|
| 1 | +use langcodec::types::Translation; |
| 2 | +use langcodec::{Codec, convert_auto}; |
| 3 | +use std::path::{Path, PathBuf}; |
| 4 | + |
| 5 | +#[derive(Clone)] |
| 6 | +struct ExpectedValue { |
| 7 | + language: &'static str, |
| 8 | + key: &'static str, |
| 9 | + value: &'static str, |
| 10 | +} |
| 11 | + |
| 12 | +struct ParseCase { |
| 13 | + name: &'static str, |
| 14 | + input_relative_path: &'static str, |
| 15 | + lang_hint: Option<&'static str>, |
| 16 | + expected_values: Vec<ExpectedValue>, |
| 17 | +} |
| 18 | + |
| 19 | +struct ConvertCase { |
| 20 | + name: &'static str, |
| 21 | + input_relative_path: &'static str, |
| 22 | + output_file_name: &'static str, |
| 23 | + output_lang_hint: Option<&'static str>, |
| 24 | + expected_values: Vec<ExpectedValue>, |
| 25 | +} |
| 26 | + |
| 27 | +fn corpus_root() -> PathBuf { |
| 28 | + Path::new(env!("CARGO_MANIFEST_DIR")) |
| 29 | + .join("..") |
| 30 | + .join("tests") |
| 31 | + .join("data") |
| 32 | + .join("lib") |
| 33 | + .join("corpus") |
| 34 | +} |
| 35 | + |
| 36 | +fn expected_en_stable_values() -> Vec<ExpectedValue> { |
| 37 | + vec![ |
| 38 | + ExpectedValue { |
| 39 | + language: "en", |
| 40 | + key: "welcome_message", |
| 41 | + value: "Hello, World!", |
| 42 | + }, |
| 43 | + ExpectedValue { |
| 44 | + language: "en", |
| 45 | + key: "xml_entities", |
| 46 | + value: "Use <tag> & value", |
| 47 | + }, |
| 48 | + ExpectedValue { |
| 49 | + language: "en", |
| 50 | + key: "comma_text", |
| 51 | + value: "alpha, beta, gamma", |
| 52 | + }, |
| 53 | + ExpectedValue { |
| 54 | + language: "en", |
| 55 | + key: "accent_text", |
| 56 | + value: "Café crème brûlée", |
| 57 | + }, |
| 58 | + ] |
| 59 | +} |
| 60 | + |
| 61 | +fn expected_fr_stable_values() -> Vec<ExpectedValue> { |
| 62 | + vec![ |
| 63 | + ExpectedValue { |
| 64 | + language: "fr", |
| 65 | + key: "welcome_message", |
| 66 | + value: "Bonjour, le monde !", |
| 67 | + }, |
| 68 | + ExpectedValue { |
| 69 | + language: "fr", |
| 70 | + key: "xml_entities", |
| 71 | + value: "Utiliser <tag> & valeur", |
| 72 | + }, |
| 73 | + ExpectedValue { |
| 74 | + language: "fr", |
| 75 | + key: "comma_text", |
| 76 | + value: "alpha, bêta, gamma", |
| 77 | + }, |
| 78 | + ExpectedValue { |
| 79 | + language: "fr", |
| 80 | + key: "accent_text", |
| 81 | + value: "Café crème brûlée", |
| 82 | + }, |
| 83 | + ] |
| 84 | +} |
| 85 | + |
| 86 | +fn with_language( |
| 87 | + expected_values: Vec<ExpectedValue>, |
| 88 | + language: &'static str, |
| 89 | +) -> Vec<ExpectedValue> { |
| 90 | + expected_values |
| 91 | + .into_iter() |
| 92 | + .map(|item| ExpectedValue { language, ..item }) |
| 93 | + .collect() |
| 94 | +} |
| 95 | + |
| 96 | +fn read_codec(path: &Path, lang_hint: Option<&str>) -> Codec { |
| 97 | + let mut codec = Codec::new(); |
| 98 | + codec |
| 99 | + .read_file_by_extension(path, lang_hint.map(|l| l.to_string())) |
| 100 | + .unwrap_or_else(|e| panic!("failed to read {}: {}", path.display(), e)); |
| 101 | + codec |
| 102 | +} |
| 103 | + |
| 104 | +fn assert_expected_values(codec: &Codec, expected_values: &[ExpectedValue], case_name: &str) { |
| 105 | + for expected in expected_values { |
