1 // Copyright 2012-2014 The Rust Project Developers. See the COPYRIGHT
2 // file at the top-level directory of this distribution and at
3 // http://rust-lang.org/COPYRIGHT.
5 // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
6 // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
7 // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
8 // option. This file may not be copied, modified, or distributed
9 // except according to those terms.
11 //! This is an Earley-like parser, without support for in-grammar nonterminals,
12 //! only by calling out to the main rust parser for named nonterminals (which it
13 //! commits to fully when it hits one in a grammar). This means that there are no
14 //! completer or predictor rules, and therefore no need to store one column per
15 //! token: instead, there's a set of current Earley items and a set of next
16 //! ones. Instead of NTs, we have a special case for Kleene star. The big-O, in
17 //! pathological cases, is worse than traditional Earley parsing, but it's an
18 //! easier fit for Macro-by-Example-style rules, and I think the overhead is
19 //! lower. (In order to prevent the pathological case, we'd need to lazily
20 //! construct the resulting `NamedMatch`es at the very end. It'd be a pain,
21 //! and require more memory to keep around old items, but it would also save
24 //! Quick intro to how the parser works:
26 //! A 'position' is a dot in the middle of a matcher, usually represented as a
27 //! dot. For example `· a $( a )* a b` is a position, as is `a $( · a )* a b`.
29 //! The parser walks through the input a character at a time, maintaining a list
30 //! of items consistent with the current position in the input string: `cur_eis`.
32 //! As it processes them, it fills up `eof_eis` with items that would be valid if
33 //! the macro invocation is now over, `bb_eis` with items that are waiting on
34 //! a Rust nonterminal like `$e:expr`, and `next_eis` with items that are waiting
35 //! on a particular token. Most of the logic concerns moving the · through the
36 //! repetitions indicated by Kleene stars. It only advances or calls out to the
37 //! real Rust parser when no `cur_eis` items remain
42 //! Start parsing a a a a b against [· a $( a )* a b].
44 //! Remaining input: a a a a b
45 //! next_eis: [· a $( a )* a b]
47 //! - - - Advance over an a. - - -
49 //! Remaining input: a a a b
50 //! cur: [a · $( a )* a b]
51 //! Descend/Skip (first item).
52 //! next: [a $( · a )* a b] [a $( a )* · a b].
54 //! - - - Advance over an a. - - -
56 //! Remaining input: a a b
57 //! cur: [a $( a · )* a b] next: [a $( a )* a · b]
58 //! Finish/Repeat (first item)
59 //! next: [a $( a )* · a b] [a $( · a )* a b] [a $( a )* a · b]
61 //! - - - Advance over an a. - - - (this looks exactly like the last step)
63 //! Remaining input: a b
64 //! cur: [a $( a · )* a b] next: [a $( a )* a · b]
65 //! Finish/Repeat (first item)
66 //! next: [a $( a )* · a b] [a $( · a )* a b] [a $( a )* a · b]
68 //! - - - Advance over an a. - - - (this looks exactly like the last step)
70 //! Remaining input: b
71 //! cur: [a $( a · )* a b] next: [a $( a )* a · b]
72 //! Finish/Repeat (first item)
73 //! next: [a $( a )* · a b] [a $( · a )* a b]
75 //! - - - Advance over a b. - - -
77 //! Remaining input: ''
78 //! eof: [a $( a )* a b ·]
81 pub use self::NamedMatch::*;
82 pub use self::ParseResult::*;
