1 | /*! |
2 | Support for printing and parsing instants using the [RFC 2822] datetime format. |
3 | |
4 | RFC 2822 is most commonly found when dealing with email messages. |
5 | |
6 | Since RFC 2822 only supports specifying a complete instant in time, the parser |
7 | and printer in this module only use [`Zoned`] and [`Timestamp`]. If you need |
8 | inexact time, you can get it from [`Zoned`] via [`Zoned::datetime`]. |
9 | |
10 | [RFC 2822]: https://datatracker.ietf.org/doc/html/rfc2822 |
11 | |
12 | # Incomplete support |
13 | |
14 | The RFC 2822 support in this crate is technically incomplete. Specifically, |
15 | it does not support parsing comments within folding whitespace. It will parse |
16 | comments after the datetime itself (including nested comments). See [Issue |
17 | #39][issue39] for an example. If you find a real world use case for parsing |
18 | comments within whitespace at any point in the datetime string, please file |
19 | an issue. That is, the main reason it isn't currently supported is because |
20 | it didn't seem worth the implementation complexity to account for it. But if |
21 | there are real world use cases that need it, then that would be sufficient |
22 | justification for adding it. |
23 | |
24 | RFC 2822 support should otherwise be complete, including support for parsing |
25 | obselete offsets. |
26 | |
27 | [issue39]: https://github.com/BurntSushi/jiff/issues/39 |
28 | |
29 | # Warning |
30 | |
31 | The RFC 2822 format only supports writing a precise instant in time |
32 | expressed via a time zone offset. It does *not* support serializing |
33 | the time zone itself. This means that if you format a zoned datetime |
34 | in a time zone like `America/New_York` and then deserialize it, the |
35 | zoned datetime you get back will be a "fixed offset" zoned datetime. |
36 | This in turn means it will not perform daylight saving time safe |
37 | arithmetic. |
38 | |
39 | Basically, you should use the RFC 2822 format if it's required (for |
40 | example, when dealing with email). But you should not choose it as a |
41 | general interchange format for new applications. |
42 | */ |
43 | |
44 | use crate::{ |
45 | civil::{Date, DateTime, Time, Weekday}, |
46 | error::{err, ErrorContext}, |
47 | fmt::{util::DecimalFormatter, Parsed, Write, WriteExt}, |
48 | tz::{Offset, TimeZone}, |
49 | util::{ |
50 | escape, parse, |
51 | rangeint::{ri8, RFrom}, |
52 | t::{self, C}, |
53 | }, |
54 | Error, Timestamp, Zoned, |
55 | }; |
56 | |
57 | /// The default date time parser that we use throughout Jiff. |
58 | pub(crate) static DEFAULT_DATETIME_PARSER: DateTimeParser = |
59 | DateTimeParser::new(); |
60 | |
61 | /// The default date time printer that we use throughout Jiff. |
62 | pub(crate) static DEFAULT_DATETIME_PRINTER: DateTimePrinter = |
63 | DateTimePrinter::new(); |
64 | |
65 | /// Convert a [`Zoned`] to an [RFC 2822] datetime string. |
66 | /// |
67 | /// This is a convenience function for using [`DateTimePrinter`]. In |
68 | /// particular, this always creates and allocates a new `String`. For writing |
69 | /// to an existing string, or converting a [`Timestamp`] to an RFC 2822 |
70 | /// datetime string, you'll need to use `DateTimePrinter`. |
71 | /// |
72 | /// [RFC 2822]: https://datatracker.ietf.org/doc/html/rfc2822 |
73 | /// |
74 | /// # Warning |
75 | /// |
76 | /// The RFC 2822 format only supports writing a precise instant in time |
77 | /// expressed via a time zone offset. It does *not* support serializing |
78 | /// the time zone itself. This means that if you format a zoned datetime |
79 | /// in a time zone like `America/New_York` and then deserialize it, the |
80 | /// zoned datetime you get back will be a "fixed offset" zoned datetime. |
81 | /// This in turn means it will not perform daylight saving time safe |
82 | /// arithmetic. |
83 | /// |
84 | /// Basically, you should use the RFC 2822 format if it's required (for |
85 | /// example, when dealing with email). But you should not choose it as a |
86 | /// general interchange format for new applications. |
87 | /// |
88 | /// # Errors |
89 | /// |
90 | /// This returns an error if the year corresponding to this timestamp cannot be |
91 | /// represented in the RFC 2822 format. For example, a negative year. |
92 | /// |
93 | /// # Example |
94 | /// |
95 | /// This example shows how to convert a zoned datetime to the RFC 2822 format: |
96 | /// |
97 | /// ``` |
98 | /// use jiff::{civil::date, fmt::rfc2822}; |
99 | /// |
100 | /// let zdt = date(2024, 6, 15).at(7, 0, 0, 0).in_tz("Australia/Tasmania" )?; |
101 | /// assert_eq!(rfc2822::to_string(&zdt)?, "Sat, 15 Jun 2024 07:00:00 +1000" ); |
102 | /// |
103 | /// # Ok::<(), Box<dyn std::error::Error>>(()) |
104 | /// ``` |
105 | #[cfg (feature = "alloc" )] |
106 | #[inline ] |
107 | pub fn to_string(zdt: &Zoned) -> Result<alloc::string::String, Error> { |
108 | let mut buf: String = alloc::string::String::new(); |
109 | DEFAULT_DATETIME_PRINTER.print_zoned(zdt, &mut buf)?; |
110 | Ok(buf) |
111 | } |
112 | |
113 | /// Parse an [RFC 2822] datetime string into a [`Zoned`]. |
114 | /// |
115 | /// This is a convenience function for using [`DateTimeParser`]. In particular, |
116 | /// this takes a `&str` while the `DateTimeParser` accepts a `&[u8]`. |
117 | /// Moreover, if any configuration options are added to RFC 2822 parsing (none |
118 | /// currently exist at time of writing), then it will be necessary to use a |
119 | /// `DateTimeParser` to toggle them. Additionally, a `DateTimeParser` is needed |
120 | /// for parsing into a [`Timestamp`]. |
121 | /// |
122 | /// [RFC 2822]: https://datatracker.ietf.org/doc/html/rfc2822 |
123 | /// |
124 | /// # Warning |
125 | /// |
126 | /// The RFC 2822 format only supports writing a precise instant in time |
127 | /// expressed via a time zone offset. It does *not* support serializing |
128 | /// the time zone itself. This means that if you format a zoned datetime |
129 | /// in a time zone like `America/New_York` and then deserialize it, the |
130 | /// zoned datetime you get back will be a "fixed offset" zoned datetime. |
131 | /// This in turn means it will not perform daylight saving time safe |
132 | /// arithmetic. |
133 | /// |
134 | /// Basically, you should use the RFC 2822 format if it's required (for |
135 | /// example, when dealing with email). But you should not choose it as a |
136 | /// general interchange format for new applications. |
137 | /// |
138 | /// # Errors |
139 | /// |
140 | /// This returns an error if the datetime string given is invalid or if it |
141 | /// is valid but doesn't fit in the datetime range supported by Jiff. For |
142 | /// example, RFC 2822 supports offsets up to 99 hours and 59 minutes, |
143 | /// but Jiff's maximum offset is 25 hours, 59 minutes and 59 seconds. |
144 | /// |
145 | /// # Example |
146 | /// |
147 | /// This example shows how serializing a zoned datetime to RFC 2822 format |
148 | /// and then deserializing will drop information: |
149 | /// |
150 | /// ``` |
151 | /// use jiff::{civil::date, fmt::rfc2822}; |
152 | /// |
153 | /// let zdt = date(2024, 7, 13) |
154 | /// .at(15, 9, 59, 789_000_000) |
155 | /// .in_tz("America/New_York" )?; |
156 | /// // The default format (i.e., Temporal) guarantees lossless |
157 | /// // serialization. |
158 | /// assert_eq!(zdt.to_string(), "2024-07-13T15:09:59.789-04:00[America/New_York]" ); |
159 | /// |
160 | /// let rfc2822 = rfc2822::to_string(&zdt)?; |
161 | /// // Notice that the time zone name and fractional seconds have been dropped! |
162 | /// assert_eq!(rfc2822, "Sat, 13 Jul 2024 15:09:59 -0400" ); |
163 | /// // And of course, if we parse it back, all that info is still lost. |
164 | /// // Which means this `zdt` cannot do DST safe arithmetic! |
165 | /// let zdt = rfc2822::parse(&rfc2822)?; |
166 | /// assert_eq!(zdt.to_string(), "2024-07-13T15:09:59-04:00[-04:00]" ); |
167 | /// |
168 | /// # Ok::<(), Box<dyn std::error::Error>>(()) |
169 | /// ``` |
170 | #[inline ] |
171 | pub fn parse(string: &str) -> Result<Zoned, Error> { |
172 | DEFAULT_DATETIME_PARSER.parse_zoned(input:string) |
173 | } |
174 | |
175 | /// A parser for [RFC 2822] datetimes. |
176 | /// |
177 | /// [RFC 2822]: https://datatracker.ietf.org/doc/html/rfc2822 |
178 | /// |
179 | /// # Warning |
180 | /// |
181 | /// The RFC 2822 format only supports writing a precise instant in time |
182 | /// expressed via a time zone offset. It does *not* support serializing |
183 | /// the time zone itself. This means that if you format a zoned datetime |
184 | /// in a time zone like `America/New_York` and then deserialize it, the |
185 | /// zoned datetime you get back will be a "fixed offset" zoned datetime. |
186 | /// This in turn means it will not perform daylight saving time safe |
187 | /// arithmetic. |
188 | /// |
189 | /// Basically, you should use the RFC 2822 format if it's required (for |
190 | /// example, when dealing with email). But you should not choose it as a |
191 | /// general interchange format for new applications. |
192 | /// |
193 | /// # Example |
194 | /// |
195 | /// This example shows how serializing a zoned datetime to RFC 2822 format |
196 | /// and then deserializing will drop information: |
197 | /// |
198 | /// ``` |
199 | /// use jiff::{civil::date, fmt::rfc2822}; |
200 | /// |
201 | /// let zdt = date(2024, 7, 13) |
202 | /// .at(15, 9, 59, 789_000_000) |
203 | /// .in_tz("America/New_York" )?; |
204 | /// // The default format (i.e., Temporal) guarantees lossless |
205 | /// // serialization. |
206 | /// assert_eq!(zdt.to_string(), "2024-07-13T15:09:59.789-04:00[America/New_York]" ); |
207 | /// |
208 | /// let rfc2822 = rfc2822::to_string(&zdt)?; |
209 | /// // Notice that the time zone name and fractional seconds have been dropped! |
210 | /// assert_eq!(rfc2822, "Sat, 13 Jul 2024 15:09:59 -0400" ); |
211 | /// // And of course, if we parse it back, all that info is still lost. |
212 | /// // Which means this `zdt` cannot do DST safe arithmetic! |
