1 | use alloc::string::String; |
2 | use alloc::vec::Vec; |
3 | use core::iter; |
4 | |
5 | use crate::error::InvalidFormatDescription; |
6 | use crate::format_description::parse::{ |
7 | attach_location, unused, Error, ErrorInner, Location, Spanned, SpannedValue, Unused, |
8 | }; |
9 | use crate::format_description::{self, modifier, BorrowedFormatItem, Component}; |
10 | |
11 | /// Parse a sequence of items from the [`strftime` format description][strftime docs]. |
12 | /// |
13 | /// The only heap allocation required is for the `Vec` itself. All components are bound to the |
14 | /// lifetime of the input. |
15 | /// |
16 | /// [strftime docs]: https://man7.org/linux/man-pages/man3/strftime.3.html |
17 | #[doc (alias = "parse_strptime_borrowed" )] |
18 | pub fn parse_strftime_borrowed( |
19 | s: &str, |
20 | ) -> Result<Vec<BorrowedFormatItem<'_>>, InvalidFormatDescription> { |
21 | let tokens: impl Iterator- >
= lex(input:s.as_bytes()); |
22 | let items: Vec> = into_items(tokens).collect::<Result<_, _>>()?; |
23 | Ok(items) |
24 | } |
25 | |
26 | /// Parse a sequence of items from the [`strftime` format description][strftime docs]. |
27 | /// |
28 | /// This requires heap allocation for some owned items. |
29 | /// |
30 | /// [strftime docs]: https://man7.org/linux/man-pages/man3/strftime.3.html |
31 | #[doc (alias = "parse_strptime_owned" )] |
32 | pub fn parse_strftime_owned( |
33 | s: &str, |
34 | ) -> Result<format_description::OwnedFormatItem, InvalidFormatDescription> { |
35 | parse_strftime_borrowed(s).map(op:Into::into) |
36 | } |
37 | |
38 | #[derive (Debug, Clone, Copy, PartialEq)] |
39 | enum Padding { |
40 | /// The default padding for a numeric component. Indicated by no character. |
41 | Default, |
42 | /// Pad a numeric component with spaces. Indicated by an underscore. |
43 | Spaces, |
44 | /// Do not pad a numeric component. Indicated by a hyphen. |
45 | None, |
46 | /// Pad a numeric component with zeroes. Indicated by a zero. |
47 | Zeroes, |
48 | } |
49 | |
50 | enum Token<'a> { |
51 | Literal(Spanned<&'a [u8]>), |
52 | Component { |
53 | _percent: Unused<Location>, |
54 | padding: Spanned<Padding>, |
55 | component: Spanned<u8>, |
56 | }, |
57 | } |
58 | |
59 | fn lex(mut input: &[u8]) -> iter::Peekable<impl Iterator<Item = Result<Token<'_>, Error>>> { |
60 | let mut iter = attach_location(input.iter()).peekable(); |
61 | |
62 | iter::from_fn(move || { |
63 | Some(Ok(match iter.next()? { |
64 | (b'%' , percent_loc) => match iter.next() { |
65 | Some((padding @ (b'_' | b'-' | b'0' ), padding_loc)) => { |
66 | let padding = match padding { |
67 | b'_' => Padding::Spaces, |
68 | b'-' => Padding::None, |
69 | b'0' => Padding::Zeroes, |
70 | _ => unreachable!(), |
71 | }; |
72 | let (&component, component_loc) = iter.next()?; |
73 | input = &input[3..]; |
74 | Token::Component { |
75 | _percent: unused(percent_loc), |
76 | padding: padding.spanned(padding_loc.to_self()), |
77 | component: component.spanned(component_loc.to_self()), |
78 | } |
79 | } |
80 | Some((&component, component_loc)) => { |
81 | input = &input[2..]; |
82 | let span = component_loc.to_self(); |
83 | Token::Component { |
84 | _percent: unused(percent_loc), |
