ridl_syntax/parser.rs
1//! The hand-written recursive-descent parser for the typl grammar
2//! (ADR-0004 §2, typl reference Appendix E).
3//!
4//! The parser consumes the flat token stream from [`crate::lex`] and builds a
5//! lossless rowan tree whose nodes match `family.ungram` exactly: every token —
6//! including whitespace and comments — lands in the tree in source order, so
7//! [`Parse::syntax`] round-trips back to the original source for both valid
8//! and broken input. On an unexpected token the parser records a
9//! [`SyntaxError`] and wraps the skipped tokens in an
10//! [`ErrorNode`](SyntaxKind::ErrorNode) rather than dropping them, so recovery
11//! never loses text and never panics.
12//!
13//! Trivia placement: leading trivia — including doc comments — is flushed at
14//! the parent level before a child node starts, so a doc comment sits as a
15//! sibling immediately before the definition it documents
16//! (`ast::HasDocComments` reads it from there).
17//!
18//! # The `[a..b]` bound policy
19//!
20//! `[3..6]` has two legal tree shapes: a pair of scalar `Literal` children
21//! (a value range) or a `Bound` child (a length bound). The parser emits a
22//! `Bound` iff the constraint attaches to a `string`/`bytes` backing and both
23//! endpoints are integer-shaped (no float token); it emits scalar `Literal`s
24//! otherwise — including when a `step` clause follows, which only the scalar
25//! shape can carry. A single literal with no `..` (`[17]`) is always a
26//! `Bound`; a float endpoint is always scalar.
27//!
28//! # Grammar over-approximations
29//!
30//! The grammar deliberately accepts more than the typl reference allows —
31//! constraints on any primitive in field position, arbitrary map key types,
32//! identifier literals in value positions, `step` on open ranges, non-integer
33//! literals in bounds and reserved entries, repeated separator commas with no
34//! member between them (`,,,`), and the `internal`/`error` modifiers on every
35//! definition kind. The checker narrows these later (TYPL-2xx); the parser
36//! must not reject them.
37//!
38//! The interaction grammar follows the same discipline: a `: return_type`
39//! after a command's params parses; timing parses on command, query, and
40//! fixed; an attr block parses on signal, event, and fixed; an init value
41//! parses on event and fixed; a stream `<T>` parses in every type position,
42//! including signal/event payloads and struct fields; a typl definition inside
43//! an interface body whose `}` still lies ahead recovers into one body-local
44//! `ErrorNode` and draws **RIDL-107** here, where the keyword is recognised,
45//! while a body with no `}` ahead reports an unclosed `{` and keeps the
46//! declarations that follow it. The other rejections are checker scope (RIDL-104/-106/-201/-301).
47//!
48//! # Profile boundary
49//!
50//! The parser runs under a [`Profile`] (ADR-0007 decision 10). In a `.typl`
51//! parse, a `Duration` token or a stray `@` anywhere emits **TYPL-302** (typl
52//! reference §2.8) and parsing continues, an interaction keyword at
53//! declaration-start position emits **TYPL-304**, and a `require`/`ensure`
54//! attribute emits **TYPL-303** with the same recovery. In a `.ridl` parse durations and `@` are ordinary tokens,
55//! and a `ReservedWord` at declaration-start position — a word of the
56//! rmdl/rsdl profiles, or a ridl clause word — emits **RIDL-403** (ridl
57//! reference §16.4). In a
58//! `.rsdl` parse a declaration of another profile at declaration-start position
59//! — a typl definition, `interface`, `service`, or any other `ReservedWord` —
60//! emits **RSDL-604** (rsdl reference §2), and an `internal` or `error`
61//! modifier before an rsdl declaration emits FORM-102. The typl and ridl
62//! declaration-start boundaries recover exactly as FORM-105 does; the rsdl one
63//! consumes a `{ … }` body whole. The stream grammar parses under both
64//! profiles: a `<T>` in type position builds a `StreamType` node everywhere,
65//! and in a `.typl` parse it additionally emits **TYPL-301** (`stream type in
66//! typl context`) and parsing continues. Leading zeros in an integer literal
67//! emit **FORM-005**. Every [`SyntaxError`] carries its diagnostic code; the
68//! coded `Diagnostic` model consumes it.
69//!
70//! # Error recovery
71//!
72//! A broken declaration does not derail the rest of the file. When a loop
73//! meets a token it cannot place, it reports one diagnostic and skips forward,
74//! wrapping the skipped run in a single [`ErrorNode`](SyntaxKind::ErrorNode),
75//! until it reaches a resynchronization point: a top-level keyword (`type`,
76//! `const`, `struct`, and the other definition starters) or a closing `}` (a
77//! tuple loop also stops at its `)`). A block body that runs into a top-level
78//! keyword treats its `{` as unclosed (FORM-103) and hands the keyword back to
79//! the top level rather than swallowing the next declaration. A family reserved
80//! word where a declared name is expected is held in an `ErrorNode` and flagged
81//! FORM-105, so the rest of the declaration still parses. Every [`SyntaxError`]
82//! carries the narrowest honest range — the offending token — while the
83//! `ErrorNode` carries the wider skipped span.
84//!
85//! # One diagnostic per source position
86//!
87//! [`Parser::expect`] never consumes on a mismatch, so a stuck cursor could
88//! otherwise re-report at the same offset on every unwind step — a closed
89//! 200-level type nesting once produced over a thousand diagnostics piled on a
90//! single token. The parser records at most one [`SyntaxError`] per token start
91//! offset. Because the cursor only ever moves forward,
92//! [`Parser::error_at_current`] suppressing an error at the offset the previous
93//! error already used is exactly "one diagnostic per source position". A run of
94//! garbage therefore collapses to one diagnostic plus one `ErrorNode`, and the
95//! pathological nesting inputs report a small, constant number of diagnostics
96//! instead of one per unwound level.
97//!
98//! The profile-boundary **TYPL-302** is the one exception: a positional
99//! FORM-101 can land on a duration or `@` token first (`[0..10ms]` fails the
100//! `]` check at `10ms`), and the profile-boundary fact is the higher-value
101//! diagnostic, so both are recorded. A TYPL-302 token is always consumed where
102//! it is raised, so exempting it can never re-introduce a flood.
103
104use rowan::{GreenNode, GreenNodeBuilder, TextRange, TextSize};
105
106use crate::keywords::{self, Profile};
107use crate::lexer::{Token, lex};
108use crate::syntax_kind::{SyntaxKind, SyntaxNode};
109
110/// A parse diagnostic: a message, the stable diagnostic code it will carry in
111/// the coded model (`FORM-…` per ADR-0007 decision 2, `TYPL-302` per typl
112/// reference §16.4), and the source range of the offending token.
113#[derive(Debug, Clone, PartialEq, Eq)]
114pub struct SyntaxError {
115 pub message: String,
116 pub code: &'static str,
117 pub range: TextRange,
118}
119
120/// The result of [`parse`]: the lossless green tree plus any diagnostics.
121///
122/// [`Parse`] compares by the identity of its green tree, which is what the
123/// salsa query graph needs to decide whether a reparse changed
124/// anything downstream.
125#[derive(Debug, Clone)]
126pub struct Parse {
127 green: GreenNode,
128 errors: Vec<SyntaxError>,
129}
130
131impl Parse {
132 /// A fresh red-tree root over the parsed green tree.
133 pub fn syntax(&self) -> SyntaxNode {
134 SyntaxNode::new_root(self.green.clone())
135 }
136
137 /// The diagnostics gathered while parsing, in source order.
138 pub fn errors(&self) -> &[SyntaxError] {
139 &self.errors
140 }
141}
142
143impl PartialEq for Parse {
144 fn eq(&self, other: &Self) -> bool {
145 self.green == other.green
146 }
147}
148
149impl Eq for Parse {}
150
151/// Parses `input` into a lossless [`Parse`] under `profile`.
152pub fn parse(input: &str, profile: Profile) -> Parse {
153 let mut parser = Parser::new(lex(input, profile), profile);
154 parser.source_file();
155 parser.finish()
156}
157
158/// Whether `kind` can be the value token of a `Literal` (`family.ungram` rule
159/// `Literal`, minus the optional leading `-`).
160fn is_value_token(kind: SyntaxKind) -> bool {
161 matches!(
162 kind,
163 SyntaxKind::IntNumber
164 | SyntaxKind::FloatNumber
165 | SyntaxKind::String
166 | SyntaxKind::TrueKw
167 | SyntaxKind::FalseKw
168 | SyntaxKind::Regex
169 | SyntaxKind::Ident
170 )
171}
172
173/// Names the literal shape a `reserved` tombstone refused, for the FORM-102
174/// message. `kind` is a value token other than `IntNumber` — the one shape
175/// typl Appendix E admits after `reserved` alongside a name.
176fn ungrammatical_reserved_noun(kind: SyntaxKind) -> &'static str {
177 match kind {
178 SyntaxKind::FloatNumber => "a floating-point literal",
179 SyntaxKind::String => "a string literal",
180 SyntaxKind::TrueKw | SyntaxKind::FalseKw => "a boolean literal",
181 SyntaxKind::Regex => "a regex literal",
182 SyntaxKind::Ident => "a constant reference",
183 _ => "a literal of that shape",
184 }
185}
186
187/// Whether `kind` starts a top-level construct. These are the
188/// resynchronization points recovery falls back to, both at the file level
189/// and when a block body runs past an unclosed `}` into the next declaration.
190/// `interface` joins the set under the ridl profile; under [`Profile::Typl`] it never
191/// occurs (the word lexes to `ReservedWord` there). The four top-level rsdl
192/// declaration keywords join it with the rsdl grammar; they occur only under
193/// [`Profile::Rsdl`]. `machine` is not among them: a machine is declared only
194/// inside a deployment body (rsdl reference Appendix B).
195fn is_top_level_start(kind: SyntaxKind) -> bool {
196 matches!(
197 kind,
198 SyntaxKind::PackageKw
199 | SyntaxKind::ImportKw
200 | SyntaxKind::InternalKw
201 | SyntaxKind::ErrorKw
202 | SyntaxKind::TypeKw
203 | SyntaxKind::ConstKw
204 | SyntaxKind::StructKw
205 | SyntaxKind::EnumKw
206 | SyntaxKind::EnumsetKw
207 | SyntaxKind::UnionKw
208 | SyntaxKind::InterfaceKw
209 | SyntaxKind::ServiceKw
210 | SyntaxKind::SystemKw
211 | SyntaxKind::ComponentKw
212 | SyntaxKind::DistributionKw
213 | SyntaxKind::DeploymentKw
214 )
215}
216
217/// Whether `kind` starts an rsdl declaration, `machine` included — what an
218/// `internal` or `error` modifier in an `.rsdl` file is refused in front of
219/// (FORM-102, rsdl reference §2).
220fn is_rsdl_declaration_start(kind: SyntaxKind) -> bool {
221 matches!(
222 kind,
223 SyntaxKind::SystemKw
224 | SyntaxKind::ComponentKw
225 | SyntaxKind::DistributionKw
226 | SyntaxKind::DeploymentKw
227 | SyntaxKind::MachineKw
228 )
229}
230
231/// The three body kinds of the rsdl containers (rsdl reference §3, §4): what
232/// starts a line, what parses it, and how the body is named in a diagnostic.
233#[derive(Clone, Copy, PartialEq, Eq)]
234enum RsdlBody {
235 /// A `system`, `distribution` or `machine` body: bare member lines.
236 Members,
237 /// A `component` body: `offers` and `requires` lines.
238 ComponentLines,
239 /// A `deployment` body: `machine` declarations.
240 Machines,
241}
242
243impl RsdlBody {
244 fn starts_line(self, kind: SyntaxKind) -> bool {
245 match self {
246 // A reference starts like a `QualifiedName` ([`Parser::at_path_segment`]).
247 Self::Members => matches!(
248 kind,
249 SyntaxKind::Ident
250 | SyntaxKind::BooleanKw
251 | SyntaxKind::IntegerKw
252 | SyntaxKind::FloatKw
253 | SyntaxKind::StringKw
254 | SyntaxKind::BytesKw
255 ),
256 Self::ComponentLines => matches!(kind, SyntaxKind::OffersKw | SyntaxKind::RequiresKw),
257 Self::Machines => kind == SyntaxKind::MachineKw,
258 }
259 }
260}
261
262/// The RSDL-604 message for a declaration of another profile at the top level
263/// of an `.rsdl` file, naming the word that starts it (rsdl reference §2).
264fn rsdl_foreign_declaration_message(word: &str) -> String {
265 format!(
266 "`{word}` starts a declaration of another profile, and an `.rsdl` file declares \
267 `system`, `component`, `distribution` and `deployment` only — a type is declared in \
268 a `.typl` file, an interface or a service in a `.ridl` file (rsdl reference §2)"
269 )
270}
271
272/// Whether `kind` starts an interaction inside an interface body — the five
273/// interaction keywords. They join the recovery sync set inside
274/// interface bodies, so garbage resynchronizes at the next interaction.
275fn is_interaction_start(kind: SyntaxKind) -> bool {
276 matches!(
277 kind,
278 SyntaxKind::SignalKw
279 | SyntaxKind::EventKw
280 | SyntaxKind::CommandKw
281 | SyntaxKind::QueryKw
282 | SyntaxKind::FixedKw
283 )
284}
285
286/// Where recovery from a rejected return type stops: the interface-body sync
287/// set plus `@`, so a timing annotation written after a broken return type is
288/// still parsed into its own node and still draws its own diagnostic.
289fn is_return_sync(kind: SyntaxKind) -> bool {
290 matches!(
291 kind,
292 SyntaxKind::RBrace | SyntaxKind::Comma | SyntaxKind::ReservedKw | SyntaxKind::At
293 ) || is_interaction_start(kind)
294 || is_top_level_start(kind)
295}
296
297/// Whether `kind` is a typl definition keyword. Inside an interface body
298/// whose `}` still lies ahead, these recover into a body-local
299/// [`ErrorNode`](SyntaxKind::ErrorNode) — the ridl reference forbids type
300/// declarations there (§14.1) and reserves RIDL-107 for them (§16.1), which
301/// the body loop raises. In a body with no `}` ahead they signal an unclosed
302/// `{` exactly like the remaining top-level keywords, so a genuinely
303/// unclosed interface keeps the declarations that follow it.
304fn is_typl_definition_start(kind: SyntaxKind) -> bool {
305 matches!(
306 kind,
307 SyntaxKind::TypeKw
308 | SyntaxKind::ConstKw
309 | SyntaxKind::StructKw
310 | SyntaxKind::EnumKw
311 | SyntaxKind::EnumsetKw
312 | SyntaxKind::UnionKw
313 )
314}
315
316/// The deepest field-type, expression, or attribute-value nesting the parser
317/// follows before it stops recursing, and the greatest height of the tree it
318/// builds for one expression — far beyond any real schema, small enough that
319/// a pathological input can overflow neither the parser's stack nor the stack
320/// of a pass that later recurses over the tree.
321const MAX_TYPE_DEPTH: usize = 128;
322
323/// The recursive-descent state: the token stream, a cursor, the tree builder,
324/// and the diagnostics.
325struct Parser<'a> {
326 tokens: Vec<Token<'a>>,
327 /// Byte offset of the start of each token, plus a final entry for the end
328 /// of input; `offsets[i]` is the offset of `tokens[i]`.
329 offsets: Vec<usize>,
330 pos: usize,
331 /// The nesting the cursor is under, bounded by [`MAX_TYPE_DEPTH`].
332 /// Outside an expression it counts the field types and attribute values
333 /// open above the cursor (each `field_type` or `attr_value` call adds
334 /// one). Inside an expression it counts the expression nodes open above
335 /// the cursor — the groups, prefixes and chain wraps the node built next
336 /// will sit under — which is what the height bound is measured against.
337 depth: usize,
338 /// Whether the expression being parsed has already drawn FORM-102. A
339 /// later part of the same expression that does not fit either goes flat
340 /// without a second diagnostic.
341 expr_refused: bool,
342 /// The profile the file parses under — where the profile-boundary
343 /// diagnostics (TYPL-302, TYPL-304, RIDL-403) are drawn.
344 profile: Profile,
345 builder: GreenNodeBuilder<'static>,
346 errors: Vec<SyntaxError>,
347}
348
349impl<'a> Parser<'a> {
350 fn new(tokens: Vec<Token<'a>>, profile: Profile) -> Self {
351 let mut offsets = Vec::with_capacity(tokens.len() + 1);
352 let mut acc = 0;
353 for token in &tokens {
354 offsets.push(acc);
355 acc += token.text.len();
356 }
357 offsets.push(acc);
358 Self {
359 tokens,
360 offsets,
361 pos: 0,
362 depth: 0,
363 expr_refused: false,
364 profile,
365 builder: GreenNodeBuilder::new(),
366 errors: Vec::new(),
367 }
368 }
369
370 fn finish(self) -> Parse {
371 Parse {
372 green: self.builder.finish(),
373 errors: self.errors,
374 }
375 }
376
377 // --- cursor over significant (non-trivia) tokens ---------------------
378
379 /// Index of the next non-trivia token at or after the cursor.
380 fn significant_pos(&self) -> usize {
381 let mut i = self.pos;
382 while self.tokens.get(i).is_some_and(|t| t.kind.is_trivia()) {
383 i += 1;
384 }
385 i
386 }
387
388 /// Kind of the next significant token, or `None` at end of input.
389 fn current(&self) -> Option<SyntaxKind> {
390 self.tokens.get(self.significant_pos()).map(|t| t.kind)
391 }
392
393 /// Kind of the `n`-th significant token ahead (`nth(0)` is [`current`]).
394 fn nth(&self, n: usize) -> Option<SyntaxKind> {
395 let mut i = self.pos;
396 let mut seen = 0;
397 loop {
398 let token = self.tokens.get(i)?;
399 if !token.kind.is_trivia() {
400 if seen == n {
401 return Some(token.kind);
402 }
403 seen += 1;
404 }
405 i += 1;
406 }
407 }
408
409 fn at(&self, kind: SyntaxKind) -> bool {
410 self.current() == Some(kind)
411 }
412
413 /// Source text of the next significant token, or `""` at end of input —
414 /// for a message that quotes the word the author wrote.
415 fn current_text(&self) -> &'a str {
416 self.tokens
417 .get(self.significant_pos())
418 .map_or("", |token| token.text)
419 }
420
421 // --- token consumption (feeds the tree, keeps losslessness) ----------
422
423 /// Attaches any pending trivia tokens to the current node, in source order.
424 fn eat_trivia(&mut self) {
425 while let Some(token) = self.tokens.get(self.pos) {
426 if token.kind.is_trivia() {
427 self.builder.token(token.kind.into(), token.text);
428 self.pos += 1;
429 } else {
430 break;
431 }
432 }
433 }
434
435 /// Flushes leading trivia to the currently open node, then starts a child
436 /// node — so a definition's doc comment stays a sibling of the definition
437 /// rather than its first child.
438 fn start(&mut self, kind: SyntaxKind) {
439 self.eat_trivia();
440 self.builder.start_node(kind.into());
441 }
442
443 /// Consumes one significant token into the tree, flushing leading trivia
444 /// first. Callers must have confirmed a token is present.
445 fn bump(&mut self) {
446 self.eat_trivia();
447 if let Some(token) = self.tokens.get(self.pos) {
448 self.builder.token(token.kind.into(), token.text);
449 self.pos += 1;
450 }
451 }
452
453 /// Consumes the current token if it is `kind`; reports FORM-101 otherwise.
454 fn expect(&mut self, kind: SyntaxKind) {
455 if self.at(kind) {
456 self.bump();
457 } else {
458 self.error_at_current("FORM-101", format!("expected {kind:?}"));
459 }
460 }
461
462 /// Records an error spanning the current significant token (or the end of
463 /// input), unless a diagnostic already sits at that start offset.
