@prenta/core 0.118.0

package/dist/content-health/cron.js

@@ -0,0 +1,167 @@
+/**
+ * Content health cron — runs `scanDocument` across published
+ * documents in chunks, upserts findings into
+ * `prenta_content_issues`, and deletes rows whose issue resolved
+ * since the previous run.
+ *
+ * ## Why chunked iteration
+ *
+ * A single full scan over a 10k-doc corpus does ~10k HTML parses
+ * plus up to ~50k internal-link probes (HEAD requests with 5s
+ * timeouts). On Vercel that quickly approaches the function
+ * wall-time cap. We batch in `BATCH_SIZE` pages and yield to the
+ * event loop between batches so a 10-minute cron can make
+ * incremental progress even when the corpus is large.
+ *
+ * ## Why upsert + delete-stale
+ *
+ * The scanner is the source of truth. After a run, any (docId, type)
+ * pair NOT emitted by the scan no longer applies — the editor
+ * fixed the issue, the document was unpublished, etc. We track
+ * pairs we've seen and DELETE the unseen ones at the end.
+ *
+ * ## Why we never delete on partial failure
+ *
+ * If the scan throws mid-run, we abandon the delete step. Better
+ * to over-report (a previously-recorded issue may already be
+ * resolved) than to silently drop a real issue because the scan
+ * crashed before reaching that document.
+ */
+import { scanDocument, checkInternalLinkReachable, } from './scanner.js';
+const DEFAULT_BATCH_SIZE = 50;
+const DEFAULT_MAX_DOCUMENTS = 500;
+export async function processContentHealthScan(db, options = {}) {
+    const maxDocuments = options.maxDocuments ?? DEFAULT_MAX_DOCUMENTS;
+    const batchSize = options.batchSize ?? DEFAULT_BATCH_SIZE;
+    const linkProbeTimeoutMs = options.linkProbeTimeoutMs ?? 5000;
+    // If neither a checker nor a siteOrigin is supplied, the link
+    // check is disabled — the scanner still runs the other three
+    // checks. Passing `checkLink: null` explicitly is the supported
+    // way to opt out (useful in CI / tests).
+    const checkLink = options.checkLink === undefined
+        ? (url) => checkInternalLinkReachable(url, linkProbeTimeoutMs)
+        : options.checkLink;
+    const result = {
+        scanned: 0,
+        inserted: 0,
+        updated: 0,
+        resolved: 0,
+        totals: {
+            MISSING_META_DESCRIPTION: 0,
+            MISSING_ALT_TEXT: 0,
+            BROKEN_INTERNAL_LINK: 0,
+            OUTDATED_CONTENT: 0,
+        },
+        truncated: false,
+    };
+    if (!db?.document || !db?.contentIssue)
+        return result;
+    // Track the (docId, type) pairs we've seen this run. After the
+    // scan completes, any row in the table NOT in this set is
+    // resolved and gets deleted.
+    const seen = new Set();
+    let cursor;
+    while (result.scanned < maxDocuments) {
+        const remaining = maxDocuments - result.scanned;
+        const take = Math.min(batchSize, remaining);
+        const batch = await db.document.findMany({
+            where: { deletedAt: null, status: 'PUBLISHED' },
+            orderBy: { id: 'asc' },
+            take,
+            ...(cursor ? { skip: 1, cursor: { id: cursor } } : {}),
+            select: { id: true, data: true, updatedAt: true, publishedAt: true },
+        });
+        if (batch.length === 0)
+            break;
+        for (const doc of batch) {
+            const issues = await scanDocument({
+                id: doc.id,
+                data: (doc.data ?? {}),
+                updatedAt: doc.updatedAt,
+                publishedAt: doc.publishedAt,
+            }, {
+                checkLink,
+                siteOrigin: options.siteOrigin ?? null,
+            });
+            for (const issue of issues) {
+                seen.add(`${doc.id}|${issue.type}`);
+                const persisted = await upsertIssue(db, doc.id, issue);
+                if (persisted === 'inserted')
+                    result.inserted++;
+                else if (persisted === 'updated')
+                    result.updated++;
+                result.totals[issue.type] += issue.count;
+            }
+            result.scanned++;
+        }
+        cursor = batch[batch.length - 1]?.id;
+        if (batch.length < take)
+            break;
+        if (result.scanned >= maxDocuments) {
+            result.truncated = true;
+            break;
+        }
+    }
+    // Resolve no-longer-applicable rows. Done in one DELETE per
+    // type so the table stays consistent even on a corpus with
+    // 10k+ resolved issues — `IN` clauses on large id lists are
+    // far slower than `NOT IN` with a (small) set of currently-seen
+    // pairs per type.
+    result.resolved = await pruneResolved(db, seen);
+    return result;
+}
+async function upsertIssue(db, documentId, issue) {
+    // We can't tell `upsert` to report whether it inserted or
+    // updated from a single call, so we attempt to `update` first
+    // and fall back to `create` on P2025 (record not found).
+    try {
+        await db.contentIssue.update({
+            where: { documentId_type: { documentId, type: issue.type } },
+            data: {
+                count: issue.count,
+                details: issue.details ?? null,
+                lastSeenAt: new Date(),
+            },
+        });
+        return 'updated';
+    }
+    catch (err) {
+        if (err?.code === 'P2025') {
+            await db.contentIssue.create({
+                data: {
+                    documentId,
+                    type: issue.type,
+                    count: issue.count,
+                    details: issue.details ?? undefined,
+                },
+            });
+            return 'inserted';
+        }
+        throw err;
+    }
+}
+/**
+ * Delete rows for (documentId, type) pairs NOT present in `seen`.
+ * Implementation: load all current rows in (id, documentId, type)
+ * tuples, filter to the unseen ones, and bulk-delete. For tables
+ * up to ~100k rows this is faster than building a giant SQL
+ * `WHERE NOT (... OR ... OR ...)` clause.
+ */
+async function pruneResolved(db, seen) {
+    const allRows = await db.contentIssue.findMany({
+        select: { id: true, documentId: true, type: true },
+    });
+    const stale = allRows.filter((r) => !seen.has(`${r.documentId}|${r.type}`)).map((r) => r.id);
+    if (stale.length === 0)
+        return 0;
+    // Chunked deleteMany — Postgres caps IN-list size around 32k
+    // parameters; 1000 is well under and easy to reason about.
