@aglyn/plugins-ai 1.0.0-beta.288 → 1.0.0-beta.290

package/src/lib/site-import/site-import-html.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../../../../../../libs/plugins/ai/src/lib/site-import/site-import-html.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 * WHAT A CRAWLED PAGE SAYS (AGL-3974, parent AGL-3973).\n *\n * \"Bring my site\" reads a person's existing pages for their words, their\n * structure and their pictures. No server HTML parser is a dependency of any\n * lib, and a page's text in order is a small question, so this is a\n * tokenizer: tags and text runs, with the few elements that matter tracked\n * and everything else read through.\n *\n * What never survives, before a single tag is walked:\n * - scripts, styles, iframes, embeds, objects, SVG, canvas, media players,\n *   templates and comments, content and all;\n * - any form with a card, bank or payment field, whole;\n * - and a page with a password field is not read at all: it is refused and\n *   reported (`loginForm`), because a sign-in page is exactly what a\n *   phishing clone wants and nothing a site import needs.\n *\n * The output is DATA. Text from a page reaches a model only fenced as\n * untrusted (`fenceUntrustedText`), never as instructions.\n *=========================================*/\n\n/** One run of a page's content, in the order the page draws it. */\nexport type SiteImportBlock =\n  | { kind: 'heading'; level: 1 | 2 | 3 | 4 | 5 | 6; text: string; region: SiteImportRegion }\n  | { kind: 'paragraph' | 'list-item' | 'quote'; text: string; region: SiteImportRegion }\n  | { kind: 'image'; url: string; alt: string; region: SiteImportRegion }\n\n/** Where on the page a block sat: the header and footer carry business details. */\nexport type SiteImportRegion = 'header' | 'main' | 'footer'\n\nexport interface SiteImportImage {\n  /** Absolute https (or http) address. */\n  url: string\n  alt: string\n  /** On the site's own origin; otherwise a CDN or a third party. The importer decides. */\n  sameSite: boolean\n}\n\nexport interface SiteImportNavLink {\n  text: string\n  url: string\n}\n\n/** A form that was kept: what it asks, never what was typed. */\nexport interface SiteImportForm {\n  fields: { type: string; name: string }[]\n}\n\nexport interface SiteImportExtract {\n  title: string\n  description: string\n  lang: string | null\n  canonical: string | null\n  blocks: SiteImportBlock[]\n  images: SiteImportImage[]\n  navLinks: SiteImportNavLink[]\n  /** Same-site page addresses, fragment-free, in first-seen order. */\n  links: string[]\n  forms: { kept: SiteImportForm[]; removed: number }\n  /** True when the page has a password field; nothing else is extracted. */\n  loginForm: boolean\n}\n\n/** Ceilings that keep one odd page from becoming the whole inventory. */\nexport const SITE_IMPORT_PAGE_LIMITS = {\n  blocks: 400,\n  blockChars: 2_000,\n  images: 100,\n  links: 300,\n  navLinks: 60,\n  forms: 10,\n  fieldsPerForm: 30,\n} as const\n\n/** Elements removed with everything inside them. */\nconst REMOVED_WITH_CONTENT = [\n  'script',\n  'style',\n  'noscript',\n  'template',\n  'iframe',\n  'frame',\n  'frameset',\n  'object',\n  'applet',\n  'svg',\n  'math',\n  'canvas',\n  'video',\n  'audio',\n]\n\n/** Form controls, removed with their content once the forms are summarized. */\nconst FORM_CONTROLS = ['textarea', 'select', 'button']\n\n/** Void elements removed wherever they appear. */\nconst REMOVED_VOID = /<(?:embed|param|source|track|link|base)\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*>/gi\n\nconst PASSWORD_INPUT = /<input\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*\\btype\\s*=\\s*[\"']?password\\b/i\n\n/**\n * A field that takes money or a bank detail: an autocomplete hint for a\n * card, or a name or id that says card, CVC, IBAN, routing or payment.\n */\nconst PAYMENT_FIELD =\n  /<(?:input|select)\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*?(?:\\bautocomplete\\s*=\\s*[\"']?\\s*cc-|\\b(?:name|id|data-stripe|data-braintree-name)\\s*=\\s*[\"']?[^\"'\\s>]*(?:card|cc[-_]?(?:num|number|exp|csc|cvv|cvc)|cvv|cvc|csc|iban|routing|account[-_]?number|payment|billing[-_]?card))/i\n\nconst BLOCK_TAGS: Record<string, 'paragraph' | 'list-item' | 'quote'> = {\n  p: 'paragraph',\n  li: 'list-item',\n  dt: 'paragraph',\n  dd: 'paragraph',\n  td: 'paragraph',\n  th: 'paragraph',\n  address: 'paragraph',\n  pre: 'paragraph',\n  figcaption: 'paragraph',\n  caption: 'paragraph',\n  label: 'paragraph',\n  blockquote: 'quote',\n}\n\n/** Tags that end a run of loose text without being a block themselves. */\nconst BREAK_TAGS = new Set([\n  'div',\n  'section',\n  'article',\n  'main',\n  'aside',\n  'header',\n  'footer',\n  'nav',\n  'ul',\n  'ol',\n  'dl',\n  'table',\n  'tr',\n  'tbody',\n  'thead',\n  'figure',\n  'form',\n  'hr',\n  'br',\n  'body',\n])\n\nconst NAMED_ENTITIES: Record<string, string> = {\n  amp: '&',\n  lt: '<',\n  gt: '>',\n  quot: '\"',\n  apos: \"'\",\n  nbsp: ' ',\n  rsquo: '’',\n  lsquo: '‘',\n  rdquo: '”',\n  ldquo: '“',\n  mdash: '—',\n  ndash: '–',\n  hellip: '…',\n  copy: '©',\n  reg: '®',\n  trade: '™',\n  middot: '·',\n  bull: '•',\n  eacute: 'é',\n  egrave: 'è',\n  aacute: 'á',\n  ntilde: 'ñ',\n  uuml: 'ü',\n  ouml: 'ö',\n  auml: 'ä',\n  deg: '°',\n  times: '×',\n}\n\n/** Character references to characters; an unknown name stays as written. */\nexport function decodeHtmlEntities(text: string): string {\n  return text.replace(/&(#x[0-9a-f]+|#\\d+|[a-z][a-z0-9]*);/gi, (whole, body: string) => {\n    if (body[0] === '#') {\n      const code =\n        body[1] === 'x' || body[1] === 'X' ? parseInt(body.slice(2), 16) : parseInt(body.slice(1), 10)\n      if (!Number.isFinite(code) || code <= 0 || code > 0x10ffff) return ''\n      if (code < 32 && code !== 9 && code !== 10 && code !== 13) return ''\n      return String.fromCodePoint(code)\n    }\n    const named = NAMED_ENTITIES[body.toLowerCase()]\n    return named ?? whole\n  })\n}\n\nconst ATTRIBUTE = /([^\\s=/>\"']+)(?:\\s*=\\s*(?:\"([^\"]*)\"|'([^']*)'|([^\\s>\"']+)))?/g\n\n/** A tag's attributes, names lowercased, values decoded. */\nexport function parseAttributes(source: string): Record<string, string> {\n  const attributes: Record<string, string> = {}\n  for (const match of source.matchAll(ATTRIBUTE)) {\n    const name = match[1].toLowerCase()\n    if (name in attributes) continue\n    attributes[name] = decodeHtmlEntities(match[2] ?? match[3] ?? match[4] ?? '')\n  }\n  return attributes\n}\n\nconst TAG = /<(\\/?)([a-zA-Z][\\w:-]*)((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/g\n\n/** Every `<meta>` tag's attributes. */\nfunction metaTags(html: string): Record<string, string>[] {\n  const out: Record<string, string>[] = []\n  for (const match of html.matchAll(/<meta\\b((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/gi)) {\n    out.push(parseAttributes(match[1]))\n  }\n  return out\n}\n\n/** The `content` of every `<meta name=\"…\">` with this name. */\nexport function readMetaValues(html: string, name: string): string[] {\n  const wanted = name.toLowerCase()\n  return metaTags(html)\n    .filter((attributes) => (attributes['name'] ?? '').toLowerCase() === wanted)\n    .map((attributes) => (attributes['content'] ?? '').trim())\n    .filter(Boolean)\n}\n\nfunction collapse(text: string): string {\n  return text.replace(/\\s+/g, ' ').trim()\n}\n\nfunction clip(text: string, max: number): string {\n  return text.length > max ? `${text.slice(0, max - 1).trimEnd()}…` : text\n}\n\nfunction removeElements(html: string, tags: readonly string[]): string {\n  let out = html\n  for (const tag of tags) {\n    const paired = new RegExp(`<${tag}\\\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*>[\\\\s\\\\S]*?<\\\\/${tag}\\\\s*>`, 'gi')\n    out = out.replace(paired, ' ')\n    // One left open runs to the end of the document, as a browser reads it.\n    const unclosed = new RegExp(`<${tag}\\\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*>[\\\\s\\\\S]*$`, 'i')\n    out = out.replace(unclosed, ' ')\n  }\n  return out\n}\n\n/** Forms with a payment field go whole; the rest are summarized, then go. */\nfunction takeForms(html: string): { html: string; kept: SiteImportForm[]; removed: number } {\n  const kept: SiteImportForm[] = []\n  let removed = 0\n  const out = html.replace(/<form\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*>[\\s\\S]*?(?:<\\/form\\s*>|$)/gi, (form) => {\n    if (PAYMENT_FIELD.test(form)) {\n      removed += 1\n      return ' '\n    }\n    if (kept.length < SITE_IMPORT_PAGE_LIMITS.forms) {\n      const fields: SiteImportForm['fields'] = []\n      for (const match of form.matchAll(/<(input|textarea|select)\\b((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/gi)) {\n        const attributes = parseAttributes(match[2])\n        const type = match[1].toLowerCase() === 'input' ? (attributes['type'] || 'text').toLowerCase() : match[1].toLowerCase()\n        if (type === 'hidden' || type === 'submit' || type === 'button' || type === 'image') continue\n        fields.push({ type, name: clip(attributes['name'] || attributes['id'] || '', 80) })\n        if (fields.length >= SITE_IMPORT_PAGE_LIMITS.fieldsPerForm) break\n      }\n      kept.push({ fields })\n    }\n    return ' '\n  })\n  return { html: out, kept, removed }\n}\n\n/** `raw` against the page, as an absolute http(s) address without a fragment. */\nexport function resolveSiteUrl(raw: string | undefined, base: string): URL | null {\n  const value = String(raw ?? '').trim()\n  if (!value || /^(?:javascript|data|mailto|tel|sms|blob|about):/i.test(value)) return null\n  try {\n    const url = new URL(value, base)\n    if (url.protocol !== 'https:' && url.protocol !== 'http:') return null\n    url.hash = ''\n    return url\n  } catch {\n    return null\n  }\n}\n\n/** The first address a `srcset` offers. */\nfunction firstSrcset(srcset: string | undefined): string | undefined {\n  const first = String(srcset ?? '').split(',')[0]?.trim().split(/\\s+/)[0]\n  return first || undefined\n}\n\n/**\n * The words, structure, pictures and links of one page.\n *\n * `isSameSite(url)` says whether an address is on the site being imported;\n * only those become `links`, and images are marked by it.\n */\nexport function extractSiteImportPage(\n  html: string,\n  pageUrl: string,\n  isSameSite: (url: URL) => boolean,\n): SiteImportExtract {\n  const empty: SiteImportExtract = {\n    title: '',\n    description: '',\n    lang: null,\n    canonical: null,\n    blocks: [],\n    images: [],\n    navLinks: [],\n    links: [],\n    forms: { kept: [], removed: 0 },\n    loginForm: false,\n  }\n  let source = String(html ?? '')\n    .replace(/<!--[\\s\\S]*?(?:-->|$)/g, ' ')\n    .replace(/<!\\[CDATA\\[[\\s\\S]*?\\]\\]>/g, ' ')\n  if (PASSWORD_INPUT.test(source)) return { ...empty, loginForm: true }\n\n  source = removeElements(source, REMOVED_WITH_CONTENT).replace(REMOVED_VOID, ' ')\n  // Forms next, so a field list is read before the controls go.\n  const forms = takeForms(source)\n  source = removeElements(forms.html, FORM_CONTROLS)\n\n  const metas = metaTags(source)\n  const metaContent = (key: string, attribute: 'name' | 'property') =>\n    collapse(metas.find((m) => (m[attribute] ?? '').toLowerCase() === key)?.['content'] ?? '')\n  const titleMatch = /<title\\b[^>]*>([\\s\\S]*?)<\\/title\\s*>/i.exec(source)\n  const title = clip(\n    collapse(decodeHtmlEntities(titleMatch?.[1] ?? '')) || metaContent('og:title', 'property'),\n    300,\n  )\n  const description = clip(\n    metaContent('description', 'name') || metaContent('og:description', 'property'),\n    1_000,\n  )\n  const htmlTag = /<html\\b((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/i.exec(source)\n  const lang = htmlTag ? parseAttributes(htmlTag[1])['lang']?.trim() || null : null\n  let canonical: string | null = null\n  for (const match of String(html ?? '').matchAll(/<link\\b((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/gi)) {\n    const attributes = parseAttributes(match[1])\n    if ((attributes['rel'] ?? '').toLowerCase().split(/\\s+/).includes('canonical')) {\n      canonical = resolveSiteUrl(attributes['href'], pageUrl)?.toString() ?? null\n      break\n    }\n  }\n\n  const bodyStart = source.search(/<body\\b/i)\n  const body = bodyStart >= 0 ? source.slice(bodyStart) : source.replace(/<head\\b[\\s\\S]*?<\\/head\\s*>/i, ' ')\n  const hasNav = /<nav\\b/i.test(body)\n\n  const blocks: SiteImportBlock[] = []\n  const images: SiteImportImage[] = []\n  const imageSeen = new Set<string>()\n  const navLinks: SiteImportNavLink[] = []\n  const links: string[] = []\n  const linkSeen = new Set<string>()\n\n  let navDepth = 0\n  let headerDepth = 0\n  let footerDepth = 0\n  let current: { kind: 'heading' | 'paragraph' | 'list-item' | 'quote'; level?: number; tag: string } | null = null\n  let buffer = ''\n  let anchor: { url: URL; text: string } | null = null\n\n  const region = (): SiteImportRegion =>\n    footerDepth > 0 ? 'footer' : headerDepth > 0 ? 'header' : 'main'\n  // A site without a <nav> keeps its menu in the header; its links are the nav.\n  const inNav = () => navDepth > 0 || (!hasNav && headerDepth > 0)\n\n  const flush = () => {\n    const text = clip(collapse(decodeHtmlEntities(buffer)), SITE_IMPORT_PAGE_LIMITS.blockChars)\n    buffer = ''\n    const kind = current?.kind ?? 'paragraph'\n    const level = current?.level\n    current = null\n    if (!text || navDepth > 0 || blocks.length >= SITE_IMPORT_PAGE_LIMITS.blocks) return\n    const last = blocks[blocks.length - 1]\n    if (last && last.kind !== 'image' && last.text === text) return\n    if (kind === 'heading') {\n      blocks.push({ kind, level: (level ?? 2) as 1 | 2 | 3 | 4 | 5 | 6, text, region: region() })\n    } else {\n      blocks.push({ kind, text, region: region() })\n    }\n  }\n\n  for (const match of body.matchAll(new RegExp(`${TAG.source}|([^<]+)|<`, 'g'))) {\n    const text = match[4]\n    if (text !