213 | /// let zdt = rfc2822::parse(&rfc2822)?; |
214 | /// assert_eq!(zdt.to_string(), "2024-07-13T15:09:59-04:00[-04:00]" ); |
215 | /// |
216 | /// # Ok::<(), Box<dyn std::error::Error>>(()) |
217 | /// ``` |
218 | #[derive (Debug)] |
219 | pub struct DateTimeParser { |
220 | relaxed_weekday: bool, |
221 | } |
222 | |
223 | impl DateTimeParser { |
224 | /// Create a new RFC 2822 datetime parser with the default configuration. |
225 | #[inline ] |
226 | pub const fn new() -> DateTimeParser { |
227 | DateTimeParser { relaxed_weekday: false } |
228 | } |
229 | |
230 | /// When enabled, parsing will permit the weekday to be inconsistent with |
231 | /// the date. When enabled, the weekday is still parsed and can result in |
232 | /// an error if it isn't _a_ valid weekday. Only the error checking for |
233 | /// whether it is _the_ correct weekday for the parsed date is disabled. |
234 | /// |
235 | /// This is sometimes useful for interaction with systems that don't do |
236 | /// strict error checking. |
237 | /// |
238 | /// This is disabled by default. And note that RFC 2822 compliance requires |
239 | /// that the weekday is consistent with the date. |
240 | /// |
241 | /// # Example |
242 | /// |
243 | /// ``` |
244 | /// use jiff::{civil::date, fmt::rfc2822}; |
245 | /// |
246 | /// let string = "Sun, 13 Jul 2024 15:09:59 -0400" ; |
247 | /// // The above normally results in an error, since 2024-07-13 is a |
248 | /// // Saturday: |
249 | /// assert!(rfc2822::parse(string).is_err()); |
250 | /// // But we can relax the error checking: |
251 | /// static P: rfc2822::DateTimeParser = rfc2822::DateTimeParser::new() |
252 | /// .relaxed_weekday(true); |
253 | /// assert_eq!( |
254 | /// P.parse_zoned(string)?, |
255 | /// date(2024, 7, 13).at(15, 9, 59, 0).in_tz("America/New_York" )?, |
256 | /// ); |
257 | /// // But note that something that isn't recognized as a valid weekday |
258 | /// // will still result in an error: |
259 | /// assert!(P.parse_zoned("Wat, 13 Jul 2024 15:09:59 -0400" ).is_err()); |
260 | /// |
261 | /// # Ok::<(), Box<dyn std::error::Error>>(()) |
262 | /// ``` |
263 | #[inline ] |
264 | pub const fn relaxed_weekday(self, yes: bool) -> DateTimeParser { |
265 | DateTimeParser { relaxed_weekday: yes, ..self } |
266 | } |
267 | |
268 | /// Parse a datetime string into a [`Zoned`] value. |
269 | /// |
270 | /// Note that RFC 2822 does not support time zone annotations. The zoned |
271 | /// datetime returned will therefore always have a fixed offset time zone. |
272 | /// |
273 | /// # Warning |
274 | /// |
275 | /// The RFC 2822 format only supports writing a precise instant in time |
276 | /// expressed via a time zone offset. It does *not* support serializing |
277 | /// the time zone itself. This means that if you format a zoned datetime |
278 | /// in a time zone like `America/New_York` and then deserialize it, the |
279 | /// zoned datetime you get back will be a "fixed offset" zoned datetime. |
280 | /// This in turn means it will not perform daylight saving time safe |
281 | /// arithmetic. |
282 | /// |
283 | /// Basically, you should use the RFC 2822 format if it's required (for |
284 | /// example, when dealing with email). But you should not choose it as a |
285 | /// general interchange format for new applications. |
286 | /// |
287 | /// # Errors |
288 | /// |
289 | /// This returns an error if the datetime string given is invalid or if it |
290 | /// is valid but doesn't fit in the datetime range supported by Jiff. For |
291 | /// example, RFC 2822 supports offsets up to 99 hours and 59 minutes, |
292 | /// but Jiff's maximum offset is 25 hours, 59 minutes and 59 seconds. |
293 | /// |
294 | /// # Example |
295 | /// |
296 | /// This shows a basic example of parsing a `Timestamp` from an RFC 2822 |
297 | /// datetime string. |
298 | /// |
299 | /// ``` |
300 | /// use jiff::fmt::rfc2822::DateTimeParser; |
301 | /// |
302 | /// static PARSER: DateTimeParser = DateTimeParser::new(); |
303 | /// |
304 | /// let zdt = PARSER.parse_zoned("Thu, 29 Feb 2024 05:34 -0500" )?; |
305 | /// assert_eq!(zdt.to_string(), "2024-02-29T05:34:00-05:00[-05:00]" ); |
306 | /// |
307 | /// # Ok::<(), Box<dyn std::error::Error>>(()) |
308 | /// ``` |
309 | pub fn parse_zoned<I: AsRef<[u8]>>( |
310 | &self, |
311 | input: I, |
312 | ) -> Result<Zoned, Error> { |
313 | let input = input.as_ref(); |
314 | let zdt = self |
315 | .parse_zoned_internal(input) |
316 | .context( |
317 | "failed to parse RFC 2822 datetime into Jiff zoned datetime" , |
318 | )? |
319 | .into_full()?; |
320 | Ok(zdt) |
321 | } |
322 | |
323 | /// Parse an RFC 2822 datetime string into a [`Timestamp`]. |
324 | /// |
325 | /// # Errors |
326 | /// |
327 | /// This returns an error if the datetime string given is invalid or if it |
328 | /// is valid but doesn't fit in the datetime range supported by Jiff. For |
329 | /// example, RFC 2822 supports offsets up to 99 hours and 59 minutes, |
330 | /// but Jiff's maximum offset is 25 hours, 59 minutes and 59 seconds. |
331 | /// |
332 | /// # Example |
333 | /// |
334 | /// This shows a basic example of parsing a `Timestamp` from an RFC 2822 |
335 | /// datetime string. |
336 | /// |
337 | /// ``` |
338 | /// use jiff::fmt::rfc2822::DateTimeParser; |
339 | /// |
340 | /// static PARSER: DateTimeParser = DateTimeParser::new(); |
341 | /// |
342 | /// let timestamp = PARSER.parse_timestamp("Thu, 29 Feb 2024 05:34 -0500" )?; |
343 | /// assert_eq!(timestamp.to_string(), "2024-02-29T10:34:00Z" ); |
344 | /// |
345 | /// # Ok::<(), Box<dyn std::error::Error>>(()) |
346 | /// ``` |
347 | pub fn parse_timestamp<I: AsRef<[u8]>>( |
348 | &self, |
349 | input: I, |
350 | ) -> Result<Timestamp, Error> { |
351 | let input = input.as_ref(); |
352 | let ts = self |
353 | .parse_timestamp_internal(input) |
354 | .context("failed to parse RFC 2822 datetime into Jiff timestamp" )? |
355 | .into_full()?; |
356 | Ok(ts) |
357 | } |
358 | |
359 | /// Parses an RFC 2822 datetime as a zoned datetime. |
360 | /// |
361 | /// Note that this doesn't check that the input has been completely |
362 | /// consumed. |
363 | #[cfg_attr (feature = "perf-inline" , inline(always))] |
364 | fn parse_zoned_internal<'i>( |
365 | &self, |
366 | input: &'i [u8], |
367 | ) -> Result<Parsed<'i, Zoned>, Error> { |
368 | let Parsed { value: (dt, offset), input } = |
369 | self.parse_datetime_offset(input)?; |
370 | let ts = offset |
371 | .to_timestamp(dt) |
372 | .context("RFC 2822 datetime out of Jiff's range" )?; |
373 | let zdt = ts.to_zoned(TimeZone::fixed(offset)); |
374 | Ok(Parsed { value: zdt, input }) |
375 | } |
376 | |
377 | /// Parses an RFC 2822 datetime as a timestamp. |
378 | /// |
379 | /// Note that this doesn't check that the input has been completely |
380 | /// consumed. |
381 | #[cfg_attr (feature = "perf-inline" , inline(always))] |
382 | fn parse_timestamp_internal<'i>( |
383 | &self, |
384 | input: &'i [u8], |
385 | ) -> Result<Parsed<'i, Timestamp>, Error> { |
386 | let Parsed { value: (dt, offset), input } = |
387 | self.parse_datetime_offset(input)?; |
388 | let ts = offset |
389 | .to_timestamp(dt) |
390 | .context("RFC 2822 datetime out of Jiff's range" )?; |
391 | Ok(Parsed { value: ts, input }) |
392 | } |
393 | |
394 | /// Parse the entirety of the given input into RFC 2822 components: a civil |
395 | /// datetime and its offset. |
396 | /// |
397 | /// This also consumes any trailing (superfluous) whitespace. |
398 | #[cfg_attr (feature = "perf-inline" , inline(always))] |
399 | fn parse_datetime_offset<'i>( |
400 | &self, |
401 | input: &'i [u8], |
402 | ) -> Result<Parsed<'i, (DateTime, Offset)>, Error> { |
403 | let input = input.as_ref(); |
404 | let Parsed { value: dt, input } = self.parse_datetime(input)?; |
405 | let Parsed { value: offset, input } = self.parse_offset(input)?; |
406 | let Parsed { input, .. } = self.skip_whitespace(input); |
407 | let input = if input.is_empty() { |
408 | input |
409 | } else { |
410 | self.skip_comment(input)?.input |
411 | }; |
412 | Ok(Parsed { value: (dt, offset), input }) |
413 | } |
414 | |
415 | /// Parses a civil datetime from an RFC 2822 string. The input may have |
416 | /// leading whitespace. |
417 | /// |
418 | /// This also parses and trailing whitespace, including requiring at least |
419 | /// one whitespace character. |
420 | /// |
421 | /// This basically parses everything except for the zone. |
422 | #[cfg_attr (feature = "perf-inline" , inline(always))] |
423 | fn parse_datetime<'i>( |
424 | &self, |
425 | input: &'i [u8], |
426 | ) -> Result<Parsed<'i, DateTime>, Error> { |
427 | if input.is_empty() { |
428 | return Err(err!( |
429 | "expected RFC 2822 datetime, but got empty string" |
430 | )); |
431 | } |
432 | let Parsed { input, .. } = self.skip_whitespace(input); |
433 | if input.is_empty() { |
434 | return Err(err!( |
435 | "expected RFC 2822 datetime, but got empty string after \ |
436 | trimming whitespace" , |
437 | )); |
438 | } |
439 | let Parsed { value: wd, input } = self.parse_weekday(input)?; |
440 | let Parsed { value: day, input } = self.parse_day(input)?; |
441 | let Parsed { value: month, input } = self.parse_month(input)?; |
442 | let Parsed { value: year, input } = self.parse_year(input)?; |
443 | |
444 | let Parsed { value: hour, input } = self.parse_hour(input)?; |
445 | let Parsed { input, .. } = self.parse_time_separator(input)?; |
446 | let Parsed { value: minute, input } = self.parse_minute(input)?; |
447 | let (second, input) = if !input.starts_with(b":" ) { |
448 | (t::Second::N::<0>(), input) |
449 | } else { |
450 | let Parsed { input, .. } = self.parse_time_separator(input)?; |
451 | let Parsed { value: second, input } = self.parse_second(input)?; |
452 | (second, input) |
453 | }; |
454 | let Parsed { input, .. } = self |
455 | .parse_whitespace(input) |
456 | .with_context(|| err!("expected whitespace after parsing time" ))?; |
457 | |
458 | let date = |
459 | Date::new_ranged(year, month, day).context("invalid date" )?; |
460 | let time = Time::new_ranged( |
461 | hour, |