464 ///
465 /// The cursor only ever moves forward, so the start offset here is
466 /// non-decreasing across calls; comparing against the last recorded error
467 /// therefore enforces one diagnostic per source position (see the module
468 /// doc). This is the flood control: on a stuck cursor the repeated,
469 /// no-progress `expect`/`bound` errors during an unwind collapse to the
470 /// first one.
471 ///
472 /// The profile-boundary **TYPL-302** is exempt: it is the higher-value
473 /// diagnostic and must survive a positional FORM-101 that lands on the
474 /// same boundary token first (`[0..10ms]` reports both). A TYPL-302 token
475 /// is always consumed at the site that raises it, so it can never be the
476 /// stuck-cursor flood source the suppression guards against.
477 fn error_at_current(&mut self, code: &'static str, message: String) {
478 let i = self.significant_pos();
479 let start = TextSize::new(self.offsets[i] as u32);
480 if code != "TYPL-302"
481 && self
482 .errors
483 .last()
484 .is_some_and(|last| last.range.start() == start)
485 {
486 return;
487 }
488 let end = self
489 .tokens
490 .get(i)
491 .map_or(start, |t| start + TextSize::of(t.text));
492 self.errors.push(SyntaxError {
493 message,
494 code,
495 range: TextRange::new(start, end),
496 });
497 }
498
499 /// The diagnostic for the current unexpected token: in a `.typl` parse,
500 /// TYPL-302 for the profile-boundary tokens (duration literals and the `@`
501 /// timing sigil, typl reference §2.8); FORM-102 otherwise. Under
502 /// [`Profile::Ridl`] durations and `@` are ordinary tokens, so they draw
503 /// the generic FORM-102 like any other token the grammar cannot place.
504 fn unexpected_token_diag(&self, context: &str) -> (&'static str, String) {
505 if self.profile == Profile::Typl {
506 match self.current() {
507 Some(SyntaxKind::Duration) => {
508 return ("TYPL-302", "duration literal in typl context".to_string());
509 }
510 Some(SyntaxKind::At) => {
511 return ("TYPL-302", "timing annotation in typl context".to_string());
512 }
513 _ => {}
514 }
515 }
516 ("FORM-102", format!("unexpected token {context}"))
517 }
518
519 /// Recovery: report the current unexpected token, then skip forward —
520 /// wrapping the skipped run in one [`SyntaxKind::ErrorNode`] — until a
521 /// token `is_sync` accepts (a resynchronization point) or the input ends.
522 /// The first token is always consumed, so the enclosing loop makes
523 /// progress and no text is dropped.
524 fn err_and_recover(&mut self, context: &str, is_sync: impl Fn(SyntaxKind) -> bool) {
525 let (code, message) = self.unexpected_token_diag(context);
526 self.error_at_current(code, message);
527 self.recover(is_sync);
528 }
529
530 /// The recovery half of [`err_and_recover`]: consumes the current token
531 /// and everything up to the next token `is_sync` accepts into one
532 /// [`SyntaxKind::ErrorNode`]. The profile-boundary arms (TYPL-304,
533 /// RIDL-403) share it, so they recover exactly as FORM-102/FORM-105 do.
534 fn recover(&mut self, is_sync: impl Fn(SyntaxKind) -> bool) {
535 self.start(SyntaxKind::ErrorNode);
536 self.bump();
537 while let Some(kind) = self.current() {
538 if is_sync(kind) {
539 break;
540 }
541 self.bump();
542 }
543 self.builder.finish_node();
544 }
545
546 // --- grammar productions --------------------------------------------
547
548 /// `SourceFile = PackageDecl? Import* Definition*`
549 ///
550 /// A missing `package` emits FORM-104 and parsing continues; extra
551 /// `package` declarations parse as further `PackageDecl` nodes (TYPL-001
552 /// is checker scope). Imports and definitions are accepted in any order —
553 /// ordering is surface convention, not tree shape.
554 fn source_file(&mut self) {
555 self.builder.start_node(SyntaxKind::SourceFile.into());
556 self.eat_trivia();
557 if self.at(SyntaxKind::PackageKw) {
558 self.package_decl();
559 } else {
560 self.error_at_current("FORM-104", "missing `package` declaration".to_string());
561 }
562 loop {
563 self.eat_trivia();
564 match self.current() {
565 None => break,
566 Some(SyntaxKind::PackageKw) => self.package_decl(),
567 Some(SyntaxKind::ImportKw) => self.import(),
568 Some(
569 SyntaxKind::InternalKw
570 | SyntaxKind::ErrorKw
571 | SyntaxKind::TypeKw
572 | SyntaxKind::ConstKw
573 | SyntaxKind::StructKw
574 | SyntaxKind::EnumKw
575 | SyntaxKind::EnumsetKw
576 | SyntaxKind::UnionKw
577 | SyntaxKind::InterfaceKw,
578 ) => {
579 if self.profile == Profile::Rsdl {
580 self.rsdl_typl_word_at_top_level();
581 } else {
582 self.definition();
583 }
584 }
585 Some(SyntaxKind::ServiceKw) => self.service_def(),
586 Some(SyntaxKind::SystemKw) => {
587 self.rsdl_container(SyntaxKind::SystemDef, RsdlBody::Members);
588 }
589 Some(SyntaxKind::ComponentKw) => {
590 self.rsdl_container(SyntaxKind::ComponentDef, RsdlBody::ComponentLines);
591 }
592 Some(SyntaxKind::DistributionKw) => {
593 self.rsdl_container(SyntaxKind::DistributionDef, RsdlBody::Members);
594 }
595 Some(SyntaxKind::DeploymentKw) => self.deployment_def(),
596 Some(SyntaxKind::ReservedWord) => {
597 if !self.profile_boundary_at_top_level() {
598 self.err_and_recover("at top level", is_top_level_start);
599 }
600 }
601 Some(_) => self.err_and_recover("at top level", is_top_level_start),
602 }
603 }
604 self.builder.finish_node();
605 }
606
607 /// The profile boundary at declaration-start position (ADR-0007 decision
608 /// 10, ridl reference §16.4). Called on a `ReservedWord` at the top level:
609 ///
610 /// - in a `.typl` parse, an interaction keyword (one of the nine ridl
611 /// words) draws **TYPL-304** instead of the generic FORM-102 — except
612 /// `require`/`ensure`, which are attributes, not declarations, and
613 /// draw **TYPL-303** (`require/ensure attribute in typl context`,
614 /// typl reference §16.4);
615 /// - in a `.ridl` parse, any `ReservedWord` — a behaviour or architecture
616 /// word of the rmdl/rsdl profiles, or a ridl clause word — draws
617 /// **RIDL-403**;
618 /// - in a `.rsdl` parse, any `ReservedWord` — `interface` and `service`
619 /// first of all — draws **RSDL-604** (rsdl reference §2), with the
620 /// brace-aware recovery of [`Parser::rsdl_foreign_declaration`]. The typl
621 /// definition keywords are active in an `.rsdl` file and reach
622 /// [`Parser::rsdl_typl_word_at_top_level`] instead.
623 ///
624 /// All recover exactly as FORM-105 does ([`Parser::recover`]: the run
625 /// lands in one `ErrorNode`, resynchronizing at the next top-level
626 /// keyword). Returns `false` — leaving the token untouched for the
627 /// caller's generic recovery — when neither boundary applies (a
628 /// uxdl/rmdl/rsdl word in a `.typl` parse stays a generic FORM-102).
629 fn profile_boundary_at_top_level(&mut self) -> bool {
630 let text = self.tokens[self.significant_pos()].text;
631 let (code, message) = match self.profile {
632 Profile::Typl => match keywords::ridl_keyword(text) {
633 None => return false,
634 Some(SyntaxKind::RequireKw | SyntaxKind::EnsureKw) => (
635 "TYPL-303",
636 "require/ensure attribute in typl context".to_string(),
637 ),
638 Some(_) => (
639 "TYPL-304",
640 format!("interaction declaration in typl context: `{text}`"),
641 ),
642 },
643 // The message names neither "behaviour" nor "user-interaction" nor
644 // "architecture": those are the catalogue's words for the three
645 // profiles, and the author wrote one word, not a taxonomy. What is
646 // wrong is that the word is spoken for; what is allowed is the two
647 // declarations a `.ridl` file makes.
648 Profile::Ridl => (
649 "RIDL-403",
650 format!(
651 "`{text}` is reserved by another profile of the family, so it cannot start a \
652 declaration here — a `.ridl` file declares `interface` and `service`. Move \
653 the declaration to the profile that owns the word, or, if `{text}` was meant \
654 as a name, choose one the family does not reserve (ridl reference §16.4)"
655 ),
656 ),
657 Profile::Rsdl => {
658 self.rsdl_foreign_declaration(0);
659 return true;
660 }
661 };
662 self.error_at_current(code, message);
663 self.recover(is_top_level_start);
664 true
665 }
666
667 /// A typl definition keyword or an `internal`/`error` modifier at the top
668 /// level of an `.rsdl` file. A modifier run in front of an rsdl declaration
669 /// keyword is FORM-102 on each modifier (rsdl reference §2: the five
670 /// declarations take no modifier), and the declaration after it still
671 /// parses. Anything else is a typl declaration, RSDL-604: the modifiers,
672 /// the keyword and the rest of the declaration land in one `ErrorNode`
673 /// ([`Parser::rsdl_foreign_declaration`]), so the declaration draws one
674 /// diagnostic, not one per word.
675 fn rsdl_typl_word_at_top_level(&mut self) {
676 let mut modifiers = 0;
677 while matches!(
678 self.nth(modifiers),
679 Some(SyntaxKind::InternalKw | SyntaxKind::ErrorKw)
680 ) {
681 modifiers += 1;
682 }
683 if modifiers > 0 && self.nth(modifiers).is_some_and(is_rsdl_declaration_start) {
684 for _ in 0..modifiers {
685 let word = self.current_text();
686 self.error_at_current(
687 "FORM-102",
688 format!(
689 "`{word}` cannot modify an rsdl declaration — the closure is \
690 workspace-wide, so every declaration it names is visible from the \
691 system's package (rsdl reference §2)"
692 ),
693 );
694 self.start(SyntaxKind::ErrorNode);
695 self.bump();
696 self.builder.finish_node();
697 }
698 return;
699 }
700 self.rsdl_foreign_declaration(modifiers);
701 }
702
703 /// RSDL-604 (rsdl reference §2): a declaration of another profile at the
704 /// top level of an `.rsdl` file, starting at the current token after
705 /// `modifiers` `internal`/`error` words. The declaration draws one
706 /// diagnostic, on its first token, naming its keyword. Recovery is
707 /// brace-aware, like [`Parser::recover_definition_in_body`]: a body
708 /// `{ … }` is consumed whole, so a ridl word inside it (`signal`) is not
709 /// taken for the next declaration; at depth 0 recovery stops at a
710 /// top-level keyword or at a reserved word, which starts the next
711 /// declaration of another profile.
712 fn rsdl_foreign_declaration(&mut self, modifiers: usize) {
713 let keyword = self.nth_text(modifiers);
714 self.error_at_current("RSDL-604", rsdl_foreign_declaration_message(keyword));
715 self.start(SyntaxKind::ErrorNode);
716 for _ in 0..=modifiers {
717 self.bump();
718 }
719 let mut depth = 0usize;
720 while let Some(kind) = self.current() {
721 match kind {
722 SyntaxKind::LBrace => depth += 1,
723 SyntaxKind::RBrace if depth > 0 => depth -= 1,
724 _ if depth == 0
725 && (is_top_level_start(kind) || kind == SyntaxKind::ReservedWord) =>
726 {
727 break;
728 }
729 _ => {}
730 }
731 self.bump();
732 }
733 self.builder.finish_node();
734 }
735
736 /// Source text of the `n`-th significant token ahead (`nth_text(0)` is
737 /// [`Parser::current_text`]), or `""` past the end of input.
738 fn nth_text(&self, n: usize) -> &'a str {
739 self.tokens[self.pos..]
740 .iter()
741 .filter(|token| !token.kind.is_trivia())
742 .nth(n)
743 .map_or("", |token| token.text)
744 }
745
746 /// The rsdl containers with no relation clause — `system`, `component`,
747 /// `distribution` and `machine` (rsdl reference §3, general form §2 Shape
748 /// 3): `kw Name AttrBlock? '{' lines '}'`.
749 fn rsdl_container(&mut self, kind: SyntaxKind, body: RsdlBody) {
750 self.start(kind);
751 self.bump(); // the declaration keyword
752 self.name();
753 if self.at(SyntaxKind::LBracket) {
754 self.attr_block();
755 }
756 self.rsdl_body(body);
757 self.builder.finish_node();
758 }
759
760 /// `DeploymentDef = 'deployment' Name 'for' system:Reference AttrBlock?
761 /// '{' (machines:MachineDef ','?)* '}'` (rsdl reference §3.4). The clause
762 /// comes before the attributes (R5). A missing `for` is FORM-101 and the
763 /// body still parses.
764 fn deployment_def(&mut self) {
765 self.start(SyntaxKind::DeploymentDef);
766 self.bump(); // 'deployment'
767 self.name();
768 if self.at(SyntaxKind::ForKw) {
769 self.bump(); // 'for'
770 self.reference();
771 } else {
772 self.error_at_current(
773 "FORM-101",
774 "expected `for` and the system this deployment places".to_string(),
775 );
776 }
777 if self.at(SyntaxKind::LBracket) {
778 self.attr_block();
779 }
780 self.rsdl_body(RsdlBody::Machines);
781 self.builder.finish_node();
782 }
783
784 /// The body loop of an rsdl container, the rsdl counterpart of
785 /// [`Parser::block_body`]: lines separated by newlines or commas, a
786 /// trailing comma permitted (R8). A token that starts no line of this body
787 /// is FORM-102 and recovery resynchronizes at the next line, comma or `}`.
788 /// A reference where a component line is expected is the one-reference-
789 /// per-line rule (rsdl reference §3.2), and its message says so.
790 fn rsdl_body(&mut self, body: RsdlBody) {
791 if !self.block_open() {
792 return;
793 }
794 loop {
795 self.eat_trivia();
796 match self.current() {
797 None => {
798 self.error_at_current("FORM-103", "unclosed `{`".to_string());
799 break;
800 }
801 Some(SyntaxKind::RBrace) => {
802 self.bump();
803 break;
804 }
805 Some(SyntaxKind::Comma) => self.bump(),
806 Some(kind) if body.starts_line(kind) => match body {
807 RsdlBody::Members => self.member_line(),
808 RsdlBody::ComponentLines => self.component_line(),
809 RsdlBody::Machines => {
810 self.rsdl_container(SyntaxKind::MachineDef, RsdlBody::Members);
811 }
812 },
813 Some(kind) if is_top_level_start(kind) => {
814 self.error_at_current("FORM-103", "unclosed `{`".to_string());
815 break;
816 }
817 Some(_) => {
818 let sync = move |kind: SyntaxKind| {
819 matches!(kind, SyntaxKind::RBrace | SyntaxKind::Comma)
820 || body.starts_line(kind)
821 || is_top_level_start(kind)
822 };
823 if body == RsdlBody::ComponentLines && self.at_path_segment() {
824 self.error_at_current(
825 "FORM-102",
826 format!(
827 "`{}` is a second reference — a component line takes one \
828 reference, so write its own `offers` or `requires` before it \
829 (rsdl reference §3.2)",
830 self.current_text()
831 ),
832 );
833 self.recover(sync);
834 } else {
835 let context = match body {
836 RsdlBody::Members => "in a member list",
837 RsdlBody::ComponentLines => "in a component body",
838 RsdlBody::Machines => "in a deployment body",
839 };
840 self.err_and_recover(context, sync);
841 }
842 }
843 }
844 }
845 }
846
847 /// `MemberLine = Reference AttrBlock?` — a bare reference in a `system`,
848 /// `distribution` or `machine` body (rsdl reference §4).
849 fn member_line(&mut self) {
850 self.start(SyntaxKind::MemberLine);
851 self.reference();
852 if self.at(SyntaxKind::LBracket) {
853 self.attr_block();
854 }
855 self.builder.finish_node();
856 }
857
858 /// `ComponentLine = ('offers' | 'requires') Reference AttrBlock?` (rsdl
859 /// reference §3.2).
860 fn component_line(&mut self) {
861 self.start(SyntaxKind::ComponentLine);
862 self.bump(); // 'offers' | 'requires'
863 self.reference();
864 if self.at(SyntaxKind::LBracket) {
865 self.attr_block();
866 }
867 self.builder.finish_node();
868 }
869
870 /// `Reference = QualifiedName` (rsdl reference §4, Appendix B). No node is
871 /// built when the reference is missing.
872 fn reference(&mut self) {
873 if !self.at_path_segment() {
874 self.error_at_current("FORM-101", "expected a reference".to_string());
875 return;
876 }
877 self.start(SyntaxKind::Reference);
878 self.qualified_name();
879 self.builder.finish_node();
880 }
881
882 /// `PackageDecl = 'package' QualifiedName`
883 fn package_decl(&mut self) {
884 self.start(SyntaxKind::PackageDecl);
885 self.bump(); // 'package'
886 self.qualified_name();
887 self.builder.finish_node();
888 }
889
890 /// `Import = 'import' QualifiedName ('as' alias:Name)?`
891 fn import(&mut self) {
892 self.start(SyntaxKind::Import);
893 self.bump(); // 'import'
894 self.qualified_name();
895 if self.at(SyntaxKind::AsKw) {
896 self.bump();
897 self.name();
898 }
899 self.builder.finish_node();
900 }
901
902 /// `Definition = TypeDef | ConstDef | StructDef | EnumDef | EnumSetDef |
903 /// UnionDef`, each with optional leading `internal`/`error` modifiers.
904 /// The definition keyword is found by looking past the modifiers; a
905 /// modifier with no definition keyword after it is an unexpected token.
906 fn definition(&mut self) {
907 let mut n = 0;
908 while matches!(
909 self.nth(n),
910 Some(SyntaxKind::InternalKw | SyntaxKind::ErrorKw)
911 ) {
912 n += 1;
913 }
914 match self.nth(n) {
915 Some(SyntaxKind::TypeKw) => self.type_def(),
916 Some(SyntaxKind::ConstKw) => self.const_def(),
917 Some(SyntaxKind::StructKw) => self.struct_def(),
918 Some(SyntaxKind::EnumKw) => self.enum_def(),
919 Some(SyntaxKind::EnumsetKw) => self.enum_set_def(),
920 Some(SyntaxKind::UnionKw) => self.union_def(),
921 Some(SyntaxKind::InterfaceKw) => self.interface_def(),
922 _ => self.err_and_recover("in a definition", is_top_level_start),
923 }
924 }
925
926 /// Consumes the `internal`/`error` modifier run into the open node.
927 fn modifiers(&mut self) {
928 while matches!(
929 self.current(),
930 Some(SyntaxKind::InternalKw | SyntaxKind::ErrorKw)
931 ) {
932 self.bump();
933 }
934 }
935
936 /// `TypeDef = 'internal'? 'error'? 'type' Name ':' Backing Constraint?
937 /// InitValue?` — the constraint is a direct child of `TypeDef` here,
938 /// never nested inside `PrimitiveType` (that position is for field
939 /// types).
940 fn type_def(&mut self) {
941 self.start(SyntaxKind::TypeDef);
942 self.modifiers();
943 self.bump(); // 'type'
944 self.name();
945 self.expect(SyntaxKind::Colon);
946 let stringish = self.backing();
947 if self.at(SyntaxKind::LBracket) {
948 self.constraint(stringish);
949 }
950 self.init_value_opt();
951 self.builder.finish_node();
952 }
953
954 /// `Backing = PrimitiveType | UnitExpr` in `type`-definition position.
