@chatpanel/events 0.102.0 → 0.103.1

package/web-search.js

@@ -1,160 +1 @@
-// Reading a search engine's results page: the rules, without the DOM.
-//
-// Scraping a SERP is mostly judgement, and the judgement is identical everywhere: which
-// engines to ask, how to build their URL safely from a template the user can edit, how to
-// unwrap the redirector every engine wraps its links in, and which of the hundred anchors on
-// the page are actually RESULTS rather than the engine's own navigation, promos and
-// newsletter sign-ups.
-//
-// None of that needs a document. What does need one is turning HTML into a list of anchors,
-// and that is the single thing each host injects: the extension has `DOMParser`, the desktop's
-// main process has neither and uses a small extractor, and a gateway that grew one could pass
-// its own. So there is ONE scraper with three front doors, rather than three scrapers.
-//
-// THE DANGEROUS PART IS THE URL, NOT THE HTML. A search template is user input that becomes a
-// fetch, so `buildSearchUrl` refuses anything that is not https and anything without a query
-// placeholder — and the caller is required to pass the host guard it already owns
-// (`assertFetchable`), because deciding whether an address is safe to fetch is not a question
-// this module is allowed to answer on its own.
-
-/**
- * The engines, in the order they are tried.
- *
- * `enabled` is what a fresh install starts with. Mojeek is absent rather than disabled:
- * tested, it returns nothing at all, and an engine that never answers is not a fallback — it
- * is latency plus a misleading "no results".
- */
-export const SEARCH_ENGINES = Object.freeze([
-  Object.freeze({ id: 'startpage', name: 'Startpage', url: 'https://www.startpage.com/sp/search?query=%s', enabled: true }),
-  Object.freeze({ id: 'duckduckgo', name: 'DuckDuckGo', url: 'https://html.duckduckgo.com/html/?q=%s', enabled: true }),
-  Object.freeze({ id: 'google', name: 'Google', url: 'https://www.google.com/search?q=%s', enabled: false }),
-  Object.freeze({ id: 'bing', name: 'Bing', url: 'https://www.bing.com/search?q=%s', enabled: false }),
-]);
-
-/** How many results to take from one engine before moving on. */
-export const RESULTS_PER_ENGINE = 5;
-
-/**
- * Turn a template into a URL, refusing the ones that are not safe to fetch.
- *
- * `assertFetchable` is INJECTED rather than implemented here: the SSRF/private-host guard is
- * one shared primitive that the client, the gateway and the bridge all call, and a second
- * opinion about what counts as a private address is how the two drift apart. Omitting it is
- * allowed only where the caller has already checked — it throws loudly rather than silently
- * skipping, so "I forgot" and "I checked elsewhere" cannot look the same.
- */
-export function buildSearchUrl(template, query, { assertFetchable } = {}) {
-  const t = String(template || '').trim();
-  if (!/^https:\/\//i.test(t)) throw new Error('Search engine URL must start with https://');
-  if (!t.includes('%s') && !t.includes('{q}')) {
-    throw new Error('Search engine URL must contain a %s (or {q}) query placeholder');
-  }
-  const url = t.replace(/%s|\{q\}/g, encodeURIComponent(String(query || '').trim()));
-  if (typeof assertFetchable === 'function') assertFetchable(url);
-  return url;
-}
-
-/**
- * Unwrap the redirector an engine wraps its results in, so what is fetched and CITED is the
- * real destination rather than the engine's tracking URL.
- */
-export function unwrapRedirect(href) {
-  try {
-    const u = new URL(href, 'https://duckduckgo.com');
-    const uddg = u.searchParams.get('uddg'); // DuckDuckGo /l/?uddg=<encoded>
-    if (uddg) return decodeURIComponent(uddg);
-    const other = u.searchParams.get('url') || u.searchParams.get('u');
-    if (other && /^https?:/i.test(other)) return other;
-    return u.href;
-  } catch {
-    return String(href || '');
-  }
-}
-
-/**
- * Hosts that appear on a SERP and are never results: the engine's own family, and the promos
- * they carry (Startpage's StartMail, Mojeek's newsletter).
