@juspay/neurolink 10.8.6 → 10.8.7

package/dist/utils/messageBuilder.js

@@ -777,6 +777,38 @@ function appendDetectedFileResult(result, file, options) {
         logger.info(`[FileDetector] ⚠️ Unknown format (metadata extracted): ${filename}`);
     }
 }
+/**
+ * Fold the `audioFiles` / `videoFiles` aliases into the unified `files` array.
+ *
+ * #284 gave audio and video their own input fields, but neither has a
+ * dedicated processor — both are meant to travel through the same
+ * auto-detecting `files` pipeline that already understands "audio"/"video"
+ * FileDetector results (see `appendDetectedFileResult`). The fold used to live
+ * inline in `buildMultimodalMessagesArray`, which meant any path that bypassed
+ * that builder never performed it and dropped the files silently: the model
+ * received the prompt alone and answered as though nothing were attached
+ * (#1259). GoogleVertex's native SDK path and `buildMultimodalOptions`
+ * (Bedrock) are both such paths, so the fold has to be callable from them.
+ *
+ * The aliases are cleared once merged, which makes the call idempotent: a
+ * provider override and the shared builder can both call this on the same
+ * options object without attaching every file twice.
+ *
+ * Mutates in place, matching `processUnifiedFilesArray` below — downstream
+ * stages all read `options.input.files`.
+ */
+export function mergeMediaFileAliases(input) {
+    if (!input.audioFiles?.length && !input.videoFiles?.length) {
+        return;
+    }
+    input.files = [
+        ...(input.files ?? []),
+        ...(input.audioFiles ?? []),
+        ...(input.videoFiles ?? []),
+    ];
+    input.audioFiles = undefined;
+    input.videoFiles = undefined;
+}
 /**
  * Process the unified files array with auto-detection.
  * Handles lazy file registration, full processing, and preview injection.
@@ -1115,18 +1147,7 @@ export async function buildMultimodalMessagesArray(options, provider, model) {
     // local const so TypeScript sees the definite (non-optional) type in the
     // rest of this function, avoiding 60+ "possibly undefined" errors.
     const inp = options.input;
-    // #284: audioFiles/videoFiles have no dedicated processor — route them
-    // through the same auto-detecting `files` pipeline that already
-    // understands "audio"/"video" FileDetector results (see
-    // appendDetectedFileResult), instead of silently dropping them once
-    // detectMultimodal() routes an audio/video-only request here.
-    if (inp.audioFiles?.length || inp.videoFiles?.length) {
-        inp.files = [
-            ...(inp.files || []),
-            ...(inp.audioFiles || []),
-            ...(inp.videoFiles || []),
-        ];
-    }
+    mergeMediaFileAliases(inp);
     // Compute provider-specific max PDF size once for consistent validation
     const pdfConfig = PDFProcessor.getProviderConfig(provider);
     const maxSize = pdfConfig

package/dist/utils/multimodalOptionsBuilder.d.ts

@@ -12,6 +12,8 @@ import type { StreamOptions } from "../types/index.js";
  *   - input.files: Auto-detected file types
  *   - input.csvFiles: CSV files for tabular data
  *   - input.pdfFiles: PDF documents (Buffer | string paths)
+ *   - input.audioFiles: Audio files (Buffer | string paths)
+ *   - input.videoFiles: Video files (Buffer | string paths)
  *   - csvOptions: CSV parsing options
  *   - systemPrompt: System-level instructions
  *   - conversationMessages: Chat history
@@ -23,7 +25,7 @@ import type { StreamOptions } from "../types/index.js";
  * @param {string} providerName - Provider identifier (e.g., "vertex", "openai", "anthropic")
  * @param {string} modelName - Model identifier (e.g., "gemini-2.5-flash", "gpt-4o")
  * @returns {object} Normalized options object with:
- *   - input: { text, images, content, files, csvFiles, pdfFiles }
+ *   - input: { text, images, content, files, csvFiles, pdfFiles, audioFiles, videoFiles }
  *   - csvOptions: CSV processing options
  *   - systemPrompt: System prompt string
  *   - conversationHistory: Message history array
@@ -49,6 +51,8 @@ export declare function buildMultimodalOptions(options: StreamOptions, providerN
         files: (string | Buffer<ArrayBufferLike> | import("../index.js").FileWithMetadata)[] | undefined;
         csvFiles: (string | Buffer<ArrayBufferLike>)[] | undefined;
         pdfFiles: (string | Buffer<ArrayBufferLike>)[] | undefined;
+        audioFiles: (string | Buffer<ArrayBufferLike>)[] | undefined;
+        videoFiles: (string | Buffer<ArrayBufferLike>)[] | undefined;
     };
     csvOptions: import("../index.js").CSVProcessorOptions | undefined;
     pdfOptions: {

package/dist/utils/multimodalOptionsBuilder.js

@@ -11,6 +11,8 @@
  *   - input.files: Auto-detected file types
  *   - input.csvFiles: CSV files for tabular data
  *   - input.pdfFiles: PDF documents (Buffer | string paths)
+ *   - input.audioFiles: Audio files (Buffer | string paths)
+ *   - input.videoFiles: Video files (Buffer | string paths)
  *   - csvOptions: CSV parsing options
  *   - systemPrompt: System-level instructions
  *   - conversationMessages: Chat history
@@ -22,7 +24,7 @@
  * @param {string} providerName - Provider identifier (e.g., "vertex", "openai", "anthropic")
  * @param {string} modelName - Model identifier (e.g., "gemini-2.5-flash", "gpt-4o")
  * @returns {object} Normalized options object with:
- *   - input: { text, images, content, files, csvFiles, pdfFiles }
+ *   - input: { text, images, content, files, csvFiles, pdfFiles, audioFiles, videoFiles }
  *   - csvOptions: CSV processing options
  *   - systemPrompt: System prompt string
  *   - conversationHistory: Message history array
@@ -49,6 +51,11 @@ export function buildMultimodalOptions(options, providerName, modelName) {
             files: options.input?.files,
             csvFiles: options.input?.csvFiles,
             pdfFiles: options.input?.pdfFiles,
+            // #1259: this is a whitelist — a field omitted here is dropped
+            // silently, and the model answers as though nothing were attached.
+            // audioFiles/videoFiles were missing, so Bedrock received neither.
+            audioFiles: options.input?.audioFiles,
+            videoFiles: options.input?.videoFiles,
         },
         csvOptions: options.csvOptions,
         pdfOptions: options.pdfOptions,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@juspay/neurolink",
-  "version": "10.8.6",
+  "version": "10.8.7",
   "packageManager": "pnpm@10.15.1",
   "description": "TypeScript AI SDK with 24+ LLM providers behind one consistent API. MCP-native (connect any MCP server), voice TTS/STT/realtime, RAG, agents, memory, context compaction. OpenAI · Anthropic · Gemini · Bedrock · Azure · Ollama · DeepSeek · NVIDIA NIM and more.",
   "author": {