agno 2.2.13__py3-none-any.whl → 2.4.3__py3-none-any.whl

agno/utils/print_response/agent.py

@@ -30,12 +30,11 @@ def print_response_stream(
     session_id: Optional[str] = None,
     session_state: Optional[Dict[str, Any]] = None,
     user_id: Optional[str] = None,
+    run_id: Optional[str] = None,
     audio: Optional[Sequence[Audio]] = None,
     images: Optional[Sequence[Image]] = None,
     videos: Optional[Sequence[Video]] = None,
     files: Optional[Sequence[File]] = None,
-    stream_events: bool = False,
-    stream_intermediate_steps: bool = False,
     knowledge_filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None,
     debug_mode: Optional[bool] = None,
     markdown: bool = False,
@@ -82,20 +81,17 @@ def print_response_stream(
 
         input_content = get_text_from_message(input)
 
-        # Consider both stream_events and stream_intermediate_steps (deprecated)
-        stream_events = stream_events or stream_intermediate_steps
-
         for response_event in agent.run(
             input=input,
             session_id=session_id,
             session_state=session_state,
             user_id=user_id,
+            run_id=run_id,
             audio=audio,
             images=images,
             videos=videos,
             files=files,
             stream=True,
-            stream_events=stream_events,
             knowledge_filters=knowledge_filters,
             debug_mode=debug_mode,
             add_history_to_context=add_history_to_context,
@@ -137,6 +133,11 @@ def print_response_stream(
                                 )
                             except Exception as e:
                                 log_warning(f"Failed to convert response to JSON: {e}")
+                        elif agent.output_schema is not None and isinstance(response_event.content, dict):
+                            try:
+                                response_content_batch = JSON(json.dumps(response_event.content), indent=2)  # type: ignore
+                            except Exception as e:
+                                log_warning(f"Failed to convert response to JSON: {e}")
                         else:
                             try:
                                 response_content_batch = JSON(json.dumps(response_event.content), indent=4)
@@ -144,6 +145,12 @@ def print_response_stream(
                                 log_warning(f"Failed to convert response to JSON: {e}")
                     if hasattr(response_event, "reasoning_content") and response_event.reasoning_content is not None:  # type: ignore
                         _response_reasoning_content += response_event.reasoning_content  # type: ignore
+
+                # Handle streaming reasoning content delta events
+                if response_event.event == RunEvent.reasoning_content_delta:  # type: ignore
+                    if hasattr(response_event, "reasoning_content") and response_event.reasoning_content is not None:  # type: ignore
+                        _response_reasoning_content += response_event.reasoning_content  # type: ignore
+
                 if hasattr(response_event, "reasoning_steps") and response_event.reasoning_steps is not None:  # type: ignore
                     reasoning_steps = response_event.reasoning_steps  # type: ignore
 
@@ -184,6 +191,7 @@ def print_response_stream(
                 show_reasoning=show_reasoning,
                 show_full_reasoning=show_full_reasoning,
                 accumulated_tool_calls=accumulated_tool_calls,
+                compression_manager=agent.compression_manager,
             )
             panels.extend(additional_panels)
             if panels:
@@ -209,6 +217,10 @@ def print_response_stream(
             live_log.update(Group(*panels))
             agent.session_summary_manager.summaries_updated = False
 
+        # Clear compression stats after final display
+        if agent.compression_manager is not None:
+            agent.compression_manager.stats.clear()
+
         response_timer.stop()
 
         # Final update to remove the "Thinking..." status
@@ -222,12 +234,11 @@ async def aprint_response_stream(
     session_id: Optional[str] = None,
     session_state: Optional[Dict[str, Any]] = None,
     user_id: Optional[str] = None,
+    run_id: Optional[str] = None,
     audio: Optional[Sequence[Audio]] = None,
     images: Optional[Sequence[Image]] = None,
     videos: Optional[Sequence[Video]] = None,
     files: Optional[Sequence[File]] = None,
-    stream_events: bool = False,
-    stream_intermediate_steps: bool = False,
     knowledge_filters: Optional[Union[Dict[str, Any], List[FilterExpr]]] = None,
     debug_mode: Optional[bool] = None,
     markdown: bool = False,
@@ -272,20 +283,17 @@ async def aprint_response_stream(
         if render:
             live_log.update(Group(*panels))
 
