aboutsummaryrefslogtreecommitdiff
path: root/xtask/src/codegen
diff options
context:
space:
mode:
Diffstat (limited to 'xtask/src/codegen')
-rw-r--r--xtask/src/codegen/gen_syntax.rs224
-rw-r--r--xtask/src/codegen/rust.ungram529
2 files changed, 749 insertions, 4 deletions
diff --git a/xtask/src/codegen/gen_syntax.rs b/xtask/src/codegen/gen_syntax.rs
index 5a18b3e2b..24e8be1fb 100644
--- a/xtask/src/codegen/gen_syntax.rs
+++ b/xtask/src/codegen/gen_syntax.rs
@@ -3,19 +3,27 @@
3//! Specifically, it generates the `SyntaxKind` enum and a number of newtype 3//! Specifically, it generates the `SyntaxKind` enum and a number of newtype
4//! wrappers around `SyntaxNode` which implement `ra_syntax::AstNode`. 4//! wrappers around `SyntaxNode` which implement `ra_syntax::AstNode`.
5 5
6use std::{collections::HashSet, fmt::Write}; 6use std::{
7 collections::{BTreeSet, HashSet},
8 fmt::Write,
9};
7 10
8use proc_macro2::{Punct, Spacing}; 11use proc_macro2::{Punct, Spacing};
9use quote::{format_ident, quote}; 12use quote::{format_ident, quote};
13use ungrammar::{Grammar, Rule};
10 14
11use crate::{ 15use crate::{
12 ast_src::{rust_ast, AstSrc, Field, FieldSrc, KindsSrc, KINDS_SRC}, 16 ast_src::{AstEnumSrc, AstNodeSrc, AstSrc, Field, FieldSrc, KindsSrc, KINDS_SRC},
13 codegen::{self, update, Mode}, 17 codegen::{self, update, Mode},
14 project_root, Result, 18 project_root, Result,
15}; 19};
16 20
17pub fn generate_syntax(mode: Mode) -> Result<()> { 21pub fn generate_syntax(mode: Mode) -> Result<()> {
18 let ast = rust_ast(); 22 let grammar = include_str!("rust.ungram")
23 .parse::<Grammar>()
24 .unwrap_or_else(|err| panic!("\n \x1b[91merror\x1b[0m: {}\n", err));
25 let ast = lower(&grammar);
26
19 let syntax_kinds_file = project_root().join(codegen::SYNTAX_KINDS); 27 let syntax_kinds_file = project_root().join(codegen::SYNTAX_KINDS);
20 let syntax_kinds = generate_syntax_kinds(KINDS_SRC)?; 28 let syntax_kinds = generate_syntax_kinds(KINDS_SRC)?;
21 update(syntax_kinds_file.as_path(), &syntax_kinds, mode)?; 29 update(syntax_kinds_file.as_path(), &syntax_kinds, mode)?;
@@ -215,7 +223,9 @@ fn generate_nodes(kinds: KindsSrc<'_>, grammar: &AstSrc) -> Result<String> {
215 .map(|kind| to_pascal_case(kind)) 223 .map(|kind| to_pascal_case(kind))
216 .filter(|name| !defined_nodes.iter().any(|&it| it == name)) 224 .filter(|name| !defined_nodes.iter().any(|&it| it == name))
217 { 225 {
218 eprintln!("Warning: node {} not defined in ast source", node); 226 drop(node)
227 // TODO: restore this
228 // eprintln!("Warning: node {} not defined in ast source", node);
219 } 229 }
220 230
221 let ast = quote! { 231 let ast = quote! {
@@ -414,6 +424,10 @@ fn to_pascal_case(s: &str) -> String {
414 buf 424 buf
415} 425}
416 426
427fn pluralize(s: &str) -> String {
428 format!("{}s", s)
429}
430
417impl Field { 431impl Field {
418 fn is_many(&self) -> bool { 432 fn is_many(&self) -> bool {
419 matches!(self, Field::Node { src: FieldSrc::Many(_), .. }) 433 matches!(self, Field::Node { src: FieldSrc::Many(_), .. })
