[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8},"actionData","errors","root",{"_107":108,"_109":110},"routes/embed.x.$handle.$name.$",{"_9":10,"_11":12,"_13":14,"_15":16,"_17":18,"_19":20,"_21":22},"handle","amir5","name","dogsspot_puppy_path","path","SOURCE-BUNDLE.part7.txt","project",{"_23":45,"_11":12,"_34":46,"_47":-5,"_48":42,"_49":50,"_51":50,"_52":53,"_54":-5,"_55":56,"_57":58,"_59":33,"_60":-5,"_61":-5,"_62":63,"_64":65,"_66":67,"_68":69,"_70":71,"_72":-5,"_73":74,"_75":74,"_76":77,"_78":77,"_79":77,"_80":74,"_81":74,"_82":83,"_84":85,"_86":87,"_88":42},"file",{"_23":24,"_11":14,"_25":26,"_27":-5,"_28":29,"_30":31,"_32":33,"_34":35,"_36":37,"_38":17,"_39":40,"_41":42,"_43":-5,"_13":44},"permalink","https://www.val.town/x/amir5/dogsspot_puppy_path/code/SOURCE-BUNDLE.part7.txt","extension","txt","id","a0c319ae-4dbe-437f-99d7-e1e158982a52","fileId","771fe81c-91b7-11f1-b53b-1607ee4eb77e","parentId","branchId","551774fb-a850-4c9d-94a1-d3509ddcf1b0","branchVersion",1221,"latest",true,"createdAt",["D",1786982956788],"version",40,"fileType","content","THE DOGS SPOT — SOURCE BUNDLE (part 7 of 47)\n\nEVERY source file in this project, engine/ included. Restoring these parts and nothing else gives\nyou a tree that builds, typechecks and runs its own test suite with no network and no val.\n\nengine/ is here as of 14 Aug 2026. It used to be left out because val 1bfa6e81-74e1-4c6c-ab32-118304a4ffff\nholds it, which was true of every engine file except engine/watch.ts: a module the val does not run,\nso it was never deployed, so it lived in one container and nothing else. A reclaim destroyed it.\nscripts/deploy.mjs --check still compares engine/ against the val file by file; that is a second\ncheck on the same files now, not the only copy of them.\n\nRESTORE (needs only node, from an empty folder). Save each part, then run this ON EACH ONE:\n\n node -e 'const fs=require(\"fs\"),p=require(\"path\");const b=fs.readFileSync(process.argv[1],\"utf8\");\n for(const s of b.split(\"\\n=====FILE \").slice(1)){const i=s.indexOf(\" =====\\n\");\n const f=s.slice(0,i),c=s.slice(i+7);fs.mkdirSync(p.dirname(f),{recursive:true});\n fs.writeFileSync(f,c.replace(/\\n$/,\"\"))}' PART.txt\n\nA record named #1, #2, ... is one FILE cut into pieces: concatenate them in order, then\nstrip one trailing newline from the last. scripts/restore.mjs does this for you.\n\nEach part is complete in itself: whole files only, no reassembly, any order.\n\nThen: node scripts/build.mjs (engine/ came out of these parts; nothing to fetch first)\n node scripts/typecheck.mjs\n node tests/run-all.mjs\n node scripts/deploy.mjs --check (pulls nothing, proves the val matches what you restored)\n node tests/negative-controls.mjs (proves the suite above can actually fail)\n\nIN THIS PART (3 of 129 files)\n engine/gads-rows.ts 42887 bytes b9c15320bd5a\n engine/gads.ts 5366 bytes c844b4ff05a5\n engine/google-ads-ingest.ts 14131 bytes 0861ffa044c1\n\n================================================================================================\n=====FILE engine/gads-rows.ts =====\n// GOOGLE ADS OFFLINE CONVERSIONS — THE ROW BUILDER, EXTRACTED SO A SECOND ENDPOINT CAN SHARE IT.\n// v1.0, 15 Aug 2026.\n//\n// WHY THIS FILE EXISTS. engine/gads.ts is the hand-driven endpoint: a human hits it with the admin\n// key, gets JSON to read or a CSV to download, and uploads that file by hand. It works and it is\n// verified live, so it is not touched. What it cannot do is be POLLED: it wants the admin key in a\n// query string, it has no cache, and a bad minute at the CRM turns into an HTTP 500 that a machine\n// on a schedule would simply record as a failed run.\n//\n// engine/gads-feed.ts is the machine-facing twin. Both need the same three things — find the\n// Google-clicked contacts, join them to real outcomes, emit Google's exact column set — so those\n// three things live here rather than being typed twice and drifting.\n//\n// THE COLUMN CONTRACT IS PINNED BY A TEST. tests/gads-feed.test.mjs asserts that the header line\n// and the Parameters line in this file appear VERBATIM in engine/gads.ts. If either file's idea of\n// Google's template changes, that test goes red instead of one endpoint quietly emitting a file\n// Google will reject with \"The selected file is missing data columns\".\n//\n// EVERYTHING HERE IS READ-ONLY AGAINST THE CRM. There is no write path in this module at all.\n\n/** The two conversion actions created in Google Ads on 15 Aug 2026. The names must match EXACTLY. */\nexport const LEAD_NAME = \"CRM Lead (Google)\";\nexport const SHOW_NAME = \"CRM Booking Showed (Google)\";\n\n// Google's own live template, fetched 15 Aug 2026 from\n// https://www.gstatic.com/conversiontracking/conversion-import-template.csv\n// SEVEN columns, not the five in most third-party write-ups. The two consent columns are new and\n// omitting them fails Preview with the unhelpful \"The selected file is missing data columns\".\nexport const CSV_HEADER =\n \"Google Click ID,Conversion Name,Conversion Time,Conversion Value,Conversion Currency,Ad User Data,Ad Personalization\";\n// THE PREAMBLE, COPIED FROM GOOGLE'S OWN TEMPLATE, AND WHY IT IS NOT DECORATION.\n// [was: PARAMS_LINE alone, `Parameters:TimeZone=+0000,,,,,,` — six commas, on the theory that the\n// Parameters row must match the header's column count. Google's template uses FOUR, and carries\n// three comment rows and a `### TEMPLATE ###` marker above it. On 16 Aug 2026 a file with the\n// six-comma row and no preamble was refused four times with \"The selected file is missing data\n// columns\"; the byte-for-byte template shape was accepted immediately and parsed all 15 rows.\n// Which half mattered was never isolated, so both are reproduced exactly rather than guessed at\n// a fifth time.]\nexport const PREAMBLE = [\n \"### INSTRUCTIONS ###,,,,\",\n '\"# IMPORTANT: Remember to set the TimeZone value in the \"\"parameters\"\" row and/or in your Conversion Time column\",,,,',\n \"# For instructions on how to set your timezones visit http://goo.gl/T1C5Ov,,,,\",\n \",,,,\",\n \"### TEMPLATE ###,,,,\",\n \"Parameters:TimeZone=+0000,,,,\",\n];\n/** Kept for callers that only want the Parameters row. Four commas, per the template. */\nexport const PARAMS_LINE = \"Parameters:TimeZone=+0000,,,,\";\n// CONSENT: BLANK, and this is a correction, not a preference.\n// [was: \"UNSPECIFIED\", with a comment asserting \"Google processes US traffic normally on\n// UNSPECIFIED.\" Google does not accept it at all. On 16 Aug 2026 every row of a 15-row file was\n// rejected with \"The value 'UNSPECIFIED' in column 'Ad User Data' is invalid.\" Blank is how the\n// column says not-specified, it makes no claim about consent collection either, and the same file\n// with blanks was accepted 14/15 minutes later. If this ever reads UNSPECIFIED again, the whole\n// scheduled feed silently rejects 100% of its rows.]\nexport const CONSENT = \"\";\n\n/** Hard cap on the lookback. Google drops any conversion uploaded more than 90 days after the\n * click; the click precedes contact creation, so the window on contact age must stay short of it. */\nexport const MAX_DAYS = 85;\n/** What the scheduled feed uses. Ten days of margin under Google's 90, because the gap between the\n * click and the contact record being created is real but unmeasured — see NOTES in gads-feed.ts. */\nexport const FEED_DAYS = 80;\n\nconst LOC = \"BQHJ19uohldYe3eqA0bl\";\nconst GHL = \"https://services.leadconnectorhq.com\";\n// Cloudflare in front of the CRM refuses non-browser clients inconsistently. Every read in this\n// project sends a Chrome UA for that reason; python urllib gets a 1010 and curl with this does not.\nconst UA =\n \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0 Safari/537.36\";\n// THE RUNAWAY GUARD, AND WHY 18 WAS ALREADY TOO SMALL.\n//\n// The walk stops when it passes the cutoff; this cap only stops it running forever if that never\n// happens. So it must sit ABOVE the number of contacts the widest legal window can hold, with room\n// to grow, or it silently becomes the thing that ends the walk — and a walk ended early drops the\n// OLDEST Google-clicked contacts, the ones nearest their 90-day expiry, first.