@growth-labs/cms 0.4.0 → 0.5.1

package/src/engine/foundry-dispatch.ts

@@ -28,6 +28,15 @@ export interface DispatchResult {
 	triggerToken: string
 }
 
+export class FoundryDispatchPrerequisiteError extends Error {
+	readonly code = 'missing_duration' as const
+
+	constructor(contentId: string) {
+		super(`Content ${contentId} requires a positive finite duration before dispatch`)
+		this.name = 'FoundryDispatchPrerequisiteError'
+	}
+}
+
 export interface CallbackInput {
 	eventId: string
 	eventKind: string
@@ -82,7 +91,7 @@ function inferVideoSourceKind(sourceUrl: string): string {
 /**
  * Dispatch a video or podcast content item to Foundry via the injected hook.
  * Mints a fresh `processing_trigger_token` (UUID), calls
- * `hook.dispatchVideo({ contentId, sourceUrl, kind })`, and writes
+ * `hook.dispatchVideo({ contentId, sourceUrl, kind, durationSeconds })`, and writes
  * `processing_correlation_id` + advances to `'queued'` in either
  * `video_content` or `podcast_content`.
  *
@@ -110,12 +119,13 @@ export async function dispatchToFoundry(
 	// Determine which media table owns this content id.
 	const videoRow = await db
 		.prepare(
-			'SELECT content_id, video_id, processing_source_url, processing_source_kind FROM video_content WHERE content_id = ?',
+			'SELECT content_id, video_id, duration_seconds, processing_source_url, processing_source_kind FROM video_content WHERE content_id = ?',
 		)
 		.bind(contentId)
 		.first<{
 			content_id: string
 			video_id: string
+			duration_seconds: number | null
 			processing_source_url: string | null
 			processing_source_kind: string | null
 		}>()
@@ -125,6 +135,7 @@ export async function dispatchToFoundry(
 	let kind: string
 	let sourceKind: string | null = null
 	let videoId: string | null = null
+	let durationSeconds: number | null
 
 	if (videoRow) {
 		table = 'video_content'
@@ -135,11 +146,18 @@ export async function dispatchToFoundry(
 		kind = 'video'
 		sourceKind = videoRow.processing_source_kind || inferVideoSourceKind(sourceUrl)
 		videoId = normalizeVideoId(videoRow.video_id, content.slug)
+		durationSeconds = videoRow.duration_seconds
 	} else {
 		const podcastRow = await db
-			.prepare('SELECT content_id, audio_r2_key FROM podcast_content WHERE content_id = ?')
+			.prepare(
+				'SELECT content_id, audio_r2_key, duration_seconds FROM podcast_content WHERE content_id = ?',
+			)
 			.bind(contentId)
-			.first<{ content_id: string; audio_r2_key: string | null }>()
+			.first<{
+				content_id: string
+				audio_r2_key: string | null
+				duration_seconds: number | null
+			}>()
 
 		if (!podcastRow) {
 			throw new Error(
@@ -154,6 +172,15 @@ export async function dispatchToFoundry(
 		kind = 'podcast'
 		sourceKind = 'r2_key'
 		videoId = normalizeVideoId(content.slug, content.slug)
+		durationSeconds = podcastRow.duration_seconds
+	}
+
+	if (
+		typeof durationSeconds !== 'number' ||
+		!Number.isFinite(durationSeconds) ||
+		durationSeconds <= 0
+	) {
+		throw new FoundryDispatchPrerequisiteError(contentId)
 	}
 
 	const triggerToken = crypto.randomUUID()
@@ -162,6 +189,7 @@ export async function dispatchToFoundry(
 		contentId,
 		sourceUrl,
 		kind,
+		durationSeconds,
 		...(sourceKind ? { sourceKind } : {}),
 		...(videoId ? { videoId } : {}),
 		title: content.title,

package/src/engine/index.ts

@@ -204,6 +204,7 @@ export {
 	type UpdateContentInput,
 	unpublishContent,
 	updateContentItem,
+	VerifiedMediaDurationConflictError,
 	type VideoInput,
 } from './publisher.js'
 // Verified-user/site reader product state. Analytics remains observation-only.