-        # Considering both stream_events and stream_intermediate_steps (deprecated)
-        stream_events = stream_events or stream_intermediate_steps
-
         result = agent.arun(
             input=input,
             session_id=session_id,
             session_state=session_state,
             user_id=user_id,
+            run_id=run_id,
             audio=audio,
             images=images,
             videos=videos,
             files=files,
             stream=True,
-            stream_events=stream_events,
             knowledge_filters=knowledge_filters,
             debug_mode=debug_mode,
             add_history_to_context=add_history_to_context,
@@ -327,6 +335,11 @@ async def aprint_response_stream(
                             response_content_batch = JSON(resp.content.model_dump_json(exclude_none=True), indent=2)  # type: ignore
                         except Exception as e:
                             log_warning(f"Failed to convert response to JSON: {e}")
+                    elif agent.output_schema is not None and isinstance(resp.content, dict):
+                        try:
+                            response_content_batch = JSON(json.dumps(resp.content), indent=2)  # type: ignore
+                        except Exception as e:
+                            log_warning(f"Failed to convert response to JSON: {e}")
                     else:
                         try:
                             response_content_batch = JSON(json.dumps(resp.content), indent=4)
@@ -335,6 +348,11 @@ async def aprint_response_stream(
                     if resp.reasoning_content is not None:  # type: ignore
                         _response_reasoning_content += resp.reasoning_content  # type: ignore
 
+                # Handle streaming reasoning content delta events
+                if resp.event == RunEvent.reasoning_content_delta:  # type: ignore
+                    if hasattr(resp, "reasoning_content") and resp.reasoning_content is not None:  # type: ignore
+                        _response_reasoning_content += resp.reasoning_content  # type: ignore
+
                 if hasattr(resp, "reasoning_steps") and resp.reasoning_steps is not None:  # type: ignore
                     reasoning_steps = resp.reasoning_steps  # type: ignore
 
@@ -377,6 +395,7 @@ async def aprint_response_stream(
                 show_reasoning=show_reasoning,
                 show_full_reasoning=show_full_reasoning,
                 accumulated_tool_calls=accumulated_tool_calls,
+                compression_manager=agent.compression_manager,
             )
             panels.extend(additional_panels)
             if panels:
@@ -402,6 +421,10 @@ async def aprint_response_stream(
             live_log.update(Group(*panels))
             agent.session_summary_manager.summaries_updated = False
 
+        # Clear compression stats after final display
+        if agent.compression_manager is not None:
+            agent.compression_manager.stats.clear()
+
         response_timer.stop()
 
         # Final update to remove the "Thinking..." status
@@ -418,6 +441,7 @@ def build_panels_stream(
     show_reasoning: bool = True,
     show_full_reasoning: bool = False,
     accumulated_tool_calls: Optional[List] = None,
+    compression_manager: Optional[Any] = None,
 ):
     panels = []
 
@@ -458,8 +482,18 @@ def build_panels_stream(
         for formatted_tool_call in formatted_tool_calls:
             tool_calls_content.append(f"• {formatted_tool_call}\n")
 
+        tool_calls_text = tool_calls_content.plain.rstrip()
+
+        # Add compression stats if available (don't clear - caller will clear after final display)
+        if compression_manager is not None and compression_manager.stats:
+            stats = compression_manager.stats
+            saved = stats.get("original_size", 0) - stats.get("compressed_size", 0)
+            orig = stats.get("original_size", 1)
+            if stats.get("tool_results_compressed", 0) > 0:
+                tool_calls_text += f"\n\ncompressed: {stats.get('tool_results_compressed', 0)} | Saved: {saved:,} chars ({saved / orig * 100:.0f}%)"
+
         tool_calls_panel = create_panel(
-            content=tool_calls_content.plain.rstrip(),
+            content=tool_calls_text,
             title="Tool Calls",
             border_style="yellow",
         )
@@ -480,11 +514,23 @@ def build_panels_stream(
         and response_event.citations is not None
         and response_event.citations.urls is not None
     ):
-        md_content = "\n".join(
+        md_lines = []
+
+        # Add search queries if present
+        if response_event.citations.search_queries:
+            md_lines.append("**Search Queries:**")
+            for query in response_event.citations.search_queries:
+                md_lines.append(f"- {query}")
+            md_lines.append("")  # Empty line before URLs
+
+        # Add URL citations
+        md_lines.extend(
             f"{i + 1}. [{citation.title or citation.url}]({citation.url})"
             for i, citation in enumerate(response_event.citations.urls)
             if citation.url  # Only include citations with valid URLs
         )
+
+        md_content = "\n".join(md_lines)
         if md_content:  # Only create panel if there are citations
             citations_panel = create_panel(
                 content=Markdown(md_content),
@@ -502,6 +548,7 @@ def print_response(
     session_id: Optional[str] = None,
     session_state: Optional[Dict[str, Any]] = None,
     user_id: Optional[str] = None,
+    run_id: Optional[str] = None,
     audio: Optional[Sequence[Audio]] = None,
     images: Optional[Sequence[Image]] = None,
     videos: Optional[Sequence[Video]] = None,
@@ -546,11 +593,13 @@ def print_response(
             session_id=session_id,
             session_state=session_state,
             user_id=user_id,
+            run_id=run_id,
             audio=audio,
             images=images,
             videos=videos,
             files=files,
             stream=False,
+            stream_events=True,
             knowledge_filters=knowledge_filters,
             debug_mode=debug_mode,
             add_history_to_context=add_history_to_context,
@@ -584,6 +633,7 @@ def print_response(
             show_full_reasoning=show_full_reasoning,
             tags_to_include_in_markdown=tags_to_include_in_markdown,
             markdown=markdown,
+            compression_manager=agent.compression_manager,
         )
         panels.extend(additional_panels)
 
