@feltdb/core 0.4.12 → 0.4.14

package/dist/create/server-source/crates/feltdb/src/workflow.rs

@@ -0,0 +1,713 @@
+use crate::FlowRef;
+/// Workflow module for durable, reactive distributed workflows
+///
+/// A workflow is a durable graph of intents and references that can execute
+/// wherever the required capabilities exist, without the workflow owning execution location.
+use serde::{Deserialize, Serialize};
+use std::collections::{HashMap, HashSet};
+use std::fmt::{Display, Formatter};
+
+/// Reference to a workflow - addressable object in the fabric
+///
+/// WorkflowRef {
+///     namespace: "workflows",
+///     workflow_id: "order-processing",
+///     version: 1,
+/// }
+///
+/// For example:
+/// - flow://workflow/order-processing@1
+/// - flow://workflow/document-ingestion@2
+/// - flow://workflow/research@1
+#[derive(Debug, Clone, Serialize, Deserialize, Eq, PartialEq, Hash)]
+pub struct WorkflowRef {
+    /// Namespace for workflows (typically "workflow")
+    pub namespace: String,
+    /// Unique identifier for the workflow
+    pub workflow_id: String,
+    /// Version of the workflow
+    pub version: u64,
+}
+
+impl WorkflowRef {
+    /// Create a new WorkflowRef
+    pub fn new(namespace: impl Into<String>, workflow_id: impl Into<String>, version: u64) -> Self {
+        Self {
+            namespace: namespace.into(),
+            workflow_id: workflow_id.into(),
+            version,
+        }
+    }
+
+    /// Get canonical string representation
+    pub fn canonical(&self) -> String {
+        format!(
+            "flow://{}/{}@{}",
+            self.namespace, self.workflow_id, self.version
+        )
+    }
+
+    /// Parse from canonical form
+    pub fn from_canonical(s: &str) -> Option<Self> {
+        if !s.starts_with("flow://") {
+            return None;
+        }
+        let s = &s[7..]; // Remove "flow://"
+
+        let (namespace, rest) = s.split_once('/')?;
+        let (workflow_id, version_str) = rest.split_once('@')?;
+        let version = version_str.parse().ok()?;
+
+        Some(Self {
+            namespace: namespace.to_string(),
+            workflow_id: workflow_id.to_string(),
+            version,
+        })
+    }
+}
+
+impl Display for WorkflowRef {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}", self.canonical())
+    }
+}
+
+/// Reference to a step within a workflow
+///
+/// Provides stable identity for workflow steps across peer failures and migrations.
+#[derive(Debug, Clone, Serialize, Deserialize, Eq, PartialEq, Hash)]
+pub struct WorkflowStepRef {
+    /// Reference to the parent workflow
+    pub workflow_ref: WorkflowRef,
+    /// Step identifier within the workflow (e.g., "step-1", "extract", "join-results")
+    pub step_id: String,
+}
+
+impl WorkflowStepRef {
+    /// Create a new WorkflowStepRef
+    pub fn new(workflow_ref: WorkflowRef, step_id: impl Into<String>) -> Self {
+        Self {
+            workflow_ref,
+            step_id: step_id.into(),
+        }
+    }
+
+    /// Get canonical string representation
+    pub fn canonical(&self) -> String {
+        format!("{}/{}", self.workflow_ref.canonical(), self.step_id)
+    }
+
+    /// Parse from canonical form
+    pub fn from_canonical(s: &str) -> Option<Self> {
+        // Expected format: flow://workflow/workflow-id@version/step-id
+        if !s.starts_with("flow://") {
+            return None;
+        }
+
+        let s = &s[7..]; // Remove "flow://"
+
+        // Find the last '/' to separate workflow part from step-id
+        let last_slash_idx = s.rfind('/')?;
+        let workflow_str = &s[..last_slash_idx];
+        let step_id = &s[last_slash_idx + 1..];
+
+        // Parse the workflow part
+        let (namespace, rest) = workflow_str.split_once('/')?;
+        let (workflow_id, version_str) = rest.split_once('@')?;
+        let version = version_str.parse().ok()?;
+
+        Some(Self {
+            workflow_ref: WorkflowRef {
