Skip to main content

structured_pkix/
time.rs

1//! The two ways a certificate writes a moment, read into one.
2//!
3//! Both preserve UTC labels, including an inserted second. The form is kept because it is
4//! part of what the certificate says: the profile requires the two-digit form
5//! through 2049 and the four-digit form afterwards, and a report that has to
6//! explain a certificate cannot explain what it did not keep.
7//!
8//! Only the fixed-width form with seconds and `Z` is accepted. The wider
9//! encodings the base standard allows, an offset from local time, minutes
10//! without seconds, a fractional part, all let one moment be written several
11//! ways, and here they would let two documents disagree about when a
12//! certificate expired while both claiming to be the same certificate.
13
14use crate::error::{Error, ErrorCode, Result};
15use crate::temporal::Instant;
16use core::fmt;
17use structured_asn1::{Element, Reader, Tag, universal};
18
19mod xsd10;
20pub use xsd10::Xsd10Label;
21
22/// UTC-qualified relation of a source label to the caller's lossless reading.
23/// This is a temporal fact, not a rule for application acceptance.
24#[derive(Debug, Clone, Copy, PartialEq, Eq)]
25pub enum Relation {
26    /// Source label is strictly before the reading.
27    Before,
28    /// Source label equals the exact reading.
29    At,
30    /// Source label is strictly after the reading.
31    After,
32    /// Supplied UTC facts cannot establish an input label.
33    Uncertain(crate::temporal::Uncertainty),
34}
35
36pub(crate) fn relation_in<W: structured_asn1::memory::Work>(
37    label: Instant,
38    now: crate::temporal::Reading,
39    utc: &crate::temporal::LeapTable<'_>,
40    execution: &W,
41    origin: (&'static str, u64),
42) -> Result<Relation> {
43    let (field, at) = origin;
44    // Two labels, three binary searches each, before calling the table core.
45    let searches = u64::from(usize::BITS - utc.transitions().len().leading_zeros() + 1);
46    execution
47        .charge_work(searches * 6 + 64)
48        .map_err(|reason| Error::resource(reason, field, at))?;
49    for compared in [label, now.start()] {
50        if let Err(reason) = utc.shift(compared, 0) {
51            return Ok(Relation::Uncertain(reason));
52        }
53    }
54    Ok(
55        if label < now.start() || (label == now.start() && now.has_subnanosecond_remainder()) {
56            Relation::Before
57        } else if label == now.start() {
58            Relation::At
59        } else {
60            Relation::After
61        },
62    )
63}
64
65/// Profile calendar conversion with producing-reader admission and location.
66pub(crate) fn read_in(
67    parent: &Reader<'_>,
68    element: &Element<'_>,
69    field: &'static str,
70) -> Result<Time> {
71    admit_in(parent, element, field)?;
72    read(element, field)
73}
74
75/// Bare DER `GeneralizedTime` retains fractional precision, unlike a
76/// certificate-validity [`Time`]. X.690 §11.7 permits canonical decimal fractions.
77/// <https://www.itu.int/rec/T-REC-X.690-202102-I/en>
78pub(crate) fn read_generalized_in(
79    parent: &Reader<'_>,
80    element: &Element<'_>,
81    field: &'static str,
82) -> Result<crate::temporal::Reading> {
83    if element.tag() != Tag::primitive(universal::GENERALIZED_TIME) {
84        return Err(Error::new(
85            ErrorCode::Encoding(structured_asn1::ErrorCode::UnexpectedTag),
86            field,
87            element.span().start,
88        ));
89    }
90    admit_in(parent, element, field)?;
91    read_label_as(element, field, Fraction::Written).map(|(reading, _)| reading)
92}
93
94fn admit_in(parent: &Reader<'_>, element: &Element<'_>, field: &'static str) -> Result<()> {
95    // Fixed calendar conversion plus four byte passes includes validation,
96    // digit conversion and the shared GeneralizedTime semantic core.
97    let work = u64::try_from(element.contents().len())
98        .ok()
99        .and_then(|length| length.checked_mul(4))
100        .and_then(|work| work.checked_add(64))
101        .ok_or_else(|| {
102            Error::resource(
103                structured_asn1::memory::MemoryError::WorkExceeded,
104                field,
105                element.span().start,
106            )
107        })?;
108    parent.charge(work).map_err(|error| {
109        Error::new(
110            ErrorCode::Encoding(error.code()),
111            field,
112            element.span().start,
113        )
114    })
115}
116
117/// Which form a time was written in.
118#[derive(Debug, Clone, Copy, PartialEq, Eq)]
119pub enum Form {
120    /// Two digits of year, which the profile requires through 2049.
121    Utc,
122    /// Four digits of year, which the profile requires from 2050.
123    Generalized,
124}
125
126/// A moment, and how the certificate wrote it.
127///
128/// Two times are equal when they name the same moment. The form is what the
129/// certificate used to say it, not part of which moment it is.
