Upload 2 files (#2)

- Upload 2 files (245fe4f92bf8ccb1f4e8e522370d407469bbec99)


Co-authored-by: John Smith <John6666@users.noreply.huggingface.co>

app.py

@@ -1,161 +1,172 @@
+import os
+import mimetypes
+import tempfile
+import urllib.parse
+import urllib.request
+
 import gradio as gr
-import google.generativeai as genai
-from typing import Optional, Tuple
-
-# Configuration
-MODEL_ID = "gemini-2.0-flash-exp-image-generation"
-
-def process_media(
-    api_key: str,
-    prompt: str,
-    img1: Optional[str],
-    img2: Optional[str],
-    img3: Optional[str]
-) -> Tuple[Optional[str], str]:
-    """
-    Handles both Text-to-Image generation and Image Editing/Understanding
-    using Google Gemini 2.0 Flash Exp.
-    """
+from PIL import Image  # kept for compatibility / future UI use
+
+from google import genai
+from google.genai import types
+
+# Config
+MODEL_ID = os.environ.get("GEMINI_MODEL", "gemini-2.5-flash-image")
+
+
+def _load_bytes(path_or_url: str) -> tuple[bytes, str]:
+    """Return (data, mime_type) from either a local path or a http(s) URL."""
+    parsed = urllib.parse.urlparse(path_or_url)
+    is_url = parsed.scheme in ("http", "https")
+
+    if is_url:
+        req = urllib.request.Request(path_or_url, headers={"User-Agent": "Mozilla/5.0"})
+        with urllib.request.urlopen(req, timeout=30) as resp:
+            data = resp.read()
+            mime_type = resp.headers.get_content_type() or "application/octet-stream"
+        if mime_type in ("application/octet-stream", "binary/octet-stream"):
+            guess, _ = mimetypes.guess_type(path_or_url)
+            if guess:
+                mime_type = guess
+        return data, mime_type
+
+    with open(path_or_url, "rb") as f:
+        data = f.read()
+    mime_type, _ = mimetypes.guess_type(path_or_url)
+    return data, (mime_type or "application/octet-stream")
+
+
+def generate_content(prompt: str, image1: str | None, image2: str | None, image3: str | None, api_key: str) -> tuple[str | None, str | None]:
+    """Generate content using the current Google GenAI SDK (google-genai)."""
+    api_key = (api_key or "").strip() or os.environ.get("GEMINI_API_KEY")
     if not api_key:
-        return None, "⚠️ Please enter your Google API Key to start."
-    
-    if not prompt:
-        return None, "⚠️ Please enter a prompt."
+        raise gr.Error("Please set Space secret GEMINI_API_KEY (recommended) or paste your key into the UI field.")
 
     try:
-        genai.configure(api_key=api_key)
-        model = genai.GenerativeModel(MODEL_ID)
-
-        # Prepare inputs: Prompt first, then any uploaded images
-        inputs = [prompt]
-        
-        uploaded_images = [img for img in [img1, img2, img3] if img is not None]
-        
-        # Append images to the input list
-        for img in uploaded_images:
-            inputs.append(img)
-
-        # Call the model
-        response = model.generate_content(inputs)
-        
-        # Resolve the response (for streaming compatibility, though we use simple generate here)
-        response.resolve()
-
-        # Extract the result
-        # The model may return an image (inline_data) or text (suggestions/analysis)
-        result_image = None
-        result_text = ""
-
-        for candidate in response.candidates:
-            for part in candidate.content.parts:
-                if part.inline_data:
-                    # Return image data (base64 encoded by Gradio automatically)
-                    result_image = part.inline_data.data
-                elif part.text:
-                    result_text += part.text
-
-        # If we got an image, return it with any accompanying text
-        if result_image:
-            return result_image, result_text if result_text else "Image generated/edited successfully."
-        
-        # If no image, return the text (likely understanding/suggestions)
-        if result_text:
-            return None, result_text
-            
-        return None, "No content generated. Try rephrasing your prompt."
+        client = genai.Client(api_key=api_key)
+
+        # Prepare request contents: images first, then instruction text
+        contents: list[types.Part | str] = []
+        for img_path in [image1, image2, image3]:
+            if img_path:
+                data, mime_type = _load_bytes(img_path)
+                contents.append(types.Part.from_bytes(data=data, mime_type=mime_type))
+
+        contents.append((prompt or "").strip() or "Describe the provided images.")
+
+        # Ask for TEXT+IMAGE when supported; retry as TEXT-only if needed.
+        try:
+            response = client.models.generate_content(
+                model=MODEL_ID,
+                contents=contents,
+                config=types.GenerateContentConfig(response_modalities=["TEXT", "IMAGE"]),
+            )
+        except Exception:
+            response = client.models.generate_content(model=MODEL_ID, contents=contents)
+
+        image_output: str | None = None
+        text_chunks: list[str] = []
+
+        if getattr(response, "candidates", None):
+            for part in response.candidates[0].content.parts:
+                if getattr(part, "text", None):
+                    text_chunks.append(part.text)
+
+                inline = getattr(part, "inline_data", None)
+                if inline and getattr(inline, "data", None) and image_output is None:
+                    filename = os.path.join(tempfile.gettempdir(), f"gemini_output_{os.urandom(4).hex()}.png")
+                    with open(filename, "wb") as f:
+                        f.write(inline.data)
+                    image_output = filename
+
+        if not text_chunks and getattr(response, "text", None):
+            text_chunks.append(response.text)
+
+        text_output = "\n".join([t for t in text_chunks if t]).strip() or None
+
+        if image_output:
+            return image_output, text_output or "Image generated successfully!"
+        if text_output:
+            return None, text_output
+        return None, "No content generated. Try a different prompt or model."
 
