build.sh

#!/usr/bin/bash

pegasus < parser.grammar > parser.json
pegasus-crystalsem -l parser.json -a parser.sem -s parser.cr

crystal test.cr

parser.grammar

token whitespace = /[ \n\t]+/ [ skip ];
token SLASH = /\//;
token LITERAL = /[A-Za-z0-9_]+/;
token DOT = /\./;
token SYMBOL = /:[A-Za-z0-9_]+/;
token STAR = /\*[A-Za-z0-9_]+/;
token LPAREN = /\(/;
token RPAREN = /\)/;
token OR = /\|/;

rule S = expressions;
rule expressions = expressions expression | expression | or;
rule expression = terminal | group | STAR;
rule terminal = SYMBOL | SLASH | LITERAL | DOT;
rule group = LPAREN expressions RPAREN;
rule or = expressions OR expression;

parser.sem

type token = $$ Token $$

typerules token = [S, expressions, expression, terminal, group, or]

init = $$ $$

rule S(0) = $$ $out = $0 $$
rule expressions(0) = $$ $out = $0 $$
rule expressions(1) = $$ $out = $0 $$
rule expressions(2) = $$ $out = $0 $$
rule expression(0) = $$ $out = $0 $$
rule expression(1) = $$ $out = $0 $$
rule expression(2) = $$ $out = $0 $$
rule terminal(0) = $$ $out = $0 $$
rule terminal(1) = $$ $out = $0 $$
rule terminal(2) = $$ $out = $0 $$
rule terminal(3) = $$ $out = $0 $$
rule group(0) = $$ $out = $0 $$
rule or(0) = $$ $out = $0 $$

test.cr

require "./parser.cr"

def print_tree(tree, indent = 0)
  indent.times { STDOUT << "  " }
  case tree
  when Pegasus::Generated::TerminalTree
    STDOUT << "Terminal: "
    STDOUT.puts tree.string
  when Pegasus::Generated::NonterminalTree
    STDOUT << "Nonterminal: " << tree.name
    STDOUT.puts
    tree.children.each { |it| print_tree(it, indent + 1) }
  end
end

paths = [
  "/path/to/some.html",
  "/path/to/some|other.html",
  "/path/to/some|other.(:format)",
  "/path/to(/track/(/*path))",
]

paths.each do |path|
  puts "=" * 50
  puts path
  puts "-" * 50
  # print_tree(Pegasus::Generated.process(path))
  puts Pegasus::Generated.process(path).inspect
  puts "=" * 50
end