955 /// Returns whether the backing is `string`/`bytes` — the bound policy
956 /// input for the constraint that may follow.
957 fn backing(&mut self) -> bool {
958 match self.current() {
959 Some(
960 kw @ (SyntaxKind::BooleanKw
961 | SyntaxKind::IntegerKw
962 | SyntaxKind::FloatKw
963 | SyntaxKind::StringKw
964 | SyntaxKind::BytesKw),
965 ) => {
966 self.start(SyntaxKind::PrimitiveType);
967 self.bump();
968 self.builder.finish_node();
969 matches!(kw, SyntaxKind::StringKw | SyntaxKind::BytesKw)
970 }
971 Some(SyntaxKind::Ident | SyntaxKind::Slash | SyntaxKind::Dot | SyntaxKind::Percent) => {
972 // `UnitExpr = ('ident' | '/' | '.' | '%')*` — a UCUM
973 // expression (`km/h`, `N.m`, `/min`, `%`). Semantic
974 // validation is the ucum module's job (TYPL-110).
975 self.start(SyntaxKind::UnitExpr);
976 while matches!(
977 self.current(),
978 Some(
979 SyntaxKind::Ident
980 | SyntaxKind::Slash
981 | SyntaxKind::Dot
982 | SyntaxKind::Percent
983 )
984 ) {
985 self.bump();
986 }
987 self.builder.finish_node();
988 false
989 }
990 _ => {
991 self.error_at_current("FORM-101", "expected a backing type".to_string());
992 false
993 }
994 }
995 }
996
997 /// `ConstDef = 'internal'? 'error'? 'const' Name (':' type_ref:PathType)?
998 /// '=' value:Literal`
999 fn const_def(&mut self) {
1000 self.start(SyntaxKind::ConstDef);
1001 self.modifiers();
1002 self.bump(); // 'const'
1003 self.name();
1004 if self.at(SyntaxKind::Colon) {
1005 self.bump();
1006 self.path_type();
1007 }
1008 self.expect(SyntaxKind::Eq);
1009 self.literal();
1010 self.builder.finish_node();
1011 }
1012
1013 /// `StructDef = 'internal'? 'error'? 'struct' Name '{' (members ','?)*
1014 /// '}'` — a member is a `FieldDef` or a `ReservedEntry`.
1015 fn struct_def(&mut self) {
1016 self.start(SyntaxKind::StructDef);
1017 self.modifiers();
1018 self.bump(); // 'struct'
1019 self.name();
1020 self.block_body("in a struct body", true, Self::field_def);
1021 self.builder.finish_node();
1022 }
1023
1024 /// `EnumDef = 'internal'? 'error'? 'enum' Name '{' ((values | reserved)
1025 /// ','?)* '}'`
1026 fn enum_def(&mut self) {
1027 self.start(SyntaxKind::EnumDef);
1028 self.modifiers();
1029 self.bump(); // 'enum'
1030 self.name();
1031 self.block_body("in an enum body", true, Self::enum_value);
1032 self.builder.finish_node();
1033 }
1034
1035 /// `EnumSetDef` — the standalone form (`enumset Name '{' bits '}'`) or
1036 /// the derived form (`enumset Name ':' backing_ref:PathType`), §9.1–§9.2.
1037 fn enum_set_def(&mut self) {
1038 self.start(SyntaxKind::EnumSetDef);
1039 self.modifiers();
1040 self.bump(); // 'enumset'
1041 self.name();
1042 if self.at(SyntaxKind::Colon) {
1043 self.bump();
1044 self.path_type();
1045 } else {
1046 self.block_body("in an enumset body", false, Self::enum_set_bit);
1047 }
1048 self.builder.finish_node();
1049 }
1050
1051 /// `UnionDef = 'internal'? 'error'? 'union' Name '{' ((arms | reserved)
1052 /// ','?)* '}'`
1053 fn union_def(&mut self) {
1054 self.start(SyntaxKind::UnionDef);
1055 self.modifiers();
1056 self.bump(); // 'union'
1057 self.name();
1058 self.block_body("in a union body", true, Self::union_arm);
1059 self.builder.finish_node();
1060 }
1061
1062 /// `UnionArm = Name ':' type_ref:PathType` — arms reference named types
1063 /// only; a primitive keyword still parses as a path segment and the
1064 /// checker rejects it (TYPL-204).
1065 fn union_arm(&mut self) {
1066 self.start(SyntaxKind::UnionArm);
1067 self.name();
1068 self.expect(SyntaxKind::Colon);
1069 self.path_type();
1070 self.builder.finish_node();
1071 }
1072
1073 // --- ridl interaction productions (ridl reference Appendix C) --
1074
1075 /// `InterfaceDef = 'internal'? 'error'? 'interface' Name '{' (members
1076 /// ','?)* '}'` — an interface body holds interactions and `reserved`
1077 /// tombstones only (ridl reference §14.1); a type declaration inside it
1078 /// is RIDL-107, raised by the body loop. The body loop is the interaction
1079 /// counterpart of [`Parser::block_body`]: members are announced by the
1080 /// five interaction keywords instead of an `Ident`, and those keywords
1081 /// are the recovery sync points, so garbage inside a body
1082 /// resynchronizes at the next interaction.
1083 fn interface_def(&mut self) {
1084 self.start(SyntaxKind::InterfaceDef);
1085 self.modifiers();
1086 self.bump(); // 'interface'
1087 self.name();
1088 if !self.block_open() {
1089 self.builder.finish_node();
1090 return;
1091 }
1092 self.interface_body("an interface", "interface");
1093 self.builder.finish_node();
1094 }
1095
1096 /// The interaction-body loop shared by an `interface` body and a
1097 /// `service` inline body: the opening `{` has already been
1098 /// consumed. Members are announced by the five interaction keywords, and
1099 /// those keywords plus the top-level starters are the recovery sync
1100 /// points, so garbage inside a body resynchronizes at the next
1101 /// interaction. A `service` inline body runs the same structural pass as
1102 /// an interface (ridl reference §14.5).
1103 ///
1104 /// `noun` and `shape` name the enclosing construct for the RIDL-107
1105 /// message — `"an interface"`/`"interface"` or `"a service"`/`"service"`.
1106 /// The two callers share this loop, and a reader is told where the stray
1107 /// declaration has to move to.
1108 fn interface_body(&mut self, noun: &str, shape: &str) {
1109 loop {
1110 self.eat_trivia();
1111 match self.current() {
1112 None => {
1113 self.error_at_current("FORM-103", "unclosed `{`".to_string());
1114 break;
1115 }
1116 Some(SyntaxKind::RBrace) => {
1117 self.bump();
1118 break;
1119 }
1120 Some(SyntaxKind::Comma) => self.bump(),
1121 Some(SyntaxKind::ReservedKw) => self.reserved_entry(),
1122 Some(SyntaxKind::SignalKw) => self.value_interaction(SyntaxKind::SignalDef),
1123 Some(SyntaxKind::EventKw) => self.value_interaction(SyntaxKind::EventDef),
1124 Some(SyntaxKind::CommandKw) => self.callable_interaction(SyntaxKind::CommandDef),
1125 Some(SyntaxKind::QueryKw) => self.callable_interaction(SyntaxKind::QueryDef),
1126 Some(SyntaxKind::FixedKw) => self.value_interaction(SyntaxKind::FixedDef),
1127 // A typl definition keyword inside the body: the body holds
1128 // interactions and tombstones only (ridl §14.1). When the
1129 // body's `}` still lies ahead, the declaration recovers into
1130 // a body-local ErrorNode — brace-aware, so a composite body
1131 // stays inside it and the members after it keep their place.
1132 // With no `}` ahead the guard fails and the keyword falls
1133 // through to the unclosed-`{` arm below, exactly like the
1134 // remaining top-level keywords (`package`, `import`,
1135 // `internal`, `error`, `interface`, `service`).
1136 //
1137 // RIDL-107 is raised here rather than by the checker over the
1138 // recovered node. The parser has already recognised exactly
1139 // what this is, so the generic FORM-102 "unexpected token in
1140 // an interface body" it used to raise was both wrong and a
1141 // second diagnostic for one mistake: every RIDL-107 arrived
1142 // paired with a contradicting FORM-102 at the same span. The
1143 // profile-boundary codes RIDL-403 and TYPL-304 are raised from
1144 // the parser for the same reason.
1145 Some(kind)
1146 if is_typl_definition_start(kind) && self.interface_body_close_ahead() =>
1147 {
1148 self.error_at_current(
1149 "RIDL-107",
1150 format!(
1151 "`{}` starts a typl declaration, and {noun} body holds interactions \
1152 and `reserved` tombstones only — move the declaration to package \
1153 level, beside the {shape} (ridl reference §14.1)",
1154 self.current_text(),
1155 ),
1156 );
1157 self.recover_definition_in_body();
1158 }
1159 // An unclosed `{`: the body ran into the next top-level
1160 // declaration. Report the missing brace and hand the
1161 // declaration back, exactly as `block_body` does.
1162 Some(kind) if is_top_level_start(kind) => {
1163 self.error_at_current("FORM-103", "unclosed `{`".to_string());
1164 break;
1165 }
1166 Some(_) => self.err_and_recover("in an interface body", |kind| {
1167 matches!(
1168 kind,
1169 SyntaxKind::RBrace | SyntaxKind::Comma | SyntaxKind::ReservedKw
1170 ) || is_interaction_start(kind)
1171 || is_top_level_start(kind)
1172 }),
1173 }
1174 }
1175 }
1176
1177 /// `ServiceDef = 'service' DottedName (':' shapes (',' shapes)* ','? |
1178 /// '{' (inline_members ','?)* '}')` — the global published declaration of
1179 /// one or more interfaces (ridl reference §14.5, ADR-0015 decision 12).
1180 /// The named form carries a comma-separated shape list after `:`; the
1181 /// inline form carries an interaction body reusing
1182 /// [`Parser::interface_body`], so the checker runs the same structural pass
1183 /// over it (RIDL-401/-402). Never both forms (ADR-0015 decision 14). A
1184 /// service takes no `internal`/`error` modifiers — it is a global, published
1185 /// contract.
1186 fn service_def(&mut self) {
1187 self.start(SyntaxKind::ServiceDef);
1188 self.bump(); // 'service'
1189 self.dotted_name();
1190 match self.current() {
1191 Some(SyntaxKind::Colon) => {
1192 self.bump(); // ':'
1193 self.service_shape_list();
1194 }
1195 Some(SyntaxKind::LBrace) => {
1196 self.bump(); // '{'
1197 self.interface_body("a service", "service");
1198 }
1199 _ => self.error_at_current(
1200 "FORM-101",
1201 "expected `:` and an interface list, or `{` and an inline shape".to_string(),
1202 ),
1203 }
1204 self.builder.finish_node();
1205 }
1206
1207 /// The comma-separated shape list of a service's named form: one
1208 /// `PathType` per shape (ADR-0015 decision 12).
1209 ///
1210 /// Commas are **required** between shapes, diverging from the family's
1211 /// optional-comma convention, and the reason is structural (ADR-0015
1212 /// decision 13): every other list in the grammar is terminated by a
1213 /// closing token — `}` for a body, `)` for a parameter list, `]` for an
1214 /// attribute block — and this one is not, because it ends where the next
1215 /// declaration begins. It would still parse without commas, since every
1216 /// top-level declaration starts with a keyword and a bare `CamelCase`
1217 /// identifier never does — but it would parse greedily, so a mistyped
1218 /// declaration on a following line (`Struct Foo {` for `struct Foo {`)
1219 /// would be absorbed as another shape and the error would surface at the
1220 /// `{` with no connection to the mistake. Ending the list at the first
1221 /// missing comma instead leaves that identifier to the top-level loop,
1222 /// whose FORM-102 points at it directly. A trailing comma stays optional.
1223 fn service_shape_list(&mut self) {
1224 // One shape: an interface reference. A primitive keyword still parses
1225 // as a path segment and the checker rejects it (RIDL-141), the E1
1226 // lenient-parser discipline. `reserved` is not a shape: a service's
1227 // list is a set with no slot to retire (lock design §9), so the list
1228 // ends before it and the top-level loop reports the stray keyword.
1229 self.path_type();
1230 // `current()` peeks past trivia without consuming it, so a list that
1231 // ends here leaves the trailing trivia outside the ServiceDef node.
1232 while self.at(SyntaxKind::Comma) {
1233 self.bump(); // ','
1234 // A trailing comma: nothing after it starts a shape.
1235 if !self.at_path_segment() {
1236 break;
1237 }
1238 self.path_type();
1239 }
1240 }
1241
1242 /// `DottedName = 'ident' ('.' 'ident')*` — a dotted global service name
1243 /// (`veh.adas.cruise`, ADR-0002 §1). Each segment is a plain identifier;
1244 /// a trailing `.` with no segment after it is FORM-101.
1245 fn dotted_name(&mut self) {
1246 self.start(SyntaxKind::DottedName);
1247 if self.at(SyntaxKind::Ident) {
1248 self.bump();
1249 while self.at(SyntaxKind::Dot) {
1250 self.bump(); // '.'
1251 if self.at(SyntaxKind::Ident) {
1252 self.bump();
1253 } else {
1254 self.error_at_current(
1255 "FORM-101",
1256 "expected a name segment after `.`".to_string(),
1257 );
1258 break;
1259 }
1260 }
1261 } else {
1262 self.error_at_current("FORM-101", "expected a service name".to_string());
1263 }
1264 self.builder.finish_node();
1265 }
1266
1267 /// Whether the interface body being parsed still closes: scans forward,
1268 /// brace-aware, for an `RBrace` at depth 0 before the end of input or an
1269 /// unambiguous new-top-level marker (`package`, `import`, `interface`).
1270 /// Decides the recovery for a typl definition keyword inside a body: a
1271 /// body-local ErrorNode when the body closes (RIDL-107),
1272 /// an unclosed-`{` report when it does not. The scan runs through the
1273 /// definition keywords themselves — several stray declarations inside
1274 /// one closed body must all recover in place.
1275 fn interface_body_close_ahead(&self) -> bool {
1276 let mut depth = 0usize;
1277 for token in self.tokens[self.significant_pos()..]
1278 .iter()
1279 .filter(|token| !token.kind.is_trivia())
1280 {
1281 match token.kind {
1282 SyntaxKind::LBrace => depth += 1,
1283 SyntaxKind::RBrace if depth == 0 => return true,
1284 SyntaxKind::RBrace => depth -= 1,
1285 SyntaxKind::PackageKw
1286 | SyntaxKind::ImportKw
1287 | SyntaxKind::InterfaceKw
1288 | SyntaxKind::ServiceKw
1289 if depth == 0 =>
1290 {
1291 return false;
1292 }
1293 _ => {}
1294 }
1295 }
1296 false
1297 }
1298
1299 /// Consumes one stray typl declaration inside an interface body into a
1300 /// single [`ErrorNode`](SyntaxKind::ErrorNode). Brace-aware, unlike
1301 /// [`Parser::recover`]: a composite body (`struct`/`enum`/… with its
1302 /// `{ … }`, commas included) is consumed whole, so the declaration's own
1303 /// `}` is never taken for the interface's closer and the members after
1304 /// it stay in the body. At depth 0 the interface sync set applies. The
1305 /// caller has confirmed the body's `}` lies ahead
1306 /// ([`Parser::interface_body_close_ahead`]), so the loop terminates
1307 /// there at the latest.
1308 fn recover_definition_in_body(&mut self) {
1309 self.start(SyntaxKind::ErrorNode);
1310 self.bump(); // the typl definition keyword
1311 let mut depth = 0usize;
1312 while let Some(kind) = self.current() {
1313 match kind {
1314 SyntaxKind::LBrace => depth += 1,
1315 SyntaxKind::RBrace if depth > 0 => depth -= 1,
1316 _ if depth == 0
1317 && (matches!(
1318 kind,
1319 SyntaxKind::RBrace | SyntaxKind::Comma | SyntaxKind::ReservedKw
1320 ) || is_interaction_start(kind)
1321 || is_top_level_start(kind)) =>
1322 {
1323 break;
1324 }
1325 _ => {}
1326 }
1327 self.bump();
1328 }
1329 self.builder.finish_node();
1330 }
1331
1332 /// The shared `kw Name ':' payload InitValue? annotations` shape of the
1333 /// three value interactions — `SignalDef`, `EventDef`, and `FixedDef`.
1334 /// The bare `= value` init comes before the timing (ADR-0008 decision
1335 /// 2). All three kinds accept an init, a timing, and an attr block, as
1336 /// the reference's Appendix C shows; the checker narrows them (FORM-102
1337 /// for an init on an event or a fixed, RIDL-106/-301).
1338 fn value_interaction(&mut self, kind: SyntaxKind) {
1339 self.start(kind);
1340 self.bump(); // 'signal' | 'event' | 'fixed'
1341 self.name();
1342 self.expect(SyntaxKind::Colon);
1343 self.field_type();
1344 self.init_value_opt();
1345 self.interaction_annotations();
1346 self.builder.finish_node();
1347 }
1348
1349 /// The shared `kw Name '(' params ')' (':' ReturnType)? annotations`
1350 /// shape of the two callable interactions — `CommandDef` and
1351 /// `QueryDef`. A command's return type is lenient here (RIDL-104,
1352 /// checker scope); a query's is mandatory — a missing `:` draws
1353 /// FORM-101 (ridl reference §6.1, §7.1).
1354 fn callable_interaction(&mut self, kind: SyntaxKind) {
1355 self.start(kind);
1356 self.bump(); // 'command' | 'query'
1357 self.name();
1358 self.param_list();
1359 if self.at(SyntaxKind::Colon) {
1360 self.bump();
1361 self.return_type();
1362 } else if kind == SyntaxKind::QueryDef {
1363 self.error_at_current("FORM-101", "expected `:` and a return type".to_string());
1364 }
1365 self.interaction_annotations();
1366 self.builder.finish_node();
1367 }
1368
1369 /// `ParamList = '(' (params:Param ','?)* ')'` — separator commas are
1370 /// direct children of the list node, mirroring the block-body
1371 /// discipline (typl reference §15.2). No node is built when the `(` is
1372 /// missing, so `CommandDef::params()` sees `None`.
1373 fn param_list(&mut self) {
1374 if !self.at(SyntaxKind::LParen) {
1375 self.error_at_current("FORM-101", "expected `(`".to_string());
1376 return;
1377 }
1378 self.start(SyntaxKind::ParamList);
1379 self.bump(); // '('
1380 loop {
1381 self.eat_trivia();
1382 match self.current() {
1383 None => {
1384 self.error_at_current("FORM-103", "unclosed `(`".to_string());
1385 break;
1386 }
1387 Some(SyntaxKind::RParen) => {
1388 self.bump();
1389 break;
1390 }
1391 Some(SyntaxKind::Comma) => self.bump(),
1392 Some(SyntaxKind::Ident) => self.param(),
1393 // An unclosed `(`: the list ran into an interaction, body,
1394 // or top-level boundary. Report it and hand the token back.
1395 Some(kind)
1396 if kind == SyntaxKind::RBrace
1397 || is_interaction_start(kind)
1398 || is_top_level_start(kind) =>
1399 {
1400 self.error_at_current("FORM-103", "unclosed `(`".to_string());
1401 break;
1402 }
1403 Some(_) => self.err_and_recover("in a parameter list", |kind| {
1404 matches!(
1405 kind,
1406 SyntaxKind::RParen
1407 | SyntaxKind::Comma
1408 | SyntaxKind::Ident
1409 | SyntaxKind::RBrace
1410 ) || is_interaction_start(kind)
1411 || is_top_level_start(kind)
1412 }),
1413 }
1414 }
1415 self.builder.finish_node();
1416 }
1417
1418 /// `Param = Name ':' (FieldType | StreamType)` — a named typl type, a
1419 /// tuple (typl §11), or a stream `<T>` (ridl reference §12.1); the
1420 /// stream parses through the shared field-type dispatch.
