Skip to main content

dfir_lang/graph/
meta_graph.rs

1#![warn(missing_docs)]
2
3extern crate proc_macro;
4
5use std::collections::{BTreeMap, BTreeSet};
6use std::fmt::Debug;
7use std::iter::FusedIterator;
8
9use itertools::Itertools;
10use proc_macro2::{Ident, Literal, Span, TokenStream};
11use quote::{ToTokens, format_ident, quote, quote_spanned};
12use serde::{Deserialize, Serialize};
13use slotmap::{Key, SecondaryMap, SlotMap, SparseSecondaryMap};
14use syn::spanned::Spanned;
15
16use super::graph_write::{Dot, GraphWrite, Mermaid};
17use super::ops::{
18    DelayType, FloType, OPERATORS, OperatorWriteOutput, WriteContextArgs, find_op_op_constraints,
19    null_write_iterator_fn,
20};
21use super::{
22    CONTEXT, Color, DiMulGraph, GRAPH, GraphEdgeId, GraphLoopId, GraphNode, GraphNodeId,
23    GraphSubgraphId, HANDOFF_NODE_STR, HandoffKind, MODULE_BOUNDARY_NODE_STR, OperatorInstance,
24    PortIndexValue, SINGLETON_SLOT_NODE_STR, Varname, change_spans, get_operator_generics,
25};
26use crate::diagnostic::{Diagnostic, Diagnostics, Level};
27use crate::pretty_span::{PrettyRowCol, PrettySpan};
28use crate::process_singletons;
29
30/// A resolved handoff reference: the target node ID plus mutability and access group info.
31#[derive(Clone, Debug, Serialize, Deserialize)]
32pub struct ResolvedHandoffRef {
33    /// The resolved target node ID (`None` if unresolved/error).
34    pub node_id: Option<GraphNodeId>,
35    /// Whether this is a mutable reference (`#mut var`).
36    pub is_mut: bool,
37    /// Optional access group for ordering (`#{N} var`).
38    pub access_group: Option<u32>,
39}
40
41/// An abstract "meta graph" representation of a DFIR graph.
42///
43/// Can be with or without subgraph partitioning, stratification, and handoff insertion. This is
44/// the meta graph used for generating Rust source code in macros from DFIR sytnax.
45///
46/// This struct has a lot of methods for manipulating the graph, vaguely grouped together in
47/// separate `impl` blocks. You might notice a few particularly specific arbitray-seeming methods
48/// in here--those are just what was needed for the compilation algorithms. If you need another
49/// method then add it.
50#[derive(Default, Debug, Serialize, Deserialize)]
51pub struct DfirGraph {
52    /// Each node type (operator or handoff).
53    nodes: SlotMap<GraphNodeId, GraphNode>,
54
55    /// Instance data corresponding to each operator node.
56    /// This field will be empty after deserialization.
57    #[serde(skip)]
58    operator_instances: SecondaryMap<GraphNodeId, OperatorInstance>,
59    /// Debugging/tracing tag for each operator node.
60    operator_tag: SecondaryMap<GraphNodeId, String>,
61    /// Graph data structure (two-way adjacency list).
62    graph: DiMulGraph<GraphNodeId, GraphEdgeId>,
63    /// Input and output port for each edge.
64    ports: SecondaryMap<GraphEdgeId, (PortIndexValue, PortIndexValue)>,
65
66    /// Which loop a node belongs to (or none for top-level).
67    node_loops: SecondaryMap<GraphNodeId, GraphLoopId>,
68    /// Which nodes belong to each loop.
69    loop_nodes: SlotMap<GraphLoopId, Vec<GraphNodeId>>,
70    /// For the loop, what is its parent (`None` for top-level).
71    loop_parent: SparseSecondaryMap<GraphLoopId, GraphLoopId>,
72    /// What loops are at the root.
73    root_loops: Vec<GraphLoopId>,
74    /// For the loop, what are its child loops.
75    loop_children: SecondaryMap<GraphLoopId, Vec<GraphLoopId>>,
76
77    /// Which subgraph each node belongs to.
78    node_subgraph: SecondaryMap<GraphNodeId, GraphSubgraphId>,
79
80    /// Which nodes belong to each subgraph.
81    subgraph_nodes: SlotMap<GraphSubgraphId, Vec<GraphNodeId>>,
82    /// Subgraph IDs in topological sort order (set during partitioning).
83    subgraph_toposort: Vec<GraphSubgraphId>,
84
85    /// Resolved handoff varnames references, per node.
86    node_handoff_references: SparseSecondaryMap<GraphNodeId, Vec<ResolvedHandoffRef>>,
87    /// What variable name each graph node belongs to (if any). For debugging (graph writing) purposes only.
88    node_varnames: SparseSecondaryMap<GraphNodeId, Varname>,
89
90    /// Delay type for handoff nodes that represent tick-boundary back-edges.
91    /// Set by `order_subgraphs` for `defer_tick` / `defer_tick_lazy`, either on handoff nodes
92    /// it injects or on existing handoff nodes that it marks as tick-boundary back-edges.
93    handoff_delay_type: SparseSecondaryMap<GraphNodeId, DelayType>,
94}
95
96/// Basic methods.
97impl DfirGraph {
98    /// Create a new empty graph.
99    pub fn new() -> Self {
100        Default::default()
101    }
102}
103
104/// Node methods.
105impl DfirGraph {
106    /// Get a node with its operator instance (if applicable).
107    pub fn node(&self, node_id: GraphNodeId) -> &GraphNode {
108        self.nodes.get(node_id).expect("Node not found.")
109    }
110
111    /// Get the `OperatorInstance` for a given node. Node must be an operator and have an
112    /// `OperatorInstance` present, otherwise will return `None`.
113    ///
114    /// Note that no operator instances will be persent after deserialization.
115    pub fn node_op_inst(&self, node_id: GraphNodeId) -> Option<&OperatorInstance> {
116        self.operator_instances.get(node_id)
117    }
118
119    /// Get the debug variable name attached to a graph node.
120    pub fn node_varname(&self, node_id: GraphNodeId) -> Option<&Varname> {
121        self.node_varnames.get(node_id)
122    }
123
124    /// Get subgraph for node.
125    pub fn node_subgraph(&self, node_id: GraphNodeId) -> Option<GraphSubgraphId> {
126        self.node_subgraph.get(node_id).copied()
127    }
128
129    /// Degree into a node, i.e. the number of predecessors.
130    pub fn node_degree_in(&self, node_id: GraphNodeId) -> usize {
131        self.graph.degree_in(node_id)
132    }
133
134    /// Degree out of a node, i.e. the number of successors.
135    pub fn node_degree_out(&self, node_id: GraphNodeId) -> usize {
136        self.graph.degree_out(node_id)
137    }
138
139    /// Successors, iterator of `(GraphEdgeId, GraphNodeId)` of outgoing edges.
140    pub fn node_successors(
141        &self,
142        src: GraphNodeId,
143    ) -> impl '_
144    + DoubleEndedIterator<Item = (GraphEdgeId, GraphNodeId)>
145    + ExactSizeIterator
146    + FusedIterator
147    + Clone
148    + Debug {
149        self.graph.successors(src)
150    }
151
152    /// Predecessors, iterator of `(GraphEdgeId, GraphNodeId)` of incoming edges.
153    pub fn node_predecessors(
154        &self,
155        dst: GraphNodeId,
156    ) -> impl '_
157    + DoubleEndedIterator<Item = (GraphEdgeId, GraphNodeId)>
158    + ExactSizeIterator
159    + FusedIterator
160    + Clone
161    + Debug {
162        self.graph.predecessors(dst)
163    }
164
165    /// Successor edges, iterator of `GraphEdgeId` of outgoing edges.
166    pub fn node_successor_edges(
167        &self,
168        src: GraphNodeId,
169    ) -> impl '_
170    + DoubleEndedIterator<Item = GraphEdgeId>
171    + ExactSizeIterator
172    + FusedIterator
173    + Clone
174    + Debug {
175        self.graph.successor_edges(src)
176    }
177
178    /// Predecessor edges, iterator of `GraphEdgeId` of incoming edges.
179    pub fn node_predecessor_edges(
180        &self,
181        dst: GraphNodeId,
182    ) -> impl '_
183    + DoubleEndedIterator<Item = GraphEdgeId>
184    + ExactSizeIterator
185    + FusedIterator
186    + Clone
187    + Debug {
188        self.graph.predecessor_edges(dst)
189    }
190
191    /// Successor nodes, iterator of `GraphNodeId`.
192    pub fn node_successor_nodes(
193        &self,
194        src: GraphNodeId,
195    ) -> impl '_
196    + DoubleEndedIterator<Item = GraphNodeId>
197    + ExactSizeIterator
198    + FusedIterator
199    + Clone
200    + Debug {
201        self.graph.successor_vertices(src)
202    }
203
204    /// Predecessor nodes, iterator of `GraphNodeId`.
205    pub fn node_predecessor_nodes(
206        &self,
207        dst: GraphNodeId,
208    ) -> impl '_
209    + DoubleEndedIterator<Item = GraphNodeId>
210    + ExactSizeIterator
211    + FusedIterator
212    + Clone
213    + Debug {
214        self.graph.predecessor_vertices(dst)
215    }
216
217    /// Iterator of node IDs `GraphNodeId`.
218    pub fn node_ids(&self) -> slotmap::basic::Keys<'_, GraphNodeId, GraphNode> {
219        self.nodes.keys()
220    }
221
222    /// Iterator over `(GraphNodeId, &Node)` pairs.
223    pub fn nodes(&self) -> slotmap::basic::Iter<'_, GraphNodeId, GraphNode> {
224        self.nodes.iter()
225    }
226
227    /// Insert a node, assigning the given varname.
228    pub fn insert_node(
229        &mut self,
230        node: GraphNode,
231        varname_opt: Option<Ident>,
232        loop_opt: Option<GraphLoopId>,
233    ) -> GraphNodeId {
234        let node_id = self.nodes.insert(node);
235        if let Some(varname) = varname_opt {
236            self.node_varnames.insert(node_id, Varname(varname));
237        }
238        if let Some(loop_id) = loop_opt {
239            self.node_loops.insert(node_id, loop_id);
240            self.loop_nodes[loop_id].push(node_id);
241        }
242        node_id
243    }
244
245    /// Insert an operator instance for the given node. Panics if already set.
246    pub fn insert_node_op_inst(&mut self, node_id: GraphNodeId, op_inst: OperatorInstance) {
247        assert!(matches!(
248            self.nodes.get(node_id),
249            Some(GraphNode::Operator(_))
250        ));
251        let old_inst = self.operator_instances.insert(node_id, op_inst);
252        assert!(old_inst.is_none());
253    }
254
255    /// Assign all operator instances if not set. Write diagnostic messages/errors into `diagnostics`.
256    pub fn insert_node_op_insts_all(&mut self, diagnostics: &mut Diagnostics) {
257        // Handle all nodes in two phases, since the helper methods take total ownership of `&self`.
258        // Possible to do in one phase, but would require accessing fields directly for partial mutable ownership.
259
260        // Collect operator instances, then assign.
261        let mut op_insts = Vec::new();
262        // Collect nodes that should be lowered to handoffs (the `handoff()`/`singleton()` pseudo-operators).
263        let mut handoff_nodes: Vec<(GraphNodeId, HandoffKind, Span)> = Vec::new();
264
265        for (node_id, node) in self.nodes() {
266            let GraphNode::Operator(operator) = node else {
267                continue;
268            };
269            if self.node_op_inst(node_id).is_some() {
270                continue;
271            };
272
273            // Recognize `handoff()`/`singleton()` pseudo-operators and lower to GraphNode::Handoff.
274            let handoff_kind = match &*operator.name_string() {
275                "handoff" => Some(HandoffKind::Vec),
276                "singleton" => Some(HandoffKind::Singleton),
277                "optional" => Some(HandoffKind::Optional),
278                _ => None,
279            };
280            if let Some(kind) = handoff_kind {
281                if !operator.args.is_empty() {
282                    diagnostics.push(Diagnostic::spanned(
283                        operator.path.span(),
284                        Level::Error,
285                        format!("`{}` takes no arguments.", operator.name_string()),
286                    ));
287                }
288                if operator.type_arguments().is_some() {
289                    diagnostics.push(Diagnostic::spanned(
290                        operator.path.span(),
291                        Level::Error,
292                        format!("`{}` takes no generic arguments.", operator.name_string()),
293                    ));
294                }
295                handoff_nodes.push((node_id, kind, operator.path.span()));
296                continue;
297            }
298
299            // Op constraints.
300            let Some(op_constraints) = find_op_op_constraints(operator) else {
301                diagnostics.push(Diagnostic::spanned(
302                    operator.path.span(),
303                    Level::Error,
304                    format!("Unknown operator `{}`", operator.name_string()),
305                ));
306                continue;
307            };
308
309            // Input and output ports.
310            let (input_ports, output_ports) = {
311                let mut input_edges: Vec<(&PortIndexValue, GraphNodeId)> = self
312                    .node_predecessors(node_id)
313                    .map(|(edge_id, pred_id)| (self.edge_ports(edge_id).1, pred_id))
314                    .collect();
315                // Ensure sorted by port index.
316                input_edges.sort();
317                let input_ports: Vec<PortIndexValue> = input_edges
318                    .into_iter()
319                    .map(|(port, _pred)| port)
320                    .cloned()
321                    .collect();
322
323                // Collect output arguments (successors).
324                let mut output_edges: Vec<(&PortIndexValue, GraphNodeId)> = self
325                    .node_successors(node_id)
326                    .map(|(edge_id, succ)| (self.edge_ports(edge_id).0, succ))
327                    .collect();
328                // Ensure sorted by port index.
329                output_edges.sort();
330                let output_ports: Vec<PortIndexValue> = output_edges
331                    .into_iter()
332                    .map(|(port, _succ)| port)
333                    .cloned()
334                    .collect();
335
336                (input_ports, output_ports)
337            };
338
339            // Generic arguments.
340            let generics = get_operator_generics(diagnostics, operator);
341            // Generic argument errors.
342            {
343                // Span of `generic_args` (if it exists), otherwise span of the operator name.