\n//\n// Counted 15 Aug 2026 against the live CRM: 1,737 contacts in the last 85 days, 25.2/day over the\n// last 30. Eighty-five days needed EIGHTEEN pages of a cap of eighteen — the live endpoint returned\n// `pages: 18` that morning, with 63 contacts of headroom, about two and a half days of it. The old\n// comment's own premise (\"~25 new contacts/day\") works out to 2,125 contacts over 85 days, which is\n// 22 pages: the cap contradicted the assumption written beside it.\n// 40 pages = 4,000 contacts = ~158 days at today's rate. Pages past the cutoff are never fetched,\n// so raising this costs nothing on any normal run.\nexport const PAGE_CAP = 40;\n\nexport type Row = {\n clickId: string;\n name: string;\n time: string;\n value: number;\n who: string;\n why: string;\n};\n\nexport type Built = {\n rows: Row[];\n scanned: number;\n pages: number;\n clicked: number;\n reachedCutoff: boolean;\n /** True when PAGE_CAP, not the cutoff, ended the walk: the file is short at the OLD end. */\n truncated: boolean;\n /** Google click ids the CSV deliberately does not carry — see the walk for why. */\n braidOnly: { name: string; dateAdded: string; kind: string; id: string }[];\n /** Clicked, but the click was NOT their first touch, so the lead row could not be dated after it.\n * No showed row ships either: without the later click timestamp, an older paid appointment can\n * also predate that click. See clickPrecedesCreation for why this is a drop and not a guess. */\n undatableLead: { name: string; dateAdded: string; id: string }[];\n /** Returning contacts whose showed outcome was withheld because the later click time is unknown. */\n unsafeShowTiming: { name: string; dateAdded: string; id: string }[];\n /** Appointment reads that failed. A scheduled build must never silently publish through these. */\n appointmentFailures: number;\n /** Duplicate Google identity keys removed locally before CSV generation. */\n duplicateRowsDropped: number;\n /** Lead rows held until their +5 minute conversion timestamp is no longer in the future. */\n futureLeadRows: number;\n /** Paid outcomes whose timestamp is invalid, before contact creation, or in the future. */\n invalidOutcomeTimes: number;\n /** Came from an ad, turned up to something free, has not bought. Deliberately NOT a showed row. */\n consultOnly: { name: string; what: string[] }[];\n /** Attended, not free, and on a calendar nobody has classified. NOT counted as a sale — see the\n * ranking note. This list is the queue of calendars somebody needs to price. */\n unclassified: { name: string; what: string; calendarId: string }[];\n /** dateAdded of the NEWEST Google-clicked contact found, or null if the walk found none.\n * gads-feed.ts turns this into daysSinceNewestGclidContact, which is half its alarm. */\n newestClickAt: string | null;\n days: number;\n since: string;\n};\n\n/** How the CRM is read. Injected so the tests can drive the whole builder without a network. */\nexport type Getter = (url: string, version: string) => Promise;\n\n// ─────────────────────────────────────────────────────────────────────────────────────────────────\n// WHICH ATTENDED APPOINTMENTS ARE ACTUALLY A SALE\n//\n// THE BUG THIS EXISTS TO KILL, stated with its receipt. Until 15 Aug 2026 this module took the\n// EARLIEST attended appointment and posted it to Google as \"CRM Booking Showed\" at $150. Read\n// against the live CRM that same evening, of the four contacts that produced a showed row:\n//\n// contact A — free phone consult picked at $150. They DID buy later (a boarding stay), so the\n// value was right by luck and the DATE was five days early, pinned to a free\n// 20-minute call instead of the sale.\n// contact B — free phone consult picked at $150. Their only other appointment is also free.\n// They never bought anything. The file invented $150 of revenue and would have\n// taught Google to go find more people like them.\n// contact C — paid boarding drop-off picked at $150. Correct.\n// contact D — paid program drop-off picked at $150. Correct.\n//\n// So 1 of 4 rows was fabricated and a 2nd was misdated: a 25% invented-revenue rate in a file that\n// was one click from being uploaded into Kim's live account. Optimising bids toward free-consult\n// takers is worse than not measuring at all, because it looks like measurement.\n//\n// HOW THE CLASSIFICATION IS DONE, AND WHY NOT BY TITLE ALONE. Calendar id is the primary key: it is\n// exact, stable across renames, and read from the live calendar list (29 calendars, 15 Aug 2026).\n// Title is a fallback for a calendar created after this list was taken.\n//\n// The title fallback deliberately does NOT match a bare /free/. Two REVENUE calendars are called\n// \"Leash Free Drop Off\" and \"Leash Free Pick-up\" — Leash Free is a paid training program — and a\n// naive /free/ classifies both as giveaways and silently deletes real sales from the file. \"free\"\n// only counts when followed by training or consult.\n// ─────────────────────────────────────────────────────────────────────────────────────────────────\n\n/** Calendars whose appointments cost the customer nothing. Ids read live 15 Aug 2026. */\nexport const FREE_CALENDARS: Record = {\n qHuYJ0NEqzibnoqf5Cw3: \"Pup Talk Phone Call\",\n bLn88LGOYYOZ3NySO1EZ: \"Free Training Day\",\n K24xHYRExFJDEGpgSNVB: \"Free Dog Training Day\",\n \"3kigJYgi2o0BapQfaqEz\": \"Free Training Consultation (Phone)\",\n DBILn7p80AzaqnsCb4XU: \"Free Training Consultation (Zoom)\",\n x9Wbc1UnBAD5JR3s0laV: \"Free Training Consultation (Facility)\",\n // Free by Kim's own written Meet & Greet policy, 12 Aug 2026.\n Uwq6x1shML23UzMSkHz8: \"Meet & Greet\",\n};\n\n/** Fallback for calendars created after the list above. See the note on \"Leash Free\". */\nexport const FREE_TITLE_RE =\n /(pup\\s*talk|meet\\s*(?:&|and)\\s*greet|free\\s+(?:dog\\s+)?training|free\\s+\\w*\\s*consult)/i;\n\n/** GHL ships BOTH `appointmentStatus` and the misspelt `appoinmentStatus` on every event. Read both,\n * because which one is populated is not ours to depend on. */\nexport function apptStatus(e: any): string {\n return String(e?.appointmentStatus ?? e?.appoinmentStatus ?? \"\");\n}\n\n/** True when the customer turned up. */\nexport function attended(e: any): boolean {\n return /showed|completed/i.test(apptStatus(e)) && !!e?.startTime;\n}\n\n/** True when attending this cost the customer nothing, so it is a LEAD signal and not a sale. */\nexport function isFreeAppointment(e: any): boolean {\n if (e?.calendarId && Object.prototype.hasOwnProperty.call(FREE_CALENDARS, e.calendarId)) return true;\n return FREE_TITLE_RE.test(String(e?.title ?? \"\"));\n}\n\n// ─────────────────────────────────────────────────────────────────────────────────────────────────\n// THE RANKING, AND THE ONE THING IT DOES NOT DECIDE\n//\n// WHAT IT CONTROLS. Exactly one showed row is emitted per contact, so the ranking chooses WHICH\n// appointment that row is stamped against — its DATE, and the sentence in `why` that justifies it.\n// It does NOT choose the money. Every showed row carries the same flat $150 floor, because Kim's\n// price list has never been supplied and this project does not invent numbers. So \"most valuable\"\n// here cannot mean \"the biggest dollar figure\" — there are no per-service dollar figures to rank.\n// It means: OF THE THINGS THIS PERSON ATTENDED, WHICH ONE BEST EVIDENCES THAT MONEY CHANGED HANDS.\n// Per-service values are [NEEDS DATA]; when they land, this ranking is where they attach.\n//\n// THE TIERS, and why they sit in this order. Read from the live calendar list — all 29 calendars on\n// location BQHJ19uohldYe3eqA0bl, fetched 15 Aug 2026, not guessed from memory:\n//\n// 4 PROGRAM Basic, Bundle, Bootcamp, Leash Free, Service Dog, Community. Each is a named\n// multi-day board-and-train program with its own Drop Off and Pick-up calendars — the\n// dog stays. A program enrolment is a larger commitment than a night of kennelling.\n// 3 BOARDING Boarding Drop Off / Pick Up, and their two Online twins. A real purchase, smaller.\n// 2 LESSON Private Lesson, DIY Private Lesson, Accelarator Weekly Lesson. Single sessions.\n// 1 UNSURE A calendar nobody has classified, and a title with nothing recognisable in it.\n// Emits NO sale — see below. Counted and named in `unclassified` so it gets fixed.