130#[derive(Debug, Clone, Copy)]
131pub struct Time {
132    instant: Instant,
133    form: Form,
134}
135
136/// Gregorian calendar components of a labelled time, not an elapsed-time proof.
137/// The year uses astronomical numbering (zero denotes 1 BCE); callers of other
138/// calendar contracts must apply their own specified year-numbering convention.
139#[derive(Debug, Clone, Copy, PartialEq, Eq)]
140pub struct Calendar {
141    /// Signed astronomical Gregorian year.
142    pub year: i64,
143    /// Month, one through twelve.
144    pub month: u8,
145    /// Day within the month.
146    pub day: u8,
147    /// Hour, zero through twenty-three.
148    pub hour: u8,
149    /// Minute, zero through fifty-nine.
150    pub minute: u8,
151    /// Labelled second, including sixty for an inserted second.
152    pub second: u8,
153    /// Fractional nanoseconds retained by this `Time` value.
154    pub nanos: u32,
155}
156
157/// The moment `99991231235959Z` names, which the profile uses to say that a
158/// certificate has no well defined expiry rather than that it expires then.
159pub const NO_WELL_DEFINED_EXPIRY: i64 = 253_402_300_799;
160
161impl Time {
162    /// Project the original UTC label into XSD1.0 year numbering.
163    /// This retains second sixty; [`Xsd10Label::at_utc`] performs the separate
164    /// XSD value normalization. It does not change this time's UTC identity.
165    pub fn xsd10_label<W: structured_asn1::memory::Work>(
166        self,
167        execution: &W,
168    ) -> core::result::Result<Xsd10Label, structured_asn1::memory::MemoryError> {
169        let calendar = self.calendar_at_minutes(0, execution)?;
170        Ok(Xsd10Label {
171            year: if calendar.year <= 0 {
172                calendar.year - 1
173            } else {
174                calendar.year
175            },
176            month: calendar.month,
177            day: calendar.day,
178            hour: calendar.hour,
179            minute: calendar.minute,
180            second: calendar.second,
181            nanos: calendar.nanos,
182        })
183    }
184
185    /// Project the original label at a signed calendar offset.
186    /// Offset minutes need not describe a valid wire timezone; this also permits
187    /// inspecting the previous local date of a midnight label without a second
188    /// calendar implementation. No UTC correction history is inferred.
189    /// The same inverse Gregorian core serves [`Self`]'s text rendering.
190    ///
191    /// # Panics
192    /// Only if that private Gregorian core violates its bounded component
193    /// invariants; parsed time labels and any i16-minute offset preserve them.
194    pub fn calendar_at_minutes<W: structured_asn1::memory::Work>(
195        self,
196        minutes: i16,
197        execution: &W,
198    ) -> core::result::Result<Calendar, structured_asn1::memory::MemoryError> {
199        execution.charge_work(64)?;
200        // Parsed Time years are 0000..9999. The complete i16-minute offset
201        // changes this bounded label by less than 23 days, so no i64 overflow.
202        let (year, month, day, hour, minute, second) =
203            civil_from_seconds(self.seconds() + i64::from(minutes) * 60);
204        Ok(Calendar {
205            year,
206            month: u8::try_from(month).expect("Gregorian month"),
207            day: u8::try_from(day).expect("Gregorian day"),
208            hour: u8::try_from(hour).expect("Gregorian hour"),
209            minute: u8::try_from(minute).expect("Gregorian minute"),
210            second: if self.is_leap_second() {
211                60
212            } else {
213                u8::try_from(second).expect("Gregorian second")
214            },
215            nanos: self.instant.nanos(),
216        })
217    }
218
219    /// POSIX label, negative before the epoch. For an inserted second this
220    /// is the preceding ordinary second's label. Use [`Self::instant`] for
221    /// comparisons and [`crate::temporal::LeapTable`] for elapsed arithmetic.
222    #[must_use]
223    pub fn seconds(self) -> i64 {
224        self.instant.seconds()
225    }
226
227    /// The lossless UTC label; parsing does not authenticate its occurrence.
228    #[must_use]
229    pub fn instant(self) -> Instant {
230        self.instant
231    }
232
233    /// Whether the encoded time labels an inserted second.
234    #[must_use]
235    pub fn is_leap_second(self) -> bool {
236        self.instant.is_leap_second()
237    }
238
239    /// Which form the certificate wrote this in.
240    #[must_use]
241    pub fn form(self) -> Form {
242        self.form
243    }
244
245    /// Whether this is the moment the profile uses to mean that no expiry was
246    /// defined. What that means for a verdict is a policy question, and this
247    /// only reports that the certificate said it.
