Add ISO 8601 parser for duration format with designators

Author: pitdicker

src/calendar_duration.rs

@@ -1,7 +1,9 @@
 use core::fmt;
 use core::num::NonZeroU32;
+use core::str;
 use core::time::Duration;
 
+use crate::format::{parse_iso8601_duration, ParseError, TOO_LONG};
 use crate::{expect, try_opt};
 
 /// ISO 8601 duration type.
@@ -117,6 +119,18 @@ impl fmt::Display for CalendarDuration {
     }
 }
 
+impl str::FromStr for CalendarDuration {
+    type Err = ParseError;
+
+    fn from_str(s: &str) -> Result<CalendarDuration, ParseError> {
+        let (s, duration) = parse_iso8601_duration(s)?;
+        if !s.is_empty() {
+            return Err(TOO_LONG);
+        }
+        Ok(duration)
+    }
+}
+
 impl CalendarDuration {
     /// Create a new duration initialized to `0`.
     ///

src/format/mod.rs

@@ -72,6 +72,7 @@ pub use locales::Locale;
 pub(crate) use locales::Locale;
 pub(crate) use parse::parse_rfc3339;
 pub use parse::{parse, parse_and_remainder};
+pub(crate) use parse_iso8601::parse_iso8601_duration;
 pub use parsed::Parsed;
 pub use strftime::StrftimeItems;
 