== undefined || match[2] === undefined) {\n      const run = text ?? '<'\n      buffer += run\n      if (anchor) anchor.text += run\n      continue\n    }\n    const closing = match[1] === '/'\n    const tag = match[2].toLowerCase()\n    const attributes = closing ? {} : parseAttributes(match[3] ?? '')\n\n    if (tag === 'a') {\n      if (closing) {\n        if (anchor) {\n          const label = clip(collapse(decodeHtmlEntities(anchor.text)), 120)\n          if (inNav() && label && navLinks.length < SITE_IMPORT_PAGE_LIMITS.navLinks) {\n            navLinks.push({ text: label, url: anchor.url.toString() })\n          }\n          anchor = null\n        }\n        continue\n      }\n      const url = resolveSiteUrl(attributes['href'], pageUrl)\n      anchor = url ? { url, text: '' } : null\n      if (url && isSameSite(url) && links.length < SITE_IMPORT_PAGE_LIMITS.links) {\n        const key = url.toString()\n        if (!linkSeen.has(key)) {\n          linkSeen.add(key)\n          links.push(key)\n        }\n      }\n      continue\n    }\n\n    if (tag === 'img' && !closing) {\n      const url = resolveSiteUrl(\n        attributes['src'] || attributes['data-src'] || firstSrcset(attributes['srcset'] || attributes['data-srcset']),\n        pageUrl,\n      )\n      if (url && images.length < SITE_IMPORT_PAGE_LIMITS.images) {\n        const key = url.toString()\n        const alt = clip(collapse(attributes['alt'] ?? ''), 300)\n        if (!imageSeen.has(key)) {\n          imageSeen.add(key)\n          images.push({ url: key, alt, sameSite: isSameSite(url) })\n          if (navDepth === 0) {\n            flush()\n            if (blocks.length < SITE_IMPORT_PAGE_LIMITS.blocks) {\n              blocks.push({ kind: 'image', url: key, alt, region: region() })\n            }\n          }\n        }\n      }\n      continue\n    }\n\n    if (tag === 'nav' || tag === 'header' || tag === 'footer') {\n      flush()\n      const step = closing ? -1 : 1\n      if (tag === 'nav') navDepth = Math.max(0, navDepth + step)\n      if (tag === 'header') headerDepth = Math.max(0, headerDepth + step)\n      if (tag === 'footer') footerDepth = Math.max(0, footerDepth + step)\n      continue\n    }\n\n    const heading = /^h([1-6])$/.exec(tag)\n    if (heading || BLOCK_TAGS[tag]) {\n      if (closing) {\n        if (current?.tag === tag) flush()\n        continue\n      }\n      flush()\n      current = heading\n        ? { kind: 'heading', level: Number(heading[1]), tag }\n        : { kind: BLOCK_TAGS[tag], tag }\n      continue\n    }\n\n    if (BREAK_TAGS.has(tag)) {\n      if (tag === 'br' && current) {\n        buffer += ' '\n        continue\n      }\n      flush()\n    }\n  }\n  flush()\n\n  return {\n    title,\n    description,\n    lang,\n    canonical,\n    blocks,\n    images,\n    navLinks,\n    links,\n    forms: { kept: forms.kept, removed: forms.removed },\n    loginForm: false,\n  }\n}\n"],"names":["SITE_IMPORT_PAGE_LIMITS","blocks","blockChars","images","links","navLinks","forms","fieldsPerForm","REMOVED_WITH_CONTENT","FORM_CONTROLS","REMOVED_VOID","PASSWORD_INPUT","PAYMENT_FIELD","BLOCK_TAGS","p","li","dt","dd","td","th","address","pre","figcaption","caption","label","blockquote","BREAK_TAGS","Set","NAMED_ENTITIES","amp","lt","gt","quot","apos","nbsp","rsquo","lsquo","rdquo","ldquo","mdash","ndash","hellip","copy","reg","trade","middot","bull","eacute","egrave","aacute","ntilde","uuml","ouml","auml","deg","times","decodeHtmlEntities","text","replace","whole","body","code","parseInt","slice","Number","isFinite","String","fromCodePoint","named","toLowerCase","ATTRIBUTE","parseAttributes","source","attributes","match","matchAll","name","TAG","metaTags","html","out","push","readMetaValues","wanted","filter","map","trim","Boolean","collapse","clip","max","length","trimEnd","removeElements","tags","tag","paired","RegExp","unclosed","takeForms","kept","removed","form","test","fields","type","resolveSiteUrl","raw","base","value","url","URL","protocol","hash","firstSrcset","srcset","first","split","undefined","extractSiteImportPage","pageUrl","isSameSite","empty","title","description","lang","canonical","loginForm","metas","metaContent","key","attribute","find","m","titleMatch","exec","htmlTag","includes","toString","bodyStart","search","hasNav","imageSeen","linkSeen","navDepth","headerDepth","footerDepth","current","buffer","anchor","region","inNav","flush","kind","level","last","run","closing","has","add","alt","sameSite","step","Math","heading"],"mappings":"AAAA;;;;;;;;;;;;;;;CAeC,GAED;;;;;;;;;;;;;;;;;;;2CAmB2C,GAE3C,iEAAiE;AA0CjE,uEAAuE,GACvE,OAAO,MAAMA,0BAA0B;IACrCC,QAAQ;IACRC,YAAY;IACZC,QAAQ;IACRC,OAAO;IACPC,UAAU;IACVC,OAAO;IACPC,eAAe;AACjB,EAAU;AAEV,kDAAkD,GAClD,MAAMC,uBAAuB;IAC3B;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;CACD;AAED,6EAA6E,GAC7E,MAAMC,gBAAgB;IAAC;IAAY;IAAU;CAAS;AAEtD,gDAAgD,GAChD,MAAMC,eAAe;AAErB,MAAMC,iBAAiB;AAEvB;;;CAGC,GACD,MAAMC,gBACJ;AAEF,MAAMC,aAAkE;IACtEC,GAAG;IACHC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,SAAS;IACTC,KAAK;IACLC,YAAY;IACZC,SAAS;IACTC,OAAO;IACPC,YAAY;AACd;AAEA,wEAAwE,GACxE,MAAMC,aAAa,IAAIC,IAAI;IACzB;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;CACD;AAED,MAAMC,iBAAyC;IAC7CC,KAAK;IACLC,IAAI;IACJC,IAAI;IACJC,MAAM;IACNC,MAAM;IACNC,MAAM;IACNC,OAAO;IACPC,OAAO;IACPC,OAAO;IACPC,OAAO;IACPC,OAAO;IACPC,OAAO;IACPC,QAAQ;IACRC,MAAM;IACNC,KAAK;IACLC,OAAO;IACPC,QAAQ;IACRC,MAAM;IACNC,QAAQ;IACRC,QAAQ;IACRC,QAAQ;IACRC,QAAQ;IACRC,MAAM;IACNC,MAAM;IACNC,MAAM;IACNC,KAAK;IACLC,OAAO;AACT;AAEA,0EAA0E,GAC1E,OAAO,SAASC,mBAAmBC,IAAY;IAC7C,OAAOA,KAAKC,OAAO,CAAC,yCAAyC,CAACC,OAAOC;QACnE,IAAIA,IAAI,CAAC,EAAE,KAAK,KAAK;YACnB,MAAMC,OACJD,IAAI,CAAC,EAAE,KAAK,OAAOA,IAAI,CAAC,EAAE,KAAK,MAAME,SAASF,KAAKG,KAAK,CAAC,IAAI,MAAMD,SAASF,KAAKG,KAAK,CAAC,IAAI;YAC7F,IAAI,CAACC,OAAOC,QAAQ,CAACJ,SAASA,QAAQ,KAAKA,OAAO,UAAU,OAAO;YACnE,IAAIA,OAAO,MAAMA,SAAS,KAAKA,SAAS,MAAMA,SAAS,IAAI,OAAO;YAClE,OAAOK,OAAOC,aAAa,CAACN;QAC9B;QACA,MAAMO,QAAQxC,cAAc,CAACgC,KAAKS,WAAW,GAAG;QAChD,OAAOD,gBAAAA,QAAST;IAClB;AACF;AAEA,MAAMW,YAAY;AAElB,0DAA0D,GAC1D,OAAO,SAASC,gBAAgBC,MAAc;IAC5C,MAAMC,aAAqC,CAAC;IAC5C,KAAK,MAAMC,SAASF,OAAOG,QAAQ,CAACL,WAAY;YAGRI,MAAAA,OAAAA;QAFtC,MAAME,OAAOF,KAAK,CAAC,EAAE,CAACL,WAAW;QACjC,IAAIO,QAAQH,YAAY;QACxBA,UAAU,CAACG,KAAK,GAAGpB,oBAAmBkB,QAAAA,SAAAA,UAAAA,KAAK,CAAC,EAAE,YAARA,UAAYA,KAAK,CAAC,EAAE,YAApBA,QAAwBA,KAAK,CAAC,EAAE,YAAhCA,OAAoC;IAC5E;IACA,OAAOD;AACT;AAEA,MAAMI,MAAM;AAEZ,qCAAqC,GACrC,SAASC,SAASC,IAAY;IAC5B,MAAMC,MAAgC,EAAE;IACxC,KAAK,MAAMN,SAASK,KAAKJ,QAAQ,CAAC,2CAA4C;QAC5EK,IAAIC,IAAI,CAACV,gBAAgBG,KAAK,CAAC,EAAE;IACnC;IACA,OAAOM;AACT;AAEA,6DAA6D,GAC7D,OAAO,SAASE,eAAeH,IAAY,EAAEH,IAAY;IACvD,MAAMO,SAASP,KAAKP,WAAW;IAC/B,OAAOS,SAASC,MACbK,MAAM,CAAC,CAACX;YAAgBA;eAAD,EAACA,mBAAAA,UAAU,CAAC,OAAO,YAAlBA,mBAAsB,IAAIJ,WAAW,OAAOc;OACpEE,GAAG,CAAC,CAACZ;YAAgBA;eAAD,EAACA,sBAAAA,UAAU,CAAC,UAAU,YAArBA,sBAAyB,IAAIa,IAAI;OACtDF,MAAM,CAACG;AACZ;AAEA,SAASC,SAAS/B,IAAY;IAC5B,OAAOA,KAAKC,OAAO,CAAC,QAAQ,KAAK4B,IAAI;AACvC;AAEA,SAASG,KAAKhC,IAAY,EAAEiC,GAAW;IACrC,OAAOjC,KAAKkC,MAAM,GAAGD,MAAM,GAAGjC,KAAKM,KAAK,CAAC,GAAG2B,MAAM,GAAGE,OAAO,GAAG,CAAC,CAAC,GAAGnC;AACtE;AAEA,SAASoC,eAAed,IAAY,EAAEe,IAAuB;IAC3D,IAAId,MAAMD;IACV,KAAK,MAAMgB,OAAOD,KAAM;QACtB,MAAME,SAAS,IAAIC,OAAO,CAAC,CAAC,EAAEF,IAAI,6CAA6C,EAAEA,IAAI,KAAK,CAAC,EAAE;QAC7Ff,MAAMA,IAAItB,OAAO,CAACsC,QAAQ;QAC1B,wEAAwE;QACxE,MAAME,WAAW,IAAID,OAAO,CAAC,CAAC,EAAEF,IAAI,yCAAyC,CAAC,EAAE;QAChFf,MAAMA,IAAItB,OAAO,CAACwC,UAAU;IAC9B;IACA,OAAOlB;AACT;AAEA,2EAA2E,GAC3E,SAASmB,UAAUpB,IAAY;IAC7B,MAAMqB,OAAyB,EAAE;IACjC,IAAIC,UAAU;IACd,MAAMrB,MAAMD,KAAKrB,OAAO,CAAC,kEAAkE,CAAC4C;QAC1F,IAAI1F,cAAc2F,IAAI,CAACD,OAAO;YAC5BD,WAAW;YACX,OAAO;QACT;QACA,IAAID,KAAKT,MAAM,GAAG3F,wBAAwBM,KAAK,EAAE;YAC/C,MAAMkG,SAAmC,EAAE;YAC3C,KAAK,MAAM9B,SAAS4B,KAAK3B,QAAQ,CAAC,8DAA+D;gBAC/F,MAAMF,aAAaF,gBAAgBG,KAAK,CAAC,EAAE;gBAC3C,MAAM+B,OAAO/B,KAAK,CAAC,EAAE,CAACL,WAAW,OAAO,UAAU,AAACI,CAAAA,UAAU,CAAC,OAAO,IAAI,MAAK,EAAGJ,WAAW,KAAKK,KAAK,CAAC,EAAE,CAACL,WAAW;gBACrH,IAAIoC,SAAS,YAAYA,SAAS,YAAYA,SAAS,YAAYA,SAAS,SAAS;gBACrFD,OAAOvB,IAAI,CAAC;oBAAEwB;oBAAM7B,MAAMa,KAAKhB,UAAU,CAAC,OAAO,IAAIA,UAAU,CAAC,KAAK,IAAI,IAAI;gBAAI;gBACjF,IAAI+B,OAAOb,MAAM,IAAI3F,wBAAwBO,aAAa,EAAE;YAC9D;YACA6F,KAAKnB,IAAI,CAAC;gBAAEuB;YAAO;QACrB;QACA,OAAO;IACT;IACA,OAAO;QAAEzB,MAAMC;QAAKoB;QAAMC;IAAQ;AACpC;AAEA,+EAA+E,GAC/E,OAAO,SAASK,eAAeC,GAAuB,EAAEC,IAAY;IAClE,MAAMC,QAAQ3C,OAAOyC,cAAAA,MAAO,IAAIrB,IAAI;IACpC,IAAI,CAACuB,SAAS,mDAAmDN,IAAI,CAACM,QAAQ,OAAO;IACrF,IAAI;QACF,MAAMC,MAAM,IAAIC,IAAIF,OAAOD;QAC3B,IAAIE,IAAIE,QAAQ,KAAK,YAAYF,IAAIE,QAAQ,KAAK,SAAS,OAAO;QAClEF,IAAIG,IAAI,GAAG;QACX,OAAOH;IACT,EAAE,eAAM;QACN,OAAO;IACT;AACF;AAEA,yCAAyC,GACzC,SAASI,YAAYC,MAA0B;QAC/BjD;IAAd,MAAMkD,SAAQlD,iBAAAA,OAAOiD,iBAAAA,SAAU,IAAIE,KAAK,CAAC,IAAI,CAAC,EAAE,qBAAlCnD,eAAoCoB,IAAI,GAAG+B,KAAK,CAAC,MAAM,CAAC,EAAE;IACxE,OAAOD,SAASE;AAClB;AAEA;;;;;CAKC,GACD,OAAO,SAASC,sBACdxC,IAAY,EACZyC,OAAe,EACfC,UAAiC;;QAqCVlD;IAnCvB,MAAMmD,QAA2B;QAC/BC,OAAO;QACPC,aAAa;QACbC,MAAM;QACNC,WAAW;QACX7H,QAAQ,EAAE;QACVE,QAAQ,EAAE;QACVE,UAAU,EAAE;QACZD,OAAO,EAAE;QACTE,OAAO;YAAE8F,MAAM,EAAE;YAAEC,SAAS;QAAE;QAC9B0B,WAAW;IACb;IACA,IAAIvD,SAASN,OAAOa,eAAAA,OAAQ,IACzBrB,OAAO,CAAC,0BAA0B,KAClCA,OAAO,CAAC,6BAA6B;IACxC,IAAI/C,eAAe4F,IAAI,CAAC/B,SAAS,OAAO,aAAKkD;QAAOK,WAAW;;IAE/DvD,SAASqB,eAAerB,QAAQhE,sBAAsBkD,OAAO,CAAChD,cAAc;IAC5E,8DAA8D;IAC9D,MAAMJ,QAAQ6F,UAAU3B;IACxBA,SAASqB,eAAevF,MAAMyE,IAAI,EAAEtE;IAEpC,MAAMuH,QAAQlD,SAASN;IACvB,MAAMyD,cAAc,CAACC,KAAaC;;YACvBH;eAATxC,kBAASwC,cAAAA,MAAMI,IAAI,CAAC,CAACC;gBAAOA;mBAAD,EAACA,eAAAA,CAAC,CAACF,UAAU,YAAZE,eAAgB,IAAIhE,WAAW,OAAO6D;+BAAzDF,WAA+D,CAAC,UAAU,mBAAI;;IACzF,MAAMM,aAAa,wCAAwCC,IAAI,CAAC/D;IAChE,MAAMmD,QAAQlC,KACZD,SAAShC,2BAAmB8E,8BAAAA,UAAY,CAAC,EAAE,mBAAI,QAAQL,YAAY,YAAY,aAC/E;IAEF,MAAML,cAAcnC,KAClBwC,YAAY,eAAe,WAAWA,YAAY,kBAAkB,aACpE;IAEF,MAAMO,UAAU,yCAAyCD,IAAI,CAAC/D;IAC9D,MAAMqD,OAAOW,UAAUjE,EAAAA,wBAAAA,gBAAgBiE,OAAO,CAAC,EAAE,CAAC,CAAC,OAAO,qBAAnCjE,sBAAqCe,IAAI,OAAM,OAAO;IAC7E,IAAIwC,YAA2B;IAC/B,KAAK,MAAMpD,SAASR,OAAOa,eAAAA,OAAQ,IAAIJ,QAAQ,CAAC,2CAA4C;YAErFF;QADL,MAAMA,aAAaF,gBAAgBG,KAAK,CAAC,EAAE;QAC3C,IAAI,EAACD,kBAAAA,UAAU,CAAC,MAAM,YAAjBA,kBAAqB,IAAIJ,WAAW,GAAGgD,KAAK,CAAC,OAAOoB,QAAQ,CAAC,cAAc;;gBAClE/B;YAAZoB,sBAAYpB,kBAAAA,eAAejC,UAAU,CAAC,OAAO,EAAE+C,6BAAnCd,gBAA6CgC,QAAQ,sBAAM;YACvE;QACF;IACF;IAEA,MAAMC,YAAYnE,OAAOoE,MAAM,CAAC;IAChC,MAAMhF,OAAO+E,aAAa,IAAInE,OAAOT,KAAK,CAAC4E,aAAanE,OAAOd,OAAO,CAAC,+BAA+B;IACtG,MAAMmF,SAAS,UAAUtC,IAAI,CAAC3C;IAE9B,MAAM3D,SAA4B,EAAE;IACpC,MAAME,SAA4B,EAAE;IACpC,MAAM2I,YAAY,IAAInH;IACtB,MAAMtB,WAAgC,EAAE;IACxC,MAAMD,QAAkB,EAAE;IAC1B,MAAM2I,WAAW,IAAIpH;IAErB,IAAIqH,WAAW;IACf,IAAIC,cAAc;IAClB,IAAIC,cAAc;IAClB,IAAIC,UAAyG;IAC7G,IAAIC,SAAS;IACb,IAAIC,SAA4C;IAEhD,MAAMC,SAAS,IACbJ,cAAc,IAAI,WAAWD,cAAc,IAAI,WAAW;IAC5D,8EAA8E;IAC9E,MAAMM,QAAQ,IAAMP,WAAW,KAAM,CAACH,UAAUI,cAAc;IAE9D,MAAMO,QAAQ;;QACZ,MAAM/F,OAAOgC,KAAKD,SAAShC,mBAAmB4F,UAAUpJ,wBAAwBE,UAAU;QAC1FkJ,SAAS;QACT,MAAMK,eAAON,2BAAAA,QAASM,IAAI,mBAAI;QAC9B,MAAMC,QAAQP,2BAAAA,QAASO,KAAK;QAC5BP,UAAU;QACV,IAAI,CAAC1F,QAAQuF,WAAW,KAAK/I,OAAO0F,MAAM,IAAI3F,wBAAwBC,MAAM,EAAE;QAC9E,MAAM0J,OAAO1J,MAAM,CAACA,OAAO0F,MAAM,GAAG,EAAE;QACtC,IAAIgE,QAAQA,KAAKF,IAAI,KAAK,WAAWE,KAAKlG,IAAI,KAAKA,MAAM;QACzD,IAAIgG,SAAS,WAAW;YACtBxJ,OAAOgF,IAAI,CAAC;gBAAEwE;gBAAMC,KAAK,EAAGA,gBAAAA,QAAS;gBAA6BjG;gBAAM6F,QAAQA;YAAS;QAC3F,OAAO;YACLrJ,OAAOgF,IAAI,CAAC;gBAAEwE;gBAAMhG;gBAAM6F,QAAQA;YAAS;QAC7C;IACF;IAEA,KAAK,MAAM5E,SAASd,KAAKe,QAAQ,CAAC,IAAIsB,OAAO,GAAGpB,IAAIL,MAAM,CAAC,UAAU,CAAC,EAAE,MAAO;YAU3BE;QATlD,MAAMjB,OAAOiB,KAAK,CAAC,EAAE;QACrB,IAAIjB,SAAS6D,aAAa5C,KAAK,CAAC,EAAE,KAAK4C,WAAW;YAChD,MAAMsC,MAAMnG,eAAAA,OAAQ;YACpB2F,UAAUQ;YACV,IAAIP,QAAQA,OAAO5F,IAAI,IAAImG;YAC3B;QACF;QACA,MAAMC,UAAUnF,KAAK,CAAC,EAAE,KAAK;QAC7B,MAAMqB,MAAMrB,KAAK,CAAC,EAAE,CAACL,WAAW;QAChC,MAAMI,aAAaoF,UAAU,CAAC,IAAItF,iBAAgBG,UAAAA,KAAK,CAAC,EAAE,YAARA,UAAY;QAE9D,IAAIqB,QAAQ,KAAK;YACf,IAAI8D,SAAS;gBACX,IAAIR,QAAQ;oBACV,MAAM7H,QAAQiE,KAAKD,SAAShC,mBAAmB6F,OAAO5F,IAAI,IAAI;oBAC9D,IAAI8F,WAAW/H,SAASnB,SAASsF,MAAM,GAAG3F,wBAAwBK,QAAQ,EAAE;wBAC1EA,SAAS4E,IAAI,CAAC;4BAAExB,MAAMjC;4BAAOsF,KAAKuC,OAAOvC,GAAG,CAAC4B,QAAQ;wBAAG;oBAC1D;oBACAW,SAAS;gBACX;gBACA;YACF;YACA,MAAMvC,MAAMJ,eAAejC,UAAU,CAAC,OAAO,EAAE+C;YAC/C6B,SAASvC,MAAM;gBAAEA;gBAAKrD,MAAM;YAAG,IAAI;YACnC,IAAIqD,OAAOW,WAAWX,QAAQ1G,MAAMuF,MAAM,GAAG3F,wBAAwBI,KAAK,EAAE;gBAC1E,MAAM8H,MAAMpB,IAAI4B,QAAQ;gBACxB,IAAI,CAACK,SAASe,GAAG,CAAC5B,MAAM;oBACtBa,SAASgB,GAAG,CAAC7B;oBACb9H,MAAM6E,IAAI,CAACiD;gBACb;YACF;YACA;QACF;QAEA,IAAInC,QAAQ,SAAS,CAAC8D,SAAS;YAC7B,MAAM/C,MAAMJ,eACVjC,UAAU,CAAC,MAAM,IAAIA,UAAU,CAAC,WAAW,IAAIyC,YAAYzC,UAAU,CAAC,SAAS,IAAIA,UAAU,CAAC,cAAc,GAC5G+C;YAEF,IAAIV,OAAO3G,OAAOwF,MAAM,GAAG3F,wBAAwBG,MAAM,EAAE;oBAE/BsE;gBAD1B,MAAMyD,MAAMpB,IAAI4B,QAAQ;gBACxB,MAAMsB,MAAMvE,KAAKD,UAASf,kBAAAA,UAAU,CAAC,MAAM,YAAjBA,kBAAqB,KAAK;gBACpD,IAAI,CAACqE,UAAUgB,GAAG,CAAC5B,MAAM;oBACvBY,UAAUiB,GAAG,CAAC7B;oBACd/H,OAAO8E,IAAI,CAAC;wBAAE6B,KAAKoB;wBAAK8B;wBAAKC,UAAUxC,WAAWX;oBAAK;oBACvD,IAAIkC,aAAa,GAAG;wBAClBQ;wBACA,IAAIvJ,OAAO0F,MAAM,GAAG3F,wBAAwBC,MAAM,EAAE;4BAClDA,OAAOgF,IAAI,CAAC;gCAAEwE,MAAM;gCAAS3C,KAAKoB;gCAAK8B;gCAAKV,QAAQA;4BAAS;wBAC/D;oBACF;gBACF;YACF;YACA;QACF;QAEA,IAAIvD,QAAQ,SAASA,QAAQ,YAAYA,QAAQ,UAAU;YACzDyD;YACA,MAAMU,OAAOL,UAAU,CAAC,IAAI;YAC5B,IAAI9D,QAAQ,OAAOiD,WAAWmB,KAAKzE,GAAG,CAAC,GAAGsD,WAAWkB;YACrD,IAAInE,QAAQ,UAAUkD,cAAckB,KAAKzE,GAAG,CAAC,GAAGuD,cAAciB;YAC9D,IAAInE,QAAQ,UAAUmD,cAAciB,KAAKzE,GAAG,CAAC,GAAGwD,cAAcgB;YAC9D;QACF;QAEA,MAAME,UAAU,aAAa7B,IAAI,CAACxC;QAClC,IAAIqE,WAAWvJ,UAAU,CAACkF,IAAI,EAAE;YAC9B,IAAI8D,SAAS;gBACX,IAAIV,CAAAA,2BAAAA,QAASpD,GAAG,MAAKA,KAAKyD;gBAC1B;YACF;YACAA;YACAL,UAAUiB,UACN;gBAAEX,MAAM;gBAAWC,OAAO1F,OAAOoG,OAAO,CAAC,EAAE;gBAAGrE;YAAI,IAClD;gBAAE0D,MAAM5I,UAAU,CAACkF,IAAI;gBAAEA;YAAI;YACjC;QACF;QAEA,IAAIrE,WAAWoI,GAAG,CAAC/D,MAAM;YACvB,IAAIA,QAAQ,QAAQoD,SAAS;gBAC3BC,UAAU;gBACV;YACF;YACAI;QACF;IACF;IACAA;IAEA,OAAO;QACL7B;QACAC;QACAC;QACAC;QACA7H;QACAE;QACAE;QACAD;QACAE,OAAO;YAAE8F,MAAM9F,MAAM8F,IAAI;YAAEC,SAAS/F,MAAM+F,OAAO;QAAC;QAClD0B,WAAW;IACb;AACF"}
+{"version":3,"sources":["../../../../../../../libs/plugins/ai/src/lib/site-import/site-import-html.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 * WHAT A CRAWLED PAGE SAYS (AGL-3974, parent AGL-3973).\n *\n * \"Bring my site\" reads a person's existing pages for their words, their\n * structure and their pictures. No server HTML parser is a dependency of any\n * lib, and a page's text in order is a small question, so this is a\n * tokenizer: tags and text runs, with the few elements that matter tracked\n * and everything else read through.\n *\n * What never survives, before a single tag is walked:\n * - scripts, styles, iframes, embeds, objects, SVG, canvas, media players,\n *   templates and comments, content and all;\n * - any form with a card, bank or payment field, whole;\n * - and a page with a password field is not read at all: it is refused and\n *   reported (`loginForm`), because a sign-in page is exactly what a\n *   phishing clone wants and nothing a site import needs.\n *\n * The output is DATA. Text from a page reaches a model only fenced as\n * untrusted (`fenceUntrustedText`), never as instructions.\n *\n * What a page is made of is kept as the page shows it (AGL-4083): words that\n * sit in tags side by side stay apart (\"Sign in\" and \"Start free\", never\n * \"Sign inStart free\"); a link is a block of its own, `mailto:` and `tel:`\n * ones too, so an address or a phone number survives; a menu entry is named\n * by its first line, not by its mega-menu description; a `<picture>`'s\n * source stands in for an `<img>` with no address; and the head's share card,\n * icons, site name and Organization structured data come along for the\n * header, the footer and the site's own facts.\n *=========================================*/\n\n/** One run of a page's content, in the order the page draws it. */\nexport type SiteImportBlock =\n  | { kind: 'heading'; level: 1 | 2 | 3 | 4 | 5 | 6; text: string; region: SiteImportRegion }\n  | { kind: 'paragraph' | 'list-item' | 'quote'; text: string; region: SiteImportRegion }\n  | { kind: 'image'; url: string; alt: string; region: SiteImportRegion }\n  /**\n   * A link the page shows outside its menus (AGL-4083): `href` is an absolute\n   * http(s) address, a `mailto:` or a `tel:`. `inline` when its words are\n   * already in the block before it (a link inside a sentence); a link on its\n   * own (a button, an email address in a card) is only this block.\n   */\n  | { kind: 'link'; text: string; href: string; region: SiteImportRegion; inline?: true }\n\n/** Where on the page a block sat: the header and footer carry business details. */\nexport type SiteImportRegion = 'header' | 'main' | 'footer'\n\nexport interface SiteImportImage {\n  /** Absolute https (or http) address. */\n  url: string\n  alt: string\n  /** On the site's own origin; otherwise a CDN or a third party. The importer decides. */\n  sameSite: boolean\n  /** Where on the page it sat, its menus included (AGL-4083): a header's picture is often the logo. */\n  region: SiteImportRegion\n}\n\n/** A `<link rel=\"icon\">` or `apple-touch-icon` of the page's head (AGL-4083). */\nexport interface SiteImportIcon {\n  url: string\n  rel: 'icon' | 'apple-touch-icon'\n  /** The larger side its `sizes` names, or `null` for none or `any`. */\n  size: number | null\n  /** Its declared type, as written. */\n  type: string | null\n}\n\n/**\n * The link to the home page in a page's header (AGL-4083): what its brand is\n * drawn with. `logo` is the first picture inside it; `text` its words, which\n * are the site's name when the header prints one.