@@ -449,6 +463,7 @@ impl Field {
449 "." => "dot", 463 "." => "dot",
450 ".." => "dotdot", 464 ".." => "dotdot",
451 "..." => "dotdotdot", 465 "..." => "dotdotdot",
466 "..=" => "dotdoteq",
452 "=>" => "fat_arrow", 467 "=>" => "fat_arrow",
453 "@" => "at", 468 "@" => "at",
454 ":" => "colon", 469 ":" => "colon",
@@ -475,3 +490,204 @@ impl Field {
475 } 490 }
476 } 491 }
477} 492}
493
494fn lower(grammar: &Grammar) -> AstSrc {
495 let mut res = AstSrc::default();
496 res.tokens = vec!["Whitespace".into(), "Comment".into(), "String".into(), "RawString".into()];
497
498 let nodes = grammar
499 .iter()
500 .filter(|&node| match grammar[node].rule {
501 Rule::Node(it) if it == node => false,
502 _ => true,
503 })
504 .collect::<Vec<_>>();
505
506 for &node in &nodes {
507 let name = grammar[node].name.clone();
508 let rule = &grammar[node].rule;
509 match lower_enum(grammar, rule) {
510 Some(variants) => {
511 let enum_src = AstEnumSrc { doc: Vec::new(), name, traits: Vec::new(), variants };
512 res.enums.push(enum_src);
513 }
514 None => {
515 let mut fields = Vec::new();
516 lower_rule(&mut fields, grammar, rule);
517 res.nodes.push(AstNodeSrc { doc: Vec::new(), name, traits: Vec::new(), fields });
518 }
519 }
520 }
521
522 deduplicate_fields(&mut res);
523 extract_enums(&mut res);
524 extract_struct_traits(&mut res);
525 extract_enum_traits(&mut res);
526 res
527}
528
529fn lower_enum(grammar: &Grammar, rule: &Rule) -> Option<Vec<String>> {
530 let alternatives = match rule {
531 Rule::Alt(it) => it,
532 _ => return None,
533 };
534 let mut variants = Vec::new();
535 for alternative in alternatives {
536 match alternative {
537 Rule::Node(it) => variants.push(grammar[*it].name.clone()),
538 _ => return None,
539 }
540 }
541 Some(variants)
542}
543
544fn lower_rule(acc: &mut Vec<Field>, grammar: &Grammar, rule: &Rule) {
545 match rule {
546 Rule::Node(node) => {
547 let field = Field::Node { name: grammar[*node].name.clone(), src: FieldSrc::Shorthand };
548 acc.push(field);
549 }
550 Rule::Token(token) => {
551 let mut name = grammar[*token].name.clone();
552 if name != "int_number" && name != "string" {
553 if "[]{}()".contains(&name) {
554 name = format!("'{}'", name);
555 }
556 let field = Field::Token(name);
557 acc.push(field);
558 }
559 }
560 Rule::Rep(inner) => {
561 if let Rule::Node(node) = &**inner {
562 let name = grammar[*node].name.clone();
563 let label = pluralize(&to_lower_snake_case(&name));
564 let field = Field::Node { name: label.clone(), src: FieldSrc::Many(name) };
565 acc.push(field);
566 return;
567 }
568 todo!("{:?}", rule)
569 }
570 Rule::Labeled { label, rule } => {
571 let node = match &**rule {
572 Rule::Rep(inner) | Rule::Opt(inner) => match &**inner {
573 Rule::Node(node) => node,
574 _ => todo!("{:?}", rule),
575 },
576 Rule::Node(node) => node,
577 _ => todo!("{:?}", rule),
578 };
579 let field = Field::Node {
580 name: label.clone(),
581 src: match &**rule {
582 Rule::Rep(_) => FieldSrc::Many(grammar[*node].name.clone()),