344                let generics_span = generics
345                    .generic_args
346                    .as_ref()
347                    .map(Spanned::span)
348                    .unwrap_or_else(|| operator.path.span());
349
350                if !op_constraints
351                    .persistence_args
352                    .contains(&generics.persistence_args.len())
353                {
354                    diagnostics.push(Diagnostic::spanned(
355                        generics.persistence_args_span().unwrap_or(generics_span),
356                        Level::Error,
357                        format!(
358                            "`{}` should have {} persistence lifetime arguments, actually has {}.",
359                            op_constraints.name,
360                            op_constraints.persistence_args.human_string(),
361                            generics.persistence_args.len()
362                        ),
363                    ));
364                }
365                if !op_constraints.type_args.contains(&generics.type_args.len()) {
366                    diagnostics.push(Diagnostic::spanned(
367                        generics.type_args_span().unwrap_or(generics_span),
368                        Level::Error,
369                        format!(
370                            "`{}` should have {} generic type arguments, actually has {}.",
371                            op_constraints.name,
372                            op_constraints.type_args.human_string(),
373                            generics.type_args.len()
374                        ),
375                    ));
376                }
377            }
378
379            op_insts.push((
380                node_id,
381                OperatorInstance {
382                    op_constraints,
383                    input_ports,
384                    output_ports,
385                    singletons_referenced: operator.singletons_referenced.clone(),
386                    generics,
387                    arguments_pre: operator.args.clone(),
388                    arguments_raw: operator.args_raw.clone(),
389                },
390            ));
391        }
392
393        for (node_id, op_inst) in op_insts {
394            self.insert_node_op_inst(node_id, op_inst);
395        }
396
397        // Replace pseudo-operator nodes with GraphNode::Handoff.
398        for (node_id, kind, span) in handoff_nodes {
399            self.nodes[node_id] = GraphNode::Handoff {
400                kind,
401                src_span: span,
402                dst_span: span,
403            };
404        }
405    }
406
407    /// Inserts a node between two existing nodes connected by the given `edge_id`.
408    ///
409    /// `edge`: (src, dst, dst_idx)
410    ///
411    /// Before: A (src) ------------> B (dst)
412    /// After:  A (src) -> X (new) -> B (dst)
413    ///
414    /// Returns the ID of X & ID of edge OUT of X.
415    ///
416    /// Note that both the edges will be new and `edge_id` will be removed. Both new edges will
417    /// get the edge type of the original edge.
418    pub fn insert_intermediate_node(
419        &mut self,
420        edge_id: GraphEdgeId,
421        new_node: GraphNode,
422    ) -> (GraphNodeId, GraphEdgeId) {
423        let span = Some(new_node.span());
424
425        // Make corresponding operator instance (if `node` is an operator).
426        let op_inst_opt = 'oc: {
427            let GraphNode::Operator(operator) = &new_node else {
428                break 'oc None;
429            };
430            let Some(op_constraints) = find_op_op_constraints(operator) else {
431                break 'oc None;
432            };
433            let (input_port, output_port) = self.ports.get(edge_id).cloned().unwrap();
434
435            let mut dummy_diagnostics = Diagnostics::new();
436            let generics = get_operator_generics(&mut dummy_diagnostics, operator);
437            assert!(dummy_diagnostics.is_empty());
438
439            Some(OperatorInstance {
440                op_constraints,
441                input_ports: vec![input_port],
442                output_ports: vec![output_port],
443                singletons_referenced: operator.singletons_referenced.clone(),
444                generics,
445                arguments_pre: operator.args.clone(),
446                arguments_raw: operator.args_raw.clone(),
447            })
448        };
449
450        // Insert new `node`.
451        let node_id = self.nodes.insert(new_node);
452        // Insert corresponding `OperatorInstance` if applicable.
453        if let Some(op_inst) = op_inst_opt {
454            self.operator_instances.insert(node_id, op_inst);
455        }
456        // Update edges to insert node within `edge_id`.
457        let (e0, e1) = self
458            .graph
459            .insert_intermediate_vertex(node_id, edge_id)
460            .unwrap();
461
462        // Update corresponding ports.
463        let (src_idx, dst_idx) = self.ports.remove(edge_id).unwrap();
464        self.ports
465            .insert(e0, (src_idx, PortIndexValue::Elided(span)));
466        self.ports
467            .insert(e1, (PortIndexValue::Elided(span), dst_idx));
468
469        (node_id, e1)
470    }
471
472    /// Remove the node `node_id` but preserves and connects the single predecessor and single successor.
473    /// Panics if the node does not have exactly one predecessor and one successor, or is not in the graph.
474    pub fn remove_intermediate_node(&mut self, node_id: GraphNodeId) {
475        assert_eq!(
476            1,
477            self.node_degree_in(node_id),
478            "Removed intermediate node must have one predecessor"
479        );
480        assert_eq!(
481            1,
482            self.node_degree_out(node_id),
483            "Removed intermediate node must have one successor"
484        );
485        assert!(
486            self.node_subgraph.is_empty() && self.subgraph_nodes.is_empty(),
487            "Should not remove intermediate node after subgraph partitioning"
488        );
489
490        assert!(self.nodes.remove(node_id).is_some());
491        let (new_edge_id, (pred_edge_id, succ_edge_id)) =
492            self.graph.remove_intermediate_vertex(node_id).unwrap();
493        self.operator_instances.remove(node_id);
494        self.node_varnames.remove(node_id);
495
496        let (src_port, _) = self.ports.remove(pred_edge_id).unwrap();
497        let (_, dst_port) = self.ports.remove(succ_edge_id).unwrap();
498        self.ports.insert(new_edge_id, (src_port, dst_port));
499    }
500
501    /// Helper method: determine the "color" (pull vs push) of a node based on its in and out degree,
502    /// excluding reference edges. If linear (1 in, 1 out), color is `None`, indicating it can be
503    /// either push or pull.
504    ///
505    /// Note that this does NOT consider `DelayType` barriers (which generally implies `Pull`).
506    pub(crate) fn node_color(&self, node_id: GraphNodeId) -> Option<Color> {
507        if matches!(self.node(node_id), GraphNode::Handoff { .. }) {
508            return Some(Color::Hoff);
509        }
510
511        // TODO(shadaj): this is a horrible hack
512        if let GraphNode::Operator(op) = self.node(node_id)
513            && (op.name_string() == "resolve_futures_blocking"
514                || op.name_string() == "resolve_futures_blocking_ordered")
515        {
516            return Some(Color::Push);
517        }
518
519        // In-degree, excluding ref-edges.
520        let inn_degree = self.node_predecessor_nodes(node_id).len();
521        // Out-degree excluding ref-edges.
522        let out_degree = self.node_successor_nodes(node_id).len();
523
524        match (inn_degree, out_degree) {
525            (0, 0) => None, // Generally should not happen, "Degenerate subgraph detected".
526            (0, 1) => Some(Color::Pull),
527            (1, 0) => Some(Color::Push),
528            (1, 1) => None, // Linear, can be either push or pull.
529            (_many, 0 | 1) => Some(Color::Pull),
530            (0 | 1, _many) => Some(Color::Push),
531            (_many, _to_many) => Some(Color::Comp),
532        }
533    }
534
535    /// Set the operator tag (for debugging/tracing).
536    pub fn set_operator_tag(&mut self, node_id: GraphNodeId, tag: String) {
537        self.operator_tag.insert(node_id, tag);
538    }
539}
540
541/// Handoff references.
542impl DfirGraph {
543    /// Set the handoff references for the `node_id` operator. Each reference corresponds to the
544    /// same index in the [`crate::parse::Operator::singletons_referenced`] vec.
545    pub fn set_node_handoff_references(
546        &mut self,
547        node_id: GraphNodeId,
548        singletons_referenced: Vec<ResolvedHandoffRef>,
549    ) -> Option<Vec<ResolvedHandoffRef>> {
550        self.node_handoff_references
551            .insert(node_id, singletons_referenced)
552    }
553
554    /// Gets the handoff references for a node. Returns an empty slice for non-operators and
555    /// operators that do not reference handoffs.
556    pub fn node_handoff_references(&self, node_id: GraphNodeId) -> &[ResolvedHandoffRef] {
557        self.node_handoff_references
558            .get(node_id)
559            .map(std::ops::Deref::deref)
560            .unwrap_or_default()
561    }
562
563    /// Collect all refs, grouped by the handoff they're pointing at, then by the access group idx `Option<u32>`.
564    pub fn node_handoff_reference_groups(&self) -> NodeHandoffReferenceGroups<'_> {
565        let mut handoff_references = NodeHandoffReferenceGroups::new();
566        for node_id in self.node_ids() {
567            if let GraphNode::Operator(operator) = self.node(node_id) {
568                let resolved = self.node_handoff_references(node_id);
569                for (resolved_ref, ref_token) in
570                    resolved.iter().zip(operator.singletons_referenced.iter())
571                {
572                    if let Some(target_nid) = resolved_ref.node_id {
573                        handoff_references
574                            .entry(target_nid)
575                            .or_default()
576                            .entry(resolved_ref.access_group)
577                            .or_default()
578                            .push((node_id, resolved_ref, ref_token.span()));
579                    }
580                }
581            }
582        }
583        handoff_references
584    }
585}
586
587/// Per-node handoff references, in turn grouped by access group.
588/// Map: handoff_node_id -> access_group -> (source `GraphNodeId`, `ResolvedHandoffRef`, `#ref` span)
589pub type NodeHandoffReferenceGroups<'a> =
590    BTreeMap<GraphNodeId, BTreeMap<Option<u32>, Vec<(GraphNodeId, &'a ResolvedHandoffRef, Span)>>>;
591
592/// Module methods.
593impl DfirGraph {
594    /// When modules are imported into a flat graph, they come with an input and output ModuleBoundary node.
595    /// The partitioner doesn't understand these nodes and will panic if it encounters them.
596    /// merge_modules removes them from the graph, stitching the input and ouput sides of the ModuleBondaries based on their ports
597    /// For example:
598    ///     source_iter([]) -> \[myport\]ModuleBoundary(input)\[my_port\] -> map(|x| x) -> ModuleBoundary(output) -> null();
599    /// in the above eaxmple, the \[myport\] port will be used to connect the source_iter with the map that is inside of the module.
600    /// The output module boundary has elided ports, this is also used to match up the input/output across the module boundary.
601    pub fn merge_modules(&mut self) -> Result<(), Diagnostic> {
602        let mod_bound_nodes = self
603            .nodes()
604            .filter(|(_nid, node)| matches!(node, GraphNode::ModuleBoundary { .. }))
605            .map(|(nid, _node)| nid)
606            .collect::<Vec<_>>();
607
608        for mod_bound_node in mod_bound_nodes {
609            self.remove_module_boundary(mod_bound_node)?;
610        }
611
612        Ok(())
613    }
614
615    /// see `merge_modules`
616    /// This function removes a singular module boundary from the graph and performs the necessary stitching to fix the graph afterward.
617    /// `merge_modules` calls this function for each module boundary in the graph.
618    fn remove_module_boundary(&mut self, mod_bound_node: GraphNodeId) -> Result<(), Diagnostic> {
619        assert!(
620            self.node_subgraph.is_empty() && self.subgraph_nodes.is_empty(),
621            "Should not remove intermediate node after subgraph partitioning"
622        );
623
624        let mut mod_pred_ports = BTreeMap::new();
625        let mut mod_succ_ports = BTreeMap::new();
626
627        for mod_out_edge in self.node_predecessor_edges(mod_bound_node) {
628            let (pred_port, succ_port) = self.edge_ports(mod_out_edge);
629            mod_pred_ports.insert(succ_port.clone(), (mod_out_edge, pred_port.clone()));
630        }
631
632        for mod_inn_edge in self.node_successor_edges(mod_bound_node) {
633            let (pred_port, succ_port) = self.edge_ports(mod_inn_edge);
634            mod_succ_ports.insert(pred_port.clone(), (mod_inn_edge, succ_port.clone()));
635        }
636
637        if mod_pred_ports.keys().collect::<BTreeSet<_>>()
638            != mod_succ_ports.keys().collect::<BTreeSet<_>>()
639        {
640            // get module boundary node
641            let GraphNode::ModuleBoundary { input, import_expr } = self.node(mod_bound_node) else {
642                panic!();
643            };
644
645            if *input {
646                return Err(Diagnostic {
647                    span: *import_expr,
648                    level: Level::Error,
649                    message: format!(
650                        "The ports into the module did not match. input: {:?}, expected: {:?}",
651                        mod_pred_ports.keys().map(|x| x.to_string()).join(", "),
652                        mod_succ_ports.keys().map(|x| x.to_string()).join(", ")
653                    ),
654                });
655            } else {
656                return Err(Diagnostic {
657                    span: *import_expr,
658                    level: Level::Error,
659                    message: format!(
660                        "The ports out of the module did not match. output: {:?}, expected: {:?}",
661                        mod_succ_ports.keys().map(|x| x.to_string()).join(", "),
662                        mod_pred_ports.keys().map(|x| x.to_string()).join(", "),
663                    ),
664                });
665            }
666        }
667
668        for (port, (pred_edge, pred_port)) in mod_pred_ports {
669            let (succ_edge, succ_port) = mod_succ_ports.remove(&port).unwrap();
670
671            let (src, _) = self.edge(pred_edge);
672            let (_, dst) = self.edge(succ_edge);
673            self.remove_edge(pred_edge);
674            self.remove_edge(succ_edge);
675
676            let new_edge_id = self.graph.insert_edge(src, dst);
677            self.ports.insert(new_edge_id, (pred_port, succ_port));
678        }
679
680        self.graph.remove_vertex(mod_bound_node);
681        self.nodes.remove(mod_bound_node);
682
683        Ok(())
684    }
685}
686
687/// Edge methods.
688impl DfirGraph {
689    /// Get the `src` and `dst` for an edge: `(src GraphNodeId, dst GraphNodeId)`.
690    pub fn edge(&self, edge_id: GraphEdgeId) -> (GraphNodeId, GraphNodeId) {
691        let (src, dst) = self.graph.edge(edge_id).expect("Edge not found.");
692        (src, dst)
693    }
694
695    /// Get the source and destination ports for an edge: `(src &PortIndexValue, dst &PortIndexValue)`.
696    pub fn edge_ports(&self, edge_id: GraphEdgeId) -> (&PortIndexValue, &PortIndexValue) {
697        let (src_port, dst_port) = self.ports.get(edge_id).expect("Edge not found.");
698        (src_port, dst_port)
699    }
700
701    /// Iterator of all edge IDs `GraphEdgeId`.
