//! FIXME: write short doc here use ra_parser::{ FragmentKind::{self, *}, ParseError, TreeSink, }; use ra_syntax::{ ast, AstNode, AstToken, NodeOrToken, Parse, SmolStr, SyntaxKind, SyntaxKind::*, SyntaxNode, SyntaxTreeBuilder, TextRange, TextUnit, T, }; use tt::buffer::{Cursor, TokenBuffer}; use crate::subtree_source::SubtreeTokenSource; use crate::ExpandError; use std::sync::atomic::{AtomicU32, Ordering}; /// Maps `tt::TokenId` to the relative range of the original token. pub struct TokenMap { /// Maps `tt::TokenId` to the *relative* source range. tokens: Vec, map_id: u32, } static TOKEN_MAP_COUNTER: AtomicU32 = AtomicU32::new(0); /// Generate an unique token map id for each instance fn make_uniq_token_map_id() -> u32 { TOKEN_MAP_COUNTER.fetch_add(1, Ordering::SeqCst) } impl std::default::Default for TokenMap { fn default() -> TokenMap { TokenMap { tokens: Default::default(), map_id: make_uniq_token_map_id() } } } /// Convert the syntax tree (what user has written) to a `TokenTree` (what macro /// will consume). pub fn ast_to_token_tree(ast: &ast::TokenTree) -> Option<(tt::Subtree, TokenMap)> { let mut token_map = TokenMap::default(); let node = ast.syntax(); let tt = convert_tt(&mut token_map, node.text_range().start(), node)?; Some((tt, token_map)) } /// Convert the syntax node to a `TokenTree` (what macro /// will consume). pub fn syntax_node_to_token_tree(node: &SyntaxNode) -> Option<(tt::Subtree, TokenMap)> { let mut token_map = TokenMap::default(); let tt = convert_tt(&mut token_map, node.text_range().start(), node)?; 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) -> token_tree_to_expr // * Pat(P) -> token_tree_to_pat // * Ty(P) -> token_tree_to_ty // * Stmts(SmallVec<[ast::Stmt; 1]>) -> token_tree_to_stmts // * Items(SmallVec<[P; 1]>) -> token_tree_to_items // // * TraitItems(SmallVec<[ast::TraitItem; 1]>) // * ImplItems(SmallVec<[ast::ImplItem; 1]>) // * ForeignItems(SmallVec<[ast::ForeignItem; 1]> fn fragment_to_syntax_node( tt: &tt::Subtree, fragment_kind: FragmentKind, ) -> Result, ExpandError> { let tmp; let tokens = match tt { tt::Subtree { delimiter: tt::Delimiter::None, token_trees } => token_trees.as_slice(), _ => { tmp = [tt.clone().into()]; &tmp[..] } }; let buffer = TokenBuffer::new(&tokens); let mut token_source = SubtreeTokenSource::new(&buffer); let mut tree_sink = TtTreeSink::new(buffer.begin()); ra_parser::parse_fragment(&mut token_source, &mut tree_sink, fragment_kind); if tree_sink.roots.len() != 1 { return Err(ExpandError::ConversionError); } //FIXME: would be cool to report errors let parse = tree_sink.inner.finish(); Ok(parse) } /// Parses the token tree (result of macro expansion) to an expression pub fn token_tree_to_expr(tt: &tt::Subtree) -> Result, ExpandError> { let parse = fragment_to_syntax_node(tt, Expr)?; parse.cast().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, ExpandError> { let parse = fragment_to_syntax_node(tt, Pattern)?; parse.cast().ok_or_else(|| crate::ExpandError::ConversionError) } /// Parses the token tree (result of macro expansion) to a Type pub fn token_tree_to_ty(tt: &tt::Subtree) -> Result, ExpandError> { let parse = fragment_to_syntax_node(tt, Type)?; parse.cast().ok_or_else(|| crate::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, ExpandError> { let parse = fragment_to_syntax_node(tt, Statements)?; parse.cast().ok_or_else(|| crate::ExpandError::ConversionError) } /// Parses the token tree (result of macro expansion) as a sequence of items pub fn token_tree_to_items(tt: &tt::Subtree) -> Result, ExpandError> { let parse = fragment_to_syntax_node(tt, Items)?; parse.cast().ok_or_else(|| crate::ExpandError::ConversionError) } impl TokenMap { pub fn relative_range_of(&self, tt: tt::TokenId) -> Option { if self.map_id != tt.map_id() { return None; } let idx = tt.token_id() as usize; self.tokens.get(idx).copied() } fn