462 | minute, |
463 | second, |
464 | t::SubsecNanosecond::N::<0>(), |
465 | ); |
466 | let dt = DateTime::from_parts(date, time); |
467 | if let Some(wd) = wd { |
468 | if !self.relaxed_weekday && wd != dt.weekday() { |
469 | return Err(err!( |
470 | "found parsed weekday of {parsed}, \ |
471 | but parsed datetime of {dt} has weekday \ |
472 | {has}" , |
473 | parsed = weekday_abbrev(wd), |
474 | has = weekday_abbrev(dt.weekday()), |
475 | )); |
476 | } |
477 | } |
478 | Ok(Parsed { value: dt, input }) |
479 | } |
480 | |
481 | /// Parses an optional weekday at the beginning of an RFC 2822 datetime. |
482 | /// |
483 | /// This expects that any optional whitespace preceding the start of an |
484 | /// optional day has been stripped and that the input has at least one |
485 | /// byte. |
486 | /// |
487 | /// When the first byte of the given input is a digit (or is empty), then |
488 | /// this returns `None`, as it implies a day is not present. But if it |
489 | /// isn't a digit, then we assume that it must be a weekday and return an |
490 | /// error based on that assumption if we couldn't recognize a weekday. |
491 | /// |
492 | /// If a weekday is parsed, then this also skips any trailing whitespace |
493 | /// (and requires at least one whitespace character). |
494 | #[cfg_attr (feature = "perf-inline" , inline(always))] |
495 | fn parse_weekday<'i>( |
496 | &self, |
497 | input: &'i [u8], |
498 | ) -> Result<Parsed<'i, Option<Weekday>>, Error> { |
499 | // An empty input is invalid, but we let that case be |
500 | // handled by the caller. Otherwise, we know there MUST |
501 | // be a present day if the first character isn't an ASCII |
502 | // digit. |
503 | if matches!(input[0], b'0' ..=b'9' ) { |
504 | return Ok(Parsed { value: None, input }); |
505 | } |
506 | if input.len() < 4 { |
507 | return Err(err!( |
508 | "expected day at beginning of RFC 2822 datetime \ |
509 | since first non-whitespace byte, {first:?}, \ |
510 | is not a digit, but given string is too short \ |
511 | (length is {length})" , |
512 | first = escape::Byte(input[0]), |
513 | length = input.len(), |
514 | )); |
515 | } |
516 | let b1 = input[0].to_ascii_lowercase(); |
517 | let b2 = input[1].to_ascii_lowercase(); |
518 | let b3 = input[2].to_ascii_lowercase(); |
519 | let wd = match &[b1, b2, b3] { |
520 | b"sun" => Weekday::Sunday, |
521 | b"mon" => Weekday::Monday, |
522 | b"tue" => Weekday::Tuesday, |
523 | b"wed" => Weekday::Wednesday, |
524 | b"thu" => Weekday::Thursday, |
525 | b"fri" => Weekday::Friday, |
526 | b"sat" => Weekday::Saturday, |
527 | _ => { |
528 | return Err(err!( |
529 | "expected day at beginning of RFC 2822 datetime \ |
530 | since first non-whitespace byte, {first:?}, \ |
531 | is not a digit, but did not recognize {got:?} \ |
532 | as a valid weekday abbreviation" , |
533 | first = escape::Byte(input[0]), |
534 | got = escape::Bytes(&input[..3]), |
535 | )); |
536 | } |
537 | }; |
538 | if input[3] != b',' { |
539 | return Err(err!( |
540 | "expected day at beginning of RFC 2822 datetime \ |
541 | since first non-whitespace byte, {first:?}, \ |
542 | is not a digit, but found {got:?} after parsed \ |
543 | weekday {wd:?} and expected a comma" , |
544 | first = escape::Byte(input[0]), |
545 | got = escape::Byte(input[3]), |
546 | wd = escape::Bytes(&input[..3]), |
547 | )); |
548 | } |
549 | let Parsed { input, .. } = |
550 | self.parse_whitespace(&input[4..]).with_context(|| { |
551 | err!( |
552 | "expected whitespace after parsing {got:?}" , |
553 | got = escape::Bytes(&input[..4]), |
554 | ) |
555 | })?; |
556 | Ok(Parsed { value: Some(wd), input }) |
557 | } |
558 | |
559 | /// Parses a 1 or 2 digit day. |
560 | /// |
561 | /// This assumes the input starts with what must be an ASCII digit (or it |
562 | /// may be empty). |
563 | /// |
564 | /// This also parses at least one mandatory whitespace character after the |
565 | /// day. |
566 | #[cfg_attr (feature = "perf-inline" , inline(always))] |
567 | fn parse_day<'i>( |
568 | &self, |
569 | input: &'i [u8], |
570 | ) -> Result<Parsed<'i, t::Day>, Error> { |
571 | if input.is_empty() { |
572 | return Err(err!("expected day, but found end of input" )); |
573 | } |
574 | let mut digits = 1; |
575 | if input.len() >= 2 && matches!(input[1], b'0' ..=b'9' ) { |
576 | digits = 2; |
577 | } |
578 | let (day, input) = input.split_at(digits); |
579 | let day = parse::i64(day).with_context(|| { |
580 | err!("failed to parse {day:?} as day" , day = escape::Bytes(day)) |
581 | })?; |
582 | let day = t::Day::try_new("day" , day).context("day is not valid" )?; |
583 | let Parsed { input, .. } = |
584 | self.parse_whitespace(input).with_context(|| { |
585 | err!("expected whitespace after parsing day {day}" ) |
586 | })?; |
587 | Ok(Parsed { value: day, input }) |
588 | } |
589 | |
590 | /// Parses an abbreviated month name. |
591 | /// |
592 | /// This assumes the input starts with what must be the beginning of a |
593 | /// month name (or the input may be empty). |
594 | /// |
595 | /// This also parses at least one mandatory whitespace character after the |
596 | /// month name. |
597 | #[cfg_attr (feature = "perf-inline" , inline(always))] |
598 | fn parse_month<'i>( |
599 | &self, |
600 | input: &'i [u8], |
601 | ) -> Result<Parsed<'i, t::Month>, Error> { |
602 | if input.is_empty() { |
603 | return Err(err!( |
604 | "expected abbreviated month name, but found end of input" |
605 | )); |
606 | } |
607 | if input.len() < 3 { |
608 | return Err(err!( |
609 | "expected abbreviated month name, but remaining input \ |
610 | is too short (remaining bytes is {length})" , |
611 | length = input.len(), |
612 | )); |
613 | } |
614 | let b1 = input[0].to_ascii_lowercase(); |
615 | let b2 = input[1].to_ascii_lowercase(); |
616 | let b3 = input[2].to_ascii_lowercase(); |
617 | let month = match &[b1, b2, b3] { |
618 | b"jan" => 1, |
619 | b"feb" => 2, |
620 | b"mar" => 3, |
621 | b"apr" => 4, |
622 | b"may" => 5, |
623 | b"jun" => 6, |
624 | b"jul" => 7, |
625 | b"aug" => 8, |
626 | b"sep" => 9, |
627 | b"oct" => 10, |
628 | b"nov" => 11, |
629 | b"dec" => 12, |
630 | _ => { |
631 | return Err(err!( |
632 | "expected abbreviated month name, \ |
633 | but did not recognize {got:?} \ |
634 | as a valid month" , |
635 | got = escape::Bytes(&input[..3]), |
636 | )); |
637 | } |
638 | }; |
639 | // OK because we just assigned a numeric value ourselves |
640 | // above, and all values are valid months. |
641 | let month = t::Month::new(month).unwrap(); |
642 | let Parsed { input, .. } = |
643 | self.parse_whitespace(&input[3..]).with_context(|| { |
644 | err!("expected whitespace after parsing month name" ) |
645 | })?; |
646 | Ok(Parsed { value: month, input }) |
647 | } |
648 | |
649 | /// Parses a 2, 3 or 4 digit year. |
650 | /// |
651 | /// This assumes the input starts with what must be an ASCII digit (or it |
652 | /// may be empty). |
653 | /// |
654 | /// This also parses at least one mandatory whitespace character after the |
655 | /// day. |
656 | /// |
657 | /// The 2 or 3 digit years are "obsolete," which we support by following |
658 | /// the rules in RFC 2822: |
659 | /// |
660 | /// > Where a two or three digit year occurs in a date, the year is to be |
661 | /// > interpreted as follows: If a two digit year is encountered whose |
662 | /// > value is between 00 and 49, the year is interpreted by adding 2000, |
663 | /// > ending up with a value between 2000 and 2049. If a two digit year is |
664 | /// > encountered with a value between 50 and 99, or any three digit year |
665 | /// > is encountered, the year is interpreted by adding 1900. |
666 | #[cfg_attr (feature = "perf-inline" , inline(always))] |
667 | fn parse_year<'i>( |
668 | &self, |
669 | input: &'i [u8], |
670 | ) -> Result<Parsed<'i, t::Year>, Error> { |
671 | let mut digits = 0; |
672 | while digits <= 3 |
673 | && !input[digits..].is_empty() |
674 | && matches!(input[digits], b'0' ..=b'9' ) |
675 | { |
676 | digits += 1; |
677 | } |
678 | if digits <= 1 { |
679 | return Err(err!( |
680 | "expected at least two ASCII digits for parsing \ |
681 | a year, but only found {digits}" , |
682 | )); |
683 | } |
684 | let (year, input) = input.split_at(digits); |
685 | let year = parse::i64(year).with_context(|| { |
686 | err!( |
687 | "failed to parse {year:?} as year \ |
688 | (a two, three or four digit integer)" , |
689 | year = escape::Bytes(year), |
690 | ) |
691 | })?; |
692 | let year = match digits { |
693 | 2 if year <= 49 => year + 2000, |
694 | 2 | 3 => year + 1900, |
695 | 4 => year, |
696 | _ => unreachable!("digits= {digits} must be 2, 3 or 4" ), |
697 | }; |
698 | let year = |
699 | t::Year::try_new("year" , year).context("year is not valid" )?; |
700 | let Parsed { input, .. } = self |
701 | .parse_whitespace(input) |
702 | .with_context(|| err!("expected whitespace after parsing year" ))?; |
703 | Ok(Parsed { value: year, input }) |
704 | } |
705 | |
706 | /// Parses a 2-digit hour. This assumes the input begins with what should |
707 | /// be an ASCII digit. (i.e., It doesn't trim leading whitespace.) |
708 | /// |
709 | /// This parses a mandatory trailing `:`, advancing the input to |
710 | /// immediately after it. |
711 | #[cfg_attr (feature = "perf-inline" , inline(always))] |
712 | fn parse_hour<'i>( |
713 | &self, |
714 | input: &'i [u8], |
715 | ) -> Result<Parsed<'i, t::Hour>, Error> { |
716 | let (hour, input) = parse::split(input, 2).ok_or_else(|| { |
717 | err!("expected two digit hour, but found end of input" ) |
718 | })?; |
719 | let hour = parse::i64(hour).with_context(|| { |
720 | err!( |
721 | "failed to parse {hour:?} as hour (a two digit integer)" , |
722 | hour = escape::Bytes(hour), |
723 | ) |
724 | })?; |
725 | let hour = |
726 | t::Hour::try_new("hour" , hour).context("hour is not valid" )?; |
727 | Ok(Parsed { value: hour, input }) |
728 | } |
729 | |
730 | /// Parses a 2-digit minute. This assumes the input begins with what should |
731 | /// be an ASCII digit. (i.e., It doesn't trim leading whitespace.) |
