@aglyn/shared-util-email 1.0.0-beta.221 → 1.0.0-beta.222

package/src/lib/hosted-page-screen.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../../../../../../libs/shared/util/email/src/lib/hosted-page-screen.ts"],"sourcesContent":["/**\n * @license\n * Copyright 2026 Aglyn LLC\n *\n * Licensed under the Apache License, Version 2.0 (the \"License\");\n * you may not use this file except in compliance with the License.\n * You may obtain a copy of the License at\n *\n *   http://www.apache.org/licenses/LICENSE-2.0\n *\n * Unless required by applicable law or agreed to in writing, software\n * distributed under the License is distributed on an \"AS IS\" BASIS,\n * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n * See the License for the specific language governing permissions and\n * limitations under the License.\n */\n\n/*==========================================\n * THE HOSTED PAGE SCREEN (AGL-3362).\n *\n * The outbound phishing screen, read over a PUBLISHED PAGE instead of an\n * email: the same brand list, the same lookalike logic, the same tiers\n * (`signalsThatHold`). A fraud actor who could not mail a lure could still\n * host one, and the incident's sibling site published a page at the\n * document-share path `/reviewfile`.\n *\n * Five shapes:\n *\n * 1. **A lookalike link or embed** (`lookalike-link`, STRONG) — any `href`,\n *    `src`, form action or URL in the page's authored content whose host\n *    wears a brand without being the brand's domain.\n * 2. **A credential field** (`credential-field`, STRONG) — a field the\n *    merchant AUTHORED that asks for a password, a card number or a\n *    one-time code: its type, its autocomplete token, or its label,\n *    placeholder or name. The platform's own sign-in, account and checkout\n *    elements render their inputs in code, not in stored props, so they are\n *    never read as a field here — which is exactly the sanctioned/unsanctioned\n *    line: what the platform collects through its own components is\n *    Aglyn's to protect, and what a page's author collects in a form of\n *    their own is not something a site may ask a stranger for.\n * 3. **A brand's call to action** (`brand-action-page`, SOFT) — one element\n *    whose text names a brand that is not the workspace's AND asks the reader\n *    to sign in, verify, confirm or open a document, on a page that links\n *    somewhere that is neither the workspace's, nor a brand's own, nor a\n *    common social or maps link.\n * 4. **A brand's lure, page-wide** (`brand-lure-page`, SOFT, AGL-3447) — the\n *    same brand and link anywhere on the page, beside an account or\n *    document-share LURE anywhere on it (\"Secure Document Access Portal\"), or\n *    a call to action that leaves the site worded as an action (\"Continue to\n *    Document\"). The 2026-10-01 phishing page put Proofpoint's name in its\n *    body, the lure in its heading and the action on its button: three\n *    elements, none of which shape 3 reads on its own. What it does not count page-wide is\n *    a bare \"sign in\" that stays on the site — a hotel that says \"find us on\n *    Booking.com\" in one place and has a member sign-in in another is two\n *    ordinary sentences.\n * 5. **A call to action that leaves, beside a lure** (`offsite-action-page`,\n *    SOFT, AGL-3447) — the page's only call to action, or one worded as an\n *    account or document action, sends visitors to a host that is not the\n *    site's own, a listed brand's or a common link, on a page that carries an\n *    account or document-share lure. No brand needed: a \"Secure Document\n *    Access Portal\" whose one button leaves for the kit is the shape whatever\n *    name it wears.\n *\n * Shapes 3-5 are soft, so they hold only for a workspace in its first\n * fortnight (`signalsThatHold`); an established workspace's page is held by\n * shapes 1 and 2 alone.\n *\n * The same reading applies to a site REDIRECT ({@link screenSiteRedirect}):\n * a rule that sends a path off the site is a page whose one call to action\n * fires on arrival, and its path is the only text it has.\n *\n * Pure: the caller hands over the composed node tree and the workspace's own\n * names and hosts, and decides what a hold does.\n *=========================================*/\n\nimport {\n  ACCOUNT_LURE_PATTERNS,\n  DOCUMENT_SHARE_LURE_PATTERNS,\n  isWorkspaceOwnBrand,\n  linkHostsIn,\n  isBrandsOwnDomain,\n  lookalikeBrandForHost,\n  PHISHING_LURE_PATTERNS,\n  PHISHING_SCREEN_BRANDS,\n  type PhishingScreenBrand,\n  type PhishingScreenSignal,\n  registrableDomain,\n  isAnyOfficialBrandDomain,\n  isCommonLinkDomain,\n  squashScreenText,\n  visibleTextOf,\n} from './outbound-phishing-screen'\n\nexport interface HostedPageScreenInput {\n  /**\n   * The page as it will be served — the composed tree, or any object that\n   * holds nodes (`{ componentId, props }`) at any depth. Walked, never\n   * trusted to have one shape: a screen, a layout's chrome, a grafted\n   * component and a form all arrive together.\n   */\n  nodes: unknown\n  /** The workspace's own names, as `screenOutboundEmail` reads them. */\n  ownNames?: readonly (string | null | undefined)[]\n  /** The workspace's own hosts — a link to one is never \"elsewhere\". */\n  ownDomains?: readonly (string | null | undefined)[]\n}\n\nexport interface HostedPageScreenVerdict {\n  signals: PhishingScreenSignal[]\n  /**\n   * Every host the page (or the redirect) links, embeds or posts to, lowercased — what the\n   * link reputation lookup reads (AGL-3451), so the page is walked once.\n   */\n  linkHosts: string[]\n}\n\n/** A node as the walk finds it. */\ninterface WalkedNode {\n  componentId: string\n  props: Record<string, unknown>\n}\n\n/** The most a page's text is read to, so a huge page costs a bounded scan. */\nconst MAX_PAGE_TEXT = 2_000_000\n/** The most nodes the walk visits. */\nconst MAX_NODES = 20_000\n\n/** Props that describe a field the author defined, rather than content. */\nconst FIELD_PROP_KEYS = ['fieldType', 'fieldName', 'inputType', 'autoComplete', 'autocomplete']\n\n/** Autocomplete tokens (WHATWG) that name a credential. */\nconst CREDENTIAL_AUTOCOMPLETE: Record<string, 'password' | 'card' | 'otp'> = {\n  'current-password': 'password',\n  'new-password': 'password',\n  'cc-number': 'card',\n  'cc-csc': 'card',\n  'cc-exp': 'card',\n  'one-time-code': 'otp',\n}\n\n/** Field wording that asks for a credential, by kind. */\nconst CREDENTIAL_LABELS: ReadonlyArray<{ field: 'password' | 'card' | 'otp'; pattern: RegExp }> = [\n  { field: 'password', pattern: /\\bpass\\s?(?:word|code|phrase)s?\\b|\\bpasswd\\b|\\bpin\\s*(?:code|number)?\\s*$/i },\n  {\n    field: 'card',\n    pattern:\n      /\\b(?:credit|debit)\\s*card\\b|\\bcard\\s*(?:number|no\\.?|#)|\\bcardnumber\\b|\\bcvv2?\\b|\\bcvc\\b|\\bcard\\s*security\\s*code\\b|\\bcard\\s*expir/i,\n  },\n  {\n    field: 'otp',\n    pattern:\n      /\\bone[-\\s]?time\\s*(?:pass\\s?code|password|code|pin)\\b|\\botp\\b|\\bverification\\s*code\\b|\\b2fa\\b|\\btwo[-\\s]?factor\\b|\\bauthenticat(?:ion|or)\\s*code\\b|\\bsms\\s*code\\b/i,\n  },\n]\n\n/** Raw HTML an author placed that holds a credential input. */\nconst HTML_CREDENTIAL_INPUTS: ReadonlyArray<{ field: 'password' | 'card' | 'otp'; pattern: RegExp }> = [\n  { field: 'password', pattern: /<input\\b[^>]*\\btype\\s*=\\s*[\"']?password/i },\n  { field: 'password', pattern: /<input\\b[^>]*\\b(?:name|autocomplete)\\s*=\\s*[\"']?(?:pass(?:word|wd)?|current-password|new-password)\\b/i },\n  { field: 'card', pattern: /<input\\b[^>]*\\b(?:name|autocomplete)\\s*=\\s*[\"']?(?:card_?number|cc-number|cc-csc|cvv|cvc)\\b/i },\n  { field: 'otp', pattern: /<input\\b[^>]*\\b(?:name|autocomplete)\\s*=\\s*[\"']?(?:otp|one-time-code|verification_?code)\\b/i },\n]\n\n/**\n * What an element asks its reader to DO with an account or a document —\n * counted beside a brand in the SAME element, or on a call to action that\n * leaves the site. The email lure phrasing counts too.\n */\nconst PAGE_ACTION_PATTERNS: readonly RegExp[] = [\n  /\\b(?:sign|log)[\\s-]?in\\b/i,\n  /\\bverify\\b/i,\n  /\\bconfirm\\s+(?:your|the)\\b/i,\n  /\\bunlock\\b/i,\n  /\\benter\\s+your\\b/i,\n  /\\b(?:view|review|open|download|access)\\s+(?:the\\s+|your\\s+)?(?:shared\\s+|secure\\s+)?(?:file|document|invoice|statement)s?\\b/i,\n  /\\b(?:continue|proceed)\\s+to\\s+(?:the\\s+|your\\s+)?(?:document|file|account|sign[\\s-]?in|log[\\s-]?in)s?\\b/i,\n]\n\n/**\n * The lures a page is read for as a WHOLE (AGL-3447): account and\n * document-share wording. Not the marketplace, booking and parcel lures,\n * which a hotel's \"guest reviews\" and a shop's delivery page carry every day\n * and which still count in one element beside a brand.\n */\nconst PAGE_LURE_PATTERNS: readonly RegExp[] = [...ACCOUNT_LURE_PATTERNS, ...DOCUMENT_SHARE_LURE_PATTERNS]\n\n/**\n * A component that renders a call to action — a button or a link, however a\n * plugin names it (`muiButton`, `muiScreenLink`, `muiLinkBox`). It is one\n * when it carries a target: an `href`, or a `screenId` on the site itself.\n */\nconst CALL_TO_ACTION_COMPONENT = /button|fab|cta|link/i\n\n/** The props a call to action's visible words live in; the first found wins. */\nconst CALL_TO_ACTION_LABEL_KEYS = ['children', 'label', 'text', 'title', 'html']\n\n/** URLs in page text, so prose is read for a brand or a lure without them. */\nconst PROSE_URL_PATTERN = /(?:https?:)?\\/\\/[^\\s\"'<>]+|\\bwww\\.[^\\s\"'<>]+/gi\n\n/** One call to action: what it says, and the host it leaves for (null: it stays). */\ninterface CallToAction {\n  label: string\n  host: string | null\n}\n\n/** Every node in the tree, breadth-first in document order, bounded. */\nfunction walkNodes(root: unknown): WalkedNode[] {\n  const found: WalkedNode[] = []\n  const seen = new Set<unknown>()\n  const stack: unknown[] = [root]\n  for (let head = 0; head < stack.length && found.length < MAX_NODES; head += 1) {\n    const value = stack[head]\n    if (!value || typeof value !== 'object' || seen.has(value)) continue\n    seen.add(value)\n    const record = value as Record<string, unknown>\n    if (typeof record['componentId'] === 'string') {\n      const props = record['props']\n      found.push({\n        componentId: record['componentId'] as string,\n        props: props && typeof props === 'object' ? (props as Record<string, unknown>) : {},\n      })\n    }\n    for (const child of Array.isArray(value) ? value : Object.values(record)) {\n      if (child && typeof child === 'object') stack.push(child)\n    }\n  }\n  return found\n}\n\n/** Every string a node's props hold, one level of arrays and objects deep. */\nfunction propStrings(props: Record<string, unknown>, skip: ReadonlySet<string> = new Set()): string[] {\n  const out: string[] = []\n  const visit = (value: unknown, depth: number) => {\n    if (typeof value === 'string') out.push(value)\n    else if (depth < 3 && value && typeof value === 'object') {\n      for (const entry of Array.isArray(value) ? value : Object.values(value as Record<string, unknown>)) {\n        visit(entry, depth + 1)\n      }\n    }\n  }\n  for (const [key, value] of Object.entries(props)) {\n    if (key === 'sx' || skip.has(key)) continue\n    visit(value, 0)\n  }\n  return out\n}\n\n/** Does this node describe a field its author defined? */\nfunction isAuthoredField(node: WalkedNode): boolean {\n  if (/field|input/i.test(node.componentId)) return true\n  return FIELD_PROP_KEYS.some((key) => node.props[key] !== undefined)\n}\n\n/** The credential a field asks for, or null. */\nfunction credentialOf(node: WalkedNode): { field: 'password' | 'card' | 'otp'; label: string } | null {\n  const text = (key: string) => (typeof node.props[key] === 'string' ? (node.props[key] as string) : '')\n  const type = [text('fieldType'), text('type'), text('inputType')].join(' ').toLowerCase()\n  const label = text('label') || text('placeholder') || text('fieldName') || text('name')\n  if (/\\bpassword\\b/.test(type)) return { field: 'password', label: label || 'password' }\n  const autocomplete = (text('autoComplete') || text('autocomplete')).toLowerCase().trim()\n  const byToken = CREDENTIAL_AUTOCOMPLETE[autocomplete]\n  if (byToken) return { field: byToken, label: label || autocomplete }\n  const wording = [text('label'), text('placeholder'), text('fieldName'), text('name'), text('helperText')]\n    .filter(Boolean)\n    .join(' ')\n  for (const { field, pattern } of CREDENTIAL_LABELS) {\n    if (pattern.test(wording)) return { field, label: (label || wording).slice(0, 80) }\n  }\n  return null\n}\n\n/**\n * The credential an input asks for, read from the two attributes that decide\n * it in the browser — its `type` and its `autocomplete` token — or null.\n * For code that builds inputs rather than storing them as page props: the\n * marketplace verifier reads a plugin bundle's `{ type: 'password' }` and\n * `setAttribute('autocomplete', 'cc-number')` with this (AGL-3362).\n */\nexport function credentialForInputAttribute(\n  attribute: string,\n  value: string,\n): 'password' | 'card' | 'otp' | null {\n  const name = String(attribute ?? '').toLowerCase()\n  const token = String(value ?? '').trim().toLowerCase()\n  if ((name === 'type' || name === 'inputtype') && token === 'password') return 'password'\n  if (name === 'autocomplete') return CREDENTIAL_AUTOCOMPLETE[token] ?? null\n  return null\n}\n\n/** The credential a piece of raw HTML asks for in an input of its own, or null. */\nexport function credentialInHtml(html: string): 'password' | 'card' | 'otp' | null {\n  const text = String(html ?? '')\n  if (!text.includes('<')) return null\n  return HTML_CREDENTIAL_INPUTS.find(({ pattern }) => pattern.test(text))?.field ?? null\n}\n\n/**\n * Does a submitted field's NAME ask for a credential — a password, a card\n * number or security code, a one-time code? The same wording the page\n * screen holds a field for, read over a name as a form posts it\n * (`card_number`, `cardNumber`, `one-time-code`).\n *\n * For the one path that does not pass a publish: a public form endpoint\n * takes whatever field names a request carries, so a page the screen never\n * saw (third-party code, a hand-built request) could still post a password\n * to it. The endpoint drops such a field rather than store it.\n */\nexport function isCredentialFieldName(name: string): boolean {\n  const words = String(name ?? '')\n    .replace(/([a-z0-9])([A-Z])/g, '$1 $2')\n    .replace(/[_\\-.[\\]]+/g, ' ')\n    .trim()\n  if (!words) return false\n  if (CREDENTIAL_AUTOCOMPLETE[words.toLowerCase().replace(/\\s+/g, '-')]) return true\n  if (/^(?:pass|pwd|ccnum|cc\\s?num|cc\\s?number|cvv2?|cvc|csc)$/i.test(words)) return true\n  return CREDENTIAL_LABELS.some(({ pattern }) => pattern.test(words))\n}\n\n/** URLs a page carries in a form Links-by-shape would miss: `//host/…`. */\nfunction protocolRelativeUrls(strings: readonly string[]): string {\n  return strings\n    .filter((value) => /^\\/\\/[a-z0-9-]+(?:\\.