Merge #1148

1148: Add token_tree_to_xxx functions r=matklad a=edwin0cheng

<del>As discus in  PR #1147 , this PR added a `mbe::MacroKind` .
Currently only 2 kind of macro are supported, `SourceFile` and `Block`.</del>

Added following functions for `tt::TokenTree` and `ast::Node` conversion:

* token_tree_to_expr
* token_tree_to_pat
* token_tree_to_ty
* token_tree_to_macro_stmts
* token_tree_to_macro_items

And added two new syntax kind:

* MACRO_ITEMS
* MACRO_STMTS

Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>

crates/ra_mbe/src/lib.rs

@@ -37,9 +37,19 @@ pub enum ExpandError {
     NoMatchingRule,
     UnexpectedToken,
     BindingError(String),
+    ConversionError,
 }
 
-pub use crate::syntax_bridge::{ast_to_token_tree, token_tree_to_ast_item_list, syntax_node_to_token_tree};
+pub use crate::syntax_bridge::{
+    ast_to_token_tree,
+    token_tree_to_ast_item_list,
+    syntax_node_to_token_tree,
+    token_tree_to_expr,
+    token_tree_to_pat,
+    token_tree_to_ty,
+    token_tree_to_macro_items,
+    token_tree_to_macro_stmts,
+};
 
 /// This struct contains AST for a single `macro_rules` definition. What might
 /// be very confusing is that AST has almost exactly the same shape as
@@ -192,23 +202,26 @@ impl_froms!(TokenTree: Leaf, Subtree);
     pub(crate) fn expand_to_syntax(
         rules: &MacroRules,
         invocation: &str,
-    ) -> ra_syntax::TreeArc<ast::SourceFile> {
+    ) -> ra_syntax::TreeArc<ast::MacroItems> {
         let expanded = expand(rules, invocation);
-        token_tree_to_ast_item_list(&expanded)
+        token_tree_to_macro_items(&expanded).unwrap()
     }
 
     pub(crate) fn assert_expansion(rules: &MacroRules, invocation: &str, expansion: &str) {
         let expanded = expand(rules, invocation);
         assert_eq!(expanded.to_string(), expansion);
 
-        let tree = token_tree_to_ast_item_list(&expanded);
+        let tree = token_tree_to_macro_items(&expanded);
 
         // Eat all white space by parse it back and forth
         let expansion = ast::SourceFile::parse(expansion);
         let expansion = syntax_node_to_token_tree(expansion.syntax()).unwrap().0;
-        let file = token_tree_to_ast_item_list(&expansion);
+        let file = token_tree_to_macro_items(&expansion);
 
-        assert_eq!(tree.syntax().debug_dump().trim(), file.syntax().debug_dump().trim());
+        assert_eq!(
+            tree.unwrap().syntax().debug_dump().trim(),
+            file.unwrap().syntax().debug_dump().trim()
+        );
     }
 
     #[test]
@@ -346,11 +359,11 @@ impl_froms!(TokenTree: Leaf, Subtree);
             ",
         );
         let expansion = expand(&rules, "structs!(Foo, Bar)");
-        let tree = token_tree_to_ast_item_list(&expansion);
+        let tree = token_tree_to_macro_items(&expansion);
         assert_eq!(
-            tree.syntax().debug_dump().trim(),
+            tree.unwrap().syntax().debug_dump().trim(),
             r#"
-SOURCE_FILE@[0; 40)
+MACRO_ITEMS@[0; 40)
   STRUCT_DEF@[0; 20)
     STRUCT_KW@[0; 6) "struct"
     NAME@[6; 9)
@@ -444,6 +457,59 @@ SOURCE_FILE@[0; 40)
         assert_expansion(&rules, "foo! { foo, bar }", "fn foo () {let a = foo ; let b = bar ;}");
     }
 