+                namespace: namespace.to_string(),
+                workflow_id: workflow_id.to_string(),
+                version,
+            },
+            step_id: step_id.to_string(),
+        })
+    }
+}
+
+impl Display for WorkflowStepRef {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}", self.canonical())
+    }
+}
+
+/// State of a workflow step in the durable execution model
+#[derive(Debug, Clone, Copy, Serialize, Deserialize, PartialEq, Eq)]
+pub enum WorkflowStepState {
+    /// Waiting for dependencies to be satisfied
+    Pending,
+    /// Dependencies satisfied, ready to execute
+    Ready,
+    /// Step has been claimed by a peer for execution
+    Claimed,
+    /// Step is currently executing
+    Running,
+    /// Step completed successfully
+    Completed,
+    /// Step failed during execution
+    Failed,
+    /// Step scheduled for retry after failure
+    Retrying,
+    /// Step is blocked, waiting for external input
+    Blocked,
+}
+
+impl Display for WorkflowStepState {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        match self {
+            Self::Pending => write!(f, "Pending"),
+            Self::Ready => write!(f, "Ready"),
+            Self::Claimed => write!(f, "Claimed"),
+            Self::Running => write!(f, "Running"),
+            Self::Completed => write!(f, "Completed"),
+            Self::Failed => write!(f, "Failed"),
+            Self::Retrying => write!(f, "Retrying"),
+            Self::Blocked => write!(f, "Blocked"),
+        }
+    }
+}
+
+/// Reason a workflow step is blocked
+#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
+pub enum BlockedReason {
+    /// Awaiting human approval
+    AwaitingApproval(String),
+    /// Awaiting human input
+    AwaitingInput(String),
+    /// Required capability not available
+    AwaitingCapability(String),
+    /// Insufficient resources
+    AwaitingResource(String),
+    /// Custom blocking reason
+    Custom(String),
+}
+
+impl Display for BlockedReason {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        match self {
+            Self::AwaitingApproval(msg) => write!(f, "AwaitingApproval: {}", msg),
+            Self::AwaitingInput(msg) => write!(f, "AwaitingInput: {}", msg),
+            Self::AwaitingCapability(cap) => write!(f, "AwaitingCapability: {}", cap),
+            Self::AwaitingResource(res) => write!(f, "AwaitingResource: {}", res),
+            Self::Custom(msg) => write!(f, "Custom: {}", msg),
+        }
+    }
+}
+
+/// State of a workflow
+#[derive(Debug, Clone, Copy, Serialize, Deserialize, PartialEq, Eq)]
+pub enum WorkflowState {
+    /// Workflow has been created but not started
+    Created,
+    /// Workflow is currently executing
+    Running,
+    /// Workflow completed successfully
+    Completed,
+    /// Workflow failed
+    Failed,
+}
+
+impl Display for WorkflowState {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        match self {
+            Self::Created => write!(f, "Created"),
+            Self::Running => write!(f, "Running"),
+            Self::Completed => write!(f, "Completed"),
+            Self::Failed => write!(f, "Failed"),
+        }
+    }
+}
+
+/// Workflow lifecycle event - becomes a durable operation
+#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
+pub enum WorkflowOperation {
+    /// Workflow was created
+    Created { workflow_ref: WorkflowRef },
+    /// Workflow started execution
+    Started { workflow_ref: WorkflowRef },
+    /// A step became ready to execute
+    StepReady { step_ref: WorkflowStepRef },
+    /// A step was claimed by a peer
+    StepClaimed {
+        step_ref: WorkflowStepRef,
+        claimed_by: String,
+    },
+    /// A step started executing
+    StepExecuting {
+        step_ref: WorkflowStepRef,
+        claimed_by: String,
+    },
+    /// A step completed successfully
+    StepCompleted {
+        step_ref: WorkflowStepRef,
