// Copyright 2012 The Rust Project Developers. See the COPYRIGHT // file at the top-level directory of this distribution and at // http://rust-lang.org/COPYRIGHT. // // Licensed under the Apache License, Version 2.0 or the MIT license // , at your // option. This file may not be copied, modified, or distributed // except according to those terms. use self::LockstepIterSize::*; use ast::Ident; use errors::Handler; use ext::tt::macro_parser::{NamedMatch, MatchedSeq, MatchedNonterminal}; use parse::token::{self, MatchNt, SubstNt, Token, NtIdent, NtTT}; use syntax_pos::{Span, DUMMY_SP}; use tokenstream::{self, TokenTree}; use util::small_vector::SmallVector; use std::rc::Rc; use std::ops::Add; use std::collections::HashMap; ///an unzipping of `TokenTree`s #[derive(Clone)] struct TtFrame { forest: TokenTree, idx: usize, dotdotdoted: bool, sep: Option, } #[derive(Clone)] struct TtReader<'a> { sp_diag: &'a Handler, /// the unzipped tree: stack: SmallVector, /* for MBE-style macro transcription */ interpolations: HashMap>, repeat_idx: Vec, repeat_len: Vec, } /// This can do Macro-By-Example transcription. On the other hand, if /// `src` contains no `TokenTree::Sequence`s, `MatchNt`s or `SubstNt`s, `interp` can /// (and should) be None. pub fn transcribe(sp_diag: &Handler, interp: Option>>, src: Vec) -> Vec { let mut r = TtReader { sp_diag: sp_diag, stack: SmallVector::one(TtFrame { forest: TokenTree::Sequence(DUMMY_SP, Rc::new(tokenstream::SequenceRepetition { tts: src, // doesn't matter. This merely holds the root unzipping. separator: None, op: tokenstream::KleeneOp::ZeroOrMore, num_captures: 0 })), idx: 0, dotdotdoted: false, sep: None, }), interpolations: match interp { /* just a convenience */ None => HashMap::new(), Some(x) => x, }, repeat_idx: Vec::new(), repeat_len: Vec::new(), }; let mut tts = Vec::new(); let mut prev_span = DUMMY_SP; while let Some(tt) = tt_next_token(&mut r, prev_span) { prev_span = tt.span(); tts.push(tt); } tts } fn lookup_cur_matched_by_matched(r: &TtReader, start: Rc) -> Rc { r.repeat_idx.iter().fold(start, |ad, idx| { match *ad { MatchedNonterminal(_) => { // end of the line; duplicate henceforth ad.clone() } MatchedSeq(ref ads, _) => ads[*idx].clone() } }) } fn lookup_cur_matched(r: &TtReader, name: Ident) -> Option> { let matched_opt = r.interpolations.get(&name).cloned(); matched_opt.map(|s| lookup_cur_matched_by_matched(r, s)) } #[derive(Clone)] enum LockstepIterSize { LisUnconstrained, LisConstraint(usize, Ident), LisContradiction(String), } impl Add for LockstepIterSize { type Output = LockstepIterSize; fn add(self, other: LockstepIterSize) -> LockstepIterSize { match self { LisUnconstrained => other, LisContradiction(_) => self, LisConstraint(l_len, ref l_id) => match other { LisUnconstrained => self.clone(), LisContradiction(_) => other, LisConstraint(r_len, _) if l_len == r_len => self.clone(), LisConstraint(r_len, r_id) => { LisContradiction(format!("inconsistent lockstep iteration: \ '{}' has {} items, but '{}' has {}", l_id, l_len, r_id, r_len)) } }, } } } fn lockstep_iter_size(t: &TokenTree, r: &TtReader) -> LockstepIterSize { match *t { TokenTree::Delimited(_, ref delimed) => { delimed.tts.iter().fold(LisUnconstrained, |size, tt| { size + lockstep_iter_size(tt, r) }) }, TokenTree::Sequence(_, ref seq) => { seq.tts.iter().fold(LisUnconstrained, |size, tt| { size + lockstep_iter_size(tt, r) }) }, TokenTree::Token(_, SubstNt(name)) | TokenTree::Token(_, MatchNt(name, _)) => match lookup_cur_matched(r, name) { Some(matched) => match *matched { MatchedNonterminal(_) => LisUnconstrained, MatchedSeq(ref ads, _) => LisConstraint(ads.len(), name), }, _ => LisUnconstrained }, TokenTree::Token(..) => LisUnconstrained, } } /// Return the next token from the TtReader. /// EFFECT: advances the reader's token field fn tt_next_token(r: &mut TtReader, prev_span: Span) -> Option { loop { let should_pop = if let Some(frame) = r.stack.last() { if frame.idx < frame.forest.len() { break; } !frame.dotdotdoted || *r.repeat_idx.last().unwrap() == *r.repeat_len.last().unwrap() - 1 } else { return None; }; /* done with this set; pop or repeat? */ if should_pop { let prev = r.stack.pop().unwrap(); if let Some(frame) = r.stack.last_mut() { frame.idx += 1; } else { return None; } if prev.dotdotdoted { r.repeat_idx.pop(); r.repeat_len.pop(); } } else { /* repeat */ *r.repeat_idx.last_mut().unwrap() += 1; r.stack.last_mut().unwrap().idx = 0; if let Some(tk) = r.stack.last().unwrap().sep.clone() { return Some(TokenTree::Token(prev_span, tk)); // repeat same span, I guess } } } loop { /* because it's easiest, this handles `TokenTree::Delimited` not starting with a `TokenTree::Token`, even though it won't happen */ let t = { let frame = r.stack.last().unwrap(); // FIXME(pcwalton): Bad copy. frame.forest.get_tt(frame.idx) }; match t { TokenTree::Sequence(sp, seq) => { // FIXME(pcwalton): Bad copy. match lockstep_iter_size(&TokenTree::Sequence(sp, seq.clone()), r) { LisUnconstrained => { panic!(r.sp_diag.span_fatal( sp.clone(), /* blame macro writer */ "attempted to repeat an expression \ containing no syntax \ variables matched as repeating at this depth")); } LisContradiction(ref msg) => { // FIXME #2887 blame macro invoker instead panic!(r.sp_diag.span_fatal(sp.clone(), &msg[..])); } LisConstraint(len, _) => { if len == 0 { if seq.op == tokenstream::KleeneOp::OneOrMore { // FIXME #2887 blame invoker panic!(r.sp_diag.span_fatal(sp.clone(), "this must repeat at least once")); } r.stack.last_mut().unwrap().idx += 1; return tt_next_token(r, prev_span); } r.repeat_len.push(len); r.repeat_idx.push(0); r.stack.push(TtFrame { idx: 0, dotdotdoted: true, sep: seq.separator.clone(), forest: TokenTree::Sequence(sp, seq), }); } } } // FIXME #2887: think about span stuff here TokenTree::Token(sp, SubstNt(ident)) => { r.stack.last_mut().unwrap().idx += 1; match lookup_cur_matched(r, ident) { None => { return Some(TokenTree::Token(sp, SubstNt(ident))); // this can't be 0 length, just like TokenTree::Delimited } Some(cur_matched) => if let MatchedNonterminal(ref nt) = *cur_matched { match **nt { // sidestep the interpolation tricks for ident because // (a) idents can be in lots of places, so it'd be a pain // (b) we actually can, since it's a token. NtIdent(ref sn) => { return Some(TokenTree::Token(sn.span, token::Ident(sn.node))); } NtTT(ref tt) => return Some(tt.clone()), _ => { // FIXME(pcwalton): Bad copy return Some(TokenTree::Token(sp, token::Interpolated(nt.clone()))); } } } else { panic!(r.sp_diag.span_fatal( sp, /* blame the macro writer */ &format!("variable '{}' is still repeating at this depth", ident))); } } } // TokenTree::Delimited or any token that can be unzipped seq @ TokenTree::Delimited(..) | seq @ TokenTree::Token(_, MatchNt(..)) => { // do not advance the idx yet r.stack.push(TtFrame { forest: seq, idx: 0, dotdotdoted: false, sep: None }); // if this could be 0-length, we'd need to potentially recur here } tt @ TokenTree::Token(..) => { r.stack.last_mut().unwrap().idx += 1; return Some(tt); } } } }