+    #[test]
+    fn test_tt_to_stmts() {
+        let rules = create_rules(
+            r#"
+        macro_rules! foo {
+            () => {
+                 let a = 0;
+                 a = 10 + 1;
+                 a
+            }
+        }
+"#,
+        );
+
+        let expanded = expand(&rules, "foo!{}");
+        let stmts = token_tree_to_macro_stmts(&expanded);
+
+        assert_eq!(
+            stmts.unwrap().syntax().debug_dump().trim(),
+            r#"MACRO_STMTS@[0; 15)
+  LET_STMT@[0; 7)
+    LET_KW@[0; 3) "let"
+    BIND_PAT@[3; 4)
+      NAME@[3; 4)
+        IDENT@[3; 4) "a"
+    EQ@[4; 5) "="
+    LITERAL@[5; 6)
+      INT_NUMBER@[5; 6) "0"
+    SEMI@[6; 7) ";"
+  EXPR_STMT@[7; 14)
+    BIN_EXPR@[7; 13)
+      PATH_EXPR@[7; 8)
+        PATH@[7; 8)
+          PATH_SEGMENT@[7; 8)
+            NAME_REF@[7; 8)
+              IDENT@[7; 8) "a"
+      EQ@[8; 9) "="
+      BIN_EXPR@[9; 13)
+        LITERAL@[9; 11)
+          INT_NUMBER@[9; 11) "10"
+        PLUS@[11; 12) "+"
+        LITERAL@[12; 13)
+          INT_NUMBER@[12; 13) "1"
+    SEMI@[13; 14) ";"
+  EXPR_STMT@[14; 15)
+    PATH_EXPR@[14; 15)
+      PATH@[14; 15)
+        PATH_SEGMENT@[14; 15)
+          NAME_REF@[14; 15)
+            IDENT@[14; 15) "a""#,
+        );
+    }
+
     // The following tests are port from intellij-rust directly
     // https://github.com/intellij-rust/intellij-rust/blob/c4e9feee4ad46e7953b1948c112533360b6087bb/src/test/kotlin/org/rust/lang/core/macros/RsMacroExpansionTest.kt
 
@@ -527,7 +593,7 @@ SOURCE_FILE@[0; 40)
 
         assert_eq!(
             expand_to_syntax(&rules, "foo! { 1 + 1  }").syntax().debug_dump().trim(),
-            r#"SOURCE_FILE@[0; 15)
+            r#"MACRO_ITEMS@[0; 15)
   FN_DEF@[0; 15)
     FN_KW@[0; 2) "fn"
     NAME@[2; 5)

crates/ra_mbe/src/syntax_bridge.rs

@@ -5,6 +5,7 @@ use ra_syntax::{
 };
 
 use crate::subtree_source::{SubtreeTokenSource, Querier};
+use crate::ExpandError;
 
 /// Maps `tt::TokenId` to the relative range of the original token.
 #[derive(Default)]
@@ -30,6 +31,71 @@ pub fn syntax_node_to_token_tree(node: &SyntaxNode) -> Option<(tt::Subtree, Toke
     Some((tt, token_map))
 }
 