+        result_ref: Option<FlowRef>,
+    },
+    /// A step failed
+    StepFailed {
+        step_ref: WorkflowStepRef,
+        error: String,
+        attempt: u32,
+    },
+    /// A step is blocked pending external input
+    StepBlocked {
+        step_ref: WorkflowStepRef,
+        reason: BlockedReason,
+    },
+    /// A step retry was scheduled
+    StepRetrying {
+        step_ref: WorkflowStepRef,
+        attempt: u32,
+    },
+    /// Workflow completed
+    Completed { workflow_ref: WorkflowRef },
+    /// Workflow failed
+    Failed {
+        workflow_ref: WorkflowRef,
+        error: String,
+    },
+}
+
+impl WorkflowOperation {
+    /// Get the affected workflow reference
+    pub fn workflow_ref(&self) -> Option<&WorkflowRef> {
+        match self {
+            Self::Created { workflow_ref } => Some(workflow_ref),
+            Self::Started { workflow_ref } => Some(workflow_ref),
+            Self::Completed { workflow_ref } => Some(workflow_ref),
+            Self::Failed { workflow_ref, .. } => Some(workflow_ref),
+            _ => None,
+        }
+    }
+
+    /// Get the affected step reference
+    pub fn step_ref(&self) -> Option<&WorkflowStepRef> {
+        match self {
+            Self::StepReady { step_ref } => Some(step_ref),
+            Self::StepClaimed { step_ref, .. } => Some(step_ref),
+            Self::StepExecuting { step_ref, .. } => Some(step_ref),
+            Self::StepCompleted { step_ref, .. } => Some(step_ref),
+            Self::StepFailed { step_ref, .. } => Some(step_ref),
+            Self::StepBlocked { step_ref, .. } => Some(step_ref),
+            Self::StepRetrying { step_ref, .. } => Some(step_ref),
+            _ => None,
+        }
+    }
+}
+
+/// Workflow graph node - represents a capability to execute
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct WorkflowNode {
+    /// Step identifier (e.g., "extract", "embed", "classify")
+    pub step_id: String,
+    /// Capability to execute (reference to a CapabilityRef)
+    pub capability: String,
+    /// Input references required for this step
+    pub inputs: Vec<String>, // Step IDs or input slot names
+    /// Output slot names produced by this step
+    pub outputs: Vec<String>,
+    /// State requirements for this step
+    pub state_requirements: HashMap<String, String>,
+    /// Metadata about this node
+    pub metadata: HashMap<String, serde_json::Value>,
+}
+
+impl WorkflowNode {
+    /// Create a new workflow node
+    pub fn new(step_id: impl Into<String>, capability: impl Into<String>) -> Self {
+        Self {
+            step_id: step_id.into(),
+            capability: capability.into(),
+            inputs: Vec::new(),
+            outputs: Vec::new(),
+            state_requirements: HashMap::new(),
+            metadata: HashMap::new(),
+        }
+    }
+
+    /// Add an input dependency
+    pub fn with_input(mut self, input: impl Into<String>) -> Self {
+        self.inputs.push(input.into());
+        self
+    }
+
+    /// Add an output slot
+    pub fn with_output(mut self, output: impl Into<String>) -> Self {
+        self.outputs.push(output.into());
+        self
+    }
+
+    /// Add a state requirement
+    pub fn with_state_requirement(
+        mut self,
+        key: impl Into<String>,
+        value: impl Into<String>,
+    ) -> Self {
+        self.state_requirements.insert(key.into(), value.into());
+        self
+    }
+}
+
+/// Workflow graph definition - durable DAG structure
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct WorkflowGraph {
+    /// Workflow reference
+    pub workflow_ref: WorkflowRef,
+    /// Nodes in the workflow
+    pub nodes: HashMap<String, WorkflowNode>,
+    /// Edges: from_step_id -> to_step_id
+    pub edges: HashMap<String, Vec<String>>,
+    /// Input references for the entire workflow
+    pub inputs: Vec<FlowRef>,
+    /// Output slot definitions
