wiki-plugin-similarity 0.13.0 → 0.13.1

package/client/similarity.js

@@ -1,4 +1,4 @@
-/* wiki-plugin-similarity - 0.13.0 - Tue, 18 Aug 2026 13:37:47 GMT */
+/* wiki-plugin-similarity - 0.13.1 - Sun, 30 Aug 2026 12:37:30 GMT */
 (()=>{var q={high:.78,medium:.68,low:.58},N=q.medium,J=10,_=t=>{let e=[],s=[],i=[],p=null,E=null,f="search",y=!1,L=!1,b=!1,k=null,C=null,S=(a,r)=>a===r||a.startsWith(r)&&/^[\s:]/.test(a.slice(r.length)),w=(a,r)=>a.slice(r.length).replace(/^\s*:?\s*/,"").trim();for(let a of t.split(`
 `)){let r=a.trim();if(!r||r.startsWith("#"))continue;let g=r.toUpperCase();if(S(g,"LIVE")){y=!0;continue}if(S(g,"SUBJECT")){L=!0;continue}if(S(g,"AUTHOR")){!e.length&&f==="search"&&(f="author");continue}if(S(g,"REPORT")){f==="search"&&(f="report");continue}if(S(g,"KEYWORD")){f==="search"&&(f="keyword");continue}if(S(g,"SITES")){f==="search"&&(f="sites");continue}if(S(g,"BUILD")){f==="search"&&(f="build");continue}if(S(g,"FORCE")){b=!0;continue}if(S(g,"GHOST")){k=w(r,"GHOST"),f==="search"&&(f="ghost");continue}if(S(g,"BUTTON")){C=w(r,"BUTTON");continue}if(S(g,"ROSTER")){let v=w(r,"ROSTER");v&&s.push(v);continue}if(S(g,"FARM")){let v=w(r,"FARM");v&&i.push(v);continue}if(S(g,"LIST")){!e.length&&f==="search"&&(f="list");continue}if(S(g,"SIMILAR")){let v=w(g,"SIMILAR").toLowerCase();p=q[v]||N,!e.length&&f==="search"&&(f="similar");continue}if(S(g,"THRESHOLD")){let v=w(r,"THRESHOLD");p=q[v.toLowerCase()]??(parseFloat(v)||N);continue}if(S(g,"LIMIT")){E=parseInt(w(r,"LIMIT"))||J;continue}e.push(["PUBLIC","LOCAL","PRIVATE","GALAXY"].includes(g)?g:r)}return{mode:f,specs:e,rosterRefs:s,farms:i,threshold:p??N,limit:E??J,live:y,subject:L,force:b,ghostUrl:k,label:C,thresholdSet:p!==null}},Y=t=>t.includes("*")||t.includes("?"),W=t=>t==="PUBLIC"||t==="LOCAL"||t==="PRIVATE"||t==="GALAXY";var X=t=>t.toLowerCase().replace(/\s+/g,"-").replace(/[^a-z0-9-]/g,""),B=new Map,pt=async(t,e)=>{if(B.has(t))return B.get(t);let s=`${e}/system/indexed-domains.json?pattern=${encodeURIComponent(t)}`,i=await fetch(s);if(!i.ok)throw new Error(`indexed-domains failed: ${i.status}`);let p=await i.json();return B.set(t,p),p},O=async(t,e)=>{t.length||(t=[window.location.hostname]);let s=new Set,i=[];for(let p of t)if(p==="*"||Y(p)||W(p))for(let E of await pt(p,e))s.has(E.domain)||(s.add(E.domain),i.push(E));else s.has(p)||(s.add(p),i.push({domain:p,page_count:null}));return i},gt=/^([a-zA-Z0-9-]+(\.[a-zA-Z0-9-]+)+|localhost)(:\d+)?$/,ht=/^ROSTER ([A-Za-z0-9.\-:]+\/[a-z0-9-]+)$/,wt=/^REFERENCES ([A-Za-z0-9.