Update Gradio_UI.py

Gradio_UI.py

@@ -13,10 +13,12 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+
 import mimetypes
 import os
 import re
 import shutil
+import uuid
 from typing import Optional
 
 from smolagents.agent_types import AgentAudio, AgentImage, AgentText, handle_agent_output_types
@@ -25,36 +27,27 @@ from smolagents.memory import MemoryStep
 from smolagents.utils import _is_package_available
 
 
-def pull_messages_from_step(
-    step_log: MemoryStep,
-):
+def pull_messages_from_step(step_log: MemoryStep):
     """Extract ChatMessage objects from agent steps with proper nesting"""
     import gradio as gr
 
     if isinstance(step_log, ActionStep):
-        # Output the step number
         step_number = f"Step {step_log.step_number}" if step_log.step_number is not None else ""
         yield gr.ChatMessage(role="assistant", content=f"**{step_number}**")
 
-        # First yield the thought/reasoning from the LLM
         if hasattr(step_log, "model_output") and step_log.model_output is not None:
-            # Clean up the LLM output
             model_output = step_log.model_output.strip()
-            # Remove any trailing <end_code> and extra backticks, handling multiple possible formats
-            model_output = re.sub(r"```\s*<end_code>", "```", model_output)  # handles ```<end_code>
-            model_output = re.sub(r"<end_code>\s*```", "```", model_output)  # handles <end_code>```
-            model_output = re.sub(r"```\s*\n\s*<end_code>", "```", model_output)  # handles ```\n<end_code>
+            model_output = re.sub(r"```\s*<end_code>", "```", model_output)
+            model_output = re.sub(r"<end_code>\s*```", "```", model_output)
+            model_output = re.sub(r"```\s*\n\s*<end_code>", "```", model_output)
             model_output = model_output.strip()
             yield gr.ChatMessage(role="assistant", content=model_output)
 
-        # For tool calls, create a parent message
         if hasattr(step_log, "tool_calls") and step_log.tool_calls is not None:
             first_tool_call = step_log.tool_calls[0]
             used_code = first_tool_call.name == "python_interpreter"
             parent_id = f"call_{len(step_log.tool_calls)}"
 
-            # Tool call becomes the parent message with timing info
-            # First we will handle arguments based on type
             args = first_tool_call.arguments
             if isinstance(args, dict):
                 content = str(args.get("answer", str(args)))
@@ -62,9 +55,8 @@ def pull_messages_from_step(
                 content = str(args).strip()
 
             if used_code:
-                # Clean up the content by removing any end code tags
-                content = re.sub(r"```.*?\n", "", content)  # Remove existing code blocks
-                content = re.sub(r"\s*<end_code>\s*", "", content)  # Remove end_code tags
+                content = re.sub(r"```.*?\n", "", content)
+                content = re.sub(r"\s*<end_code>\s*", "", content)
                 content = content.strip()
                 if not content.startswith("```python"):
                     content = f"```python\n{content}\n```"
@@ -80,10 +72,9 @@ def pull_messages_from_step(
             )
             yield parent_message_tool
 
-            # Nesting execution logs under the tool call if they exist
             if hasattr(step_log, "observations") and (
                 step_log.observations is not None and step_log.observations.strip()
-            ):  # Only yield execution logs if there's actual content
+            ):
                 log_content = step_log.observations.strip()
                 if log_content:
                     log_content = re.sub(r"^Execution logs:\s*", "", log_content)
@@ -93,7 +84,6 @@ def pull_messages_from_step(
                         metadata={"title": "📝 Execution Logs", "parent_id": parent_id, "status": "done"},
                     )
 
-            # Nesting any errors under the tool call
             if hasattr(step_log, "error") and step_log.error is not None:
                 yield gr.ChatMessage(
                     role="assistant",
@@ -101,14 +91,11 @@ def pull_messages_from_step(
                     metadata={"title": "💥 Error", "parent_id": parent_id, "status": "done"},
                 )
 
-            # Update parent message metadata to done status without yielding a new message
             parent_message_tool.metadata["status"] = "done"
 
-        # Handle standalone errors but not from tool calls
         elif hasattr(step_log, "error") and step_log.error is not None:
             yield gr.ChatMessage(role="assistant", content=str(step_log.error), metadata={"title": "💥 Error"})
 