702    pub fn edge_ids(&self) -> slotmap::basic::Keys<'_, GraphEdgeId, (GraphNodeId, GraphNodeId)> {
703        self.graph.edge_ids()
704    }
705
706    /// Iterator over all edges: `(GraphEdgeId, (src GraphNodeId, dst GraphNodeId))`.
707    pub fn edges(
708        &self,
709    ) -> impl '_
710    + ExactSizeIterator<Item = (GraphEdgeId, (GraphNodeId, GraphNodeId))>
711    + FusedIterator
712    + Clone
713    + Debug {
714        self.graph.edges()
715    }
716
717    /// Insert an edge between nodes thru the given ports.
718    pub fn insert_edge(
719        &mut self,
720        src: GraphNodeId,
721        src_port: PortIndexValue,
722        dst: GraphNodeId,
723        dst_port: PortIndexValue,
724    ) -> GraphEdgeId {
725        let edge_id = self.graph.insert_edge(src, dst);
726        self.ports.insert(edge_id, (src_port, dst_port));
727        edge_id
728    }
729
730    /// Removes an edge and its corresponding ports and edge type info.
731    pub fn remove_edge(&mut self, edge: GraphEdgeId) {
732        let (_src, _dst) = self.graph.remove_edge(edge).unwrap();
733        let (_src_port, _dst_port) = self.ports.remove(edge).unwrap();
734    }
735}
736
737/// Subgraph methods.
738impl DfirGraph {
739    /// Nodes belonging to the given subgraph.
740    pub fn subgraph(&self, subgraph_id: GraphSubgraphId) -> &Vec<GraphNodeId> {
741        self.subgraph_nodes
742            .get(subgraph_id)
743            .expect("Subgraph not found.")
744    }
745
746    /// Iterator over all subgraph IDs.
747    pub fn subgraph_ids(&self) -> slotmap::basic::Keys<'_, GraphSubgraphId, Vec<GraphNodeId>> {
748        self.subgraph_nodes.keys()
749    }
750
751    /// Subgraph IDs in topological sort order.
752    pub fn subgraph_toposort(&self) -> &[GraphSubgraphId] {
753        &self.subgraph_toposort
754    }
755
756    /// Set the topological sort order for subgraphs.
757    pub fn set_subgraph_toposort(&mut self, order: Vec<GraphSubgraphId>) {
758        self.subgraph_toposort = order;
759    }
760
761    /// Iterator over all subgraphs, ID and members: `(GraphSubgraphId, Vec<GraphNodeId>)`.
762    pub fn subgraphs(&self) -> slotmap::basic::Iter<'_, GraphSubgraphId, Vec<GraphNodeId>> {
763        self.subgraph_nodes.iter()
764    }
765
766    /// Create a subgraph consisting of `node_ids`. Returns an error if any of the nodes are already in a subgraph.
767    pub fn insert_subgraph(
768        &mut self,
769        node_ids: Vec<GraphNodeId>,
770    ) -> Result<GraphSubgraphId, (GraphNodeId, GraphSubgraphId)> {
771        // Check none are already in subgraphs
772        for &node_id in node_ids.iter() {
773            if let Some(&old_sg_id) = self.node_subgraph.get(node_id) {
774                return Err((node_id, old_sg_id));
775            }
776        }
777        let subgraph_id = self.subgraph_nodes.insert_with_key(|sg_id| {
778            for &node_id in node_ids.iter() {
779                self.node_subgraph.insert(node_id, sg_id);
780            }
781            node_ids
782        });
783
784        Ok(subgraph_id)
785    }
786
787    /// Removes a node from its subgraph. Returns true if the node was in a subgraph.
788    pub fn remove_from_subgraph(&mut self, node_id: GraphNodeId) -> bool {
789        if let Some(old_sg_id) = self.node_subgraph.remove(node_id) {
790            self.subgraph_nodes[old_sg_id].retain(|&other_node_id| other_node_id != node_id);
791            true
792        } else {
793            false
794        }
795    }
796
797    /// Gets the delay type for a handoff node, if set.
798    pub fn handoff_delay_type(&self, node_id: GraphNodeId) -> Option<DelayType> {
799        self.handoff_delay_type.get(node_id).copied()
800    }
801
802    /// Sets the delay type for a handoff node.
803    pub fn set_handoff_delay_type(&mut self, node_id: GraphNodeId, delay_type: DelayType) {
804        self.handoff_delay_type.insert(node_id, delay_type);
805    }
806
807    /// Helper: finds the first index in `subgraph_nodes` where it transitions from pull to push.
808    fn find_pull_to_push_idx(&self, subgraph_nodes: &[GraphNodeId]) -> usize {
809        subgraph_nodes
810            .iter()
811            .position(|&node_id| {
812                self.node_color(node_id)
813                    .is_some_and(|color| Color::Pull != color)
814            })
815            .unwrap_or(subgraph_nodes.len())
816    }
817}
818
819/// Display/output methods.
820impl DfirGraph {
821    /// Helper to generate a deterministic `Ident` for the given node.
822    fn node_as_ident(&self, node_id: GraphNodeId, is_pred: bool) -> Ident {
823        let name = match &self.nodes[node_id] {
824            GraphNode::Operator(_) => format!("op_{:?}", node_id.data()),
825            GraphNode::Handoff {
826                kind: HandoffKind::Vec,
827                ..
828            } => format!(
829                "hoff_{:?}_{}",
830                node_id.data(),
831                if is_pred { "recv" } else { "send" }
832            ),
833            GraphNode::Handoff {
834                kind: HandoffKind::Singleton | HandoffKind::Optional,
835                ..
836            } => format!(
837                "singleton_{:?}_{}",
838                node_id.data(),
839                if is_pred { "recv" } else { "send" }
840            ),
841            GraphNode::ModuleBoundary { .. } => panic!(),
842        };
843        let span = match (is_pred, &self.nodes[node_id]) {
844            (_, GraphNode::Operator(operator)) => operator.span(),
845            (true, &GraphNode::Handoff { src_span, .. }) => src_span,
846            (false, &GraphNode::Handoff { dst_span, .. }) => dst_span,
847            (_, GraphNode::ModuleBoundary { .. }) => panic!(),
848        };
849        Ident::new(&name, span)
850    }
851
852    /// Helper to generate the main buffer `Ident` for a handoff node.
853    fn hoff_buf_ident(&self, hoff_id: GraphNodeId, span: Span) -> Ident {
854        Ident::new(&format!("hoff_{:?}_buf", hoff_id.data()), span)
855    }
856
857    /// Helper to generate the back (double-buffer) `Ident` for a handoff node.
858    fn hoff_back_ident(&self, hoff_id: GraphNodeId, span: Span) -> Ident {
859        Ident::new(&format!("hoff_{:?}_back", hoff_id.data()), span)
860    }
861
862    /// Resolve the handoff references via [`Self::node_handoff_references`] for the given `node_id`.
863    /// Returns token streams for each reference:
864    /// - For HandoffKind::Singleton: `buf.as_ref().unwrap()` (shared, `&T`) or
865    ///   `buf.as_mut().unwrap()` (mutable, `&mut T`)
866    /// - For HandoffKind::Optional: `&buf` (shared, `&Option<T>`) or
867    ///   `&mut buf` (mutable, `&mut Option<T>`)
868    /// - For HandoffKind::Vec: `&buf` (shared, `&Vec<T>`) or
869    ///   `&mut buf` (mutable, `&mut Vec<T>`)
870    fn helper_resolve_singletons(&self, node_id: GraphNodeId, span: Span) -> Vec<TokenStream> {
871        self.node_handoff_references(node_id)
872            .iter()
873            .map(|resolved_ref| {
874                // TODO(mingwei): this `expect` should be caught in error checking
875                let ref_node_id = resolved_ref
876                    .node_id
877                    .expect("Expected singleton to be resolved but was not, this is a bug.");
878                let is_mut = resolved_ref.is_mut;
879                match self.node(ref_node_id) {
880                    GraphNode::Handoff {
881                        kind: HandoffKind::Singleton,
882                        ..
883                    } => {
884                        let buf_ident = self.hoff_buf_ident(ref_node_id, span);
885                        if is_mut {
886                            quote_spanned! {span=> #buf_ident.as_mut().unwrap() }
887                        } else {
888                            quote_spanned! {span=> #buf_ident.as_ref().unwrap() }
889                        }
890                    }
891                    GraphNode::Handoff {
892                        kind: HandoffKind::Optional | HandoffKind::Vec,
893                        ..
894                    } => {
895                        let buf_ident = self.hoff_buf_ident(ref_node_id, span);
896                        if is_mut {
897                            quote_spanned! {span=> &mut #buf_ident }
898                        } else {
899                            quote_spanned! {span=> &#buf_ident }
900                        }
901                    }
902                    _ => {
903                        unreachable!("Only handoff nodes should be reachable as handoff references")
904                    }
905                }
906            })
907            .collect::<Vec<_>>()
908    }
909
910    /// Returns each subgraph's receive and send handoffs.
911    /// `Map<GraphSubgraphId, (recv handoffs, send handoffs)>`
912    fn helper_collect_subgraph_handoffs(
913        &self,
914    ) -> SecondaryMap<GraphSubgraphId, (Vec<GraphNodeId>, Vec<GraphNodeId>)> {
915        // Get data on handoff src and dst subgraphs.
916        let mut subgraph_handoffs: SecondaryMap<
917            GraphSubgraphId,
918            (Vec<GraphNodeId>, Vec<GraphNodeId>),
919        > = self
920            .subgraph_nodes
921            .keys()
922            .map(|k| (k, Default::default()))
923            .collect();
924
925        // For each handoff/singleton node, add it to the `send`/`recv` lists for the corresponding subgraphs.
926        for (hoff_id, hoff) in self.nodes() {
927            if !matches!(hoff, GraphNode::Handoff { .. }) {
928                continue;
929            }
930            // Receivers from the handoff. (Should really only be one).
931            for (_edge, succ_id) in self.node_successors(hoff_id) {
932                let succ_sg = self
933                    .node_subgraph(succ_id)
934                    .expect("bug: successor not in subgraph, may be a doubled/adjacent handoff");
935                subgraph_handoffs[succ_sg].0.push(hoff_id);
936            }
937            // Senders into the handoff. (Should really only be one).
938            for (_edge, pred_id) in self.node_predecessors(hoff_id) {
939                let pred_sg = self
940                    .node_subgraph(pred_id)
941                    .expect("bug: predecessor not in subgraph, may be a doubled/adjacent handoff");
942                subgraph_handoffs[pred_sg].1.push(hoff_id);
943            }
944        }
945
946        subgraph_handoffs
947    }
948
949    /// Compute the output handoffs exiting each loop (sender inside, receiver outside).
950    /// Returns a map from loop ID to the list of handoff node IDs that exit that loop.
951    fn helper_loop_output_handoffs(&self) -> SecondaryMap<GraphLoopId, Vec<GraphNodeId>> {
952        let mut loop_hoffs_out = SecondaryMap::<GraphLoopId, Vec<GraphNodeId>>::new();
953
954        for (hoff_id, hoff) in self.nodes() {
955            if !matches!(hoff, GraphNode::Handoff { .. }) {
956                continue;
957            }
958
959            let loop_pred = self
960                .node_predecessors(hoff_id)
961                .next()
962                .and_then(|(_, pred)| self.node_loop(pred));
963            let loop_succ = self
964                .node_successors(hoff_id)
965                .next()
966                .and_then(|(_, succ)| self.node_loop(succ));
967
968            if let Some(loop_pred) = loop_pred
969                && loop_succ == self.loop_parent(loop_pred)
970            {
971                // Pred is inside a child loop, succ is in the parent/outer.
972                loop_hoffs_out
973                    .entry(loop_pred)
974                    .expect("loop removed")
975                    .or_default()
976                    .push(hoff_id);
977            }
978        }
979
980        loop_hoffs_out
981    }
982
983    /// Returns true if `node_loop` is `loop_id` or a (transitive) child of `loop_id`.
984    fn is_inside_loop(&self, node_loop: Option<GraphLoopId>, loop_id: GraphLoopId) -> bool {
985        let mut current = node_loop;
986        while let Some(l) = current {
987            if l == loop_id {
988                return true;
989            }
990            current = self.loop_parent(l);
991        }
992        false
993    }
994
995    /// Emit a loop gate: wraps `child_body` in the appropriate control structure
996    /// and appends the result + swap code to `output`.
997    ///
998    /// - **Root-level loops** (no parent) are fused with the tick: they emit an `if`
999    ///   so the body runs at most once per tick when their entry condition is met.
1000    /// - **Nested loops** (have a parent loop) emit a `while` so they can iterate
1001    ///   until fixpoint (driven by `defer_tick` back-edges).
1002    /// - If there are no gate checks, the body is emitted unconditionally.
1003    fn emit_loop_gate(
1004        &self,
1005        loop_id: GraphLoopId,
1006        child_body: TokenStream,
1007        loop_input_handoffs: &SecondaryMap<GraphLoopId, Vec<GraphNodeId>>,
1008        back_edge_hoffs_and_lazyness: &SparseSecondaryMap<GraphNodeId, bool>,
1009        loop_swap_code: &std::collections::HashMap<GraphLoopId, Vec<TokenStream>>,
1010        output: &mut TokenStream,
1011    ) {
1012        // Get swap code for this loop's defer_tick handoffs.
1013        let swap_code = loop_swap_code
1014            .get(&loop_id)
1015            .map(|v| v.as_slice())
1016            .unwrap_or(&[]);
1017
1018        // Root-level loops are fused with the tick: emit `if` instead of `while`.
1019        let is_root_loop = self.loop_parent(loop_id).is_none();
1020
1021        // Build the gate condition from entry handoffs (excluding lazy windowing operators).
1022        let entry_handoffs = loop_input_handoffs.get(loop_id).expect("loop missing");
1023        let mut gate_checks: Vec<TokenStream> = entry_handoffs
1024            .iter()
1025            .filter(|&&hoff_id| {
1026                // Check if the successor (windowing operator) is lazy.
1027                // If so, exclude from the gate — it doesn't trigger the loop.
