oxedyne/fe2o3/fe2o3_datime/src/format/rfc9557.rs
15.3 KiB, 28 runs
created by r1870400018:8456, which is this file's identity for as long as the history lasts, whatever it is later renamed to
download · who wrote it · its history
| 1 | //! RFC 9557 Internet Extended Date/Time Format (IXDTF) serialisation capabilities. |
| 2 | //! |
| 3 | //! This module implements timezone-preserving serialisation according to RFC 9557, |
| 4 | //! which extends RFC 3339 timestamps with additional timezone information. |
| 5 | //! |
| 6 | //! RFC 9557 allows timestamps to include both UTC offset and IANA timezone identifier, |
| 7 | //! enabling proper handling of daylight saving transitions and timezone-aware calculations. |
| 8 | //! |
| 9 | //! # Examples |
| 10 | //! |
| 11 | //! ```ignore |
| 12 | //! use oxedyne_fe2o3_datime::{CalClock, CalClockZone, format::rfc9557::Rfc9557Format}; |
| 13 | //! |
| 14 | //! let zone = res!(CalClockZone::new("America/New_York")); |
| 15 | //! let calclock = res!(CalClock::new(2024, 6, 15, 14, 30, 0, 0, zone)); |
| 16 | //! |
| 17 | //! // Standard RFC 3339 format |
| 18 | //! let rfc3339 = res!(calclock.to_rfc9557_basic()); |
| 19 | //! // "2024-06-15T14:30:00.000000000-04:00" |
| 20 | //! |
| 21 | //! // Extended RFC 9557 format with timezone preservation |
| 22 | //! let rfc9557 = res!(calclock.to_rfc9557_extended()); |
| 23 | //! // "2024-06-15T14:30:00.000000000-04:00[America/New_York]" |
| 24 | //! ``` |
| 25 | //! |
| 26 | //! [Written with AI entirely](https://need2know.ai/entirely-ai/code)\ |
| 27 | //! Anthropic Claude |
| 28 | |
| 29 | use crate::{ |
| 30 | time::{CalClock, CalClockZone}, |
| 31 | clock::ClockTime, |
| 32 | calendar::CalendarDate, |
| 33 | }; |
| 34 | |
| 35 | use oxedyne_fe2o3_core::prelude::*; |
| 36 | |
| 37 | #[derive(Clone, Debug, PartialEq)] |
| 38 | pub struct Rfc9557Config { |
| 39 | pub include_timezone_name: bool, // IANA name in square brackets |
| 40 | pub always_include_nanoseconds: bool, // even when zero |
| 41 | pub use_z_for_utc: bool, // Z rather than +00:00 |
| 42 | pub include_precision_indicators: bool, |
| 43 | pub precision_level: PrecisionLevel, |
| 44 | } |
| 45 | |
| 46 | #[derive(Clone, Debug, PartialEq)] |
| 47 | pub enum PrecisionLevel { |
| 48 | // The indicator character each level writes. |
| 49 | Exact, // none |
| 50 | Approximate, // ~ |
| 51 | Uncertain, // % |
| 52 | Around, // @ |
| 53 | Between, // * |
| 54 | } |
| 55 | |
| 56 | impl Default for Rfc9557Config { |
| 57 | fn default() -> Self { |
| 58 | Self { |
| 59 | include_timezone_name: true, |
| 60 | always_include_nanoseconds: false, |
| 61 | use_z_for_utc: true, |
| 62 | include_precision_indicators: false, |
| 63 | precision_level: PrecisionLevel::Exact, |
| 64 | } |
| 65 | } |
| 66 | } |
| 67 | |
| 68 | impl PrecisionLevel { |
| 69 | pub fn indicator(&self) -> Option<char> { |
| 70 | match self { |
| 71 | Self::Exact => None, |
| 72 | Self::Approximate => Some('~'), |
| 73 | Self::Uncertain => Some('%'), |
| 74 | Self::Around => Some('@'), |
| 75 | Self::Between => Some('*'), |
| 76 | } |
| 77 | } |
| 78 | } |
| 79 | |
| 80 | pub trait Rfc9557Format { |
| 81 | fn to_rfc9557(&self) -> Outcome<String>; |
| 82 | |
| 83 | fn to_rfc9557_with_config(&self, config: &Rfc9557Config) -> Outcome<String>; |
