wiki-plugin-similarity 0.21.1 → 0.21.2

package/client/similarity.js

@@ -1,4 +1,4 @@
-/* wiki-plugin-similarity - 0.21.1 - Fri, 04 Sep 2026 19:48:21 GMT */
+/* wiki-plugin-similarity - 0.21.2 - Fri, 04 Sep 2026 20:04:24 GMT */
 (()=>{var dt={high:.78,medium:.68,low:.58},ct=dt.medium,yt=10,ht=t=>{let e=[],n=[],s=[],r=null,l=null,u="search",g=!1,k=!1,x=!1,P=null,D=null,R=null,M=null,j=(o,m)=>o===m||o.startsWith(m)&&/^[\s:]/.test(o.slice(m.length)),L=(o,m)=>o.slice(m.length).replace(/^\s*:?\s*/,"").trim();for(let o of t.split(`
 `)){let m=o.trim();if(!m||m.startsWith("#"))continue;let w=m.toUpperCase();if(j(w,"LIVE")){g=!0;continue}if(j(w,"SUBJECT")){k=!0;continue}if(j(w,"AUTHOR")){!e.length&&u==="search"&&(u="author");continue}if(j(w,"REPORT")){u==="search"&&(u="report");continue}if(j(w,"KEYWORD")){u==="search"&&(u="keyword");continue}if(j(w,"SITES")){u==="search"&&(u="sites");continue}if(j(w,"BUILD")){u==="search"&&(u="build");continue}if(j(w,"FORCE")){x=!0;continue}if(j(w,"GHOST")){P=L(m,"GHOST"),u==="search"&&(u="ghost");continue}if(j(w,"BUTTON")){D=L(m,"BUTTON");continue}if(j(w,"BATCH")){let E=L(m,"BATCH").toLowerCase();R=E==="off"||E==="no"||E==="0"?0:parseInt(E)||50;continue}if(j(w,"ALGORITHM")){M=L(m,"ALGORITHM").replace(/^\[\[|\]\]$/g,"").trim()||"search-algorithm",u==="search"&&!e.length&&(u="algorithm");continue}if(j(w,"ROSTER")){let E=L(m,"ROSTER");E&&n.push(E);continue}if(j(w,"FARM")){let E=L(m,"FARM");E&&s.push(E);continue}if(j(w,"LIST")){!e.length&&u==="search"&&(u="list");continue}if(j(w,"STATUS")){u==="search"&&(u="status");continue}if(j(w,"SIMILAR")){let E=L(w,"SIMILAR").toLowerCase();r=dt[E]||ct,!e.length&&u==="search"&&(u="similar");continue}if(j(w,"THRESHOLD")){let E=L(m,"THRESHOLD");r=dt[E.toLowerCase()]??(parseFloat(E)||ct);continue}if(j(w,"LIMIT")){l=parseInt(L(m,"LIMIT"))||yt;continue}e.push(["PUBLIC","LOCAL","PRIVATE","GALAXY","SITE"].includes(w)?w:m)}return{mode:u,specs:e,rosterRefs:n,farms:s,threshold:r??ct,limit:l??yt,live:g,subject:k,force:x,ghostUrl:P,label:D,batch:R,algorithm:M,thresholdSet:r!==null}},X=t=>t.includes("*")||t.includes("?"),Z=t=>t==="PUBLIC"||t==="LOCAL"||t==="PRIVATE"||t==="GALAXY";var xt=t=>t.toLowerCase().replace(/\s+/g,"-").replace(/[^a-z0-9-]/g,""),ut=new Map,Qt=async(t,e)=>{if(ut.has(t))return ut.get(t);let n=`${e}/system/indexed-domains.json?pattern=${encodeURIComponent(t)}`,s=await fetch(n);if(!s.ok)throw new Error(`indexed-domains failed: ${s.status}`);let r=await s.json();return ut.set(t,r),r},K=async(t,e)=>{t.length||(t=[window.location.hostname]);let n=new Set,s=[];for(let r of t)if(r==="*"||X(r)||Z(r))for(let l of await Qt(r,e))n.has(l.domain)||(n.add(l.domain),s.push(l));else n.has(r)||(n.add(r),s.push({domain:r,page_count:null}));return s},te=/^([a-zA-Z0-9-]+(\.[a-zA-Z0-9-]+)+|localhost)(:\d+)?$/,ee=/^ROSTER ([A-Za-z0-9.\-:]+\/[a-z0-9-]+)$/,ne=/^REFERENCES ([A-Za-z0-9.\-:]+\/[a-z0-9-]+)$/,bt=async t=>{let e=t.indexOf("/"),n=await fetch(`//${t.slice(0,e)}/${t.slice(e+1)}.json`);if(!n.ok)throw new Error(`roster page ${t} failed: ${n.status}`);return n.json()},St=async(t,e,n)=>{if(n.has(t))return;n.add(t);let s;try{s=await bt(t)}catch(r){console.warn(r.message);return}for(let r of s.story||[])if(r.type==="roster")for(let l of(r.text||"").split(/\r?\n/)){let u=l.trim();if(!u)continue;let g=u.match(te);if(g){e.add(g[0]);continue}let k=u.match(ee);if(k){await St(k[1],e,n);continue}let x=u.match(ne);if(x)try{let P=await bt(x[1]);for(let D of P.story||[])D.type==="reference"&&D.site&&e.add(D.site)}catch(P){console.warn(P.message)}}},se=async t=>{let e=new Set,n=new Set;for(let s of t)await St(s,e,n);return[...e]},vt=async(t,e)=>{let n=typeof window<"u"&&window.location?window.location.hostname:null,s=t.map(r=>r.toUpperCase()==="SITE"&&n?n:r);return e.length?