pest_meta/optimizer/factorizer.rs
1// pest. The Elegant Parser
2// Copyright (c) 2018 DragoČ™ Tiselice
3//
4// Licensed under the Apache License, Version 2.0
5// <LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0> or the MIT
6// license <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
7// option. All files in the project carrying such notice may not be copied,
8// modified, or distributed except according to those terms.
9
10use crate::ast::*;
11
12pub fn factor(rule: Rule) -> Rule {
13 let Rule { name, ty, expr } = rule;
14 Rule {
15 name,
16 ty,
17 expr: expr.map_top_down(|expr| {
18 // TODO: Use box syntax when it gets stabilized.
19 match expr {
20 Expr::Choice(lhs, rhs) => match (*lhs, *rhs) {
21 (Expr::Seq(l1, r1), Expr::Seq(l2, r2)) => {
22 if l1 == l2 {
23 Expr::Seq(l1, Box::new(Expr::Choice(r1, r2)))
24 } else {
25 Expr::Choice(Box::new(Expr::Seq(l1, r1)), Box::new(Expr::Seq(l2, r2)))
26 }
27 }
28 // Converts `(rule ~ rest) | rule` to `rule ~ rest?`, avoiding trying to match `rule` twice.
29 // This is only done for atomic rules, because other rule types have implicit whitespaces.
30 // FIXME: "desugar" implicit whitespace rules before applying any optimizations
31 (Expr::Seq(l1, l2), r)
32 if matches!(ty, RuleType::Atomic | RuleType::CompoundAtomic) =>
33 {
34 if *l1 == r {
35 Expr::Seq(l1, Box::new(Expr::Opt(l2)))
36 } else {
37 Expr::Choice(Box::new(Expr::Seq(l1, l2)), Box::new(r))
38 }
39 }
40 // Converts `rule | (rule ~ rest)` to `rule` since `(rule ~ rest)`
41 // will never match if `rule` didn't.
42 (l, Expr::Seq(r1, r2)) => {
43 if l == *r1 {
44 l
45 } else {
46 Expr::Choice(Box::new(l), Box::new(Expr::Seq(r1, r2)))
47 }
48 }
49 (lhs, rhs) => Expr::Choice(Box::new(lhs), Box::new(rhs)),
50 },
51 expr => expr,
52 }
53 }),
54 }
55}