248    #[must_use]
249    pub fn is_no_well_defined_expiry(self) -> bool {
250        self.seconds() == NO_WELL_DEFINED_EXPIRY && !self.is_leap_second()
251    }
252}
253
254impl PartialEq for Time {
255    fn eq(&self, other: &Self) -> bool {
256        self.instant == other.instant
257    }
258}
259
260impl Eq for Time {}
261
262impl PartialOrd for Time {
263    fn partial_cmp(&self, other: &Self) -> Option<core::cmp::Ordering> {
264        Some(self.cmp(other))
265    }
266}
267
268impl Ord for Time {
269    fn cmp(&self, other: &Self) -> core::cmp::Ordering {
270        self.instant.cmp(&other.instant)
271    }
272}
273
274impl fmt::Display for Time {
275    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
276        let (year, month, day, hour, minute, ordinary) = civil_from_seconds(self.seconds());
277        let second = if self.is_leap_second() { 60 } else { ordinary };
278        write!(
279            f,
280            "{year:04}-{month:02}-{day:02}T{hour:02}:{minute:02}:{second:02}Z"
281        )
282    }
283}
284
285/// Reads an element that a schema says is a time.
286///
287/// The tag chooses the form, so a field declared as a choice between the two is
288/// read by handing whichever element was found to this.
289pub(crate) fn read(element: &Element<'_>, field: &'static str) -> Result<Time> {
290    read_as(element, field, Fraction::Absent)
291}
292
293/// Whether a fractional-seconds element may close the value.
294#[derive(Clone, Copy, PartialEq, Eq)]
295pub(crate) enum Fraction {
296    /// The profile writes a certificate's validity dates without one
297    /// (RFC 5280 4.1.2.5.2), and this reader answers with one moment, which
298    /// a fraction would not change.
299    Absent,
300    /// The encoding writes one where there is one to write, with no trailing
301    /// zeros and none at all where it stands for zero (X.690 11.7.3). A
302    /// value preserved inside somebody else's payload is written under that
303    /// rule and not under the profile's narrower one.
304    Written,
305}
306
307/// Reads a time under whichever of the two rules the caller is asking about.
308pub(crate) fn read_as(
309    element: &Element<'_>,
310    field: &'static str,
311    fraction: Fraction,
312) -> Result<Time> {
313    let (reading, form) = read_label_as(element, field, fraction)?;
314    Ok(Time {
315        instant: reading.start(),
316        form,
317    })
318}
319
320fn read_label_as(
321    element: &Element<'_>,
322    field: &'static str,
323    fraction: Fraction,
324) -> Result<(crate::temporal::Reading, Form)> {
325    let fail = |code| Error::new(code, field, element.span().start);
326
327    let form = if element.tag() == Tag::primitive(universal::UTC_TIME) {
328        Form::Utc
329    } else if element.tag() == Tag::primitive(universal::GENERALIZED_TIME) {
330        Form::Generalized
331    } else {
332        return Err(fail(ErrorCode::Encoding(
333            structured_asn1::ErrorCode::UnexpectedTag,
334        )));
335    };
336
337    let contents = element.contents();
338    let (year, rest) = match form {
339        Form::Utc => {
340            let (digits, rest) =
341                split(contents, 2).ok_or_else(|| fail(ErrorCode::MalformedTime))?;
342            let two = number(digits).ok_or_else(|| fail(ErrorCode::MalformedTime))?;
343            // The profile's pivot: a two-digit year is read into the window
344            // 1950 to 2049, so a certificate written in 1998 does not become
345            // one written in 2098.
346            (if two >= 50 { 1900 + two } else { 2000 + two }, rest)
347        }
348        Form::Generalized => {
349            // Any year, including one the profile tells issuers to write as a
350            // UTCTime instead. That rule binds the issuer: "CAs conforming to
351            // this profile MUST always encode certificate validity dates
352            // through the year 2049 as UTCTime". The sentence after it binds
353            // this reader the other way: "Conforming applications MUST be
354            // able to process validity dates that are encoded in either
355            // UTCTime or GeneralizedTime" (RFC 5280 4.1.2.5). A date written
356            // in the wrong form still names one moment, unambiguously, and
357            // refusing to read it would leave a person holding a document
358            // this product declined to explain over its issuer's spelling.
359            let (digits, rest) =
360                split(contents, 4).ok_or_else(|| fail(ErrorCode::MalformedTime))?;
361            let four = number(digits).ok_or_else(|| fail(ErrorCode::MalformedTime))?;
362            (four, rest)
363        }
364    };
365
366    let ([month, day, hour, minute, second], tail) =
367        fields_of(rest).ok_or_else(|| fail(ErrorCode::MalformedTime))?;
368
369    // A fraction closes the value where one is written at all: only the wider
370    // form has one, and only where the caller is reading under the encoding's
371    // rule rather than the profile's.
372    let closed = match tail.split_first() {
373        Some((b'.', digits)) if fraction == Fraction::Written && form == Form::Generalized => {
374            // At least one digit, no trailing zero, and the Z after them.
