Merge #885

885: Parse token trees directy r=matklad a=matklad

This takes advantage of the recent macro refactoring to directly parse token stream into a syntax tree.

Co-authored-by: Aleksey Kladov <[email protected]>

3 files changed, 207 insertions, 4 deletions

diff --git a/crates/ra_mbe/Cargo.toml b/crates/ra_mbe/Cargo.toml
index e7b8660e7..6e785f570 100644
--- a/crates/ra_mbe/Cargo.toml
+++ b/crates/ra_mbe/Cargo.toml
@@ -6,6 +6,7 @@ authors = ["rust-analyzer developers"]
 
 [dependencies]
 ra_syntax = { path = "../ra_syntax" }
+ra_parser = { path = "../ra_parser" }
 tt = { path = "../ra_tt", package = "ra_tt" }
 
 rustc-hash = "1.0.0"
diff --git a/crates/ra_mbe/src/lib.rs b/crates/ra_mbe/src/lib.rs
index cdca3cafb..907402f5f 100644
--- a/crates/ra_mbe/src/lib.rs
+++ b/crates/ra_mbe/src/lib.rs
@@ -24,7 +24,7 @@ use ra_syntax::SmolStr;
 
 pub use tt::{Delimiter, Punct};
 
-pub use crate::syntax_bridge::ast_to_token_tree;
+pub use crate::syntax_bridge::{ast_to_token_tree, token_tree_to_ast_item_list};
 
 /// 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
@@ -164,14 +164,18 @@ impl_froms!(TokenTree: Leaf, Subtree);
         crate::MacroRules::parse(&definition_tt).unwrap()
     }
 
-    fn assert_expansion(rules: &MacroRules, invocation: &str, expansion: &str) {
+    fn expand(rules: &MacroRules, invocation: &str) -> tt::Subtree {
         let source_file = ast::SourceFile::parse(invocation);
         let macro_invocation =
             source_file.syntax().descendants().find_map(ast::MacroCall::cast).unwrap();
 
         let (invocation_tt, _) = ast_to_token_tree(macro_invocation.token_tree().unwrap()).unwrap();
 
-        let expanded = rules.expand(&invocation_tt).unwrap();
+        rules.expand(&invocation_tt).unwrap()
+    }
+
+    fn assert_expansion(rules: &MacroRules, invocation: &str, expansion: &str) {
+        let expanded = expand(rules, invocation);
         assert_eq!(expanded.to_string(), expansion);
     }
 
@@ -268,4 +272,57 @@ impl_froms!(TokenTree: Leaf, Subtree);
         assert_expansion(&rules, "foo! { Foo,# Bar }", "struct Foo ; struct Bar ;");
     }
 
+    #[test]
+    fn expand_to_item_list() {
+        let rules = create_rules(
+            "
+            macro_rules! structs {
+                ($($i:ident),*) => {
+                    $(struct $i { field: u32 } )*
+                }
+            }
+            ",
+        );
+        let expansion = expand(&rules, "structs!(Foo, Bar)");
+        let tree = token_tree_to_ast_item_list(&expansion);
+        assert_eq!(
+            tree.syntax().debug_dump().trim(),
+            r#"
+SOURCE_FILE@[0; 40)
+  STRUCT_DEF@[0; 20)
+    STRUCT_KW@[0; 6)
+    NAME@[6; 9)
+      IDENT@[6; 9) "Foo"
+    NAMED_FIELD_DEF_LIST@[9; 20)
+      L_CURLY@[9; 10)
+      NAMED_FIELD_DEF@[10; 19)
+        NAME@[10; 15)
+          IDENT@[10; 15) "field"
+        COLON@[15; 16)
+        PATH_TYPE@[16; 19)
+          PATH@[16; 19)
+            PATH_SEGMENT@[16; 19)
+              NAME_REF@[16; 19)
+                IDENT@[16; 19) "u32"
+      R_CURLY@[19; 20)
+  STRUCT_DEF@[20; 40)
+    STRUCT_KW@[20; 26)
+    NAME@[26; 29)
+      IDENT@[26; 29) "Bar"
+    NAMED_FIELD_DEF_LIST@[29; 40)
+      L_CURLY@[29; 30)
+      NAMED_FIELD_DEF@[30; 39)
+        NAME@[30; 35)
+          IDENT@[30; 35) "field"
+        COLON@[35; 36)
+        PATH_TYPE@[36; 39)
+          PATH@[36; 39)
+            PATH_SEGMENT@[36; 39)
+              NAME_REF@[36; 39)
+                IDENT@[36; 39) "u32"
+      R_CURLY@[39; 40)"#
+                .trim()
+        );
+    }
+
 }
diff --git a/crates/ra_mbe/src/syntax_bridge.rs b/crates/ra_mbe/src/syntax_bridge.rs
index 848c785f8..c1472bbe5 100644
--- a/crates/ra_mbe/src/syntax_bridge.rs
+++ b/crates/ra_mbe/src/syntax_bridge.rs
@@ -1,5 +1,6 @@
+use ra_parser::{TokenSource, TreeSink, ParseError};
 use ra_syntax::{
-    AstNode, SyntaxNode, TextRange,
+    AstNode, SyntaxNode, TextRange, SyntaxKind, SmolStr, SyntaxTreeBuilder, TreeArc,
     ast, SyntaxKind::*, TextUnit
 };
 
@@ -19,6 +20,15 @@ pub fn ast_to_token_tree(ast: &ast::TokenTree) -> Option<(tt::Subtree, TokenMap)
     Some((tt, token_map))
 }
 