src/format/parse_iso8601.rs

@@ -1,5 +1,164 @@
 use super::scan;
 use super::{ParseResult, INVALID, OUT_OF_RANGE};
+use crate::CalendarDuration;
+
+/// Parser for the ISO 8601 duration format with designators.
+///
+/// Supported formats:
+/// - `Pnn̲Ynn̲Mnn̲DTnn̲Hnn̲Mnn̲S`
+/// - `Pnn̲W`
+///
+/// Any number-designator pair may be missing when zero, as long as there is at least one pair.
+/// The last pair may contain a decimal fraction instead of an integer.
+///
+/// - Fractional years will be expressed in months.
+/// - Fractional weeks will be expressed in days.
+/// - Fractional hours, minutes or seconds will be expressed in minutes, seconds and nanoseconds.
+pub(crate) fn parse_iso8601_duration(mut s: &str) -> ParseResult<(&str, CalendarDuration)> {
+    macro_rules! consume {
+        ($e:expr) => {{
+            $e.map(|(s_, v)| {
+                s = s_;
+                v
+            })
+        }};
+    }
+
+    s = scan::char(s, b'P')?;
+    let mut duration = CalendarDuration::new();
+
+    let mut next = consume!(Decimal::parse(s)).ok();
+    if let Some(val) = next {
+        if s.as_bytes().first() == Some(&b'W') {
+            s = &s[1..];
+            // Nothing is allowed after a week value
+            return Ok((s, duration.with_days(val.mul(7)?)));
+        }
+        if s.as_bytes().first() == Some(&b'Y') {
+            s = &s[1..];
+            duration = duration.with_months(val.mul(12)?);
+            if val.fraction.is_some() {
+                return Ok((s, duration));
+            }
+            next = consume!(Decimal::parse(s)).ok();
+        }
+    }
+
+    if let Some(val) = next {
+        if s.as_bytes().first() == Some(&b'M') {
+            s = &s[1..];
+            let months = duration.months().checked_add(val.integer()?).ok_or(OUT_OF_RANGE)?;
+            duration = duration.with_months(months);
+            next = consume!(Decimal::parse(s)).ok();
+        }
+    }
+
+    if let Some(val) = next {
+        if s.as_bytes().first() == Some(&b'D') {
+            s = &s[1..];
+            duration = duration.with_days(val.integer()?);
+            next = None;
+        }
+    }
+
+    if next.is_some() {
+        // We have numbers without a matching designator.
+        return Err(INVALID);
+    }
+
+    if s.as_bytes().first() == Some(&b'T') {
+        duration = consume!(parse_iso8601_duration_time(s, duration))?
+    }
+    Ok((s, duration))
+}
+
+/// Parser for the time part of the ISO 8601 duration format with designators.
+pub(crate) fn parse_iso8601_duration_time(
+    mut s: &str,
+    duration: CalendarDuration,
+) -> ParseResult<(&str, CalendarDuration)> {
+    macro_rules! consume_or_return {
+        ($e:expr, $return:expr) => {{
+            match $e {
+                Ok((s_, next)) => {
+                    s = s_;
+                    next
+                }
+                Err(_) => return $return,
+            }
+        }};
+    }
+    fn set_hms_nano(
+        duration: CalendarDuration,
+        hours: u32,
+        minutes: u32,
+        seconds: u32,
+        nanoseconds: u32,
+    ) -> ParseResult<CalendarDuration> {
+        let duration = match (hours, minutes) {
+            (0, 0) => duration.with_seconds(seconds),
+            _ => duration.with_hms(hours, minutes, seconds).ok_or(OUT_OF_RANGE)?,
+        };
+        Ok(duration.with_nanos(nanoseconds).unwrap())
+    }
+
+    s = scan::char(s, b'T')?;
+    let mut hours = 0;
+    let mut minutes = 0;
+    let mut incomplete = true; // at least one component is required
+
+    let (s_, mut next) = Decimal::parse(s)?;
+    s = s_;
+    if s.as_bytes().first() == Some(&b'H') {
+        s = &s[1..];
+        incomplete = false;
+        match next.integer() {
+            Ok(h) => hours = h,
+            _ => {
+                let (secs, nanos) = next.mul_with_nanos(3600)?;
+                let mins = secs / 60;
+                let secs = (secs % 60) as u32;
+                let minutes = u32::try_from(mins).map_err(|_| OUT_OF_RANGE)?;
+                return Ok((s, set_hms_nano(duration, 0, minutes, secs, nanos)?));
+            }
+        }
+        next = consume_or_return!(
+            Decimal::parse(s),
+            Ok((s, set_hms_nano(duration, hours, minutes, 0, 0)?))
+        );
+    }
+
+    if s.as_bytes().first() == Some(&b'M') {
+        s = &s[1..];
+        incomplete = false;
+        match next.integer() {
+            Ok(m) => minutes = m,
+            _ => {
+                let (secs, nanos) = next.mul_with_nanos(60)?;
+                let mins = secs / 60;
+                let secs = (secs % 60) as u32;
+                minutes = u32::try_from(mins).map_err(|_| OUT_OF_RANGE)?;
+                return Ok((s, set_hms_nano(duration, hours, minutes, secs, nanos)?));
+            }
+        }
+        next = consume_or_return!(
+            Decimal::parse(s),
+            Ok((s, set_hms_nano(duration, hours, minutes, 0, 0)?))
+        );
+    }
+
+    if s.as_bytes().first() == Some(&b'S') {
+        s = &s[1..];
+        let (secs, nanos) = next.mul_with_nanos(1)?;
+        let secs = u32::try_from(secs).map_err(|_| OUT_OF_RANGE)?;
+        return Ok((s, set_hms_nano(duration, hours, minutes, secs, nanos)?));
+    }
+
+    if incomplete {
+        return Err(INVALID);
+    }
+    Ok((s, set_hms_nano(duration, hours, minutes, 0, 0)?))
+}
 
 /// Helper type for parsing decimals (as in an ISO 8601 duration).
 #[derive(Copy, Clone)]
@@ -96,7 +255,7 @@ impl Fraction {
         let huge = self.0 * unit + div / 2;
         let whole = huge / POW10[15];
         let fraction_as_nanos = (huge % POW10[15]) / div;
-        dbg!(whole as i64, fraction_as_nanos as i64)
+        (whole as i64, fraction_as_nanos as i64)
     }
 }
 
@@ -121,8 +280,9 @@ const POW10: [u64; 16] = [
 
 #[cfg(test)]
 mod tests {
-    use super::Fraction;
-    use crate::format::INVALID;
+    use super::{parse_iso8601_duration, parse_iso8601_duration_time, Fraction};
+    use crate::format::{INVALID, OUT_OF_RANGE, TOO_SHORT};
+    use crate::CalendarDuration;
 
