agno 1.8.0__py3-none-any.whl → 2.0.0a1__py3-none-any.whl

agno/workflow/types.py

@@ -0,0 +1,482 @@
+from dataclasses import dataclass
+from enum import Enum
+from typing import Any, Dict, List, Optional, Union
+
+from fastapi import WebSocket
+from pydantic import BaseModel
+
+from agno.media import AudioArtifact, File, ImageArtifact, VideoArtifact
+from agno.models.metrics import Metrics
+from agno.utils.log import log_warning
+
+
+@dataclass
+class WorkflowExecutionInput:
+    """Input data for a step execution"""
+
+    input: Optional[Union[str, Dict[str, Any], List[Any], BaseModel]] = None
+
+    additional_data: Optional[Dict[str, Any]] = None
+
+    # Media inputs
+    images: Optional[List[ImageArtifact]] = None
+    videos: Optional[List[VideoArtifact]] = None
+    audio: Optional[List[AudioArtifact]] = None
+    files: Optional[List[File]] = None
+
+    def get_input_as_string(self) -> Optional[str]:
+        """Convert input to string representation"""
+        if self.input is None:
+            return None
+
+        if isinstance(self.input, str):
+            return self.input
+        elif isinstance(self.input, BaseModel):
+            return self.input.model_dump_json(indent=2, exclude_none=True)
+        elif isinstance(self.input, (dict, list)):
+            import json
+
+            return json.dumps(self.input, indent=2, default=str)
+        else:
+            return str(self.input)
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary"""
+        input_dict: Optional[Union[str, Dict[str, Any], List[Any]]] = None
+        if self.input is not None:
+            if isinstance(self.input, BaseModel):
+                input_dict = self.input.model_dump(exclude_none=True)
+            elif isinstance(self.input, (dict, list)):
+                input_dict = self.input
+            else:
+                input_dict = str(self.input)
+
+        return {
+            "input": input_dict,
+            "additional_data": self.additional_data,
+            "images": [img.to_dict() for img in self.images] if self.images else None,
+            "videos": [vid.to_dict() for vid in self.videos] if self.videos else None,
+            "audio": [aud.to_dict() for aud in self.audio] if self.audio else None,
+        }
+
+
+@dataclass
+class StepInput:
+    """Input data for a step execution"""
+
+    input: Optional[Union[str, Dict[str, Any], List[Any], BaseModel]] = None
+
+    previous_step_content: Optional[Any] = None
+    previous_step_outputs: Optional[Dict[str, "StepOutput"]] = None
+
+    additional_data: Optional[Dict[str, Any]] = None
+
+    # Media inputs
+    images: Optional[List[ImageArtifact]] = None
+    videos: Optional[List[VideoArtifact]] = None
+    audio: Optional[List[AudioArtifact]] = None
+    files: Optional[List[File]] = None
+
+    def get_input_as_string(self) -> Optional[str]:
+        """Convert input to string representation"""
+        if self.input is None:
+            return None
+
+        if isinstance(self.input, str):
+            return self.input
+        elif isinstance(self.input, BaseModel):
+            return self.input.model_dump_json(indent=2, exclude_none=True)
+        elif isinstance(self.input, (dict, list)):
+            import json
+
+            return json.dumps(self.input, indent=2, default=str)
+        else:
+            return str(self.input)
+
+    def get_step_output(self, step_name: str) -> Optional["StepOutput"]:
+        """Get output from a specific previous step by name"""
+        if not self.previous_step_outputs:
+            return None
+        return self.previous_step_outputs.get(step_name)
+
+    def get_step_content(self, step_name: str) -> Optional[Union[str, Dict[str, str]]]:
+        """Get content from a specific previous step by name
+
+        For parallel steps, if you ask for the parallel step name, returns a dict
+        with {step_name: content} for each sub-step.
+        For other nested steps (Condition, Router, Loop, Steps), returns the deepest content.