732 | #[cfg_attr (feature = "perf-inline" , inline(always))] |
733 | fn parse_minute<'i>( |
734 | &self, |
735 | input: &'i [u8], |
736 | ) -> Result<Parsed<'i, t::Minute>, Error> { |
737 | let (minute, input) = parse::split(input, 2).ok_or_else(|| { |
738 | err!("expected two digit minute, but found end of input" ) |
739 | })?; |
740 | let minute = parse::i64(minute).with_context(|| { |
741 | err!( |
742 | "failed to parse {minute:?} as minute (a two digit integer)" , |
743 | minute = escape::Bytes(minute), |
744 | ) |
745 | })?; |
746 | let minute = t::Minute::try_new("minute" , minute) |
747 | .context("minute is not valid" )?; |
748 | Ok(Parsed { value: minute, input }) |
749 | } |
750 | |
751 | /// Parses a 2-digit second. This assumes the input begins with what should |
752 | /// be an ASCII digit. (i.e., It doesn't trim leading whitespace.) |
753 | #[cfg_attr (feature = "perf-inline" , inline(always))] |
754 | fn parse_second<'i>( |
755 | &self, |
756 | input: &'i [u8], |
757 | ) -> Result<Parsed<'i, t::Second>, Error> { |
758 | let (second, input) = parse::split(input, 2).ok_or_else(|| { |
759 | err!("expected two digit second, but found end of input" ) |
760 | })?; |
761 | let mut second = parse::i64(second).with_context(|| { |
762 | err!( |
763 | "failed to parse {second:?} as second (a two digit integer)" , |
764 | second = escape::Bytes(second), |
765 | ) |
766 | })?; |
767 | if second == 60 { |
768 | second = 59; |
769 | } |
770 | let second = t::Second::try_new("second" , second) |
771 | .context("second is not valid" )?; |
772 | Ok(Parsed { value: second, input }) |
773 | } |
774 | |
775 | /// Parses a time zone offset (including obsolete offsets like EDT). |
776 | /// |
777 | /// This assumes the offset must begin at the beginning of `input`. That |
778 | /// is, any leading whitespace should already have been trimmed. |
779 | #[cfg_attr (feature = "perf-inline" , inline(always))] |
780 | fn parse_offset<'i>( |
781 | &self, |
782 | input: &'i [u8], |
783 | ) -> Result<Parsed<'i, Offset>, Error> { |
784 | type ParsedOffsetHours = ri8<0, { t::SpanZoneOffsetHours::MAX }>; |
785 | type ParsedOffsetMinutes = ri8<0, { t::SpanZoneOffsetMinutes::MAX }>; |
786 | |
787 | let sign = input.get(0).copied().ok_or_else(|| { |
788 | err!( |
789 | "expected sign for time zone offset, \ |
790 | (or a legacy time zone name abbreviation), \ |
791 | but found end of input" , |
792 | ) |
793 | })?; |
794 | let sign = if sign == b'+' { |
795 | t::Sign::N::<1>() |
796 | } else if sign == b'-' { |
797 | t::Sign::N::<-1>() |
798 | } else { |
799 | return self.parse_offset_obsolete(input); |
800 | }; |
801 | let input = &input[1..]; |
802 | let (hhmm, input) = parse::split(input, 4).ok_or_else(|| { |
803 | err!( |
804 | "expected at least 4 digits for time zone offset \ |
805 | after sign, but found only {len} bytes remaining" , |
806 | len = input.len(), |
807 | ) |
808 | })?; |
809 | |
810 | let hh = parse::i64(&hhmm[0..2]).with_context(|| { |
811 | err!( |
812 | "failed to parse hours from time zone offset {hhmm}" , |
813 | hhmm = escape::Bytes(hhmm) |
814 | ) |
815 | })?; |
816 | let hh = ParsedOffsetHours::try_new("zone-offset-hours" , hh) |
817 | .context("time zone offset hours are not valid" )?; |
818 | let hh = t::SpanZoneOffset::rfrom(hh); |
819 | |
820 | let mm = parse::i64(&hhmm[2..4]).with_context(|| { |
821 | err!( |
822 | "failed to parse minutes from time zone offset {hhmm}" , |
823 | hhmm = escape::Bytes(hhmm) |
824 | ) |
825 | })?; |
826 | let mm = ParsedOffsetMinutes::try_new("zone-offset-minutes" , mm) |
827 | .context("time zone offset minutes are not valid" )?; |
828 | let mm = t::SpanZoneOffset::rfrom(mm); |
829 | |
830 | let seconds = hh * C(3_600) + mm * C(60); |
831 | let offset = Offset::from_seconds_ranged(seconds * sign); |
832 | Ok(Parsed { value: offset, input }) |
833 | } |
834 | |
835 | /// Parses an obsolete time zone offset. |
836 | #[inline (never)] |
837 | fn parse_offset_obsolete<'i>( |
838 | &self, |
839 | input: &'i [u8], |
840 | ) -> Result<Parsed<'i, Offset>, Error> { |
841 | let mut letters = [0; 5]; |
842 | let mut len = 0; |
843 | while len <= 4 |
844 | && !input[len..].is_empty() |
845 | && !is_whitespace(input[len]) |
846 | { |
847 | letters[len] = input[len].to_ascii_lowercase(); |
848 | len += 1; |
849 | } |
850 | if len == 0 { |
851 | return Err(err!( |
852 | "expected obsolete RFC 2822 time zone abbreviation, \ |
853 | but found no remaining non-whitespace characters \ |
854 | after time" , |
855 | )); |
856 | } |
857 | let offset = match &letters[..len] { |
858 | b"ut" | b"gmt" | b"z" => Offset::UTC, |
859 | b"est" => Offset::constant(-5), |
860 | b"edt" => Offset::constant(-4), |
861 | b"cst" => Offset::constant(-6), |
862 | b"cdt" => Offset::constant(-5), |
863 | b"mst" => Offset::constant(-7), |
864 | b"mdt" => Offset::constant(-6), |
865 | b"pst" => Offset::constant(-8), |
866 | b"pdt" => Offset::constant(-7), |
867 | name => { |
868 | if name.len() == 1 |
869 | && matches!(name[0], b'a' ..=b'i' | b'k' ..=b'z' ) |
870 | { |
871 | // Section 4.3 indicates these as military time: |
872 | // |
873 | // > The 1 character military time zones were defined in |
874 | // > a non-standard way in [RFC822] and are therefore |
875 | // > unpredictable in their meaning. The original |
876 | // > definitions of the military zones "A" through "I" are |
877 | // > equivalent to "+0100" through "+0900" respectively; |
878 | // > "K", "L", and "M" are equivalent to "+1000", "+1100", |
879 | // > and "+1200" respectively; "N" through "Y" are |
880 | // > equivalent to "-0100" through "-1200" respectively; |
881 | // > and "Z" is equivalent to "+0000". However, because of |
882 | // > the error in [RFC822], they SHOULD all be considered |
883 | // > equivalent to "-0000" unless there is out-of-band |
884 | // > information confirming their meaning. |
885 | // |
886 | // So just treat them as UTC. |
887 | Offset::UTC |
888 | } else if name.len() >= 3 |
889 | && name.iter().all(|&b| matches!(b, b'a' ..=b'z' )) |
890 | { |
891 | // Section 4.3 also says that anything that _looks_ like a |
892 | // zone name should just be -0000 too: |
893 | // |
894 | // > Other multi-character (usually between 3 and 5) |
895 | // > alphabetic time zones have been used in Internet |
896 | // > messages. Any such time zone whose meaning is not |
897 | // > known SHOULD be considered equivalent to "-0000" |
898 | // > unless there is out-of-band information confirming |
899 | // > their meaning. |
900 | Offset::UTC |
901 | } else { |
902 | // But anything else we throw our hands up I guess. |
903 | return Err(err!( |
904 | "expected obsolete RFC 2822 time zone abbreviation, \ |
905 | but found {found:?}" , |
906 | found = escape::Bytes(&input[..len]), |
907 | )); |
908 | } |
909 | } |
910 | }; |
911 | Ok(Parsed { value: offset, input: &input[len..] }) |
912 | } |
913 | |
914 | /// Parses a time separator. This returns an error if one couldn't be |
915 | /// found. |
916 | #[cfg_attr (feature = "perf-inline" , inline(always))] |
917 | fn parse_time_separator<'i>( |
918 | &self, |
919 | input: &'i [u8], |
920 | ) -> Result<Parsed<'i, ()>, Error> { |
921 | if input.is_empty() { |
922 | return Err(err!( |
923 | "expected time separator of ':', but found end of input" , |
924 | )); |
925 | } |
926 | if input[0] != b':' { |
927 | return Err(err!( |
928 | "expected time separator of ':', but found {got}" , |
929 | got = escape::Byte(input[0]), |
930 | )); |
931 | } |
932 | Ok(Parsed { value: (), input: &input[1..] }) |
933 | } |
934 | |
935 | /// Parses at least one whitespace character. If no whitespace was found, |
936 | /// then this returns an error. |
937 | #[cfg_attr (feature = "perf-inline" , inline(always))] |
938 | fn parse_whitespace<'i>( |
939 | &self, |
940 | input: &'i [u8], |
941 | ) -> Result<Parsed<'i, ()>, Error> { |
942 | let oldlen = input.len(); |
943 | let parsed = self.skip_whitespace(input); |
944 | let newlen = parsed.input.len(); |
945 | if oldlen == newlen { |
946 | return Err(err!( |
947 | "expected at least one whitespace character (space or tab), \ |
948 | but found none" , |
949 | )); |
950 | } |
951 | Ok(parsed) |
952 | } |
953 | |
954 | /// Skips over any ASCII whitespace at the beginning of `input`. |
955 | /// |
956 | /// This returns the input unchanged if it does not begin with whitespace. |
957 | #[cfg_attr (feature = "perf-inline" , inline(always))] |
958 | fn skip_whitespace<'i>(&self, mut input: &'i [u8]) -> Parsed<'i, ()> { |
959 | while input.first().map_or(false, |&b| is_whitespace(b)) { |
960 | input = &input[1..]; |
961 | } |
962 | Parsed { value: (), input } |
963 | } |
964 | |
965 | /// This attempts to parse and skip any trailing "comment" in an RFC 2822 |
966 | /// datetime. |
967 | /// |
968 | /// This is a bit more relaxed than what RFC 2822 specifies. We basically |
969 | /// just try to balance parenthesis and skip over escapes. |
970 | /// |
971 | /// This assumes that if a comment exists, its opening parenthesis is at |
972 | /// the beginning of `input`. That is, any leading whitespace has been |
973 | /// stripped. |
974 | #[inline (never)] |
975 | fn skip_comment<'i>( |
976 | &self, |
977 | mut input: &'i [u8], |
978 | ) -> Result<Parsed<'i, ()>, Error> { |
979 | if !input.starts_with(b"(" ) { |
980 | return Ok(Parsed { value: (), input }); |
981 | } |
982 | input = &input[1..]; |
983 | let mut depth: u8 = 1; |
984 | let mut escape = false; |
985 | for byte in input.iter().copied() { |
986 | input = &input[1..]; |
987 | if escape { |
988 | escape = false; |
989 | } else if byte == b' \\' { |
990 | escape = true; |
991 | } else if byte == b')' { |
992 | // I believe this error case is actually impossible, since as |
993 | // soon as we hit 0, we break out. If there is more "comment," |
994 | // then it will flag an error as unparsed input. |
995 | depth = depth.checked_sub(1).ok_or_else(|| { |
996 | err!( |
997 | "found closing parenthesis in comment with \ |
998 | no matching opening parenthesis" |