583 _ => FieldSrc::Optional(grammar[*node].name.clone()),
584 },
585 };
586 acc.push(field);
587 }
588 Rule::Seq(rules) | Rule::Alt(rules) => {
589 for rule in rules {
590 lower_rule(acc, grammar, rule)
591 }
592 }
593 Rule::Opt(rule) => lower_rule(acc, grammar, rule),
594 }
595}
596
597fn deduplicate_fields(ast: &mut AstSrc) {
598 eprintln!();
599 for node in &mut ast.nodes {
600 let mut i = 0;
601 'outer: while i < node.fields.len() {
602 for j in 0..i {
603 let f1 = &node.fields[i];
604 let f2 = &node.fields[j];
605 if f1 == f2 {
606 node.fields.remove(i);
607 continue 'outer;
608 }
609 }
610 i += 1;
611 }
612 }
613}
614
615fn extract_enums(ast: &mut AstSrc) {
616 for node in &mut ast.nodes {
617 for enm in &ast.enums {
618 let mut to_remove = Vec::new();
619 for (i, field) in node.fields.iter().enumerate() {
620 let ty = field.ty().to_string();
621 if enm.variants.iter().any(|it| it == &ty) {
622 to_remove.push(i);
623 }
624 }
625 if to_remove.len() == enm.variants.len() {
626 node.remove_field(to_remove);
627 node.fields.push(Field::Node { name: enm.name.clone(), src: FieldSrc::Shorthand });
628 }
629 }
630 }
631}
632
633fn extract_struct_traits(ast: &mut AstSrc) {
634 let traits: &[(&str, &[&str])] = &[
635 ("AttrsOwner", &["attrs"]),
636 ("NameOwner", &["name"]),
637 ("VisibilityOwner", &["visibility"]),
638 ("TypeParamsOwner", &["type_param_list", "where_clause"]),
639 ("TypeBoundsOwner", &["type_bound_list", "colon_token"]),
640 ("ModuleItemOwner", &["items"]),
641 ("TypeAscriptionOwner", &["ascribed_type"]),
642 ("LoopBodyOwner", &["label", "loop_body"]),
643 ("ArgListOwner", &["arg_list"]),
644 ];
645
646 for node in &mut ast.nodes {
647 for (name, methods) in traits {
648 extract_struct_trait(node, name, methods);
649 }
650 }
651}
652
653fn extract_struct_trait(node: &mut AstNodeSrc, trait_name: &str, methods: &[&str]) {
654 let mut to_remove = Vec::new();
655 for (i, field) in node.fields.iter().enumerate() {
656 let method_name = field.method_name().to_string();
657 if methods.iter().any(|&it| it == &method_name) {
658 to_remove.push(i);
659 }
660 }
661 if to_remove.len() == methods.len() {
662 node.traits.push(trait_name.to_string());
663 node.remove_field(to_remove);
664 }
665}
666
667fn extract_enum_traits(ast: &mut AstSrc) {
668 for enm in &mut ast.enums {
669 let nodes = &ast.nodes;
670 let mut variant_traits = enm
671 .variants
672 .iter()
673 .map(|var| nodes.iter().find(|it| &it.name == var).unwrap())
674 .map(|node| node.traits.iter().cloned().collect::<BTreeSet<_>>());
675
676 let mut enum_traits = match variant_traits.next() {
677 Some(it) => it,
678 None => continue,
679 };
680 for traits in variant_traits {
681 enum_traits = enum_traits.intersection(&traits).cloned().collect();
682 }
683 enm.traits = enum_traits.into_iter().collect();
684 }
685}
686
687impl AstNodeSrc {
688 fn remove_field(&mut self, to_remove: Vec<usize>) {
689 to_remove.into_iter().rev().for_each(|idx| {
690 self.fields.remove(idx);
691 });
