Implement builtin#format_args, using rustc's format_args parser

Veykril · Veykril · commit a491b92bca45 · 2023-09-05T19:06:15.000+02:00
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/crates/hir-def/Cargo.toml b/crates/hir-def/Cargo.toml
@@ -33,6 +33,7 @@ triomphe.workspace = true
 
 rustc_abi = { version = "0.0.20221221", package = "hkalbasi-rustc-ap-rustc_abi", default-features = false }
 rustc_index = { version = "0.0.20221221", package = "hkalbasi-rustc-ap-rustc_index", default-features = false }
+rustc_lexer = { version = "0.1.0", package = "ra-ap-rustc_lexer" }
 
 # local deps
 stdx.workspace = true
diff --git a/crates/hir-def/src/body/lower.rs b/crates/hir-def/src/body/lower.rs
@@ -29,9 +29,13 @@ use crate::{
     db::DefDatabase,
     expander::Expander,
     hir::{
-        dummy_expr_id, Array, Binding, BindingAnnotation, BindingId, BindingProblems, CaptureBy,
-        ClosureKind, Expr, ExprId, InlineAsm, Label, LabelId, Literal, LiteralOrConst, MatchArm,
-        Movability, OffsetOf, Pat, PatId, RecordFieldPat, RecordLitField, Statement,
+        dummy_expr_id,
+        format_args::{
+            self, FormatArgs, FormatArgument, FormatArgumentKind, FormatArgumentsCollector,
+        },
+        Array, Binding, BindingAnnotation, BindingId, BindingProblems, CaptureBy, ClosureKind,
+        Expr, ExprId, InlineAsm, Label, LabelId, Literal, LiteralOrConst, MatchArm, Movability,
+        OffsetOf, Pat, PatId, RecordFieldPat, RecordLitField, Statement,
     },
     item_scope::BuiltinShadowMode,
     lang_item::LangItem,
@@ -649,15 +653,58 @@ impl ExprCollector<'_> {
             }
             ast::Expr::UnderscoreExpr(_) => self.alloc_expr(Expr::Underscore, syntax_ptr),
             ast::Expr::AsmExpr(e) => {
-                let expr = Expr::InlineAsm(InlineAsm { e: self.collect_expr_opt(e.expr()) });
-                self.alloc_expr(expr, syntax_ptr)
+                let e = self.collect_expr_opt(e.expr());
+                self.alloc_expr(Expr::InlineAsm(InlineAsm { e }), syntax_ptr)
             }
             ast::Expr::OffsetOfExpr(e) => {
                 let container = Interned::new(TypeRef::from_ast_opt(&self.ctx(), e.ty()));
                 let fields = e.fields().map(|it| it.as_name()).collect();
                 self.alloc_expr(Expr::OffsetOf(OffsetOf { container, fields }), syntax_ptr)
             }
-            ast::Expr::FormatArgsExpr(_) => self.missing_expr(),
+            ast::Expr::FormatArgsExpr(f) => {
+                let mut args = FormatArgumentsCollector::new();
+                f.args().for_each(|arg| {
+                    args.add(FormatArgument {
+                        kind: match arg.name() {
+                            Some(name) => FormatArgumentKind::Named(name.as_name()),
+                            None => FormatArgumentKind::Normal,
+                        },
+                        expr: self.collect_expr_opt(arg.expr()),
+                    });
+                });
+                let template = f.template();
+                let fmt_snippet = template.as_ref().map(ToString::to_string);
+                let expr = self.collect_expr_opt(f.template());
+                if let Expr::Literal(Literal::String(_)) = self.body[expr] {
+                    let source = self.source_map.expr_map_back[expr].clone();
+                    let is_direct_literal = source.file_id == self.expander.current_file_id;
+                    if let ast::Expr::Literal(l) =
+                        source.value.to_node(&self.db.parse_or_expand(source.file_id))
+                    {
+                        if let ast::LiteralKind::String(s) = l.kind() {
+                            return Some(self.alloc_expr(
+                                Expr::FormatArgs(format_args::parse(
+                                    expr,
+                                    &s,
+                                    fmt_snippet,
+                                    args,
+                                    is_direct_literal,
+                                )),
+                                syntax_ptr,
+                            ));
+                        }
+                    }
+                }
+
+                self.alloc_expr(
+                    Expr::FormatArgs(FormatArgs {
+                        template_expr: expr,
+                        template: Default::default(),
+                        arguments: args.finish(),
+                    }),
+                    syntax_ptr,
+                )
+            }
         })
     }
 
