Merge pull request #205 from yuankunzhang/refactor-timestamp-parsing

refactor: replace floating-point seconds with precise integer representation

src/items/builder.rs

@@ -3,7 +3,7 @@
 
 use chrono::{DateTime, Datelike, FixedOffset, NaiveDate, TimeZone, Timelike};
 
-use super::{date, relative, time, timezone, weekday, year};
+use super::{date, epoch, relative, time, timezone, weekday, year};
 
 /// The builder is used to construct a DateTime object from various components.
 /// The parser creates a `DateTimeBuilder` object with the parsed components,
@@ -13,7 +13,7 @@ use super::{date, relative, time, timezone, weekday, year};
 #[derive(Debug, Default)]
 pub(crate) struct DateTimeBuilder {
     base: Option<DateTime<FixedOffset>>,
-    timestamp: Option<f64>,
+    timestamp: Option<epoch::Timestamp>,
     date: Option<date::Date>,
     time: Option<time::Time>,
     weekday: Option<weekday::Weekday>,
@@ -35,7 +35,7 @@ impl DateTimeBuilder {
 
     /// Sets a timestamp value. Timestamp values are exclusive to other date/time
     /// items (date, time, weekday, timezone, relative adjustments).
-    pub(super) fn set_timestamp(mut self, ts: f64) -> Result<Self, &'static str> {
+    pub(super) fn set_timestamp(mut self, ts: epoch::Timestamp) -> Result<Self, &'static str> {
         if self.timestamp.is_some() {
             return Err("timestamp cannot appear more than once");
         } else if self.date.is_some()
@@ -148,15 +148,15 @@ impl DateTimeBuilder {
         self.set_time(time)
     }
 
-    fn build_from_timestamp(ts: f64, tz: &FixedOffset) -> Option<DateTime<FixedOffset>> {
-        // TODO: How to make the fract -> nanosecond conversion more precise?
-        // Maybe considering using the
-        // [rust_decimal](https://crates.io/crates/rust_decimal) crate?
-        match chrono::Utc.timestamp_opt(ts as i64, (ts.fract() * 10f64.powi(9)).round() as u32) {
+    fn build_from_timestamp(
+        ts: epoch::Timestamp,
+        tz: &FixedOffset,
+    ) -> Option<DateTime<FixedOffset>> {
+        match chrono::Utc.timestamp_opt(ts.second, ts.nanosecond) {
             chrono::MappedLocalTime::Single(t) => Some(t.with_timezone(tz)),
             chrono::MappedLocalTime::Ambiguous(earliest, _latest) => {
-                // TODO: When there is a fold in the local time, which value
-                // do we choose? For now, we use the earliest one.
+                // When there is a fold in the local time, we use the earliest
+                // one.
                 Some(earliest.with_timezone(tz))
             }
             chrono::MappedLocalTime::None => None, // Invalid timestamp
@@ -210,6 +210,7 @@ impl DateTimeBuilder {
             hour,
             minute,
             second,
+            nanosecond,
             ref offset,
         }) = self.time
         {
@@ -224,8 +225,8 @@ impl DateTimeBuilder {
                 dt.day(),
                 hour,
                 minute,
-                second as u32,
-                (second.fract() * 10f64.powi(9)).round() as u32,
+                second,
+                nanosecond,
                 offset,
             )?;
         }

src/items/combined.rs

@@ -55,7 +55,8 @@ mod tests {
             time: Time {
                 hour: 10,
                 minute: 10,
-                second: 55.0,
+                second: 55,
+                nanosecond: 0,
                 offset: None,
             },
         });

src/items/epoch.rs

@@ -15,35 +15,147 @@
 //! > ‘@1483228800’ represents 2017-01-01 00:00:00 UTC, and there is no way to
 //! > represent the intervening leap second 2016-12-31 23:59:60 UTC.
 