\n */\nexport interface SiteImportBrandMark {\n  logo: { url: string; alt: string } | null\n  text: string\n  region: SiteImportRegion\n}\n\n/**\n * The page's Organization (or LocalBusiness) structured data (AGL-4083), as\n * written: the site's own statement of its name, address, phone, email and\n * profiles. Every field is the empty string, or an empty list, when absent.\n */\nexport interface SiteImportOrganization {\n  name: string\n  description: string\n  url: string\n  logo: string\n  telephone: string\n  email: string\n  /** One line: street, town, region and postal code, country. */\n  address: string\n  sameAs: string[]\n}\n\nexport interface SiteImportNavLink {\n  text: string\n  url: string\n}\n\n/** A form that was kept: what it asks, never what was typed. */\nexport interface SiteImportForm {\n  fields: { type: string; name: string }[]\n}\n\nexport interface SiteImportExtract {\n  title: string\n  description: string\n  lang: string | null\n  canonical: string | null\n  /** The share card, `og:image` (or `twitter:image`), as an absolute address (AGL-4083). */\n  ogImage: string | null\n  /** The site's own name, `og:site_name` (or `application-name`); empty when absent. */\n  siteName: string\n  icons: SiteImportIcon[]\n  brand: SiteImportBrandMark | null\n  organization: SiteImportOrganization | null\n  blocks: SiteImportBlock[]\n  images: SiteImportImage[]\n  navLinks: SiteImportNavLink[]\n  /** Same-site page addresses, fragment-free, in first-seen order. */\n  links: string[]\n  forms: { kept: SiteImportForm[]; removed: number }\n  /** True when the page has a password field; nothing else is extracted. */\n  loginForm: boolean\n}\n\n/** Ceilings that keep one odd page from becoming the whole inventory. */\nexport const SITE_IMPORT_PAGE_LIMITS = {\n  /** A pricing page's plan table runs past 400 (AGL-4083); a wave's read keeps fewer, under its own budget. */\n  blocks: 2_000,\n  /** A legal page's paragraph runs past 2,000 characters. */\n  blockChars: 10_000,\n  images: 100,\n  links: 300,\n  /** A mega-menu's entries and the footer's columns together. */\n  navLinks: 150,\n  forms: 10,\n  fieldsPerForm: 30,\n  icons: 12,\n} as const\n\n/** Elements removed with everything inside them. */\nconst REMOVED_WITH_CONTENT = [\n  'script',\n  'style',\n  'noscript',\n  'template',\n  'iframe',\n  'frame',\n  'frameset',\n  'object',\n  'applet',\n  'svg',\n  'math',\n  'canvas',\n  'video',\n  'audio',\n]\n\n/** Form controls, removed with their content once the forms are summarized. */\nconst FORM_CONTROLS = ['textarea', 'select', 'button']\n\n/** Void elements removed wherever they appear. */\nconst REMOVED_VOID = /<(?:embed|param|source|track|link|base)\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*>/gi\n\nconst PASSWORD_INPUT = /<input\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*\\btype\\s*=\\s*[\"']?password\\b/i\n\n/**\n * A field that takes money or a bank detail: an autocomplete hint for a\n * card, or a name or id that says card, CVC, IBAN, routing or payment.\n */\nconst PAYMENT_FIELD =\n  /<(?:input|select)\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*?(?:\\bautocomplete\\s*=\\s*[\"']?\\s*cc-|\\b(?:name|id|data-stripe|data-braintree-name)\\s*=\\s*[\"']?[^\"'\\s>]*(?:card|cc[-_]?(?:num|number|exp|csc|cvv|cvc)|cvv|cvc|csc|iban|routing|account[-_]?number|payment|billing[-_]?card))/i\n\nconst BLOCK_TAGS: Record<string, 'paragraph' | 'list-item' | 'quote'> = {\n  p: 'paragraph',\n  li: 'list-item',\n  dt: 'paragraph',\n  dd: 'paragraph',\n  td: 'paragraph',\n  th: 'paragraph',\n  address: 'paragraph',\n  pre: 'paragraph',\n  figcaption: 'paragraph',\n  caption: 'paragraph',\n  label: 'paragraph',\n  blockquote: 'quote',\n}\n\n/** Tags that end a run of loose text without being a block themselves. */\nconst BREAK_TAGS = new Set([\n  'div',\n  'section',\n  'article',\n  'main',\n  'aside',\n  'header',\n  'footer',\n  'nav',\n  'ul',\n  'ol',\n  'dl',\n  'table',\n  'tr',\n  'tbody',\n  'thead',\n  'figure',\n  'form',\n  'hr',\n  'br',\n  'body',\n])\n\n/**\n * Tags a browser lays out apart from the words beside them (AGL-4083): two\n * links, or a link and a picture, side by side are two runs of words, never\n * one (\"Sign in\" and \"Start free\", not \"Sign inStart free\"). Every other tag\n * inside a block styles words where they are (a `<span>` or a `<strong>`\n * mid-sentence) and adds no space; block-level tags already end their block.\n */\nconst RUN_BOUNDARY = new Set(['a', 'img', 'button', 'input', 'select', 'textarea', 'picture', 'video', 'audio', 'iframe', 'object', 'embed', 'canvas', 'svg', 'meter', 'progress', 'output'])\n\n/** Where one run of words ends and the next begins, resolved when a block is written. */\nconst RUN_BREAK = '\\u0001'\n\n/**\n * The runs of a block joined as a reader sees them: a space between two\n * runs, except before closing punctuation or after an opening one, where\n * the page wrote none (\"<span>Austin</span>, Texas\" is \"Austin, Texas\").\n */\nconst RUN_BREAKS = new RegExp(`[\\\\s${RUN_BREAK}]*${RUN_BREAK}[\\\\s${RUN_BREAK}]*`, 'g')\n\nfunction joinRuns(text: string): string {\n  if (!text.includes(RUN_BREAK)) return text\n  return text.replace(RUN_BREAKS, (match: string, offset: number, whole: string) => {\n    if (/\\s/.test(match)) return ' '\n    const before = whole[offset - 1] ?? ''\n    const after = whole[offset + match.length] ?? ''\n    if (!before || !after) return ''\n    if (/[.,;:!?%)\\]}\\u2019\\u201d\\u00bb]/.test(after) || /[([{\\u2018\\u201c\\u00ab]/.test(before)) return ''\n    // A straight quote opens after a space or a bracket, and closes before one.\n    if (/[\"']/.test(before) && /^$|[\\s([{]/.test(whole[offset - 2] ?? '')) return ''\n    if (/[\"']/.test(after) && /^$|[\\s.,;:!?)\\]}]/.test(whole[offset + match.length + 1] ?? '')) return ''\n    return ' '\n  })\n}\n\nconst NAMED_ENTITIES: Record<string, string> = {\n  amp: '&',\n  lt: '<',\n  gt: '>',\n  quot: '\"',\n  apos: \"'\",\n  nbsp: ' ',\n  rsquo: '’',\n  lsquo: '‘',\n  rdquo: '”',\n  ldquo: '“',\n  mdash: '—',\n  ndash: '–',\n  hellip: '…',\n  copy: '©',\n  reg: '®',\n  trade: '™',\n  middot: '·',\n  bull: '•',\n  eacute: 'é',\n  egrave: 'è',\n  aacute: 'á',\n  ntilde: 'ñ',\n  uuml: 'ü',\n  ouml: 'ö',\n  auml: 'ä',\n  deg: '°',\n  times: '×',\n}\n\n/** Character references to characters; an unknown name stays as written. */\nexport function decodeHtmlEntities(text: string): string {\n  return text.replace(/&(#x[0-9a-f]+|#\\d+|[a-z][a-z0-9]*);/gi, (whole, body: string) => {\n    if (body[0] === '#') {\n      const code =\n        body[1] === 'x' || body[1] === 'X' ? parseInt(body.slice(2), 16) : parseInt(body.slice(1), 10)\n      if (!Number.isFinite(code) || code <= 0 || code > 0x10ffff) return ''\n      if (code < 32 && code !== 9 && code !== 10 && code !== 13) return ''\n      return String.fromCodePoint(code)\n    }\n    const named = NAMED_ENTITIES[body.toLowerCase()]\n    return named ?? whole\n  })\n}\n\nconst ATTRIBUTE = /([^\\s=/>\"']+)(?:\\s*=\\s*(?:\"([^\"]*)\"|'([^']*)'|([^\\s>\"']+)))?/g\n\n/** A tag's attributes, names lowercased, values decoded. */\nexport function parseAttributes(source: string): Record<string, string> {\n  const attributes: Record<string, string> = {}\n  for (const match of source.matchAll(ATTRIBUTE)) {\n    const name = match[1].toLowerCase()\n    if (name in attributes) continue\n    attributes[name] = decodeHtmlEntities(match[2] ?? match[3] ?? match[4] ?? '')\n  }\n  return attributes\n}\n\nconst TAG = /<(\\/?)([a-zA-Z][\\w:-]*)((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/g\n\n/** Every `<meta>` tag's attributes. */\nfunction metaTags(html: string): Record<string, string>[] {\n  const out: Record<string, string>[] = []\n  for (const match of html.matchAll(/<meta\\b((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/gi)) {\n    out.push(parseAttributes(match[1]))\n  }\n  return out\n}\n\n/** The `content` of every `<meta name=\"…\">` with this name. */\nexport function readMetaValues(html: string, name: string): string[] {\n  const wanted = name.toLowerCase()\n  return metaTags(html)\n    .filter((attributes) => (attributes['name'] ?? '').toLowerCase() === wanted)\n    .map((attributes) => (attributes['content'] ?? '').trim())\n    .filter(Boolean)\n}\n\nfunction collapse(text: string): string {\n  return joinRuns(text).replace(/\\s+/g, ' ').trim()\n}\n\nfunction clip(text: string, max: number): string {\n  return text.length > max ? `${text.slice(0, max - 1).trimEnd()}…` : text\n}\n\nfunction removeElements(html: string, tags: readonly string[]): string {\n  let out = html\n  for (const tag of tags) {\n    const paired = new RegExp(`<${tag}\\\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*>[\\\\s\\\\S]*?<\\\\/${tag}\\\\s*>`, 'gi')\n    out = out.replace(paired, ' ')\n    // One left open runs to the end of the document, as a browser reads it.\n    const unclosed = new RegExp(`<${tag}\\\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*>[\\\\s\\\\S]*$`, 'i')\n    out = out.replace(unclosed, ' ')\n  }\n  return out\n}\n\n/** Forms with a payment field go whole; the rest are summarized, then go. */\nfunction takeForms(html: string): { html: string; kept: SiteImportForm[]; removed: number } {\n  const kept: SiteImportForm[] = []\n  let removed = 0\n  const out = html.replace(/<form\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*>[\\s\\S]*?(?:<\\/form\\s*>|$)/gi, (form) => {\n    if (PAYMENT_FIELD.test(form)) {\n      removed += 1\n      return ' '\n    }\n    if (kept.length < SITE_IMPORT_PAGE_LIMITS.forms) {\n      const fields: SiteImportForm['fields'] = []\n      for (const match of form.matchAll(/<(input|textarea|select)\\b((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/gi)) {\n        const attributes = parseAttributes(match[2])\n        const type = match[1].toLowerCase() === 'input' ? (attributes['type'] || 'text').toLowerCase() : match[1].toLowerCase()\n        if (type === 'hidden' || type === 'submit' || type === 'button' || type === 'image') continue\n        // A field no visitor sees is a trap for bots or a widget's shadow, not a question (AGL-4083).\n        if ((attributes['aria-hidden'] ?? '').toLowerCase() === 'true' || attributes['tabindex'] === '-1') continue\n        fields.push({ type, name: clip(attributes['name'] || attributes['id'] || '', 80) })\n        if (fields.length >= SITE_IMPORT_PAGE_LIMITS.fieldsPerForm) break\n      }\n      kept.push({ fields })\n    }\n    return ' '\n  })\n  return { html: out, kept, removed }\n}\n\n/** `raw` against the page, as an absolute http(s) address without a fragment. */\nexport function resolveSiteUrl(raw: string | undefined, base: string): URL | null {\n  const value = String(raw ?? '').trim()\n  if (!value || /^(?:javascript|data|mailto|tel|sms|blob|about):/i.test(value)) return null\n  try {\n    const url = new URL(value, base)\n    if (url.protocol !== 'https:' && url.protocol !== 'http:') return null\n    url.hash = ''\n    return url\n  } catch {\n    return null\n  }\n}\n\n/** The first address a `srcset` offers. */\nfunction firstSrcset(srcset: string | undefined): string | undefined {\n  const first = String(srcset ?? '').split(',')[0]?.trim().split(/\\s+/)[0]\n  return first || undefined\n}\n\n/**\n * Where a link a page shows goes (AGL-4083): an absolute http(s) address\n * without its fragment, a `mailto:` with an address in it, or a `tel:` with\n * a number in it; `null` for anything else (a script, a jump on the page).\n */\nexport function resolveSiteLink(raw: string | undefined, base: string): string | null {\n  const value = String(raw ?? '').trim()\n  const mail = /^mailto:([^?\\s<>\"']+@[^?\\s<>\"']+)/i.exec(value)\n  if (mail) return `mailto:${decodeURIComponent(mail[1]).toLowerCase()}`\n  const phone = /^tel:([+\\d][\\d\\s().-]{2,30})$/i.exec(value)\n  if (phone) return `tel:${phone[1].replace(/[\\s().-]/g, '')}`\n  if (value.startsWith('#')) return null\n  return resolveSiteUrl(value, base)?.toString() ?? null\n}\n\n/**\n * A `<picture>` whose `<img>` has no address of its own (a placeholder, or\n * none at all) is read from its first `<source>` (AGL-4083); a `<picture>`\n * whose `<img>` has one keeps it. The sources go with the void elements.\n */\nfunction readPictures(html: string): string {\n  return html.replace(/<picture\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*>([\\s\\S]*?)<\\/picture\\s*>/gi, (whole, inner: string) => {\n    const img = /<img\\b((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/i.exec(inner)\n    const attributes = img ? parseAttributes(img[1]) : {}\n    const own = attributes['src'] || attributes['data-src'] || firstSrcset(attributes['srcset'] || attributes['data-srcset'])\n    if (own && !/^data:/i.test(own.trim())) return whole\n    let source: string | undefined\n    for (const match of inner.matchAll(/<source\\b((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/gi)) {\n      const sourceAttributes = parseAttributes(match[1])\n      source = firstSrcset(sourceAttributes['srcset'] || sourceAttributes['data-srcset'])\n      if (source) break\n    }\n    if (!source) return whole\n    const quote = (value: string) => value.replace(/&/g, '&amp;').replace(/\"/g, '&quot;').replace(/</g, '&lt;')\n    return ` <img src=\"${quote(source)}\" alt=\"${quote(attributes['alt'] ?? '')}\"> `\n  })\n}\n\n/** The head's icons, the largest first within each kind. */\nfunction readIcons(html: string, pageUrl: string): SiteImportIcon[] {\n  const icons: SiteImportIcon[] = []\n  for (const match of html.matchAll(/<link\\b((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/gi)) {\n    const attributes = parseAttributes(match[1])\n    const rels = (attributes['rel'] ?? '').toLowerCase().split(/\\s+/)\n    const rel = rels.includes('apple-touch-icon') || rels.includes('apple-touch-icon-precomposed')\n      ? 'apple-touch-icon'\n      : rels.includes('icon')\n        ? 'icon'\n        : null\n    if (!rel) continue\n    const url = resolveSiteUrl(attributes['href'], pageUrl)\n    if (!url) continue\n    const sizes: string[] = [...((attributes['sizes'] ?? '').match(/(\\d+)\\s*x\\s*(\\d+)/gi) ?? [])]\n    const size = sizes.reduce<number>((most, entry) => Math.max(most, ...entry.split(/x/i).map((side) => Number(side) || 0)), 0)\n    icons.push({ url: url.toString(), rel, size: size || null, type: attributes['type']?.trim() || null })\n    if (icons.length >= SITE_IMPORT_PAGE_LIMITS.icons) break\n  }\n  return icons.sort((a, b) => (a.rel === b.rel ? (b.size ?? 0) - (a.size ?? 0) : a.rel === 'apple-touch-icon' ? -1 : 1))\n}\n\nconst ORGANIZATION_TYPE = /(?:^|\\b)(?:Organization|Corporation|LocalBusiness|NGO|EducationalOrganization|MedicalOrganization|SportsOrganization|Store|Restaurant|ProfessionalService|HomeAndConstructionBusiness|HealthAndBeautyBusiness|FoodEstablishment|AutomotiveBusiness|LegalService|FinancialService|Dentist|Physician|LodgingBusiness|EntertainmentBusiness)$/\n\n/**\n * The first Organization (or kind of local business) the page's structured\n * data describes, read before scripts are thrown away. Malformed data, or\n * none, is `null`; nothing in it is ever run.