dfir_lang/graph/
mod.rs

1//! Graph representation stages for DFIR graphs.
2
3use std::borrow::Cow;
4use std::hash::Hash;
5
6use proc_macro2::{Ident, Span, TokenStream};
7use quote::ToTokens;
8use serde::{Deserialize, Serialize};
9use slotmap::new_key_type;
10use syn::punctuated::Punctuated;
11use syn::spanned::Spanned;
12use syn::{Expr, ExprPath, GenericArgument, Token, Type};
13
14use self::ops::{OperatorConstraints, Persistence};
15use crate::diagnostic::{Diagnostic, Diagnostics, Level};
16use crate::parse::{DfirCode, IndexInt, Operator, PortIndex, Ported};
17use crate::pretty_span::PrettySpan;
18
19mod di_mul_graph;
20mod eliminate_extra_unions_tees;
21mod flat_graph_builder;
22mod flat_to_partitioned;
23mod graph_write;
24mod meta_graph;
25mod meta_graph_debugging;
26
27use std::fmt::Display;
28
29pub use di_mul_graph::DiMulGraph;
30pub use eliminate_extra_unions_tees::eliminate_extra_unions_tees;
31pub use flat_graph_builder::{FlatGraphBuilder, FlatGraphBuilderOutput};
32pub use flat_to_partitioned::partition_graph;
33pub use meta_graph::{DfirGraph, WriteConfig, WriteGraphType};
34
35pub mod graph_algorithms;
36pub mod ops;
37
38new_key_type! {
39    /// ID to identify a node (operator or handoff) in [`DfirGraph`].
40    pub struct GraphNodeId;
41
42    /// ID to identify an edge.
43    pub struct GraphEdgeId;
44
45    /// ID to identify a subgraph in [`DfirGraph`].
46    pub struct GraphSubgraphId;
47
48    /// ID to identify a loop block in [`DfirGraph`].
49    pub struct GraphLoopId;
50}
51
52impl GraphSubgraphId {
53    /// Generate a deterministic `Ident` for the given loop ID.
54    pub fn as_ident(self, span: Span) -> Ident {
55        use slotmap::Key;
56        Ident::new(&format!("sgid_{:?}", self.data()), span)
57    }
58}
59
60impl GraphLoopId {
61    /// Generate a deterministic `Ident` for the given loop ID.
62    pub fn as_ident(self, span: Span) -> Ident {
63        use slotmap::Key;
64        Ident::new(&format!("loop_{:?}", self.data()), span)
65    }
66}
67
68/// Context identifier as a string.
69const CONTEXT: &str = "context";
70/// Runnable DFIR graph object identifier as a string.
71const GRAPH: &str = "df";
72
73const HANDOFF_NODE_STR: &str = "handoff";
74const MODULE_BOUNDARY_NODE_STR: &str = "module_boundary";
75
76mod serde_syn {
77    use serde::{Deserialize, Deserializer, Serializer};
78
79    pub fn serialize<S, T>(value: &T, serializer: S) -> Result<S::Ok, S::Error>
80    where
81        S: Serializer,
82        T: quote::ToTokens,
83    {
84        serializer.serialize_str(&value.to_token_stream().to_string())
85    }
86
87    pub fn deserialize<'de, D, T>(deserializer: D) -> Result<T, D::Error>
88    where
89        D: Deserializer<'de>,
90        T: syn::parse::Parse,
91    {
92        let s = String::deserialize(deserializer)?;
93        syn::parse_str(&s).map_err(<D::Error as serde::de::Error>::custom)
94    }
95}
96
97/// A variable name assigned to a pipeline in DFIR syntax.
98///
99/// Fundamentally a serializable/deserializable wrapper around [`syn::Ident`].
100#[derive(Clone, Debug, Serialize, Deserialize, PartialOrd, Ord, PartialEq, Eq, Hash)]
101pub struct Varname(#[serde(with = "serde_syn")] pub Ident);
102
103/// A node, corresponding to an operator or a handoff.
104#[derive(Clone, Serialize, Deserialize)]
105pub enum GraphNode {
106    /// An operator.
107    Operator(#[serde(with = "serde_syn")] Operator),
108    /// A handoff point, used between subgraphs (or within a subgraph to break a cycle).
109    Handoff {
110        /// The span of the input into the handoff.
111        #[serde(skip, default = "Span::call_site")]
112        src_span: Span,
113        /// The span of the output out of the handoff.
114        #[serde(skip, default = "Span::call_site")]
115        dst_span: Span,
116    },
117
118    /// Module Boundary, used for importing modules. Only exists prior to partitioning.
119    ModuleBoundary {
120        /// If this module is an input or output boundary.
121        input: bool,
122
123        /// The span of the import!() expression that imported this module.