\n// 0 FREE Never a sale. FREE_CALENDARS above, plus the title fallback.\n//\n// WHY \"UNSURE\" DOES NOT PRODUCE A SALE, which is a behaviour change worth stating plainly. The old\n// rule was \"not free, therefore revenue\", so any calendar created after this list was written fell\n// straight into the $150 bucket. That is the SAME shape as the bug this file exists to kill: a\n// thing nobody had checked was assumed to be money. Five live calendars sit in exactly that state\n// today — Pack Walks, Community Group Lesson, P.U.P.S Club Community lesson, and the two remaining\n// class bookings — and nobody here knows whether attending one costs the customer a penny. Not one\n// of the 13 Google-clicked contacts has ever attended any of them, so this costs nothing today.\n// The two ways to be wrong are not symmetric: calling a free thing revenue INVENTS money in Kim's\n// live account, calling a revenue thing unsure merely understates and says so out loud.\n//\n// TIE-BREAK 1, WITHIN A TIER: THE DROP-OFF, NOT THE PICK-UP. A drop-off and its pick-up are ONE\n// purchase logged twice (Stephen: 30 Jul drop, 3 Aug pick. Lauren: 18 Jul drop, 8 Aug pick). The\n// drop-off is when the customer actually converted, and it sits further from Google's\n// 90-days-after-the-click rejection cliff. Louie has only a pick-up on record, so a pick-up still\n// counts when it is the only evidence there is.\n// TIE-BREAK 2: EARLIEST. Same reason — distance from the cliff, and first purchase is the conversion.\n//\n// WHAT THIS CHANGED IN THE OUTPUT, MEASURED RATHER THAN CLAIMED: nothing. Against all 13 live\n// Google-clicked contacts on 15 Aug 2026, the tiered rule and the previous earliest-revenue rule\n// select the identical three appointments, because no contact has ever attended two DIFFERENT\n// revenue tiers. This is a change of GUARANTEE, not of output, and saying otherwise would be the\n// \"built is not proven\" mistake in miniature.\n// ─────────────────────────────────────────────────────────────────────────────────────────────────\n\nexport const RANK = { FREE: 0, UNSURE: 1, LESSON: 2, BOARDING: 3, PROGRAM: 4 } as const;\n\n/** Calendar id -> tier. Ids read live from the location's 29 calendars, 15 Aug 2026. */\nexport const CALENDAR_RANK: Record = {\n // 4 — named multi-day board-and-train programs, each a Drop Off / Pick-up pair.\n \"34JtodEqRp3K2wLp0a0y\": RANK.PROGRAM, // Basic Drop Off\n U53Ci7ndlS0NlkAa6vya: RANK.PROGRAM, // Basic Pick-up\n Zqzo53ckFZafZcaUKyOM: RANK.PROGRAM, // Bundle Drop Off\n \"2sAT9Q61WM2WNTqLqcGj\": RANK.PROGRAM, // Bundle Pick-up\n Qpaf5pzYxNg17K5IRwAl: RANK.PROGRAM, // Bootcamp Pick Up\n drCumXztAXB1MlgbNyqX: RANK.PROGRAM, // Bootcamp Pick-up\n MXqoZqw2t3ewo1Oxja2m: RANK.PROGRAM, // Leash Free Drop Off — \"Free\" in the name, paid program\n QBN6Y6U6lgDufXHz6B2l: RANK.PROGRAM, // Leash Free Pick-up — likewise\n U0sp9FfaU9qJOiWp1Upb: RANK.PROGRAM, // Service Dog Drop Off\n \"8rQdqxN39H6Db3Duf5ZX\": RANK.PROGRAM, // Service Dog Pick-up\n WMnuQPTsY8tz3JaxqPPf: RANK.PROGRAM, // Community Drop Off\n DN6b9L0gVEwBI80v7Ctk: RANK.PROGRAM, // Community Pick-up\n // 3 — the boarding (kennel) stay.\n wS5N8WN4BbzznaLjEg1N: RANK.BOARDING, // Boarding Drop Off\n ZmmjQJszkRMUltfEbumB: RANK.BOARDING, // Boarding Drop Off - Online\n \"1FnbK7pQp1ViZWIzX9SR\": RANK.BOARDING, // Boarding Pick Up\n bN6wWGJa0qKq0QGRg4CC: RANK.BOARDING, // Boarding Pick Up - Online\n // 2 — single paid sessions.\n dZwsM5n1Mu56898lAUrU: RANK.LESSON, // Private Lesson\n \"3XRkHVefDdcol153WXn7\": RANK.LESSON, // DIY - Private Lesson\n zR7eIQ6mToz4YOZIGIl0: RANK.LESSON, // Accelarator Weekly Lesson\n // DELIBERATELY ABSENT, so they land in UNSURE and get reported rather than assumed:\n // F9NJIG5VJO4GkHpJYg9q Pack Walks\n // cCmCBJGScp2R0JvfYlCj Community Group Lesson\n // PF1pYhk5QIrhBD33wUk0 P.U.P.S Club Community lesson\n // Nobody has told us whether these cost the customer anything. When somebody does, add them here.\n};\n\n/** Title fallback for a calendar created after the list above. Ordered: first match wins. */\nexport const TITLE_RANK: [RegExp, number][] = [\n [/(bootcamp|board\\s*(?:&|and)\\s*train|basic|bundle|leash\\s*free|service\\s*dog|accelerat)/i, RANK.PROGRAM],\n [/(boarding|day\\s*ca?re)/i, RANK.BOARDING],\n [/(private\\s*lesson|group\\s*lesson|pack\\s*walk|weekly\\s*lesson)/i, RANK.LESSON],\n];\n\n/** Which tier an appointment sits in. Calendar id first (exact, survives renames), title second. */\nexport function saleRank(e: any): number {\n if (isFreeAppointment(e)) return RANK.FREE;\n // A present-but-unknown id is a new source record, not permission to infer money from prose.\n // Title fallback is only for legacy events that genuinely carry no calendar id.\n if (e?.calendarId) return CALENDAR_RANK[e.calendarId] ?? RANK.UNSURE;\n const title = String(e?.title ?? \"\");\n for (const [re, rank] of TITLE_RANK) if (re.test(title)) return rank;\n return RANK.UNSURE;\n}\n\n/** The six live \"collect the dog at the end\" calendars. Ids read 15 Aug 2026, same list as above. */\nexport const PICKUP_CALENDARS = new Set([\n \"U53Ci7ndlS0NlkAa6vya\", // Basic Pick-up\n \"2sAT9Q61WM2WNTqLqcGj\", // Bundle Pick-up\n \"Qpaf5pzYxNg17K5IRwAl\", // Bootcamp Pick Up\n \"drCumXztAXB1MlgbNyqX\", // Bootcamp Pick-up\n \"QBN6Y6U6lgDufXHz6B2l\", // Leash Free Pick-up\n \"8rQdqxN39H6Db3Duf5ZX\", // Service Dog Pick-up\n \"DN6b9L0gVEwBI80v7Ctk\", // Community Pick-up\n \"1FnbK7pQp1ViZWIzX9SR\", // Boarding Pick Up\n \"bN6wWGJa0qKq0QGRg4CC\", // Boarding Pick Up - Online\n]);\n\n/** True when this is the START of the service rather than the collection at the end of it.\n * Calendar id is exact; the title fallback covers \"Pickup\", \"Pick Up\" and \"Pick-Up\" alike, all\n * three of which appear verbatim in the live records. Anything that is neither — a lesson, a\n * single session — is its own start, which is why the default is true. */\nexport function isServiceStart(e: any): boolean {\n if (e?.calendarId && PICKUP_CALENDARS.has(e.calendarId)) return false;\n return !/pick\\s*[-_ ]?\\s*up/i.test(String(e?.title ?? \"\"));\n}\n\n/**\n * The one appointment that represents the sale, or null if the contact has not bought.\n * Highest tier wins; within a tier the drop-off beats the pick-up; then earliest. See the ranking\n * note above for why each of those three is the way round it is.\n */\nexport function pickSaleAppointment(events: any[]): any | null {\n const revenue = (events ?? []).filter((e) => attended(e) && saleRank(e) > RANK.UNSURE);\n if (!revenue.length) return null;\n revenue.sort((a, b) =>\n saleRank(b) - saleRank(a) ||\n Number(isServiceStart(b)) - Number(isServiceStart(a)) ||\n parseGhlTime(a.startTime).getTime() - parseGhlTime(b.startTime).getTime()\n );\n return revenue[0];\n}\n\n/** The real one. Same headers as every other read in this project. */\nexport function ghlGetter(token: string): Getter {\n if (!token) throw new Error(\"GHL_TOKEN is required\");\n return async (url: string, version: string) => {\n const target = new URL(url, GHL);\n if (target.origin !== GHL || !target.pathname.startsWith(\"/contacts/\")) {\n throw new Error(\"refusing a CRM pagination URL outside the contacts API\");\n }\n const r = await fetch(target.toString(), {\n headers: {\n Authorization: `Bearer ${token}`,\n Version: version,\n \"User-Agent\": UA,\n Accept: \"application/json\",\n },\n });\n if (!r.ok) throw new Error(`${r.status} ${url.slice(0, 120)}`);\n return r.json();\n };\n}\n\n/**\n * Google click ids, wherever GoHighLevel actually keeps them.\n *\n * THE CENSUS THIS IS BUILT ON. Measured 15 Aug 2026 against ALL 5,770 contacts in the CRM — every\n * page of GET /contacts/, not a sample — because the previous matcher was written from the docs and\n * the docs are wrong about the field name:\n *\n * attributions[].utmGclid 81 entries across 50 contacts. The structured field, and the\n * NAME is utmGclid, not gclid — so the old matcher's\n * `\"gclid\":\"…\"` alternative matched ZERO of the 5,770 records\n * in the payload this walk actually reads. Every hit it ever\n * scored came from the other branch. (It does match the DETAIL\n * endpoint's attributionSource.gclid, on 12 of 1,884 checked —\n * but nothing here reads that endpoint.)