[a-z0-9-]+)+/i.test(value.trim()))\n    .map((value) => `https:${value.trim()}`)\n    .join('\\n')\n}\n\n/**\n * Screen one page as it will be served. The caller applies the tiers with\n * `signalsThatHold` and decides what a hold does.\n */\nexport function screenHostedPage(input: HostedPageScreenInput): HostedPageScreenVerdict {\n  const signals: PhishingScreenSignal[] = []\n  const nodes = walkNodes(input.nodes)\n  const ownNames = squashScreenText((input.ownNames ?? []).filter(Boolean).join(' '))\n  const ownHosts = (input.ownDomains ?? [])\n    .map((domain) => String(domain ?? '').trim().toLowerCase().replace(/\\.+$/, ''))\n    .filter((domain) => domain.includes('.'))\n  const ownRegistrables = new Set(ownHosts.map(registrableDomain))\n  const isOwnHost = (host: string) => ownHosts.some((own) => host === own || host.endsWith(`.${own}`))\n\n  const nodeStrings = nodes.map((node) => propStrings(node.props))\n  let pageText = nodeStrings.map((strings) => strings.join('\\n')).join('\\n')\n  pageText = `${pageText}\\n${protocolRelativeUrls(nodeStrings.flat())}`.slice(0, MAX_PAGE_TEXT)\n\n  // 1. A lookalike host anywhere the page links, embeds or posts to.\n  const linkHosts = linkHostsIn(pageText)\n  for (const host of linkHosts) {\n    const brand = lookalikeBrandForHost(host)\n    if (brand && !isBrandsOwnDomain(brand, ownRegistrables)) {\n      signals.push({ code: 'lookalike-link', brand: brand.id, host })\n    }\n  }\n\n  // 2. A credential field the page's author defined.\n  const credentialSeen = new Set<string>()\n  const noteCredential = (field: 'password' | 'card' | 'otp', label: string) => {\n    if (credentialSeen.has(field)) return\n    credentialSeen.add(field)\n    signals.push({ code: 'credential-field', field, label: label.slice(0, 80) })\n  }\n  nodes.forEach((node, index) => {\n    if (isAuthoredField(node)) {\n      const credential = credentialOf(node)\n      if (credential) noteCredential(credential.field, credential.label)\n    }\n    for (const html of nodeStrings[index]) {\n      if (!html.includes('<')) continue\n      for (const { field, pattern } of HTML_CREDENTIAL_INPUTS) {\n        if (pattern.test(html)) noteCredential(field, 'an input in the page’s own HTML')\n      }\n    }\n  })\n\n  const isElsewhere = (host: string) => {\n    if (isOwnHost(host)) return false\n    const registrable = registrableDomain(host)\n    return !isAnyOfficialBrandDomain(registrable) && !isCommonLinkDomain(registrable)\n  }\n  const foreign = linkHosts.find(isElsewhere)\n  const actions = [...PAGE_ACTION_PATTERNS, ...PHISHING_LURE_PATTERNS]\n  const named = new Set<string>()\n  const isOthersBrand = (brand: PhishingScreenBrand) =>\n    !named.has(brand.id) && !isWorkspaceOwnBrand(brand, ownNames, ownRegistrables)\n\n  // 3. A brand's call to action, in one element, on a page that links away.\n  if (foreign) {\n    for (const strings of nodeStrings) {\n      const text = proseOf(strings.join(' '))\n      if (!text) continue\n      const action = firstMatch(actions, text)\n      if (!action) continue\n      for (const brand of PHISHING_SCREEN_BRANDS) {\n        if (isOthersBrand(brand) && brand.mention.test(text)) {\n          named.add(brand.id)\n          signals.push({ code: 'brand-action-page', brand: brand.id, action: action.slice(0, 80) })\n        }\n      }\n    }\n  }\n\n  const prose = proseOf(pageText)\n  const pageLure = firstMatch(PAGE_LURE_PATTERNS, prose)\n  const ctas = callsToAction(nodes)\n  const leaving = ctas.filter((cta): cta is CallToAction & { host: string } =>\n    Boolean(cta.host && isElsewhere(cta.host)),\n  )\n  const wordedLeaving = leaving.find((cta) => firstMatch(actions, cta.label))\n\n  // 4. A brand's lure, read over the page as a whole, on a page that links away.\n  const brandLure = pageLure ?? (wordedLeaving ? firstMatch(actions, wordedLeaving.label) : undefined)\n  if (foreign && brandLure) {\n    const host = (wordedLeaving ?? leaving[0])?.host ?? foreign\n    for (const brand of PHISHING_SCREEN_BRANDS) {\n      if (isOthersBrand(brand) && brand.mention.test(prose)) {\n        named.add(brand.id)\n        signals.push({ code: 'brand-lure-page', brand: brand.id, lure: brandLure.slice(0, 120), host })\n      }\n    }\n  }\n\n  // 5. The page's call to action leaves the site, beside a lure. Its only\n  //    one, or the one worded as an account or document action.\n  const primary = wordedLeaving ?? (ctas.length === 1 ? leaving[0] : undefined)\n  if (primary && pageLure) {\n    signals.push({\n      code: 'offsite-action-page',\n      action: (primary.label || primary.host).slice(0, 80),\n      lure: pageLure.slice(0, 120),\n      host: primary.host,\n    })\n  }\n\n  return { signals, linkHosts }\n}\n\n/**\n * What a visitor reads in page text: its markup taken out the way the email\n * screen takes it out ({@link visibleTextOf}), so a custom HTML block or a\n * rich-text prop cannot split a brand or a lure letter by letter with tags\n * (AGL-3453), and its bare URLs taken out too. Links are still read from\n * the raw text, where an href lives.\n */\nfunction proseOf(text: string): string {\n  return visibleTextOf(text).replace(PROSE_URL_PATTERN, ' ').replace(/\\s+/g, ' ').trim()\n}\n\n/** The first pattern's match in the text, or undefined. */\nfunction firstMatch(patterns: readonly RegExp[], text: string): string | undefined {\n  if (!text) return undefined\n  for (const pattern of patterns) {\n    const found = text.match(pattern)?.[0]\n    if (found) return found\n  }\n  return undefined\n}\n\n/** The words a call to action shows, from its label prop, markup taken out. */\nfunction callToActionLabel(props: Record<string, unknown>): string {\n  for (const key of CALL_TO_ACTION_LABEL_KEYS) {\n    const value = props[key]\n    if (typeof value === 'string' && value.trim()) {\n      return visibleTextOf(value).split('\\n')[0]\n    }\n  }\n  return ''\n}\n\n/**\n * Every call to action on the page: each button or link that carries a\n * target. A `screenId` is a page of the site and takes precedence over an\n * `href`, as the button renders it; a relative, `mailto:` or `tel:` href has\n * no web host and stays too.\n */\nfunction callsToAction(nodes: readonly WalkedNode[]): CallToAction[] {\n  const found: CallToAction[] = []\n  for (const node of nodes) {\n    if (!CALL_TO_ACTION_COMPONENT.test(node.componentId)) continue\n    const screenId = typeof node.props['screenId'] === 'string' ? node.props['screenId'].trim() : ''\n    const href = typeof node.props['href'] === 'string' ? node.props['href'].trim() : ''\n    if (!screenId && !href) continue\n    const absolute = href.startsWith('//') ? `https:${href}` : href\n    found.push({\n      label: callToActionLabel(node.props),\n      host: screenId ? null : (linkHostsIn(absolute)[0] ?? null),\n    })\n  }\n  return found\n}\n\nexport interface SiteRedirectScreenInput {\n  /** The path the rule answers — `/secure-document` — or its pattern. */\n  source: string\n  /** Where it sends the visitor, after any `$n` substitution. */\n  destination: string\n  /** The workspace's own names, as `screenHostedPage` reads them. */\n  ownNames?: readonly (string | null | undefined)[]\n  /** The workspace's own hosts — a redirect to one never leaves. */\n  ownDomains?: readonly (string | null | undefined)[]\n}\n\n/**\n * Screen one site redirect (AGL-3447): a rule that sends a path off the site\n * is a call to action that fires on arrival, so it is read with the page's\n * rules over the one text it has, its path.\n *\n * - `lookalike-link` (STRONG) — the destination wears a brand it is not.\n * - `offsite-redirect` (SOFT) — the destination is not the site's own, a\n *   listed brand's or a common link, and the path reads as an account or\n *   document-share lure (`/secure-document-access`).\n * - `brand-lure-link` (SOFT) — the same, and the path names a brand that is\n *   not the workspace's (`/paypal-verify-your-account`), the email screen's\n *   three-part lure.\n *\n * The caller applies the tiers with `signalsThatHold`, with the workspace's\n * age, and decides what a hold does.\n */\nexport function screenSiteRedirect(input: SiteRedirectScreenInput): HostedPageScreenVerdict {\n  const signals: PhishingScreenSignal[] = []\n  const destination = String(input.destination ?? '').trim()\n  const hosts = linkHostsIn(destination.startsWith('//') ? `https:${destination}` : destination)\n  if (!hosts.length) return { signals, linkHosts: hosts }\n  const ownNames = squashScreenText((input.ownNames ?? []).filter(Boolean).join(' '))\n  const ownHosts = (input.ownDomains ?? [])\n    .map((domain) => String(domain ?? '').trim().toLowerCase().replace(/\\.+$/, ''))\n    .filter((domain) => domain.includes('.'))\n  const ownRegistrables = new Set(ownHosts.map(registrableDomain))\n\n  for (const host of hosts) {\n    const brand = lookalikeBrandForHost(host)\n    if (brand && !isBrandsOwnDomain(brand, ownRegistrables)) {\n      signals.push({ code: 'lookalike-link', brand: brand.id, host })\n    }\n  }\n\n  // The first host is where the browser goes; the rest is a userinfo disguise.\n  const host = hosts[0]\n  const registrable = registrableDomain(host)\n  const leaves =\n    !ownHosts.some((own) => host === own || host.endsWith(`.${own}`)) &&\n    !isAnyOfficialBrandDomain(registrable) &&\n    !isCommonLinkDomain(registrable)\n  if (!leaves) return { signals, linkHosts: hosts }\n\n  const source = String(input.source ?? '').trim()\n  const words = pathWords(source)\n  const lure = firstMatch(PAGE_LURE_PATTERNS, words)\n  if (lure) {\n    signals.push({ code: 'offsite-redirect', source: source.slice(0, 120), lure: lure.slice(0, 120), host })\n  }\n  const anyLure = lure ?? firstMatch(PHISHING_LURE_PATTERNS, words)\n  if (anyLure) {\n    const brand = PHISHING_SCREEN_BRANDS.find(\n      (candidate) =>\n        candidate.mention.test(words) && !isWorkspaceOwnBrand(candidate, ownNames, ownRegistrables),\n    )\n    if (brand) {\n      signals.push({ code: 'brand-lure-link', brand: brand.id, lure: anyLure.slice(0, 120), host })\n    }\n  }\n  return { signals, linkHosts: hosts }\n}\n\n/** A path read as words: `/secure-document_access` → `secure document access`. */\nfunction pathWords(path: string): string {\n  let decoded = path\n  try {\n    decoded = decodeURIComponent(path)\n  } catch {\n    // A malformed escape is read as typed.\n  }\n  return decoded.replace(/[^a-z0-9]+/gi, ' ').trim()\n}\n"],"names":["ACCOUNT_LURE_PATTERNS","DOCUMENT_SHARE_LURE_PATTERNS","isWorkspaceOwnBrand","linkHostsIn","isBrandsOwnDomain","lookalikeBrandForHost","PHISHING_LURE_PATTERNS","PHISHING_SCREEN_BRANDS","registrableDomain","isAnyOfficialBrandDomain","isCommonLinkDomain","squashScreenText","visibleTextOf","MAX_PAGE_TEXT","MAX_NODES","FIELD_PROP_KEYS","CREDENTIAL_AUTOCOMPLETE","CREDENTIAL_LABELS","field","pattern","HTML_CREDENTIAL_INPUTS","PAGE_ACTION_PATTERNS","PAGE_LURE_PATTERNS","CALL_TO_ACTION_COMPONENT","CALL_TO_ACTION_LABEL_KEYS","PROSE_URL_PATTERN","walkNodes","root","found","seen","Set","stack","head","length","value","has","add","record","props","push","componentId","child","Array","isArray","Object","values","propStrings","skip","out","visit","depth","entry","key","entries","isAuthoredField","node","test","some","undefined","credentialOf","text","type","join","toLowerCase","label","autocomplete","trim","byToken","wording","filter","Boolean","slice","credentialForInputAttribute","attribute","name","String","token","credentialInHtml","html","includes","find","isCredentialFieldName","words","replace","protocolRelativeUrls","strings","map","screenHostedPage","input","signals","nodes","ownNames","ownHosts","ownDomains","domain","ownRegistrables","isOwnHost","host","own","endsWith","nodeStrings","pageText","flat","linkHosts","brand","code","id","credentialSeen","noteCredential","forEach","index","credential","isElsewhere","registrable","foreign","actions","named","isOthersBrand","proseOf","action","firstMatch","mention","prose","pageLure","ctas","callsToAction","leaving","cta","wordedLeaving","brandLure","lure","primary","patterns","match","callToActionLabel","split","screenId","href","absolute","startsWith","screenSiteRedirect","destination","hosts","leaves","source","pathWords","anyLure","candidate","path","decoded","decodeURIComponent"],"mappings":"AAAA;;;;;;;;;;;;;;;CAeC,GAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;2CAwD2C,GAE3C,SACEA,qBAAqB,EACrBC,4BAA4B,EAC5BC,mBAAmB,EACnBC,WAAW,EACXC,iBAAiB,EACjBC,qBAAqB,EACrBC,sBAAsB,EACtBC,sBAAsB,EAGtBC,iBAAiB,EACjBC,wBAAwB,EACxBC,kBAAkB,EAClBC,gBAAgB,EAChBC,aAAa,QACR,gCAA4B;AA+BnC,4EAA4E,GAC5E,MAAMC,gBAAgB;AACtB,oCAAoC,GACpC,MAAMC,YAAY;AAElB,yEAAyE,GACzE,MAAMC,kBAAkB;IAAC;IAAa;IAAa;IAAa;IAAgB;CAAe;AAE/F,yDAAyD,GACzD,MAAMC,0BAAuE;IAC3E,oBAAoB;IACpB,gBAAgB;IAChB,aAAa;IACb,UAAU;IACV,UAAU;IACV,iBAAiB;AACnB;AAEA,uDAAuD,GACvD,MAAMC,oBAA4F;IAChG;QAAEC,OAAO;QAAYC,SAAS;IAA6E;IAC3G;QACED,OAAO;QACPC,SACE;IACJ;IACA;QACED,OAAO;QACPC,SACE;IACJ;CACD;AAED,6DAA6D,GAC7D,MAAMC,yBAAiG;IACrG;QAAEF,OAAO;QAAYC,SAAS;IAA2C;IACzE;QAAED,OAAO;QAAYC,SAAS;IAAwG;IACtI;QAAED,OAAO;QAAQC,SAAS;IAA+F;IACzH;QAAED,OAAO;QAAOC,SAAS;IAA8F;CACxH;AAED;;;;CAIC,GACD,MAAME,uBAA0C;IAC9C;IACA;IACA;IACA;IACA;IACA;IACA;CACD;AAED;;;;;CAKC,GACD,MAAMC,qBAAwC;OAAItB;OAA0BC;CAA6B;AAEzG;;;;CAIC,GACD,MAAMsB,2BAA2B;AAEjC,8EAA8E,GAC9E,MAAMC,4BAA4B;IAAC;IAAY;IAAS;IAAQ;IAAS;CAAO;AAEhF,4EAA4E,GAC5E,MAAMC,oBAAoB;AAQ1B,sEAAsE,GACtE,SAASC,UAAUC,IAAa;IAC9B,MAAMC,QAAsB,EAAE;IAC9B,MAAMC,OAAO,IAAIC;IACjB,MAAMC,QAAmB;QAACJ;KAAK;IAC/B,IAAK,IAAIK,OAAO,GAAGA,OAAOD,MAAME,MAAM,IAAIL,MAAMK,MAAM,GAAGnB,WAAWkB,QAAQ,EAAG;QAC7E,MAAME,QAAQH,KAAK,CAACC,KAAK;QACzB,IAAI,CAACE,SAAS,OAAOA,UAAU,YAAYL,KAAKM,GAAG,CAACD,QAAQ;QAC5DL,KAAKO,GAAG,CAACF;QACT,MAAMG,SAASH;QACf,IAAI,OAAOG,MAAM,CAAC,cAAc,KAAK,UAAU;YAC7C,MAAMC,QAAQD,MAAM,CAAC,QAAQ;YAC7BT,MAAMW,IAAI,CAAC;gBACTC,aAAaH,MAAM,CAAC,cAAc;gBAClCC,OAAOA,SAAS,OAAOA,UAAU,WAAYA,QAAoC,CAAC;YACpF;QACF;QACA,KAAK,MAAMG,SAASC,MAAMC,OAAO,CAACT,SAASA,QAAQU,OAAOC,MAAM,CAACR,QAAS;YACxE,IAAII,SAAS,OAAOA,UAAU,UAAUV,MAAMQ,IAAI,CAACE;QACrD;IACF;IACA,OAAOb;AACT;AAEA,4EAA4E,GAC5E,SAASkB,YAAYR,KAA8B,EAAES,OAA4B,IAAIjB,KAAK;IACxF,MAAMkB,MAAgB,EAAE;IACxB,MAAMC,QAAQ,CAACf,OAAgBgB;QAC7B,IAAI,OAAOhB,UAAU,UAAUc,IAAIT,IAAI,CAACL;aACnC,IAAIgB,QAAQ,KAAKhB,SAAS,OAAOA,UAAU,UAAU;YACxD,KAAK,MAAMiB,SAAST,MAAMC,OAAO,CAACT,SAASA,QAAQU,OAAOC,MAAM,CAACX,OAAmC;gBAClGe,MAAME,OAAOD,QAAQ;YACvB;QACF;IACF;IACA,KAAK,MAAM,CAACE,KAAKlB,MAAM,IAAIU,OAAOS,OAAO,CAACf,OAAQ;QAChD,IAAIc,QAAQ,QAAQL,KAAKZ,GAAG,CAACiB,MAAM;QACnCH,MAAMf,OAAO;IACf;IACA,OAAOc;AACT;AAEA,wDAAwD,GACxD,SAASM,gBAAgBC,IAAgB;IACvC,IAAI,eAAeC,IAAI,CAACD,KAAKf,WAAW,GAAG,OAAO;IAClD,OAAOzB,gBAAgB0C,IAAI,CAAC,CAACL,MAAQG,KAAKjB,KAAK,CAACc,IAAI,KAAKM;AAC3D;AAEA,8CAA8C,GAC9C,SAASC,aAAaJ,IAAgB;IACpC,MAAMK,OAAO,CAACR,MAAiB,OAAOG,KAAKjB,KAAK,CAACc,IAAI,KAAK,WAAYG,KAAKjB,KAAK,CAACc,IAAI,GAAc;IACnG,MAAMS,OAAO;QAACD,KAAK;QAAcA,KAAK;QAASA,KAAK;KAAa,CAACE,IAAI,CAAC,KAAKC,WAAW;IACvF,MAAMC,QAAQJ,KAAK,YAAYA,KAAK,kBAAkBA,KAAK,gBAAgBA,KAAK;IAChF,IAAI,eAAeJ,IAAI,CAACK,OAAO,OAAO;QAAE3C,OAAO;QAAY8C,OAAOA,SAAS;IAAW;IACtF,MAAMC,eAAe,AAACL,CAAAA,KAAK,mBAAmBA,KAAK,eAAc,EAAGG,WAAW,GAAGG,IAAI;IACtF,MAAMC,UAAUnD,uBAAuB,CAACiD,aAAa;IACrD,IAAIE,