-        # Calculate duration and token information
         step_footnote = f"{step_number}"
         if hasattr(step_log, "input_token_count") and hasattr(step_log, "output_token_count"):
             token_str = (
@@ -118,17 +105,68 @@ def pull_messages_from_step(
         if hasattr(step_log, "duration"):
             step_duration = f" | Duration: {round(float(step_log.duration), 2)}" if step_log.duration else None
             step_footnote += step_duration
+
         step_footnote = f"""<span style="color: #bbbbc2; font-size: 12px;">{step_footnote}</span> """
         yield gr.ChatMessage(role="assistant", content=f"{step_footnote}")
         yield gr.ChatMessage(role="assistant", content="-----")
 
 
-def stream_to_gradio(
-    agent,
-    task: str,
-    reset_agent_memory: bool = False,
-    additional_args: Optional[dict] = None,
-):
+def _save_agent_image(final_answer: AgentImage) -> str:
+    """
+    Convert AgentImage into a real PNG path so Gradio can render it.
+    Works for:
+      - AgentImage.to_string() returning path
+      - PIL Images stored inside .value, .image, .data, .pil_image
+      - raw bytes stored inside .value/.data
+    """
+    os.makedirs("generated_images", exist_ok=True)
+
+    # If to_string() already points to a valid file, use it
+    img_str = final_answer.to_string()
+    if isinstance(img_str, str) and os.path.exists(img_str):
+        return img_str
+
+    # Otherwise save it ourselves
+    img_path = os.path.join("generated_images", f"image_{uuid.uuid4().hex[:8]}.png")
+
+    pil_img = None
+
+    # Try common fields
+    for attr in ["value", "image", "data", "pil_image"]:
+        if hasattr(final_answer, attr):
+            candidate = getattr(final_answer, attr)
+            if candidate is None:
+                continue
+            if hasattr(candidate, "save"):  # PIL Image
+                pil_img = candidate
+                break
+
+    # Try method
+    if pil_img is None and hasattr(final_answer, "to_pil"):
+        try:
+            pil_img = final_answer.to_pil()
+        except Exception:
+            pil_img = None
+
+    if pil_img is not None:
+        pil_img.save(img_path)
+        return img_path
+
+    # Fallback: raw bytes
+    for attr in ["value", "image", "data"]:
+        if hasattr(final_answer, attr):
+            candidate = getattr(final_answer, attr)
+            if isinstance(candidate, (bytes, bytearray)):
+                with open(img_path, "wb") as f:
+                    f.write(candidate)
+                return img_path
+
+    # If we couldn't save anything, return a placeholder path
+    # (Gradio will show broken image, but app won't crash)
+    return img_path
+
+
+def stream_to_gradio(agent, task: str, reset_agent_memory: bool = False, additional_args: Optional[dict] = None):
     """Runs an agent with the given task and streams the messages from the agent as gradio ChatMessages."""
     if not _is_package_available("gradio"):
         raise ModuleNotFoundError(
@@ -140,7 +178,6 @@ def stream_to_gradio(
     total_output_tokens = 0
 
     for step_log in agent.run(task, stream=True, reset=reset_agent_memory, additional_args=additional_args):
-        # Track tokens if model provides them
         if hasattr(agent.model, "last_input_token_count"):
             total_input_tokens += agent.model.last_input_token_count
             total_output_tokens += agent.model.last_output_token_count
@@ -148,44 +185,22 @@ def stream_to_gradio(
                 step_log.input_token_count = agent.model.last_input_token_count
                 step_log.output_token_count = agent.model.last_output_token_count
 
-        for message in pull_messages_from_step(
-            step_log,
-        ):
+        for message in pull_messages_from_step(step_log):
             yield message
 