package/src/engine/published-content.ts

@@ -17,6 +17,14 @@ export const MAX_PUBLISHED_CONTENT_PAGE_SIZE = 100
 export const MAX_PUBLISHED_CONTENT_PAYLOAD_BYTES = 4 * 1024 * 1024
 export const MAX_PUBLISHED_CONTENT_PAGE_BYTES = 4 * 1024 * 1024
 
+// D1/SQLite caps a single prepared statement at SQLITE_MAX_VARIABLE_NUMBER (100)
+// bound variables. The hydrate query binds two variables per selected outline
+// (`ci.id` + `ci.published_revision_id`), so a page of 51+ outlines in one
+// statement exceeds the cap and D1 rejects it with `too many SQL variables`.
+// Hydrate in chunks of at most this many outlines (=> <=100 bound variables) so
+// a full MAX_PUBLISHED_CONTENT_PAGE_SIZE page reads across multiple statements.
+const MAX_HYDRATE_CHUNK_OUTLINES = 50
+
 export type PublishedContentReadErrorCode =
 	| 'published_content_input_invalid'
 	| 'published_revision_missing'
@@ -534,25 +542,33 @@ LIMIT ?`,
 		}
 		if (selected.length === 0) return { items: [], nextCursor: null }
 
-		const selectedPredicates = selected.map(() => '(ci.id = ? AND ci.published_revision_id = ?)')
-		const selectedValues = selected.flatMap((outline) => [
-			outline.content_id,
-			outline.pointer_revision_id,
-		])
-		const result = await db
-			.prepare(
-				`${SELECT_PUBLISHED_CONTENT}
+		// Hydrate the selected outlines in bound-variable-safe chunks. Each chunk
+		// binds two variables per outline; capping at MAX_HYDRATE_CHUNK_OUTLINES
+		// keeps every statement at or under D1's 100-variable limit. Rows merge
+		// into one pointer-keyed map, so ordering and the per-item integrity
+		// checks below are identical to a single-statement hydrate.
+		const rowsByPointer = new Map<string, PublishedContentRow>()
+		for (let offset = 0; offset < selected.length; offset += MAX_HYDRATE_CHUNK_OUTLINES) {
+			const chunk = selected.slice(offset, offset + MAX_HYDRATE_CHUNK_OUTLINES)
+			const chunkPredicates = chunk.map(() => '(ci.id = ? AND ci.published_revision_id = ?)')
+			const chunkValues = chunk.flatMap((outline) => [
+				outline.content_id,
+				outline.pointer_revision_id,
+			])
+			const chunkResult = await db
+				.prepare(
+					`${SELECT_PUBLISHED_CONTENT}
 JOIN content_revisions cr ON cr.id = ci.published_revision_id
 WHERE ci.status = 'published'
   AND ci.deleted_at IS NULL
-  AND (${selectedPredicates.join(' OR ')})`,
-			)
-			.bind(...selectedValues)
-			.all<PublishedContentRow>()
-		const rows = result.results ?? []
-		const rowsByPointer = new Map(
-			rows.map((row) => [`${row.content_id}\0${row.pointer_revision_id}`, row]),
-		)
+  AND (${chunkPredicates.join(' OR ')})`,
+				)
+				.bind(...chunkValues)
+				.all<PublishedContentRow>()
+			for (const row of chunkResult.results ?? []) {
+				rowsByPointer.set(`${row.content_id}\0${row.pointer_revision_id}`, row)
+			}
+		}
 		const items = selected.map((outline) => {
 			const row = rowsByPointer.get(`${outline.content_id}\0${outline.pointer_revision_id}`)
 			if (

package/src/engine/publisher.ts

@@ -181,6 +181,118 @@ function isBodyBackedContentInput(input: CreateContentInput): input is BodyBacke
 	return isBodyBackedContentType(input.type)
 }
 