+    let deleted = 0;
+    for (let i = 0; i < stale.length; i += 1000) {
+        const slice = stale.slice(i, i + 1000);
+        const r = await db.contentIssue.deleteMany({ where: { id: { in: slice } } });
+        deleted += r.count;
+    }
+    return deleted;
+}
+//# sourceMappingURL=cron.js.map

package/dist/content-health/cron.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"cron.js","sourceRoot":"","sources":["../../src/content-health/cron.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4BG;AAEH,OAAO,EACL,YAAY,EACZ,0BAA0B,GAG3B,MAAM,cAAc,CAAA;AAwCrB,MAAM,kBAAkB,GAAG,EAAE,CAAA;AAC7B,MAAM,qBAAqB,GAAG,GAAG,CAAA;AAEjC,MAAM,CAAC,KAAK,UAAU,wBAAwB,CAC5C,EAAY,EACZ,UAAoC,EAAE;IAEtC,MAAM,YAAY,GAAG,OAAO,CAAC,YAAY,IAAI,qBAAqB,CAAA;IAClE,MAAM,SAAS,GAAG,OAAO,CAAC,SAAS,IAAI,kBAAkB,CAAA;IACzD,MAAM,kBAAkB,GAAG,OAAO,CAAC,kBAAkB,IAAI,IAAI,CAAA;IAE7D,8DAA8D;IAC9D,6DAA6D;IAC7D,gEAAgE;IAChE,yCAAyC;IACzC,MAAM,SAAS,GACb,OAAO,CAAC,SAAS,KAAK,SAAS;QAC7B,CAAC,CAAC,CAAC,GAAW,EAAE,EAAE,CAAC,0BAA0B,CAAC,GAAG,EAAE,kBAAkB,CAAC;QACtE,CAAC,CAAC,OAAO,CAAC,SAAS,CAAA;IAEvB,MAAM,MAAM,GAA4B;QACtC,OAAO,EAAE,CAAC;QACV,QAAQ,EAAE,CAAC;QACX,OAAO,EAAE,CAAC;QACV,QAAQ,EAAE,CAAC;QACX,MAAM,EAAE;YACN,wBAAwB,EAAE,CAAC;YAC3B,gBAAgB,EAAE,CAAC;YACnB,oBAAoB,EAAE,CAAC;YACvB,gBAAgB,EAAE,CAAC;SACpB;QACD,SAAS,EAAE,KAAK;KACjB,CAAA;IAED,IAAI,CAAC,EAAE,EAAE,QAAQ,IAAI,CAAC,EAAE,EAAE,YAAY;QAAE,OAAO,MAAM,CAAA;IAErD,+DAA+D;IAC/D,0DAA0D;IAC1D,6BAA6B;IAC7B,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;IAE9B,IAAI,MAA0B,CAAA;IAC9B,OAAO,MAAM,CAAC,OAAO,GAAG,YAAY,EAAE,CAAC;QACrC,MAAM,SAAS,GAAG,YAAY,GAAG,MAAM,CAAC,OAAO,CAAA;QAC/C,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,SAAS,EAAE,SAAS,CAAC,CAAA;QAC3C,MAAM,KAAK,GAKN,MAAM,EAAE,CAAC,QAAQ,CAAC,QAAQ,CAAC;YAC9B,KAAK,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,MAAM,EAAE,WAAW,EAAE;YAC/C,OAAO,EAAE,EAAE,EAAE,EAAE,KAAK,EAAE;YACtB,IAAI;YACJ,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,EAAE,MAAM,EAAE,EAAE,EAAE,EAAE,MAAM,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;YACtD,MAAM,EAAE,EAAE,EAAE,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,SAAS,EAAE,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE;SACrE,CAAC,CAAA;QACF,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;YAAE,MAAK;QAE7B,KAAK,MAAM,GAAG,IAAI,KAAK,EAAE,CAAC;YACxB,MAAM,MAAM,GAAG,MAAM,YAAY,CAC/B;gBACE,EAAE,EAAE,GAAG,CAAC,EAAE;gBACV,IAAI,EAAE,CAAC,GAAG,CAAC,IAAI,IAAI,EAAE,CAA4B;gBACjD,SAAS,EAAE,GAAG,CAAC,SAAS;gBACxB,WAAW,EAAE,GAAG,CAAC,WAAW;aAC7B,EACD;gBACE,SAAS;gBACT,UAAU,EAAE,OAAO,CAAC,UAAU,IAAI,IAAI;aACvC,CACF,CAAA;YAED,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;gBAC3B,IAAI,CAAC,GAAG,CAAC,GAAG,GAAG,CAAC,EAAE,IAAI,KAAK,CAAC,IAAI,EAAE,CAAC,CAAA;gBACnC,MAAM,SAAS,GAAG,MAAM,WAAW,CAAC,EAAE,EAAE,GAAG,CAAC,EAAE,EAAE,KAAK,CAAC,CAAA;gBACtD,IAAI,SAAS,KAAK,UAAU;oBAAE,MAAM,CAAC,QAAQ,EAAE,CAAA;qBAC1C,IAAI,SAAS,KAAK,SAAS;oBAAE,MAAM,CAAC,OAAO,EAAE,CAAA;gBAClD,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,KAAK,CAAC,KAAK,CAAA;YAC1C,CAAC;YACD,MAAM,CAAC,OAAO,EAAE,CAAA;QAClB,CAAC;QACD,MAAM,GAAG,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,EAAE,EAAE,CAAA;QACpC,IAAI,KAAK,CAAC,MAAM,GAAG,IAAI;YAAE,MAAK;QAC9B,IAAI,MAAM,CAAC,OAAO,IAAI,YAAY,EAAE,CAAC;YACnC,MAAM,CAAC,SAAS,GAAG,IAAI,CAAA;YACvB,MAAK;QACP,CAAC;IACH,CAAC;IAED,4DAA4D;IAC5D,2DAA2D;IAC3D,4DAA4D;IAC5D,gEAAgE;IAChE,kBAAkB;IAClB,MAAM,CAAC,QAAQ,GAAG,MAAM,aAAa,CAAC,EAAE,EAAE,IAAI,CAAC,CAAA;IAE/C,OAAO,MAAM,CAAA;AACf,CAAC;AAED,KAAK,UAAU,WAAW,CACxB,EAAY,EACZ,UAAkB,EAClB,KAAmB;IAEnB,0DAA0D;IAC1D,8DAA8D;IAC9D,yDAAyD;IACzD,IAAI,CAAC;QACH,MAAM,EAAE,CAAC,YAAY,CAAC,MAAM,CAAC;YAC3B,KAAK,EAAE,EAAE,eAAe,EAAE,EAAE,UAAU,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,EAAE,EAAE;YAC5D,IAAI,EAAE;gBACJ,KAAK,EAAE,KAAK,CAAC,KAAK;gBAClB,OAAO,EAAE,KAAK,CAAC,OAAO,IAAI,IAAI;gBAC9B,UAAU,EAAE,IAAI,IAAI,EAAE;aACvB;SACF,CAAC,CAAA;QACF,OAAO,SAAS,CAAA;IAClB,CAAC;IAAC,OAAO,GAAQ,EAAE,CAAC;QAClB,IAAI,GAAG,EAAE,IAAI,KAAK,OAAO,EAAE,CAAC;YAC1B,MAAM,EAAE,CAAC,YAAY,CAAC,MAAM,CAAC;gBAC3B,IAAI,EAAE;oBACJ,UAAU;oBACV,IAAI,EAAE,KAAK,CAAC,IAAI;oBAChB,KAAK,EAAE,KAAK,CAAC,KAAK;oBAClB,OAAO,EAAE,KAAK,CAAC,OAAO,IAAI,SAAS;iBACpC;aACF,CAAC,CAAA;YACF,OAAO,UAAU,CAAA;QACnB,CAAC;QACD,MAAM,GAAG,CAAA;IACX,CAAC;AACH,CAAC;AAED;;;;;;GAMG;AACH,KAAK,UAAU,aAAa,CAAC,EAAY,EAAE,IAAiB;IAC1D,MAAM,OAAO,GACX,MAAM,EAAE,CAAC,YAAY,CAAC,QAAQ,CAAC;QAC7B,MAAM,EAAE,EAAE,EAAE,EAAE,IAAI,EAAE,UAAU,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE;KACnD,CAAC,CAAA;IACJ,MAAM,KAAK,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,UAAU,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAA;IAC5F,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,CAAC,CAAA;IAChC,6DAA6D;IAC7D,2DAA2D;IAC3D,IAAI,OAAO,GAAG,CAAC,CAAA;IACf,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,IAAI,IAAI,EAAE,CAAC;QAC5C,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,CAAA;QACtC,MAAM,CAAC,GAAG,MAAM,EAAE,CAAC,YAAY,CAAC,UAAU,CAAC,EAAE,KAAK,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC,CAAA;QAC5E,OAAO,IAAI,CAAC,CAAC,KAAK,CAAA;IACpB,CAAC;IACD,OAAO,OAAO,CAAA;AAChB,CAAC"}

