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 | |
10 | use crate::ast::*; |
11 | |
12 | pub 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 | match expr { |
19 | Expr::Choice(lhs, rhs) => match (*lhs, *rhs) { |
20 | (Expr::Seq(l1, r1), Expr::Seq(l2, r2)) => { |
21 | if l1 == l2 { |
22 | Expr::Seq(l1, Box::new(Expr::Choice(r1, r2))) |
23 | } else { |
24 | Expr::Choice(Box::new(Expr::Seq(l1, r1)), Box::new(Expr::Seq(l2, r2))) |
25 | } |
26 | } |
27 | // Converts `(rule ~ rest) | rule` to `rule ~ rest?`, avoiding trying to match `rule` twice. |
28 | // This is only done for atomic rules, because other rule types have implicit whitespaces. |
29 | // FIXME: "desugar" implicit whitespace rules before applying any optimizations |
30 | (Expr::Seq(l1, l2), r) |
31 | if matches!(ty, RuleType::Atomic | RuleType::CompoundAtomic) => |
32 | { |
33 | if *l1 == r { |
34 | Expr::Seq(l1, Box::new(Expr::Opt(l2))) |
35 | } else { |
36 | Expr::Choice(Box::new(Expr::Seq(l1, l2)), Box::new(r)) |
37 | } |
38 | } |
39 | // Converts `rule | (rule ~ rest)` to `rule` since `(rule ~ rest)` |
40 | // will never match if `rule` didn't. |
41 | (l, Expr::Seq(r1, r2)) => { |
42 | if l == *r1 { |
43 | l |
44 | } else { |
45 | Expr::Choice(Box::new(l), Box::new(Expr::Seq(r1, r2))) |
46 | } |
47 | } |
48 | (lhs, rhs) => Expr::Choice(Box::new(lhs), Box::new(rhs)), |
49 | }, |
50 | expr => expr, |
51 | } |
52 | }), |
53 | } |
54 | } |
55 | |