| 1 | //! Whale nickname pools and locale-aware display-name helpers for sub-agents. |
| 2 | |
| 3 | /// Whale species used as friendly names for sub-agents in the UI. The full |
| 4 | /// Cetacea infraorder — baleen whales (Mysticeti), toothed whales |
| 5 | /// (Odontoceti), plus select dolphin species (family Delphinidae) that |
| 6 | /// don't conflate with existing agent type labels. Porpoises (Phocoenidae) |
| 7 | /// are excluded because their name doesn't carry well as a friendly label. |
| 8 | /// |
| 9 | /// English and Simplified-Chinese names are stored as adjacent pairs. Name |
| 10 | /// selection follows the active session locale; it never mixes languages in |
| 11 | /// one session. Smaller curated pools below cover every other shipped locale. |
| 12 | /// |
| 13 | /// Taxonomy source: Society for Marine Mammalogy (2025). |
| 14 | pub const WHALE_NICKNAMES: &[&str] = &[ |
| 15 | "Blue", |
| 16 | "蓝鲸", |
| 17 | "Humpback", |
| 18 | "座头鲸", |
| 19 | "Sperm", |
| 20 | "抹香鲸", |
| 21 | "Fin", |
| 22 | "长须鲸", |
| 23 | "Sei", |
| 24 | "塞鲸", |
| 25 | "Bryde's", |
| 26 | "布氏鲸", |
| 27 | "Minke", |
| 28 | "小须鲸", |
| 29 | "Antarctic Minke", |
| 30 | "南极小须鲸", |
| 31 | "Pygmy Right", |
| 32 | "小露脊鲸", |
| 33 | "Omura's", |
| 34 | "大村鲸", |
| 35 | "Eden's", |
| 36 | "艾氏鲸", |
| 37 | "Rice's", |
| 38 | "赖斯鲸", |
| 39 | "Gray", |
| 40 | "灰鲸", |
| 41 | "Bowhead", |
| 42 | "弓头鲸", |
| 43 | "North Atlantic Right", |
| 44 | "北大西洋露脊鲸", |
| 45 | "North Pacific Right", |
| 46 | "北太平洋露脊鲸", |
| 47 | "Southern Right", |
| 48 | "南露脊鲸", |
| 49 | "Beluga", |
| 50 | "白鲸", |
| 51 | "Narwhal", |
| 52 | "独角鲸", |
| 53 | "Orca", |
| 54 | "虎鲸", |
| 55 | "Pilot", |
| 56 | "领航鲸", |
| 57 | "False Killer", |
| 58 | "伪虎鲸", |
| 59 | "Pygmy Killer", |
| 60 | "小虎鲸", |
| 61 | "Melon-headed", |
| 62 | "瓜头鲸", |
| 63 | "Beaked", |
| 64 | "喙鲸", |
| 65 | "Cuvier's Beaked", |
| 66 | "柯氏喙鲸", |
| 67 | "Baird's Beaked", |
| 68 | "贝氏喙鲸", |
| 69 | "Blainville's Beaked", |
| 70 | "柏氏喙鲸", |
| 71 | "Ginkgo-toothed Beaked", |
| 72 | "银杏齿喙鲸", |
| 73 | "Strap-toothed", |
| 74 | "带齿喙鲸", |
| 75 | "Stejneger's Beaked", |
| 76 | "斯氏喙鲸", |
| 77 | "Dwarf Sperm", |
| 78 | "小抹香鲸", |
| 79 | "Pygmy Sperm", |
| 80 | "侏儒抹香鲸", |
| 81 | "Rough-toothed", |
| 82 | "糙齿海豚", |
| 83 | "Atlantic Spotted", |
| 84 | "大西洋斑海豚", |
| 85 | "Pantropical Spotted", |
| 86 | "热带斑海豚", |
| 87 | "Spinner", |
| 88 | "长吻飞旋海豚", |
| 89 | "Clymene", |
| 90 | "短吻飞旋海豚", |
| 91 | "Striped", |
| 92 | "条纹海豚", |
| 93 | "Common Bottlenose", |
| 94 | "宽吻海豚", |
| 95 | "Indo-Pacific Bottlenose", |
| 96 | "印太瓶鼻海豚", |
| 97 | "Risso's", |
| 98 | "灰海豚", |
| 99 | "Commerson's", |
| 100 | "花斑海豚", |
| 101 | "Chilean", |
| 102 | "智利海豚", |
| 103 | "Heaviside's", |
| 104 | "海氏矮海豚", |
| 105 | "Hector's", |
| 106 | "赫氏矮海豚", |
| 107 | "Amazon River", |
| 108 | "亚马逊河豚", |
| 109 | "Ganges River", |
| 110 | "恒河豚", |
| 111 | "Indus River", |
| 112 | "印度河豚", |