| 84 | |
| 85 | /// RFC 3339 only, with no timezone name. |
| 86 | fn to_rfc9557_basic(&self) -> Outcome<String>; |
| 87 | |
| 88 | /// Appends the IANA timezone name in square brackets. |
| 89 | fn to_rfc9557_extended(&self) -> Outcome<String>; |
| 90 | |
| 91 | fn from_rfc9557(input: &str) -> Outcome<Self> |
| 92 | where |
| 93 | Self: Sized; |
| 94 | } |
| 95 | |
| 96 | impl Rfc9557Format for CalClock { |
| 97 | fn to_rfc9557(&self) -> Outcome<String> { |
| 98 | self.to_rfc9557_with_config(&Rfc9557Config::default()) |
| 99 | } |
| 100 | |
| 101 | fn to_rfc9557_with_config(&self, config: &Rfc9557Config) -> Outcome<String> { |
| 102 | // Build the basic timestamp part. |
| 103 | let date_part = format!("{:04}-{:02}-{:02}", |
| 104 | self.year(), self.month(), self.day()); |
| 105 | |
| 106 | // Format time with conditional nanoseconds. |
| 107 | let time_part = if config.always_include_nanoseconds || self.nanosecond() > 0 { |
| 108 | format!("{:02}:{:02}:{:02}.{:09}", |
| 109 | self.hour(), self.minute(), self.second(), self.nanosecond()) |
| 110 | } else { |
| 111 | format!("{:02}:{:02}:{:02}", |
| 112 | self.hour(), self.minute(), self.second()) |
| 113 | }; |
| 114 | |
| 115 | // Get timezone offset. |
| 116 | let offset_millis = res!(self.zone().offset_millis_at_time(res!(self.to_millis()))); |
| 117 | let offset_hours = offset_millis / (60 * 60 * 1000); |
| 118 | let offset_minutes = (offset_millis.abs() % (60 * 60 * 1000)) / (60 * 1000); |
| 119 | |
| 120 | // Format offset part. |
| 121 | let offset_part = if offset_millis == 0 && config.use_z_for_utc { |
| 122 | "Z".to_string() |
| 123 | } else { |
| 124 | format!("{:+03}:{:02}", offset_hours, offset_minutes) |
| 125 | }; |
| 126 | |
| 127 | // Add precision indicator if configured. |
| 128 | let precision_suffix = if config.include_precision_indicators { |
| 129 | config.precision_level.indicator() |
| 130 | .map(|c| c.to_string()) |
| 131 | .unwrap_or_default() |
| 132 | } else { |
| 133 | String::new() |
| 134 | }; |
| 135 | |
| 136 | // Add timezone name if configured. |
| 137 | let timezone_suffix = if config.include_timezone_name { |
| 138 | format!("[{}]", self.zone().id()) |
| 139 | } else { |
| 140 | String::new() |
| 141 | }; |
| 142 | |
| 143 | Ok(format!("{}T{}{}{}{}", |
| 144 | date_part, time_part, offset_part, precision_suffix, timezone_suffix)) |
| 145 | } |
| 146 | |
| 147 | fn to_rfc9557_basic(&self) -> Outcome<String> { |
| 148 | let config = Rfc9557Config { |
| 149 | include_timezone_name: false, |
| 150 | ..Default::default() |
| 151 | }; |
| 152 | self.to_rfc9557_with_config(&config) |
| 153 | } |
| 154 | |
| 155 | fn to_rfc9557_extended(&self) -> Outcome<String> { |
| 156 | let config = Rfc9557Config { |
| 157 | include_timezone_name: true, |
| 158 | ..Default::default() |
| 159 | }; |
| 160 | self.to_rfc9557_with_config(&config) |
| 161 | } |
| 162 | |
| 163 | fn from_rfc9557(input: &str) -> Outcome<Self> { |
| 164 | // Parse RFC 9557 format: YYYY-MM-DDTHH:MM:SS[.nnnnnnnnn][±HH:MM|Z][~|%|@|*][timezone] |
| 165 | let input = input.trim(); |
| 166 | |
| 167 | // Extract timezone name if present (in square brackets at the end). |
| 168 | let (timestamp_part, timezone_name) = if let Some(bracket_start) = input.rfind('[') { |