-    final_answer = step_log  # Last log is the run's final_answer
+    final_answer = step_log
     final_answer = handle_agent_output_types(final_answer)
 
     if isinstance(final_answer, AgentText):
-        yield gr.ChatMessage(
-            role="assistant",
-            content=f"**Final answer:**\n{final_answer.to_string()}\n",
-        )
+        yield gr.ChatMessage(role="assistant", content=f"**Final answer:**\n{final_answer.to_string()}\n")
+
     elif isinstance(final_answer, AgentImage):
-        img_path = final_answer.to_string()
-    
-        # If the returned string isn't a valid file path, save the image ourselves
-        if not isinstance(img_path, str) or not os.path.exists(img_path):
-            os.makedirs("generated_images", exist_ok=True)
-            img_path = os.path.join("generated_images", "generated_image.png")
-    
-            # Try to save AgentImage.value (usually PIL image)
-            try:
-                final_answer.value.save(img_path)
-            except Exception:
-                # Fallback for raw bytes
-                with open(img_path, "wb") as f:
-                    f.write(final_answer.value)
-    
-        yield gr.ChatMessage(
-            role="assistant",
-            content={"path": img_path, "mime_type": "image/png"},
-        )
+        img_path = _save_agent_image(final_answer)
+        yield gr.ChatMessage(role="assistant", content={"path": img_path, "mime_type": "image/png"})
+
     elif isinstance(final_answer, AgentAudio):
-        yield gr.ChatMessage(
-            role="assistant",
-            content={"path": final_answer.to_string(), "mime_type": "audio/wav"},
-        )
+        yield gr.ChatMessage(role="assistant", content={"path": final_answer.to_string(), "mime_type": "audio/wav"})
+
     else:
         yield gr.ChatMessage(role="assistant", content=f"**Final answer:** {str(final_answer)}")
 
@@ -206,7 +221,6 @@ class GradioUI:
 
     def interact_with_agent(self, prompt, messages):
         import gradio as gr
-
         messages.append(gr.ChatMessage(role="user", content=prompt))
         yield messages
         for msg in stream_to_gradio(self.agent, task=prompt, reset_agent_memory=False):
@@ -224,9 +238,7 @@ class GradioUI:
             "text/plain",
         ],
     ):
-        """
-        Handle file uploads, default allowed types are .pdf, .docx, and .txt
-        """
+        """Handle file uploads, default allowed types are .pdf, .docx, and .txt"""
         import gradio as gr
 
         if file is None:
@@ -240,23 +252,18 @@ class GradioUI:
         if mime_type not in allowed_file_types:
             return gr.Textbox("File type disallowed", visible=True), file_uploads_log
 
-        # Sanitize file name
         original_name = os.path.basename(file.name)
-        sanitized_name = re.sub(
-            r"[^\w\-.]", "_", original_name
-        )  # Replace any non-alphanumeric, non-dash, or non-dot characters with underscores
+        sanitized_name = re.sub(r"[^\w\-.]", "_", original_name)
 
         type_to_ext = {}
         for ext, t in mimetypes.types_map.items():
             if t not in type_to_ext:
                 type_to_ext[t] = ext
 
-        # Ensure the extension correlates to the mime type
         sanitized_name = sanitized_name.split(".")[:-1]
         sanitized_name.append("" + type_to_ext[mime_type])
         sanitized_name = "".join(sanitized_name)
 
-        # Save the uploaded file to the specified folder
         file_path = os.path.join(self.file_upload_folder, os.path.basename(sanitized_name))
         shutil.copy(file.name, file_path)
 
@@ -275,7 +282,7 @@ class GradioUI:
 
     def launch(self, **kwargs):
         import gradio as gr
-    
+
         with gr.Blocks(fill_height=True) as demo:
             stored_messages = gr.State([])
             file_uploads_log = gr.State([])
@@ -289,7 +296,7 @@ class GradioUI:
                 resizable=True,
                 scale=1,
             )
-    
+
             if self.file_upload_folder is not None:
                 upload_file = gr.File(label="Upload a file")
                 upload_status = gr.Textbox(label="Upload Status", interactive=False, visible=False)
@@ -298,18 +305,17 @@ class GradioUI:
                     [upload_file, file_uploads_log],
                     [upload_status, file_uploads_log],
                 )
-    
+
             text_input = gr.Textbox(lines=1, label="Chat Message")
             text_input.submit(
                 self.log_user_message,
                 [text_input, file_uploads_log],
                 [stored_messages, text_input],
             ).then(self.interact_with_agent, [stored_messages, chatbot], [chatbot])
-    
+
         # Disable share on Spaces automatically
         is_spaces = os.environ.get("SPACE_ID") is not None
         demo.launch(debug=True, share=not is_spaces, **kwargs)
 
 
-
-__all__ = ["stream_to_gradio", "GradioUI"]
+__all__ = ["stream_to_gradio", "GradioUI"]