375            matches!(digits.split_last(), Some((b'Z', fraction)) if !fraction.is_empty()
376                && fraction.iter().all(u8::is_ascii_digit)
377                && fraction.last() != Some(&b'0'))
378        }
379        _ => tail == b"Z",
380    };
381    if !closed {
382        return Err(fail(ErrorCode::MalformedTime));
383    }
384
385    if form == Form::Generalized {
386        // X.680 (2021) §46.3 incorporates ISO 8601 time-of-day notation,
387        // including inserted-second labels. X.690 §11.7 restricts spelling,
388        // not UTC occurrence. After the spelling checks above, use the same
389        // calendar/label semantics as the temporal reader; a LeapTable alone
390        // can establish occurrence. UTCTime has its own 00..59 grammar below.
391        // https://www.itu.int/rec/T-REC-X.680-202102-I/en
392        // https://www.itu.int/rec/T-REC-X.690-202102-I/en
393        let reading = crate::temporal::generalized(element, field)?;
394        return Ok((reading, form));
395    }
396
397    if !(1..=12).contains(&month) || day < 1 || day > days_in_month(year, month) {
398        return Err(fail(ErrorCode::TimeOutOfRange));
399    }
400    // X.680 (2021) §47.3(b)(2): "ss is seconds (00 to 59)" for UTCTime.
401    // https://www.itu.int/rec/T-REC-X.680-202102-I/en
402    if hour > 23 || minute > 59 || second > 59 {
403        return Err(fail(ErrorCode::TimeOutOfRange));
404    }
405
406    // Every field is bounded by the checks above and the year by its digit
407    // count, so the seconds count is at most the year 9999 in seconds and
408    // nothing here can overflow.
409    let days = days_from_civil(i64::from(year), month, day);
410    let seconds =
411        days * 86_400 + i64::from(hour) * 3_600 + i64::from(minute) * 60 + i64::from(second);
412
413    let instant = Instant::new(seconds, 0).ok_or_else(|| fail(ErrorCode::TimeOutOfRange))?;
414    Ok((instant.into(), form))
415}
416
417/// Useful time syntax under BER, independently of a PKIX validity field.
418/// X.680 (2021) §§46.2–46.3, 47.3; X.690 §8.25. CER/DER §§11.7–11.8
419/// narrow this syntax and are deliberately applied only by the DER reader.
420/// <https://www.itu.int/rec/T-REC-X.680-202102-I/en>
421pub(crate) fn ber_fits(contents: &[u8], generalized: bool) -> bool {
422    let year_width = if generalized { 4 } else { 2 };
423    let Some((year, rest)) = split(contents, year_width) else {
424        return false;
425    };
426    let Some(year) = number(year) else {
427        return false;
428    };
429    let (clock, zone) = match rest
430        .iter()
431        .position(|byte| matches!(byte, b'Z' | b'+' | b'-'))
432    {
433        Some(at) => rest.split_at(at),
434        None => (rest, &[][..]),
435    };
436    if !ber_zone(zone, generalized) {
437        return false;
438    }
439    let digits = match clock.iter().position(|byte| matches!(byte, b'.' | b',')) {
440        Some(at) if generalized => {
441            let (digits, fraction) = clock.split_at(at);
442            if fraction.len() < 2 || !fraction[1..].iter().all(u8::is_ascii_digit) {
443                return false;
444            }
445            digits
446        }
447        Some(_) => return false,
448        None => clock,
449    };
450    let Some(([month, day, hour], tail)) = fields_of(digits) else {
451        return false;
452    };
453    if month == 0 || month > 12 || day == 0 || day > days_in_month(year, month) || hour > 23 {
454        return false;
455    }
456    // UTCTime gives only the low two year digits, not a century pivot. The
457    // cycle check allows February 29 for yy=00 (some possible centuries are
458    // leap years); selecting a century belongs to the interpreting schema.
459    if tail.is_empty() {
460        return generalized;
461    }
462    let Some(([minute], tail)) = fields_of(tail) else {
463        return false;
464    };
465    if minute > 59 {
466        return false;
467    }
468    if tail.is_empty() {
469        return true;
470    }
471    let Some(([second], [])) = fields_of(tail) else {
472        return false;
473    };
474    second <= if generalized { 60 } else { 59 }
475}
476
477fn ber_zone(zone: &[u8], generalized: bool) -> bool {
478    if zone.is_empty() {
479        return generalized;
480    }
481    if zone == b"Z" {
482        return true;
483    }
484    let Some((b'+' | b'-', digits)) = zone.split_first() else {
485        return false;
486    };
487    let Some(([hour], tail)) = fields_of(digits) else {
488        return false;
489    };
490    if hour > 23 {
491        return false;
492    }
493    if tail.is_empty() {
494        return generalized;
495    }
496    matches!(fields_of(tail), Some(([minute], [])) if minute <= 59)
497}
498
499/// Splits off `count` bytes, which must all be digits.
500pub(super) fn split(contents: &[u8], count: usize) -> Option<(&[u8], &[u8])> {
501    let (digits, rest) = contents.split_at_checked(count)?;
502    digits
503        .iter()
504        .all(u8::is_ascii_digit)
505        .then_some((digits, rest))
506}
507
508/// Reads a run of digits already known to be digits.
