\n let tagline = (biz.slogan || biz.alternateName || \"\").trim();\n let taglineSource = tagline ? \"schema.slogan\" : \"\";\n if (!tagline) {\n const title = (html.match(/<title[^>]*>([\\s\\S]*?)<\\/title>/i)?.[1] || \"\")\n .replace(/\\s+/g, \" \").trim();\n const parts = title.split(/\\s+[|\\u2013\\u2014\\u00b7\\u2022-]\\s+/).map((s) =>\n s.trim()\n ).filter(Boolean);\n const cand = parts.slice(1).find((p) =>\n /\\s/.test(p) && p.length >= 12 && p.length <= 70 && !/^https?:/i.test(p)\n );\n if (cand) {\n tagline = cand;\n taglineSource = \"<title>\";\n }\n }\n\n // Years in business — foundingDate, then \"since YYYY\", then \"N years\"\n let founded = \"\", years = null, yearsSource = \"\";\n const fdYear = (biz.foundingDate || \"\").toString().match(/(?:19|20)\\d{2}/)\n ?.[0];\n if (fdYear) {\n founded = fdYear;\n yearsSource = \"schema.foundingDate\";\n }\n if (!founded) {\n const since = text.match(\n /\\b(?:since|est\\.?|established(?:\\s+in)?)\\s+((?:19|20)\\d{2})/i,\n );\n if (since) {\n founded = since[1];\n yearsSource = \"'since YYYY' on page\";\n }\n }\n if (founded) years = new Date().getFullYear() - parseInt(founded, 10);\n if (years == null) {\n const yrs = text.match(\n /\\b(\\d{1,2})\\+?\\s*years?\\s+(?:of\\s+)?(?:experience|in\\s+business|serving|cleaning)/i,\n );\n if (yrs) {\n years = parseInt(yrs[1], 10);\n yearsSource = \"'N years' on page\";\n }\n }\n\n // Owner — Person node, founder, then a first-person intro on the page\n const person = schema?.person || null;\n let owner = person?.name || biz.founder?.name || biz.employee?.name || \"\";\n let ownerSource = owner ? \"schema\" : \"\";\n if (!owner) {\n const hi = text.match(\n /\\b(?:hi|hello)[,!]?\\s+(?:I'?m|my name is)\\s+([A-Z][a-z]+(?:\\s+[A-Z][a-z]+)?)/,\n );\n if (hi) {\n owner = hi[1];\n ownerSource = \"'Hi, I'm ...' on page\";\n }\n }\n\n // Rating + review count — schema first, then hero-badge text.\n // Note: hero text like \"760+ reviews\" is an approximation; keep it verbatim\n // and flag for GBP verification rather than silently coercing to an integer.\n let rating = biz.aggregateRating?.ratingValue\n ? String(biz.aggregateRating.ratingValue)\n : \"\";\n let ratingSource = rating ? \"schema.aggregateRating\" : \"\";\n let reviewCount = biz.aggregateRating?.reviewCount\n ? String(biz.aggregateRating.reviewCount)\n : \"\";\n let reviewCountSource = reviewCount ? \"schema.aggregateRating\" : \"\";\n let reviewCountApprox = false;\n if (!rating) {\n // \"4.7 rating on Google\", \"4.7 stars\", \"rated 4.7\", \"4.7/5\"\n const rm =\n text.match(\n /\\b([0-5](?:\\.\\d)?)\\s*(?:\\/\\s*5\\b|stars?\\b|rating\\b|out of 5)/i,\n ) ||\n text.match(/\\brat(?:ed|ing)\\s*(?:of\\s*)?([0-5](?:\\.\\d)?)\\b/i);\n if (rm && parseFloat(rm[1]) >= 3 && parseFloat(rm[1]) <= 5) {\n rating = rm[1];\n ratingSource = \"hero text on page\";\n }\n }\n if (!reviewCount) {\n // \"760+ Reviews\", \"760 Google reviews\", \"over 760 reviews\"\n const cm = text.match(\n /\\b(?:over\\s+)?(\\d{1,3}(?:,\\d{3})*|\\d+)\\s*(\\+)?\\s*(?:google\\s+)?reviews?\\b/i,\n );\n if (cm) {\n const n = cm[1].replace(/,/g, \"\");\n if (parseInt(n, 10) >= 1) {\n reviewCount = cm[2] ? `${n}+` : n;\n reviewCountApprox = !!cm[2];\n reviewCountSource = \"hero text on page\";\n }\n }\n }\n\n return {\n tagline,\n taglineSource,\n founded,\n years,\n yearsSource,\n owner,\n ownerTitle: person?.jobTitle || \"\",\n ownerSource,\n rating,\n ratingSource,\n reviewCount,\n reviewCountSource,\n reviewCountApprox,\n description: (biz.description || \"\").trim(),\n };\n}\n\nexport function extractLogoAndFavicon(html, baseUrl) {\n const abs = (href) => {\n if (!href) return \"\";\n try {\n if (href.startsWith(\"data:\")) return href.slice(0, 40) + \"…\";\n if (href.startsWith(\"http\")) return href;\n return new URL(href, baseUrl).href;\n } catch {\n return href;\n }\n };\n\n // Favicon — check link tags in order of preference\n const faviconRx =\n /<link[^>]+rel=[\"'][^\"']*(?:shortcut icon|icon)[^\"']*[\"'][^>]*href=[\"']([^\"']+)[\"']|<link[^>]+href=[\"']([^\"']+)[\"'][^>]+rel=[\"'][^\"']*(?:shortcut icon|icon)[^\"']*[\"']/gi;\n const favicons = [];\n let fm;\n while ((fm = faviconRx.exec(html)) !== null) {\n const href = abs(fm[1] || fm[2]);\n if (href && !favicons.includes(href)) favicons.push(href);\n }\n // Apple touch icon as fallback\n if (!favicons.length) {\n const appleM = html.match(\n /<link[^>]+rel=[\"']apple-touch-icon[\"'][^>]*href=[\"']([^\"']+)[\"']/i,\n );\n if (appleM) favicons.push(abs(appleM[1]));\n }\n\n // Logo — several strategies, most-specific first.\n const logos = [];\n const pushLogo = (href) => {\n const a = abs(href);\n if (a && !logos.includes(a) && !a.startsWith(\"data:\")) logos.push(a);\n };\n // 1. <img> whose class/alt/id/src mentions logo or brand\n const logoRx =\n /(?:class=[\"'][^\"']*(?:logo|brand|site-title)[^\"']*[\"'][^>]*>[\\s\\S]{0,300}?)?<img[^>]+src=[\"']([^\"']+)[\"'][^>]*(?:class=[\"'][^\"']*logo[^\"']*[\"']|alt=[\"'][^\"']*logo[^\"']*[\"']|id=[\"'][^\"']*logo[^\"']*[\"'])|<(?:a|div)[^>]+class=[\"'][^\"']*logo[^\"']*[\"'][^>]*>[\\s\\S]{0,200}?<img[^>]+src=[\"']([^\"']+)[\"']/gi;\n let lm;\n while ((lm = logoRx.exec(html)) !== null) pushLogo(lm[1] || lm[2]);\n // 2. Any <img> whose src filename contains \"logo\"\n (html.match(/<img[^>]+src=[\"']([^\"']*logo[^\"']*)[\"']/gi) || [])\n .forEach((m) => pushLogo(m.match(/src=[\"']([^\"']+)[\"']/i)?.[1]));\n // 3. Inline <svg> inside a .logo / .brand wrapper → note as inline (can't URL it)\n const hasInlineSvgLogo =\n /class=[\"'][^\"']*(?:logo|brand|site-title)[^\"']*[\"'][^>]*>[\\s\\S]{0,200}?<svg/i\n .test(html);\n // 4. First <img> inside <header> (many builders use a bare logo img, no class)\n if (!logos.length) {\n const header = (html.match(/<header[\\s\\S]*?<\\/header>/i) || [\"\"])[0] ||\n (html.match(\n /<(?:nav|div)[^>]+(?:id|class)=[\"'][^\"']*(?:header|masthead|navbar|site-header)[^\"']*[\"'][\\s\\S]*?<\\/(?:nav|div)>/i,\n ) || [\"\"])[0];\n const firstImg = header.match(\n /<img[^>]+(?:src|data-src)=[\"']([^\"']+)[\"'][^>]*>/i,\n );\n if (firstImg) pushLogo(firstImg[1]);\n // srcset fallback\n if (!logos.length) {\n const ss = header.match(/<img[^>]+srcset=[\"']([^\"',\\s]+)/i);\n if (ss) pushLogo(ss[1]);\n }\n }\n // 5. schema logo / publisher logo as last resort\n if (!logos.length) {\n const schemaLogo = html.match(\n /\"logo\"\\s*:\\s*(?:{[^}]*\"url\"\\s*:\\s*\"([^\"]+)\"|\"([^\"]+)\")/i,\n );\n if (schemaLogo) pushLogo(schemaLogo[1] || schemaLogo[2]);\n }\n\n // OG image from meta as hero candidate\n const ogM = html.match(\n /<meta[^>]+property=[\"']og:image[\"'][^>]*content=[\"']([^\"']+)[\"']|<meta[^>]+content=[\"']([^\"']+)[\"'][^>]+property=[\"']og:image[\"']/i,\n );\n const ogImage = ogM ? abs(ogM[1] || ogM[2]) : \"\";\n\n return {\n favicon: favicons[0] || \"\",\n favicons,\n logo: logos[0] || \"\",\n logos,\n hasInlineSvgLogo,\n ogImage,\n };\n}\n\nexport function extractContactFromHTML(html, baseUrl) {\n // Phone — from tel: links (most reliable) + text patterns in header/footer\n const telLinks = [\n ...new Set(\n (html.match(/href=[\"']tel:([^\"']+)[\"']/gi) || [])\n .map((m) => m.match(/tel:([^\"']+)/i)?.[1]?.trim().replace(/\\s/g, \"\"))\n .filter(Boolean),\n ),\n ];\n const phonePattern = /\$?\\d{3}\$?[-.\\s]\\d{3}[-.\\s]\\d{4}/g;\n // Scan header and footer specifically for text phone numbers\n const headerFooterHtml = [\n (html.match(/<header[\\s\\S]*?<\\/header>/i) || [\"\"])[0],\n (html.match(/<footer[\\s\\S]*?<\\/footer>/i) || [\"\"])[0],\n (html.match(\n /<div[^>]+class=[\"'][^\"']*banner[^\"']*[\"'][^>]*>[\\s\\S]{0,1000}/i,\n ) || [\"\"])[0],\n ].join(\" \");\n const textPhones = [\n ...new Set(\n (headerFooterHtml.match(phonePattern) || []).map((p) => p.trim()),\n ),\n ];\n const phones = [...new Set([...telLinks, ...textPhones])];\n\n // Email — three layers: mailto: links, Cloudflare-obfuscated, plain text.\n const emailSet = new Set();\n const addEmail = (e) => {\n const v = (e || \"\").trim().toLowerCase();\n if (!v || !/^[a-z0-9._%+-]+@[a-z0-9.-]+\\.[a-z]{2,}$/.test(v)) return;\n // Skip asset filenames misread as emails and common false positives\n if (/@\\d+x\\b|\\.(png|jpe?g|gif|webp|svg|css|js)$/i.test(v)) return;\n if (\n /^(?:example|email|your|name|user|info)@(?:example|domain|email)\\./.test(\n v,\n )\n ) return;\n emailSet.add(v);\n };\n\n // 1. mailto: links (most reliable)\n (html.match(/href=[\"']mailto:([^\"'? ]+)/gi) || [])\n .forEach((m) => addEmail(m.match(/mailto:([^\"'? ]+)/i)?.[1]));\n\n // 2. Cloudflare email obfuscation — data-cfemail=\"HEX\" or /cdn-cgi/l/email-protection#HEX\n // First hex byte is the XOR key; remaining bytes are the char codes.\n const decodeCfEmail = (hex) => {\n try {\n const key = parseInt(hex.slice(0, 2), 16);\n let out = \"\";\n for (let i = 2; i < hex.length; i += 2) {\n out += String.fromCharCode(parseInt(hex.slice(i, i + 2), 16) ^ key);\n }\n return out;\n } catch {\n return \"\";\n }\n };\n (html.match(/data-cfemail=[\"']([0-9a-fA-F]+)[\"']/gi) || [])\n .forEach((m) =>\n addEmail(decodeCfEmail(m.match(/[\"']([0-9a-fA-F]+)[\"']/)[1]))\n );\n (html.match(/\\/cdn-cgi\\/l\\/email-protection#([0-9a-fA-F]+)/gi) || [])\n .forEach((m) => addEmail(decodeCfEmail(m.match(/#([0-9a-fA-F]+)/)[1])));\n\n // 3. Plain-text emails anywhere on the page (footers often omit mailto:)\n (html.replace(/<script[\\s\\S]*?<\\/script>/gi, \" \").replace(\n /<style[\\s\\S]*?<\\/style>/gi,\n \" \",\n )\n .match(/[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}/g) || [])\n .forEach(addEmail);\n\n const emails = [...emailSet];\n\n // Address — <address> tag first, then footer heuristic\n const addrTagMatch = html.match(/<address[^>]*>([\\s\\S]*?)<\\/address>/i);\n const addrFromTag = addrTagMatch\n ? addrTagMatch[1].replace(/<[^>]+>/g, \" \").replace(/\\s+/g, \" \").trim()\n : \"\";\n\n // Footer — strip scripts/styles then grab text\n const footerRaw = (html.match(/<footer[\\s\\S]*?<\\/footer>/i) || [\"\"])[0];\n const footerText = footerRaw.replace(/<script[\\s\\S]*?<\\/script>/gi, \"\")\n .replace(/<style[\\s\\S]*?<\\/style>/gi, \"\").replace(/<[^>]+>/g, \" \").replace(\n /\\s+/g,\n \" \",\n ).trim().slice(0, 800);\n\n // Social links — href-based. Two hazards worth naming, because the naive\n // version of this gets them both wrong:\n //\n // 1. SHARE BUTTONS. Almost every site carries links like\n // twitter.com/intent/tweet, linkedin.com/shareArticle, wa.me/?text=,\n // pinterest.com/pin/create/button. Those are the *reader* sharing the\n // page, not the business's profile. Only facebook's sharer was excluded\n // before, so any site with a share bar could land a share URL in\n // twitterLink at high confidence — a wrong value that looks right.\n //\n // 2. POSITION. A business's own profiles sit in the header or footer;\n // share widgets sit in the article body. So the footer is scanned\n // separately and wins over a whole-document match.\n //\n // Keys here are the *audit's* vocabulary; buildIntakePrefill maps them onto\n // the registry's `*Link` names. Don't rename them here without doing that.\n const SOCIAL_HOSTS = [\n [\"facebook\", /facebook\\.com/i],\n [\"instagram\", /instagram\\.com/i],\n [\"youtube\", /youtube\\.com/i],\n [\"twitter\", /(?:twitter|x)\\.com/i],\n [\"linkedin\", /linkedin\\.com/i],\n [\"pinterest\", /pinterest\\.[a-z.]{2,6}/i],\n [\"tiktok\", /tiktok\\.com/i],\n [\"yelp\", /yelp\\.com/i],\n [\"nextdoor\", /nextdoor\\.com/i],\n [\"angi\", /(?:angi\\.com|angieslist\\.com|homeadvisor\\.com)/i],\n [\"thumbtack\", /thumbtack\\.com/i],\n [\"whatsapp\", /(?:wa\\.me|(?:api|web|chat)\\.whatsapp\\.com)/i],\n ];\n // Generic share-intent shapes, applied to every host.\n const SHARE_RX =\n /\\/(?:sharer|share\\.php|shareArticle|sharing\\/|intent\\/|pin\\/create|submit)\\b/i;\n // Per-host \"is this actually a profile\" tests. A missing entry means the\n // host has no known false-positive shape and anything on it is accepted.\n const PROFILE_OK = {\n facebook: (u) => !/\\/(?:sharer|dialog)\\b/i.test(u),\n // /p/, /reel/ and /explore/ are individual posts, not the account.\n instagram: (u) =>\n !/instagram\\.com\\/(?:p|reel|reels|explore|stories)\\//i.test(u),\n // An embedded video is not a channel link.\n youtube: (u) => !/\\/(?:watch|embed|shorts)\\b/i.test(u),\n twitter: (u) => !/\\/(?:intent|share|i)\\//i.test(u),\n // Personal profiles (/in/) are the owner, not the business — still useful,\n // but a company page is the better answer when both are present.\n linkedin: (u) => /linkedin\\.com\\/(?:company|in|school)\\//i.test(u),\n pinterest: (u) => !/\\/pin\\//i.test(u),\n tiktok: (u) => !/\\/video\\//i.test(u),\n // yelp.com alone is also an asset/widget host; only /biz/ is the listing.\n yelp: (u) => /yelp\\.com\\/biz\\//i.test(u),\n // wa.me/15555550142 is click-to-chat; wa.me/?text= is a share button.\n // The number is what distinguishes them.\n whatsapp: (u) => /(?:wa\\.me\\/|phone=)\\+?\\d{7,}/i.test(u),\n };\n const cleanSocial = (url, key) => {\n // Tracking params (?fbclid=, ?hl=en) aren't part of the profile URL.\n // WhatsApp is the exception: its query string carries the phone number.\n if (key === \"whatsapp\") return url;\n return url.split(\"?\")[0].replace(/#.*$/, \"\");\n };\n const collectSocial = (scope) => {\n const found = {};\n if (!scope) return found;\n const rx = /href=[\"'](https?:\\/\\/[^\"'\\s>]+)[\"']/gi;\n let m;\n while ((m = rx.exec(scope)) !