from"../chunk-N224PFBI.js";import{a as d,f as S}from"../chunk-Q4DKP5DO.js";import{a as v}from"../chunk-FQHALFRR.js";import"../chunk-64TVBQFR.js";import{c as j}from"../chunk-W6T7MOKY.js";import"../chunk-FUUTGGJS.js";import{homedir as P}from"os";import{join as k}from"path";import{existsSync as B,mkdirSync as C,readFileSync as D,writeFileSync as O}from"fs";import{createHash as z}from"crypto";var f="Xenova/all-MiniLM-L6-v2",w="q8",b=32,g=null;async function K(){return g||(g=(async()=>{let t;try{t=await import("@huggingface/transformers")}catch(i){throw new Error(`Semantic search requires the embedding runtime, which is an optional dependency.
 Install it with:
   npm install @huggingface/transformers
-(underlying import error: ${i.message})`)}let{pipeline:t,env:r}=n;r.cacheDir=y(F(),".kanonak","models");let e=await t("feature-extraction",p);return async i=>(await e(i,{pooling:"mean",normalize:!0})).tolist()})()),f}function z(n,t){let r=0;for(let e=0;e<n.length;e++)r+=n[e]*t[e];return r}function D(n,t){return t?n.version?t.version?k(n.version,t.version)>0:!0:!1:!0}function K(n,t){let r=d(n,t),e=[r.label],i=x(t);if(i.length>0){let s=j(n,i[0]),o=s?d(n,s).label:i[0].name;o&&e.push(`\u2014 a ${o}.`)}return r.summary&&e.push(r.summary),e.join(" ")}function U(n){let t=new Map;for(let r of n){if(!(r instanceof v))continue;let e=c(r);if(!e)continue;let i=h(e),s=t.get(i);(!s||D(e,c(s)))&&t.set(i,r)}return[...t.values()].sort((r,e)=>{let i=c(r)?h(c(r)):r.name,s=c(e)?h(c(e)):e.name;return i.localeCompare(s)})}var b=class{entries=[];get size(){return this.entries.length}async build(t,r={}){let e=U(t);if(e.length===0)return this.entries=[],{size:0,cached:!1};let i=e.map(u=>K(t,u)),s=r.cacheDir,o;if(s){let u=P("sha256").update(p+`
+(underlying import error: ${i.message})`)}let{pipeline:r,env:n}=t;n.cacheDir=k(P(),".kanonak","models");let e=await r("feature-extraction",f,{dtype:w});return async i=>(await e(i,{pooling:"mean",normalize:!0})).tolist()})()),g}function H(t,r){let n=0;for(let e=0;e<t.length;e++)n+=t[e]*r[e];return n}function U(t,r){return r?t.version?r.version?v(t.version,r.version)>0:!0:!1:!0}function F(t,r){let n=p(t,r),e=[n.label],i=x(r);if(i.length>0){let s=S(t,i[0]),c=s?p(t,s).label:i[0].name;c&&e.push(`\u2014 a ${c}.`)}return n.summary&&e.push(n.summary),e.join(" ")}function M(t){let r=new Map;for(let n of t){if(!(n instanceof j))continue;let e=m(n);if(!e)continue;let i=d(e),s=r.get(i);(!s||U(e,m(s)))&&r.set(i,n)}return[...r.values()].sort((n,e)=>{let i=m(n)?d(m(n)):n.name,s=m(e)?d(m(e)):e.name;return i.localeCompare(s)})}var y=class{entries=[];get size(){return this.entries.length}async build(r,n={}){let e=M(r);if(n.include&&(e=e.filter(n.include)),e.length===0)return this.entries=[],{size:0,cached:!1};let i=e.map(o=>F(r,o)),s=n.cacheDir,c;if(s){let o=z("sha256").update(f+"|"+w+`
 `+i.join(`
-`)).digest("hex").slice(0,32);o=y(s,`index-${u}.json`);let m=H(o);if(m&&m.model===p&&m.vectors.length===e.length)return this.entries=e.map((g,w)=>({subject:g,namespace:g.namespace||"",vector:m.vectors[w]})),{size:this.entries.length,cached:!0}}let l=await(await S())(i);return this.entries=e.map((u,m)=>({subject:u,namespace:u.namespace||"",vector:l[m]})),o&&M(o,{model:p,vectors:l}),{size:this.entries.length,cached:!1}}async query(t,r){if(this.entries.length===0)return[];let e=await S(),[i]=await e([t]),s=r?.scope;return this.entries.filter(a=>!s||a.namespace.startsWith(s)).map(a=>({subject:a.subject,score:z(i,a.vector)})).sort((a,l)=>l.score-a.score).slice(0,r?.limit??20)}};function H(n){try{return E(n)?JSON.parse(C(n,"utf-8")):null}catch{return null}}function M(n,t){try{B(y(n,".."),{recursive:!0}),O(n,JSON.stringify(t))}catch{}}export{b as SearchIndex,K as embedTextFor};
+`)).digest("hex").slice(0,32);c=k(s,`index-${o}.json`);let u=T(c);if(u&&u.model===f&&u.vectors.length===e.length)return this.entries=e.map((h,E)=>({subject:h,namespace:h.namespace||"",vector:u.vectors[E]})),{size:this.entries.length,cached:!0}}let a=await K(),l=[];for(let o=0;o<i.length;o+=b){let u=await a(i.slice(o,o+b));for(let h of u)l.push(h);n.onProgress?.(Math.min(o+b,i.length),i.length)}return this.entries=e.map((o,u)=>({subject:o,namespace:o.namespace||"",vector:l[u]})),c&&q(c,{model:f,vectors:l}),{size:this.entries.length,cached:!1}}async query(r,n){if(this.entries.length===0)return[];let e=await K(),[i]=await e([r]),s=n?.scope;return this.entries.filter(a=>!s||a.namespace.startsWith(s)).map(a=>({subject:a.subject,score:H(i,a.vector)})).sort((a,l)=>l.score-a.score).slice(0,n?.limit??20)}};function T(t){try{return B(t)?JSON.parse(D(t,"utf-8")):null}catch{return null}}function q(t,r){try{C(k(t,".."),{recursive:!0}),O(t,JSON.stringify(r))}catch{}}export{y as SearchIndex,F as embedTextFor};

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@kanonak-protocol/sdk",
-  "version": "3.50.0",
+  "version": "3.51.0",
   "description": "Kanonak Protocol SDK - Document repository and parsing implementations for TypeScript",
   "type": "module",
   "main": "./dist/index.js",
@@ -121,7 +121,7 @@
     "yaml-parser"
   ],
   "dependencies": {
-    "@kanonak-protocol/types": "^3.50.0",
+    "@kanonak-protocol/types": "^3.51.0",
     "ignore": "^7.0.5",
     "js-yaml": "^4.1.0",
     "yaml": "^2.7.0"