-use winnow::{combinator::preceded, ModalResult, Parser};
+use winnow::{
+    ascii::digit1,
+    combinator::{opt, preceded},
+    token::one_of,
+    ModalResult, Parser,
+};
 
-use super::primitive::{float, s};
+use super::primitive::{dec_uint, s};
 
-/// Parse a timestamp in the form of `@1234567890`.
-pub fn parse(input: &mut &str) -> ModalResult<f64> {
-    s(preceded("@", float)).parse_next(input)
+/// Represents a timestamp with nanosecond accuracy.
+///
+/// # Invariants
+///
+/// - `nanosecond` is always in the range of `0..1_000_000_000`.
+/// - Negative timestamps are represented by a negative `second` value and a
+///   positive `nanosecond` value.
+#[derive(Debug, PartialEq)]
+pub(crate) struct Timestamp {
+    pub(crate) second: i64,
+    pub(crate) nanosecond: u32,
+}
+
+/// Parse a timestamp in the form of `1234567890` or `-1234567890.12345` or
+/// `1234567890,12345`.
+pub(crate) fn parse(input: &mut &str) -> ModalResult<Timestamp> {
+    (s("@"), opt(s(one_of(['-', '+']))), sec_and_nsec)
+        .verify_map(|(_, sign, (sec, nsec))| {
+            let sec = i64::try_from(sec).ok()?;
+            let (second, nanosecond) = match (sign, nsec) {
+                (Some('-'), 0) => (-sec, 0),
+                // Truncate towards minus infinity.
+                (Some('-'), _) => ((-sec).checked_sub(1)?, 1_000_000_000 - nsec),
+                _ => (sec, nsec),
+            };
+            Some(Timestamp { second, nanosecond })
+        })
+        .parse_next(input)
+}
+
+/// Parse a second value in the form of `1234567890` or `1234567890.12345` or
+/// `1234567890,12345`.
+///
+/// The first part represents whole seconds. The optional second part represents
+/// fractional seconds, parsed as a nanosecond value from up to 9 digits
+/// (padded with zeros on the right if fewer digits are present). If the second
+/// part is omitted, it defaults to 0 nanoseconds.
+pub(super) fn sec_and_nsec(input: &mut &str) -> ModalResult<(u64, u32)> {
+    (s(dec_uint), opt(preceded(one_of(['.', ',']), digit1)))
+        .verify_map(|(sec, opt_nsec_str)| match opt_nsec_str {
+            Some(nsec_str) if nsec_str.len() >= 9 => Some((sec, nsec_str[..9].parse().ok()?)),
+            Some(nsec_str) => {
+                let multiplier = 10_u32.pow(9 - nsec_str.len() as u32);
+                Some((sec, nsec_str.parse::<u32>().ok()?.checked_mul(multiplier)?))
+            }
+            None => Some((sec, 0)),
+        })
+        .parse_next(input)
 }
 
 #[cfg(test)]
 mod tests {
-    use super::parse;
+    use super::*;
 
-    fn float_eq(a: f64, b: f64) -> bool {
-        (a - b).abs() < f64::EPSILON
+    #[test]
+    fn sec_and_nsec_test() {
+        let mut input = "1234567890";
+        assert_eq!(sec_and_nsec(&mut input).unwrap(), (1234567890, 0));
+
+        let mut input = "1234567890.12345";
+        assert_eq!(sec_and_nsec(&mut input).unwrap(), (1234567890, 123450000));
+
+        let mut input = "1234567890,12345";
+        assert_eq!(sec_and_nsec(&mut input).unwrap(), (1234567890, 123450000));
+
+        let mut input = "1234567890.1234567890123";
+        assert_eq!(sec_and_nsec(&mut input).unwrap(), (1234567890, 123456789));
     }
 
     #[test]
-    fn float() {
+    fn timestamp() {
         let mut input = "@1234567890";
-        assert!(float_eq(parse(&mut input).unwrap(), 1234567890.0));
+        assert_eq!(
+            parse(&mut input).unwrap(),
+            Timestamp {
+                second: 1234567890,
+                nanosecond: 0,
+            }
+        );
+
+        let mut input = "@ 1234567890";
+        assert_eq!(
+            parse(&mut input).unwrap(),
+            Timestamp {
+                second: 1234567890,
+                nanosecond: 0,
+            }
+        );
+
+        let mut input = "@ -1234567890";
+        assert_eq!(
+            parse(&mut input).unwrap(),
+            Timestamp {
+                second: -1234567890,
+                nanosecond: 0,
+            }
+        );
+
+        let mut input = "@ - 1234567890";
+        assert_eq!(
+            parse(&mut input).unwrap(),
+            Timestamp {
+                second: -1234567890,
+                nanosecond: 0,
+            }
+        );
 