- *
- * Deliberately NOT a broad denylist — real results legitimately point at YouTube, Yahoo and
- * Google properties, and filtering those would be filtering the web.
- */
-const JUNK_HOSTS = /(^|\.)(startpage\.com|startmail\.com|startpage\.dev|mojeek\.com|buttondown\.(com|email)|ecosia\.org|duckduckgo\.com|search\.brave\.com|qwant\.com)$/i;
-
-/** Is this a link off the engine's own page, or part of the engine itself? */
-export function isResultHost(hostname, engineHost = '') {
-  const host = String(hostname || '').toLowerCase();
-  if (!host) return false;
-  if (JUNK_HOSTS.test(host)) return false;
-  const engineDomain = String(engineHost || '').split('.').slice(-2).join('.').toLowerCase();
-  if (engineDomain && (host === engineDomain || host.endsWith(`.${engineDomain}`))) return false;
-  return true;
-}
-
-/**
- * Turn a page's anchors into results.
- *
- * `anchors` is what the host's extractor produced: `{ href, text, snippet?, chrome? }`, where
- * `chrome` marks an anchor that sat inside a nav, footer or promo block. Everything after this
- * point is identical in every client, which is the reason the module exists.
- *
- * Deduped on origin+path rather than the full URL, so the same page offered twice with
- * different tracking parameters counts once.
- */
-export function pickResults(anchors, { engineHost = '', limit = RESULTS_PER_ENGINE, fallback = false } = {}) {
-  const out = [];
-  const seen = new Set();
-  for (const a of anchors || []) {
-    const href = unwrapRedirect(a?.href || '');
-    let u;
-    try { u = new URL(href); } catch { continue; }
-    if (u.protocol !== 'https:' && u.protocol !== 'http:') continue;
-    if (!isResultHost(u.hostname, engineHost)) continue;
-    // The generic sweep sees the WHOLE page, so it is filtered harder: an anchor inside the
-    // page's own chrome is navigation, not a result.
-    if (fallback && a.chrome) continue;
-    const title = String(a?.text || '').replace(/\s+/g, ' ').trim();
-    // A link whose visible text is a word or two is an icon, a "next", or a breadcrumb.
-    if (title.length < 6) continue;
-    const key = u.origin + u.pathname;
-    if (seen.has(key)) continue;
-    seen.add(key);
-    out.push({ url: href, title, snippet: String(a?.snippet || '').replace(/\s+/g, ' ').trim() });
-    if (out.length >= limit) break;
-  }
-  return out;
-}
-
-/**
- * Merge what several engines returned into one list.
- *
- * Engine ORDER is preserved rather than interleaved: the first engine is the one the user put
- * first, and a result it ranked third is more likely to be wanted than the second engine's
- * first. Deduped across engines on origin+path, so two engines agreeing shows once.
- */
-export function mergeEngineResults(perEngine, limit = 8) {
-  const out = [];
-  const seen = new Set();
-  for (const results of perEngine || []) {
-    for (const r of results || []) {
-      let key;
-      try { const u = new URL(r.url); key = u.origin + u.pathname; } catch { key = r.url; }
-      if (seen.has(key)) continue;
-      seen.add(key);
-      out.push(r);
-      if (out.length >= limit) return out;
-    }
-  }
-  return out;
-}
-
-/**
- * The engines to try, in order: the enabled ones first, then the rest.
- *
- * A search that found nothing should escalate to the engines the user configured but left off
- * rather than report failure — being wrong about which engine works today is much more likely
- * than the web having no answer.
