w4c-workflows-api/Services/Nodes/NodeGraphRunner.cs
2026-09-12 01:02:46 +03:00

459 lines
18 KiB
C#

using System.Text.Json.Nodes;
using w4c_workflows.Models.Credentials;
using w4c_workflows.Models.Nodes;
using w4c_workflows.Services.Nodes.Interpolation;
using w4c_workflows.Services.Security;
namespace w4c_workflows.Services.Nodes;
/// <summary>Ambient data a node run needs beyond the graph itself.</summary>
public sealed record NodeRunEnvironment
{
public string TenantId { get; init; } = string.Empty;
public string RunId { get; init; } = string.Empty;
/// <summary>Workflow variables exposed to expressions as <c>$env</c>.</summary>
public JsonObject Variables { get; init; } = new();
/// <summary>Decrypted credentials keyed by the blueprint alias the step uses.</summary>
public IReadOnlyDictionary<string, CredentialData> Credentials { get; init; } =
new Dictionary<string, CredentialData>();
/// <summary>Directory code-executing nodes (<c>core.code</c>) resolve their entry file from.</summary>
public string? WorkingDirectory { get; init; }
/// <summary>
/// Run-scoped mutable state shared by all nodes of one run. Loop nodes use it
/// to remember their iteration cursor; it is created fresh per run.
/// </summary>
public IDictionary<string, object?> State { get; init; } = new Dictionary<string, object?>();
/// <summary>
/// Invokes another workflow in-process from a <c>core.executeWorkflow</c>
/// node. Null when the host does not support sub-workflows (e.g. a bare
/// <see cref="NodeGraphRunner"/> in a unit test); the node then fails cleanly.
/// </summary>
public ISubWorkflowInvoker? SubWorkflows { get; init; }
/// <summary>Nesting depth of this run: 0 top-level, +1 per sub-workflow call.</summary>
public int Depth { get; init; }
/// <summary>
/// Workflow ids from the root run down to (and including) this run, used to
/// reject recursive sub-workflow cycles before they can loop forever.
/// </summary>
public IReadOnlyList<Guid> Ancestry { get; init; } = Array.Empty<Guid>();
public IServiceProvider? Services { get; init; }
}
/// <summary>Result of running a node graph, with per-node outputs and a run-level failure.</summary>
public sealed record NodeGraphRunResult
{
public required IReadOnlyDictionary<string, IReadOnlyList<IReadOnlyList<FlowItem>>> OutputsByNode { get; init; }
/// <summary>Node ids in the order they were executed.</summary>
public required IReadOnlyList<string> ExecutionOrder { get; init; }
public NodeFailure? Failure { get; init; }
public bool Succeeded => Failure == null;
/// <summary>Items emitted by a node on a given output port (empty when absent).</summary>
public IReadOnlyList<FlowItem> OutputOf(string nodeId, int port = 0)
=> OutputsByNode.TryGetValue(nodeId, out var ports) && port >= 0 && port < ports.Count
? ports[port]
: Array.Empty<FlowItem>();
}
/// <summary>
/// Drives a <see cref="NodeGraph"/> to completion: accumulates items on each
/// input port, fires a node once every incoming edge has delivered, resolves the
/// node's parameters per item, runs the registered executor, stamps provenance
/// on produced items and forwards them along the outgoing edges.
///
/// Invariants guaranteed by <see cref="NodeGraphCompiler"/>: exactly one entry
/// node and no cycles, so a node fires exactly once.
/// </summary>
public sealed class NodeGraphRunner
{
private readonly NodeExecutorRegistry _executors;
private readonly NodeParameterInterpolator _interpolator;
private readonly NodePermissionPolicy? _permissions;
private readonly ILogger<NodeGraphRunner>? _logger;
public NodeGraphRunner(
NodeExecutorRegistry executors,
NodeParameterInterpolator interpolator,
ILogger<NodeGraphRunner>? logger = null,
NodePermissionPolicy? permissions = null)
{
_executors = executors;
_interpolator = interpolator;
_logger = logger;
_permissions = permissions;
}
public async Task<NodeGraphRunResult> RunAsync(
NodeGraph graph,
IReadOnlyList<FlowItem> seed,
NodeRunEnvironment? environment = null,
INodeRunListener? listener = null,
CancellationToken ct = default)
{
environment ??= new NodeRunEnvironment();
var outputs = new Dictionary<string, List<IReadOnlyList<FlowItem>>>(StringComparer.Ordinal);
var order = new List<string>();
// Loop-back edges do not count toward readiness: they re-trigger the loop
// node after its body has run, rather than being a normal predecessor.