124        /// The value of this span when the ModuleBoundary node is still inside the module is Span::call_site()
125        /// TODO: This could one day reference into the module file itself?
126        #[serde(skip, default = "Span::call_site")]
127        import_expr: Span,
128    },
129}
130impl GraphNode {
131    /// Return the node as a human-readable string.
132    pub fn to_pretty_string(&self) -> Cow<'static, str> {
133        match self {
134            GraphNode::Operator(op) => op.to_pretty_string().into(),
135            GraphNode::Handoff { .. } => HANDOFF_NODE_STR.into(),
136            GraphNode::ModuleBoundary { .. } => MODULE_BOUNDARY_NODE_STR.into(),
137        }
138    }
139
140    /// Return the name of the node as a string, excluding parenthesis and op source code.
141    pub fn to_name_string(&self) -> Cow<'static, str> {
142        match self {
143            GraphNode::Operator(op) => op.name_string().into(),
144            GraphNode::Handoff { .. } => HANDOFF_NODE_STR.into(),
145            GraphNode::ModuleBoundary { .. } => MODULE_BOUNDARY_NODE_STR.into(),
146        }
147    }
148
149    /// Return the source code span of the node (for operators) or input/otput spans for handoffs.
150    pub fn span(&self) -> Span {
151        match self {
152            Self::Operator(op) => op.span(),
153            &Self::Handoff {
154                src_span, dst_span, ..
155            } => src_span.join(dst_span).unwrap_or(src_span),
156            Self::ModuleBoundary { import_expr, .. } => *import_expr,
157        }
158    }
159}
160impl std::fmt::Debug for GraphNode {
161    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
162        match self {
163            Self::Operator(operator) => {
164                write!(f, "Node::Operator({} span)", PrettySpan(operator.span()))
165            }
166            Self::Handoff { .. } => write!(f, "Node::Handoff"),
167            Self::ModuleBoundary { input, .. } => {
168                write!(f, "Node::ModuleBoundary{{input: {}}}", input)
169            }
170        }
171    }
172}
173
174/// Meta-data relating to operators which may be useful throughout the compilation process.
175///
176/// This data can be generated from the graph, but it is useful to have it readily available
177/// pre-computed as many algorithms use the same info. Stuff like port names, arguments, and the
178/// [`OperatorConstraints`] for the operator.
179///
180/// Because it is derived from the graph itself, there can be "cache invalidation"-esque issues
181/// if this data is not kept in sync with the graph.
182#[derive(Clone, Debug)]
183pub struct OperatorInstance {
184    /// Name of the operator (will match [`OperatorConstraints::name`]).
185    pub op_constraints: &'static OperatorConstraints,
186    /// Port values used as this operator's input.
187    pub input_ports: Vec<PortIndexValue>,
188    /// Port values used as this operator's output.
189    pub output_ports: Vec<PortIndexValue>,
190    /// Singleton references within the operator arguments.
191    pub singletons_referenced: Vec<Ident>,
192
193    /// Generic arguments.
194    pub generics: OpInstGenerics,
195    /// Arguments provided by the user into the operator as arguments.
196    /// I.e. the `a, b, c` in `-> my_op(a, b, c) -> `.
197    ///
198    /// These arguments do not include singleton postprocessing codegen. Instead use
199    /// [`ops::WriteContextArgs::arguments`].
200    pub arguments_pre: Punctuated<Expr, Token![,]>,
201    /// Unparsed arguments, for singleton parsing.
202    pub arguments_raw: TokenStream,
203}
204
205/// Operator generic arguments, split into specific categories.
206#[derive(Clone, Debug)]
207pub struct OpInstGenerics {
208    /// Operator generic (type or lifetime) arguments.
209    pub generic_args: Option<Punctuated<GenericArgument, Token![,]>>,
210    /// Lifetime persistence arguments. Corresponds to a prefix of [`Self::generic_args`].
211    pub persistence_args: Vec<Persistence>,
212    /// Type persistence arguments. Corersponds to a (suffix) of [`Self::generic_args`].
213    pub type_args: Vec<Type>,
214}
215
216/// Gets the generic arguments for the operator.
217///
218/// This helper method is useful due to the special handling of persistence lifetimes (`'static`,
219/// `'tick`, `'mutable`) which must come before other generic parameters.
220pub fn get_operator_generics(diagnostics: &mut Diagnostics, operator: &Operator) -> OpInstGenerics {
221    // Generic arguments.
