@umituz/react-native-ai-pruna-provider 1.0.6 → 1.0.7

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@umituz/react-native-ai-pruna-provider",
-  "version": "1.0.6",
+  "version": "1.0.7",
   "description": "Pruna AI provider for React Native - implements IAIProvider interface for unified AI generation",
   "main": "./src/index.ts",
   "types": "./src/index.ts",

package/src/domain/entities/pruna.types.ts

@@ -80,6 +80,8 @@ export interface PrunaPredictionInput {
   readonly image?: string;
   readonly images?: readonly string[];
   readonly reference_image?: string;
+  /** Audio file for p-video (base64 or URL). When provided, duration is determined by audio length. */
+  readonly audio?: string;
   readonly duration?: number;
   readonly resolution?: PrunaResolution;
   readonly fps?: number;

package/src/infrastructure/services/pruna-api-client.ts

@@ -18,22 +18,58 @@ import { generationLogCollector } from "../utils/log-collector";
 const TAG = 'pruna-api';
 
 /**
- * Upload a base64 image to Pruna's file storage.
- * p-video requires a file URL (not raw base64).
+ * Detect MIME type from raw binary bytes.
+ * Supports image (png, jpeg, webp) and audio (mp3, wav, flac, m4a/aac) formats.
+ */
+function detectMimeType(bytes: Uint8Array): string {
+  // JPEG
+  if (bytes[0] === 0xFF && bytes[1] === 0xD8) return 'image/jpeg';
+  // PNG
+  if (bytes[0] === 0x89 && bytes[1] === 0x50) return 'image/png';
+  // RIFF container — WAV or WebP
+  if (bytes[0] === 0x52 && bytes[1] === 0x49 && bytes[2] === 0x46 && bytes[3] === 0x46) {
+    if (bytes.length > 11 && bytes[8] === 0x57 && bytes[9] === 0x41 && bytes[10] === 0x56 && bytes[11] === 0x45) return 'audio/wav';
+    return 'image/webp';
+  }
+  // MP3 with ID3 tag
+  if (bytes[0] === 0x49 && bytes[1] === 0x44 && bytes[2] === 0x33) return 'audio/mpeg';
+  // MP3 sync word (0xFF followed by 0xE0-0xFF)
+  if (bytes[0] === 0xFF && (bytes[1] & 0xE0) === 0xE0 && bytes[1] !== 0xFF) return 'audio/mpeg';
+  // FLAC
+  if (bytes[0] === 0x66 && bytes[1] === 0x4C && bytes[2] === 0x61 && bytes[3] === 0x43) return 'audio/flac';
+  // M4A/AAC (MP4 container — ftyp box at offset 4)
+  if (bytes.length > 7 && bytes[4] === 0x66 && bytes[5] === 0x74 && bytes[6] === 0x79 && bytes[7] === 0x70) return 'audio/mp4';
+  // Default
+  return 'image/png';
+}
+
+const MIME_EXTENSION_MAP: Record<string, string> = {
+  'image/png': 'png',
+  'image/jpeg': 'jpg',
+  'image/webp': 'webp',
+  'audio/mpeg': 'mp3',
+  'audio/wav': 'wav',
+  'audio/flac': 'flac',
+  'audio/mp4': 'm4a',
+};
+
+/**
+ * Upload a base64 file (image or audio) to Pruna's file storage.
+ * p-video requires file URLs (not raw base64).
  * Returns the HTTPS file URL to use in predictions.
  */
-export async function uploadImageToFiles(
+export async function uploadFileToStorage(
   base64Data: string,
   apiKey: string,
   sessionId: string,
 ): Promise<string> {
   // Already a URL — return as-is
   if (base64Data.startsWith('http')) {
-    generationLogCollector.log(sessionId, TAG, 'Image already a URL, skipping upload');
+    generationLogCollector.log(sessionId, TAG, 'File already a URL, skipping upload');
     return base64Data;
   }
 
-  generationLogCollector.log(sessionId, TAG, 'Uploading image to Pruna file storage...');
+  generationLogCollector.log(sessionId, TAG, 'Uploading file to Pruna storage...');
 
