test: add images to link previews

apps/sim/app/api/link-preview/route.ts

@@ -21,16 +21,17 @@ const MAX_RESPONSE_BYTES = 256 * 1024
 const MAX_REDIRECTS = 3
 const TITLE_MAX_CHARS = 200
 const DESCRIPTION_MAX_CHARS = 300
+const IMAGE_URL_MAX_CHARS = 2048
 const CACHE_TTL_SECONDS = 24 * 60 * 60
 const NEGATIVE_CACHE_TTL_SECONDS = 60 * 60
-const CACHE_KEY_PREFIX = 'link-preview:v1:'
+const CACHE_KEY_PREFIX = 'link-preview:v2:'
 
 /**
  * Parses preview metadata from the fetched document (already capped at
  * MAX_RESPONSE_BYTES); cheerio handles attribute order, quoting, and entity
  * decoding.
  */
-function parsePreview(html: string): LinkPreview {
+function parsePreview(html: string, baseUrl: string): LinkPreview {
   const $ = cheerio.load(html)
 
   const meta = (key: string): string | null => {
@@ -42,12 +43,35 @@ function parsePreview(html: string): LinkPreview {
     meta('og:title') ?? meta('twitter:title') ?? ($('title').first().text().trim() || null)
   const description = meta('og:description') ?? meta('twitter:description') ?? meta('description')
   const siteName = meta('og:site_name')
+  const rawImageUrl = meta('og:image') ?? meta('twitter:image')
 
-  if (!title && !description && !siteName) return null
+  let imageUrl: string | null = null
+  if (rawImageUrl) {
+    try {
+      const resolvedUrl = new URL(rawImageUrl, baseUrl)
+      if (resolvedUrl.href.length <= IMAGE_URL_MAX_CHARS) {
+        imageUrl = resolvedUrl.href
+      } else {
+        logger.warn('Image URL exceeds maximum length; dropping', {
+          rawImageUrl: truncate(rawImageUrl, 100),
+          resolvedLength: resolvedUrl.href.length,
+          maxLength: IMAGE_URL_MAX_CHARS,
+        })
+      }
+    } catch (error) {
+      logger.warn('Failed to resolve image URL', {
+        rawImageUrl: truncate(rawImageUrl, 100),
+        error: getErrorMessage(error),
+      })
+    }
+  }
+
+  if (!title && !description && !siteName && !imageUrl) return null
   return {
     title: title ? truncate(title, TITLE_MAX_CHARS) : null,
     description: description ? truncate(description, DESCRIPTION_MAX_CHARS) : null,
     siteName: siteName ? truncate(siteName, TITLE_MAX_CHARS) : null,
+    imageUrl,
   }
 }
 
@@ -66,7 +90,7 @@ async function fetchPreview(url: string): Promise<LinkPreview> {
   if (!contentType.includes('text/html') && !contentType.includes('application/xhtml+xml')) {
     return null
   }
-  return parsePreview(await response.text())
+  return parsePreview(await response.text(), response.url)
 }
 
 export const GET = withRouteHandler(async (request: NextRequest) => {

apps/sim/app/api/tools/agiloft/attach/route.test.ts

@@ -55,6 +55,7 @@ function mockSecureFetchResponse(body: {
     statusText: '',
     headers: new Headers(),
     body: null,
+    url: 'https://example.com',
     text: async () => body.text ?? '',
     json: async () => body.json ?? {},
     arrayBuffer: async () => new ArrayBuffer(0),

apps/sim/app/api/tools/agiloft/retrieve/route.test.ts

@@ -40,6 +40,7 @@ function mockSecureFetchResponse(body: {
     statusText: '',
     headers: body.headers ?? new Headers(),
     body: null,
+    url: 'https://example.com',
     text: async () => body.text ?? '',
     json: async () => body.json ?? {},
     arrayBuffer: async () => body.arrayBuffer ?? new ArrayBuffer(0),

apps/sim/app/api/tools/stt/route.test.ts

@@ -51,6 +51,7 @@ function mockSecureFetchResponse(body: { ok?: boolean; contentType?: string }) {
     statusText: '',
     headers: new Headers({ 'content-type': body.contentType ?? 'audio/mpeg' }),
     body: null,
+    url: 'https://example.com',
     text: async () => '',
     json: async () => ({}),
     arrayBuffer: async () => new ArrayBuffer(8),

apps/sim/lib/api/contracts/link-preview.ts

@@ -16,6 +16,7 @@ export const linkPreviewResponseSchema = z.object({
       title: z.string().nullable(),
       description: z.string().nullable(),
       siteName: z.string().nullable(),
+      imageUrl: z.string().nullable(),
     })
     .nullable(),
 })

apps/sim/lib/core/security/input-validation.server.ts

@@ -462,6 +462,7 @@ export interface SecureFetchResponse {
   statusText: string
   headers: SecureFetchHeaders
   body: ReadableStream<Uint8Array> | null
+  url: string
   text: () => Promise<string>
   json: () => Promise<unknown>
   arrayBuffer: () => Promise<ArrayBuffer>
@@ -1170,6 +1171,7 @@ export async function secureFetchWithPinnedIP(
               statusText: res.statusMessage || '',
               headers: new SecureFetchHeaders(headersRecord, setCookieArray),
               body: null,
+              url,
               text: async () => '',
               json: async () => ({}),
               arrayBuffer: async () => new ArrayBuffer(0),
@@ -1249,6 +1251,7 @@ export async function secureFetchWithPinnedIP(
         statusText: res.statusMessage || '',
         headers: new SecureFetchHeaders(headersRecord, setCookieArray),
         body,
+        url,
         text: async () => (await readBodyAsBuffer()).toString('utf-8'),
         json: async () => JSON.parse((await readBodyAsBuffer()).toString('utf-8')),
         arrayBuffer: async () => {

apps/sim/lib/knowledge/documents/utils.test.ts

@@ -26,6 +26,7 @@ function fakeResponse(
     statusText: `status-${status}`,
     headers: { get: (name: string) => headers[name.toLowerCase()] ?? null },
     body: null,
+    url: 'https://example.com',
     text: async () => options.body ?? '',
     json: async () => JSON.parse(options.body ?? '{}'),
     arrayBuffer: async () => new ArrayBuffer(0),