509pub(super) fn number(digits: &[u8]) -> Option<u32> {
510    digits.iter().try_fold(0u32, |value, digit| {
511        value
512            .checked_mul(10)
513            .and_then(|value| value.checked_add(u32::from(digit - b'0')))
514    })
515}
516
517/// Where a year falls in the four-hundred-year cycle the calendar repeats
518/// on, which is what every question about its length answers to.
519///
520/// The Gregorian calendar comes back to the same shape every four hundred
521/// years: the leap rule is stated in fours, hundreds and four hundreds, and
522/// four hundred years hold a whole number of weeks. So a year of any width
523/// is answered by its last four digits, which is what makes a year too wide
524/// for a counter no less readable than a short one. The sign turns the count
525/// the other way round the cycle.
526fn in_cycle(digits: &str, negative: bool) -> u32 {
527    let counted = digits.as_bytes().iter().fold(0u32, |cycle, digit| {
528        (cycle * 10 + u32::from(digit - b'0')) % 400
529    });
530    if negative {
531        (400 - counted) % 400
532    } else {
533        counted
534    }
535}
536
537/// How many weeks a year is written in: fifty-two, and fifty-three where the
538/// year is the long one.
539///
540/// A year is long where it begins on a Thursday, and where it is a leap year
541/// beginning on a Wednesday (ISO 8601 4.1.4, which the week components of
542/// X.680 Table 7 are written in). Both are said by the day the first of
543/// January falls on, which is what the count below is: the days from the
544/// start of the calendar to that year, over seven.
545fn weeks_in_year(year: u32) -> u32 {
546    let opens = |year: u32| (year + year / 4 - year / 100 + year / 400) % 7;
547    // The second reading needs the year before, which the first year of the
548    // calendar does not have; it opens the calendar rather than continuing
549    // one, so only the first reading applies to it.
550    let long = opens(year) == 4 || year.checked_sub(1).is_some_and(|before| opens(before) == 3);
551    if long { 53 } else { 52 }
552}
553
554fn is_leap(year: u32) -> bool {
555    year.is_multiple_of(4) && (!year.is_multiple_of(100) || year.is_multiple_of(400))
556}
557
558pub(super) fn days_in_month(year: u32, month: u32) -> u32 {
559    match month {
560        1 | 3 | 5 | 7 | 8 | 10 | 12 => 31,
561        4 | 6 | 9 | 11 => 30,
562        2 if is_leap(year) => 29,
563        2 => 28,
564        _ => 0,
565    }
566}
567
568/// Days from the epoch to a date, by the shift-the-year-to-March method: with
569/// March first, the four-year, hundred-year and four-hundred-year corrections
570/// all land at the end of the year and the leap day needs no special case.
571///
572/// Every value is bounded by the calendar checks above, so ordinary arithmetic
573/// cannot overflow here.
574pub(super) fn days_from_civil(year: i64, month: u32, day: u32) -> i64 {
575    let month = i64::from(month);
576    let year = if month <= 2 { year - 1 } else { year };
577    let era = if year >= 0 { year } else { year - 399 } / 400;
578    let year_of_era = year - era * 400;
579    let day_of_year = (153 * (month + if month > 2 { -3 } else { 9 }) + 2) / 5 + i64::from(day) - 1;
580    let day_of_era = year_of_era * 365 + year_of_era / 4 - year_of_era / 100 + day_of_year;
581    // 719468 is the number of days from the epoch back to the start of the era
582    // this counts from.
583    era * 146_097 + day_of_era - 719_468
584}
585
586/// The inverse, for rendering. Bounded the same way.
587pub(super) fn civil_from_seconds(seconds: i64) -> (i64, i64, i64, i64, i64, i64) {
588    let days = seconds.div_euclid(86_400);
589    let within = seconds.rem_euclid(86_400);
590
591    let shifted = days + 719_468;
592    let era = if shifted >= 0 {
593        shifted
594    } else {
595        shifted - 146_096
596    } / 146_097;
597    let day_of_era = shifted - era * 146_097;
598    let year_of_era =
599        (day_of_era - day_of_era / 1_460 + day_of_era / 36_524 - day_of_era / 146_096) / 365;
600    let year = year_of_era + era * 400;
601    let day_of_year = day_of_era - (365 * year_of_era + year_of_era / 4 - year_of_era / 100);
602    let shifted_month = (5 * day_of_year + 2) / 153;
603    let day = day_of_year - (153 * shifted_month + 2) / 5 + 1;
604    let month = if shifted_month < 10 {
605        shifted_month + 3
606    } else {
607        shifted_month - 9
608    };
609
610    (
611        if month <= 2 { year + 1 } else { year },
612        month,
613        day,
614        within / 3_600,
615        within / 60 % 60,
616        within % 60,
617    )
618}
619
620/// Which of the types defined in X.680 38.4 a value claims to be, for the
621/// four whose notation is settled enough to hold it to.