package/dist/content-health/scanner.d.ts

@@ -0,0 +1,119 @@
+/**
+ * Content health scanner — produces the issue rows that power the
+ * dashboard "Content Health" card.
+ *
+ * ## Why a separate module from the cron / endpoint
+ *
+ * The scanner is pure logic — give it a document, get back an array
+ * of issues. The cron drives it across the corpus and persists the
+ * results; the endpoint reads the persisted results. Keeping the
+ * scanner free of I/O dependencies makes it trivially testable
+ * (~10 µs per call, no DB) and lets future callers (CI checks, a
+ * pre-publish hook, etc.) reuse it without dragging in cron
+ * infrastructure.
+ *
+ * ## Why HEAD probes for the link checker
+ *
+ * Internal links 404 in two flavors: a hard 404 from the framework
+ * (page genuinely doesn't exist) and a soft 404 where the framework
+ * returns 200 with a "not found" template. We only catch the first.
+ * That's fine — the dashboard's job is to surface the high-confidence
+ * issues, not chase ambiguous soft-404 heuristics. The scanner can
+ * be extended with response-body matching later if the editor
+ * feedback warrants it.
+ *
+ * ## Why we count alt-less images per document, not per image
+ *
+ * The dashboard surfaces total issue counts. Storing one row per
+ * image would blow up the issue table for pages with 20+
+ * unannotated photos and provide no extra value (the editor still
+ * gets a single "12 images need alt text" row in the UI). We
+ * collapse to one row per (document, type) with `count` carrying
+ * the per-document magnitude.
+ */
+export type ContentIssueKind = 'MISSING_META_DESCRIPTION' | 'MISSING_ALT_TEXT' | 'BROKEN_INTERNAL_LINK' | 'OUTDATED_CONTENT';
+export interface ScannerInput {
+    id: string;
+    /**
+     * Document's `data` JSON. Fields we read:
+     *   - `metaDescription`, `seoDescription` (optional)
+     *   - `body`, `content`, `richText`, `html` (any string field used
+     *     as the primary HTML payload — we try each, longest wins)
+     */
+    data: Record<string, unknown>;
+    updatedAt: Date;
+    publishedAt?: Date | null;
+}
+export interface ScannerIssue {
+    type: ContentIssueKind;
+    count: number;
+    details?: Record<string, unknown>;
+}
+export interface ScannerOptions {
+    /**
+     * Cut-off (in days) for the OUTDATED_CONTENT check. A document
+     * whose `updatedAt` is older than this is flagged. Default 90.
+     */
+    outdatedDays?: number;
+    /**
+     * Link-checker callback. Resolves to `true` if the link is
+     * reachable, `false` if broken. Pass `null` to disable the
+     * link-check (useful in tests). The default cron supplies
+     * `checkInternalLinkReachable` from this module.
+     */
+    checkLink?: ((url: string) => Promise<boolean>) | null;
+    /**
+     * Origin used to resolve relative URLs and identify
+     * "internal" links. Required for the link checker — passing
+     * `null` here is the same as passing `checkLink: null` (no
+     * link probing).
+     */
+    siteOrigin?: string | null;
+}
+/**
+ * Count <img> tags in the HTML that are missing alt text, excluding images
+ * correctly marked decorative (empty alt + role="presentation"/aria-hidden).
+ * Delegates to the shared, decorative-aware analyzer so every SEO surface
+ * (analysis, audit engine, this scanner, the cron scan) agrees.
+ */
+export declare function countMissingAltImages(html: string): number;
+/**
+ * Extract candidate URLs from <a href="..."> attributes. Returns
+ * the raw href values; classification (internal vs external,
+ * absolute vs relative) happens in the caller.
+ */
+export declare function extractLinks(html: string): string[];
+/**
+ * Decide whether a URL is "internal" relative to `siteOrigin`.
+ * Internal links include:
+ *   - relative paths: `/about`, `./page`, `../foo`
+ *   - same-origin absolute URLs: `https://example.com/page`
+ *
+ * Excluded:
+ *   - external URLs (different origin)
+ *   - hash-only anchors (`#section`)
+ *   - non-http schemes (`mailto:`, `tel:`, `javascript:`, etc.)
+ *   - protocol-relative URLs to a different host (`//other.com/x`)
+ */
+export declare function classifyLink(url: string, siteOrigin: string): {
+    isInternal: boolean;
+    resolved: string | null;
+};
+/**
+ * Single-link reachability probe. Default implementation used by
+ * the cron — issues a HEAD with a 5s timeout, falls back to GET if
+ * HEAD returns 405/501 (some frameworks reject HEAD).