\n */\nfunction readOrganization(html: string): SiteImportOrganization | null {\n  const nodes: Record<string, unknown>[] = []\n  const visit = (value: unknown, depth: number) => {\n    if (depth > 4 || nodes.length > 50 || !value || typeof value !== 'object') return\n    if (Array.isArray(value)) {\n      for (const entry of value) visit(entry, depth + 1)\n      return\n    }\n    const node = value as Record<string, unknown>\n    nodes.push(node)\n    if (node['@graph']) visit(node['@graph'], depth + 1)\n  }\n  for (const match of html.matchAll(/<script\\b[^>]*\\btype\\s*=\\s*[\"']?application\\/ld\\+json[\"']?[^>]*>([\\s\\S]*?)<\\/script\\s*>/gi)) {\n    try {\n      visit(JSON.parse(match[1]), 0)\n    } catch {\n      // Structured data that does not parse says nothing.\n    }\n  }\n  const typeOf = (node: Record<string, unknown>) => ([] as unknown[]).concat(node['@type'] ?? []).map(String)\n  const node = nodes.find((entry) => typeOf(entry).some((type) => ORGANIZATION_TYPE.test(type)))\n  if (!node) return null\n  const text = (value: unknown, max = 300): string => (typeof value === 'string' ? clip(collapse(value), max) : '')\n  const urlOf = (value: unknown): string => {\n    const raw = typeof value === 'string' ? value : value && typeof value === 'object' ? (value as Record<string, unknown>)['url'] : ''\n    return typeof raw === 'string' && /^https?:\\/\\//i.test(raw.trim()) ? raw.trim() : ''\n  }\n  const contact = ([] as unknown[]).concat(node['contactPoint'] ?? []).find((entry) => entry && typeof entry === 'object') as\n    | Record<string, unknown>\n    | undefined\n  const address = (() => {\n    const value = node['address']\n    if (typeof value === 'string') return text(value)\n    if (!value || typeof value !== 'object') return ''\n    const parts = value as Record<string, unknown>\n    const region = [text(parts['addressRegion']), text(parts['postalCode'])].filter(Boolean).join(' ')\n    const country = parts['addressCountry']\n    return [text(parts['streetAddress']), text(parts['addressLocality']), region, typeof country === 'string' ? text(country) : text((country as Record<string, unknown> | null)?.['name'])]\n      .filter(Boolean)\n      .join(', ')\n  })()\n  return {\n    name: text(node['name']),\n    description: text(node['description'], 1_000),\n    url: urlOf(node['url']),\n    logo: urlOf(node['logo']),\n    telephone: text(node['telephone']) || text(contact?.['telephone']),\n    email: (text(node['email']) || text(contact?.['email'])).replace(/^mailto:/i, ''),\n    address,\n    sameAs: ([] as unknown[])\n      .concat(node['sameAs'] ?? [])\n      .map((entry) => urlOf(entry))\n      .filter(Boolean)\n      .slice(0, 20),\n  }\n}\n\n/**\n * The words, structure, pictures and links of one page.\n *\n * `isSameSite(url)` says whether an address is on the site being imported;\n * only those become `links`, and images are marked by it.\n */\nexport function extractSiteImportPage(\n  html: string,\n  pageUrl: string,\n  isSameSite: (url: URL) => boolean,\n): SiteImportExtract {\n  const empty: SiteImportExtract = {\n    title: '',\n    description: '',\n    lang: null,\n    canonical: null,\n    ogImage: null,\n    siteName: '',\n    icons: [],\n    brand: null,\n    organization: null,\n    blocks: [],\n    images: [],\n    navLinks: [],\n    links: [],\n    forms: { kept: [], removed: 0 },\n    loginForm: false,\n  }\n  let source = String(html ?? '')\n    .replace(/<!--[\\s\\S]*?(?:-->|$)/g, ' ')\n    .replace(/<!\\[CDATA\\[[\\s\\S]*?\\]\\]>/g, ' ')\n  if (PASSWORD_INPUT.test(source)) return { ...empty, loginForm: true }\n\n  // Read before scripts go: the structured data is a script's body.\n  const organization = readOrganization(source)\n  const icons = readIcons(source, pageUrl)\n  source = removeElements(readPictures(source), REMOVED_WITH_CONTENT).replace(REMOVED_VOID, ' ')\n  // Forms next, so a field list is read before the controls go.\n  const forms = takeForms(source)\n  source = removeElements(forms.html, FORM_CONTROLS)\n\n  const metas = metaTags(source)\n  const metaContent = (key: string, attribute: 'name' | 'property') =>\n    collapse(metas.find((m) => (m[attribute] ?? '').toLowerCase() === key)?.['content'] ?? '')\n  const titleMatch = /<title\\b[^>]*>([\\s\\S]*?)<\\/title\\s*>/i.exec(source)\n  const title = clip(\n    collapse(decodeHtmlEntities(titleMatch?.[1] ?? '')) || metaContent('og:title', 'property'),\n    300,\n  )\n  const description = clip(\n    metaContent('description', 'name') || metaContent('og:description', 'property'),\n    1_000,\n  )\n  const ogImage =\n    resolveSiteUrl(metaContent('og:image', 'property') || metaContent('og:image:url', 'property') || metaContent('twitter:image', 'name'), pageUrl)?.toString() ??\n    null\n  const siteName = clip(metaContent('og:site_name', 'property') || metaContent('application-name', 'name'), 120)\n  const htmlTag = /<html\\b((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/i.exec(source)\n  const lang = htmlTag ? parseAttributes(htmlTag[1])['lang']?.trim() || null : null\n  let canonical: string | null = null\n  for (const match of String(html ?? '').matchAll(/<link\\b((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/gi)) {\n    const attributes = parseAttributes(match[1])\n    if ((attributes['rel'] ?? '').toLowerCase().split(/\\s+/).includes('canonical')) {\n      canonical = resolveSiteUrl(attributes['href'], pageUrl)?.toString() ?? null\n      break\n    }\n  }\n\n  const bodyStart = source.search(/<body\\b/i)\n  const body = bodyStart >= 0 ? source.slice(bodyStart) : source.replace(/<head\\b[\\s\\S]*?<\\/head\\s*>/i, ' ')\n  const hasNav = /<nav\\b/i.test(body)\n\n  const blocks: SiteImportBlock[] = []\n  const images: SiteImportImage[] = []\n  const imageSeen = new Set<string>()\n  const navLinks: SiteImportNavLink[] = []\n  const links: string[] = []\n  const linkSeen = new Set<string>()\n  let brand: SiteImportBrandMark | null = null\n\n  let navDepth = 0\n  let headerDepth = 0\n  let footerDepth = 0\n  let current: { kind: 'heading' | 'paragraph' | 'list-item' | 'quote'; level?: number; tag: string } | null = null\n  let buffer = ''\n  /** The link being read: its menu address (http(s) only), where it goes, its words and its first picture. */\n  let anchor: {\n    url: URL | null\n    href: string | null\n    text: string\n    /** Its words split where a block-level tag inside it began, for a menu entry's first line. */\n    lines: string[]\n    image: { url: string; alt: string } | null\n    /** Where in `buffer` its words begin. */\n    start: number\n    /** Its words ran across a block's end: some of them are in a block already. */\n    crossed: boolean\n    region: SiteImportRegion\n  } | null = null\n  /** The links read into the block being built, written after it. */\n  let pending: Array<{ text: string; href: string; start: number; end: number; crossed: boolean; region: SiteImportRegion }> = []\n\n  const region = (): SiteImportRegion =>\n    footerDepth > 0 ? 'footer' : headerDepth > 0 ? 'header' : 'main'\n  // A site without a <nav> keeps its menu in the header; its links are the nav.\n  const inNav = () => navDepth > 0 || (!hasNav && headerDepth > 0)\n  const room = () => blocks.length < SITE_IMPORT_PAGE_LIMITS.blocks\n\n  const flush = () => {\n    const raw = buffer\n    const written = pending\n    buffer = ''\n    pending = []\n    if (anchor && collapse(decodeHtmlEntities(raw.slice(anchor.start)))) anchor.crossed = true\n    if (anchor) anchor.start = 0\n    const kind = current?.kind ?? 'paragraph'\n    const level = current?.level\n    const loose = current === null\n    current = null\n    if (navDepth > 0) return\n    const text = clip(collapse(decodeHtmlEntities(raw)), SITE_IMPORT_PAGE_LIMITS.blockChars)\n    // Loose words that are only links (a row of buttons, an address in a\n    // card) are those links, each a block of its own; words around a link\n    // keep their block, and the link follows it.\n    let rest = raw\n    for (const link of [...written].sort((a, b) => b.start - a.start)) rest = `${rest.slice(0, link.start)}${rest.slice(link.end)}`\n    const onlyLinks = loose && written.length > 0 && !collapse(decodeHtmlEntities(rest))\n    if (text && !onlyLinks && room()) {\n      const last = blocks[blocks.length - 1]\n      if (!(last && last.kind !== 'image' && last.kind !== 'link' && last.text === text)) {\n        if (kind === 'heading') {\n          blocks.push({ kind, level: (level ?? 2) as 1 | 2 | 3 | 4 | 5 | 6, text, region: region() })\n        } else {\n          blocks.push({ kind, text, region: region() })\n        }\n      }\n    }\n    for (const link of written) {\n      if (!room()) break\n      const last = blocks[blocks.length - 1]\n      if (last?.kind === 'link' && last.href === link.href && last.text === link.text) continue\n      const inline = !onlyLinks || link.crossed\n      blocks.push({ kind: 'link', text: link.text, href: link.href, region: link.region, ...(inline ? { inline: true as const } : {}) })\n    }\n  }\n\n  /** A boundary between two runs of words: in the block, and in the link being read. */\n  const boundary = (block: boolean) => {\n    buffer += RUN_BREAK\n    if (!anchor) return\n    anchor.text += RUN_BREAK\n    if (block && anchor.lines[anchor.lines.length - 1] !== '') anchor.lines.push('')\n  }\n\n  for (const match of body.matchAll(new RegExp(`${TAG.source}|([^<]+)|<`, 'g'))) {\n    const text = match[4]\n    if (text !== undefined || match[2] === undefined) {\n      const run = text ?? '<'\n      buffer += run\n      if (anchor) {\n        anchor.text += run\n        anchor.lines[anchor.lines.length - 1] += run\n      }\n      continue\n    }\n    const closing = match[1] === '/'\n    const tag = match[2].toLowerCase()\n    const attributes = closing ? {} : parseAttributes(match[3] ?? '')\n\n    if (tag === 'a') {\n      if (closing) {\n        if (anchor) {\n          const label = clip(collapse(decodeHtmlEntities(anchor.text)), 300)\n          const firstLine = clip(collapse(decodeHtmlEntities(anchor.lines.find((line) => collapse(decodeHtmlEntities(line))) ?? '')), 120)\n          if (inNav() && anchor.url && firstLine && navLinks.length < SITE_IMPORT_PAGE_LIMITS.navLinks) {\n            navLinks.push({ text: firstLine, url: anchor.url.toString() })\n          }\n          if (!brand && anchor.region !== 'footer' && (headerDepth > 0 || navDepth > 0) && anchor.url && isSameSite(anchor.url) && anchor.url.pathname === '/') {\n            brand = { logo: anchor.image, text: label, region: anchor.region }\n          }\n          if (!inNav() && anchor.href && label) {\n            pending.push({ text: label, href: anchor.href, start: anchor.start, end: buffer.length, crossed: anchor.crossed, region: anchor.region })\n          }\n          anchor = null\n        }\n        boundary(false)\n        continue\n      }\n      boundary(false)\n      const url = resolveSiteUrl(attributes['href'], pageUrl)\n      anchor = {\n        url,\n        href: resolveSiteLink(attributes['href'], pageUrl),\n        text: '',\n        lines: [''],\n        image: null,\n        start: buffer.length,\n        crossed: false,\n        region: region(),\n      }\n      if (url && isSameSite(url) && links.length < SITE_IMPORT_PAGE_LIMITS.links) {\n        const key = url.toString()\n        if (!linkSeen.has(key)) {\n          linkSeen.add(key)\n          links.push(key)\n        }\n      }\n      continue\n    }\n\n    if (tag === 'img' && !closing) {\n      const url = resolveSiteUrl(\n        attributes['src'] || attributes['data-src'] || firstSrcset(attributes['srcset'] || attributes['data-srcset']),\n        pageUrl,\n      )\n      boundary(false)\n      if (url && images.length < SITE_IMPORT_PAGE_LIMITS.images) {\n        const key = url.toString()\n        const alt = clip(collapse(attributes['alt'] ?? ''), 300)\n        if (anchor && !anchor.image) anchor.image = { url: key, alt }\n        if (!imageSeen.has(key)) {\n          imageSeen.add(key)\n          images.push({ url: key, alt, sameSite: isSameSite(url), region: region() })\n          if (navDepth === 0) {\n            flush()\n            if (room()) blocks.push({ kind: 'image', url: key, alt, region: region() })\n          }\n        }\n      }\n      continue\n    }\n\n    if (tag === 'nav' || tag === 'header' || tag === 'footer') {\n      flush()\n      boundary(true)\n      const step = closing ? -1 : 1\n      if (tag === 'nav') navDepth = Math.max(0, navDepth + step)\n      if (tag === 'header') headerDepth = Math.max(0, headerDepth + step)\n      if (tag === 'footer') footerDepth = Math.max(0, footerDepth + step)\n      continue\n    }\n\n    const heading = /^h([1-6])$/.exec(tag)\n    if (heading || BLOCK_TAGS[tag]) {\n      if (closing) {\n        if (current?.tag === tag) flush()\n        boundary(true)\n        continue\n      }\n      flush()\n      boundary(true)\n      current = heading\n        ? { kind: 'heading', level: Number(heading[1]), tag }\n        : { kind: BLOCK_TAGS[tag], tag }\n      continue\n    }\n\n    if (BREAK_TAGS.has(tag)) {\n      if (tag === 'br' && current) {\n        buffer += ' '\n        if (anchor) anchor.text += ' '\n        continue\n      }\n      flush()\n      boundary(true)\n      continue\n    }\n\n    if (RUN_BOUNDARY.has(tag)) boundary(false)\n  }\n  flush()\n\n  return {\n    title,\n    description,\n    lang,\n    canonical,\n    ogImage,\n    siteName,\n    icons,\n    brand,\n    organization,\n    blocks,\n    images,\n    navLinks,\n    links,\n    forms: { kept: forms.kept, removed: forms.removed },\n    loginForm: false,\n  }\n}\n"],"names":["SITE_IMPORT_PAGE_LIMITS","blocks","blockChars","images","links","navLinks","forms","fieldsPerForm","icons","REMOVED_WITH_CONTENT","FORM_CONTROLS","REMOVED_VOID","PASSWORD_INPUT","PAYMENT_FIELD","BLOCK_TAGS","p","li","dt","dd","td","th","address","pre","figcaption","caption","label","blockquote","BREAK_TAGS","Set","RUN_BOUNDARY","RUN_BREAK","RUN_BREAKS","RegExp","joinRuns","text","includes","replace","match","offset","whole","test","before","after","length","NAMED_ENTITIES","amp","lt","gt","quot","apos","nbsp","rsquo","lsquo","rdquo","ldquo","mdash","ndash","hellip","copy","reg","trade","middot","bull","eacute","egrave","aacute","ntilde","uuml","ouml","auml","deg","times","decodeHtmlEntities","body","code","parseInt","slice","Number","isFinite","String","fromCodePoint","named","toLowerCase","ATTRIBUTE","parseAttributes","source","attributes","matchAll","name","TAG","metaTags","html","out","push","readMetaValues","wanted","filter","map","trim","Boolean","collapse","clip","max","trimEnd","removeElements","tags","tag","paired","unclosed","takeForms","kept","removed","form","fields","type","resolveSiteUrl","raw","base","value","url","URL","protocol","hash","firstSrcset","srcset","first","split","undefined","resolveSiteLink","mail","exec","decodeURIComponent","phone","startsWith","toString","readPictures","inner","img","own","sourceAttributes","quote","readIcons","pageUrl","rels","rel","sizes","size","reduce","most","entry","Math","side","sort","a","b","ORGANIZATION_TYPE","readOrganization","node","nodes","visit","depth","Array","isArray","JSON","parse","typeOf","concat","find","some","urlOf","contact","parts","region","join","country","description","logo","telephone","email","sameAs","extractSiteImportPage","isSameSite","empty","title","lang","canonical","ogImage","siteName","brand","organization","loginForm","metas","metaContent","key","attribute","m","titleMatch","htmlTag","bodyStart","search","hasNav","imageSeen","linkSeen","navDepth","headerDepth","footerDepth","current","buffer","anchor","pending","inNav","room","flush","written","start","crossed","kind","level","loose","rest","link","end","onlyLinks","last","href","inline","boundary","block","lines","run","closing","firstLine","line","pathname","image","has","add","alt","sameSite","step","heading"],"mappings":"AAAA;;;;;;;;;;;;;;;CAeC,GAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;2CA4B2C,GAE3C,iEAAiE;AAgGjE,uEAAuE,GACvE,OAAO,MAAMA,0BAA0B;IACrC,2GAA2G,GAC3GC,QAAQ;IACR,yDAAyD,GACzDC,YAAY;IACZC,QAAQ;IACRC,OAAO;IACP,6DAA6D,GAC7DC,UAAU;IACVC,OAAO;IACPC,eAAe;IACfC,OAAO;AACT,EAAU;AAEV,kDAAkD,GAClD,MAAMC,uBAAuB;IAC3B;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;CACD;AAED,6EAA6E,GAC7E,MAAMC,gBAAgB;IAAC;IAAY;IAAU;CAAS;AAEtD,gDAAgD,GAChD,MAAMC,eAAe;AAErB,MAAMC,iBAAiB;AAEvB;;;CAGC,GACD,MAAMC,gBACJ;AAEF,MAAMC,aAAkE;IACtEC,GAAG;IACHC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,SAAS;IACTC,KAAK;IACLC,YAAY;IACZC,SAAS;IACTC,OAAO;IACPC,YAAY;AACd;AAEA,wEAAwE,GACxE,MAAMC,aAAa,IAAIC,IAAI;IACzB;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;CACD;AAED;;;;;;CAMC,GACD,MAAMC,eAAe,IAAID,IAAI;IAAC;IAAK;IAAO;IAAU;IAAS;IAAU;IAAY;IAAW;IAAS;IAAS;IAAU;IAAU;IAAS;IAAU;IAAO;IAAS;IAAY;CAAS;AAE5L,uFAAuF,GACvF,MAAME,YAAY;AAElB;;;;CAIC,GACD,MAAMC,aAAa,IAAIC,OAAO,CAAC,IAAI,EAAEF,UAAU,EAAE,EAAEA,UAAU,IAAI,EAAEA,UAAU,EAAE,CAAC,EAAE;AAElF,SAASG,SAASC,IAAY;IAC5B,IAAI,CAACA,KAAKC,QAAQ,CAACL,YAAY,OAAOI;IACtC,OAAOA,KAAKE,OAAO,CAACL,YAAY,CAACM,OAAeC,QAAgBC;YAE/CA,SACDA,UAI+BA,UACMA;QAPnD,IAAI,KAAKC,IAAI,CAACH,QAAQ,OAAO;QAC7B,MAAMI,UAASF,UAAAA,KAAK,CAACD,SAAS,EAAE,YAAjBC,UAAqB;QACpC,MAAMG,SAAQH,WAAAA,KAAK,CAACD,SAASD,MAAMM,MAAM,CAAC,YAA5BJ,WAAgC;QAC9C,IAAI,CAACE,UAAU,CAACC,OAAO,OAAO;QAC9B,IAAI,kCAAkCF,IAAI,CAACE,UAAU,0BAA0BF,IAAI,CAACC,SAAS,OAAO;QACpG,4EAA4E;QAC5E,IAAI,OAAOD,IAAI,CAACC,WAAW,aAAaD,IAAI,EAACD,WAAAA,KAAK,CAACD,SAAS,EAAE,YAAjBC,WAAqB,KAAK,OAAO;QAC9E,IAAI,OAAOC,IAAI,CAACE,UAAU,oBAAoBF,IAAI,EAACD,WAAAA,KAAK,CAACD,SAASD,MAAMM,MAAM,GAAG,EAAE,YAAhCJ,WAAoC,KAAK,OAAO;QACnG,OAAO;IACT;AACF;AAEA,MAAMK,iBAAyC;IAC7CC,KAAK;IACLC,IAAI;IACJC,IAAI;IACJC,MAAM;IACNC,MAAM;IACNC,MAAM;IACNC,OAAO;IACPC,OAAO;IACPC,OAAO;IACPC,OAAO;IACPC,OAAO;IACPC,OAAO;IACPC,QAAQ;IACRC,MAAM;IACNC,KAAK;IACLC,OAAO;IACPC,QAAQ;IACRC,MAAM;IACNC,QAAQ;IACRC,QAAQ;IACRC,QAAQ;IACRC,QAAQ;IACRC,MAAM;IACNC,MAAM;IACNC,MAAM;IACNC,KAAK;IACLC,OAAO;AACT;AAEA,0EAA0E,GAC1E,OAAO,SAASC,mBAAmBtC,IAAY;IAC7C,OAAOA,KAAKE,OAAO,CAAC,yCAAyC,CAACG,OAAOkC;QACnE,IAAIA,IAAI,CAAC,EAAE,KAAK,KAAK;YACnB,MAAMC,OACJD,IAAI,CAAC,EAAE,KAAK,OAAOA,IAAI,CAAC,EAAE,KAAK,MAAME,SAASF,KAAKG,KAAK,CAAC,IAAI,MAAMD,SAASF,KAAKG,KAAK,CAAC,IAAI;YAC7F,IAAI,CAACC,OAAOC,QAAQ,CAACJ,SAASA,QAAQ,KAAKA,OAAO,UAAU,OAAO;YACnE,IAAIA,OAAO,MAAMA,SAAS,KAAKA,SAAS,MAAMA,SAAS,IAAI,OAAO;YAClE,OAAOK,OAAOC,aAAa,CAACN;QAC9B;QACA,MAAMO,QAAQrC,cAAc,CAAC6B,KAAKS,WAAW,GAAG;QAChD,OAAOD,gBAAAA,QAAS1C;IAClB;AACF;AAEA,MAAM4C,YAAY;AAElB,0DAA0D,GAC1D,OAAO,SAASC,gBAAgBC,MAAc;IAC5C,MAAMC,aAAqC,CAAC;IAC5C,KAAK,MAAMjD,SAASgD,OAAOE,QAAQ,CAACJ,WAAY;YAGR9C,MAAAA,OAAAA;QAFtC,MAAMmD,OAAOnD,KAAK,CAAC,EAAE,CAAC6C,WAAW;QACjC,IAAIM,QAAQF,YAAY;QACxBA,UAAU,CAACE,KAAK,GAAGhB,oBAAmBnC,QAAAA,SAAAA,UAAAA,KAAK,CAAC,EAAE,YAARA,UAAYA,KAAK,CAAC,EAAE,YAApBA,QAAwBA,KAAK,CAAC,EAAE,YAAhCA,OAAoC;IAC5E;IACA,OAAOiD;AACT;AAEA,MAAMG,MAAM;AAEZ,qCAAqC,GACrC,SAASC,SAASC,IAAY;IAC5B,MAAMC,MAAgC,EAAE;IACxC,KAAK,MAAMvD,SAASsD,KAAKJ,QAAQ,CAAC,2CAA4C;QAC5EK,IAAIC,IAAI,CAACT,gBAAgB/C,KAAK,CAAC,EAAE;IACnC;IACA,OAAOuD;AACT;AAEA,6DAA6D,GAC7D,OAAO,SAASE,eAAeH,IAAY,EAAEH,IAAY;IACvD,MAAMO,SAASP,KAAKN,WAAW;IAC/B,OAAOQ,SAASC,MACbK,MAAM,CAAC,CAACV;YAAgBA;eAAD,EAACA,mBAAAA,UAAU,CAAC,OAAO,YAAlBA,mBAAsB,IAAIJ,WAAW,OAAOa;OACpEE,GAAG,CAAC,CAACX;YAAgBA;eAAD,EAACA,sBAAAA,UAAU,CAAC,UAAU,YAArBA,sBAAyB,IAAIY,IAAI;OACtDF,MAAM,CAACG;AACZ;AAEA,SAASC,SAASlE,IAAY;IAC5B,OAAOD,SAASC,MAAME,OAAO,CAAC,QAAQ,KAAK8D,IAAI;AACjD;AAEA,SAASG,KAAKnE,IAAY,EAAEoE,GAAW;IACrC,OAAOpE,KAAKS,MAAM,GAAG2D,MAAM,GAAGpE,KAAK0C,KAAK,CAAC,GAAG0B,MAAM,GAAGC,OAAO,GAAG,CAAC,CAAC,GAAGrE;AACtE;AAEA,SAASsE,eAAeb,IAAY,EAAEc,IAAuB;IAC3D,IAAIb,MAAMD;IACV,KAAK,MAAMe,OAAOD,KAAM;QACtB,MAAME,SAAS,IAAI3E,OAAO,CAAC,CAAC,EAAE0E,IAAI,6CAA6C,EAAEA,IAAI,KAAK,CAAC,EAAE;QAC7Fd,MAAMA,IAAIxD,OAAO,CAACuE,QAAQ;QAC1B,wEAAwE;QACxE,MAAMC,WAAW,IAAI5E,OAAO,CAAC,CAAC,EAAE0E,IAAI,yCAAyC,CAAC,EAAE;QAChFd,MAAMA,IAAIxD,OAAO,CAACwE,UAAU;IAC9B;IACA,OAAOhB;AACT;AAEA,2EAA2E,GAC3E,SAASiB,UAAUlB,IAAY;IAC7B,MAAMmB,OAAyB,EAAE;IACjC,IAAIC,UAAU;IACd,MAAMnB,MAAMD,KAAKvD,OAAO,CAAC,kEAAkE,CAAC4E;QAC1F,IAAInG,cAAc2B,IAAI,CAACwE,OAAO;YAC5BD,WAAW;YACX,OAAO;QACT;QACA,IAAID,KAAKnE,MAAM,GAAG3C,wBAAwBM,KAAK,EAAE;YAC/C,MAAM2G,SAAmC,EAAE;YAC3C,KAAK,MAAM5E,SAAS2E,KAAKzB,QAAQ,CAAC,8DAA+D;oBAK1FD;gBAJL,MAAMA,aAAaF,gBAAgB/C,KAAK,CAAC,EAAE;gBAC3C,MAAM6E,OAAO7E,KAAK,CAAC,EAAE,CAAC6C,WAAW,OAAO,UAAU,AAACI,CAAAA,UAAU,CAAC,OAAO,IAAI,MAAK,EAAGJ,WAAW,KAAK7C,KAAK,CAAC,EAAE,CAAC6C,WAAW;gBACrH,IAAIgC,SAAS,YAAYA,SAAS,YAAYA,SAAS,YAAYA,SAAS,SAAS;gBACrF,8FAA8F;gBAC9F,IAAI,EAAC5B,yBAAAA,UAAU,CAAC,cAAc,YAAzBA,yBAA6B,IAAIJ,WAAW,OAAO,UAAUI,UAAU,CAAC,WAAW,KAAK,MAAM;gBACnG2B,OAAOpB,IAAI,CAAC;oBAAEqB;oBAAM1B,MAAMa,KAAKf,UAAU,CAAC,OAAO,IAAIA,UAAU,CAAC,KAAK,IAAI,IAAI;gBAAI;gBACjF,IAAI2B,OAAOtE,MAAM,IAAI3C,wBAAwBO,aAAa,EAAE;YAC9D;YACAuG,KAAKjB,IAAI,CAAC;gBAAEoB;YAAO;QACrB;QACA,OAAO;IACT;IACA,OAAO;QAAEtB,MAAMC;QAAKkB;QAAMC;IAAQ;AACpC;AAEA,+EAA+E,GAC/E,OAAO,SAASI,eAAeC,GAAuB,EAAEC,IAAY;IAClE,MAAMC,QAAQvC,OAAOqC,cAAAA,MAAO,IAAIlB,IAAI;IACpC,IAAI,CAACoB,SAAS,mDAAmD9E,IAAI,CAAC8E,QAAQ,OAAO;IACrF,IAAI;QACF,MAAMC,MAAM,IAAIC,IAAIF,OAAOD;QAC3B,IAAIE,IAAIE,QAAQ,KAAK,YAAYF,IAAIE,QAAQ,KAAK,SAAS,OAAO;QAClEF,IAAIG,IAAI,GAAG;QACX,OAAOH;IACT,EAAE,eAAM;QACN,OAAO;IACT;AACF;AAEA,yCAAyC,GACzC,SAASI,YAAYC,MAA0B;QAC/B7C;IAAd,MAAM8C,SAAQ9C,iBAAAA,OAAO6C,iBAAAA,SAAU,IAAIE,KAAK,CAAC,IAAI,CAAC,EAAE,qBAAlC/C,eAAoCmB,IAAI,GAAG4B,KAAK,CAAC,MAAM,CAAC,EAAE;IACxE,OAAOD,SAASE;AAClB;AAEA;;;;CAIC,GACD,OAAO,SAASC,gBAAgBZ,GAAuB,EAAEC,IAAY;;QAO5DF;IANP,MAAMG,QAAQvC,OAAOqC,cAAAA,MAAO,IAAIlB,IAAI;IACpC,MAAM+B,OAAO,qCAAqCC,IAAI,CAACZ;IACvD,IAAIW,MAAM,OAAO,CAAC,OAAO,EAAEE,mBAAmBF,IAAI,CAAC,EAAE,EAAE/C,WAAW,IAAI;IACtE,MAAMkD,QAAQ,iCAAiCF,IAAI,CAACZ;IACpD,IAAIc,OAAO,OAAO,CAAC,IAAI,EAAEA,KAAK,CAAC,EAAE,CAAChG,OAAO,CAAC,aAAa,KAAK;IAC5D,IAAIkF,MAAMe,UAAU,CAAC,MAAM,OAAO;IAClC,gBAAOlB,kBAAAA,eAAeG,OAAOD,0BAAtBF,gBAA6BmB,QAAQ,qBAAM;AACpD;AAEA;;;;CAIC,GACD,SAASC,aAAa5C,IAAY;IAChC,OAAOA,KAAKvD,OAAO,CAAC,oEAAoE,CAACG,OAAOiG;YAa5ClD;QAZlD,MAAMmD,MAAM,wCAAwCP,IAAI,CAACM;QACzD,MAAMlD,aAAamD,MAAMrD,gBAAgBqD,GAAG,CAAC,EAAE,IAAI,CAAC;QACpD,MAAMC,MAAMpD,UAAU,CAAC,MAAM,IAAIA,UAAU,CAAC,WAAW,IAAIqC,YAAYrC,UAAU,CAAC,SAAS,IAAIA,UAAU,CAAC,cAAc;QACxH,IAAIoD,OAAO,CAAC,UAAUlG,IAAI,CAACkG,IAAIxC,IAAI,KAAK,OAAO3D;QAC/C,IAAI8C;QACJ,KAAK,MAAMhD,SAASmG,MAAMjD,QAAQ,CAAC,6CAA8C;YAC/E,MAAMoD,mBAAmBvD,gBAAgB/C,KAAK,CAAC,EAAE;YACjDgD,SAASsC,YAAYgB,gBAAgB,CAAC,SAAS,IAAIA,gBAAgB,CAAC,cAAc;YAClF,IAAItD,QAAQ;QACd;QACA,IAAI,CAACA,QAAQ,OAAO9C;QACpB,MAAMqG,QAAQ,CAACtB,QAAkBA,MAAMlF,OAAO,CAAC,MAAM,SAASA,OAAO,CAAC,MAAM,UAAUA,OAAO,CAAC,MAAM;QACpG,OAAO,CAAC,WAAW,EAAEwG,MAAMvD,QAAQ,OAAO,EAAEuD,OAAMtD,kBAAAA,UAAU,CAAC,MAAM,YAAjBA,kBAAqB,IAAI,GAAG,CAAC;IACjF;AACF;AAEA,0DAA0D,GAC1D,SAASuD,UAAUlD,IAAY,EAAEmD,OAAe;IAC9C,MAAMtI,QAA0B,EAAE;IAClC,KAAK,MAAM6B,SAASsD,KAAKJ,QAAQ,CAAC,2CAA4C;YAE9DD,iBASe,QAACA;YAEmCA;QAZjE,MAAMA,aAAaF,gBAAgB/C,KAAK,CAAC,EAAE;QAC3C,MAAM0G,OAAO,EAACzD,kBAAAA,UAAU,CAAC,MAAM,YAAjBA,kBAAqB,IAAIJ,WAAW,GAAG4C,KAAK,CAAC;QAC3D,MAAMkB,MAAMD,KAAK5G,QAAQ,CAAC,uBAAuB4G,KAAK5G,QAAQ,CAAC,kCAC3D,qBACA4G,KAAK5G,QAAQ,CAAC,UACZ,SACA;QACN,IAAI,CAAC6G,KAAK;QACV,MAAMzB,MAAMJ,eAAe7B,UAAU,CAAC,OAAO,EAAEwD;QAC/C,IAAI,CAACvB,KAAK;QACV,MAAM0B,QAAkB;gBAAK,SAAA,EAAC3D,oBAAAA,UAAU,CAAC,QAAQ,YAAnBA,oBAAuB,IAAIjD,KAAK,CAAC,kCAAlC,SAA4D,EAAE;SAAE;QAC7F,MAAM6G,OAAOD,MAAME,MAAM,CAAS,CAACC,MAAMC,QAAUC,KAAKhD,GAAG,CAAC8C,SAASC,MAAMvB,KAAK,CAAC,MAAM7B,GAAG,CAAC,CAACsD,OAAS1E,OAAO0E,SAAS,KAAK;QAC1H/I,MAAMqF,IAAI,CAAC;YAAE0B,KAAKA,IAAIe,QAAQ;YAAIU;YAAKE,MAAMA,QAAQ;YAAMhC,MAAM5B,EAAAA,mBAAAA,UAAU,CAAC,OAAO,qBAAlBA,iBAAoBY,IAAI,OAAM;QAAK;QACpG,IAAI1F,MAAMmC,MAAM,IAAI3C,wBAAwBQ,KAAK,EAAE;IACrD;IACA,OAAOA,MAAMgJ,IAAI,CAAC,CAACC,GAAGC;YAA0BA,SAAgBD;eAAnCA,EAAET,GAAG,KAAKU,EAAEV,GAAG,GAAG,EAACU,UAAAA,EAAER,IAAI,YAANQ,UAAU,OAAMD,UAAAA,EAAEP,IAAI,YAANO,UAAU,KAAKA,EAAET,GAAG,KAAK,qBAAqB,CAAC,IAAI;;AACrH;AAEA,MAAMW,oBAAoB;AAE1B;;;;CAIC,GACD,SAASC,iBAAiBjE,IAAY;QA2BKkE,oBAuB7BA;IAjDZ,MAAMC,QAAmC,EAAE;IAC3C,MAAMC,QAAQ,CAACzC,OAAgB0C;QAC7B,IAAIA,QAAQ,KAAKF,MAAMnH,MAAM,GAAG,MAAM,CAAC2E,SAAS,OAAOA,UAAU,UAAU;QAC3E,IAAI2C,MAAMC,OAAO,CAAC5C,QAAQ;YACxB,KAAK,MAAM+B,SAAS/B,MAAOyC,MAAMV,OAAOW,QAAQ;YAChD;QACF;QACA,MAAMH,OAAOvC;QACbwC,MAAMjE,IAAI,CAACgE;QACX,IAAIA,IAAI,CAAC,SAAS,EAAEE,MAAMF,IAAI,CAAC,SAAS,EAAEG,QAAQ;IACpD;IACA,KAAK,MAAM3H,SAASsD,KAAKJ,QAAQ,CAAC,6FAA8F;QAC9H,IAAI;YACFwE,MAAMI,KAAKC,KAAK,CAAC/H,KAAK,CAAC,EAAE,GAAG;QAC9B,EAAE,eAAM;QACN,oDAAoD;QACtD;IACF;IACA,MAAMgI,SAAS,CAACR;YAA2DA;eAAzB,AAAC,EAAE,CAAeS,MAAM,EAACT,aAAAA,IAAI,CAAC,QAAQ,YAAbA,aAAiB,EAAE,EAAE5D,GAAG,CAAClB;;IACpG,MAAM8E,OAAOC,MAAMS,IAAI,CAAC,CAAClB,QAAUgB,OAAOhB,OAAOmB,IAAI,CAAC,CAACtD,OAASyC,kBAAkBnH,IAAI,CAAC0E;IACvF,IAAI,CAAC2C,MAAM,OAAO;IAClB,MAAM3H,OAAO,CAACoF,OAAgBhB,MAAM,GAAG,GAAc,OAAOgB,UAAU,WAAWjB,KAAKD,SAASkB,QAAQhB,OAAO;IAC9G,MAAMmE,QAAQ,CAACnD;QACb,MAAMF,MAAM,OAAOE,UAAU,WAAWA,QAAQA,SAAS,OAAOA,UAAU,WAAW,AAACA,KAAiC,CAAC,MAAM,GAAG;QACjI,OAAO,OAAOF,QAAQ,YAAY,gBAAgB5E,IAAI,CAAC4E,IAAIlB,IAAI,MAAMkB,IAAIlB,IAAI,KAAK;IACpF;IACA,MAAMwE,UAAU,AAAC,EAAE,CAAeJ,MAAM,EAACT,qBAAAA,IAAI,CAAC,eAAe,YAApBA,qBAAwB,EAAE,EAAEU,IAAI,CAAC,CAAClB,QAAUA,SAAS,OAAOA,UAAU;IAG/G,MAAMhI,UAAU,AAAC,CAAA;QACf,MAAMiG,QAAQuC,IAAI,CAAC,UAAU;QAC7B,IAAI,OAAOvC,UAAU,UAAU,OAAOpF,KAAKoF;QAC3C,IAAI,CAACA,SAAS,OAAOA,UAAU,UAAU,OAAO;QAChD,MAAMqD,QAAQrD;QACd,MAAMsD,SAAS;YAAC1I,KAAKyI,KAAK,CAAC,gBAAgB;YAAGzI,KAAKyI,KAAK,CAAC,aAAa;SAAE,CAAC3E,MAAM,CAACG,SAAS0E,IAAI,CAAC;QAC9F,MAAMC,UAAUH,KAAK,CAAC,iBAAiB;QACvC,OAAO;YAACzI,KAAKyI,KAAK,CAAC,gBAAgB;YAAGzI,KAAKyI,KAAK,CAAC,kBAAkB;YAAGC;YAAQ,OAAOE,YAAY,WAAW5I,KAAK4I,WAAW5I,KAAM4I,2BAAD,AAACA,OAA4C,CAAC,OAAO;SAAE,CACrL9E,MAAM,CAACG,SACP0E,IAAI,CAAC;IACV,CAAA;IACA,OAAO;QACLrF,MAAMtD,KAAK2H,IAAI,CAAC,OAAO;QACvBkB,aAAa7I,KAAK2H,IAAI,CAAC,cAAc,EAAE;QACvCtC,KAAKkD,MAAMZ,IAAI,CAAC,MAAM;QACtBmB,MAAMP,MAAMZ,IAAI,CAAC,OAAO;QACxBoB,WAAW/I,KAAK2H,IAAI,CAAC,YAAY,KAAK3H,KAAKwI,2BAAAA,OAAS,CAAC,YAAY;QACjEQ,OAAO,AAAChJ,CAAAA,KAAK2H,IAAI,CAAC,QAAQ,KAAK3H,KAAKwI,2BAAAA,OAAS,CAAC,QAAQ,CAAA,EAAGtI,OAAO,CAAC,aAAa;QAC9Ef;QACA8J,QAAQ,AAAC,EAAE,CACRb,MAAM,EAACT,eAAAA,IAAI,CAAC,SAAS,YAAdA,eAAkB,EAAE,EAC3B5D,GAAG,CAAC,CAACoD,QAAUoB,MAAMpB,QACrBrD,MAAM,CAACG,SACPvB,KAAK,CAAC,GAAG;IACd;AACF;AAEA;;;;;CAKC,GACD,OAAO,SAASwG,sBACdzF,IAAY,EACZmD,OAAe,EACfuC,UAAiC;;QA6C/BlE,iBAIqB/B;IA/CvB,MAAMkG,QAA2B;QAC/BC,OAAO;QACPR,aAAa;QACbS,MAAM;QACNC,WAAW;QACXC,SAAS;QACTC,UAAU;QACVnL,OAAO,EAAE;QACToL,OAAO;QACPC,cAAc;QACd5L,QAAQ,EAAE;QACVE,QAAQ,EAAE;QACVE,UAAU,EAAE;QACZD,OAAO,EAAE;QACTE,OAAO;YAAEwG,MAAM,EAAE;YAAEC,SAAS;QAAE;QAC9B+E,WAAW;IACb;IACA,IAAIzG,SAASN,OAAOY,eAAAA,OAAQ,IACzBvD,OAAO,CAAC,0BAA0B,KAClCA,OAAO,CAAC,6BAA6B;IACxC,IAAIxB,eAAe4B,IAAI,CAAC6C,SAAS,OAAO,aAAKiG;QAAOQ,WAAW;;IAE/D,kEAAkE;IAClE,MAAMD,eAAejC,iBAAiBvE;IACtC,MAAM7E,QAAQqI,UAAUxD,QAAQyD;IAChCzD,SAASmB,eAAe+B,aAAalD,SAAS5E,sBAAsB2B,OAAO,CAACzB,cAAc;IAC1F,8DAA8D;IAC9D,MAAML,QAAQuG,UAAUxB;IACxBA,SAASmB,eAAelG,MAAMqF,IAAI,EAAEjF;IAEpC,MAAMqL,QAAQrG,SAASL;IACvB,MAAM2G,cAAc,CAACC,KAAaC;;YACvBH;eAAT3F,kBAAS2F,cAAAA,MAAMxB,IAAI,CAAC,CAAC4B;gBAAOA;mBAAD,EAACA,eAAAA,CAAC,CAACD,UAAU,YAAZC,eAAgB,IAAIjH,WAAW,OAAO+G;+BAAzDF,WAA+D,CAAC,UAAU,mBAAI;;IACzF,MAAMK,aAAa,wCAAwClE,IAAI,CAAC7C;IAChE,MAAMkG,QAAQlF,KACZD,SAAS5B,2BAAmB4H,8BAAAA,UAAY,CAAC,EAAE,mBAAI,QAAQJ,YAAY,YAAY,aAC/E;IAEF,MAAMjB,cAAc1E,KAClB2F,YAAY,eAAe,WAAWA,YAAY,kBAAkB,aACpE;IAEF,MAAMN,oBACJvE,kBAAAA,eAAe6E,YAAY,YAAY,eAAeA,YAAY,gBAAgB,eAAeA,YAAY,iBAAiB,SAASlD,6BAAvI3B,gBAAiJmB,QAAQ,sBACzJ;IACF,MAAMqD,WAAWtF,KAAK2F,YAAY,gBAAgB,eAAeA,YAAY,oBAAoB,SAAS;IAC1G,MAAMK,UAAU,yCAAyCnE,IAAI,CAAC7C;IAC9D,MAAMmG,OAAOa,UAAUjH,EAAAA,wBAAAA,gBAAgBiH,OAAO,CAAC,EAAE,CAAC,CAAC,OAAO,qBAAnCjH,sBAAqCc,IAAI,OAAM,OAAO;IAC7E,IAAIuF,YAA2B;IAC/B,KAAK,MAAMpJ,SAAS0C,OAAOY,eAAAA,OAAQ,IAAIJ,QAAQ,CAAC,2CAA4C;YAErFD;QADL,MAAMA,aAAaF,gBAAgB/C,KAAK,CAAC,EAAE;QAC3C,IAAI,EAACiD,kBAAAA,UAAU,CAAC,MAAM,YAAjBA,kBAAqB,IAAIJ,WAAW,GAAG4C,KAAK,CAAC,OAAO3F,QAAQ,CAAC,cAAc;;gBAClEgF;YAAZsE,sBAAYtE,mBAAAA,eAAe7B,UAAU,CAAC,OAAO,EAAEwD,6BAAnC3B,iBAA6CmB,QAAQ,sBAAM;YACvE;QACF;IACF;IAEA,MAAMgE,YAAYjH,OAAOkH,MAAM,CAAC;IAChC,MAAM9H,OAAO6H,aAAa,IAAIjH,OAAOT,KAAK,CAAC0H,aAAajH,OAAOjD,OAAO,CAAC,+BAA+B;IACtG,MAAMoK,SAAS,UAAUhK,IAAI,CAACiC;IAE9B,MAAMxE,SAA4B,EAAE;IACpC,MAAME,SAA4B,EAAE;IACpC,MAAMsM,YAAY,IAAI7K;IACtB,MAAMvB,WAAgC,EAAE;IACxC,MAAMD,QAAkB,EAAE;IAC1B,MAAMsM,WAAW,IAAI9K;IACrB,IAAIgK,QAAoC;IAExC,IAAIe,WAAW;IACf,IAAIC,cAAc;IAClB,IAAIC,cAAc;IAClB,IAAIC,UAAyG;IAC7G,IAAIC,SAAS;IACb,0GAA0G,GAC1G,IAAIC,SAYO;IACX,iEAAiE,GACjE,IAAIC,UAAyH,EAAE;IAE/H,MAAMrC,SAAS,IACbiC,cAAc,IAAI,WAAWD,cAAc,IAAI,WAAW;IAC5D,8EAA8E;IAC9E,MAAMM,QAAQ,IAAMP,WAAW,KAAM,CAACH,UAAUI,cAAc;IAC9D,MAAMO,OAAO,IAAMlN,OAAO0C,MAAM,GAAG3C,wBAAwBC,MAAM;IAEjE,MAAMmN,QAAQ;;QACZ,MAAMhG,MAAM2F;QACZ,MAAMM,UAAUJ;QAChBF,SAAS;QACTE,UAAU,EAAE;QACZ,IAAID,UAAU5G,SAAS5B,mBAAmB4C,IAAIxC,KAAK,CAACoI,OAAOM,KAAK,KAAKN,OAAOO,OAAO,GAAG;QACtF,IAAIP,QAAQA,OAAOM,KAAK,GAAG;QAC3B,MAAME,eAAOV,2BAAAA,QAASU,IAAI,mBAAI;QAC9B,MAAMC,QAAQX,2BAAAA,QAASW,KAAK;QAC5B,MAAMC,QAAQZ,YAAY;QAC1BA,UAAU;QACV,IAAIH,WAAW,GAAG;QAClB,MAAMzK,OAAOmE,KAAKD,SAAS5B,mBAAmB4C,OAAOpH,wBAAwBE,UAAU;QACvF,qEAAqE;QACrE,sEAAsE;QACtE,6CAA6C;QAC7C,IAAIyN,OAAOvG;QACX,KAAK,MAAMwG,QAAQ;eAAIP;SAAQ,CAAC7D,IAAI,CAAC,CAACC,GAAGC,IAAMA,EAAE4D,KAAK,GAAG7D,EAAE6D,KAAK,EAAGK,OAAO,GAAGA,KAAK/I,KAAK,CAAC,GAAGgJ,KAAKN,KAAK,IAAIK,KAAK/I,KAAK,CAACgJ,KAAKC,GAAG,GAAG;QAC/H,MAAMC,YAAYJ,SAASL,QAAQ1K,MAAM,GAAG,KAAK,CAACyD,SAAS5B,mBAAmBmJ;QAC9E,IAAIzL,QAAQ,CAAC4L,aAAaX,QAAQ;YAChC,MAAMY,OAAO9N,MAAM,CAACA,OAAO0C,MAAM,GAAG,EAAE;YACtC,IAAI,CAAEoL,CAAAA,QAAQA,KAAKP,IAAI,KAAK,WAAWO,KAAKP,IAAI,KAAK,UAAUO,KAAK7L,IAAI,KAAKA,IAAG,GAAI;gBAClF,IAAIsL,SAAS,WAAW;oBACtBvN,OAAO4F,IAAI,CAAC;wBAAE2H;wBAAMC,KAAK,EAAGA,gBAAAA,QAAS;wBAA6BvL;wBAAM0I,QAAQA;oBAAS;gBAC3F,OAAO;oBACL3K,OAAO4F,IAAI,CAAC;wBAAE2H;wBAAMtL;wBAAM0I,QAAQA;oBAAS;gBAC7C;YACF;QACF;QACA,KAAK,MAAMgD,QAAQP,QAAS;YAC1B,IAAI,CAACF,QAAQ;YACb,MAAMY,OAAO9N,MAAM,CAACA,OAAO0C,MAAM,GAAG,EAAE;YACtC,IAAIoL,CAAAA,wBAAAA,KAAMP,IAAI,MAAK,UAAUO,KAAKC,IAAI,KAAKJ,KAAKI,IAAI,IAAID,KAAK7L,IAAI,KAAK0L,KAAK1L,IAAI,EAAE;YACjF,MAAM+L,SAAS,CAACH,aAAaF,KAAKL,OAAO;YACzCtN,OAAO4F,IAAI,CAAC;gBAAE2H,MAAM;gBAAQtL,MAAM0L,KAAK1L,IAAI;gBAAE8L,MAAMJ,KAAKI,IAAI;gBAAEpD,QAAQgD,KAAKhD,MAAM;eAAMqD,SAAS;gBAAEA,QAAQ;YAAc,IAAI,CAAC;QAC/H;IACF;IAEA,oFAAoF,GACpF,MAAMC,WAAW,CAACC;QAChBpB,UAAUjL;QACV,IAAI,CAACkL,QAAQ;QACbA,OAAO9K,IAAI,IAAIJ;QACf,IAAIqM,SAASnB,OAAOoB,KAAK,CAACpB,OAAOoB,KAAK,CAACzL,MAAM,GAAG,EAAE,KAAK,IAAIqK,OAAOoB,KAAK,CAACvI,IAAI,CAAC;IAC/E;IAEA,KAAK,MAAMxD,SAASoC,KAAKc,QAAQ,CAAC,IAAIvD,OAAO,GAAGyD,IAAIJ,MAAM,CAAC,UAAU,CAAC,EAAE,MAAO;YAa3BhD;QAZlD,MAAMH,OAAOG,KAAK,CAAC,EAAE;QACrB,IAAIH,SAAS6F,aAAa1F,KAAK,CAAC,EAAE,KAAK0F,WAAW;YAChD,MAAMsG,MAAMnM,eAAAA,OAAQ;YACpB6K,UAAUsB;YACV,IAAIrB,QAAQ;gBACVA,OAAO9K,IAAI,IAAImM;gBACfrB,OAAOoB,KAAK,CAACpB,OAAOoB,KAAK,CAACzL,MAAM,GAAG,EAAE,IAAI0L;YAC3C;YACA;QACF;QACA,MAAMC,UAAUjM,KAAK,CAAC,EAAE,KAAK;QAC7B,MAAMqE,MAAMrE,KAAK,CAAC,EAAE,CAAC6C,WAAW;QAChC,MAAMI,aAAagJ,UAAU,CAAC,IAAIlJ,iBAAgB/C,UAAAA,KAAK,CAAC,EAAE,YAARA,UAAY;QAE9D,IAAIqE,QAAQ,KAAK;YACf,IAAI4H,SAAS;gBACX,IAAItB,QAAQ;wBAEyCA;oBADnD,MAAMvL,QAAQ4E,KAAKD,SAAS5B,mBAAmBwI,OAAO9K,IAAI,IAAI;oBAC9D,MAAMqM,YAAYlI,KAAKD,SAAS5B,oBAAmBwI,qBAAAA,OAAOoB,KAAK,CAAC7D,IAAI,CAAC,CAACiE,OAASpI,SAAS5B,mBAAmBgK,mBAAxDxB,qBAAmE,MAAM;oBAC5H,IAAIE,WAAWF,OAAOzF,GAAG,IAAIgH