83 use self::TokenTreeOrTokenTreeVec::*;
86 use syntax_pos::{self, BytePos, Span};
88 use errors::FatalError;
89 use ext::tt::quoted::{self, TokenTree};
90 use parse::{Directory, ParseSess};
91 use parse::parser::{PathStyle, Parser};
92 use parse::token::{self, DocComment, Token, Nonterminal};
95 use tokenstream::TokenStream;
96 use util::small_vector::SmallVector;
100 use std::collections::HashMap;
101 use std::collections::hash_map::Entry::{Vacant, Occupied};
103 // To avoid costly uniqueness checks, we require that `MatchSeq` always has
107 enum TokenTreeOrTokenTreeVec {
109 TtSeq(Vec<TokenTree>),
112 impl TokenTreeOrTokenTreeVec {
113 fn len(&self) -> usize {
115 TtSeq(ref v) => v.len(),
116 Tt(ref tt) => tt.len(),
120 fn get_tt(&self, index: usize) -> TokenTree {
122 TtSeq(ref v) => v[index].clone(),
123 Tt(ref tt) => tt.get_tt(index),
128 /// an unzipping of `TokenTree`s
130 struct MatcherTtFrame {
131 elts: TokenTreeOrTokenTreeVec,
137 stack: Vec<MatcherTtFrame>,
138 top_elts: TokenTreeOrTokenTreeVec,
141 up: Option<Box<MatcherPos>>,
142 matches: Vec<Rc<Vec<NamedMatch>>>,
150 fn push_match(&mut self, idx: usize, m: NamedMatch) {
151 let matches = Rc::make_mut(&mut self.matches[idx]);
156 pub type NamedParseResult = ParseResult<HashMap<Ident, Rc<NamedMatch>>>;
158 pub fn count_names(ms: &[TokenTree]) -> usize {
159 ms.iter().fold(0, |count, elt| {
161 TokenTree::Sequence(_, ref seq) => seq.num_captures,
162 TokenTree::Delimited(_, ref delim) => count_names(&delim.tts),
163 TokenTree::MetaVar(..) => 0,
164 TokenTree::MetaVarDecl(..) => 1,
165 TokenTree::Token(..) => 0,
170 fn initial_matcher_pos(ms: Vec<TokenTree>, lo: BytePos) -> Box<MatcherPos> {
171 let match_idx_hi = count_names(&ms[..]);
172 let matches = create_matches(match_idx_hi);
173 Box::new(MatcherPos {
182 match_hi: match_idx_hi,
187 /// `NamedMatch` is a pattern-match result for a single `token::MATCH_NONTERMINAL`:
188 /// so it is associated with a single ident in a parse, and all
189 /// `MatchedNonterminal`s in the `NamedMatch` have the same nonterminal type
190 /// (expr, item, etc). Each leaf in a single `NamedMatch` corresponds to a
191 /// single `token::MATCH_NONTERMINAL` in the `TokenTree` that produced it.
193 /// The in-memory structure of a particular `NamedMatch` represents the match
194 /// that occurred when a particular subset of a matcher was applied to a
195 /// particular token tree.
197 /// The width of each `MatchedSeq` in the `NamedMatch`, and the identity of
198 /// the `MatchedNonterminal`s, will depend on the token tree it was applied
199 /// to: each `MatchedSeq` corresponds to a single `TTSeq` in the originating
200 /// token tree. The depth of the `NamedMatch` structure will therefore depend
201 /// only on the nesting depth of `ast::TTSeq`s in the originating
202 /// token tree it was derived from.
204 #[derive(Debug, Clone)]
205 pub enum NamedMatch {
206 MatchedSeq(Rc<Vec<NamedMatch>>, syntax_pos::Span),
207 MatchedNonterminal(Rc<Nonterminal>)
210 fn nameize<I: Iterator<Item=NamedMatch>>(sess: &ParseSess, ms: &[TokenTree], mut res: I)
211 -> NamedParseResult {
212 fn n_rec<I: Iterator<Item=NamedMatch>>(sess: &ParseSess, m: &TokenTree, mut res: &mut I,
213 ret_val: &mut HashMap<Ident, Rc<NamedMatch>>)
214 -> Result<(), (syntax_pos::Span, String)> {
216 TokenTree::Sequence(_, ref seq) => {
217 for next_m in &seq.tts {
218 n_rec(sess, next_m, res.by_ref(), ret_val)?