| 169 | if let Some(bracket_end) = input.rfind(']') { |
| 170 | if bracket_end > bracket_start { |
| 171 | let timezone_part = &input[bracket_start + 1..bracket_end]; |
| 172 | let timestamp_part = &input[..bracket_start]; |
| 173 | (timestamp_part, Some(timezone_part)) |
| 174 | } else { |
| 175 | (input, None) |
| 176 | } |
| 177 | } else { |
| 178 | (input, None) |
| 179 | } |
| 180 | } else { |
| 181 | (input, None) |
| 182 | }; |
| 183 | |
| 184 | // Remove precision indicators if present. |
| 185 | let timestamp_part = timestamp_part.trim_end_matches(['~', '%', '@', '*']); |
| 186 | |
| 187 | // Manual parsing instead of relying on parse_iso. |
| 188 | // Split into date and time parts. |
| 189 | let parts: Vec<&str> = timestamp_part.split('T').collect(); |
| 190 | if parts.len() != 2 { |
| 191 | return Err(err!("Invalid timestamp format: {}", timestamp_part; Invalid, Input)); |
| 192 | } |
| 193 | |
| 194 | let date_part = parts[0]; |
| 195 | let time_part_with_offset = parts[1]; |
| 196 | |
| 197 | // Parse date. |
| 198 | let date_components: Vec<&str> = date_part.split('-').collect(); |
| 199 | if date_components.len() != 3 { |
| 200 | return Err(err!("Invalid date format: {}", date_part; Invalid, Input)); |
| 201 | } |
| 202 | |
| 203 | let year = res!(date_components[0].parse::<i32>() |
| 204 | .map_err(|_| err!("Invalid year: {}", date_components[0]; Invalid, Input))); |
| 205 | let month = res!(date_components[1].parse::<u8>() |
| 206 | .map_err(|_| err!("Invalid month: {}", date_components[1]; Invalid, Input))); |
| 207 | let day = res!(date_components[2].parse::<u8>() |
| 208 | .map_err(|_| err!("Invalid day: {}", date_components[2]; Invalid, Input))); |
| 209 | |
| 210 | // Extract offset and time. |
| 211 | let (time_part, _offset_part) = if time_part_with_offset.ends_with('Z') { |
| 212 | (&time_part_with_offset[..time_part_with_offset.len() - 1], "Z") |
| 213 | } else if let Some(plus_pos) = time_part_with_offset.rfind('+') { |
| 214 | (&time_part_with_offset[..plus_pos], &time_part_with_offset[plus_pos..]) |
| 215 | } else if let Some(minus_pos) = time_part_with_offset.rfind('-') { |
| 216 | (&time_part_with_offset[..minus_pos], &time_part_with_offset[minus_pos..]) |
| 217 | } else { |
| 218 | (time_part_with_offset, "") |
| 219 | }; |
| 220 | |
| 221 | // Parse time components. |
| 222 | let time_components: Vec<&str> = time_part.split(':').collect(); |
| 223 | if time_components.len() < 3 { |
| 224 | return Err(err!("Invalid time format: {}", time_part; Invalid, Input)); |
| 225 | } |
| 226 | |
| 227 | let hour = res!(time_components[0].parse::<u8>() |
| 228 | .map_err(|_| err!("Invalid hour: {}", time_components[0]; Invalid, Input))); |
| 229 | let minute = res!(time_components[1].parse::<u8>() |
| 230 | .map_err(|_| err!("Invalid minute: {}", time_components[1]; Invalid, Input))); |
| 231 | |
| 232 | // Handle seconds with optional fractional part. |
| 233 | let (second, nanosecond) = if let Some(dot_pos) = time_components[2].find('.') { |
| 234 | let second_part = &time_components[2][..dot_pos]; |
| 235 | let fraction_part = &time_components[2][dot_pos + 1..]; |
| 236 | |
| 237 | let second = res!(second_part.parse::<u8>() |