| 113 | "La Plata", |
| 114 | "拉普拉塔河豚", |
| 115 | "Franciscana", |
| 116 | "拉河豚", |
| 117 | ]; |
| 118 | |
| 119 | pub(super) const WHALE_NICKNAMES_JA: &[&str] = &[ |
| 120 | "シロナガスクジラ", |
| 121 | "ザトウクジラ", |
| 122 | "マッコウクジラ", |
| 123 | "ナガスクジラ", |
| 124 | "イワシクジラ", |
| 125 | "ミンククジラ", |
| 126 | "コククジラ", |
| 127 | "ホッキョククジラ", |
| 128 | "シロイルカ", |
| 129 | "イッカク", |
| 130 | "シャチ", |
| 131 | "ゴンドウクジラ", |
| 132 | ]; |
| 133 | |
| 134 | pub(super) const WHALE_NICKNAMES_ZH_HANT: &[&str] = &[ |
| 135 | "藍鯨", |
| 136 | "座頭鯨", |
| 137 | "抹香鯨", |
| 138 | "長鬚鯨", |
| 139 | "塞鯨", |
| 140 | "布氏鯨", |
| 141 | "小鬚鯨", |
| 142 | "灰鯨", |
| 143 | "弓頭鯨", |
| 144 | "白鯨", |
| 145 | "獨角鯨", |
| 146 | "虎鯨", |
| 147 | ]; |
| 148 | |
| 149 | pub(super) const WHALE_NICKNAMES_PT_BR: &[&str] = &[ |
| 150 | "Azul", |
| 151 | "Jubarte", |
| 152 | "Cachalote", |
| 153 | "Baleia-fin", |
| 154 | "Baleia-sei", |
| 155 | "Baleia-de-bryde", |
| 156 | "Baleia-minke", |
| 157 | "Cinzenta", |
| 158 | "Baleia-franca", |
| 159 | "Beluga", |
| 160 | "Narval", |
| 161 | "Orca", |
| 162 | ]; |
| 163 | |
| 164 | pub(super) const WHALE_NICKNAMES_ES_419: &[&str] = &[ |
| 165 | "Azul", |
| 166 | "Jorobada", |
| 167 | "Cachalote", |
| 168 | "Rorcual común", |
| 169 | "Rorcual sei", |
| 170 | "Rorcual de Bryde", |
| 171 | "Rorcual aliblanco", |
| 172 | "Gris", |
| 173 | "Ballena franca", |
| 174 | "Beluga", |
| 175 | "Narval", |
| 176 | "Orca", |
| 177 | ]; |
| 178 | |
| 179 | pub(super) const WHALE_NICKNAMES_VI: &[&str] = &[ |
| 180 | "Cá voi xanh", |
| 181 | "Cá voi lưng gù", |
| 182 | "Cá nhà táng", |
| 183 | "Cá voi vây", |
| 184 | "Cá voi Sei", |
| 185 | "Cá voi Bryde", |
| 186 | "Cá voi Minke", |
| 187 | "Cá voi xám", |
| 188 | "Cá voi đầu cong", |
| 189 | "Cá voi trắng", |
| 190 | "Kỳ lân biển", |
| 191 | "Cá voi sát thủ", |
| 192 | ]; |
| 193 | |
| 194 | pub(super) const WHALE_NICKNAMES_KO: &[&str] = &[ |
| 195 | "대왕고래", |
| 196 | "혹등고래", |
| 197 | "향유고래", |
| 198 | "참고래", |
| 199 | "보리고래", |
| 200 | "브라이드고래", |
| 201 | "밍크고래", |
| 202 | "귀신고래", |
| 203 | "북극고래", |
| 204 | "흰고래", |
| 205 | "외뿔고래", |
| 206 | "범고래", |
| 207 | ]; |
| 208 | |
| 209 | pub(super) const WHALE_NICKNAMES_CA: &[&str] = &[ |
| 210 | "Balena blava", |
| 211 | "Balena geperuda", |
| 212 | "Catxalot", |
| 213 | "Rorcual comú", |
| 214 | "Rorcual boreal", |
| 215 | "Balena minke", |
| 216 | "Balena franca", |
| 217 | "Balena de Groenlàndia", |
| 218 | "Beluga", |
| 219 | "Narval", |
| 220 | "Orca", |
| 221 | "Calderó", |
| 222 | ]; |
| 223 | |
| 224 | pub(super) const WHALE_NICKNAMES_DE: &[&str] = &[ |
| 225 | "Blauwal", |
| 226 | "Buckelwal", |
| 227 | "Pottwal", |
| 228 | "Finnwal", |
| 229 | "Seiwal", |
| 230 | "Zwergwal", |
| 231 | "Glattwal", |
| 232 | "Grönlandwal", |
| 233 | "Beluga", |
| 234 | "Narwal", |
| 235 | "Orca", |
| 236 | "Pilotwal", |
| 237 | ]; |
| 238 | |
| 239 | pub(super) const WHALE_NICKNAMES_FR: &[&str] = &[ |