1028                let is_lazy = self
1029                    .node_successors(hoff_id)
1030                    .next()
1031                    .and_then(|(_, succ)| self.node_op_inst(succ))
1032                    .is_some_and(|op_inst| {
1033                        op_inst.op_constraints.flo_type == Some(FloType::WindowingLazy)
1034                    });
1035                !is_lazy
1036            })
1037            .map(|&hoff_id| {
1038                let span = self.node(hoff_id).span();
1039                let buf_ident = self.hoff_buf_ident(hoff_id, span);
1040                if back_edge_hoffs_and_lazyness.contains_key(hoff_id) {
1041                    let back_ident = self.hoff_back_ident(hoff_id, span);
1042                    quote_spanned! {span=> !#back_ident.is_empty() }
1043                } else {
1044                    quote_spanned! {span=> !#buf_ident.is_empty() }
1045                }
1046            })
1047            .collect();
1048
1049        // Non-lazy defer_tick back-buffers also contribute to the gate (nested loops only).
1050        if !is_root_loop {
1051            for (hoff_id, hoff) in self.nodes() {
1052                if !matches!(hoff, GraphNode::Handoff { .. }) {
1053                    continue;
1054                }
1055                let Some(delay_type) = self.handoff_delay_type(hoff_id) else {
1056                    continue;
1057                };
1058                if delay_type != DelayType::Loop {
1059                    continue;
1060                }
1061                // Check this handoff belongs to loop_id.
1062                let hoff_loop = self
1063                    .node_successors(hoff_id)
1064                    .next()
1065                    .and_then(|(_, succ)| self.node_subgraph(succ))
1066                    .and_then(|sg| self.subgraph_loop(sg));
1067                if hoff_loop != Some(loop_id) {
1068                    continue;
1069                }
1070                let span = self.node(hoff_id).span();
1071                let back_ident = self.hoff_back_ident(hoff_id, span);
1072                gate_checks.push(quote_spanned! {span=> !#back_ident.is_empty() });
1073            }
1074        }
1075
1076        // For root-level loops: non-lazy defer_tick back-buffers also contribute to the gate.
1077        // This ensures the loop fires on the next tick when data was deferred via defer_tick.
1078        if is_root_loop {
1079            for (hoff_id, hoff) in self.nodes() {
1080                if !matches!(hoff, GraphNode::Handoff { .. }) {
1081                    continue;
1082                }
1083                let Some(delay_type) = self.handoff_delay_type(hoff_id) else {
1084                    continue;
1085                };
1086                if delay_type != DelayType::Tick {
1087                    continue;
1088                }
1089                // Check this handoff's consumer is inside this root-level loop.
1090                let hoff_loop = self
1091                    .node_successors(hoff_id)
1092                    .next()
1093                    .and_then(|(_, succ)| self.node_subgraph(succ))
1094                    .and_then(|sg| self.subgraph_loop(sg));
1095                if hoff_loop != Some(loop_id) {
1096                    continue;
1097                }
1098                let span = self.node(hoff_id).span();
1099                let back_ident = self.hoff_back_ident(hoff_id, span);
1100                gate_checks.push(quote_spanned! {span=> !#back_ident.is_empty() });
1101            }
1102        }
1103
1104        // An eager windowing operator (`batch_eager()`) forces the loop to fire unconditionally,
1105        // even when its windowed input is empty. It is only valid at the entry of a root-level
1106        // loop (disallowed in nested loops during validation, since forcing a nested loop to
1107        // always fire would prevent its fixpoint iteration from terminating).
1108        let has_eager = entry_handoffs.iter().any(|&hoff_id| {
1109            self.node_successors(hoff_id)
1110                .next()
1111                .and_then(|(_, succ)| self.node_op_inst(succ))
1112                .is_some_and(|op_inst| {
1113                    op_inst.op_constraints.flo_type == Some(FloType::WindowingEager)
1114                })
1115        });
1116
1117        if has_eager && is_root_loop {
1118            // Eager entry: always run the loop body (gate forced true).
1119            output.extend(child_body);
1120            output.extend(quote! { #( #swap_code )* });
1121        } else if gate_checks.is_empty() {
1122            // No entry handoffs — always run.
1123            output.extend(child_body);
1124            output.extend(quote! { #( #swap_code )* });
1125        } else if is_root_loop {
1126            // Root-level loop: fused with tick, fire at most once.
1127            output.extend(quote! {
1128                #[allow(clippy::nonminimal_bool, reason = "codegen")]
1129                if false #( || #gate_checks )* {
1130                    #child_body
1131                    #( #swap_code )*
1132                }
1133            });
1134        } else {
1135            // Nested loop: iterate until fixpoint.
1136            output.extend(quote! {
1137                #[allow(clippy::nonminimal_bool, reason = "codegen")]
1138                while false #( || #gate_checks )* {
1139                    #child_body
1140                    #( #swap_code )*
1141                }
1142            });
1143        }
1144    }
1145
1146    /// Compute the input handoffs into each loop (predecessor outside, successor inside).
1147    fn helper_loop_input_handoffs(&self) -> SecondaryMap<GraphLoopId, Vec<GraphNodeId>> {
1148        let mut loop_hoffs_inn = SecondaryMap::<GraphLoopId, Vec<GraphNodeId>>::new();
1149
1150        // Check each handoff node.
1151        for (hoff_id, hoff) in self.nodes() {
1152            if !matches!(hoff, GraphNode::Handoff { .. }) {
1153                continue;
1154            }
1155
1156            // Get the loop context of the predecessor and successor.
1157            let loop_pred = self
1158                .node_predecessors(hoff_id)
1159                .next()
1160                .and_then(|(_, pred)| self.node_loop(pred));
1161            let loop_succ = self
1162                .node_successors(hoff_id)
1163                .next()
1164                .and_then(|(_, succ)| self.node_loop(succ));
1165
1166            if let Some(loop_succ) = loop_succ
1167                && loop_pred == self.loop_parent(loop_succ)
1168            {
1169                // Pred is parent/outer loop of succ.
1170                loop_hoffs_inn
1171                    .entry(loop_succ)
1172                    .expect("loop removed")
1173                    .or_default()
1174                    .push(hoff_id);
1175            }
1176        }
1177
1178        loop_hoffs_inn
1179    }
1180
1181    /// Emit this graph as runnable Rust source code tokens that execute inline.
1182    /// Generates a flat `async move |df: &mut Context|` closure where subgraph
1183    /// blocks are inlined in topological order, using local `Vec<T>` buffers
1184    /// instead of runtime handoffs. Each call to the closure runs one tick.
1185    ///
1186    /// The generated code block evaluates to a `Dfir` instance wrapping the
1187    /// closure. Operator prologues run at construction time on the `Context`
1188    /// before it is moved into `Dfir::new`. `Dfir` provides the `Context`
1189    /// to the closure on each tick run.
1190    ///
1191    /// # Errors
1192    ///
1193    /// Returns all diagnostics as `Err(diagnostics)` if any are errors
1194    /// (leaving `&mut diagnostics` empty).
1195    pub fn as_code(
1196        &self,
1197        root: &TokenStream,
1198        include_type_guards: bool,
1199        prefix: TokenStream,
1200        diagnostics: &mut Diagnostics,
1201    ) -> Result<TokenStream, Diagnostics> {
1202        self.as_code_with_options(root, include_type_guards, true, prefix, diagnostics)
1203    }
1204
1205    /// Like [`Self::as_code`], but with `include_meta` controlling whether
1206    /// the runtime meta graph + diagnostics JSON blobs are baked into the
1207    /// generated `Dfir::new(...)` call.
1208    ///
1209    /// The simulator calls Dfir::new() on each iteration, and as a part of that
1210    /// it does parsing of the metagraph and diganostics blob. One of them causes spans to get allocated,
1211    /// each time a span is allocated, some threadlocal u32 is being incremented, and, on a long simulator run,
1212    /// the u32 overflows and panics.
1213    pub fn as_code_with_options(
1214        &self,
1215        root: &TokenStream,
1216        include_type_guards: bool,
1217        include_meta: bool,
1218        prefix: TokenStream,
1219        diagnostics: &mut Diagnostics,
1220    ) -> Result<TokenStream, Diagnostics> {
1221        let df = Ident::new(GRAPH, Span::call_site());
1222        let context = Ident::new(CONTEXT, Span::call_site());
1223        // Tick-local bump-allocated Vec handoff declarations (inside the tick closure).
1224        let bump_ident = Ident::new("__dfir_bump", Span::call_site());
1225
1226        // 1. Collect all handoff nodes.
1227        let handoff_nodes = self
1228            .nodes
1229            .iter()
1230            .filter_map(|(node_id, node)| match node {
1231                &GraphNode::Handoff {
1232                    kind,
1233                    src_span,
1234                    dst_span,
1235                } => Some((node_id, kind, (src_span, dst_span))),
1236                GraphNode::Operator(_) => None,
1237                GraphNode::ModuleBoundary { .. } => panic!(),
1238            })
1239            .collect::<Vec<_>>();
1240
1241        // Determine which handoff nodes are tick-boundary (defer_tick) back-edges.
1242        // These must remain as captured Vec<T> since they persist across ticks.
1243        // All other Vec handoffs will be bump-allocated (tick-local).
1244        let back_edge_hoffs_and_lazyness = handoff_nodes
1245            .iter()
1246            .map(|&(node_id, _, _)| node_id)
1247            .filter_map(|node_id| {
1248                let delay_type = self.handoff_delay_type(node_id)?;
1249                Some((
1250                    node_id,
1251                    matches!(delay_type, DelayType::TickLazy | DelayType::LoopLazy),
1252                ))
1253            })
1254            .collect::<SparseSecondaryMap<_, _>>();
1255
1256        // Back buffer idents, buf idents, and if they are lazy.
1257        let back_buffer_idents_laziness = handoff_nodes
1258            .iter()
1259            .filter_map(|&(hoff_id, _kind, (src_span, dst_span))| {
1260                back_edge_hoffs_and_lazyness.get(hoff_id).map(|&is_lazy| {
1261                    let span = src_span.join(dst_span).unwrap_or(src_span);
1262                    let back_ident = self.hoff_back_ident(hoff_id, span);
1263                    let buf_ident = self.hoff_buf_ident(hoff_id, span);
1264                    (back_ident, buf_ident, is_lazy)
1265                })
1266            })
1267            .collect::<Vec<_>>();
1268
1269        // Generate swap code for tick-boundary (defer_tick / defer_tick_lazy) handoffs.
1270        // At the end of each tick, swap the regular buffer and back buffer so the
1271        // consumer reads last tick's data from the back buffer.
1272        // Only tick-level swaps go here; loop-level swaps are emitted inside the loop gate.
1273        // IMPORTANT: For defer_tick handoffs whose consumer is inside a root-level loop,
1274        // the swap is emitted inside the `if` gate (via loop_swap_code), not at tick level.
1275        let back_edge_swap_code = handoff_nodes
1276            .iter()
1277            .filter(|&&(node_id, _kind, _)| {
1278                self.handoff_delay_type(node_id)
1279                    .is_some_and(|dt| matches!(dt, DelayType::Tick | DelayType::TickLazy))
1280            })
1281            .filter(|&&(hoff_id, _kind, _)| {
1282                // Exclude handoffs whose consumer is inside a root-level loop.
1283                // Those get their swap emitted inside the loop gate.
1284                let consumer_loop = self
1285                    .node_successors(hoff_id)
1286                    .next()
1287                    .and_then(|(_, succ)| self.node_subgraph(succ))
1288                    .and_then(|sg| self.subgraph_loop(sg));
1289                if let Some(loop_id) = consumer_loop {
1290                    // If it's a root-level loop, don't include in tick-level swap.
1291                    self.loop_parent(loop_id).is_some()
1292                } else {
1293                    // No loop context: emit at tick level (original behavior).
1294                    true
1295                }
1296            })
1297            .map(|&(hoff_id, _kind, _)| {
1298                let span = self.nodes[hoff_id].span();
1299                let buf_ident = self.hoff_buf_ident(hoff_id, span);
1300                let back_ident = self.hoff_back_ident(hoff_id, span);
1301                quote_spanned! {span=>
1302                    ::std::mem::swap(&mut #buf_ident, &mut #back_ident);
1303                }
1304            })
1305            .collect::<Vec<_>>();
1306
1307        // Collect per-loop swap code for defer_tick / defer_tick_lazy handoffs.
1308        // AND defer_tick / defer_tick_lazy handoffs inside root-level loops.
1309        // Keyed by the loop ID of the consumer (successor) of the handoff.
1310        let mut loop_swap_code: std::collections::HashMap<GraphLoopId, Vec<TokenStream>> =
1311            std::collections::HashMap::new();
1312        for &(hoff_id, _kind, _) in handoff_nodes.iter() {
1313            let Some(delay_type) = self.handoff_delay_type(hoff_id) else {
1314                continue;
1315            };
1316            // Find the loop this handoff belongs to (from its consumer's loop context).
1317            let loop_id = self
1318                .node_successors(hoff_id)
1319                .next()
1320                .and_then(|(_, succ)| self.node_subgraph(succ))
1321                .and_then(|sg| self.subgraph_loop(sg));
1322            let Some(loop_id) = loop_id else {
1323                continue;
1324            };
1325            let include = match delay_type {
1326                DelayType::Loop | DelayType::LoopLazy => true,
1327                DelayType::Tick | DelayType::TickLazy => {
1328                    // Only include in loop swap if this is a root-level loop.
1329                    self.loop_parent(loop_id).is_none()
1330                }
1331            };
1332            if !include {
1333                continue;
1334            }
1335            let span = self.nodes[hoff_id].span();
1336            let buf_ident = self.hoff_buf_ident(hoff_id, span);
1337            let back_ident = self.hoff_back_ident(hoff_id, span);
1338            loop_swap_code
1339                .entry(loop_id)
1340                .or_default()
1341                .push(quote_spanned! {span=>
1342                    ::std::mem::swap(&mut #buf_ident, &mut #back_ident);
1343                });
1344        }
1345
1346        // 2. Collect per-subgraph recv & send handoffs.
1347        let subgraph_handoffs = self.helper_collect_subgraph_handoffs();
1348
1349        // 3. Use pre-computed subgraph topological order.
1350        let all_subgraphs: Vec<_> = self
1351            .subgraph_toposort()
1352            .iter()
1353            .map(|&sg_id| (sg_id, self.subgraph(sg_id)))
1354            .collect();
1355
1356        // TODO(mingwei): If a handoff has no pipe consumers we should drop it as soon as possible, after all reference
1357        // consumers. Right now we just let these handoffs die at the end of the tick.
1358
1359        let mut op_prologue_code = Vec::new();
1360        let mut op_tick_end_code = Vec::new();
1361
1362        // Stack-based hierarchical code generation.