== null) {\n const url = m[1];\n if (SHARE_RX.test(url)) continue;\n for (const [key, host] of SOCIAL_HOSTS) {\n if (found[key] || !host.test(url)) continue;\n if (PROFILE_OK[key] && !PROFILE_OK[key](url)) continue;\n found[key] = cleanSocial(url, key);\n }\n }\n return found;\n };\n // Footer spread last so it wins — see hazard 2 above.\n const social = { ...collectSocial(html), ...collectSocial(footerRaw) };\n\n // Footer nav links — useful for understanding site structure\n const footerLinks = [];\n if (footerRaw) {\n const aRx = /<a[^>]+href=[\"']([^\"'#]+)[\"'][^>]*>([\\s\\S]*?)<\\/a>/gi;\n let am;\n while ((am = aRx.exec(footerRaw)) !== null) {\n const href = am[1].trim();\n const text = am[2].replace(/<[^>]+>/g, \" \").replace(/\\s+/g, \" \").trim();\n if (\n text && href && !href.startsWith(\"tel:\") &&\n !href.startsWith(\"mailto:\") && text.length < 60\n ) {\n footerLinks.push({ href, text });\n }\n }\n }\n\n return {\n phones,\n emails,\n address: addrFromTag,\n addresses: addrFromTag\n ? [addrFromTag]\n : extractAddressesFromText(footerText),\n footerText,\n social,\n footerLinks,\n hours: extractHoursFromText(footerText),\n };\n}\n\nexport function analyzePhones(schemaPhone, contactInfo, html) {\n const counts = new Map(), seen = new Map();\n const note = (raw, weight) => {\n const n = normPhone(raw);\n if (n.length < 10) return;\n counts.set(n, (counts.get(n) || 0) + weight);\n if (!seen.has(n)) seen.set(n, (raw || \"\").trim());\n };\n (contactInfo?.phones || []).forEach((p) => note(p, 2));\n ((html || \"\").match(/href=[\"']tel:([^\"']+)[\"']/gi) || [])\n .forEach((m) => note(m.replace(/.*tel:/i, \"\").replace(/[\"']/g, \"\"), 1));\n const schemaNorm = normPhone(schemaPhone);\n if (schemaNorm) note(schemaPhone, 0);\n\n const ranked = [...counts.entries()].sort((a, b) => b[1] - a[1]);\n const display = ranked[0]?.[0] || schemaNorm || \"\";\n const hasDiscrepancy = !!schemaNorm && !!display && schemaNorm !== display;\n return {\n all: ranked.map(([n, c]) => ({\n number: fmtPhone(n),\n raw: seen.get(n) || n,\n hits: c,\n inSchema: n === schemaNorm,\n })),\n display: display ? fmtPhone(display) : \"\",\n schema: schemaNorm ? fmtPhone(schemaNorm) : \"\",\n hasDiscrepancy,\n likelyTracking: hasDiscrepancy ? fmtPhone(display) : \"\",\n verdict: hasDiscrepancy\n ? \"Displayed number differs from schema — likely a call-tracking number\"\n : ranked.length > 1\n ? \"Multiple numbers found — confirm which is primary\"\n : ranked.length === 1\n ? \"Single consistent number\"\n : \"No number found\",\n };\n}\n\nexport function extractHomepageHero(html) {\n const strip = (s) =>\n (s || \"\")\n .replace(/<[^>]+>/g, \" \").replace(/ /gi, \" \").replace(/&/gi, \"&\")\n .replace(/’|’/gi, \"\\u2019\").replace(\n /–|–/gi,\n \"\\u2013\",\n )\n .replace(/&[a-z]+;/gi, \" \").replace(/\\s+/g, \" \").trim();\n // Ignore header/nav/footer chrome — hero copy lives in the body\n let body = html\n .replace(/<header[\\s\\S]*?<\\/header>/gi, \" \")\n .replace(/<nav[\\s\\S]*?<\\/nav>/gi, \" \")\n .replace(/<footer[\\s\\S]*?<\\/footer>/gi, \" \")\n .replace(/<script[\\s\\S]*?<\\/script>/gi, \" \")\n .replace(/<style[\\s\\S]*?<\\/style>/gi, \" \");\n const h1 = strip((body.match(/<h1[^>]*>([\\s\\S]*?)<\\/h1>/i) || [])[1]).slice(\n 0,\n 160,\n );\n const h2 = strip((body.match(/<h2[^>]*>([\\s\\S]*?)<\\/h2>/i) || [])[1]).slice(\n 0,\n 200,\n );\n // First substantial <p> after the h1\n let intro = \"\";\n const afterH1 = h1 ? body.slice(body.search(/<h1[^>]*>/i)) : body;\n const pRx = /<p[^>]*>([\\s\\S]*?)<\\/p>/gi;\n let pm;\n while ((pm = pRx.exec(afterH1)) !== null) {\n const t = strip(pm[1]);\n if (t.length >= 60) {\n intro = t.slice(0, 400);\n break;\n }\n }\n return { h1, h2, intro };\n}",[163],"70cc6548-5b3b-4394-a6b2-365009c783c3","discovery.ts","9713bd2c-8c73-11f1-8ef8-1607ee4eb77e",22,["D",1785457221412],4,"/**\n * ksite-discovery / discovery.mjs\n *\n * Orchestrates one homepage pass: fetch, run every extractor, resolve DNS/RDAP.\n *\n * Part of a set split for Val.town's 80,000-character-per-file limit. The\n * single-file build is otherwise identical — nothing here is Val-specific.\n */\n\nimport {\n detectEmailProvider,\n detectHosting,\n detectPlatform,\n extractGoogleVerification,\n extractKsiteVersion,\n extractSPFInfo,\n fetchAllDNS,\n fetchRDAP,\n fetchViaProxy,\n} from \"./core.ts\";\nimport {\n analyzePhones,\n detectCRM,\n detectFontRoles,\n extractBrandBasics,\n extractContactFromHTML,\n extractCSSUrls,\n extractCSSVars,\n extractFonts,\n extractGBP,\n extractHomepageHero,\n extractLogoAndFavicon,\n extractNav,\n extractSchema,\n scanOrgs,\n scanPerks,\n scanThirdParty,\n} from \"./extract-html.ts\";\n\nexport async function fetchSiteDiscovery(siteUrl, ksiteHint) {\n const result = {\n schema: null,\n nav: [],\n crm: null,\n cssVars: {},\n fonts: null,\n fontRoles: null,\n dns: {},\n rdap: null,\n hosting: null,\n cms: null,\n ksite: null,\n emailProvider: null,\n spf: null,\n googleVerification: null,\n thirdParty: [],\n gbp: null,\n assets: null,\n contactInfo: null,\n brand: null,\n perks: [],\n orgs: [],\n phoneAnalysis: null,\n homepageHero: null,\n error: null,\n };\n let html = \"\";\n try {\n html = await fetchViaProxy(siteUrl);\n result.schema = extractSchema(html);\n result.nav = extractNav(html);\n result.crm = detectCRM(html);\n result.thirdParty = scanThirdParty(html);\n result.gbp = extractGBP(result.schema, html);\n result.assets = extractLogoAndFavicon(html, siteUrl);\n result.contactInfo = extractContactFromHTML(html);\n const cssUrls = extractCSSUrls(html, siteUrl);\n const cssBlobs = [];\n // Inline <style> first. Small sites — the ones most likely to be on a\n // page builder — often declare their :root custom properties inline and\n // have no external stylesheet worth fetching, so scanning only linked CSS\n // returned no brand colours at all for them.\n const inlineCss = (html.match(/<style[^>]*>([\\s\\S]*?)<\\/style>/gi) || [])\n .map((b) => b.replace(/^<style[^>]*>/i, \"\").replace(/<\\/style>$/i, \"\"))\n .join(\"\\n\");\n if (inlineCss.trim()) {\n Object.assign(result.cssVars, extractCSSVars(inlineCss));\n cssBlobs.push({ url: \"(inline <style>)\", css: inlineCss });\n }\n for (const url of cssUrls.slice(0, 3)) {\n try {\n const css = await fetchViaProxy(url);\n Object.assign(result.cssVars, extractCSSVars(css));\n cssBlobs.push({ url, css });\n } catch {}\n }\n const siteDomain = (() => {\n try {\n return new URL(siteUrl).hostname.replace(/^www\\./, \"\");\n } catch {\n return \"\";\n }\n })();\n result.fonts = extractFonts(html, cssBlobs, siteDomain);\n result.fontRoles = detectFontRoles(html, cssBlobs);\n result.brand = extractBrandBasics(result.schema, html);\n result.perks = scanPerks(html);\n result.orgs = scanOrgs(html);\n result.phoneAnalysis = analyzePhones(\n result.schema?.business?.telephone,\n result.contactInfo,\n html,\n );\n result.homepageHero = extractHomepageHero(html);\n } catch (e) {\n result.error = e.message;\n }\n\n // ksite detection — use the stamp captured at load; if absent, check sitemap.xml\n // and the homepage's <meta generator> as a self-contained fallback.\n result.ksite = ksiteHint?.isKsite ? ksiteHint : null;\n if (!result.ksite) {\n const genM = html.match(\n /<meta[^>]+name=[\"']generator[\"'][^>]+content=[\"']\\s*ksite\\s*([^\"']*)[\"']/i,\n );\n if (genM) {\n result.ksite = {\n isKsite: true,\n version: (genM[1] || \"\").trim(),\n source: \"meta generator\",\n };\n }\n }\n if (!result.ksite) {\n try {\n const origin = new URL(siteUrl).origin;\n const smXml = await fetchViaProxy(`${origin}/sitemap.xml`);\n const stamp = extractKsiteVersion(smXml);\n if (stamp) result.ksite = { ...stamp, source: \"sitemap.xml\" };\n } catch {}\n }\n\n try {\n const domain = new URL(siteUrl).hostname.replace(/^www\\./, \"\");\n result.dns = await fetchAllDNS(domain);\n result.hosting = detectHosting(result.dns);\n result.cms = detectPlatform(html, result.dns, result.ksite);\n result.emailProvider = detectEmailProvider(result.dns.MX || []);\n result.spf = extractSPFInfo(result.dns.TXT || []);\n result.googleVerification = extractGoogleVerification(\n result.dns.TXT || [],\n html,\n );\n result.rdap = await fetchRDAP(domain);\n } catch {}\n return result;\n}",[170],"9dbde85d-564d-4ff0-b8b6-8c2e2d56aae4","core.ts","1a9ef656-8c75-11f1-9c91-1607ee4eb77e",7,["D",1785457040283],"/**\n * ksite-discovery / core.mjs\n *\n * Fetching, DNS, URL classification, sitemap parsing, shared text helpers.\n *\n * Part of a set split for Val.town's 80,000-character-per-file limit. The\n * single-file build is otherwise identical — nothing here is Val-specific.\n */\n\nexport const UA =\n \"Mozilla/5.0 (compatible; ksite-auditor/1.0; +https://kobami.com)\";\n\nexport const TIMEOUT_MS = 20000;\n\n// Same service the browser build uses. Page extraction reads markdown, not HTML.\nexport const PAGE_API = \"https://html2markdown.val.run/\";\n\n// Server-side there is no same-origin policy, so the whole proxy chain the\n// browser build needs (corsproxy.io → allorigins → codetabs, each with its own\n// rate limit and failure mode) collapses into one direct fetch. This is the\n// single biggest reliability win from moving off the browser.\nexport async function fetchViaProxy(url) {\n const r = await fetch(url, {\n redirect: \"follow\",\n headers: {\n // Some hosts serve a different page, or a challenge, to an unknown agent.\n \"user-agent\": UA,\n \"accept\":\n \"text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8\",\n \"accept-language\": \"en-US,en;q=0.9\",\n },\n signal: AbortSignal.timeout(TIMEOUT_MS),\n });\n if (!r.ok) throw new Error(`${r.status} ${r.statusText} — ${url}`);\n return r.text();\n}\n\nexport const US_STATES =\n /-(al|ak|az|ar|ca|co|ct|de|fl|ga|hi|id|il|in|ia|ks|ky|la|me|md|ma|mi|mn|ms|mo|mt|ne|nv|nh|nj|nm|ny|nc|nd|oh|ok|or|pa|ri|sc|sd|tn|tx|ut|vt|va|wa|wv|wi|wy)$/;\n\nexport function extractCityFromSlug(slug) {\n let city = slug.replace(US_STATES, \"\");\n for (\n const p of [\n \"house-cleaning-services-\",\n \"house-cleaning-service-\",\n \"house-cleaning-\",\n \"cleaning-services-company-\",\n \"cleaning-services-\",\n \"cleaning-service-\",\n \"maid-services-\",\n \"maid-service-\",\n \"home-cleaning-services-\",\n \"home-cleaning-\",\n \"residential-cleaning-services-\",\n \"residential-cleaning-\",\n ]\n ) {\n if (city.startsWith(p)) {\n city = city.slice(p.length);\n break;\n }\n }\n return city || slug;\n}\n\nexport function classify(rawUrl) {\n try {\n const u = new URL(rawUrl);\n const p = u.pathname.toLowerCase();\n const segs = p.split(\"/\").filter(Boolean);\n const last = segs[segs.length - 1] || \"\";\n if (/\\/(category|tag|author)\\//.test(p)) {\n return { type: \"wp-taxonomy\", status: \"404\", ksite: \"\" };\n }\n if (/\\/page\\/\\d+/.test(p)) {\n return { type: \"wp-pagination\", status: \"404\", ksite: \"\" };\n }\n if (/\\/feed\\/?$/.test(p)) {\n return { type: \"wp-feed\", status: \"404\", ksite: \"\" };\n }\n if (u.search.includes(\"p=\")) {\n return { type: \"wp-qp\", status: \"404\", ksite: \"\" };\n }\n if (!segs.length) return { type: \"homepage\", status: \"keep\", ksite: \"/\" };\n if (\n /\\/(service-areas?|locations?|areas?-we-serve|cities-we-serve)\\b/.test(p)\n ) {\n return {\n type: \"service-area\",\n status: \"keep\",\n ksite: segs.length > 1 ? `/service-areas/${last}` : \"/service-areas\",\n };\n }\n if (\n segs.length === 1 && US_STATES.test(last) &&\n /clean|maid|house|service/.test(last)\n ) {\n return {\n type: \"service-area\",\n status: \"keep\",\n ksite: `/service-areas/${extractCityFromSlug(last)}`,\n };\n }\n if (/^\\/services\\//.test(p) && segs.length > 1) {\n return { type: \"service\", status: \"keep\", ksite: `/services/${last}` };\n }\n if (segs.length >= 1 && /cleaning|maid|housekeeping|sanitiz/.test(last)) {\n return { type: \"service\", status: \"keep\", ksite: `/services/${last}` };\n }\n if (/\\babout\\b/.test(p)) {\n return { type: \"about\", status: \"keep\", ksite: \"/about-us\" };\n }\n if (/\\bcontact\\b/.test(p)) {\n return { type: \"contact\", status: \"keep\", ksite: \"/contact-us\" };\n }\n if (/estimate|free-quote|get-a-quote/.test(p)) {\n return { type: \"estimate\", status: \"keep\", ksite: \"/estimate\" };\n }\n if (/\\bbook\\b/.test(p)) {\n return { type: \"booking\", status: \"keep\", ksite: \"/book\" };\n }\n if (/thank/.test(p)) {\n return { type: \"thank-you\", status: \"keep\", ksite: \"/thank-you\" };\n }\n if (/\\bfaqs?\\b/.test(p)) {\n return { type: \"faq\", status: \"keep\", ksite: \"/faqs\" };\n }\n if (/employ|career|\\bjobs?\\b/.test(p)) {\n return { type: \"employment\", status: \"keep\", ksite: \"/employment\" };\n }\n if (/gift/.test(p)) {\n return { type: \"gift-cards\", status: \"keep\", ksite: \"/gift-cards\" };\n }\n if (/site-?map/.test(p)) {\n return { type: \"system\", status: \"keep\", ksite: \"/sitemap\" };\n }\n if (\n /privacy|terms|legal|cookie|policies|policy|disclaimer|accessibility/\n .test(p)\n ) {\n return { type: \"policies\", status: \"review\", ksite: \"\" };\n }\n if (/\\breviews?