85 | padding: Padding::Default.spanned(span), |
86 | component: component.spanned(span), |
87 | } |
88 | } |
89 | None => { |
90 | return Some(Err(Error { |
91 | _inner: unused(percent_loc.error("unexpected end of input" )), |
92 | public: InvalidFormatDescription::Expected { |
93 | what: "valid escape sequence" , |
94 | index: percent_loc.byte as _, |
95 | }, |
96 | })); |
97 | } |
98 | }, |
99 | (_, start_location) => { |
100 | let mut bytes = 1; |
101 | let mut end_location = start_location; |
102 | |
103 | while let Some((_, location)) = iter.next_if(|&(&byte, _)| byte != b'%' ) { |
104 | end_location = location; |
105 | bytes += 1; |
106 | } |
107 | |
108 | let value = &input[..bytes]; |
109 | input = &input[bytes..]; |
110 | |
111 | Token::Literal(value.spanned(start_location.to(end_location))) |
112 | } |
113 | })) |
114 | }) |
115 | .peekable() |
116 | } |
117 | |
118 | fn into_items<'iter, 'token: 'iter>( |
119 | mut tokens: iter::Peekable<impl Iterator<Item = Result<Token<'token>, Error>> + 'iter>, |
120 | ) -> impl Iterator<Item = Result<BorrowedFormatItem<'token>, Error>> + 'iter { |
121 | iter::from_fn(move || { |
122 | let next: Token<'token> = match tokens.next()? { |
123 | Ok(token: Token<'token>) => token, |
124 | Err(err: Error) => return Some(Err(err)), |
125 | }; |
126 | |
127 | Some(match next { |
128 | Token::Literal(spanned: Spanned<&[u8]>) => Ok(BorrowedFormatItem::Literal(*spanned)), |
129 | Token::Component { |
130 | _percent: Unused, |
131 | padding: Spanned, |
132 | component: Spanned, |
133 | } => parse_component(padding, component), |
134 | }) |
135 | }) |
136 | } |
137 | |
138 | fn parse_component( |
139 | padding: Spanned<Padding>, |
140 | component: Spanned<u8>, |
141 | ) -> Result<BorrowedFormatItem<'static>, Error> { |
142 | let padding_or_default = |padding: Padding, default| match padding { |
143 | Padding::Default => default, |
144 | Padding::Spaces => modifier::Padding::Space, |
145 | Padding::None => modifier::Padding::None, |
146 | Padding::Zeroes => modifier::Padding::Zero, |
147 | }; |
148 | |
149 | /// Helper macro to create a component. |
150 | macro_rules! component { |
151 | ($name:ident { $($inner:tt)* }) => { |
152 | BorrowedFormatItem::Component(Component::$name(modifier::$name { |
153 | $($inner)* |
154 | })) |
155 | } |
156 | } |
157 | |
158 | Ok(match *component { |
159 | b'%' => BorrowedFormatItem::Literal(b"%" ), |
160 | b'a' => component!(Weekday { |
161 | repr: modifier::WeekdayRepr::Short, |
162 | one_indexed: true, |
163 | case_sensitive: true, |
164 | }), |
165 | b'A' => component!(Weekday { |
166 | repr: modifier::WeekdayRepr::Long, |
167 | one_indexed: true, |
168 | case_sensitive: true, |
169 | }), |
170 | b'b' | b'h' => component!(Month { |
171 | repr: modifier::MonthRepr::Short, |
172 | padding: modifier::Padding::Zero, |
173 | case_sensitive: true, |
174 | }), |
175 | b'B' => component!(Month { |
176 | repr: modifier::MonthRepr::Long, |
177 | padding: modifier::Padding::Zero, |
178 | case_sensitive: true, |
179 | }), |
180 | b'c' => BorrowedFormatItem::Compound(&[ |
181 | component!(Weekday { |
182 | repr: modifier::WeekdayRepr::Short, |