+ */
+export declare function checkInternalLinkReachable(url: string, timeoutMs?: number): Promise<boolean>;
+/**
+ * Run all checks on a single document. The link checker is
+ * concurrency-bounded to `LINK_CHECK_CONCURRENCY` per document so a
+ * page with 200 links doesn't open 200 sockets at once.
+ *
+ * Each issue type appears at most once in the returned array. The
+ * shape `{ type, count, details }` lets the caller upsert by
+ * `(documentId, type)` and rely on `count` for the magnitude
+ * surfaced in the dashboard.
+ */
+export declare function scanDocument(doc: ScannerInput, opts?: ScannerOptions): Promise<ScannerIssue[]>;
+//# sourceMappingURL=scanner.d.ts.map

package/dist/content-health/scanner.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"scanner.d.ts","sourceRoot":"","sources":["../../src/content-health/scanner.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAgCG;AAMH,MAAM,MAAM,gBAAgB,GACxB,0BAA0B,GAC1B,kBAAkB,GAClB,sBAAsB,GACtB,kBAAkB,CAAA;AAEtB,MAAM,WAAW,YAAY;IAC3B,EAAE,EAAE,MAAM,CAAA;IACV;;;;;OAKG;IACH,IAAI,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAA;IAC7B,SAAS,EAAE,IAAI,CAAA;IACf,WAAW,CAAC,EAAE,IAAI,GAAG,IAAI,CAAA;CAC1B;AAED,MAAM,WAAW,YAAY;IAC3B,IAAI,EAAE,gBAAgB,CAAA;IACtB,KAAK,EAAE,MAAM,CAAA;IACb,OAAO,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAA;CAClC;AAED,MAAM,WAAW,cAAc;IAC7B;;;OAGG;IACH,YAAY,CAAC,EAAE,MAAM,CAAA;IACrB;;;;;OAKG;IACH,SAAS,CAAC,EAAE,CAAC,CAAC,GAAG,EAAE,MAAM,KAAK,OAAO,CAAC,OAAO,CAAC,CAAC,GAAG,IAAI,CAAA;IACtD;;;;;OAKG;IACH,UAAU,CAAC,EAAE,MAAM,GAAG,IAAI,CAAA;CAC3B;AAsCD;;;;;GAKG;AACH,wBAAgB,qBAAqB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAE1D;AAED;;;;GAIG;AACH,wBAAgB,YAAY,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,CAUnD;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,YAAY,CAC1B,GAAG,EAAE,MAAM,EACX,UAAU,EAAE,MAAM,GACjB;IAAE,UAAU,EAAE,OAAO,CAAC;IAAC,QAAQ,EAAE,MAAM,GAAG,IAAI,CAAA;CAAE,CA8BlD;AAED;;;;GAIG;AACH,wBAAsB,0BAA0B,CAAC,GAAG,EAAE,MAAM,EAAE,SAAS,SAAO,GAAG,OAAO,CAAC,OAAO,CAAC,CAwBhG;AAID;;;;;;;;;GASG;AACH,wBAAsB,YAAY,CAChC,GAAG,EAAE,YAAY,EACjB,IAAI,GAAE,cAAmB,GACxB,OAAO,CAAC,YAAY,EAAE,CAAC,CA4DzB"}

package/dist/content-health/scanner.js

@@ -0,0 +1,257 @@
+/**
+ * Content health scanner — produces the issue rows that power the
+ * dashboard "Content Health" card.
+ *
+ * ## Why a separate module from the cron / endpoint
+ *
+ * The scanner is pure logic — give it a document, get back an array
+ * of issues. The cron drives it across the corpus and persists the
+ * results; the endpoint reads the persisted results. Keeping the
+ * scanner free of I/O dependencies makes it trivially testable
+ * (~10 µs per call, no DB) and lets future callers (CI checks, a
+ * pre-publish hook, etc.) reuse it without dragging in cron
+ * infrastructure.
+ *
+ * ## Why HEAD probes for the link checker
+ *
+ * Internal links 404 in two flavors: a hard 404 from the framework
+ * (page genuinely doesn't exist) and a soft 404 where the framework
+ * returns 200 with a "not found" template. We only catch the first.
+ * That's fine — the dashboard's job is to surface the high-confidence
+ * issues, not chase ambiguous soft-404 heuristics. The scanner can
+ * be extended with response-body matching later if the editor
+ * feedback warrants it.
+ *
+ * ## Why we count alt-less images per document, not per image
+ *
+ * The dashboard surfaces total issue counts. Storing one row per
+ * image would blow up the issue table for pages with 20+
+ * unannotated photos and provide no extra value (the editor still
+ * gets a single "12 images need alt text" row in the UI). We
+ * collapse to one row per (document, type) with `count` carrying
+ * the per-document magnitude.
+ */
+import { countMissingAltImages as sharedCountMissingAltImages } from '../seo/alt-text.js';
+// ─── Helpers ───────────────────────────────────────────────────────────────
+/**
+ * Return the longest plausible HTML payload from a document. CMS
+ * collections use different field names (`body`, `content`,
+ * `richText`, `html`) so we try each and pick the longest — that
+ * resolves the common case where a document has both a plain
+ * `excerpt` and a rich `body`, and we want to scan the body.
+ */
+function extractHtml(data) {
+    const candidates = ['body', 'content', 'richText', 'richtext', 'html', 'description'];
+    let best = '';
+    for (const key of candidates) {
+        const v = data[key];
+        if (typeof v === 'string' && v.length > best.length)
+            best = v;
+    }
+    return best;
+}
+const META_DESCRIPTION_KEYS = ['metaDescription', 'seoDescription', 'meta_description'];
+function getMetaDescription(data) {
+    for (const key of META_DESCRIPTION_KEYS) {
+        const v = data[key];
+        if (typeof v === 'string' && v.trim().length > 0)
+            return v.trim();
+    }
+    // Nested SEO objects (e.g. data.seo.description).
+    const seo = data['seo'];
+    if (seo && typeof seo === 'object') {
+        const obj = seo;
+        const v = obj['description'] ?? obj['metaDescription'];
+        if (typeof v === 'string' && v.trim().length > 0)
+            return v.trim();
+    }
+    return null;
+}
+/**
+ * Count <img> tags in the HTML that are missing alt text, excluding images
+ * correctly marked decorative (empty alt + role="presentation"/aria-hidden).
+ * Delegates to the shared, decorative-aware analyzer so every SEO surface
+ * (analysis, audit engine, this scanner, the cron scan) agrees.