1421 fn param(&mut self) {
1422 self.start(SyntaxKind::Param);
1423 self.name();
1424 self.expect(SyntaxKind::Colon);
1425 self.field_type();
1426 self.builder.finish_node();
1427 }
1428
1429 /// `ReturnType = PathType | TupleType | StreamType | FallibleType` —
1430 /// the four return shapes (ridl reference §7.1). A named type followed
1431 /// by `|` extends into `FallibleType = ok '|' err` (general form §6.1,
1432 /// ADR-0008 decision 1). No node is built when no shape starts here, so
1433 /// `return_type()` accessors see `None`.
1434 fn return_type(&mut self) {
1435 self.eat_trivia();
1436 match self.current() {
1437 Some(SyntaxKind::LParen) => {
1438 self.start(SyntaxKind::ReturnType);
1439 self.tuple_type();
1440 self.builder.finish_node();
1441 }
1442 Some(SyntaxKind::Lt) => {
1443 self.start(SyntaxKind::ReturnType);
1444 self.stream_type();
1445 self.builder.finish_node();
1446 }
1447 _ if self.at_path_segment() => {
1448 self.start(SyntaxKind::ReturnType);
1449 let checkpoint = self.builder.checkpoint();
1450 self.path_type();
1451 if self.at(SyntaxKind::Pipe) {
1452 self.builder
1453 .start_node_at(checkpoint, SyntaxKind::FallibleType.into());
1454 self.bump(); // '|'
1455 self.path_type();
1456 self.builder.finish_node();
1457 }
1458 self.builder.finish_node();
1459 }
1460 _ => {
1461 self.error_at_current(
1462 "FORM-101",
1463 "a return type must be a named type, a named-field tuple `(a: A, b: B)`, a \
1464 stream `<T>`, or a fallible `T | E` (ridl reference §7.1)"
1465 .to_string(),
1466 );
1467 // Consume what stands here into an `ErrorNode` rather than
1468 // leaving it for `interaction_annotations`. A rejected return
1469 // shape most often starts with `[` — a collection, `[Level;
1470 // 0..8]` — and leaving it behind made the annotation parser
1471 // read it as this member's attribute block, so one mistake drew
1472 // a FORM-102 and a FORM-106 "unknown attribute key `Level`" on
1473 // top of the real diagnostic. `@` stays in the sync set so a
1474 // timing annotation written after a broken return is still
1475 // parsed and still reaches its own check.
1476 if self.current().is_some_and(|kind| !is_return_sync(kind)) {
1477 self.recover(is_return_sync);
1478 }
1479 }
1480 }
1481 }
1482
1483 /// `StreamType = '<' (PathType | 'string' | 'bytes') '>'` — the element
1484 /// is a named type, or a bare raw `string`/`bytes` token (ridl
1485 /// reference §12.2, the one exception to typl §15.3). Callers have
1486 /// confirmed the `<`.
1487 fn stream_type(&mut self) {
1488 self.start(SyntaxKind::StreamType);
1489 self.bump(); // '<'
1490 if matches!(
1491 self.current(),
1492 Some(SyntaxKind::StringKw | SyntaxKind::BytesKw)
1493 ) && self.nth(1) == Some(SyntaxKind::Gt)
1494 {
1495 self.bump(); // the raw element keyword
1496 } else if self.at_path_segment() {
1497 self.path_type();
1498 } else {
1499 self.error_at_current("FORM-101", "expected an element type".to_string());
1500 }
1501 self.expect(SyntaxKind::Gt);
1502 self.builder.finish_node();
1503 }
1504
1505 /// The lenient trailing annotations of an interaction: at most one
1506 /// [`Timing`](SyntaxKind::Timing) and at most one
1507 /// [`AttrBlock`](SyntaxKind::AttrBlock), in either order. Which kinds
1508 /// may carry which annotation is checker scope (RIDL-104/-106/-301); the
1509 /// parser accepts both on every interaction.
1510 fn interaction_annotations(&mut self) {
1511 let mut seen_timing = false;
1512 let mut seen_attrs = false;
1513 loop {
1514 match self.current() {
1515 Some(SyntaxKind::At) if !seen_timing => {
1516 seen_timing = true;
1517 self.timing();
1518 }
1519 Some(SyntaxKind::LBracket) if !seen_attrs => {
1520 seen_attrs = true;
1521 self.attr_block();
1522 }
1523 _ => break,
1524 }
1525 }
1526 }
1527
1528 /// `Timing = '@' ('duration' | '[' TimingRange ']')` — strict periodic
1529 /// or range (ridl reference §9). Callers have confirmed the `@`.
1530 fn timing(&mut self) {
1531 self.start(SyntaxKind::Timing);
1532 self.bump(); // '@'
1533 if self.at(SyntaxKind::Duration) {
1534 self.bump();
1535 } else if self.at(SyntaxKind::LBracket) {
1536 self.bump();
1537 self.timing_range();
1538 self.expect(SyntaxKind::RBracket);
1539 } else {
1540 self.error_at_current("FORM-101", "expected a duration or `[`".to_string());
1541 }
1542 self.builder.finish_node();
1543 }
1544
1545 /// `TimingRange = 'duration'? '..' 'duration'?` — both bounds or the
1546 /// half-open forms `min..` and `..max` (ridl reference §9). A range
1547 /// with neither bound is a structural error (FORM-101).
1548 fn timing_range(&mut self) {
1549 self.start(SyntaxKind::TimingRange);
1550 let has_min = self.at(SyntaxKind::Duration);
1551 if has_min {
1552 self.bump();
1553 }
1554 self.expect(SyntaxKind::DotDot);
1555 let has_max = self.at(SyntaxKind::Duration);
1556 if has_max {
1557 self.bump();
1558 }
1559 if !has_min && !has_max {
1560 self.error_at_current("FORM-101", "expected a duration".to_string());
1561 }
1562 self.builder.finish_node();
1563 }
1564
1565 /// `AttrBlock = '[' (attributes:Attribute ','?)* ']'` — the single
1566 /// attribute production of the general form (§4.2–§4.3). Separator
1567 /// commas are direct children of the block node, mirroring the
1568 /// block-body discipline. gf §4.2 requires at least one attribute per
1569 /// block: an empty `[]` still parses losslessly into an `AttrBlock`
1570 /// node and FORM-101 (`expected an attribute`) lands on the closing
1571 /// bracket. An unclosed block stops at an interaction or body boundary
1572 /// (FORM-103) instead of swallowing the rest of the interface.
1573 fn attr_block(&mut self) {
1574 self.start(SyntaxKind::AttrBlock);
1575 self.bump(); // '['
1576 let mut saw_attribute = false;
1577 loop {
1578 match self.current() {
1579 None => {
1580 self.error_at_current("FORM-103", "unclosed `[`".to_string());
1581 break;
1582 }
1583 Some(SyntaxKind::RBracket) => {
1584 if !saw_attribute {
1585 self.error_at_current("FORM-101", "expected an attribute".to_string());
1586 }
1587 self.bump();
1588 break;
1589 }
1590 Some(SyntaxKind::Comma) => self.bump(),
1591 // A `ReservedWord` in key position goes through
1592 // [`Parser::name`], which holds it in an `ErrorNode` with
1593 // FORM-105 — the E1 reserved-word discipline — so the rest
1594 // of the attribute (`= value`) still parses.
1595 Some(
1596 SyntaxKind::Ident
1597 | SyntaxKind::RequireKw
1598 | SyntaxKind::EnsureKw
1599 | SyntaxKind::ReservedWord,
1600 ) => {
1601 saw_attribute = true;
1602 self.attribute();
1603 }
1604 Some(kind)
1605 if kind == SyntaxKind::RBrace
1606 || is_interaction_start(kind)
1607 || is_top_level_start(kind) =>
1608 {
1609 self.error_at_current("FORM-103", "unclosed `[`".to_string());
1610 break;
1611 }
1612 Some(_) => self.err_and_recover("in an attribute block", |kind| {
1613 matches!(
1614 kind,
1615 SyntaxKind::RBracket
1616 | SyntaxKind::Comma
1617 | SyntaxKind::Ident
1618 | SyntaxKind::RequireKw
1619 | SyntaxKind::EnsureKw
1620 | SyntaxKind::ReservedWord
1621 | SyntaxKind::RBrace
1622 ) || is_interaction_start(kind)
1623 || is_top_level_start(kind)
1624 }),
1625 }
1626 }
1627 self.builder.finish_node();
1628 }
1629
1630 /// `Attribute = key | key '=' const_value | ('require'|'ensure') expr`
1631 /// — the three forms of gf §4.2. One grammar production for all of
1632 /// them; which keys are legal where is checker scope (gf §4.3,
1633 /// FORM-106/107/108). Callers have confirmed the leading token is an
1634 /// `Ident` or a predicate keyword.
1635 fn attribute(&mut self) {
1636 self.start(SyntaxKind::Attribute);
1637 match self.current() {
1638 Some(SyntaxKind::RequireKw | SyntaxKind::EnsureKw) => {
1639 self.bump(); // 'require' | 'ensure'
1640 self.expr();
1641 }
1642 _ => {
1643 self.name(); // the attr key
1644 // A backend key `namespace.key` (rsdl reference §5, Appendix
1645 // B `key`). Only the rsdl profile admits the dotted key; under
1646 // typl and ridl the `.` stays an unexpected token in the block.
1647 if self.profile == Profile::Rsdl && self.at(SyntaxKind::Dot) {
1648 self.bump(); // '.'
1649 self.name();
1650 }
1651 if self.at(SyntaxKind::Eq) {
1652 self.bump();
1653 self.attr_value();
1654 }
1655 }
1656 }
1657 self.builder.finish_node();
1658 }
1659
1660 /// `const_value = literal | SCREAMING_SNAKE_ID | '(' const_value (','
1661 /// const_value)* ')'` (gf §4.2), as an [`AttrValue`](SyntaxKind::AttrValue)
1662 /// node: one `Literal` child (an `ident` value token is the constant
1663 /// reference), or a parenthesised list of nested `AttrValue`s. The
1664 /// depth guard bounds pathological list nesting, sharing the
1665 /// [`MAX_TYPE_DEPTH`] budget with the type and expression grammars.
1666 fn attr_value(&mut self) {
1667 if self.depth >= MAX_TYPE_DEPTH {
1668 self.error_at_current(
1669 "FORM-102",
1670 format!("value nesting deeper than {MAX_TYPE_DEPTH} levels"),
1671 );
1672 self.start(SyntaxKind::ErrorNode);
1673 self.bump();
1674 self.builder.finish_node();
1675 return;
1676 }
1677 self.depth += 1;
1678 self.attr_value_inner();
1679 self.depth -= 1;
1680 }
1681
1682 fn attr_value_inner(&mut self) {
1683 self.eat_trivia();
1684 if self.at(SyntaxKind::LParen) {
1685 self.start(SyntaxKind::AttrValue);
1686 self.bump(); // '('
1687 loop {
1688 self.eat_trivia();
1689 match self.current() {
1690 None => {
1691 self.error_at_current("FORM-103", "unclosed `(`".to_string());
1692 break;
1693 }
1694 Some(SyntaxKind::RParen) => {
1695 self.bump();
1696 break;
1697 }
1698 Some(SyntaxKind::Comma) => self.bump(),
1699 Some(SyntaxKind::LParen) => self.attr_value(),
1700 Some(kind)
1701 if kind == SyntaxKind::RBracket
1702 || kind == SyntaxKind::RBrace
1703 || is_interaction_start(kind)
1704 || is_top_level_start(kind) =>
1705 {
1706 self.error_at_current("FORM-103", "unclosed `(`".to_string());
1707 break;
1708 }
1709 _ if self.literal_len_at(0) > 0 => self.attr_value(),
1710 Some(_) => self.err_and_recover("in an attribute value", |kind| {
1711 matches!(
1712 kind,
1713 SyntaxKind::RParen
1714 | SyntaxKind::Comma
1715 | SyntaxKind::LParen
1716 | SyntaxKind::RBracket
1717 | SyntaxKind::RBrace
1718 ) || is_interaction_start(kind)
1719 || is_top_level_start(kind)
1720 }),
1721 }
1722 }
1723 self.builder.finish_node();
1724 } else if self.literal_len_at(0) > 0 {
1725 self.start(SyntaxKind::AttrValue);
1726 self.literal();
1727 self.builder.finish_node();
1728 } else {
1729 self.error_at_current("FORM-101", "expected a value".to_string());
1730 }
1731 }
1732
1733 // --- the guaranteed-subset expression grammar ------------------
1734 //
1735 // The expr-core specification §3.1 productions, one function per
1736 // precedence level, loosest to tightest: `||` — `&&` — comparison —
1737 // additive — multiplicative — unary — member access. Binary levels are
1738 // left-associative (each iteration wraps the run so far in a fresh
1739 // BinaryExpr from the same checkpoint); a comparison never chains.
1740 // Inside an expression `<` is always the comparison operator — the
1741 // stream-type reading of `<` exists only in param-type and return-type
1742 // position, and the two positions never overlap.
1743 //
1744 // The height bound (driftsys/ridl#346). A left-associative chain builds
1745 // a tree as high as its term count, and a pass that recurses over the
1746 // tree needs stack in proportion to that height, so the parser bounds
1747 // the height of every expression tree at MAX_TYPE_DEPTH nodes. Every
1748 // function below returns the height of the subtree it built (1 for a
1749 // node whose children are all tokens, 0 when no node was built), and
1750 // `depth` counts the expression nodes open above the cursor. A node is
1751 // built only when `depth + height <= MAX_TYPE_DEPTH` (`expr_fits`): a
1752 // chain checks the wrap before consuming its operator, from the height
1753 // of the run so far plus one, and parses the right operand one level
1754 // deeper, so the operand's own checks account for the wrap above it; a
1755 // member step is checked like a chain wrap; a group or a prefix is
1756 // opened only when it and one leaf below it fit.
1757 // The invariant holds by induction over the build order: each node's
1758 // check covers the node itself, and a child built earlier sits one level
1759 // deeper under a parent whose height is at least the child's plus one.
1760 // A leaf is never checked, because every node above it was admitted
1761 // with room for one — a chain wrap counts its left operand as at least
1762 // one high even when that operand is missing. A
1763 // budget charged per parse call and released on return would not bound
1764 // the tree, because chains closed inside groups, and chains that are the
1765 // first operand of a looser chain, stack on the left spine.
1766 //
1767 // Past the limit the expression draws one FORM-102 and the rest of the
1768 // current group goes flat, as tokens under one ErrorNode: the tree stays
1769 // lossless and a later part of the same expression that does not fit
1770 // either is consumed the same way without a second diagnostic.
1771
1772 /// `expr = or_expr` — the entry point, called from a predicate
1773 /// attribute; a parenthesised group recurses through [`Parser::primary`].
1774 /// An attribute block is never inside a type or a value, so the
1775 /// expression starts at depth 0 and a leaf always fits below it.
1776 fn expr(&mut self) {
1777 self.expr_refused = false;
1778 self.or_expr();
1779 }
1780
1781 /// Whether an expression node `height` levels high fits at the current
1782 /// depth. Past the limit, reports FORM-102 at the current token unless
1783 /// the expression has already drawn one, so that one overlong expression
1784 /// draws one diagnostic whatever its shape. The message says how the
1785 /// levels are counted — one per binary operator, member access,
1786 /// parenthesised group and prefix — because the nesting of a flat
1787 /// operator chain is not visible in the source.
1788 fn expr_fits(&mut self, height: usize) -> bool {
1789 if self.depth + height <= MAX_TYPE_DEPTH {
1790 return true;
1791 }
1792 if !self.expr_refused {
1793 self.expr_refused = true;
1794 self.error_at_current(
1795 "FORM-102",
1796 format!(
1797 "expression nesting deeper than {MAX_TYPE_DEPTH} levels (each operator, \
1798 member access, parenthesised group and prefix nests one level)"
1799 ),
1800 );
1801 }
1802 false
1803 }
1804
1805 /// Consumes the rest of the current expression level as flat children
1806 /// of one [`ErrorNode`](SyntaxKind::ErrorNode): every token up to the
1807 /// `)` that closes the enclosing group, the `]` or `,` that ends the
1808 /// attribute, a brace, the end of input, a token that starts another
1809 /// construct, or the start of the next attribute, with or without a
1810 /// comma before it — a `require` or `ensure`, a reserved word in key
1811 /// position, or a key, which is an identifier where the expression has
1812 /// just ended in an operand and no operator can follow. A group opened
1813 /// inside is consumed through its own `)`.
1814 fn skip_expr_rest(&mut self) {
1815 self.start(SyntaxKind::ErrorNode);
1816 self.skip_expr_tokens();
1817 self.builder.finish_node();
1818 }
1819
1820 /// The token loop of [`Parser::skip_expr_rest`], without the node.
1821 fn skip_expr_tokens(&mut self) {
1822 let mut open = 0usize;
1823 let mut after_operand = false;
1824 while let Some(kind) = self.current() {
1825 match kind {
1826 SyntaxKind::RBracket
1827 | SyntaxKind::Comma
1828 | SyntaxKind::LBrace
1829 | SyntaxKind::RBrace
1830 | SyntaxKind::RequireKw
1831 | SyntaxKind::EnsureKw
1832 | SyntaxKind::ReservedWord => break,
1833 SyntaxKind::RParen if open == 0 => break,
1834 SyntaxKind::Ident if after_operand => break,
1835 _ if is_interaction_start(kind) || is_top_level_start(kind) => break,
1836 SyntaxKind::RParen => open -= 1,
1837 SyntaxKind::LParen => open += 1,
1838 _ => {}
1839 }
1840 after_operand = matches!(
1841 kind,
1842 SyntaxKind::Ident
1843 | SyntaxKind::IntNumber
1844 | SyntaxKind::FloatNumber
1845 | SyntaxKind::String
1846 | SyntaxKind::TrueKw
1847 | SyntaxKind::FalseKw
1848 | SyntaxKind::Duration
1849 | SyntaxKind::RParen
1850 );
1851 self.bump();
1852 }
1853 }
1854
1855 /// The shared loop of the four chaining binary levels: `operand { op
1856 /// operand }`, each `op` a token `is_op` accepts, each iteration wrapping
1857 /// the run so far in a [`BinaryExpr`](SyntaxKind::BinaryExpr) from the
1858 /// same checkpoint. Returns the height of the tree built.
1859 fn chain(&mut self, is_op: fn(SyntaxKind) -> bool, operand: fn(&mut Self) -> usize) -> usize {
1860 self.eat_trivia();
1861 let checkpoint = self.builder.checkpoint();
1862 let mut height = operand(self);
1863 while self.current().is_some_and(is_op) {
1864 // A missing left operand (FORM-101, already reported) has height
1865 // 0; the wrap still holds the right operand, so it is at least
1866 // two high, and the check must say so for the leaf below to fit.
1867 if !self.expr_fits(height.max(1) + 1) {
1868 self.skip_expr_rest();
1869 break;
1870 }
1871 self.builder
1872 .start_node_at(checkpoint, SyntaxKind::BinaryExpr.into());
1873 self.bump();
1874 self.depth += 1;
1875 let rhs = operand(self);
1876 self.depth -= 1;
1877 self.builder.finish_node();
1878 height = 1 + height.max(rhs);
1879 }
1880 height
1881 }
1882
1883 /// `or_expr = and_expr { '||' and_expr }`
1884 fn or_expr(&mut self) -> usize {
1885 self.chain(|kind| kind == SyntaxKind::PipePipe, Self::and_expr)
1886 }
1887
1888 /// `and_expr = cmp_expr { '&&' cmp_expr }`
1889 fn and_expr(&mut self) -> usize {
1890 self.chain(|kind| kind == SyntaxKind::AmpAmp, Self::cmp_expr)
1891 }
1892
1893 /// `cmp_expr = add_expr [ cmp_op add_expr ]` — at most one comparison:
1894 /// `a < b < c` is a parse error (write `a < b && b < c`), reported by
1895 /// whatever context the leftover operator lands in.
