@reddoorla/maintenance 0.97.0 → 0.98.0

package/dist/chunk-DZZ5KRTB.js

@@ -0,0 +1,23 @@
+// src/prospect/daily-cap.ts
+var PROSPECT_AUDIT_DAILY_CAP = 25;
+var DAILY_CAP_LOOKBACK = PROSPECT_AUDIT_DAILY_CAP * 2;
+var DAY_MS = 24 * 60 * 60 * 1e3;
+function countAuditsInDailyWindow(recent, now) {
+  const dayAgo = now.getTime() - DAY_MS;
+  return recent.filter((r) => Date.parse(r.created_at) >= dayAgo).length;
+}
+function isOverDailyCap(count) {
+  return count >= PROSPECT_AUDIT_DAILY_CAP;
+}
+function dailyCapMessage(count, cap = PROSPECT_AUDIT_DAILY_CAP) {
+  return `${count} audits have run in the last 24 hours (cap ${cap}). This is a runaway brake \u2014 if the run is genuinely needed, raise PROSPECT_AUDIT_DAILY_CAP.`;
+}
+
+export {
+  PROSPECT_AUDIT_DAILY_CAP,
+  DAILY_CAP_LOOKBACK,
+  countAuditsInDailyWindow,
+  isOverDailyCap,
+  dailyCapMessage
+};
+//# sourceMappingURL=chunk-DZZ5KRTB.js.map

package/dist/chunk-DZZ5KRTB.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/prospect/daily-cap.ts"],"sourcesContent":["/**\n * The prospect-audit runaway brake — the numbers and the arithmetic, in one\n * place both callers reach.\n *\n * It lived inside `src/dashboard/prospect-audit-trigger.ts`, which meant it\n * guarded the dashboard dispatch path and nothing else, while every batch to\n * date — including the 29-site corpus — went through `prospect-audit` on the\n * CLI, which had no cap at all (MED-15 of the 2026-09-02 review).\n *\n * That is the same shape #618 already fixed once in this feature, and said so\n * in `src/cli/commands/prospect-audit.ts` while fixing it: *\"The sole private-\n * address guard was in `triggerProspectAudit` — one layer at the far end of the\n * chain — so anyone running the CLI directly, or any future second caller, had\n * none.\"* Identical sentence, different guard.\n *\n * Dependency-free on purpose: the CLI's command modules are lazily loaded and\n * must not drag the GitHub/dispatch layer in behind a constant, and the count\n * needs nothing from a row but its timestamp.\n */\n\n/** Anything with a creation timestamp — `ProspectAuditListItem` satisfies it.\n *  Structural so this module needs no import from `src/db`. */\nexport type DatedAudit = { created_at: string };\n\n/** Most audits that may be STARTED in any rolling 24 hours (#612 review).\n *\n *  The duplicate window stops the SAME url being re-run; nothing stopped\n *  DISTINCT urls. One authenticated session could dispatch ~30/minute against\n *  30 hostnames indefinitely, and one audit is structurally an Opus call plus\n *  up to 28 Sonnet calls with up to 112 billed web searches, a 20-page double\n *  crawl, a 3-pass Lighthouse and a PDF render, inside a billed Actions job.\n *\n *  25/day is far above real use (this is one operator clicking a button) and far\n *  below a number that could quietly cost hundreds. A cap that never binds in\n *  normal operation is the point: it is a runaway brake, not a quota. */\nexport const PROSPECT_AUDIT_DAILY_CAP = 25;\n\n/** Lookback for the daily cap. Must exceed the cap so the count can actually\n *  reach it — a lookback at or below the cap would make the limit unreachable\n *  and the brake permanently disengaged, which is exactly the kind of guard\n *  that reads as working while doing nothing. */\nexport const DAILY_CAP_LOOKBACK = PROSPECT_AUDIT_DAILY_CAP * 2;\n\n/** The rolling window itself. */\nconst DAY_MS = 24 * 60 * 60 * 1000;\n\n/** How many of `recent` were created in the 24 hours before `now`.\n *\n *  `recent` is expected to be the newest `DAILY_CAP_LOOKBACK` audits; the\n *  caller does the fetching so this stays pure and testable from either side. */\nexport function countAuditsInDailyWindow(recent: readonly DatedAudit[], now: Date): number {\n  const dayAgo = now.getTime() - DAY_MS;\n  return recent.filter((r) => Date.parse(r.created_at) >= dayAgo).length;\n}\n\n/** Whether that count trips the brake. */\nexport function isOverDailyCap(count: number): boolean {\n  return count >= PROSPECT_AUDIT_DAILY_CAP;\n}\n\n/** The refusal, worded once so the dashboard's 429 body and the CLI's stderr\n *  say the same thing. Carries both numbers: a bare \"refused\" leaves the\n *  operator guessing whether they hit a limit or broke something. */\nexport function dailyCapMessage(count: number, cap: number = PROSPECT_AUDIT_DAILY_CAP): string {\n  return `${count} audits have run in the last 24 hours (cap ${cap}). This is a runaway brake — if the run is genuinely needed, raise PROSPECT_AUDIT_DAILY_CAP.`;\n}\n"],"mappings":";AAmCO,IAAM,2BAA2B;AAMjC,IAAM,qBAAqB,2BAA2B;AAG7D,IAAM,SAAS,KAAK,KAAK,KAAK;AAMvB,SAAS,yBAAyB,QAA+B,KAAmB;AACzF,QAAM,SAAS,IAAI,QAAQ,IAAI;AAC/B,SAAO,OAAO,OAAO,CAAC,MAAM,KAAK,MAAM,EAAE,UAAU,KAAK,MAAM,EAAE;AAClE;AAGO,SAAS,eAAe,OAAwB;AACrD,SAAO,SAAS;AAClB;AAKO,SAAS,gBAAgB,OAAe,MAAc,0BAAkC;AAC7F,SAAO,GAAG,KAAK,8CAA8C,GAAG;AAClE;","names":[]}