1363        // Each entry is (loop_id, body_tokens) for an open loop context.
1364        // The "current output" is always the innermost open context (or root).
1365        let mut loop_stack: Vec<(GraphLoopId, TokenStream)> = Vec::new();
1366        let mut current_output = TokenStream::new();
1367
1368        // Pre-compute loop gate data.
1369        let loop_input_handoffs = self.helper_loop_input_handoffs();
1370        let loop_output_handoffs = self.helper_loop_output_handoffs();
1371
1372        {
1373            for &(subgraph_id, subgraph_nodes) in all_subgraphs.iter() {
1374                let sg_loop = self.subgraph_loop(subgraph_id);
1375
1376                // Transition loop contexts: close loops we've exited, open loops we've entered.
1377                // Close loops until we're at the right level.
1378                while let Some(&(top_loop, _)) = loop_stack.last() {
1379                    if sg_loop == Some(top_loop) || self.is_inside_loop(sg_loop, top_loop) {
1380                        break;
1381                    }
1382                    // Pop: wrap the body in a loop gate and append to parent.
1383                    let (closed_loop, child_body) = loop_stack.pop().unwrap();
1384                    let target = if let Some((_, parent_body)) = loop_stack.last_mut() {
1385                        parent_body
1386                    } else {
1387                        &mut current_output
1388                    };
1389                    self.emit_loop_gate(
1390                        closed_loop,
1391                        child_body,
1392                        &loop_input_handoffs,
1393                        &back_edge_hoffs_and_lazyness,
1394                        &loop_swap_code,
1395                        target,
1396                    );
1397                }
1398
1399                // Open new loops if we've descended.
1400                if let Some(target_loop) = sg_loop
1401                    && loop_stack.last().map(|&(l, _)| l) != Some(target_loop)
1402                {
1403                    // Find the path of loops to open (from outermost to target).
1404                    let mut path = Vec::new();
1405                    let mut cur = Some(target_loop);
1406                    while let Some(l) = cur {
1407                        if loop_stack.last().map(|&(top, _)| top) == Some(l) {
1408                            break;
1409                        }
1410                        path.push(l);
1411                        cur = self.loop_parent(l);
1412                    }
1413                    // Push in outermost-first order, emitting exit-handoff declarations
1414                    // to the parent level before the while loop.
1415                    for &loop_id in path.iter().rev() {
1416                        // Declare exit-handoff buffers at the current (parent) level.
1417                        if let Some(exit_hoffs) = loop_output_handoffs.get(loop_id) {
1418                            let exit_hoff_decls = exit_hoffs.iter().map(|&hoff_id| {
1419                                let span = self.nodes[hoff_id].span();
1420                                let buf_ident = self.hoff_buf_ident(hoff_id, span);
1421                                let GraphNode::Handoff { kind, .. } = self.node(hoff_id) else {
1422                                    panic!()
1423                                };
1424                                match kind {
1425                                    HandoffKind::Vec => quote_spanned! {span=>
1426                                        let mut #buf_ident = #root::bumpalo::collections::Vec::new_in(&#bump_ident);
1427                                    },
1428                                    HandoffKind::Singleton | HandoffKind::Optional => quote_spanned! {span=>
1429                                        let mut #buf_ident = ::std::option::Option::None;
1430                                    },
1431                                }
1432                            });
1433                            let target = if let Some((_, body)) = loop_stack.last_mut() {
1434                                body
1435                            } else {
1436                                &mut current_output
1437                            };
1438                            target.extend(quote! { #( #exit_hoff_decls )* });
1439                        }
1440                        loop_stack.push((loop_id, TokenStream::new()));
1441                    }
1442                }
1443                let sg_metrics_ffi = subgraph_id.data().as_ffi();
1444                let (recv_hoffs, send_hoffs) = &subgraph_handoffs[subgraph_id];
1445
1446                // Generate buffer ident helpers for this subgraph's handoffs.
1447                let recv_port_idents: Vec<Ident> = recv_hoffs
1448                    .iter()
1449                    .map(|&hoff_id| self.node_as_ident(hoff_id, true))
1450                    .collect();
1451                let send_port_idents: Vec<Ident> = send_hoffs
1452                    .iter()
1453                    .map(|&hoff_id| self.node_as_ident(hoff_id, false))
1454                    .collect();
1455
1456                // Map handoff node IDs to buffer idents.
1457                let recv_buf_idents: Vec<Ident> = recv_hoffs
1458                    .iter()
1459                    .map(|&hoff_id| self.hoff_buf_ident(hoff_id, self.nodes[hoff_id].span()))
1460                    .collect();
1461                let send_buf_idents: Vec<Ident> = send_hoffs
1462                    .iter()
1463                    .map(|&hoff_id| self.hoff_buf_ident(hoff_id, self.nodes[hoff_id].span()))
1464                    .collect();
1465
1466                // Handoff kinds
1467                let recv_kinds = recv_hoffs
1468                    .iter()
1469                    .map(|&hoff_id| {
1470                        let GraphNode::Handoff { kind, .. } = self.node(hoff_id) else {
1471                            panic!()
1472                        };
1473                        *kind
1474                    })
1475                    .collect::<Vec<_>>();
1476                let send_kinds = send_hoffs
1477                    .iter()
1478                    .map(|&hoff_id| {
1479                        let GraphNode::Handoff { kind, .. } = self.node(hoff_id) else {
1480                            panic!()
1481                        };
1482                        *kind
1483                    })
1484                    .collect::<Vec<_>>();
1485
1486                // Recv port code: drain from buffer into iterator, tracking if non-empty.
1487                // For back-edge (defer_tick) handoffs, drain from the back buffer instead.
1488                // Also update handoff metrics (measured at recv, not send — see graph.rs).
1489                let recv_port_code: Vec<TokenStream> = recv_port_idents
1490                    .iter()
1491                    .zip(recv_buf_idents.iter())
1492                    .zip(recv_kinds.iter())
1493                    .zip(recv_hoffs.iter())
1494                    .map(|(((port_ident, buf_ident), &kind), &hoff_id)| {
1495                        let hoff_ffi = hoff_id.data().as_ffi();
1496                        // Use call_site span for internal identifiers to avoid
1497                        // hygiene issues when invoked through declarative macros
1498                        // (e.g. dfir_expect_warnings!). TODO(#2781): define these once.
1499                        let work_done = Ident::new("__dfir_work_done", Span::call_site());
1500                        let metrics = Ident::new("__dfir_metrics", Span::call_site());
1501
1502                        // Compute len and drain expressions based on handoff kind.
1503                        let (len_expr, drain_expr) = match kind {
1504                            HandoffKind::Singleton | HandoffKind::Optional => (
1505                                quote! { if #buf_ident.is_some() { 1usize } else { 0usize } },
1506                                quote! { #root::dfir_pipes::pull::iter(#buf_ident.take().into_iter()) },
1507                            ),
1508                            HandoffKind::Vec => {
1509                                // Special asymmetric handling for defer tick handoffs, which are double-buffered.
1510                                // The producer writes to the regular buffer; at end-of-tick the buffers are swapped,
1511                                // so the consumer drains from the back buffer (here).
1512                                let drain_ident = if back_edge_hoffs_and_lazyness.contains_key(hoff_id) {
1513                                    &self.hoff_back_ident(hoff_id, buf_ident.span())
1514                                } else {
1515                                    buf_ident
1516                                };
1517                                (
1518                                    quote! { #drain_ident.len() },
1519                                    quote! { #root::dfir_pipes::pull::iter(#drain_ident.drain(..)) },
1520                                )
1521                            }
1522                        };
1523
1524                        quote_spanned! {port_ident.span()=>
1525                            {
1526                                let hoff_len = #len_expr;
1527                                if hoff_len > 0 {
1528                                    #work_done = true;
1529                                }
1530                                let hoff_metrics = &#metrics.handoffs[
1531                                    #root::slotmap::KeyData::from_ffi(#hoff_ffi).into()
1532                                ];
1533                                hoff_metrics.total_items_count.update(|x| x + hoff_len);
1534                                hoff_metrics.curr_items_count.set(hoff_len);
1535                            }
1536                            let #port_ident = #drain_expr;
1537                        }
1538                    })
1539                    .collect();
1540
1541                // Send port code: push into buffer.
1542                let send_port_code: Vec<TokenStream> = send_port_idents
1543                    .iter()
1544                    .zip(send_buf_idents.iter())
1545                    .zip(send_kinds.iter())
1546                    .map(|((port_ident, buf_ident), &kind)| {
1547                        match kind {
1548                            HandoffKind::Singleton => {
1549                                // Singleton slot: store exactly one item, panic on duplicate.
1550                                quote_spanned! {port_ident.span()=>
1551                                    let #port_ident = #root::dfir_pipes::push::for_each(|__item| {
1552                                        if #buf_ident.replace(__item).is_some() {
1553                                            panic!("singleton() received more than one item");
1554                                        }
1555                                    });
1556                                }
1557                            }
1558                            HandoffKind::Optional => {
1559                                // Optional slot: store at most one item, panic on duplicate.
1560                                quote_spanned! {port_ident.span()=>
1561                                    let #port_ident = #root::dfir_pipes::push::for_each(|__item| {
1562                                        if #buf_ident.replace(__item).is_some() {
1563                                            panic!("optional() received more than one item");
1564                                        }
1565                                    });
1566                                }
1567                            }
1568                            HandoffKind::Vec => {
1569                                quote_spanned! {port_ident.span()=>
1570                                    // TODO(mingwei): use `#root::dfir_pipes::push::vec_push`?
1571                                    let #port_ident = #root::dfir_pipes::push::for_each(|item| { #buf_ident.push(item); });
1572                                }
1573                            }
1574                        }
1575                    })
1576                    .collect();
1577
1578                // All nodes in a subgraph should be in the same loop.
1579                let loop_id = self.node_loop(subgraph_nodes[0]);
1580
1581                let mut subgraph_op_iter_code = Vec::new();
1582                let mut subgraph_op_iter_after_code = Vec::new();
1583                {
1584                    let pull_to_push_idx = self.find_pull_to_push_idx(subgraph_nodes);
1585
1586                    let (pull_half, push_half) = subgraph_nodes.split_at(pull_to_push_idx);
1587                    let nodes_iter = pull_half.iter().chain(push_half.iter().rev());
1588
1589                    for (idx, &node_id) in nodes_iter.enumerate() {
1590                        let node = &self.nodes[node_id];
1591                        assert!(
1592                            matches!(node, GraphNode::Operator(_)),
1593                            "Handoffs are not part of subgraphs."
1594                        );
1595                        let op_inst = &self.operator_instances[node_id];
1596
1597                        let op_span = node.span();
1598                        let op_name = op_inst.op_constraints.name;
1599                        // Use op's span for root. #root is expected to be correct, any errors should span back to the op gen.
1600                        let root = change_spans(root.clone(), op_span);
1601                        let op_constraints = OPERATORS
1602                            .iter()
1603                            .find(|op| op_name == op.name)
1604                            .unwrap_or_else(|| panic!("Failed to find op: {}", op_name));
1605
1606                        let ident = self.node_as_ident(node_id, false);
1607
1608                        {
1609                            // TODO clean this up.
1610                            // Collect input arguments (predecessors).
1611                            let mut input_edges = self
1612                                .graph
1613                                .predecessor_edges(node_id)
1614                                .map(|edge_id| (self.edge_ports(edge_id).1, edge_id))
1615                                .collect::<Vec<_>>();
1616                            // Ensure sorted by port index.
1617                            input_edges.sort();
1618
1619                            let inputs = input_edges
1620                                .iter()
1621                                .map(|&(_port, edge_id)| {
1622                                    let (pred, _) = self.edge(edge_id);
1623                                    self.node_as_ident(pred, true)
1624                                })
1625                                .collect::<Vec<_>>();
1626
1627                            // Collect output arguments (successors).
1628                            let mut output_edges = self
1629                                .graph
1630                                .successor_edges(node_id)
1631                                .map(|edge_id| (&self.ports[edge_id].0, edge_id))
1632                                .collect::<Vec<_>>();
1633                            // Ensure sorted by port index.