\n * attributions[].gbraid 31 entries across 20 contacts. Always beside a gclid today.\n * attributions[].pageUrl \"?gclid=\" 81 entries. The ONLY reason extraction worked at all.\n * attributionSource.gclid, the shape GET /contacts/{id} returns. That endpoint returns\n * lastAttributionSource.gclid no attributions[], and the list endpoint returns no\n * attributionSource: the two shapes never co-occur, so a\n * matcher that knows only one of them is one refactor away\n * from finding nothing at all.\n * wbraid 0 in the whole database — but iOS/in-app traffic emits it.\n *\n * WHY THAT WAS FRAGILE RATHER THAN MERELY UGLY. Extraction rested entirely on the click id being\n * DUPLICATED into a URL. It is duplicated in 81 of 81 entries today — and pageUrl is absent from\n * most attribution entries the list endpoint returns at all (6 of 190 on a sampled page carried\n * one). A sparse entry holding utmGclid and no pageUrl was worth nothing, silently: no error, the\n * contact simply is not in the file. On a SCHEDULED feed nobody is watching when that happens.\n *\n * So: match the KEY under any prefix (utmGclid, utm_gclid, lastGclid, gclid) OR the query parameter\n * in any of its three encodings, and return EVERY id found rather than the first — the caller must\n * tell a gclid from a braid, because Google's columns do.\n *\n * Value charset [A-Za-z0-9_-] is verified, not assumed: all 50 live gclids (55–92 chars) and all 31\n * gbraids (34 chars) fall inside it. The bound is 20–200 rather than 20–140 so a longer id is\n * rejected outright instead of being truncated into a plausible, wrong one.\n */\nexport function extractClickIds(raw: string): { gclid?: string; gbraid?: string; wbraid?: string } {\n const found: { gclid?: string; gbraid?: string; wbraid?: string } = {};\n for (const kind of [\"gclid\", \"wbraid\", \"gbraid\"] as const) {\n const m =\n raw.match(new RegExp(`\"[A-Za-z_]*${kind}\"\\\\s*:\\\\s*\"([A-Za-z0-9_-]{20,200})\"`, \"i\")) ||\n // No word-boundary guard on this branch, deliberately. A boundary looks tidier and it broke\n // the doubly-encoded case the moment it was tested: in \"...%253Fgclid%253DCj0…\" the character\n // before \"gclid\" is the F of %253F, which is alphanumeric, so a boundary rejects a real click\n // id. Nothing needs the guard either — \"fbclid\" does not contain \"gclid\", and any token that\n // ends in \"gclid=\" IS a gclid. Proven both ways against the live 5,770 and against synthetic\n // placements before this line was written.\n raw.match(new RegExp(`${kind}(?:=|%3D|%253D)([A-Za-z0-9_-]{20,200})`, \"i\"));\n if (m) found[kind] = m[1];\n }\n return found;\n}\n\n/** Kept as a thin wrapper so any caller written against the old single-id shape still compiles. */\nexport function extractClickId(raw: string): { kind: string; id: string } | null {\n const f = extractClickIds(raw);\n if (f.gclid) return { kind: \"gclid\", id: f.gclid };\n if (f.gbraid) return { kind: \"gbraid\", id: f.gbraid };\n if (f.wbraid) return { kind: \"wbraid\", id: f.wbraid };\n return null;\n}\n\n// TIMEZONE. GHL is inconsistent and it cost us a wrong file on 15 Aug 2026. Contact `dateAdded` is UTC\n// with a Z. Appointment `startTime` is a NAIVE LOCAL string with no offset, in America/New_York. val.town\n// runs TZ=UTC, so `new Date(naive)` read Eastern wall-clock as UTC and stamped every booking FOUR HOURS\n// EARLY. Detected by checking each appointment against the UTC creation time of the contact who booked it:\n// read as UTC, every booking preceded the person who made it. Never patch this with a flat +4h — it breaks\n// when EST returns in November. This module is imported by gads-feed.ts, which serves an endpoint Google\n// polls on a schedule, so a silent 4h error here would repeat forever unattended.\nconst BIZ_TZ = \"America/New_York\";\n\n/** Offset of `tz` from UTC in ms at a given instant. Positive = ahead of UTC. */\nfunction tzOffsetMs(tz: string, atUtcMs: number): number {\n const dtf = new Intl.DateTimeFormat(\"en-US\", {\n timeZone: tz, hour12: false,\n year: \"numeric\", month: \"2-digit\", day: \"2-digit\",\n hour: \"2-digit\", minute: \"2-digit\", second: \"2-digit\",\n });\n const p: Record = {};\n for (const part of dtf.formatToParts(new Date(atUtcMs))) p[part.type] = part.value;\n return Date.UTC(+p.year, +p.month - 1, +p.day, +p.hour % 24, +p.minute, +p.second) - atUtcMs;\n}\n\n/** True when the string carries its own zone — trailing Z, or a +hh:mm / -hh:mm offset. */\nfunction hasExplicitZone(s: string): boolean {\n return /(?:Z|[+-]\\d{2}:?\\d{2})\\s*$/.test(s.trim());\n}\n\n/** Parse a GHL timestamp to a real instant, reading naive strings as business-local, DST-aware. */\nexport function parseGhlTime(s: string): Date {\n const t = String(s).trim();\n if (hasExplicitZone(t)) return new Date(t);\n const m = t.match(/^(\\d{4})-(\\d{2})-(\\d{2})[T ](\\d{2}):(\\d{2})(?::(\\d{2}))?/);\n if (!m) return new Date(t);\n const [, Y, Mo, D, H, Mi, Se] = m;\n const wall = Date.UTC(+Y, +Mo - 1, +D, +H, +Mi, +(Se ?? 0));\n let guess = wall;\n for (let i = 0; i < 2; i++) guess = wall - tzOffsetMs(BIZ_TZ, guess);\n return new Date(guess);\n}\n\n/** Google's documented CSV time shape. No per-row zone: the Parameters line pins UTC for the file. */\n/**\n * DID THE GOOGLE CLICK HAPPEN BEFORE THIS CONTACT EXISTED? (16 Aug 2026.)\n *\n * The lead row's timestamp is `dateAdded + 5 minutes`, and the comment above buildRows justifies it\n * with \"the click precedes contact creation\". That is true of a FIRST touch and false of a returning\n * visitor: a contact created in June who clicks an ad in July still carries that July gclid, and\n * stamping their lead conversion in June puts it before its own click.\n *\n * Google rejects exactly that, in those words — \"The conversion can't occur before the click\" — and\n * it did, on 16 Aug 2026, on one row of a 15-row file.\n *\n * GoHighLevel marks the first attribution `isFirst: true` and OMITS the key on every later one. That\n * is measured, not assumed: across 4,500 live contacts there are 81 attributions carrying a gclid,\n * 43 with `isFirst: true`, 38 with the key absent, and ZERO with `isFirst: false`. The rejected row's\n * contact is in the second group and has two attributions — a first touch with no click id, and a\n * later paid-search chat session carrying the gclid.\n *\n * So this returns true ONLY on an explicit `isFirst === true` beside the click id. Absent, false, or\n * an unreadable shape all mean \"cannot establish\", and the caller drops the lead row rather than\n * date it wrongly. Failing towards \"no row\" costs a signal; failing the other way puts a row Google\n * will reject into a file somebody assumes was accepted.\n */\nexport function clickPrecedesCreation(contact: unknown): boolean {\n const atts = (contact as any)?.attributions;\n if (!Array.isArray(atts)) return false;\n for (const a of atts) {\n if (a?.isFirst !== true) continue;\n if (extractClickIds(JSON.stringify(a ?? {})).gclid) return true;\n }\n return false;\n}\n\nexport function gTime(iso: string, plusMinutes = 0): string {\n const d = new Date(parseGhlTime(iso).getTime() + plusMinutes * 60_000);\n const p = (n: number) => String(n).padStart(2, \"0\");\n return `${d.getUTCFullYear()}-${p(d.getUTCMonth() + 1)}-${p(d.getUTCDate())} ` +\n `${p(d.getUTCHours())}:${p(d.getUTCMinutes())}:${p(d.getUTCSeconds())}`;\n}\n\n/** Byte-for-byte the file engine/gads.ts serves: same quoting, same column order, same consent. */\nexport function toCsv(rows: Row[]): string {\n if (!Array.isArray(rows)) throw new Error(\"conversion rows are not an array\");\n const identities = new Set();\n for (const [i, row] of rows.entries()) {\n if (!/^[A-Za-z0-9_-]{20,200}$/.test(String(row?.clickId ?? \"\"))) throw new Error(`row ${i + 1} has an invalid click id`);\n if (row?.name !== LEAD_NAME && row?.name !== SHOW_NAME) throw new Error(`row ${i + 1} has an unknown conversion name`);\n if (!