SAAS,OAAO;QAAEjD,OAAOiD;QAASH,OAAOA,SAASC;IAAa;IACnE,MAAMG,UAAU;QAACR,KAAK;QAAUA,KAAK;QAAgBA,KAAK;QAAcA,KAAK;QAASA,KAAK;KAAc,CACtGS,MAAM,CAACC,SACPR,IAAI,CAAC;IACR,KAAK,MAAM,EAAE5C,KAAK,EAAEC,OAAO,EAAE,IAAIF,kBAAmB;QAClD,IAAIE,QAAQqC,IAAI,CAACY,UAAU,OAAO;YAAElD;YAAO8C,OAAO,AAACA,CAAAA,SAASI,OAAM,EAAGG,KAAK,CAAC,GAAG;QAAI;IACpF;IACA,OAAO;AACT;AAEA;;;;;;CAMC,GACD,OAAO,SAASC,4BACdC,SAAiB,EACjBvC,KAAa;QAKuBlB;IAHpC,MAAM0D,OAAOC,OAAOF,oBAAAA,YAAa,IAAIV,WAAW;IAChD,MAAMa,QAAQD,OAAOzC,gBAAAA,QAAS,IAAIgC,IAAI,GAAGH,WAAW;IACpD,IAAI,AAACW,CAAAA,SAAS,UAAUA,SAAS,WAAU,KAAME,UAAU,YAAY,OAAO;IAC9E,IAAIF,SAAS,gBAAgB,QAAO1D,iCAAAA,uBAAuB,CAAC4D,MAAM,YAA9B5D,iCAAkC;IACtE,OAAO;AACT;AAEA,iFAAiF,GACjF,OAAO,SAAS6D,iBAAiBC,IAAY;;QAGpC1D;IAFP,MAAMwC,OAAOe,OAAOG,eAAAA,OAAQ;IAC5B,IAAI,CAAClB,KAAKmB,QAAQ,CAAC,MAAM,OAAO;IAChC,gBAAO3D,+BAAAA,uBAAuB4D,IAAI,CAAC,CAAC,EAAE7D,OAAO,EAAE,GAAKA,QAAQqC,IAAI,CAACI,2BAA1DxC,6BAAkEF,KAAK,mBAAI;AACpF;AAEA;;;;;;;;;;CAUC,GACD,OAAO,SAAS+D,sBAAsBP,IAAY;IAChD,MAAMQ,QAAQP,OAAOD,eAAAA,OAAQ,IAC1BS,OAAO,CAAC,sBAAsB,SAC9BA,OAAO,CAAC,eAAe,KACvBjB,IAAI;IACP,IAAI,CAACgB,OAAO,OAAO;IACnB,IAAIlE,uBAAuB,CAACkE,MAAMnB,WAAW,GAAGoB,OAAO,CAAC,QAAQ,KAAK,EAAE,OAAO;IAC9E,IAAI,2DAA2D3B,IAAI,CAAC0B,QAAQ,OAAO;IACnF,OAAOjE,kBAAkBwC,IAAI,CAAC,CAAC,EAAEtC,OAAO,EAAE,GAAKA,QAAQqC,IAAI,CAAC0B;AAC9D;AAEA,yEAAyE,GACzE,SAASE,qBAAqBC,OAA0B;IACtD,OAAOA,QACJhB,MAAM,CAAC,CAACnC,QAAU,oCAAoCsB,IAAI,CAACtB,MAAMgC,IAAI,KACrEoB,GAAG,CAAC,CAACpD,QAAU,CAAC,MAAM,EAAEA,MAAMgC,IAAI,IAAI,EACtCJ,IAAI,CAAC;AACV;AAEA;;;CAGC,GACD,OAAO,SAASyB,iBAAiBC,KAA4B;QAGxBA,iBACjBA;IAHlB,MAAMC,UAAkC,EAAE;IAC1C,MAAMC,QAAQhE,UAAU8D,MAAME,KAAK;IACnC,MAAMC,WAAWhF,iBAAiB,EAAC6E,kBAAAA,MAAMG,QAAQ,YAAdH,kBAAkB,EAAE,EAAEnB,MAAM,CAACC,SAASR,IAAI,CAAC;IAC9E,MAAM8B,WAAW,EAACJ,oBAAAA,MAAMK,UAAU,YAAhBL,oBAAoB,EAAE,EACrCF,GAAG,CAAC,CAACQ,SAAWnB,OAAOmB,iBAAAA,SAAU,IAAI5B,IAAI,GAAGH,WAAW,GAAGoB,OAAO,CAAC,QAAQ,KAC1Ed,MAAM,CAAC,CAACyB,SAAWA,OAAOf,QAAQ,CAAC;IACtC,MAAMgB,kBAAkB,IAAIjE,IAAI8D,SAASN,GAAG,CAAC9E;IAC7C,MAAMwF,YAAY,CAACC,OAAiBL,SAASnC,IAAI,CAAC,CAACyC,MAAQD,SAASC,OAAOD,KAAKE,QAAQ,CAAC,CAAC,CAAC,EAAED,KAAK;IAElG,MAAME,cAAcV,MAAMJ,GAAG,CAAC,CAAC/B,OAAST,YAAYS,KAAKjB,KAAK;IAC9D,IAAI+D,WAAWD,YAAYd,GAAG,CAAC,CAACD,UAAYA,QAAQvB,IAAI,CAAC,OAAOA,IAAI,CAAC;IACrEuC,WAAW,GAAGA,SAAS,EAAE,EAAEjB,qBAAqBgB,YAAYE,IAAI,KAAK,CAAC/B,KAAK,CAAC,GAAG1D;IAE/E,mEAAmE;IACnE,MAAM0F,YAAYpG,YAAYkG;IAC9B,KAAK,MAAMJ,QAAQM,UAAW;QAC5B,MAAMC,QAAQnG,sBAAsB4F;QACpC,IAAIO,SAAS,CAACpG,kBAAkBoG,OAAOT,kBAAkB;YACvDN,QAAQlD,IAAI,CAAC;gBAAEkE,MAAM;gBAAkBD,OAAOA,MAAME,EAAE;gBAAET;YAAK;QAC/D;IACF;IAEA,mDAAmD;IACnD,MAAMU,iBAAiB,IAAI7E;IAC3B,MAAM8E,iBAAiB,CAAC1F,OAAoC8C;QAC1D,IAAI2C,eAAexE,GAAG,CAACjB,QAAQ;QAC/ByF,eAAevE,GAAG,CAAClB;QACnBuE,QAAQlD,IAAI,CAAC;YAAEkE,MAAM;YAAoBvF;YAAO8C,OAAOA,MAAMO,KAAK,CAAC,GAAG;QAAI;IAC5E;IACAmB,MAAMmB,OAAO,CAAC,CAACtD,MAAMuD;QACnB,IAAIxD,gBAAgBC,OAAO;YACzB,MAAMwD,aAAapD,aAAaJ;YAChC,IAAIwD,YAAYH,eAAeG,WAAW7F,KAAK,EAAE6F,WAAW/C,KAAK;QACnE;QACA,KAAK,MAAMc,QAAQsB,WAAW,CAACU,MAAM,CAAE;YACrC,IAAI,CAAChC,KAAKC,QAAQ,CAAC,MAAM;YACzB,KAAK,MAAM,EAAE7D,KAAK,EAAEC,OAAO,EAAE,IAAIC,uBAAwB;gBACvD,IAAID,QAAQqC,IAAI,CAACsB,OAAO8B,eAAe1F,OAAO;YAChD;QACF;IACF;IAEA,MAAM8F,cAAc,CAACf;QACnB,IAAID,UAAUC,OAAO,OAAO;QAC5B,MAAMgB,cAAczG,kBAAkByF;QACtC,OAAO,CAACxF,yBAAyBwG,gBAAgB,CAACvG,mBAAmBuG;IACvE;IACA,MAAMC,UAAUX,UAAUvB,IAAI,CAACgC;IAC/B,MAAMG,UAAU;WAAI9F;WAAyBf;KAAuB;IACpE,MAAM8G,QAAQ,IAAItF;IAClB,MAAMuF,gBAAgB,CAACb,QACrB,CAACY,MAAMjF,GAAG,CAACqE,MAAME,EAAE,KAAK,CAACxG,oBAAoBsG,OAAOb,UAAUI;IAEhE,0EAA0E;IAC1E,IAAImB,SAAS;QACX,KAAK,MAAM7B,WAAWe,YAAa;YACjC,MAAMxC,OAAO0D,QAAQjC,QAAQvB,IAAI,CAAC;YAClC,IAAI,CAACF,MAAM;YACX,MAAM2D,SAASC,WAAWL,SAASvD;YACnC,IAAI,CAAC2D,QAAQ;YACb,KAAK,MAAMf,SAASjG,uBAAwB;gBAC1C,IAAI8G,cAAcb,UAAUA,MAAMiB,OAAO,CAACjE,IAAI,CAACI,OAAO;oBACpDwD,MAAMhF,GAAG,CAACoE,MAAME,EAAE;oBAClBjB,QAAQlD,IAAI,CAAC;wBAAEkE,MAAM;wBAAqBD,OAAOA,MAAME,EAAE;wBAAEa,QAAQA,OAAOhD,KAAK,CAAC,GAAG;oBAAI;gBACzF;YACF;QACF;IACF;IAEA,MAAMmD,QAAQJ,QAAQjB;IACtB,MAAMsB,WAAWH,WAAWlG,oBAAoBoG;IAChD,MAAME,OAAOC,cAAcnC;IAC3B,MAAMoC,UAAUF,KAAKvD,MAAM,CAAC,CAAC0D,MAC3BzD,QAAQyD,IAAI9B,IAAI,IAAIe,YAAYe,IAAI9B,IAAI;IAE1C,MAAM+B,gBAAgBF,QAAQ9C,IAAI,CAAC,CAAC+C,MAAQP,WAAWL,SAASY,IAAI/D,KAAK;IAEzE,+EAA+E;IAC/E,MAAMiE,YAAYN,mBAAAA,WAAaK,gBAAgBR,WAAWL,SAASa,cAAchE,KAAK,IAAIN;IAC1F,IAAIwD,WAAWe,WAAW;;YACVD;QAAd,MAAM/B,gBAAQ+B,QAAAA,wBAAAA,gBAAiBF,OAAO,CAAC,EAAE,qBAA5B,AAACE,MAA8B/B,IAAI,mBAAIiB;QACpD,KAAK,MAAMV,SAASjG,uBAAwB;YAC1C,IAAI8G,cAAcb,UAAUA,MAAMiB,OAAO,CAACjE,IAAI,CAACkE,QAAQ;gBACrDN,MAAMhF,GAAG,CAACoE,MAAME,EAAE;gBAClBjB,QAAQlD,IAAI,CAAC;oBAAEkE,MAAM;oBAAmBD,OAAOA,MAAME,EAAE;oBAAEwB,MAAMD,UAAU1D,KAAK,CAAC,GAAG;oBAAM0B;gBAAK;YAC/F;QACF;IACF;IAEA,wEAAwE;IACxE,8DAA8D;IAC9D,MAAMkC,UAAUH,wBAAAA,gBAAkBJ,KAAK3F,MAAM,KAAK,IAAI6F,OAAO,CAAC,EAAE,GAAGpE;IACnE,IAAIyE,WAAWR,UAAU;QACvBlC,QAAQlD,IAAI,CAAC;YACXkE,MAAM;YACNc,QAAQ,AAACY,CAAAA,QAAQnE,KAAK,IAAImE,QAAQlC,IAAI,AAAD,EAAG1B,KAAK,CAAC,GAAG;YACjD2D,MAAMP,SAASpD,KAAK,CAAC,GAAG;YACxB0B,MAAMkC,QAAQlC,IAAI;QACpB;IACF;IAEA,OAAO;QAAER;QAASc;IAAU;AAC9B;AAEA;;;;;;CAMC,GACD,SAASe,QAAQ1D,IAAY;IAC3B,OAAOhD,cAAcgD,MAAMuB,OAAO,CAAC1D,mBAAmB,KAAK0D,OAAO,CAAC,QAAQ,KAAKjB,IAAI;AACtF;AAEA,yDAAyD,GACzD,SAASsD,WAAWY,QAA2B,EAAExE,IAAY;IAC3D,IAAI,CAACA,MAAM,OAAOF;IAClB,KAAK,MAAMvC,WAAWiH,SAAU;YAChBxE;QAAd,MAAMhC,SAAQgC,cAAAA,KAAKyE,KAAK,CAAClH,6BAAXyC,WAAqB,CAAC,EAAE;QACtC,IAAIhC,OAAO,OAAOA;IACpB;IACA,OAAO8B;AACT;AAEA,6EAA6E,GAC7E,SAAS4E,kBAAkBhG,KAA8B;IACvD,KAAK,MAAMc,OAAO5B,0BAA2B;QAC3C,MAAMU,QAAQI,KAAK,CAACc,IAAI;QACxB,IAAI,OAAOlB,UAAU,YAAYA,MAAMgC,IAAI,IAAI;YAC7C,OAAOtD,cAAcsB,OAAOqG,KAAK,CAAC,KAAK,CAAC,EAAE;QAC5C;IACF;IACA,OAAO;AACT;AAEA;;;;;CAKC,GACD,SAASV,cAAcnC,KAA4B;IACjD,MAAM9D,QAAwB,EAAE;IAChC,KAAK,MAAM2B,QAAQmC,MAAO;YAQGvF;QAP3B,IAAI,CAACoB,yBAAyBiC,IAAI,CAACD,KAAKf,WAAW,GAAG;QACtD,MAAMgG,WAAW,OAAOjF,KAAKjB,KAAK,CAAC,WAAW,KAAK,WAAWiB,KAAKjB,KAAK,CAAC,WAAW,CAAC4B,IAAI,KAAK;QAC9F,MAAMuE,OAAO,OAAOlF,KAAKjB,KAAK,CAAC,OAAO,KAAK,WAAWiB,KAAKjB,KAAK,CAAC,OAAO,CAAC4B,IAAI,KAAK;QAClF,IAAI,CAACsE,YAAY,CAACC,MAAM;QACxB,MAAMC,WAAWD,KAAKE,UAAU,CAAC,QAAQ,CAAC,MAAM,EAAEF,MAAM,GAAGA;QAC3D7G,MAAMW,IAAI,CAAC;YACTyB,OAAOsE,kBAAkB/E,KAAKjB,KAAK;YACnC2D,MAAMuC,WAAW,QAAQrI,gBAAAA,YAAYuI,SAAS,CAAC,EAAE,YAAxBvI,gBAA4B;QACvD;IACF;IACA,OAAOyB;AACT;AAaA;;;;;;;;;;;;;;;CAeC,GACD,OAAO,SAASgH,mBAAmBpD,KAA8B;QAEpCA,oBAGQA,iBACjBA,mBAqBIA;IA1BtB,MAAMC,UAAkC,EAAE;IAC1C,MAAMoD,cAAclE,QAAOa,qBAAAA,MAAMqD,WAAW,YAAjBrD,qBAAqB,IAAItB,IAAI;IACxD,MAAM4E,QAAQ3I,YAAY0I,YAAYF,UAAU,CAAC,QAAQ,CAAC,MAAM,EAAEE,aAAa,GAAGA;IAClF,IAAI,CAACC,MAAM7G,MAAM,EAAE,OAAO;QAAEwD;QAASc,WAAWuC;IAAM;IACtD,MAAMnD,WAAWhF,iBAAiB,EAAC6E,kBAAAA,MAAMG,QAAQ,YAAdH,kBAAkB,EAAE,EAAEnB,MAAM,CAACC,SAASR,IAAI,CAAC;IAC9E,MAAM8B,WAAW,EAACJ,oBAAAA,MAAMK,UAAU,YAAhBL,oBAAoB,EAAE,EACrCF,GAAG,CAAC,CAACQ,SAAWnB,OAAOmB,iBAAAA,SAAU,IAAI5B,IAAI,GAAGH,WAAW,GAAGoB,OAAO,CAAC,QAAQ,KAC1Ed,MAAM,CAAC,CAACyB,SAAWA,OAAOf,QAAQ,CAAC;IACtC,MAAMgB,kBAAkB,IAAIjE,IAAI8D,SAASN,GAAG,CAAC9E;IAE7C,KAAK,MAAMyF,QAAQ6C,MAAO;QACxB,MAAMtC,QAAQnG,sBAAsB4F;QACpC,IAAIO,SAAS,CAACpG,kBAAkBoG,OAAOT,kBAAkB;YACvDN,QAAQlD,IAAI,CAAC;gBAAEkE,MAAM;gBAAkBD,OAAOA,MAAME,EAAE;gBAAET;YAAK;QAC/D;IACF;IAEA,6EAA6E;IAC7E,MAAMA,OAAO6C,KAAK,CAAC,EAAE;IACrB,MAAM7B,cAAczG,kBAAkByF;IACtC,MAAM8C,SACJ,CAACnD,SAASnC,IAAI,CAAC,CAACyC,MAAQD,SAASC,OAAOD,KAAKE,QAAQ,CAAC,CAAC,CAAC,EAAED,KAAK,MAC/D,CAACzF,yBAAyBwG,gBAC1B,CAACvG,mBAAmBuG;IACtB,IAAI,CAAC8B,QAAQ,OAAO;QAAEtD;QAASc,WAAWuC;IAAM;IAEhD,MAAME,SAASrE,QAAOa,gBAAAA,MAAMwD,MAAM,YAAZxD,gBAAgB,IAAItB,IAAI;IAC9C,MAAMgB,QAAQ+D,UAAUD;IACxB,MAAMd,OAAOV,WAAWlG,oBAAoB4D;IAC5C,IAAIgD,MAAM;QACRzC,QAAQlD,IAAI,CAAC;YAAEkE,MAAM;YAAoBuC,QAAQA,OAAOzE,KAAK,CAAC,GAAG;YAAM2D,MAAMA,KAAK3D,KAAK,CAAC,GAAG;YAAM0B;QAAK;IACxG;IACA,MAAMiD,UAAUhB,eAAAA,OAAQV,WAAWlH,wBAAwB4E;IAC3D,IAAIgE,SAAS;QACX,MAAM1C,QAAQjG,uBAAuByE,IAAI,CACvC,CAACmE,YACCA,UAAU1B,OAAO,CAACjE,IAAI,CAAC0B,UAAU,CAAChF,oBAAoBiJ,WAAWxD,UAAUI;QAE/E,IAAIS,OAAO;YACTf,QAAQlD,IAAI,CAAC;gBAAEkE,MAAM;gBAAmBD,OAAOA,MAAME,EAAE;gBAAEwB,MAAMgB,QAAQ3E,KAAK,CAAC,GAAG;gBAAM0B;YAAK;QAC7F;IACF;IACA,OAAO;QAAER;QAASc,WAAWuC;IAAM;AACrC;AAEA,gFAAgF,GAChF,SAASG,UAAUG,IAAY;IAC7B,IAAIC,UAAUD;IACd,IAAI;QACFC,UAAUC,mBAAmBF;IAC/B,EAAE,eAAM;IACN,uCAAuC;IACzC;IACA,OAAOC,QAAQlE,OAAO,CAAC,gBAAgB,KAAKjB,IAAI;AAClD"}
+{"version":3,"sources":["../../../../../../../libs/shared/util/email/src/lib/hosted-page-screen.ts"],"sourcesContent":["/**\n * @license\n * Copyright 2026 Aglyn LLC\n *\n * Licensed under the Apache License, Version 2.0 (the \"License\");\n * you may not use this file except in compliance with the License.\n * You may obtain a copy of the License at\n *\n *   http://www.apache.org/licenses/LICENSE-2.0\n *\n * Unless required by applicable law or agreed to in writing, software\n * distributed under the License is distributed on an \"AS IS\" BASIS,\n * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n * See the License for the specific language governing permissions and\n * limitations under the License.\n */\n\n/*==========================================\n * THE HOSTED PAGE SCREEN (AGL-3362).\n *\n * The outbound phishing screen, read over a PUBLISHED PAGE instead of an\n * email: the same brand list, the same lookalike logic, the same tiers\n * (`signalsThatHold`). A fraud actor who could not mail a lure could still\n * host one, and the incident's sibling site published a page at the\n * document-share path `/reviewfile`.\n *\n * Five shapes:\n *\n * 1. **A lookalike link or embed** (`lookalike-link`, STRONG) — any `href`,\n *    `src`, form action or URL in the page's authored content whose host\n *    wears a brand without being the brand's domain.\n * 2. **A credential field** (`credential-field`, STRONG) — a field the\n *    merchant AUTHORED that asks for a password, a card number or a\n *    one-time code: its type, its autocomplete token, or its label,\n *    placeholder or name. The platform's own sign-in, account and checkout\n *    elements render their inputs in code, not in stored props, so they are\n *    never read as a field here — which is exactly the sanctioned/unsanctioned\n *    line: what the platform collects through its own components is\n *    Aglyn's to protect, and what a page's author collects in a form of\n *    their own is not something a site may ask a stranger for.\n * 3. **A brand's call to action** (`brand-action-page`, SOFT) — one element\n *    whose text names a brand that is not the workspace's AND asks the reader\n *    to sign in, verify, confirm or open a document, on a page that links\n *    somewhere that is neither the workspace's, nor a brand's own, nor a\n *    common social or maps link.\n * 4. **A brand's lure, page-wide** (`brand-lure-page`, SOFT, AGL-3447) — the\n *    same brand and link anywhere on the page, beside an account or\n *    document-share LURE anywhere on it (\"Secure Document Access Portal\"), or\n *    a call to action that leaves the site worded as an action (\"Continue to\n *    Document\"). The 2026-10-01 phishing page put Proofpoint's name in its\n *    body, the lure in its heading and the action on its button: three\n *    elements, none of which shape 3 reads on its own. What it does not count page-wide is\n *    a bare \"sign in\" that stays on the site — a hotel that says \"find us on\n *    Booking.com\" in one place and has a member sign-in in another is two\n *    ordinary sentences.\n * 5. **A call to action that leaves, beside a lure** (`offsite-action-page`,\n *    SOFT, AGL-3447) — the page's only call to action, or one worded as an\n *    account or document action, sends visitors to a host that is not the\n *    site's own, a listed brand's or a common link, on a page that carries an\n *    account or document-share lure. No brand needed: a \"Secure Document\n *    Access Portal\" whose one button leaves for the kit is the shape whatever\n *    name it wears.\n *\n * Shapes 3-5 are soft, so they hold only for a workspace in its first\n * fortnight (`signalsThatHold`); an established workspace's page is held by\n * shapes 1 and 2 alone.\n *\n * The same reading applies to a site REDIRECT ({@link screenSiteRedirect}):\n * a rule that sends a path off the site is a page whose one call to action\n * fires on arrival, and its path is the only text it has.\n *\n * Pure: the caller hands over the composed node tree and the workspace's own\n * names and hosts, and decides what a hold does.\n *=========================================*/\n\nimport {\n  ACCOUNT_LURE_PATTERNS,\n  DOCUMENT_SHARE_LURE_PATTERNS,\n  isWorkspaceOwnBrand,\n  linkHostsIn,\n  linkUrlsIn,\n  isBrandsOwnDomain,\n  lookalikeBrandForHost,\n  PHISHING_LURE_PATTERNS,\n  PHISHING_SCREEN_BRANDS,\n  type PhishingScreenBrand,\n  type PhishingScreenSignal,\n  registrableDomain,\n  isAnyOfficialBrandDomain,\n  isCommonLinkDomain,\n  squashScreenText,\n  visibleTextOf,\n} from './outbound-phishing-screen'\n\nexport interface HostedPageScreenInput {\n  /**\n   * The page as it will be served — the composed tree, or any object that\n   * holds nodes (`{ componentId, props }`) at any depth. Walked, never\n   * trusted to have one shape: a screen, a layout's chrome, a grafted\n   * component and a form all arrive together.\n   */\n  nodes: unknown\n  /** The workspace's own names, as `screenOutboundEmail` reads them. */\n  ownNames?: readonly (string | null | undefined)[]\n  /** The workspace's own hosts — a link to one is never \"elsewhere\". */\n  ownDomains?: readonly (string | null | undefined)[]\n}\n\nexport interface HostedPageScreenVerdict {\n  signals: PhishingScreenSignal[]\n  /**\n   * Every host the page (or the redirect) links, embeds or posts to, lowercased — what the\n   * link reputation lookup reads (AGL-3451), so the page is walked once.\n   */\n  linkHosts: string[]\n  /**\n   * The same links whole, as written (`linkUrlsIn`), for the lookup's\n   * `'url'` mode (AGL-3459). Query strings are still on them:\n   * `foreignLinksForReputation` drops them before anything is looked up or\n   * stored.\n   */\n  linkUrls: string[]\n}\n\n/** A node as the walk finds it. */\ninterface WalkedNode {\n  componentId: string\n  props: Record<string, unknown>\n}\n\n/** The most a page's text is read to, so a huge page costs a bounded scan. */\nconst MAX_PAGE_TEXT = 2_000_000\n/** The most nodes the walk visits. */\nconst MAX_NODES = 20_000\n\n/** Props that describe a field the author defined, rather than content. */\nconst FIELD_PROP_KEYS = ['fieldType', 'fieldName', 'inputType', 'autoComplete', 'autocomplete']\n\n/** Autocomplete tokens (WHATWG) that name a credential. */\nconst CREDENTIAL_AUTOCOMPLETE: Record<string, 'password' | 'card' | 'otp'> = {\n  'current-password': 'password',\n  'new-password': 'password',\n  'cc-number': 'card',\n  'cc-csc': 'card',\n  'cc-exp': 'card',\n  'one-time-code': 'otp',\n}\n\n/** Field wording that asks for a credential, by kind. */\nconst CREDENTIAL_LABELS: ReadonlyArray<{ field: 'password' | 'card' | 'otp'; pattern: RegExp }> = [\n  { field: 'password', pattern: /\\bpass\\s?(?:word|code|phrase)s?\\b|\\bpasswd\\b|\\bpin\\s*(?:code|number)?\\s*$/i },\n  {\n    field: 'card',\n    pattern:\n      /\\b(?:credit|debit)\\s*card\\b|\\bcard\\s*(?:number|no\\.?|#)|\\bcardnumber\\b|\\bcvv2?