|
1 | | -// Placeholder: real tests to be added. |
| 1 | +// This file is part of the uutils awk package. |
| 2 | +// |
| 3 | +// For the full copyright and license information, please view the LICENSE |
| 4 | +// files that was distributed with this source code. |
| 5 | + |
| 6 | +use std::io::Write; |
| 7 | + |
| 8 | +use crate::{Context, Extra, Identifier, Lexer, Token}; |
| 9 | +use bumpalo::{ |
| 10 | + Bump, |
| 11 | + collections::{CollectIn, Vec}, |
| 12 | +}; |
| 13 | + |
| 14 | +fn lex<'a>( |
| 15 | + src: &'a [u8], |
| 16 | + arena: &'a Bump, |
| 17 | + posix_strict: bool, |
| 18 | + gnu_strict: bool, |
| 19 | +) -> Vec<'a, Token<'a>> { |
| 20 | + Lexer::with_extras( |
| 21 | + src, |
| 22 | + Extra { |
| 23 | + ctx: Context::AcceptExpression, |
| 24 | + arena, |
| 25 | + posix_strict, |
| 26 | + gnu_strict, |
| 27 | + }, |
| 28 | + ) |
| 29 | + .collect_in::<Result<Vec<_>, _>>(arena) |
| 30 | + .unwrap() |
| 31 | +} |
| 32 | + |
| 33 | +#[test] |
| 34 | +fn lexer_test_newlines_non_posix() { |
| 35 | + let mixed = " \t \n \t\n\n \\\n \t"; |
| 36 | + let arena = Bump::new(); |
| 37 | + let mut str = Vec::new_in(&arena); |
| 38 | + for tok in ["BEGIN", "{", "else", "do", "&&", "||", "?", ":", ","] { |
| 39 | + write!(str, "{tok}{mixed}").unwrap(); |
| 40 | + } |
| 41 | + str.push(b'}'); |
| 42 | + assert_eq!( |
| 43 | + &lex(&str, &arena, false, false), |
| 44 | + &[ |
| 45 | + Token::BeginPattern, |
| 46 | + Token::Newline, |
| 47 | + Token::Newline, |
| 48 | + Token::OpenBrace, |
| 49 | + Token::Else, |
| 50 | + Token::Do, |
| 51 | + Token::BooleanAnd, |
| 52 | + Token::BooleanOr, |
| 53 | + Token::QuestionMark, |
| 54 | + Token::Colon, |
| 55 | + Token::Comma, |
| 56 | + Token::ClosedBrace |
| 57 | + ] |
| 58 | + ); |
| 59 | +} |
| 60 | + |
| 61 | +#[test] |
| 62 | +#[should_panic] |
| 63 | +fn lexer_test_newlines_posix() { |
| 64 | + let mixed = " \t \n \t\n\n \\\n \t"; |
| 65 | + let arena = Bump::new(); |
| 66 | + let mut str = Vec::new_in(&arena); |
| 67 | + for tok in ["BEGIN", "{", "else", "do", "&&", "||", "?", ":", ","] { |
| 68 | + write!(str, "{tok}{mixed}").unwrap(); |
| 69 | + } |
| 70 | + str.push(b'}'); |
| 71 | + lex(&str, &arena, true, false); |
| 72 | +} |
| 73 | + |
| 74 | +#[test] |
| 75 | +fn lexer_test_collapsible_delimiters() { |
| 76 | + let arena = Bump::new(); |
| 77 | + let str = b";\\\n;;;\n\n\n\n;;;\n\\\n\n"; |
| 78 | + assert_eq!( |
| 79 | + &lex(str, &arena, false, false), |
| 80 | + &[ |
| 81 | + Token::Semicolon, |
| 82 | + Token::Semicolon, |
| 83 | + Token::Newline, |
| 84 | + Token::Semicolon, |
| 85 | + Token::Newline, |
| 86 | + Token::Newline, |
| 87 | + ] |
| 88 | + ); |
| 89 | +} |
| 90 | + |
| 91 | +#[test] |
| 92 | +fn lexer_test_multiline() { |
| 93 | + let arena = Bump::new(); |
| 94 | + let str = b"\"aaaa\\\nbbbb\", /ccc\\\nd/"; |
| 95 | + assert_eq!( |
| 96 | + &lex(str, &arena, false, false), |
| 97 | + &[ |
| 98 | + Token::String(b"aaaabbbb".into()), |
| 99 | + Token::Comma, |
| 100 | + Token::Regex(b"cccd".into()) |
| 101 | + ] |
| 102 | + ); |
| 103 | +} |
| 104 | + |
| 105 | +#[test] |
| 106 | +#[should_panic] |
| 107 | +fn lexer_test_uu_extensions() { |
| 108 | + let arena = Bump::new(); |