\-:]+\/[a-z0-9-]+)$/,Z=async t=>{let e=t.indexOf("/"),s=await fetch(`//${t.slice(0,e)}/${t.slice(e+1)}.json`);if(!s.ok)throw new Error(`roster page ${t} failed: ${s.status}`);return s.json()},Q=async(t,e,s)=>{if(s.has(t))return;s.add(t);let i;try{i=await Z(t)}catch(p){console.warn(p.message);return}for(let p of i.story||[])if(p.type==="roster")for(let E of(p.text||"").split(/\r?\n/)){let f=E.trim();if(!f)continue;let y=f.match(gt);if(y){e.add(y[0]);continue}let L=f.match(ht);if(L){await Q(L[1],e,s);continue}let b=f.match(wt);if(b)try{let k=await Z(b[1]);for(let C of k.story||[])C.type==="reference"&&C.site&&e.add(C.site)}catch(k){console.warn(k.message)}}},yt=async t=>{let e=new Set,s=new Set;for(let i of t)await Q(i,e,s);return[...e]},tt=async(t,e)=>e.length?[...t,...await yt(e)]:t;var G=new Map,$t=t=>`${window.location.origin}/system/semantic-vectors.json?domain=${encodeURIComponent(t)}`,et=async t=>{if(G.has(t))return G.get(t);let e=await fetch($t(t));if(!e.ok)return[];let s=await e.json();return G.set(t,s),s},U=async(t,e)=>{let s=await fetch(`${e}/system/embed.json`,{method:"POST",headers:{"Content-Type":"application/json"},body:JSON.stringify({text:t})});if(!s.ok)throw new Error(`embed failed: ${s.status}`);return(await s.json()).vector},st=async(t,e)=>{let i=(await et(e)).find(p=>p.slug===t);return i?i.vector:null},M=(t,e,{threshold:s,limit:i,excludeSlug:p,excludeDomain:E})=>{let f=[];for(let{domain:y,pages:L}of e)for(let{slug:b,title:k,vector:C}of L){if(b===p&&y===E)continue;let S=0;for(let w=0;w<t.length;w++)S+=t[w]*C[w];S>=s&&f.push({domain:y,slug:b,title:k,score:S})}return f.sort((y,L)=>L.score-y.score),f.slice(0,i)},z=async(t,e)=>{let s=await O(t,e);return(await Promise.all(s.map(async({domain:p})=>({domain:p,pages:await et(p)})))).filter(p=>p.pages.length>0)};var nt=t=>{let e=t.parents(".page:first"),s=e.prev(".page"),i=s.length?s:e,p=(i.attr("id")||"").split("_rev")[0],E=i.data("site")||window.location.hostname,f=i.data("data")||{},y=f.title||i.find(".title").text().trim()||p,L=[y,...(f.story||[]).filter(b=>b.type==="markdown"||b.type==="paragraph").map(b=>(b.text||"").trim()).filter(Boolean)].join(`
 `).slice(0,2e3);return{slug:p,site:E,title:y,text:L,isSelf:!s.length}},T=t=>`Subject: ${t.title}${t.isSelf?" (this page)":` @ ${t.site}`}`;var it=t=>`sim-cache-${t}`,ot=t=>{try{let e=JSON.parse(localStorage.getItem(it(t.id))||"null");return e?.text===(t.text||"")?e:null}catch{return null}},P=(t,e)=>{try{localStorage.setItem(it(t.id),JSON.stringify({text:t.text||"",ts:Date.now(),...e}))}catch{}},at=t=>{let e=Math.floor((Date.now()-t)/1e3);return e<60?`${e}s ago`:e<3600?`${Math.floor(e/60)}m ago`:e<86400?`${Math.floor(e/3600)}h ago`:`${Math.floor(e/86400)}d ago`};var D=`