+        """
+        step_output = self.get_step_output(step_name)
+        if not step_output:
+            return None
+
+        # Check if this is a parallel step with nested steps
+        if step_output.step_type == "Parallel" and step_output.steps:
+            # Return dict with {step_name: content} for each sub-step
+            parallel_content = {}
+            for sub_step in step_output.steps:
+                if sub_step.step_name and sub_step.content:
+                    # Check if this sub-step has its own nested steps (like Condition -> Research Step)
+                    if sub_step.steps and len(sub_step.steps) > 0:
+                        # This is a composite step (like Condition) - get content from its nested steps
+                        for nested_step in sub_step.steps:
+                            if nested_step.step_name and nested_step.content:
+                                parallel_content[nested_step.step_name] = str(nested_step.content)
+                    else:
+                        # This is a direct step - use its content
+                        parallel_content[sub_step.step_name] = str(sub_step.content)
+            return parallel_content if parallel_content else str(step_output.content)
+
+        # For other nested step types (Condition, Router, Loop, Steps), get the deepest content
+        elif step_output.steps and len(step_output.steps) > 0:
+            # This is a nested step structure - recursively get the deepest content
+            return self._get_deepest_step_content(step_output.steps[-1])
+
+        # Regular step, return content directly
+        return step_output.content  # type: ignore[return-value]
+
+    def _get_deepest_step_content(self, step_output: "StepOutput") -> Optional[Union[str, Dict[str, str]]]:
+        """Helper method to recursively extract deepest content from nested steps"""
+        # If this step has nested steps, go deeper
+        if step_output.steps and len(step_output.steps) > 0:
+            return self._get_deepest_step_content(step_output.steps[-1])
+
+        # Return the content of this step
+        return step_output.content  # type: ignore[return-value]
+
+    def get_all_previous_content(self) -> str:
+        """Get concatenated content from all previous steps"""
+        if not self.previous_step_outputs:
+            return ""
+
+        content_parts = []
+        for step_name, output in self.previous_step_outputs.items():
+            if output.content:
+                content_parts.append(f"=== {step_name} ===\n{output.content}")
+
+        return "\n\n".join(content_parts)
+
+    def get_last_step_content(self) -> Optional[str]:
+        """Get content from the most recent step (for backward compatibility)"""
+        if not self.previous_step_outputs:
+            return None
+
+        last_output = list(self.previous_step_outputs.values())[-1] if self.previous_step_outputs else None
+        if not last_output:
+            return None
+
+        # Use the helper method to get the deepest content
+        return self._get_deepest_step_content(last_output)  # type: ignore[return-value]
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary"""
+        # Handle the unified message field
+        input_dict: Optional[Union[str, Dict[str, Any], List[Any]]] = None
+        if self.input is not None:
+            if isinstance(self.input, BaseModel):
+                input_dict = self.input.model_dump(exclude_none=True)
+            elif isinstance(self.input, (dict, list)):
+                input_dict = self.input
+            else:
+                input_dict = str(self.input)
+
+        previous_step_content_str: Optional[str] = None
+        # Handle previous_step_content (keep existing logic)
+        if isinstance(self.previous_step_content, BaseModel):
+            previous_step_content_str = self.previous_step_content.model_dump_json(indent=2, exclude_none=True)
+        elif isinstance(self.previous_step_content, dict):
+            import json
+
+            previous_step_content_str = json.dumps(self.previous_step_content, indent=2, default=str)
+        elif self.previous_step_content:
+            previous_step_content_str = str(self.previous_step_content)
+
+        # Convert previous_step_outputs to serializable format (keep existing logic)
+        previous_steps_dict = {}
+        if self.previous_step_outputs:
+            for step_name, output in self.previous_step_outputs.items():
+                previous_steps_dict[step_name] = output.to_dict()
+
+        return {
+            "input": input_dict,
+            "previous_step_outputs": previous_steps_dict,
+            "previous_step_content": previous_step_content_str,
+            "additional_data": self.additional_data,
+            "images": [img.to_dict() for img in self.images] if self.images else None,