| 238 | .map_err(|_| err!("Invalid second: {}", second_part; Invalid, Input))); |
| 239 | |
| 240 | // Parse fractional seconds, padding or truncating to 9 digits. |
| 241 | let mut fraction_str = fraction_part.to_string(); |
| 242 | fraction_str.truncate(9); // Max 9 digits for nanoseconds. |
| 243 | while fraction_str.len() < 9 { |
| 244 | fraction_str.push('0'); // Pad with zeros. |
| 245 | } |
| 246 | |
| 247 | let nanosecond = res!(fraction_str.parse::<u32>() |
| 248 | .map_err(|_| err!("Invalid nanosecond: {}", fraction_str; Invalid, Input))); |
| 249 | |
| 250 | (second, nanosecond) |
| 251 | } else { |
| 252 | let second = res!(time_components[2].parse::<u8>() |
| 253 | .map_err(|_| err!("Invalid second: {}", time_components[2]; Invalid, Input))); |
| 254 | (second, 0) |
| 255 | }; |
| 256 | |
| 257 | // Determine timezone. |
| 258 | let zone = if let Some(tz_name) = timezone_name { |
| 259 | res!(CalClockZone::new(tz_name)) |
| 260 | } else { |
| 261 | CalClockZone::utc() |
| 262 | }; |
| 263 | |
| 264 | Self::new(year, month, day, hour, minute, second, nanosecond, zone) |
| 265 | } |
| 266 | } |
| 267 | |
| 268 | impl Rfc9557Format for ClockTime { |
| 269 | fn to_rfc9557(&self) -> Outcome<String> { |
| 270 | self.to_rfc9557_with_config(&Rfc9557Config::default()) |
| 271 | } |
| 272 | |
| 273 | fn to_rfc9557_with_config(&self, config: &Rfc9557Config) -> Outcome<String> { |
| 274 | // Format time component only. |
| 275 | let time_part = if config.always_include_nanoseconds || self.nanosecond().of() > 0 { |
| 276 | format!("{:02}:{:02}:{:02}.{:09}", |
| 277 | self.hour().of(), self.minute().of(), self.second().of(), self.nanosecond().of()) |
| 278 | } else { |
| 279 | format!("{:02}:{:02}:{:02}", |
| 280 | self.hour().of(), self.minute().of(), self.second().of()) |
| 281 | }; |
| 282 | |
| 283 | // Add precision indicator if configured. |
| 284 | let precision_suffix = if config.include_precision_indicators { |
| 285 | config.precision_level.indicator() |
| 286 | .map(|c| c.to_string()) |
| 287 | .unwrap_or_default() |
| 288 | } else { |
| 289 | String::new() |
| 290 | }; |
| 291 | |
| 292 | // Add timezone name if configured. |
| 293 | let timezone_suffix = if config.include_timezone_name { |
| 294 | format!("[{}]", self.zone().id()) |
| 295 | } else { |
| 296 | String::new() |
| 297 | }; |
| 298 | |
| 299 | Ok(format!("{}{}{}", time_part, precision_suffix, timezone_suffix)) |
| 300 | } |
| 301 | |
| 302 | fn to_rfc9557_basic(&self) -> Outcome<String> { |
| 303 | let config = Rfc9557Config { |
| 304 | include_timezone_name: false, |
| 305 | ..Default::default() |
| 306 | }; |
| 307 | self.to_rfc9557_with_config(&config) |
| 308 | } |
| 309 | |
| 310 | fn to_rfc9557_extended(&self) -> Outcome<String> { |
| 311 | let config = Rfc9557Config { |
| 312 | include_timezone_name: true, |
| 313 | ..Default::default() |
| 314 | }; |
| 315 | self.to_rfc9557_with_config(&config) |
| 316 | } |
| 317 | |
| 318 | fn from_rfc9557(input: &str) -> Outcome<Self> { |
| 319 | // For time-only parsing, we'll use UTC as default and then update if timezone specified. |
| 320 | let input = input.trim(); |
| 321 | |
| 322 | // Extract timezone name if present. |
| 323 | let (time_part, timezone_name) = if let Some(bracket_start) = input.rfind('[') { |