- */
-export function engineOrder(engines = SEARCH_ENGINES) {
-  const list = (Array.isArray(engines) ? engines : []).filter((e) => e && e.url && !e.retired);
-  return [...list.filter((e) => e.enabled !== false), ...list.filter((e) => e.enabled === false)];
-}
+const f=Object.freeze([Object.freeze({id:"startpage",name:"Startpage",url:"https://www.startpage.com/sp/search?query=%s",enabled:!0}),Object.freeze({id:"duckduckgo",name:"DuckDuckGo",url:"https://html.duckduckgo.com/html/?q=%s",enabled:!0}),Object.freeze({id:"google",name:"Google",url:"https://www.google.com/search?q=%s",enabled:!1}),Object.freeze({id:"bing",name:"Bing",url:"https://www.bing.com/search?q=%s",enabled:!1})]),h=5;function m(n,r,{assertFetchable:e}={}){const t=String(n||"").trim();if(!/^https:\/\//i.test(t))throw new Error("Search engine URL must start with https://");if(!t.includes("%s")&&!t.includes("{q}"))throw new Error("Search engine URL must contain a %s (or {q}) query placeholder");const s=t.replace(/%s|\{q\}/g,encodeURIComponent(String(r||"").trim()));return typeof e=="function"&&e(s),s}function p(n){try{const r=new URL(n,"https://duckduckgo.com"),e=r.searchParams.get("uddg");if(e)return decodeURIComponent(e);const t=r.searchParams.get("url")||r.searchParams.get("u");return t&&/^https?:/i.test(t)?t:r.href}catch{return String(n||"")}}const g=/(^|\.)(startpage\.com|startmail\.com|startpage\.dev|mojeek\.com|buttondown\.(com|email)|ecosia\.org|duckduckgo\.com|search\.brave\.com|qwant\.com)$/i;function d(n,r=""){const e=String(n||"").toLowerCase();if(!e||g.test(e))return!1;const t=String(r||"").split(".").slice(-2).join(".").toLowerCase();return!(t&&(e===t||e.endsWith(`.${t}`)))}function w(n,{engineHost:r="",limit:e=h,fallback:t=!1}={}){const s=[],c=new Set;for(const o of n||[]){const a=p(o?.href||"");let i;try{i=new URL(a)}catch{continue}if(i.protocol!=="https:"&&i.protocol!=="http:"||!d(i.hostname,r)||t&&o.chrome)continue;const u=String(o?.text||"").replace(/\s+/g," ").trim();if(u.length<6)continue;const l=i.origin+i.pathname;if(!c.has(l)&&(c.add(l),s.push({url:a,title:u,snippet:String(o?.snippet||"").replace(/\s+/g," ").trim()}),s.length>=e))break}return s}function S(n,r=8){const e=[],t=new Set;for(const s of n||[])for(const c of s||[]){let o;try{const a=new URL(c.url);o=a.origin+a.pathname}catch{o=c.url}if(!t.has(o)&&(t.add(o),e.push(c),e.length>=r))return e}return e}function b(n=f){const r=(Array.isArray(n)?n:[]).filter(e=>e&&e.url&&!e.retired);return[...r.filter(e=>e.enabled!==!1),...r.filter(e=>e.enabled===!1)]}export{h as RESULTS_PER_ENGINE,f as SEARCH_ENGINES,m as buildSearchUrl,b as engineOrder,d as isResultHost,S as mergeEngineResults,w as pickResults,p as unwrapRedirect};

package/widget.js

@@ -1,153 +1 @@
-// WIDGETS — small apps the user asks for and the model builds, which then live in THEIR
-// ChatPanel. A timer, a calculator, a sticky note, a unit converter, a habit tracker: things
-// nobody should have to file a feature request for. Two people's ChatPanel can differ
-// entirely, without either of them writing code and without us shipping a release.
-//
-// A widget is NOT a capability. A capability is reviewed and approved before it runs, so it
-// can be trusted to act. A widget is written by a model on a user's whim, so it is treated as
-// exactly what it is — untrusted code — and given the smallest surface that still makes it
-// useful:
-//
-//   • its own state, and nothing else's. `state` is a private object keyed by widget id. A
-//     timer remembers where it got to; a sticky note remembers its text. No widget can read
-//     another's, and none can read the user's notes, meetings or chats.
-//   • no network, no chrome.*, no DOM outside its sandbox — enforced by the host, not by
-//     good behaviour here.
-//
-// Anything beyond that is a GRANT: a widget may REQUEST capabilities in its manifest, and
-// those do nothing until the user approves them. Requesting is not receiving — `grants` is
-// stored separately from the manifest precisely so a widget cannot edit its own permissions
-// by rewriting its own code.
-//
-// Pure and host-free: no DOM, no storage, no postMessage. The client owns the transport and
-// the persistence; this owns what is legal.