diff --git a/crates/hir-def/src/body/pretty.rs b/crates/hir-def/src/body/pretty.rs
@@ -156,6 +156,11 @@ impl Printer<'_> {
             Expr::Missing => w!(self, "�"),
             Expr::Underscore => w!(self, "_"),
             Expr::InlineAsm(_) => w!(self, "builtin#asm(_)"),
+            Expr::FormatArgs(_fmt_args) => {
+                w!(self, "builtin#format_args(");
+                // FIXME
+                w!(self, ")");
+            }
             Expr::OffsetOf(offset_of) => {
                 w!(self, "builtin#offset_of(");
                 self.print_type_ref(&offset_of.container);
diff --git a/crates/hir-def/src/hir.rs b/crates/hir-def/src/hir.rs
@@ -13,6 +13,7 @@
 //! See also a neighboring `body` module.
 
 pub mod type_ref;
+pub mod format_args;
 
 use std::fmt;
 
@@ -24,6 +25,7 @@ use syntax::ast;
 
 use crate::{
     builtin_type::{BuiltinFloat, BuiltinInt, BuiltinUint},
+    hir::format_args::{FormatArgs, FormatArgumentKind},
     path::{GenericArgs, Path},
     type_ref::{Mutability, Rawness, TypeRef},
     BlockId, ConstBlockId,
@@ -117,7 +119,6 @@ impl From<ast::LiteralKind> for Literal {
     fn from(ast_lit_kind: ast::LiteralKind) -> Self {
         use ast::LiteralKind;
         match ast_lit_kind {
-            // FIXME: these should have actual values filled in, but unsure on perf impact
             LiteralKind::IntNumber(lit) => {
                 if let builtin @ Some(_) = lit.suffix().and_then(BuiltinFloat::from_suffix) {
                     Literal::Float(
@@ -283,6 +284,7 @@ pub enum Expr {
     Underscore,
     OffsetOf(OffsetOf),
     InlineAsm(InlineAsm),
+    FormatArgs(FormatArgs),
 }
 
 #[derive(Debug, Clone, PartialEq, Eq)]
@@ -355,7 +357,15 @@ impl Expr {
         match self {
             Expr::Missing => {}
             Expr::Path(_) | Expr::OffsetOf(_) => {}
-            Expr::InlineAsm(e) => f(e.e),
+            Expr::InlineAsm(it) => f(it.e),
+            Expr::FormatArgs(it) => {
+                f(it.template_expr);
+                it.arguments
+                    .arguments
+                    .iter()
+                    .filter(|it| !matches!(it.kind, FormatArgumentKind::Captured(_)))
+                    .for_each(|it| f(it.expr));
+            }
             Expr::If { condition, then_branch, else_branch } => {
                 f(*condition);
                 f(*then_branch);
diff --git a/crates/hir-expand/src/name.rs b/crates/hir-expand/src/name.rs
@@ -54,6 +54,12 @@ impl Name {
         Name(Repr::Text(text))
     }
 