| 324 | if let Some(bracket_end) = input.rfind(']') { |
| 325 | if bracket_end > bracket_start { |
| 326 | let timezone_part = &input[bracket_start + 1..bracket_end]; |
| 327 | let time_part = &input[..bracket_start]; |
| 328 | (time_part, Some(timezone_part)) |
| 329 | } else { |
| 330 | (input, None) |
| 331 | } |
| 332 | } else { |
| 333 | (input, None) |
| 334 | } |
| 335 | } else { |
| 336 | (input, None) |
| 337 | }; |
| 338 | |
| 339 | // Remove precision indicators. |
| 340 | let time_part = time_part.trim_end_matches(['~', '%', '@', '*']); |
| 341 | |
| 342 | // Parse time components. |
| 343 | let parts: Vec<&str> = time_part.split(':').collect(); |
| 344 | if parts.len() < 3 { |
| 345 | return Err(err!("Invalid time format: {}", input; Invalid, Input)); |
| 346 | } |
| 347 | |
| 348 | let hour = res!(parts[0].parse::<u8>() |
| 349 | .map_err(|_| err!("Invalid hour: {}", parts[0]; Invalid, Input))); |
| 350 | let minute = res!(parts[1].parse::<u8>() |
| 351 | .map_err(|_| err!("Invalid minute: {}", parts[1]; Invalid, Input))); |
| 352 | |
| 353 | // Handle seconds with optional fractional part. |
| 354 | let (second, nanosecond) = if let Some(dot_pos) = parts[2].find('.') { |
| 355 | let second_part = &parts[2][..dot_pos]; |
| 356 | let fraction_part = &parts[2][dot_pos + 1..]; |
| 357 | |
| 358 | let second = res!(second_part.parse::<u8>() |
| 359 | .map_err(|_| err!("Invalid second: {}", second_part; Invalid, Input))); |
| 360 | |
| 361 | // Parse fractional seconds, padding or truncating to 9 digits. |
| 362 | let mut fraction_str = fraction_part.to_string(); |
| 363 | fraction_str.truncate(9); // Max 9 digits for nanoseconds. |
| 364 | while fraction_str.len() < 9 { |
| 365 | fraction_str.push('0'); // Pad with zeros. |
| 366 | } |
| 367 | |
| 368 | let nanosecond = res!(fraction_str.parse::<u32>() |
| 369 | .map_err(|_| err!("Invalid nanosecond: {}", fraction_str; Invalid, Input))); |
| 370 | |
| 371 | (second, nanosecond) |
| 372 | } else { |
| 373 | let second = res!(parts[2].parse::<u8>() |
| 374 | .map_err(|_| err!("Invalid second: {}", parts[2]; Invalid, Input))); |
| 375 | (second, 0) |
| 376 | }; |
| 377 | |
| 378 | // Determine timezone. |
| 379 | let zone = if let Some(tz_name) = timezone_name { |
| 380 | res!(CalClockZone::new(tz_name)) |
| 381 | } else { |
| 382 | CalClockZone::utc() |
| 383 | }; |
| 384 | |
| 385 | Self::new(hour, minute, second, nanosecond, zone) |
| 386 | } |
| 387 | } |
| 388 | |
| 389 | impl Rfc9557Format for CalendarDate { |
| 390 | fn to_rfc9557(&self) -> Outcome<String> { |
| 391 | self.to_rfc9557_with_config(&Rfc9557Config::default()) |
| 392 | } |
| 393 | |
| 394 | fn to_rfc9557_with_config(&self, config: &Rfc9557Config) -> Outcome<String> { |
| 395 | // Format date component only. |
| 396 | let date_part = format!("{:04}-{:02}-{:02}", |
| 397 | self.year(), self.month(), self.day()); |
| 398 | |
| 399 | // Add timezone name if configured. |
| 400 | let timezone_suffix = if config.include_timezone_name { |
| 401 | format!("[{}]", self.zone().id()) |
| 402 | } else { |
| 403 | String::new() |
| 404 | }; |
| 405 | |
| 406 | Ok(format!("{}{}", date_part, timezone_suffix)) |
| 407 | } |
| 408 | |
| 409 | fn to_rfc9557_basic(&self) -> Outcome<String> { |