\\b|\\bcvc\\b|\\bcard\\s*security\\s*code\\b|\\bcard\\s*expir/i,\n  },\n  {\n    field: 'otp',\n    pattern:\n      /\\bone[-\\s]?time\\s*(?:pass\\s?code|password|code|pin)\\b|\\botp\\b|\\bverification\\s*code\\b|\\b2fa\\b|\\btwo[-\\s]?factor\\b|\\bauthenticat(?:ion|or)\\s*code\\b|\\bsms\\s*code\\b/i,\n  },\n]\n\n/** Raw HTML an author placed that holds a credential input. */\nconst HTML_CREDENTIAL_INPUTS: ReadonlyArray<{ field: 'password' | 'card' | 'otp'; pattern: RegExp }> = [\n  { field: 'password', pattern: /<input\\b[^>]*\\btype\\s*=\\s*[\"']?password/i },\n  { field: 'password', pattern: /<input\\b[^>]*\\b(?:name|autocomplete)\\s*=\\s*[\"']?(?:pass(?:word|wd)?|current-password|new-password)\\b/i },\n  { field: 'card', pattern: /<input\\b[^>]*\\b(?:name|autocomplete)\\s*=\\s*[\"']?(?:card_?number|cc-number|cc-csc|cvv|cvc)\\b/i },\n  { field: 'otp', pattern: /<input\\b[^>]*\\b(?:name|autocomplete)\\s*=\\s*[\"']?(?:otp|one-time-code|verification_?code)\\b/i },\n]\n\n/**\n * What an element asks its reader to DO with an account or a document —\n * counted beside a brand in the SAME element, or on a call to action that\n * leaves the site. The email lure phrasing counts too.\n */\nconst PAGE_ACTION_PATTERNS: readonly RegExp[] = [\n  /\\b(?:sign|log)[\\s-]?in\\b/i,\n  /\\bverify\\b/i,\n  /\\bconfirm\\s+(?:your|the)\\b/i,\n  /\\bunlock\\b/i,\n  /\\benter\\s+your\\b/i,\n  /\\b(?:view|review|open|download|access)\\s+(?:the\\s+|your\\s+)?(?:shared\\s+|secure\\s+)?(?:file|document|invoice|statement)s?\\b/i,\n  /\\b(?:continue|proceed)\\s+to\\s+(?:the\\s+|your\\s+)?(?:document|file|account|sign[\\s-]?in|log[\\s-]?in)s?\\b/i,\n]\n\n/**\n * The lures a page is read for as a WHOLE (AGL-3447): account and\n * document-share wording. Not the marketplace, booking and parcel lures,\n * which a hotel's \"guest reviews\" and a shop's delivery page carry every day\n * and which still count in one element beside a brand.\n */\nconst PAGE_LURE_PATTERNS: readonly RegExp[] = [...ACCOUNT_LURE_PATTERNS, ...DOCUMENT_SHARE_LURE_PATTERNS]\n\n/**\n * A component that renders a call to action — a button or a link, however a\n * plugin names it (`muiButton`, `muiScreenLink`, `muiLinkBox`). It is one\n * when it carries a target: an `href`, or a `screenId` on the site itself.\n */\nconst CALL_TO_ACTION_COMPONENT = /button|fab|cta|link/i\n\n/** The props a call to action's visible words live in; the first found wins. */\nconst CALL_TO_ACTION_LABEL_KEYS = ['children', 'label', 'text', 'title', 'html']\n\n/** URLs in page text, so prose is read for a brand or a lure without them. */\nconst PROSE_URL_PATTERN = /(?:https?:)?\\/\\/[^\\s\"'<>]+|\\bwww\\.[^\\s\"'<>]+/gi\n\n/** One call to action: what it says, and the host it leaves for (null: it stays). */\ninterface CallToAction {\n  label: string\n  host: string | null\n}\n\n/** Every node in the tree, breadth-first in document order, bounded. */\nfunction walkNodes(root: unknown): WalkedNode[] {\n  const found: WalkedNode[] = []\n  const seen = new Set<unknown>()\n  const stack: unknown[] = [root]\n  for (let head = 0; head < stack.length && found.length < MAX_NODES; head += 1) {\n    const value = stack[head]\n    if (!value || typeof value !== 'object' || seen.has(value)) continue\n    seen.add(value)\n    const record = value as Record<string, unknown>\n    if (typeof record['componentId'] === 'string') {\n      const props = record['props']\n      found.push({\n        componentId: record['componentId'] as string,\n        props: props && typeof props === 'object' ? (props as Record<string, unknown>) : {},\n      })\n    }\n    for (const child of Array.isArray(value) ? value : Object.values(record)) {\n      if (child && typeof child === 'object') stack.push(child)\n    }\n  }\n  return found\n}\n\n/** Every string a node's props hold, one level of arrays and objects deep. */\nfunction propStrings(props: Record<string, unknown>, skip: ReadonlySet<string> = new Set()): string[] {\n  const out: string[] = []\n  const visit = (value: unknown, depth: number) => {\n    if (typeof value === 'string') out.push(value)\n    else if (depth < 3 && value && typeof value === 'object') {\n      for (const entry of Array.isArray(value) ? value : Object.values(value as Record<string, unknown>)) {\n        visit(entry, depth + 1)\n      }\n    }\n  }\n  for (const [key, value] of Object.entries(props)) {\n    if (key === 'sx' || skip.has(key)) continue\n    visit(value, 0)\n  }\n  return out\n}\n\n/** Does this node describe a field its author defined? */\nfunction isAuthoredField(node: WalkedNode): boolean {\n  if (/field|input/i.test(node.componentId)) return true\n  return FIELD_PROP_KEYS.some((key) => node.props[key] !== undefined)\n}\n\n/** The credential a field asks for, or null. */\nfunction credentialOf(node: WalkedNode): { field: 'password' | 'card' | 'otp'; label: string } | null {\n  const text = (key: string) => (typeof node.props[key] === 'string' ? (node.props[key] as string) : '')\n  const type = [text('fieldType'), text('type'), text('inputType')].join(' ').toLowerCase()\n  const label = text('label') || text('placeholder') || text('fieldName') || text('name')\n  if (/\\bpassword\\b/.test(type)) return { field: 'password', label: label || 'password' }\n  const autocomplete = (text('autoComplete') || text('autocomplete')).toLowerCase().trim()\n  const byToken = CREDENTIAL_AUTOCOMPLETE[autocomplete]\n  if (byToken) return { field: byToken, label: label || autocomplete }\n  const wording = [text('label'), text('placeholder'), text('fieldName'), text('name'), text('helperText')]\n    .filter(Boolean)\n    .join(' ')\n  for (const { field, pattern } of CREDENTIAL_LABELS) {\n    if (pattern.test(wording)) return { field, label: (label || wording).slice(0, 80) }\n  }\n  return null\n}\n\n/**\n * The credential an input asks for, read from the two attributes that decide\n * it in the browser — its `type` and its `autocomplete` token — or null.\n * For code that builds inputs rather than storing them as page props: the\n * marketplace verifier reads a plugin bundle's `{ type: 'password' }` and\n * `setAttribute('autocomplete', 'cc-number')` with this (AGL-3362).\n */\nexport function credentialForInputAttribute(\n  attribute: string,\n  value: string,\n): 'password' | 'card' | 'otp' | null {\n  const name = String(attribute ?? '').toLowerCase()\n  const token = String(value ?? '').trim().toLowerCase()\n  if ((name === 'type' || name === 'inputtype') && token === 'password') return 'password'\n  if (name === 'autocomplete') return CREDENTIAL_AUTOCOMPLETE[token] ?? null\n  return null\n}\n\n/** The credential a piece of raw HTML asks for in an input of its own, or null. */\nexport function credentialInHtml(html: string): 'password' | 'card' | 'otp' | null {\n  const text = String(html ?? '')\n  if (!text.includes('<')) return null\n  return HTML_CREDENTIAL_INPUTS.find(({ pattern }) => pattern.test(text))?.field ?? null\n}\n\n/**\n * Does a submitted field's NAME ask for a credential — a password, a card\n * number or security code, a one-time code? The same wording the page\n * screen holds a field for, read over a name as a form posts it\n * (`card_number`, `cardNumber`, `one-time-code`).\n *\n * For the one path that does not pass a publish: a public form endpoint\n * takes whatever field names a request carries, so a page the screen never\n * saw (third-party code, a hand-built request) could still post a password\n * to it. The endpoint drops such a field rather than store it.\n */\nexport function isCredentialFieldName(name: string): boolean {\n  const words = String(name ?? '')\n    .replace(/([a-z0-9])([A-Z])/g, '$1 $2')\n    .replace(/[_\\-.[\\]]+/g, ' ')\n    .trim()\n  if (!words) return false\n  if (CREDENTIAL_AUTOCOMPLETE[words.toLowerCase().replace(/\\s+/g, '-')]) return true\n  if (/^(?:pass|pwd|ccnum|cc\\s?num|cc\\s?number|cvv2?|cvc|csc)$/i.test(words)) return true\n  return CREDENTIAL_LABELS.some(({ pattern }) => pattern.test(words))\n}\n\n/** URLs a page carries in a form Links-by-shape would miss: `//host/…`. */\nfunction protocolRelativeUrls(strings: readonly string[]): string {\n  return strings\n    .filter((value) => /^\\/\\/[a-z0-9-]+(?:\\.[a-z0-9-]+)+/i.test(value.trim()))\n    .map((value) => `https:${value.trim()}`)\n    .join('\\n')\n}\n\n/**\n * Screen one page as it will be served. The caller applies the tiers with\n * `signalsThatHold` and decides what a hold does.\n */\nexport function screenHostedPage(input: HostedPageScreenInput): HostedPageScreenVerdict {\n  const signals: PhishingScreenSignal[] = []\n  const nodes = walkNodes(input.nodes)\n  const ownNames = squashScreenText((input.ownNames ?? []).filter(Boolean).join(' '))\n  const ownHosts = (input.ownDomains ?? [])\n    .map((domain) => String(domain ?? '').trim().toLowerCase().replace(/\\.+$/, ''))\n    .filter((domain) => domain.includes('.'))\n  const ownRegistrables = new Set(ownHosts.map(registrableDomain))\n  const isOwnHost = (host: string) => ownHosts.some((own) => host === own || host.endsWith(`.${own}`))\n\n  const nodeStrings = nodes.map((node) => propStrings(node.props))\n  let pageText = nodeStrings.map((strings) => strings.join('\\n')).join('\\n')\n  pageText = `${pageText}\\n${protocolRelativeUrls(nodeStrings.flat())}`.slice(0, MAX_PAGE_TEXT)\n\n  // 1. A lookalike host anywhere the page links, embeds or posts to.\n  const linkHosts = linkHostsIn(pageText)\n  for (const host of linkHosts) {\n    const brand = lookalikeBrandForHost(host)\n    if (brand && !isBrandsOwnDomain(brand, ownRegistrables)) {\n      signals.push({ code: 'lookalike-link', brand: brand.id, host })\n    }\n  }\n\n  // 2. A credential field the page's author defined.\n  const credentialSeen = new Set<string>()\n  const noteCredential = (field: 'password' | 'card' | 'otp', label: string) => {\n    if (credentialSeen.has(field)) return\n    credentialSeen.add(field)\n    signals.push({ code: 'credential-field', field, label: label.slice(0, 80) })\n  }\n  nodes.forEach((node, index) => {\n    if (isAuthoredField(node)) {\n      const credential = credentialOf(node)\n      if (credential) noteCredential(credential.field, credential.label)\n    }\n    for (const html of nodeStrings[index]) {\n      if (!html.includes('<')) continue\n      for (const { field, pattern } of HTML_CREDENTIAL_INPUTS) {\n        if (pattern.test(html)) noteCredential(field, 'an input in the page’s own HTML')\n      }\n    }\n  })\n\n  const isElsewhere = (host: string) => {\n    if (isOwnHost(host)) return false\n    const registrable = registrableDomain(host)\n    return !isAnyOfficialBrandDomain(registrable) && !isCommonLinkDomain(registrable)\n  }\n  const foreign = linkHosts.find(isElsewhere)\n  const actions = [...PAGE_ACTION_PATTERNS, ...PHISHING_LURE_PATTERNS]\n  const named = new Set<string>()\n  const isOthersBrand = (brand: PhishingScreenBrand) =>\n    !named.has(brand.id) && !isWorkspaceOwnBrand(brand, ownNames, ownRegistrables)\n\n  // 3. A brand's call to action, in one element, on a page that links away.\n  if (foreign) {\n    for (const strings of nodeStrings) {\n      const text = proseOf(strings.join(' '))\n      if (!text) continue\n      const action = firstMatch(actions, text)\n      if (!action) continue\n      for (const brand of PHISHING_SCREEN_BRANDS) {\n        if (isOthersBrand(brand) && brand.mention.test(text)) {\n          named.add(brand.id)\n          signals.push({ code: 'brand-action-page', brand: brand.id, action: action.slice(0, 80) })\n        }\n      }\n    }\n  }\n\n  const prose = proseOf(pageText)\n  const pageLure = firstMatch(PAGE_LURE_PATTERNS, prose)\n  const ctas = callsToAction(nodes)\n  const leaving = ctas.filter((cta): cta is CallToAction & { host: string } =>\n    Boolean(cta.host && isElsewhere(cta.host)),\n  )\n  const wordedLeaving = leaving.find((cta) => firstMatch(actions, cta.label))\n\n  // 4. A brand's lure, read over the page as a whole, on a page that links away.\n  const brandLure = pageLure ?? (wordedLeaving ? firstMatch(actions, wordedLeaving.label) : undefined)\n  if (foreign && brandLure) {\n    const host = (wordedLeaving ?? leaving[0])?.host ?? foreign\n    for (const brand of PHISHING_SCREEN_BRANDS) {\n      if (isOthersBrand(brand) && brand.mention.test(prose)) {\n        named.add(brand.id)\n        signals.push({ code: 'brand-lure-page', brand: brand.id, lure: brandLure.slice(0, 120), host })\n      }\n    }\n  }\n\n  // 5. The page's call to action leaves the site, beside a lure. Its only\n  //    one, or the one worded as an account or document action.\n  const primary = wordedLeaving ?? (ctas.length === 1 ? leaving[0] : undefined)\n  if (primary && pageLure) {\n    signals.push({\n      code: 'offsite-action-page',\n      action: (primary.label || primary.host).slice(0, 80),\n      lure: pageLure.slice(0, 120),\n      host: primary.host,\n    })\n  }\n\n  return { signals, linkHosts, linkUrls: linkUrlsIn(pageText) }\n}\n\n/**\n * What a visitor reads in page text: its markup taken out the way the email\n * screen takes it out ({@link visibleTextOf}), so a custom HTML block or a\n * rich-text prop cannot split a brand or a lure letter by letter with tags\n * (AGL-3453), and its bare URLs taken out too. Links are still read from\n * the raw text, where an href lives.\n */\nfunction proseOf(text: string): string {\n  return visibleTextOf(text).replace(PROSE_URL_PATTERN, ' ').replace(/\\s+/g, ' ').trim()\n}\n\n/** The first pattern's match in the text, or undefined. */\nfunction firstMatch(patterns: readonly RegExp[], text: string): string | undefined {\n  if (!text) return undefined\n  for (const pattern of patterns) {\n    const found = text.match(pattern)?.[0]\n    if (found) return found\n  }\n  return undefined\n}\n\n/** The words a call to action shows, from its label prop, markup taken out. */\nfunction callToActionLabel(props: Record<string, unknown>): string {\n  for (const key of CALL_TO_ACTION_LABEL_KEYS) {\n    const value = props[key]\n    if (typeof value === 'string' && value.trim()) {\n      return visibleTextOf(value).split('\\n')[0]\n    }\n  }\n  return ''\n}\n\n/**\n * Every call to action on the page: each button or link that carries a\n * target. A `screenId` is a page of the site and takes precedence over an\n * `href`, as the button renders it; a relative, `mailto:` or `tel:` href has\n * no web host and stays too.\n */\nfunction callsToAction(nodes: readonly WalkedNode[]): CallToAction[] {\n  const found: CallToAction[] = []\n  for (const node of nodes) {\n    if (!CALL_TO_ACTION_COMPONENT.test(node.componentId)) continue\n    const screenId = typeof node.props['screenId'] === 'string' ? node.props['screenId'].trim() : ''\n    const href = typeof node.props['href'] === 'string' ? node.props['href'].trim() : ''\n    if (!screenId && !href) continue\n    const absolute = href.startsWith('//') ? `https:${href}` : href\n    found.push({\n      label: callToActionLabel(node.props),\n      host: screenId ? null : (linkHostsIn(absolute)[0] ?? null),\n    })\n  }\n  return found\n}\n\nexport interface SiteRedirectScreenInput {\n  /** The path the rule answers — `/secure-document` — or its pattern. */\n  source: string\n  /** Where it sends the visitor, after any `$n` substitution. */\n  destination: string\n  /** The workspace's own names, as `screenHostedPage` reads them. */\n  ownNames?: readonly (string | null | undefined)[]\n  /** The workspace's own hosts — a redirect to one never leaves. */\n  ownDomains?: readonly (string | null | undefined)[]\n}\n\n/**\n * Screen one site redirect (AGL-3447): a rule that sends a path off the site\n * is a call to action that fires on arrival, so it is read with the page's\n * rules over the one text it has, its path.