692 }
693}
diff --git a/xtask/src/codegen/rust.ungram b/xtask/src/codegen/rust.ungram
new file mode 100644
index 000000000..8a3eb7b29
--- /dev/null
+++ b/xtask/src/codegen/rust.ungram
@@ -0,0 +1,529 @@
1SourceFile =
2 Attr*
3 items:ModuleItem*
4
5FnDef =
6 Attr* Visibility? Abi? 'const' 'default' 'async' 'unsafe' 'fn' Name TypeParamList?
7 ParamList RetType?
8 WhereClause?
9 (body:BlockExpr | ';')
10
11RetType =
12 '->' TypeRef
13
14StructDef =
15 Attr* Visibility? 'struct' Name TypeParamList? (
16 WhereClause? (RecordFieldDefList | ';')
17 | TupleFieldDefList WhereClause? ';'
18 )
19
20UnionDef =
21 Attr* Visibility? 'union' Name TypeParamList? WhereClause?
22 RecordFieldDefList
23
24RecordFieldDefList =
25 '{' fields:RecordFieldDef* '}'
26
27RecordFieldDef =
28 Attr* Visibility? Name ':' ascribed_type:TypeRef
29
30TupleFieldDefList =
31 '(' fields:TupleFieldDef* ')'
32
33TupleFieldDef =
34 Attr* Visibility? Name TypeRef
35
36FieldDefList =
37 RecordFieldDefList
38| TupleFieldDefList
39
40EnumDef =
41 Attr* Visibility? 'enum' Name TypeParamList? WhereClause?
42 variant_list:EnumVariantList
43
44EnumVariantList =
45 '{' variants:EnumVariant* '}'
46
47EnumVariant =
48 Attr* Visibility? Name FieldDefList ('=' Expr)?
49
50TraitDef =
51 Attr* Visibility? 'unsafe'? 'auto'? 'trait' Name TypeParamList
52 (':' TypeBoundList?)? WhereClause
53 ItemList
54
55Module =
56 Attr* Visibility? 'mod' Name
57 (ItemList | ';')
58
59ItemList =
60 '{'
61 AssocItem*
62 items:ModuleItem*
63 '}'
64
65ConstDef =
66 Attr* Visibility? 'default'? 'const' Name ':' ascribed_type:TypeRef
67 '=' body:Expr ';'
68
69StaticDef =
70 Attr* Visibility? 'static'? 'mut'? 'static' Name ':' ascribed_type:TypeRef
71 '=' body:Expr ';'
72
73TypeAliasDef =
74 Attr* Visibility? 'default'? 'type' Name TypeParamList? WhereClause? (':' TypeBoundList?)?
75 '=' TypeRef ';'
76
77ImplDef =
78 Attr* Visibility? 'const'? 'default'? 'unsafe'? 'impl' TypeParamList? '!'? 'for'
79 WhereClause?
80 ItemList
81
82ParenType =
83 '(' TypeRef ')'
84
85TupleType =
86 '(' fields:TypeRef* ')'
87
88NeverType =
89 '!'
90
91PathType =
92 Path
93
94PointerType =
95 '*' ('const' | 'mut') TypeRef
96
97ArrayType =
98 '[' TypeRef ';' Expr ']'
99
100SliceType =
101 '[' TypeRef ']'
102
103ReferenceType =
104 '&' 'lifetime'? 'mut'? TypeRef
105
106PlaceholderType =
107 '_'
108
109FnPointerType =
110 Abi 'unsafe'? 'fn' ParamList RetType?
111
112ForType =
113 'for' TypeParamList TypeRef
114
115ImplTraitType =
116 'impl' TypeBoundList
117
118DynTraitType =
119 'dyn' TypeBoundList
120
121TupleExpr =
122 Attr* '(' Expr* ')'
123
124ArrayExpr =
125 Attr* '[' (Expr* | Expr ';' Expr) ']'
126
127ParenExpr =
128 Attr* '(' Expr ')'
129
130PathExpr =
131 Path
132
133LambdaExpr =
134 Attr* 'static'? 'async'? 'move'? ParamList RetType?
135 body:Expr
136
137IfExpr =
138 Attr* 'if' Condition
139
140Condition =
141 'let' Pat '=' Expr
142| Expr
143
144EffectExpr =
145 Attr* Label? ('try' | 'unsafe' | 'async') BlockExpr
146
147LoopExpr =
148 Attr* Label? 'loop'
149 loop_body:BlockExpr?
150
151ForExpr =
152 Attr* Label? 'for' Pat 'in' iterable:Expr
153 loop_body:BlockExpr?
154
155WhileExpr =
156 Attr* Label? 'while' Condition
157 loop_body:BlockExpr?