+function assertValidMediaDuration(value: unknown): asserts value is number | null | undefined {
+	if (value === null || value === undefined) return
+	if (
+		typeof value !== 'number' ||
+		!Number.isFinite(value) ||
+		!Number.isInteger(value) ||
+		value <= 0
+	) {
+		throw new Error('Media duration must be a positive integer when supplied')
+	}
+}
+
+export class VerifiedMediaDurationConflictError extends Error {
+	readonly code = 'verified_duration_conflict'
+
+	constructor() {
+		super('Media duration conflicts with verified upload duration')
+		this.name = 'VerifiedMediaDurationConflictError'
+	}
+}
+
+type VerifiedMediaKind = 'video' | 'podcast'
+
+function parseVerifiedMediaDuration(variantsJson: string | null): number | null {
+	if (!variantsJson) return null
+	try {
+		const parsed: unknown = JSON.parse(variantsJson)
+		if (!parsed || typeof parsed !== 'object' || Array.isArray(parsed)) return null
+		const durationSeconds = (parsed as Record<string, unknown>).durationSeconds
+		return typeof durationSeconds === 'number' &&
+			Number.isFinite(durationSeconds) &&
+			Number.isInteger(durationSeconds) &&
+			durationSeconds > 0
+			? durationSeconds
+			: null
+	} catch {
+		return null
+	}
+}
+
+async function loadVerifiedMediaDuration(
+	db: D1Database,
+	kind: VerifiedMediaKind,
+	source: string | null | undefined,
+): Promise<number | null> {
+	if (!source) return null
+
+	const isHttpUrl = /^https?:\/\//i.test(source)
+	const statement =
+		kind === 'video'
+			? isHttpUrl
+				? db
+						.prepare(
+							`SELECT variants_json
+							 FROM media_assets
+							 WHERE asset_type = 'video'
+							   AND source_type = 'editor-video-source'
+							   AND (url = ? OR public_url = ?)`,
+						)
+						.bind(source, source)
+				: db
+						.prepare(
+							`SELECT variants_json
+							 FROM media_assets
+							 WHERE asset_type = 'video'
+							   AND source_type = 'editor-video-source'
+							   AND (id = ? OR object_key = ?)`,
+						)
+						.bind(source, source)
+			: db
+					.prepare(
+						`SELECT variants_json
+						 FROM media_assets
+						 WHERE asset_type = 'audio'
+						   AND source_type = 'editor-upload'
+						   AND (id = ? OR object_key = ?)`,
+					)
+					.bind(source, source)
+
+	const rows = await statement.all<{ variants_json: string | null }>()
+	const durations = new Set<number>()
+	for (const row of rows.results ?? []) {
+		const duration = parseVerifiedMediaDuration(row.variants_json)
+		if (duration !== null) durations.add(duration)
+	}
+
+	if (durations.size > 1) {
+		// Multiple exact package-owned identities disagreeing is unsafe to guess.
+		throw new VerifiedMediaDurationConflictError()
+	}
+	return durations.values().next().value ?? null
+}
+
+async function resolveMediaDuration(
+	db: D1Database,
+	kind: VerifiedMediaKind,
+	source: string | null | undefined,
+	suppliedDuration: unknown,
+): Promise<number | null> {
+	assertValidMediaDuration(suppliedDuration)
+	const verifiedDuration = await loadVerifiedMediaDuration(db, kind, source)
+	if (verifiedDuration === null) return suppliedDuration ?? null
+	if (
+		suppliedDuration !== null &&
+		suppliedDuration !== undefined &&
+		suppliedDuration !== verifiedDuration
+	) {
+		throw new VerifiedMediaDurationConflictError()
+	}
+	return verifiedDuration
+}
+
 export interface UpdateContentInput extends Partial<BaseContentInput> {
 	content?: Partial<
 		| ArticleInput['content']
@@ -197,6 +309,23 @@ export interface UpdateContentInput extends Partial<BaseContentInput> {
 	aiLockedFields?: string[]
 }
 