+    // FIXME: See above, unfortunately some places really need this right now
+    #[doc(hidden)]
+    pub const fn new_text_dont_use(text: SmolStr) -> Name {
+        Name(Repr::Text(text))
+    }
+
     pub fn new_tuple_field(idx: usize) -> Name {
         Name(Repr::TupleField(idx))
     }
diff --git a/crates/hir-ty/src/infer/closure.rs b/crates/hir-ty/src/infer/closure.rs
@@ -9,7 +9,10 @@ use chalk_ir::{
 };
 use hir_def::{
     data::adt::VariantData,
-    hir::{Array, BinaryOp, BindingId, CaptureBy, Expr, ExprId, Pat, PatId, Statement, UnaryOp},
+    hir::{
+        format_args::FormatArgumentKind, Array, BinaryOp, BindingId, CaptureBy, Expr, ExprId, Pat,
+        PatId, Statement, UnaryOp,
+    },
     lang_item::LangItem,
     resolver::{resolver_for_expr, ResolveValueResult, ValueNs},
     DefWithBodyId, FieldId, HasModule, VariantId,
@@ -453,6 +456,14 @@ impl InferenceContext<'_> {
     fn walk_expr_without_adjust(&mut self, tgt_expr: ExprId) {
         match &self.body[tgt_expr] {
             Expr::OffsetOf(_) => (),
+            Expr::FormatArgs(fa) => {
+                self.walk_expr_without_adjust(fa.template_expr);
+                fa.arguments
+                    .arguments
+                    .iter()
+                    .filter(|it| !matches!(it.kind, FormatArgumentKind::Captured(_)))
+                    .for_each(|it| self.walk_expr_without_adjust(it.expr));
+            }
             Expr::InlineAsm(e) => self.walk_expr_without_adjust(e.e),
             Expr::If { condition, then_branch, else_branch } => {
                 self.consume_expr(*condition);
diff --git a/crates/hir-ty/src/infer/expr.rs b/crates/hir-ty/src/infer/expr.rs
@@ -9,7 +9,8 @@ use chalk_ir::{cast::Cast, fold::Shift, DebruijnIndex, Mutability, TyVariableKin
 use hir_def::{
     generics::TypeOrConstParamData,
     hir::{
-        ArithOp, Array, BinaryOp, ClosureKind, Expr, ExprId, LabelId, Literal, Statement, UnaryOp,
+        format_args::FormatArgumentKind, ArithOp, Array, BinaryOp, ClosureKind, Expr, ExprId,
+        LabelId, Literal, Statement, UnaryOp,
     },
     lang_item::{LangItem, LangItemTarget},
     path::{GenericArg, GenericArgs},
@@ -848,6 +849,25 @@ impl InferenceContext<'_> {
                 self.infer_expr_no_expect(it.e);
                 self.result.standard_types.unit.clone()
             }
+            Expr::FormatArgs(fa) => {
+                fa.arguments
+                    .arguments
+                    .iter()
+                    .filter(|it| !matches!(it.kind, FormatArgumentKind::Captured(_)))
+                    .for_each(|it| _ = self.infer_expr_no_expect(it.expr));
+
+                match self
+                    .resolve_lang_item(LangItem::FormatArguments)
+                    .and_then(|it| it.as_struct())
+                {
+                    Some(s) => {
+                        // NOTE: This struct has a lifetime parameter, but we don't currently emit
+                        // those to chalk
+                        TyKind::Adt(AdtId(s.into()), Substitution::empty(Interner)).intern(Interner)
+                    }
+                    None => self.err_ty(),
+                }
+            }
         };
         // use a new type variable if we got unknown here
         let ty = self.insert_type_vars_shallow(ty);
diff --git a/crates/hir-ty/src/infer/mutability.rs b/crates/hir-ty/src/infer/mutability.rs
@@ -3,7 +3,10 @@
 