/^\\d{4}-\\d{2}-\\d{2} \\d{2}:\\d{2}:\\d{2}$/.test(String(row?.time ?? \"\")) ||\n !Number.isFinite(Date.parse(`${row.time.replace(\" \", \"T\")}Z`))) throw new Error(`row ${i + 1} has an invalid conversion time`);\n if (!Number.isFinite(row?.value) || row.value < 0 || row.value > 1_000_000_000) throw new Error(`row ${i + 1} has an invalid conversion value`);\n const identity = `${row.clickId}\\u0000${row.name}\\u0000${row.time}`;\n if (identities.has(identity)) throw new Error(`row ${i + 1} duplicates a Google conversion identity`);\n identities.add(identity);\n }\n const esc = (s: string) => `\"${String(s).replace(/\"/g, '\"\"')}\"`;\n const lines = [\n ...PREAMBLE,\n CSV_HEADER,\n ...rows.map((r) =>\n [esc(r.clickId), esc(r.name), esc(r.time), r.value, \"USD\", CONSENT, CONSENT].join(\",\")\n ),\n ];\n return lines.join(\"\\n\") + \"\\n\";\n}\n\n/**\n * Walk contacts newest-first to the cutoff, keep the ones carrying a Google click id, and join each\n * to its real outcomes.\n *\n * TRAPS ENCODED HERE so nobody relearns them:\n * * Conversion Time must be AFTER the click. The click precedes contact creation, so a lead row\n * is dateAdded + 5 minutes and a showed row is the appointment start, later still.\n * * GET /opportunities/search with date_added_start/end returns HTTP 200 and ZERO rows silently\n * (proved 15 Aug 2026). Nothing here uses those filters: contacts are paginated by\n * meta.nextPageUrl and cut off locally against dateAdded.\n * * One \"showed\" row per contact — the earliest attended appointment that the customer PAID for.\n * Free consults never produce one; see pickSaleAppointment and the receipt above it. Google\n * de-dupes on (click id, conversion name, time), so a later second stay must not become a\n * second sale in the same row; it restates the first.\n * * $150 on a showed row is a deliberately conservative FLOOR, not a price. A boarding stay and a\n * Basic training program are not worth the same and Kim's price list has never been supplied.\n * Understating is the safe direction — Google under-bids on a low value and over-bids on an\n * invented one — but this stays wrong-and-known until real values land. Do not report it to Kim\n * as revenue.\n * * One unreadable contact must not sink the whole file. Appointment reads are individually\n * caught, and the count of failures is returned rather than thrown.\n */\nexport async function buildRows(\n get: Getter,\n opts: { days?: number; leadValue?: number; showValue?: number; now?: number } = {},\n): Promise {\n const rawDays = opts.days ?? FEED_DAYS;\n if (!Number.isFinite(Number(rawDays)) || Number(rawDays) <= 0) throw new Error(\"days must be a positive number\");\n const days = Math.min(Math.max(Math.floor(Number(rawDays)), 1), MAX_DAYS);\n const value = (raw: unknown, fallback: number, label: string) => {\n const n = raw == null ? fallback : Number(raw);\n if (!Number.isFinite(n) || n < 0 || n > 1_000_000_000) throw new Error(`${label} is invalid`);\n return n;\n };\n const leadValue = value(opts.leadValue, 5, \"leadValue\");\n const showValue = value(opts.showValue, 150, \"showValue\");\n const now = opts.now ?? Date.now();\n if (!Number.isFinite(now)) throw new Error(\"now is invalid\");\n const cutoff = now - days * 86_400_000;\n\n const hits: { id: string; name: string; dateAdded: string; addedMs: number; firstTouch: boolean; click: { kind: string; id: string } }[] = [];\n // Clicked contacts whose lead row cannot be dated safely. Counted, never silently dropped.\n const undatableLead: { name: string; dateAdded: string; id: string }[] = [];\n // Contacts whose ONLY Google identifier is a gbraid or wbraid. Google's \"Google Click ID\" column\n // is a GCLID column; a braid posted into it is rejected. Zero contacts in the CRM are in this\n // state today (31 gbraids, every one beside a gclid) — but the old code would have written one\n // into the file as if it were a click id, and a rejected row inside an otherwise accepted file\n // is exactly the failure nobody goes looking for. Counted and returned instead.\n const braidOnly: { name: string; dateAdded: string; kind: string; id: string }[] = [];\n const consultOnly: { name: string; what: string[] }[] = [];\n const unclassified: { name: string; what: string; calendarId: string }[] = [];\n const unsafeShowTiming: { name: string; dateAdded: string; id: string }[] = [];\n let appointmentFailures = 0, futureLeadRows = 0, invalidOutcomeTimes = 0;\n let url: string | null = `${GHL}/contacts/?locationId=${LOC}&limit=100`;\n let pages = 0, scanned = 0, reachedCutoff = false;\n const pageUrls = new Set();\n const contactIds = new Set();\n let previousAdded = Number.POSITIVE_INFINITY;\n while (url && pages < PAGE_CAP && !reachedCutoff) {\n if (pageUrls.has(url)) throw new Error(\"contacts pagination repeated a URL\");\n pageUrls.add(url);\n const page = await get(url, \"2021-07-28\");\n if (!page || !Array.isArray(page.contacts)) throw new Error(`contacts page ${pages + 1} has no contacts array`);\n const contacts = page.contacts;\n for (const c of contacts) {\n scanned++;\n const dateAdded = String(c?.dateAdded ?? \"\");\n const added = Date.parse(dateAdded);\n if (!Number.isFinite(added)) throw new Error(`contact ${String(c?.id ?? \"(missing)\")} has an invalid dateAdded`);\n if (added > now + 5 * 60_000) throw new Error(`contact ${String(c?.id ?? \"(missing)\")} has a future dateAdded`);\n if (added > previousAdded) throw new Error(\"contacts source is not newest-first; refusing an incomplete cutoff walk\");\n previousAdded = added;\n // Finish validating this page even after the first old row. That is how we can prove a later\n // in-window row was not hidden behind an out-of-order cutoff boundary.\n if (added < cutoff) { reachedCutoff = true; continue; }\n const contactId = String(c?.id ?? \"\").trim();\n if (!contactId) throw new Error(\"a contact in the conversion window has no stable id\");\n if (contactIds.has(contactId)) throw new Error(`contacts source repeated id ${contactId}`);\n contactIds.add(contactId);\n const ids = extractClickIds(JSON.stringify(c));\n const who = [c?.firstName, c?.lastName].filter(Boolean).join(\" \") || \"(no name)\";\n if (ids.gclid) {\n hits.push({ id: contactId, name: who, dateAdded, addedMs: added, firstTouch: clickPrecedesCreation(c), click: { kind: \"gclid\", id: ids.gclid } });\n } else if (ids.gbraid || ids.wbraid) {\n braidOnly.push({\n name: who,\n dateAdded,\n kind: ids.gbraid ? \"gbraid\" : \"wbraid\",\n id: (ids.gbraid ?? ids.wbraid) as string,\n });\n }\n }\n const next = page?.meta?.nextPageUrl ?? null;\n if (next !== null && typeof next !== \"string\") throw new Error(`contacts page ${pages + 1} has an invalid nextPageUrl`);\n if (next) {\n const target = new URL(next, GHL);\n if (target.origin !== GHL || !target.pathname.startsWith(\"/contacts/\")) {\n throw new Error(\"contacts pagination pointed outside the CRM contacts API\");\n }\n url = target.toString();\n } else url = null;\n pages++;\n }\n\n const rows: Row[] = [];\n for (const h of hits) {\n // A later click on an existing contact has no timestamp in the list payload. That makes BOTH\n // rows unsafe: dateAdded can predate the click, and the selected paid appointment can also be\n // older than it. Keep the contact visible in diagnostics and emit nothing until the click time\n // is available. This is the exact edge case behind Google's \"conversion before click\" reject.\n if (!h.firstTouch) {\n undatableLead.push({ name: h.name, dateAdded: h.dateAdded, id: h.click.id });\n unsafeShowTiming.push({ name: h.name, dateAdded: h.dateAdded, id: h.click.id });\n continue;\n }\n const leadAt = h.addedMs + 5 * 60_000;\n if (leadAt > now) {\n futureLeadRows++;\n continue;\n }\n rows.push({\n clickId: h.click.id,\n name: LEAD_NAME,\n time: gTime(h.dateAdded, 5),\n value: leadValue,\n who: h.name,\n why: `contact created ${h.dateAdded} carrying a ${h.click.kind}, and that click was their first touch`,\n });\n try {\n const appts = await get(`${GHL}/contacts/${h.id}/appointments`, \"2021-04-15\");\n if (!appts || !Array.isArray(appts.events)) throw new Error(\"appointments response has no events array\");\n const events = appts.events;\n // Validate every known paid event, not just whichever one the ranker selects. Otherwise a\n // valid event can sort ahead of a malformed sibling and make an incomplete source look safe.