Allow parsers to accept unvalidated UTF8 (#295)

Partially resolves #275

We still validate for utf16, ideally the utf16 parsing would be
available in the `ixdtf` crate.

Author: HalidOdat

src/builtins/core/date.rs

@@ -431,6 +431,22 @@ impl PlainDate {
         partial.calendar.date_from_partial(&partial, overflow)
     }
 
+    // Converts a UTF-8 encoded string into a `PlainDate`.
+    pub fn from_utf8(s: &[u8]) -> TemporalResult<Self> {
+        let parse_record = parse_date_time(s)?;
+
+        let calendar = parse_record
+            .calendar
+            .map(Calendar::try_from_utf8)
+            .transpose()?
+            .unwrap_or_default();
+
+        // Assertion: PlainDate must exist on a DateTime parse.
+        let date = parse_record.date.temporal_unwrap()?;
+
+        Self::try_new(date.year, date.month, date.day, calendar)
+    }
+
     /// Creates a date time with values from a `PartialDate`.
     pub fn with(
         &self,
@@ -726,18 +742,7 @@ impl FromStr for PlainDate {
     type Err = TemporalError;
 
     fn from_str(s: &str) -> Result<Self, Self::Err> {
-        let parse_record = parse_date_time(s)?;
-
-        let calendar = parse_record
-            .calendar
-            .map(Calendar::try_from_utf8)
-            .transpose()?
-            .unwrap_or_default();
-
-        // Assertion: PlainDate must exist on a DateTime parse.
-        let date = parse_record.date.temporal_unwrap()?;
-
-        Self::try_new(date.year, date.month, date.day, calendar)
+        Self::from_utf8(s.as_bytes())
     }
 }
 

src/builtins/core/datetime.rs

@@ -449,6 +449,34 @@ impl PlainDateTime {
         Self::from_date_and_time(date, PlainTime::new_unchecked(iso_time))
     }
 
+    // Converts a UTF-8 encoded string into a `PlainDateTime`.
+    pub fn from_utf8(s: &[u8]) -> TemporalResult<Self> {
+        let parse_record = parse_date_time(s)?;
+
+        let calendar = parse_record
+            .calendar
+            .map(Calendar::try_from_utf8)
+            .transpose()?
+            .unwrap_or_default();
+
+        let time = parse_record
+            .time
+            .map(IsoTime::from_time_record)
+            .transpose()?
+            .unwrap_or_default();
+
+        let parsed_date = parse_record.date.temporal_unwrap()?;
+
+        let date = IsoDate::new_with_overflow(
+            parsed_date.year,
+            parsed_date.month,
+            parsed_date.day,
+            ArithmeticOverflow::Reject,
+        )?;
+
+        Ok(Self::new_unchecked(IsoDateTime::new(date, time)?, calendar))
+    }
+
     /// Creates a new `DateTime` with the fields of a `PartialDateTime`.
     ///
     /// ```rust
@@ -816,30 +844,7 @@ impl FromStr for PlainDateTime {
     type Err = TemporalError;
 
     fn from_str(s: &str) -> Result<Self, Self::Err> {
-        let parse_record = parse_date_time(s)?;
-
-        let calendar = parse_record
-            .calendar
-            .map(Calendar::try_from_utf8)
-            .transpose()?
-            .unwrap_or_default();
-
-        let time = parse_record
-            .time
-            .map(IsoTime::from_time_record)
-            .transpose()?
-            .unwrap_or_default();
-
-        let parsed_date = parse_record.date.temporal_unwrap()?;
-
-        let date = IsoDate::new_with_overflow(
-            parsed_date.year,
-            parsed_date.month,
-            parsed_date.day,
-            ArithmeticOverflow::Reject,
-        )?;
-
-        Ok(Self::new_unchecked(IsoDateTime::new(date, time)?, calendar))
+        Self::from_utf8(s.as_bytes())
     }
 }
 