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "wiki-plugin-similarity",
-  "version": "0.13.0",
+  "version": "0.13.1",
   "description": "Federated Wiki plugin — semantic similarity search across wiki domains using prebuilt vector indexes",
   "keywords": [
     "wiki",

package/server/galaxy-vectors.js

@@ -10,10 +10,12 @@
 // at index time because off-farm pages have no local page JSON for
 // search-report's enrich() to read.
 //
-// Farm vectors (149 MB across the whole farm) are cheap enough to re-read per
-// query; galaxy scale is not. This module holds deserialized per-site entries
-// in a byte-capped LRU keyed by file mtime, so a repeat query over a followed
-// roster costs no JSON parsing at all.
+// This module holds deserialized per-site entries in a byte-capped LRU keyed
+// by file path + mtime, so a repeat query costs no JSON parsing at all. Since
+// 0.13.1 the FARM vectors ride the same cache: "cheap enough to re-read per
+// query" proved false at 400 sites / 150 MB — the synchronous re-parse
+// blocked the event loop for tens of seconds and 502'd concurrent page saves
+// (Semantic ReIndex Plan, Phase 1; the yellow-halo bug, 2026-08-30).
 
 const fs   = require('node:fs')
 const path = require('node:path')
@@ -28,29 +30,28 @@ const galaxyRoot = () => {
   return path.join(process.env.HOME || '/tmp', '.cache', 'wiki-similarity', 'galaxy-vectors')
 }
 
-// domain → {mtimeMs, bytes, pages}
+// file path → {mtimeMs, bytes, pages} — keyed by path, not domain, so the
+// same domain name in different farms (or farm vs galaxy) can never collide.
 const cache = new Map()
 let cachedBytes = 0
 
 const evictUntil = budget => {
-  for (const [domain, entry] of cache) {
+  for (const [file, entry] of cache) {
     if (cachedBytes <= budget) break
-    cache.delete(domain)
+    cache.delete(file)
     cachedBytes -= entry.bytes
   }
 }
 
-// Same contract as search-report's loadVectors(farm, domain) — returns the
-// entry array (with enrichment fields) or [].
-const loadGalaxyVectors = (farm, domain) => {
-  const file = path.join(farm, domain, 'status', 'semantic-vectors.json')
+// The cache core: one vectors FILE → its entry array (or []), mtime-fresh.
+const loadVectorsFile = file => {
   let stat
   try { stat = fs.statSync(file) } catch { return [] }
 
-  const hit = cache.get(domain)
+  const hit = cache.get(file)
   if (hit && hit.mtimeMs === stat.mtimeMs) {
-    cache.delete(domain)          // refresh LRU position
-    cache.set(domain, hit)
+    cache.delete(file)            // refresh LRU position
+    cache.set(file, hit)
     return hit.pages
   }
 
@@ -58,17 +59,23 @@ const loadGalaxyVectors = (farm, domain) => {
   try { pages = JSON.parse(fs.readFileSync(file, 'utf8')) } catch { return [] }
   if (!Array.isArray(pages)) return []
 
-  if (hit) { cache.delete(domain); cachedBytes -= hit.bytes }
+  if (hit) { cache.delete(file); cachedBytes -= hit.bytes }
   cachedBytes += stat.size
-  cache.set(domain, { mtimeMs: stat.mtimeMs, bytes: stat.size, pages })
+  cache.set(file, { mtimeMs: stat.mtimeMs, bytes: stat.size, pages })
   evictUntil(CACHE_BYTES)
   return pages
 }
 
+// Same contract as the old loadVectors(farm, domain) — farm or galaxy tree.
+const loadVectorsCached = (farm, domain) =>
+  loadVectorsFile(path.join(farm, domain, 'status', 'semantic-vectors.json'))
+
+const loadGalaxyVectors = loadVectorsCached
+
 const galaxyCacheStats = () => ({
   sites: cache.size,
   bytes: cachedBytes,
   capBytes: CACHE_BYTES,
 })
 
-module.exports = { galaxyRoot, loadGalaxyVectors, galaxyCacheStats }
+module.exports = { galaxyRoot, loadGalaxyVectors, loadVectorsCached, loadVectorsFile, galaxyCacheStats }