622#[derive(Clone, Copy, PartialEq, Eq)]
623pub(crate) enum Defined {
624    /// `DATE`, a year, a month and a day.
625    Date,
626    /// `TIME-OF-DAY`, an hour, a minute and a second.
627    TimeOfDay,
628    /// `DATE-TIME`, the two of them.
629    DateTime,
630    /// `DURATION`, a length of time rather than a moment.
631    Duration,
632    /// `TIME` itself, whose notation is the whole of the interchange format
633    /// and is read here only as the characters it is written from.
634    Any,
635}
636
637/// Whether the contents of one of the newer time types are the notation that
638/// type writes.
639///
640/// The contents are the value notation in UTF-8, with the characters X.690
641/// 8.26 names removed: the hyphens of a date, the colons of a time, the T
642/// between them, the P before a duration. What is left is what this reads.
643/// A value that is not the notation is not a value of the type, and one
644/// preserved for somebody else must not claim a type it does not have.
645pub(crate) fn defined_fits(kind: Defined, contents: &[u8]) -> bool {
646    match kind {
647        Defined::Date => date_fits(contents),
648        Defined::TimeOfDay => time_of_day_fits(contents),
649        Defined::DateTime => contents
650            .split_at_checked(8)
651            .is_some_and(|(date, time)| date_fits(date) && time_of_day_fits(time)),
652        Defined::Duration => duration_fits(contents),
653        Defined::Any => core::str::from_utf8(contents).is_ok_and(any_fits),
654    }
655}
656
657/// The unrestricted type, whose notation is the whole of what the property
658/// settings can choose: a point in time, a duration, an interval of the two,
659/// and any of those repeated (X.680 Table 7).
660///
661/// Written in the extended formats, which keep their separators, and X.690
662/// 8.26.1.2 takes only the quotation marks off, so what arrives here is the
663/// notation as the table writes it.
664fn any_fits(text: &str) -> bool {
665    // Recurrence opens the value: R, how many times if it is not unlimited,
666    // and then the interval it repeats.
667    let interval = match text.strip_prefix('R') {
668        Some(after) => {
669            let count = after.bytes().take_while(u8::is_ascii_digit).count();
670            let (repeats, tail) = after.split_at(count);
671            if repeats.starts_with('0') && repeats.len() > 1 {
672                return false;
673            }
674            let Some(rest) = tail.strip_prefix('/') else {
675                return false;
676            };
677            rest
678        }
679        None => text,
680    };
681    match interval.split_once('/') {
682        // An interval is a start point and an end point, a start point and a
683        // duration, or a duration and an end point (Table 7, "Time
684        // interval"). Two durations name no interval: nothing says when.
685        Some((start, end)) => match (part_fits(start), part_fits(end)) {
686            (Some(starts), Some(ends)) => !(starts == Part::Duration && ends == Part::Duration),
687            _ => false,
688        },
689        None => part_fits(interval).is_some(),
690    }
691}
692
693/// Which of the two an end of an interval, or a value standing alone, is.
694#[derive(Clone, Copy, PartialEq, Eq)]
695enum Part {
696    Point,
697    Duration,
698}
699
700/// Whether one end of a value is a point in time or a length of one.
701fn part_fits(text: &str) -> Option<Part> {
702    if let Some(duration) = text.strip_prefix('P') {
703        return duration_fits(duration.as_bytes()).then_some(Part::Duration);
704    }
705    point_fits(text).then_some(Part::Point)
706}
707
708/// Whether text is a point in time: a date, a time of day, or both with the
709/// T between them, and the zone the value is read in if it carries one.
710fn point_fits(text: &str) -> bool {
711    match text.split_once('T') {
712        Some((date, time)) => date_part_fits(date) && zoned_time_fits(time),
713        // The hours component of a time standing alone is written without
714        // the T (Table 7, "Hours component" with Basic=Time), so what is
715        // left is one or the other.
716        None => date_part_fits(text) || zoned_time_fits(text),
717    }
718}
719
720/// Whether text is a time of day and, where it carries one, the zone it is
721/// read in: the UTC designator, or the difference from it (Table 7, the last
722/// two rows). The zone follows the time and nothing else, so it is looked
723/// for here rather than at the end of the whole point.
724fn zoned_time_fits(text: &str) -> bool {
725    let stated = match text.strip_suffix('Z') {
726        Some(rest) => rest,
727        None => match text.rfind(['+', '-']) {
728            Some(at) if difference_fits(&text[at..]) => &text[..at],
729            _ => text,
730        },
731    };
732    time_part_fits(stated)
733}
734
735/// Whether text is the difference from UTC: a sign, hours, and minutes where
736/// the difference is not a whole number of hours.