1896 fn cmp_expr(&mut self) -> usize {
1897 self.eat_trivia();
1898 let checkpoint = self.builder.checkpoint();
1899 let mut height = self.add_expr();
1900 if matches!(
1901 self.current(),
1902 Some(
1903 SyntaxKind::EqEq
1904 | SyntaxKind::Neq
1905 | SyntaxKind::Lt
1906 | SyntaxKind::Le
1907 | SyntaxKind::Gt
1908 | SyntaxKind::Ge
1909 )
1910 ) {
1911 if !self.expr_fits(height.max(1) + 1) {
1912 self.skip_expr_rest();
1913 return height;
1914 }
1915 self.builder
1916 .start_node_at(checkpoint, SyntaxKind::BinaryExpr.into());
1917 self.bump();
1918 self.depth += 1;
1919 let rhs = self.add_expr();
1920 self.depth -= 1;
1921 self.builder.finish_node();
1922 height = 1 + height.max(rhs);
1923 }
1924 height
1925 }
1926
1927 /// `add_expr = mul_expr { ('+' | '-') mul_expr }`
1928 fn add_expr(&mut self) -> usize {
1929 self.chain(
1930 |kind| matches!(kind, SyntaxKind::Plus | SyntaxKind::Minus),
1931 Self::mul_expr,
1932 )
1933 }
1934
1935 /// `mul_expr = unary_expr { ('*' | '/' | '%') unary_expr }`
1936 fn mul_expr(&mut self) -> usize {
1937 self.chain(
1938 |kind| {
1939 matches!(
1940 kind,
1941 SyntaxKind::Star | SyntaxKind::Slash | SyntaxKind::Percent
1942 )
1943 },
1944 Self::unary_expr,
1945 )
1946 }
1947
1948 /// `unary_expr = [ '!' | '-' ] postfix_expr` — at most one prefix.
1949 fn unary_expr(&mut self) -> usize {
1950 if matches!(self.current(), Some(SyntaxKind::Bang | SyntaxKind::Minus)) {
1951 if !self.expr_fits(2) {
1952 self.skip_expr_rest();
1953 return 1;
1954 }
1955 self.start(SyntaxKind::PrefixExpr);
1956 self.bump();
1957 self.depth += 1;
1958 let operand = self.postfix_expr();
1959 self.depth -= 1;
1960 self.builder.finish_node();
1961 1 + operand
1962 } else {
1963 self.postfix_expr()
1964 }
1965 }
1966
1967 /// `postfix_expr = primary { '.' member }` — each `.member` step wraps
1968 /// the run so far in a [`MemberExpr`](SyntaxKind::MemberExpr), so
1969 /// `filter.severity` and `GearPosition.PARK` nest left-associatively.
1970 fn postfix_expr(&mut self) -> usize {
1971 self.eat_trivia();
1972 let checkpoint = self.builder.checkpoint();
1973 let mut height = self.primary();
1974 while self.at(SyntaxKind::Dot) {
1975 if !self.expr_fits(height + 1) {
1976 self.skip_expr_rest();
1977 break;
1978 }
1979 self.builder
1980 .start_node_at(checkpoint, SyntaxKind::MemberExpr.into());
1981 self.bump(); // '.'
1982 if self.at(SyntaxKind::Ident) {
1983 self.bump();
1984 } else {
1985 self.error_at_current("FORM-101", "expected a member name".to_string());
1986 }
1987 self.builder.finish_node();
1988 height += 1;
1989 }
1990 height
1991 }
1992
1993 /// `primary = literal | duration_lit | path_head | '(' expr ')'` — a
1994 /// [`LiteralExpr`](SyntaxKind::LiteralExpr) (typl literal or duration —
1995 /// zero durations are legal in expression position, expr-core spec
1996 /// §3.1), a [`PathExpr`](SyntaxKind::PathExpr), or a parenthesised
1997 /// group. No node is built when no primary starts here, so a missing
1998 /// operand reports FORM-101 without consuming the boundary token. The
1999 /// group is where the grammar recurses, so the height bound is also what
2000 /// keeps a run of `(` from overflowing the parser's own stack: a group
2001 /// that does not fit is consumed flat through its `)`.
2002 fn primary(&mut self) -> usize {
2003 match self.current() {
2004 Some(
2005 SyntaxKind::IntNumber
2006 | SyntaxKind::FloatNumber
2007 | SyntaxKind::String
2008 | SyntaxKind::TrueKw
2009 | SyntaxKind::FalseKw
2010 | SyntaxKind::Duration,
2011 ) => {
2012 self.start(SyntaxKind::LiteralExpr);
2013 if self.at(SyntaxKind::IntNumber) {
2014 self.flag_leading_zeros();
2015 }
2016 self.bump();
2017 self.builder.finish_node();
2018 1
2019 }
2020 Some(SyntaxKind::Ident) => {
2021 self.start(SyntaxKind::PathExpr);
2022 self.bump();
2023 self.builder.finish_node();
2024 1
2025 }
2026 Some(SyntaxKind::LParen) => {
2027 if !self.expr_fits(2) {
2028 self.start(SyntaxKind::ErrorNode);
2029 self.bump(); // '('
2030 self.skip_expr_tokens();
2031 if self.at(SyntaxKind::RParen) {
2032 self.bump();
2033 }
2034 self.builder.finish_node();
2035 return 1;
2036 }
2037 self.start(SyntaxKind::ParenExpr);
2038 self.bump();
2039 self.depth += 1;
2040 let inner = self.or_expr();
2041 self.depth -= 1;
2042 self.expect(SyntaxKind::RParen);
2043 self.builder.finish_node();
2044 1 + inner
2045 }
2046 _ => {
2047 self.error_at_current("FORM-101", "expected an expression".to_string());
2048 0
2049 }
2050 }
2051 }
2052
2053 /// `EnumValue = Name '=' value:Literal`
2054 fn enum_value(&mut self) {
2055 self.value_assignment(SyntaxKind::EnumValue);
2056 }
2057
2058 /// `EnumSetBit = Name '=' value:Literal`
2059 fn enum_set_bit(&mut self) {
2060 self.value_assignment(SyntaxKind::EnumSetBit);
2061 }
2062
2063 /// The shared `Name '=' Literal` shape of enum values and enumset bits.
2064 fn value_assignment(&mut self, kind: SyntaxKind) {
2065 self.start(kind);
2066 self.name();
2067 self.expect(SyntaxKind::Eq);
2068 self.literal();
2069 self.builder.finish_node();
2070 }
2071
2072 /// The shared block-body loop (§15.2): members separated by newlines or
2073 /// commas, trailing comma permitted, separator commas placed as direct
2074 /// children of the block node between members. `member` parses the
2075 /// member a leading `Ident` announces; `allow_reserved` admits
2076 /// `ReservedEntry` members (structs, enums, unions — not enumsets).
2077 fn block_body(&mut self, context: &str, allow_reserved: bool, member: fn(&mut Self)) {
2078 if !self.block_open() {
2079 return;
2080 }
2081 loop {
2082 self.eat_trivia();
2083 match self.current() {
2084 None => {
2085 self.error_at_current("FORM-103", "unclosed `{`".to_string());
2086 break;
2087 }
2088 Some(SyntaxKind::RBrace) => {
2089 self.bump();
2090 break;
2091 }
2092 Some(SyntaxKind::Comma) => self.bump(),
2093 Some(SyntaxKind::ReservedKw) if allow_reserved => self.reserved_entry(),
2094 Some(SyntaxKind::Ident) => member(self),
2095 // An unclosed `{`: the body ran into the next top-level
2096 // declaration. Report the missing brace and let `source_file`
2097 // resynchronize instead of swallowing the declaration.
2098 Some(kind) if is_top_level_start(kind) => {
2099 self.error_at_current("FORM-103", "unclosed `{`".to_string());
2100 break;
2101 }
2102 Some(_) => self.err_and_recover(context, |kind| {
2103 matches!(
2104 kind,
2105 SyntaxKind::RBrace | SyntaxKind::Comma | SyntaxKind::Ident
2106 ) || (allow_reserved && kind == SyntaxKind::ReservedKw)
2107 || is_top_level_start(kind)
2108 }),
2109 }
2110 }
2111 }
2112
2113 /// Consumes the `{` opening a block body; reports FORM-101 and returns
2114 /// `false` when it is missing, so the caller skips the body loop instead
2115 /// of swallowing the rest of the file.
2116 fn block_open(&mut self) -> bool {
2117 if self.at(SyntaxKind::LBrace) {
2118 self.bump();
2119 true
2120 } else {
2121 self.error_at_current("FORM-101", "expected `{`".to_string());
2122 false
2123 }
2124 }
2125
2126 /// `FieldDef = Name ':' FieldType InitValue?`
2127 fn field_def(&mut self) {
2128 self.start(SyntaxKind::FieldDef);
2129 self.name();
2130 self.expect(SyntaxKind::Colon);
2131 self.field_type();
2132 self.init_value_opt();
2133 self.builder.finish_node();
2134 }
2135
2136 /// `ReservedEntry = 'reserved' (Name | int_lit)` — the tombstone (§7.4).
2137 /// typl Appendix E spells it `reserved = "reserved" ( camelCase_id |
2138 /// int_lit )`. The production is body-agnostic and this parser keeps it
2139 /// that way: both forms parse in every body that admits a tombstone.
2140 ///
2141 /// What each form *means* is per body, and **nothing narrows it** — there
2142 /// is no per-body check anywhere, so this comment describes the language,
2143 /// not an enforced rule. The name form is meaningful everywhere (TYPL-210
2144 /// in a struct, union or enum; RIDL-401 in an interface). The integer form
2145 /// is meaningful in an `enum`, where it retires a wire value (TYPL-210),
2146 /// and in an interaction body, where the nameless spelling records the
2147 /// ordinal it protects (`veh-cluster/cluster/evolution.ridl`). In a
2148 /// `struct` or `union` body it is **inert**: members there carry no
2149 /// explicit value, so nothing can ever match it and no retired name is
2150 /// recorded. `test_data/parser/ok/grammar_overapprox.typl` carries that
2151 /// case as an over-approximation awaiting a checker rule; see
2152 /// `veh-cluster/NOTES` for the residual and what closing it needs.
2153 ///
2154 /// Any other literal — a string, a boolean, a float, a regex, a bare
2155 /// constant reference — is outside the grammar and draws FORM-102 here
2156 /// rather than parsing as a `Literal`. Lowering keeps only a `Name` or an
2157 /// integer, so an ungrammatical literal would otherwise consume the
2158 /// ordinal while recording neither: the tombstone survives but the
2159 /// retired identity it exists to record is destroyed, with no diagnostic.
2160 /// The offending token is held in an `ErrorNode` so the tree stays
2161 /// lossless and the body loop resumes at the next member.
2162 fn reserved_entry(&mut self) {
2163 self.start(SyntaxKind::ReservedEntry);
2164 self.bump(); // 'reserved'
2165 if self.at(SyntaxKind::Ident) {
2166 self.name();
2167 } else if let Some(kind) = self.literal_value_kind_at(0) {
2168 if kind == SyntaxKind::IntNumber {
2169 self.literal();
2170 } else {
2171 self.start(SyntaxKind::ErrorNode);
2172 // A leading `-` belongs to the literal but is not what makes
2173 // it ungrammatical, so it is consumed before the diagnostic is
2174 // raised: the caret then lands on the value token the message
2175 // names, as it already does for every single-token shape.
2176 if self.at(SyntaxKind::Minus) {
2177 self.bump();
2178 }
2179 self.error_at_current(
2180 "FORM-102",
2181 format!(
2182 "a `reserved` tombstone takes a retired member name or a retired integer value, not {}",
2183 ungrammatical_reserved_noun(kind),
2184 ),
2185 );
2186 self.bump();
2187 self.builder.finish_node();
2188 }
2189 } else {
2190 self.error_at_current(
2191 "FORM-101",
2192 "expected a name or an integer value".to_string(),
2193 );
2194 }
2195 self.builder.finish_node();
2196 }
2197
2198 /// `FieldType = PathType | PrimitiveType | TupleType | ArrayType |
2199 /// MapType | OptionalType` — the type of a field, tuple field, or
2200 /// collection element. Each `?` suffix wraps the type parsed so far in an
2201 /// `OptionalType` node. The depth guard keeps a pathological nesting of
2202 /// collection and tuple types from overflowing the stack — the parser
2203 /// must never panic.
2204 fn field_type(&mut self) {
2205 if self.depth >= MAX_TYPE_DEPTH {
2206 self.error_at_current(
2207 "FORM-102",
2208 format!("type nesting deeper than {MAX_TYPE_DEPTH} levels"),
2209 );
2210 self.start(SyntaxKind::ErrorNode);
2211 self.bump();
2212 self.builder.finish_node();
2213 return;
2214 }
2215 self.depth += 1;
2216 self.field_type_inner();
2217 self.depth -= 1;
2218 }
2219
2220 fn field_type_inner(&mut self) {
2221 self.eat_trivia();
2222 let checkpoint = self.builder.checkpoint();
2223 match self.current() {
2224 Some(
2225 kw @ (SyntaxKind::BooleanKw
2226 | SyntaxKind::IntegerKw
2227 | SyntaxKind::FloatKw
2228 | SyntaxKind::StringKw
2229 | SyntaxKind::BytesKw),
2230 ) => {
2231 // The inline constrained primitive (`integer [0..6]`,
2232 // §15.3): in field position the constraint nests inside
2233 // PrimitiveType, unlike the `type`-definition position.
2234 self.start(SyntaxKind::PrimitiveType);
2235 self.bump();
2236 if self.at(SyntaxKind::LBracket) {
2237 let stringish = matches!(kw, SyntaxKind::StringKw | SyntaxKind::BytesKw);
2238 self.constraint(stringish);
2239 }
2240 self.builder.finish_node();
2241 }
2242 Some(SyntaxKind::Ident) => self.path_type(),
2243 Some(SyntaxKind::LParen) => self.tuple_type(),
2244 Some(SyntaxKind::LBracket) => self.array_or_map(checkpoint),
2245 Some(SyntaxKind::Lt) => {
2246 // The stream container `<T>` (ridl reference §12) parses in
2247 // every type position under both profiles; in a `.typl`
2248 // parse it is the profile boundary — TYPL-301 — and parsing
2249 // continues (ADR-0007 decision 10). Under `Profile::Ridl`
2250 // the positions where a stream is not allowed (struct
2251 // fields, signal and event payloads) are checker scope.
2252 if self.profile == Profile::Typl {
2253 self.error_at_current("TYPL-301", "stream type in typl context".to_string());
2254 }
2255 self.stream_type();
2256 }
2257 _ => {
2258 self.error_at_current("FORM-101", "expected a type".to_string());
2259 return;
2260 }
2261 }
2262 // Each `?` wraps one OptionalType node, so an unbounded run would
2263 // build a tree deep enough to overflow the stack when it is later
2264 // traversed or dropped. The wrap count shares the MAX_TYPE_DEPTH
2265 // bound: past it, the remaining `?` tokens land flat in a single
2266 // ErrorNode with one FORM-102.
2267 let mut wraps = 0;
2268 while self.at(SyntaxKind::Question) {
2269 if wraps >= MAX_TYPE_DEPTH {
2270 self.error_at_current(
2271 "FORM-102",
2272 format!("type nesting deeper than {MAX_TYPE_DEPTH} levels"),
2273 );
2274 self.start(SyntaxKind::ErrorNode);
2275 while self.at(SyntaxKind::Question) {
2276 self.bump();
2277 }
2278 self.builder.finish_node();
2279 break;
2280 }
2281 self.builder
2282 .start_node_at(checkpoint, SyntaxKind::OptionalType.into());
2283 self.bump();
2284 self.builder.finish_node();
2285 wraps += 1;
2286 }
2287 }
2288
2289 /// `TupleType = '(' fields:TupleField (',' fields:TupleField)* ')'` —
2290 /// separator commas are direct children of the tuple node.
2291 fn tuple_type(&mut self) {
2292 self.start(SyntaxKind::TupleType);
2293 self.bump(); // '('
2294 loop {
2295 self.eat_trivia();
2296 match self.current() {
2297 None => {
2298 self.error_at_current("FORM-103", "unclosed `(`".to_string());
2299 break;
2300 }
2301 Some(SyntaxKind::RParen) => {
2302 self.bump();
2303 break;
2304 }
2305 Some(SyntaxKind::Comma) => self.bump(),
2306 Some(SyntaxKind::Ident) => self.tuple_field(),
2307 // An unclosed `(`: the tuple ran into a block or top-level
2308 // boundary. Report it and let the enclosing loop recover.
2309 Some(kind) if kind == SyntaxKind::RBrace || is_top_level_start(kind) => {
2310 self.error_at_current("FORM-103", "unclosed `(`".to_string());
2311 break;
2312 }
2313 Some(_) => self.err_and_recover("in a tuple type", |kind| {
2314 matches!(
2315 kind,
2316 SyntaxKind::RParen
2317 | SyntaxKind::Comma
2318 | SyntaxKind::Ident
2319 | SyntaxKind::RBrace
2320 ) || is_top_level_start(kind)
2321 }),
2322 }
2323 }
2324 self.builder.finish_node();
2325 }
2326
2327 /// `TupleField = Name ':' FieldType`
2328 fn tuple_field(&mut self) {
2329 self.start(SyntaxKind::TupleField);
2330 self.name();
2331 self.expect(SyntaxKind::Colon);
2332 self.field_type();
2333 self.builder.finish_node();
2334 }
2335
2336 /// `ArrayType = '[' element:FieldType ';' Bound ']'` or
2337 /// `MapType = '[' key:FieldType ':' value:FieldType ';' Bound ']'` — the
2338 /// node kind is decided by the token after the first type, then wrapped
2339 /// around everything from the checkpoint (the `[`).
2340 fn array_or_map(&mut self, checkpoint: rowan::Checkpoint) {
2341 self.bump(); // '['
2342 self.field_type(); // the element or key type
2343 if self.at(SyntaxKind::Colon) {
2344 self.builder
2345 .start_node_at(checkpoint, SyntaxKind::MapType.into());
2346 self.bump(); // ':'
2347 self.field_type(); // the value type
2348 } else {
2349 self.builder
2350 .start_node_at(checkpoint, SyntaxKind::ArrayType.into());
2351 }
2352 self.expect(SyntaxKind::Semicolon);
2353 self.bound();
2354 self.expect(SyntaxKind::RBracket);
2355 self.builder.finish_node();
2356 }
2357
2358 /// `Constraint = '[' … ']'` — see the module doc for the bound policy.
2359 fn constraint(&mut self, stringish: bool) {
2360 self.start(SyntaxKind::Constraint);
2361 self.bump(); // '['
2362 self.constraint_body(stringish);
2363 self.expect(SyntaxKind::RBracket);
2364 self.builder.finish_node();
2365 }
2366
2367 /// The three constraint shapes behind the brackets: a scalar range with
2368 /// optional step, a length bound with optional match, and a bare match.
2369 fn constraint_body(&mut self, stringish: bool) {
2370 match self.current() {
2371 Some(SyntaxKind::MatchKw) => {
2372 self.bump();
2373 self.literal();
2374 }
2375 Some(SyntaxKind::DotDot) => {
2376 // `[..max]` — an open lower bound is always the scalar shape.