221 TokenTree::Delimited(_, ref delim) => {
222 for next_m in &delim.tts {
223 n_rec(sess, next_m, res.by_ref(), ret_val)?;
226 TokenTree::MetaVarDecl(span, _, id) if id.name == keywords::Invalid.name() => {
227 if sess.missing_fragment_specifiers.borrow_mut().remove(&span) {
228 return Err((span, "missing fragment specifier".to_string()));
231 TokenTree::MetaVarDecl(sp, bind_name, _) => {
232 match ret_val.entry(bind_name) {
234 // FIXME(simulacrum): Don't construct Rc here
235 spot.insert(Rc::new(res.next().unwrap()));
238 return Err((sp, format!("duplicated bind name: {}", bind_name)))
242 TokenTree::MetaVar(..) | TokenTree::Token(..) => (),
248 let mut ret_val = HashMap::new();
250 match n_rec(sess, m, res.by_ref(), &mut ret_val) {
252 Err((sp, msg)) => return Error(sp, msg),
259 pub enum ParseResult<T> {
261 /// Arm failed to match. If the second parameter is `token::Eof`, it
262 /// indicates an unexpected end of macro invocation. Otherwise, it
263 /// indicates that no rules expected the given token.
264 Failure(syntax_pos::Span, Token),
265 /// Fatal error (malformed macro?). Abort compilation.
266 Error(syntax_pos::Span, String)
269 pub fn parse_failure_msg(tok: Token) -> String {
271 token::Eof => "unexpected end of macro invocation".to_string(),
272 _ => format!("no rules expected the token `{}`", pprust::token_to_string(&tok)),
276 /// Perform a token equality check, ignoring syntax context (that is, an unhygienic comparison)
277 fn token_name_eq(t1 : &Token, t2 : &Token) -> bool {
278 if let (Some(id1), Some(id2)) = (t1.ident(), t2.ident()) {
280 } else if let (&token::Lifetime(id1), &token::Lifetime(id2)) = (t1, t2) {
287 fn create_matches(len: usize) -> Vec<Rc<Vec<NamedMatch>>> {
288 (0..len).into_iter().map(|_| Rc::new(Vec::new())).collect()
291 fn inner_parse_loop(sess: &ParseSess,
292 cur_eis: &mut SmallVector<Box<MatcherPos>>,
293 next_eis: &mut Vec<Box<MatcherPos>>,
294 eof_eis: &mut SmallVector<Box<MatcherPos>>,
295 bb_eis: &mut SmallVector<Box<MatcherPos>>,
297 span: syntax_pos::Span)
299 while let Some(mut ei) = cur_eis.pop() {
300 // When unzipped trees end, remove them
301 while ei.idx >= ei.top_elts.len() {
302 match ei.stack.pop() {
303 Some(MatcherTtFrame { elts, idx }) => {
312 let len = ei.top_elts.len();
314 // at end of sequence
316 // We are repeating iff there is a parent
318 // Disregarding the separator, add the "up" case to the tokens that should be
320 // (remove this condition to make trailing seps ok)
322 let mut new_pos = ei.up.clone().unwrap();
324 // update matches (the MBE "parse tree") by appending
325 // each tree as a subtree.
327 // Only touch the binders we have actually bound
328 for idx in ei.match_lo..ei.match_hi {
329 let sub = ei.matches[idx].clone();
330 new_pos.push_match(idx, MatchedSeq(sub, Span { lo: ei.sp_lo, ..span }));
333 new_pos.match_cur = ei.match_hi;
335 cur_eis.push(new_pos);
338 // Check if we need a separator
339 if idx == len && ei.sep.is_some() {
340 // We have a separator, and it is the current token.
341 if ei.sep.as_ref().map(|sep| token_name_eq(token, sep)).unwrap_or(false) {
345 } else { // we don't need a separator
346 ei.match_cur = ei.match_lo;
351 // We aren't repeating, so we must be potentially at the end of the input.