1634                            output_edges.sort();
1635
1636                            let outputs = output_edges
1637                                .iter()
1638                                .map(|&(_port, edge_id)| {
1639                                    let (_, succ) = self.edge(edge_id);
1640                                    self.node_as_ident(succ, false)
1641                                })
1642                                .collect::<Vec<_>>();
1643
1644                            let is_pull = idx < pull_to_push_idx;
1645
1646                            // There's a bit of dark magic hidden in `Span`s... you'd think it's just a `file:line:column`,
1647                            // but it has one extra bit of info for _name resolution_, used for `Ident`s. `Span::call_site()`
1648                            // has the (unhygienic) resolution we want, an ident is just solely determined by its string name,
1649                            // which is what you'd expect out of unhygienic proc macros like this. Meanwhile, declarative macros
1650                            // use `Span::mixed_site()` which is weird and I don't understand it. It turns out that if you call
1651                            // the dfir syntax proc macro from _within_ a declarative macro then `op_span` will have the
1652                            // bad `Span::mixed_site()` name resolution and cause "Cannot find value `df/context`" errors. So
1653                            // we call `.resolved_at()` to fix resolution back to `Span::call_site()`. -Mingwei
1654                            let df_local = &Ident::new(GRAPH, op_span.resolved_at(df.span()));
1655                            let context = &Ident::new(CONTEXT, op_span.resolved_at(context.span()));
1656
1657                            let singletons_resolved =
1658                                self.helper_resolve_singletons(node_id, op_span);
1659
1660                            let arguments = &process_singletons::postprocess_singletons(
1661                                op_inst.arguments_raw.clone(),
1662                                singletons_resolved,
1663                            );
1664
1665                            let source_tag = 'a: {
1666                                if let Some(tag) = self.operator_tag.get(node_id).cloned() {
1667                                    break 'a tag;
1668                                }
1669
1670                                if proc_macro::is_available() {
1671                                    let op_span = op_span.unwrap();
1672                                    break 'a format!(
1673                                        "loc_{}_{}_{}_{}_{}",
1674                                        crate::pretty_span::make_source_path_relative(
1675                                            &op_span.file()
1676                                        )
1677                                        .display()
1678                                        .to_string()
1679                                        .replace(|x: char| !x.is_ascii_alphanumeric(), "_"),
1680                                        op_span.start().line(),
1681                                        op_span.start().column(),
1682                                        op_span.end().line(),
1683                                        op_span.end().column(),
1684                                    );
1685                                }
1686
1687                                format!(
1688                                    "loc_nopath_{}_{}_{}_{}",
1689                                    op_span.start().line,
1690                                    op_span.start().column,
1691                                    op_span.end().line,
1692                                    op_span.end().column
1693                                )
1694                            };
1695
1696                            let work_fn = format_ident!(
1697                                "{}__{}__{}",
1698                                ident,
1699                                op_name,
1700                                source_tag,
1701                                span = op_span
1702                            );
1703                            let work_fn_async = format_ident!("{}__async", work_fn, span = op_span);
1704
1705                            let context_args = WriteContextArgs {
1706                                root: &root,
1707                                df_ident: df_local,
1708                                context,
1709                                subgraph_id,
1710                                node_id,
1711                                loop_id,
1712                                op_span,
1713                                op_tag: self.operator_tag.get(node_id).cloned(),
1714                                work_fn: &work_fn,
1715                                work_fn_async: &work_fn_async,
1716                                ident: &ident,
1717                                is_pull,
1718                                inputs: &inputs,
1719                                outputs: &outputs,
1720                                op_name,
1721                                op_inst,
1722                                arguments,
1723                            };
1724
1725                            let write_result =
1726                                (op_constraints.write_fn)(&context_args, diagnostics);
1727                            let OperatorWriteOutput {
1728                                write_prologue,
1729                                write_iterator,
1730                                write_iterator_after,
1731                                write_tick_end,
1732                            } = write_result.unwrap_or_else(|()| {
1733                                assert!(
1734                                    diagnostics.has_error(),
1735                                    "Operator `{}` returned `Err` but emitted no diagnostics, this is a bug.",
1736                                    op_name,
1737                                );
1738                                OperatorWriteOutput {
1739                                    write_iterator: null_write_iterator_fn(&context_args),
1740                                    ..Default::default()
1741                                }
1742                            });
1743
1744                            op_prologue_code.push(syn::parse_quote! {
1745                                #[allow(dead_code, non_snake_case, reason = "codegen")]
1746                                #[inline(always)]
1747                                fn #work_fn<T>(thunk: impl ::std::ops::FnOnce() -> T) -> T {
1748                                    thunk()
1749                                }
1750
1751                                #[allow(dead_code, non_snake_case, reason = "codegen")]
1752                                #[inline(always)]
1753                                async fn #work_fn_async<T>(
1754                                    thunk: impl ::std::future::Future<Output = T>,
1755                                ) -> T {
1756                                    thunk.await
1757                                }
1758                            });
1759                            op_prologue_code.push(write_prologue);
1760                            op_tick_end_code.push(write_tick_end);
1761                            subgraph_op_iter_code.push(write_iterator);
1762
1763                            if include_type_guards {
1764                                let type_guard = if is_pull {
1765                                    quote_spanned! {op_span=>
1766                                        let #ident = {
1767                                            #[allow(non_snake_case)]
1768                                            #[inline(always)]
1769                                            pub fn #work_fn<Item, Input>(input: Input)
1770                                                -> impl #root::dfir_pipes::pull::Pull<Item = Item, Meta = (), CanPend = Input::CanPend, CanEnd = Input::CanEnd>
1771                                            where
1772                                                Input: #root::dfir_pipes::pull::Pull<Item = Item, Meta = ()>,
1773                                            {
1774                                                #root::pin_project_lite::pin_project! {
1775                                                    #[repr(transparent)]
1776                                                    struct Pull<Item, Input: #root::dfir_pipes::pull::Pull<Item = Item>> {
1777                                                        #[pin]
1778                                                        inner: Input
1779                                                    }
1780                                                }
1781
1782                                                impl<Item, Input> #root::dfir_pipes::pull::Pull for Pull<Item, Input>
1783                                                where
1784                                                    Input: #root::dfir_pipes::pull::Pull<Item = Item>,
1785                                                {
1786                                                    type Ctx<'ctx> = Input::Ctx<'ctx>;
1787
1788                                                    type Item = Item;
1789                                                    type Meta = Input::Meta;
1790                                                    type CanPend = Input::CanPend;
1791                                                    type CanEnd = Input::CanEnd;
1792
1793                                                    #[inline(always)]
1794                                                    fn pull(
1795                                                        self: ::std::pin::Pin<&mut Self>,
1796                                                        ctx: &mut Self::Ctx<'_>,
1797                                                    ) -> #root::dfir_pipes::pull::PullStep<Self::Item, Self::Meta, Self::CanPend, Self::CanEnd> {
1798                                                        #root::dfir_pipes::pull::Pull::pull(self.project().inner, ctx)
1799                                                    }
1800
1801                                                    #[inline(always)]
1802                                                    fn size_hint(&self) -> (usize, Option<usize>) {
1803                                                        #root::dfir_pipes::pull::Pull::size_hint(&self.inner)
1804                                                    }
1805                                                }
1806
1807                                                Pull {
1808                                                    inner: input
1809                                                }
1810                                            }
1811                                            #work_fn::<_, _>( #ident )
1812                                        };
1813                                    }
1814                                } else {
1815                                    quote_spanned! {op_span=>
1816                                        let #ident = {
1817                                            #[allow(non_snake_case)]
1818                                            #[inline(always)]
1819                                            pub fn #work_fn<Item, Psh>(psh: Psh) -> impl #root::dfir_pipes::push::Push<Item, (), CanPend = Psh::CanPend>
1820                                            where
1821                                                Psh: #root::dfir_pipes::push::Push<Item, ()>
1822                                            {
1823                                                #root::pin_project_lite::pin_project! {
1824                                                    #[repr(transparent)]
1825                                                    struct PushGuard<Psh> {
1826                                                        #[pin]
1827                                                        inner: Psh,
1828                                                    }
1829                                                }
1830
1831                                                impl<Item, Psh> #root::dfir_pipes::push::Push<Item, ()> for PushGuard<Psh>
1832                                                where
1833                                                    Psh: #root::dfir_pipes::push::Push<Item, ()>,
1834                                                {
1835                                                    type Ctx<'ctx> = Psh::Ctx<'ctx>;
1836
1837                                                    type CanPend = Psh::CanPend;
1838
1839                                                    #[inline(always)]
1840                                                    fn poll_ready(
1841                                                        self: ::std::pin::Pin<&mut Self>,
1842                                                        ctx: &mut Self::Ctx<'_>,
1843                                                    ) -> #root::dfir_pipes::push::PushStep<Self::CanPend> {
1844                                                        #root::dfir_pipes::push::Push::poll_ready(self.project().inner, ctx)
1845                                                    }
1846
1847                                                    #[inline(always)]
1848                                                    fn start_send(
1849                                                        self: ::std::pin::Pin<&mut Self>,
1850                                                        item: Item,
1851                                                        meta: (),
1852                                                    ) {
1853                                                        #root::dfir_pipes::push::Push::start_send(self.project().inner, item, meta)
1854                                                    }
1855
1856                                                    #[inline(always)]
1857                                                    fn poll_finalize(
1858                                                        self: ::std::pin::Pin<&mut Self>,
1859                                                        ctx: &mut Self::Ctx<'_>,
1860                                                    ) -> #root::dfir_pipes::push::PushStep<Self::CanPend> {
1861                                                        #root::dfir_pipes::push::Push::poll_finalize(self.project().inner, ctx)
1862                                                    }
1863
1864                                                    #[inline(always)]
1865                                                    fn size_hint(
1866                                                        self: ::std::pin::Pin<&mut Self>,
1867                                                        hint: (usize, Option<usize>),
1868                                                    ) {
1869                                                        #root::dfir_pipes::push::Push::size_hint(self.project().inner, hint)
1870                                                    }
1871                                                }
1872
1873                                                PushGuard {
1874                                                    inner: psh
1875                                                }
1876                                            }
1877                                            #work_fn( #ident )
1878                                        };
1879                                    }
1880                                };
1881                                subgraph_op_iter_code.push(type_guard);
1882                            }
1883                            subgraph_op_iter_after_code.push(write_iterator_after);
1884                        }
1885                    }
1886
1887                    {
1888                        // Determine pull and push halves of the `Pivot`.
1889                        let pull_ident = if 0 < pull_to_push_idx {
1890                            self.node_as_ident(subgraph_nodes[pull_to_push_idx - 1], false)
1891                        } else {
1892                            // Entire subgraph is push (with a single recv/pull handoff input).
1893                            recv_port_idents[0].clone()
1894                        };
1895
1896                        #[rustfmt::skip]
1897                        let push_ident = if let Some(&node_id) =
1898                            subgraph_nodes.get(pull_to_push_idx)
1899                        {
1900                            self.node_as_ident(node_id, false)
1901                        } else if 1 == send_port_idents.len() {
1902                            // Entire subgraph is pull (with a single send/push handoff output).
1903                            send_port_idents[0].clone()
1904                        } else {
1905                            diagnostics.push(Diagnostic::spanned(
1906                                pull_ident.span(),
1907                                Level::Error,
1908                                "Degenerate subgraph detected, is there a disconnected `null()` or other degenerate pipeline somewhere?",
1909                            ));
1910                            continue;
1911                        };
1912
1913                        // Pivot span is combination of pull and push spans (or if not possible, just take the push).
1914                        let pivot_span = pull_ident
1915                            .span()
1916                            .join(push_ident.span())
1917                            .unwrap_or_else(|| push_ident.span());
1918                        let pivot_fn_ident = Ident::new(
1919                            &format!("pivot_run_sg_{:?}", subgraph_id.data()),
1920                            pivot_span,
1921                        );
1922                        let root = change_spans(root.clone(), pivot_span);
1923                        subgraph_op_iter_code.push(quote_spanned! {pivot_span=>
1924                            #[inline(always)]
1925                            fn #pivot_fn_ident<Pul, Psh, Item>(pull: Pul, push: Psh)
1926                                -> impl ::std::future::Future<Output = ()>
1927                            where
1928                                Pul: #root::dfir_pipes::pull::Pull<Item = Item>,
1929                                Psh: #root::dfir_pipes::push::Push<Item, Pul::Meta>,
1930                            {
1931                                #root::dfir_pipes::pull::Pull::send_push(pull, push)
1932                            }
1933                            (#pivot_fn_ident)(#pull_ident, #push_ident).await;
1934                        });
1935                    }
1936                };
1937
1938                // Each subgraph block is an async block so it can be individually instrumented.
1939                // Note: this ident is for the subgraph future, not a runtime SubgraphId binding
1940                // (unlike the scheduled path's `sg_ident`).
1941                let sg_fut_ident = subgraph_id.as_ident(Span::call_site());
1942
1943                // Generate send-side curr_items_count updates (after subgraph runs).
1944                let send_metrics_code = send_hoffs
1945                    .iter()
1946                    .zip(send_buf_idents.iter())
1947                    .zip(send_kinds.iter())
1948                    .map(|((&hoff_id, buf_ident), &kind)| {
1949                        let hoff_ffi = hoff_id.data().as_ffi();
1950                        let len_expr = match kind {
1951                            HandoffKind::Singleton | HandoffKind::Optional => {
1952                                quote! { if #buf_ident.is_some() { 1 } else { 0 } }
1953                            }
1954                            HandoffKind::Vec => {
1955                                quote! { #buf_ident.len() }
1956                            }
1957                        };
1958                        quote! {
1959                            __dfir_metrics.handoffs[
1960                                #root::slotmap::KeyData::from_ffi(#hoff_ffi).into()
1961                            ].curr_items_count.set(#len_expr);
1962                        }
1963                    })
1964                    .collect::<Vec<_>>();
1965
1966                // Create the handoffs we are about to push to (send).
1967                // Exit handoffs (sender inside a loop, receiver in parent) are already declared
1968                // before the while loop, so skip them here.
1969                let send_hoff_make_code = send_buf_idents.iter()
1970                    .zip(send_kinds.iter())
1971                    .zip(send_hoffs.iter())
1972                    .filter_map(|((buf_ident, &kind), &hoff_id)| {
1973                        let span = buf_ident.span();
1974                        if back_edge_hoffs_and_lazyness.contains_key(hoff_id) {
1975                            // Defer_tick send buffers are declared outside the tick closure
1976                            // as std::vec::Vec for O(1) swap. Just clear here.
1977                            Some(quote_spanned! {span=>
1978                                #buf_ident.clear();
1979                            })
1980                        } else {
1981                            // Check if this is a loop-exit handoff: sender is in a loop,
1982                            // receiver is in the parent (already declared outside the while loop).
1983                            let receiver_loop = self
1984                                .node_successors(hoff_id)
1985                                .next()
1986                                .and_then(|(_, succ)| self.node_loop(succ));
1987                            let is_exit = if let Some(sender_loop) = sg_loop {
1988                                receiver_loop == self.loop_parent(sender_loop)
1989                            } else {
1990                                false
1991                            };
1992                            if is_exit {
1993                                // Exit handoff: buffer already declared at parent level.
1994                                None
1995                            } else {
1996                                Some(match kind {
1997                                    HandoffKind::Vec => quote_spanned! {span=>
1998                                        let mut #buf_ident = #root::bumpalo::collections::Vec::new_in(&#bump_ident);
1999                                    },
2000                                    HandoffKind::Singleton | HandoffKind::Optional => quote_spanned! {span=>
2001                                        let mut #buf_ident = ::std::option::Option::None;
2002                                    },
2003                                })
2004                            }
2005                        }
2006                    })
2007                    .collect::<Vec<_>>();
2008                // Drop the handoffs we just drained (recv).
2009                // TODO(mingwei): we could use `.into_iter()` instead of `.drain(..)` to consume the handoffs directly.
2010                // This only works for handoffs within the tick, though, not `defer_tick` handoffs.
2011                let recv_hoff_drop_code = recv_buf_idents
2012                    .iter()
2013                    .zip(recv_hoffs.iter())
2014                    .filter(|&(_, &hoff_id)| !back_edge_hoffs_and_lazyness.contains_key(hoff_id))
2015                    .map(|(buf_ident, _)| {
2016                        let span = buf_ident.span();
2017                        quote_spanned! {span=>
2018                            let _ = #buf_ident;
2019                        }
2020                    });
2021
2022                // Emit subgraph block to the current loop level (top of stack or root).
2023                let sg_block = quote! {
2024                    // Create the handoffs we are about to push to (send).