\n *\n * - `lookalike-link` (STRONG) — the destination wears a brand it is not.\n * - `offsite-redirect` (SOFT) — the destination is not the site's own, a\n *   listed brand's or a common link, and the path reads as an account or\n *   document-share lure (`/secure-document-access`).\n * - `brand-lure-link` (SOFT) — the same, and the path names a brand that is\n *   not the workspace's (`/paypal-verify-your-account`), the email screen's\n *   three-part lure.\n *\n * The caller applies the tiers with `signalsThatHold`, with the workspace's\n * age, and decides what a hold does.\n */\nexport function screenSiteRedirect(input: SiteRedirectScreenInput): HostedPageScreenVerdict {\n  const signals: PhishingScreenSignal[] = []\n  const destination = String(input.destination ?? '').trim()\n  const absolute = destination.startsWith('//') ? `https:${destination}` : destination\n  const hosts = linkHostsIn(absolute)\n  const linkUrls = linkUrlsIn(absolute)\n  if (!hosts.length) return { signals, linkHosts: hosts, linkUrls }\n  const ownNames = squashScreenText((input.ownNames ?? []).filter(Boolean).join(' '))\n  const ownHosts = (input.ownDomains ?? [])\n    .map((domain) => String(domain ?? '').trim().toLowerCase().replace(/\\.+$/, ''))\n    .filter((domain) => domain.includes('.'))\n  const ownRegistrables = new Set(ownHosts.map(registrableDomain))\n\n  for (const host of hosts) {\n    const brand = lookalikeBrandForHost(host)\n    if (brand && !isBrandsOwnDomain(brand, ownRegistrables)) {\n      signals.push({ code: 'lookalike-link', brand: brand.id, host })\n    }\n  }\n\n  // The first host is where the browser goes; the rest is a userinfo disguise.\n  const host = hosts[0]\n  const registrable = registrableDomain(host)\n  const leaves =\n    !ownHosts.some((own) => host === own || host.endsWith(`.${own}`)) &&\n    !isAnyOfficialBrandDomain(registrable) &&\n    !isCommonLinkDomain(registrable)\n  if (!leaves) return { signals, linkHosts: hosts, linkUrls }\n\n  const source = String(input.source ?? '').trim()\n  const words = pathWords(source)\n  const lure = firstMatch(PAGE_LURE_PATTERNS, words)\n  if (lure) {\n    signals.push({ code: 'offsite-redirect', source: source.slice(0, 120), lure: lure.slice(0, 120), host })\n  }\n  const anyLure = lure ?? firstMatch(PHISHING_LURE_PATTERNS, words)\n  if (anyLure) {\n    const brand = PHISHING_SCREEN_BRANDS.find(\n      (candidate) =>\n        candidate.mention.test(words) && !isWorkspaceOwnBrand(candidate, ownNames, ownRegistrables),\n    )\n    if (brand) {\n      signals.push({ code: 'brand-lure-link', brand: brand.id, lure: anyLure.slice(0, 120), host })\n    }\n  }\n  return { signals, linkHosts: hosts, linkUrls }\n}\n\n/** A path read as words: `/secure-document_access` → `secure document access`. */\nfunction pathWords(path: string): string {\n  let decoded = path\n  try {\n    decoded = decodeURIComponent(path)\n  } catch {\n    // A malformed escape is read as typed.\n  }\n  return decoded.replace(/[^a-z0-9]+/gi, ' ').trim()\n}\n"],"names":["ACCOUNT_LURE_PATTERNS","DOCUMENT_SHARE_LURE_PATTERNS","isWorkspaceOwnBrand","linkHostsIn","linkUrlsIn","isBrandsOwnDomain","lookalikeBrandForHost","PHISHING_LURE_PATTERNS","PHISHING_SCREEN_BRANDS","registrableDomain","isAnyOfficialBrandDomain","isCommonLinkDomain","squashScreenText","visibleTextOf","MAX_PAGE_TEXT","MAX_NODES","FIELD_PROP_KEYS","CREDENTIAL_AUTOCOMPLETE","CREDENTIAL_LABELS","field","pattern","HTML_CREDENTIAL_INPUTS","PAGE_ACTION_PATTERNS","PAGE_LURE_PATTERNS","CALL_TO_ACTION_COMPONENT","CALL_TO_ACTION_LABEL_KEYS","PROSE_URL_PATTERN","walkNodes","root","found","seen","Set","stack","head","length","value","has","add","record","props","push","componentId","child","Array","isArray","Object","values","propStrings","skip","out","visit","depth","entry","key","entries","isAuthoredField","node","test","some","undefined","credentialOf","text","type","join","toLowerCase","label","autocomplete","trim","byToken","wording","filter","Boolean","slice","credentialForInputAttribute","attribute","name","String","token","credentialInHtml","html","includes","find","isCredentialFieldName","words","replace","protocolRelativeUrls","strings","map","screenHostedPage","input","signals","nodes","ownNames","ownHosts","ownDomains","domain","ownRegistrables","isOwnHost","host","own","endsWith","nodeStrings","pageText","flat","linkHosts","brand","code","id","credentialSeen","noteCredential","forEach","index","credential","isElsewhere","registrable","foreign","actions","named","isOthersBrand","proseOf","action","firstMatch","mention","prose","pageLure","ctas","callsToAction","leaving","cta","wordedLeaving","brandLure","lure","primary","linkUrls","patterns","match","callToActionLabel","split","screenId","href","absolute","startsWith","screenSiteRedirect","destination","hosts","leaves","source","pathWords","anyLure","candidate","path","decoded","decodeURIComponent"],"mappings":"AAAA;;;;;;;;;;;;;;;CAeC,GAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;2CAwD2C,GAE3C,SACEA,qBAAqB,EACrBC,4BAA4B,EAC5BC,mBAAmB,EACnBC,WAAW,EACXC,UAAU,EACVC,iBAAiB,EACjBC,qBAAqB,EACrBC,sBAAsB,EACtBC,sBAAsB,EAGtBC,iBAAiB,EACjBC,wBAAwB,EACxBC,kBAAkB,EAClBC,gBAAgB,EAChBC,aAAa,QACR,gCAA4B;AAsCnC,4EAA4E,GAC5E,MAAMC,gBAAgB;AACtB,oCAAoC,GACpC,MAAMC,YAAY;AAElB,yEAAyE,GACzE,MAAMC,kBAAkB;IAAC;IAAa;IAAa;IAAa;IAAgB;CAAe;AAE/F,yDAAyD,GACzD,MAAMC,0BAAuE;IAC3E,oBAAoB;IACpB,gBAAgB;IAChB,aAAa;IACb,UAAU;IACV,UAAU;IACV,iBAAiB;AACnB;AAEA,uDAAuD,GACvD,MAAMC,oBAA4F;IAChG;QAAEC,OAAO;QAAYC,SAAS;IAA6E;IAC3G;QACED,OAAO;QACPC,SACE;IACJ;IACA;QACED,OAAO;QACPC,SACE;IACJ;CACD;AAED,6DAA6D,GAC7D,MAAMC,yBAAiG;IACrG;QAAEF,OAAO;QAAYC,SAAS;IAA2C;IACzE;QAAED,OAAO;QAAYC,SAAS;IAAwG;IACtI;QAAED,OAAO;QAAQC,SAAS;IAA+F;IACzH;QAAED,OAAO;QAAOC,SAAS;IAA8F;CACxH;AAED;;;;CAIC,GACD,MAAME,uBAA0C;IAC9C;IACA;IACA;IACA;IACA;IACA;IACA;CACD;AAED;;;;;CAKC,GACD,MAAMC,qBAAwC;OAAIvB;OAA0BC;CAA6B;AAEzG;;;;CAIC,GACD,MAAMuB,2BAA2B;AAEjC,8EAA8E,GAC9E,MAAMC,4BAA4B;IAAC;IAAY;IAAS;IAAQ;IAAS;CAAO;AAEhF,4EAA4E,GAC5E,MAAMC,oBAAoB;AAQ1B,sEAAsE,GACtE,SAASC,UAAUC,IAAa;IAC9B,MAAMC,QAAsB,EAAE;IAC9B,MAAMC,OAAO,IAAIC;IACjB,MAAMC,QAAmB;QAACJ;KAAK;IAC/B,IAAK,IAAIK,OAAO,GAAGA,OAAOD,MAAME,MAAM,IAAIL,MAAMK,MAAM,GAAGnB,WAAWkB,QAAQ,EAAG;QAC7E,MAAME,QAAQH,KAAK,CAACC,KAAK;QACzB,IAAI,CAACE,SAAS,OAAOA,UAAU,YAAYL,KAAKM,GAAG,CAACD,QAAQ;QAC5DL,KAAKO,GAAG,CAACF;QACT,MAAMG,SAASH;QACf,IAAI,OAAOG,MAAM,CAAC,cAAc,KAAK,UAAU;YAC7C,MAAMC,QAAQD,MAAM,CAAC,QAAQ;YAC7BT,MAAMW,IAAI,CAAC;gBACTC,aAAaH,MAAM,CAAC,cAAc;gBAClCC,OAAOA,SAAS,OAAOA,UAAU,WAAYA,QAAoC,CAAC;YACpF;QACF;QACA,KAAK,MAAMG,SAASC,MAAMC,OAAO,CAACT,SAASA,QAAQU,OAAOC,MAAM,CAACR,QAAS;YACxE,IAAII,SAAS,OAAOA,UAAU,UAAUV,MAAMQ,IAAI,CAACE;QACrD;IACF;IACA,OAAOb;AACT;AAEA,4EAA4E,GAC5E,SAASkB,YAAYR,KAA8B,EAAES,OAA4B,IAAIjB,KAAK;IACxF,MAAMkB,MAAgB,EAAE;IACxB,MAAMC,QAAQ,CAACf,OAAgBgB;QAC7B,IAAI,OAAOhB,UAAU,UAAUc,IAAIT,IAAI,CAACL;aACnC,IAAIgB,QAAQ,KAAKhB,SAAS,OAAOA,UAAU,UAAU;YACxD,KAAK,MAAMiB,SAAST,MAAMC,OAAO,CAACT,SAASA,QAAQU,OAAOC,MAAM,CAACX,OAAmC;gBAClGe,MAAME,OAAOD,QAAQ;YACvB;QACF;IACF;IACA,KAAK,MAAM,CAACE,KAAKlB,MAAM,IAAIU,OAAOS,OAAO,CAACf,OAAQ;QAChD,IAAIc,QAAQ,QAAQL,KAAKZ,GAAG,CAACiB,MAAM;QACnCH,MAAMf,OAAO;IACf;IACA,OAAOc;AACT;AAEA,wDAAwD,GACxD,SAASM,gBAAgBC,IAAgB;IACvC,IAAI,eAAeC,IAAI,CAACD,KAAKf,WAAW,GAAG,OAAO;IAClD,OAAOzB,gBAAgB0C,IAAI,CAAC,CAACL,MAAQG,KAAKjB,KAAK,CAACc,IAAI,KAAKM;AAC3D;AAEA,8CAA8C,GAC9C,SAASC,aAAaJ,IAAgB;IACpC,MAAMK,OAAO,CAACR,MAAiB,OAAOG,KAAKjB,KAAK,CAACc,IAAI,KAAK,WAAYG,KAAKjB,KAAK,CAACc,IAAI,GAAc;IACnG,MAAMS,OAAO;QAACD,KAAK;QAAcA,KAAK;QAASA,KAAK;KAAa,CAACE,IAAI,CAAC,KAAKC,WAAW;IACvF,MAAMC,QAAQJ,KAAK,YAAYA,KAAK,kBAAkBA,KAAK,gBAAgBA,KAAK;IAChF,IAAI,eAAeJ,IAAI,CAACK,OAAO,OAAO;QAAE3C,OAAO;QAAY8C,OAAOA,SAAS;IAAW;IACtF,MAAMC,eAAe,AAACL,CAAAA,KAAK,mBAAmBA,KAAK,eAAc,EAAGG,WAAW,GAAGG,IAAI;IACtF,MAAMC,UAAUnD,uBAAuB,CAACiD,aAAa;IACrD,IAAIE,SAAS,OAAO;QAAEjD,OAAOiD;QAASH,OAAOA,SAASC;IAAa;IACnE,MAAMG,UAAU;QAACR,KAAK;QAAUA,KAAK;QAAgBA,KAAK;QAAcA,KAAK;QAASA,KAAK;KAAc,CACtGS,MAAM,CAACC,SACPR,IAAI,CAAC;IACR,KAAK,MAAM,EAAE5C,KAAK,EAAEC,OAAO,EAAE,IAAIF,kBAAmB;QAClD,IAAIE,QAAQqC,IAAI,CAACY,UAAU,OAAO;YAAElD;YAAO8C,OAAO,AAACA,CAAAA,SAASI,OAAM,EAAGG,KAAK,CAAC,GAAG;QAAI;IACpF;IACA,OAAO;AACT;AAEA;;;;;;CAMC,GACD,OAAO,SAASC,4BACdC,SAAiB,EACjBvC,KAAa;QAKuBlB;IAHpC,MAAM0D,OAAOC,OAAOF,oBAAAA,YAAa,IAAIV,WAAW;IAChD,MAAMa,QAAQD,OAAOzC,gBAAAA,QAAS,IAAIgC,IAAI,GAAGH,WAAW;IACpD,IAAI,AAACW,CAAAA,SAAS,UAAUA,SAAS,WAAU,KAAME,UAAU,YAAY,OAAO;IAC9E,IAAIF,SAAS,gBAAgB,QAAO1D,iCAAAA,uBAAuB,CAAC4D,MAAM,YAA9B5D,iCAAkC;IACtE,OAAO;AACT;AAEA,iFAAiF,GACjF,OAAO,SAAS6D,iBAAiBC,IAAY;;QAGpC1D;IAFP,MAAMwC,OAAOe,OAAOG,eAAAA,OAAQ;IAC5B,IAAI,CAAClB,KAAKmB,QAAQ,CAAC,MAAM,OAAO;IAChC,gBAAO3D,+BAAAA,uBAAuB4D,IAAI,CAAC,CAAC,EAAE7D,OAAO,EAAE,GAAKA,QAAQqC,IAAI,CAACI,2BAA1DxC,6BAAkEF,KAAK,mBAAI;AACpF;AAEA;;;;;;;;;;CAUC,GACD,OAAO,SAAS+D,sBAAsBP,IAAY;IAChD,MAAMQ,QAAQP,OAAOD,eAAAA,OAAQ,IAC1BS,OAAO,CAAC,sBAAsB,SAC9BA,OAAO,CAAC,eAAe,KACvBjB,IAAI;IACP,IAAI,CAACgB,OAAO,OAAO;IACnB,IAAIlE,uBAAuB,CAACkE,MAAMnB,WAAW,GAAGoB,OAAO,CAAC,QAAQ,KAAK,EAAE,OAAO;IAC9E,IAAI,2DAA2D3B,IAAI,CAAC0B,QAAQ,OAAO;IACnF,OAAOjE,kBAAkBwC,IAAI,CAAC,CAAC,EAAEtC,OAAO,EAAE,GAAKA,QAAQqC,IAAI,CAAC0B;AAC9D;AAEA,yEAAyE,GACzE,SAASE,qBAAqBC,OAA0B;IACtD,OAAOA,QACJhB,MAAM,CAAC,CAACnC,QAAU,oCAAoCsB,IAAI,CAACtB,MAAMgC,IAAI,KACrEoB,GAAG,CAAC,CAACpD,QAAU,CAAC,MAAM,EAAEA,MAAMgC,IAAI,IAAI,EACtCJ,IAAI,CAAC;AACV;AAEA;;;CAGC,GACD,OAAO,SAASyB,iBAAiBC,KAA4B;QAGxBA,iBACjBA;IAHlB,MAAMC,UAAkC,EAAE;IAC1C,MAAMC,QAAQhE,UAAU8D,MAAME,KAAK;IACnC,MAAMC,WAAWhF,iBAAiB,EAAC6E,kBAAAA,MAAMG,QAAQ,YAAdH,kBAAkB,EAAE,EAAEnB,MAAM,CAACC,SAASR,IAAI,CAAC;IAC9E,MAAM8B,WAAW,EAACJ,oBAAAA,MAAMK,UAAU,YAAhBL,oBAAoB,EAAE,EACrCF,GAAG,CAAC,CAACQ,SAAWnB,OAAOmB,iBAAAA,SAAU,IAAI5B,IAAI,GAAGH,WAAW,GAAGoB,OAAO,CAAC,QAAQ,KAC1Ed,MAAM,CAAC,CAACyB,SAAWA,OAAOf,QAAQ,CAAC;IACtC,MAAMgB,kBAAkB,IAAIjE,IAAI8D,SAASN,GAAG,CAAC9E;IAC7C,MAAMwF,YAAY,CAACC,OAAiBL,SAASnC,IAAI,CAAC,CAACyC,MAAQD,SAASC,OAAOD,KAAKE,QAAQ,CAAC,CAAC,CAAC,EAAED,KAAK;IAElG,MAAME,cAAcV,MAAMJ,GAAG,CAAC,CAAC/B,OAAST,YAAYS,KAAKjB,KAAK;IAC9D,IAAI+D,WAAWD,YAAYd,GAAG,CAAC,CAACD,UAAYA,QAAQvB,IAAI,CAAC,OAAOA,IAAI,CAAC;IACrEuC,WAAW,GAAGA,SAAS,EAAE,EAAEjB,qBAAqBgB,YAAYE,IAAI,KAAK,CAAC/B,KAAK,CAAC,GAAG1D;IAE/E,mEAAmE;IACnE,MAAM0F,YAAYrG,YAAYmG;IAC9B,KAAK,MAAMJ,QAAQM,UAAW;QAC5B,MAAMC,QAAQnG,sBAAsB4F;QACpC,IAAIO,SAAS,CAACpG,kBAAkBoG,OAAOT,kBAAkB;YACvDN,QAAQlD,IAAI,CAAC;gBAAEkE,MAAM;gBAAkBD,OAAOA,MAAME,EAAE;gBAAET;YAAK;QAC/D;IACF;IAEA,mDAAmD;IACnD,MAAMU,iBAAiB,IAAI7E;IAC3B,MAAM8E,iBAAiB,CAAC1F,OAAoC8C;QAC1D,IAAI2C,eAAexE,GAAG,CAACjB,QAAQ;QAC/ByF,eAAevE,GAAG,CAAClB;QACnBuE,QAAQlD,IAAI,CAAC;YAAEkE,MAAM;YAAoBvF;YAAO8C,OAAOA,MAAMO,KAAK,CAAC,GAAG;QAAI;IAC5E;IACAmB,MAAMmB,OAAO,CAAC,CAACtD,MAAMuD;QACnB,IAAIxD,gBAAgBC,OAAO;YACzB,MAAMwD,aAAapD,aAAaJ;YAChC,IAAIwD,YAAYH,eAAeG,WAAW7F,KAAK,EAAE6F,WAAW/C,KAAK;QACnE;QACA,KAAK,MAAMc,QAAQsB,WAAW,CAACU,MAAM,CAAE;YACrC,IAAI,CAAChC,KAAKC,QAAQ,CAAC,MAAM;YACzB,KAAK,MAAM,EAAE7D,KAAK,EAAEC,OAAO,EAAE,IAAIC,uBAAwB;gBACvD,IAAID,QAAQqC,IAAI,CAACsB,OAAO8B,eAAe1F,OAAO;YAChD;QACF;IACF;IAEA,MAAM8F,cAAc,CAACf;QACnB,IAAID,UAAUC,OAAO,OAAO;QAC5B,MAAMgB,cAAczG,kBAAkByF;QACtC,OAAO,CAACxF,yBAAyBwG,gBAAgB,CAACvG,mBAAmBuG;IACvE;IACA,MAAMC,UAAUX,UAAUvB,IAAI,CAACgC;IAC/B,MAAMG,UAAU;WAAI9F;WAAyBf;KAAuB;IACpE,MAAM8G,QAAQ,IAAItF;IAClB,MAAMuF,gBAAgB,CAACb,QACrB,CAACY,MAAMjF,GAAG,CAACqE,MAAME,EAAE,KAAK,CAACzG,oBAAoBuG,OAAOb,UAAUI;IAEhE,0EAA0E;IAC1E,IAAImB,SAAS;QACX,KAAK,MAAM7B,WAAWe,YAAa;YACjC,MAAMxC,OAAO0D,QAAQjC,QAAQvB,IAAI,CAAC;YAClC,IAAI,CAACF,MAAM;YACX,MAAM2D,SAASC,WAAWL,SAASvD;YACnC,IAAI,CAAC2D,QAAQ;YACb,KAAK,MAAMf,SAASjG,uBAAwB;gBAC1C,IAAI8G,cAAcb,UAAUA,MAAMiB,OAAO,CAACjE,IAAI,CAACI,OAAO;oBACpDwD,MAAMhF,GAAG,CAACoE,MAAME,EAAE;oBAClBjB,QAAQlD,IAAI,CAAC;wBAAEkE,MAAM;wBAAqBD,OAAOA,MAAME,EAAE;wBAAEa,QAAQA,OAAOhD,KAAK,CAAC,GAAG;oBAAI;gBACzF;YACF;QACF;IACF;IAEA,MAAMmD,QAAQJ,QAAQjB;IACtB,MAAMsB,WAAWH,WAAWlG,oBAAoBoG;IAChD,MAAME,OAAOC,cAAcnC;IAC3B,MAAMoC,UAAUF,KAAKvD,MAAM,CAAC,CAAC0D,MAC3BzD,QAAQyD,IAAI9B,IAAI,IAAIe,YAAYe,IAAI9B,IAAI;IAE1C,MAAM+B,gBAAgBF,QAAQ9C,IAAI,CAAC,CAAC+C,MAAQP,WAAWL,SAASY,IAAI/D,KAAK;IAEzE,+EAA+E;IAC/E,MAAMiE,YAAYN,mBAAAA,WAAaK,gBAAgBR,WAAWL,SAASa,cAAchE,KAAK,IAAIN;IAC1F,IAAIwD,WAAWe,WAAW;;YACVD;QAAd,MAAM/B,gBAAQ+B,QAAAA,wBAAAA,gBAAiBF,OAAO,CAAC,EAAE,qBAA5B,AAACE,MAA8B/B,IAAI,mBAAIiB;QACpD,KAAK,MAAMV,SAASjG,uBAAwB;YAC1C,IAAI8G,cAAcb,UAAUA,MAAMiB,OAAO,CAACjE,IAAI,CAACkE,QAAQ;gBACrDN,MAAMhF,GAAG,CAACoE,MAAME,EAAE;gBAClBjB,QAAQlD,IAAI,CAAC;oBAAEkE,MAAM;oBAAmBD,OAAOA,MAAME,EAAE;oBAAEwB,MAAMD,UAAU1D,KAAK,CAAC,GAAG;oBAAM0B;gBAAK;YAC/F;QACF;IACF;IAEA,wEAAwE;IACxE,8DAA8D;IAC9D,MAAMkC,UAAUH,wBAAAA,gBAAkBJ,KAAK3F,MAAM,KAAK,IAAI6F,OAAO,CAAC,EAAE,GAAGpE;IACnE,IAAIyE,WAAWR,UAAU;QACvBlC,QAAQlD,IAAI,CAAC;YACXkE,MAAM;YACNc,QAAQ,AAACY,CAAAA,QAAQnE,KAAK,IAAImE,QAAQlC,IAAI,AAAD,EAAG1B,KAAK,CAAC,GAAG;YACjD2D,MAAMP,SAASpD,KAAK,CAAC,GAAG;YACxB0B,MAAMkC,QAAQlC,IAAI;QACpB;IACF;IAEA,OAAO;QAAER;QAASc;QAAW6B,UAAUjI,WAAWkG;IAAU;AAC9D;AAEA;;;;;;CAMC,GACD,SAASiB,QAAQ1D,IAAY;IAC3B,OAAOhD,cAAcgD,MAAMuB,OAAO,CAAC1D,mBAAmB,KAAK0D,OAAO,CAAC,QAAQ,KAAKjB,IAAI;AACtF;AAEA,yDAAyD,GACzD,SAASsD,WAAWa,QAA2B,EAAEzE,IAAY;IAC3D,IAAI,CAACA,MAAM,OAAOF;IAClB,KAAK,MAAMvC,WAAWkH,SAAU;YAChBzE;QAAd,MAAMhC,SAAQgC,cAAAA,KAAK0E,KAAK,CAACnH,6BAAXyC,WAAqB,CAAC,EAAE;QACtC,IAAIhC,OAAO,OAAOA;IACpB;IACA,OAAO8B;AACT;AAEA,6EAA6E,GAC7E,SAAS6E,kBAAkBjG,KAA8B;IACvD,KAAK,MAAMc,OAAO5B,0BAA2B;QAC3C,MAAMU,QAAQI,KAAK,CAACc,IAAI;QACxB,IAAI,OAAOlB,UAAU,YAAYA,MAAMgC,IAAI,IAAI;YAC7C,OAAOtD,cAAcsB,OAAOsG,KAAK,CAAC,KAAK,CAAC,EAAE;QAC5C;IACF;IACA,OAAO;AACT;AAEA;;;;;CAKC,GACD,SAASX,cAAcnC,KAA4B;IACjD,MAAM9D,QAAwB,EAAE;IAChC,KAAK,MAAM2B,QAAQmC,MAAO;YAQGxF;QAP3B,IAAI,CAACqB,yBAAyBiC,IAAI,CAACD,KAAKf,WAAW,GAAG;QACtD,MAAMiG,WAAW,OAAOlF,KAAKjB,KAAK,CAAC,WAAW,KAAK,WAAWiB,KAAKjB,KAAK,CAAC,WAAW,CAAC4B,IAAI,KAAK;QAC9F,MAAMwE,OAAO,OAAOnF,KAAKjB,KAAK,CAAC,OAAO,KAAK,WAAWiB,KAAKjB,KAAK,CAAC,OAAO,CAAC4B,IAAI,KAAK;QAClF,IAAI,CAACuE,YAAY,CAACC,MAAM;QACxB,MAAMC,WAAWD,KAAKE,UAAU,CAAC,QAAQ,CAAC,MAAM,EAAEF,MAAM,GAAGA;QAC3D9G,MAAMW,IAAI,CAAC;YACTyB,OAAOuE,kBAAkBhF,KAAKjB,KAAK;YACnC2D,MAAMwC,WAAW,QAAQvI,gBAAAA,YAAYyI,SAAS,CAAC,EAAE,YAAxBzI,gBAA4B;QACvD;IACF;IACA,OAAO0B;AACT;AAaA;;;;;;;;;;;;;;;CAeC,GACD,OAAO,SAASiH,mBAAmBrD,KAA8B;QAEpCA,oBAKQA,iBACjBA,mBAqBIA;IA5BtB,MAAMC,UAAkC,EAAE;IAC1C,MAAMqD,cAAcnE,QAAOa,qBAAAA,MAAMsD,WAAW,YAAjBtD,qBAAqB,IAAItB,IAAI;IACxD,MAAMyE,WAAWG,YAAYF,UAAU,CAAC,QAAQ,CAAC,MAAM,EAAEE,aAAa,GAAGA;IACzE,MAAMC,QAAQ7I,YAAYyI;IAC1B,MAAMP,WAAWjI,WAAWwI;IAC5B,IAAI,CAACI,MAAM9G,MAAM,EAAE,OAAO;QAAEwD;QAASc,WAAWwC;QAAOX;IAAS;IAChE,MAAMzC,WAAWhF,iBAAiB,EAAC6E,kBAAAA,MAAMG,QAAQ,YAAdH,kBAAkB,EAAE,EAAEnB,MAAM,CAACC,SAASR,IAAI,CAAC;IAC9E,MAAM8B,WAAW,EAACJ,oBAAAA,MAAMK,UAAU,YAAhBL,oBAAoB,EAAE,EACrCF,GAAG,CAAC,CAACQ,SAAWnB,OAAOmB,iBAAAA,SAAU,IAAI5B,IAAI,GAAGH,WAAW,GAAGoB,OAAO,CAAC,QAAQ,KAC1Ed,MAAM,CAAC,CAACyB,SAAWA,OAAOf,QAAQ,CAAC;IACtC,MAAMgB,kBAAkB,IAAIjE,IAAI8D,SAASN,GAAG,CAAC9E;IAE7C,KAAK,MAAMyF,QAAQ8C,MAAO;QACxB,MAAMvC,QAAQnG,sBAAsB4F;QACpC,IAAIO,SAAS,CAACpG,kBAAkBoG,OAAOT,kBAAkB;YACvDN,QAAQlD,IAAI,CAAC;gBAAEkE,MAAM;gBAAkBD,OAAOA,MAAME,EAAE;gBAAET;YAAK;QAC/D;IACF;IAEA,6EAA6E;IAC7E,MAAMA,OAAO8C,KAAK,CAAC,EAAE;IACrB,MAAM9B,cAAczG,kBAAkByF;IACtC,MAAM+C,SACJ,CAACpD,SAASnC,IAAI,CAAC,CAACyC,MAAQD,SAASC,OAAOD,KAAKE,QAAQ,CAAC,CAAC,CAAC,EAAED,KAAK,MAC/D,CAACzF,yBAAyBwG,gBAC1B,CAACvG,mBAAmBuG;IACtB,IAAI,CAAC+B,QAAQ,OAAO;QAAEvD;QAASc,WAAWwC;QAAOX;IAAS;IAE1D,MAAMa,SAAStE,QAAOa,gBAAAA,MAAMyD,MAAM,YAAZzD,gBAAgB,IAAItB,IAAI;IAC9C,MAAMgB,QAAQgE,UAAUD;IACxB,MAAMf,OAAOV,WAAWlG,oBAAoB4D;IAC5C,IAAIgD,MAAM;QACRzC,QAAQlD,IAAI,CAAC;YAAEkE,MAAM;YAAoBwC,QAAQA,OAAO1E,KAAK,CAAC,GAAG;YAAM2D,MAAMA,KAAK3D,KAAK,CAAC,GAAG;YAAM0B;QAAK;IACxG;IACA,MAAMkD,UAAUjB,eAAAA,OAAQV,WAAWlH,wBAAwB4E;IAC3D,IAAIiE,SAAS;QACX,MAAM3C,QAAQjG,uBAAuByE,IAAI,CACvC,CAACoE,YACCA,UAAU3B,OAAO,CAACjE,IAAI,CAAC0B,UAAU,CAACjF,oBAAoBmJ,WAAWzD,UAAUI;QAE/E,IAAIS,OAAO;YACTf,QAAQlD,IAAI,CAAC;gBAAEkE,MAAM;gBAAmBD,OAAOA,MAAME,EAAE;gBAAEwB,MAAMiB,QAAQ5E,KAAK,CAAC,GAAG;gBAAM0B;YAAK;QAC7F;IACF;IACA,OAAO;QAAER;QAASc,WAAWwC;QAAOX;IAAS;AAC/C;AAEA,gFAAgF,GAChF,SAASc,UAAUG,IAAY;IAC7B,IAAIC,UAAUD;IACd,IAAI;QACFC,UAAUC,mBAAmBF;IAC/B,EAAE,eAAM;IACN,uCAAuC;IACzC;IACA,OAAOC,QAAQnE,OAAO,CAAC,gBAAgB,KAAKjB,IAAI;AAClD"}