183 | one_indexed: true, |
184 | case_sensitive: true, |
185 | }), |
186 | BorrowedFormatItem::Literal(b" " ), |
187 | component!(Month { |
188 | repr: modifier::MonthRepr::Short, |
189 | padding: modifier::Padding::Zero, |
190 | case_sensitive: true, |
191 | }), |
192 | BorrowedFormatItem::Literal(b" " ), |
193 | component!(Day { |
194 | padding: modifier::Padding::Space |
195 | }), |
196 | BorrowedFormatItem::Literal(b" " ), |
197 | component!(Hour { |
198 | padding: modifier::Padding::Zero, |
199 | is_12_hour_clock: false, |
200 | }), |
201 | BorrowedFormatItem::Literal(b":" ), |
202 | component!(Minute { |
203 | padding: modifier::Padding::Zero, |
204 | }), |
205 | BorrowedFormatItem::Literal(b":" ), |
206 | component!(Second { |
207 | padding: modifier::Padding::Zero, |
208 | }), |
209 | BorrowedFormatItem::Literal(b" " ), |
210 | component!(Year { |
211 | padding: modifier::Padding::Zero, |
212 | repr: modifier::YearRepr::Full, |
213 | iso_week_based: false, |
214 | sign_is_mandatory: false, |
215 | }), |
216 | ]), |
217 | b'C' => component!(Year { |
218 | padding: padding_or_default(*padding, modifier::Padding::Zero), |
219 | repr: modifier::YearRepr::Century, |
220 | iso_week_based: false, |
221 | sign_is_mandatory: false, |
222 | }), |
223 | b'd' => component!(Day { |
224 | padding: padding_or_default(*padding, modifier::Padding::Zero), |
225 | }), |
226 | b'D' => BorrowedFormatItem::Compound(&[ |
227 | component!(Month { |
228 | repr: modifier::MonthRepr::Numerical, |
229 | padding: modifier::Padding::Zero, |
230 | case_sensitive: true, |
231 | }), |
232 | BorrowedFormatItem::Literal(b"/" ), |
233 | component!(Day { |
234 | padding: modifier::Padding::Zero, |
235 | }), |
236 | BorrowedFormatItem::Literal(b"/" ), |
237 | component!(Year { |
238 | padding: modifier::Padding::Zero, |
239 | repr: modifier::YearRepr::LastTwo, |
240 | iso_week_based: false, |
241 | sign_is_mandatory: false, |
242 | }), |
243 | ]), |
244 | b'e' => component!(Day { |
245 | padding: padding_or_default(*padding, modifier::Padding::Space), |
246 | }), |
247 | b'F' => BorrowedFormatItem::Compound(&[ |
248 | component!(Year { |
249 | padding: modifier::Padding::Zero, |
250 | repr: modifier::YearRepr::Full, |
251 | iso_week_based: false, |
252 | sign_is_mandatory: false, |
253 | }), |
254 | BorrowedFormatItem::Literal(b"-" ), |
255 | component!(Month { |
256 | padding: modifier::Padding::Zero, |
257 | repr: modifier::MonthRepr::Numerical, |
258 | case_sensitive: true, |
259 | }), |
260 | BorrowedFormatItem::Literal(b"-" ), |
261 | component!(Day { |
262 | padding: modifier::Padding::Zero, |
263 | }), |
264 | ]), |
265 | b'g' => component!(Year { |
266 | padding: padding_or_default(*padding, modifier::Padding::Zero), |
267 | repr: modifier::YearRepr::LastTwo, |
268 | iso_week_based: true, |
269 | sign_is_mandatory: false, |
270 | }), |
271 | b'G' => component!(Year { |
272 | padding: modifier::Padding::Zero, |
273 | repr: modifier::YearRepr::Full, |
274 | iso_week_based: true, |
275 | sign_is_mandatory: false, |
276 | }), |
277 | b'H' => component!(Hour { |
278 | padding: padding_or_default(*padding, modifier::Padding::Zero), |
279 | is_12_hour_clock: false, |