999 | ) |
1000 | })?; |
1001 | if depth == 0 { |
1002 | break; |
1003 | } |
1004 | } else if byte == b'(' { |
1005 | depth = depth.checked_add(1).ok_or_else(|| { |
1006 | err!("found too many nested parenthesis in comment" ) |
1007 | })?; |
1008 | } |
1009 | } |
1010 | if depth > 0 { |
1011 | return Err(err!( |
1012 | "found opening parenthesis in comment with \ |
1013 | no matching closing parenthesis" |
1014 | )); |
1015 | } |
1016 | Ok(self.skip_whitespace(input)) |
1017 | } |
1018 | } |
1019 | |
1020 | /// A printer for [RFC 2822] datetimes. |
1021 | /// |
1022 | /// This printer converts an in memory representation of a precise instant in |
1023 | /// time to an RFC 2822 formatted string. That is, [`Zoned`] or [`Timestamp`], |
1024 | /// since all other datetime types in Jiff are inexact. |
1025 | /// |
1026 | /// [RFC 2822]: https://datatracker.ietf.org/doc/html/rfc2822 |
1027 | /// |
1028 | /// # Warning |
1029 | /// |
1030 | /// The RFC 2822 format only supports writing a precise instant in time |
1031 | /// expressed via a time zone offset. It does *not* support serializing |
1032 | /// the time zone itself. This means that if you format a zoned datetime |
1033 | /// in a time zone like `America/New_York` and then deserialize it, the |
1034 | /// zoned datetime you get back will be a "fixed offset" zoned datetime. |
1035 | /// This in turn means it will not perform daylight saving time safe |
1036 | /// arithmetic. |
1037 | /// |
1038 | /// Basically, you should use the RFC 2822 format if it's required (for |
1039 | /// example, when dealing with email). But you should not choose it as a |
1040 | /// general interchange format for new applications. |
1041 | /// |
1042 | /// # Example |
1043 | /// |
1044 | /// This example shows how to convert a zoned datetime to the RFC 2822 format: |
1045 | /// |
1046 | /// ``` |
1047 | /// use jiff::{civil::date, fmt::rfc2822::DateTimePrinter}; |
1048 | /// |
1049 | /// const PRINTER: DateTimePrinter = DateTimePrinter::new(); |
1050 | /// |
1051 | /// let zdt = date(2024, 6, 15).at(7, 0, 0, 0).in_tz("Australia/Tasmania" )?; |
1052 | /// |
1053 | /// let mut buf = String::new(); |
1054 | /// PRINTER.print_zoned(&zdt, &mut buf)?; |
1055 | /// assert_eq!(buf, "Sat, 15 Jun 2024 07:00:00 +1000" ); |
1056 | /// |
1057 | /// # Ok::<(), Box<dyn std::error::Error>>(()) |
1058 | /// ``` |
1059 | /// |
1060 | /// # Example: using adapters with `std::io::Write` and `std::fmt::Write` |
1061 | /// |
1062 | /// By using the [`StdIoWrite`](super::StdIoWrite) and |
1063 | /// [`StdFmtWrite`](super::StdFmtWrite) adapters, one can print datetimes |
1064 | /// directly to implementations of `std::io::Write` and `std::fmt::Write`, |
1065 | /// respectively. The example below demonstrates writing to anything |
1066 | /// that implements `std::io::Write`. Similar code can be written for |
1067 | /// `std::fmt::Write`. |
1068 | /// |
1069 | /// ```no_run |
1070 | /// use std::{fs::File, io::{BufWriter, Write}, path::Path}; |
1071 | /// |
1072 | /// use jiff::{civil::date, fmt::{StdIoWrite, rfc2822::DateTimePrinter}}; |
1073 | /// |
1074 | /// let zdt = date(2024, 6, 15).at(7, 0, 0, 0).in_tz("Asia/Kolkata" )?; |
1075 | /// |
1076 | /// let path = Path::new("/tmp/output" ); |
1077 | /// let mut file = BufWriter::new(File::create(path)?); |
1078 | /// DateTimePrinter::new().print_zoned(&zdt, StdIoWrite(&mut file)).unwrap(); |
1079 | /// file.flush()?; |
1080 | /// assert_eq!( |
1081 | /// std::fs::read_to_string(path)?, |
1082 | /// "Sat, 15 Jun 2024 07:00:00 +0530" , |
1083 | /// ); |
1084 | /// |
1085 | /// # Ok::<(), Box<dyn std::error::Error>>(()) |
1086 | /// ``` |
1087 | #[derive (Debug)] |
1088 | pub struct DateTimePrinter { |
1089 | // The RFC 2822 printer has no configuration at present. |
1090 | _private: (), |
1091 | } |
1092 | |
1093 | impl DateTimePrinter { |
1094 | /// Create a new RFC 2822 datetime printer with the default configuration. |
1095 | #[inline ] |
1096 | pub const fn new() -> DateTimePrinter { |
1097 | DateTimePrinter { _private: () } |
1098 | } |
1099 | |
1100 | /// Format a `Zoned` datetime into a string. |
1101 | /// |
1102 | /// This never emits `-0000` as the offset in the RFC 2822 format. If you |
1103 | /// desire a `-0000` offset, use [`DateTimePrinter::print_timestamp`] via |
1104 | /// [`Zoned::timestamp`]. |
1105 | /// |
1106 | /// Moreover, since RFC 2822 does not support fractional seconds, this |
1107 | /// routine prints the zoned datetime as if truncating any fractional |
1108 | /// seconds. |
1109 | /// |
1110 | /// This is a convenience routine for [`DateTimePrinter::print_zoned`] |
1111 | /// with a `String`. |
1112 | /// |
1113 | /// # Warning |
1114 | /// |
1115 | /// The RFC 2822 format only supports writing a precise instant in time |
1116 | /// expressed via a time zone offset. It does *not* support serializing |
1117 | /// the time zone itself. This means that if you format a zoned datetime |
1118 | /// in a time zone like `America/New_York` and then deserialize it, the |
1119 | /// zoned datetime you get back will be a "fixed offset" zoned datetime. |
1120 | /// This in turn means it will not perform daylight saving time safe |
1121 | /// arithmetic. |
1122 | /// |
1123 | /// Basically, you should use the RFC 2822 format if it's required (for |
1124 | /// example, when dealing with email). But you should not choose it as a |
1125 | /// general interchange format for new applications. |
1126 | /// |
1127 | /// # Errors |
1128 | /// |
1129 | /// This can return an error if the year corresponding to this timestamp |
1130 | /// cannot be represented in the RFC 2822 format. For example, a negative |
1131 | /// year. |
1132 | /// |
1133 | /// # Example |
1134 | /// |
1135 | /// ``` |
1136 | /// use jiff::{civil::date, fmt::rfc2822::DateTimePrinter}; |
1137 | /// |
1138 | /// const PRINTER: DateTimePrinter = DateTimePrinter::new(); |
1139 | /// |
1140 | /// let zdt = date(2024, 6, 15).at(7, 0, 0, 0).in_tz("America/New_York" )?; |
1141 | /// assert_eq!( |
1142 | /// PRINTER.zoned_to_string(&zdt)?, |
1143 | /// "Sat, 15 Jun 2024 07:00:00 -0400" , |
1144 | /// ); |
1145 | /// |
1146 | /// # Ok::<(), Box<dyn std::error::Error>>(()) |
1147 | /// ``` |
1148 | #[cfg (feature = "alloc" )] |
1149 | pub fn zoned_to_string( |
1150 | &self, |
1151 | zdt: &Zoned, |
1152 | ) -> Result<alloc::string::String, Error> { |
1153 | let mut buf = alloc::string::String::with_capacity(4); |
1154 | self.print_zoned(zdt, &mut buf)?; |
1155 | Ok(buf) |
1156 | } |
1157 | |
1158 | /// Format a `Timestamp` datetime into a string. |
1159 | /// |
1160 | /// This always emits `-0000` as the offset in the RFC 2822 format. If you |
1161 | /// desire a `+0000` offset, use [`DateTimePrinter::print_zoned`] with a |
1162 | /// zoned datetime with [`TimeZone::UTC`]. |
1163 | /// |
1164 | /// Moreover, since RFC 2822 does not support fractional seconds, this |
1165 | /// routine prints the timestamp as if truncating any fractional seconds. |
1166 | /// |
1167 | /// This is a convenience routine for [`DateTimePrinter::print_timestamp`] |
1168 | /// with a `String`. |
1169 | /// |
1170 | /// # Errors |
1171 | /// |
1172 | /// This returns an error if the year corresponding to this |
1173 | /// timestamp cannot be represented in the RFC 2822 format. For example, a |
1174 | /// negative year. |
1175 | /// |
1176 | /// # Example |
1177 | /// |
1178 | /// ``` |
1179 | /// use jiff::{fmt::rfc2822::DateTimePrinter, Timestamp}; |
1180 | /// |
1181 | /// let timestamp = Timestamp::from_second(1) |
1182 | /// .expect("one second after Unix epoch is always valid" ); |
1183 | /// assert_eq!( |
1184 | /// DateTimePrinter::new().timestamp_to_string(×tamp)?, |
1185 | /// "Thu, 1 Jan 1970 00:00:01 -0000" , |
1186 | /// ); |
1187 | /// |
1188 | /// # Ok::<(), Box<dyn std::error::Error>>(()) |
1189 | /// ``` |
1190 | #[cfg (feature = "alloc" )] |
1191 | pub fn timestamp_to_string( |
1192 | &self, |
1193 | timestamp: &Timestamp, |
1194 | ) -> Result<alloc::string::String, Error> { |
1195 | let mut buf = alloc::string::String::with_capacity(4); |
1196 | self.print_timestamp(timestamp, &mut buf)?; |
1197 | Ok(buf) |
1198 | } |
1199 | |
1200 | /// Format a `Timestamp` datetime into a string in a way that is explicitly |
1201 | /// compatible with [RFC 9110]. This is typically useful in contexts where |
1202 | /// strict compatibility with HTTP is desired. |
1203 | /// |
1204 | /// This always emits `GMT` as the offset and always uses two digits for |
1205 | /// the day. This results in a fixed length format that always uses 29 |
1206 | /// characters. |
1207 | /// |
1208 | /// Since neither RFC 2822 nor RFC 9110 supports fractional seconds, this |
1209 | /// routine prints the timestamp as if truncating any fractional seconds. |
1210 | /// |
1211 | /// This is a convenience routine for |
1212 | /// [`DateTimePrinter::print_timestamp_rfc9110`] with a `String`. |
1213 | /// |
1214 | /// # Errors |
1215 | /// |
1216 | /// This returns an error if the year corresponding to this timestamp |
1217 | /// cannot be represented in the RFC 2822 or RFC 9110 format. For example, |
1218 | /// a negative year. |
1219 | /// |
1220 | /// # Example |
1221 | /// |
1222 | /// ``` |
1223 | /// use jiff::{fmt::rfc2822::DateTimePrinter, Timestamp}; |
1224 | /// |
1225 | /// let timestamp = Timestamp::from_second(1) |
1226 | /// .expect("one second after Unix epoch is always valid" ); |
1227 | /// assert_eq!( |
1228 | /// DateTimePrinter::new().timestamp_to_rfc9110_string(×tamp)?, |
1229 | /// "Thu, 01 Jan 1970 00:00:01 GMT" , |
1230 | /// ); |
1231 | /// |
1232 | /// # Ok::<(), Box<dyn std::error::Error>>(()) |
1233 | /// ``` |
1234 | /// |
1235 | /// [RFC 9110]: https://datatracker.ietf.org/doc/html/rfc9110#section-5.6.7-15 |
1236 | #[cfg (feature = "alloc" )] |
1237 | pub fn timestamp_to_rfc9110_string( |
1238 | &self, |
1239 | timestamp: &Timestamp, |
1240 | ) -> Result<alloc::string::String, Error> { |
1241 | let mut buf = alloc::string::String::with_capacity(29); |