\\b/.test(p)) {\n return { type: \"unknown\", status: \"keep\", ksite: \"/reviews\" };\n }\n if (/checklist/.test(p)) {\n return { type: \"unknown\", status: \"review\", ksite: \"\" };\n }\n if (/blog|news|post|article/.test(p)) {\n return { type: \"blog\", status: \"review\", ksite: \"\" };\n }\n return { type: \"unknown\", status: \"review\", ksite: \"\" };\n } catch {\n return { type: \"unknown\", status: \"review\", ksite: \"\" };\n }\n}\n\nexport function extractKsiteVersion(xml) {\n if (!xml) return null;\n const m = xml.match(\n //i,\n );\n if (!m) return null;\n return { isKsite: true, version: (m[1] || \"\").trim() };\n}\n\n// The browser build used DOMParser here. It's the only DOM dependency in the\n// discovery path, and sitemap XML is shallow enough that regex covers it —\n// which keeps this module at zero runtime dependencies. If you ever need real\n// XML handling, swap in linkedom rather than reaching for jsdom.\nexport function parseSitemap(xml) {\n const tag = (block, name) => {\n const m = block.match(\n new RegExp(`<${name}[^>]*>([\\\\s\\\\S]*?)</${name}>`, \"i\"),\n );\n if (!m) return \"\";\n return m[1]\n .replace(/<!\\[CDATA\\[([\\s\\S]*?)\\]\\]>/g, \"$1\")\n .replace(/</g, \"<\").replace(/>/g, \">\")\n .replace(/"/g, '\"').replace(/'/g, \"'\").replace(/&/g, \"&\")\n .trim();\n };\n const isIndex = /<sitemapindex[\\s>]/i.test(xml);\n const blocks =\n xml.match(\n isIndex\n ? /<sitemap[\\s>][\\s\\S]*?<\\/sitemap>/gi\n : /<url[\\s>][\\s\\S]*?<\\/url>/gi,\n ) || [];\n const entries = blocks.map((b) => ({\n loc: tag(b, \"loc\"),\n lastmod: tag(b, \"lastmod\"),\n changefreq: tag(b, \"changefreq\"),\n priority: tag(b, \"priority\"),\n })).filter((e) => e.loc);\n return { isIndex, entries };\n}\n\nexport async function fetchDNS(domain, type) {\n const r = await fetch(\n `https://cloudflare-dns.com/dns-query?name=${\n encodeURIComponent(domain)\n }&type=${type}`,\n { headers: { \"Accept\": \"application/dns-json\" } },\n );\n if (!r.ok) throw new Error(`DNS ${type} ${r.status}`);\n const d = await r.json();\n return d.Answer || [];\n}\n\nexport async function fetchAllDNS(domain) {\n const out = {};\n await Promise.all([\"A\", \"MX\", \"NS\", \"TXT\"].map(async (type) => {\n try {\n out[type] = await fetchDNS(domain, type);\n } catch {\n out[type] = [];\n }\n }));\n return out;\n}\n\nexport async function fetchRDAP(domain) {\n const parse = (d) => {\n const ents = d.entities || [];\n const vcardName = (ent) => {\n const fn = (ent?.vcardArray?.[1] || []).find((f) => f[0] === \"fn\");\n return fn?.[3] || ent?.handle || \"\";\n };\n const ev = (action) =>\n (d.events || []).find((e) => e.eventAction === action)?.eventDate || \"\";\n const created = ev(\"registration\");\n let ageYears = null;\n if (created) {\n const ms = Date.now() - new Date(created).getTime();\n if (!isNaN(ms)) ageYears = Math.floor(ms / 31557600000);\n }\n return {\n registrar: vcardName(\n ents.find((e) => (e.roles || []).includes(\"registrar\")),\n ),\n registrant: vcardName(\n ents.find((e) => (e.roles || []).includes(\"registrant\")),\n ),\n created,\n expires: ev(\"expiration\"),\n changed: ev(\"last changed\"),\n ageYears,\n status: (d.status || []).join(\", \"),\n };\n };\n const url = `https://rdap.org/domain/${encodeURIComponent(domain)}`;\n try {\n const r = await fetch(url);\n if (!r.ok) throw new Error(`RDAP ${r.status}`);\n return parse(await r.json());\n } catch {\n try {\n return parse(JSON.parse(await fetchViaProxy(url)));\n } catch {\n return null;\n }\n }\n}\n\nexport function detectHosting(dns) {\n const ns = (dns.NS || []).map((r) => (r.data || \"\").toLowerCase()).join(\" \");\n const a = (dns.A || []).map((r) => r.data || \"\").join(\" \");\n if (ns.includes(\"cloudflare\")) return \"Cloudflare\";\n if (ns.includes(\"wpengine\")) return \"WP Engine\";\n if (ns.includes(\"kinsta\")) return \"Kinsta\";\n if (ns.includes(\"netlify\")) return \"Netlify\";\n if (ns.includes(\"vercel\")) return \"Vercel\";\n if (ns.includes(\"siteground\")) return \"SiteGround\";\n if (ns.includes(\"bluehost\")) return \"Bluehost\";\n if (ns.includes(\"godaddy\")) return \"GoDaddy\";\n if (ns.includes(\"amazonaws\")) return \"AWS\";\n if (ns.includes(\"fastly\")) return \"Fastly\";\n if (ns.includes(\"flywheel\")) return \"Flywheel\";\n if (ns.includes(\"pressable\")) return \"Pressable\";\n if (/104\\.1[6-9]\\.|104\\.[2-9]\\d\\./.test(a)) return \"Cloudflare (proxy)\";\n return null;\n}\n\nexport function detectPlatform(html, dns, ksiteHint) {\n const ns = (dns?.NS || []).map((r) => (r.data || \"\").toLowerCase()).join(\" \");\n const a = (dns?.A || []).map((r) => r.data || \"\").join(\" \");\n const h = (html || \"\").toLowerCase();\n\n // ksite — our own framework. Highest priority, deterministic.\n // Primary signal: build stamp captured from sitemap.xml.\n // Secondary: <meta name=\"generator\" content=\"ksite ...\"> if present in HTML.\n if (ksiteHint?.isKsite) {\n return ksiteHint.version\n ? `ksite ${ksiteHint.version}`\n : \"ksite\";\n }\n const genM =\n (html || \"\").match(\n /<meta[^>]+name=[\"']generator[\"'][^>]+content=[\"']\\s*ksite\\s*([^\"']*)[\"']/i,\n ) ||\n (html || \"\").match(\n /<meta[^>]+content=[\"']\\s*ksite\\s*([^\"']*)[\"'][^>]+name=[\"']generator[\"']/i,\n );\n if (genM) return genM[1]?.trim() ? `ksite ${genM[1].trim()}` : \"ksite\";\n\n // DNS-based (most reliable)\n if (/wixdns\\.net/.test(ns)) return \"Wix\";\n if (/squarespace\\.com/.test(ns)) return \"Squarespace\";\n if (/webflow\\.com/.test(ns)) return \"Webflow\";\n if (/shopify\\.com/.test(ns)) return \"Shopify\";\n if (/myshopify\\.com/.test(ns)) return \"Shopify\";\n if (/hubspot\\.com/.test(ns)) return \"HubSpot CMS\";\n if (/wpengine\\.com/.test(ns)) return \"WordPress (WP Engine)\";\n if (/kinsta\\.com/.test(ns)) return \"WordPress (Kinsta)\";\n if (/flywheel\\.com/.test(ns)) return \"WordPress (Flywheel)\";\n if (/pressable\\.com/.test(ns)) return \"WordPress (Pressable)\";\n\n // A-record IP ranges\n if (/23\\.227\\.(3[89]|[4-9]\\d)\\./.test(a)) return \"Shopify\";\n if (/185\\.230\\.(6[0-3]|160|161|162|163)\\./.test(a)) return \"Wix\";\n\n // HTML signals\n if (/wix\\.com|wixstatic\\.com|wixsite\\.com/.test(h)) return \"Wix\";\n if (/squarespace\\.com|static1\\.squarespace/.test(h)) return \"Squarespace\";\n if (/webflow\\.io|webflow\\.com\\/css/.test(h)) return \"Webflow\";\n if (/shopify\\.com|cdn\\.shopify/.test(h)) return \"Shopify\";\n if (/cdn\\.hubspot\\.com|hs-sites\\.com/.test(h)) return \"HubSpot CMS\";\n if (/ghost\\.org|ghost\\.io/.test(h)) return \"Ghost\";\n if (/sites\\.google\\.com|googleapis.*sites/.test(h)) return \"Google Sites\";\n if (/godaddysites\\.com/.test(h)) return \"GoDaddy Website Builder\";\n if (/weebly\\.com/.test(h)) return \"Weebly\";\n if (/jimdo\\.com/.test(h)) return \"Jimdo\";\n if (/elementor|wp-content\\/plugins\\/elementor/.test(h)) {\n return \"WordPress (Elementor)\";\n }\n if (/wp-content|wp-includes|xmlrpc\\.php/.test(h)) return \"WordPress\";\n if (/drupal\\.org|drupal\\.js|drupal\\.settings/.test(h)) return \"Drupal\";\n if (/joomla/.test(h)) return \"Joomla\";\n\n return \"Unknown\";\n}\n\nexport function detectEmailProvider(mxRecords) {\n const mx = mxRecords.map((r) => (r.data || \"\").toLowerCase()).join(\" \");\n if (!mxRecords.length) return \"None / Unknown\";\n if (/google|googlemail/.test(mx)) return \"Google Workspace\";\n if (/outlook|protection\\.outlook|hotmail/.test(mx)) return \"Microsoft 365\";\n if (/zoho/.test(mx)) return \"Zoho Mail\";\n if (/mailgun/.test(mx)) return \"Mailgun\";\n if (/sendgrid/.test(mx)) return \"SendGrid\";\n if (/amazonses|amazonaws/.test(mx)) return \"Amazon SES\";\n if (/privateemail/.test(mx)) return \"Namecheap Mail\";\n if (/bluehost/.test(mx)) return \"Bluehost Mail\";\n if (/siteground/.test(mx)) return \"SiteGround Mail\";\n if (/godaddy/.test(mx)) return \"GoDaddy Mail\";\n if (/infusionmail|keap/.test(mx)) return \"Keap / Infusionsoft\";\n return mxRecords[0]?.data?.split(\" \").pop() || \"Unknown\";\n}\n\nexport function extractSPFInfo(txtRecords) {\n const spf =\n txtRecords.map((r) => r.data || \"\").find((t) => t.includes(\"v=spf1\")) || \"\";\n const includes = [...spf.matchAll(/include:([^\\s]+)/g)].map((m) => m[1]);\n return { raw: spf, includes };\n}\n\nexport function extractGoogleVerification(txtRecords, html) {\n const fromTxt = txtRecords.map((r) => r.data || \"\").find((t) =>\n t.startsWith(\"google-site-verification=\")\n );\n const fromHtml = html?.match(\n /content=[\"']([a-zA-Z0-9_-]{30,})[\"'][^>]*name=[\"']google-site-verification[\"']|name=[\"']google-site-verification[\"'][^>]*content=[\"']([a-zA-Z0-9_-]{30,})[\"']/,\n )?.[1];\n return fromTxt?.replace(\"google-site-verification=\", \"\") || fromHtml || null;\n}\n\nexport const THIRD_PARTY_CHECKS = [\n // Analytics\n {\n name: \"Google Analytics 4\",\n cat: \"analytics\",\n rx: /G-[A-Z0-9]{6,}/,\n extract: (h) => h.match(/G-[A-Z0-9]{6,}/)?.[0],\n },\n {\n name: \"Google Analytics (UA)\",\n cat: \"analytics\",\n rx: /UA-\\d{4,}-\\d+/,\n extract: (h) => h.match(/UA-\\d{4,}-\\d+/)?.[0],\n },\n {\n name: \"Google Tag Manager\",\n cat: \"analytics\",\n rx: /GTM-[A-Z0-9]+|googletagmanager\\.com/,\n extract: (h) => h.match(/GTM-[A-Z0-9]+/)?.[0],\n },\n {\n name: \"Hotjar\",\n cat: \"analytics\",\n rx: /hotjar\\.com/,\n extract: (h) => h.match(/hjid[:\\s]+(\\d+)/)?.[1],\n },\n {\n name: \"Microsoft Clarity\",\n cat: \"analytics\",\n rx: /clarity\\.ms/,\n extract: null,\n },\n // Ads\n {\n name: \"Facebook / Meta Pixel\",\n cat: \"ads\",\n rx:\n /connect\\.facebook\\.net\\/en_US\\/fbevents|fbq\\s*\\(\\s*['\"]init|facebook\\.com\\/tr\\?id=/,\n extract: (h) =>\n h.match(\n /fbq\\s*\\(\\s*['\"]init['\"],\\s*['\"](\\d+)|facebook\\.com\\/tr\\?id=(\\d+)/,\n )?.[1] || h.match(/facebook\\.com\\/tr\\?id=(\\d+)/)?.[1],\n },\n {\n name: \"Google Ads\",\n cat: \"ads\",\n rx: /AW-\\d+|googleadservices/,\n extract: (h) => h.match(/AW-\\d+/)?.[0],\n },\n // CRM / Marketing\n {\n name: \"HubSpot\",\n cat: \"crm\",\n rx: /hubspot\\.com|hs-analytics/,\n extract: null,\n },\n {\n name: \"ActiveCampaign\",\n cat: \"crm\",\n rx: /activecampaign\\.com/,\n extract: null,\n },\n {\n name: \"Mailchimp\",\n cat: \"email\",\n rx: /mailchimp\\.com|chimpified/,\n extract: null,\n },\n // Chat\n { name: \"Intercom\", cat: \"chat\", rx: /intercom\\.io/, extract: null },\n { name: \"Drift\", cat: \"chat\", rx: /drift\\.com/, extract: null },\n { name: \"Zendesk\", cat: \"chat\", rx: /zendesk\\.com/, extract: null },\n { name: \"Crisp\", cat: \"chat\", rx: /crisp\\.chat/, extract: null },\n { name: \"Tidio\", cat: \"chat\", rx: /tidio\\.com/, extract: null },\n { name: \"Tawk.to\", cat: \"chat\", rx: /tawk\\.to/, extract: null },\n { name: \"Smith.ai\", cat: \"chat\", rx: /smith\\.ai/, extract: null },\n // Reviews\n { name: \"Podium\", cat: \"reviews\", rx: /podium\\.com/, extract: null },\n { name: \"Birdeye\", cat: \"reviews\", rx: /birdeye\\.com/, extract: null },\n { name: \"NiceJob\", cat: \"reviews\", rx: /nicejob\\.com/, extract: null },\n { name: \"Broadly\", cat: \"reviews\", rx: /broadly\\.com/, extract: null },\n {\n name: \"Yelp Widget\",\n cat: \"reviews\",\n rx: /yelp-biz-badge|yelp\\.com\\/biz_widget/,\n extract: null,\n },\n // SEO / Other\n { name: \"Yoast SEO\", cat: \"seo\", rx: /yoast\\.com|wpseo/, extract: null },\n {\n name: \"RankMath\",\n cat: \"seo\",\n rx: /rank-math\\.com|rankmath/,\n extract: null,\n },\n { name: \"Plausible\", cat: \"analytics\", rx: /plausible\\.io/, extract: null },\n { name: \"CallRail\", cat: \"tracking\", rx: /callrail\\.com/, extract: null },\n {\n name: \"CallTrackingMetrics\",\n cat: \"tracking\",\n rx: /calltrackingmetrics\\.com/,\n extract: null,\n },\n];\n\nexport function htmlToText(html) {\n return (html || \"\")\n .replace(/<script[\\s\\S]*?<\\/script>/gi, \" \")\n .replace(/<style[\\s\\S]*?<\\/style>/gi, \" \")\n .replace(/<[^>]+>/g, \" \")\n .replace(/ /gi, \" \").replace(/&/gi, \"&\")\n .replace(/\\s+/g, \" \");\n}\n\nexport function normPhone(p) {\n return (p || \"\").replace(/\\D/g, \"\").replace(/^1(?=\\d{10}$)/, \"\");\n}\n\nexport function fmtPhone(d) {\n return d.length === 10\n ? `(${d.slice(0, 3)}) ${d.slice(3, 6)}-${d.slice(6)}`\n : d;\n}\n\nexport function extractAddressesFromText(text) {\n const addrRx =\n /\\d{1,6}\\s+[A-Z][a-zA-Z0-9\\s.,'#-]{5,80}(?:Street|St|Avenue|Ave|Road|Rd|Boulevard|Blvd|Drive|Dr|Lane|Ln|Way|Court|Ct|Place|Pl|Suite|Ste|Northeast|Northwest|Southeast|Southwest|NE|NW|SE|SW)[^\\n]{0,80}(?:[A-Z][a-zA-Z\\s]+,\\s*[A-Z]{2}\\s*\\d{5})/gi;\n return [\n ...new Set((text.match(addrRx) || []).map((a) =>\n a.replace(/\\s+/g, \" \").trim()\n )),\n ];\n}\n\nexport function extractHoursFromText(text) {\n const hours = {};\n const timeP = `\\\\d{1,2}(?::\\\\d{2})?\\\\s*(?:am|pm)`;\n const timeRange = `${timeP}\\\\s*[-–]\\\\s*${timeP}`;\n const dayFull = `monday|tuesday|wednesday|thursday|friday|saturday|sunday`;\n const dayShort = `mon|tue|wed|thu|fri|sat|sun`;\n const dayAny = `(?:${dayFull}|${dayShort})`;\n const expand = (from, to) => {\n const order = [\n \"Monday\",\n \"Tuesday\",\n \"Wednesday\",\n \"Thursday\",\n \"Friday\",\n \"Saturday\",\n \"Sunday\",\n ];\n const norm = (d) => {\n const map = {\n mon: \"Monday\",\n tue: \"Tuesday\",\n wed: \"Wednesday\",\n thu: \"Thursday\",\n fri: \"Friday\",\n sat: \"Saturday\",\n sun: \"Sunday\",\n monday: \"Monday\",\n tuesday: \"Tuesday\",\n wednesday: \"Wednesday\",\n thursday: \"Thursday\",\n friday: \"Friday\",\n saturday: \"Saturday\",\n sunday: \"Sunday\",\n };\n return map[d.toLowerCase()] || d;\n };\n const f = norm(from), t = norm(to);\n const fi = order.indexOf(f), ti = order.indexOf(t);\n if (fi < 0 || ti < 0) return [f];\n return fi <= ti\n ? order.slice(fi, ti + 1)\n : [...order.slice(fi), ...order.slice(0, ti + 1)];\n };\n\n // Format 1: day[-day]: time (Mon: 8am-5pm / Monday - Friday: 8am - 5pm)\n const rx1 = new RegExp(\n `(${dayAny})\\\\s*(?:[-–]\\\\s*(${dayAny}))?