src/builtins/core/duration.rs

@@ -257,6 +257,113 @@ impl Duration {
         )
     }
 
+    // Converts a UTF-8 encoded string into a `Duration`.
+    pub fn from_utf8(s: &[u8]) -> TemporalResult<Self> {
+        let parse_record = IsoDurationParser::from_utf8(s)
+            .parse()
+            .map_err(|e| TemporalError::range().with_message(format!("{e}")))?;
+
+        let (hours, minutes, seconds, millis, micros, nanos) = match parse_record.time {
+            Some(TimeDurationRecord::Hours { hours, fraction }) => {
+                let unadjusted_fraction =
+                    fraction.and_then(|x| x.to_nanoseconds()).unwrap_or(0) as u64;
+                let fractional_hours_ns = unadjusted_fraction * 3600;
+                let minutes = fractional_hours_ns.div_euclid(60 * 1_000_000_000);
+                let fractional_minutes_ns = fractional_hours_ns.rem_euclid(60 * 1_000_000_000);
+
+                let seconds = fractional_minutes_ns.div_euclid(1_000_000_000);
+                let fractional_seconds = fractional_minutes_ns.rem_euclid(1_000_000_000);
+
+                let milliseconds = fractional_seconds.div_euclid(1_000_000);
+                let rem = fractional_seconds.rem_euclid(1_000_000);
+
+                let microseconds = rem.div_euclid(1_000);
+                let nanoseconds = rem.rem_euclid(1_000);
+
+                (
+                    hours,
+                    minutes,
+                    seconds,
+                    milliseconds,
+                    microseconds,
+                    nanoseconds,
+                )
+            }
+            // Minutes variant is defined as { hours: u32, minutes: u32, fraction: u64 }
+            Some(TimeDurationRecord::Minutes {
+                hours,
+                minutes,
+                fraction,
+            }) => {
+                let unadjusted_fraction =
+                    fraction.and_then(|x| x.to_nanoseconds()).unwrap_or(0) as u64;
+                let fractional_minutes_ns = unadjusted_fraction * 60;
+                let seconds = fractional_minutes_ns.div_euclid(1_000_000_000);
+                let fractional_seconds = fractional_minutes_ns.rem_euclid(1_000_000_000);
+
+                let milliseconds = fractional_seconds.div_euclid(1_000_000);
+                let rem = fractional_seconds.rem_euclid(1_000_000);
+
+                let microseconds = rem.div_euclid(1_000);
+                let nanoseconds = rem.rem_euclid(1_000);
+
+                (
+                    hours,
+                    minutes,
+                    seconds,
+                    milliseconds,
+                    microseconds,
+                    nanoseconds,
+                )
+            }
+            // Seconds variant is defined as { hours: u32, minutes: u32, seconds: u32, fraction: u32 }
+            Some(TimeDurationRecord::Seconds {
+                hours,
+                minutes,
+                seconds,
+                fraction,
+            }) => {
+                let ns = fraction.and_then(|x| x.to_nanoseconds()).unwrap_or(0);
+                let milliseconds = ns.div_euclid(1_000_000);
+                let rem = ns.rem_euclid(1_000_000);
+
+                let microseconds = rem.div_euclid(1_000);
+                let nanoseconds = rem.rem_euclid(1_000);
+
+                (
+                    hours,
+                    minutes,
+                    seconds,
+                    milliseconds as u64,
+                    microseconds as u64,
+                    nanoseconds as u64,
+                )
+            }
+            None => (0, 0, 0, 0, 0, 0),
+        };
+
+        let (years, months, weeks, days) = if let Some(date) = parse_record.date {
+            (date.years, date.months, date.weeks, date.days)
+        } else {
+            (0, 0, 0, 0)
+        };
+
+        let sign = parse_record.sign as i64;
+
+        Self::new(
+            years as i64 * sign,
+            months as i64 * sign,
+            weeks as i64 * sign,
+            days as i64 * sign,
+            hours as i64 * sign,
+            minutes as i64 * sign,
+            seconds as i64 * sign,
+            millis as i64 * sign,
+            micros as i128 * sign as i128,
+            nanos as i128 * sign as i128,
+        )
+    }
+
     /// Return if the Durations values are within their valid ranges.
     #[inline]
     #[must_use]