+// The following items are what `rustc` macro can be parsed into :
+// link: https://github.com/rust-lang/rust/blob/9ebf47851a357faa4cd97f4b1dc7835f6376e639/src/libsyntax/ext/expand.rs#L141
+// * Expr(P<ast::Expr>)                     -> token_tree_to_expr
+// * Pat(P<ast::Pat>)                       -> token_tree_to_pat
+// * Ty(P<ast::Ty>)                         -> token_tree_to_ty
+// * Stmts(SmallVec<[ast::Stmt; 1]>)        -> token_tree_to_stmts
+// * Items(SmallVec<[P<ast::Item>; 1]>)     -> token_tree_to_items
+//
+// * TraitItems(SmallVec<[ast::TraitItem; 1]>)
+// * ImplItems(SmallVec<[ast::ImplItem; 1]>)
+// * ForeignItems(SmallVec<[ast::ForeignItem; 1]>
+//
+//
+
+/// Parses the token tree (result of macro expansion) to an expression
+pub fn token_tree_to_expr(tt: &tt::Subtree) -> Result<TreeArc<ast::Expr>, ExpandError> {
+    let token_source = SubtreeTokenSource::new(tt);
+    let mut tree_sink = TtTreeSink::new(token_source.querier());
+    ra_parser::parse_expr(&token_source, &mut tree_sink);
+    let syntax = tree_sink.inner.finish();
+    ast::Expr::cast(&syntax)
+        .map(|m| m.to_owned())
+        .ok_or_else(|| crate::ExpandError::ConversionError)
+}
+
+/// Parses the token tree (result of macro expansion) to a Pattern
+pub fn token_tree_to_pat(tt: &tt::Subtree) -> Result<TreeArc<ast::Pat>, ExpandError> {
+    let token_source = SubtreeTokenSource::new(tt);
+    let mut tree_sink = TtTreeSink::new(token_source.querier());
+    ra_parser::parse_pat(&token_source, &mut tree_sink);
+    let syntax = tree_sink.inner.finish();
+    ast::Pat::cast(&syntax).map(|m| m.to_owned()).ok_or_else(|| ExpandError::ConversionError)
+}
+
+/// Parses the token tree (result of macro expansion) to a Type
+pub fn token_tree_to_ty(tt: &tt::Subtree) -> Result<TreeArc<ast::TypeRef>, ExpandError> {
+    let token_source = SubtreeTokenSource::new(tt);
+    let mut tree_sink = TtTreeSink::new(token_source.querier());
+    ra_parser::parse_ty(&token_source, &mut tree_sink);
+    let syntax = tree_sink.inner.finish();
+    ast::TypeRef::cast(&syntax).map(|m| m.to_owned()).ok_or_else(|| ExpandError::ConversionError)
+}
+
+/// Parses the token tree (result of macro expansion) as a sequence of stmts
+pub fn token_tree_to_macro_stmts(
+    tt: &tt::Subtree,
+) -> Result<TreeArc<ast::MacroStmts>, ExpandError> {
+    let token_source = SubtreeTokenSource::new(tt);
+    let mut tree_sink = TtTreeSink::new(token_source.querier());
+    ra_parser::parse_macro_stmts(&token_source, &mut tree_sink);
+    let syntax = tree_sink.inner.finish();
+    ast::MacroStmts::cast(&syntax).map(|m| m.to_owned()).ok_or_else(|| ExpandError::ConversionError)
+}
+
+/// Parses the token tree (result of macro expansion) as a sequence of items
+pub fn token_tree_to_macro_items(
+    tt: &tt::Subtree,
+) -> Result<TreeArc<ast::MacroItems>, ExpandError> {
+    let token_source = SubtreeTokenSource::new(tt);
+    let mut tree_sink = TtTreeSink::new(token_source.querier());
+    ra_parser::parse_macro_items(&token_source, &mut tree_sink);
+    let syntax = tree_sink.inner.finish();
+    ast::MacroItems::cast(&syntax).map(|m| m.to_owned()).ok_or_else(|| ExpandError::ConversionError)
+}
+
 /// Parses the token tree (result of macro expansion) as a sequence of items
 pub fn token_tree_to_ast_item_list(tt: &tt::Subtree) -> TreeArc<ast::SourceFile> {
     let token_source = SubtreeTokenSource::new(tt);

crates/ra_parser/src/grammar.rs

@@ -49,6 +49,27 @@ pub(crate) fn root(p: &mut Parser) {
     m.complete(p, SOURCE_FILE);
 }
 
+pub(crate) fn macro_items(p: &mut Parser) {
+    let m = p.start();
+    items::mod_contents(p, false);
+    m.complete(p, MACRO_ITEMS);
+}
+
+pub(crate) fn macro_stmts(p: &mut Parser) {
+    let m = p.start();
+
+    while !p.at(EOF) {
+        if p.current() == SEMI {
+            p.bump();
+            continue;
+        }
+
+        expressions::stmt(p, expressions::StmtWithSemi::Optional);
+    }
+
+    m.complete(p, MACRO_STMTS);
+}
+
 pub(crate) fn path(p: &mut Parser) {
     paths::type_path(p);
 }
@@ -66,6 +87,11 @@ pub(crate) fn pattern(p: &mut Parser) {
 }
 
 pub(crate) fn stmt(p: &mut Parser, with_semi: bool) {
+    let with_semi = match with_semi {
+        true => expressions::StmtWithSemi::Yes,
+        false => expressions::StmtWithSemi::No,
+    };
+
     expressions::stmt(p, with_semi)
 }
 

crates/ra_parser/src/grammar/expressions.rs

@@ -4,6 +4,12 @@ pub(crate) use self::atom::match_arm_list;
 pub(super) use self::atom::{literal, LITERAL_FIRST};
 use super::*;
 