355 match ei.top_elts.get_tt(idx) {
356 /* need to descend into sequence */
357 TokenTree::Sequence(sp, seq) => {
358 if seq.op == quoted::KleeneOp::ZeroOrMore {
359 // Examine the case where there are 0 matches of this sequence
360 let mut new_ei = ei.clone();
361 new_ei.match_cur += seq.num_captures;
363 for idx in ei.match_cur..ei.match_cur + seq.num_captures {
364 new_ei.push_match(idx, MatchedSeq(Rc::new(vec![]), sp));
366 cur_eis.push(new_ei);
369 // Examine the case where there is at least one match of this sequence
370 let matches = create_matches(ei.matches.len());
371 cur_eis.push(Box::new(MatcherPos {
373 sep: seq.separator.clone(),
376 match_lo: ei.match_cur,
377 match_cur: ei.match_cur,
378 match_hi: ei.match_cur + seq.num_captures,
381 top_elts: Tt(TokenTree::Sequence(sp, seq)),
384 TokenTree::MetaVarDecl(span, _, id) if id.name == keywords::Invalid.name() => {
385 if sess.missing_fragment_specifiers.borrow_mut().remove(&span) {
386 return Error(span, "missing fragment specifier".to_string());
389 TokenTree::MetaVarDecl(_, _, id) => {
390 // Built-in nonterminals never start with these tokens,
391 // so we can eliminate them from consideration.
392 if may_begin_with(&*id.name.as_str(), token) {
396 seq @ TokenTree::Delimited(..) | seq @ TokenTree::Token(_, DocComment(..)) => {
397 let lower_elts = mem::replace(&mut ei.top_elts, Tt(seq));
399 ei.stack.push(MatcherTtFrame {
406 TokenTree::Token(_, ref t) if token_name_eq(t, token) => {
410 TokenTree::Token(..) | TokenTree::MetaVar(..) => {}
418 pub fn parse(sess: &ParseSess,
421 directory: Option<Directory>,
422 recurse_into_modules: bool)
423 -> NamedParseResult {
424 let mut parser = Parser::new(sess, tts, directory, recurse_into_modules, true);
425 let mut cur_eis = SmallVector::one(initial_matcher_pos(ms.to_owned(), parser.span.lo));
426 let mut next_eis = Vec::new(); // or proceed normally
429 let mut bb_eis = SmallVector::new(); // black-box parsed by parser.rs
430 let mut eof_eis = SmallVector::new();
431 assert!(next_eis.is_empty());
433 match inner_parse_loop(sess, &mut cur_eis, &mut next_eis, &mut eof_eis, &mut bb_eis,
434 &parser.token, parser.span) {
436 Failure(sp, tok) => return Failure(sp, tok),
437 Error(sp, msg) => return Error(sp, msg),
440 // inner parse loop handled all cur_eis, so it's empty
441 assert!(cur_eis.is_empty());
443 /* error messages here could be improved with links to orig. rules */
444 if token_name_eq(&parser.token, &token::Eof) {
445 if eof_eis.len() == 1 {
446 let matches = eof_eis[0].matches.iter_mut().map(|mut dv| {
447 Rc::make_mut(dv).pop().unwrap()
449 return nameize(sess, ms, matches);
450 } else if eof_eis.len() > 1 {
451 return Error(parser.span, "ambiguity: multiple successful parses".to_string());
453 return Failure(parser.span, token::Eof);
455 } else if (!bb_eis.is_empty() && !next_eis.is_empty()) || bb_eis.len() > 1 {
456 let nts = bb_eis.iter().map(|ei| match ei.top_elts.get_tt(ei.idx) {
457 TokenTree::MetaVarDecl(_, bind, name) => {
458 format!("{} ('{}')", name, bind)
461 }).collect::<Vec<String>>().join(" or ");
463 return Error(parser.span, format!(
464 "local ambiguity: multiple parsing options: {}",
465 match next_eis.len() {
466 0 => format!("built-in NTs {}.", nts),
467 1 => format!("built-in NTs {} or 1 other option.", nts),
468 n => format!("built-in NTs {} or {} other options.", nts, n),
471 } else if bb_eis.is_empty() && next_eis.is_empty() {
472 return Failure(parser.span, parser.token);
473 } else if !next_eis.is_empty() {
474 /* Now process the next token */
475 cur_eis.extend(next_eis.drain(..));
477 } else /* bb_eis.len() == 1 */ {
478 let mut ei = bb_eis.pop().unwrap();
479 if let TokenTree::MetaVarDecl(span, _, ident) = ei.top_elts.get_tt(ei.idx) {
480 let match_cur = ei.match_cur;
481 ei.push_match(match_cur,
482 MatchedNonterminal(Rc::new(parse_nt(&mut parser, span, &ident.name.as_str()))));
491 assert!(!cur_eis.is_empty());
495 /// Checks whether a non-terminal may begin with a particular token.