\n const invalidPaid = new Set();\n for (const event of events) {\n if (!attended(event) || saleRank(event) <= RANK.UNSURE) continue;\n const eventAt = parseGhlTime(String(event?.startTime ?? \"\")).getTime();\n if (!Number.isFinite(eventAt) || eventAt <= h.addedMs || eventAt > now) {\n invalidOutcomeTimes++;\n invalidPaid.add(event);\n }\n }\n const sale = pickSaleAppointment(events.filter((event: any) => !invalidPaid.has(event)));\n if (sale) {\n const saleAt = parseGhlTime(String(sale.startTime ?? \"\")).getTime();\n if (!Number.isFinite(saleAt) || saleAt <= h.addedMs || saleAt > now) {\n invalidOutcomeTimes++;\n } else {\n rows.push({\n clickId: h.click.id,\n name: SHOW_NAME,\n time: gTime(sale.startTime),\n value: showValue,\n who: h.name,\n why: `attended paid service \"${String(sale.title ?? \"appointment\").trim()}\" ` +\n `at ${sale.startTime} (calendar ${sale.calendarId})`,\n });\n }\n } else {\n // NOT a silent drop. A contact who attended only free things is the funnel working — they\n // came from an ad, they turned up, they have not bought YET. It is a real state and the\n // reason this contact is absent from the showed rows, so it is reported rather than lost.\n const freeAttended = events.filter((e: any) => attended(e) && isFreeAppointment(e));\n if (freeAttended.length) {\n consultOnly.push({\n name: h.name,\n what: freeAttended.map((e: any) =>\n `${FREE_CALENDARS[e.calendarId] ?? e.title ?? \"free appointment\"} ${e.startTime}`\n ),\n });\n }\n }\n // Reported whether or not a sale was found: an unclassified calendar is a question, and a\n // contact can easily have both a known sale and an unpriced appointment beside it.\n for (const e of events) {\n if (attended(e) && saleRank(e) === RANK.UNSURE) {\n unclassified.push({\n name: h.name,\n what: `${String(e.title ?? \"appointment\").trim()} ${e.startTime}`,\n calendarId: String(e.calendarId ?? \"\"),\n });\n }\n }\n } catch {\n // Count it. gads-feed.ts treats any non-zero count as an incomplete build and serves the last\n // complete cache instead, so a partial CRM outage cannot silently become Google's new truth.\n appointmentFailures++;\n }\n }\n\n // Google also de-duplicates this identity, but the unattended feed must know its own output is\n // unique before relying on a downstream safety net. Preserve the first occurrence deterministically.\n const seen = new Set();\n const uniqueRows: Row[] = [];\n let duplicateRowsDropped = 0;\n for (const row of rows) {\n const key = `${row.clickId}\\u0000${row.name}\\u0000${row.time}`;\n if (seen.has(key)) { duplicateRowsDropped++; continue; }\n seen.add(key);\n uniqueRows.push(row);\n }\n\n return {\n rows: uniqueRows,\n scanned,\n pages,\n clicked: hits.length,\n reachedCutoff,\n // The walk ran out of pages before it reached the cutoff, so the file is SHORT — and short at\n // the old end, where the contacts closest to their 90-day expiry live. On a scheduled feed this\n // is the difference between \"no conversions this week\" and \"we stopped looking\".\n truncated: !reachedCutoff && !!url && pages >= PAGE_CAP,\n braidOnly,\n undatableLead,\n unsafeShowTiming,\n appointmentFailures,\n duplicateRowsDropped,\n futureLeadRows,\n invalidOutcomeTimes,\n consultOnly,\n unclassified,\n // The walk runs newest-first, so the FIRST hit is the newest — but it is computed by comparing\n // dates rather than trusting that order, because \"the contacts feed is sorted descending\" is an\n // observation about one day's data and this value feeds an alarm that has to hold on every day.\n newestClickAt: hits.reduce(\n (best, h) => (!best || Date.parse(h.dateAdded) > Date.parse(best) ? h.dateAdded : best),\n null,\n ),\n days,\n since: new Date(cutoff).toISOString(),\n };\n}\n\n\n=====FILE engine/gads.ts =====\n// HUMAN GOOGLE ADS OFFLINE-CONVERSION PREVIEW.\n//\n// This endpoint is intentionally a thin authenticated view over gads-rows.ts. The scheduled feed\n// and the file a human inspects must share one extractor, one source walk, one de-duplication key,\n// and one CSV serializer; otherwise the preview can approve bytes the recurring job never sends.\n\nimport {\n buildRows,\n ghlGetter,\n toCsv,\n LEAD_NAME,\n SHOW_NAME,\n MAX_DAYS,\n PAGE_CAP,\n} from \"./gads-rows.ts\";\n\nexport { parseGhlTime } from \"./gads-rows.ts\";\n\ndeclare const Deno: { env: { get(name: string): string | undefined } };\n\nasync function sha256Hex(value: string): Promise {\n const digest = await crypto.subtle.digest(\"SHA-256\", new TextEncoder().encode(value));\n return [...new Uint8Array(digest)].map((b) => b.toString(16).padStart(2, \"0\")).join(\"\");\n}\n\nasync function authorised(candidate: string): Promise {\n const expected = (Deno.env.get(\"NURTURE_ADMIN_KEY_SHA256\") ?? \"\").trim().toLowerCase();\n if (!candidate || !/^[0-9a-f]{64}$/.test(expected)) return false;\n const actual = await sha256Hex(candidate);\n let difference = actual.length ^ expected.length;\n for (let i = 0; i < Math.min(actual.length, expected.length); i++) {\n difference |= actual.charCodeAt(i) ^ expected.charCodeAt(i);\n }\n return difference === 0;\n}\n\nfunction queryNumber(u: URL, name: string, fallback: number): number {\n const raw = u.searchParams.get(name);\n if (raw === null) return fallback;\n if (!raw.trim()) throw new Error(`${name} must be a number`);\n const value = Number(raw);\n if (!Number.isFinite(value)) throw new Error(`${name} must be a finite number`);\n return value;\n}\n\nconst errorResponse = (message: string, status: number) =>\n Response.json({ ok: false, error: message }, {\n status,\n headers: { \"Cache-Control\": \"no-store\", \"X-Content-Type-Options\": \"nosniff\" },\n });\n\nexport default async function (req: Request): Promise {\n const u = new URL(req.url);\n if (!(await authorised(u.searchParams.get(\"k\") ?? \"\"))) return new Response(\"Not found\", { status: 404 });\n if (req.method !== \"GET\" && req.method !== \"HEAD\") return new Response(\"Not found\", { status: 404 });\n\n const token = (Deno.env.get(\"GHL_TOKEN\") ?? \"\").trim();\n if (!token) return errorResponse(\"GHL_TOKEN is not configured\", 503);\n\n let days: number, leadValue: number, showValue: number;\n try {\n days = queryNumber(u, \"days\", MAX_DAYS);\n leadValue = queryNumber(u, \"leadValue\", 5);\n showValue = queryNumber(u, \"showValue\", 150);\n } catch (error) {\n return errorResponse(String(error).replace(/^Error:\\s*/, \"\"), 400);\n }\n\n const generatedAt = new Date().toISOString();\n let built: Awaited>;\n try {\n built = await buildRows(ghlGetter(token), { days, leadValue, showValue });\n } catch (error) {\n return errorResponse(`Google conversion source read failed: ${String(error).replace(/^Error:\\s*/, \"\").slice(0, 240)}`, 503);\n }\n\n const incomplete: string[] = [];\n if (built.truncated) incomplete.push(`contact walk hit its ${PAGE_CAP}-page cap before the cutoff`);\n if (built.appointmentFailures) incomplete.push(`${built.appointmentFailures} appointment read(s) failed`);\n if (built.invalidOutcomeTimes) incomplete.push(`${built.invalidOutcomeTimes} paid outcome timestamp(s) were invalid`);\n if (incomplete.length) {\n return errorResponse(`Refusing a partial Google conversion result: ${incomplete.join(\"; \")}`, 503);\n }\n\n const sourceReceipt = {\n source: \"GoHighLevel contacts and appointments\",\n readAt: generatedAt,\n lookbackDays: built.days,\n since: built.since,\n contactsScanned: built.scanned,\n pagesRead: built.pages,\n reachedCutoff: built.reachedCutoff,\n };\n\n if (u.searchParams.get(\"csv\") === \"1\") {\n if (!built.rows.length) return errorResponse(\"No complete conversion rows are available\", 503);\n let csv: string;\n try { csv = toCsv(built.rows); }\n catch (error) { return errorResponse(`Conversion file validation failed: ${String(error).slice(0, 200)}`, 503); }\n return new Response(req.method === \"HEAD\" ? null : csv, {\n headers: {\n \"Content-Type\": \"text/csv; charset=utf-8\",\n \"Content-Disposition\": `attachment; filename=\"dogsspot-google-offline-${generatedAt.slice(0, 10)}.csv\"`,\n \"Cache-Control\": \"no-store\",\n \"X-Content-Type-Options\": \"nosniff\",\n \"X-Feed-Source\": \"GoHighLevel\",\n \"X-Feed-As-At\": generatedAt,\n \"X-Feed-Rows\": String(built.rows.length),\n },\n });\n }\n\n return Response.json({\n ok: true,\n source: sourceReceipt,\n conversionActions: [LEAD_NAME, SHOW_NAME],\n googleClickedContacts: built.clicked,\n uploadRows: built.rows.length,\n duplicateRowsDropped: built.duplicateRowsDropped,\n withheld: {\n braidOnly: built.braidOnly,\n undatableLead: built.undatableLead,\n unsafeShowTiming: built.unsafeShowTiming,\n futureLeadRows: built.futureLeadRows,\n consultOnly: built.consultOnly,\n unclassified: built.unclassified,\n },\n rows: built.rows,\n uploadHow:\n `Google Ads → Goals → Conversions → Uploads → + → Upload a file. The conversion actions must ` +\n `already exist as imports from clicks and match ${LEAD_NAME} and ${SHOW_NAME} exactly.