| 410 | let config = Rfc9557Config { |
| 411 | include_timezone_name: false, |
| 412 | ..Default::default() |
| 413 | }; |
| 414 | self.to_rfc9557_with_config(&config) |
| 415 | } |
| 416 | |
| 417 | fn to_rfc9557_extended(&self) -> Outcome<String> { |
| 418 | let config = Rfc9557Config { |
| 419 | include_timezone_name: true, |
| 420 | ..Default::default() |
| 421 | }; |
| 422 | self.to_rfc9557_with_config(&config) |
| 423 | } |
| 424 | |
| 425 | fn from_rfc9557(input: &str) -> Outcome<Self> { |
| 426 | let input = input.trim(); |
| 427 | |
| 428 | // Extract timezone name if present. |
| 429 | let (date_part, timezone_name) = if let Some(bracket_start) = input.rfind('[') { |
| 430 | if let Some(bracket_end) = input.rfind(']') { |
| 431 | if bracket_end > bracket_start { |
| 432 | let timezone_part = &input[bracket_start + 1..bracket_end]; |
| 433 | let date_part = &input[..bracket_start]; |
| 434 | (date_part, Some(timezone_part)) |
| 435 | } else { |
| 436 | (input, None) |
| 437 | } |
| 438 | } else { |
| 439 | (input, None) |
| 440 | } |
| 441 | } else { |
| 442 | (input, None) |
| 443 | }; |
| 444 | |
| 445 | // Parse date components. |
| 446 | let parts: Vec<&str> = date_part.split('-').collect(); |
| 447 | if parts.len() != 3 { |
| 448 | return Err(err!("Invalid date format: {}", input; Invalid, Input)); |
| 449 | } |
| 450 | |
| 451 | let year = res!(parts[0].parse::<i32>() |
| 452 | .map_err(|_| err!("Invalid year: {}", parts[0]; Invalid, Input))); |
| 453 | let month = res!(parts[1].parse::<u8>() |
| 454 | .map_err(|_| err!("Invalid month: {}", parts[1]; Invalid, Input))); |
| 455 | let day = res!(parts[2].parse::<u8>() |
| 456 | .map_err(|_| err!("Invalid day: {}", parts[2]; Invalid, Input))); |
| 457 | |
| 458 | // Determine timezone. |
| 459 | let zone = if let Some(tz_name) = timezone_name { |
| 460 | res!(CalClockZone::new(tz_name)) |
| 461 | } else { |
| 462 | CalClockZone::utc() |
| 463 | }; |
| 464 | |
| 465 | Self::new(year, month, day, zone) |
| 466 | } |
| 467 | } |
| 468 | |
| 469 | pub mod utils { |
| 470 | use super::*; |
| 471 | |
| 472 | pub fn validate_rfc9557(input: &str) -> Outcome<()> { |
| 473 | // Try to parse as CalClock to validate format. |
| 474 | match CalClock::from_rfc9557(input) { |
| 475 | Ok(_) => Ok(()), |
| 476 | Err(_) => Err(err!("Invalid RFC 9557 format: {}", input; Invalid, Input)), |
| 477 | } |
| 478 | } |
| 479 | |
| 480 | pub fn extract_timezone_name(input: &str) -> Option<&str> { |
| 481 | if let Some(bracket_start) = input.rfind('[') { |
| 482 | if let Some(bracket_end) = input.rfind(']') { |
| 483 | if bracket_end > bracket_start { |
| 484 | return Some(&input[bracket_start + 1..bracket_end]); |
| 485 | } |
| 486 | } |
| 487 | } |
| 488 | None |
| 489 | } |
| 490 | |
| 491 | pub fn extract_precision_indicator(input: &str) -> Option<PrecisionLevel> { |
| 492 | if input.ends_with('~') { |
| 493 | Some(PrecisionLevel::Approximate) |
| 494 | } else if input.ends_with('%') { |
| 495 | Some(PrecisionLevel::Uncertain) |
| 496 | } else if input.ends_with('@') { |
| 497 | Some(PrecisionLevel::Around) |
| 498 | } else if input.ends_with('*') { |
| 499 | Some(PrecisionLevel::Between) |
| 500 | } else { |
| 501 | None |
| 502 | } |
| 503 | } |
| 504 | } |