+            "videos": [vid.to_dict() for vid in self.videos] if self.videos else None,
+            "audio": [aud.to_dict() for aud in self.audio] if self.audio else None,
+            "files": [file for file in self.files] if self.files else None,
+        }
+
+
+@dataclass
+class StepOutput:
+    """Output data from a step execution"""
+
+    step_name: Optional[str] = None
+    step_id: Optional[str] = None
+    step_type: Optional[str] = None
+    executor_type: Optional[str] = None
+    executor_name: Optional[str] = None
+    # Primary output
+    content: Optional[Union[str, Dict[str, Any], List[Any], BaseModel, Any]] = None
+
+    # Link to the run ID of the step execution
+    step_run_id: Optional[str] = None
+
+    # Media outputs
+    images: Optional[List[ImageArtifact]] = None
+    videos: Optional[List[VideoArtifact]] = None
+    audio: Optional[List[AudioArtifact]] = None
+    files: Optional[List[File]] = None
+
+    # Metrics for this step execution
+    metrics: Optional[Metrics] = None
+
+    success: bool = True
+    error: Optional[str] = None
+
+    stop: bool = False
+
+    steps: Optional[List["StepOutput"]] = None
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary"""
+        # Handle the unified content field
+        content_dict: Optional[Union[str, Dict[str, Any], List[Any]]] = None
+        if self.content is not None:
+            if isinstance(self.content, BaseModel):
+                content_dict = self.content.model_dump(exclude_none=True)
+            elif isinstance(self.content, (dict, list)):
+                content_dict = self.content
+            else:
+                content_dict = str(self.content)
+
+        result = {
+            "content": content_dict,
+            "step_name": self.step_name,
+            "step_id": self.step_id,
+            "step_type": self.step_type,
+            "executor_type": self.executor_type,
+            "executor_name": self.executor_name,
+            "step_run_id": self.step_run_id,
+            "images": [img.to_dict() for img in self.images] if self.images else None,
+            "videos": [vid.to_dict() for vid in self.videos] if self.videos else None,
+            "audio": [aud.to_dict() for aud in self.audio] if self.audio else None,
+            "metrics": self.metrics.to_dict() if self.metrics else None,
+            "success": self.success,
+            "error": self.error,
+            "stop": self.stop,
+            "files": [file for file in self.files] if self.files else None,
+        }
+
+        # Add nested steps if they exist
+        if self.steps:
+            result["steps"] = [step.to_dict() for step in self.steps]
+
+        return result
+
+    @classmethod
+    def from_dict(cls, data: Dict[str, Any]) -> "StepOutput":
+        """Create StepOutput from dictionary"""
+        # Reconstruct media artifacts
+        images = data.get("images")
+        if images:
+            images = [ImageArtifact.model_validate(img) for img in images]
+
+        videos = data.get("videos")
+        if videos:
+            videos = [VideoArtifact.model_validate(vid) for vid in videos]
+
+        audio = data.get("audio")
+        if audio:
+            audio = [AudioArtifact.model_validate(aud) for aud in audio]
+
+        files = data.get("files")
+        if files:
+            files = [File.model_validate(file) for file in files]
+
+        metrics_data = data.get("metrics")
+        metrics = None
+        if metrics_data:
+            if isinstance(metrics_data, dict):
+                # Convert dict to Metrics object
+                from agno.models.metrics import Metrics
+
+                metrics = Metrics(**metrics_data)
+            else:
+                # Already a Metrics object
+                metrics = metrics_data
+
+        # Handle nested steps
+        steps_data = data.get("steps")
+        steps = None
+        if steps_data:
+            steps = [cls.from_dict(step_data) for step_data in steps_data]
+
+        return cls(
+            step_name=data.get("step_name"),
+            step_id=data.get("step_id"),
+            step_type=data.get("step_type"),
+            executor_type=data.get("executor_type"),
+            executor_name=data.get("executor_name"),
+            content=data.get("content"),
+            step_run_id=data.get("step_run_id"),
+            images=images,
+            videos=videos,
+            audio=audio,
+            files=files,
+            metrics=metrics,
+            success=data.get("success", True),
+            error=data.get("error"),
+            stop=data.get("stop", False),
+            steps=steps,
+        )
+
+
+@dataclass
+class StepMetrics:
+    """Metrics for a single step execution"""
+
+    step_name: str
+    executor_type: str  # "agent", "team", etc.