`,\n }, { headers: { \"Cache-Control\": \"no-store\", \"X-Content-Type-Options\": \"nosniff\" } });\n}\n\n\n=====FILE engine/google-ads-ingest.ts =====\n// PII-FREE GOOGLE ADS SNAPSHOT INGEST.\n//\n// A Google Ads Script posts aggregate campaign metrics here. The endpoint stores campaign names,\n// status, budgets and totals only. Tracking surfaces consume the same validated snapshot and always\n// expose its source timestamp. Older replays cannot replace newer truth.\n\nexport const GOOGLE_ADS_BLOB = \"google_ads_snapshot_v1\";\nexport const GOOGLE_ADS_HISTORY_BLOB = \"google_ads_history_v1\";\nexport const SNAPSHOT_FRESH_MS = 2 * 3_600_000;\nexport const SNAPSHOT_MAX_INGEST_AGE_MS = 30 * 60_000;\nexport const SNAPSHOT_FUTURE_TOLERANCE_MS = 5 * 60_000;\nconst MAX_BODY_BYTES = 200_000;\n\nexport type CampaignMetric = {\n id: string;\n name: string;\n status: \"ENABLED\" | \"PAUSED\" | \"REMOVED\" | \"UNKNOWN\";\n type: string;\n dailyBudget: number;\n today: { impressions: number; clicks: number; cost: number; conversions: number };\n last7Days: { impressions: number; clicks: number; cost: number; conversions: number };\n};\n\nexport type GoogleAdsSnapshot = {\n schema: 1;\n generatedAt: string;\n accountTimeZone: string;\n currency: string;\n campaigns: CampaignMetric[];\n};\n\nexport type SnapshotFreshness = {\n state: \"fresh\" | \"stale\" | \"future\" | \"missing\" | \"invalid\";\n snapshot: GoogleAdsSnapshot | null;\n ageMs: number | null;\n reason: string;\n};\n\nconst METRIC_KEYS = [\"impressions\", \"clicks\", \"cost\", \"conversions\"] as const;\nconst COUNT_KEYS = new Set([\"impressions\", \"clicks\"]);\nconst MAX_COUNT = 1_000_000_000;\nconst MAX_MONEY = 1_000_000_000;\nconst BAD_TEXT = /[\\u0000-\\u001f\\u007f\\u202a-\\u202e\\u2066-\\u2069\\u200e\\u200f\\u061c]/i;\n\nconst assertKeys = (value: Record, allowed: readonly string[], label: string): void => {\n const allow = new Set(allowed);\n const unexpected = Object.keys(value).find((key) => !allow.has(key));\n if (unexpected) throw new Error(`${label} contains an unexpected field: ${unexpected}`);\n};\n\n/** Reject any payload shape that could accidentally turn this aggregate store into a lead dump. */\nexport function assertNoPii(value: unknown): void {\n const walk = (v: unknown, key = \"\") => {\n if (/email|phone|contact|gclid|gbraid|wbraid|address|first.?name|last.?name|url|query|customer/i.test(key)) {\n throw new Error(`PII-like key is not allowed: ${key}`);\n }\n if (typeof v === \"string\") {\n if (/@/.test(v) || /(?:\\+\\d[\\d\\s().-]{8,}\\d|\$\\d{3}\$\\s*\\d{3}|\\d{3}[-.\\s]\\d{3}[-.\\s]\\d{4})/.test(v) ||\n /https?:\\/\\//i.test(v) || /(?:gclid|gbraid|wbraid)=/i.test(v)) {\n throw new Error(\"PII-like value is not allowed\");\n }\n return;\n }\n if (Array.isArray(v)) return v.forEach((x) => walk(x, key));\n if (v && typeof v === \"object\") {\n for (const [k, x] of Object.entries(v as Record)) walk(x, k);\n }\n };\n walk(value);\n}\n\nconst finiteNumber = (v: unknown, max: number): number => {\n if (typeof v !== \"number\") throw new Error(\"metric must be a JSON number\");\n const n = v;\n if (!Number.isFinite(n) || n < 0 || n > max) throw new Error(\"metric is outside the accepted range\");\n return n;\n};\n\nconst cleanText = (v: unknown, label: string, max: number): string => {\n if (typeof v !== \"string\") throw new Error(`invalid ${label}`);\n const s = v.trim();\n if (!s || s.length > max || BAD_TEXT.test(s)) throw new Error(`invalid ${label}`);\n return s;\n};\n\nconst validTimeZone = (v: unknown): string => {\n const s = cleanText(v, \"accountTimeZone\", 80);\n try { new Intl.DateTimeFormat(\"en-US\", { timeZone: s }).format(new Date(0)); }\n catch { throw new Error(\"invalid accountTimeZone\"); }\n return s;\n};\n\nexport function validateSnapshot(raw: unknown): GoogleAdsSnapshot {\n assertNoPii(raw);\n const v = raw as any;\n if (!v || typeof v !== \"object\" || Array.isArray(v)) throw new Error(\"snapshot must be an object\");\n assertKeys(v, [\"schema\", \"generatedAt\", \"accountTimeZone\", \"currency\", \"campaigns\"], \"snapshot\");\n if (v?.schema !== 1) throw new Error(\"unsupported schema\");\n if (typeof v.generatedAt !== \"string\" || !/^\\d{4}-\\d{2}-\\d{2}T\\d{2}:\\d{2}:\\d{2}(?:\\.\\d{1,3})?Z$/.test(v.generatedAt)) {\n throw new Error(\"invalid generatedAt\");\n }\n const generated = Date.parse(v.generatedAt);\n if (!Number.isFinite(generated) || new Date(generated).toISOString() !== v.generatedAt) throw new Error(\"invalid generatedAt\");\n const accountTimeZone = validTimeZone(v.accountTimeZone);\n const currency = typeof v.currency === \"string\" ? v.currency : \"\";\n if (!/^[A-Z]{3}$/.test(currency)) throw new Error(\"invalid currency\");\n if (!Array.isArray(v.campaigns) || v.campaigns.length < 1 || v.campaigns.length > 100) {\n throw new Error(\"campaigns must contain 1 to 100 rows\");\n }\n\n const campaigns = v.campaigns.map((c: any): CampaignMetric => {\n if (!c || typeof c !== \"object\" || Array.isArray(c)) throw new Error(\"invalid campaign row\");\n assertKeys(c, [\"id\", \"name\", \"status\", \"type\", \"dailyBudget\", \"today\", \"last7Days\"], \"campaign\");\n const id = typeof c.id === \"string\" ? c.id : \"\";\n const name = cleanText(c?.name, \"campaign name\", 160);\n const status = typeof c.status === \"string\" ? c.status.toUpperCase() : \"\";\n const type = cleanText(c?.type, \"campaign type\", 80);\n if (!/^\\d{1,30}$/.test(id)) throw new Error(\"invalid campaign id\");\n if (![\"ENABLED\", \"PAUSED\", \"REMOVED\", \"UNKNOWN\"].includes(status)) throw new Error(\"invalid campaign status\");\n const dailyBudget = finiteNumber(c?.dailyBudget, MAX_MONEY);\n const period = (key: \"today\" | \"last7Days\") => {\n if (!c?.[key] || typeof c[key] !== \"object\" || Array.isArray(c[key])) throw new Error(`invalid ${key} metrics`);\n assertKeys(c[key], METRIC_KEYS, key);\n const out: Record = {};\n for (const metric of METRIC_KEYS) {\n const n = finiteNumber(c[key][metric], COUNT_KEYS.has(metric) ? MAX_COUNT : MAX_MONEY);\n if (COUNT_KEYS.has(metric) && !Number.isInteger(n)) throw new Error(`${key}.${metric} must be an integer`);\n out[metric] = +n.toFixed(2);\n }\n return out as CampaignMetric[typeof key];\n };\n return {\n id, name, status: status as CampaignMetric[\"status\"], type,\n dailyBudget: +dailyBudget.toFixed(2), today: period(\"today\"), last7Days: period(\"last7Days\"),\n };\n });\n\n const ids = new Set();\n for (const c of campaigns) {\n if (ids.has(c.id)) throw new Error(\"duplicate campaign id\");\n ids.add(c.id);\n }\n return {\n schema: 1,\n generatedAt: new Date(generated).toISOString(),\n accountTimeZone,\n currency,\n campaigns,\n };\n}\n\n/** Structural validation plus the clock state used by every reader. */\nexport function snapshotFreshness(\n raw: unknown,\n nowMs: number = Date.now(),\n maxAgeMs: number = SNAPSHOT_FRESH_MS,\n): SnapshotFreshness {\n if (raw == null) return { state: \"missing\", snapshot: null, ageMs: null, reason: \"No Google Ads snapshot has been stored.\" };\n let snapshot: GoogleAdsSnapshot;\n try { snapshot = validateSnapshot(raw); }\n catch (e) { return { state: \"invalid\", snapshot: null, ageMs: null, reason: String(e).replace(/^Error:\\s*/, \"\").slice(0, 180) }; }\n const generated = Date.parse(snapshot.generatedAt);\n if (!Number.isFinite(nowMs) || nowMs < 0 || !Number.isFinite(maxAgeMs) || maxAgeMs < 0) {\n return { state: \"invalid\", snapshot: null, ageMs: null, reason: \"The freshness clock is invalid.\" };\n }\n if (generated > nowMs + SNAPSHOT_FUTURE_TOLERANCE_MS) {\n return { state: \"future\", snapshot, ageMs: generated - nowMs, reason: \"The source timestamp is materially in the future.\" };\n }\n const ageMs = Math.max(0, nowMs - generated);\n return ageMs <= maxAgeMs\n ? { state: \"fresh\", snapshot, ageMs, reason: \"The Google Ads Script snapshot is current.