         let mut input = "@1234567890.12345";
-        assert!(float_eq(parse(&mut input).unwrap(), 1234567890.12345));
+        assert_eq!(
+            parse(&mut input).unwrap(),
+            Timestamp {
+                second: 1234567890,
+                nanosecond: 123450000,
+            }
+        );
 
         let mut input = "@1234567890,12345";
-        assert!(float_eq(parse(&mut input).unwrap(), 1234567890.12345));
+        assert_eq!(
+            parse(&mut input).unwrap(),
+            Timestamp {
+                second: 1234567890,
+                nanosecond: 123450000,
+            }
+        );
 
         let mut input = "@-1234567890.12345";
-        assert_eq!(parse(&mut input).unwrap(), -1234567890.12345);
+        assert_eq!(
+            parse(&mut input).unwrap(),
+            Timestamp {
+                second: -1234567891,
+                nanosecond: 876550000,
+            }
+        );
     }
 }

src/items/mod.rs

@@ -58,7 +58,7 @@ use crate::ParseDateTimeError;
 
 #[derive(PartialEq, Debug)]
 pub(crate) enum Item {
-    Timestamp(f64),
+    Timestamp(epoch::Timestamp),
     DateTime(combined::DateTime),
     Date(date::Date),
     Time(time::Time),

src/items/primitive.rs

@@ -3,8 +3,10 @@
 
 //! Primitive combinators.
 
+use std::str::FromStr;
+
 use winnow::{
-    ascii::{digit1, multispace0},
+    ascii::{digit1, multispace0, Uint},
     combinator::{alt, delimited, not, opt, peek, preceded, repeat, separated},
     error::{ContextError, ParserError, StrContext, StrContextValue},
     stream::AsChar,
@@ -100,8 +102,9 @@ where
 ///
 /// See the rationale for `dec_int` for why we don't use
 /// `winnow::ascii::dec_uint`.
-pub(super) fn dec_uint<'a, E>(input: &mut &'a str) -> winnow::Result<u32, E>
+pub(super) fn dec_uint<'a, O, E>(input: &mut &'a str) -> winnow::Result<O, E>
 where
+    O: Uint + FromStr,
     E: ParserError<&'a str>,
 {
     digit1

src/items/time.rs

@@ -40,11 +40,12 @@
 use winnow::{
     combinator::{alt, opt, preceded},
     error::ErrMode,
-    seq, ModalResult, Parser,
+    ModalResult, Parser,
 };
 
 use super::{
-    primitive::{colon, ctx_err, dec_uint, float, s},
+    epoch::sec_and_nsec,
+    primitive::{colon, ctx_err, dec_uint, s},
     timezone::{timezone_num, Offset},
 };
 
@@ -52,7 +53,8 @@ use super::{
 pub(crate) struct Time {
     pub hour: u32,
     pub minute: u32,
-    pub second: f64,
+    pub second: u32,
+    pub nanosecond: u32,
     pub offset: Option<Offset>,
 }
 
@@ -74,16 +76,24 @@ pub(super) fn iso(input: &mut &str) -> ModalResult<Time> {
         (hour24, timezone_num).map(|(hour, offset)| Time {
             hour,
             minute: 0,
-            second: 0.0,
+            second: 0,
+            nanosecond: 0,
             offset: Some(offset),
         }),
-        seq!(Time {
-            hour: hour24,
-            _: colon,
-            minute: minute,
-            second: opt(preceded(colon, second)).map(|s| s.unwrap_or(0.0)),
-            offset: opt(timezone_num)
-        }),
+        (
+            hour24,
+            colon,
+            minute,
+            opt(preceded(colon, second)),
+            opt(timezone_num),
+        )
+            .map(|(hour, _, minute, sec_nsec, offset)| Time {
+                hour,
+                minute,
+                second: sec_nsec.map_or(0, |(s, _)| s),
+                nanosecond: sec_nsec.map_or(0, |(_, ns)| ns),
+                offset,
+            }),
     ))
     .parse_next(input)
 }
@@ -92,7 +102,7 @@ pub(super) fn iso(input: &mut &str) -> ModalResult<Time> {
 ///
 /// The hours are restricted to 12 or lower in this format
 fn am_pm_time(input: &mut &str) -> ModalResult<Time> {
-    let (h, m, s, meridiem) = (
+    let (h, m, sec_nsec, meridiem) = (
         hour12,
         opt(preceded(colon, minute)),
         opt(preceded(colon, second)),
@@ -118,7 +128,8 @@ fn am_pm_time(input: &mut &str) -> ModalResult<Time> {
     Ok(Time {
         hour: h,
         minute: m.unwrap_or(0),
-        second: s.unwrap_or(0.0),
+        second: sec_nsec.map_or(0, |(s, _)| s),
+        nanosecond: sec_nsec.map_or(0, |(_, ns)| ns),
         offset: None,
     })
 }
@@ -139,14 +150,9 @@ pub(super) fn minute(input: &mut &str) -> ModalResult<u32> {
 }
 
