Tesseract.js OCR Missing File Type Validation

package.json

@@ -57,6 +57,7 @@
     "drizzle-orm": "^0.45.2",
     "embla-carousel-react": "8.6.0",
     "file-saver": "2.0.5",
+    "file-type": "^22.0.1",
     "framer-motion": "^12.40.0",
     "google-tts-api": "^0.0.6",
     "groq-sdk": "1.1.1",

src/app/api/ask-ai/route.ts

@@ -6,7 +6,7 @@ import { db } from "@/configs/db";
 import { membershipsTable, usersTable, aiSessionsTable } from "@/configs/schema";
 import { enforceApiRateLimit } from "@/lib/ratelimit/api-rate-limit";
 import { aiLimiter } from "@/lib/ratelimit/ratelimit";
-import { AI_REQUEST_MAX_BYTES } from "@/lib/ai/ai-image-validation";
+import { AI_REQUEST_MAX_BYTES, validateAiImageDataUrl, type AiImageValidationResult } from "@/lib/ai/ai-image-validation";
 import { buildSystemMessages } from "@/lib/ai/socratic-prompt";
 import { buildErrorResponse } from "@/lib/errors/error-handler";
 import type { AIMode } from "@/types/ai-chat";
@@ -65,15 +65,12 @@ export async function POST(req: Request): Promise<NextResponse> {
         : "";
 
   if (body.imageBase64 !== undefined) {
-    const img = body.imageBase64 as string;
-    const validMime = /^data:image\/(png|jpe?g|webp);base64,/.test(img);
-    if (!validMime) {
+    const result = await validateAiImageDataUrl(body.imageBase64);
+    if (!result.ok) {
+      const err = result as Extract<AiImageValidationResult, { ok: false }>;
       return NextResponse.json(
-        {
-          error: "Please upload a valid PNG, JPG, or WEBP image.",
-          code: "INVALID_IMAGE_PAYLOAD",
-        },
-        { status: 422 }
+        { error: err.error, code: err.code },
+        { status: err.status }
       );
     }
   }

src/lib/ai/ai-image-validation.ts

@@ -1,3 +1,5 @@
+import { fileTypeFromBuffer } from 'file-type';
+
 export const AI_REQUEST_MAX_BYTES = 4 * 1024 * 1024;
 export const AI_IMAGE_MAX_BYTES = 3 * 1024 * 1024;
 
@@ -44,9 +46,42 @@ export function getBase64DecodedByteLength(base64Data: string) {
     return Math.floor((base64Data.length * 3) / 4) - padding;
 }
 
-export function validateAiImageDataUrl(
+/**
+ * Verify that the decoded base64 payload's magic bytes match the declared
+ * MIME type.  This prevents clients from spoofing the data URI prefix while
+ * sending a different (potentially malicious) binary format.
+ */
+async function validateMagicBytes(
+    base64Data: string,
+    declaredMime: string
+): Promise<AiImageValidationResult | null> {
+    const buffer = Buffer.from(base64Data, 'base64');
+    const detected = await fileTypeFromBuffer(buffer);
+
+    if (!detected) {
+        return {
+            ok: false,
+            status: 422,
+            code: 'INVALID_IMAGE_PAYLOAD',
+            error: 'Cannot verify image format. Please upload a valid PNG, JPG, or WEBP image.',
+        };
+    }
+
+    if (detected.mime !== declaredMime) {
+        return {
+            ok: false,
+            status: 422,
+            code: 'IMAGE_MIME_MISMATCH',
+            error: `Declared image type (${declaredMime}) does not match actual content (${detected.mime}).`,
+        };
+    }
+
+    return null;
+}
+
+export async function validateAiImageDataUrl(
     imageBase64: unknown
-): AiImageValidationResult {
+): Promise<AiImageValidationResult> {
     if (typeof imageBase64 !== "string") {
         return {
             ok: false,
@@ -108,6 +143,9 @@ export function validateAiImageDataUrl(
         };
     }
 
+    const magicError = await validateMagicBytes(base64Data, mimeType);
+    if (magicError) return magicError;
+
     return {
         ok: true,
         dataUrl: imageBase64,