2025                    #( #send_hoff_make_code )*
2026
2027                    let #sg_fut_ident = async {
2028                        let #context = &#df;
2029                        #( #recv_port_code )*
2030                        #( #send_port_code )*
2031                        #( #subgraph_op_iter_code )*
2032                        #( #subgraph_op_iter_after_code )*
2033                    };
2034                    {
2035                        // Instrument w/ the subgraph metrics.
2036                        let sg_metrics = &__dfir_metrics.subgraphs[
2037                            #root::slotmap::KeyData::from_ffi(#sg_metrics_ffi).into()
2038                        ];
2039                        #root::scheduled::metrics::InstrumentSubgraph::new(
2040                            #sg_fut_ident, sg_metrics
2041                        ).await;
2042                        sg_metrics.total_run_count.update(|x| x + 1);
2043
2044                        // Update send (output) handoff metrics.
2045                        #( #send_metrics_code )*
2046
2047                        // Drop the handoffs we just drained (recv).
2048                        #( #recv_hoff_drop_code )*
2049                    }
2050                };
2051                if let Some((_, body)) = loop_stack.last_mut() {
2052                    body.extend(sg_block);
2053                } else {
2054                    current_output.extend(sg_block);
2055                }
2056            }
2057        }
2058
2059        // Close any remaining open loops.
2060        let gated_subgraph_code = {
2061            while let Some((closed_loop, child_body)) = loop_stack.pop() {
2062                let target = if let Some((_, parent_body)) = loop_stack.last_mut() {
2063                    parent_body
2064                } else {
2065                    &mut current_output
2066                };
2067                self.emit_loop_gate(
2068                    closed_loop,
2069                    child_body,
2070                    &loop_input_handoffs,
2071                    &back_edge_hoffs_and_lazyness,
2072                    &loop_swap_code,
2073                    target,
2074                );
2075            }
2076            current_output
2077        };
2078
2079        if diagnostics.has_error() {
2080            return Err(std::mem::take(diagnostics));
2081        }
2082        let _ = diagnostics; // Ensure no more diagnostics may be added after checking for errors.
2083
2084        let (meta_graph_arg, diagnostics_arg) = if include_meta {
2085            let meta_graph_json = serde_json::to_string(&self).unwrap();
2086            let meta_graph_json = Literal::string(&meta_graph_json);
2087
2088            let serde_diagnostics: Vec<_> = diagnostics.iter().map(Diagnostic::to_serde).collect();
2089            let diagnostics_json = serde_json::to_string(&*serde_diagnostics).unwrap();
2090            let diagnostics_json = Literal::string(&diagnostics_json);
2091
2092            (
2093                quote! { Some(#meta_graph_json) },
2094                quote! { Some(#diagnostics_json) },
2095            )
2096        } else {
2097            (quote! { None }, quote! { None })
2098        };
2099
2100        // Generate metrics initialization: one entry per handoff and per subgraph.
2101        let metrics_init_code = {
2102            let handoff_inits = handoff_nodes.iter().map(|&(node_id, _, _)| {
2103                let ffi = node_id.data().as_ffi();
2104                quote! {
2105                    dfir_metrics.handoffs.insert(
2106                        #root::slotmap::KeyData::from_ffi(#ffi).into(),
2107                        ::std::default::Default::default(),
2108                    );
2109                }
2110            });
2111            let subgraph_inits = all_subgraphs.iter().map(|&(sg_id, _)| {
2112                let ffi = sg_id.data().as_ffi();
2113                quote! {
2114                    dfir_metrics.subgraphs.insert(
2115                        #root::slotmap::KeyData::from_ffi(#ffi).into(),
2116                        ::std::default::Default::default(),
2117                    );
2118                }
2119            });
2120            handoff_inits.chain(subgraph_inits).collect::<Vec<_>>()
2121        };
2122
2123        // For creating back-buffer handoff vecs.
2124        let back_buffer_idents = back_buffer_idents_laziness
2125            .iter()
2126            .map(|(back_ident, _, _)| back_ident);
2127        // For creating the send-side buffer for defer_tick handoffs (also outside the closure).
2128        let defer_tick_buf_idents = back_buffer_idents_laziness
2129            .iter()
2130            .map(|(_, buf_ident, _)| buf_ident);
2131        // For checking if we should start the next tick (`schedule_subgraph`):
2132        // Collect the ident to check for each non-lazy back-edge handoff.
2133        // - For defer_tick handoffs in a root-level loop: check `back` (swap happened inside `if`)
2134        // - For all others: check `buf` (original behavior; tick-level swap hasn't happened yet)
2135        let non_lazy_schedule_idents: Vec<&Ident> = handoff_nodes
2136            .iter()
2137            .filter_map(|&(hoff_id, _, _)| {
2138                let delay_type = self.handoff_delay_type(hoff_id)?;
2139                // Only non-lazy.
2140                if matches!(delay_type, DelayType::TickLazy | DelayType::LoopLazy) {
2141                    return None;
2142                }
2143                let span = self.nodes[hoff_id].span();
2144                let expected_back_ident = self.hoff_back_ident(hoff_id, span);
2145                let entry = back_buffer_idents_laziness
2146                    .iter()
2147                    .find(|(back_ident, _, _)| *back_ident == expected_back_ident)?;
2148
2149                // For defer_tick inside a root-level loop, check `back`.
2150                if delay_type == DelayType::Tick {
2151                    let consumer_loop = self
2152                        .node_successors(hoff_id)
2153                        .next()
2154                        .and_then(|(_, succ)| self.node_subgraph(succ))
2155                        .and_then(|sg| self.subgraph_loop(sg));
2156                    if consumer_loop.is_some_and(|lid| self.loop_parent(lid).is_none()) {
2157                        return Some(&entry.0); // back ident
2158                    }
2159                }
2160                Some(&entry.1) // buf ident
2161            })
2162            .collect();
2163
2164        // Prologues and buffer declarations persist across ticks (outside the closure).
2165        // Subgraph blocks run each tick (inside the closure).
2166        Ok(quote! {
2167            {
2168                #prefix
2169
2170                use #root::{var_expr, var_args};
2171
2172                let __dfir_wake_state = ::std::sync::Arc::new(
2173                    #root::scheduled::context::WakeState::default()
2174                );
2175
2176                let __dfir_metrics = {
2177                    let mut dfir_metrics = #root::scheduled::metrics::DfirMetrics::default();
2178                    #( #metrics_init_code )*
2179                    ::std::rc::Rc::new(dfir_metrics)
2180                };
2181
2182                #[allow(unused_mut)]
2183                let mut #df = #root::scheduled::context::Context::new(
2184                    ::std::clone::Clone::clone(&__dfir_wake_state),
2185                    __dfir_metrics,
2186                );
2187
2188                #( #op_prologue_code )*
2189
2190                // For tick-boundary handoffs (`defer_tick` / `defer_tick_lazy`), declare both the
2191                // send buffer and the "back" buffer as std::vec::Vec outside the tick closure.
2192                // This enables O(1) mem::swap at end of tick for double-buffering.
2193                #( let mut #back_buffer_idents = ::std::vec::Vec::new(); )*
2194                #( let mut #defer_tick_buf_idents = ::std::vec::Vec::new(); )*
2195
2196                // Bump allocator for handoffs (except for back-edge handoffs, above).
2197                let mut #bump_ident = #root::bumpalo::Bump::new();
2198
2199                // Pre-set to true so the first tick always returns true
2200                // (matching Dfir pre-scheduling behavior). Subsequent ticks
2201                // start false (from take()) and are set true by recv port code
2202                // if any handoff buffer has data.
2203                let mut __dfir_work_done = true;
2204                #[allow(unused_qualifications, unused_mut, unused_variables, clippy::await_holding_refcell_ref, clippy::deref_addrof)]
2205                let __dfir_inline_tick = async move |#df: &mut #root::scheduled::context::Context| {
2206                    // Reset arena between ticks (start-of-tick)
2207                    #bump_ident.reset();
2208
2209                    {
2210                        let __dfir_metrics = #df.metrics();
2211
2212                        #gated_subgraph_code
2213
2214                        // For non-lazy defer_tick: if any deferred buffer has data,
2215                        // signal that another tick should run.
2216                        #[allow(clippy::nonminimal_bool, reason = "codegen")]
2217                        if false #( || !#non_lazy_schedule_idents.is_empty() )* {
2218                            #df.schedule_subgraph(true);
2219                        }
2220
2221                        // Double-buffer swap for defer_tick handoffs: move last tick's producer output (regular buffer)
2222                        // into the back buffer for the consumer to drain.
2223                        #( #back_edge_swap_code )*
2224                    }
2225
2226                    // End-of-tick per-operator state handling (i.e. 'tick persistence).
2227                    #( #op_tick_end_code )*
2228
2229                    #df.__end_tick();
2230
2231                    ::std::mem::take(&mut __dfir_work_done)
2232                };
2233                #root::scheduled::context::Dfir::new(
2234                    __dfir_inline_tick,
2235                    #df,
2236                    #meta_graph_arg,
2237                    #diagnostics_arg,
2238                )
2239            }
2240        })
2241    }
2242
2243    /// Color mode (pull vs. push, handoff vs. comp) for nodes. Some nodes can be push *OR* pull;
2244    /// those nodes will not be set in the returned map.
2245    pub fn node_color_map(&self) -> SparseSecondaryMap<GraphNodeId, Color> {
2246        let mut node_color_map: SparseSecondaryMap<GraphNodeId, Color> = self
2247            .node_ids()
2248            .filter_map(|node_id| {
2249                let op_color = self.node_color(node_id)?;
2250                Some((node_id, op_color))
2251            })
2252            .collect();
2253
2254        // Fill in rest via subgraphs.
2255        for sg_nodes in self.subgraph_nodes.values() {
2256            let pull_to_push_idx = self.find_pull_to_push_idx(sg_nodes);
2257
2258            for (idx, node_id) in sg_nodes.iter().copied().enumerate() {
2259                let is_pull = idx < pull_to_push_idx;
2260                node_color_map.insert(node_id, if is_pull { Color::Pull } else { Color::Push });
2261            }
2262        }
2263
2264        node_color_map
2265    }
2266
2267    /// Writes this graph as mermaid into a string.
2268    pub fn to_mermaid(&self, write_config: &WriteConfig) -> String {
2269        let mut output = String::new();
2270        self.write_mermaid(&mut output, write_config).unwrap();
2271        output
2272    }
2273
2274    /// Writes this graph as mermaid into the given `Write`.
2275    pub fn write_mermaid(
2276        &self,
2277        output: impl std::fmt::Write,
2278        write_config: &WriteConfig,
2279    ) -> std::fmt::Result {
2280        let mut graph_write = Mermaid::new(output);
2281        self.write_graph(&mut graph_write, write_config)
2282    }
2283
2284    /// Writes this graph as DOT (graphviz) into a string.
2285    pub fn to_dot(&self, write_config: &WriteConfig) -> String {
2286        let mut output = String::new();
2287        let mut graph_write = Dot::new(&mut output);
2288        self.write_graph(&mut graph_write, write_config).unwrap();
2289        output
2290    }
2291
2292    /// Writes this graph as DOT (graphviz) into the given `Write`.
2293    pub fn write_dot(
2294        &self,
2295        output: impl std::fmt::Write,
2296        write_config: &WriteConfig,
2297    ) -> std::fmt::Result {
2298        let mut graph_write = Dot::new(output);
2299        self.write_graph(&mut graph_write, write_config)
2300    }
2301
2302    /// Write out this graph using the given `GraphWrite`. E.g. `Mermaid` or `Dot.
2303    pub(crate) fn write_graph<W>(
2304        &self,
2305        mut graph_write: W,
2306        write_config: &WriteConfig,
2307    ) -> Result<(), W::Err>
2308    where
2309        W: GraphWrite,
2310    {
2311        fn helper_edge_label(
2312            src_port: &PortIndexValue,
2313            dst_port: &PortIndexValue,
2314        ) -> Option<String> {
2315            let src_label = match src_port {
2316                PortIndexValue::Path(path) => Some(path.to_token_stream().to_string()),
2317                PortIndexValue::Int(index) => Some(index.value.to_string()),
2318                _ => None,
2319            };
2320            let dst_label = match dst_port {
2321                PortIndexValue::Path(path) => Some(path.to_token_stream().to_string()),
2322                PortIndexValue::Int(index) => Some(index.value.to_string()),
2323                _ => None,
2324            };
2325            let label = match (src_label, dst_label) {
2326                (Some(l1), Some(l2)) => Some(format!("{}\n{}", l1, l2)),
2327                (Some(l1), None) => Some(l1),
2328                (None, Some(l2)) => Some(l2),
2329                (None, None) => None,
2330            };
2331            label
2332        }
2333
2334        // Make node color map one time.
2335        let node_color_map = self.node_color_map();
2336
2337        // Write prologue.
2338        graph_write.write_prologue()?;
2339
2340        // Define nodes.
2341        let mut skipped_handoffs = BTreeSet::new();
2342        for (node_id, node) in self.nodes() {
2343            if matches!(node, GraphNode::Handoff { .. }) && write_config.no_handoffs {
2344                skipped_handoffs.insert(node_id);
2345                continue;
2346            }
2347            graph_write.write_node_definition(
2348                node_id,
2349                &if write_config.op_short_text {
2350                    node.to_name_string()
2351                } else if write_config.op_text_no_imports {
2352                    // Remove any lines that start with "use" (imports)
2353                    let full_text = node.to_pretty_string();
2354                    let mut output = String::new();
2355                    for sentence in full_text.split('\n') {
2356                        if sentence.trim().starts_with("use") {
2357                            continue;
2358                        }
2359                        output.push('\n');
2360                        output.push_str(sentence);
2361                    }
2362                    output.into()
2363                } else {
2364                    node.to_pretty_string()
2365                },
2366                if write_config.no_pull_push {
2367                    None
2368                } else {
2369                    node_color_map.get(node_id).copied()
2370                },
2371            )?;
2372        }
2373
2374        // Write edges.
2375        for (edge_id, (src_id, mut dst_id)) in self.edges() {
2376            // Handling for if `write_config.no_handoffs` true.
2377            if skipped_handoffs.contains(&src_id) {
2378                continue;
2379            }
2380
2381            let (src_port, mut dst_port) = self.edge_ports(edge_id);
2382            if skipped_handoffs.contains(&dst_id) {
2383                // The destination is a hidden handoff. If it has a successor, skip through.