-
-import { EventError } from './event.js';
-
-const str = (v) => typeof v === 'string' && v.length > 0;
-const MAX_HTML = 512 * 1024;   // a small app, not a bundled framework
-const MAX_STATE = 256 * 1024;  // a note, a lap list — not a database
-
-export const WIDGET_SURFACES = Object.freeze(['panel', 'chat']);
-
-/**
- * Validate a widget MANIFEST — everything a user sees before deciding to keep it.
- */
-export function validateWidget(w) {
-  if (!w || typeof w !== 'object') throw new EventError('SHAPE', 'widget must be an object');
-  if (!str(w.id)) throw new EventError('SHAPE', 'widget.id required');
-  if (!/^[a-z0-9][a-z0-9_-]{0,63}$/.test(w.id)) {
-    throw new EventError('SHAPE', 'widget.id must be lowercase [a-z0-9_-], max 64 chars');
-  }
-  if (!str(w.name)) throw new EventError('SHAPE', 'widget.name required');
-  if (!str(w.html)) throw new EventError('SHAPE', 'widget.html required');
-  if (w.html.length > MAX_HTML) throw new EventError('SHAPE', `widget.html exceeds ${MAX_HTML} bytes`);
-  // An icon NAME (the client maps it to its own icon set), so a pinned widget is
-  // recognisable at a glance rather than being one of five identical marks.
-  if (w.icon != null && !(str(w.icon) && /^[a-z][a-z0-9-]{0,31}$/.test(w.icon))) {
-    throw new EventError('SHAPE', 'widget.icon must be an icon name like "timer"');
-  }
-  if (w.surface != null && !WIDGET_SURFACES.includes(w.surface)) {
-    throw new EventError('SHAPE', `widget.surface must be one of ${WIDGET_SURFACES}`);
-  }
-  if (w.height != null && !(Number.isInteger(w.height) && w.height > 0 && w.height <= 2000)) {
-    throw new EventError('SHAPE', 'widget.height must be a positive integer <= 2000');
-  }
-  // Requesting is not receiving. This only records what the widget WANTS; the grant lives
-  // outside the manifest so rewriting the widget can never widen its own permissions.
-  if (w.requests != null && !(Array.isArray(w.requests) && w.requests.every(str))) {
-    throw new EventError('SHAPE', 'widget.requests must be string[] (capability ids it asks for)');
-  }
-  return w;
-}
-
-/**
- * Validate a message a widget sent. Untrusted input: nothing is believed, and the widget's
- * own id is supplied by the HOST (which knows which frame sent it), never read from the
- * message — otherwise a widget could name someone else's id and read their state.
- */
-export function validateWidgetMessage(msg, { widgetId, grants = [] } = {}) {
-  if (!str(widgetId)) throw new EventError('SHAPE', 'host must supply the sending widget id');
-  if (!msg || typeof msg !== 'object') throw new EventError('SHAPE', 'widget message must be an object');
-  if (!str(msg.callId)) throw new EventError('SHAPE', 'widget message needs a callId');
-
-  switch (msg.op) {
-    case 'state.get':
-      return { op: 'state.get', widgetId, callId: msg.callId };
-
-    case 'state.set': {
-      if (msg.state === undefined) throw new EventError('SHAPE', 'state.set needs state');
-      let size = 0;
-      try { size = JSON.stringify(msg.state ?? null).length; }
-      catch { throw new EventError('SHAPE', 'widget state must be JSON-serialisable'); }
-      if (size > MAX_STATE) throw new EventError('SHAPE', `widget state exceeds ${MAX_STATE} bytes`);
-      return { op: 'state.set', widgetId, callId: msg.callId, state: msg.state };
-    }
-
-    case 'invoke': {
-      if (!str(msg.capability)) throw new EventError('SHAPE', 'invoke must name a capability');
-      // THE REFUSAL THAT MATTERS. A widget gets what the user granted it and nothing else —
-      // checked here, before any kernel guard has to have an opinion, and against the stored
-      // grants rather than against anything the widget or its manifest claims.