280 | }), |
281 | b'I' => component!(Hour { |
282 | padding: padding_or_default(*padding, modifier::Padding::Zero), |
283 | is_12_hour_clock: true, |
284 | }), |
285 | b'j' => component!(Ordinal { |
286 | padding: padding_or_default(*padding, modifier::Padding::Zero), |
287 | }), |
288 | b'k' => component!(Hour { |
289 | padding: padding_or_default(*padding, modifier::Padding::Space), |
290 | is_12_hour_clock: false, |
291 | }), |
292 | b'l' => component!(Hour { |
293 | padding: padding_or_default(*padding, modifier::Padding::Space), |
294 | is_12_hour_clock: true, |
295 | }), |
296 | b'm' => component!(Month { |
297 | padding: padding_or_default(*padding, modifier::Padding::Zero), |
298 | repr: modifier::MonthRepr::Numerical, |
299 | case_sensitive: true, |
300 | }), |
301 | b'M' => component!(Minute { |
302 | padding: padding_or_default(*padding, modifier::Padding::Zero), |
303 | }), |
304 | b'n' => BorrowedFormatItem::Literal(b" \n" ), |
305 | b'O' => { |
306 | return Err(Error { |
307 | _inner: unused(ErrorInner { |
308 | _message: "unsupported modifier" , |
309 | _span: component.span, |
310 | }), |
311 | public: InvalidFormatDescription::NotSupported { |
312 | what: "modifier" , |
313 | context: "" , |
314 | index: component.span.start.byte as _, |
315 | }, |
316 | }) |
317 | } |
318 | b'p' => component!(Period { |
319 | is_uppercase: true, |
320 | case_sensitive: true |
321 | }), |
322 | b'P' => component!(Period { |
323 | is_uppercase: false, |
324 | case_sensitive: true |
325 | }), |
326 | b'r' => BorrowedFormatItem::Compound(&[ |
327 | component!(Hour { |
328 | padding: modifier::Padding::Zero, |
329 | is_12_hour_clock: true, |
330 | }), |
331 | BorrowedFormatItem::Literal(b":" ), |
332 | component!(Minute { |
333 | padding: modifier::Padding::Zero, |
334 | }), |
335 | BorrowedFormatItem::Literal(b":" ), |
336 | component!(Second { |
337 | padding: modifier::Padding::Zero, |
338 | }), |
339 | BorrowedFormatItem::Literal(b" " ), |
340 | component!(Period { |
341 | is_uppercase: true, |
342 | case_sensitive: true, |
343 | }), |
344 | ]), |
345 | b'R' => BorrowedFormatItem::Compound(&[ |
346 | component!(Hour { |
347 | padding: modifier::Padding::Zero, |
348 | is_12_hour_clock: false, |
349 | }), |
350 | BorrowedFormatItem::Literal(b":" ), |
351 | component!(Minute { |
352 | padding: modifier::Padding::Zero, |
353 | }), |
354 | ]), |
355 | b's' => component!(UnixTimestamp { |
356 | precision: modifier::UnixTimestampPrecision::Second, |
357 | sign_is_mandatory: false, |
358 | }), |
359 | b'S' => component!(Second { |
360 | padding: padding_or_default(*padding, modifier::Padding::Zero), |
361 | }), |
362 | b't' => BorrowedFormatItem::Literal(b" \t" ), |
363 | b'T' => BorrowedFormatItem::Compound(&[ |
364 | component!(Hour { |
365 | padding: modifier::Padding::Zero, |
366 | is_12_hour_clock: false, |
367 | }), |
368 | BorrowedFormatItem::Literal(b":" ), |
369 | component!(Minute { |
370 | padding: modifier::Padding::Zero, |
371 | }), |
372 | BorrowedFormatItem::Literal(b":" ), |
373 | component!(Second { |
374 | padding: modifier::Padding::Zero, |
375 | }), |
376 | ]), |
377 | b'u' => component!(Weekday { |
378 | repr: modifier::WeekdayRepr::Monday, |