\" }\n : { state: \"stale\", snapshot, ageMs, reason: \"The Google Ads Script snapshot has stopped updating.\" };\n}\n\n/** New ingests must be near-real-time. Exact retries are handled before this check. */\nexport function assertSnapshotCurrent(snapshot: GoogleAdsSnapshot, nowMs: number = Date.now()): void {\n const generated = Date.parse(snapshot.generatedAt);\n if (generated > nowMs + SNAPSHOT_FUTURE_TOLERANCE_MS) throw new Error(\"generatedAt is too far in the future\");\n if (nowMs - generated > SNAPSHOT_MAX_INGEST_AGE_MS) throw new Error(\"generatedAt is too old to become current truth\");\n}\n\nasync function sha256Hex(s: string): Promise {\n const b = await crypto.subtle.digest(\"SHA-256\", new TextEncoder().encode(s));\n return [...new Uint8Array(b)].map((x) => x.toString(16).padStart(2, \"0\")).join(\"\");\n}\n\nasync function authorised(key: string): Promise {\n const expected = (Deno.env.get(\"GADS_INGEST_SHA256\") ?? \"\").trim().toLowerCase();\n if (!key || !/^[0-9a-f]{64}$/.test(expected)) return false;\n const got = await sha256Hex(key);\n let diff = got.length ^ expected.length;\n for (let i = 0; i < Math.min(got.length, expected.length); i++) diff |= got.charCodeAt(i) ^ expected.charCodeAt(i);\n return diff === 0;\n}\n\nclass BodyTooLarge extends Error {}\n\nasync function readBody(req: Request): Promise {\n const declaredRaw = req.headers.get(\"content-length\");\n if (declaredRaw != null) {\n const declared = Number(declaredRaw);\n if (!Number.isInteger(declared) || declared < 0) throw new Error(\"invalid content-length\");\n if (declared > MAX_BODY_BYTES) throw new BodyTooLarge(\"payload too large\");\n }\n if (!req.body) throw new Error(\"empty body\");\n const reader = req.body.getReader();\n const chunks: Uint8Array[] = [];\n let total = 0;\n while (true) {\n const { value, done } = await reader.read();\n if (done) break;\n total += value.byteLength;\n if (total > MAX_BODY_BYTES) {\n await reader.cancel().catch(() => undefined);\n throw new BodyTooLarge(\"payload too large\");\n }\n chunks.push(value);\n }\n if (!total) throw new Error(\"empty body\");\n const bytes = new Uint8Array(total);\n let offset = 0;\n for (const chunk of chunks) { bytes.set(chunk, offset); offset += chunk.byteLength; }\n return new TextDecoder(\"utf-8\", { fatal: true }).decode(bytes);\n}\n\nconst noStore = { \"Cache-Control\": \"no-store\", \"X-Content-Type-Options\": \"nosniff\" };\nconst empty204 = (result: string, at: string) => new Response(null, {\n status: 204,\n headers: { ...noStore, \"X-Ingest-Result\": result, \"X-Snapshot-At\": at },\n});\n\n// Blob reads and writes have no compare-and-swap primitive. Serialize requests handled by this\n// isolate so two different source timestamps cannot race and leave the older one as current.\nlet ingestTail: Promise = Promise.resolve();\nfunction serialiseIngest(work: () => Promise): Promise {\n const run = ingestTail.then(work, work);\n ingestTail = run.then(() => undefined, () => undefined);\n return run;\n}\n\nexport default async function (req: Request): Promise {\n if (req.method !== \"POST\") return new Response(\"Not found\", { status: 404, headers: noStore });\n const u = new URL(req.url);\n const key = u.pathname.split(\"/\").filter(Boolean)[0] ?? \"\";\n if (!(await authorised(key))) return new Response(\"Not found\", { status: 404, headers: noStore });\n if (!/^application\\/json(?:\\s*;|$)/i.test(req.headers.get(\"content-type\") ?? \"\")) {\n return new Response(\"Content-Type must be application/json\", { status: 415, headers: noStore });\n }\n\n let raw: unknown;\n try { raw = JSON.parse(await readBody(req)); }\n catch (e) {\n const status = e instanceof BodyTooLarge ? 413 : 400;\n return Response.json({ ok: false, error: String(e).replace(/^Error:\\s*/, \"\").slice(0, 180) }, { status, headers: noStore });\n }\n\n let snapshot: GoogleAdsSnapshot;\n try { snapshot = validateSnapshot(raw); }\n catch (e) { return Response.json({ ok: false, error: String(e).slice(0, 180) }, { status: 400, headers: noStore }); }\n\n return serialiseIngest(async () => { try {\n const { blob } = await import(\"https://esm.town/v/std/blob\");\n const existingRaw = await blob.getJSON(GOOGLE_ADS_BLOB).catch(() => null);\n let existing: GoogleAdsSnapshot | null = null;\n if (existingRaw != null) {\n try { existing = validateSnapshot(existingRaw); }\n catch { existing = null; } // a valid new source reading repairs an invalid current blob\n }\n if (existing) {\n const incomingAt = Date.parse(snapshot.generatedAt), existingAt = Date.parse(existing.generatedAt);\n if (incomingAt === existingAt) {\n if (JSON.stringify(snapshot) === JSON.stringify(existing)) return empty204(\"duplicate\", snapshot.generatedAt);\n return Response.json({ ok: false, error: \"conflicting snapshot at the same generatedAt\" }, { status: 409, headers: noStore });\n }\n if (incomingAt < existingAt) {\n return Response.json({ ok: false, error: \"older snapshot cannot replace newer truth\" }, { status: 409, headers: noStore });\n }\n }\n assertSnapshotCurrent(snapshot);\n\n // Val's blob API returns `null` for a key that has never existed; the offline harness used to\n // model that as a rejected promise. Both shapes mean \"first snapshot\", while a non-null,\n // non-array value is still corruption and must fail closed.\n const historyRaw = await blob.getJSON(GOOGLE_ADS_HISTORY_BLOB).catch(() => null);\n if (historyRaw != null && !Array.isArray(historyRaw)) {\n throw new Error(`Google Ads history blob is not an array (${Object.prototype.toString.call(historyRaw)})`);\n }\n const history: GoogleAdsSnapshot[] = (historyRaw ?? []).map((x) => validateSnapshot(x));\n const day = snapshot.generatedAt.slice(0, 10);\n const nextHistory = [snapshot, ...history.filter((x) => x.generatedAt.slice(0, 10) !== day)]\n .sort((a, b) => Date.parse(b.generatedAt) - Date.parse(a.generatedAt))\n .slice(0, 90);\n\n // History first: if it cannot persist, current truth remains untouched and the source retries.\n await blob.setJSON(GOOGLE_ADS_HISTORY_BLOB, nextHistory);\n await blob.setJSON(GOOGLE_ADS_BLOB, snapshot);\n return empty204(\"stored\", snapshot.generatedAt);\n } catch (e) {\n return Response.json({ ok: false, error: `storage unavailable: ${String(e).replace(/^Error:\\s*/, \"\").slice(0, 140)}` },\n { status: 503, headers: noStore });\n } });\n}\n\n","deleted",false,"label",[14],"1bfa6e81-74e1-4c6c-ab32-118304a4ffff",["D",1785429259053],"deletedAt","legacyEnvVars","privacy","public","httpPrivacy","description","Puppy path nurture engine v1.2 — The Dogs Spot","imageUrl","userId","d26e64e6-cd39-4be1-b5b4-1de11cc555f8","pinnedFileId","ad633fec-8e73-11f1-b800-1607ee4eb77e","pinPreview","pinMigrated","screenshotVersion","likeCount",0,"orgId","019f56fd-bd59-724f-8163-80ffcf8235dd","tursoName","1bfa6e8174e14c6cab32118304a4ffff","tursoHostname","1bfa6e8174e14c6cab32118304a4ffff-val-town.aws-us-east-2.turso.io","tursoMeasuredAt",["D",1789519842404],"pinned","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm","user",{"_9":10,"_106":102},"org",{"_9":10,"_99":100,"_101":102,"_55":56,"_103":104,"_105":42},"branch",{"_36":89,"_11":90,"_23":29,"_91":45,"_92":-5,"_93":-5,"_94":-5,"_95":-5,"_96":-5,"_97":12,"_98":10},"isOwned",1233,"main","projectId","forkedBranchId","forkedBranchVersion","forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","projectName","userHandle","displayName","Amir","avatar","https://lh3.googleusercontent.com/a/ACg8ocILk3bwhcvrzozUWm7gxlKoBm7ZGtsgIkesW3ysOOsJ1rSQuA=s96-c","tier","free","canRestrictVals","profileImageUrl","ctx",{"_82":-5,"_140":141,"_142":143,"_144":145,"_146":42,"_147":42,"_148":42,"_149":33,"_150":-5,"_151":-7,"_152":-7,"_153":63,"_154":42,"_155":42,"_156":-5,"_157":42,"_158":42,"_159":42,"_160":-5},"ENV",{"_111":112,"_113":114,"_115":90,"_116":117,"_118":119,"_120":121,"_122":123,"_124":125,"_126":127,"_128":129,"_130":131,"_132":133,"_134":135,"_136":137,"_138":139},"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","orgs",[],"preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","preferencesCloseBrackets","preferencesChatPanelState","preferencesDprint","preferencesDenofmt","notificationCount","flagDenols","isBusinessOrgMember","businessOrgHandle","isBusinessOrgAdmin","isHigherTownieTier","navCollapsed","currentOrgId"]