     #[test]
     fn test_parse_fraction() {
@@ -138,4 +298,128 @@ mod tests {
         let (_, fraction) = Fraction::parse(",5").unwrap();
         assert_eq!(fraction.mul_with_nanos(1), (0, 500_000_000));
     }
+
+    #[test]
+    fn test_parse_duration_time() {
+        let parse = parse_iso8601_duration_time;
+        let d = CalendarDuration::new();
+
+        assert_eq!(parse("T12H", d), Ok(("", d.with_hms(12, 0, 0).unwrap())));
+        assert_eq!(parse("T12.25H", d), Ok(("", d.with_hms(12, 15, 0).unwrap())));
+        assert_eq!(parse("T12,25H", d), Ok(("", d.with_hms(12, 15, 0).unwrap())));
+        assert_eq!(parse("T34M", d), Ok(("", d.with_hms(0, 34, 0).unwrap())));
+        assert_eq!(parse("T34.25M", d), Ok(("", d.with_hms(0, 34, 15).unwrap())));
+        assert_eq!(parse("T56S", d), Ok(("", d.with_seconds(56))));
+        assert_eq!(parse("T0.789S", d), Ok(("", d.with_millis(789).unwrap())));
+        assert_eq!(parse("T0,789S", d), Ok(("", d.with_millis(789).unwrap())));
+        assert_eq!(parse("T12H34M", d), Ok(("", d.with_hms(12, 34, 0).unwrap())));
+        assert_eq!(parse("T12H34M60S", d), Ok(("", d.with_hms(12, 34, 60).unwrap())));
+        assert_eq!(
+            parse("T12H34M56.789S", d),
+            Ok(("", d.with_hms(12, 34, 56).unwrap().with_millis(789).unwrap()))
+        );
+        assert_eq!(parse("T12H56S", d), Ok(("", d.with_hms(12, 0, 56).unwrap())));
+        assert_eq!(parse("T34M56S", d), Ok(("", d.with_hms(0, 34, 56).unwrap())));
+
+        // Data after a fraction is ignored
+        assert_eq!(parse("T12.5H16M", d), Ok(("16M", d.with_hms(12, 30, 0).unwrap())));
+        assert_eq!(parse("T12H16.5M30S", d), Ok(("30S", d.with_hms(12, 16, 30).unwrap())));
+
+        // Zero values
+        assert_eq!(parse("T0H", d), Ok(("", d)));
+        assert_eq!(parse("T0M", d), Ok(("", d)));
+        assert_eq!(parse("T0S", d), Ok(("", d)));
+        assert_eq!(parse("T0,0S", d), Ok(("", d)));
+
+        // Empty or invalid values
+        assert_eq!(parse("T", d), Err(TOO_SHORT));
+        assert_eq!(parse("TH", d), Err(INVALID));
+        assert_eq!(parse("TM", d), Err(INVALID));
+        assert_eq!(parse("TS", d), Err(INVALID));
+        assert_eq!(parse("T.5S", d), Err(INVALID));
+        assert_eq!(parse("T,5S", d), Err(INVALID));
+
+        // Date components
+        assert_eq!(parse("T5W", d), Err(INVALID));
+        assert_eq!(parse("T5Y", d), Err(INVALID));
+        assert_eq!(parse("T5D", d), Err(INVALID));
+
+        // Max values
+        assert_eq!(parse("T1118481H", d), Ok(("", d.with_hms(1118481, 0, 0).unwrap())));
+        assert_eq!(parse("T1118482H", d), Err(OUT_OF_RANGE));
+        assert_eq!(parse("T1118481.05H", d), Ok(("", d.with_hms(1118481, 3, 0).unwrap())));
+        assert_eq!(parse("T1118481.5H", d), Err(OUT_OF_RANGE));
+        assert_eq!(parse("T67108863M", d), Ok(("", d.with_hms(0, u32::MAX >> 6, 0).unwrap())));
+        assert_eq!(parse("T67108864M", d), Err(OUT_OF_RANGE));
+        assert_eq!(parse("T67108863.25M", d), Ok(("", d.with_hms(0, u32::MAX >> 6, 15).unwrap())));