 use chalk_ir::Mutability;
 use hir_def::{
-    hir::{Array, BinaryOp, BindingAnnotation, Expr, ExprId, PatId, Statement, UnaryOp},
+    hir::{
+        format_args::FormatArgumentKind, Array, BinaryOp, BindingAnnotation, Expr, ExprId, PatId,
+        Statement, UnaryOp,
+    },
     lang_item::LangItem,
 };
 use hir_expand::name;
@@ -37,6 +40,13 @@ impl InferenceContext<'_> {
             Expr::Missing => (),
             Expr::InlineAsm(e) => self.infer_mut_expr_without_adjust(e.e, Mutability::Not),
             Expr::OffsetOf(_) => (),
+            Expr::FormatArgs(fa) => {
+                fa.arguments
+                    .arguments
+                    .iter()
+                    .filter(|it| !matches!(it.kind, FormatArgumentKind::Captured(_)))
+                    .for_each(|arg| self.infer_mut_expr_without_adjust(arg.expr, Mutability::Not));
+            }
             &Expr::If { condition, then_branch, else_branch } => {
                 self.infer_mut_expr(condition, Mutability::Not);
                 self.infer_mut_expr(then_branch, Mutability::Not);
diff --git a/crates/hir-ty/src/mir/lower.rs b/crates/hir-ty/src/mir/lower.rs
@@ -376,6 +376,9 @@ impl<'ctx> MirLowerCtx<'ctx> {
             Expr::InlineAsm(_) => {
                 not_supported!("builtin#asm")
             }
+            Expr::FormatArgs(_) => {
+                not_supported!("builtin#format_args")
+            }
             Expr::Missing => {
                 if let DefWithBodyId::FunctionId(f) = self.owner {
                     let assoc = f.lookup(self.db.upcast());
diff --git a/crates/hir-ty/src/tests/simple.rs b/crates/hir-ty/src/tests/simple.rs
@@ -3612,3 +3612,25 @@ fn main() {
 "#,
     );
 }
+
+#[test]
+fn builtin_format_args() {
+    check_infer(
+        r#"
+#[lang = "format_arguments"]
+pub struct Arguments<'a>;
+fn main() {
+    let are = "are";
+    builtin#format_args("hello {} friends, we {are} {0}{last}", "fancy", last = "!");
+}
+"#,
+        expect![[r#"
+            65..175 '{     ...!"); }': ()
+            75..78 'are': &str
+            81..86 '"are"': &str
+            92..172 'builti...= "!")': Arguments<'_>
+            152..159 '"fancy"': &str
+            168..171 '"!"': &str
+        "#]],
+    );
+}
diff --git a/crates/parser/src/grammar/expressions/atom.rs b/crates/parser/src/grammar/expressions/atom.rs
@@ -219,7 +219,7 @@ fn tuple_expr(p: &mut Parser<'_>) -> CompletedMarker {
 // test builtin_expr
 // fn foo() {
 //     builtin#asm(0);
-//     builtin#format_args(0);
+//     builtin#format_args("", 0, 1, a = 2 + 3, a + b);
 //     builtin#offset_of(Foo, bar.baz.0);
 // }
 fn builtin_expr(p: &mut Parser<'_>) -> Option<CompletedMarker> {
@@ -249,6 +249,24 @@ fn builtin_expr(p: &mut Parser<'_>) -> Option<CompletedMarker> {
         p.bump_remap(T![format_args]);
         p.expect(T!['(']);
         expr(p);
+        if p.eat(T![,]) {
+            while !p.at(EOF) && !p.at(T![')']) {
+                let m = p.start();
+                if p.at(IDENT) && p.nth_at(1, T![=]) {
+                    name(p);
+                    p.bump(T![=]);
+                }
+                if expr(p).is_none() {
+                    m.abandon(p);
+                    break;
+                }
+                m.complete(p, FORMAT_ARGS_ARG);
+
+                if !p.at(T![')']) {
+                    p.expect(T![,]);
+                }
+            }
+        }
         p.expect(T![')']);
         Some(m.complete(p, FORMAT_ARGS_EXPR))
     } else if p.at_contextual_kw(T![asm]) {
diff --git a/crates/parser/src/syntax_kind/generated.rs b/crates/parser/src/syntax_kind/generated.rs
@@ -210,6 +210,7 @@ pub enum SyntaxKind {
     OFFSET_OF_EXPR,
     ASM_EXPR,
     FORMAT_ARGS_EXPR,
+    FORMAT_ARGS_ARG,
     CALL_EXPR,
     INDEX_EXPR,
     METHOD_CALL_EXPR,
diff --git a/crates/parser/test_data/parser/inline/ok/0207_builtin_expr.rs b/crates/parser/test_data/parser/inline/ok/0207_builtin_expr.rs
@@ -1,5 +1,5 @@
 fn foo() {
     builtin#asm(0);
-    builtin#format_args(0);
+    builtin#format_args("", 0, 1, a = 2 + 3, a + b);
     builtin#offset_of(Foo, bar.baz.0);
 }
diff --git a/crates/syntax/rust.ungram b/crates/syntax/rust.ungram
@@ -382,7 +382,13 @@ AsmExpr =
   Attr* 'builtin' '#' 'asm' '(' Expr ')'
 
 FormatArgsExpr =
-  Attr* 'builtin' '#' 'format_args' '(' ')'
+  Attr* 'builtin' '#' 'format_args' '('
+  template:Expr
+  (',' args:(FormatArgsArg (',' FormatArgsArg)* ','?)? )?
+  ')'
+
+FormatArgsArg =
+  (Name '=')? Expr
 
 MacroExpr =
   MacroCall
diff --git a/crates/syntax/src/ast/generated/nodes.rs b/crates/syntax/src/ast/generated/nodes.rs
diff --git a/crates/syntax/src/tests/ast_src.rs b/crates/syntax/src/tests/ast_src.rs

Original file line number	Diff line number	Diff line change
`@@ -54,6 +54,12 @@ impl Name {`
`54`	`54`	`Name(Repr::Text(text))`
`55`	`55`	`}`
`56`	`56`
	`57`	`+ // FIXME: See above, unfortunately some places really need this right now`
	`58`	`+ #[doc(hidden)]`
	`59`	`+ pub const fn new_text_dont_use(text: SmolStr) -> Name {`
	`60`	`+ Name(Repr::Text(text))`
	`61`	`+ }`
	`62`	`+`
`57`	`63`	`pub fn new_tuple_field(idx: usize) -> Name {`
`58`	`64`	`Name(Repr::TupleField(idx))`
`59`	`65`	`}`
Original file line number	Diff line number	Diff line change
`@@ -1,5 +1,5 @@`
`1`	`1`	`fn foo() {`
`2`	`2`	`builtin#asm(0);`
`3`		`- builtin#format_args(0);`
	`3`	`+ builtin#format_args("", 0, 1, a = 2 + 3, a + b);`
`4`	`4`	`builtin#offset_of(Foo, bar.baz.0);`
`5`	`5`	`}`