497 /// Returning `false` is a *stability guarantee* that such a matcher will *never* begin with that
498 /// token. Be conservative (return true) if not sure.
499 fn may_begin_with(name: &str, token: &Token) -> bool {
500 /// Checks whether the non-terminal may contain a single (non-keyword) identifier.
501 fn may_be_ident(nt: &token::Nonterminal) -> bool {
503 token::NtItem(_) | token::NtBlock(_) | token::NtVis(_) => false,
509 "expr" => token.can_begin_expr(),
510 "ty" => token.can_begin_type(),
511 "ident" => token.is_ident(),
512 "vis" => match *token { // The follow-set of :vis + "priv" keyword + interpolated
513 Token::Comma | Token::Ident(_) | Token::Interpolated(_) => true,
514 _ => token.can_begin_type(),
516 "block" => match *token {
517 Token::OpenDelim(token::Brace) => true,
518 Token::Interpolated(ref nt) => match nt.0 {
525 token::NtVis(_) => false, // none of these may start with '{'.
530 "path" | "meta" => match *token {
531 Token::ModSep | Token::Ident(_) => true,
532 Token::Interpolated(ref nt) => match nt.0 {
533 token::NtPath(_) | token::NtMeta(_) => true,
534 _ => may_be_ident(&nt.0),
538 "pat" => match *token {
539 Token::Ident(_) | // box, ref, mut, and other identifiers (can stricten)
540 Token::OpenDelim(token::Paren) | // tuple pattern
541 Token::OpenDelim(token::Bracket) | // slice pattern
542 Token::BinOp(token::And) | // reference
543 Token::BinOp(token::Minus) | // negative literal
544 Token::AndAnd | // double reference
545 Token::Literal(..) | // literal
546 Token::DotDot | // range pattern (future compat)
547 Token::DotDotDot | // range pattern (future compat)
548 Token::ModSep | // path
549 Token::Lt | // path (UFCS constant)
550 Token::BinOp(token::Shl) | // path (double UFCS)
551 Token::Underscore => true, // placeholder
552 Token::Interpolated(ref nt) => may_be_ident(&nt.0),
556 token::CloseDelim(_) => false,
562 fn parse_nt<'a>(p: &mut Parser<'a>, sp: Span, name: &str) -> Nonterminal {
564 return token::NtTT(p.parse_token_tree());
566 // check at the beginning and the parser checks after each bump
567 p.process_potential_macro_variable();
569 "item" => match panictry!(p.parse_item()) {
570 Some(i) => token::NtItem(i),
572 p.fatal("expected an item keyword").emit();
576 "block" => token::NtBlock(panictry!(p.parse_block())),
577 "stmt" => match panictry!(p.parse_stmt()) {
578 Some(s) => token::NtStmt(s),
580 p.fatal("expected a statement").emit();
584 "pat" => token::NtPat(panictry!(p.parse_pat())),
585 "expr" => token::NtExpr(panictry!(p.parse_expr())),
586 "ty" => token::NtTy(panictry!(p.parse_ty())),
587 // this could be handled like a token, since it is one
588 "ident" => match p.token {
589 token::Ident(sn) => {
591 token::NtIdent(Spanned::<Ident>{node: sn, span: p.prev_span})
594 let token_str = pprust::token_to_string(&p.token);
595 p.fatal(&format!("expected ident, found {}",
596 &token_str[..])).emit();
601 token::NtPath(panictry!(p.parse_path(PathStyle::Type)))
603 "meta" => token::NtMeta(panictry!(p.parse_meta_item())),
604 "vis" => token::NtVis(panictry!(p.parse_visibility(true))),
605 // this is not supposed to happen, since it has been checked
606 // when compiling the macro.
607 _ => p.span_bug(sp, "invalid fragment specifier")