+    executor_name: str
+    metrics: Optional[Metrics] = None
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary"""
+        return {
+            "step_name": self.step_name,
+            "executor_type": self.executor_type,
+            "executor_name": self.executor_name,
+            "metrics": self.metrics.to_dict() if self.metrics else None,
+        }
+
+    @classmethod
+    def from_dict(cls, data: Dict[str, Any]) -> "StepMetrics":
+        """Create StepMetrics from dictionary"""
+
+        # Handle metrics properly
+        metrics_data = data.get("metrics")
+        metrics = None
+        if metrics_data:
+            if isinstance(metrics_data, dict):
+                # Convert dict to Metrics object
+                from agno.models.metrics import Metrics
+
+                metrics = Metrics(**metrics_data)
+            else:
+                # Already a Metrics object
+                metrics = metrics_data
+
+        return cls(
+            step_name=data["step_name"],
+            executor_type=data["executor_type"],
+            executor_name=data["executor_name"],
+            metrics=metrics,
+        )
+
+
+@dataclass
+class WorkflowMetrics:
+    """Complete metrics for a workflow execution"""
+
+    steps: Dict[str, StepMetrics]
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary"""
+        return {
+            "steps": {name: step.to_dict() for name, step in self.steps.items()},
+        }
+
+    @classmethod
+    def from_dict(cls, data: Dict[str, Any]) -> "WorkflowMetrics":
+        """Create WorkflowMetrics from dictionary"""
+        steps = {name: StepMetrics.from_dict(step_data) for name, step_data in data["steps"].items()}
+
+        return cls(
+            steps=steps,
+        )
+
+
+@dataclass
+class WebSocketHandler:
+    """Generic WebSocket handler for real-time workflow events"""
+
+    websocket: Optional[WebSocket] = None
+
+    def format_sse_event(self, json_data: str) -> str:
+        """Parse JSON data into SSE-compliant format.
+
+        Args:
+            json_data: JSON string containing the event data
+
+        Returns:
+            SSE-formatted response with event type and data
+        """
+        import json
+
+        try:
+            # Parse the JSON to extract the event type
+            data = json.loads(json_data)
+            event_type = data.get("event", "message")
+
+            # Format as SSE: event: <event_type>\ndata: <json_data>\n\n
+            return f"event: {event_type}\ndata: {json_data}\n\n"
+        except (json.JSONDecodeError, KeyError):
+            # Fallback to generic message event if parsing fails
+            return f"event: message\ndata: {json_data}\n\n"
+
+    async def handle_event(self, event: Any) -> None:
+        """Handle an event object - serializes and sends via WebSocket"""
+        if not self.websocket:
+            return
+
+        try:
+            if hasattr(event, "to_dict"):
+                data = event.to_dict()
+            elif hasattr(event, "__dict__"):
+                data = event.__dict__
+            elif isinstance(event, dict):
+                data = event
+            else:
+                data = {"type": "message", "content": str(event)}
+
+            import json
+
+            await self.websocket.send_text(self.format_sse_event(json.dumps(data)))
+
+        except Exception as e:
+            log_warning(f"Failed to handle WebSocket event: {e}")
+
+    async def handle_text(self, message: str) -> None:
+        """Handle a plain text message"""
+        if not self.websocket:
+            return
+
+        try:
+            await self.websocket.send_text(self.format_sse_event(message))
+        except Exception as e:
+            log_warning(f"Failed to send WebSocket text: {e}")
+
+    async def handle_dict(self, data: Dict[str, Any]) -> None:
+        """Handle a dictionary directly"""
+        if not self.websocket:
+            return
+
+        try:
+            import json
+
+            await self.websocket.send_text(self.format_sse_event(json.dumps(data)))
+        except Exception as e:
+            log_warning(f"Failed to send WebSocket dict: {e}")
+
+
+class StepType(str, Enum):
+    FUNCTION = "Function"
+    STEP = "Step"
+    STEPS = "Steps"
+    LOOP = "Loop"
+    PARALLEL = "Parallel"
+    CONDITION = "Condition"
+    ROUTER = "Router"