1242 | self.print_timestamp_rfc9110(timestamp, &mut buf)?; |
1243 | Ok(buf) |
1244 | } |
1245 | |
1246 | /// Print a `Zoned` datetime to the given writer. |
1247 | /// |
1248 | /// This never emits `-0000` as the offset in the RFC 2822 format. If you |
1249 | /// desire a `-0000` offset, use [`DateTimePrinter::print_timestamp`] via |
1250 | /// [`Zoned::timestamp`]. |
1251 | /// |
1252 | /// Moreover, since RFC 2822 does not support fractional seconds, this |
1253 | /// routine prints the zoned datetime as if truncating any fractional |
1254 | /// seconds. |
1255 | /// |
1256 | /// # Warning |
1257 | /// |
1258 | /// The RFC 2822 format only supports writing a precise instant in time |
1259 | /// expressed via a time zone offset. It does *not* support serializing |
1260 | /// the time zone itself. This means that if you format a zoned datetime |
1261 | /// in a time zone like `America/New_York` and then deserialize it, the |
1262 | /// zoned datetime you get back will be a "fixed offset" zoned datetime. |
1263 | /// This in turn means it will not perform daylight saving time safe |
1264 | /// arithmetic. |
1265 | /// |
1266 | /// Basically, you should use the RFC 2822 format if it's required (for |
1267 | /// example, when dealing with email). But you should not choose it as a |
1268 | /// general interchange format for new applications. |
1269 | /// |
1270 | /// # Errors |
1271 | /// |
1272 | /// This returns an error when writing to the given [`Write`] |
1273 | /// implementation would fail. Some such implementations, like for `String` |
1274 | /// and `Vec<u8>`, never fail (unless memory allocation fails). |
1275 | /// |
1276 | /// This can also return an error if the year corresponding to this |
1277 | /// timestamp cannot be represented in the RFC 2822 format. For example, a |
1278 | /// negative year. |
1279 | /// |
1280 | /// # Example |
1281 | /// |
1282 | /// ``` |
1283 | /// use jiff::{civil::date, fmt::rfc2822::DateTimePrinter}; |
1284 | /// |
1285 | /// const PRINTER: DateTimePrinter = DateTimePrinter::new(); |
1286 | /// |
1287 | /// let zdt = date(2024, 6, 15).at(7, 0, 0, 0).in_tz("America/New_York" )?; |
1288 | /// |
1289 | /// let mut buf = String::new(); |
1290 | /// PRINTER.print_zoned(&zdt, &mut buf)?; |
1291 | /// assert_eq!(buf, "Sat, 15 Jun 2024 07:00:00 -0400" ); |
1292 | /// |
1293 | /// # Ok::<(), Box<dyn std::error::Error>>(()) |
1294 | /// ``` |
1295 | pub fn print_zoned<W: Write>( |
1296 | &self, |
1297 | zdt: &Zoned, |
1298 | wtr: W, |
1299 | ) -> Result<(), Error> { |
1300 | self.print_civil_with_offset(zdt.datetime(), Some(zdt.offset()), wtr) |
1301 | } |
1302 | |
1303 | /// Print a `Timestamp` datetime to the given writer. |
1304 | /// |
1305 | /// This always emits `-0000` as the offset in the RFC 2822 format. If you |
1306 | /// desire a `+0000` offset, use [`DateTimePrinter::print_zoned`] with a |
1307 | /// zoned datetime with [`TimeZone::UTC`]. |
1308 | /// |
1309 | /// Moreover, since RFC 2822 does not support fractional seconds, this |
1310 | /// routine prints the timestamp as if truncating any fractional seconds. |
1311 | /// |
1312 | /// # Errors |
1313 | /// |
1314 | /// This returns an error when writing to the given [`Write`] |
1315 | /// implementation would fail. Some such implementations, like for `String` |
1316 | /// and `Vec<u8>`, never fail (unless memory allocation fails). |
1317 | /// |
1318 | /// This can also return an error if the year corresponding to this |
1319 | /// timestamp cannot be represented in the RFC 2822 format. For example, a |
1320 | /// negative year. |
1321 | /// |
1322 | /// # Example |
1323 | /// |
1324 | /// ``` |
1325 | /// use jiff::{fmt::rfc2822::DateTimePrinter, Timestamp}; |
1326 | /// |
1327 | /// let timestamp = Timestamp::from_second(1) |
1328 | /// .expect("one second after Unix epoch is always valid" ); |
1329 | /// |
1330 | /// let mut buf = String::new(); |
1331 | /// DateTimePrinter::new().print_timestamp(×tamp, &mut buf)?; |
1332 | /// assert_eq!(buf, "Thu, 1 Jan 1970 00:00:01 -0000" ); |
1333 | /// |
1334 | /// # Ok::<(), Box<dyn std::error::Error>>(()) |
1335 | /// ``` |
1336 | pub fn print_timestamp<W: Write>( |
1337 | &self, |
1338 | timestamp: &Timestamp, |
1339 | wtr: W, |
1340 | ) -> Result<(), Error> { |
1341 | let dt = TimeZone::UTC.to_datetime(*timestamp); |
1342 | self.print_civil_with_offset(dt, None, wtr) |
1343 | } |
1344 | |
1345 | /// Print a `Timestamp` datetime to the given writer in a way that is |
1346 | /// explicitly compatible with [RFC 9110]. This is typically useful in |
1347 | /// contexts where strict compatibility with HTTP is desired. |
1348 | /// |
1349 | /// This always emits `GMT` as the offset and always uses two digits for |
1350 | /// the day. This results in a fixed length format that always uses 29 |
1351 | /// characters. |
1352 | /// |
1353 | /// Since neither RFC 2822 nor RFC 9110 supports fractional seconds, this |
1354 | /// routine prints the timestamp as if truncating any fractional seconds. |
1355 | /// |
1356 | /// # Errors |
1357 | /// |
1358 | /// This returns an error when writing to the given [`Write`] |
1359 | /// implementation would fail. Some such implementations, like for `String` |
1360 | /// and `Vec<u8>`, never fail (unless memory allocation fails). |
1361 | /// |
1362 | /// This can also return an error if the year corresponding to this |
1363 | /// timestamp cannot be represented in the RFC 2822 or RFC 9110 format. For |
1364 | /// example, a negative year. |
1365 | /// |
1366 | /// # Example |
1367 | /// |
1368 | /// ``` |
1369 | /// use jiff::{fmt::rfc2822::DateTimePrinter, Timestamp}; |
1370 | /// |
1371 | /// let timestamp = Timestamp::from_second(1) |
1372 | /// .expect("one second after Unix epoch is always valid" ); |
1373 | /// |
1374 | /// let mut buf = String::new(); |
1375 | /// DateTimePrinter::new().print_timestamp_rfc9110(×tamp, &mut buf)?; |
1376 | /// assert_eq!(buf, "Thu, 01 Jan 1970 00:00:01 GMT" ); |
1377 | /// |
1378 | /// # Ok::<(), Box<dyn std::error::Error>>(()) |
1379 | /// ``` |
1380 | /// |
1381 | /// [RFC 9110]: https://datatracker.ietf.org/doc/html/rfc9110#section-5.6.7-15 |
1382 | pub fn print_timestamp_rfc9110<W: Write>( |
1383 | &self, |
1384 | timestamp: &Timestamp, |
1385 | wtr: W, |
1386 | ) -> Result<(), Error> { |
1387 | self.print_civil_always_utc(timestamp, wtr) |
1388 | } |
1389 | |
1390 | fn print_civil_with_offset<W: Write>( |
1391 | &self, |
1392 | dt: DateTime, |
1393 | offset: Option<Offset>, |
1394 | mut wtr: W, |
1395 | ) -> Result<(), Error> { |
1396 | static FMT_DAY: DecimalFormatter = DecimalFormatter::new(); |
1397 | static FMT_YEAR: DecimalFormatter = DecimalFormatter::new().padding(4); |
1398 | static FMT_TIME_UNIT: DecimalFormatter = |
1399 | DecimalFormatter::new().padding(2); |
1400 | |
1401 | if dt.year() < 0 { |
1402 | // RFC 2822 actually says the year must be at least 1900, but |
1403 | // other implementations (like Chrono) allow any positive 4-digit |
1404 | // year. |
1405 | return Err(err!( |
1406 | "datetime {dt} has negative year, \ |
1407 | which cannot be formatted with RFC 2822" , |
1408 | )); |
1409 | } |
1410 | |
1411 | wtr.write_str(weekday_abbrev(dt.weekday()))?; |
1412 | wtr.write_str(", " )?; |
1413 | wtr.write_int(&FMT_DAY, dt.day())?; |
1414 | wtr.write_str(" " )?; |
1415 | wtr.write_str(month_name(dt.month()))?; |
1416 | wtr.write_str(" " )?; |
1417 | wtr.write_int(&FMT_YEAR, dt.year())?; |
1418 | wtr.write_str(" " )?; |
1419 | wtr.write_int(&FMT_TIME_UNIT, dt.hour())?; |
1420 | wtr.write_str(":" )?; |
1421 | wtr.write_int(&FMT_TIME_UNIT, dt.minute())?; |
1422 | wtr.write_str(":" )?; |
1423 | wtr.write_int(&FMT_TIME_UNIT, dt.second())?; |
1424 | wtr.write_str(" " )?; |
1425 | |
1426 | let Some(offset) = offset else { |
1427 | wtr.write_str("-0000" )?; |
1428 | return Ok(()); |
1429 | }; |
1430 | wtr.write_str(if offset.is_negative() { "-" } else { "+" })?; |
1431 | let mut hours = offset.part_hours_ranged().abs().get(); |
1432 | let mut minutes = offset.part_minutes_ranged().abs().get(); |
1433 | // RFC 2822, like RFC 3339, requires that time zone offsets are an |
1434 | // integral number of minutes. While rounding based on seconds doesn't |
1435 | // seem clearly indicated, we choose to do that here. An alternative |
1436 | // would be to return an error. It isn't clear how important this is in |
1437 | // practice though. |
1438 | if offset.part_seconds_ranged().abs() >= C(30) { |
1439 | if minutes == 59 { |
1440 | hours = hours.saturating_add(1); |
1441 | minutes = 0; |
1442 | } else { |
1443 | minutes = minutes.saturating_add(1); |
1444 | } |
1445 | } |
1446 | wtr.write_int(&FMT_TIME_UNIT, hours)?; |
1447 | wtr.write_int(&FMT_TIME_UNIT, minutes)?; |
1448 | Ok(()) |
1449 | } |
1450 | |
1451 | fn print_civil_always_utc<W: Write>( |
1452 | &self, |
1453 | timestamp: &Timestamp, |
1454 | mut wtr: W, |
1455 | ) -> Result<(), Error> { |
1456 | static FMT_DAY: DecimalFormatter = DecimalFormatter::new().padding(2); |
1457 | static FMT_YEAR: DecimalFormatter = DecimalFormatter::new().padding(4); |
1458 | static FMT_TIME_UNIT: DecimalFormatter = |
1459 | DecimalFormatter::new().padding(2); |
1460 | |
1461 | let dt = TimeZone::UTC.to_datetime(*timestamp); |
1462 | if dt.year() < 0 { |
1463 | // RFC 2822 actually says the year must be at least 1900, but |
1464 | // other implementations (like Chrono) allow any positive 4-digit |
1465 | // year. |
1466 | return Err(err!( |
1467 | "datetime {dt} has negative year, \ |
1468 | which cannot be formatted with RFC 2822" , |
1469 | )); |
1470 | } |
1471 | |
1472 | wtr.write_str(weekday_abbrev(dt.weekday()))?; |
1473 | wtr.write_str(", " )?; |
1474 | wtr.write_int(&FMT_DAY, dt.day())?; |
1475 | wtr.write_str(" " )?; |
1476 | wtr.write_str(month_name(dt.month()))?; |
1477 | wtr.write_str(" " )?; |
1478 | wtr.write_int(&FMT_YEAR, dt.year())?; |
1479 | wtr.write_str(" " )?; |
1480 | wtr.write_int(&FMT_TIME_UNIT, dt.hour())?; |
1481 | wtr.write_str(":" )?; |
1482 | wtr.write_int(&FMT_TIME_UNIT, dt.minute())?; |