2384                // If it has 0 successors (ref-only singleton), drop this edge entirely —
2385                // the data dependency is captured via the reference edge instead.
2386                let mut handoff_succs = self.node_successors(dst_id);
2387                if handoff_succs.len() == 0 {
2388                    continue;
2389                }
2390                let (succ_edge, succ_node) = handoff_succs.next().unwrap();
2391                dst_id = succ_node;
2392                dst_port = self.edge_ports(succ_edge).1;
2393            }
2394
2395            let label = helper_edge_label(src_port, dst_port);
2396            let delay_type = self
2397                .node_op_inst(dst_id)
2398                .and_then(|op_inst| (op_inst.op_constraints.input_delaytype_fn)(dst_port));
2399            graph_write.write_edge(src_id, dst_id, delay_type, label.as_deref(), false)?;
2400        }
2401
2402        // Write reference edges.
2403        if !write_config.no_references {
2404            for dst_id in self.node_ids() {
2405                for src_ref_id in self
2406                    .node_handoff_references(dst_id)
2407                    .iter()
2408                    .filter_map(|r| r.node_id)
2409                {
2410                    // When handoffs are hidden, resolve through to the predecessor of
2411                    // the singleton handoff so the edge points from the actual writer.
2412                    let resolved_src = if skipped_handoffs.contains(&src_ref_id) {
2413                        self.node_predecessor_nodes(src_ref_id).next()
2414                    } else {
2415                        Some(src_ref_id)
2416                    };
2417                    let Some(resolved_src) = resolved_src else {
2418                        continue;
2419                    };
2420                    let label = None;
2421                    graph_write.write_edge(resolved_src, dst_id, None, label, true)?;
2422                }
2423            }
2424        }
2425
2426        // The following code is a little bit tricky. Generally, the graph has the hierarchy:
2427        // `loop -> subgraph -> varname -> node`. However, each of these can be disabled via the `write_config`. To
2428        // handle both the enabled and disabled case, this code is structured as a series of nested loops. If the layer
2429        // is disabled, then the HashMap<Option<KEY>, Vec<VALUE>> will only have a single key (`None`) with a
2430        // corresponding `Vec` value containing everything. This way no special handling is needed for the next layer.
2431
2432        // Loop -> Subgraphs
2433        let loop_subgraphs = self.subgraph_ids().map(|sg_id| {
2434            let loop_id = if write_config.no_loops {
2435                None
2436            } else {
2437                self.subgraph_loop(sg_id)
2438            };
2439            (loop_id, sg_id)
2440        });
2441        let loop_subgraphs = into_group_map(loop_subgraphs);
2442        for (loop_id, subgraph_ids) in loop_subgraphs {
2443            if let Some(loop_id) = loop_id {
2444                graph_write.write_loop_start(loop_id)?;
2445            }
2446
2447            // Subgraph -> Varnames.
2448            let subgraph_varnames_nodes = subgraph_ids.into_iter().flat_map(|sg_id| {
2449                self.subgraph(sg_id).iter().copied().map(move |node_id| {
2450                    let opt_sg_id = if write_config.no_subgraphs {
2451                        None
2452                    } else {
2453                        Some(sg_id)
2454                    };
2455                    (opt_sg_id, (self.node_varname(node_id), node_id))
2456                })
2457            });
2458            let subgraph_varnames_nodes = into_group_map(subgraph_varnames_nodes);
2459            for (sg_id, varnames) in subgraph_varnames_nodes {
2460                if let Some(sg_id) = sg_id {
2461                    graph_write.write_subgraph_start(sg_id)?;
2462                }
2463
2464                // Varnames -> Nodes.
2465                let varname_nodes = varnames.into_iter().map(|(varname, node)| {
2466                    let varname = if write_config.no_varnames {
2467                        None
2468                    } else {
2469                        varname
2470                    };
2471                    (varname, node)
2472                });
2473                let varname_nodes = into_group_map(varname_nodes);
2474                for (varname, node_ids) in varname_nodes {
2475                    if let Some(varname) = varname {
2476                        graph_write.write_varname_start(&varname.0.to_string(), sg_id)?;
2477                    }
2478
2479                    // Write all nodes.
2480                    for node_id in node_ids {
2481                        graph_write.write_node(node_id)?;
2482                    }
2483
2484                    if varname.is_some() {
2485                        graph_write.write_varname_end()?;
2486                    }
2487                }
2488
2489                if sg_id.is_some() {
2490                    graph_write.write_subgraph_end()?;
2491                }
2492            }
2493
2494            if loop_id.is_some() {
2495                graph_write.write_loop_end()?;
2496            }
2497        }
2498
2499        // Write epilogue.
2500        graph_write.write_epilogue()?;
2501
2502        Ok(())
2503    }
2504
2505    /// Convert back into surface syntax.
2506    pub fn surface_syntax_string(&self) -> String {
2507        let mut string = String::new();
2508        self.write_surface_syntax(&mut string).unwrap();
2509        string
2510    }
2511
2512    /// Convert back into surface syntax.
2513    pub fn write_surface_syntax(&self, write: &mut impl std::fmt::Write) -> std::fmt::Result {
2514        for (key, node) in self.nodes.iter() {
2515            match node {
2516                GraphNode::Operator(op) => {
2517                    writeln!(write, "_{:?} = {};", key.data(), op.to_token_stream())?;
2518                }
2519                GraphNode::Handoff {
2520                    kind: HandoffKind::Vec,
2521                    ..
2522                } => {
2523                    writeln!(write, "_{:?} = handoff();", key.data())?;
2524                }
2525                GraphNode::Handoff {
2526                    kind: HandoffKind::Singleton,
2527                    ..
2528                } => {
2529                    writeln!(write, "_{:?} = singleton();", key.data())?;
2530                }
2531                GraphNode::Handoff {
2532                    kind: HandoffKind::Optional,
2533                    ..
2534                } => {
2535                    writeln!(write, "_{:?} = optional();", key.data())?;
2536                }
2537                GraphNode::ModuleBoundary { .. } => panic!(),
2538            }
2539        }
2540        writeln!(write)?;
2541        for (e, (src_key, dst_key)) in self.graph.edges() {
2542            let (src_port, dst_port) = self.edge_ports(e);
2543            let src_port_str = if src_port.is_specified() {
2544                format!("[{}]", src_port)
2545            } else {
2546                String::new()
2547            };
2548            let dst_port_str = if dst_port.is_specified() {
2549                format!("[{}]", dst_port)
2550            } else {
2551                String::new()
2552            };
2553            writeln!(
2554                write,
2555                "_{:?}{} -> {}_{:?};",
2556                src_key.data(),
2557                src_port_str,
2558                dst_port_str,
2559                dst_key.data()
2560            )?;
2561        }
2562        Ok(())
2563    }
2564
2565    /// Convert into a [mermaid](https://mermaid-js.github.io/) graph. Ignores subgraphs.
2566    pub fn mermaid_string_flat(&self) -> String {
2567        let mut string = String::new();
2568        self.write_mermaid_flat(&mut string).unwrap();
2569        string
2570    }
2571
2572    /// Convert into a [mermaid](https://mermaid-js.github.io/) graph. Ignores subgraphs.
2573    pub fn write_mermaid_flat(&self, write: &mut impl std::fmt::Write) -> std::fmt::Result {
2574        writeln!(write, "flowchart TB")?;
2575        for (key, node) in self.nodes.iter() {
2576            match node {
2577                GraphNode::Operator(operator) => writeln!(
2578                    write,
2579                    "    %% {span}\n    {id:?}[\"{row_col} <tt>{code}</tt>\"]",
2580                    span = PrettySpan(node.span()),
2581                    id = key.data(),
2582                    row_col = PrettyRowCol(node.span()),
2583                    code = operator
2584                        .to_token_stream()
2585                        .to_string()
2586                        .replace('&', "&amp;")
2587                        .replace('<', "&lt;")
2588                        .replace('>', "&gt;")
2589                        .replace('"', "&quot;")
2590                        .replace('\n', "<br>"),
2591                ),
2592                GraphNode::Handoff {
2593                    kind: HandoffKind::Vec,
2594                    ..
2595                } => {
2596                    writeln!(write, r#"    {:?}{{"{}"}}"#, key.data(), HANDOFF_NODE_STR)
2597                }
2598                GraphNode::Handoff {
2599                    kind: HandoffKind::Singleton | HandoffKind::Optional,
2600                    ..
2601                } => {
2602                    writeln!(
2603                        write,
2604                        r#"    {:?}{{"{}"}}"#,
2605                        key.data(),
2606                        SINGLETON_SLOT_NODE_STR
2607                    )
2608                }
2609                GraphNode::ModuleBoundary { .. } => {
2610                    writeln!(
2611                        write,
2612                        r#"    {:?}{{"{}"}}"#,
2613                        key.data(),
2614                        MODULE_BOUNDARY_NODE_STR
2615                    )
2616                }
2617            }?;
2618        }
2619        writeln!(write)?;
2620        for (_e, (src_key, dst_key)) in self.graph.edges() {
2621            writeln!(write, "    {:?}-->{:?}", src_key.data(), dst_key.data())?;
2622        }
2623        Ok(())
2624    }
2625}
2626
2627/// Loops
2628impl DfirGraph {
2629    /// Iterator over all loop IDs.
2630    pub fn loop_ids(&self) -> slotmap::basic::Keys<'_, GraphLoopId, Vec<GraphNodeId>> {
2631        self.loop_nodes.keys()
2632    }
2633
2634    /// Iterator over all loops, ID and members: `(GraphLoopId, Vec<GraphNodeId>)`.
2635    pub fn loops(&self) -> slotmap::basic::Iter<'_, GraphLoopId, Vec<GraphNodeId>> {
2636        self.loop_nodes.iter()
2637    }
2638
2639    /// Get a loop's member nodes.
2640    pub fn loop_nodes(&self, loop_id: GraphLoopId) -> &[GraphNodeId] {
2641        self.loop_nodes.get(loop_id).unwrap()
2642    }
2643
2644    /// Create a new loop context, with the given parent loop (or `None`).
2645    pub fn insert_loop(&mut self, parent_loop: Option<GraphLoopId>) -> GraphLoopId {
2646        let loop_id = self.loop_nodes.insert(Vec::new());
2647        self.loop_children.insert(loop_id, Vec::new());
2648        if let Some(parent_loop) = parent_loop {
2649            self.loop_parent.insert(loop_id, parent_loop);
2650            self.loop_children
2651                .get_mut(parent_loop)
2652                .unwrap()
2653                .push(loop_id);
2654        } else {
2655            self.root_loops.push(loop_id);
2656        }
2657        loop_id
2658    }
2659
2660    /// Get a node's loop context (or `None` for root).
2661    pub fn node_loop(&self, node_id: GraphNodeId) -> Option<GraphLoopId> {
2662        self.node_loops.get(node_id).copied()
2663    }
2664
2665    /// Get a subgraph's loop context (or `None` for root).
2666    pub fn subgraph_loop(&self, subgraph_id: GraphSubgraphId) -> Option<GraphLoopId> {
2667        let &node_id = self.subgraph(subgraph_id).first().unwrap();
2668        let out = self.node_loop(node_id);
2669        debug_assert!(
2670            self.subgraph(subgraph_id)
2671                .iter()
2672                .all(|&node_id| self.node_loop(node_id) == out),
2673            "Subgraph nodes should all have the same loop context."
2674        );
2675        out
2676    }
2677
2678    /// Get a loop context's parent loop context (or `None` for root).
2679    pub fn loop_parent(&self, loop_id: GraphLoopId) -> Option<GraphLoopId> {
2680        self.loop_parent.get(loop_id).copied()
2681    }
2682
2683    /// Get a loop context's child loops.
2684    pub fn loop_children(&self, loop_id: GraphLoopId) -> &Vec<GraphLoopId> {
2685        self.loop_children.get(loop_id).unwrap()
2686    }
2687
2688    /// Get root-level loops (those with no parent loop).
2689    pub fn root_loops(&self) -> &[GraphLoopId] {
2690        &self.root_loops
2691    }
2692}
2693
2694/// Configuration for writing graphs.
2695#[derive(Clone, Debug, Default)]
2696#[cfg_attr(feature = "clap-derive", derive(clap::Args))]
2697pub struct WriteConfig {
2698    /// Subgraphs will not be rendered if set.
2699    #[cfg_attr(feature = "clap-derive", arg(long))]
2700    pub no_subgraphs: bool,
2701    /// Variable names will not be rendered if set.
2702    #[cfg_attr(feature = "clap-derive", arg(long))]
2703    pub no_varnames: bool,
2704    /// Will not render pull/push shapes if set.
2705    #[cfg_attr(feature = "clap-derive", arg(long))]
2706    pub no_pull_push: bool,
2707    /// Will not render handoffs if set.
2708    #[cfg_attr(feature = "clap-derive", arg(long))]
2709    pub no_handoffs: bool,
2710    /// Will not render singleton references if set.
2711    #[cfg_attr(feature = "clap-derive", arg(long))]
2712    pub no_references: bool,
2713    /// Will not render loops if set.
2714    #[cfg_attr(feature = "clap-derive", arg(long))]
2715    pub no_loops: bool,
2716
2717    /// Op text will only be their name instead of the whole source.
2718    #[cfg_attr(feature = "clap-derive", arg(long))]
2719    pub op_short_text: bool,
2720    /// Op text will exclude any line that starts with "use".
2721    #[cfg_attr(feature = "clap-derive", arg(long))]
2722    pub op_text_no_imports: bool,
2723}
2724
2725/// Enum for choosing between mermaid and dot graph writing.
2726#[derive(Copy, Clone, Debug)]
2727#[cfg_attr(feature = "clap-derive", derive(clap::Parser, clap::ValueEnum))]
2728pub enum WriteGraphType {
2729    /// Mermaid graphs.
2730    Mermaid,
2731    /// Dot (Graphviz) graphs.
2732    Dot,
2733}
2734
2735/// [`itertools::Itertools::into_group_map`], but for `BTreeMap`.
2736fn into_group_map<K, V>(iter: impl IntoIterator<Item = (K, V)>) -> BTreeMap<K, Vec<V>>
2737where
2738    K: Ord,
2739{
2740    let mut out: BTreeMap<_, Vec<_>> = BTreeMap::new();
2741    for (k, v) in iter {
2742        out.entry(k).or_default().push(v);
2743    }
2744    out
2745}