+pub(super) enum StmtWithSemi {
+    Yes,
+    No,
+    Optional,
+}
+
 const EXPR_FIRST: TokenSet = LHS_FIRST;
 
 pub(super) fn expr(p: &mut Parser) -> BlockLike {
@@ -48,7 +54,7 @@ fn is_expr_stmt_attr_allowed(kind: SyntaxKind) -> bool {
     }
 }
 
-pub(super) fn stmt(p: &mut Parser, with_semi: bool) {
+pub(super) fn stmt(p: &mut Parser, with_semi: StmtWithSemi) {
     // test block_items
     // fn a() { fn b() {} }
     let m = p.start();
@@ -111,13 +117,23 @@ pub(super) fn stmt(p: &mut Parser, with_semi: bool) {
         //     }
         //     test!{}
         // }
-        if with_semi {
+
-            if blocklike.is_block() {
+        match with_semi {
-                p.eat(SEMI);
+            StmtWithSemi::Yes => {
-            } else {
+                if blocklike.is_block() {
-                p.expect(SEMI);
+                    p.eat(SEMI);
+                } else {
+                    p.expect(SEMI);
+                }
+            }
+            StmtWithSemi::No => {}
+            StmtWithSemi::Optional => {
+                if p.at(SEMI) {
+                    p.eat(SEMI);
+                }
             }
         }
+
         m.complete(p, EXPR_STMT);
     }
 
@@ -128,7 +144,7 @@ pub(super) fn stmt(p: &mut Parser, with_semi: bool) {
     //     let c = 92;
     //     let d: i32 = 92;
     // }
-    fn let_stmt(p: &mut Parser, m: Marker, with_semi: bool) {
+    fn let_stmt(p: &mut Parser, m: Marker, with_semi: StmtWithSemi) {
         assert!(p.at(LET_KW));
         p.bump();
         patterns::pattern(p);
@@ -139,8 +155,16 @@ pub(super) fn stmt(p: &mut Parser, with_semi: bool) {
             expressions::expr(p);
         }
 
-        if with_semi {
+        match with_semi {
-            p.expect(SEMI);
+            StmtWithSemi::Yes => {
+                p.expect(SEMI);
+            }
+            StmtWithSemi::No => {}
+            StmtWithSemi::Optional => {
+                if p.at(SEMI) {
+                    p.eat(SEMI);
+                }
+            }
         }
         m.complete(p, LET_STMT);
     }
@@ -160,7 +184,7 @@ pub(crate) fn expr_block_contents(p: &mut Parser) {
             continue;
         }
 
-        stmt(p, true)
+        stmt(p, StmtWithSemi::Yes)
     }
 }
 

crates/ra_parser/src/lib.rs

@@ -98,6 +98,14 @@ pub fn parse_item(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink)
     parse_from_tokens(token_source, tree_sink, grammar::item);
 }
 
+pub fn parse_macro_items(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink) {
+    parse_from_tokens(token_source, tree_sink, grammar::macro_items);
+}
+
+pub fn parse_macro_stmts(token_source: &dyn TokenSource, tree_sink: &mut dyn TreeSink) {
+    parse_from_tokens(token_source, tree_sink, grammar::macro_stmts);
+}
+
 /// A parsing function for a specific braced-block.
 pub struct Reparser(fn(&mut parser::Parser));
 