1483 | wtr.write_str(":" )?; |
1484 | wtr.write_int(&FMT_TIME_UNIT, dt.second())?; |
1485 | wtr.write_str(" " )?; |
1486 | wtr.write_str("GMT" )?; |
1487 | Ok(()) |
1488 | } |
1489 | } |
1490 | |
1491 | fn weekday_abbrev(wd: Weekday) -> &'static str { |
1492 | match wd { |
1493 | Weekday::Sunday => "Sun" , |
1494 | Weekday::Monday => "Mon" , |
1495 | Weekday::Tuesday => "Tue" , |
1496 | Weekday::Wednesday => "Wed" , |
1497 | Weekday::Thursday => "Thu" , |
1498 | Weekday::Friday => "Fri" , |
1499 | Weekday::Saturday => "Sat" , |
1500 | } |
1501 | } |
1502 | |
1503 | fn month_name(month: i8) -> &'static str { |
1504 | match month { |
1505 | 1 => "Jan" , |
1506 | 2 => "Feb" , |
1507 | 3 => "Mar" , |
1508 | 4 => "Apr" , |
1509 | 5 => "May" , |
1510 | 6 => "Jun" , |
1511 | 7 => "Jul" , |
1512 | 8 => "Aug" , |
1513 | 9 => "Sep" , |
1514 | 10 => "Oct" , |
1515 | 11 => "Nov" , |
1516 | 12 => "Dec" , |
1517 | _ => unreachable!("invalid month value {month}" ), |
1518 | } |
1519 | } |
1520 | |
1521 | /// Returns true if the given byte is "whitespace" as defined by RFC 2822. |
1522 | /// |
1523 | /// From S2.2.2: |
1524 | /// |
1525 | /// > Many of these tokens are allowed (according to their syntax) to be |
1526 | /// > introduced or end with comments (as described in section 3.2.3) as well |
1527 | /// > as the space (SP, ASCII value 32) and horizontal tab (HTAB, ASCII value |
1528 | /// > 9) characters (together known as the white space characters, WSP), and |
1529 | /// > those WSP characters are subject to header "folding" and "unfolding" as |
1530 | /// > described in section 2.2.3. |
1531 | /// |
1532 | /// In other words, ASCII space or tab. |
1533 | /// |
1534 | /// With all that said, it seems odd to limit this to just spaces or tabs, so |
1535 | /// we relax this and let it absorb any kind of ASCII whitespace. This also |
1536 | /// handles, I believe, most cases of "folding" whitespace. (By treating `\r` |
1537 | /// and `\n` as whitespace.) |
1538 | fn is_whitespace(byte: u8) -> bool { |
1539 | byte.is_ascii_whitespace() |
1540 | } |
1541 | |
1542 | #[cfg (feature = "alloc" )] |
1543 | #[cfg (test)] |
1544 | mod tests { |
1545 | use alloc::string::{String, ToString}; |
1546 | |
1547 | use crate::civil::date; |
1548 | |
1549 | use super::*; |
1550 | |
1551 | #[test ] |
1552 | fn ok_parse_basic() { |
1553 | let p = |input| DateTimeParser::new().parse_zoned(input).unwrap(); |
1554 | |
1555 | insta::assert_debug_snapshot!( |
1556 | p("Wed, 10 Jan 2024 05:34:45 -0500" ), |
1557 | @"2024-01-10T05:34:45-05:00[-05:00]" , |
1558 | ); |
1559 | insta::assert_debug_snapshot!( |
1560 | p("Tue, 9 Jan 2024 05:34:45 -0500" ), |
1561 | @"2024-01-09T05:34:45-05:00[-05:00]" , |
1562 | ); |
1563 | insta::assert_debug_snapshot!( |
1564 | p("Tue, 09 Jan 2024 05:34:45 -0500" ), |
1565 | @"2024-01-09T05:34:45-05:00[-05:00]" , |
1566 | ); |
1567 | insta::assert_debug_snapshot!( |
1568 | p("10 Jan 2024 05:34:45 -0500" ), |
1569 | @"2024-01-10T05:34:45-05:00[-05:00]" , |
1570 | ); |
1571 | insta::assert_debug_snapshot!( |
1572 | p("10 Jan 2024 05:34 -0500" ), |
1573 | @"2024-01-10T05:34:00-05:00[-05:00]" , |
1574 | ); |
1575 | insta::assert_debug_snapshot!( |
1576 | p("10 Jan 2024 05:34:45 +0500" ), |
1577 | @"2024-01-10T05:34:45+05:00[+05:00]" , |
1578 | ); |
1579 | insta::assert_debug_snapshot!( |
1580 | p("Thu, 29 Feb 2024 05:34 -0500" ), |
1581 | @"2024-02-29T05:34:00-05:00[-05:00]" , |
1582 | ); |
1583 | |
1584 | // leap second constraining |
1585 | insta::assert_debug_snapshot!( |
1586 | p("10 Jan 2024 05:34:60 -0500" ), |
1587 | @"2024-01-10T05:34:59-05:00[-05:00]" , |
1588 | ); |
1589 | } |
1590 | |
1591 | #[test ] |
1592 | fn ok_parse_obsolete_zone() { |
1593 | let p = |input| DateTimeParser::new().parse_zoned(input).unwrap(); |
1594 | |
1595 | insta::assert_debug_snapshot!( |
1596 | p("Wed, 10 Jan 2024 05:34:45 EST" ), |
1597 | @"2024-01-10T05:34:45-05:00[-05:00]" , |
1598 | ); |
1599 | insta::assert_debug_snapshot!( |
1600 | p("Wed, 10 Jan 2024 05:34:45 EDT" ), |
1601 | @"2024-01-10T05:34:45-04:00[-04:00]" , |
1602 | ); |
1603 | insta::assert_debug_snapshot!( |
1604 | p("Wed, 10 Jan 2024 05:34:45 CST" ), |
1605 | @"2024-01-10T05:34:45-06:00[-06:00]" , |
1606 | ); |
1607 | insta::assert_debug_snapshot!( |
1608 | p("Wed, 10 Jan 2024 05:34:45 CDT" ), |
1609 | @"2024-01-10T05:34:45-05:00[-05:00]" , |
1610 | ); |
1611 | insta::assert_debug_snapshot!( |
1612 | p("Wed, 10 Jan 2024 05:34:45 mst" ), |
1613 | @"2024-01-10T05:34:45-07:00[-07:00]" , |
1614 | ); |
1615 | insta::assert_debug_snapshot!( |
1616 | p("Wed, 10 Jan 2024 05:34:45 mdt" ), |
1617 | @"2024-01-10T05:34:45-06:00[-06:00]" , |
1618 | ); |
1619 | insta::assert_debug_snapshot!( |
1620 | p("Wed, 10 Jan 2024 05:34:45 pst" ), |
1621 | @"2024-01-10T05:34:45-08:00[-08:00]" , |
1622 | ); |
1623 | insta::assert_debug_snapshot!( |
1624 | p("Wed, 10 Jan 2024 05:34:45 pdt" ), |
1625 | @"2024-01-10T05:34:45-07:00[-07:00]" , |
1626 | ); |
1627 | |
1628 | // Various things that mean UTC. |
1629 | insta::assert_debug_snapshot!( |
1630 | p("Wed, 10 Jan 2024 05:34:45 UT" ), |
1631 | @"2024-01-10T05:34:45+00:00[UTC]" , |
1632 | ); |
1633 | insta::assert_debug_snapshot!( |
1634 | p("Wed, 10 Jan 2024 05:34:45 Z" ), |
1635 | @"2024-01-10T05:34:45+00:00[UTC]" , |
1636 | ); |
1637 | insta::assert_debug_snapshot!( |
1638 | p("Wed, 10 Jan 2024 05:34:45 gmt" ), |
1639 | @"2024-01-10T05:34:45+00:00[UTC]" , |
1640 | ); |
1641 | |
1642 | // Even things that are unrecognized just get treated as having |
1643 | // an offset of 0. |
1644 | insta::assert_debug_snapshot!( |
1645 | p("Wed, 10 Jan 2024 05:34:45 XXX" ), |
1646 | @"2024-01-10T05:34:45+00:00[UTC]" , |
1647 | ); |
1648 | insta::assert_debug_snapshot!( |
1649 | p("Wed, 10 Jan 2024 05:34:45 ABCDE" ), |
1650 | @"2024-01-10T05:34:45+00:00[UTC]" , |
1651 | ); |
1652 | insta::assert_debug_snapshot!( |
1653 | p("Wed, 10 Jan 2024 05:34:45 FUCK" ), |
1654 | @"2024-01-10T05:34:45+00:00[UTC]" , |
1655 | ); |
1656 | } |
1657 | |
1658 | // whyyyyyyyyyyyyy |
1659 | #[test ] |
1660 | fn ok_parse_comment() { |
1661 | let p = |input| DateTimeParser::new().parse_zoned(input).unwrap(); |
1662 | |
1663 | insta::assert_debug_snapshot!( |
1664 | p("Wed, 10 Jan 2024 05:34:45 -0500 (wat)" ), |
1665 | @"2024-01-10T05:34:45-05:00[-05:00]" , |
1666 | ); |
1667 | insta::assert_debug_snapshot!( |
1668 | p("Wed, 10 Jan 2024 05:34:45 -0500 (w(a)t)" ), |
1669 | @"2024-01-10T05:34:45-05:00[-05:00]" , |
1670 | ); |
1671 | insta::assert_debug_snapshot!( |
1672 | p(r"Wed, 10 Jan 2024 05:34:45 -0500 (w\(a\)t)" ), |
1673 | @"2024-01-10T05:34:45-05:00[-05:00]" , |
1674 | ); |
1675 | } |
1676 | |
1677 | #[test ] |
1678 | fn ok_parse_whitespace() { |
1679 | let p = |input| DateTimeParser::new().parse_zoned(input).unwrap(); |
1680 | |
1681 | insta::assert_debug_snapshot!( |
1682 | p("Wed, 10 \t Jan \n\r\n\n 2024 05:34:45 -0500" ), |
1683 | @"2024-01-10T05:34:45-05:00[-05:00]" , |
1684 | ); |
1685 | insta::assert_debug_snapshot!( |
1686 | p("Wed, 10 Jan 2024 05:34:45 -0500 " ), |
1687 | @"2024-01-10T05:34:45-05:00[-05:00]" , |
1688 | ); |
1689 | } |
1690 | |
1691 | #[test ] |
1692 | fn err_parse_invalid() { |
1693 | let p = |input| { |
1694 | DateTimeParser::new().parse_zoned(input).unwrap_err().to_string() |
1695 | }; |
1696 | |
1697 | insta::assert_snapshot!( |
1698 | p("Thu, 10 Jan 2024 05:34:45 -0500" ), |
1699 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: found parsed weekday of Thu, but parsed datetime of 2024-01-10T05:34:45 has weekday Wed" , |
1700 | ); |
1701 | insta::assert_snapshot!( |
1702 | p("Wed, 29 Feb 2023 05:34:45 -0500" ), |
1703 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: invalid date: parameter 'day' with value 29 is not in the required range of 1..=28" , |
1704 | ); |
1705 | insta::assert_snapshot!( |
1706 | p("Mon, 31 Jun 2024 05:34:45 -0500" ), |
1707 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: invalid date: parameter 'day' with value 31 is not in the required range of 1..=30" , |
1708 | ); |
1709 | insta::assert_snapshot!( |
1710 | p("Tue, 32 Jun 2024 05:34:45 -0500" ), |
1711 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: day is not valid: parameter 'day' with value 32 is not in the required range of 1..=31" , |
1712 | ); |
1713 | insta::assert_snapshot!( |
1714 | p("Sun, 30 Jun 2024 24:00:00 -0500" ), |
1715 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: hour is not valid: parameter 'hour' with value 24 is not in the required range of 0..=23" , |
1716 | ); |
1717 | } |
1718 | |
1719 | #[test ] |
1720 | fn err_parse_incomplete() { |
1721 | let p = |input| { |
1722 | DateTimeParser::new().parse_zoned(input).unwrap_err().to_string() |
1723 | }; |
1724 | |
1725 | insta::assert_snapshot!( |
1726 | p("" ), |
1727 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected RFC 2822 datetime, but got empty string" , |
1728 | ); |
1729 | insta::assert_snapshot!( |
1730 | p(" " ), |
1731 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected RFC 2822 datetime, but got empty string after trimming whitespace" , |
1732 | ); |
1733 | insta::assert_snapshot!( |
1734 | p("Wat" ), |
1735 | @r###"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected day at beginning of RFC 2822 datetime since first non-whitespace byte, "W", is not a digit, but given string is too short (length is 3)"### , |
1736 | ); |
1737 | insta::assert_snapshot!( |
1738 | p("Wed" ), |
1739 | @r###"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected day at beginning of RFC 2822 datetime since first non-whitespace byte, "W", is not a digit, but given string is too short (length is 3)"### , |
1740 | ); |
1741 | insta::assert_snapshot!( |
1742 | p("Wat, " ), |
1743 | @r###"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected day at beginning of RFC 2822 datetime since first non-whitespace byte, "W", is not a digit, but did not recognize "Wat" as a valid weekday abbreviation"### , |
1744 | ); |
1745 | insta::assert_snapshot!( |
1746 | p("Wed, " ), |
1747 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected day, but found end of input" , |