222    let generic_args = operator.type_arguments().cloned();
223    let persistence_args = generic_args.iter().flatten().map_while(|generic_arg| match generic_arg {
224            GenericArgument::Lifetime(lifetime) => {
225                match &*lifetime.ident.to_string() {
226                    "none" => Some(Persistence::None),
227                    "loop" => Some(Persistence::Loop),
228                    "tick" => Some(Persistence::Tick),
229                    "static" => Some(Persistence::Static),
230                    "mutable" => Some(Persistence::Mutable),
231                    _ => {
232                        diagnostics.push(Diagnostic::spanned(
233                            generic_arg.span(),
234                            Level::Error,
235                            format!("Unknown lifetime generic argument `'{}`, expected `'none`, `'loop`, `'tick`, `'static`, or `'mutable`.", lifetime.ident),
236                        ));
237                        // TODO(mingwei): should really keep going and not short circuit?
238                        None
239                    }
240                }
241            },
242            _ => None,
243        }).collect::<Vec<_>>();
244    let type_args = generic_args
245        .iter()
246        .flatten()
247        .skip(persistence_args.len())
248        .map_while(|generic_arg| match generic_arg {
249            GenericArgument::Type(typ) => Some(typ),
250            _ => None,
251        })
252        .cloned()
253        .collect::<Vec<_>>();
254
255    OpInstGenerics {
256        generic_args,
257        persistence_args,
258        type_args,
259    }
260}
261
262/// Push, Pull, Comp, or Hoff polarity.
263#[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord, Serialize, Deserialize)]
264pub enum Color {
265    /// Pull (green)
266    Pull,
267    /// Push (blue)
268    Push,
269    /// Computation (yellow)
270    Comp,
271    /// Handoff (grey) -- not a color for operators, inserted between subgraphs.
272    Hoff,
273}
274
275/// Helper struct for [`PortIndex`] which keeps span information for elided ports.
276#[derive(Clone, Debug, Serialize, Deserialize)]
277pub enum PortIndexValue {
278    /// An integer value: `[0]`, `[1]`, etc. Can be negative although we don't use that (2023-08-16).
279    Int(#[serde(with = "serde_syn")] IndexInt),
280    /// A name or path. `[pos]`, `[neg]`, etc. Can use `::` separators but we don't use that (2023-08-16).
281    Path(#[serde(with = "serde_syn")] ExprPath),
282    /// Elided, unspecified port. We have this variant, rather than wrapping in `Option`, in order
283    /// to preserve the `Span` information.
284    Elided(#[serde(skip)] Option<Span>),
285}
286impl PortIndexValue {
287    /// For a [`Ported`] value like `[port_in]name[port_out]`, get the `port_in` and `port_out` as
288    /// [`PortIndexValue`]s.
289    pub fn from_ported<Inner>(ported: Ported<Inner>) -> (Self, Inner, Self)
290    where
291        Inner: Spanned,
292    {
293        let ported_span = Some(ported.inner.span());
294        let port_inn = ported
295            .inn
296            .map(|idx| idx.index.into())
297            .unwrap_or_else(|| Self::Elided(ported_span));
298        let inner = ported.inner;
299        let port_out = ported
300            .out
301            .map(|idx| idx.index.into())
302            .unwrap_or_else(|| Self::Elided(ported_span));
303        (port_inn, inner, port_out)
304    }
305
306    /// Returns `true` if `self` is not [`PortIndexValue::Elided`].
307    pub fn is_specified(&self) -> bool {
308        !matches!(self, Self::Elided(_))
309    }
310
311    /// Returns whichever of the two ports are specified.
312    /// If both are [`Self::Elided`], returns [`Self::Elided`].
313    /// If both are specified, returns `Err(self)`.
314    #[allow(clippy::allow_attributes, reason = "Only triggered on nightly.")]
315    #[allow(
316        clippy::result_large_err,
317        reason = "variants are same size, error isn't to be propagated."
318    )]
319    pub fn combine(self, other: Self) -> Result<Self, Self> {
320        match (self.is_specified(), other.is_specified()) {
321            (false, _other) => Ok(other),
322            (true, false) => Ok(self),
323            (true, true) => Err(self),
324        }
325    }
326
327    /// Formats self as a human-readable string for error messages.
328    pub fn as_error_message_string(&self) -> String {
329        match self {
330            PortIndexValue::Int(n) => format!("`{}`", n.value),
331            PortIndexValue::Path(path) => format!("`{}`", path.to_token_stream()),
332            PortIndexValue::Elided(_) => "<elided>".to_owned(),
333        }
334    }
335
336    /// Returns the span of this port value.