737fn difference_fits(text: &str) -> bool {
738    let Some(rest) = text.strip_prefix(['+', '-']) else {
739        return false;
740    };
741    match rest.split_once(':') {
742        Some((hours, minutes)) => {
743            two_digits(hours).is_some_and(|hours| hours <= 23)
744                && two_digits(minutes).is_some_and(|minutes| minutes <= 59)
745        }
746        None => two_digits(rest).is_some_and(|hours| hours <= 23),
747    }
748}
749
750/// Whether text is a date in the extended format: a year, and then a month
751/// and a day, a week and a day of it, or a day of the year (Table 7, the
752/// year, month, week and day components).
753fn date_part_fits(text: &str) -> bool {
754    let signed = text.starts_with(['+', '-']);
755    let rest = if signed { &text[1..] } else { text };
756    let mut parts = rest.split('-');
757    let Some(year) = parts.next() else {
758        return false;
759    };
760    // Four digits, or more where the settings ask for a wider year.
761    if year.len() < 4 || !year.bytes().all(|byte| byte.is_ascii_digit()) {
762        return false;
763    }
764    let Some(next) = parts.next() else {
765        return parts.next().is_none();
766    };
767    // How far the days and the weeks of this year run is the year's to say,
768    // and a year of any width says it: the calendar repeats every four
769    // hundred years, so where in that cycle the year falls is the whole of
770    // the answer.
771    let counted = in_cycle(year, signed);
772    if let Some(week) = next.strip_prefix('W') {
773        // A week, and then which day of it, one digit through seven.
774        let last = weeks_in_year(counted);
775        let weeks = two_digits(week).is_some_and(|week| (1..=last).contains(&week));
776        return weeks
777            && match parts.next() {
778                Some(day) => {
779                    day.len() == 1
780                        && day.bytes().all(|byte| (b'1'..=b'7').contains(&byte))
781                        && parts.next().is_none()
782                }
783                None => true,
784            };
785    }
786    if next.len() == 3 {
787        // The day of the year, which stands in for the month and the day,
788        // and runs to the length of that year.
789        let last = if is_leap(counted) { 366 } else { 365 };
790        return next.bytes().all(|byte| byte.is_ascii_digit())
791            && number(next.as_bytes()).is_some_and(|day| (1..=last).contains(&day))
792            && parts.next().is_none();
793    }
794    let Some(month) = two_digits(next).filter(|month| (1..=12).contains(month)) else {
795        return false;
796    };
797    match parts.next() {
798        Some(day) => {
799            let Some(day) = two_digits(day) else {
800                return false;
801            };
802            // The calendar decides the last day of the month, and the year
803            // decides February.
804            let last = days_in_month(counted, month);
805            day >= 1 && day <= last && parts.next().is_none()
806        }
807        None => true,
808    }
809}
810
811/// Whether text is a time of day in the extended format: hours, and then
812/// minutes and seconds, the last component present carrying a fraction where
813/// the value is written to that accuracy (Table 7).
814fn time_part_fits(text: &str) -> bool {
815    let mut parts = text.split(':');
816    let bounds = [23u32, 59, 60];
817    let mut carried = None;
818    let mut counted = 0usize;
819    let mut ending = false;
820    for (piece, most) in parts.by_ref().zip(bounds) {
821        let (digits, fraction) = match piece.split_once(['.', ',']) {
822            Some((digits, fraction)) => (digits, Some(fraction)),
823            None => (piece, None),
824        };
825        let Some(value) = two_digits(digits) else {
826            return false;
827        };
828        // A fraction says how accurately the value is written, so the
829        // component carrying one is the last (ISO 8601, 4.2.2.4).
830        if let Some(fraction) = fraction {
831            if fraction.is_empty() || !fraction.bytes().all(|byte| byte.is_ascii_digit()) {
832                return false;
833            }
834            carried = Some(counted);
835        }
836        counted += 1;
837        if carried.is_some_and(|at| at + 1 != counted) {
838            return false;
839        }
840        // Midnight at the end of a day is the twenty-fourth hour, which is
841        // the one value past the bounds (Table 7, "Hours component"). It
842        // names one instant rather than an hour a value can sit inside, so
843        // what follows it is zero and nothing carries a fraction: a
844        // fraction of that hour, or of a component after it, is a moment
845        // past the end of the day.
846        if ending && fraction.is_some() {
847            return false;
848        }
849        if counted == 1 && value == 24 {
850            ending = true;
851            if fraction.is_some() {
852                return false;
853            }
854            continue;
855        }
856        if value > most || (ending && value != 0) {
857            return false;
858        }
859    }
860    counted > 0 && parts.next().is_none()
861}
862
863/// Two digits read as a number, which is how every component but the year
864/// and the day of the year is written.
865fn two_digits(text: &str) -> Option<u32> {
866    (text.len() == 2 && text.bytes().all(|byte| byte.is_ascii_digit()))
867        .then(|| number(text.as_bytes()))?
868}
869
870/// Eight digits naming a day of the calendar (`Date=YMD Year=Basic`).