| 106 | + let entry = codec |
| 107 | + .find_entry(expected.key, expected.language) |
| 108 | + .unwrap_or_else(|| { |
| 109 | + panic!( |
| 110 | + "{case_name}: missing key '{}' for language '{}'", |
| 111 | + expected.key, expected.language |
| 112 | + ) |
| 113 | + }); |
| 114 | + |
| 115 | + match &entry.value { |
| 116 | + Translation::Singular(actual) => { |
| 117 | + assert_eq!( |
| 118 | + actual, expected.value, |
| 119 | + "{case_name}: value mismatch for {}:{}", |
| 120 | + expected.language, expected.key |
| 121 | + ); |
| 122 | + } |
| 123 | + other => panic!( |
| 124 | + "{case_name}: expected singular value for {}:{}, got {:?}", |
| 125 | + expected.language, expected.key, other |
| 126 | + ), |
| 127 | + } |
| 128 | + } |
| 129 | +} |
| 130 | + |
| 131 | +#[test] |
| 132 | +fn parse_edge_case_corpora_table_driven() { |
| 133 | + let root = corpus_root(); |
| 134 | + let mut csv_and_tsv_expected = expected_en_stable_values(); |
| 135 | + csv_and_tsv_expected.extend(expected_fr_stable_values()); |
| 136 | + |
| 137 | + let parse_cases = vec![ |
| 138 | + ParseCase { |
| 139 | + name: "strings corpus parse", |
| 140 | + input_relative_path: "en.lproj/Localizable.strings", |
| 141 | + lang_hint: None, |
| 142 | + expected_values: { |
| 143 | + let mut expected = expected_en_stable_values(); |
| 144 | + expected.push(ExpectedValue { |
| 145 | + language: "en", |
| 146 | + key: "quoted_text", |
| 147 | + value: "He said \\\"Hello\\\"", |
| 148 | + }); |
| 149 | + expected.push(ExpectedValue { |
| 150 | + language: "en", |
| 151 | + key: "apostrophe_text", |
| 152 | + value: "Don't stop", |
| 153 | + }); |
| 154 | + expected |
| 155 | + }, |
| 156 | + }, |
| 157 | + ParseCase { |
| 158 | + name: "android corpus parse", |
| 159 | + input_relative_path: "values-en/strings.xml", |
| 160 | + lang_hint: None, |
| 161 | + expected_values: { |
| 162 | + let mut expected = expected_en_stable_values(); |
| 163 | + expected.push(ExpectedValue { |
| 164 | + language: "en", |
| 165 | + key: "quoted_text", |
| 166 | + value: "He said \"Hello\"", |
| 167 | + }); |
| 168 | + expected.push(ExpectedValue { |
| 169 | + language: "en", |
| 170 | + key: "apostrophe_text", |
| 171 | + value: "Don\\'t stop", |
| 172 | + }); |
| 173 | + expected |
| 174 | + }, |
| 175 | + }, |
| 176 | + ParseCase { |
| 177 | + name: "csv corpus parse", |
| 178 | + input_relative_path: "corpus.csv", |
| 179 | + lang_hint: None, |
| 180 | + expected_values: csv_and_tsv_expected.clone(), |
| 181 | + }, |
| 182 | + ParseCase { |
| 183 | + name: "tsv corpus parse", |
| 184 | + input_relative_path: "corpus.tsv", |
| 185 | + lang_hint: None, |
| 186 | + expected_values: csv_and_tsv_expected, |
| 187 | + }, |
| 188 | + ]; |
| 189 | + |
| 190 | + for case in parse_cases { |
| 191 | + let input_path = root.join(case.input_relative_path); |
| 192 | + let codec = read_codec(&input_path, case.lang_hint); |
| 193 | + assert_expected_values(&codec, &case.expected_values, case.name); |
| 194 | + } |
| 195 | +} |
| 196 | + |
| 197 | +#[test] |
| 198 | +fn convert_edge_case_corpora_table_driven() { |
| 199 | + let root = corpus_root(); |