var incoming = graph.Nodes.ToDictionary(
n => n.Id,
n => graph.EdgesTo(n.Id).Count(e => !e.IsLoopBack),
StringComparer.Ordinal);
var arrived = graph.Nodes.ToDictionary(n => n.Id, _ => 0, StringComparer.Ordinal);
var inputs = graph.Nodes.ToDictionary(n => n.Id, NewInputPorts, StringComparer.Ordinal);
var exhausted = new HashSet<string>(StringComparer.Ordinal);
inputs[graph.EntryNodeId][0].AddRange(seed);
var ready = new Queue<string>();
ready.Enqueue(graph.EntryNodeId);
var hasLoops = graph.Edges.Any(e => e.IsLoopBack);
var maxSteps = hasLoops
? Math.Max(1024, graph.Nodes.Count * 256)
: Math.Max(64, graph.Nodes.Count * 64);
var steps = 0;
while (ready.Count > 0)
{
ct.ThrowIfCancellationRequested();
if (++steps > maxSteps)
return Failure(outputs, order, new NodeFailure("node graph exceeded its execution budget", "budget"));
var nodeId = ready.Dequeue();
var node = graph.Require(nodeId);
if (!_executors.CanRun(node.Blueprint.Type))
return Failure(outputs, order,
new NodeFailure($"no executor is installed for node type '{node.Blueprint.Type}'", "missing_executor"));
order.Add(nodeId);
var nodeInputs = inputs[nodeId].Select(port => (IReadOnlyList<FlowItem>)port).ToList();
await NotifyStarted(listener, node, nodeInputs, ct);
// Node permissions are enforced right before invocation, so a graph
// persisted before a policy change still cannot run a barred node.
// A denial is a normal node failure: it routes to the error output or
// fails the run, exactly like an executor error.
var permission = _permissions?.Evaluate(node.Blueprint, environment.TenantId);
var outcome = permission is { Allowed: false }
? NodeExecutionOutcome.Failed(
permission.Reason ?? $"node type '{node.Blueprint.Type}' is not permitted",
permission.Code ?? "node_not_permitted")
: await InvokeAsync(node, nodeInputs, environment, outputs, graph, ct);
var nodeFailure = outcome.Failure;
IReadOnlyList<IReadOnlyList<FlowItem>>? errorPorts = null;
if (!outcome.Succeeded)
{
var errorPort = node.ErrorOutputIndex;
if (errorPort < 0 && !node.ContinueOnFail)
{
await NotifyFinished(listener, node, Array.Empty<IReadOnlyList<FlowItem>>(), nodeFailure, ct);
return Failure(outputs, order, outcome.Failure!);
}
errorPorts = BuildErrorOutputs(node, nodeInputs, outcome.Failure!, errorPort);
}
var ports = StampAndNormalize(errorPorts ?? outcome.Outputs, node);
outputs[nodeId] = ports;
await NotifyFinished(listener, node, ports, nodeFailure, ct);
if (outcome.LoopComplete)
exhausted.Add(nodeId);
Propagate(graph, nodeId, ports, inputs, arrived, incoming, ready, exhausted, outcome.LoopComplete);
// Consume this node's inputs so a loop can gather them again next
// iteration; cleared after use, never before.
arrived[nodeId] = 0;
foreach (var port in inputs[nodeId])
port.Clear();
}
_logger?.LogDebug("Node graph run finished: {Count} nodes executed", order.Count);
return new NodeGraphRunResult { OutputsByNode = Freeze(outputs), ExecutionOrder = order };
}
private async Task<NodeExecutionOutcome> InvokeAsync(
NodeGraphNode node,
IReadOnlyList<IReadOnlyList<FlowItem>> inputs,
NodeRunEnvironment environment,
IReadOnlyDictionary<string, List<IReadOnlyList<FlowItem>>> outputs,
NodeGraph graph,
CancellationToken ct)
{
var executor = _executors.Resolve(node.Blueprint.Type)
?? throw new InvalidOperationException($"no executor for '{node.Blueprint.Type}'");
var loopBackInput = graph.EdgesTo(node.Id).Any(e => e.IsLoopBack);
// Nodes with several inputs, or an explicit all-items run mode, receive the
// whole array at once; the default is one invocation per input item.