+interface VideoUpdatePlan {
+	script: string
+	videoId: string
+	durationSeconds: number | null
+	thumbnailImageId: string | null
+	sourceUrl: string | null
+	sourceKind: string | null
+	sourceChanged: boolean
+}
+
+interface PodcastUpdatePlan {
+	transcript: string
+	audioR2Key: string
+	durationSeconds: number | null
+	processingTriggerToken: string | null
+}
+
 /**
  * The shape `getContentItem` returns from `SELECT * FROM content_items`.
  * Note this is the engine's read-side row (nullable `featured`/timestamps as
@@ -416,6 +545,22 @@ export async function createContent(
 	db: D1Database,
 	input: CreateContentInput,
 ): Promise<{ id: string; slug: string }> {
+	const resolvedMediaDuration =
+		input.type === 'video'
+			? await resolveMediaDuration(
+					db,
+					'video',
+					input.content.processingSourceUrl,
+					input.content.durationSeconds,
+				)
+			: input.type === 'podcast'
+				? await resolveMediaDuration(
+						db,
+						'podcast',
+						input.content.audioR2Key,
+						input.content.durationSeconds,
+					)
+				: null
 	const id = crypto.randomUUID()
 	const slug = await ensureUniqueSlug(db, input.slug)
 	const visibility = input.visibility || 'free'
@@ -502,7 +647,7 @@ export async function createContent(
 				id,
 				input.content.script,
 				input.content.videoId,
-				input.content.durationSeconds || null,
+				resolvedMediaDuration,
 				input.content.thumbnailImageId || null,
 				initialVideoProcessingState,
 				input.content.processingSourceUrl || null,
@@ -517,12 +662,7 @@ export async function createContent(
 				`INSERT INTO podcast_content (content_id, transcript, audio_r2_key, duration_seconds)
          VALUES (?, ?, ?, ?)`,
 			)
-			.bind(
-				id,
-				input.content.transcript,
-				input.content.audioR2Key,
-				input.content.durationSeconds || null,
-			)
+			.bind(id, input.content.transcript, input.content.audioR2Key, resolvedMediaDuration)
 			.run()
 	}
 
@@ -552,6 +692,123 @@ export async function updateContentItem(
 	if (!existing) return null
 
 	const hasOwn = (source: object, key: string) => Object.hasOwn(source, key)
+	let videoUpdatePlan: VideoUpdatePlan | null | undefined
+	let podcastUpdatePlan: PodcastUpdatePlan | null | undefined
+
+	// Resolve all media identity/duration invariants before the base content row,
+	// tags, or relations can be mutated. D1 does not provide a transaction on the
+	// structural surface used here, so validation must be a strict preflight.
+	if (input.content && existing.type === 'video') {
+		const current = await db
+			.prepare(
+				`SELECT script, video_id, duration_seconds, thumbnail_image_id,
+					processing_source_url, processing_source_kind
+				 FROM video_content
+				 WHERE content_id = ?
+				 LIMIT 1`,
+			)
+			.bind(id)
+			.first<{
+				script: string
+				video_id: string
+				duration_seconds: number | null
+				thumbnail_image_id: string | null
+				processing_source_url: string | null
+				processing_source_kind: string | null
+			}>()
+
+		if (current) {