     except Exception as e:
-        error_msg = f"Error: {str(e)}"
-        if "API key" in str(e).lower():
-            error_msg = "Invalid API Key. Please check your Google AI Studio key."
-        return None, error_msg
+        raise gr.Error(f"Error generating content: {type(e).__name__}: {str(e)}")
+
 
-# Gradio 6 Application
+# Gradio app
 with gr.Blocks() as demo:
-    
-    # Header Section
-    gr.Markdown(
-        """
-        # 🎨 Gemini 2.0 Flash Studio
-        **Image Generation & Editing powered by Google DeepMind**
-        
-        Drag and drop up to 3 images to edit or understand them, or just use text to generate new art.
-        """
-    )
-    
-    gr.HTML(
-        '<div style="text-align: center; margin-bottom: 20px;">'
-        '<a href="https://huggingface.co/spaces/akhaliq/anycoder" target="_blank" style="color: #6366f1; font-weight: bold; text-decoration: none;">Built with anycoder</a>'
-        '</div>'
+    gr.HTML('''        <div style="text-align: center; margin-bottom: 20px;">
+      <h1>✨ Gemini Flash Studio</h1>
+      <p>Image Generation & Editing powered by the Gemini API</p>
+      <p style="opacity:0.8;">Model: <code>{model}</code> (override with <code>GEMINI_MODEL</code>)</p>
+      <a href="https://huggingface.co/spaces/akhaliq/anycoder" target="_blank"
+         style="color: #007bff; text-decoration: none; font-weight: bold;">Built with anycoder</a>
+    </div>
+    '''.format(model=MODEL_ID))
+
+    gr.Markdown("### Configuration")
+    with gr.Row():
+        api_key_input = gr.Textbox(
+            label="Gemini API Key (optional if GEMINI_API_KEY secret is set)",
+            type="password",
+            placeholder="If you set GEMINI_API_KEY in Space Secrets, you can leave this empty.",
+            scale=4,
+        )
+
+    gr.Markdown("### Inputs")
+    with gr.Row():
+        img_input_1 = gr.Image(label="Image 1 (Drag & Drop)", type="filepath", sources=["upload", "clipboard"], height=300)
+        img_input_2 = gr.Image(label="Image 2 (Drag & Drop)", type="filepath", sources=["upload", "clipboard"], height=300)
+        img_input_3 = gr.Image(label="Image 3 (Drag & Drop)", type="filepath", sources=["upload", "clipboard"], height=300)
+
+    prompt_input = gr.Textbox(
+        label="✍️ Prompt",
+        placeholder="Describe the image you want to generate or how you want to edit the uploaded images...",
+        lines=3,
+        show_copy_button=True,
     )
 