+ */
+export function countMissingAltImages(html) {
+    return sharedCountMissingAltImages(html);
+}
+/**
+ * Extract candidate URLs from <a href="..."> attributes. Returns
+ * the raw href values; classification (internal vs external,
+ * absolute vs relative) happens in the caller.
+ */
+export function extractLinks(html) {
+    if (!html)
+        return [];
+    const out = [];
+    const re = /<a\b[^>]*\shref\s*=\s*("([^"]*)"|'([^']*)')/gi;
+    let match;
+    while ((match = re.exec(html)) !== null) {
+        const url = (match[2] ?? match[3] ?? '').trim();
+        if (url.length > 0)
+            out.push(url);
+    }
+    return out;
+}
+/**
+ * Decide whether a URL is "internal" relative to `siteOrigin`.
+ * Internal links include:
+ *   - relative paths: `/about`, `./page`, `../foo`
+ *   - same-origin absolute URLs: `https://example.com/page`
+ *
+ * Excluded:
+ *   - external URLs (different origin)
+ *   - hash-only anchors (`#section`)
+ *   - non-http schemes (`mailto:`, `tel:`, `javascript:`, etc.)
+ *   - protocol-relative URLs to a different host (`//other.com/x`)
+ */
+export function classifyLink(url, siteOrigin) {
+    const trimmed = url.trim();
+    if (trimmed.length === 0)
+        return { isInternal: false, resolved: null };
+    if (trimmed.startsWith('#'))
+        return { isInternal: false, resolved: null };
+    // Reject non-http schemes outright.
+    const schemeMatch = /^([a-z][a-z0-9+.-]*):/i.exec(trimmed);
+    if (schemeMatch) {
+        const scheme = schemeMatch[1].toLowerCase();
+        if (scheme !== 'http' && scheme !== 'https')
+            return { isInternal: false, resolved: null };
+    }
+    let parsedOrigin;
+    try {
+        parsedOrigin = new URL(siteOrigin);
+    }
+    catch {
+        return { isInternal: false, resolved: null };
+    }
+    try {
+        const resolved = new URL(trimmed, parsedOrigin);
+        if (resolved.protocol !== 'http:' && resolved.protocol !== 'https:') {
+            return { isInternal: false, resolved: null };
+        }
+        return {
+            isInternal: resolved.origin === parsedOrigin.origin,
+            resolved: resolved.toString(),
+        };
+    }
+    catch {
+        return { isInternal: false, resolved: null };
+    }
+}
+/**
+ * Single-link reachability probe. Default implementation used by
+ * the cron — issues a HEAD with a 5s timeout, falls back to GET if
+ * HEAD returns 405/501 (some frameworks reject HEAD).
+ */
+export async function checkInternalLinkReachable(url, timeoutMs = 5000) {
+    const controller = new AbortController();
+    const t = setTimeout(() => controller.abort(), timeoutMs);
+    try {
+        const head = await fetch(url, {
+            method: 'HEAD',
+            redirect: 'follow',
+            signal: controller.signal,
+        });
+        if (head.status === 405 || head.status === 501) {
+            // Method not allowed / not implemented → retry with GET.
+            const get = await fetch(url, {
+                method: 'GET',
+                redirect: 'follow',
+                signal: controller.signal,
+            });
+            return get.status < 400;
+        }
+        return head.status < 400;
+    }
+    catch {
+        return false;
+    }
+    finally {
+        clearTimeout(t);
+    }
+}
+// ─── Scanner ───────────────────────────────────────────────────────────────
+/**
+ * Run all checks on a single document. The link checker is
+ * concurrency-bounded to `LINK_CHECK_CONCURRENCY` per document so a
+ * page with 200 links doesn't open 200 sockets at once.
+ *
+ * Each issue type appears at most once in the returned array. The
+ * shape `{ type, count, details }` lets the caller upsert by
+ * `(documentId, type)` and rely on `count` for the magnitude
+ * surfaced in the dashboard.
+ */
+export async function scanDocument(doc, opts = {}) {
+    const outdatedDays = opts.outdatedDays ?? 90;
+    const issues = [];
+    // 1. Missing meta description --------------------------------------------
+    if (!getMetaDescription(doc.data)) {
+        issues.push({ type: 'MISSING_META_DESCRIPTION', count: 1 });
+    }
+    // 2. Missing alt text ----------------------------------------------------
+    const html = extractHtml(doc.data);
+    const altLessImages = countMissingAltImages(html);
+    if (altLessImages > 0) {
+        issues.push({ type: 'MISSING_ALT_TEXT', count: altLessImages });
+    }
+    // 3. Outdated content ----------------------------------------------------
+    const referenceTime = doc.publishedAt ?? doc.updatedAt;
+    const ageMs = Date.now() - referenceTime.getTime();
+    if (ageMs > outdatedDays * 24 * 60 * 60 * 1000) {
+        issues.push({
+            type: 'OUTDATED_CONTENT',
+            count: 1,
+            details: {
+                ageDays: Math.floor(ageMs / (24 * 60 * 60 * 1000)),
+                threshold: outdatedDays,
+            },
+        });
+    }
+    // 4. Broken internal links ----------------------------------------------
+    if (opts.checkLink && opts.siteOrigin) {
+        const links = extractLinks(html);
+        const internalUrls = new Set();
+        for (const raw of links) {
+            const { isInternal, resolved } = classifyLink(raw, opts.siteOrigin);
+            if (isInternal && resolved)
+                internalUrls.add(resolved);
+        }
+        const broken = [];
+        if (internalUrls.size > 0) {
+            const checker = opts.checkLink;
+            await Promise.all(chunk([...internalUrls], LINK_CHECK_CONCURRENCY).map(async (group) => {
+                const results = await Promise.all(group.map((u) => checker(u).catch(() => false)));
+                group.forEach((u, i) => {
+                    if (!results[i])
+                        broken.push(u);
+                });
+            }));
+        }
+        if (broken.length > 0) {
+            issues.push({
+                type: 'BROKEN_INTERNAL_LINK',
+                count: broken.length,
+                details: { urls: broken.slice(0, 10) }, // cap stored details
+            });
+        }
+    }
+    return issues;
+}
+/**
+ * Concurrency cap for the per-document link checker. Vercel
+ * functions are wall-time bounded; opening 5 concurrent sockets
+ * keeps us under the typical 60s soft cap even for pages with
+ * 100+ internal links.