package/dist/{chunk-U2ZI6G7H.js → chunk-GRF2RYEO.js}

@@ -26,7 +26,7 @@ import {
 } from "./chunk-LUF52EIY.js";
 import {
   runAudits
-} from "./chunk-L3S3QZXL.js";
+} from "./chunk-DXZRWW2F.js";
 import {
   siteLabel
 } from "./chunk-XXTZBPUY.js";
@@ -161,4 +161,4 @@ export {
   DEFAULT_INIT_STEPS,
   init
 };
-//# sourceMappingURL=chunk-U2ZI6G7H.js.map
+//# sourceMappingURL=chunk-GRF2RYEO.js.map

package/dist/{chunk-QFSL2L2B.js → chunk-L5F3WYIY.js}

@@ -2,6 +2,7 @@ import {
   AI_AGENTS,
   CLASSICAL_AGENTS,
   USER_AGENT,
+  anchorsTruncated,
   buildJourney,
   canonicalizeUrl,
   crawlSite,
@@ -16,7 +17,7 @@ import {
   sleep,
   summarizeAccessibility,
   usablePages
-} from "./chunk-LAJ3FF54.js";
+} from "./chunk-5UI4FERC.js";
 import {
   checkGoal
 } from "./chunk-T7ZUUOZ2.js";
@@ -26,8 +27,12 @@ import {
 