+			const videoContent = input.content as Partial<VideoInput['content']>
+			const sourceUrl = hasOwn(videoContent, 'processingSourceUrl')
+				? (videoContent.processingSourceUrl ?? null)
+				: current.processing_source_url
+			const sourceKind = hasOwn(videoContent, 'processingSourceKind')
+				? (videoContent.processingSourceKind ?? null)
+				: current.processing_source_kind
+			const sourceChanged =
+				sourceUrl !== current.processing_source_url || sourceKind !== current.processing_source_kind
+			const durationProvided = hasOwn(videoContent, 'durationSeconds')
+			const resolvedDuration = await resolveMediaDuration(
+				db,
+				'video',
+				sourceUrl,
+				durationProvided ? videoContent.durationSeconds : undefined,
+			)
+
+			videoUpdatePlan = {
+				script: hasOwn(videoContent, 'script')
+					? (videoContent.script ?? current.script)
+					: current.script,
+				videoId: hasOwn(videoContent, 'videoId')
+					? (videoContent.videoId ?? current.video_id)
+					: current.video_id,
+				durationSeconds:
+					durationProvided || sourceChanged
+						? resolvedDuration
+						: (resolvedDuration ?? current.duration_seconds),
+				thumbnailImageId: hasOwn(videoContent, 'thumbnailImageId')
+					? (videoContent.thumbnailImageId ?? null)
+					: current.thumbnail_image_id,
+				sourceUrl,
+				sourceKind,
+				sourceChanged,
+			}
+		} else {
+			videoUpdatePlan = null
+		}
+	}
+
+	if (input.content && existing.type === 'podcast') {
+		const current = await db
+			.prepare(
+				`SELECT transcript, audio_r2_key, duration_seconds, processing_trigger_token
+				 FROM podcast_content
+				 WHERE content_id = ?
+				 LIMIT 1`,
+			)
+			.bind(id)
+			.first<{
+				transcript: string
+				audio_r2_key: string
+				duration_seconds: number | null
+				processing_trigger_token: string | null
+			}>()
+
+		if (current) {
+			const podcastContent = input.content as Partial<PodcastInput['content']>
+			const audioR2Key = hasOwn(podcastContent, 'audioR2Key')
+				? (podcastContent.audioR2Key ?? current.audio_r2_key)
+				: current.audio_r2_key
+			const sourceChanged = audioR2Key !== current.audio_r2_key
+			const durationProvided = hasOwn(podcastContent, 'durationSeconds')
+			const resolvedDuration = await resolveMediaDuration(
+				db,
+				'podcast',
+				audioR2Key,
+				durationProvided ? podcastContent.durationSeconds : undefined,
+			)
+			const replacementTranscriptProvided =
+				hasOwn(podcastContent, 'transcript') && typeof podcastContent.transcript === 'string'
+
+			podcastUpdatePlan = {
+				transcript: sourceChanged
+					? replacementTranscriptProvided
+						? (podcastContent.transcript as string)
+						: ''
+					: replacementTranscriptProvided
+						? (podcastContent.transcript as string)
+						: current.transcript,
+				audioR2Key,
+				durationSeconds:
+					durationProvided || sourceChanged
+						? resolvedDuration
+						: (resolvedDuration ?? current.duration_seconds),
+				processingTriggerToken: sourceChanged ? null : current.processing_trigger_token,
+			}
+		} else {
+			podcastUpdatePlan = null
+		}
+	}
 