[...s,...await se(e)]:s};var mt=new Map,oe=t=>`${window.location.origin}/system/semantic-vectors.json?domain=${encodeURIComponent(t)}`,kt=async t=>{if(mt.has(t))return mt.get(t);let e=await fetch(oe(t));if(!e.ok)return[];let n=await e.json();return mt.set(t,n),n},Q=async(t,e)=>{let n=await fetch(`${e}/system/embed.json`,{method:"POST",headers:{"Content-Type":"application/json"},body:JSON.stringify({text:t})});if(!n.ok)throw new Error(`embed failed: ${n.status}`);return(await n.json()).vector},Lt=async(t,e)=>{let s=(await kt(e)).find(r=>r.slug===t);return s?s.vector:null},tt=(t,e,{threshold:n,limit:s,excludeSlug:r,excludeDomain:l})=>{let u=[];for(let{domain:g,pages:k}of e)for(let{slug:x,title:P,vector:D}of k){if(x===r&&g===l)continue;let R=0;for(let M=0;M<t.length;M++)R+=t[M]*D[M];R>=n&&u.push({domain:g,slug:x,title:P,score:R})}return u.sort((g,k)=>k.score-g.score),u.slice(0,s)},et=async(t,e)=>{let n=await K(t,e);return(await Promise.all(n.map(async({domain:r})=>({domain:r,pages:await kt(r)})))).filter(r=>r.pages.length>0)};var Et=t=>{let e=t.parents(".page:first"),n=e.prev(".page"),s=n.length?n:e,r=(s.attr("id")||"").split("_rev")[0],l=s.data("site")||window.location.hostname,u=s.data("data")||{},g=u.title||s.find(".title").text().trim()||r,k=[g,...(u.story||[]).filter(x=>x.type==="markdown"||x.type==="paragraph").map(x=>(x.text||"").trim()).filter(Boolean)].join(`
 `).slice(0,2e3);return{slug:r,site:l,title:g,text:k,isSelf:!n.length}},G=t=>`Subject: ${t.title}${t.isSelf?" (this page)":` @ ${t.site}`}`;var Ct=t=>`sim-cache-${t}`,jt=t=>{try{let e=JSON.parse(localStorage.getItem(Ct(t.id))||"null");return e?.text===(t.text||"")?e:null}catch{return null}},W=(t,e)=>{try{localStorage.setItem(Ct(t.id),JSON.stringify({text:t.text||"",ts:Date.now(),...e}))}catch{}},Tt=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 ie=()=>{let t=new Set;try{for(let e of Object.keys(window.wiki?.neighborhood||{}))t.add(e);document.querySelectorAll(".page").forEach(e=>{let n=window.$?window.$(e).data("site"):null;t.add(n||window.location.hostname)})}catch{}return[...t]},It=async(t,e)=>{let n=await fetch(t,{method:"POST",headers:{"Content-Type":"application/json"},body:JSON.stringify(e)});if(!n.ok){let s=new Error(`${t.split("/").pop()} failed: ${n.status}`);throw s.status=n.status,s.body=await n.json().catch(()=>({})),s}return n.json()},ae=(t,e,n)=>{let s=t.slice(0,n).map(l=>`${l.site} ${l.slug}`).join("|"),r=e.slice(0,n).map(l=>`${l.site} ${l.slug}`).join("|");return s===r},At=async t=>{let{origin:e,query:n,seeded:s={},specs:r,limit:l,batch:u,algorithm:g,render:k}=t,x=a=>t.status&&t.status(a),P={roster:[...t.roster||[],...g&&g.liked||[]],neighborhood:ie(),followed:!0,...t.prefer||{}};x("Ranking sites\u2026");let D=await It(`${e}/system/site-rank.json`,{query:n,...s,domains:r,prefer:P,algorithm:g||{},batch:u||50}),R=D.batches||[],M=new Set(R.flat().map(a=>a.toLowerCase())),j=(r||[]).filter(a=>typeof a=="string"&&/\./.test(a)&&!/[*?]/.test(a)&&!["PUBLIC","LOCAL","PRIVATE","GALAXY"].includes(a.toUpperCase())&&!M.has(a.toLowerCase()));j.length&&R.push(j);let o={searched:0,total:(D.count||0)+j.length||R.reduce((a,p)=>a+p.length,0),batches:R.length,done:0,stopped:!1,converged:!1,running:!0,unindexed:D.unindexed||[],answered:{local:0},siteIndex:D.siteIndex||null,siteInfo:new Map((D.sites||[]).map(a=>[a.domain,a]))},m=new Map,w=[],E=0,J=()=>{o.stopped=!0};o.stop=J;let V=async a=>{let p={query:n,...s,domains:a,limit:Math.max(l*3,30),flat:!0};t.thresholdSet&&(p.threshold=t.threshold);let c=await It(`${e}/system/search-report.json`,p);for(let d of c.results||[]){let y=`${d.site} ${d.slug}`,v=m.get(y);(!v||v.score<d.score)&&m.set(y,d)}let f=0;for(let d of c.peers||[])o.answered[d.host]=(o.answered[d.host]||0)+(d.domains||0),f+=d.domains||0;o.answered.local+=Math.max(0,(c.stats&&c.stats.domains||0)-f);let h=[...m.values()].sort((d,y)=>y.score-d.score),i=ae(w,h,l);w=h,o.searched+=a.length,o.done+=1,E=i?E+1:0},lt=()=>R.slice(o.done),Y=async(a=!1)=>{for(o.running=!0,o.stopped=!1;o.done<R.length&&!o.stopped;)if(await V(R[o.done]),k(w,o),!a&&E>=3&&o.done<R.length){o.converged=!0;break}return o.running=!1,k(w,o),w};return o.continueAll=()=>(o.converged=!1,E=-1/0,Y(!0)),await Y(!1),{results:w,state:o,remaining:lt}};var st="similarity:learned",re=new Set(["liked","visited","neighbourhood","neighborhood","followed","centroid","fresh","reliable"]),nt=/^([a-zA-Z0-9-]+(\.[a-zA-Z0-9-]+)+|localhost)(:\d+)?$/,ft=t=>{let e={weights:{},always:[],never:[],batch:null,trust:null};for(let n of(t||"").split(`