+ */
+const LINK_CHECK_CONCURRENCY = 5;
+function chunk(arr, size) {
+    if (size <= 0)
+        return [arr];
+    const out = [];
+    for (let i = 0; i < arr.length; i += size)
+        out.push(arr.slice(i, i + size));
+    return out;
+}
+//# sourceMappingURL=scanner.js.map

package/dist/content-health/scanner.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"scanner.js","sourceRoot":"","sources":["../../src/content-health/scanner.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAgCG;AAEH,OAAO,EAAE,qBAAqB,IAAI,2BAA2B,EAAE,MAAM,oBAAoB,CAAA;AAmDzF,8EAA8E;AAE9E;;;;;;GAMG;AACH,SAAS,WAAW,CAAC,IAA6B;IAChD,MAAM,UAAU,GAAG,CAAC,MAAM,EAAE,SAAS,EAAE,UAAU,EAAE,UAAU,EAAE,MAAM,EAAE,aAAa,CAAC,CAAA;IACrF,IAAI,IAAI,GAAG,EAAE,CAAA;IACb,KAAK,MAAM,GAAG,IAAI,UAAU,EAAE,CAAC;QAC7B,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAA;QACnB,IAAI,OAAO,CAAC,KAAK,QAAQ,IAAI,CAAC,CAAC,MAAM,GAAG,IAAI,CAAC,MAAM;YAAE,IAAI,GAAG,CAAC,CAAA;IAC/D,CAAC;IACD,OAAO,IAAI,CAAA;AACb,CAAC;AAED,MAAM,qBAAqB,GAAG,CAAC,iBAAiB,EAAE,gBAAgB,EAAE,kBAAkB,CAAC,CAAA;AAEvF,SAAS,kBAAkB,CAAC,IAA6B;IACvD,KAAK,MAAM,GAAG,IAAI,qBAAqB,EAAE,CAAC;QACxC,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAA;QACnB,IAAI,OAAO,CAAC,KAAK,QAAQ,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC;YAAE,OAAO,CAAC,CAAC,IAAI,EAAE,CAAA;IACnE,CAAC;IACD,kDAAkD;IAClD,MAAM,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,CAAA;IACvB,IAAI,GAAG,IAAI,OAAO,GAAG,KAAK,QAAQ,EAAE,CAAC;QACnC,MAAM,GAAG,GAAG,GAA8B,CAAA;QAC1C,MAAM,CAAC,GAAG,GAAG,CAAC,aAAa,CAAC,IAAI,GAAG,CAAC,iBAAiB,CAAC,CAAA;QACtD,IAAI,OAAO,CAAC,KAAK,QAAQ,IAAI,CAAC,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC;YAAE,OAAO,CAAC,CAAC,IAAI,EAAE,CAAA;IACnE,CAAC;IACD,OAAO,IAAI,CAAA;AACb,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,qBAAqB,CAAC,IAAY;IAChD,OAAO,2BAA2B,CAAC,IAAI,CAAC,CAAA;AAC1C,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,YAAY,CAAC,IAAY;IACvC,IAAI,CAAC,IAAI;QAAE,OAAO,EAAE,CAAA;IACpB,MAAM,GAAG,GAAa,EAAE,CAAA;IACxB,MAAM,EAAE,GAAG,+CAA+C,CAAA;IAC1D,IAAI,KAA6B,CAAA;IACjC,OAAO,CAAC,KAAK,GAAG,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC;QACxC,MAAM,GAAG,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,KAAK,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;QAC/C,IAAI,GAAG,CAAC,MAAM,GAAG,CAAC;YAAE,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;IACnC,CAAC;IACD,OAAO,GAAG,CAAA;AACZ,CAAC;AAED;;;;;;;;;;;GAWG;AACH,MAAM,UAAU,YAAY,CAC1B,GAAW,EACX,UAAkB;IAElB,MAAM,OAAO,GAAG,GAAG,CAAC,IAAI,EAAE,CAAA;IAC1B,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,UAAU,EAAE,KAAK,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAA;IACtE,IAAI,OAAO,CAAC,UAAU,CAAC,GAAG,CAAC;QAAE,OAAO,EAAE,UAAU,EAAE,KAAK,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAA;IACzE,oCAAoC;IACpC,MAAM,WAAW,GAAG,wBAAwB,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;IAC1D,IAAI,WAAW,EAAE,CAAC;QAChB,MAAM,MAAM,GAAG,WAAW,CAAC,CAAC,CAAE,CAAC,WAAW,EAAE,CAAA;QAC5C,IAAI,MAAM,KAAK,MAAM,IAAI,MAAM,KAAK,OAAO;YAAE,OAAO,EAAE,UAAU,EAAE,KAAK,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAA;IAC3F,CAAC;IAED,IAAI,YAAiB,CAAA;IACrB,IAAI,CAAC;QACH,YAAY,GAAG,IAAI,GAAG,CAAC,UAAU,CAAC,CAAA;IACpC,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,EAAE,UAAU,EAAE,KAAK,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAA;IAC9C,CAAC;IAED,IAAI,CAAC;QACH,MAAM,QAAQ,GAAG,IAAI,GAAG,CAAC,OAAO,EAAE,YAAY,CAAC,CAAA;QAC/C,IAAI,QAAQ,CAAC,QAAQ,KAAK,OAAO,IAAI,QAAQ,CAAC,QAAQ,KAAK,QAAQ,EAAE,CAAC;YACpE,OAAO,EAAE,UAAU,EAAE,KAAK,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAA;QAC9C,CAAC;QACD,OAAO;YACL,UAAU,EAAE,QAAQ,CAAC,MAAM,KAAK,YAAY,CAAC,MAAM;YACnD,QAAQ,EAAE,QAAQ,CAAC,QAAQ,EAAE;SAC9B,CAAA;IACH,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,EAAE,UAAU,EAAE,KAAK,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAA;IAC9C,CAAC;AACH,CAAC;AAED;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,0BAA0B,CAAC,GAAW,EAAE,SAAS,GAAG,IAAI;IAC5E,MAAM,UAAU,GAAG,IAAI,eAAe,EAAE,CAAA;IACxC,MAAM,CAAC,GAAG,UAAU,CAAC,GAAG,EAAE,CAAC,UAAU,CAAC,KAAK,EAAE,EAAE,SAAS,CAAC,CAAA;IACzD,IAAI,CAAC;QACH,MAAM,IAAI,GAAG,MAAM,KAAK,CAAC,GAAG,EAAE;YAC5B,MAAM,EAAE,MAAM;YACd,QAAQ,EAAE,QAAQ;YAClB,MAAM,EAAE,UAAU,CAAC,MAAM;SAC1B,CAAC,CAAA;QACF,IAAI,IAAI,CAAC,MAAM,KAAK,GAAG,IAAI,IAAI,CAAC,MAAM,KAAK,GAAG,EAAE,CAAC;YAC/C,yDAAyD;YACzD,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,GAAG,EAAE;gBAC3B,MAAM,EAAE,KAAK;gBACb,QAAQ,EAAE,QAAQ;gBAClB,MAAM,EAAE,UAAU,CAAC,MAAM;aAC1B,CAAC,CAAA;YACF,OAAO,GAAG,CAAC,MAAM,GAAG,GAAG,CAAA;QACzB,CAAC;QACD,OAAO,IAAI,CAAC,MAAM,GAAG,GAAG,CAAA;IAC1B,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,KAAK,CAAA;IACd,CAAC;YAAS,CAAC;QACT,YAAY,CAAC,CAAC,CAAC,CAAA;IACjB,CAAC;AACH,CAAC;AAED,8EAA8E;AAE9E;;;;;;;;;GASG;AACH,MAAM,CAAC,KAAK,UAAU,YAAY,CAChC,GAAiB,EACjB,OAAuB,EAAE;IAEzB,MAAM,YAAY,GAAG,IAAI,CAAC,YAAY,IAAI,EAAE,CAAA;IAC5C,MAAM,MAAM,GAAmB,EAAE,CAAA;IAEjC,2EAA2E;IAC3E,IAAI,CAAC,kBAAkB,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;QAClC,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,0BAA0B,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC,CAAA;IAC7D,CAAC;IAED,2EAA2E;IAC3E,MAAM,IAAI,GAAG,WAAW,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;IAClC,MAAM,aAAa,GAAG,qBAAqB,CAAC,IAAI,CAAC,CAAA;IACjD,IAAI,aAAa,GAAG,CAAC,EAAE,CAAC;QACtB,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,kBAAkB,EAAE,KAAK,EAAE,aAAa,EAAE,CAAC,CAAA;IACjE,CAAC;IAED,2EAA2E;IAC3E,MAAM,aAAa,GAAG,GAAG,CAAC,WAAW,IAAI,GAAG,CAAC,SAAS,CAAA;IACtD,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,EAAE,GAAG,aAAa,CAAC,OAAO,EAAE,CAAA;IAClD,IAAI,KAAK,GAAG,YAAY,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,IAAI,EAAE,CAAC;QAC/C,MAAM,CAAC,IAAI,CAAC;YACV,IAAI,EAAE,kBAAkB;YACxB,KAAK,EAAE,CAAC;YACR,OAAO,EAAE;gBACP,OAAO,EAAE,IAAI,CAAC,KAAK,CAAC,KAAK,GAAG,CAAC,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,IAAI,CAAC,CAAC;gBAClD,SAAS,EAAE,YAAY;aACxB;SACF,CAAC,CAAA;IACJ,CAAC;IAED,0EAA0E;IAC1E,IAAI,IAAI,CAAC,SAAS,IAAI,IAAI,CAAC,UAAU,EAAE,CAAC;QACtC,MAAM,KAAK,GAAG,YAAY,CAAC,IAAI,CAAC,CAAA;QAChC,MAAM,YAAY,GAAG,IAAI,GAAG,EAAU,CAAA;QACtC,KAAK,MAAM,GAAG,IAAI,KAAK,EAAE,CAAC;YACxB,MAAM,EAAE,UAAU,EAAE,QAAQ,EAAE,GAAG,YAAY,CAAC,GAAG,EAAE,IAAI,CAAC,UAAU,CAAC,CAAA;YACnE,IAAI,UAAU,IAAI,QAAQ;gBAAE,YAAY,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAA;QACxD,CAAC;QACD,MAAM,MAAM,GAAa,EAAE,CAAA;QAC3B,IAAI,YAAY,CAAC,IAAI,GAAG,CAAC,EAAE,CAAC;YAC1B,MAAM,OAAO,GAAG,IAAI,CAAC,SAAS,CAAA;YAC9B,MAAM,OAAO,CAAC,GAAG,CACf,KAAK,CAAC,CAAC,GAAG,YAAY,CAAC,EAAE,sBAAsB,CAAC,CAAC,GAAG,CAAC,KAAK,EAAE,KAAK,EAAE,EAAE;gBACnE,MAAM,OAAO,GAAG,MAAM,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,EAAE,CAAC,KAAK,CAAC,CAAC,CAAC,CAAA;gBAClF,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE;oBACrB,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC;wBAAE,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAA;gBACjC,CAAC,CAAC,CAAA;YACJ,CAAC,CAAC,CACH,CAAA;QACH,CAAC;QACD,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACtB,MAAM,CAAC,IAAI,CAAC;gBACV,IAAI,EAAE,sBAAsB;gBAC5B,KAAK,EAAE,MAAM,CAAC,MAAM;gBACpB,OAAO,EAAE,EAAE,IAAI,EAAE,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,EAAE,EAAE,qBAAqB;aAC9D,CAAC,CAAA;QACJ,CAAC;IACH,CAAC;IAED,OAAO,MAAM,CAAA;AACf,CAAC;AAED;;;;;GAKG;AACH,MAAM,sBAAsB,GAAG,CAAC,CAAA;AAEhC,SAAS,KAAK,CAAI,GAAQ,EAAE,IAAY;IACtC,IAAI,IAAI,IAAI,CAAC;QAAE,OAAO,CAAC,GAAG,CAAC,CAAA;IAC3B,MAAM,GAAG,GAAU,EAAE,CAAA;IACrB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC,IAAI,IAAI;QAAE,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,CAAC,CAAA;IAC3E,OAAO,GAAG,CAAA;AACZ,CAAC"}

package/dist/cron/index.d.ts

@@ -0,0 +1,90 @@
+/**
+ * Cron handlers for scheduled platform jobs.
+ *
+ * The endpoints in `api/handlers.ts` validate `Authorization: Bearer ${CRON_SECRET}`
+ * before invoking these — Vercel Cron sends that header automatically when
+ * `CRON_SECRET` is defined in the project's environment. See:
+ * https://vercel.com/docs/cron-jobs/manage-cron-jobs#securing-cron-jobs
+ *
+ * Each handler is **idempotent** and **bounded** — safe to invoke from any
+ * scheduler (Vercel Cron, GitHub Actions, EventBridge, k8s CronJob, etc.) and
+ * safe to invoke twice in the same window.