-      if (!grants.includes(msg.capability)) {
-        throw new EventError('DENIED',
-          `widget "${widgetId}" has no grant for "${msg.capability}" — the user must approve it first`);
-      }
-      if (msg.args != null && (typeof msg.args !== 'object' || Array.isArray(msg.args))) {
-        throw new EventError('SHAPE', 'invoke args must be an object');
-      }
-      return { op: 'invoke', widgetId, callId: msg.callId, capability: msg.capability, args: msg.args || {} };
-    }
-
-    default:
-      throw new EventError('SHAPE', `unknown widget op "${msg.op}"`);
-  }
-}
-
-/**
- * The permissions a widget actually holds: the intersection of what it asked for and what the
- * user approved. Asking for more later cannot grant it — a request that was never approved
- * stays absent, so an updated widget re-asking silently gains nothing.
- */
-export function effectiveGrants(widget, approved = []) {
-  const asked = new Set(widget?.requests || []);
-  return (approved || []).filter((id) => asked.has(id));
-}
-
-// Pick an icon for a widget from what it is called. A pinned widget sits in a narrow strip
-// next to the others, so five identical marks are worse than no icon at all — the point of
-// pinning is to find the thing without reading.
-//
-// Returns an icon NAME, not a glyph: the client owns how it is drawn (the extension resolves
-// these through its vendored set, a mobile client through its own), and a name survives that
-// mapping in a way an emoji does not. Every name here exists in the extension's icon set.
-const ICON_WORDS = [
-  [/\b(timers?|pomodoro|stopwatch|countdowns?|intervals?)\b/i, 'timer'],
-  [/\b(calc|calculators?|math|arithmetic|tip)\b/i, 'hash'],
-  [/\b(notes?|sticky|scratch|memos?|journals?)\b/i, 'notebook-pen'],
-  [/\b(todos?|tasks?|checklists?|habits?|trackers?)\b/i, 'list-checks'],
-  [/\b(charts?|graphs?|stats?|metrics?|dashboards?)\b/i, 'bar-chart-3'],
-  [/\b(convert|converters?|units?|currency|exchange|weights?)\b/i, 'scale'],
-  [/\b(calendars?|schedules?|agendas?|dates?)\b/i, 'calendar'],
-  [/\b(clocks?|time|timezones?|zones?)\b/i, 'clock'],
-  [/\b(goals?|targets?|focus|okrs?)\b/i, 'target'],
-  [/\b(ideas?|brainstorm|prompts?)\b/i, 'lightbulb'],
-  [/\b(dice|random|rolls?|roller|coins?|shuffle)\b/i, 'zap'],
-  [/\b(search|find|lookup)\b/i, 'search'],
-  [/\b(web|urls?|links?|browser)\b/i, 'globe'],
-  [/\b(mood|mind|memory|brain)\b/i, 'brain'],
-  [/\b(password|secrets?|vault|lock)\b/i, 'lock'],
-  [/\b(quotes?|sayings?)\b/i, 'quote'],
-  [/\b(meetings?|standups?|people|team)\b/i, 'users'],
-  [/\b(music|player|sound|audio)\b/i, 'play'],
-  [/\b(photos?|images?|gallery)\b/i, 'image'],
-  [/\b(files?|documents?|docs?)\b/i, 'file-text'],
-];
-
-/** The widget's own icon if it declared one, else one derived from its name. */
-export function widgetIcon(widget) {
-  if (widget?.icon) return widget.icon;
-  const name = String(widget?.name || '');
-  for (const [re, iconName] of ICON_WORDS) if (re.test(name)) return iconName;
-  return 'app-window'; // generic, but never the mark the shelf itself uses
-}