var allItems = node.EffectiveRunMode == NodeRunMode.AllItems || node.Blueprint.Inputs.Count > 1;
if (allItems)
{
var parameters = ResolveParameters(node, null, 0, environment, outputs, graph);
return await executor.RunAsync(
BuildContext(node, parameters, inputs, environment, 0, loopBackInput), ct);
}
var mainItems = inputs.Count > 0 ? inputs[0] : Array.Empty<FlowItem>();
// A node with no incoming items is skipped, except the entry node: like a
// trigger/source it runs once with an empty item so it can generate data.
var isEntry = string.Equals(node.Id, graph.EntryNodeId, StringComparison.Ordinal);
if (mainItems.Count == 0 && !isEntry)
return NodeExecutionOutcome.Empty;
if (mainItems.Count == 0)
mainItems = new[] { new FlowItem { Json = new JsonObject() } };
var accumulated = new List<List<FlowItem>>();
for (var index = 0; index < mainItems.Count; index++)
{
var parameters = ResolveParameters(node, mainItems[index], index, environment, outputs, graph);
var outcome = await executor.RunAsync(
BuildContext(node, parameters, new[] { (IReadOnlyList<FlowItem>)new[] { mainItems[index] } }, environment, index, loopBackInput),
ct);
if (!outcome.Succeeded)
return outcome;
while (accumulated.Count < outcome.Outputs.Count)
accumulated.Add(new List<FlowItem>());
for (var port = 0; port < outcome.Outputs.Count; port++)
accumulated[port].AddRange(outcome.Outputs[port]);
}
return new NodeExecutionOutcome
{
Outputs = accumulated.Select(list => (IReadOnlyList<FlowItem>)list).ToList(),
};
}
private NodeExecutionContext BuildContext(
NodeGraphNode node,
JsonObject parameters,
IReadOnlyList<IReadOnlyList<FlowItem>> inputs,
NodeRunEnvironment environment,
int itemIndex,
bool loopBackInput = false)
=> new()
{
Blueprint = node.Blueprint,
Parameters = parameters,
Inputs = inputs,
Credentials = environment.Credentials,
Environment = environment.Variables,
TenantId = environment.TenantId,
RunId = environment.RunId,
TaskId = node.Id,
NodeName = node.Id,
ItemIndex = itemIndex,
RunIndex = 0,
WorkingDirectory = environment.WorkingDirectory,
LoopBackInput = loopBackInput,
State = environment.State,
SubWorkflows = environment.SubWorkflows,
Depth = environment.Depth,
Services = environment.Services,
};
private JsonObject ResolveParameters(
NodeGraphNode node,
FlowItem? item,
int itemIndex,
NodeRunEnvironment environment,
IReadOnlyDictionary<string, List<IReadOnlyList<FlowItem>>> outputs,
NodeGraph graph)
{
var scope = new InterpolationScope
{
Item = item?.Json,
Parameters = node.Parameters,
Environment = environment.Variables,
ItemIndex = itemIndex,
RunIndex = 0,
NodeItems = name => ResolveNodeItems(name, outputs, graph),
};
return _interpolator.ResolveObject(node.Parameters, scope);
}
private static IReadOnlyList<FlowItem> ResolveNodeItems(
string name,
IReadOnlyDictionary<string, List<IReadOnlyList<FlowItem>>> outputs,
NodeGraph graph)
{
if (outputs.TryGetValue(name, out var direct) && direct.Count > 0)
return direct[0];
// Fall back to the blueprint display name, so $("Send message") also works.
var byName = graph.Nodes.FirstOrDefault(
n => string.Equals(n.Blueprint.DisplayName, name, StringComparison.Ordinal));
if (byName != null && outputs.TryGetValue(byName.Id, out var named) && named.Count > 0)
return named[0];
return Array.Empty<FlowItem>();
}
// ------------------------------------------------------------------ routing
private static void Propagate(
NodeGraph graph,
string nodeId,
IReadOnlyList<IReadOnlyList<FlowItem>> ports,
IReadOnlyDictionary<string, List<FlowItem>[]> inputs,
IDictionary<string, int> arrived,
IReadOnlyDictionary<string, int> incoming,
Queue<string> ready,
ISet<string> exhausted,
bool skipEmptyPorts)
{
for (var port = 0; port < ports.Count; port++)
{
var produced = ports[port];
// On a loop node's final ("done") invocation the loop port is empty:
// skip it so the body is not re-triggered for a no-op iteration.