2377 self.bump();
2378 self.literal();
2379 self.step_opt();
2380 }
2381 _ if self.literal_len_at(0) > 0 => self.constraint_led_by_literal(stringish),
2382 _ => self.error_at_current("FORM-101", "expected a constraint".to_string()),
2383 }
2384 }
2385
2386 /// A constraint whose body starts with a literal: decide between the
2387 /// length-`Bound` shape and the scalar-range shape (module doc policy).
2388 fn constraint_led_by_literal(&mut self, stringish: bool) {
2389 let first_len = self.literal_len_at(0);
2390 match self.nth(first_len) {
2391 // A single literal with no `..` is always a length bound:
2392 // `[17]`, `[8]`, `[17 match P]`.
2393 Some(SyntaxKind::RBracket | SyntaxKind::MatchKw) | None => {
2394 self.bound();
2395 self.match_opt();
2396 }
2397 Some(SyntaxKind::DotDot) => {
2398 let second_len = self.literal_len_at(first_len + 1);
2399 let after_second = self.nth(first_len + 1 + second_len);
2400 let bound_shape = stringish
2401 && second_len > 0
2402 && !self.literal_is_float_at(0)
2403 && !self.literal_is_float_at(first_len + 1)
2404 && matches!(
2405 after_second,
2406 Some(SyntaxKind::RBracket | SyntaxKind::MatchKw) | None
2407 );
2408 if bound_shape {
2409 self.bound();
2410 self.match_opt();
2411 } else {
2412 self.literal(); // min
2413 self.bump(); // '..'
2414 if self.literal_len_at(0) > 0 {
2415 self.literal(); // max — absent on `[min..]`
2416 }
2417 self.step_opt();
2418 }
2419 }
2420 _ => {
2421 // Not a legal shape after the literal; parse the literal as a
2422 // length bound and let the closing-bracket check report.
2423 self.bound();
2424 self.match_opt();
2425 }
2426 }
2427 }
2428
2429 /// `Bound = min:Literal ('..' max:Literal)?`
2430 fn bound(&mut self) {
2431 if self.literal_len_at(0) == 0 {
2432 self.error_at_current("FORM-101", "expected a bound".to_string());
2433 return;
2434 }
2435 self.start(SyntaxKind::Bound);
2436 self.literal();
2437 if self.at(SyntaxKind::DotDot) {
2438 self.bump();
2439 self.literal();
2440 }
2441 self.builder.finish_node();
2442 }
2443
2444 /// `('step' step:Literal)?`
2445 fn step_opt(&mut self) {
2446 if self.at(SyntaxKind::StepKw) {
2447 self.bump();
2448 self.literal();
2449 }
2450 }
2451
2452 /// `('match' match_pattern:Literal)?`
2453 fn match_opt(&mut self) {
2454 if self.at(SyntaxKind::MatchKw) {
2455 self.bump();
2456 self.literal();
2457 }
2458 }
2459
2460 /// `InitValue = '=' Literal`, when present.
2461 fn init_value_opt(&mut self) {
2462 if self.at(SyntaxKind::Eq) {
2463 self.start(SyntaxKind::InitValue);
2464 self.bump();
2465 self.literal();
2466 self.builder.finish_node();
2467 }
2468 }
2469
2470 /// How many significant tokens the literal starting at the `n`-th
2471 /// significant position spans: 2 for `-` + value, 1 for a bare value, 0
2472 /// when no literal starts there.
2473 fn literal_len_at(&self, n: usize) -> usize {
2474 match self.nth(n) {
2475 Some(SyntaxKind::Minus) => match self.nth(n + 1) {
2476 Some(kind) if is_value_token(kind) => 2,
2477 _ => 0,
2478 },
2479 Some(kind) if is_value_token(kind) => 1,
2480 _ => 0,
2481 }
2482 }
2483
2484 /// The value token of the literal starting at the `n`-th significant
2485 /// position, looking past a leading `-`. `None` when no literal starts
2486 /// there. Lets a caller admit one literal shape and refuse the rest.
2487 fn literal_value_kind_at(&self, n: usize) -> Option<SyntaxKind> {
2488 match self.nth(n) {
2489 Some(SyntaxKind::Minus) => self.nth(n + 1).filter(|kind| is_value_token(*kind)),
2490 Some(kind) if is_value_token(kind) => Some(kind),
2491 _ => None,
2492 }
2493 }
2494
2495 /// Whether the literal starting at the `n`-th significant position has a
2496 /// float value token.
2497 fn literal_is_float_at(&self, n: usize) -> bool {
2498 match self.nth(n) {
2499 Some(SyntaxKind::FloatNumber) => true,
2500 Some(SyntaxKind::Minus) => matches!(self.nth(n + 1), Some(SyntaxKind::FloatNumber)),
2501 _ => false,
2502 }
2503 }
2504
2505 /// `Literal = '-'? ('int_number' | 'float_number' | 'string_lit' | 'true'
2506 /// | 'false' | 'regex' | 'ident')` — an `ident` is a constant reference
2507 /// in a value position. A duration literal here in a `.typl` parse is the
2508 /// profile boundary (TYPL-302); an integer with leading zeros is FORM-005.
2509 /// Under [`Profile::Ridl`] a duration is an ordinary token that is simply
2510 /// not a literal, so it falls through to the generic missing-value path.
2511 fn literal(&mut self) {
2512 if self.profile == Profile::Typl && self.at(SyntaxKind::Duration) {
2513 self.error_at_current("TYPL-302", "duration literal in typl context".to_string());
2514 self.start(SyntaxKind::ErrorNode);
2515 self.bump();
2516 self.builder.finish_node();
2517 return;
2518 }
2519 if self.literal_len_at(0) == 0 {
2520 self.error_at_current("FORM-101", "expected a value".to_string());
2521 return;
2522 }
2523 self.start(SyntaxKind::Literal);
2524 if self.at(SyntaxKind::Minus) {
2525 self.bump();
2526 }
2527 if self.at(SyntaxKind::IntNumber) {
2528 self.flag_leading_zeros();
2529 }
2530 self.bump();
2531 self.builder.finish_node();
2532 }
2533
2534 /// FORM-005: an integer literal with leading zeros (typl reference §2.4)
2535 /// lexes as one token and is flagged here.
2536 fn flag_leading_zeros(&mut self) {
2537 let i = self.significant_pos();
2538 let text = self.tokens[i].text;
2539 if text.len() > 1 && text.starts_with('0') {
2540 self.error_at_current("FORM-005", "integer literal has leading zeros".to_string());
2541 }
2542 }
2543
2544 /// A declared name — a single `Ident` wrapped in a `Name` node. A family
2545 /// reserved word is held in an `ErrorNode` and flagged FORM-105; otherwise
2546 /// no node is built when the name is missing, so accessors see `None`
2547 /// rather than an empty `Name`.
2548 fn name(&mut self) {
2549 if self.at(SyntaxKind::Ident) {
2550 self.start(SyntaxKind::Name);
2551 self.bump();
2552 self.builder.finish_node();
2553 } else if self.at(SyntaxKind::ReservedWord) {
2554 // A family reserved word (typl reference §1.4) used where a
2555 // declared name is expected: hold it in an `ErrorNode` so the
2556 // rest of the declaration still parses, and flag FORM-105.
2557 let message = format!(
2558 "reserved word `{}` used as identifier",
2559 self.tokens[self.significant_pos()].text,
2560 );
2561 self.error_at_current("FORM-105", message);
2562 self.start(SyntaxKind::ErrorNode);
2563 self.bump();
2564 self.builder.finish_node();
2565 } else {
2566 self.error_at_current("FORM-101", "expected a name".to_string());
2567 }
2568 }
2569
2570 /// Whether the current token can be a path segment. Primitive keywords
2571 /// are admitted as segments so `const MAX_GEAR : integer = 6` (typl
2572 /// reference Appendix B) parses; the checker narrows where a primitive
2573 /// is not allowed (TYPL-204 and the related composite checks).
2574 fn at_path_segment(&self) -> bool {
2575 matches!(
2576 self.current(),
2577 Some(
2578 SyntaxKind::Ident
2579 | SyntaxKind::BooleanKw
2580 | SyntaxKind::IntegerKw
2581 | SyntaxKind::FloatKw
2582 | SyntaxKind::StringKw
2583 | SyntaxKind::BytesKw
2584 )
2585 )
2586 }
2587
2588 /// `PathType = QualifiedName` — a type reference in type position. No
2589 /// node is built when the reference is missing.
2590 fn path_type(&mut self) {
2591 if !self.at_path_segment() {
2592 self.error_at_current("FORM-101", "expected a type name".to_string());
2593 return;
2594 }
2595 self.start(SyntaxKind::PathType);
2596 self.qualified_name();
2597 self.builder.finish_node();
2598 }
2599
2600 /// `QualifiedName = 'ident' ('.' 'ident')*`
2601 fn qualified_name(&mut self) {
2602 if !self.at_path_segment() {
2603 self.error_at_current("FORM-101", "expected a name".to_string());
2604 return;
2605 }
2606 self.start(SyntaxKind::QualifiedName);
2607 self.bump();
2608 while self.at(SyntaxKind::Dot) {
2609 self.bump();
2610 if self.at_path_segment() {
2611 self.bump();
2612 } else {
2613 self.error_at_current("FORM-101", "expected a name after `.`".to_string());
2614 break;
2615 }
2616 }
2617 self.builder.finish_node();
2618 }
2619}
2620
2621#[cfg(test)]
2622mod tests {
2623 use super::*;
2624
2625 const FIXTURE: &str = include_str!("../fixtures/walking_skeleton.typl");
2626
2627 #[test]
2628 fn fixture_round_trips_losslessly() {
2629 let parse = parse(FIXTURE, Profile::Typl);
2630 assert_eq!(
2631 parse.syntax().text().to_string(),
2632 FIXTURE,
2633 "the tree text must reproduce the source byte for byte",
2634 );
2635 assert!(
2636 parse.errors().is_empty(),
2637 "the fixture must parse without errors, got: {:?}",
2638 parse.errors(),
2639 );
2640 }
2641
2642 #[test]
2643 fn mangled_input_round_trips_and_reports_errors() {
2644 let input = "type 123 :: [";
2645 let parse = parse(input, Profile::Typl);
2646 assert_eq!(
2647 parse.syntax().text().to_string(),
2648 input,
2649 "recovery must not drop any text",
2650 );
2651 assert!(
2652 !parse.errors().is_empty(),
2653 "broken input must produce at least one error",
2654 );
2655 }
2656
2657 #[test]
2658 fn parse_equality_is_green_node_identity() {
2659 assert_eq!(
2660 parse(FIXTURE, Profile::Typl),
2661 parse(FIXTURE, Profile::Typl),
2662 "parsing identical input twice must compare equal",
2663 );
2664 assert_ne!(
2665 parse(FIXTURE, Profile::Typl),
2666 parse("package fixtures\ntype X: m", Profile::Typl),
2667 "parsing different input must not compare equal",
2668 );
2669 }
2670
2671 /// The codes of the recorded errors, in source order.
2672 fn error_codes(input: &str) -> Vec<&'static str> {
2673 parse(input, Profile::Typl)
2674 .errors()
2675 .iter()
2676 .map(|e| e.code)
2677 .collect()
2678 }
2679
2680 #[test]
2681 fn missing_package_flags_form_104() {
2682 let codes = error_codes("type Speed: km/h\n");
2683 assert_eq!(codes, vec!["FORM-104"]);
2684 }
2685
2686 #[test]
2687 fn leading_zero_integer_flags_form_005() {
2688 let codes = error_codes("package p\nconst BAD = 042\n");
2689 assert_eq!(codes, vec!["FORM-005"]);
2690 }
2691
2692 #[test]
2693 fn duration_literal_flags_typl_302() {
2694 let parse = parse("package p\nconst BAD = 10ms\n", Profile::Typl);
2695 assert_eq!(
2696 parse.errors().iter().map(|e| e.code).collect::<Vec<_>>(),
2697 vec!["TYPL-302"],
2698 );
2699 assert_eq!(
2700 parse.syntax().text().to_string(),
2701 "package p\nconst BAD = 10ms\n",
2702 "the rejected duration token must stay in the tree",
2703 );
2704 }
2705
2706 #[test]
2707 fn stray_at_sigil_flags_typl_302() {
2708 let codes = error_codes("package p\n@\n");
2709 assert_eq!(codes, vec!["TYPL-302"]);
2710 }
2711
2712 // Under `Profile::Ridl` durations and `@` are ordinary tokens — no TYPL-302
2713 // fires anywhere.
2714 #[test]
2715 fn duration_and_at_draw_no_typl_302_under_ridl() {
2716 let parsed = parse("package p\nconst BAD = 10ms\n", Profile::Ridl);
2717 assert!(
2718 parsed.errors().iter().all(|e| e.code != "TYPL-302"),
2719 "no TYPL-302 under Ridl, got: {:?}",
2720 parsed.errors(),
2721 );
2722 let parsed = parse("package p\n@\n", Profile::Ridl);
2723 assert_eq!(
2724 parsed.errors().iter().map(|e| e.code).collect::<Vec<_>>(),
2725 vec!["FORM-102"],
2726 "a stray `@` under Ridl is a generic unexpected token",
2727 );
2728 }
2729
2730 // An interaction keyword at declaration-start in a `.typl` parse is the
2731 // profile boundary — TYPL-304, recovering like FORM-105 does (ErrorNode,
2732 // resync at the next top-level keyword).
2733 #[test]
2734 fn interaction_keyword_in_typl_flags_typl_304_and_recovers() {
2735 let input = "package p\ninterface X {}\ntype Fine: m\n";
2736 let parsed = parse(input, Profile::Typl);
2737 assert_eq!(
2738 parsed.syntax().text().to_string(),
2739 input,
2740 "TYPL-304 recovery must stay lossless",
2741 );
2742 assert_eq!(
2743 parsed.errors().iter().map(|e| e.code).collect::<Vec<_>>(),
2744 vec!["TYPL-304"],
2745 );
2746 assert_eq!(def_names_in(Profile::Typl, input), vec!["Fine"]);
2747
2748 // The same boundary fires for every interaction kind.
2749 let codes: Vec<_> = parse("package p\nsignal speed: Speed\n", Profile::Typl)
2750 .errors()
2751 .iter()
2752 .map(|e| e.code)
2753 .collect();
2754 assert_eq!(codes, vec!["TYPL-304"]);
2755 }
2756
2757 // The stream grammar parses under both profiles; in a `.typl` parse a `<T>`
2758 // in type position is the profile boundary — TYPL-301 — and the StreamType
2759 // node is still built, losslessly.
2760 #[test]
2761 fn stream_type_in_typl_flags_typl_301_and_keeps_parsing() {
2762 let input = "package p\nstruct S {\n f : <Frame>\n g : integer [0..1]\n}\n";
2763 let parsed = parse(input, Profile::Typl);
2764 assert_eq!(
2765 parsed.syntax().text().to_string(),
2766 input,
2767 "the TYPL-301 boundary must stay lossless",
2768 );
2769 assert_eq!(
2770 parsed.errors().iter().map(|e| e.code).collect::<Vec<_>>(),
2771 vec!["TYPL-301"],
2772 );
2773 let has_stream = parsed
2774 .syntax()
2775 .descendants()
2776 .any(|node| node.kind() == SyntaxKind::StreamType);
2777 assert!(has_stream, "the stream still parses into a StreamType node");
2778
2779 // Under `Profile::Ridl` the same field parses with no error at all —
2780 // stream placement is checker scope there.
2781 let parsed = parse(input, Profile::Ridl);
2782 assert!(
2783 parsed.errors().is_empty(),
2784 "no parser error under Ridl, got: {:?}",
2785 parsed.errors(),
2786 );
2787 }
2788
2789 // A reserved word of another profile at declaration-start in a `.ridl` parse
2790 // is RIDL-403, with the same recovery.
2791 #[test]
2792 fn reserved_word_in_ridl_flags_ridl_403_and_recovers() {
2793 let input = "package p\nmodel X {}\ntype Fine: m\n";
2794 let parsed = parse(input, Profile::Ridl);
2795 assert_eq!(
2796 parsed.syntax().text().to_string(),
2797 input,
2798 "RIDL-403 recovery must stay lossless",
2799 );
2800 assert_eq!(
2801 parsed.errors().iter().map(|e| e.code).collect::<Vec<_>>(),
2802 vec!["RIDL-403"],
2803 );
2804 assert_eq!(def_names_in(Profile::Ridl, input), vec!["Fine"]);
2805 }
2806
2807 // rsdl reference §2, the boundary seen from the other two profiles: an
2808 // rsdl declaration in a `.typl` file draws the generic FORM-102 (the typl
2809 // boundary names ridl words only), and in a `.ridl` file RIDL-403. Both
2810 // recover at the next top-level keyword.
2811 #[test]
2812 fn rsdl_declaration_in_typl_and_ridl_draws_their_existing_codes() {
2813 let input = "package p\ndistribution Adas { Cruise }\ntype Fine: m\n";
2814 for (profile, code) in [(Profile::Typl, "FORM-102"), (Profile::Ridl, "RIDL-403")] {
2815 let parsed = parse(input, profile);
2816 assert_eq!(parsed.syntax().text().to_string(), input);
2817 assert_eq!(
2818 parsed.errors().iter().map(|e| e.code).collect::<Vec<_>>(),
2819 vec![code],
2820 "under {profile:?}",
2821 );
2822 assert_eq!(
2823 def_names_in(profile, input),
2824 vec!["Fine"],
2825 "under {profile:?}"
2826 );
2827 }
2828 }
2829
2830 // rsdl reference §2 and §3.1: under the rsdl profile the five declarations
2831 // parse clean, and a declaration of another profile is RSDL-604 — once per
2832 // declaration, with the rsdl declaration after it still parsed.
2833 #[test]
2834 fn rsdl_profile_parses_its_declarations_and_rejects_the_others() {
2835 let clean = "package p\nsystem Vehicle { Cruise, veh.diag.access }\n";
2836 assert_eq!(parse(clean, Profile::Rsdl).errors(), &[]);
2837
2838 for foreign in [
2839 "type Speed: km/h",
2840 "internal type Speed: km/h",
2841 "struct S { a: integer [0..1] }",
2842 "interface I { signal s: S @10ms }",
2843 "service veh.a.b : I",
2844 ] {
2845 let input = format!("package p\n{foreign}\nsystem Vehicle {{ Cruise }}\n");
2846 let parsed = parse(&input, Profile::Rsdl);
2847 assert_eq!(parsed.syntax().text().to_string(), input);
2848 assert_eq!(
2849 parsed.errors().iter().map(|e| e.code).collect::<Vec<_>>(),
2850 vec!["RSDL-604"],
2851 "`{foreign}`",
2852 );
2853 assert!(
2854 parsed
2855 .syntax()
2856 .children()
2857 .any(|node| node.kind() == SyntaxKind::SystemDef),
2858 "the system after `{foreign}` still parses",
2859 );
2860 }
2861 }
2862
2863 // rsdl reference §5: a backend key `namespace.key` is an rsdl attribute
2864 // key. Under ridl the `.` stays an unexpected token.
2865 #[test]
2866 fn a_backend_key_parses_under_rsdl_only() {
2867 let rsdl = "package p\nsystem V [ linux.cpuset = (2, 3), linux.realtime ] { A }\n";
2868 assert_eq!(parse(rsdl, Profile::Rsdl).errors(), &[]);
2869
2870 let ridl = "package p\ninterface I {\n command c() [ linux.cpuset = 1 ]\n}\n";
2871 assert_eq!(
2872 parse(ridl, Profile::Ridl)
2873 .errors()
2874 .iter()
2875 .map(|e| e.code)
2876 .collect::<Vec<_>>(),
2877 vec!["FORM-102"],
2878 );
2879 }
2880
2881 /// Recovery from a rejected return type stops *before* a sync token — it
2882 /// never consumes the construct that follows.