+import{EventError as t}from"./event.js";const s=e=>typeof e=="string"&&e.length>0,o=512*1024,n=256*1024,c=Object.freeze(["panel","chat"]);function u(e){if(!e||typeof e!="object")throw new t("SHAPE","widget must be an object");if(!s(e.id))throw new t("SHAPE","widget.id required");if(!/^[a-z0-9][a-z0-9_-]{0,63}$/.test(e.id))throw new t("SHAPE","widget.id must be lowercase [a-z0-9_-], max 64 chars");if(!s(e.name))throw new t("SHAPE","widget.name required");if(!s(e.html))throw new t("SHAPE","widget.html required");if(e.html.length>o)throw new t("SHAPE",`widget.html exceeds ${o} bytes`);if(e.icon!=null&&!(s(e.icon)&&/^[a-z][a-z0-9-]{0,31}$/.test(e.icon)))throw new t("SHAPE",'widget.icon must be an icon name like "timer"');if(e.surface!=null&&!c.includes(e.surface))throw new t("SHAPE",`widget.surface must be one of ${c}`);if(e.height!=null&&!(Number.isInteger(e.height)&&e.height>0&&e.height<=2e3))throw new t("SHAPE","widget.height must be a positive integer <= 2000");if(e.requests!=null&&!(Array.isArray(e.requests)&&e.requests.every(s)))throw new t("SHAPE","widget.requests must be string[] (capability ids it asks for)");return e}function h(e,{widgetId:i,grants:a=[]}={}){if(!s(i))throw new t("SHAPE","host must supply the sending widget id");if(!e||typeof e!="object")throw new t("SHAPE","widget message must be an object");if(!s(e.callId))throw new t("SHAPE","widget message needs a callId");switch(e.op){case"state.get":return{op:"state.get",widgetId:i,callId:e.callId};case"state.set":{if(e.state===void 0)throw new t("SHAPE","state.set needs state");let r=0;try{r=JSON.stringify(e.state??null).length}catch{throw new t("SHAPE","widget state must be JSON-serialisable")}if(r>n)throw new t("SHAPE",`widget state exceeds ${n} bytes`);return{op:"state.set",widgetId:i,callId:e.callId,state:e.state}}case"invoke":{if(!s(e.capability))throw new t("SHAPE","invoke must name a capability");if(!a.includes(e.capability))throw new t("DENIED",`widget "${i}" has no grant for "${e.capability}" — the user must approve it first`);if(e.args!=null&&(typeof e.args!="object"||Array.isArray(e.args)))throw new t("SHAPE","invoke args must be an object");return{op:"invoke",widgetId:i,callId:e.callId,capability:e.capability,args:e.args||{}}}default:throw new t("SHAPE",`unknown widget op "${e.op}"`)}}function d(e,i=[]){const a=new Set(e?.requests||[]);return(i||[]).filter(r=>a.has(r))}const l=[[/\b(timers?|pomodoro|stopwatch|countdowns?|intervals?)\b/i,"timer"],[/\b(calc|calculators?|math|arithmetic|tip)\b/i,"hash"],[/\b(notes?|sticky|scratch|memos?|journals?)\b/i,"notebook-pen"],[/\b(todos?|tasks?|checklists?|habits?|trackers?)\b/i,"list-checks"],[/\b(charts?|graphs?|stats?|metrics?|dashboards?)\b/i,"bar-chart-3"],[/\b(convert|converters?|units?|currency|exchange|weights?)\b/i,"scale"],[/\b(calendars?|schedules?|agendas?|dates?)\b/i,"calendar"],[/\b(clocks?|time|timezones?|zones?)\b/i,"clock"],[/\b(goals?|targets?|focus|okrs?)\b/i,"target"],[/\b(ideas?|brainstorm|prompts?)\b/i,"lightbulb"],[/\b(dice|random|rolls?|roller|coins?|shuffle)\b/i,"zap"],[/\b(search|find|lookup)\b/i,"search"],[/\b(web|urls?|links?|browser)\b/i,"globe"],[/\b(mood|mind|memory|brain)\b/i,"brain"],[/\b(password|secrets?|vault|lock)\b/i,"lock"],[/\b(quotes?|sayings?)\b/i,"quote"],[/\b(meetings?|standups?|people|team)\b/i,"users"],[/\b(music|player|sound|audio)\b/i,"play"],[/\b(photos?|images?|gallery)\b/i,"image"],[/\b(files?|documents?|docs?)\b/i,"file-text"]];function f(e){if(e?.icon)return e.icon;const i=String(e?.name||"");for(const[a,r]of l)if(a.test(i))return r;return"app-window"}export{c as WIDGET_SURFACES,d as effectiveGrants,u as validateWidget,h as validateWidgetMessage,f as widgetIcon};