   // Strip data URI prefix if present
   const raw = base64Data.includes('base64,') ? base64Data.split('base64,')[1] : base64Data;
@@ -42,7 +78,7 @@ export async function uploadImageToFiles(
   try {
     binaryStr = atob(raw);
   } catch {
-    throw new Error("Invalid image format. Please provide base64 or a valid URL.");
+    throw new Error("Invalid file format. Please provide base64 or a valid URL.");
   }
 
   const bytes = new Uint8Array(binaryStr.length);
@@ -50,13 +86,9 @@ export async function uploadImageToFiles(
     bytes[i] = binaryStr.charCodeAt(i);
   }
 
-  // Detect MIME from first bytes
-  let mime = 'image/png';
-  if (bytes[0] === 0xFF && bytes[1] === 0xD8) mime = 'image/jpeg';
-  else if (bytes[0] === 0x52 && bytes[1] === 0x49) mime = 'image/webp';
-
+  const mime = detectMimeType(bytes);
+  const ext = MIME_EXTENSION_MAP[mime] || mime.split('/')[1];
   const blob = new Blob([bytes], { type: mime });
-  const ext = mime.split('/')[1];
   const formData = new FormData();
   formData.append('content', blob, `upload.${ext}`);
 
@@ -84,6 +116,9 @@ export async function uploadImageToFiles(
   return fileUrl;
 }
 
+/** @deprecated Use uploadFileToStorage instead */
+export const uploadImageToFiles = uploadFileToStorage;
+
 /**
  * Strip base64 data URI prefix, returning raw base64 string.
  * If input is already a URL, returns it unchanged.

package/src/infrastructure/services/pruna-input-builder.ts

@@ -5,12 +5,12 @@
  * Each Pruna model has strict schema requirements:
  *   p-image:      { prompt, aspect_ratio? }
  *   p-image-edit: { images: string[], prompt, aspect_ratio? }
- *   p-video:      { image: string (URL), prompt, duration, resolution, fps, draft, aspect_ratio, prompt_upsampling }
+ *   p-video:      { image: string (URL), prompt, duration, resolution, fps, draft, aspect_ratio, prompt_upsampling, audio? }
  */
 
 import type { PrunaModelId, PrunaAspectRatio, PrunaResolution } from "../../domain/entities/pruna.types";
 import { P_VIDEO_DEFAULTS, DEFAULT_ASPECT_RATIO } from "./pruna-provider.constants";
-import { uploadImageToFiles, stripBase64Prefix } from "./pruna-api-client";
+import { uploadFileToStorage, stripBase64Prefix } from "./pruna-api-client";
 import { generationLogCollector } from "../utils/log-collector";
 
 const TAG = 'pruna-input-builder';
@@ -115,7 +115,7 @@ async function buildVideoInput(
 
   // Upload base64 to file storage if needed (p-video requires HTTPS URL)
   generationLogCollector.log(sessionId, TAG, 'p-video: preparing image for video generation...');
-  const fileUrl = await uploadImageToFiles(rawImage, apiKey, sessionId);
+  const fileUrl = await uploadFileToStorage(rawImage, apiKey, sessionId);
 
   const duration = (input.duration as number) ?? P_VIDEO_DEFAULTS.duration;
   const resolution = (input.resolution as PrunaResolution) ?? P_VIDEO_DEFAULTS.resolution;
@@ -135,6 +135,15 @@ async function buildVideoInput(
     prompt_upsampling: promptUpsampling,
   };
 
+  // Handle audio input — upload to file storage if base64, pass URL if already remote
+  const rawAudio = input.audio as string | undefined;
+  if (rawAudio) {
+    generationLogCollector.log(sessionId, TAG, 'p-video: preparing audio for video generation...');
+    const audioUrl = await uploadFileToStorage(rawAudio, apiKey, sessionId);
+    payload.audio = audioUrl;
+    generationLogCollector.log(sessionId, TAG, 'p-video: audio attached — duration will be determined by audio length');
+  }
+
   if (input.disable_safety_checker !== undefined) payload.disable_safety_checker = input.disable_safety_checker;
 
   return payload;