+ */
+import { schedulingCronHandler } from '../scheduling/index.js';
+import type { PrentaDB } from '../db/client-types.js';
+type PrismaDB = PrentaDB;
+export interface CleanupOptions {
+    /** Hard-delete revoked / expired sessions older than this. Default: 7d. */
+    sessionRetentionMs?: number;
+    /** Hard-delete audit log entries older than this. Default: 90d. */
+    auditLogRetentionMs?: number;
+    /** Hard-delete soft-deleted documents older than this. Default: 30d. */
+    trashRetentionMs?: number;
+    /** Hard-delete used / expired password reset tokens older than this. Default: 1d. */
+    passwordResetRetentionMs?: number;
+    /**
+     * Hard-delete form submissions older than this (PII retention). Opt-in:
+     * `0` (the default) disables forms retention entirely. Wire from
+     * `forms.retentionDays` in the CMS config.
+     */
+    formSubmissionRetentionMs?: number;
+    /** Hard-delete document version rows older than this. Default: 90d. */
+    versionRetentionMs?: number;
+    /** Hard-delete webhook delivery logs older than this. Default: 30d. */
+    webhookDeliveryRetentionMs?: number;
+    /** Hard-delete API request metric buckets older than this. Default: 30d. */
+    apiRequestMetricRetentionMs?: number;
+}
+export interface CleanupResult {
+    sessionsDeleted: number;
+    auditLogsDeleted: number;
+    documentsDeleted: number;
+    passwordResetTokensDeleted: number;
+    formSubmissionsDeleted: number;
+    versionsDeleted: number;
+    webhookDeliveriesDeleted: number;
+    apiRequestMetricsDeleted: number;
+}
+/**
+ * Delete stale rows from session, audit log, trash, and password-reset tables.
+ *
+ * Each deletion is wrapped in its own try/catch so a missing model (e.g. an
+ * older Prisma schema without `passwordResetToken`) doesn't fail the entire
+ * job — partial cleanup is still useful and the caller logs the count.
+ */
+export declare function processCleanup(db: PrismaDB, options?: CleanupOptions): Promise<CleanupResult>;
+export interface SeoScanIssue {
+    documentId: string;
+    title: string;
+    slug: string;
+    problems: string[];
+}
+export interface SeoScanResult {
+    total: number;
+    pagesWithIssues: number;
+    totalProblems: number;
+    issues: SeoScanIssue[];
+}
+/**
+ * Run the same SEO checks as `POST /seo/scan`, but headlessly so a cron
+ * (or a CLI / external scheduler) can invoke it without an admin session.
+ *
+ * Bounded by `maxDocuments` to avoid OOM / timeout on large catalogs —
+ * defaults to 5000, which is well above the typical Vercel function memory
+ * envelope while staying safely under the 60s cron execution limit.
+ */
+export declare function processSeoScan(db: PrismaDB, options?: {
+    maxDocuments?: number;
+    allowedCollections?: string[];
+}): Promise<SeoScanResult>;
+/** Re-export for convenience so the route handler can import everything from one module. */
+export { schedulingCronHandler };
+/**
+ * Validate a Vercel-style cron Authorization header against `CRON_SECRET`.
+ *
+ * Returns `false` if the env var is missing — that's deliberately fail-closed
+ * so a misconfigured deploy can't silently expose cron endpoints to the public
+ * internet. Comparison is constant-time to defeat byte-by-byte timing attacks.
+ */
+export declare function isAuthorizedCronRequest(authHeader: string | null | undefined): boolean;
+//# sourceMappingURL=index.d.ts.map

package/dist/cron/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/cron/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAGH,OAAO,EAAE,qBAAqB,EAAE,MAAM,wBAAwB,CAAA;AAC9D,OAAO,KAAK,EAAE,QAAQ,EAAE,MAAM,uBAAuB,CAAA;AAQrD,KAAK,QAAQ,GAAG,QAAQ,CAAA;AAExB,MAAM,WAAW,cAAc;IAC7B,2EAA2E;IAC3E,kBAAkB,CAAC,EAAE,MAAM,CAAA;IAC3B,mEAAmE;IACnE,mBAAmB,CAAC,EAAE,MAAM,CAAA;IAC5B,wEAAwE;IACxE,gBAAgB,CAAC,EAAE,MAAM,CAAA;IACzB,qFAAqF;IACrF,wBAAwB,CAAC,EAAE,MAAM,CAAA;IACjC;;;;OAIG;IACH,yBAAyB,CAAC,EAAE,MAAM,CAAA;IAClC,uEAAuE;IACvE,kBAAkB,CAAC,EAAE,MAAM,CAAA;IAC3B,uEAAuE;IACvE,0BAA0B,CAAC,EAAE,MAAM,CAAA;IACnC,4EAA4E;IAC5E,2BAA2B,CAAC,EAAE,MAAM,CAAA;CACrC;AAeD,MAAM,WAAW,aAAa;IAC5B,eAAe,EAAE,MAAM,CAAA;IACvB,gBAAgB,EAAE,MAAM,CAAA;IACxB,gBAAgB,EAAE,MAAM,CAAA;IACxB,0BAA0B,EAAE,MAAM,CAAA;IAClC,sBAAsB,EAAE,MAAM,CAAA;IAC9B,eAAe,EAAE,MAAM,CAAA;IACvB,wBAAwB,EAAE,MAAM,CAAA;IAChC,wBAAwB,EAAE,MAAM,CAAA;CACjC;AAcD;;;;;;GAMG;AACH,wBAAsB,cAAc,CAClC,EAAE,EAAE,QAAQ,EACZ,OAAO,GAAE,cAAmB,GAC3B,OAAO,CAAC,aAAa,CAAC,CA2GxB;AAED,MAAM,WAAW,YAAY;IAC3B,UAAU,EAAE,MAAM,CAAA;IAClB,KAAK,EAAE,MAAM,CAAA;IACb,IAAI,EAAE,MAAM,CAAA;IACZ,QAAQ,EAAE,MAAM,EAAE,CAAA;CACnB;AAED,MAAM,WAAW,aAAa;IAC5B,KAAK,EAAE,MAAM,CAAA;IACb,eAAe,EAAE,MAAM,CAAA;IACvB,aAAa,EAAE,MAAM,CAAA;IACrB,MAAM,EAAE,YAAY,EAAE,CAAA;CACvB;AAED;;;;;;;GAOG;AACH,wBAAsB,cAAc,CAClC,EAAE,EAAE,QAAQ,EACZ,OAAO,GAAE;IAAE,YAAY,CAAC,EAAE,MAAM,CAAC;IAAC,kBAAkB,CAAC,EAAE,MAAM,EAAE,CAAA;CAAO,GACrE,OAAO,CAAC,aAAa,CAAC,CAgGxB;AAED,4FAA4F;AAC5F,OAAO,EAAE,qBAAqB,EAAE,CAAA;AAEhC;;;;;;GAMG;AACH,wBAAgB,uBAAuB,CAAC,UAAU,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,GAAG,OAAO,CAStF"}