| 109 | + lex(b"@concurrent", &arena, false, true); |
| 110 | +} |
| 111 | + |
| 112 | +#[test] |
| 113 | +fn lexer_test_gnu_pattern() { |
| 114 | + let arena = Bump::new(); |
| 115 | + assert_eq!( |
| 116 | + &lex(b"BEGINFILE ENDFILE", &arena, true, false), |
| 117 | + &[ |
| 118 | + Token::Identifier(Identifier { |
| 119 | + namespace: None, |
| 120 | + literal: "BEGINFILE" |
| 121 | + }), |
| 122 | + Token::Identifier(Identifier { |
| 123 | + namespace: None, |
| 124 | + literal: "ENDFILE" |
| 125 | + }) |
| 126 | + ] |
| 127 | + ); |
| 128 | +} |
| 129 | + |
| 130 | +#[test] |
| 131 | +fn lexer_test_floats() { |
| 132 | + let arena = Bump::new(); |
| 133 | + let str = b"1 20. 0. .3 2e4 -3.e2 5e+1 2.1e-3"; |
| 134 | + assert_eq!( |
| 135 | + &lex(str, &arena, false, false), |
| 136 | + &[ |
| 137 | + Token::Number(1.), |
| 138 | + Token::Number(20.), |
| 139 | + Token::Number(0.), |
| 140 | + Token::Number(0.3), |
| 141 | + Token::Number(2e4), |
| 142 | + Token::Number(-3e2), |
| 143 | + Token::Number(5e1), |
| 144 | + Token::Number(2.1e-3) |
| 145 | + ] |
| 146 | + ); |
| 147 | +} |
| 148 | + |
| 149 | +#[test] |
| 150 | +fn lexer_test_directive_escaping() { |
| 151 | + let arena = Bump::new(); |
| 152 | + let str = br#" @include "aa\"a\ta" @nsinclude "b\"\nb" "#; |
| 153 | + assert_eq!( |
| 154 | + &lex(str, &arena, false, false), |
| 155 | + &[ |
| 156 | + Token::IncludeDirective(b"aa\"a\ta".into()), |
| 157 | + Token::NsIncludeDirective(b"b\"\nb".into()) |
| 158 | + ] |
| 159 | + ); |
| 160 | +} |
| 161 | + |
| 162 | +#[test] |
| 163 | +#[should_panic] |
| 164 | +fn lexer_test_ident_rules_non_posix() { |
| 165 | + let arena = Bump::new(); |
| 166 | + lex(b"@namespace \"1a\"; a::1a", &arena, false, false); |
| 167 | +} |
| 168 | + |
| 169 | +#[test] |
| 170 | +#[should_panic] |
| 171 | +fn lexer_test_ident_rules_posix() { |
| 172 | + let arena = Bump::new(); |
| 173 | + lex(b"@namespace \"foo\"; foo::a", &arena, true, false); |
| 174 | +} |
| 175 | + |
| 176 | +#[test] |
| 177 | +fn lexer_test_general_tokens() { |
| 178 | + let arena = Bump::new(); |
| 179 | + let str = br#" |
| 180 | + @load "lib1.so.1" |
| 181 | + BEGIN { print a + 1 } |
| 182 | + /2\..*/; |
| 183 | + END { $1 == foo::bar } |
| 184 | + "#; |
| 185 | + assert_eq!( |
| 186 | + &lex(str, &arena, false, false), |
| 187 | + &[ |
| 188 | + Token::Newline, |
| 189 | + Token::LoadDirective(b"lib1.so.1".into()), |
| 190 | + Token::Newline, |
| 191 | + Token::BeginPattern, |
| 192 | + Token::OpenBrace, |
| 193 | + Token::Print, |
| 194 | + Token::Identifier(Identifier { |
| 195 | + namespace: None, |
| 196 | + literal: "a" |
| 197 | + }), |
| 198 | + Token::Plus, |
| 199 | + Token::Number(1.), |
| 200 | + Token::ClosedBrace, |
| 201 | + Token::Newline, |
| 202 | + Token::Regex(b"2\\..*".into()), |
| 203 | + Token::Semicolon, |
| 204 | + Token::Newline, |
| 205 | + Token::EndPattern, |
| 206 | + Token::OpenBrace, |
| 207 | + Token::Record, |
| 208 | + Token::Number(1.), |
| 209 | + Token::EqualTo, |
| 210 | + Token::Identifier(Identifier { |
| 211 | + namespace: Some("foo"), |
| 212 | + literal: "bar" |
| 213 | + }), |
| 214 | + Token::ClosedBrace, |
| 215 | + Token::Newline |
| 216 | + ] |
| 217 | + ); |
| 218 | +} |
0 commit comments