 // src/prospect/consistency.ts
 var PHONE_IN_TEXT = /(?<!\d)(?:\+?1[\s.-]?)?\(?\d{3}\)?[\s.-]?\d{3}[\s.-]?\d{4}(?!\d)/g;
-var COPYRIGHT_YEAR = /(?:©|&copy;|copyright)([^\d<>]{0,30}?)(?:(\d{4})\s*[-–—]\s*)?(\d{4})/gi;
-var SENTENCE_BREAK = /[.!?]\s/;
+var NAME_GAP = "(?:[\\s.,&'\u2019-]*[A-Z][A-Za-z.'\u2019&-]{0,19}){0,6}[\\s.,&'\u2019-]*";
+var COPYRIGHT_YEAR = new RegExp(
+  `(?:\xA9|&[Cc]opy;|&COPY;|[Cc]opyright|COPYRIGHT)(${NAME_GAP})(?:(\\d{4})\\s*[-\u2013\u2014]\\s*)?(\\d{4})`,
+  "g"
+);
+var SENTENCE_BREAK = /[.!?]\s+[A-Z]/;
 function normalizePhone(raw) {
   const digits = raw.replace(/\D/g, "");
   const national = digits.length === 11 && digits.startsWith("1") ? digits.slice(1) : digits;
@@ -66,9 +71,12 @@ function checkConsistency(pages) {
         record(emails, address.toLowerCase(), address, page.url, true);
         continue;
       }
-      hrefs.add(href);
+      const abs = resolveNavigable(href, page.url);
+      if (!abs) continue;
+      const key = canonicalizeUrl(abs);
+      if (key) hrefs.add(key);
     }
-    linkSets.push({ url: page.url, hrefs });
+    linkSets.push({ url: page.url, hrefs, truncated: anchorsTruncated(extract) });
     for (const match of extract.text.matchAll(PHONE_IN_TEXT)) {
       const raw = match[0];
       if (!raw) continue;
@@ -81,24 +89,26 @@ function checkConsistency(pages) {
       if (year >= 1990 && year <= 2100) years.add(year);
     }
   }
+  const templateSets = linkSets.filter((p) => !p.truncated);
   const counts = /* @__PURE__ */ new Map();
-  for (const { hrefs } of linkSets) {
+  for (const { hrefs } of templateSets) {
     for (const href of hrefs) counts.set(href, (counts.get(href) ?? 0) + 1);
   }
-  const threshold = Math.ceil(linkSets.length * 0.6);
+  const threshold = Math.ceil(templateSets.length * 0.6);
   const sharedNav = new Set(
     [...counts.entries()].filter(([, n]) => n >= threshold).map(([href]) => href)
   );
-  const canJudgeTemplate = linkSets.length >= 3 && sharedNav.size > 0;
+  const canJudgeTemplate = templateSets.length >= 3 && sharedNav.size > 0;
   const sortedYears = [...years].sort((a, b) => a - b);
   return {
     phones: [...phones.values()],
     emails: [...emails.values()],
     copyrightYears: sortedYears,
     newestCopyrightYear: sortedYears.at(-1) ?? null,
-    pagesOffTemplate: canJudgeTemplate ? linkSets.filter((p) => ![...sharedNav].some((h) => p.hrefs.has(h))).map((p) => p.url) : [],
+    pagesOffTemplate: canJudgeTemplate ? templateSets.filter((p) => ![...sharedNav].some((h) => p.hrefs.has(h))).map((p) => p.url) : [],
     sharedNavLinks: sharedNav.size,
-    pagesExamined: linkSets.length
+    pagesExamined: linkSets.length,
+    pagesWithTruncatedAnchors: linkSets.filter((p) => p.truncated).map((p) => p.url)
   };
 }
 