+    pub outputs: Vec<String>,
+    /// Metadata
+    pub metadata: HashMap<String, serde_json::Value>,
+}
+
+impl WorkflowGraph {
+    /// Create a new workflow graph
+    pub fn new(workflow_ref: WorkflowRef) -> Self {
+        Self {
+            workflow_ref,
+            nodes: HashMap::new(),
+            edges: HashMap::new(),
+            inputs: Vec::new(),
+            outputs: Vec::new(),
+            metadata: HashMap::new(),
+        }
+    }
+
+    /// Add a node to the graph
+    pub fn add_node(mut self, node: WorkflowNode) -> Self {
+        self.nodes.insert(node.step_id.clone(), node);
+        self
+    }
+
+    /// Add an edge from one step to another
+    pub fn add_edge(mut self, from_step: impl Into<String>, to_step: impl Into<String>) -> Self {
+        let from = from_step.into();
+        let to = to_step.into();
+        self.edges.entry(from).or_insert_with(Vec::new).push(to);
+        self
+    }
+
+    /// Add an input reference
+    pub fn with_input(mut self, input: FlowRef) -> Self {
+        self.inputs.push(input);
+        self
+    }
+
+    /// Validate the graph structure (basic DAG check)
+    pub fn validate(&self) -> Result<(), String> {
+        // Check all edges reference existing nodes
+        for (from, tos) in &self.edges {
+            if !self.nodes.contains_key(from) {
+                return Err(format!("Edge source '{}' does not exist", from));
+            }
+            for to in tos {
+                if !self.nodes.contains_key(to) {
+                    return Err(format!("Edge target '{}' does not exist", to));
+                }
+            }
+        }
+
+        // Check input dependencies reference existing nodes or inputs
+        for node in self.nodes.values() {
+            for input in &node.inputs {
+                if !self.nodes.contains_key(input) && !input.starts_with("input:") {
+                    return Err(format!(
+                        "Node '{}' has input dependency '{}' that doesn't exist",
+                        node.step_id, input
+                    ));
+                }
+            }
+        }
+
+        // Simple cycle detection using DFS
+        self.detect_cycles()?;
+
+        Ok(())
+    }
+
+    /// Detect cycles in the graph using DFS
+    fn detect_cycles(&self) -> Result<(), String> {
+        let mut visited: HashSet<String> = HashSet::new();
+        let mut rec_stack: HashSet<String> = HashSet::new();
+
+        for node_id in self.nodes.keys() {
+            if !visited.contains(node_id) {
+                self.dfs(node_id, &mut visited, &mut rec_stack)?;
+            }
+        }
+
+        Ok(())
+    }
+
+    /// DFS helper for cycle detection
+    fn dfs(
+        &self,
+        node_id: &str,
+        visited: &mut HashSet<String>,
+        rec_stack: &mut HashSet<String>,
+    ) -> Result<(), String> {
+        visited.insert(node_id.to_string());
+        rec_stack.insert(node_id.to_string());
+
+        if let Some(neighbors) = self.edges.get(node_id) {
+            for neighbor in neighbors {
+                if !visited.contains(neighbor) {
+                    self.dfs(neighbor, visited, rec_stack)?;
+                } else if rec_stack.contains(neighbor) {
+                    return Err(format!("Cycle detected: {} -> {}", node_id, neighbor));
+                }
+            }
+        }
+
+        rec_stack.remove(node_id);
+        Ok(())
+    }
+
+    /// Get nodes with no dependencies (entry points)
+    pub fn entry_nodes(&self) -> Vec<String> {
+        self.nodes
+            .keys()
+            .filter(|node_id| {
+                if let Some(node) = self.nodes.get(*node_id) {
+                    // A node is an entry point if all its inputs come from workflow inputs
+                    node.inputs.is_empty()
+                        || node.inputs.iter().all(|inp| inp.starts_with("input:"))
+                } else {
+                    false