| 240 | "Baleine bleue", |
| 241 | "Baleine à bosse", |
| 242 | "Cachalot", |
| 243 | "Rorqual commun", |
| 244 | "Rorqual boréal", |
| 245 | "Petit rorqual", |
| 246 | "Baleine franche", |
| 247 | "Baleine boréale", |
| 248 | "Béluga", |
| 249 | "Narval", |
| 250 | "Orque", |
| 251 | "Globicéphale", |
| 252 | ]; |
| 253 | |
| 254 | pub(super) const WHALE_NICKNAMES_ID: &[&str] = &[ |
| 255 | "Paus biru", |
| 256 | "Paus bungkuk", |
| 257 | "Paus sperma", |
| 258 | "Paus sirip", |
| 259 | "Paus sei", |
| 260 | "Paus minke", |
| 261 | "Paus sikat", |
| 262 | "Paus bowhead", |
| 263 | "Beluga", |
| 264 | "Narwal", |
| 265 | "Orca", |
| 266 | "Paus pilot", |
| 267 | ]; |
| 268 | |
| 269 | pub(super) const WHALE_NICKNAMES_HI: &[&str] = &[ |
| 270 | "नीली व्हेल", |
| 271 | "कूबड़ व्हेल", |
| 272 | "स्पर्म व्हेल", |
| 273 | "फिन व्हेल", |
| 274 | "सेई व्हेल", |
| 275 | "मिंक व्हेल", |
| 276 | "राइट व्हेल", |
| 277 | "बोहेड व्हेल", |
| 278 | "बेलुगा", |
| 279 | "नार्व्हल", |
| 280 | "ओर्का", |
| 281 | "पायलट व्हेल", |
| 282 | ]; |
| 283 | |
| 284 | pub(super) const WHALE_NICKNAMES_RU: &[&str] = &[ |
| 285 | "Синий кит", |
| 286 | "Горбатый кит", |
| 287 | "Кашалот", |
| 288 | "Финвал", |
| 289 | "Сейвал", |
| 290 | "Малый полосатик", |
| 291 | "Гладкий кит", |
| 292 | "Гренландский кит", |
| 293 | "Белуха", |
| 294 | "Нарвал", |
| 295 | "Косатка", |
| 296 | "Гринда", |
| 297 | ]; |
| 298 | |
| 299 | pub(super) const WHALE_NICKNAMES_UK: &[&str] = &[ |
| 300 | "Синій кит", |
| 301 | "Горбатий кит", |
| 302 | "Кашалот", |
| 303 | "Фінвал", |
| 304 | "Сейвал", |
| 305 | "Малий смугач", |
| 306 | "Гладкий кит", |
| 307 | "Гренландський кит", |
| 308 | "Белуга", |
| 309 | "Нарвал", |
| 310 | "Косатка", |
| 311 | "Гринда", |
| 312 | ]; |
| 313 | |
| 314 | /// Return a deterministic whale name in the active UI locale. |
| 315 | #[must_use] |
| 316 | pub fn whale_name_for_id_in_locale(id: &str, locale_tag: &str) -> String { |
| 317 | use std::hash::{Hash, Hasher}; |
| 318 | let mut hasher = std::collections::hash_map::DefaultHasher::new(); |
| 319 | id.hash(&mut hasher); |
| 320 | let hash = hasher.finish() as usize; |
| 321 | let normalized = locale_tag.trim().to_ascii_lowercase(); |
| 322 | |
| 323 | let localized_pool = match normalized.as_str() { |
| 324 | "ja" => Some(WHALE_NICKNAMES_JA), |
| 325 | "zh-hant" => Some(WHALE_NICKNAMES_ZH_HANT), |
| 326 | "pt-br" => Some(WHALE_NICKNAMES_PT_BR), |
| 327 | "es-419" => Some(WHALE_NICKNAMES_ES_419), |
| 328 | "vi" => Some(WHALE_NICKNAMES_VI), |
| 329 | "ko" => Some(WHALE_NICKNAMES_KO), |
| 330 | "ca" => Some(WHALE_NICKNAMES_CA), |
| 331 | "de" => Some(WHALE_NICKNAMES_DE), |
| 332 | "fr" => Some(WHALE_NICKNAMES_FR), |
| 333 | "id" => Some(WHALE_NICKNAMES_ID), |
| 334 | "hi" => Some(WHALE_NICKNAMES_HI), |
| 335 | "ru" => Some(WHALE_NICKNAMES_RU), |
| 336 | "uk" => Some(WHALE_NICKNAMES_UK), |