@@ -734,7 +744,7 @@ function ownHost(url) {
     return null;
   }
 }
-function analyzeAnswerSpace(answers, prospectUrl) {
+function analyzeAnswerSpace(answers, prospectUrl, categoryAttempted) {
   const category = answers.filter((a) => a.kind === "category");
   const host = ownHost(prospectUrl);
   const counts = /* @__PURE__ */ new Map();
@@ -774,7 +784,7 @@ function analyzeAnswerSpace(answers, prospectUrl) {
   const own = ownIndex === -1 ? void 0 : topSources[ownIndex];
   return {
     answersWithCitations,
-    queriesAsked: category.length,
+    queriesAsked: categoryAttempted,
     citationsTotal,
     distinctDomains: topSources.length,
     topSources,
@@ -1045,7 +1055,7 @@ async function runVisibilityProbes(input, engines, opts = {}) {
     // a pure read of citations already recorded, so computing it here keeps the
     // ask loop to one job and lets the same function be re-run over a stored
     // report if we ever backfill.
-    answerSpace: analyzeAnswerSpace(answers, input.url)
+    answerSpace: analyzeAnswerSpace(answers, input.url, categoryAttempted)
   };
 }
 function perplexityEngine(apiKey, fetchImpl = fetch) {
@@ -1455,12 +1465,15 @@ ${a.fullAnswer}
 function normalize(s) {
   return s.replace(/\s+/g, " ").trim().toLowerCase();
 }
-function distinctive(terms) {
+function distinctive(terms, claim) {
+  const namesInClaim = new Set(
+    scanWords(claim).filter((w) => w.capitalised && !w.opensSentence).map((w) => w.lower)
+  );
   return terms.map((t) => t.trim()).filter((t) => {
     if (t.length < 5) return false;
     const multiWord = /\s/.test(t);
     const hasNumber = /\d/.test(t);
-    const properNoun = /^[A-Z]/.test(t);
+    const properNoun = namesInClaim.has(t.toLowerCase());
     return multiWord || hasNumber || properNoun;
   });
 }
@@ -1477,7 +1490,7 @@ function backstopAbsent(assertions, siteText, searchTermsByClaim, siteFullyRead)
     if (a.verdict !== "absent") return a;
     const flat = siteText.replace(/\s+/g, " ");
     let scattered = null;
-    for (const term of distinctive(searchTermsByClaim.get(a.claim) ?? [])) {
+    for (const term of distinctive(searchTermsByClaim.get(a.claim) ?? [], a.claim)) {
       const found = findTerm(haystack, term);
       if (found === null) continue;
       if (found === "exact") {
@@ -1538,21 +1551,53 @@ var WEAK_WORDS = /* @__PURE__ */ new Set([
   "based"
 ]);
 function quoteSupportsClaim(claim, quote) {
-  const words = (t) => {
+  const tokens = (t) => {
     const out = /* @__PURE__ */ new Set();
-    for (const w of t.split(/[^A-Za-z0-9]+/)) {
-      if (w === "") continue;
-      const lower = w.toLowerCase();
-      if (WEAK_WORDS.has(lower)) continue;
-      const properNoun = /^[A-Z]/.test(w) && w.length >= 3;
-      if (properNoun || lower.length >= 4) out.add(lower);
+    const names = /* @__PURE__ */ new Set();
+    const run = [];
+    const flushRun = () => {
+      if (run.length > 0) names.add(run.join(" "));
+      run.length = 0;
+    };
+    for (const w of scanWords(t)) {
+      if (WEAK_WORDS.has(w.lower)) {
+        flushRun();
+        continue;
+      }
+      const properNoun = w.capitalised && !w.opensSentence && w.raw.length >= 2;
+      if (properNoun) run.push(w.lower);
+      else flushRun();
+      if (properNoun || w.lower.length >= 4) out.add(w.lower);
     }
-    return out;
+    flushRun();
+    return { words: out, names };
   };
-  const inQuote = words(quote);
+  const c = tokens(claim);
+  const q = tokens(quote);
   let shared = 0;
-  for (const w of words(claim)) if (inQuote.has(w)) shared += 1;
-  return shared >= 2;
+  for (const w of c.words) if (q.words.has(w)) shared += 1;
+  if (shared >= 2) return true;
+  for (const n of c.names) if (q.names.has(n)) return true;
+  return false;
+}
+function scanWords(text) {
+  const out = [];
+  let opensSentence = true;
+  let index = 0;
+  for (const m of text.matchAll(/[A-Za-z0-9]+/g)) {
+    const raw = m[0];
+    const gap = text.slice(index, m.index);
+    if (out.length > 0 && /[.!?\n\r]/.test(gap)) opensSentence = true;
+    index = m.index + raw.length;
+    out.push({
+      raw,
+      lower: raw.toLowerCase(),
+      capitalised: /^[A-Z]/.test(raw),
+      opensSentence
+    });
+    opensSentence = false;
+  }
+  return out;
 }
 function verifyQuotes(raw, answerText, siteText, answerOf, prospectDomain, searchTermsByClaim = /* @__PURE__ */ new Map()) {
   const answers = normalize(answerText);
@@ -1564,7 +1609,7 @@ function verifyQuotes(raw, answerText, siteText, answerOf, prospectDomain, searc
     let verdict = a.verdict;
     let siteQuote = a.siteQuote;
     if (a.verdict === "contradicted") {
-      for (const term of distinctive(searchTermsByClaim.get(a.claim) ?? [])) {
+      for (const term of distinctive(searchTermsByClaim.get(a.claim) ?? [], a.claim)) {
         if (findTerm(site, term) !== "exact") continue;
         const at = site.indexOf(normalize(term));
         verdict = "confirmed";
@@ -1970,7 +2015,7 @@ function claudeCodeEngine(run = defaultClaudeCodeRun) {
 function defaultLighthouseDeps() {
   return {
     async audit(site) {
-      const { lighthouseAudit } = await import("./lighthouse-52WSPZ5D.js");
+      const { lighthouseAudit } = await import("./lighthouse-RFMFREEX.js");
       return lighthouseAudit({ site });
     }
   };
@@ -2442,7 +2487,12 @@ async function checkBasics(crawl, deps) {
       // Only meaningful on an https site, and only over the images the extract
       // records — stylesheets and scripts are not captured, so the report must
       // not claim to have checked them.
-      measured: isHttps && usable.pages.length > 0,
+      //
+      // `imageSrcsMeasured`, not `pages.length > 0`: the loop above reads
+      // `extract.imageSrcs ?? []`, and that field is absent on every report
+      // stored before it existed. Counting those as measured published "we
+      // checked and found none" from a measurement never taken.
+      measured: isHttps && usable.imageSrcsMeasured,
       imageUrls: [...insecureImages].slice(0, 12),
       imagesSeen
     },
@@ -3014,4 +3064,4 @@ export {
   mergeFixes,
   runProspectAudit
 };
-//# sourceMappingURL=chunk-QFSL2L2B.js.map
+//# sourceMappingURL=chunk-L5F3WYIY.js.map