+/// 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 = TtTokenSource::new(tt);
+    let mut tree_sink = TtTreeSink::new(&token_source.tokens);
+    ra_parser::parse(&token_source, &mut tree_sink);
+    let syntax = tree_sink.inner.finish();
+    ast::SourceFile::cast(&syntax).unwrap().to_owned()
+}
+
 impl TokenMap {
     pub fn relative_range_of(&self, tt: tt::TokenId) -> Option<TextRange> {
         let idx = tt.0 as usize;
@@ -84,3 +94,138 @@ fn convert_tt(
     let res = tt::Subtree { delimiter, token_trees };
     Some(res)
 }
+
+struct TtTokenSource {
+    tokens: Vec<TtToken>,
+}
+
+struct TtToken {
+    kind: SyntaxKind,
+    is_joint_to_next: bool,
+    text: SmolStr,
+}
+
+impl TtTokenSource {
+    fn new(tt: &tt::Subtree) -> TtTokenSource {
+        let mut res = TtTokenSource { tokens: Vec::new() };
+        res.convert_subtree(tt);
+        res
+    }
+    fn convert_subtree(&mut self, sub: &tt::Subtree) {
+        self.push_delim(sub.delimiter, false);
+        sub.token_trees.iter().for_each(|tt| self.convert_tt(tt));
+        self.push_delim(sub.delimiter, true)
+    }
+    fn convert_tt(&mut self, tt: &tt::TokenTree) {
+        match tt {
+            tt::TokenTree::Leaf(leaf) => self.convert_leaf(leaf),
+            tt::TokenTree::Subtree(sub) => self.convert_subtree(sub),
+        }
+    }
+    fn convert_leaf(&mut self, leaf: &tt::Leaf) {
+        let tok = match leaf {
+            tt::Leaf::Literal(l) => TtToken {
+                kind: SyntaxKind::INT_NUMBER, // FIXME
+                is_joint_to_next: false,
+                text: l.text.clone(),
+            },
+            tt::Leaf::Punct(p) => {
+                let kind = match p.char {
+                    // lexer may produce combpund tokens for these ones
+                    '.' => DOT,
+                    ':' => COLON,
+                    '=' => EQ,
+                    '!' => EXCL,
+                    '-' => MINUS,
+                    c => SyntaxKind::from_char(c).unwrap(),
+                };
+                let text = {
+                    let mut buf = [0u8; 4];
+                    let s: &str = p.char.encode_utf8(&mut buf);
+                    SmolStr::new(s)
+                };
+                TtToken { kind, is_joint_to_next: p.spacing == tt::Spacing::Joint, text }
+            }
+            tt::Leaf::Ident(ident) => {
+                let kind = SyntaxKind::from_keyword(ident.text.as_str()).unwrap_or(IDENT);
+                TtToken { kind, is_joint_to_next: false, text: ident.text.clone() }
+            }
+        };
+        self.tokens.push(tok)
+    }
+    fn push_delim(&mut self, d: tt::Delimiter, closing: bool) {
+        let (kinds, texts) = match d {
+            tt::Delimiter::Parenthesis => ([L_PAREN, R_PAREN], "()"),
+            tt::Delimiter::Brace => ([L_CURLY, R_CURLY], "{}"),
+            tt::Delimiter::Bracket => ([L_BRACK, R_BRACK], "[]"),
+            tt::Delimiter::None => return,
+        };
+        let idx = closing as usize;
+        let kind = kinds[idx];
+        let text = &texts[idx..texts.len() - (1 - idx)];
+        let tok = TtToken { kind, is_joint_to_next: false, text: SmolStr::new(text) };
+        self.tokens.push(tok)
+    }
+}
+
+impl TokenSource for TtTokenSource {
+    fn token_kind(&self, pos: usize) -> SyntaxKind {
+        if let Some(tok) = self.tokens.get(pos) {
+            tok.kind
+        } else {
+            SyntaxKind::EOF
+        }
+    }
+    fn is_token_joint_to_next(&self, pos: usize) -> bool {
+        self.tokens[pos].is_joint_to_next
+    }
+    fn is_keyword(&self, pos: usize, kw: &str) -> bool {
+        self.tokens[pos].text == *kw
+    }
+}
+
+#[derive(Default)]
+struct TtTreeSink<'a> {
+    buf: String,
+    tokens: &'a [TtToken],
+    text_pos: TextUnit,
+    token_pos: usize,
+    inner: SyntaxTreeBuilder,
+}
+
+impl<'a> TtTreeSink<'a> {
+    fn new(tokens: &'a [TtToken]) -> TtTreeSink {
+        TtTreeSink {
+            buf: String::new(),
+            tokens,
+            text_pos: 0.into(),
+            token_pos: 0,
+            inner: SyntaxTreeBuilder::default(),
+        }
+    }
+}
+
+impl<'a> TreeSink for TtTreeSink<'a> {
+    fn leaf(&mut self, kind: SyntaxKind, n_tokens: u8) {
+        for _ in 0..n_tokens {
+            self.buf += self.tokens[self.token_pos].text.as_str();
+            self.token_pos += 1;
+        }
+        self.text_pos += TextUnit::of_str(&self.buf);
+        let text = SmolStr::new(self.buf.as_str());
+        self.buf.clear();
+        self.inner.leaf(kind, text)
+    }
+
+    fn start_branch(&mut self, kind: SyntaxKind) {
+        self.inner.start_branch(kind);
+    }
+
+    fn finish_branch(&mut self) {
+        self.inner.finish_branch();
+    }
+
+    fn error(&mut self, error: ParseError) {
+        self.inner.error(error, self.text_pos)
+    }
+}