379 | one_indexed: true, |
380 | case_sensitive: true, |
381 | }), |
382 | b'U' => component!(WeekNumber { |
383 | padding: padding_or_default(*padding, modifier::Padding::Zero), |
384 | repr: modifier::WeekNumberRepr::Sunday, |
385 | }), |
386 | b'V' => component!(WeekNumber { |
387 | padding: padding_or_default(*padding, modifier::Padding::Zero), |
388 | repr: modifier::WeekNumberRepr::Iso, |
389 | }), |
390 | b'w' => component!(Weekday { |
391 | repr: modifier::WeekdayRepr::Sunday, |
392 | one_indexed: true, |
393 | case_sensitive: true, |
394 | }), |
395 | b'W' => component!(WeekNumber { |
396 | padding: padding_or_default(*padding, modifier::Padding::Zero), |
397 | repr: modifier::WeekNumberRepr::Monday, |
398 | }), |
399 | b'x' => BorrowedFormatItem::Compound(&[ |
400 | component!(Month { |
401 | repr: modifier::MonthRepr::Numerical, |
402 | padding: modifier::Padding::Zero, |
403 | case_sensitive: true, |
404 | }), |
405 | BorrowedFormatItem::Literal(b"/" ), |
406 | component!(Day { |
407 | padding: modifier::Padding::Zero |
408 | }), |
409 | BorrowedFormatItem::Literal(b"/" ), |
410 | component!(Year { |
411 | padding: modifier::Padding::Zero, |
412 | repr: modifier::YearRepr::LastTwo, |
413 | iso_week_based: false, |
414 | sign_is_mandatory: false, |
415 | }), |
416 | ]), |
417 | b'X' => BorrowedFormatItem::Compound(&[ |
418 | component!(Hour { |
419 | padding: modifier::Padding::Zero, |
420 | is_12_hour_clock: false, |
421 | }), |
422 | BorrowedFormatItem::Literal(b":" ), |
423 | component!(Minute { |
424 | padding: modifier::Padding::Zero, |
425 | }), |
426 | BorrowedFormatItem::Literal(b":" ), |
427 | component!(Second { |
428 | padding: modifier::Padding::Zero, |
429 | }), |
430 | ]), |
431 | b'y' => component!(Year { |
432 | padding: padding_or_default(*padding, modifier::Padding::Zero), |
433 | repr: modifier::YearRepr::LastTwo, |
434 | iso_week_based: false, |
435 | sign_is_mandatory: false, |
436 | }), |
437 | b'Y' => component!(Year { |
438 | padding: modifier::Padding::Zero, |
439 | repr: modifier::YearRepr::Full, |
440 | iso_week_based: false, |
441 | sign_is_mandatory: false, |
442 | }), |
443 | b'z' => BorrowedFormatItem::Compound(&[ |
444 | component!(OffsetHour { |
445 | sign_is_mandatory: true, |
446 | padding: modifier::Padding::Zero, |
447 | }), |
448 | component!(OffsetMinute { |
449 | padding: modifier::Padding::Zero, |
450 | }), |
451 | ]), |
452 | b'Z' => { |
453 | return Err(Error { |
454 | _inner: unused(ErrorInner { |
455 | _message: "unsupported component" , |
456 | _span: component.span, |
457 | }), |
458 | public: InvalidFormatDescription::NotSupported { |
459 | what: "component" , |
460 | context: "" , |
461 | index: component.span.start.byte as _, |
462 | }, |
463 | }) |
464 | } |
465 | _ => { |
466 | return Err(Error { |
467 | _inner: unused(ErrorInner { |
468 | _message: "invalid component" , |
469 | _span: component.span, |
470 | }), |
471 | public: InvalidFormatDescription::InvalidComponentName { |
472 | name: String::from_utf8_lossy(&[*component]).into_owned(), |
473 | index: component.span.start.byte as _, |
474 | }, |
475 | }) |
476 | } |
477 | }) |
478 | } |
479 | |