\\\\s*[:\\\\s]\\\\s*(${timeRange}|closed|by\\\\s+appointment)`,\n \"gi\",\n );\n let m;\n while ((m = rx1.exec(text)) !== null) {\n const days = m[2] ? expand(m[1], m[2]) : [m[1]];\n const time = m[3].trim();\n days.forEach((d) => {\n if (!hours[d]) hours[d] = time;\n });\n }\n\n // Format 2: time day[-day] (8:30am - 5:00pm Mon - Fri)\n const rx2 = new RegExp(\n `(${timeRange})\\\\s+(${dayAny})\\\\s*(?:[-–]\\\\s*(${dayAny}))?`,\n \"gi\",\n );\n while ((m = rx2.exec(text)) !== null) {\n const time = m[1].trim();\n const days = m[3] ? expand(m[2], m[3]) : [m[2]];\n days.forEach((d) => {\n if (!hours[d]) hours[d] = time;\n });\n }\n\n // Format 3: Mon-Fri alone on a line following a time range on prev line\n const lines = text.split(/\\n|\\r|<br\\s*\\/?>/i).map((l) =>\n l.replace(/<[^>]+>/g, \"\").trim()\n ).filter(Boolean);\n for (let i = 0; i < lines.length - 1; i++) {\n const timeLine = lines[i].match(new RegExp(`^(${timeRange})$`, \"i\"));\n const dayLine = lines[i + 1].match(\n new RegExp(`^(${dayAny})\\\\s*[-–]\\\\s*(${dayAny})$`, \"i\"),\n );\n if (timeLine && dayLine) {\n const time = timeLine[1].trim();\n expand(dayLine[1], dayLine[2]).forEach((d) => {\n if (!hours[d]) hours[d] = time;\n });\n }\n }\n\n return hours;\n}\n\nexport function toTitleCase(slug) {\n return slug.replace(/-/g, \" \").replace(/\\b\\w/g, (c) => c.toUpperCase());\n}\n\nexport function psiScore(val) {\n return val != null ? Math.round(val * 100) : null;\n}",[178],"pulls","branches","forks","envVars",{"_191":151,"_193":63},"connections",{"_191":63,"_192":63},"count","unfilledCount","emptyCount",{"_43":177,"_32":178,"_30":179,"_136":-5,"_41":42,"_40":180,"_138":25,"_38":220,"_13":159,"_33":34,"_35":182,"_24":25,"_142":-5,"_202":221,"_27":222},{"_43":169,"_32":170,"_30":171,"_136":-5,"_41":42,"_40":172,"_138":25,"_38":217,"_13":174,"_33":34,"_35":175,"_24":25,"_142":-5,"_202":218,"_27":219},{"_43":162,"_32":163,"_30":164,"_136":-5,"_41":42,"_40":165,"_138":25,"_38":214,"_13":159,"_33":34,"_35":167,"_24":25,"_142":-5,"_202":215,"_27":216},{"_43":154,"_32":155,"_30":156,"_136":-5,"_41":42,"_40":157,"_138":25,"_38":211,"_13":159,"_33":34,"_35":160,"_24":25,"_142":-5,"_202":212,"_27":213},{"_43":147,"_32":148,"_30":57,"_136":-5,"_41":42,"_40":149,"_138":25,"_38":208,"_13":151,"_33":34,"_35":152,"_24":25,"_142":45,"_202":209,"_27":210},{"_43":144,"_32":28,"_30":31,"_136":-5,"_41":42,"_40":14,"_138":25,"_38":205,"_13":140,"_33":34,"_35":37,"_24":25,"_142":-5,"_202":206,"_27":207},{"_43":133,"_32":134,"_30":135,"_136":-5,"_41":42,"_40":137,"_138":59,"_38":201,"_13":140,"_33":34,"_35":141,"_24":25,"_142":-5,"_202":203,"_27":204},["D",1785457297789],"children",[],[134],["D",1785457310217],[],[28],["D",1785457189282],[],[148],["D",1785457108786],[],[155],["D",1785457192588],[],[163],["D",1785457221412],[],[170],["D",1785457040283],[],[178],["D",1785456196393],{"_97":96,"_104":100},{"_97":96,"_98":-5,"_99":100,"_18":55,"_101":102,"_103":25},{"_13":86,"_32":87,"_43":42,"_88":44,"_89":-5,"_90":-5,"_91":-5,"_92":-5,"_93":-5,"_94":45,"_95":96},"ctx",{"_79":-5,"_123":260,"_261":262,"_263":264,"_265":25,"_266":25,"_267":25,"_268":59,"_269":-5,"_270":-7,"_271":-7,"_272":63,"_273":25,"_274":25,"_275":-5,"_276":25,"_277":25,"_278":25,"_279":-5},"ENV",{"_231":232,"_233":234,"_235":87,"_236":237,"_238":239,"_240":241,"_242":243,"_244":245,"_246":247,"_248":249,"_250":251,"_252":253,"_254":255,"_256":257,"_258":259},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","e87eb04e707cf64e7e532c8e8ec395f209e17f15","RENDER_GIT_BRANCH","POSTHOG_API_TOKEN","phc_nmoyuAkStEZuUp69rYlg1DZgrIt19Gx0Y4Hwf8tGgJP","POSTHOG_API_HOST","https://wart.val.town","WEB_API_TEMPLATE","https://{val}.web.val.run","APP_URL","https://www.val.town","MODULE_URL","https://esm.town","VANITY_WEB_API_TEMPLATE","https://{label}.val.run/","DENOLS_URL","https://lsp.val.town","NODE_ENV","production","CLOUDFLARE_TURNSTILE_SITE_KEY","0x4AAAAAACtaIZJH7s9rBMZ6","GOOGLE_CLIENT_ID","875989696899-jtsqkhlbdv5jrlsk8631p7ipjg3c6o62.apps.googleusercontent.com","STRIPE_PUBLISHABLE_KEY","pk_live_51N2zvWEpZTTLipIkVEwXrUG9AT0yeOOEYxszarZT1jpsZV58Ly6wlg3SltZpHN26CUalOBmUSY7adQRQiSkCZz2500qSXNkf8Y","API_URL","https://api.val.town",[],"preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","preferencesCloseBrackets","preferencesChatPanelState","preferencesDprint","preferencesDenofmt","notificationCount","flagDenols","isBusinessOrgMember","businessOrgHandle","isBusinessOrgAdmin","isHigherTownieTier","navCollapsed","currentOrgId"] </article></body></html>

[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8,"_9":10},"actionData","errors","root",{"_227":228,"_229":230},"routes/_app.x.$handle.$name",{"_97":96,"_32":45,"_105":87,"_27":53,"_13":14,"_106":-5,"_11":107,"_108":109,"_110":111,"_112":113,"_114":115,"_20":-5,"_116":-5,"_117":-5,"_118":25,"_119":120,"_121":25,"_122":-5,"_123":-5,"_124":125},"routes/_app.x.$handle.$name.branch.$branch.version.$version.diff",{"_11":12,"_13":14,"_15":16,"_17":-5,"_18":-5},"project",{"_43":44,"_32":45,"_38":46,"_47":-5,"_48":25,"_49":50,"_51":50,"_52":53,"_54":-5,"_18":55,"_56":57,"_58":59,"_60":-5,"_61":-5,"_62":63,"_64":65,"_66":-5,"_67":-5,"_68":-5,"_69":-5,"_70":71,"_72":71,"_73":74,"_75":74,"_76":74,"_77":71,"_78":71,"_79":80,"_81":82,"_83":84,"_85":25},"version",24,"changes",[19],"aiChatInfo","userId",{"_20":21,"_22":23,"_24":25,"_26":25,"_27":28,"_29":28},"file",{"_30":31,"_32":28,"_33":34,"_35":37,"_38":39,"_40":14,"_41":42,"_24":25},"previous",{"_30":31,"_32":28,"_33":34,"_35":36},"deleted",false,"deletedHere","path","mod.ts","latestPath","fileId","716c9268-8c75-11f1-b03b-1607ee4eb77e","name","fileType","script","content","/**\n * ksite-discovery / mod.mjs — entry point.\n *\n * import { runDiscovery } from \"./mod.mjs\";\n * const record = await runDiscovery(\"https://example.com\");\n *\n * Split across five files for Val.town's 80,000-character-per-file limit:\n * core.mjs fetch, DNS, classification, sitemap parsing\n * extract-html.mjs everything that reads raw HTML\n * extract-page.mjs the markdown side (html2markdown payloads)\n * discovery.mjs one homepage pass, all extractors\n * prefill.mjs registry key + enum mapping\n *\n * Zero dependencies. Needs fetch and AbortSignal.timeout.\n */\n\nimport {\n classify,\n fetchViaProxy,\n PAGE_API,\n parseSitemap,\n TIMEOUT_MS,\n} from \"./core.mjs\";\nimport { fetchSiteDiscovery } from \"./discovery.mjs\";\nimport { extractPageData } from \"./extract-page.mjs\";\nimport { buildIntakePrefill } from \"./prefill.mjs\";\n\n/**\n * Fetch a site's sitemap, run discovery over the homepage, and return the\n * registry-shaped intake prefill alongside the raw discovery record.\n *\n * @param {string} siteUrl e.g. \"https://bluebirdhomecleaning.com\"\n * @param {object} [opts]\n * @param {number} [opts.maxPages=40] cap on pages fetched for per-page extraction\n * @param {boolean} [opts.fetchPages=true] set false for a homepage-only pass\n * @param {string} [opts.pageSelector] CSS selector passed to the markdown API\n * @param {string} [opts.sitemapUrl] explicit sitemap; skips robots.txt discovery\n */\nexport async function runDiscovery(siteUrl, opts = {}) {\n const {\n maxPages = 40,\n fetchPages = true,\n pageSelector = \"\",\n sitemapUrl = \"\",\n } = opts;\n const origin = new URL(siteUrl).origin;\n\n // 1. Locate a sitemap. robots.txt is the authoritative pointer and costs one\n // request; the common paths are the fallback for sites that don't declare\n // it. /sitemap.xml is not a safe assumption — WordPress has shipped\n // /wp-sitemap.xml since 5.5 and Yoast uses /sitemap_index.xml.\n let rows = [];\n let sitemapError = null;\n let sitemapUsed = null;\n\n const candidates = [];\n if (sitemapUrl) {\n candidates.push(sitemapUrl);\n } else {\n try {\n const robots = await fetchViaProxy(`${origin}/robots.txt`);\n for (const line of robots.split(/\\r?\\n/)) {\n const m = line.match(/^\\s*sitemap:\\s*(\\S+)/i);\n if (m) candidates.push(m[1].trim());\n }\n } catch {}\n candidates.push(\n `${origin}/sitemap.xml`,\n `${origin}/sitemap_index.xml`,\n `${origin}/wp-sitemap.xml`,\n `${origin}/sitemap-index.xml`,\n );\n }\n\n for (const candidate of candidates) {\n try {\n const xml = await fetchViaProxy(candidate);\n const { isIndex, entries } = parseSitemap(xml);\n if (!entries.length) continue;\n let locs = entries.map((e) => e.loc);\n if (isIndex) {\n const nested = [];\n for (const loc of locs.slice(0, 5)) {\n try {\n const sub = parseSitemap(await fetchViaProxy(loc));\n nested.push(...sub.entries.map((e) => e.loc));\n } catch {}\n }\n locs = nested;\n }\n if (!locs.length) continue;\n rows = locs.map((url) => ({ url, ...classify(url) }));\n sitemapUsed = candidate;\n break;\n } catch (e) {\n sitemapError = e.message;\n }\n }\n\n // No sitemap is normal for a small site, not a failure. Audit the homepage.\n if (!rows.length) {\n rows = [{ url: siteUrl, ...classify(siteUrl) }];\n } else {\n sitemapError = null;\n }\n\n // 2. Homepage discovery.\n const discovery = await fetchSiteDiscovery(siteUrl, null);\n\n // 3. Per-page extraction. extractPageData() consumes markdown, not HTML —\n // the browser build gets it from the html2markdown Val.town service, and\n // there's no reason to change that here. Sequential on purpose: a burst of\n // parallel requests from a datacentre IP is what gets an audit rate-limited\n // or served a challenge page.\n if (fetchPages) {\n for (const row of rows.slice(0, maxPages)) {\n try {\n const q = `${PAGE_API}?path=${encodeURIComponent(row.url)}` +\n (pageSelector ? `&select=${encodeURIComponent(pageSelector)}` : \"\");\n const res = await fetch(q, { signal: AbortSignal.timeout(TIMEOUT_MS) });\n if (!res.ok) throw new Error(`page API ${res.status}`);\n row.pageData = await res.json();\n row.extracted = extractPageData(row.pageData, row.url);\n row.fetchState = \"done\";\n } catch (e) {\n row.fetchState = \"error\";\n row.error = e.message;\n }\n }\n }\n\n // 4. Registry-shaped output.\n const prefill = buildIntakePrefill({\n discovery,\n rows,\n siteSummary: null,\n siteLabel: siteUrl,\n perf: null,\n });\n\n return {\n siteUrl,\n sitemapUsed,\n sitemapError,\n rowCount: rows.length,\n discovery,\n rows,\n ...prefill,\n };\n}\n\nexport * from \"./core.mjs\";\nexport * from \"./extract-html.mjs\";\nexport * from \"./extract-page.mjs\";\nexport * from \"./discovery.mjs\";\nexport * from \"./prefill.mjs\";","/**\n * ksite-discovery / mod.mjs — entry point.\n *\n * import { runDiscovery } from \"./mod.mjs\";\n * const record = await runDiscovery(\"https://example.com\");\n *\n * Split across five files for Val.town's 80,000-character-per-file limit:\n * core.mjs fetch, DNS, classification, sitemap parsing\n * extract-html.mjs everything that reads raw HTML\n * extract-page.mjs the markdown side (html2markdown payloads)\n * discovery.mjs one homepage pass, all extractors\n * prefill.mjs registry key + enum mapping\n *\n * Zero dependencies. Needs fetch and AbortSignal.timeout.\n */\n\nimport {\n classify,\n fetchViaProxy,\n PAGE_API,\n parseSitemap,\n TIMEOUT_MS,\n} from \"./core.ts\";\nimport { fetchSiteDiscovery } from \"./discovery.ts\";\nimport { extractPageData } from \"./extract-page.ts\";\nimport { buildIntakePrefill } from \"./prefill.ts\";\n\n/**\n * Fetch a site's sitemap, run discovery over the homepage, and return the\n * registry-shaped intake prefill alongside the raw discovery record.\n *\n * @param {string} siteUrl e.g. \"https://bluebirdhomecleaning.com\"\n * @param {object} [opts]\n * @param {number} [opts.maxPages=40] cap on pages fetched for per-page extraction\n * @param {boolean} [opts.fetchPages=true] set false for a homepage-only pass\n * @param {string} [opts.pageSelector] CSS selector passed to the markdown API\n * @param {string} [opts.sitemapUrl] explicit sitemap; skips robots.txt discovery\n */\nexport async function runDiscovery(siteUrl, opts = {}) {\n const {\n maxPages = 40,\n fetchPages = true,\n pageSelector = \"\",\n sitemapUrl = \"\",\n } = opts;\n const origin = new URL(siteUrl).origin;\n\n // 1. Locate a sitemap. robots.txt is the authoritative pointer and costs one\n // request; the common paths are the fallback for sites that don't declare\n // it. /sitemap.xml is not a safe assumption — WordPress has shipped\n // /wp-sitemap.xml since 5.5 and Yoast uses /sitemap_index.xml.\n let rows = [];\n let sitemapError = null;\n let sitemapUsed = null;\n\n const candidates = [];\n if (sitemapUrl) {\n candidates.push(sitemapUrl);\n } else {\n try {\n const robots = await fetchViaProxy(`${origin}/robots.txt`);\n for (const line of robots.split(/\\r?\\n/)) {\n const m = line.match(/^\\s*sitemap:\\s*(\\S+)/i);\n if (m) candidates.push(m[1].trim());\n }\n } catch {}\n candidates.push(\n `${origin}/sitemap.xml`,\n `${origin}/sitemap_index.xml`,\n `${origin}/wp-sitemap.xml`,\n `${origin}/sitemap-index.xml`,\n );\n }\n\n for (const candidate of candidates) {\n try {\n const xml = await fetchViaProxy(candidate);\n const { isIndex, entries } = parseSitemap(xml);\n if (!entries.length) continue;\n let locs = entries.map((e) => e.loc);\n if (isIndex) {\n const nested = [];\n for (const loc of locs.slice(0, 5)) {\n try {\n const sub = parseSitemap(await fetchViaProxy(loc));\n nested.push(...sub.entries.map((e) => e.loc));\n } catch {}\n }\n locs = nested;\n }\n if (!locs.length) continue;\n rows = locs.map((url) => ({ url, ...classify(url) }));\n sitemapUsed = candidate;\n break;\n } catch (e) {\n sitemapError = e.message;\n }\n }\n\n // No sitemap is normal for a small site, not a failure. Audit the homepage.