package/src/lib/link-reputation.d.ts

@@ -18,27 +18,49 @@ import type { PhishingScreenSignal } from './outbound-phishing-screen';
 /** The threat lists a host is looked up against. */
 export type LinkThreatType = 'SOCIAL_ENGINEERING' | 'MALWARE' | 'UNWANTED_SOFTWARE';
 export declare const LINK_THREAT_TYPES: readonly LinkThreatType[];
-/** A host the list names, and what it names it for. */
+/**
+ * A host the list names, or one address on it, and what it names it for.
+ * `url` is set when the listing is the address's, not the host's.
+ */
 export interface HostReputationHit {
     host: string;
     threats: LinkThreatType[];
+    /** The address the list names (scheme, host and path), when it is one page on a clean host. */
+    url?: string;
 }
-/** What one lookup answered, host by host. */
+/** What one lookup answered, host by host and address by address. */
 export interface LinkReputationAnswer {
-    /** Hosts the list names. */
+    /** Hosts, and addresses, the list names. */
     hits: HostReputationHit[];
-    /** Hosts the list does not name. */
+    /** Hosts and addresses the list does not name. */
     clean: string[];
-    /** Hosts nobody could answer for: an error, a timeout, a switched-off lookup. */
+    /** Hosts and addresses nobody could answer for: an error, a timeout, a switched-off lookup. */
     unknown: string[];
 }
 export interface LinkReputationLookupOptions {
     /** How long the caller will wait, overall. */
     deadlineMs?: number;
+    /**
+     * The links' addresses, as {@link normalizeReputationLink} reduced them —
+     * looked up only in `'url'` mode, and only on the hosts asked about.
+     */
+    urls?: readonly string[];
 }
 export type LinkReputationLookup = (hosts: readonly string[], options?: LinkReputationLookupOptions) => Promise<LinkReputationAnswer>;
 /** The most foreign hosts one page, message or redirect is looked up for. */
 export declare const MAX_REPUTATION_HOSTS_PER_LOOKUP = 20;
+/**
+ * The most link addresses (path included) one page, message or redirect is
+ * looked up for in `'url'` mode, beside its hosts (AGL-3459). With the hosts,
+ * at most forty calls, all inside the same per-page deadline.
+ */
+export declare const MAX_REPUTATION_URLS_PER_LOOKUP = 20;
+/**
+ * The longest path an address keeps. Longer is cut back to the last `/`
+ * within it, so what is looked up is still a path the link really starts
+ * with — Web Risk lists a kit by its path prefix.
+ */
+export declare const MAX_REPUTATION_URL_PATH_LENGTH = 512;
 /** Installs the lookup. Called once, from `@aglyn/tenant-data-admin`. */
 export declare function setLinkReputationLookup(lookup: LinkReputationLookup | null): void;
 /** The installed lookup, or null. */
@@ -47,6 +69,52 @@ export declare function getLinkReputationLookup(): LinkReputationLookup | null;
 export declare function resetLinkReputationLookupForTests(): void;
 /** A host as the lookup takes it, or null for anything that is not one. */
 export declare function normalizeReputationHost(host: string): string | null;
+/** A link reduced to what may be looked up: its host, and its address without query or fragment. */
+export interface ReputationLink {
+    host: string;
+    /** `scheme://host[:port]/path` — never a query string, a fragment or a `user:password@`. */
+    url: string;
+}
+/**
+ * A link reduced to what Web Risk may be asked about it (AGL-3459), or null
+ * for anything that is not an `http(s)` link to a real host.
+ *
+ * - `www.…`, `//…` and a bare host are read as `https://`;
+ * - the query string, the fragment and any `user:password@` are dropped —
+ *   that is where a recipient's address and a session's token live;
+ * - the host is lowercased, a default port dropped, and `.` and `..`
+ *   segments resolved (WHATWG URL rules);
+ * - the path stops before the first segment carrying an email address or an
+ *   unrendered merge tag, and is cut back to a `/` within
+ *   {@link MAX_REPUTATION_URL_PATH_LENGTH}.
+ */
+export declare function normalizeReputationLink(link: string): ReputationLink | null;
+/** Is this address a host's front door — nothing a host lookup did not already ask? */
+export declare function isRootReputationUrl(url: string): boolean;
+/**
+ * The addresses worth asking about beside `hosts`: those on one of them
+ * whose path is more than `/`, each once, taken a host at a time — every
+ * host's first address, then every host's second — up to `max`
+ * ({@link MAX_REPUTATION_URLS_PER_LOOKUP}). A link that does not normalize
+ * is skipped.
+ */
+export declare function pickReputationUrls(links: readonly string[], hosts: readonly string[], max?: number): string[];
+/** The foreign hosts a set of links points at, and the addresses on them worth asking about. */
+export interface ForeignReputationLinks {
+    hosts: string[];
+    urls: string[];
+}
+/**
+ * {@link foreignHostsForReputation} over links — hosts or whole links — and
+ * the addresses on those hosts ({@link pickReputationUrls}). The addresses
+ * are only ever looked up in `'url'` mode; the lookup decides.
+ */
+export declare function foreignLinksForReputation(links: readonly string[], options?: {
+    ownDomains?: readonly (string | null | undefined)[];
+    excludeDomains?: readonly (string | null | undefined)[];
+    max?: number;
+    maxUrls?: number;
+}): ForeignReputationLinks;
 /**
  * The hosts worth asking about: foreign ones. Not the workspace's own, not
  * a domain under one of `excludeDomains` (the platform's own), not a listed
@@ -63,11 +131,13 @@ export declare function foreignHostsForReputation(hosts: readonly string[], opti
 /** A listing, as the screen's signal. */
 export declare function webRiskSignals(hits: readonly HostReputationHit[]): PhishingScreenSignal[];
 /**
- * Look the foreign hosts up through the installed lookup and answer the
- * signals they carry. Nothing installed, nothing foreign, an error: no
- * signals. Never throws.
+ * Look the foreign links up through the installed lookup and answer the
+ * signals they carry. `links` are hosts or whole links (`linkUrlsIn`): the
+ * hosts are asked about, and their addresses too when the lookup is in
+ * `'url'` mode. Nothing installed, nothing foreign, an error: no signals.
+ * Never throws.
  */
-export declare function linkReputationSignals(hosts: readonly string[], options?: {
+export declare function linkReputationSignals(links: readonly string[], options?: {
     ownDomains?: readonly (string | null | undefined)[];
     excludeDomains?: readonly (string | null | undefined)[];
     deadlineMs?: number;

package/src/lib/link-reputation.js

@@ -37,18 +37,47 @@
  * A lookup that errors or times out says nothing about the host. It never
  * holds by itself; the lookup reports the host as unknown and logs a warning.
  *
- * ## Only the host
+ * ## The host, and — only when switched on — the path
  *
- * The lookup sees `https://<host>/` and nothing else: never the path, the
- * query, the page or the message. A host is enough to catch a domain that
- * exists to harvest, and it is the least a third party can be told.
- *=========================================*/ import { isAnyOfficialBrandDomain, isCommonLinkDomain, registrableDomain } from "./outbound-phishing-screen.js";
+ * By default (`platformSettings/webRisk.lookupMode: 'host'`) the lookup sees
+ * `https://<host>/` and nothing else. A host is enough to catch a domain
+ * that exists to harvest, and it is the least a third party can be told.
+ *
+ * A phishing kit as often sits at a path on a compromised site that is
+ * otherwise legitimate, which a host lookup reads as clean. In `'url'` mode
+ * (AGL-3459) each link's address is looked up too, as
+ * {@link normalizeReputationLink} reduces it: scheme, host and path. Never
+ * the query string or the fragment, which is where a recipient's address
+ * and a session's token live; never a `user:password@`; and the path stops
+ * before any segment that carries an email address or an unrendered merge
+ * tag. Sending paths makes Google a recipient for a new purpose, so the
+ * switch is flipped only once the Subprocessors page names it.
+ *
+ * ## The budget
+ *
+ * One page, message or redirect asks about at most
+ * {@link MAX_REPUTATION_HOSTS_PER_LOOKUP} hosts and, in `'url'` mode, at most
+ * {@link MAX_REPUTATION_URLS_PER_LOOKUP} addresses, deduplicated and taken a
+ * host at a time ({@link pickReputationUrls}), so a long list of links to
+ * one site cannot crowd out the one link to another.
+ *=========================================*/ import { _ as _extends } from "@swc/helpers/_/_extends";
+import { isAnyOfficialBrandDomain, isCommonLinkDomain, registrableDomain } from "./outbound-phishing-screen.js";
 export const LINK_THREAT_TYPES = [
     'SOCIAL_ENGINEERING',
     'MALWARE',
     'UNWANTED_SOFTWARE'
 ];
 /** The most foreign hosts one page, message or redirect is looked up for. */ export const MAX_REPUTATION_HOSTS_PER_LOOKUP = 20;
+/**
+ * The most link addresses (path included) one page, message or redirect is
+ * looked up for in `'url'` mode, beside its hosts (AGL-3459). With the hosts,
+ * at most forty calls, all inside the same per-page deadline.
+ */ export const MAX_REPUTATION_URLS_PER_LOOKUP = 20;
+/**
+ * The longest path an address keeps. Longer is cut back to the last `/`
+ * within it, so what is looked up is still a path the link really starts
+ * with — Web Risk lists a kit by its path prefix.
+ */ export const MAX_REPUTATION_URL_PATH_LENGTH = 512;
 let installedLookup = null;
 /** Installs the lookup. Called once, from `@aglyn/tenant-data-admin`. */ export function setLinkReputationLookup(lookup) {
     installedLookup = lookup;
@@ -64,6 +93,99 @@ const HOST_PATTERN = /^[a-z0-9](?:[a-z0-9-]{0,62}[a-z0-9])?(?:\.[a-z0-9](?:[a-z0
     const value = String(host != null ? host : '').trim().toLowerCase().replace(/\.+$/, '');
     return value.length <= 253 && HOST_PATTERN.test(value) ? value : null;
 }
+/** A path segment that carries an email address, or a merge tag nobody has filled in. */ const PRIVATE_SEGMENT = /@|%40|\{|%7b|\}|%7d/i;
+/**
+ * A link reduced to what Web Risk may be asked about it (AGL-3459), or null
+ * for anything that is not an `http(s)` link to a real host.
+ *
+ * - `www.…`, `//…` and a bare host are read as `https://`;
+ * - the query string, the fragment and any `user:password@` are dropped —
+ *   that is where a recipient's address and a session's token live;
+ * - the host is lowercased, a default port dropped, and `.` and `..`
+ *   segments resolved (WHATWG URL rules);
+ * - the path stops before the first segment carrying an email address or an
+ *   unrendered merge tag, and is cut back to a `/` within
+ *   {@link MAX_REPUTATION_URL_PATH_LENGTH}.
+ */ export function normalizeReputationLink(link) {
+    let text = String(link != null ? link : '').trim();
+    if (!text) return null;
+    if (text.startsWith('//')) text = `https:${text}`;
+    else if (/^www\./i.test(text)) text = `https://${text}`;
+    else if (/^[a-z][a-z0-9+.-]*:(?!\d)/i.test(text)) {
+        if (!/^https?:\/\//i.test(text)) return null;
+    } else text = `https://${text}`;
+    let parsed;
+    try {
+        parsed = new URL(text);
+    } catch (unused) {
+        return null;
+    }
+    if (parsed.protocol !== 'https:' && parsed.protocol !== 'http:') return null;
+    const host = normalizeReputationHost(parsed.hostname);
+    if (!host) return null;
+    const kept = [];
+    for (const segment of parsed.pathname.split('/')){
+        if (PRIVATE_SEGMENT.test(segment)) {
+            kept.push('');
+            break;
+        }
+        kept.push(segment);
+    }
+    let path = kept.join('/') || '/';
+    if (path.length > MAX_REPUTATION_URL_PATH_LENGTH) {
+        path = path.slice(0, path.lastIndexOf('/', MAX_REPUTATION_URL_PATH_LENGTH - 1) + 1);
+    }
+    const port = parsed.port ? `:${parsed.port}` : '';
+    return {
+        host,
+        url: `${parsed.protocol}//${host}${port}${path}`
+    };
+}
+/** Is this address a host's front door — nothing a host lookup did not already ask? */ export function isRootReputationUrl(url) {
+    return /^https?:\/\/[^/]+\/$/.test(url);
+}
+/**
+ * The addresses worth asking about beside `hosts`: those on one of them
+ * whose path is more than `/`, each once, taken a host at a time — every
+ * host's first address, then every host's second — up to `max`
+ * ({@link MAX_REPUTATION_URLS_PER_LOOKUP}). A link that does not normalize
+ * is skipped.
+ */ export function pickReputationUrls(links, hosts, max = MAX_REPUTATION_URLS_PER_LOOKUP) {
+    const byHost = new Map(hosts.map((host)=>[
+            host,
+            []
+        ]));
+    const seen = new Set();
+    for (const raw of links){
+        const link = normalizeReputationLink(raw);
+        if (!link || seen.has(link.url) || isRootReputationUrl(link.url)) continue;
+        const list = byHost.get(link.host);
+        if (!list) continue;
+        seen.add(link.url);
+        list.push(link.url);
+    }
+    const picked = [];
+    for(let round = 0; picked.length < max; round += 1){
+        const before = picked.length;
+        for (const list of byHost.values()){
+            if (round < list.length && picked.length < max) picked.push(list[round]);
+        }
+        if (picked.length === before) break;
+    }
+    return picked;
+}
+/**
+ * {@link foreignHostsForReputation} over links — hosts or whole links — and
+ * the addresses on those hosts ({@link pickReputationUrls}). The addresses
+ * are only ever looked up in `'url'` mode; the lookup decides.