+                }
+            })
+            .cloned()
+            .collect()
+    }
+
+    /// Get nodes that depend on a given step
+    pub fn dependents(&self, step_id: &str) -> Vec<String> {
+        self.edges
+            .get(step_id)
+            .map(|deps| deps.clone())
+            .unwrap_or_default()
+    }
+
+    /// Get nodes that a given step depends on
+    pub fn dependencies(&self, step_id: &str) -> Vec<String> {
+        if let Some(node) = self.nodes.get(step_id) {
+            node.inputs
+                .iter()
+                .filter(|inp| !inp.starts_with("input:"))
+                .cloned()
+                .collect()
+        } else {
+            Vec::new()
+        }
+    }
+}
+
+/// Runtime state for a workflow instance
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct WorkflowInstance {
+    /// Reference to the workflow definition
+    pub workflow_ref: WorkflowRef,
+    /// Unique instance ID for this execution
+    pub instance_id: String,
+    /// Overall workflow state
+    pub state: WorkflowState,
+    /// Current state of each step
+    pub step_states: HashMap<String, WorkflowStepState>,
+    /// Which peer claimed each step
+    pub step_claims: HashMap<String, String>,
+    /// Results produced by each step
+    pub step_results: HashMap<String, FlowRef>,
+    /// Failed attempts for each step
+    pub step_failures: HashMap<String, Vec<String>>,
+    /// Blocked reason for each step
+    pub step_blocked_reasons: HashMap<String, BlockedReason>,
+    /// Timestamp created (ms since UNIX epoch)
+    pub created_ms: u128,
+    /// Timestamp started (ms since UNIX epoch)
+    pub started_ms: Option<u128>,
+    /// Timestamp completed (ms since UNIX epoch)
+    pub completed_ms: Option<u128>,
+}
+
+impl WorkflowInstance {
+    /// Create a new workflow instance
+    pub fn new(workflow_ref: WorkflowRef, instance_id: impl Into<String>) -> Self {
+        Self {
+            workflow_ref,
+            instance_id: instance_id.into(),
+            state: WorkflowState::Created,
+            step_states: HashMap::new(),
+            step_claims: HashMap::new(),
+            step_results: HashMap::new(),
+            step_failures: HashMap::new(),
+            step_blocked_reasons: HashMap::new(),
+            created_ms: crate::now_ms(),
+            started_ms: None,
+            completed_ms: None,
+        }
+    }
+
+    /// Get the state of a step
+    pub fn get_step_state(&self, step_id: &str) -> Option<WorkflowStepState> {
+        self.step_states.get(step_id).copied()
+    }
+
+    /// Set the state of a step
+    pub fn set_step_state(&mut self, step_id: impl Into<String>, state: WorkflowStepState) {
+        self.step_states.insert(step_id.into(), state);
+    }
+
+    /// Mark a step as claimed
+    pub fn claim_step(
+        &mut self,
+        step_id: impl Into<String>,
+        claimed_by: impl Into<String>,
+    ) -> Result<(), String> {
+        let step_id = step_id.into();
+        let claimed_by = claimed_by.into();
+
+        match self.get_step_state(&step_id) {
+            Some(WorkflowStepState::Ready) => {
+                self.set_step_state(step_id.clone(), WorkflowStepState::Claimed);
+                self.step_claims.insert(step_id, claimed_by);
+                Ok(())
+            }
+            Some(state) => Err(format!("Cannot claim step in state {}", state)),
+            None => Err(format!("Step not found: {}", step_id)),
+        }
+    }
+
+    /// Record a step completion
+    pub fn complete_step(
+        &mut self,
+        step_id: impl Into<String>,
+        result_ref: FlowRef,
+    ) -> Result<(), String> {
+        let step_id = step_id.into();
+        self.set_step_state(step_id.clone(), WorkflowStepState::Completed);
+        self.step_results.insert(step_id, result_ref);
+        Ok(())
+    }
+
+    /// Record a step failure