\n if (!rows.length) {\n rows = [{ url: siteUrl, ...classify(siteUrl) }];\n } else {\n sitemapError = null;\n }\n\n // 2. Homepage discovery.\n const discovery = await fetchSiteDiscovery(siteUrl, null);\n\n // 3. Per-page extraction. extractPageData() consumes markdown, not HTML —\n // the browser build gets it from the html2markdown Val.town service, and\n // there's no reason to change that here. Sequential on purpose: a burst of\n // parallel requests from a datacentre IP is what gets an audit rate-limited\n // or served a challenge page.\n if (fetchPages) {\n for (const row of rows.slice(0, maxPages)) {\n try {\n const q = `${PAGE_API}?path=${encodeURIComponent(row.url)}` +\n (pageSelector ? `&select=${encodeURIComponent(pageSelector)}` : \"\");\n const res = await fetch(q, { signal: AbortSignal.timeout(TIMEOUT_MS) });\n if (!res.ok) throw new Error(`page API ${res.status}`);\n row.pageData = await res.json();\n row.extracted = extractPageData(row.pageData, row.url);\n row.fetchState = \"done\";\n } catch (e) {\n row.fetchState = \"error\";\n row.error = e.message;\n }\n }\n }\n\n // 4. Registry-shaped output.\n const prefill = buildIntakePrefill({\n discovery,\n rows,\n siteSummary: null,\n siteLabel: siteUrl,\n perf: null,\n });\n\n return {\n siteUrl,\n sitemapUsed,\n sitemapError,\n rowCount: rows.length,\n discovery,\n rows,\n ...prefill,\n };\n}\n\nexport * from \"./core.mjs\";\nexport * from \"./extract-html.mjs\";\nexport * from \"./extract-page.mjs\";\nexport * from \"./discovery.mjs\";\nexport * from \"./prefill.mjs\";","createdAt",["D",1785457310217],"branchVersion","branchId","b3fec883-81b7-4085-b377-aa59c244fd21","id","69ac1545-bee9-44a1-9b12-f30bfa51163d","site-discovery",["D",1785456196393],"deletedAt","legacyEnvVars","privacy","public","httpPrivacy","description","","imageUrl","81943710-260d-11f0-91a9-01fde9046867","pinnedFileId","3afecbd0-8c73-11f1-9bf1-1607ee4eb77e","pinPreview",true,"pinMigrated","screenshotVersion","likeCount",0,"orgId","ccea884f-1c02-4098-abb0-afc80dcce4e9","tursoName","tursoHostname","tursoMeasuredAt","pinned","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm","user",{"_97":96,"_104":100},"org",{"_97":96,"_98":-5,"_99":100,"_18":55,"_101":102,"_103":25},"branch",{"_13":86,"_32":87,"_43":42,"_88":44,"_89":-5,"_90":-5,"_91":-5,"_92":-5,"_93":-5,"_94":45,"_95":96},"isOwned",32,"main","projectId","forkedBranchId","forkedBranchVersion","forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","projectName","userHandle","kobami","handle","displayName","avatar","https://img.clerk.com/eyJ0eXBlIjoiZGVmYXVsdCIsImlpZCI6Imluc18yTm10cGhQUklwU0tqZHNiOHNnNGd2ZjJURFUiLCJyaWQiOiJ1c2VyXzJ3U3p1WWRNUDlzTjNVSXBRa2U5VUVIU2xocCJ9","tier","free","canRestrictVals","profileImageUrl","branchName","adoptOrgId",{"_43":44,"_32":45,"_38":223,"_47":-5,"_48":25,"_49":50,"_51":50,"_52":53,"_54":-5,"_18":55,"_56":57,"_58":59,"_60":-5,"_61":-5,"_62":63,"_64":65,"_66":-5,"_67":-5,"_68":-5,"_69":-5,"_70":71,"_72":71,"_73":74,"_75":74,"_76":74,"_77":71,"_78":71,"_79":224,"_81":225,"_83":226,"_85":25},"tree",[194,195,196,197,198,199,200],"files",[194,195,196,197,198,199,200],"counts",{"_184":63,"_185":151,"_186":63,"_187":188,"_189":190},"contents",[126,127,128,129,130,131,132],"upstreamProject","upstreamBranch","emptyEnvVars","flags",[],"hasHttpFiles","grants","orgs","customDomains",{},{"_43":177,"_32":178,"_30":179,"_136":-5,"_41":42,"_40":180,"_138":25,"_38":181,"_13":159,"_33":34,"_35":182,"_24":25,"_142":-5,"_27":183},{"_43":169,"_32":170,"_30":171,"_136":-5,"_41":42,"_40":172,"_138":25,"_38":173,"_13":174,"_33":34,"_35":175,"_24":25,"_142":-5,"_27":176},{"_43":162,"_32":163,"_30":164,"_136":-5,"_41":42,"_40":165,"_138":25,"_38":166,"_13":159,"_33":34,"_35":167,"_24":25,"_142":-5,"_27":168},{"_43":154,"_32":155,"_30":156,"_136":-5,"_41":42,"_40":157,"_138":25,"_38":158,"_13":159,"_33":34,"_35":160,"_24":25,"_142":-5,"_27":161},{"_43":147,"_32":148,"_30":57,"_136":-5,"_41":42,"_40":149,"_138":25,"_38":150,"_13":151,"_33":34,"_35":152,"_24":25,"_142":45,"_27":153},{"_43":144,"_32":28,"_30":31,"_136":-5,"_41":42,"_40":14,"_138":25,"_38":145,"_13":140,"_33":34,"_35":37,"_24":25,"_142":-5,"_27":146},{"_43":133,"_32":134,"_30":135,"_136":-5,"_41":42,"_40":137,"_138":59,"_38":139,"_13":140,"_33":34,"_35":141,"_24":25,"_142":-5,"_27":143},"dd22073e-5330-4fe3-8d19-10d2b32621c0","prefill.ts","61912c64-8c75-11f1-87f3-1607ee4eb77e","parentId",23,"latest",["D",1785457297789],3,"/**\n * ksite-discovery / prefill.mjs\n *\n * Maps discovery output onto intake-registry.yaml keys and enum slugs.\n *\n * Part of a set split for Val.town's 80,000-character-per-file limit. The\n * single-file build is otherwise identical — nothing here is Val-specific.\n */\n\nimport { extractCityFromSlug, psiScore, toTitleCase, UA } from \"./core.ts\";\n\nexport const INTAKE_INTEGRATIONS = {\n \"MaidTech\": \"maidtech\",\n \"ZenMaid\": \"zenmaid\",\n \"HouseCall Pro\": \"housecallpro\",\n \"Jobber\": \"jobber\",\n \"BookingKoala\": \"bookingkoala\",\n \"Google Analytics 4\": \"ga4\",\n \"Google Analytics (UA)\": \"ga4\",\n \"Google Tag Manager\": \"gtm\",\n \"Facebook / Meta Pixel\": \"meta-pixel\",\n \"Meta / Facebook\": \"meta-pixel\",\n \"Podium\": \"review-widget\",\n \"Birdeye\": \"review-widget\",\n \"NiceJob\": \"review-widget\",\n \"Broadly\": \"review-widget\",\n \"Yelp Widget\": \"review-widget\",\n \"Trustpilot\": \"review-widget\",\n \"Trustindex\": \"review-widget\",\n \"Google Reviews widget\": \"review-widget\",\n};\n\nexport const INTAKE_INTEGRATION_CATS = {\n chat: \"chat-widget\",\n reviews: \"review-widget\",\n};\n\nexport const INTAKE_PAGE_MAP = {\n about: \"about\",\n contact: \"contact\",\n estimate: \"estimate\",\n booking: \"booking\",\n \"service-area\": \"service-areas\",\n faq: \"faqs\",\n employment: \"employment\",\n \"gift-cards\": \"gift-cards\",\n \"thank-you\": \"thank-you\",\n blog: \"blog\",\n policies: \"policies\",\n};\n\nexport const INTAKE_PERK_MAP = {\n \"Eco-Friendly Products\": \"eco\",\n \"Client Portal\": \"portal\",\n \"Background-Checked Employees\": \"background-checked\",\n \"Insured\": \"insured\",\n \"Bonded\": \"bonded\",\n \"English Speaking\": \"english\",\n \"Spanish Speaking\": \"spanish\",\n \"Cashless Payment\": \"cashless\",\n \"Licensed\": \"licensed\",\n \"Pet-Friendly\": \"pet-friendly\",\n \"Satisfaction Guarantee\": \"guarantee\",\n \"Same-Day Service\": \"same-day\",\n \"Free Estimates\": \"free-estimates\",\n \"Locally Owned\": \"locally-owned\",\n \"Family Owned\": \"family-owned\",\n \"Veteran Owned\": \"veteran-owned\",\n \"Woman Owned\": \"woman-owned\",\n \"Uniformed Staff\": \"uniformed\",\n \"Supplies Provided\": \"supplies-provided\",\n \"Online Booking\": \"online-booking\",\n \"Recurring Discounts\": \"recurring-discounts\",\n};\n\nexport const INTAKE_ORG_MAP = {\n \"Cleaning for a Reason\": \"cfr\",\n \"Chamber of Commerce\": \"chamber\",\n \"Better Business Bureau\": \"bbb\",\n \"ISSA\": \"issa\",\n \"ARCSI\": \"arcsi\",\n};\n\nexport function platformSlug(cms) {\n if (!cms) return \"\";\n if (/wordpress/i.test(cms)) return \"wordpress\";\n if (/^wix/i.test(cms)) return \"wix\";\n if (/squarespace/i.test(cms)) return \"squarespace\";\n if (/godaddy/i.test(cms)) return \"godaddy\";\n return \"other\";\n}\n\nexport function buildIntakePrefill(\n { discovery, rows = [], siteSummary, siteLabel, perf = null },\n) {\n const d = discovery || {};\n const biz = d.schema?.business || {};\n const ci = d.contactInfo || {};\n const brand = d.brand || {};\n const pa = d.phoneAnalysis || {};\n const vars = d.cssVars || {};\n const gbp = d.gbp || {};\n\n const fields = {};\n // Provenance envelope. `value: null` and `confidence: null` travel together —\n // an absent value doesn't have low confidence, it has no confidence. That\n // distinction is what lets gaps() sort the report: a null confidence ranks\n // last and lands in the Enter bucket, a low one ranks third and lands in\n // Confirm. Emitting \"ask\" as a confidence broke that sort silently.\n const set = (key, value, confidence, evidence) => {\n const empty = value == null || value === \"\" ||\n (Array.isArray(value) && !value.length) ||\n (typeof value === \"object\" && !Array.isArray(value) &&\n !Object.keys(value).length);\n fields[key] = empty\n ? {\n value: null,\n source: \"default\",\n confidence: null,\n evidence: null,\n confirmed_by: null,\n blocked: false,\n blocked_reason: null,\n }\n : {\n value,\n source: \"audit\",\n confidence,\n evidence: evidence || null,\n confirmed_by: null,\n blocked: false,\n blocked_reason: null,\n };\n };\n const firstOf = (key, cands) => {\n for (const [v, c, s] of cands) {\n const ok = Array.isArray(v) ? v.length : (v != null && v !== \"\");\n if (ok) return set(key, v, c, s);\n }\n return set(key, null, null, \"\");\n };\n\n const rowOf = (t) =>\n rows.find((r) => r.type === t && r.status !== \"404\") || null;\n const rowsOf = (t) => rows.filter((r) => r.type === t && r.status !== \"404\");\n // Prefer a fully-fetched homepage row; fall back to the hero Discover captured\n const homeRow = rowOf(\"homepage\");\n const isFetchedHome = homeRow?.fetchState === \"done\" ||\n !!homeRow?.pageData?.markdown;\n const hero = d.homepageHero || {};\n const hx = {\n h1: homeRow?.extracted?.h1 || hero.h1 || \"\",\n h2: homeRow?.extracted?.h2 || hero.h2 || \"\",\n intro: homeRow?.extracted?.intro || hero.intro || \"\",\n };\n\n let origin = \"\", domain = \"\";\n try {\n const u = new URL(rowOf(\"homepage\")?.url || siteLabel);\n origin = u.origin;\n domain = u.hostname.replace(/^www\\./, \"\");\n } catch {}\n\n const addr = [\n biz.address?.streetAddress,\n biz.address?.addressLocality,\n [biz.address?.addressRegion, biz.address?.postalCode].filter(Boolean).join(\n \" \",\n ),\n ].filter(Boolean).join(\", \");\n\n const schemaSvcs = (d.schema?.services || []).filter((s) => s.name)\n .map((s) => ({ name: s.name, description: s.description || \"\" }));\n // Nav labels → path, so a service page can be named from the menu (the\n // business's own wording) even when the page itself hasn't been fetched.\n const navByPath = {};\n (d.nav || []).forEach((n) => {\n if (!n.text || !n.href) return;\n try {\n const p = new URL(n.href, origin || undefined).pathname.replace(\n /\\/+$/,\n \"\",\n );\n if (p && !(p in navByPath)) navByPath[p] = n.text.trim();\n } catch {}\n });\n const rowSvcs = rowsOf(\"service\")\n .map((r) => {\n let path = \"\";\n try {\n path = new URL(r.url).pathname.replace(/\\/+$/, \"\");\n } catch {}\n return {\n name: navByPath[path] || r.extracted?.h1 || \"\",\n description: r.extracted?.intro || \"\",\n };\n })\n .filter((s) => s.name);\n\n const schemaRvs = (d.schema?.reviews || []).filter((r) => r.text)\n .map((r) => ({\n name: r.name || \"\",\n source: \"google\",\n rating: Number(r.rating) || 5,\n review: r.text,\n }));\n const pageRvs = rows.flatMap((r) => r.extracted?.reviews || [])\n .map((r) => ({\n name: r.name || \"\",\n source: \"\",\n rating: r.rating || 5,\n review: r.text || \"\",\n }))\n .filter((r) => r.review);\n\n const areas = [\n ...new Set(\n rowsOf(\"service-area\").map((r) => {\n const last = (() => {\n try {\n return new URL(r.url).pathname.split(\"/\").filter(Boolean).pop() ||\n \"\";\n } catch {\n return \"\";\n }\n })();\n return last\n ? toTitleCase(extractCityFromSlug(last))\n : (r.extracted?.h1 || \"\");\n }).filter(Boolean),\n ),\n ];\n\n // Legal pages — lifted verbatim, never rewritten\n const polRows = rows.filter((r) =>\n r.type === \"policies\" && r.pageData?.markdown\n );\n const polText = (r) =>\n (r.pageData.markdown || \"\").replace(/^#.*$/gm, \"\").replace(\n /\\n{3,}/g,\n \"\\n\\n\",\n ).trim().slice(0, 4000);\n const disclaimerRow =\n polRows.find((r) => /disclaimer|terms|liability/i.test(r.url)) || null;\n const policiesRow =\n polRows.find((r) =>\n r !== disclaimerRow && /polic|privacy|cancellation/i.test(r.url)\n ) || null;\n\n const hours = (siteSummary?.hours && Object.keys(siteSummary.hours).length)\n ? siteSummary.hours\n : (ci.hours && Object.keys(ci.hours).length ? ci.hours : {});\n\n const color1 = vars[\"--color-primary\"] || vars[\"--primary\"] ||\n vars[\"--color-brand\"] || \"\";\n const color2 = vars[\"--color-secondary\"] || vars[\"--secondary\"] ||\n vars[\"--color-accent\"] || vars[\"--accent\"] || \"\";\n\n // Nav CTA — last conversion-flavoured item wins (usually the button)\n const ctaItem = [...(d.nav || [])].reverse()\n .find((n) =>\n /quote|estimate|book|schedule|get started/i.test(n.text || \"\")\n ) || null;\n let ctaDest = \"\";\n if (ctaItem?.href) {\n const hit = rows.find((r) => {\n try {\n return new URL(r.url).pathname ===\n new URL(ctaItem.href, origin || undefined).pathname;\n } catch {\n return false;\n }\n });\n if (hit) ctaDest = { booking: \"book\" }[hit.type] || hit.type;\n }\n\n const pages = [\n ...new Set(\n rows.filter((r) => r.status !