+ */ export function foreignLinksForReputation(links, options = {}) {
+    const parsed = links.map((link)=>normalizeReputationLink(link)).filter((link)=>Boolean(link));
+    const hosts = foreignHostsForReputation(parsed.map((link)=>link.host), options);
+    return {
+        hosts,
+        urls: pickReputationUrls(parsed.map((link)=>link.url), hosts, options.maxUrls)
+    };
+}
 /**
  * The hosts worth asking about: foreign ones. Not the workspace's own, not
  * a domain under one of `excludeDomains` (the platform's own), not a listed
@@ -92,28 +214,33 @@ const HOST_PATTERN = /^[a-z0-9](?:[a-z0-9-]{0,62}[a-z0-9])?(?:\.[a-z0-9](?:[a-z0
     return out;
 }
 /** A listing, as the screen's signal. */ export function webRiskSignals(hits) {
-    return hits.map((hit)=>({
+    return hits.map((hit)=>_extends({
             code: 'web-risk-link',
             host: hit.host,
             threats: [
                 ...new Set(hit.threats)
             ].sort()
-        }));
+        }, hit.url ? {
+            url: hit.url
+        } : {}));
 }
 /**
- * Look the foreign hosts up through the installed lookup and answer the
- * signals they carry. Nothing installed, nothing foreign, an error: no
- * signals. Never throws.
- */ export async function linkReputationSignals(hosts, options = {}) {
+ * Look the foreign links up through the installed lookup and answer the
+ * signals they carry. `links` are hosts or whole links (`linkUrlsIn`): the
+ * hosts are asked about, and their addresses too when the lookup is in
+ * `'url'` mode. Nothing installed, nothing foreign, an error: no signals.
+ * Never throws.
+ */ export async function linkReputationSignals(links, options = {}) {
     const lookup = installedLookup;
     if (!lookup) return [];
-    const foreign = foreignHostsForReputation(hosts, options);
-    if (!foreign.length) return [];
+    const foreign = foreignLinksForReputation(links, options);
+    if (!foreign.hosts.length) return [];
     try {
-        const answer = await lookup(foreign, {
-            deadlineMs: options.deadlineMs
+        const answer = await lookup(foreign.hosts, {
+            deadlineMs: options.deadlineMs,
+            urls: foreign.urls
         });
-        return webRiskSignals(answer.hits.filter((hit)=>foreign.includes(hit.host)));
+        return webRiskSignals(answer.hits.filter((hit)=>foreign.hosts.includes(hit.host) && (!hit.url || foreign.urls.includes(hit.url))));
     } catch (error) {
         console.warn('[link-reputation] the lookup failed — no signal', error);
         return [];

package/src/lib/link-reputation.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../../../../../../libs/shared/util/email/src/lib/link-reputation.ts"],"sourcesContent":["/**\n * @license\n * Copyright 2026 Aglyn LLC\n *\n * Licensed under the Apache License, Version 2.0 (the \"License\");\n * you may not use this file except in compliance with the License.\n * You may obtain a copy of the License at\n *\n *   http://www.apache.org/licenses/LICENSE-2.0\n *\n * Unless required by applicable law or agreed to in writing, software\n * distributed under the License is distributed on an \"AS IS\" BASIS,\n * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n * See the License for the specific language governing permissions and\n * limitations under the License.\n */\n\n/*==========================================\n * LINK REPUTATION (AGL-3451): what the phishing screen asks about a host\n * nobody's rules can judge.\n *\n * The screen's own rules catch a page or a message that LOOKS like a lure —\n * a brand's lookalike host, a password field, a brand beside a sign-in. The\n * 2026-10-01 harvesters linked to `temps-juenes.com` and\n * `conservascaorvi.com`, which wear no brand at all, so nothing held them.\n * A host already known to be bad anywhere is caught here instead: every\n * FOREIGN host a page, an email or a redirect points at is looked up against\n * a reputation list (Google Web Risk), and a listed one is a\n * `web-risk-link` signal — STRONG, so it holds for every workspace.\n *\n * Split like the send seam's gate: the pure half lives here (which hosts are\n * foreign, how a listing becomes a signal), and the lookup — a network call\n * and a cache — is installed by `@aglyn/tenant-data-admin`\n * (`web-risk.ts`) through {@link setLinkReputationLookup}. Nothing installed\n * means nothing looked up.\n *\n * ## Not evidence when it fails\n *\n * A lookup that errors or times out says nothing about the host. It never\n * holds by itself; the lookup reports the host as unknown and logs a warning.\n *\n * ## Only the host\n *\n * The lookup sees `https://<host>/` and nothing else: never the path, the\n * query, the page or the message. A host is enough to catch a domain that\n * exists to harvest, and it is the least a third party can be told.\n *=========================================*/\n\nimport type { PhishingScreenSignal } from './outbound-phishing-screen'\nimport {\n  isAnyOfficialBrandDomain,\n  isCommonLinkDomain,\n  registrableDomain,\n} from './outbound-phishing-screen'\n\n/** The threat lists a host is looked up against. */\nexport type LinkThreatType = 'SOCIAL_ENGINEERING' | 'MALWARE' | 'UNWANTED_SOFTWARE'\n\nexport const LINK_THREAT_TYPES: readonly LinkThreatType[] = [\n  'SOCIAL_ENGINEERING',\n  'MALWARE',\n  'UNWANTED_SOFTWARE',\n]\n\n/** A host the list names, and what it names it for. */\nexport interface HostReputationHit {\n  host: string\n  threats: LinkThreatType[]\n}\n\n/** What one lookup answered, host by host. */\nexport interface LinkReputationAnswer {\n  /** Hosts the list names. */\n  hits: HostReputationHit[]\n  /** Hosts the list does not name. */\n  clean: string[]\n  /** Hosts nobody could answer for: an error, a timeout, a switched-off lookup. */\n  unknown: string[]\n}\n\nexport interface LinkReputationLookupOptions {\n  /** How long the caller will wait, overall. */\n  deadlineMs?: number\n}\n\nexport type LinkReputationLookup = (\n  hosts: readonly string[],\n  options?: LinkReputationLookupOptions,\n) => Promise<LinkReputationAnswer>\n\n/** The most foreign hosts one page, message or redirect is looked up for. */\nexport const MAX_REPUTATION_HOSTS_PER_LOOKUP = 20\n\nlet installedLookup: LinkReputationLookup | null = null\n\n/** Installs the lookup. Called once, from `@aglyn/tenant-data-admin`. */\nexport function setLinkReputationLookup(lookup: LinkReputationLookup | null): void {\n  installedLookup = lookup\n}\n\n/** The installed lookup, or null. */\nexport function getLinkReputationLookup(): LinkReputationLookup | null {\n  return installedLookup\n}\n\n/** Test seam: forget any installed lookup. */\nexport function resetLinkReputationLookupForTests(): void {\n  installedLookup = null\n}\n\nconst HOST_PATTERN = /^[a-z0-9](?:[a-z0-9-]{0,62}[a-z0-9])?(?:\\.[a-z0-9](?:[a-z0-9-]{0,62}[a-z0-9])?)+$/\n\n/** A host as the lookup takes it, or null for anything that is not one. */\nexport function normalizeReputationHost(host: string): string | null {\n  const value = String(host ?? '')\n    .trim()\n    .toLowerCase()\n    .replace(/\\.+$/, '')\n  return value.length <= 253 && HOST_PATTERN.test(value) ? value : null\n}\n\n/**\n * The hosts worth asking about: foreign ones. Not the workspace's own, not\n * a domain under one of `excludeDomains` (the platform's own), not a listed\n * brand's own domain and not a common social, maps or review link — those\n * are judged by the screen's rules, and asking about them would only spend\n * lookups. Deduplicated, in the order found, at most\n * {@link MAX_REPUTATION_HOSTS_PER_LOOKUP} unless `max` says otherwise.\n */\nexport function foreignHostsForReputation(\n  hosts: readonly string[],\n  options: {\n    ownDomains?: readonly (string | null | undefined)[]\n    excludeDomains?: readonly (string | null | undefined)[]\n    max?: number\n  } = {},\n): string[] {\n  const under = (list: readonly (string | null | undefined)[] | undefined) =>\n    (list ?? [])\n      .map((domain) => normalizeReputationHost(String(domain ?? '')))\n      .filter((domain): domain is string => Boolean(domain))\n  const own = [...under(options.ownDomains), ...under(options.excludeDomains)]\n  const max = options.max ?? MAX_REPUTATION_HOSTS_PER_LOOKUP\n  const out: string[] = []\n  for (const raw of hosts) {\n    if (out.length >= max) break\n    const host = normalizeReputationHost(raw)\n    if (!host || out.includes(host)) continue\n    if (own.some((domain) => host === domain || host.endsWith(`.${domain}`))) continue\n    const registrable = registrableDomain(host)\n    if (isAnyOfficialBrandDomain(registrable) || isCommonLinkDomain(registrable)) continue\n    out.push(host)\n  }\n  return out\n}\n\n/** A listing, as the screen's signal. */\nexport function webRiskSignals(hits: readonly HostReputationHit[]): PhishingScreenSignal[] {\n  return hits.map((hit) => ({\n    code: 'web-risk-link' as const,\n    host: hit.host,\n    threats: [...new Set(hit.threats)].sort(),\n  }))\n}\n\n/**\n * Look the foreign hosts up through the installed lookup and answer the\n * signals they carry. Nothing installed, nothing foreign, an error: no\n * signals. Never throws.\n */\nexport async function linkReputationSignals(\n  hosts: readonly string[],\n  options: {\n    ownDomains?: readonly (string | null | undefined)[]\n    excludeDomains?: readonly (string | null | undefined)[]\n    deadlineMs?: number\n  } = {},\n): Promise<PhishingScreenSignal[]> {\n  const lookup = installedLookup\n  if (!lookup) return []\n  const foreign = foreignHostsForReputation(hosts, options)\n  if (!foreign.length) return []\n  try {\n    const answer = await lookup(foreign, { deadlineMs: options.deadlineMs })\n    return webRiskSignals(answer.hits.filter((hit) => foreign.includes(hit.host)))\n  } catch (error) {\n    console.warn('[link-reputation] the lookup failed — no signal', error)\n    return []\n  }\n}\n"],"names":["isAnyOfficialBrandDomain","isCommonLinkDomain","registrableDomain","LINK_THREAT_TYPES","MAX_REPUTATION_HOSTS_PER_LOOKUP","installedLookup","setLinkReputationLookup","lookup","getLinkReputationLookup","resetLinkReputationLookupForTests","HOST_PATTERN","normalizeReputationHost","host","value","String","trim","toLowerCase","replace","length","test","foreignHostsForReputation","hosts","options","under","list","map","domain","filter","Boolean","own","ownDomains","excludeDomains","max","out","raw","includes","some","endsWith","registrable","push","webRiskSignals","hits","hit","code","threats","Set","sort","linkReputationSignals","foreign","answer","deadlineMs","error","console","warn"],"mappings":"AAAA;;;;;;;;;;;;;;;CAeC,GAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;2CA6B2C,GAG3C,SACEA,wBAAwB,EACxBC,kBAAkB,EAClBC,iBAAiB,QACZ,gCAA4B;AAKnC,OAAO,MAAMC,oBAA+C;IAC1D;IACA;IACA;CACD,CAAA;AA4BD,2EAA2E,GAC3E,OAAO,MAAMC,kCAAkC,GAAE;AAEjD,IAAIC,kBAA+C;AAEnD,uEAAuE,GACvE,OAAO,SAASC,wBAAwBC,MAAmC;IACzEF,kBAAkBE;AACpB;AAEA,mCAAmC,GACnC,OAAO,SAASC;IACd,OAAOH;AACT;AAEA,4CAA4C,GAC5C,OAAO,SAASI;IACdJ,kBAAkB;AACpB;AAEA,MAAMK,eAAe;AAErB,yEAAyE,GACzE,OAAO,SAASC,wBAAwBC,IAAY;IAClD,MAAMC,QAAQC,OAAOF,eAAAA,OAAQ,IAC1BG,IAAI,GACJC,WAAW,GACXC,OAAO,CAAC,QAAQ;IACnB,OAAOJ,MAAMK,MAAM,IAAI,OAAOR,aAAaS,IAAI,CAACN,SAASA,QAAQ;AACnE;AAEA;;;;;;;CAOC,GACD,OAAO,SAASO,0BACdC,KAAwB,EACxBC,UAII,CAAC,CAAC;QAOMA;IALZ,MAAMC,QAAQ,CAACC,OACb,CAACA,eAAAA,OAAQ,EAAE,EACRC,GAAG,CAAC,CAACC,SAAWf,wBAAwBG,OAAOY,iBAAAA,SAAU,MACzDC,MAAM,CAAC,CAACD,SAA6BE,QAAQF;IAClD,MAAMG,MAAM;WAAIN,MAAMD,QAAQQ,UAAU;WAAMP,MAAMD,QAAQS,cAAc;KAAE;IAC5E,MAAMC,OAAMV,eAAAA,QAAQU,GAAG,YAAXV,eAAelB;IAC3B,MAAM6B,MAAgB,EAAE;IACxB,KAAK,MAAMC,OAAOb,MAAO;QACvB,IAAIY,IAAIf,MAAM,IAAIc,KAAK;QACvB,MAAMpB,OAAOD,wBAAwBuB;QACrC,IAAI,CAACtB,QAAQqB,IAAIE,QAAQ,CAACvB,OAAO;QACjC,IAAIiB,IAAIO,IAAI,CAAC,CAACV,SAAWd,SAASc,UAAUd,KAAKyB,QAAQ,CAAC,CAAC,CAAC,EAAEX,QAAQ,IAAI;QAC1E,MAAMY,cAAcpC,kBAAkBU;QACtC,IAAIZ,yBAAyBsC,gBAAgBrC,mBAAmBqC,cAAc;QAC9EL,IAAIM,IAAI,CAAC3B;IACX;IACA,OAAOqB;AACT;AAEA,uCAAuC,GACvC,OAAO,SAASO,eAAeC,IAAkC;IAC/D,OAAOA,KAAKhB,GAAG,CAAC,CAACiB,MAAS,CAAA;YACxBC,MAAM;YACN/B,MAAM8B,IAAI9B,IAAI;YACdgC,SAAS;mBAAI,IAAIC,IAAIH,IAAIE,OAAO;aAAE,CAACE,IAAI;QACzC,CAAA;AACF;AAEA;;;;CAIC,GACD,OAAO,eAAeC,sBACpB1B,KAAwB,EACxBC,UAII,CAAC,CAAC;IAEN,MAAMf,SAASF;IACf,IAAI,CAACE,QAAQ,OAAO,EAAE;IACtB,MAAMyC,UAAU5B,0BAA0BC,OAAOC;IACjD,IAAI,CAAC0B,QAAQ9B,MAAM,EAAE,OAAO,EAAE;IAC9B,IAAI;QACF,MAAM+B,SAAS,MAAM1C,OAAOyC,SAAS;YAAEE,YAAY5B,QAAQ4B,UAAU;QAAC;QACtE,OAAOV,eAAeS,OAAOR,IAAI,CAACd,MAAM,CAAC,CAACe,MAAQM,QAAQb,QAAQ,CAACO,IAAI9B,IAAI;IAC7E,EAAE,OAAOuC,OAAO;QACdC,QAAQC,IAAI,CAAC,mDAAmDF;QAChE,OAAO,EAAE;IACX;AACF"}
+{"version":3,"sources":["../../../../../../../libs/shared/util/email/src/lib/link-reputation.ts"],"sourcesContent":["/**\n * @license\n * Copyright 2026 Aglyn LLC\n *\n * Licensed under the Apache License, Version 2.0 (the \"License\");\n * you may not use this file except in compliance with the License.\n * You may obtain a copy of the License at\n *\n *   http://www.apache.org/licenses/LICENSE-2.0\n *\n * Unless required by applicable law or agreed to in writing, software\n * distributed under the License is distributed on an \"AS IS\" BASIS,\n * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n * See the License for the specific language governing permissions and\n * limitations under the License.\n */\n\n/*==========================================\n * LINK REPUTATION (AGL-3451): what the phishing screen asks about a host\n * nobody's rules can judge.\n *\n * The screen's own rules catch a page or a message that LOOKS like a lure —\n * a brand's lookalike host, a password field, a brand beside a sign-in. The\n * 2026-10-01 harvesters linked to `temps-juenes.com` and\n * `conservascaorvi.com`, which wear no brand at all, so nothing held them.\n * A host already known to be bad anywhere is caught here instead: every\n * FOREIGN host a page, an email or a redirect points at is looked up against\n * a reputation list (Google Web Risk), and a listed one is a\n * `web-risk-link` signal — STRONG, so it holds for every workspace.\n *\n * Split like the send seam's gate: the pure half lives here (which hosts are\n * foreign, how a listing becomes a signal), and the lookup — a network call\n * and a cache — is installed by `@aglyn/tenant-data-admin`\n * (`web-risk.ts`) through {@link setLinkReputationLookup}. Nothing installed\n * means nothing looked up.\n *\n * ## Not evidence when it fails\n *\n * A lookup that errors or times out says nothing about the host. It never\n * holds by itself; the lookup reports the host as unknown and logs a warning.\n *\n * ## The host, and — only when switched on — the path\n *\n * By default (`platformSettings/webRisk.lookupMode: 'host'`) the lookup sees\n * `https://<host>/` and nothing else. A host is enough to catch a domain\n * that exists to harvest, and it is the least a third party can be told.\n *\n * A phishing kit as often sits at a path on a compromised site that is\n * otherwise legitimate, which a host lookup reads as clean. In `'url'` mode\n * (AGL-3459) each link's address is looked up too, as\n * {@link normalizeReputationLink} reduces it: scheme, host and path. Never\n * the query string or the fragment, which is where a recipient's address\n * and a session's token live; never a `user:password@`; and the path stops\n * before any segment that carries an email address or an unrendered merge\n * tag. Sending paths makes Google a recipient for a new purpose, so the\n * switch is flipped only once the Subprocessors page names it.\n *\n * ## The budget\n *\n * One page, message or redirect asks about at most\n * {@link MAX_REPUTATION_HOSTS_PER_LOOKUP} hosts and, in `'url'` mode, at most\n * {@link MAX_REPUTATION_URLS_PER_LOOKUP} addresses, deduplicated and taken a\n * host at a time ({@link pickReputationUrls}), so a long list of links to\n * one site cannot crowd out the one link to another.\n *=========================================*/\n\nimport type { PhishingScreenSignal } from './outbound-phishing-screen'\nimport {\n  isAnyOfficialBrandDomain,\n  isCommonLinkDomain,\n  registrableDomain,\n} from './outbound-phishing-screen'\n\n/** The threat lists a host is looked up against. */\nexport type LinkThreatType = 'SOCIAL_ENGINEERING' | 'MALWARE' | 'UNWANTED_SOFTWARE'\n\nexport const LINK_THREAT_TYPES: readonly LinkThreatType[] = [\n  'SOCIAL_ENGINEERING',\n  'MALWARE',\n  'UNWANTED_SOFTWARE',\n]\n\n/**\n * A host the list names, or one address on it, and what it names it for.