 	const slug = input.slug ? await ensureUniqueSlug(db, input.slug, id) : existing.slug
 	const visibility = input.visibility || existing.visibility
@@ -735,37 +992,7 @@ export async function updateContentItem(
 		}
 
 		if (existing.type === 'video') {
-			const current = await db
-				.prepare(
-					`
-          SELECT script, video_id, duration_seconds, thumbnail_image_id,
-            processing_source_url, processing_source_kind
-          FROM video_content
-          WHERE content_id = ?
-          LIMIT 1
-          `,
-				)
-				.bind(id)
-				.first<{
-					script: string
-					video_id: string
-					duration_seconds: number | null
-					thumbnail_image_id: string | null
-					processing_source_url: string | null
-					processing_source_kind: string | null
-				}>()
-			if (!current) return { slug }
-
-			const videoContent = input.content as Partial<VideoInput['content']>
-			const nextSourceUrl = hasOwn(videoContent, 'processingSourceUrl')
-				? (videoContent.processingSourceUrl ?? null)
-				: current.processing_source_url
-			const nextSourceKind = hasOwn(videoContent, 'processingSourceKind')
-				? (videoContent.processingSourceKind ?? null)
-				: current.processing_source_kind
-			const sourceChanged =
-				nextSourceUrl !== current.processing_source_url ||
-				nextSourceKind !== current.processing_source_kind
+			if (!videoUpdatePlan) return { slug }
 			await db
 				.prepare(
 					`
@@ -790,78 +1017,51 @@ export async function updateContentItem(
             hls_manifest_url = CASE WHEN ? THEN NULL ELSE hls_manifest_url END,
             hls_poster_url = CASE WHEN ? THEN NULL ELSE hls_poster_url END
           WHERE content_id = ?
-          `,
+		  `,
 				)
 				.bind(
-					hasOwn(videoContent, 'script') ? (videoContent.script ?? current.script) : current.script,
-					hasOwn(videoContent, 'videoId')
-						? (videoContent.videoId ?? current.video_id)
-						: current.video_id,
-					hasOwn(videoContent, 'durationSeconds')
-						? (videoContent.durationSeconds ?? null)
-						: current.duration_seconds,
-					hasOwn(videoContent, 'thumbnailImageId')
-						? (videoContent.thumbnailImageId ?? null)
-						: current.thumbnail_image_id,
-					nextSourceUrl,
-					nextSourceKind,
-					sourceChanged ? 1 : 0,
-					sourceChanged ? 1 : 0,
-					sourceChanged ? 1 : 0,
-					sourceChanged ? 1 : 0,
-					sourceChanged ? 1 : 0,
-					sourceChanged ? 1 : 0,
-					sourceChanged ? 1 : 0,
-					sourceChanged ? 1 : 0,
-					sourceChanged ? 1 : 0,
-					sourceChanged ? 1 : 0,
-					sourceChanged ? 1 : 0,
-					sourceChanged ? 1 : 0,
-					sourceChanged ? 1 : 0,
+					videoUpdatePlan.script,
+					videoUpdatePlan.videoId,
+					videoUpdatePlan.durationSeconds,
+					videoUpdatePlan.thumbnailImageId,
+					videoUpdatePlan.sourceUrl,
+					videoUpdatePlan.sourceKind,
+					videoUpdatePlan.sourceChanged ? 1 : 0,
+					videoUpdatePlan.sourceChanged ? 1 : 0,
+					videoUpdatePlan.sourceChanged ? 1 : 0,
+					videoUpdatePlan.sourceChanged ? 1 : 0,
+					videoUpdatePlan.sourceChanged ? 1 : 0,
+					videoUpdatePlan.sourceChanged ? 1 : 0,
+					videoUpdatePlan.sourceChanged ? 1 : 0,
+					videoUpdatePlan.sourceChanged ? 1 : 0,
+					videoUpdatePlan.sourceChanged ? 1 : 0,
+					videoUpdatePlan.sourceChanged ? 1 : 0,
+					videoUpdatePlan.sourceChanged ? 1 : 0,
+					videoUpdatePlan.sourceChanged ? 1 : 0,
+					videoUpdatePlan.sourceChanged ? 1 : 0,
 					id,
 				)
 				.run()
 		}
 
 		if (existing.type === 'podcast') {
-			const current = await db
-				.prepare(
-					`
-          SELECT transcript, audio_r2_key, duration_seconds
-          FROM podcast_content
-          WHERE content_id = ?
-          LIMIT 1
-          `,
-				)
-				.bind(id)
-				.first<{
-					transcript: string
-					audio_r2_key: string
-					duration_seconds: number | null
-				}>()
-			if (!current) return { slug }
-
-			const podcastContent = input.content as Partial<PodcastInput['content']>
+			if (!podcastUpdatePlan) return { slug }
 			await db
 				.prepare(
 					`
-          UPDATE podcast_content SET
-            transcript = ?,
-            audio_r2_key = ?,
-            duration_seconds = ?
-          WHERE content_id = ?
-          `,
+		  UPDATE podcast_content SET
+		    transcript = ?,
+		    audio_r2_key = ?,
+		    duration_seconds = ?,
+		    processing_trigger_token = ?
+		  WHERE content_id = ?
+		  `,
 				)
 				.bind(
-					hasOwn(podcastContent, 'transcript')
-						? (podcastContent.transcript ?? current.transcript)
-						: current.transcript,
-					hasOwn(podcastContent, 'audioR2Key')
-						? (podcastContent.audioR2Key ?? current.audio_r2_key)
-						: current.audio_r2_key,
-					hasOwn(podcastContent, 'durationSeconds')
-						? (podcastContent.durationSeconds ?? null)
-						: current.duration_seconds,
+					podcastUpdatePlan.transcript,
+					podcastUpdatePlan.audioR2Key,
+					podcastUpdatePlan.durationSeconds,
+					podcastUpdatePlan.processingTriggerToken,
 					id,
 				)
 				.run()

package/src/providers/types.ts

@@ -237,6 +237,7 @@ export interface FoundryJobSpec {
 	sourceUrl: string
 	sourceKind?: string
 	kind: string
+	durationSeconds: number
 	videoId?: string
 	title?: string
 	description?: string | null