== \"404\").map((r) => INTAKE_PAGE_MAP[r.type])\n .filter(Boolean),\n ),\n ];\n\n // Integrations — map onto the form's vocabulary, overflow into other_intg\n const tp = d.thirdParty || [];\n const mapped = new Set(), extra = [];\n if (d.crm?.platform && d.crm.platform !== \"Unknown\") {\n const slug = INTAKE_INTEGRATIONS[d.crm.platform];\n if (slug) mapped.add(slug);\n else extra.push(d.crm.platform);\n }\n tp.forEach((t) => {\n if ([\"cdn\", \"seo\", \"cms\", \"monitoring\"].includes(t.cat)) return; // infrastructure, not an integration\n // Vendor name first, then the category — any chat vendor is chat-widget,\n // any review vendor is review-widget, and the specific name survives in\n // evidence rather than in the enum.\n const slug = INTAKE_INTEGRATIONS[t.name] || INTAKE_INTEGRATION_CATS[t.cat];\n if (slug) mapped.add(slug);\n else extra.push(t.name);\n });\n const integrationEvidence = tp.length\n ? `script / embed detection: ${tp.map((t) => t.name).join(\", \")}`\n : \"script / embed detection\";\n\n // Access — presence is detectable, credentials are not. Flip the question:\n // don't ask \"do you use GA\", ask \"we found GA4 G-XXXX, do you have login?\"\n const findTp = (rx) => tp.find((t) => rx.test(t.name));\n const ga = findTp(/Google Analytics/i),\n gtm = findTp(/Tag Manager/i),\n ads = findTp(/Google Ads/i);\n const nsList = (d.dns?.NS || []).map((r) => r.data || \"\");\n const cfNs = nsList.find((x) => /cloudflare/i.test(x)) || \"\";\n const accessDetected = {\n cf: {\n present: !!cfNs || !!findTp(/Cloudflare/i),\n evidence: cfNs\n ? `NS: ${cfNs}`\n : (findTp(/Cloudflare/i) ? \"Cloudflare assets on page\" : \"\"),\n },\n gsc: {\n present: !!d.googleVerification,\n evidence: d.googleVerification || \"\",\n },\n ga: { present: !!ga, evidence: ga?.id || (ga ? \"detected on page\" : \"\") },\n gtm: {\n present: !!gtm,\n evidence: gtm?.id || (gtm ? \"detected on page\" : \"\"),\n },\n ads: {\n present: !!ads,\n evidence: ads?.id || (ads ? \"detected on page\" : \"\"),\n },\n gbp: {\n present: !!(gbp.mapsUrl || gbp.reviewsUrl),\n evidence: gbp.mapsUrl || gbp.reviewsUrl || \"\",\n },\n };\n\n // ── FAQs — schema nodes first, then anything scraped off the pages ────────\n const faqKey = (f) =>\n (f.question || \"\").trim().toLowerCase().replace(/\\s+/g, \" \");\n const dedupeFaq = (list) => {\n const seen = new Set(), out = [];\n for (const f of list) {\n const k = faqKey(f);\n if (!k || !f.answer || seen.has(k)) continue;\n seen.add(k);\n out.push({\n question: f.question.trim(),\n answer: String(f.answer).trim(),\n });\n }\n return out;\n };\n const schemaFaqs = dedupeFaq(d.schema?.faqs || []);\n const pageFaqs = dedupeFaq(rows.flatMap((r) => r.extracted?.faqs || []));\n const allFaqs = dedupeFaq([...schemaFaqs, ...pageFaqs]);\n\n // ── perks → registry slugs. Twelve of the eighteen detectors have no slug;\n // they're named in evidence so the finding survives the enum gap.\n const perkSlugs = [\n ...new Set((d.perks || []).map((p) => INTAKE_PERK_MAP[p]).filter(Boolean)),\n ];\n const perkOrphans = (d.perks || []).filter((p) => !INTAKE_PERK_MAP[p]);\n // Should always be empty now that every detector has a slug. Kept as a\n // tripwire: if a detector is added without a map entry, it shows up here\n // instead of vanishing.\n const perkEvidence = \"trust-badge scan\" +\n (perkOrphans.length\n ? ` — DETECTED BUT UNMAPPED (add to INTAKE_PERK_MAP): ${\n perkOrphans.join(\", \")\n }`\n : \"\");\n\n // ── orgs → slugs + chamber_name + orgs_other. Never a joined string: one\n // field carrying card selections, a chamber name and free text can't be\n // reliably split on read (decision #8).\n const orgSlugs = [\n ...new Set((d.orgs || []).map((o) => INTAKE_ORG_MAP[o]).filter(Boolean)),\n ];\n const orgOrphans = (d.orgs || []).filter((o) => !INTAKE_ORG_MAP[o]);\n // Which chamber, not just that there is one. Chambers name themselves\n // consistently enough to pattern-match off the page text.\n const allMarkdown = rows.map((r) => r.pageData?.markdown || \"\").join(\"\\n\");\n const chamberName = orgSlugs.includes(\"chamber\")\n ? (allMarkdown.match(\n /\\b([A-Z][\\w.'&-]*(?:\\s+[A-Z][\\w.'&-]*){0,4}\\s+(?:Area\\s+|Regional\\s+|County\\s+)?Chamber of Commerce)\\b/,\n )?.[1] || \"\")\n : \"\";\n\n // ── founding date — a year, emitted quoted at build (decision #3).\n // brand.founded is an actual year off schema.foundingDate or \"since YYYY\".\n // Deriving it from \"N years in business\" drifts by one for most of the year,\n // so that path is low confidence with the arithmetic shown.\n const foundedDerived = (!brand.founded && brand.years != null)\n ? String(new Date().getFullYear() - Number(brand.years))\n : \"\";\n\n // ── lead_flow — partially inferable, never confidently. /contact/ maps to\n // two flows and the pricing variant isn't visible in markup at all\n // (projection-map: flowToPath is non-reversible). It's a Stage 5 STOP GATE,\n // so a confident wrong answer costs more than an honest uncertain one.\n const hasEstimateP = !!rowOf(\"estimate\");\n const hasContactP = !!rowOf(\"contact\");\n const hasBookingP = !!rowOf(\"booking\") ||\n !!(d.crm?.embedUrl || d.crm?.embedType);\n const leadFlow = hasEstimateP\n ? \"estimate\"\n : (hasBookingP && hasContactP)\n ? \"contact-booking\"\n : hasContactP\n ? \"contact-basic\"\n : \"\";\n const leadFlowWhy = hasEstimateP\n ? \"estimate page in sitemap — cannot tell estimate from estimate-pricing\"\n : (hasBookingP && hasContactP)\n ? \"contact page + booking embed detected\"\n : hasContactP\n ? \"contact page, no booking embed detected\"\n : \"\";\n\n // ── infrastructure not previously surfaced ────────────────────────────────\n const mxRecords = (d.dns?.MX || []).map((r) => r.data).filter(Boolean);\n const blogCount =\n rows.filter((r) => r.type === \"blog\" && r.status !== \"404\").length;\n // PSI mobile performance score, 0-100. Only present once the perf audit runs.\n // Named to avoid shadowing the top-level psiScore() helper — if this local is\n // ever removed, the outer function would silently take its place and\n // page_speed would be assigned a function reference instead of a number.\n const psiMobileScore = (() => {\n const raw = perf?.psi?.mobile?.lighthouseResult?.categories?.performance\n ?.score;\n return typeof raw === \"number\" ? Math.round(raw * 100) : null;\n })();\n const socialOf = (key, rx) =>\n ci.social?.[key] || (biz.sameAs || []).find((u) => rx.test(u)) || \"\";\n\n // ── Field map, in intake-form order (canonical intake keys) ───────────────\n // Identity\n firstOf(\"company_name\", [[biz.name, \"high\", \"schema.name\"], [\n siteSummary?.detectedCompanyName,\n \"medium\",\n \"page heuristics\",\n ]]);\n firstOf(\"company_phone\", [[pa.schema, \"high\", \"schema.telephone\"], [\n biz.telephone,\n \"high\",\n \"schema.telephone\",\n ], [ci.phones?.[0], \"medium\", \"header/footer\"]]);\n firstOf(\"company_email\", [[biz.email, \"high\", \"schema.email\"], [\n ci.emails?.[0],\n \"medium\",\n \"mailto: link\",\n ]]);\n firstOf(\"company_address\", [[addr, \"high\", \"schema.address\"], [\n ci.addresses?.[0],\n \"medium\",\n \"

/ footer\",\n ]]);\n firstOf(\"company_logo\", [[d.assets?.logo, \"high\", \" / schema logo\"]]);\n firstOf(\"url\", [[origin, \"high\", \"sitemap\"]]);\n firstOf(\"domain\", [[domain, \"high\", \"sitemap\"]]);\n // Rating and review count are low by policy, not by detection quality: schema\n // aggregateRating goes stale against GBP on almost every site we audit.\n firstOf(\"company_rating\", [[\n biz.aggregateRating?.ratingValue?.toString(),\n \"low\",\n \"schema.aggregateRating — confirm against GBP\",\n ], [brand.rating, \"low\", `${brand.ratingSource} — confirm against GBP`]]);\n firstOf(\"company_reviewCount\", [[\n biz.aggregateRating?.reviewCount?.toString(),\n \"low\",\n \"schema.aggregateRating — confirm against GBP\",\n ], [\n brand.reviewCount,\n \"low\",\n `${brand.reviewCountSource}${\n brand.reviewCountApprox\n ? \" — approximate ('+'), confirm exact count in GBP\"\n : \" — confirm against GBP\"\n }`,\n ]]);\n firstOf(\"company_foundingDate\", [[\n brand.founded,\n /schema/.test(brand.yearsSource || \"\") ? \"high\" : \"medium\",\n brand.yearsSource || \"\",\n ], [\n foundedDerived,\n \"low\",\n `derived: ${new Date().getFullYear()} − ${brand.years} years (${\n brand.yearsSource || \"page text\"\n }) — confirm, drifts by one`,\n ]]);\n firstOf(\"company_hours\", [[hours, \"medium\", \"footer / schema hours\"]]);\n firstOf(\"areas\", [[areas, \"high\", \"service-area URLs\"]]);\n firstOf(\"company_founder_name\", [[\n brand.owner,\n brand.ownerSource === \"schema\" ? \"high\" : \"medium\",\n brand.ownerSource || \"\",\n ]]);\n firstOf(\"owner_bio\", [[\n rowOf(\"about\")?.extracted?.intro,\n \"medium\",\n \"about page intro\",\n ]]);\n\n // Social — every registry social key, not just the two we happened to emit\n firstOf(\"facebookLink\", [[\n socialOf(\"facebook\", /facebook\\.com/i),\n \"high\",\n \"footer link / schema.sameAs\",\n ]]);\n firstOf(\"instagramLink\", [[\n socialOf(\"instagram\", /instagram\\.com/i),\n \"high\",\n \"footer link / schema.sameAs\",\n ]]);\n firstOf(\"youtubeLink\", [[\n socialOf(\"youtube\", /youtube\\.com|youtu\\.be/i),\n \"high\",\n \"footer link / schema.sameAs\",\n ]]);\n firstOf(\"linkedinLink\", [[\n socialOf(\"linkedin\", /linkedin\\.com/i),\n \"high\",\n \"footer link / schema.sameAs\",\n ]]);\n firstOf(\"twitterLink\", [[\n socialOf(\"twitter\", /twitter\\.com|x\\.com/i),\n \"high\",\n \"footer link / schema.sameAs\",\n ]]);\n firstOf(\"tiktokLink\", [[\n socialOf(\"tiktok\", /tiktok\\.com/i),\n \"high\",\n \"footer link / schema.sameAs\",\n ]]);\n firstOf(\"pinterestLink\", [[\n socialOf(\"pinterest\", /pinterest\\./i),\n \"high\",\n \"footer link / schema.sameAs\",\n ]]);\n firstOf(\"yelpLink\", [[\n socialOf(\"yelp\", /yelp\\.com/i),\n \"high\",\n \"footer link / schema.sameAs\",\n ]]);\n firstOf(\"nextdoorLink\", [[\n socialOf(\"nextdoor\", /nextdoor\\.com/i),\n \"high\",\n \"footer link / schema.sameAs\",\n ]]);\n firstOf(\"angiLink\", [[\n socialOf(\"angi\", /angi\\.com|angieslist\\.com|homeadvisor\\.com/i),\n \"high\",\n \"footer link / schema.sameAs\",\n ]]);\n firstOf(\"thumbtackLink\", [[\n socialOf(\"thumbtack\", /thumbtack\\.com/i),\n \"high\",\n \"footer link / schema.sameAs\",\n ]]);\n firstOf(\"whatsappLink\", [[\n socialOf(\"whatsapp\", /wa\\.me|whatsapp\\.com/i),\n \"high\",\n \"footer link / schema.sameAs\",\n ]]);\n firstOf(\"googleLink\", [[gbp.mapsUrl, \"high\", \"schema.sameAs\"], [\n gbp.reviewsUrl,\n \"medium\",\n \"schema.hasMap\",\n ]]);\n firstOf(\"review_link\", [[gbp.reviewsUrl, \"medium\", \"schema.hasMap\"]]);\n\n // Content\n const heroSrc = isFetchedHome ? \"homepage row\" : \"Discover (homepage HTML)\";\n firstOf(\"hero_title\", [[hx.h1, \"high\", `H1 · ${heroSrc}`]]);\n firstOf(\"hero_subtitle\", [[hx.h2, \"high\", `H2 · ${heroSrc}`]]);\n firstOf(\"hero_body\", [[hx.intro, \"high\", `intro paragraph · ${heroSrc}`]]);\n set(\"benefits\", null, null, \"\");\n firstOf(\"company_services\", [[schemaSvcs, \"high\", \"schema Service nodes\"], [\n rowSvcs,\n \"medium\",\n \"service pages (nav label / H1)\",\n ]]);\n firstOf(\"company_faqs\", [[schemaFaqs, \"high\", \"schema FAQPage nodes\"], [\n allFaqs,\n \"medium\",\n \"Q/A pairs scraped from page markup\",\n ]]);\n firstOf(\"disclaimer\", [[\n disclaimerRow ? polText(disclaimerRow) : \"\",\n \"medium\",\n disclaimerRow ? `verbatim from ${disclaimerRow.url}` : \"\",\n ]]);\n firstOf(\"policies\", [[\n policiesRow ? polText(policiesRow) : \"\",\n \"medium\",\n policiesRow ? `verbatim from ${policiesRow.url}` : \"\",\n ]]);\n firstOf(\"perks\", [[perkSlugs, \"medium\", perkEvidence]]);\n firstOf(\"reviews\", [[schemaRvs, \"high\", \"schema Review nodes\"], [\n pageRvs,\n \"medium\",\n \"page text\",\n ]]);\n // If no review text was scraped but a review WIDGET is on the page, the text is\n // JS-rendered and unreachable by a static fetch. Say so explicitly so a null\n // doesn't read as \"no reviews\" — the client provides them or we pull from GBP.\n const reviewWidget = (d.thirdParty || []).find((t) => t.cat === \"reviews\");\n if (!fields.reviews.value && reviewWidget) {\n fields.reviews = {\n value: null,\n source: \"audit\",\n confidence: null,\n confirmed_by: null,\n evidence: `${reviewWidget.name} review widget detected in page HTML`,\n blocked: true,\n blocked_reason:\n `Review text is rendered by the ${reviewWidget.name} widget and is not reachable by a static fetch. Pull from the Google Business Profile or collect from the client — this is not something a fuller crawl will fix.`,\n };\n }\n firstOf(\"orgs\", [[orgSlugs, \"medium\", \"affiliation scan\"]]);\n firstOf(\"chamber_name\", [[\n chamberName,\n \"medium\",\n \"chamber name matched in page text\",\n ]]);\n firstOf(\"orgs_other\", [[\n orgOrphans.join(\"\\n\"),\n \"medium\",\n \"detected affiliations with no registry slug\",\n ]]);\n set(\"story\", null, null, \"\");\n firstOf(\"tagline\", [[\n brand.tagline,\n brand.taglineSource === \"schema.slogan\" ? \"high\" : \"medium\",\n brand.taglineSource || \"\",\n ]]);\n set(\"differentiator\", null, null, \"\");\n firstOf(\"site_image\", [[\n homeRow?.extracted?.ogImage,\n \"high\",\n \"og:image on the homepage\",\n ]]);\n\n // Design\n firstOf(\"color1\", [[color1, \"high\", \"CSS custom property\"]]);\n firstOf(\"color2\", [[color2, \"medium\", \"CSS custom property\"]]);\n firstOf(\"font_primary\", [[\n d.fontRoles?.heading,\n \"medium\",\n \"font-family on h1–h6\",\n ]]);\n firstOf(\"font_secondary\", [[\n d.fontRoles?.body,\n \"medium\",\n \"font-family on body\",\n ]]);\n set(\"hero_style\", null, null, \"\");\n set(\"feel\", null, null, \"\");\n set(\"inspiration\", null, null, \"\");\n\n // Lead flow\n firstOf(\"lead_flow\", [[leadFlow, \"low\", leadFlowWhy]]);\n firstOf(\"leads_buttonText\", [[\n ctaItem?.text?.trim(),\n \"medium\",\n \"nav CTA label\",\n ]]);\n // data_pid gates Stage 8's test-lead check and only exists on a site that\n // already runs MaidTech — this is the one place in the pipeline that can find\n // an existing one. It is NOT the embed; keep them apart (decision #5).