\n * `url` is set when the listing is the address's, not the host's.\n */\nexport interface HostReputationHit {\n  host: string\n  threats: LinkThreatType[]\n  /** The address the list names (scheme, host and path), when it is one page on a clean host. */\n  url?: string\n}\n\n/** What one lookup answered, host by host and address by address. */\nexport interface LinkReputationAnswer {\n  /** Hosts, and addresses, the list names. */\n  hits: HostReputationHit[]\n  /** Hosts and addresses the list does not name. */\n  clean: string[]\n  /** Hosts and addresses nobody could answer for: an error, a timeout, a switched-off lookup. */\n  unknown: string[]\n}\n\nexport interface LinkReputationLookupOptions {\n  /** How long the caller will wait, overall. */\n  deadlineMs?: number\n  /**\n   * The links' addresses, as {@link normalizeReputationLink} reduced them —\n   * looked up only in `'url'` mode, and only on the hosts asked about.\n   */\n  urls?: readonly string[]\n}\n\nexport type LinkReputationLookup = (\n  hosts: readonly string[],\n  options?: LinkReputationLookupOptions,\n) => Promise<LinkReputationAnswer>\n\n/** The most foreign hosts one page, message or redirect is looked up for. */\nexport const MAX_REPUTATION_HOSTS_PER_LOOKUP = 20\n/**\n * The most link addresses (path included) one page, message or redirect is\n * looked up for in `'url'` mode, beside its hosts (AGL-3459). With the hosts,\n * at most forty calls, all inside the same per-page deadline.\n */\nexport const MAX_REPUTATION_URLS_PER_LOOKUP = 20\n/**\n * The longest path an address keeps. Longer is cut back to the last `/`\n * within it, so what is looked up is still a path the link really starts\n * with — Web Risk lists a kit by its path prefix.\n */\nexport const MAX_REPUTATION_URL_PATH_LENGTH = 512\n\nlet installedLookup: LinkReputationLookup | null = null\n\n/** Installs the lookup. Called once, from `@aglyn/tenant-data-admin`. */\nexport function setLinkReputationLookup(lookup: LinkReputationLookup | null): void {\n  installedLookup = lookup\n}\n\n/** The installed lookup, or null. */\nexport function getLinkReputationLookup(): LinkReputationLookup | null {\n  return installedLookup\n}\n\n/** Test seam: forget any installed lookup. */\nexport function resetLinkReputationLookupForTests(): void {\n  installedLookup = null\n}\n\nconst HOST_PATTERN = /^[a-z0-9](?:[a-z0-9-]{0,62}[a-z0-9])?(?:\\.[a-z0-9](?:[a-z0-9-]{0,62}[a-z0-9])?)+$/\n\n/** A host as the lookup takes it, or null for anything that is not one. */\nexport function normalizeReputationHost(host: string): string | null {\n  const value = String(host ?? '')\n    .trim()\n    .toLowerCase()\n    .replace(/\\.+$/, '')\n  return value.length <= 253 && HOST_PATTERN.test(value) ? value : null\n}\n\n/** A link reduced to what may be looked up: its host, and its address without query or fragment. */\nexport interface ReputationLink {\n  host: string\n  /** `scheme://host[:port]/path` — never a query string, a fragment or a `user:password@`. */\n  url: string\n}\n\n/** A path segment that carries an email address, or a merge tag nobody has filled in. */\nconst PRIVATE_SEGMENT = /@|%40|\\{|%7b|\\}|%7d/i\n\n/**\n * A link reduced to what Web Risk may be asked about it (AGL-3459), or null\n * for anything that is not an `http(s)` link to a real host.\n *\n * - `www.…`, `//…` and a bare host are read as `https://`;\n * - the query string, the fragment and any `user:password@` are dropped —\n *   that is where a recipient's address and a session's token live;\n * - the host is lowercased, a default port dropped, and `.` and `..`\n *   segments resolved (WHATWG URL rules);\n * - the path stops before the first segment carrying an email address or an\n *   unrendered merge tag, and is cut back to a `/` within\n *   {@link MAX_REPUTATION_URL_PATH_LENGTH}.\n */\nexport function normalizeReputationLink(link: string): ReputationLink | null {\n  let text = String(link ?? '').trim()\n  if (!text) return null\n  if (text.startsWith('//')) text = `https:${text}`\n  else if (/^www\\./i.test(text)) text = `https://${text}`\n  else if (/^[a-z][a-z0-9+.-]*:(?!\\d)/i.test(text)) {\n    if (!/^https?:\\/\\//i.test(text)) return null\n  } else text = `https://${text}`\n  let parsed: URL\n  try {\n    parsed = new URL(text)\n  } catch {\n    return null\n  }\n  if (parsed.protocol !== 'https:' && parsed.protocol !== 'http:') return null\n  const host = normalizeReputationHost(parsed.hostname)\n  if (!host) return null\n  const kept: string[] = []\n  for (const segment of parsed.pathname.split('/')) {\n    if (PRIVATE_SEGMENT.test(segment)) {\n      kept.push('')\n      break\n    }\n    kept.push(segment)\n  }\n  let path = kept.join('/') || '/'\n  if (path.length > MAX_REPUTATION_URL_PATH_LENGTH) {\n    path = path.slice(0, path.lastIndexOf('/', MAX_REPUTATION_URL_PATH_LENGTH - 1) + 1)\n  }\n  const port = parsed.port ? `:${parsed.port}` : ''\n  return { host, url: `${parsed.protocol}//${host}${port}${path}` }\n}\n\n/** Is this address a host's front door — nothing a host lookup did not already ask? */\nexport function isRootReputationUrl(url: string): boolean {\n  return /^https?:\\/\\/[^/]+\\/$/.test(url)\n}\n\n/**\n * The addresses worth asking about beside `hosts`: those on one of them\n * whose path is more than `/`, each once, taken a host at a time — every\n * host's first address, then every host's second — up to `max`\n * ({@link MAX_REPUTATION_URLS_PER_LOOKUP}). A link that does not normalize\n * is skipped.\n */\nexport function pickReputationUrls(\n  links: readonly string[],\n  hosts: readonly string[],\n  max: number = MAX_REPUTATION_URLS_PER_LOOKUP,\n): string[] {\n  const byHost = new Map<string, string[]>(hosts.map((host) => [host, []]))\n  const seen = new Set<string>()\n  for (const raw of links) {\n    const link = normalizeReputationLink(raw)\n    if (!link || seen.has(link.url) || isRootReputationUrl(link.url)) continue\n    const list = byHost.get(link.host)\n    if (!list) continue\n    seen.add(link.url)\n    list.push(link.url)\n  }\n  const picked: string[] = []\n  for (let round = 0; picked.length < max; round += 1) {\n    const before = picked.length\n    for (const list of byHost.values()) {\n      if (round < list.length && picked.length < max) picked.push(list[round])\n    }\n    if (picked.length === before) break\n  }\n  return picked\n}\n\n/** The foreign hosts a set of links points at, and the addresses on them worth asking about. */\nexport interface ForeignReputationLinks {\n  hosts: string[]\n  urls: string[]\n}\n\n/**\n * {@link foreignHostsForReputation} over links — hosts or whole links — and\n * the addresses on those hosts ({@link pickReputationUrls}). The addresses\n * are only ever looked up in `'url'` mode; the lookup decides.\n */\nexport function foreignLinksForReputation(\n  links: readonly string[],\n  options: {\n    ownDomains?: readonly (string | null | undefined)[]\n    excludeDomains?: readonly (string | null | undefined)[]\n    max?: number\n    maxUrls?: number\n  } = {},\n): ForeignReputationLinks {\n  const parsed = links\n    .map((link) => normalizeReputationLink(link))\n    .filter((link): link is ReputationLink => Boolean(link))\n  const hosts = foreignHostsForReputation(\n    parsed.map((link) => link.host),\n    options,\n  )\n  return { hosts, urls: pickReputationUrls(parsed.map((link) => link.url), hosts, options.maxUrls) }\n}\n\n/**\n * The hosts worth asking about: foreign ones. Not the workspace's own, not\n * a domain under one of `excludeDomains` (the platform's own), not a listed\n * brand's own domain and not a common social, maps or review link — those\n * are judged by the screen's rules, and asking about them would only spend\n * lookups. Deduplicated, in the order found, at most\n * {@link MAX_REPUTATION_HOSTS_PER_LOOKUP} unless `max` says otherwise.\n */\nexport function foreignHostsForReputation(\n  hosts: readonly string[],\n  options: {\n    ownDomains?: readonly (string | null | undefined)[]\n    excludeDomains?: readonly (string | null | undefined)[]\n    max?: number\n  } = {},\n): string[] {\n  const under = (list: readonly (string | null | undefined)[] | undefined) =>\n    (list ?? [])\n      .map((domain) => normalizeReputationHost(String(domain ?? '')))\n      .filter((domain): domain is string => Boolean(domain))\n  const own = [...under(options.ownDomains), ...under(options.excludeDomains)]\n  const max = options.max ?? MAX_REPUTATION_HOSTS_PER_LOOKUP\n  const out: string[] = []\n  for (const raw of hosts) {\n    if (out.length >= max) break\n    const host = normalizeReputationHost(raw)\n    if (!host || out.includes(host)) continue\n    if (own.some((domain) => host === domain || host.endsWith(`.${domain}`))) continue\n    const registrable = registrableDomain(host)\n    if (isAnyOfficialBrandDomain(registrable) || isCommonLinkDomain(registrable)) continue\n    out.push(host)\n  }\n  return out\n}\n\n/** A listing, as the screen's signal. */\nexport function webRiskSignals(hits: readonly HostReputationHit[]): PhishingScreenSignal[] {\n  return hits.map((hit) => ({\n    code: 'web-risk-link' as const,\n    host: hit.host,\n    threats: [...new Set(hit.threats)].sort(),\n    ...(hit.url ? { url: hit.url } : {}),\n  }))\n}\n\n/**\n * Look the foreign links up through the installed lookup and answer the\n * signals they carry. `links` are hosts or whole links (`linkUrlsIn`): the\n * hosts are asked about, and their addresses too when the lookup is in\n * `'url'` mode. Nothing installed, nothing foreign, an error: no signals.\n * Never throws.\n */\nexport async function linkReputationSignals(\n  links: readonly string[],\n  options: {\n    ownDomains?: readonly (string | null | undefined)[]\n    excludeDomains?: readonly (string | null | undefined)[]\n    deadlineMs?: number\n  } = {},\n): Promise<PhishingScreenSignal[]> {\n  const lookup = installedLookup\n  if (!lookup) return []\n  const foreign = foreignLinksForReputation(links, options)\n  if (!foreign.hosts.length) return []\n  try {\n    const answer = await lookup(foreign.hosts, { deadlineMs: options.deadlineMs, urls: foreign.urls })\n    return webRiskSignals(\n      answer.hits.filter(\n        (hit) => foreign.hosts.includes(hit.host) && (!hit.url || foreign.urls.includes(hit.url)),\n      ),\n    )\n  } catch (error) {\n    console.warn('[link-reputation] the lookup failed — no signal', error)\n    return []\n  }\n}\n"],"names":["isAnyOfficialBrandDomain","isCommonLinkDomain","registrableDomain","LINK_THREAT_TYPES","MAX_REPUTATION_HOSTS_PER_LOOKUP","MAX_REPUTATION_URLS_PER_LOOKUP","MAX_REPUTATION_URL_PATH_LENGTH","installedLookup","setLinkReputationLookup","lookup","getLinkReputationLookup","resetLinkReputationLookupForTests","HOST_PATTERN","normalizeReputationHost","host","value","String","trim","toLowerCase","replace","length","test","PRIVATE_SEGMENT","normalizeReputationLink","link","text","startsWith","parsed","URL","protocol","hostname","kept","segment","pathname","split","push","path","join","slice","lastIndexOf","port","url","isRootReputationUrl","pickReputationUrls","links","hosts","max","byHost","Map","map","seen","Set","raw","has","list","get","add","picked","round","before","values","foreignLinksForReputation","options","filter","Boolean","foreignHostsForReputation","urls","maxUrls","under","domain","own","ownDomains","excludeDomains","out","includes","some","endsWith","registrable","webRiskSignals","hits","hit","code","threats","sort","linkReputationSignals","foreign","answer","deadlineMs","error","console","warn"],"mappings":"AAAA;;;;;;;;;;;;;;;CAeC,GAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;2CA+C2C;AAG3C,SACEA,wBAAwB,EACxBC,kBAAkB,EAClBC,iBAAiB,QACZ,gCAA4B;AAKnC,OAAO,MAAMC,oBAA+C;IAC1D;IACA;IACA;CACD,CAAA;AAsCD,2EAA2E,GAC3E,OAAO,MAAMC,kCAAkC,GAAE;AACjD;;;;CAIC,GACD,OAAO,MAAMC,iCAAiC,GAAE;AAChD;;;;CAIC,GACD,OAAO,MAAMC,iCAAiC,IAAG;AAEjD,IAAIC,kBAA+C;AAEnD,uEAAuE,GACvE,OAAO,SAASC,wBAAwBC,MAAmC;IACzEF,kBAAkBE;AACpB;AAEA,mCAAmC,GACnC,OAAO,SAASC;IACd,OAAOH;AACT;AAEA,4CAA4C,GAC5C,OAAO,SAASI;IACdJ,kBAAkB;AACpB;AAEA,MAAMK,eAAe;AAErB,yEAAyE,GACzE,OAAO,SAASC,wBAAwBC,IAAY;IAClD,MAAMC,QAAQC,OAAOF,eAAAA,OAAQ,IAC1BG,IAAI,GACJC,WAAW,GACXC,OAAO,CAAC,QAAQ;IACnB,OAAOJ,MAAMK,MAAM,IAAI,OAAOR,aAAaS,IAAI,CAACN,SAASA,QAAQ;AACnE;AASA,uFAAuF,GACvF,MAAMO,kBAAkB;AAExB;;;;;;;;;;;;CAYC,GACD,OAAO,SAASC,wBAAwBC,IAAY;IAClD,IAAIC,OAAOT,OAAOQ,eAAAA,OAAQ,IAAIP,IAAI;IAClC,IAAI,CAACQ,MAAM,OAAO;IAClB,IAAIA,KAAKC,UAAU,CAAC,OAAOD,OAAO,CAAC,MAAM,EAAEA,MAAM;SAC5C,IAAI,UAAUJ,IAAI,CAACI,OAAOA,OAAO,CAAC,QAAQ,EAAEA,MAAM;SAClD,IAAI,6BAA6BJ,IAAI,CAACI,OAAO;QAChD,IAAI,CAAC,gBAAgBJ,IAAI,CAACI,OAAO,OAAO;IAC1C,OAAOA,OAAO,CAAC,QAAQ,EAAEA,MAAM;IAC/B,IAAIE;IACJ,IAAI;QACFA,SAAS,IAAIC,IAAIH;IACnB,EAAE,eAAM;QACN,OAAO;IACT;IACA,IAAIE,OAAOE,QAAQ,KAAK,YAAYF,OAAOE,QAAQ,KAAK,SAAS,OAAO;IACxE,MAAMf,OAAOD,wBAAwBc,OAAOG,QAAQ;IACpD,IAAI,CAAChB,MAAM,OAAO;IAClB,MAAMiB,OAAiB,EAAE;IACzB,KAAK,MAAMC,WAAWL,OAAOM,QAAQ,CAACC,KAAK,CAAC,KAAM;QAChD,IAAIZ,gBAAgBD,IAAI,CAACW,UAAU;YACjCD,KAAKI,IAAI,CAAC;YACV;QACF;QACAJ,KAAKI,IAAI,CAACH;IACZ;IACA,IAAII,OAAOL,KAAKM,IAAI,CAAC,QAAQ;IAC7B,IAAID,KAAKhB,MAAM,GAAGd,gCAAgC;QAChD8B,OAAOA,KAAKE,KAAK,CAAC,GAAGF,KAAKG,WAAW,CAAC,KAAKjC,iCAAiC,KAAK;IACnF;IACA,MAAMkC,OAAOb,OAAOa,IAAI,GAAG,CAAC,CAAC,EAAEb,OAAOa,IAAI,EAAE,GAAG;IAC/C,OAAO;QAAE1B;QAAM2B,KAAK,GAAGd,OAAOE,QAAQ,CAAC,EAAE,EAAEf,OAAO0B,OAAOJ,MAAM;IAAC;AAClE;AAEA,qFAAqF,GACrF,OAAO,SAASM,oBAAoBD,GAAW;IAC7C,OAAO,uBAAuBpB,IAAI,CAACoB;AACrC;AAEA;;;;;;CAMC,GACD,OAAO,SAASE,mBACdC,KAAwB,EACxBC,KAAwB,EACxBC,MAAczC,8BAA8B;IAE5C,MAAM0C,SAAS,IAAIC,IAAsBH,MAAMI,GAAG,CAAC,CAACnC,OAAS;YAACA;YAAM,EAAE;SAAC;IACvE,MAAMoC,OAAO,IAAIC;IACjB,KAAK,MAAMC,OAAOR,MAAO;QACvB,MAAMpB,OAAOD,wBAAwB6B;QACrC,IAAI,CAAC5B,QAAQ0B,KAAKG,GAAG,CAAC7B,KAAKiB,GAAG,KAAKC,oBAAoBlB,KAAKiB,GAAG,GAAG;QAClE,MAAMa,OAAOP,OAAOQ,GAAG,CAAC/B,KAAKV,IAAI;QACjC,IAAI,CAACwC,MAAM;QACXJ,KAAKM,GAAG,CAAChC,KAAKiB,GAAG;QACjBa,KAAKnB,IAAI,CAACX,KAAKiB,GAAG;IACpB;IACA,MAAMgB,SAAmB,EAAE;IAC3B,IAAK,IAAIC,QAAQ,GAAGD,OAAOrC,MAAM,GAAG0B,KAAKY,SAAS,EAAG;QACnD,MAAMC,SAASF,OAAOrC,MAAM;QAC5B,KAAK,MAAMkC,QAAQP,OAAOa,MAAM,GAAI;YAClC,IAAIF,QAAQJ,KAAKlC,MAAM,IAAIqC,OAAOrC,MAAM,GAAG0B,KAAKW,OAAOtB,IAAI,CAACmB,IAAI,CAACI,MAAM;QACzE;QACA,IAAID,OAAOrC,MAAM,KAAKuC,QAAQ;IAChC;IACA,OAAOF;AACT;AAQA;;;;CAIC,GACD,OAAO,SAASI,0BACdjB,KAAwB,EACxBkB,UAKI,CAAC,CAAC;IAEN,MAAMnC,SAASiB,MACZK,GAAG,CAAC,CAACzB,OAASD,wBAAwBC,OACtCuC,MAAM,CAAC,CAACvC,OAAiCwC,QAAQxC;IACpD,MAAMqB,QAAQoB,0BACZtC,OAAOsB,GAAG,CAAC,CAACzB,OAASA,KAAKV,IAAI,GAC9BgD;IAEF,OAAO;QAAEjB;QAAOqB,MAAMvB,mBAAmBhB,OAAOsB,GAAG,CAAC,CAACzB,OAASA,KAAKiB,GAAG,GAAGI,OAAOiB,QAAQK,OAAO;IAAE;AACnG;AAEA;;;;;;;CAOC,GACD,OAAO,SAASF,0BACdpB,KAAwB,EACxBiB,UAII,CAAC,CAAC;QAOMA;IALZ,MAAMM,QAAQ,CAACd,OACb,CAACA,eAAAA,OAAQ,EAAE,EACRL,GAAG,CAAC,CAACoB,SAAWxD,wBAAwBG,OAAOqD,iBAAAA,SAAU,MACzDN,MAAM,CAAC,CAACM,SAA6BL,QAAQK;IAClD,MAAMC,MAAM;WAAIF,MAAMN,QAAQS,UAAU;WAAMH,MAAMN,QAAQU,cAAc;KAAE;IAC5E,MAAM1B,OAAMgB,eAAAA,QAAQhB,GAAG,YAAXgB,eAAe1D;IAC3B,MAAMqE,MAAgB,EAAE;IACxB,KAAK,MAAMrB,OAAOP,MAAO;QACvB,IAAI4B,IAAIrD,MAAM,IAAI0B,KAAK;QACvB,MAAMhC,OAAOD,wBAAwBuC;QACrC,IAAI,CAACtC,QAAQ2D,IAAIC,QAAQ,CAAC5D,OAAO;QACjC,IAAIwD,IAAIK,IAAI,CAAC,CAACN,SAAWvD,SAASuD,UAAUvD,KAAK8D,QAAQ,CAAC,CAAC,CAAC,EAAEP,QAAQ,IAAI;QAC1E,MAAMQ,cAAc3E,kBAAkBY;QACtC,IAAId,yBAAyB6E,gBAAgB5E,mBAAmB4E,cAAc;QAC9EJ,IAAItC,IAAI,CAACrB;IACX;IACA,OAAO2D;AACT;AAEA,uCAAuC,GACvC,OAAO,SAASK,eAAeC,IAAkC;IAC/D,OAAOA,KAAK9B,GAAG,CAAC,CAAC+B,MAAS;YACxBC,MAAM;YACNnE,MAAMkE,IAAIlE,IAAI;YACdoE,SAAS;mBAAI,IAAI/B,IAAI6B,IAAIE,OAAO;aAAE,CAACC,IAAI;WACnCH,IAAIvC,GAAG,GAAG;YAAEA,KAAKuC,IAAIvC,GAAG;QAAC,IAAI,CAAC;AAEtC;AAEA;;;;;;CAMC,GACD,OAAO,eAAe2C,sBACpBxC,KAAwB,EACxBkB,UAII,CAAC,CAAC;IAEN,MAAMrD,SAASF;IACf,IAAI,CAACE,QAAQ,OAAO,EAAE;IACtB,MAAM4E,UAAUxB,0BAA0BjB,OAAOkB;IACjD,IAAI,CAACuB,QAAQxC,KAAK,CAACzB,MAAM,EAAE,OAAO,EAAE;IACpC,IAAI;QACF,MAAMkE,SAAS,MAAM7E,OAAO4E,QAAQxC,KAAK,EAAE;YAAE0C,YAAYzB,QAAQyB,UAAU;YAAErB,MAAMmB,QAAQnB,IAAI;QAAC;QAChG,OAAOY,eACLQ,OAAOP,IAAI,CAAChB,MAAM,CAChB,CAACiB,MAAQK,QAAQxC,KAAK,CAAC6B,QAAQ,CAACM,IAAIlE,IAAI,KAAM,CAAA,CAACkE,IAAIvC,GAAG,IAAI4C,QAAQnB,IAAI,CAACQ,QAAQ,CAACM,IAAIvC,GAAG,CAAA;IAG7F,EAAE,OAAO+C,OAAO;QACdC,QAAQC,IAAI,CAAC,mDAAmDF;QAChE,OAAO,EAAE;IACX;AACF"}