@@ -618,6 +668,7 @@ async def aprint_response(
     session_id: Optional[str] = None,
     session_state: Optional[Dict[str, Any]] = None,
     user_id: Optional[str] = None,
+    run_id: Optional[str] = None,
     audio: Optional[Sequence[Audio]] = None,
     images: Optional[Sequence[Image]] = None,
     videos: Optional[Sequence[Video]] = None,
@@ -662,11 +713,13 @@ async def aprint_response(
             session_id=session_id,
             session_state=session_state,
             user_id=user_id,
+            run_id=run_id,
             audio=audio,
             images=images,
             videos=videos,
             files=files,
             stream=False,
+            stream_events=True,
             knowledge_filters=knowledge_filters,
             debug_mode=debug_mode,
             add_history_to_context=add_history_to_context,
@@ -700,6 +753,7 @@ async def aprint_response(
             show_full_reasoning=show_full_reasoning,
             tags_to_include_in_markdown=tags_to_include_in_markdown,
             markdown=markdown,
+            compression_manager=agent.compression_manager,
         )
         panels.extend(additional_panels)
 
@@ -736,6 +790,7 @@ def build_panels(
     show_full_reasoning: bool = False,
     tags_to_include_in_markdown: Optional[Set[str]] = None,
     markdown: bool = False,
+    compression_manager: Optional[Any] = None,
 ):
     panels = []
 
@@ -787,8 +842,19 @@ def build_panels(
         for formatted_tool_call in formatted_tool_calls:
             tool_calls_content.append(f"• {formatted_tool_call}\n")
 
+        tool_calls_text = tool_calls_content.plain.rstrip()
+
+        # Add compression stats if available
+        if compression_manager is not None and compression_manager.stats:
+            stats = compression_manager.stats
+            saved = stats.get("original_size", 0) - stats.get("compressed_size", 0)
+            orig = stats.get("original_size", 1)
+            if stats.get("tool_results_compressed", 0) > 0:
+                tool_calls_text += f"\n\ncompressed: {stats.get('tool_results_compressed', 0)} | Saved: {saved:,} chars ({saved / orig * 100:.0f}%)"
+            compression_manager.stats.clear()
+
         tool_calls_panel = create_panel(
-            content=tool_calls_content.plain.rstrip(),
+            content=tool_calls_text,
             title="Tool Calls",
             border_style="yellow",
         )
@@ -807,6 +873,11 @@ def build_panels(
                 response_content_batch = JSON(run_response.content.model_dump_json(exclude_none=True), indent=2)
             except Exception as e:
                 log_warning(f"Failed to convert response to JSON: {e}")
+        elif output_schema is not None and isinstance(run_response.content, dict):
+            try:
+                response_content_batch = JSON(json.dumps(run_response.content), indent=2)
+            except Exception as e:
+                log_warning(f"Failed to convert response to JSON: {e}")
         else:
             try:
                 response_content_batch = JSON(json.dumps(run_response.content), indent=4)
@@ -826,11 +897,23 @@ def build_panels(
         and run_response.citations is not None
         and run_response.citations.urls is not None
     ):
-        md_content = "\n".join(
+        md_lines = []
+
+        # Add search queries if present
+        if run_response.citations.search_queries:
+            md_lines.append("**Search Queries:**")
+            for query in run_response.citations.search_queries:
+                md_lines.append(f"- {query}")
+            md_lines.append("")  # Empty line before URLs
+
+        # Add URL citations
+        md_lines.extend(
             f"{i + 1}. [{citation.title or citation.url}]({citation.url})"
             for i, citation in enumerate(run_response.citations.urls)
             if citation.url  # Only include citations with valid URLs
         )
+
+        md_content = "\n".join(md_lines)
         if md_content:  # Only create panel if there are citations
             citations_panel = create_panel(
                 content=Markdown(md_content),