158
159ContinueExpr =
160 Attr* 'continue' 'lifetime'?
161
162BreakExpr =
163 Attr* 'break' 'lifetime'? Expr?
164
165Label =
166 'lifetime'
167
168BlockExpr =
169 Attr* Label
170 '{'
171 items:ModuleItem*
172 statements:Stmt*
173 Expr?
174 '}'
175
176ReturnExpr =
177 Attr* 'return' Expr
178
179CallExpr =
180 Attr* Expr ArgList
181
182MethodCallExpr =
183 Attr* Expr '.' NameRef TypeArgList? ArgList
184
185ArgList =
186 '(' args:Expr* ')'
187
188FieldExpr =
189 Attr* Expr '.' NameRef
190
191IndexExpr =
192 Attr* '[' ']'
193
194AwaitExpr =
195 Attr* Expr '.' 'await'
196
197TryExpr =
198 Attr* Expr '?'
199
200CastExpr =
201 Attr* Expr 'as' TypeRef
202
203RefExpr =
204 Attr* '&' ('raw' | 'mut' | 'const') Expr
205
206PrefixExpr =
207 Attr* Expr
208
209BoxExpr =
210 Attr* 'box' Expr
211
212RangeExpr =
213 Attr*
214
215BinExpr =
216 Attr*
217
218Literal =
219 'int_number'
220
221MatchExpr =
222 Attr* 'match' Expr MatchArmList
223
224MatchArmList =
225 '{' arms:MatchArm* '}'
226
227MatchArm =
228 Attr* Pat guard:MatchGuard? '=>' Expr
229
230MatchGuard =
231 'if' Expr
232
233RecordLit =
234 Path RecordFieldList
235
236RecordFieldList =
237 '{'
238 fields:RecordField*
239 ('..' spread:Expr)?
240 '}'
241
242RecordField =
243 Attr* NameRef (':' Expr)?
244
245OrPat =
246 Pat*
247
248ParenPat =
249 '(' Pat ')'
250
251RefPat =
252 '&' 'mut'? Pat
253
254BoxPat =
255 'box' Path
256
257BindPat =
258 Attr* 'ref'? 'mut'? Name ('@' Pat)?
259
260PlaceholderPat =
261 '_'
262
263DotDotPat =
264 '..'
265
266PathPat =
267 Path
268
269SlicePat =
270 '[' args:Pat* ']'
271
272RangePat =
273 '..' | '..='
274
275LiteralPat =
276 Literal
277
278MacroPat =
279 MacroCall
280
281RecordPat =
282 Path RecordFieldPatList
283
284RecordFieldPatList =
285 '{'
286 record_field_pats:RecordFieldPat*
287 BindPat*
288 '..'?
289 '}'
290
291RecordFieldPat =
292 Attr* NameRef ':' Pat
293
294TupleStructPat =
295 Path '(' args:Pat* ')'
296
297TuplePat =
298 '(' args:Pat* ')'
299
300Visibility =
301 'pub' ('(' 'super' | 'self' | 'crate' | 'in' Path ')')?
302
303Name =
304 'ident'
305
306NameRef =
307 'ident' | 'int_number'
308
309MacroCall =
310 Attr* Path '!' Name? TokenTree ';'?
311
312MacroDef =
313 Name TokenTree
314
315TokenTree =
316 '(' ')' | '{' '}' | '[' ']'
317
318MacroItems =
319 items:ModuleItem*
320
321MacroStmts =
322 statements:Stmt*
323 Expr?
324
325Attr =
326 '#' '!'? '[' Path ('=' input:AttrInput)? ']'
327
328TypeParamList =
329 '<'
330 TypeParam*
331 LifetimeParam*
332 ConstParam*
333 '>'
334
335TypeParam =
336 Attr* Name (':' TypeBoundList?)?
337 ('=' default_type:TypeRef)?
338
339ConstParam =
340 Attr* 'const' Name ':' ascribed_type:TypeRef
341 ('=' default_val:Expr)?