\n firstOf(\"data_pid\", [[\n d.crm?.pid,\n \"high\",\n \"data-pid on the MaidTech widget\",\n ]]);\n firstOf(\"booking_embed\", [[\n d.crm?.embedUrl,\n \"high\",\n `${d.crm?.platform || \"booking\"} ${\n d.crm?.embedType || \"embed\"\n } in page HTML`,\n ]]);\n set(\"lead_emails\", null, null, \"\");\n set(\"crm_forward\", null, null, \"\");\n set(\"crm_name\", null, null, \"\");\n set(\"response_time\", null, null, \"\");\n\n // Build scope\n firstOf(\"pages\", [[pages, \"high\", \"sitemap classification\"]]);\n firstOf(\"integrations\", [[[...mapped], \"high\", integrationEvidence]]);\n firstOf(\"other_intg\", [[\n extra.join(\", \"),\n \"medium\",\n \"unmapped third-party scripts\",\n ]]);\n\n // Ops / infrastructure\n firstOf(\"registrar\", [[d.rdap?.registrar, \"high\", \"RDAP\"]]);\n firstOf(\"host\", [[d.hosting, \"high\", \"A / NS records\"]]);\n set(\"domain_access\", null, null, \"\");\n firstOf(\"email_host\", [[d.emailProvider, \"high\", \"MX records\"]]);\n firstOf(\"mx_provider\", [[\n d.emailProvider,\n \"high\",\n mxRecords.length ? `MX: ${mxRecords.join(\", \")}` : \"MX records\",\n ]]);\n set(\"crm_email\", null, null, \"\");\n // The finding here is the discrepancy itself; the displayed number is the\n // evidence for it, not a field of its own (display_phone was removed).\n firstOf(\"phone_setup\", [[\n pa.hasDiscrepancy ? \"We use a call tracking number for marketing\" : \"\",\n \"medium\",\n pa.hasDiscrepancy ? `displayed ${pa.display} ≠ schema ${pa.schema}` : \"\",\n ]]);\n firstOf(\"site_type\", [[\"existing\", \"high\", \"auditing a live site\"]]);\n firstOf(\"prior_platform\", [[\n platformSlug(d.cms),\n \"high\",\n d.cms ? `platform fingerprint: ${d.cms}` : \"platform fingerprint\",\n ]]);\n firstOf(\"blog_post_count\", [[blogCount, \"high\", \"sitemap classification\"]]);\n firstOf(\"page_speed\", [[\n psiMobileScore,\n \"high\",\n \"PageSpeed Insights — mobile performance score\",\n ]]);\n // access → present keys prefill as \"yes\"; undetected stay null (absence of a\n // signal doesn't prove the tool is absent). Client confirms login/access.\n const anyAccess = Object.values(accessDetected).some((v) => v.present);\n const accessValue = {};\n Object.entries(accessDetected).forEach(([k, v]) => {\n accessValue[k] = v.present ? \"yes\" : null;\n });\n fields.access = {\n value: anyAccess ? accessValue : null,\n source: anyAccess ? \"audit\" : \"default\",\n confidence: anyAccess ? \"low\" : null,\n evidence: anyAccess\n ? \"presence detected — credentials/access must be confirmed with the client\"\n : null,\n confirmed_by: null,\n blocked: false,\n blocked_reason: null,\n detected: accessDetected,\n };\n set(\"access_notes\", null, null, \"\");\n\n const prefill = {};\n Object.entries(fields).forEach(([k, v]) => {\n prefill[k] = v.value;\n });\n\n // Registry confidence scale. `none` is the count of unanswered keys — it is\n // deliberately not a confidence level, just a tally of nulls.\n const byConfidence = { high: 0, medium: 0, low: 0, none: 0 };\n Object.values(fields).forEach((f) => {\n const bucket = f.confidence || \"none\";\n byConfidence[bucket] = (byConfidence[bucket] || 0) + 1;\n });\n const total = Object.keys(fields).length;\n const filled = total - byConfidence.none;\n\n // ── Fetch-gap accounting — distinguish \"not on this site\" from \"not fetched\" ─\n // A page present in the table but not yet fetched blocks specific fields.\n // Without this, a null looks the same whether the data is absent or just\n // uncollected. The homepage hero is exempt: Discover already captured it.\n const isFetched = (r) => r?.fetchState === \"done\" || !!r?.pageData?.markdown;\n const GATED = {\n about: { types: [\"about\"], fields: [\"owner_bio\"] },\n service: { types: [\"service\"], fields: [\"services (descriptions)\"] },\n policies: { types: [\"policies\"], fields: [\"disclaimer\", \"policies\"] },\n };\n const fetchGaps = [];\n for (const g of Object.values(GATED)) {\n const present = rows.filter((r) =>\n g.types.includes(r.type) && r.status !== \"404\"\n );\n const unfetched = present.filter((r) => !isFetched(r));\n if (present.length && unfetched.length) {\n fetchGaps.push({\n pages: unfetched.map((r) => {\n try {\n return new URL(r.url).pathname;\n } catch {\n return r.url;\n }\n }),\n blocks: g.fields,\n });\n }\n }\n const unfetchedCount = fetchGaps.reduce((n, g) => n + g.pages.length, 0);\n const heroFromDiscoveryOnly = !isFetched(homeRow) &&\n !!(hero.h1 || hero.h2 || hero.intro);\n\n return {\n _meta: {\n source: siteLabel || origin,\n domain,\n generatedAt: new Date().toISOString(),\n schemaVersion: \"registry-1\",\n registry:\n \"intake-registry.yaml — key names and enum values are the registry's; this file must not invent either\",\n usage:\n \"`prefill` drops straight into the record. `fields` carries the full provenance envelope { value, source, confidence, evidence, confirmed_by }. `ask` lists keys with no value (Enter bucket). `verify` lists low-confidence values a human must check (Confirm bucket). `blocked` lists keys a static fetch cannot reach. Nothing is guessed — unknown is null, never a plausible-looking value.\",\n },\n coverage: {\n total,\n filled,\n pct: total ? Math.round(filled / total * 100) : 0,\n byConfidence,\n unfetchedCount,\n heroFromDiscoveryOnly,\n },\n // ask = no value at all (registry Enter bucket). verify = a value we found\n // but don't trust (registry Confirm bucket, confidence low).\n ask: Object.entries(fields).filter(([, f]) =>\n f.value === null && !f.blocked\n ).map(([k]) => k),\n verify: Object.entries(fields).filter(([, f]) => f.confidence === \"low\")\n .map(([k]) => k),\n blocked: Object.entries(fields).filter(([, f]) => f.blocked).map((\n [k, f],\n ) => ({ field: k, reason: f.blocked_reason })),\n fetchGaps,\n detected: {\n access: accessDetected,\n phones: pa.all || [],\n fontRoles: d.fontRoles || null,\n },\n prefill,\n fields,\n };\n}","label",[134],"2b11aabe-86ad-4d8c-89b6-3d1d5617d4e9",["D",1785457310217],[28],"c3b1fe98-a357-4e1b-b948-5431738c8ef2","main.ts",19,["D",1785457189282],1,"/**\n * ksite audit endpoint — Val.town HTTP val\n * ---------------------------------------------------------------------------\n * Wraps runDiscovery() from the discovery module and returns the registry-shaped\n * intake prefill as JSON.\n *\n * SETUP\n * 1. This must be a multi-file Val.town PROJECT, not a set of separate vals.\n * The discovery module is ~104k characters, over Val.town's 80k-per-file\n * limit, so it ships as six files that import each other relatively —\n * and separate vals can't do relative imports. Drop these in alongside\n * this file:\n *\n * core.mjs fetch, DNS, URL classification, sitemap parsing\n * extract-html.mjs everything that reads raw HTML\n * extract-page.mjs the markdown side (html2markdown payloads)\n * discovery.mjs one homepage pass, all extractors\n * prefill.mjs registry key + enum mapping\n * mod.mjs entry point — the only one you import\n *\n * 2. Set AUDIT_TOKEN in the project's environment variables. Without it the\n * endpoint is an open URL-fetching proxy that anyone can point anywhere.\n *\n * USAGE\n * GET /?url=https://example.com\n * GET /?url=https://example.com&sitemap=https://example.com/wp-sitemap.xml\n * GET /?url=https://example.com&pages=false homepage only, ~3s\n * GET /?url=https://example.com&maxPages=15\n * GET /?url=https://example.com&format=prefill flat values only\n * POST / { \"url\": \"...\", \"maxPages\": 15, \"sitemap\": \"...\" }\n *\n * Authorization: Bearer $AUDIT_TOKEN (or ?token= for quick browser tests)\n *\n * RESPONSE\n * { ok, ms, siteUrl, sitemapUsed, rowCount, coverage, prefill, fields,\n * ask, verify, blocked, _meta, discovery?, rows? }\n *\n * prefill flat key -> value, drops straight into the intake record\n * fields per key: { value, source, confidence, evidence, confirmed_by,\n * blocked, blocked_reason }\n * ask keys with no value (Enter bucket)\n * verify low-confidence values (Confirm bucket)\n * blocked keys a static fetch can't reach, with reasons\n */\n\nimport { runDiscovery } from \"./mod.mjs\";\n\n// A page-fetching crawl is sequential and Val.town caps how long an HTTP val\n// may run. 25 pages is comfortable; raise it only if you've checked your plan's\n// limit, and prefer pages=false when you just want the homepage signals.\nconst MAX_PAGES_CAP = 60;\nconst DEFAULT_MAX_PAGES = 25;\n\nconst CORS = {\n \"access-control-allow-origin\": \"*\",\n \"access-control-allow-methods\": \"GET, POST, OPTIONS\",\n \"access-control-allow-headers\": \"authorization, content-type\",\n};\n\nconst json = (body: unknown, status = 200) =>\n new Response(JSON.stringify(body, null, 2), {\n status,\n headers: { \"content-type\": \"application/json; charset=utf-8\", ...CORS },\n });\n\n/**\n * This endpoint fetches whatever URL it's handed, so it must not be usable as a\n * probe against private infrastructure. Blocks non-HTTP schemes and the obvious\n * internal ranges. Note this is a literal check on the hostname — it does not\n * stop a public hostname that resolves to a private address (DNS rebinding). If\n * that matters, put the val behind a fixed allowlist of client domains instead.\n */\nfunction assertPublicUrl(raw: string): URL {\n let u: URL;\n try {\n u = new URL(raw);\n } catch {\n throw new Error(`Not a valid URL: ${raw}`);\n }\n if (u.protocol !== \"http:\" && u.protocol !== \"https:\") {\n throw new Error(`Only http and https are allowed, got ${u.protocol}`);\n }\n const h = u.hostname.toLowerCase();\n const blocked = h === \"localhost\" ||\n h === \"0.0.0.0\" ||\n h.endsWith(\".local\") ||\n h.endsWith(\".internal\") ||\n h === \"[::1]\" ||\n /^127\\./.test(h) ||\n /^10\\./.test(h) ||\n /^192\\.168\\./.test(h) ||\n /^169\\.254\\./.test(h) ||\n /^172\\.(1[6-9]|2\\d|3[01])\\./.test(h);\n if (blocked) throw new Error(`Refusing to fetch a private address: ${h}`);\n return u;\n}\n\nfunction authorized(req: Request, url: URL): boolean {\n const expected = Deno.env.get(\"AUDIT_TOKEN\");\n // Fail closed. An unset token is a misconfiguration, not \"auth disabled\".\n if (!expected) return false;\n const header = req.headers.get(\"authorization\") || \"\";\n const bearer = header.replace(/^Bearer\\s+/i, \"\").trim();\n return bearer === expected || url.searchParams.get(\"token\") === expected;\n}\n\nexport default async function (req: Request): Promise {\n if (req.method === \"OPTIONS\") {\n return new Response(null, { status: 204, headers: CORS });\n }\n\n const url = new URL(req.url);\n\n if (!Deno.env.get(\"AUDIT_TOKEN\")) {\n return json(\n { ok: false, error: \"AUDIT_TOKEN is not set on this val\" },\n 500,\n );\n }\n if (!authorized(req, url)) {\n return json({ ok: false, error: \"Unauthorized\" }, 401);\n }\n\n // Params from the query string or a JSON body, whichever the caller used.\n let p: Record = Object.fromEntries(url.searchParams);\n if (req.method === \"POST\") {\n try {\n p = { ...p, ...(await req.json()) };\n } catch {\n return json({ ok: false, error: \"Body was not valid JSON\" }, 400);\n }\n }\n\n const target = p.url || p.site || p.siteUrl;\n if (!target) {\n return json({ ok: false, error: \"Missing `url`\" }, 400);\n }\n\n let siteUrl: URL, sitemapUrl = \"\";\n try {\n siteUrl = assertPublicUrl(String(target));\n if (p.sitemap) sitemapUrl = assertPublicUrl(String(p.sitemap)).href;\n } catch (e) {\n return json({ ok: false, error: (e as Error).message }, 400);\n }\n\n const fetchPages = String(p.pages ?? \"true\") !== \"false\";\n const maxPages = Math.min(\n Number.isFinite(+p.maxPages) && +p.maxPages > 0\n ? +p.maxPages\n : DEFAULT_MAX_PAGES,\n MAX_PAGES_CAP,\n );\n\n const started = Date.now();\n try {\n const out = await runDiscovery(siteUrl.href, {\n maxPages,\n fetchPages,\n sitemapUrl,\n pageSelector: p.selector ? String(p.selector) : \"\",\n });\n\n const base = {\n ok: true,\n ms: Date.now() - started,\n siteUrl: out.siteUrl,\n sitemapUsed: out.sitemapUsed,\n sitemapError: out.sitemapError,\n rowCount: out.rowCount,\n coverage: out.coverage,\n };\n\n // `prefill` is the small payload n8n usually wants to map onto the record.\n // The default includes provenance; `full` adds the raw discovery and rows,\n // which is large and mostly useful for debugging a bad extraction.\n if (p.format === \"prefill\") return json({ ...base, prefill: out.prefill });\n\n const body: Record = {\n ...base,\n prefill: out.prefill,\n fields: out.fields,\n ask: out.ask,\n verify: out.verify,\n blocked: out.blocked,\n _meta: out._meta,\n };\n if (p.format === \"full\") {\n body.discovery = out.discovery;\n body.rows = out.rows;\n }\n return json(body);\n } catch (e) {\n // Return the partial context rather than a bare 500 — knowing it died after\n // 40s on page 12 is the difference between a fix and a guess.