@growth-labs/seo 0.5.0 → 0.5.1

package/src/utils/sitemap.ts

@@ -105,9 +105,11 @@ ${entries}
 
 export function generateVideoSitemap(items: ContentItem[]): string {
 	const entries = items
-		.filter((item) => item.video)
+		.filter((item): item is ContentItem & { video: NonNullable<ContentItem['video']> } =>
+			Boolean(item.video && (item.video.contentUrl || item.video.embedUrl)),
+		)
 		.map((item) => {
-			const v = item.video!
+			const v = item.video
 			const contentUrlTag = v.contentUrl
 				? `\n      <video:content_loc>${escapeXml(v.contentUrl)}</video:content_loc>`
 				: ''

package/src/utils/validation.ts

@@ -7,6 +7,11 @@ export interface PageValidationOptions {
 	titleMaxLength: number
 	descriptionMaxLength: number
 	heroMinWidth?: number
+	pagePath?: string
+	requireH1?: boolean
+	requireHeroImage?: boolean
+	requireArticleSchema?: boolean
+	requireMaxImagePreviewLarge?: boolean
 }
 
 /**
@@ -139,7 +144,7 @@ export function validatePage(html: string, options: PageValidationOptions): Vali
 		html.match(/<meta\s[^>]*property=["']og:image["'][^>]*>/i) ??
 		html.match(/<meta\s[^>]*property=og:image[^>]*>/i)
 	if (!ogImageMatch) {
-		warnings.push('Missing og:image meta tag')
+		pushIssue(options.requireHeroImage, errors, warnings, 'Missing hero image og:image meta tag')
 	}
 
 	// og:image:width check against heroMinWidth
@@ -154,18 +159,27 @@ export function validatePage(html: string, options: PageValidationOptions): Vali
 			if (widthContentMatch) {
 				const width = Number(widthContentMatch[1])
 				if (!Number.isNaN(width) && width < options.heroMinWidth) {
-					warnings.push(`Hero image width ${width}px is below minimum ${options.heroMinWidth}px`)
+					pushIssue(
+						options.requireHeroImage,
+						errors,
+						warnings,
+						`Hero image width ${width}px is below minimum ${options.heroMinWidth}px`,
+					)
 				}
+			} else if (options.requireHeroImage) {
+				errors.push(`Hero image width is missing; minimum is ${options.heroMinWidth}px`)
 			}
+		} else if (options.requireHeroImage && ogImageMatch) {
+			errors.push(`Hero image width is missing; minimum is ${options.heroMinWidth}px`)
 		}
 	}
 
 	// H1 checks
 	const h1Matches = html.match(/<h1[\s>]/gi) ?? []
 	if (h1Matches.length === 0) {
-		warnings.push('Missing H1 tag')
+		pushIssue(options.requireH1, errors, warnings, 'Missing H1 tag')
 	} else if (h1Matches.length > 1) {
-		warnings.push(`Multiple H1 tags found (${h1Matches.length})`)
+		pushIssue(options.requireH1, errors, warnings, `Multiple H1 tags found (${h1Matches.length})`)
 	}
 
 	// JSON-LD presence check
@@ -174,9 +188,86 @@ export function validatePage(html: string, options: PageValidationOptions): Vali
 		warnings.push('Missing JSON-LD structured data')
 	}
 
+	if (
+		options.requireArticleSchema &&
+		isLikelyArticlePath(options.pagePath) &&
+		!hasArticleJsonLd(html)
+	) {
+		errors.push('Missing valid Article JSON-LD for article route')
+	}
+
+	if (options.requireMaxImagePreviewLarge && !hasMaxImagePreviewLarge(html)) {
+		errors.push('Missing robots max-image-preview:large directive')
+	}
+
 	return { errors, warnings }
 }
 
+function pushIssue(
+	asError: boolean | undefined,
+	errors: string[],
+	warnings: string[],
+	message: string,
+) {
+	if (asError) errors.push(message)
+	else warnings.push(message)
+}
+
+function isLikelyArticlePath(pagePath: string | undefined): boolean {
+	if (!pagePath) return false
+	return /\/(article|articles|news|story|stories)\//i.test(pagePath)
+}
+
+function hasArticleJsonLd(html: string): boolean {
+	for (const rawJson of extractJsonLdBodies(html)) {
+		try {
+			const parsed = JSON.parse(rawJson) as unknown
+			if (hasArticleType(parsed)) return true
+		} catch {}
+	}
+	return false
+}
+
+function extractJsonLdBodies(html: string): string[] {
+	const bodies: string[] = []
+	const pattern = /<script\s[^>]*type=["']application\/ld\+json["'][^>]*>([\s\S]*?)<\/script>/gi
+	for (const match of html.matchAll(pattern)) {
+		if (match[1]) bodies.push(match[1].trim())
+	}
+	return bodies
+}
+
+function hasArticleType(value: unknown): boolean {
+	if (Array.isArray(value)) return value.some(hasArticleType)
+	if (!isRecord(value)) return false
+
+	const type = value['@type']
+	if (type === 'Article' || type === 'NewsArticle' || type === 'BlogPosting') return true
+	if (
+		Array.isArray(type) &&
+		type.some((item) => item === 'Article' || item === 'NewsArticle' || item === 'BlogPosting')
+	) {
+		return true
+	}
+
+	return hasArticleType(value['@graph'])
+}
+
+function hasMaxImagePreviewLarge(html: string): boolean {
+	const metaTags = html.match(/<meta\s+[^>]*>/gi) ?? []
+	for (const tag of metaTags) {
+		if (getHtmlAttr(tag, 'name')?.toLowerCase() !== 'robots') continue
+		const content = getHtmlAttr(tag, 'content')?.toLowerCase() ?? ''
+		const directives = content.split(',').map((part) => part.trim())
+		if (directives.includes('max-image-preview:large')) return true
+	}
+	return false
+}
+
+function isRecord(value: unknown): value is Record<string, unknown> {
+	return typeof value === 'object' && value !== null
+}
+
 function isNoindexMetaRefreshRedirect(html: string): boolean {
 	const metaTags = html.match(/<meta\s+[^>]*>/gi) ?? []
 	const hasRefresh = metaTags.some(