+    generate_btn = gr.Button("✨ Generate", variant="primary", size="lg")
+
+    gr.Markdown("### Output")
     with gr.Row():
-        with gr.Column(scale=1):
-            gr.Markdown("### 🔑 Configuration")
-            api_key_input = gr.Textbox(
-                label="Google API Key",
-                type="password",
-                placeholder="Get your key at ai.google.dev",
-                info="Your key is stored locally and not sent to any server other than Google's API."
-            )
-            
-            gr.Markdown("### 🖼️ Inputs (Optional)")
-            gr.Markdown("Drag images here to edit or analyze:")
-            with gr.Row():
-                img_input_1 = gr.Image(type="pil", label="Image 1")
-                img_input_2 = gr.Image(type="pil", label="Image 2")
-                img_input_3 = gr.Image(type="pil", label="Image 3")
-            
-            prompt_input = gr.Textbox(
-                label="✨ Prompt",
-                placeholder="Describe what you want to generate or how you want to edit the images...",
-                lines=3,
-                max_lines=5
-            )
-            
-            submit_btn = gr.Button("Generate / Edit", variant="primary", size="lg")
-
-        with gr.Column(scale=1):
-            gr.Markdown("### 🚀 Output")
-            with gr.Tabs():
-                with gr.TabItem("Generated Image"):
-                    output_image = gr.Image(label="Result", type="pil")
-                with gr.TabItem("Analysis & Suggestions"):
-                    output_text = gr.Textbox(
-                        label="Details / Suggestions",
-                        lines=15,
-                        placeholder="AI suggestions and analysis will appear here...",
-                        interactive=False
-                    )
-
-    # Footer / Instructions
-    gr.Markdown(
-        """
-        ---
-        ### 💡 Tips
-        1. **Text-to-Image:** Enter only a prompt to create a new image.
-        2. **Image Editing:** Upload images and describe the changes (e.g., "Change the background to a cyberpunk city").
-        3. **Understanding:** Upload an image and ask questions about it.
-        """
+        with gr.Column():
+            output_image = gr.Image(label="Generated Image", type="filepath", height=400)
+        with gr.Column():
+            output_text = gr.Textbox(label="Response / Description", lines=10)
+
+    gr.Examples(
+        examples=[
+            ["Generate a futuristic cyberpunk city with neon lights", None, None, None],
+            ["Describe what is in this image", "https://gradio-builds.s3.amazonaws.com/assets/cheetah-003.jpg", None, None],
+            ["Edit this image: make it look like a painting", None, None, None],
+        ],
+        inputs=[prompt_input, img_input_1, img_input_2, img_input_3],
     )
 
-    # Event Listener
-    submit_btn.click(
-        fn=process_media,
-        inputs=[api_key_input, prompt_input, img_input_1, img_input_2, img_input_3],
+    generate_btn.click(
+        fn=generate_content,
+        inputs=[prompt_input, img_input_1, img_input_2, img_input_3, api_key_input],
         outputs=[output_image, output_text],
-        api_visibility="public"
+        api_visibility="public",
     )
 
-# Launch with Gradio 6 specific parameters
-demo.launch(
-    theme=gr.themes.Soft(
-        primary_hue="indigo",
-        secondary_hue="purple",
-        font=gr.themes.GoogleFont("Inter"),
-        text_size="lg",
-        spacing_size="lg",
-        radius_size="md"
-    ),
-    footer_links=[{"label": "Built with anycoder", "url": "https://huggingface.co/spaces/akhaliq/anycoder"}]
-)
+    demo.launch(
+        theme=gr.themes.Soft(
+            primary_hue="indigo",
+            secondary_hue="cyan",
+            neutral_hue="slate",
+            font=gr.themes.GoogleFont("Inter"),
+            text_size="lg",
+            spacing_size="lg",
+            radius_size="md",
+        ).set(
+            button_primary_background_fill="*primary_500",
+            button_primary_background_fill_hover="*primary_600",
+            button_primary_text_color="white",
+            block_border_width="0px",
+            block_shadow="*shadow_drop_lg",
+        ),
+        footer_links=[{"label": "Built with anycoder", "url": "https://huggingface.co/spaces/akhaliq/anycoder"}],
+    )

requirements.txt

@@ -1,2 +1,3 @@
-gradio>=6.0.2
+Pillow
 google-genai
+gradio>=6.0.2