871fn date_fits(contents: &[u8]) -> bool {
872    let Some((year, rest)) = split(contents, 4) else {
873        return false;
874    };
875    let Some(([month, day], [])) = fields_of(rest) else {
876        return false;
877    };
878    let Some(year) = number(year) else {
879        return false;
880    };
881    (1..=12).contains(&month) && day >= 1 && day <= days_in_month(year, month)
882}
883
884/// Six digits naming a time of day (`Time=HMS`).
885fn time_of_day_fits(contents: &[u8]) -> bool {
886    let Some(([hour, minute, second], [])) = fields_of(contents) else {
887        return false;
888    };
889    // Midnight is written at either end of the day, so the hour reaches
890    // twenty-four where the rest of the value is zero (X.680 38.4.2), and
891    // the second reaches sixty because the interchange format writes a leap
892    // second that way.
893    let midnight = hour == 24 && minute == 0 && second == 0;
894    (hour <= 23 || midnight) && minute <= 59 && second <= 60
895}
896
897/// A duration, as X.690 8.26.5 leaves it: the notation without its leading P,
898/// so a count and its designator for each component that is present, largest
899/// first, with the time components after a T.
900fn duration_fits(contents: &[u8]) -> bool {
901    let Ok(text) = core::str::from_utf8(contents) else {
902        return false;
903    };
904    // The week form stands alone: it is one of the three alternatives and
905    // does not mix with the others (X.680 38.3.6.1).
906    if text.ends_with('W') {
907        return components_fit(text, b"W").is_some_and(|weeks| weeks.any);
908    }
909    let (date, time) = match text.split_once('T') {
910        Some((date, time)) => (date, Some(time)),
911        None => (text, None),
912    };
913    // A T with nothing after it names no component.
914    if time.is_some_and(str::is_empty) {
915        return false;
916    }
917    let Some(dated) = components_fit(date, b"YMD") else {
918        return false;
919    };
920    // A fraction ends the duration, so a date part that carries one may not
921    // be followed by a time part either (X.680 38.3.6.13).
922    if dated.fractional && time.is_some() {
923        return false;
924    }
925    let Some(timed) = time.map_or(Some(Components::NONE), |time| components_fit(time, b"HMS"))
926    else {
927        return false;
928    };
929    // Something has to be there: a duration of no components is no length of
930    // time.
931    dated.any || timed.any
932}
933
934/// What a run of designations turned out to be.
935#[derive(Clone, Copy)]
936struct Components {
937    /// Whether there were any at all, which is a form the caller decides
938    /// about rather than a malformation on its own.
939    any: bool,
940    /// Whether the last one carried a fraction, which closes the duration.
941    fractional: bool,
942}
943
944impl Components {
945    const NONE: Self = Self {
946        any: false,
947        fractional: false,
948    };
949}
950
951/// Whether a run is designations in the order given, each designator used at
952/// most once: a count, an optional fraction, and the letter naming the
953/// component (X.680 38.3.6.2 to 38.3.6.13). `None` says the run is not
954/// designations at all.
955fn components_fit(text: &str, order: &[u8]) -> Option<Components> {
956    let mut rest = text.as_bytes();
957    let mut left = order;
958    let mut seen = Components::NONE;
959    while !rest.is_empty() {
960        // The count opens with a zero only where it is the single digit zero
961        // (38.3.6.11), so the accuracy a value is written to is written one
962        // way.
963        let count = rest.iter().take_while(|byte| byte.is_ascii_digit()).count();
964        if count == 0 || (rest[0] == b'0' && count > 1) {
965            return None;
966        }
967        let after = &rest[count..];
968        let (after, fractional) = match after.split_first() {
969            Some((b'.' | b',', tail)) => {
970                let digits = tail.iter().take_while(|byte| byte.is_ascii_digit()).count();
971                if digits == 0 {
972                    return None;
973                }
974                (&tail[digits..], true)
975            }
976            _ => (after, false),
977        };
978        let (designator, tail) = after.split_first()?;
979        let at = left.iter().position(|kind| kind == designator)?;
980        left = &left[at + 1..];
981        // "If a designation contains a fractional part, there shall be no
982        // following designation" (38.3.6.13): the fraction says how
983        // accurately the duration is written, and a component after it
984        // would say something more accurate again.
985        if seen.fractional {
986            return None;
987        }
988        seen = Components {
989            any: true,
990            fractional,
991        };
992        rest = tail;
993    }
994    Some(seen)
995}
996
997/// The two-digit fields at the front of a value, and whatever follows them
998/// for the caller to hold to its own rule.
999pub(super) fn fields_of<const N: usize>(contents: &[u8]) -> Option<([u32; N], &[u8])> {
1000    let mut values = [0u32; N];
1001    let mut rest = contents;
1002    for value in &mut values {
1003        let (digits, tail) = split(rest, 2)?;
1004        *value = number(digits)?;
1005        rest = tail;
1006    }
1007    Some((values, rest))
1008}
1009
1010#[cfg(test)]
1011mod tests;