342
343LifetimeParam =
344 Attr* 'lifetime'
345
346TypeBound =
347 'lifetime' | 'const'? TypeRef
348
349TypeBoundList =
350 bounds:TypeBound*
351
352WherePred =
353 ('for' TypeParamList)? ('lifetime' | TypeRef) ':' TypeBoundList
354
355WhereClause =
356 'where' predicates:WherePred*
357
358Abi =
359 'string'
360
361ExprStmt =
362 Attr* Expr ';'
363
364LetStmt =
365 Attr* 'let' Pat (':' ascribed_type:TypeRef)
366 '=' initializer:Expr ';'
367
368ParamList =
369 '(' SelfParam Param* ')'
370
371SelfParam =
372 Attr* ('&' 'lifetime'?)? 'mut'? 'self' (':' ascribed_type:TypeRef)
373
374Param =
375 Attr* Pat (':' ascribed_type:TypeRef)
376| '...'
377
378UseItem =
379 Attr* Visibility? 'use' UseTree ';'
380
381UseTree =
382 Path ('::' ('*' | UseTreeList)) Alias?
383
384UseTreeList =
385 '{' UseTree* '}'
386
387Alias =
388 'as' Name
389
390ExternCrateItem =
391 Attr* Visibility? 'extern' 'crate' (NameRef | 'self') Alias? ';'
392
393Path =
394 (qualifier:Path '::')? segment:PathSegment
395
396PathSegment =
397 '::' | 'crate' | 'self' | 'super'
398| '<' NameRef TypeArgList ParamList RetType PathType '>'
399
400TypeArgList =
401 '::'? '<'
402 TypeArg*
403 LifetimeArg*
404 AssocTypeArg*
405 ConstArg*
406 '>'
407
408TypeArg =
409 TypeRef
410
411AssocTypeArg =
412 NameRef (':' TypeBoundList | '=' TypeRef)
413
414LifetimeArg =
415 'lifetime'
416
417ConstArg =
418 Literal | BlockExpr BlockExpr
419
420ExternBlock =
421 Attr* Abi ExternItemList
422
423ExternItemList =
424 '{' extern_items:ExternItem* '}'
425
426MetaItem =
427 Path '=' AttrInput nested_meta_items:MetaItem*
428
429NominalDef =
430 StructDef
431| EnumDef
432| UnionDef
433
434TypeRef =
435 ParenType
436| TupleType
437| NeverType
438| PathType
439| PointerType
440| ArrayType
441| SliceType
442| ReferenceType
443| PlaceholderType
444| FnPointerType
445| ForType
446| ImplTraitType
447| DynTraitType
448
449AssocItem =
450 FnDef
451| TypeAliasDef
452| ConstDef
453
454ExternItem =
455 FnDef | StaticDef
456
457ModuleItem =
458 StructDef
459| UnionDef
460| EnumDef
461| FnDef
462| TraitDef
463| TypeAliasDef
464| ImplDef
465| UseItem
466| ExternCrateItem
467| ConstDef
468| StaticDef
469| Module
470| MacroCall
471| ExternBlock
472
473AttrInput =
474 Literal
475| TokenTree
476
477Stmt =
478 LetStmt
479| ExprStmt
480
481Pat =
482 OrPat
483| ParenPat
484| RefPat
485| BoxPat
486| BindPat
487| PlaceholderPat
488| DotDotPat
489| PathPat
490| RecordPat
491| TupleStructPat
492| TuplePat
493| SlicePat
494| RangePat
495| LiteralPat
496| MacroPat
497
498Expr =
499 TupleExpr
500| ArrayExpr
501| ParenExpr
502| PathExpr
503| LambdaExpr
504| IfExpr
505| LoopExpr
506| ForExpr
507| WhileExpr
508| ContinueExpr
509| BreakExpr
510| Label
511| BlockExpr
512| ReturnExpr
513| MatchExpr
514| RecordLit
515| CallExpr
516| IndexExpr
517| MethodCallExpr
518| FieldExpr
519| AwaitExpr
520| TryExpr
521| EffectExpr
522| CastExpr
523| RefExpr
524| PrefixExpr
525| RangeExpr
526| BinExpr
527| Literal
528| MacroCall
529| BoxExpr