crates/ra_parser/src/syntax_kind/generated.rs

@@ -233,6 +233,8 @@ pub enum SyntaxKind {
     ARG_LIST,
     TYPE_BOUND,
     TYPE_BOUND_LIST,
+    MACRO_ITEMS,
+    MACRO_STMTS,
     // Technical kind so that we can cast from u16 safely
     #[doc(hidden)]
     __LAST,
@@ -592,6 +594,8 @@ impl SyntaxKind {
             ARG_LIST => &SyntaxInfo { name: "ARG_LIST" },
             TYPE_BOUND => &SyntaxInfo { name: "TYPE_BOUND" },
             TYPE_BOUND_LIST => &SyntaxInfo { name: "TYPE_BOUND_LIST" },
+            MACRO_ITEMS => &SyntaxInfo { name: "MACRO_ITEMS" },
+            MACRO_STMTS => &SyntaxInfo { name: "MACRO_STMTS" },
             TOMBSTONE => &SyntaxInfo { name: "TOMBSTONE" },
             EOF => &SyntaxInfo { name: "EOF" },
             __LAST => &SyntaxInfo { name: "__LAST" },

crates/ra_syntax/src/ast/generated.rs

@@ -1770,6 +1770,72 @@ impl MacroCall {
     }
 }
 
+// MacroItems
+#[derive(Debug, PartialEq, Eq, Hash)]
+#[repr(transparent)]
+pub struct MacroItems {
+    pub(crate) syntax: SyntaxNode,
+}
+unsafe impl TransparentNewType for MacroItems {
+    type Repr = rowan::SyntaxNode;
+}
+
+impl AstNode for MacroItems {
+    fn cast(syntax: &SyntaxNode) -> Option<&Self> {
+        match syntax.kind() {
+            MACRO_ITEMS => Some(MacroItems::from_repr(syntax.into_repr())),
+            _ => None,
+        }
+    }
+    fn syntax(&self) -> &SyntaxNode { &self.syntax }
+}
+
+impl ToOwned for MacroItems {
+    type Owned = TreeArc<MacroItems>;
+    fn to_owned(&self) -> TreeArc<MacroItems> { TreeArc::cast(self.syntax.to_owned()) }
+}
+
+
+impl ast::ModuleItemOwner for MacroItems {}
+impl ast::FnDefOwner for MacroItems {}
+impl MacroItems {}
+
+// MacroStmts
+#[derive(Debug, PartialEq, Eq, Hash)]
+#[repr(transparent)]
+pub struct MacroStmts {
+    pub(crate) syntax: SyntaxNode,
+}
+unsafe impl TransparentNewType for MacroStmts {
+    type Repr = rowan::SyntaxNode;
+}
+
+impl AstNode for MacroStmts {
+    fn cast(syntax: &SyntaxNode) -> Option<&Self> {
+        match syntax.kind() {
+            MACRO_STMTS => Some(MacroStmts::from_repr(syntax.into_repr())),
+            _ => None,
+        }
+    }
+    fn syntax(&self) -> &SyntaxNode { &self.syntax }
+}
+
+impl ToOwned for MacroStmts {
+    type Owned = TreeArc<MacroStmts>;
+    fn to_owned(&self) -> TreeArc<MacroStmts> { TreeArc::cast(self.syntax.to_owned()) }
+}
+
+
+impl MacroStmts {
+    pub fn statements(&self) -> impl Iterator<Item = &Stmt> {
+        super::children(self)
+    }
+
+    pub fn expr(&self) -> Option<&Expr> {
+        super::child_opt(self)
+    }
+}
+
 // MatchArm
 #[derive(Debug, PartialEq, Eq, Hash)]
 #[repr(transparent)]

crates/ra_syntax/src/grammar.ron

@@ -247,6 +247,10 @@ Grammar(
         "ARG_LIST",
         "TYPE_BOUND",
         "TYPE_BOUND_LIST",
+
+        // macro related
+        "MACRO_ITEMS",
+        "MACRO_STMTS",
     ],
     ast: {
         "SourceFile": (
@@ -668,5 +672,16 @@ Grammar(
         "TypeArg": (options: ["TypeRef"]),
         "AssocTypeArg": (options: ["NameRef", "TypeRef"]),
         "LifetimeArg": (),
+
+        "MacroItems": (
+            traits: [ "ModuleItemOwner", "FnDefOwner" ],            
+        ),
+
+        "MacroStmts" : (
+            options: [ "Expr" ],
+            collections: [
+                ["statements", "Stmt"],
+            ],
+        )
     },
 )