| 337 | _ => None, |
| 338 | }; |
| 339 | if let Some(pool) = localized_pool { |
| 340 | return pool[hash % pool.len()].to_string(); |
| 341 | } |
| 342 | |
| 343 | debug_assert_eq!(WHALE_NICKNAMES.len() % 2, 0); |
| 344 | let pair_count = WHALE_NICKNAMES.len() / 2; |
| 345 | let pair = hash % pair_count; |
| 346 | let language_offset = usize::from(normalized == "zh-hans"); |
| 347 | let idx = pair * 2 + language_offset; |
| 348 | WHALE_NICKNAMES[idx].to_string() |
| 349 | } |
| 350 | |
| 351 | /// Assign a unique locale-matched whale name for an agent ID. |
| 352 | /// If the deterministic name is taken, appends a numeric suffix (for example, |
| 353 | /// `Orca (2)`). |
| 354 | #[must_use] |
| 355 | pub fn assign_unique_whale_name_in_locale( |
| 356 | id: &str, |
| 357 | active_names: &std::collections::HashSet<String>, |
| 358 | locale_tag: &str, |
| 359 | ) -> String { |
| 360 | let base = whale_name_for_id_in_locale(id, locale_tag); |
| 361 | if !active_names.contains(&base) { |
| 362 | return base; |
| 363 | } |
| 364 | // Deterministic suffix from the same hash to keep it stable |
| 365 | use std::hash::{Hash, Hasher}; |
| 366 | let mut hasher = std::collections::hash_map::DefaultHasher::new(); |
| 367 | id.hash(&mut hasher); |
| 368 | let suffix_seed = hasher.finish(); |
| 369 | for i in 2.. { |
| 370 | let candidate = format!("{base} ({i})"); |
| 371 | if !active_names.contains(&candidate) { |
| 372 | return candidate; |
| 373 | } |
| 374 | // Vary the probe using the seed |
| 375 | let probe = (suffix_seed.wrapping_add(i as u64)) % 100; |
| 376 | let candidate2 = format!("{base} ({probe})"); |
| 377 | if !active_names.contains(&candidate2) { |
| 378 | return candidate2; |
| 379 | } |
| 380 | } |
| 381 | // Fallback (should never reach here) |
| 382 | format!("{base} ({})", id.get(..4).unwrap_or("?")) |
| 383 | } |
| 384 | |
| 385 | /// Return the unsuffixed whale label when `name` could have been generated for |
| 386 | /// this exact agent id in a shipped locale. Numeric collision suffixes are |
| 387 | /// presentation-only and do not make the label user-authored. |
| 388 | pub(super) fn generated_whale_name_base<'a>(agent_id: &str, name: &'a str) -> Option<&'a str> { |
| 389 | let name = name.trim(); |
| 390 | if name.is_empty() { |
| 391 | return None; |
| 392 | } |
| 393 | let base = name |
| 394 | .rsplit_once(" (") |
| 395 | .and_then(|(base, suffix)| { |
| 396 | suffix |
| 397 | .strip_suffix(')') |
| 398 | .filter(|number| !number.is_empty() && number.chars().all(|ch| ch.is_ascii_digit())) |
| 399 | .map(|_| base) |
| 400 | }) |
| 401 | .unwrap_or(name); |
| 402 | |
| 403 | // With no persisted provenance bit, the narrowest truthful test is whether |
| 404 | // this exact agent id could have generated the label in a shipped locale. |
| 405 | // A user-authored label that happens to be a whale word for some other id |