2883 ///
2884 /// `return_type` reports and then recovers, and `recover` always bumps the
2885 /// current token, so without the `!is_return_sync(current)` guard a
2886 /// `query a() :` written immediately before a `}` eats the interface's own
2887 /// closing brace and the file then reports an unclosed `{`. Each row here
2888 /// is one member of the sync set, and each asserts on what *survived* — a
2889 /// diagnostic-count assertion cannot see over-consumption, because
2890 /// swallowing the `}` produces one extra diagnostic and swallowing an
2891 /// interaction produces none at all.
2892 #[test]
2893 fn a_rejected_return_type_never_consumes_the_construct_after_it() {
2894 for (sync, body, survivor) in [
2895 (
2896 "the next interaction keyword",
2897 " query broken() :
2898 signal speed : Speed @10ms
2899",
2900 SyntaxKind::SignalDef,
2901 ),
2902 (
2903 "a `reserved` tombstone",
2904 " query broken() :
2905 reserved legacyTemp
2906",
2907 SyntaxKind::ReservedEntry,
2908 ),
2909 (
2910 "the body's own closing brace",
2911 " query broken() :
2912",
2913 SyntaxKind::InterfaceDef,
2914 ),
2915 ] {
2916 let input = format!(
2917 "package p
2918interface I {{
2919{body}}}
2920"
2921 );
2922 let parsed = parse(&input, Profile::Ridl);
2923 assert_eq!(
2924 parsed.syntax().text().to_string(),
2925 input,
2926 "{sync}: recovery must stay lossless",
2927 );
2928 assert_eq!(
2929 parsed.errors().iter().map(|e| e.code).collect::<Vec<_>>(),
2930 vec!["FORM-101"],
2931 "{sync}: the broken return type is the only diagnostic — an unclosed `{{` here means the `}}` was consumed",
2932 );
2933 assert!(
2934 parsed
2935 .syntax()
2936 .descendants()
2937 .any(|node| node.kind() == survivor),
2938 "{sync}: the {survivor:?} after the broken return type is gone",
2939 );
2940 }
2941 }
2942
2943 #[test]
2944 fn pathological_type_nesting_does_not_panic_or_drop_text() {
2945 let input = format!("package p\nstruct S {{ f : {} }}\n", "[".repeat(300));
2946 let parse = parse(&input, Profile::Typl);
2947 assert_eq!(
2948 parse.syntax().text().to_string(),
2949 input,
2950 "deep nesting must stay lossless",
2951 );
2952 assert!(!parse.errors().is_empty());
2953 }
2954
2955 #[test]
2956 fn pathological_question_repetition_does_not_panic_or_drop_text() {
2957 // Each `?` wraps one OptionalType; an unbounded run used to build a
2958 // tree deep enough to overflow the stack on traversal or drop.
2959 let input = format!("package p\nstruct S {{ f : T{} }}\n", "?".repeat(30_000));
2960 let parse = parse(&input, Profile::Typl);
2961 assert_eq!(
2962 parse.syntax().text().to_string(),
2963 input,
2964 "the capped optional run must stay lossless",
2965 );
2966 assert_eq!(
2967 parse.errors().iter().map(|e| e.code).collect::<Vec<_>>(),
2968 vec!["FORM-102"],
2969 "the cap must report exactly one error",
2970 );
2971 }
2972
2973 #[test]
2974 fn misplaced_error_modifier_parses_clean() {
2975 // TYPL-212 is checker scope: the grammar admits `error` on every
2976 // definition kind.
2977 let codes = error_codes("package p\nerror type Legacy : integer [0..1]\n");
2978 assert_eq!(codes, Vec::<&str>::new());
2979 }
2980
2981 #[test]
2982 fn error_ranges_cover_the_offending_token() {
2983 // `042` sits at bytes 22..25 of the input.
2984 let parse = parse("package p\nconst BAD = 042\n", Profile::Typl);
2985 let error = &parse.errors()[0];
2986 assert_eq!(error.code, "FORM-005");
2987 assert_eq!(
2988 error.range,
2989 TextRange::new(TextSize::new(22), TextSize::new(25)),
2990 );
2991 }
2992
2993 /// The declared names of every top-level definition, in source order —
2994 /// the recovery proof that real declaration nodes survive garbage.
2995 fn def_names(input: &str) -> Vec<String> {
2996 def_names_in(Profile::Typl, input)
2997 }
2998
2999 /// [`def_names`] under an explicit profile.
3000 fn def_names_in(profile: Profile, input: &str) -> Vec<String> {
3001 use crate::ast::{AstNode, HasName, SourceFile};
3002 let file = SourceFile::cast(parse(input, profile).syntax()).expect("root is a SourceFile");
3003 file.definitions()
3004 .filter_map(|def| Some(def.name()?.ident_token()?.text().to_string()))
3005 .collect()
3006 }
3007
3008 #[test]
3009 fn top_level_garbage_resyncs_to_the_next_declaration() {
3010 // Valid, then garbage, then valid: the run of stray tokens is wrapped
3011 // in one ErrorNode and the declaration after it is a real node again.
3012 let input = "package p\ntype Good: km/h\n] } garbage 9\ntype AlsoGood: integer [0..1]\n";
3013 let parse = parse(input, Profile::Typl);
3014 assert_eq!(
3015 parse.syntax().text().to_string(),
3016 input,
3017 "resync must stay lossless",
3018 );
3019 assert_eq!(def_names(input), vec!["Good", "AlsoGood"]);
3020 assert_eq!(error_codes(input), vec!["FORM-102"]);
3021 }
3022
3023 #[test]
3024 fn unclosed_brace_resyncs_to_the_next_declaration() {
3025 // A struct body that never closes hands the next `type` back to the
3026 // top level (FORM-103) instead of swallowing it.
3027 let input = "package p\nstruct Broken {\n gear: integer\n\ntype After: km/h\n";
3028 let parse = parse(input, Profile::Typl);
3029 assert_eq!(
3030 parse.syntax().text().to_string(),
3031 input,
3032 "unclosed-brace recovery must stay lossless",
3033 );
3034 assert_eq!(def_names(input), vec!["Broken", "After"]);
3035 assert_eq!(error_codes(input), vec!["FORM-103"]);
3036 }
3037
3038 #[test]
3039 fn reserved_word_as_name_flags_form_105_and_keeps_parsing() {
3040 // `signal` is a family reserved word (typl reference §1.4); used as a
3041 // type name it is held in an ErrorNode and the backing still parses.
3042 let input = "package p\ntype signal: integer [0..1]\ntype Fine: integer [0..1]\n";
3043 assert_eq!(error_codes(input), vec!["FORM-105"]);
3044 // Only `Fine` has a real Name; `signal` sits in an ErrorNode.
3045 assert_eq!(def_names(input), vec!["Fine"]);
3046 }
3047
3048 #[test]
3049 fn unclosed_brackets_do_not_flood_diagnostics() {
3050 // 300 unclosed `[` in a field type: before the one-error-per-offset
3051 // rule the deep-recursion unwind piled hundreds of diagnostics on the
3052 // stuck tokens. Recovery now collapses the run to a handful.
3053 let input = format!("package p\nstruct S {{ f : {} }}\n", "[".repeat(300));
3054 let parse = parse(&input, Profile::Typl);
3055 assert_eq!(
3056 parse.syntax().text().to_string(),
3057 input,
3058 "the unclosed-bracket flood must stay lossless",
3059 );
3060 assert!(
3061 parse.errors().len() <= 4,
3062 "expected a bounded diagnostic count, got {}: {:?}",
3063 parse.errors().len(),
3064 parse.errors(),
3065 );
3066 }
3067
3068 #[test]
3069 fn duration_in_constraint_reports_both_form_101_and_typl_302() {
3070 // A duration or `@` inside a constraint fails a positional check on the
3071 // same token first; the profile-boundary TYPL-302 must still fire — it
3072 // is exempt from the one-diagnostic-per-offset suppression, so both
3073 // codes are recorded (matching the honest init-position behavior).
3074 assert_eq!(
3075 error_codes("package p\ntype X: integer [0..10ms]\n"),
3076 vec!["FORM-101", "TYPL-302"],
3077 );
3078 assert_eq!(
3079 error_codes("package p\ntype X: integer [10ms]\n"),
3080 vec!["FORM-101", "TYPL-302"],
3081 );
3082 assert_eq!(
3083 error_codes("package p\ntype X: integer [0..5 @ 3]\n"),
3084 vec!["FORM-101", "TYPL-302"],
3085 );
3086 }
3087
3088 #[test]
3089 fn closed_deep_nesting_does_not_flood_diagnostics() {
3090 // A balanced 200-level nested array, deeper than MAX_TYPE_DEPTH: the
3091 // closed shape once produced over a thousand diagnostics on unwind.
3092 let deep = format!("{}integer{}", "[".repeat(200), ";1]".repeat(200));
3093 let input = format!("package p\nstruct S {{ f : {deep} }}\n");
3094 let parse = parse(&input, Profile::Typl);
3095 assert_eq!(
3096 parse.syntax().text().to_string(),
3097 input,
3098 "the deep-nesting flood must stay lossless",
3099 );
3100 assert!(
3101 parse.errors().len() <= 4,
3102 "expected a bounded diagnostic count, got {}: {:?}",
3103 parse.errors().len(),
3104 parse.errors(),
3105 );
3106 }
3107
3108 /// typl Appendix E: `reserved = "reserved" ( camelCase_id | int_lit )`.
3109 /// The literal shapes outside that production draw FORM-102 in every body
3110 /// that admits a tombstone, under both profiles. Before this rule the
3111 /// entry parsed as an ordinary `Literal`, lowered to a slot with neither a
3112 /// name nor a value, and reported nothing.
3113 #[test]
3114 fn ungrammatical_reserved_literal_flags_form_102_in_every_body() {
3115 let bodies = [
3116 (
3117 Profile::Typl,
3118 "struct S {\n a : X\n reserved {}\n b : X\n}",
3119 ),
3120 (
3121 Profile::Typl,
3122 "enum E {\n A = 1\n reserved {}\n B = 2\n}",
3123 ),
3124 (
3125 Profile::Typl,
3126 "union U {\n a : X\n reserved {}\n b : X\n}",
3127 ),
3128 (
3129 Profile::Ridl,
3130 "interface I {\n signal a : X\n reserved {}\n signal b : X\n}",
3131 ),
3132 (
3133 Profile::Ridl,
3134 "service p.svc {\n signal a : X\n reserved {}\n signal b : X\n}",
3135 ),
3136 ];
3137 for (profile, shape) in bodies {
3138 for literal in ["\"oldName\"", "true", "false", "1.5", "-1.5", "-MAX"] {
3139 let input = format!("package p\n{}\n", shape.replace("{}", literal));
3140 let parsed = parse(&input, profile);
3141 assert_eq!(
3142 parsed.syntax().text().to_string(),
3143 input,
3144 "refusing `reserved {literal}` must stay lossless in {shape}",
3145 );
3146 assert_eq!(
3147 parsed.errors().iter().map(|e| e.code).collect::<Vec<_>>(),
3148 vec!["FORM-102"],
3149 "`reserved {literal}` in {shape} must draw exactly one FORM-102",
3150 );
3151 assert!(
3152 parsed.errors()[0].message.starts_with(
3153 "a `reserved` tombstone takes a retired member name or a retired \
3154 integer value, not "
3155 ),
3156 "the message must lead with what a tombstone does take, got {:?}",
3157 parsed.errors()[0].message,
3158 );
3159 // The entry keeps its place as a member and the refused token
3160 // is held losslessly, so the member after it is still parsed
3161 // in position and no ordinal shifts.
3162 let entry = parsed
3163 .syntax()
3164 .descendants()
3165 .find(|node| node.kind() == SyntaxKind::ReservedEntry)
3166 .expect("the tombstone still forms a ReservedEntry");
3167 assert!(
3168 entry
3169 .children()
3170 .any(|child| child.kind() == SyntaxKind::ErrorNode),
3171 "the refused literal is held in an ErrorNode, got {entry:#?}",
3172 );
3173 assert!(
3174 entry
3175 .children()
3176 .all(|child| child.kind() != SyntaxKind::Literal),
3177 "the refused literal must not form a Literal, got {entry:#?}",
3178 );
3179 }
3180 }
3181 }
3182
3183 /// The two grammatical tombstone forms keep parsing, in every body and
3184 /// under both profiles: a name, a bare integer, and a negative integer
3185 /// (`int_lit = "-"? [0-9]+`). Narrowing the production must not narrow it
3186 /// past what Appendix E writes.
3187 #[test]
3188 fn grammatical_reserved_forms_still_parse_in_every_body() {
3189 let bodies = [
3190 (
3191 Profile::Typl,
3192 "struct S {\n a : X\n reserved {}\n b : X\n}",
3193 ),
3194 (
3195 Profile::Typl,
3196 "enum E {\n A = 1\n reserved {}\n B = 2\n}",
3197 ),
3198 (
3199 Profile::Typl,
3200 "union U {\n a : X\n reserved {}\n b : X\n}",
3201 ),
3202 (
3203 Profile::Ridl,
3204 "interface I {\n signal a : X\n reserved {}\n signal b : X\n}",
3205 ),
3206 (
3207 Profile::Ridl,
3208 "service p.svc {\n signal a : X\n reserved {}\n signal b : X\n}",
3209 ),
3210 ];
3211 for (profile, shape) in bodies {
3212 for literal in ["legacyChecksum", "3", "-3"] {
3213 let input = format!("package p\n{}\n", shape.replace("{}", literal));
3214 let parsed = parse(&input, profile);
3215 assert!(
3216 parsed.errors().is_empty(),
3217 "`reserved {literal}` in {shape} must parse clean, got {:?}",
3218 parsed.errors(),
3219 );
3220 let entry = parsed
3221 .syntax()
3222 .descendants()
3223 .find(|node| node.kind() == SyntaxKind::ReservedEntry)
3224 .expect("the tombstone forms a ReservedEntry");
3225 let kept = entry
3226 .children()
3227 .any(|child| matches!(child.kind(), SyntaxKind::Name | SyntaxKind::Literal));
3228 assert!(
3229 kept,
3230 "`reserved {literal}` must keep the retired identity, got {entry:#?}",
3231 );
3232 }
3233 }
3234 }
3235
3236 /// ADR-0015 decision 13: commas are required between service shapes,
3237 /// because the list has no closing token — it ends where the next
3238 /// declaration begins. A missing comma therefore ends the list, and the
3239 /// following bare `CamelCase` identifier is reported at the top level,
3240 /// **at the identifier**: the greedy alternative would absorb a mistyped
3241 /// `Struct Foo {` as another shape and surface the error at the `{` with
3242 /// no connection to the mistake.
3243 #[test]
3244 fn a_missing_comma_ends_the_shape_list_at_the_next_identifier() {
3245 let input = "package p\nservice p.svc : DoorControl\nStruct Foo {\n a : X\n}\n";
3246 let parsed = parse(input, Profile::Ridl);
3247 assert_eq!(
3248 parsed.syntax().text().to_string(),
3249 input,
3250 "the ended list must stay lossless",
3251 );
3252 // The ServiceDef holds exactly the one shape before the missing comma.
3253 let service = parsed
3254 .syntax()
3255 .descendants()
3256 .find(|node| node.kind() == SyntaxKind::ServiceDef)
3257 .expect("the service parses");
3258 assert_eq!(
3259 service
3260 .children()
3261 .filter(|child| child.kind() == SyntaxKind::PathType)
3262 .count(),
3263 1,
3264 "`Struct` must not be absorbed as a second shape, got {service:#?}",
3265 );
3266 // The error points at `Struct` — the identifier itself, where the
3267 // mistake is — never at the `{` after it.
3268 let struct_offset = input.find("Struct").expect("the fixture spells it") as u32;
3269 let error = &parsed.errors()[0];
3270 assert_eq!(
3271 u32::from(error.range.start()),
3272 struct_offset,
3273 "the diagnostic lands on the identifier, got: {:?}",
3274 parsed.errors(),
3275 );
3276 }
3277
3278 /// The trailing comma stays optional (ADR-0015 decision 13), and every
3279 /// shape is a direct `PathType` child of the ServiceDef. `reserved` is not
3280 /// a shape (lock design §9): the list ends at the comma before it, the
3281 /// stray keyword is reported, and nothing after it joins the list.
3282 #[test]
3283 fn a_shape_list_admits_a_trailing_comma_and_ends_before_reserved() {
3284 let input = "package p\nservice p.svc : DoorControl, MotorControl,\n";
3285 let parsed = parse(input, Profile::Ridl);
3286 assert!(
3287 parsed.errors().is_empty(),
3288 "the list parses clean, got: {:?}",
3289 parsed.errors(),
3290 );
3291 assert_eq!(
3292 shape_kinds(&parsed),
3293 [SyntaxKind::PathType, SyntaxKind::PathType]
3294 );
3295
3296 let input = "package p\nservice p.svc : DoorControl, reserved LegacyDiag, MotorControl,\n";
3297 let parsed = parse(input, Profile::Ridl);
3298 assert!(
3299 !parsed.errors().is_empty(),
3300 "`reserved` in a service's list is a parse error",
3301 );
3302 assert_eq!(
3303 shape_kinds(&parsed),
3304 [SyntaxKind::PathType],
3305 "the list ends before the stray keyword",
3306 );
3307 }
3308
3309 /// The kinds of the shape children of the first ServiceDef in `parsed`.
3310 fn shape_kinds(parsed: &Parse) -> Vec<SyntaxKind> {
3311 parsed
3312 .syntax()
3313 .descendants()
3314 .find(|node| node.kind() == SyntaxKind::ServiceDef)
3315 .expect("the service parses")
3316 .children()
3317 .filter(|child| {
3318 matches!(
3319 child.kind(),
3320 SyntaxKind::PathType | SyntaxKind::ReservedEntry
3321 )
3322 })
3323 .map(|child| child.kind())
3324 .collect()
3325 }
3326
3327 /// Nothing at all after `reserved` is a missing token, not a wrong one:
3328 /// FORM-101, naming both admitted forms.
3329 #[test]
3330 fn reserved_with_no_target_flags_form_101() {
3331 let parsed = parse("package p\nstruct S {\n reserved\n}\n", Profile::Typl);
3332 assert_eq!(
3333 parsed.errors().iter().map(|e| e.code).collect::<Vec<_>>(),
3334 vec!["FORM-101"],
3335 );
3336 assert_eq!(
3337 parsed.errors()[0].message,
3338 "expected a name or an integer value",
3339 );
3340 }
3341
3342 // --- operator chains (driftsys/ridl#346) --------------------------------
3343
3344 /// A contract expression in a minimal ridl file.
3345 fn contract_file(expr: &str) -> String {
3346 format!(
3347 "package app\ntype N : integer [0..10]\ninterface I {{\n command c(a: N) [ require {expr} ] @[..50ms]\n}}\n"
3348 )
3349 }
3350
3351 /// A flat chain of `terms` operands under `op`, closed into a boolean
3352 /// where the operator is not one already.
3353 fn flat_chain(op: &str, terms: usize) -> String {
3354 match op {
3355 "||" | "&&" => vec!["a == 1"; terms].join(&format!(" {op} ")),
3356 "." => format!("a{} == 1", ".b".repeat(terms - 1)),
3357 _ => format!("{} == 1", vec!["a"; terms].join(&format!(" {op} "))),
3358 }
3359 }
3360
3361 /// The deepest node below `root`, counted without recursion, so that
3362 /// measuring a tree cannot itself exhaust the stack.