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "wiki-plugin-similarity",
-  "version": "0.21.1",
+  "version": "0.21.2",
   "description": "Federated Wiki plugin — semantic similarity search across wiki domains using prebuilt vector indexes",
   "keywords": [
     "wiki",

package/pages/about-similarity-plugin

@@ -130,6 +130,11 @@
       "type": "markdown",
       "id": "ac44fae5937e27eb",
       "text": "**0.21.1** \u2014 three fixes from the first day of the second peer. A farm's copy of a peer's Site Index is checked **hourly** by conditional GET, and always once at startup, so a restart never serves a copy the peer has since rebuilt \u2014 it was six hours, and a 304 reset the clock, so a new indexer's sites could take most of a day to reach the Cafe. A domain named explicitly in a batched item is **searched whether or not the Site Index knows it**, as one more batch, since the cascade finds the peer that holds it. And a batched item's ready line reports how far the Site Index reaches rather than how many domains this host holds, which on a farm with no galaxy tree read as a misleading zero."
+    },
+    {
+      "type": "markdown",
+      "id": "99d1f2910f05b38d",
+      "text": "**0.21.2** \u2014 a site answered two hops away was counted twice in the *answered by* line, once under the peer that relayed it and once under the peer that held it. The relay's count now excludes what its own peers answered."
     }
   ],
   "plugin": "similarity",
@@ -490,6 +495,17 @@
       },
       "after": "e2cde2572a99224e",
       "date": 1788551290021
+    },
+    {
+      "type": "add",
+      "id": "99d1f2910f05b38d",
+      "item": {
+        "type": "markdown",
+        "id": "99d1f2910f05b38d",
+        "text": "**0.21.2** \u2014 a site answered two hops away was counted twice in the *answered by* line, once under the peer that relayed it and once under the peer that held it. The relay's count now excludes what its own peers answered."
+      },
+      "after": "ac44fae5937e27eb",
+      "date": 1788552261437
     }
   ]
 }

package/server/cascade.js

@@ -42,9 +42,15 @@ const askCascade = async ({ peers, missing, body, hops = 0, via = [], post, time
         continue
       }
       for (const r of ans.body.results) out.results.push({ ...r, via: r.via || host })
+      // A peer's stats include what its own peers answered; count those
+      // under the host that really held them, not twice.
       const s = ans.body.stats || {}
-      out.peers.push({ host, domains: s.domains || 0, pages: s.pages || 0 })
-      for (const p of ans.body.peers || []) out.peers.push(p)   // what it asked in turn
+      const nested = ans.body.peers || []
+      const nestedDomains = nested.reduce((n, p) => n + (p.domains || 0), 0)
+      const nestedPages = nested.reduce((n, p) => n + (p.pages || 0), 0)
+      out.peers.push({ host, domains: Math.max(0, (s.domains || 0) - nestedDomains),
+                       pages: Math.max(0, (s.pages || 0) - nestedPages) })
+      for (const p of nested) out.peers.push(p)   // what it asked in turn
       // A peer that says what it held lets the walk go on with the rest; an
       // older peer that does not is taken to have answered for everything.
       const held = Array.isArray(ans.body.held) ? new Set(ans.body.held) : null