| 406 | // remains explicit. An exact deterministic match is inherently ambiguous |
| 407 | // and stays classified as generated for backward compatibility. |
| 408 | crate::localization::Locale::shipped() |
| 409 | .iter() |
| 410 | .any(|locale| whale_name_for_id_in_locale(agent_id, locale.tag()) == base) |
| 411 | .then_some(base) |
| 412 | } |
| 413 | |
| 414 | /// Derive the generated whale labels shown for a set of workers from their |
| 415 | /// locale-neutral ids and the active UI language. |
| 416 | /// |
| 417 | /// Persisted `nickname` values predate locale-scoped naming and may contain a |
| 418 | /// whale label chosen under another language. Those generated values are |
| 419 | /// deliberately ignored here. A nickname that this agent id could not have |
| 420 | /// generated is an explicit custom label and remains intact, even when it is a |
| 421 | /// whale word from a built-in pool. |
| 422 | #[must_use] |
| 423 | pub(crate) fn localized_whale_display_names<'a>( |
| 424 | agents: impl IntoIterator<Item = (&'a str, Option<&'a str>)>, |
| 425 | locale_tag: &str, |
| 426 | ) -> std::collections::HashMap<String, String> { |
| 427 | let mut by_id = std::collections::BTreeMap::<String, Option<String>>::new(); |
| 428 | for (agent_id, nickname) in agents { |
| 429 | if agent_id.trim().is_empty() { |
| 430 | continue; |
| 431 | } |
| 432 | let nickname = nickname |
| 433 | .map(str::trim) |
| 434 | .filter(|name| !name.is_empty()) |
| 435 | .map(str::to_string); |
| 436 | by_id |
| 437 | .entry(agent_id.to_string()) |
| 438 | .and_modify(|existing| { |
| 439 | if existing.is_none() && nickname.is_some() { |
| 440 | *existing = nickname.clone(); |
| 441 | } |
| 442 | }) |
| 443 | .or_insert(nickname); |
| 444 | } |
| 445 | |
| 446 | let mut names = std::collections::HashMap::with_capacity(by_id.len()); |
| 447 | let mut active_names = std::collections::HashSet::new(); |
| 448 | |
| 449 | // Reserve explicit labels first so generated names never shadow them. |
| 450 | for (agent_id, nickname) in &by_id { |
| 451 | let Some(nickname) = nickname |
| 452 | .as_deref() |
| 453 | .filter(|name| generated_whale_name_base(agent_id, name).is_none()) |
| 454 | else { |
| 455 | continue; |
| 456 | }; |
| 457 | active_names.insert(nickname.to_string()); |
| 458 | names.insert(agent_id.clone(), nickname.to_string()); |
| 459 | } |
| 460 | |
| 461 | // BTreeMap iteration makes collision suffix ownership stable even when |
| 462 | // manager/progress event order changes between frames or session loads. |
| 463 | for agent_id in by_id.keys() { |
| 464 | if names.contains_key(agent_id) { |
| 465 | continue; |
| 466 | } |
| 467 | let name = assign_unique_whale_name_in_locale(agent_id, &active_names, locale_tag); |
| 468 | active_names.insert(name.clone()); |
| 469 | names.insert(agent_id.clone(), name); |
| 470 | } |
| 471 | |
| 472 | names |
| 473 | } |
| 474 |