alloc(&mut self, relative_range: TextRange) -> tt::TokenId { let id = self.tokens.len(); self.tokens.push(relative_range); tt::TokenId::new(id as u32, self.map_id) } } /// Returns the textual content of a doc comment block as a quoted string /// That is, strips leading `///` (or `/**`, etc) /// and strips the ending `*/` /// And then quote the string, which is needed to convert to `tt::Literal` fn doc_comment_text(comment: &ast::Comment) -> SmolStr { let prefix_len = comment.prefix().len(); let mut text = &comment.text()[prefix_len..]; // Remove ending "*/" if comment.kind().shape == ast::CommentShape::Block { text = &text[0..text.len() - 2]; } // Quote the string // Note that `tt::Literal` expect an escaped string let text = format!("{:?}", text.escape_default().to_string()); text.into() } fn convert_doc_comment(token: &ra_syntax::SyntaxToken) -> Option> { let comment = ast::Comment::cast(token.clone())?; let doc = comment.kind().doc?; // Make `doc="\" Comments\"" let mut meta_tkns = Vec::new(); meta_tkns.push(mk_ident("doc")); meta_tkns.push(mk_punct('=')); meta_tkns.push(mk_doc_literal(&comment)); // Make `#![]` let mut token_trees = Vec::new(); token_trees.push(mk_punct('#')); if let ast::CommentPlacement::Inner = doc { token_trees.push(mk_punct('!')); } token_trees.push(tt::TokenTree::from(tt::Subtree { delimiter: tt::Delimiter::Bracket, token_trees: meta_tkns, })); return Some(token_trees); // Helper functions fn mk_ident(s: &str) -> tt::TokenTree { tt::TokenTree::from(tt::Leaf::from(tt::Ident { text: s.into(), id: tt::TokenId::unspecified(), })) } fn mk_punct(c: char) -> tt::TokenTree { tt::TokenTree::from(tt::Leaf::from(tt::Punct { char: c, spacing: tt::Spacing::Alone })) } fn mk_doc_literal(comment: &ast::Comment) -> tt::TokenTree { let lit = tt::Literal { text: doc_comment_text(comment) }; tt::TokenTree::from(tt::Leaf::from(lit)) } } fn convert_tt( token_map: &mut TokenMap, global_offset: TextUnit, tt: &SyntaxNode, ) -> Option { // This tree is empty if tt.first_child_or_token().is_none() { return Some(tt::Subtree { token_trees: vec![], delimiter: tt::Delimiter::None }); } let first_child = tt.first_child_or_token()?; let last_child = tt.last_child_or_token()?; let (delimiter, skip_first) = match (first_child.kind(), last_child.kind()) { (T!['('], T![')']) => (tt::Delimiter::Parenthesis, true), (T!['{'], T!['}']) => (tt::Delimiter::Brace, true), (T!['['], T![']']) => (tt::Delimiter::Bracket, true), _ => (tt::Delimiter::None, false), }; let mut token_trees = Vec::new(); let mut child_iter = tt.children_with_tokens().skip(skip_first as usize).peekable(); while let Some(child) = child_iter.next() { if skip_first && (child == first_child || child == last_child) { continue; } match child { NodeOrToken::Token(token) => { if let Some(doc_tokens) = convert_doc_comment(&token) { token_trees.extend(doc_tokens); } else if token.kind().is_trivia() { continue; } else if token.kind().is_punct() { assert!(token.text().len() == 1, "Input ast::token punct must be single char."); let char = token.text().chars().next().unwrap(); let spacing = match child_iter.peek() { Some(NodeOrToken::Token(token)) => { if token.kind().is_punct() { tt::Spacing::Joint } else { tt::Spacing::Alone } } _ => tt::Spacing::Alone, }; token_trees.push(tt::Leaf::from(tt::Punct { char, spacing }).into()); } else { let child: tt::TokenTree = if token.kind() == T![true] || token.kind() == T![false] { tt::Leaf::from(tt::Literal { text: token.text().clone() }).into() } else if token.kind().is_keyword() || token.kind() == IDENT || token.kind() == LIFETIME { let relative_range = token.text_range() - global_offset; let id = token_map.alloc(relative_range); let text = token.text().clone(); tt::Leaf::from(tt::Ident { text, id }).into() } else if token.kind().is_literal() { tt::Leaf::from(tt::Literal { text: token.text().clone() }).into() } else { return None; }; token_trees.push(child); } } NodeOrToken::Node(node) => { let child = convert_tt(token_map, global_offset, &node)?.into(); token_trees.push(child); } }; } let res = tt::Subtree { delimiter, token_trees }; Some(res) } struct TtTreeSink<'a> { buf: String, cursor: Cursor<'a>, text_pos: TextUnit, inner: SyntaxTreeBuilder, // Number of roots // Use for detect ill-form tree which is not single root roots: smallvec::SmallVec<[usize; 1]>, } impl<'a> TtTreeSink<'a> { fn new(cursor: Cursor<'a>) -> Self { TtTreeSink { buf: String::new(), cursor, text_pos: 0.into(), inner: SyntaxTreeBuilder::default(), roots: smallvec::SmallVec::new(), } } } fn delim_to_str(d: tt::Delimiter, closing: bool) -> SmolStr { let texts = match d { tt::Delimiter::Parenthesis => "()", tt::Delimiter::Brace => "{}", tt::Delimiter::Bracket => "[]", tt::Delimiter::None => "", }; let idx = closing as usize; let text = if !texts.is_empty() { &texts[idx..texts.len() - (1 - idx)] } else { "" }; text.into() } impl<'a> TreeSink for TtTreeSink<'a> { fn token(&mut self, kind: SyntaxKind, n_tokens: u8) { if kind == L_DOLLAR || kind == R_DOLLAR { self.cursor = self.cursor.bump_subtree(); return; } for _ in 0..n_tokens { if self.cursor.eof() { break; } match self.cursor.token_tree() { Some(tt::TokenTree::Leaf(leaf)) => { self.cursor = self.cursor.bump(); self.buf += &format!("{}", leaf); } Some(tt::TokenTree::Subtree(subtree)) => { self.cursor = self.cursor.subtree().unwrap(); self.buf += &delim_to_str(subtree.delimiter, false); } None => { if let Some(parent) = self.cursor.end() { self.cursor = self.cursor.bump(); self.buf += &delim_to_str(parent.delimiter, true); } } }; } self.text_pos += TextUnit::of_str(&self.buf); let text = SmolStr::new(self.buf.as_str()); self.buf.clear(); self.inner.token(kind, text); // Add whitespace between adjoint puncts let next = self.cursor.bump(); if let ( Some(tt::TokenTree::Leaf(tt::Leaf::Punct(curr))), Some(tt::TokenTree::Leaf(tt::Leaf::Punct(_))), ) = (self.cursor.token_tree(), next.token_tree()) { if curr.spacing == tt::Spacing::Alone { self.inner.token(WHITESPACE, " ".into()); } } } fn start_node(&mut self, kind: SyntaxKind) { self.inner.start_node(kind); match self.roots.last_mut() { None | Some(0) => self.roots.push(1), Some(ref mut n) => **n += 1, }; } fn finish_node(&mut self) { self.inner.finish_node(); *self.roots.last_mut().unwrap() -= 1; } fn error(&mut self, error: ParseError) { self.inner.error(error, self.text_pos) } } #[cfg(test)] mod tests { use super::*; use crate::tests::{create_rules, expand}; use ra_parser::TokenSource; #[test] fn convert_tt_token_source() { let rules = create_rules( r#" macro_rules! literals { ($i:ident) => { { let a = 'c'; let c = 1000; let f = 12E+99_f64; let s = "rust1"; } } } "#, ); let expansion = expand(&rules, "literals!(foo);"); let tts = &[expansion.into()]; let buffer = tt::buffer::TokenBuffer::new(tts); let mut tt_src = SubtreeTokenSource::new(&buffer); let mut tokens = vec![]; while tt_src.current().kind != EOF { tokens.push((tt_src.current().kind, tt_src.text())); tt_src.bump(); } // [${] // [let] [a] [=] ['c'] [;] assert_eq!(tokens[2 + 3].1, "'c'"); assert_eq!(tokens[2 + 3].0, CHAR); // [let] [c] [=] [1000] [;] assert_eq!(tokens[2 + 5 + 3].1, "1000"); assert_eq!(tokens[2 + 5 + 3].0, INT_NUMBER); // [let] [f] [=] [12E+99_f64] [;] assert_eq!(tokens[2 + 10 + 3].1, "12E+99_f64"); assert_eq!(tokens[2 + 10 + 3].0, FLOAT_NUMBER); // [let] [s] [=] ["rust1"] [;] assert_eq!(tokens[2 + 15 + 3].1, "\"rust1\""); assert_eq!(tokens[2 + 15 + 3].0, STRING); } #[test] fn stmts_token_trees_to_expr_is_err() { let rules = create_rules( r#" macro_rules! stmts { () => { let a = 0; let b = 0; let c = 0; let d = 0; } } "#, ); let expansion = expand(&rules, "stmts!();"); assert!(token_tree_to_expr(&expansion).is_err()); } }