3363 fn tree_depth(root: &SyntaxNode) -> usize {
3364 let (mut depth, mut deepest) = (0usize, 0usize);
3365 for event in root.preorder() {
3366 match event {
3367 rowan::WalkEvent::Enter(_) => {
3368 depth += 1;
3369 deepest = deepest.max(depth);
3370 }
3371 rowan::WalkEvent::Leave(_) => depth -= 1,
3372 }
3373 }
3374 deepest
3375 }
3376
3377 /// Runs `body` on a thread with a 64 MiB stack. Without the fix a chain
3378 /// builds a tree thousands of levels deep, and dropping it recurses; the
3379 /// large stack keeps that from aborting the whole test binary, so the
3380 /// test fails on its assertions instead.
3381 fn on_a_large_stack(body: impl FnOnce() + Send + 'static) {
3382 std::thread::Builder::new()
3383 .stack_size(64 << 20)
3384 .spawn(body)
3385 .expect("spawn the test thread")
3386 .join()
3387 .expect("the test thread must not panic");
3388 }
3389
3390 /// The bound every expression tree must stay within, measured from the
3391 /// file root. A checker pass that recurses over the tree is sized against
3392 /// it; a flat chain once built a tree as deep as its term count.
3393 const TREE_DEPTH_BOUND: usize = 2 * MAX_TYPE_DEPTH;
3394
3395 #[test]
3396 fn a_flat_operator_chain_past_the_depth_limit_draws_one_form_102() {
3397 on_a_large_stack(|| {
3398 for op in ["||", "&&", "+", "-", "*", "/", "%", "."] {
3399 let input = contract_file(&flat_chain(op, 1000));
3400 let parsed = parse(&input, Profile::Ridl);
3401 let form_102 = parsed
3402 .errors()
3403 .iter()
3404 .filter(|error| error.code == "FORM-102")
3405 .count();
3406 assert_eq!(
3407 form_102,
3408 1,
3409 "a 1000-term `{op}` chain draws exactly one FORM-102: {:?}",
3410 parsed.errors(),
3411 );
3412 assert_eq!(
3413 parsed.syntax().text().to_string(),
3414 input,
3415 "the refused `{op}` chain must stay lossless",
3416 );
3417 let depth = tree_depth(&parsed.syntax());
3418 assert!(
3419 depth <= TREE_DEPTH_BOUND,
3420 "a 1000-term `{op}` chain builds a tree {depth} levels deep",
3421 );
3422 }
3423 });
3424 }
3425
3426 /// The height of the expression tree under `root`: the most expression
3427 /// nodes on one path from the root to a leaf, counted without recursion.
3428 /// Only the expression kinds count, so the declaration nodes above the
3429 /// attribute and the flat `ErrorNode` of a refusal do not.
3430 fn expr_height(root: &SyntaxNode) -> usize {
3431 let is_expr = |kind: SyntaxKind| {
3432 matches!(
3433 kind,
3434 SyntaxKind::BinaryExpr
3435 | SyntaxKind::PrefixExpr
3436 | SyntaxKind::MemberExpr
3437 | SyntaxKind::PathExpr
3438 | SyntaxKind::ParenExpr
3439 | SyntaxKind::LiteralExpr
3440 )
3441 };
3442 let (mut depth, mut deepest) = (0usize, 0usize);
3443 for event in root.preorder() {
3444 match event {
3445 rowan::WalkEvent::Enter(node) if is_expr(node.kind()) => {
3446 depth += 1;
3447 deepest = deepest.max(depth);
3448 }
3449 rowan::WalkEvent::Leave(node) if is_expr(node.kind()) => depth -= 1,
3450 _ => {}
3451 }
3452 }
3453 deepest
3454 }
3455
3456 /// The bound as an author meets it, over every shape that raises the
3457 /// height, at every size around the limit and one far past it: the tree
3458 /// built is never higher than `MAX_TYPE_DEPTH`, and FORM-102 is drawn
3459 /// exactly when the complete tree would not fit. Each generator states
3460 /// the height its complete tree has at `n`; where that fits, the built
3461 /// tree must have exactly that height, which checks the formula. The
3462 /// right-leaning shapes are the ones a bound that forgets to parse a
3463 /// right operand one level deeper would let through; the group and
3464 /// prefix towers are the ones a bound that opens a group or a prefix
3465 /// without room for a leaf below it would let through.
3466 #[test]
3467 fn the_expression_height_bound_holds_for_every_shape() {
3468 type Shape = (&'static str, fn(usize) -> String, fn(usize) -> usize);
3469 let shapes: [Shape; 15] = [
3470 ("left chain", |n| vec!["a"; n].join(" + "), |n| n),
3471 (
3472 "`||` chain of comparisons",
3473 |n| vec!["a == 1"; n].join(" || "),
3474 |n| n + 1,
3475 ),
3476 (
3477 "member steps",
3478 |n| format!("a{}", ".b".repeat(n - 1)),
3479 |n| n,
3480 ),
3481 (
3482 "groups around a leaf",
3483 |n| format!("{}a{}", "(".repeat(n - 1), ")".repeat(n - 1)),
3484 |n| n,
3485 ),
3486 (
3487 "groups around a negated leaf",
3488 |n| format!("{}-a{}", "(".repeat(n - 1), ")".repeat(n - 1)),
3489 |n| n + 1,
3490 ),
3491 (
3492 "groups around a `!` leaf",
3493 |n| format!("{}!a{}", "(".repeat(n - 1), ")".repeat(n - 1)),
3494 |n| n + 1,
3495 ),
3496 (
3497 "right-nested chain",
3498 |n| {
3499 let mut expr = "a".to_string();
3500 for _ in 1..n {
3501 expr = format!("a + ({expr})");
3502 }
3503 expr
3504 },
3505 |n| 2 * n - 1,
3506 ),
3507 (
3508 "comparison with a grouped chain on the right",
3509 |n| format!("a == ({})", vec!["a"; n].join(" + ")),
3510 |n| n + 2,
3511 ),
3512 (
3513 "`!(` tower",
3514 |n| format!("{}a{}", "!(".repeat(n - 1), ")".repeat(n - 1)),
3515 |n| 2 * n - 1,
3516 ),
3517 (
3518 "`-(` tower",
3519 |n| format!("{}a{}", "-(".repeat(n - 1), ")".repeat(n - 1)),
3520 |n| 2 * n - 1,
3521 ),
3522 (
3523 "prefixed operands in a chain",
3524 |n| vec!["-a"; n].join(" * "),
3525 |n| n + 1,
3526 ),
3527 (
3528 "grouped member steps as a right operand",
3529 |n| format!("a + (a{})", ".b".repeat(n - 1)),
3530 |n| n + 2,
3531 ),
3532 (
3533 "`||` right-nested",
3534 |n| {
3535 let mut expr = "a == 1".to_string();
3536 for _ in 1..n {
3537 expr = format!("a == 1 || ({expr})");
3538 }
3539 expr
3540 },
3541 |n| 2 * n,
3542 ),
3543 (
3544 "a deep right operand, then a long left chain",
3545 |n| {
3546 format!(
3547 "a + ({}a{}){}",
3548 "(".repeat(60),
3549 ")".repeat(60),
3550 " + a".repeat(n - 1)
3551 )
3552 },
3553 // The group written around the 60 counts too: 61 groups, a
3554 // leaf, and the wrap, then one level per further term.
3555 |n| n + 62,
3556 ),
3557 (
3558 "a prefix tower, then a chain",
3559 |n| {
3560 format!(
3561 "{}a == 1{}{}",
3562 "!(".repeat(30),
3563 ")".repeat(30),
3564 " && a == 1".repeat(n - 1)
3565 )
3566 },
3567 |n| n + 61,
3568 ),
3569 ];
3570 for (shape, build, complete_height) in shapes {
3571 // The smallest sizes check the formula; the sizes whose complete
3572 // height is within eight levels of the limit check the bound on
3573 // both sides of it; 300 is far past it.
3574 let mut sizes: Vec<usize> = (1..)
3575 .take_while(|&n| complete_height(n) <= MAX_TYPE_DEPTH + 8)
3576 .filter(|&n| n <= 3 || complete_height(n) + 8 >= MAX_TYPE_DEPTH)
3577 .collect();
3578 sizes.push(300);
3579 for n in sizes {
3580 let input = contract_file(&build(n));
3581 let parsed = parse(&input, Profile::Ridl);
3582 let complete = complete_height(n);
3583 let built = expr_height(&parsed.syntax());
3584 let form_102 = parsed
3585 .errors()
3586 .iter()
3587 .filter(|error| error.code == "FORM-102")
3588 .count();
3589 assert!(
3590 built <= MAX_TYPE_DEPTH,
3591 "the {shape} at n = {n} builds a tree {built} high",
3592 );
3593 assert_eq!(
3594 parsed.syntax().text().to_string(),
3595 input,
3596 "{shape} at n = {n}"
3597 );
3598 if complete <= MAX_TYPE_DEPTH {
3599 assert!(
3600 parsed.errors().is_empty(),
3601 "the {shape} at n = {n} fits in {complete} levels: {:?}",
3602 parsed.errors(),
3603 );
3604 assert_eq!(
3605 built, complete,
3606 "the {shape} at n = {n} builds its complete tree"
3607 );
3608 } else {
3609 assert_eq!(
3610 form_102,
3611 1,
3612 "the {shape} at n = {n} would be {complete} high: {:?}",
3613 parsed.errors(),
3614 );
3615 }
3616 }
3617 }
3618 }
3619
3620 /// Past the limit the diagnostics are exactly one FORM-102: the refused
3621 /// group is consumed through its own `)` together with the groups inside
3622 /// it, the chain after the outermost group goes flat without a second
3623 /// diagnostic, and the next interaction parses as usual.
3624 #[test]
3625 fn a_refused_group_draws_only_form_102_and_the_next_interaction_parses() {
3626 use crate::ast::AstNode;
3627 let groups = MAX_TYPE_DEPTH + 2;
3628 let input = format!(
3629 "package app\ntype N : integer [0..10]\ninterface I {{\n \
3630 command c(a: N) [ require {}(a) + (a){} + a + a ] @[..50ms]\n \
3631 command d(a: N) [ require a == 1 ] @[..50ms]\n}}\n",
3632 "(".repeat(groups),
3633 ")".repeat(groups),
3634 );
3635 let parsed = parse(&input, Profile::Ridl);
3636 let codes: Vec<&str> = parsed.errors().iter().map(|error| error.code).collect();
3637 assert_eq!(codes, vec!["FORM-102"]);
3638 assert_eq!(parsed.syntax().text().to_string(), input);
3639 let root = parsed.syntax();
3640 assert!(expr_height(&root) <= MAX_TYPE_DEPTH);
3641 assert_eq!(
3642 root.descendants()
3643 .filter(|node| node.kind() == SyntaxKind::CommandDef)
3644 .count(),
3645 2,
3646 );
3647 let last = root
3648 .descendants()
3649 .filter_map(crate::ast::Attribute::cast)
3650 .last()
3651 .expect("the second command's attribute");
3652 assert_eq!(
3653 last.expr().expect("its expression").syntax().text(),
3654 "a == 1",
3655 );
3656 }
3657
3658 /// A chain whose left operand is missing still keeps the bound: the
3659 /// wrap holds only its right operand, and counting the missing side as
3660 /// zero high would let the leaf below the deepest wrap sit one level
3661 /// past the limit. The shapes put the wrap at the deepest group the
3662 /// bound admits.
3663 #[test]
3664 fn a_missing_left_operand_keeps_the_bound() {
3665 for inner in ["+ a", "== a", "|| a == 1", "+ a == a"] {
3666 let groups = MAX_TYPE_DEPTH - 1;
3667 let input = contract_file(&format!(
3668 "{}{inner}{}",
3669 "(".repeat(groups),
3670 ")".repeat(groups)
3671 ));
3672 let parsed = parse(&input, Profile::Ridl);
3673 let built = expr_height(&parsed.syntax());
3674 assert!(
3675 built <= MAX_TYPE_DEPTH,
3676 "`{inner}` under {groups} groups builds a tree {built} high: {:?}",
3677 parsed.errors(),
3678 );
3679 assert!(
3680 parsed.errors().iter().any(|error| error.code == "FORM-101"),
3681 "the missing operand is reported: {:?}",
3682 parsed.errors(),
3683 );
3684 assert_eq!(parsed.syntax().text().to_string(), input);
3685 }
3686 }
3687
3688 /// A refused expression ends where the expression would have ended: the
3689 /// next attribute in the block — after a comma, a key on its own line, or
3690 /// an `ensure` on the same line with no comma — and the later
3691 /// declarations are parsed as usual, including a syntax error in one of
3692 /// them. Each overlong expression draws its own FORM-102. The stop at a
3693 /// `require`, a reserved word and a key after each kind of operand is
3694 /// pinned by the test after this one.
3695 #[test]
3696 fn a_refused_expression_does_not_swallow_what_follows_it() {
3697 use crate::ast::AstNode;
3698 let long = flat_chain("+", 200);
3699 let input = format!(
3700 "package app\ntype N : integer [0..10]\ninterface I {{\n \
3701 command c(a: N) [ require {long}, require a == 1 ] @[..50ms]\n \
3702 command d(a: N) [ require {long}\n key = 1\n require a == 1 ] @[..50ms]\n \
3703 command e(a: N) [ require a == ] @[..50ms]\n \
3704 command f(a: N) [ require {long} ensure a == 1 ] @[..50ms]\n}}\n"
3705 );
3706 let parsed = parse(&input, Profile::Ridl);
3707 assert_eq!(parsed.syntax().text().to_string(), input);
3708 let codes: Vec<&str> = parsed.errors().iter().map(|error| error.code).collect();
3709 assert_eq!(
3710 codes.iter().filter(|code| **code == "FORM-102").count(),
3711 3,
3712 "one FORM-102 per overlong expression: {codes:?}",
3713 );
3714 assert!(
3715 codes.contains(&"FORM-101"),
3716 "the syntax error in the later declaration is still reported: {codes:?}",
3717 );
3718 let spans: Vec<_> = parsed
3719 .errors()
3720 .iter()
3721 .filter(|error| error.code == "FORM-102")
3722 .map(|error| error.range.start())
3723 .collect();
3724 assert!(
3725 spans.windows(2).all(|pair| pair[0] < pair[1]),
3726 "each FORM-102 sits at its own expression: {spans:?}",
3727 );
3728 let root = parsed.syntax();
3729 assert_eq!(
3730 root.descendants()
3731 .filter(|node| node.kind() == SyntaxKind::CommandDef)
3732 .count(),
3733 4,
3734 );
3735 let attributes: Vec<crate::ast::Attribute> = root
3736 .descendants()
3737 .filter_map(crate::ast::Attribute::cast)
3738 .collect();
3739 assert_eq!(attributes.len(), 8, "{attributes:?}");
3740 assert_eq!(
3741 attributes
3742 .iter()
3743 .filter(|attribute| attribute
3744 .expr()
3745 .is_some_and(|expr| expr.syntax().text() == "a == 1"))
3746 .count(),
3747 3,
3748 "the `a == 1` after each refused chain is its own attribute",
3749 );
3750 assert!(
3751 attributes.iter().any(|attribute| attribute
3752 .syntax()
3753 .text()
3754 .to_string()
3755 .starts_with("key")),
3756 "the key after the refused chain is its own attribute",
3757 );
3758 }
3759
3760 /// Each way the skipped run of a refused chain can end before the next
3761 /// attribute, with no comma: a `require` directly after the chain, a
3762 /// reserved word in key position (which still draws its own FORM-105),
3763 /// and a key after a chain that ends in a number, a name or a group.
3764 #[test]
3765 fn a_refused_chain_stops_at_every_kind_of_next_attribute() {
3766 use crate::ast::AstNode;
3767 let numbers = flat_chain("+", 200);
3768 let names = vec!["a"; 200].join(" + ");
3769 let groups = vec!["(a)"; 200].join(" + ");
3770 let cases = [
3771 ("a `require`", format!("{numbers}\n require a == 1")),
3772 ("a reserved word", format!("{numbers}\n states = 1")),
3773 ("a key after a number", format!("{numbers}\n key = 1")),
3774 ("a key after a name", format!("{names}\n key = 1")),
3775 ("a key after a group", format!("{groups}\n key = 1")),
3776 ];
3777 for (case, body) in cases {
3778 let input = format!(
3779 "package app\ntype N : integer [0..10]\ninterface I {{\n \
3780 command c(a: N) [ require {body} ] @[..50ms]\n}}\n"
3781 );
3782 let parsed = parse(&input, Profile::Ridl);
3783 assert_eq!(parsed.syntax().text().to_string(), input, "{case}");
3784 let codes: Vec<&str> = parsed.errors().iter().map(|error| error.code).collect();
3785 let expected: &[&str] = if case == "a reserved word" {
3786 &["FORM-102", "FORM-105"]
3787 } else {
3788 &["FORM-102"]
3789 };
3790 assert_eq!(codes, expected, "{case}");
3791 let attributes: Vec<String> = parsed
3792 .syntax()
3793 .descendants()
3794 .filter_map(crate::ast::Attribute::cast)
3795 .map(|attribute| attribute.syntax().text().to_string())
3796 .collect();
3797 assert_eq!(
3798 attributes.len(),
3799 2,
3800 "{case}: the next attribute is its own node: {attributes:?}",
3801 );
3802 let next = attributes[1].trim();
3803 let starts = match case {
3804 "a `require`" => "require a == 1",
3805 "a reserved word" => "states = 1",
3806 _ => "key = 1",
3807 };
3808 assert_eq!(next, starts, "{case}");
3809 }
3810 }
3811
3812 #[test]
3813 fn a_long_chain_under_the_depth_limit_parses_clean() {
3814 for op in ["||", "&&", "+", "-", "*", "/", "%", "."] {
3815 let input = contract_file(&flat_chain(op, 100));
3816 let parsed = parse(&input, Profile::Ridl);
3817 assert!(
3818 parsed.errors().is_empty(),
3819 "a 100-term `{op}` chain is accepted: {:?}",
3820 parsed.errors(),
3821 );
3822 }
3823 }
3824
3825 #[test]
3826 fn chains_stacked_through_parentheses_and_precedence_stay_bounded() {
3827 on_a_large_stack(|| {
3828 // Each parenthesis level closes a chain of its own, so a bound
3829 // that is charged while a chain is parsed and released when it
3830 // returns lets the chains stack on the left spine.
3831 let mut nested = "a".to_string();
3832 for _ in 0..120 {
3833 nested = format!("({nested}{})", " + a".repeat(119));
3834 }
3835 // One chain per precedence level, each the first operand of the
3836 // next: member, multiplicative, additive, comparison, `&&`, `||`.
3837 let ladder = format!(
3838 "a{}{}{} == a{}{}",
3839 ".b".repeat(119),
3840 " * a".repeat(119),
3841 " + a".repeat(119),
3842 " && a == a".repeat(119),
3843 " || a == a".repeat(119),
3844 );
3845 for (shape, expr) in [("nested", format!("{nested} == 1")), ("ladder", ladder)] {
3846 let input = contract_file(&expr);
3847 let parsed = parse(&input, Profile::Ridl);
3848 assert!(
3849 parsed.errors().iter().any(|error| error.code == "FORM-102"),
3850 "the {shape} shape draws FORM-102: {:?}",
3851 parsed.errors(),
3852 );
3853 assert!(
3854 parsed.errors().len() <= 4,
3855 "the {shape} shape floods {} diagnostics",
3856 parsed.errors().len(),
3857 );
3858 assert_eq!(parsed.syntax().text().to_string(), input);
3859 let depth = tree_depth(&parsed.syntax());
3860 assert!(
3861 depth <= TREE_DEPTH_BOUND,
3862 "the {shape} shape builds a tree {depth} levels deep",
3863 );
3864 }
3865 });
3866 }
3867}