,aAAalO,SAASsC,MAAM,GAAG3C,wBAAwBK,QAAQ,EAAE;wBAC5FA,SAASwF,IAAI,CAAC;4BAAE3D,MAAMqM;4BAAWhH,KAAKyF,OAAOzF,GAAG,CAACe,QAAQ;wBAAG;oBAC9D;oBACA,IAAI,CAACsD,SAASoB,OAAOpC,MAAM,KAAK,YAAagC,CAAAA,cAAc,KAAKD,WAAW,CAAA,KAAMK,OAAOzF,GAAG,IAAI8D,WAAW2B,OAAOzF,GAAG,KAAKyF,OAAOzF,GAAG,CAACkH,QAAQ,KAAK,KAAK;wBACpJ7C,QAAQ;4BAAEZ,MAAMgC,OAAO0B,KAAK;4BAAExM,MAAMT;4BAAOmJ,QAAQoC,OAAOpC,MAAM;wBAAC;oBACnE;oBACA,IAAI,CAACsC,WAAWF,OAAOgB,IAAI,IAAIvM,OAAO;wBACpCwL,QAAQpH,IAAI,CAAC;4BAAE3D,MAAMT;4BAAOuM,MAAMhB,OAAOgB,IAAI;4BAAEV,OAAON,OAAOM,KAAK;4BAAEO,KAAKd,OAAOpK,MAAM;4BAAE4K,SAASP,OAAOO,OAAO;4BAAE3C,QAAQoC,OAAOpC,MAAM;wBAAC;oBACzI;oBACAoC,SAAS;gBACX;gBACAkB,SAAS;gBACT;YACF;YACAA,SAAS;YACT,MAAM3G,MAAMJ,eAAe7B,UAAU,CAAC,OAAO,EAAEwD;YAC/CkE,SAAS;gBACPzF;gBACAyG,MAAMhG,gBAAgB1C,UAAU,CAAC,OAAO,EAAEwD;gBAC1C5G,MAAM;gBACNkM,OAAO;oBAAC;iBAAG;gBACXM,OAAO;gBACPpB,OAAOP,OAAOpK,MAAM;gBACpB4K,SAAS;gBACT3C,QAAQA;YACV;YACA,IAAIrD,OAAO8D,WAAW9D,QAAQnH,MAAMuC,MAAM,GAAG3C,wBAAwBI,KAAK,EAAE;gBAC1E,MAAM6L,MAAM1E,IAAIe,QAAQ;gBACxB,IAAI,CAACoE,SAASiC,GAAG,CAAC1C,MAAM;oBACtBS,SAASkC,GAAG,CAAC3C;oBACb7L,MAAMyF,IAAI,CAACoG;gBACb;YACF;YACA;QACF;QAEA,IAAIvF,QAAQ,SAAS,CAAC4H,SAAS;YAC7B,MAAM/G,MAAMJ,eACV7B,UAAU,CAAC,MAAM,IAAIA,UAAU,CAAC,WAAW,IAAIqC,YAAYrC,UAAU,CAAC,SAAS,IAAIA,UAAU,CAAC,cAAc,GAC5GwD;YAEFoF,SAAS;YACT,IAAI3G,OAAOpH,OAAOwC,MAAM,GAAG3C,wBAAwBG,MAAM,EAAE;oBAE/BmF;gBAD1B,MAAM2G,MAAM1E,IAAIe,QAAQ;gBACxB,MAAMuG,MAAMxI,KAAKD,UAASd,kBAAAA,UAAU,CAAC,MAAM,YAAjBA,kBAAqB,KAAK;gBACpD,IAAI0H,UAAU,CAACA,OAAO0B,KAAK,EAAE1B,OAAO0B,KAAK,GAAG;oBAAEnH,KAAK0E;oBAAK4C;gBAAI;gBAC5D,IAAI,CAACpC,UAAUkC,GAAG,CAAC1C,MAAM;oBACvBQ,UAAUmC,GAAG,CAAC3C;oBACd9L,OAAO0F,IAAI,CAAC;wBAAE0B,KAAK0E;wBAAK4C;wBAAKC,UAAUzD,WAAW9D;wBAAMqD,QAAQA;oBAAS;oBACzE,IAAI+B,aAAa,GAAG;wBAClBS;wBACA,IAAID,QAAQlN,OAAO4F,IAAI,CAAC;4BAAE2H,MAAM;4BAASjG,KAAK0E;4BAAK4C;4BAAKjE,QAAQA;wBAAS;oBAC3E;gBACF;YACF;YACA;QACF;QAEA,IAAIlE,QAAQ,SAASA,QAAQ,YAAYA,QAAQ,UAAU;YACzD0G;YACAc,SAAS;YACT,MAAMa,OAAOT,UAAU,CAAC,IAAI;YAC5B,IAAI5H,QAAQ,OAAOiG,WAAWrD,KAAKhD,GAAG,CAAC,GAAGqG,WAAWoC;YACrD,IAAIrI,QAAQ,UAAUkG,cAActD,KAAKhD,GAAG,CAAC,GAAGsG,cAAcmC;YAC9D,IAAIrI,QAAQ,UAAUmG,cAAcvD,KAAKhD,GAAG,CAAC,GAAGuG,cAAckC;YAC9D;QACF;QAEA,MAAMC,UAAU,aAAa9G,IAAI,CAACxB;QAClC,IAAIsI,WAAWlO,UAAU,CAAC4F,IAAI,EAAE;YAC9B,IAAI4H,SAAS;gBACX,IAAIxB,CAAAA,2BAAAA,QAASpG,GAAG,MAAKA,KAAK0G;gBAC1Bc,SAAS;gBACT;YACF;YACAd;YACAc,SAAS;YACTpB,UAAUkC,UACN;gBAAExB,MAAM;gBAAWC,OAAO5I,OAAOmK,OAAO,CAAC,EAAE;gBAAGtI;YAAI,IAClD;gBAAE8G,MAAM1M,UAAU,CAAC4F,IAAI;gBAAEA;YAAI;YACjC;QACF;QAEA,IAAI/E,WAAWgN,GAAG,CAACjI,MAAM;YACvB,IAAIA,QAAQ,QAAQoG,SAAS;gBAC3BC,UAAU;gBACV,IAAIC,QAAQA,OAAO9K,IAAI,IAAI;gBAC3B;YACF;YACAkL;YACAc,SAAS;YACT;QACF;QAEA,IAAIrM,aAAa8M,GAAG,CAACjI,MAAMwH,SAAS;IACtC;IACAd;IAEA,OAAO;QACL7B;QACAR;QACAS;QACAC;QACAC;QACAC;QACAnL;QACAoL;QACAC;QACA5L;QACAE;QACAE;QACAD;QACAE,OAAO;YAAEwG,MAAMxG,MAAMwG,IAAI;YAAEC,SAASzG,MAAMyG,OAAO;QAAC;QAClD+E,WAAW;IACb;AACF"}

package/src/lib/site-import/site-import-images.d.ts

@@ -17,6 +17,7 @@
 import type { PluginMediaIngest } from '@aglyn/aglyn/plugin-manager/plugin-media-ingest';
 import { type PublicFetchOptions, type PublicFetchResult } from '../jobs/ai-theme-brand-inputs';
 import type { AiSiteImportPage } from '../model/ai-site-import';
+import { type SiteImportPage } from './site-import-crawl';
 /** The largest picture brought in. */
 export declare const SITE_IMPORT_IMAGE_MAX_BYTES: number;
 /** A picture smaller than this is a spacer or a tracking pixel, not a photo. */
@@ -52,3 +53,24 @@ export interface SiteImportPhotosDeps {
  * library already holds by key. Never throws for anything a picture does.
  */
 export declare function bringInSiteImportPhotos(input: SiteImportPhotosInput, deps: SiteImportPhotosDeps): Promise<NonNullable<AiSiteImportPage['photos']>>;
+/**
+ * A picture that may be the site's logo, where it was found. `type` is what
+ * its address or its declared type says: `svg`, a raster, or `null` when
+ * neither says (a media library address with no extension); the ingest
+ * reads the answer's own type either way.
+ */
+export interface SiteImportLogoCandidate {
+    url: string;
+    alt: string;
+    source: 'header' | 'footer' | 'structured-data' | 'apple-touch-icon' | 'icon';
+    type: 'svg' | 'raster' | null;
+}
+/** A logo's address as it is fetched: https, no fragment or tracking parameter. An SVG is a logo, unlike a photo. */
+export declare function siteImportLogoUrl(raw: string): string | null;
+/**
+ * The pictures that may be the site's logo (AGL-4083), the likeliest first:
+ * the picture in the header's link home, any header picture that says it is
+ * the logo or the site's name, the footer's, the structured data's logo,
+ * then the largest touch icon and the largest icon. Each address once.
+ */
+export declare function siteImportLogoCandidates(page: Pick<SiteImportPage, 'brand' | 'images' | 'icons' | 'siteName' | 'organization' | 'title'>): SiteImportLogoCandidate[];

package/src/lib/site-import/site-import-images.js

@@ -136,7 +136,9 @@ const EXTENSION = {
                     height: held.height
                 } : {}, held.alt || image.alt ? {
                     alt: held.alt || image.alt
-                } : {}));
+                } : {}, {
+                    from: image.url
+                }));
                 continue;
             }
             const fetched = await fetchBytes(url, _extends({
@@ -184,7 +186,10 @@ const EXTENSION = {
                 height: stored.height
             } : {}, image.alt ? {
                 alt: image.alt.slice(0, 250)
-            } : {}));
+            } : {}, {
+                // Where it sat on the old page, so a page copied as it is places it there (AGL-4084).
+                from: image.url
+            }));
         } catch (error) {
             console.warn('[site-import] a photo could not be brought in', {
                 hostId: input.hostId,
@@ -200,5 +205,76 @@ const EXTENSION = {
         assets
     };
 }
+/** A logo's address as it is fetched: https, no fragment or tracking parameter. An SVG is a logo, unlike a photo. */ export function siteImportLogoUrl(raw) {
+    let url;
+    try {
+        url = new URL(String(raw != null ? raw : '').trim());
+    } catch (unused) {
+        return null;
+    }
+    if (url.protocol === 'http:') url.protocol = 'https:';
+    if (url.protocol !== 'https:' || url.username || url.password) return null;
+    if (url.port && url.port !== '443') return null;
+    url.hash = '';
+    for (const key of [
+        ...url.searchParams.keys()
+    ]){
+        if (TRACKING_PARAM.test(key)) url.searchParams.delete(key);
+    }
+    return url.toString();
+}
+function logoTypeOf(url, declared) {
+    const type = (declared != null ? declared : '').toLowerCase();
+    if (type.includes('svg')) return 'svg';
+    if (/^image\/(?:png|jpe?g|webp|gif|x-icon|vnd\.microsoft\.icon)/.test(type)) return 'raster';
+    let path;
+    try {
+        path = new URL(url).pathname.toLowerCase();
+    } catch (unused) {
+        return null;
+    }
+    if (path.endsWith('.svg')) return 'svg';
+    if (/\.(?:png|jpe?g|webp|gif|ico)$/.test(path)) return 'raster';
+    return null;
+}
+/**
+ * The pictures that may be the site's logo (AGL-4083), the likeliest first:
+ * the picture in the header's link home, any header picture that says it is
+ * the logo or the site's name, the footer's, the structured data's logo,
+ * then the largest touch icon and the largest icon. Each address once.
+ */ export function siteImportLogoCandidates(page) {
+    var _page_organization, _page_brand, _page_organization1;
+    const name = (page.siteName || ((_page_organization = page.organization) == null ? void 0 : _page_organization.name) || '').trim().toLowerCase();
+    const says = (image)=>{
+        const alt = image.alt.trim().toLowerCase();
+        return /\blogo\b/i.test(image.alt) || /logo/i.test(image.url) || !!name && (alt === name || alt === `${name} logo`);
+    };
+    const out = [];
+    const add = (raw, alt, source, declared)=>{
+        const url = siteImportLogoUrl(raw);
+        if (!url || out.some((entry)=>entry.url === url)) return;
+        out.push({
+            url,
+            alt: alt.slice(0, 250),
+            source,
+            type: logoTypeOf(url, declared)
+        });
+    };
+    if (((_page_brand = page.brand) == null ? void 0 : _page_brand.logo) && page.brand.region !== 'footer') add(page.brand.logo.url, page.brand.logo.alt, 'header');
+    for (const image of page.images)if (image.region === 'header' && says(image)) add(image.url, image.alt, 'header');
+    for (const image of page.images)if (image.region === 'footer' && says(image)) add(image.url, image.alt, 'footer');
+    if ((_page_organization1 = page.organization) == null ? void 0 : _page_organization1.logo) add(page.organization.logo, page.organization.name, 'structured-data');
+    const touch = page.icons.filter((icon)=>icon.rel === 'apple-touch-icon').sort((a, b)=>{
+        var _b_size, _a_size;
+        return ((_b_size = b.size) != null ? _b_size : 0) - ((_a_size = a.size) != null ? _a_size : 0);
+    })[0];
+    if (touch) add(touch.url, '', 'apple-touch-icon', touch.type);
+    const icon = page.icons.filter((entry)=>entry.rel === 'icon').sort((a, b)=>{
+        var _b_size, _a_size;
+        return ((_b_size = b.size) != null ? _b_size : 0) - ((_a_size = a.size) != null ? _a_size : 0);
+    })[0];
+    if (icon) add(icon.url, '', 'icon', icon.type);
+    return out;
+}
 
 //# sourceMappingURL=site-import-images.js.map

package/src/lib/site-import/site-import-images.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../../../../../../../libs/plugins/ai/src/lib/site-import/site-import-images.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 * AN IMPORTED PAGE'S PHOTOS, INTO THE SITE'S OWN MEDIA LIBRARY (AGL-3976).\n *\n * Every picture goes through the media library's server door\n * (`pluginMediaIngest().ingest`), which checks the member, the site's\n * lockdown, the bytes' structure, the takedown list and the storage band\n * exactly as an upload does. This module only fetches and names:\n *\n * - **Never a second copy.** A picture's SOURCE KEY is `import:` and a hash\n *   of its address with any fragment and tracking parameter removed. The\n *   library is asked for that key first (`findImportedPhoto`); a logo every\n *   page shows is stored once and placed everywhere. An ingest that cannot\n *   answer the question stores nothing, rather than risk the second copy.\n * - **The safe fetch.** Bytes are read through `fetchPublicTextDetailed`,\n *   the SSRF-guarded fetch the crawler uses (public https only, the socket\n *   pinned, every redirect checked again), as AglynBot, capped in size and\n *   time, and only as JPEG, PNG, WebP or GIF: the raster types the library\n *   takes from a server.\n *=========================================*/\n\nimport type { PluginMediaIngest } from '@aglyn/aglyn/plugin-manager/plugin-media-ingest'\nimport { createHash } from 'node:crypto'\nimport {\n  fetchPublicTextDetailed,\n  type PublicFetchOptions,\n  type PublicFetchResult,\n} from '../jobs/ai-theme-brand-inputs'\nimport type { AiSiteImportPage, AiSiteImportPhoto } from '../model/ai-site-import'\nimport { AGLYN_BOT_USER_AGENT } from './site-import-crawl'\n\n/** The largest picture brought in. */\nexport const SITE_IMPORT_IMAGE_MAX_BYTES = 8 * 1024 * 1024\n\n/** A picture smaller than this is a spacer or a tracking pixel, not a photo. */\nexport const SITE_IMPORT_IMAGE_MIN_BYTES = 1_024\n\n/** One picture's own clock. */\nexport const SITE_IMPORT_IMAGE_TIMEOUT_MS = 8_000\n\nconst RASTER = /^image\\/(?:jpeg|jpg|pjpeg|png|webp|gif)\\b/i\n\nconst NOT_A_PHOTO = /\\.(?:svg|ico|bmp|tiff?|avif|heic)$/i\n\nconst TRACKING_PARAM = /^(?:utm_[a-z]+|fbclid|gclid|msclkid|_ga|ref)$/i\n\nconst EXTENSION: Record<string, string> = {\n  'image/jpeg': 'jpg',\n  'image/jpg': 'jpg',\n  'image/pjpeg': 'jpg',\n  'image/png': 'png',\n  'image/webp': 'webp',\n  'image/gif': 'gif',\n}\n\n/** A picture's address as it is fetched and keyed: https, no fragment or tracking; `null` for one that is not a photo. */\nexport function siteImportImageUrl(raw: string): string | null {\n  let url: URL\n  try {\n    url = new URL(String(raw ?? '').trim())\n  } catch {\n    return null\n  }\n  if (url.protocol === 'http:') url.protocol = 'https:'\n  if (url.protocol !== 'https:' || url.username || url.password) return null\n  if (url.port && url.port !== '443') return null\n  if (NOT_A_PHOTO.test(url.pathname)) return null\n  url.hash = ''\n  for (const key of [...url.searchParams.keys()]) {\n    if (TRACKING_PARAM.test(key)) url.searchParams.delete(key)\n  }\n  return url.toString()\n}\n\n/** The key a site's library holds an imported picture by. */\nexport function siteImportImageKey(url: string): string {\n  return `import:${createHash('sha256').update(url).digest('hex').slice(0, 32)}`\n}\n\n/** A file name for the library, from the picture's address and type. */\nexport function siteImportImageFileName(url: string, contentType: string): string {\n  let base = 'photo'\n  try {\n    const last = decodeURIComponent(new URL(url).pathname.split('/').filter(Boolean).pop() ?? '')\n    base = last.replace(/\\.[a-z0-9]+$/i, '').replace(/[^a-zA-Z0-9 _-]+/g, '-').replace(/-+/g, '-').replace(/^-|-$/g, '') || 'photo'\n  } catch {\n    // An address that does not decode keeps the plain name.\n  }\n  const type = contentType.split(';')[0].trim().toLowerCase()\n  return `${base.slice(0, 80)}.${EXTENSION[type] ?? 'jpg'}`\n}\n\nexport interface SiteImportPhotosInput {\n  hostId: string\n  /** The member the photos are stored as: the job's creator. */\n  uid: string\n  domain: string\n  images: ReadonlyArray<{ url: string; alt: string }>\n  signal?: AbortSignal\n  /** When the pass must stop asking for more. */\n  deadlineAt: number\n}\n\nexport interface SiteImportPhotosDeps {\n  ingest: PluginMediaIngest | null\n  fetch?: (url: string, options: PublicFetchOptions) => Promise<PublicFetchResult>\n  now?: () => number\n}\n\n/**\n * Brings one page's pictures into the site's library, reusing any the\n * library already holds by key. Never throws for anything a picture does.\n */\nexport async function bringInSiteImportPhotos(\n  input: SiteImportPhotosInput,\n  deps: SiteImportPhotosDeps,\n): Promise<NonNullable<AiSiteImportPage['photos']>> {\n  const now = deps.now ?? Date.now\n  const fetchBytes = deps.fetch ?? fetchPublicTextDetailed\n  const assets: AiSiteImportPhoto[] = []\n  let brought = 0\n  let reused = 0\n  let skipped = 0\n  const ingest = deps.ingest\n  // No library door, or one that cannot say what it already holds: nothing\n  // is stored, because a picture stored without that check could be a copy.\n  if (!ingest?.findImportedPhoto) {\n    return { brought, reused, skipped: input.images.length, assets }\n  }\n  const seen = new Set<string>()\n  for (const image of input.images) {\n    const url = siteImportImageUrl(image.url)\n    if (!url || seen.has(url)) {\n      if (!url) skipped += 1\n      continue\n    }\n    seen.add(url)\n    if (input.signal?.aborted || now() >= input.deadlineAt) {\n      skipped += 1\n      continue\n    }\n    const sourceKey = siteImportImageKey(url)\n    try {\n      const held = await ingest.findImportedPhoto({ hostId: input.hostId, sourceKey })\n      if (held) {\n        reused += 1\n        assets.push({\n          mediaId: held.mediaId,\n          src: held.src,\n          ...(held.width ? { width: held.width } : {}),\n          ...(held.height ? { height: held.height } : {}),\n          ...