src/builtins/core/instant.rs

@@ -163,6 +163,51 @@ impl Instant {
         Self::try_new(epoch_nanos)
     }
 
+    // Converts a UTF-8 encoded string into a `Instant`.
+    pub fn from_utf8(s: &[u8]) -> TemporalResult<Self> {
+        let ixdtf_record = parse_instant(s)?;
+
+        // Find the offset
+        let ns_offset = match ixdtf_record.offset {
+            UtcOffsetRecordOrZ::Offset(offset) => {
+                let ns = offset
+                    .fraction
+                    .and_then(|x| x.to_nanoseconds())
+                    .unwrap_or(0);
+                (offset.hour as i64 * NANOSECONDS_PER_HOUR
+                    + i64::from(offset.minute) * NANOSECONDS_PER_MINUTE
+                    + i64::from(offset.second) * NANOSECONDS_PER_SECOND
+                    + i64::from(ns))
+                    * offset.sign as i64
+            }
+            UtcOffsetRecordOrZ::Z => 0,
+        };
+
+        let time_nanoseconds = ixdtf_record
+            .time
+            .fraction
+            .and_then(|x| x.to_nanoseconds())
+            .unwrap_or(0);
+        let (millisecond, rem) = time_nanoseconds.div_rem_euclid(&1_000_000);
+        let (microsecond, nanosecond) = rem.div_rem_euclid(&1_000);
+
+        let balanced = IsoDateTime::balance(
+            ixdtf_record.date.year,
+            ixdtf_record.date.month.into(),
+            ixdtf_record.date.day.into(),
+            ixdtf_record.time.hour.into(),
+            ixdtf_record.time.minute.into(),
+            ixdtf_record.time.second.clamp(0, 59).into(),
+            millisecond.into(),
+            microsecond.into(),
+            i128::from(nanosecond) - i128::from(ns_offset),
+        );
+
+        let nanoseconds = balanced.as_nanoseconds()?;
+
+        Ok(Self(nanoseconds))
+    }
+
     /// Adds a `Duration` to the current `Instant`, returning an error if the `Duration`
     /// contains a `DateDuration`.
     #[inline]
@@ -274,48 +319,9 @@ impl Instant {
 
 impl FromStr for Instant {
     type Err = TemporalError;
-    fn from_str(s: &str) -> Result<Self, Self::Err> {
-        let ixdtf_record = parse_instant(s)?;
-
-        // Find the offset
-        let ns_offset = match ixdtf_record.offset {
-            UtcOffsetRecordOrZ::Offset(offset) => {
-                let ns = offset
-                    .fraction
-                    .and_then(|x| x.to_nanoseconds())
-                    .unwrap_or(0);
-                (offset.hour as i64 * NANOSECONDS_PER_HOUR
-                    + i64::from(offset.minute) * NANOSECONDS_PER_MINUTE
-                    + i64::from(offset.second) * NANOSECONDS_PER_SECOND
-                    + i64::from(ns))
-                    * offset.sign as i64
-            }
-            UtcOffsetRecordOrZ::Z => 0,
-        };
-
-        let time_nanoseconds = ixdtf_record
-            .time
-            .fraction
-            .and_then(|x| x.to_nanoseconds())
-            .unwrap_or(0);
-        let (millisecond, rem) = time_nanoseconds.div_rem_euclid(&1_000_000);
-        let (microsecond, nanosecond) = rem.div_rem_euclid(&1_000);
 
-        let balanced = IsoDateTime::balance(
-            ixdtf_record.date.year,
-            ixdtf_record.date.month.into(),
-            ixdtf_record.date.day.into(),
-            ixdtf_record.time.hour.into(),
-            ixdtf_record.time.minute.into(),
-            ixdtf_record.time.second.clamp(0, 59).into(),
-            millisecond.into(),
-            microsecond.into(),
-            i128::from(nanosecond) - i128::from(ns_offset),
-        );
-
-        let nanoseconds = balanced.as_nanoseconds()?;
-
-        Ok(Self(nanoseconds))
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        Self::from_utf8(s.as_bytes())
     }
 }