S3 upload output

app.py

@@ -1,9 +1,13 @@
+import uuid
 import gradio as gr
 import boto3
 import os
+from PIL import Image
 
 hf_token = os.environ.get("HF_TOKEN")
 tasomi_model = os.environ.get("TASOMI_MODEL")
+s3_bucket = os.environ.get("IMCAPSULE_EXTERNAL_BUCKET")
+
 m=gr.load(tasomi_model, src="models",hf_token=hf_token)
 
 def detect_language(text):
@@ -15,6 +19,15 @@ def detect_language(text):
     detected_language = response['Languages'][0]['LanguageCode']
     return detected_language
 
+def upload_file(file_path):
+    s3 = boto3.client('s3')
+    key='public/' + os.path.basename(file_path)
+    # Upload the file
+    s3.upload_file(file_path, s3_bucket, key)
+    s3_path = f"{s3.meta.endpoint_url}/{s3_bucket}/{key}"
+    print(f'File uploaded successfully. s3_path= {s3_path} ')
+    return s3_path
+
 def translate_and_predict(text):
     """Translates text and generates an image using Stable Diffusion."""
     source_language_code=detect_language(text)
@@ -25,15 +38,19 @@ def translate_and_predict(text):
     translation_result = translate.translate_text(Text=text, SourceLanguageCode=source_language_code, TargetLanguageCode=target_language_code)
     translated_text = translation_result.get('TranslatedText')
     print("translated_text:" , translated_text)
-    image_path=m(translated_text)
-    return image_path
+    gen_image=m(translated_text)
+    webp_image = Image.open(gen_image)
+    converted_image_path = f"./tasomi-{uuid.uuid4().hex}.jpg"
+    webp_image.save(converted_image_path, "JPEG")
+    s3_path = upload_file(converted_image_path)
+    return s3_path,s3_path
 
 # Define the Gradio interface
 iface = gr.Interface(
     fn=translate_and_predict,  # Since we're only displaying an image, there's no function to call
     inputs=["text"],  # Input: text, source language, target language
-    outputs="image",  # Output: generated image
-    title="Image Viewer",
+    outputs=["image","text"],  # Output: generated image
+    title="Tasomi Images",
 )
 
 # Launch the interface