337    pub fn span(&self) -> Span {
338        match self {
339            PortIndexValue::Int(x) => x.span(),
340            PortIndexValue::Path(x) => x.span(),
341            PortIndexValue::Elided(span) => span.unwrap_or_else(Span::call_site),
342        }
343    }
344}
345impl From<PortIndex> for PortIndexValue {
346    fn from(value: PortIndex) -> Self {
347        match value {
348            PortIndex::Int(x) => Self::Int(x),
349            PortIndex::Path(x) => Self::Path(x),
350        }
351    }
352}
353impl PartialEq for PortIndexValue {
354    fn eq(&self, other: &Self) -> bool {
355        match (self, other) {
356            (Self::Int(l0), Self::Int(r0)) => l0 == r0,
357            (Self::Path(l0), Self::Path(r0)) => l0 == r0,
358            (Self::Elided(_), Self::Elided(_)) => true,
359            _else => false,
360        }
361    }
362}
363impl Eq for PortIndexValue {}
364impl PartialOrd for PortIndexValue {
365    fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
366        Some(self.cmp(other))
367    }
368}
369impl Ord for PortIndexValue {
370    fn cmp(&self, other: &Self) -> std::cmp::Ordering {
371        match (self, other) {
372            (Self::Int(s), Self::Int(o)) => s.cmp(o),
373            (Self::Path(s), Self::Path(o)) => s
374                .to_token_stream()
375                .to_string()
376                .cmp(&o.to_token_stream().to_string()),
377            (Self::Elided(_), Self::Elided(_)) => std::cmp::Ordering::Equal,
378            (Self::Int(_), Self::Path(_)) => std::cmp::Ordering::Less,
379            (Self::Path(_), Self::Int(_)) => std::cmp::Ordering::Greater,
380            (_, Self::Elided(_)) => std::cmp::Ordering::Less,
381            (Self::Elided(_), _) => std::cmp::Ordering::Greater,
382        }
383    }
384}
385
386impl Display for PortIndexValue {
387    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
388        match self {
389            PortIndexValue::Int(x) => write!(f, "{}", x.to_token_stream()),
390            PortIndexValue::Path(x) => write!(f, "{}", x.to_token_stream()),
391            PortIndexValue::Elided(_) => write!(f, "[]"),
392        }
393    }
394}
395
396/// Output of [`build_dfir_code`].
397pub struct BuildDfirCodeOutput {
398    /// The now-partitioned graph.
399    pub partitioned_graph: DfirGraph,
400    /// The Rust source code tokens for the DFIR.
401    pub code: TokenStream,
402    /// Any (non-error) diagnostics emitted.
403    pub diagnostics: Diagnostics,
404}
405
406/// The main function of this module. Compiles a [`DfirCode`] AST into a [`DfirGraph`] and
407/// source code, or [`Diagnostic`] errors.
408pub fn build_dfir_code(
409    dfir_code: DfirCode,
410    root: &TokenStream,
411) -> Result<BuildDfirCodeOutput, Diagnostics> {
412    let flat_graph_builder = FlatGraphBuilder::from_dfir(dfir_code);
413
414    let FlatGraphBuilderOutput {
415        mut flat_graph,
416        uses,
417        mut diagnostics,
418    } = flat_graph_builder.build()?;
419
420    let () = match flat_graph.merge_modules() {
421        Ok(()) => (),
422        Err(d) => {
423            diagnostics.push(d);
424            return Err(diagnostics);
425        }
426    };
427
428    eliminate_extra_unions_tees(&mut flat_graph);
429    let partitioned_graph = match partition_graph(flat_graph) {
430        Ok(partitioned_graph) => partitioned_graph,
431        Err(d) => {
432            diagnostics.push(d);
433            return Err(diagnostics);
434        }
435    };
436
437    let code =
438        partitioned_graph.as_code(root, true, quote::quote! { #( #uses )* }, &mut diagnostics)?;
439
440    // Success
441    Ok(BuildDfirCodeOutput {
442        partitioned_graph,
443        code,
444        diagnostics,
445    })
446}
447
448/// Changes all of token's spans to `span`, recursing into groups.
449fn change_spans(tokens: TokenStream, span: Span) -> TokenStream {
450    use proc_macro2::{Group, TokenTree};
451    tokens
452        .into_iter()
453        .map(|token| match token {
454            TokenTree::Group(mut group) => {
455                group.set_span(span);
456                TokenTree::Group(Group::new(
457                    group.delimiter(),
458                    change_spans(group.stream(), span),
459                ))
460            }
461            TokenTree::Ident(mut ident) => {
462                ident.set_span(span.resolved_at(ident.span()));
463                TokenTree::Ident(ident)
464            }
465            TokenTree::Punct(mut punct) => {
466                punct.set_span(span);
467                TokenTree::Punct(punct)
468            }
469            TokenTree::Literal(mut literal) => {
470                literal.set_span(span);
471                TokenTree::Literal(literal)
472            }
473        })
474        .collect()
475}
dfir_lang/graph/mod.rs

dfir_lang/graph/
mod.rs