(held.alt || image.alt ? { alt: held.alt || image.alt } : {}),\n        })\n        continue\n      }\n      const fetched = await fetchBytes(url, {\n        maxBytes: SITE_IMPORT_IMAGE_MAX_BYTES,\n        accept: RASTER,\n        acceptHeader: 'image/webp,image/png,image/jpeg,image/gif;q=0.9',\n        userAgent: AGLYN_BOT_USER_AGENT,\n        timeoutMs: Math.max(1, Math.min(SITE_IMPORT_IMAGE_TIMEOUT_MS, input.deadlineAt - now())),\n        refuseOver: true,\n        binary: true,\n        ...(input.signal ? { signal: input.signal } : {}),\n      })\n      if (!fetched.ok || !fetched.data || fetched.data.byteLength < SITE_IMPORT_IMAGE_MIN_BYTES) {\n        skipped += 1\n        continue\n      }\n      const contentType = fetched.contentType.split(';')[0].trim().toLowerCase().replace('image/jpg', 'image/jpeg').replace('image/pjpeg', 'image/jpeg')\n      const stored = await ingest.ingest({\n        hostId: input.hostId,\n        uid: input.uid,\n        fileName: siteImportImageFileName(url, contentType),\n        contentType,\n        bytes: fetched.data,\n        ...(image.alt ? { alt: image.alt.slice(0, 250) } : {}),\n        importedFrom: { key: sourceKey, url, domain: input.domain },\n      })\n      if (!stored.ok) {\n        skipped += 1\n        continue\n      }\n      brought += 1\n      assets.push({\n        mediaId: stored.mediaId,\n        src: stored.src,\n        ...(stored.width ? { width: stored.width } : {}),\n        ...(stored.height ? { height: stored.height } : {}),\n        ...(image.alt ? { alt: image.alt.slice(0, 250) } : {}),\n      })\n    } catch (error) {\n      console.warn('[site-import] a photo could not be brought in', { hostId: input.hostId, error })\n      skipped += 1\n    }\n  }\n  return { brought, reused, skipped, assets }\n}\n"],"names":["createHash","fetchPublicTextDetailed","AGLYN_BOT_USER_AGENT","SITE_IMPORT_IMAGE_MAX_BYTES","SITE_IMPORT_IMAGE_MIN_BYTES","SITE_IMPORT_IMAGE_TIMEOUT_MS","RASTER","NOT_A_PHOTO","TRACKING_PARAM","EXTENSION","siteImportImageUrl","raw","url","URL","String","trim","protocol","username","password","port","test","pathname","hash","key","searchParams","keys","delete","toString","siteImportImageKey","update","digest","slice","siteImportImageFileName","contentType","base","last","decodeURIComponent","split","filter","Boolean","pop","replace","type","toLowerCase","bringInSiteImportPhotos","input","deps","now","Date","fetchBytes","fetch","assets","brought","reused","skipped","ingest","findImportedPhoto","images","length","seen","Set","image","has","add","signal","aborted","deadlineAt","sourceKey","held","hostId","push","mediaId","src","width","height","alt","fetched","maxBytes","accept","acceptHeader","userAgent","timeoutMs","Math","max","min","refuseOver","binary","ok","data","byteLength","stored","uid","fileName","bytes","importedFrom","domain","error","console","warn"],"mappings":"AAAA;;;;;;;;;;;;;;;CAeC,GAED;;;;;;;;;;;;;;;;;;2CAkB2C;AAG3C,SAASA,UAAU,QAAQ,cAAa;AACxC,SACEC,uBAAuB,QAGlB,mCAA+B;AAEtC,SAASC,oBAAoB,QAAQ,yBAAqB;AAE1D,oCAAoC,GACpC,OAAO,MAAMC,8BAA8B,IAAI,OAAO,KAAI;AAE1D,8EAA8E,GAC9E,OAAO,MAAMC,8BAA8B,KAAK;AAEhD,6BAA6B,GAC7B,OAAO,MAAMC,+BAA+B,KAAK;AAEjD,MAAMC,SAAS;AAEf,MAAMC,cAAc;AAEpB,MAAMC,iBAAiB;AAEvB,MAAMC,YAAoC;IACxC,cAAc;IACd,aAAa;IACb,eAAe;IACf,aAAa;IACb,cAAc;IACd,aAAa;AACf;AAEA,wHAAwH,GACxH,OAAO,SAASC,mBAAmBC,GAAW;IAC5C,IAAIC;IACJ,IAAI;QACFA,MAAM,IAAIC,IAAIC,OAAOH,cAAAA,MAAO,IAAII,IAAI;IACtC,EAAE,eAAM;QACN,OAAO;IACT;IACA,IAAIH,IAAII,QAAQ,KAAK,SAASJ,IAAII,QAAQ,GAAG;IAC7C,IAAIJ,IAAII,QAAQ,KAAK,YAAYJ,IAAIK,QAAQ,IAAIL,IAAIM,QAAQ,EAAE,OAAO;IACtE,IAAIN,IAAIO,IAAI,IAAIP,IAAIO,IAAI,KAAK,OAAO,OAAO;IAC3C,IAAIZ,YAAYa,IAAI,CAACR,IAAIS,QAAQ,GAAG,OAAO;IAC3CT,IAAIU,IAAI,GAAG;IACX,KAAK,MAAMC,OAAO;WAAIX,IAAIY,YAAY,CAACC,IAAI;KAAG,CAAE;QAC9C,IAAIjB,eAAeY,IAAI,CAACG,MAAMX,IAAIY,YAAY,CAACE,MAAM,CAACH;IACxD;IACA,OAAOX,IAAIe,QAAQ;AACrB;AAEA,2DAA2D,GAC3D,OAAO,SAASC,mBAAmBhB,GAAW;IAC5C,OAAO,CAAC,OAAO,EAAEZ,WAAW,UAAU6B,MAAM,CAACjB,KAAKkB,MAAM,CAAC,OAAOC,KAAK,CAAC,GAAG,KAAK;AAChF;AAEA,sEAAsE,GACtE,OAAO,SAASC,wBAAwBpB,GAAW,EAAEqB,WAAmB;QASvCxB;IAR/B,IAAIyB,OAAO;IACX,IAAI;YAC8B;QAAhC,MAAMC,OAAOC,oBAAmB,6BAAA,IAAIvB,IAAID,KAAKS,QAAQ,CAACgB,KAAK,CAAC,KAAKC,MAAM,CAACC,SAASC,GAAG,cAApD,6BAA0D;QAC1FN,OAAOC,KAAKM,OAAO,CAAC,iBAAiB,IAAIA,OAAO,CAAC,qBAAqB,KAAKA,OAAO,CAAC,OAAO,KAAKA,OAAO,CAAC,UAAU,OAAO;IAC1H,EAAE,eAAM;IACN,wDAAwD;IAC1D;IACA,MAAMC,OAAOT,YAAYI,KAAK,CAAC,IAAI,CAAC,EAAE,CAACtB,IAAI,GAAG4B,WAAW;IACzD,OAAO,GAAGT,KAAKH,KAAK,CAAC,GAAG,IAAI,CAAC,GAAEtB,kBAAAA,SAAS,CAACiC,KAAK,YAAfjC,kBAAmB,OAAO;AAC3D;AAmBA;;;CAGC,GACD,OAAO,eAAemC,wBACpBC,KAA4B,EAC5BC,IAA0B;QAEdA,WACOA;IADnB,MAAMC,OAAMD,YAAAA,KAAKC,GAAG,YAARD,YAAYE,KAAKD,GAAG;IAChC,MAAME,cAAaH,cAAAA,KAAKI,KAAK,YAAVJ,cAAc7C;IACjC,MAAMkD,SAA8B,EAAE;IACtC,IAAIC,UAAU;IACd,IAAIC,SAAS;IACb,IAAIC,UAAU;IACd,MAAMC,SAAST,KAAKS,MAAM;IAC1B,yEAAyE;IACzE,0EAA0E;IAC1E,IAAI,EAACA,0BAAAA,OAAQC,iBAAiB,GAAE;QAC9B,OAAO;YAAEJ;YAASC;YAAQC,SAAST,MAAMY,MAAM,CAACC,MAAM;YAAEP;QAAO;IACjE;IACA,MAAMQ,OAAO,IAAIC;IACjB,KAAK,MAAMC,SAAShB,MAAMY,MAAM,CAAE;YAO5BZ;QANJ,MAAMjC,MAAMF,mBAAmBmD,MAAMjD,GAAG;QACxC,IAAI,CAACA,OAAO+C,KAAKG,GAAG,CAAClD,MAAM;YACzB,IAAI,CAACA,KAAK0C,WAAW;YACrB;QACF;QACAK,KAAKI,GAAG,CAACnD;QACT,IAAIiC,EAAAA,gBAAAA,MAAMmB,MAAM,qBAAZnB,cAAcoB,OAAO,KAAIlB,SAASF,MAAMqB,UAAU,EAAE;YACtDZ,WAAW;YACX;QACF;QACA,MAAMa,YAAYvC,mBAAmBhB;QACrC,IAAI;YACF,MAAMwD,OAAO,MAAMb,OAAOC,iBAAiB,CAAC;gBAAEa,QAAQxB,MAAMwB,MAAM;gBAAEF;YAAU;YAC9E,IAAIC,MAAM;gBACRf,UAAU;gBACVF,OAAOmB,IAAI,CAAC;oBACVC,SAASH,KAAKG,OAAO;oBACrBC,KAAKJ,KAAKI,GAAG;mBACTJ,KAAKK,KAAK,GAAG;oBAAEA,OAAOL,KAAKK,KAAK;gBAAC,IAAI,CAAC,GACtCL,KAAKM,MAAM,GAAG;oBAAEA,QAAQN,KAAKM,MAAM;gBAAC,IAAI,CAAC,GACzCN,KAAKO,GAAG,IAAId,MAAMc,GAAG,GAAG;oBAAEA,KAAKP,KAAKO,GAAG,IAAId,MAAMc,GAAG;gBAAC,IAAI,CAAC;gBAEhE;YACF;YACA,MAAMC,UAAU,MAAM3B,WAAWrC,KAAK;gBACpCiE,UAAU1E;gBACV2E,QAAQxE;gBACRyE,cAAc;gBACdC,WAAW9E;gBACX+E,WAAWC,KAAKC,GAAG,CAAC,GAAGD,KAAKE,GAAG,CAAC/E,8BAA8BwC,MAAMqB,UAAU,GAAGnB;gBACjFsC,YAAY;gBACZC,QAAQ;eACJzC,MAAMmB,MAAM,GAAG;gBAAEA,QAAQnB,MAAMmB,MAAM;YAAC,IAAI,CAAC;YAEjD,IAAI,CAACY,QAAQW,EAAE,IAAI,CAACX,QAAQY,IAAI,IAAIZ,QAAQY,IAAI,CAACC,UAAU,GAAGrF,6BAA6B;gBACzFkD,WAAW;gBACX;YACF;YACA,MAAMrB,cAAc2C,QAAQ3C,WAAW,CAACI,KAAK,CAAC,IAAI,CAAC,EAAE,CAACtB,IAAI,GAAG4B,WAAW,GAAGF,OAAO,CAAC,aAAa,cAAcA,OAAO,CAAC,eAAe;YACrI,MAAMiD,SAAS,MAAMnC,OAAOA,MAAM,CAAC;gBACjCc,QAAQxB,MAAMwB,MAAM;gBACpBsB,KAAK9C,MAAM8C,GAAG;gBACdC,UAAU5D,wBAAwBpB,KAAKqB;gBACvCA;gBACA4D,OAAOjB,QAAQY,IAAI;eACf3B,MAAMc,GAAG,GAAG;gBAAEA,KAAKd,MAAMc,GAAG,CAAC5C,KAAK,CAAC,GAAG;YAAK,IAAI,CAAC;gBACpD+D,cAAc;oBAAEvE,KAAK4C;oBAAWvD;oBAAKmF,QAAQlD,MAAMkD,MAAM;gBAAC;;YAE5D,IAAI,CAACL,OAAOH,EAAE,EAAE;gBACdjC,WAAW;gBACX;YACF;YACAF,WAAW;YACXD,OAAOmB,IAAI,CAAC;gBACVC,SAASmB,OAAOnB,OAAO;gBACvBC,KAAKkB,OAAOlB,GAAG;eACXkB,OAAOjB,KAAK,GAAG;gBAAEA,OAAOiB,OAAOjB,KAAK;YAAC,IAAI,CAAC,GAC1CiB,OAAOhB,MAAM,GAAG;gBAAEA,QAAQgB,OAAOhB,MAAM;YAAC,IAAI,CAAC,GAC7Cb,MAAMc,GAAG,GAAG;gBAAEA,KAAKd,MAAMc,GAAG,CAAC5C,KAAK,CAAC,GAAG;YAAK,IAAI,CAAC;QAExD,EAAE,OAAOiE,OAAO;YACdC,QAAQC,IAAI,CAAC,iDAAiD;gBAAE7B,QAAQxB,MAAMwB,MAAM;gBAAE2B;YAAM;YAC5F1C,WAAW;QACb;IACF;IACA,OAAO;QAAEF;QAASC;QAAQC;QAASH;IAAO;AAC5C"}
+{"version":3,"sources":["../../../../../../../libs/plugins/ai/src/lib/site-import/site-import-images.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 * AN IMPORTED PAGE'S PHOTOS, INTO THE SITE'S OWN MEDIA LIBRARY (AGL-3976).\n *\n * Every picture goes through the media library's server door\n * (`pluginMediaIngest().ingest`), which checks the member, the site's\n * lockdown, the bytes' structure, the takedown list and the storage band\n * exactly as an upload does. This module only fetches and names:\n *\n * - **Never a second copy.** A picture's SOURCE KEY is `import:` and a hash\n *   of its address with any fragment and tracking parameter removed. The\n *   library is asked for that key first (`findImportedPhoto`); a logo every\n *   page shows is stored once and placed everywhere. An ingest that cannot\n *   answer the question stores nothing, rather than risk the second copy.\n * - **The safe fetch.** Bytes are read through `fetchPublicTextDetailed`,\n *   the SSRF-guarded fetch the crawler uses (public https only, the socket\n *   pinned, every redirect checked again), as AglynBot, capped in size and\n *   time, and only as JPEG, PNG, WebP or GIF: the raster types the library\n *   takes from a server.\n *=========================================*/\n\nimport type { PluginMediaIngest } from '@aglyn/aglyn/plugin-manager/plugin-media-ingest'\nimport { createHash } from 'node:crypto'\nimport {\n  fetchPublicTextDetailed,\n  type PublicFetchOptions,\n  type PublicFetchResult,\n} from '../jobs/ai-theme-brand-inputs'\nimport type { AiSiteImportPage, AiSiteImportPhoto } from '../model/ai-site-import'\nimport { AGLYN_BOT_USER_AGENT, type SiteImportPage } from './site-import-crawl'\n\n/** The largest picture brought in. */\nexport const SITE_IMPORT_IMAGE_MAX_BYTES = 8 * 1024 * 1024\n\n/** A picture smaller than this is a spacer or a tracking pixel, not a photo. */\nexport const SITE_IMPORT_IMAGE_MIN_BYTES = 1_024\n\n/** One picture's own clock. */\nexport const SITE_IMPORT_IMAGE_TIMEOUT_MS = 8_000\n\nconst RASTER = /^image\\/(?:jpeg|jpg|pjpeg|png|webp|gif)\\b/i\n\nconst NOT_A_PHOTO = /\\.(?:svg|ico|bmp|tiff?|avif|heic)$/i\n\nconst TRACKING_PARAM = /^(?:utm_[a-z]+|fbclid|gclid|msclkid|_ga|ref)$/i\n\nconst EXTENSION: Record<string, string> = {\n  'image/jpeg': 'jpg',\n  'image/jpg': 'jpg',\n  'image/pjpeg': 'jpg',\n  'image/png': 'png',\n  'image/webp': 'webp',\n  'image/gif': 'gif',\n}\n\n/** A picture's address as it is fetched and keyed: https, no fragment or tracking; `null` for one that is not a photo. */\nexport function siteImportImageUrl(raw: string): string | null {\n  let url: URL\n  try {\n    url = new URL(String(raw ?? '').trim())\n  } catch {\n    return null\n  }\n  if (url.protocol === 'http:') url.protocol = 'https:'\n  if (url.protocol !== 'https:' || url.username || url.password) return null\n  if (url.port && url.port !== '443') return null\n  if (NOT_A_PHOTO.test(url.pathname)) return null\n  url.hash = ''\n  for (const key of [...url.searchParams.keys()]) {\n    if (TRACKING_PARAM.test(key)) url.searchParams.delete(key)\n  }\n  return url.toString()\n}\n\n/** The key a site's library holds an imported picture by. */\nexport function siteImportImageKey(url: string): string {\n  return `import:${createHash('sha256').update(url).digest('hex').slice(0, 32)}`\n}\n\n/** A file name for the library, from the picture's address and type. */\nexport function siteImportImageFileName(url: string, contentType: string): string {\n  let base = 'photo'\n  try {\n    const last = decodeURIComponent(new URL(url).pathname.split('/').filter(Boolean).pop() ?? '')\n    base = last.replace(/\\.[a-z0-9]+$/i, '').replace(/[^a-zA-Z0-9 _-]+/g, '-').replace(/-+/g, '-').replace(/^-|-$/g, '') || 'photo'\n  } catch {\n    // An address that does not decode keeps the plain name.\n  }\n  const type = contentType.split(';')[0].trim().toLowerCase()\n  return `${base.slice(0, 80)}.${EXTENSION[type] ?? 'jpg'}`\n}\n\nexport interface SiteImportPhotosInput {\n  hostId: string\n  /** The member the photos are stored as: the job's creator. */\n  uid: string\n  domain: string\n  images: ReadonlyArray<{ url: string; alt: string }>\n  signal?: AbortSignal\n  /** When the pass must stop asking for more. */\n  deadlineAt: number\n}\n\nexport interface SiteImportPhotosDeps {\n  ingest: PluginMediaIngest | null\n  fetch?: (url: string, options: PublicFetchOptions) => Promise<PublicFetchResult>\n  now?: () => number\n}\n\n/**\n * Brings one page's pictures into the site's library, reusing any the\n * library already holds by key. Never throws for anything a picture does.\n */\nexport async function bringInSiteImportPhotos(\n  input: SiteImportPhotosInput,\n  deps: SiteImportPhotosDeps,\n): Promise<NonNullable<AiSiteImportPage['photos']>> {\n  const now = deps.now ?? Date.now\n  const fetchBytes = deps.fetch ?? fetchPublicTextDetailed\n  const assets: AiSiteImportPhoto[] = []\n  let brought = 0\n  let reused = 0\n  let skipped = 0\n  const ingest = deps.ingest\n  // No library door, or one that cannot say what it already holds: nothing\n  // is stored, because a picture stored without that check could be a copy.\n  if (!ingest?.findImportedPhoto) {\n    return { brought, reused, skipped: input.images.length, assets }\n  }\n  const seen = new Set<string>()\n  for (const image of input.images) {\n    const url = siteImportImageUrl(image.url)\n    if (!url || seen.has(url)) {\n      if (!url) skipped += 1\n      continue\n    }\n    seen.add(url)\n    if (input.signal?.aborted || now() >= input.deadlineAt) {\n      skipped += 1\n      continue\n    }\n    const sourceKey = siteImportImageKey(url)\n    try {\n      const held = await ingest.findImportedPhoto({ hostId: input.hostId, sourceKey })\n      if (held) {\n        reused += 1\n        assets.push({\n          mediaId: held.mediaId,\n          src: held.src,\n          ...(held.width ? { width: held.width } : {}),\n          ...(held.height ? { height: held.height } : {}),\n          ...(held.alt || image.alt ? { alt: held.alt || image.alt } : {}),\n          from: image.url,\n        })\n        continue\n      }\n      const fetched = await fetchBytes(url, {\n        maxBytes: SITE_IMPORT_IMAGE_MAX_BYTES,\n        accept: RASTER,\n        acceptHeader: 'image/webp,image/png,image/jpeg,image/gif;q=0.9',\n        userAgent: AGLYN_BOT_USER_AGENT,\n        timeoutMs: Math.max(1, Math.min(SITE_IMPORT_IMAGE_TIMEOUT_MS, input.deadlineAt - now())),\n        refuseOver: true,\n        binary: true,\n        ...(input.signal ? { signal: input.signal } : {}),\n      })\n      if (!fetched.ok || !fetched.data || fetched.data.byteLength < SITE_IMPORT_IMAGE_MIN_BYTES) {\n        skipped += 1\n        continue\n      }\n      const contentType = fetched.contentType.split(';')[0].trim().toLowerCase().replace('image/jpg', 'image/jpeg').replace('image/pjpeg', 'image/jpeg')\n      const stored = await ingest.ingest({\n        hostId: input.hostId,\n        uid: input.uid,\n        fileName: siteImportImageFileName(url, contentType),\n        contentType,\n        bytes: fetched.data,\n        ...(image.alt ? { alt: image.alt.slice(0, 250) } : {}),\n        importedFrom: { key: sourceKey, url, domain: input.domain },\n      })\n      if (!stored.ok) {\n        skipped += 1\n        continue\n      }\n      brought += 1\n      assets.push({\n        mediaId: stored.mediaId,\n        src: stored.src,\n        ...(stored.width ? { width: stored.width } : {}),\n        ...(stored.height ? { height: stored.height } : {}),\n        ...(image.alt ? { alt: image.alt.slice(0, 250) } : {}),\n        // Where it sat on the old page, so a page copied as it is places it there (AGL-4084).\n        from: image.url,\n      })\n    } catch (error) {\n      console.warn('[site-import] a photo could not be brought in', { hostId: input.hostId, error })\n      skipped += 1\n    }\n  }\n  return { brought, reused, skipped, assets }\n}\n\n/*------------------------------------------\n * The site's logo (AGL-4083)\n *-----------------------------------------*/\n\n/**\n * A picture that may be the site's logo, where it was found. `type` is what\n * its address or its declared type says: `svg`, a raster, or `null` when\n * neither says (a media library address with no extension); the ingest\n * reads the answer's own type either way.