if (skipEmptyPorts && produced.Count == 0)
continue;
foreach (var edge in graph.EdgesFrom(nodeId, port))
{
// An exhausted loop node is not re-triggered by its loop-back edge.
if (exhausted.Contains(edge.ToNodeId) && edge.IsLoopBack)
continue;
var targetInputs = inputs[edge.ToNodeId];
if (edge.ToInput >= 0 && edge.ToInput < targetInputs.Length)
targetInputs[edge.ToInput].AddRange(produced);
arrived[edge.ToNodeId]++;
if (arrived[edge.ToNodeId] == incoming[edge.ToNodeId])
ready.Enqueue(edge.ToNodeId);
}
}
}
// ------------------------------------------------------------------ listeners
private static Task NotifyStarted(
INodeRunListener? listener,
NodeGraphNode node,
IReadOnlyList<IReadOnlyList<FlowItem>> inputs,
CancellationToken ct)
=> listener == null ? Task.CompletedTask : listener.NodeStartedAsync(node, inputs, ct);
private static Task NotifyFinished(
INodeRunListener? listener,
NodeGraphNode node,
IReadOnlyList<IReadOnlyList<FlowItem>> outputs,
NodeFailure? failure,
CancellationToken ct)
=> listener == null ? Task.CompletedTask : listener.NodeFinishedAsync(node, outputs, failure, ct);
private List<IReadOnlyList<FlowItem>> BuildErrorOutputs(
NodeGraphNode node,
IReadOnlyList<IReadOnlyList<FlowItem>> inputs,
NodeFailure failure,
int errorPort)
{
var portCount = Math.Max(1, node.Blueprint.Outputs.Count);
var ports = Enumerable.Range(0, portCount).Select(_ => (IReadOnlyList<FlowItem>)new List<FlowItem>()).ToList();
if (errorPort < 0 || errorPort >= ports.Count)
return ports;
var errorJson = new JsonObject
{
["message"] = failure.Message,
["code"] = failure.Code,
["description"] = failure.Description,
["httpStatus"] = failure.HttpStatus,
};
var source = inputs.FirstOrDefault(list => list.Count > 0);
var items = new List<FlowItem>();
if (source == null || source.Count == 0)
{
items.Add(new FlowItem { Json = new JsonObject { ["error"] = errorJson } });
}
else
{
foreach (var item in source)
{
var json = (JsonObject)item.Json.DeepClone();
json["error"] = errorJson.DeepClone();
items.Add(new FlowItem { Json = json, Binary = item.Binary, Origin = item.Origin });
}
}
ports[errorPort] = items;
return ports;
}
// ------------------------------------------------------------------ shaping
private static List<IReadOnlyList<FlowItem>> StampAndNormalize(
IReadOnlyList<IReadOnlyList<FlowItem>> produced,
NodeGraphNode node)
{
var portCount = Math.Max(1, node.Blueprint.Outputs.Count);
var ports = new List<IReadOnlyList<FlowItem>>(portCount);
for (var port = 0; port < portCount; port++)
{
var items = port < produced.Count ? produced[port] : Array.Empty<FlowItem>();
var stamped = new List<FlowItem>(items.Count);
for (var index = 0; index < items.Count; index++)
{
var item = items[index];
stamped.Add(item.Origin == null
? item with { Origin = new FlowItemOrigin(node.Id, port, 0, index) }
: item);
}
ports.Add(stamped);
}
return ports;
}
private static List<FlowItem>[] NewInputPorts(NodeGraphNode node)
{
var ports = new List<FlowItem>[Math.Max(1, node.Blueprint.Inputs.Count)];
for (var i = 0; i < ports.Length; i++)
ports[i] = new List<FlowItem>();
return ports;
}
private static IReadOnlyDictionary<string, IReadOnlyList<IReadOnlyList<FlowItem>>> Freeze(
IReadOnlyDictionary<string, List<IReadOnlyList<FlowItem>>> outputs)
=> outputs.ToDictionary(
pair => pair.Key,
pair => (IReadOnlyList<IReadOnlyList<FlowItem>>)pair.Value,
StringComparer.Ordinal);
private static NodeGraphRunResult Failure(
IReadOnlyDictionary<string, List<IReadOnlyList<FlowItem>>> outputs,
IReadOnlyList<string> order,
NodeFailure failure)
=> new()
{
OutputsByNode = Freeze(outputs),
ExecutionOrder = order.ToList(),
Failure = failure,
};
}