| 200 | + let output_dir = tempfile::tempdir().expect("create temp output dir"); |
| 201 | + |
| 202 | + let csv_default_expected = with_language(expected_en_stable_values(), "default"); |
| 203 | + |
| 204 | + let convert_cases = vec![ |
| 205 | + ConvertCase { |
| 206 | + name: "strings -> android", |
| 207 | + input_relative_path: "en.lproj/Localizable.strings", |
| 208 | + output_file_name: "from_strings.xml", |
| 209 | + output_lang_hint: Some("en"), |
| 210 | + expected_values: expected_en_stable_values(), |
| 211 | + }, |
| 212 | + ConvertCase { |
| 213 | + name: "strings -> xcstrings", |
| 214 | + input_relative_path: "en.lproj/Localizable.strings", |
| 215 | + output_file_name: "from_strings.xcstrings", |
| 216 | + output_lang_hint: None, |
| 217 | + expected_values: expected_en_stable_values(), |
| 218 | + }, |
| 219 | + ConvertCase { |
| 220 | + name: "strings -> csv", |
| 221 | + input_relative_path: "en.lproj/Localizable.strings", |
| 222 | + output_file_name: "from_strings.csv", |
| 223 | + output_lang_hint: None, |
| 224 | + expected_values: csv_default_expected.clone(), |
| 225 | + }, |
| 226 | + ConvertCase { |
| 227 | + name: "strings -> tsv", |
| 228 | + input_relative_path: "en.lproj/Localizable.strings", |
| 229 | + output_file_name: "from_strings.tsv", |
| 230 | + output_lang_hint: None, |
| 231 | + expected_values: csv_default_expected, |
| 232 | + }, |
| 233 | + ConvertCase { |
| 234 | + name: "android -> strings", |
| 235 | + input_relative_path: "values-en/strings.xml", |
| 236 | + output_file_name: "from_android.strings", |
| 237 | + output_lang_hint: Some("en"), |
| 238 | + expected_values: expected_en_stable_values(), |
| 239 | + }, |
| 240 | + ConvertCase { |
| 241 | + name: "csv -> xcstrings", |
| 242 | + input_relative_path: "corpus.csv", |
| 243 | + output_file_name: "from_csv.xcstrings", |
| 244 | + output_lang_hint: None, |
| 245 | + expected_values: { |
| 246 | + let mut expected = expected_en_stable_values(); |
| 247 | + expected.extend(expected_fr_stable_values()); |
| 248 | + expected |
| 249 | + }, |
| 250 | + }, |
| 251 | + ConvertCase { |
| 252 | + name: "tsv -> xcstrings", |
| 253 | + input_relative_path: "corpus.tsv", |
| 254 | + output_file_name: "from_tsv.xcstrings", |
| 255 | + output_lang_hint: None, |
| 256 | + expected_values: { |
| 257 | + let mut expected = expected_en_stable_values(); |
| 258 | + expected.extend(expected_fr_stable_values()); |
| 259 | + expected |
| 260 | + }, |
| 261 | + }, |
| 262 | + ]; |
| 263 | + |
| 264 | + for case in convert_cases { |
| 265 | + let input_path = root.join(case.input_relative_path); |
| 266 | + let output_path = output_dir.path().join(case.output_file_name); |
| 267 | + |
| 268 | + convert_auto(&input_path, &output_path).unwrap_or_else(|e| { |
| 269 | + panic!( |
| 270 | + "{}: conversion failed from {} to {}: {}", |
| 271 | + case.name, |
| 272 | + input_path.display(), |
| 273 | + output_path.display(), |
| 274 | + e |
| 275 | + ) |
| 276 | + }); |
| 277 | + |
| 278 | + let codec = read_codec(&output_path, case.output_lang_hint); |
| 279 | + assert_expected_values(&codec, &case.expected_values, case.name); |
| 280 | + } |
| 281 | +} |
0 commit comments