\n */\nexport interface SiteImportLogoCandidate {\n  url: string\n  alt: string\n  source: 'header' | 'footer' | 'structured-data' | 'apple-touch-icon' | 'icon'\n  type: 'svg' | 'raster' | null\n}\n\n/** A logo's address as it is fetched: https, no fragment or tracking parameter. An SVG is a logo, unlike a photo. */\nexport function siteImportLogoUrl(raw: string): string | null {\n  let url: URL\n  try {\n    url = new URL(String(raw ?? '').trim())\n  } catch {\n    return null\n  }\n  if (url.protocol === 'http:') url.protocol = 'https:'\n  if (url.protocol !== 'https:' || url.username || url.password) return null\n  if (url.port && url.port !== '443') return null\n  url.hash = ''\n  for (const key of [...url.searchParams.keys()]) {\n    if (TRACKING_PARAM.test(key)) url.searchParams.delete(key)\n  }\n  return url.toString()\n}\n\nfunction logoTypeOf(url: string, declared?: string | null): SiteImportLogoCandidate['type'] {\n  const type = (declared ?? '').toLowerCase()\n  if (type.includes('svg')) return 'svg'\n  if (/^image\\/(?:png|jpe?g|webp|gif|x-icon|vnd\\.microsoft\\.icon)/.test(type)) return 'raster'\n  let path: string\n  try {\n    path = new URL(url).pathname.toLowerCase()\n  } catch {\n    return null\n  }\n  if (path.endsWith('.svg')) return 'svg'\n  if (/\\.(?:png|jpe?g|webp|gif|ico)$/.test(path)) return 'raster'\n  return null\n}\n\n/**\n * The pictures that may be the site's logo (AGL-4083), the likeliest first:\n * the picture in the header's link home, any header picture that says it is\n * the logo or the site's name, the footer's, the structured data's logo,\n * then the largest touch icon and the largest icon. Each address once.\n */\nexport function siteImportLogoCandidates(\n  page: Pick<SiteImportPage, 'brand' | 'images' | 'icons' | 'siteName' | 'organization' | 'title'>,\n): SiteImportLogoCandidate[] {\n  const name = (page.siteName || page.organization?.name || '').trim().toLowerCase()\n  const says = (image: { url: string; alt: string }) => {\n    const alt = image.alt.trim().toLowerCase()\n    return /\\blogo\\b/i.test(image.alt) || /logo/i.test(image.url) || (!!name && (alt === name || alt === `${name} logo`))\n  }\n  const out: SiteImportLogoCandidate[] = []\n  const add = (raw: string, alt: string, source: SiteImportLogoCandidate['source'], declared?: string | null) => {\n    const url = siteImportLogoUrl(raw)\n    if (!url || out.some((entry) => entry.url === url)) return\n    out.push({ url, alt: alt.slice(0, 250), source, type: logoTypeOf(url, declared) })\n  }\n  if (page.brand?.logo && page.brand.region !== 'footer') add(page.brand.logo.url, page.brand.logo.alt, 'header')\n  for (const image of page.images) if (image.region === 'header' && says(image)) add(image.url, image.alt, 'header')\n  for (const image of page.images) if (image.region === 'footer' && says(image)) add(image.url, image.alt, 'footer')\n  if (page.organization?.logo) add(page.organization.logo, page.organization.name, 'structured-data')\n  const touch = page.icons.filter((icon) => icon.rel === 'apple-touch-icon').sort((a, b) => (b.size ?? 0) - (a.size ?? 0))[0]\n  if (touch) add(touch.url, '', 'apple-touch-icon', touch.type)\n  const icon = page.icons.filter((entry) => entry.rel === 'icon').sort((a, b) => (b.size ?? 0) - (a.size ?? 0))[0]\n  if (icon) add(icon.url, '', 'icon', icon.type)\n  return out\n}\n"],"names":["createHash","fetchPublicTextDetailed","AGLYN_BOT_USER_AGENT","SITE_IMPORT_IMAGE_MAX_BYTES","SITE_IMPORT_IMAGE_MIN_BYTES","SITE_IMPORT_IMAGE_TIMEOUT_MS","RASTER","NOT_A_PHOTO","TRACKING_PARAM","EXTENSION","siteImportImageUrl","raw","url","URL","String","trim","protocol","username","password","port","test","pathname","hash","key","searchParams","keys","delete","toString","siteImportImageKey","update","digest","slice","siteImportImageFileName","contentType","base","last","decodeURIComponent","split","filter","Boolean","pop","replace","type","toLowerCase","bringInSiteImportPhotos","input","deps","now","Date","fetchBytes","fetch","assets","brought","reused","skipped","ingest","findImportedPhoto","images","length","seen","Set","image","has","add","signal","aborted","deadlineAt","sourceKey","held","hostId","push","mediaId","src","width","height","alt","from","fetched","maxBytes","accept","acceptHeader","userAgent","timeoutMs","Math","max","min","refuseOver","binary","ok","data","byteLength","stored","uid","fileName","bytes","importedFrom","domain","error","console","warn","siteImportLogoUrl","logoTypeOf","declared","includes","path","endsWith","siteImportLogoCandidates","page","name","siteName","organization","says","out","source","some","entry","brand","logo","region","touch","icons","icon","rel","sort","a","b","size"],"mappings":"AAAA;;;;;;;;;;;;;;;CAeC,GAED;;;;;;;;;;;;;;;;;;2CAkB2C;AAG3C,SAASA,UAAU,QAAQ,cAAa;AACxC,SACEC,uBAAuB,QAGlB,mCAA+B;AAEtC,SAASC,oBAAoB,QAA6B,yBAAqB;AAE/E,oCAAoC,GACpC,OAAO,MAAMC,8BAA8B,IAAI,OAAO,KAAI;AAE1D,8EAA8E,GAC9E,OAAO,MAAMC,8BAA8B,KAAK;AAEhD,6BAA6B,GAC7B,OAAO,MAAMC,+BAA+B,KAAK;AAEjD,MAAMC,SAAS;AAEf,MAAMC,cAAc;AAEpB,MAAMC,iBAAiB;AAEvB,MAAMC,YAAoC;IACxC,cAAc;IACd,aAAa;IACb,eAAe;IACf,aAAa;IACb,cAAc;IACd,aAAa;AACf;AAEA,wHAAwH,GACxH,OAAO,SAASC,mBAAmBC,GAAW;IAC5C,IAAIC;IACJ,IAAI;QACFA,MAAM,IAAIC,IAAIC,OAAOH,cAAAA,MAAO,IAAII,IAAI;IACtC,EAAE,eAAM;QACN,OAAO;IACT;IACA,IAAIH,IAAII,QAAQ,KAAK,SAASJ,IAAII,QAAQ,GAAG;IAC7C,IAAIJ,IAAII,QAAQ,KAAK,YAAYJ,IAAIK,QAAQ,IAAIL,IAAIM,QAAQ,EAAE,OAAO;IACtE,IAAIN,IAAIO,IAAI,IAAIP,IAAIO,IAAI,KAAK,OAAO,OAAO;IAC3C,IAAIZ,YAAYa,IAAI,CAACR,IAAIS,QAAQ,GAAG,OAAO;IAC3CT,IAAIU,IAAI,GAAG;IACX,KAAK,MAAMC,OAAO;WAAIX,IAAIY,YAAY,CAACC,IAAI;KAAG,CAAE;QAC9C,IAAIjB,eAAeY,IAAI,CAACG,MAAMX,IAAIY,YAAY,CAACE,MAAM,CAACH;IACxD;IACA,OAAOX,IAAIe,QAAQ;AACrB;AAEA,2DAA2D,GAC3D,OAAO,SAASC,mBAAmBhB,GAAW;IAC5C,OAAO,CAAC,OAAO,EAAEZ,WAAW,UAAU6B,MAAM,CAACjB,KAAKkB,MAAM,CAAC,OAAOC,KAAK,CAAC,GAAG,KAAK;AAChF;AAEA,sEAAsE,GACtE,OAAO,SAASC,wBAAwBpB,GAAW,EAAEqB,WAAmB;QASvCxB;IAR/B,IAAIyB,OAAO;IACX,IAAI;YAC8B;QAAhC,MAAMC,OAAOC,oBAAmB,6BAAA,IAAIvB,IAAID,KAAKS,QAAQ,CAACgB,KAAK,CAAC,KAAKC,MAAM,CAACC,SAASC,GAAG,cAApD,6BAA0D;QAC1FN,OAAOC,KAAKM,OAAO,CAAC,iBAAiB,IAAIA,OAAO,CAAC,qBAAqB,KAAKA,OAAO,CAAC,OAAO,KAAKA,OAAO,CAAC,UAAU,OAAO;IAC1H,EAAE,eAAM;IACN,wDAAwD;IAC1D;IACA,MAAMC,OAAOT,YAAYI,KAAK,CAAC,IAAI,CAAC,EAAE,CAACtB,IAAI,GAAG4B,WAAW;IACzD,OAAO,GAAGT,KAAKH,KAAK,CAAC,GAAG,IAAI,CAAC,GAAEtB,kBAAAA,SAAS,CAACiC,KAAK,YAAfjC,kBAAmB,OAAO;AAC3D;AAmBA;;;CAGC,GACD,OAAO,eAAemC,wBACpBC,KAA4B,EAC5BC,IAA0B;QAEdA,WACOA;IADnB,MAAMC,OAAMD,YAAAA,KAAKC,GAAG,YAARD,YAAYE,KAAKD,GAAG;IAChC,MAAME,cAAaH,cAAAA,KAAKI,KAAK,YAAVJ,cAAc7C;IACjC,MAAMkD,SAA8B,EAAE;IACtC,IAAIC,UAAU;IACd,IAAIC,SAAS;IACb,IAAIC,UAAU;IACd,MAAMC,SAAST,KAAKS,MAAM;IAC1B,yEAAyE;IACzE,0EAA0E;IAC1E,IAAI,EAACA,0BAAAA,OAAQC,iBAAiB,GAAE;QAC9B,OAAO;YAAEJ;YAASC;YAAQC,SAAST,MAAMY,MAAM,CAACC,MAAM;YAAEP;QAAO;IACjE;IACA,MAAMQ,OAAO,IAAIC;IACjB,KAAK,MAAMC,SAAShB,MAAMY,MAAM,CAAE;YAO5BZ;QANJ,MAAMjC,MAAMF,mBAAmBmD,MAAMjD,GAAG;QACxC,IAAI,CAACA,OAAO+C,KAAKG,GAAG,CAAClD,MAAM;YACzB,IAAI,CAACA,KAAK0C,WAAW;YACrB;QACF;QACAK,KAAKI,GAAG,CAACnD;QACT,IAAIiC,EAAAA,gBAAAA,MAAMmB,MAAM,qBAAZnB,cAAcoB,OAAO,KAAIlB,SAASF,MAAMqB,UAAU,EAAE;YACtDZ,WAAW;YACX;QACF;QACA,MAAMa,YAAYvC,mBAAmBhB;QACrC,IAAI;YACF,MAAMwD,OAAO,MAAMb,OAAOC,iBAAiB,CAAC;gBAAEa,QAAQxB,MAAMwB,MAAM;gBAAEF;YAAU;YAC9E,IAAIC,MAAM;gBACRf,UAAU;gBACVF,OAAOmB,IAAI,CAAC;oBACVC,SAASH,KAAKG,OAAO;oBACrBC,KAAKJ,KAAKI,GAAG;mBACTJ,KAAKK,KAAK,GAAG;oBAAEA,OAAOL,KAAKK,KAAK;gBAAC,IAAI,CAAC,GACtCL,KAAKM,MAAM,GAAG;oBAAEA,QAAQN,KAAKM,MAAM;gBAAC,IAAI,CAAC,GACzCN,KAAKO,GAAG,IAAId,MAAMc,GAAG,GAAG;oBAAEA,KAAKP,KAAKO,GAAG,IAAId,MAAMc,GAAG;gBAAC,IAAI,CAAC;oBAC9DC,MAAMf,MAAMjD,GAAG;;gBAEjB;YACF;YACA,MAAMiE,UAAU,MAAM5B,WAAWrC,KAAK;gBACpCkE,UAAU3E;gBACV4E,QAAQzE;gBACR0E,cAAc;gBACdC,WAAW/E;gBACXgF,WAAWC,KAAKC,GAAG,CAAC,GAAGD,KAAKE,GAAG,CAAChF,8BAA8BwC,MAAMqB,UAAU,GAAGnB;gBACjFuC,YAAY;gBACZC,QAAQ;eACJ1C,MAAMmB,MAAM,GAAG;gBAAEA,QAAQnB,MAAMmB,MAAM;YAAC,IAAI,CAAC;YAEjD,IAAI,CAACa,QAAQW,EAAE,IAAI,CAACX,QAAQY,IAAI,IAAIZ,QAAQY,IAAI,CAACC,UAAU,GAAGtF,6BAA6B;gBACzFkD,WAAW;gBACX;YACF;YACA,MAAMrB,cAAc4C,QAAQ5C,WAAW,CAACI,KAAK,CAAC,IAAI,CAAC,EAAE,CAACtB,IAAI,GAAG4B,WAAW,GAAGF,OAAO,CAAC,aAAa,cAAcA,OAAO,CAAC,eAAe;YACrI,MAAMkD,SAAS,MAAMpC,OAAOA,MAAM,CAAC;gBACjCc,QAAQxB,MAAMwB,MAAM;gBACpBuB,KAAK/C,MAAM+C,GAAG;gBACdC,UAAU7D,wBAAwBpB,KAAKqB;gBACvCA;gBACA6D,OAAOjB,QAAQY,IAAI;eACf5B,MAAMc,GAAG,GAAG;gBAAEA,KAAKd,MAAMc,GAAG,CAAC5C,KAAK,CAAC,GAAG;YAAK,IAAI,CAAC;gBACpDgE,cAAc;oBAAExE,KAAK4C;oBAAWvD;oBAAKoF,QAAQnD,MAAMmD,MAAM;gBAAC;;YAE5D,IAAI,CAACL,OAAOH,EAAE,EAAE;gBACdlC,WAAW;gBACX;YACF;YACAF,WAAW;YACXD,OAAOmB,IAAI,CAAC;gBACVC,SAASoB,OAAOpB,OAAO;gBACvBC,KAAKmB,OAAOnB,GAAG;eACXmB,OAAOlB,KAAK,GAAG;gBAAEA,OAAOkB,OAAOlB,KAAK;YAAC,IAAI,CAAC,GAC1CkB,OAAOjB,MAAM,GAAG;gBAAEA,QAAQiB,OAAOjB,MAAM;YAAC,IAAI,CAAC,GAC7Cb,MAAMc,GAAG,GAAG;gBAAEA,KAAKd,MAAMc,GAAG,CAAC5C,KAAK,CAAC,GAAG;YAAK,IAAI,CAAC;gBACpD,sFAAsF;gBACtF6C,MAAMf,MAAMjD,GAAG;;QAEnB,EAAE,OAAOqF,OAAO;YACdC,QAAQC,IAAI,CAAC,iDAAiD;gBAAE9B,QAAQxB,MAAMwB,MAAM;gBAAE4B;YAAM;YAC5F3C,WAAW;QACb;IACF;IACA,OAAO;QAAEF;QAASC;QAAQC;QAASH;IAAO;AAC5C;AAmBA,mHAAmH,GACnH,OAAO,SAASiD,kBAAkBzF,GAAW;IAC3C,IAAIC;IACJ,IAAI;QACFA,MAAM,IAAIC,IAAIC,OAAOH,cAAAA,MAAO,IAAII,IAAI;IACtC,EAAE,eAAM;QACN,OAAO;IACT;IACA,IAAIH,IAAII,QAAQ,KAAK,SAASJ,IAAII,QAAQ,GAAG;IAC7C,IAAIJ,IAAII,QAAQ,KAAK,YAAYJ,IAAIK,QAAQ,IAAIL,IAAIM,QAAQ,EAAE,OAAO;IACtE,IAAIN,IAAIO,IAAI,IAAIP,IAAIO,IAAI,KAAK,OAAO,OAAO;IAC3CP,IAAIU,IAAI,GAAG;IACX,KAAK,MAAMC,OAAO;WAAIX,IAAIY,YAAY,CAACC,IAAI;KAAG,CAAE;QAC9C,IAAIjB,eAAeY,IAAI,CAACG,MAAMX,IAAIY,YAAY,CAACE,MAAM,CAACH;IACxD;IACA,OAAOX,IAAIe,QAAQ;AACrB;AAEA,SAAS0E,WAAWzF,GAAW,EAAE0F,QAAwB;IACvD,MAAM5D,OAAO,CAAC4D,mBAAAA,WAAY,IAAI3D,WAAW;IACzC,IAAID,KAAK6D,QAAQ,CAAC,QAAQ,OAAO;IACjC,IAAI,6DAA6DnF,IAAI,CAACsB,OAAO,OAAO;IACpF,IAAI8D;IACJ,IAAI;QACFA,OAAO,IAAI3F,IAAID,KAAKS,QAAQ,CAACsB,WAAW;IAC1C,EAAE,eAAM;QACN,OAAO;IACT;IACA,IAAI6D,KAAKC,QAAQ,CAAC,SAAS,OAAO;IAClC,IAAI,gCAAgCrF,IAAI,CAACoF,OAAO,OAAO;IACvD,OAAO;AACT;AAEA;;;;;CAKC,GACD,OAAO,SAASE,yBACdC,IAAgG;QAEjEA,oBAW3BA,aAGAA;IAdJ,MAAMC,OAAO,AAACD,CAAAA,KAAKE,QAAQ,MAAIF,qBAAAA,KAAKG,YAAY,qBAAjBH,mBAAmBC,IAAI,KAAI,EAAC,EAAG7F,IAAI,GAAG4B,WAAW;IAChF,MAAMoE,OAAO,CAAClD;QACZ,MAAMc,MAAMd,MAAMc,GAAG,CAAC5D,IAAI,GAAG4B,WAAW;QACxC,OAAO,YAAYvB,IAAI,CAACyC,MAAMc,GAAG,KAAK,QAAQvD,IAAI,CAACyC,MAAMjD,GAAG,KAAM,CAAC,CAACgG,QAASjC,CAAAA,QAAQiC,QAAQjC,QAAQ,GAAGiC,KAAK,KAAK,CAAC,AAAD;IACpH;IACA,MAAMI,MAAiC,EAAE;IACzC,MAAMjD,MAAM,CAACpD,KAAagE,KAAasC,QAA2CX;QAChF,MAAM1F,MAAMwF,kBAAkBzF;QAC9B,IAAI,CAACC,OAAOoG,IAAIE,IAAI,CAAC,CAACC,QAAUA,MAAMvG,GAAG,KAAKA,MAAM;QACpDoG,IAAI1C,IAAI,CAAC;YAAE1D;YAAK+D,KAAKA,IAAI5C,KAAK,CAAC,GAAG;YAAMkF;YAAQvE,MAAM2D,WAAWzF,KAAK0F;QAAU;IAClF;IACA,IAAIK,EAAAA,cAAAA,KAAKS,KAAK,qBAAVT,YAAYU,IAAI,KAAIV,KAAKS,KAAK,CAACE,MAAM,KAAK,UAAUvD,IAAI4C,KAAKS,KAAK,CAACC,IAAI,CAACzG,GAAG,EAAE+F,KAAKS,KAAK,CAACC,IAAI,CAAC1C,GAAG,EAAE;IACtG,KAAK,MAAMd,SAAS8C,KAAKlD,MAAM,CAAE,IAAII,MAAMyD,MAAM,KAAK,YAAYP,KAAKlD,QAAQE,IAAIF,MAAMjD,GAAG,EAAEiD,MAAMc,GAAG,EAAE;IACzG,KAAK,MAAMd,SAAS8C,KAAKlD,MAAM,CAAE,IAAII,MAAMyD,MAAM,KAAK,YAAYP,KAAKlD,QAAQE,IAAIF,MAAMjD,GAAG,EAAEiD,MAAMc,GAAG,EAAE;IACzG,KAAIgC,sBAAAA,KAAKG,YAAY,qBAAjBH,oBAAmBU,IAAI,EAAEtD,IAAI4C,KAAKG,YAAY,CAACO,IAAI,EAAEV,KAAKG,YAAY,CAACF,IAAI,EAAE;IACjF,MAAMW,QAAQZ,KAAKa,KAAK,CAAClF,MAAM,CAAC,CAACmF,OAASA,KAAKC,GAAG,KAAK,oBAAoBC,IAAI,CAAC,CAACC,GAAGC;YAAOA,SAAgBD;eAAjB,EAACC,UAAAA,EAAEC,IAAI,YAAND,UAAU,OAAMD,UAAAA,EAAEE,IAAI,YAANF,UAAU;MAAG,CAAC,EAAE;IAC3H,IAAIL,OAAOxD,IAAIwD,MAAM3G,GAAG,EAAE,IAAI,oBAAoB2G,MAAM7E,IAAI;IAC5D,MAAM+E,OAAOd,KAAKa,KAAK,CAAClF,MAAM,CAAC,CAAC6E,QAAUA,MAAMO,GAAG,KAAK,QAAQC,IAAI,CAAC,CAACC,GAAGC;YAAOA,SAAgBD;eAAjB,EAACC,UAAAA,EAAEC,IAAI,YAAND,UAAU,OAAMD,UAAAA,EAAEE,IAAI,YAANF,UAAU;MAAG,CAAC,EAAE;IAChH,IAAIH,MAAM1D,IAAI0D,KAAK7G,GAAG,EAAE,IAAI,QAAQ6G,KAAK/E,IAAI;IAC7C,OAAOsE;AACT"}

package/src/lib/site-import/site-import-page.d.ts

@@ -16,9 +16,11 @@
  */
 import { type AiSiteImportLeftOutReason, type AiSiteImportListedPage, type AiSiteImportPageRead } from '../model/ai-site-import';
 import type { SiteImportPage, SiteImportSkipReason } from './site-import-crawl';
+import type { SiteImportBlock, SiteImportOrganization } from './site-import-html';
+import { type SiteImportLogoCandidate } from './site-import-images';
 /** A line of untrusted text safe to show outside a fence: neutralized, one line, cut short. */
 export declare function siteImportLabel(text: string, max?: number): string;
-/** How many words a page says, its title and headings included. */
+/** How many words a page says, its title and headings included; a link inside a sentence is counted once. */
 export declare function siteImportPageWords(page: Pick<SiteImportPage, 'title' | 'blocks'>): number;
 /**
  * A page's name on the new site: "Home" for the home page, else its browser
@@ -42,7 +44,47 @@ export declare function siteImportPageImages(page: Pick<SiteImportPage, 'blocks'
 export declare function siteImportPageSections(page: Pick<SiteImportPage, 'title' | 'blocks'>): string[];
 /** The page's text in order, as a prompt shows it inside its fence. */
 export declare function siteImportPageText(page: Pick<SiteImportPage, 'title' | 'description' | 'blocks'>): string;
-/** The page as a wave reads it. */
+/** The page as a wave reads it: its main region's blocks whole, within the read's budget. */
 export declare function siteImportPageRead(page: SiteImportPage, at: string): AiSiteImportPageRead;
+/**
+ * How the source's header draws its brand: its logo alone (aglyn.com), its
+ * logo with its name beside it, or its name alone. `text` is the name as the
+ * header prints it, or the site's own name when it prints nothing at all.
+ */
+export interface SiteImportHeaderBrand {
+    kind: 'logo-only' | 'logo-and-name' | 'name-only';
+    logo: SiteImportLogoCandidate | null;
+    /** Whether the header prints the name as words. */
+    name: boolean;
+    text: string;
+}
+/** The site's own name: its head's, its structured data's, else its home page's title before a separator. */
+export declare function siteImportSiteName(home: Pick<SiteImportPage, 'siteName' | 'organization' | 'title'>): string;
+/**
+ * The home page's header brand (AGL-4083), as the source shows it: the link
+ * home in its header, the picture inside it, and whether words sit beside
+ * the picture, inside the link or next to it. A header with no picture link
+ * home is the name alone.
+ */
+export declare function siteImportHeaderBrand(home: Pick<SiteImportPage, 'brand' | 'images' | 'icons' | 'siteName' | 'organization' | 'title' | 'blocks'>): SiteImportHeaderBrand;
+/** The facts a site's header and footer carry, as the source writes them. */
+export interface SiteImportSiteFacts {
+    header: SiteImportBlock[];
+    footer: SiteImportBlock[];
+    organization: SiteImportOrganization | null;
+    /** Email addresses, from its structured data and its `mailto:` links. */
+    emails: string[];
+    /** Phone numbers as written, from its structured data and its `tel:` links. */
+    phones: string[];
+    /** Its profiles on other sites: its structured data's `sameAs` and its links to them. */
+    social: string[];
+}
+/**
+ * The home page's own facts (AGL-4083), word for word: its header and footer
+ * blocks, its Organization structured data, and the email addresses, phone
+ * numbers and profiles they hold. Nothing here is invented or reworded; a
+ * fact the page does not state is not here.
+ */
+export declare function siteImportSiteFacts(home: Pick<SiteImportPage, 'blocks' | 'organization'>): SiteImportSiteFacts;
 /** Why the crawler could not read a page, as an import leaves it out; `status` is the answer an `http-error` got. */
 export declare function siteImportLeftOutReason(reason: SiteImportSkipReason, status?: number): AiSiteImportLeftOutReason;