+        assert_eq!(parse("T4294967295S", d), Ok(("", d.with_seconds(u32::MAX))));
+        assert_eq!(parse("T4294967296S", d), Err(OUT_OF_RANGE));
+        assert_eq!(
+            parse("T4294967295.25S", d),
+            Ok(("", d.with_seconds(u32::MAX).with_millis(250).unwrap()))
+        );
+        assert_eq!(
+            parse("T4294967295.999999999S", d),
+            Ok(("", d.with_seconds(u32::MAX).with_nanos(999_999_999).unwrap()))
+        );
+        assert_eq!(parse("T4294967295.9999999995S", d), Err(OUT_OF_RANGE));
+        assert_eq!(parse("T12H34M61S", d), Err(OUT_OF_RANGE));
+    }
+
+    #[test]
+    fn test_parse_duration() {
+        let d = CalendarDuration::new();
+        assert_eq!(
+            parse_iso8601_duration("P12Y"),
+            Ok(("", d.with_years_and_months(12, 0).unwrap()))
+        );
+        assert_eq!(parse_iso8601_duration("P34M"), Ok(("", d.with_months(34))));
+        assert_eq!(parse_iso8601_duration("P56D"), Ok(("", d.with_days(56))));
+        assert_eq!(parse_iso8601_duration("P78W"), Ok(("", d.with_weeks_and_days(78, 0).unwrap())));
+
+        // Fractional date values
+        assert_eq!(
+            parse_iso8601_duration("P1.25Y"),
+            Ok(("", d.with_years_and_months(1, 3).unwrap()))
+        );
+        assert_eq!(
+            parse_iso8601_duration("P1.99Y"),
+            Ok(("", d.with_years_and_months(2, 0).unwrap()))
+        );
+        assert_eq!(parse_iso8601_duration("P1.4W"), Ok(("", d.with_days(10))));
+        assert_eq!(parse_iso8601_duration("P1.95W"), Ok(("", d.with_days(14))));
+        assert_eq!(parse_iso8601_duration("P1.5M"), Err(INVALID));
+        assert_eq!(parse_iso8601_duration("P1.5D"), Err(INVALID));
+
+        // Data after a fraction is ignored
+        assert_eq!(
+            parse_iso8601_duration("P1.25Y5D"),
+            Ok(("5D", d.with_years_and_months(1, 3).unwrap()))
+        );
+        assert_eq!(
+            parse_iso8601_duration("P1.25YT3H"),
+            Ok(("T3H", d.with_years_and_months(1, 3).unwrap()))
+        );
+
+        // Zero values
+        assert_eq!(parse_iso8601_duration("P0Y"), Ok(("", d)));
+        assert_eq!(parse_iso8601_duration("P0M"), Ok(("", d)));
+        assert_eq!(parse_iso8601_duration("P0W"), Ok(("", d)));
+        assert_eq!(parse_iso8601_duration("P0D"), Ok(("", d)));
+        assert_eq!(parse_iso8601_duration("PT0M"), Ok(("", d)));
+        assert_eq!(parse_iso8601_duration("PT0S"), Ok(("", d)));
+
+        // Invalid designator at a position where another designator can be expected.
+        assert_eq!(parse_iso8601_duration("P12Y12Y"), Err(INVALID));
+        assert_eq!(parse_iso8601_duration("P12M12M"), Err(INVALID));
+        assert_eq!(parse_iso8601_duration("P12M12Y"), Err(INVALID));
+
+        // Trailing data
+        assert_eq!(
+            parse_iso8601_duration("P12W34D"),
+            Ok(("34D", d.with_weeks_and_days(12, 0).unwrap()))
+        );
+        assert_eq!(parse_iso8601_duration("P12D12D"), Ok(("12D", d.with_days(12))));
+        assert_eq!(parse_iso8601_duration("P12D12Y"), Ok(("12Y", d.with_days(12))));
+    }
 }