1748 | ); |
1749 | insta::assert_snapshot!( |
1750 | p("Wed, 1" ), |
1751 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected whitespace after parsing day 1: expected at least one whitespace character (space or tab), but found none" , |
1752 | ); |
1753 | insta::assert_snapshot!( |
1754 | p("Wed, 10" ), |
1755 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected whitespace after parsing day 10: expected at least one whitespace character (space or tab), but found none" , |
1756 | ); |
1757 | insta::assert_snapshot!( |
1758 | p("Wed, 10 J" ), |
1759 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected abbreviated month name, but remaining input is too short (remaining bytes is 1)" , |
1760 | ); |
1761 | insta::assert_snapshot!( |
1762 | p("Wed, 10 Wat" ), |
1763 | @r###"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected abbreviated month name, but did not recognize "Wat" as a valid month"### , |
1764 | ); |
1765 | insta::assert_snapshot!( |
1766 | p("Wed, 10 Jan" ), |
1767 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected whitespace after parsing month name: expected at least one whitespace character (space or tab), but found none" , |
1768 | ); |
1769 | insta::assert_snapshot!( |
1770 | p("Wed, 10 Jan 2" ), |
1771 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected at least two ASCII digits for parsing a year, but only found 1" , |
1772 | ); |
1773 | insta::assert_snapshot!( |
1774 | p("Wed, 10 Jan 2024" ), |
1775 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected whitespace after parsing year: expected at least one whitespace character (space or tab), but found none" , |
1776 | ); |
1777 | insta::assert_snapshot!( |
1778 | p("Wed, 10 Jan 2024 05" ), |
1779 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected time separator of ':', but found end of input" , |
1780 | ); |
1781 | insta::assert_snapshot!( |
1782 | p("Wed, 10 Jan 2024 053" ), |
1783 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected time separator of ':', but found 3" , |
1784 | ); |
1785 | insta::assert_snapshot!( |
1786 | p("Wed, 10 Jan 2024 05:34" ), |
1787 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected whitespace after parsing time: expected at least one whitespace character (space or tab), but found none" , |
1788 | ); |
1789 | insta::assert_snapshot!( |
1790 | p("Wed, 10 Jan 2024 05:34:" ), |
1791 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected two digit second, but found end of input" , |
1792 | ); |
1793 | insta::assert_snapshot!( |
1794 | p("Wed, 10 Jan 2024 05:34:45" ), |
1795 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected whitespace after parsing time: expected at least one whitespace character (space or tab), but found none" , |
1796 | ); |
1797 | insta::assert_snapshot!( |
1798 | p("Wed, 10 Jan 2024 05:34:45 J" ), |
1799 | @r###"failed to parse RFC 2822 datetime into Jiff zoned datetime: expected obsolete RFC 2822 time zone abbreviation, but found "J""### , |
1800 | ); |
1801 | } |
1802 | |
1803 | #[test ] |
1804 | fn err_parse_comment() { |
1805 | let p = |input| { |
1806 | DateTimeParser::new().parse_zoned(input).unwrap_err().to_string() |
1807 | }; |
1808 | |
1809 | insta::assert_snapshot!( |
1810 | p(r"Wed, 10 Jan 2024 05:34:45 -0500 (wa)t)" ), |
1811 | @r###"parsed value '2024-01-10T05:34:45-05:00[-05:00]', but unparsed input "t)" remains (expected no unparsed input)"### , |
1812 | ); |
1813 | insta::assert_snapshot!( |
1814 | p(r"Wed, 10 Jan 2024 05:34:45 -0500 (wa(t)" ), |
1815 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: found opening parenthesis in comment with no matching closing parenthesis" , |
1816 | ); |
1817 | insta::assert_snapshot!( |
1818 | p(r"Wed, 10 Jan 2024 05:34:45 -0500 (w" ), |
1819 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: found opening parenthesis in comment with no matching closing parenthesis" , |
1820 | ); |
1821 | insta::assert_snapshot!( |
1822 | p(r"Wed, 10 Jan 2024 05:34:45 -0500 (" ), |
1823 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: found opening parenthesis in comment with no matching closing parenthesis" , |
1824 | ); |
1825 | insta::assert_snapshot!( |
1826 | p(r"Wed, 10 Jan 2024 05:34:45 -0500 ( " ), |
1827 | @"failed to parse RFC 2822 datetime into Jiff zoned datetime: found opening parenthesis in comment with no matching closing parenthesis" , |
1828 | ); |
1829 | } |
1830 | |
1831 | #[test ] |
1832 | fn ok_print_zoned() { |
1833 | if crate::tz::db().is_definitively_empty() { |
1834 | return; |
1835 | } |
1836 | |
1837 | let p = |zdt: &Zoned| -> String { |
1838 | let mut buf = String::new(); |
1839 | DateTimePrinter::new().print_zoned(&zdt, &mut buf).unwrap(); |
1840 | buf |
1841 | }; |
1842 | |
1843 | let zdt = date(2024, 1, 10) |
1844 | .at(5, 34, 45, 0) |
1845 | .in_tz("America/New_York" ) |
1846 | .unwrap(); |
1847 | insta::assert_snapshot!(p(&zdt), @"Wed, 10 Jan 2024 05:34:45 -0500" ); |
1848 | |
1849 | let zdt = date(2024, 2, 5) |
1850 | .at(5, 34, 45, 0) |
1851 | .in_tz("America/New_York" ) |
1852 | .unwrap(); |
1853 | insta::assert_snapshot!(p(&zdt), @"Mon, 5 Feb 2024 05:34:45 -0500" ); |
1854 | |
1855 | let zdt = date(2024, 7, 31) |
1856 | .at(5, 34, 45, 0) |
1857 | .in_tz("America/New_York" ) |
1858 | .unwrap(); |
1859 | insta::assert_snapshot!(p(&zdt), @"Wed, 31 Jul 2024 05:34:45 -0400" ); |
1860 | |
1861 | let zdt = date(2024, 3, 5).at(5, 34, 45, 0).in_tz("UTC" ).unwrap(); |
1862 | // Notice that this prints a +0000 offset. |
1863 | // But when printing a Timestamp, a -0000 offset is used. |
1864 | // This is because in the case of Timestamp, the "true" |
1865 | // offset is not known. |
1866 | insta::assert_snapshot!(p(&zdt), @"Tue, 5 Mar 2024 05:34:45 +0000" ); |
1867 | } |
1868 | |
1869 | #[test ] |
1870 | fn ok_print_timestamp() { |
1871 | if crate::tz::db().is_definitively_empty() { |
1872 | return; |
1873 | } |
1874 | |
1875 | let p = |ts: Timestamp| -> String { |
1876 | let mut buf = String::new(); |
1877 | DateTimePrinter::new().print_timestamp(&ts, &mut buf).unwrap(); |
1878 | buf |
1879 | }; |
1880 | |
1881 | let ts = date(2024, 1, 10) |
1882 | .at(5, 34, 45, 0) |
1883 | .in_tz("America/New_York" ) |
1884 | .unwrap() |
1885 | .timestamp(); |
1886 | insta::assert_snapshot!(p(ts), @"Wed, 10 Jan 2024 10:34:45 -0000" ); |
1887 | |
1888 | let ts = date(2024, 2, 5) |
1889 | .at(5, 34, 45, 0) |
1890 | .in_tz("America/New_York" ) |
1891 | .unwrap() |
1892 | .timestamp(); |
1893 | insta::assert_snapshot!(p(ts), @"Mon, 5 Feb 2024 10:34:45 -0000" ); |
1894 | |
1895 | let ts = date(2024, 7, 31) |
1896 | .at(5, 34, 45, 0) |
1897 | .in_tz("America/New_York" ) |
1898 | .unwrap() |
1899 | .timestamp(); |
1900 | insta::assert_snapshot!(p(ts), @"Wed, 31 Jul 2024 09:34:45 -0000" ); |
1901 | |
1902 | let ts = date(2024, 3, 5) |
1903 | .at(5, 34, 45, 0) |
1904 | .in_tz("UTC" ) |
1905 | .unwrap() |
1906 | .timestamp(); |
1907 | // Notice that this prints a +0000 offset. |
1908 | // But when printing a Timestamp, a -0000 offset is used. |
1909 | // This is because in the case of Timestamp, the "true" |
1910 | // offset is not known. |
1911 | insta::assert_snapshot!(p(ts), @"Tue, 5 Mar 2024 05:34:45 -0000" ); |
1912 | } |
1913 | |
1914 | #[test ] |
1915 | fn ok_print_rfc9110_timestamp() { |
1916 | if crate::tz::db().is_definitively_empty() { |
1917 | return; |
1918 | } |
1919 | |
1920 | let p = |ts: Timestamp| -> String { |
1921 | let mut buf = String::new(); |
1922 | DateTimePrinter::new() |
1923 | .print_timestamp_rfc9110(&ts, &mut buf) |
1924 | .unwrap(); |
1925 | buf |
1926 | }; |
1927 | |
1928 | let ts = date(2024, 1, 10) |
1929 | .at(5, 34, 45, 0) |
1930 | .in_tz("America/New_York" ) |
1931 | .unwrap() |
1932 | .timestamp(); |
1933 | insta::assert_snapshot!(p(ts), @"Wed, 10 Jan 2024 10:34:45 GMT" ); |
1934 | |
1935 | let ts = date(2024, 2, 5) |
1936 | .at(5, 34, 45, 0) |
1937 | .in_tz("America/New_York" ) |
1938 | .unwrap() |
1939 | .timestamp(); |
1940 | insta::assert_snapshot!(p(ts), @"Mon, 05 Feb 2024 10:34:45 GMT" ); |
1941 | |
1942 | let ts = date(2024, 7, 31) |
1943 | .at(5, 34, 45, 0) |
1944 | .in_tz("America/New_York" ) |
1945 | .unwrap() |
1946 | .timestamp(); |
1947 | insta::assert_snapshot!(p(ts), @"Wed, 31 Jul 2024 09:34:45 GMT" ); |
1948 | |
1949 | let ts = date(2024, 3, 5) |
1950 | .at(5, 34, 45, 0) |
1951 | .in_tz("UTC" ) |
1952 | .unwrap() |
1953 | .timestamp(); |
1954 | // Notice that this prints a +0000 offset. |
1955 | // But when printing a Timestamp, a -0000 offset is used. |
1956 | // This is because in the case of Timestamp, the "true" |
1957 | // offset is not known. |
1958 | insta::assert_snapshot!(p(ts), @"Tue, 05 Mar 2024 05:34:45 GMT" ); |
1959 | } |
1960 | |
1961 | #[test ] |
1962 | fn err_print_zoned() { |
1963 | if crate::tz::db().is_definitively_empty() { |
1964 | return; |
1965 | } |
1966 | |
1967 | let p = |zdt: &Zoned| -> String { |
1968 | let mut buf = String::new(); |
1969 | DateTimePrinter::new() |
1970 | .print_zoned(&zdt, &mut buf) |
1971 | .unwrap_err() |
1972 | .to_string() |
1973 | }; |
1974 | |
1975 | let zdt = date(-1, 1, 10) |
1976 | .at(5, 34, 45, 0) |
1977 | .in_tz("America/New_York" ) |
1978 | .unwrap(); |
1979 | insta::assert_snapshot!(p(&zdt), @"datetime -000001-01-10T05:34:45 has negative year, which cannot be formatted with RFC 2822" ); |
1980 | } |
1981 | |
1982 | #[test ] |
1983 | fn err_print_timestamp() { |
1984 | if crate::tz::db().is_definitively_empty() { |
1985 | return; |
1986 | } |
1987 | |
1988 | let p = |ts: Timestamp| -> String { |
1989 | let mut buf = String::new(); |
1990 | DateTimePrinter::new() |
1991 | .print_timestamp(&ts, &mut buf) |
1992 | .unwrap_err() |
1993 | .to_string() |
1994 | }; |
1995 | |
1996 | let ts = date(-1, 1, 10) |
1997 | .at(5, 34, 45, 0) |
1998 | .in_tz("America/New_York" ) |
1999 | .unwrap() |
2000 | .timestamp(); |
2001 | insta::assert_snapshot!(p(ts), @"datetime -000001-01-10T10:30:47 has negative year, which cannot be formatted with RFC 2822" ); |
2002 | } |
2003 | } |
2004 | |