\n return json(\n {\n ok: false,\n ms: Date.now() - started,\n siteUrl: siteUrl.href,\n error: (e as Error).message,\n hint:\n \"If this timed out, retry with pages=false for a homepage-only pass, \" +\n \"or lower maxPages.\",\n },\n 502,\n );\n }\n}",[148],"92bb446d-a871-4412-b916-890f7eff26c3","extract-page.ts","4a3f7fe8-8c75-11f1-bbcc-1607ee4eb77e",12,["D",1785457108786],2,"/**\n * ksite-discovery / extract-page.mjs\n *\n * The markdown side: per-page extraction of the html2markdown payload.\n *\n * Part of a set split for Val.town's 80,000-character-per-file limit. The\n * single-file build is otherwise identical — nothing here is Val-specific.\n */\n\nimport { extractHoursFromText } from \"./core.mjs\";\n\nexport function extractCleaningChecklist(md) {\n const tableRx = /(\\|.+\\|\\n\\|[-|: ]+\\|\\n(?:\\|.+\\|\\n?)+)/g;\n let tm;\n const tables = [];\n while ((tm = tableRx.exec(md)) !== null) {\n const rows = tm[1].split(\"\\n\").filter((r) =>\n r.trim() && !/^\\|[-: |]+\\|$/.test(r)\n );\n if (rows.length < 4) continue;\n const headers = rows[0].split(\"|\").slice(1, -1).map((h) => h.trim()).filter(\n Boolean,\n );\n const svcKeywords =\n /residential|deep|move|office|standard|premium|recurring|cleaning|service/i;\n if (!headers.slice(1).some((h) => svcKeywords.test(h))) continue;\n\n const sections = [];\n let cur = null;\n rows.slice(1).forEach((row) => {\n const cells = row.split(\"|\").slice(1, -1).map((c) => c.trim());\n if (!cells.length) return;\n const task = cells[0].replace(/\\*\\*/g, \"\").trim();\n const vals = cells.slice(1);\n // Section header — bold or all-caps with no service values\n if (/^[A-Z &/]+$/.test(task) || vals.every((v) => !v)) {\n cur = { title: task, tasks: [] };\n sections.push(cur);\n return;\n }\n if (!cur) {\n cur = { title: \"General\", tasks: [] };\n sections.push(cur);\n }\n cur.tasks.push({\n task,\n included: vals.map((v) => {\n if (!v) return null;\n if (/^\\*+$/.test(v) || /✓|✔|yes|included/i.test(v)) return \"included\";\n if (/add.?on|extra|optional/i.test(v)) return \"addon\";\n return v;\n }),\n });\n });\n if (sections.some((s) => s.tasks.length > 0)) {\n tables.push({ services: headers.slice(1), sections });\n }\n }\n return tables;\n}\n\nexport function extractPageData(pageData, siteUrl) {\n const empty = {\n h1: \"\",\n h2: \"\",\n intro: \"\",\n wordCount: 0,\n phones: [],\n emails: [],\n internalLinks: [],\n hasForm: false,\n reviewsDetected: false,\n faqsDetected: false,\n faqs: [],\n reviews: [],\n hours: {},\n ogImage: \"\",\n noindex: false,\n cleaningChecklists: [],\n };\n if (!pageData?.markdown) return empty;\n const md = pageData.markdown;\n const fm = pageData.frontmatter || \"\";\n\n const h1m = md.match(/^#{1}\\s+(.+)$/m);\n const h1 = h1m ? h1m[1].replace(/[*_[\\]`]/g, \"\").trim() : \"\";\n\n const h2m = md.match(/^#{2}\\s+(.+)$/m);\n const h2 = h2m ? h2m[1].replace(/[*_[\\]`]/g, \"\").trim() : \"\";\n\n // First substantial paragraph after the H1 — the hero body copy\n let intro = \"\";\n if (h1m) {\n const rest = md.slice(md.indexOf(h1m[0]) + h1m[0].length);\n for (const para of rest.split(/\\n\\s*\\n/)) {\n if (/^\\s*#/.test(para)) continue;\n const t = para.replace(/!\\[.*?\\]\$.*?\$/g, \"\")\n .replace(/\\[([^\\]]+)\\]\$[^)]+\$/g, \"$1\")\n .replace(/^[>*\\-\\s]+/, \"\").replace(/[*_`]/g, \"\")\n .replace(/\\s+/g, \" \").trim();\n if (t.length >= 60) {\n intro = t.slice(0, 400);\n break;\n }\n }\n }\n\n const clean = md.replace(/!\\[.*?\\]\$.*?\$/g, \"\").replace(\n /\\[([^\\]]+)\\]\$[^)]+\$/g,\n \"$1\",\n ).replace(/```[\\s\\S]*?```/g, \"\").replace(/[#*`_>~|]/g, \"\").replace(\n /https?:\\/\\/\\S+/g,\n \"\",\n ).replace(/\\s+/g, \" \").trim();\n const wordCount = clean ? clean.split(/\\s+/).filter(Boolean).length : 0;\n\n const phones = [\n ...new Set((md.match(/\$?\\d{3}\$?[-.\\s]\\d{3}[-.\\s]\\d{4}/g) || []).map((p) =>\n p.trim()\n )),\n ];\n const emails = [\n ...new Set(\n md.match(/[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\\.[a-zA-Z]{2,}/g) || [],\n ),\n ];\n\n let baseDomain = \"\";\n try {\n baseDomain = new URL(siteUrl).hostname;\n } catch {}\n const allLinks = (md.match(/\\]\$([^)#\\s]+)\$/g) || []).map((m) =>\n m.slice(2, -1)\n );\n const internalLinks = [\n ...new Set(allLinks.filter((l) =>\n l.startsWith(\"/\") || (baseDomain && l.includes(baseDomain))\n )),\n ];\n\n const hasForm =\n /First Name|Last Name|Phone\\s*\\\\?\\*|Email\\s*\\\\?\\*|Your Name|Zip\\s*\\\\?\\*/i\n .test(md);\n const reviewsDetected = /\\d[\\s]*\\/[\\s]*5|5[\\s-]star|★{2,}/i.test(md);\n const questionCount = (md.match(/\\?/g) || []).length;\n const faqsDetected = questionCount >= 4 && wordCount > 80;\n\n // Extract FAQs — heading-question followed by answer paragraph\n const faqs = [];\n const faqRx =\n /^(#{2,4})\\s+(.{8,120}\\?)\\s*\\n+((?:(?!#{1,4}\\s)[\\s\\S])+?)(?=\\n#{1,4}\\s|$)/gm;\n let fm2;\n while ((fm2 = faqRx.exec(md)) !== null && faqs.length < 25) {\n const q = fm2[2].replace(/[*_[\\]`]/g, \"\").trim();\n const a = fm2[3].replace(/[*_[\\]`#]/g, \"\").replace(/\\n+/g, \" \").trim()\n .slice(0, 500);\n if (q && a && a.split(\" \").length > 4) {\n faqs.push({ question: q, answer: a });\n }\n }\n // Fallback: plain question lines\n if (faqs.length < 3) {\n const plainRx =\n /^([A-Z][^?\\n]{8,100}\\?)\\s*\\n+((?:(?![A-Z][^?\\n]{8,100}\\?)[\\s\\S]){20,400}?)(?=\\n[A-Z][^?\\n]{8,100}\\?|\\n#|$)/gm;\n let pm;\n while ((pm = plainRx.exec(md)) !== null && faqs.length < 25) {\n const q = pm[1].replace(/[*_[\\]`]/g, \"\").trim();\n const a = pm[2].replace(/[*_[\\]`#]/g, \"\").replace(/\\n+/g, \" \").trim()\n .slice(0, 500);\n if (\n q && a && a.split(\" \").length > 4 && !faqs.find((f) => f.question === q)\n ) faqs.push({ question: q, answer: a });\n }\n }\n\n // Extract reviews — quoted/blockquoted text attributed to a Capitalized name.\n // Accepts straight + curly quotes and hyphen / en-dash / em-dash attribution.\n const reviews = [];\n const revRx =\n /(?:[\"“”]([^\"“”]{20,300})[\"“”]|(?:^>?\\s*)(.{20,300}))[\\s\\S]{0,100}?[-–—]\\s*([A-Z][a-zA-Z.']+(?:\\s+[A-Z][a-zA-Z.']+)?)/gm;\n let rv;\n while ((rv = revRx.exec(md)) !== null && reviews.length < 15) {\n const text = (rv[1] || rv[2] || \"\").replace(/[*_[\\]]/g, \"\").trim();\n const name = rv[3]?.trim();\n if (name && text && text.split(\" \").length > 5 && !/^\\s*#/.test(rv[0])) {\n if (!reviews.find((r) => r.name === name)) {\n reviews.push({ name, rating: 5, text: text.slice(0, 300) });\n }\n }\n }\n\n const hours = extractHoursFromText(md);\n\n const ogImgMatch = fm.match(/\"og:image\":\\s*\"([^\"]+)\"/);\n const ogImage = ogImgMatch?.[1] || \"\";\n const noindex = /[\"']robots[\"'][\\s\\S]{0,60}noindex/i.test(fm);\n const cleaningChecklists = extractCleaningChecklist(md);\n\n return {\n h1,\n h2,\n intro,\n wordCount,\n phones,\n emails,\n internalLinks,\n hasForm,\n reviewsDetected,\n faqsDetected,\n faqs,\n reviews,\n hours,\n ogImage,\n noindex,\n cleaningChecklists,\n };\n}",[155],"75dc6bbb-cf68-4898-9a9e-823828829ffd","extract-html.ts","4263d1c0-8c75-11f1-a2d5-1607ee4eb77e",20,["D",1785457192588],"/**\n * ksite-discovery / extract-html.mjs\n *\n * Everything that reads a raw HTML string: schema, nav, contact, brand, assets, CSS, third-party scripts.\n *\n * Part of a set split for Val.town's 80,000-character-per-file limit. The\n * single-file build is otherwise identical — nothing here is Val-specific.\n */\n\nimport {\n extractAddressesFromText,\n extractHoursFromText,\n fmtPhone,\n htmlToText,\n normPhone,\n THIRD_PARTY_CHECKS,\n} from \"./core.mjs\";\nimport { buildIntakePrefill } from \"./prefill.mjs\";\n\nexport function extractSchema(html) {\n const schemas = [];\n const rx =\n /]+type=[\"']application\\/ld\\+json[\"'][^>]*>([\\s\\S]*?)<\\/script>/gi;\n let m;\n while ((m = rx.exec(html)) !== null) {\n try {\n schemas.push(JSON.parse(m[1].trim()));\n } catch {}\n }\n const all = [];\n schemas.forEach((s) => {\n if (s[\"@graph\"]) all.push(...s[\"@graph\"]);\n else all.push(s);\n });\n\n const ofType = (types) =>\n all.filter((s) => {\n const t = Array.isArray(s[\"@type\"]) ? s[\"@type\"] : [s[\"@type\"] || \"\"];\n return types.some((type) =>\n t.some((st) => st === type || st.includes(type))\n );\n });\n const first = (types) => ofType(types)[0] || null;\n\n // Business — a page may carry several business-ish nodes (e.g. a sparse\n // Organization AND a rich LocalBusiness with rating/address/phone). Graph\n // order is not priority, so score candidates by completeness + type\n // specificity and take the best, then merge in fields from the others so a\n // value present on any business node is available.\n const BIZ_TYPES = [\n \"LocalBusiness\",\n \"HouseCleaning\",\n \"CleaningService\",\n \"HomeAndConstructionBusiness\",\n \"ProfessionalService\",\n \"Organization\",\n \"Corporation\",\n ];\n const bizCandidates = ofType(BIZ_TYPES);\n const SPECIFIC = new Set([\n \"LocalBusiness\",\n \"HouseCleaning\",\n \"CleaningService\",\n \"HomeAndConstructionBusiness\",\n \"ProfessionalService\",\n ]);\n const bizScore = (n) => {\n let s = 0;\n if (n.aggregateRating) s += 5;\n if (n.address) s += 3;\n if (n.telephone) s += 2;\n if (n.geo) s += 2;\n if (n.openingHours || n.openingHoursSpecification) s += 2;\n if (n.email) s += 1;\n if (n.priceRange) s += 1;\n if (n.hasOfferCatalog || n.makesOffer) s += 1;\n // Type specificity — a LocalBusiness outranks a bare Organization\n const types = Array.isArray(n[\"@type\"]) ? n[\"@type\"] : [n[\"@type\"] || \"\"];\n if (types.some((t) => SPECIFIC.has(t))) s += 4;\n return s;\n };\n const ranked = [...bizCandidates].sort((a, b) => bizScore(b) - bizScore(a));\n let business = ranked[0] || all.find((s) =>\n s[\"@type\"]?.toString().includes(\"Business\")\n ) || null;\n // Backfill: if the top node is missing a field another business node has, borrow it\n if (business && ranked.length > 1) {\n business = { ...business };\n const BACKFILL = [\n \"aggregateRating\",\n \"address\",\n \"telephone\",\n \"geo\",\n \"openingHours\",\n \"openingHoursSpecification\",\n \"email\",\n \"priceRange\",\n \"sameAs\",\n \"foundingDate\",\n \"founder\",\n \"logo\",\n \"image\",\n ];\n for (const node of ranked.slice(1)) {\n for (const f of BACKFILL) {\n if (business[f] == null && node[f] != null) business[f] = node[f];\n }\n }\n }\n\n // FAQs\n const faqs = ofType([\"FAQPage\"]).flatMap((fp) => {\n const items = fp.mainEntity || [];\n return (Array.isArray(items) ? items : [items]).map((item) => ({\n question: item.name || \"\",\n answer: item.acceptedAnswer?.text || item.acceptedAnswer?.[\"@value\"] ||\n \"\",\n })).filter((f) => f.question);\n });\n\n // Services\n const services = ofType([\n \"Service\",\n \"Product\",\n \"Offer\",\n \"ItemList\",\n \"OfferCatalog\",\n ]).flatMap((s) => {\n if ([\"ItemList\", \"OfferCatalog\"].includes(s[\"@type\"])) {\n return (s.itemListElement || s.hasOfferCatalog || []).map((i) => ({\n name: i.name || i.item?.name || \"\",\n url: i.url || i.item?.url || \"\",\n description: i.description || i.item?.description || \"\",\n })).filter((i) => i.name);\n }\n return [{\n name: s.name || \"\",\n url: s.url || \"\",\n description: s.description || \"\",\n price: s.offers?.price || \"\",\n }].filter((i) => i.name);\n });\n\n // Reviews — standalone + embedded in business\n const standaloneReviews = ofType([\"Review\", \"UserReview\"]).map((r) => ({\n name: r.author?.name || \"\",\n rating: r.reviewRating?.ratingValue || \"\",\n text: r.reviewBody || r.description || \"\",\n date: r.datePublished || \"\",\n }));\n const bizReviews =\n (business?.review\n ? (Array.isArray(business.review) ? business.review : [business.review])\n : [])\n .map((r) => ({\n name: r.author?.name || r.name || \"\",\n rating: r.reviewRating?.ratingValue || \"\",\n text: r.reviewBody || r.description || \"\",\n date: r.datePublished || \"\",\n }));\n const reviews = [...standaloneReviews, ...bizReviews].filter((r) =>\n r.name || r.text\n );\n\n // Breadcrumbs\n const breadcrumbs = ofType([\"BreadcrumbList\"]).flatMap((bl) =>\n (bl.itemListElement || []).map((i) => ({\n position: i.position || 0,\n name: i.name || i.item?.name || \"\",\n url: i.item?.[\"@id\"] || i.item || \"\",\n }))\n ).sort((a, b) => a.position - b.position);\n\n // WebSite, WebPage, Article\n const person = first([\"Person\"]);\n const website = first([\"WebSite\"]);\n const webpage = first([\n \"WebPage\",\n \"AboutPage\",\n \"ContactPage\",\n \"CollectionPage\",\n \"ItemPage\",\n ]);\n const article = first([\"Article\", \"BlogPosting\", \"NewsArticle\"]);\n\n // Events\n const events = ofType([\"Event\"]).map((e) => ({\n name: e.name || \"\",\n startDate: e.startDate || \"\",\n location: e.location?.name || \"\",\n url: e.url || \"\",\n })).filter((e) => e.name);\n\n // SameAs\n const sameAs = business?.sameAs\n ? (Array.isArray(business.sameAs) ? business.sameAs : [business.sameAs])\n : [];\n\n // All @types found\n const types = [\n ...new Set(\n all.flatMap((s) =>\n Array.isArray(s[\"@type\"]) ? s[\"@type\"] : [s[\"@type\"] || \"\"]\n ).filter(Boolean),\n ),\n ];\n\n return {\n raw: all,\n types,\n business,\n person,\n faqs,\n services,\n reviews,\n breadcrumbs,\n website,\n webpage,\n article,\n events,\n sameAs,\n };\n}\n\nexport function extractNav(html) {\n // Try

Diff · v24