+    pub fn fail_step(&mut self, step_id: impl Into<String>, error: String) -> Result<(), String> {
+        let step_id = step_id.into();
+        self.set_step_state(step_id.clone(), WorkflowStepState::Failed);
+        self.step_failures
+            .entry(step_id)
+            .or_insert_with(Vec::new)
+            .push(error);
+        Ok(())
+    }
+
+    /// Check if workflow is terminal (won't change state)
+    pub fn is_terminal(&self) -> bool {
+        matches!(self.state, WorkflowState::Completed | WorkflowState::Failed)
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_workflow_ref_canonical() {
+        let r = WorkflowRef::new("workflow", "order-processing", 1);
+        assert_eq!(r.canonical(), "flow://workflow/order-processing@1");
+    }
+
+    #[test]
+    fn test_workflow_ref_parse() {
+        let s = "flow://workflow/order-processing@1";
+        let r = WorkflowRef::from_canonical(s).unwrap();
+        assert_eq!(r.namespace, "workflow");
+        assert_eq!(r.workflow_id, "order-processing");
+        assert_eq!(r.version, 1);
+    }
+
+    #[test]
+    fn test_workflow_step_ref_canonical() {
+        let wf = WorkflowRef::new("workflow", "test", 1);
+        let step = WorkflowStepRef::new(wf, "step-1");
+        assert_eq!(step.canonical(), "flow://workflow/test@1/step-1");
+    }
+
+    #[test]
+    fn test_workflow_step_ref_parse() {
+        let s = "flow://workflow/test@1/step-1";
+        let step = WorkflowStepRef::from_canonical(s).unwrap();
+        assert_eq!(step.step_id, "step-1");
+        assert_eq!(step.workflow_ref.workflow_id, "test");
+    }
+
+    #[test]
+    fn test_workflow_graph_validation() {
+        let wf = WorkflowRef::new("workflow", "test", 1);
+        let graph = WorkflowGraph::new(wf)
+            .add_node(WorkflowNode::new("step-1", "capability-a"))
+            .add_node(WorkflowNode::new("step-2", "capability-b").with_input("step-1"))
+            .add_edge("step-1", "step-2");
+
+        assert!(graph.validate().is_ok());
+    }
+
+    #[test]
+    fn test_workflow_graph_cycle_detection() {
+        let wf = WorkflowRef::new("workflow", "test", 1);
+        let graph = WorkflowGraph::new(wf)
+            .add_node(WorkflowNode::new("step-1", "capability-a"))
+            .add_node(WorkflowNode::new("step-2", "capability-b"))
+            .add_edge("step-1", "step-2")
+            .add_edge("step-2", "step-1");
+
+        assert!(graph.validate().is_err());
+    }
+
+    #[test]
+    fn test_workflow_graph_entry_nodes() {
+        let wf = WorkflowRef::new("workflow", "test", 1);
+        let graph = WorkflowGraph::new(wf)
+            .add_node(WorkflowNode::new("step-1", "capability-a"))
+            .add_node(WorkflowNode::new("step-2", "capability-b").with_input("step-1"))
+            .add_edge("step-1", "step-2");
+
+        let entries = graph.entry_nodes();
+        assert_eq!(entries.len(), 1);
+        assert!(entries.contains(&"step-1".to_string()));
+    }
+
+    #[test]
+    fn test_workflow_instance_lifecycle() {
+        let wf = WorkflowRef::new("workflow", "test", 1);
+        let mut instance = WorkflowInstance::new(wf, "instance-1");
+
+        instance.set_step_state("step-1", WorkflowStepState::Pending);
+        assert_eq!(
+            instance.get_step_state("step-1"),
+            Some(WorkflowStepState::Pending)
+        );
+
+        instance.set_step_state("step-1", WorkflowStepState::Ready);
+        let claim_result = instance.claim_step("step-1", "peer-a");
+        assert!(claim_result.is_ok());
+        assert_eq!(
+            instance.get_step_state("step-1"),
+            Some(WorkflowStepState::Claimed)
+        );
+    }
+
+    #[test]
+    fn test_workflow_blocked_reason() {
+        let reason = BlockedReason::AwaitingApproval("Needs manager approval".to_string());
+        assert_eq!(
+            reason.to_string(),
+            "AwaitingApproval: Needs manager approval"
+        );
+    }
+}