 /// Parse a number of seconds (preceded by whitespace)
-fn second(input: &mut &str) -> ModalResult<f64> {
-    s(float)
-        .verify(|x| *x < 60.0)
-        .map(|x| {
-            // Truncates the fractional part of seconds to 9 digits.
-            let factor = 10f64.powi(9);
-            (x * factor).trunc() / factor
-        })
+fn second(input: &mut &str) -> ModalResult<(u32, u32)> {
+    sec_and_nsec
+        .verify_map(|(s, ns)| if s < 60 { Some((s as u32, ns)) } else { None })
         .parse_next(input)
 }
 
@@ -159,7 +165,8 @@ mod tests {
         let reference = Time {
             hour: 20,
             minute: 2,
-            second: 0.0,
+            second: 0,
+            nanosecond: 0,
             offset: None,
         };
 
@@ -201,7 +208,8 @@ mod tests {
         let reference = Time {
             hour: 11,
             minute: 0,
-            second: 0.0,
+            second: 0,
+            nanosecond: 0,
             offset: None,
         };
 
@@ -223,12 +231,50 @@ mod tests {
         }
     }
 
+    #[test]
+    fn nanoseconds() {
+        let reference = Time {
+            hour: 11,
+            minute: 0,
+            second: 0,
+            nanosecond: 123450000,
+            offset: None,
+        };
+
+        for mut s in ["11:00:00.12345", "11:00:00.12345am"] {
+            let old_s = s.to_owned();
+            assert_eq!(
+                parse(&mut s).ok(),
+                Some(reference.clone()),
+                "Format string: {old_s}"
+            );
+        }
+
+        let reference = Time {
+            hour: 11,
+            minute: 0,
+            second: 0,
+            nanosecond: 123456789,
+            offset: None,
+        };
+
+        for mut s in ["11:00:00.123456789", "11:00:00.1234567890123"] {
+            let old_s = s.to_owned();
+            assert_eq!(
+                parse(&mut s).ok(),
+                Some(reference.clone()),
+                "Format string: {old_s}"
+            );
+        }
+    }
+
     #[test]
     fn noon() {
         let reference = Time {
             hour: 12,
             minute: 0,
-            second: 0.0,
+            second: 0,
+            nanosecond: 0,
             offset: None,
         };
 
@@ -254,7 +300,8 @@ mod tests {
         let reference = Time {
             hour: 0,
             minute: 0,
-            second: 0.0,
+            second: 0,
+            nanosecond: 0,
             offset: None,
         };
 
@@ -273,7 +320,8 @@ mod tests {
         let reference = Time {
             hour: 1,
             minute: 23,
-            second: 0.0,
+            second: 0,
+            nanosecond: 0,
             offset: Some(Offset {
                 negative: false,
                 hours: 5,
@@ -303,7 +351,8 @@ mod tests {
         let reference = Time {
             hour: 3,
             minute: 45,
-            second: 0.0,
+            second: 0,
+            nanosecond: 0,
             offset: Some(Offset {
                 negative: false,
                 hours: 5,
@@ -335,7 +384,8 @@ mod tests {
         let reference = Time {
             hour: 3,
             minute: 45,
-            second: 0.0,
+            second: 0,
+            nanosecond: 0,
             offset: Some(Offset {
                 negative: false,
                 hours: 0,
@@ -366,7 +416,8 @@ mod tests {
         let reference = Time {
             hour: 3,
             minute: 45,
-            second: 0.0,
+            second: 0,
+            nanosecond: 0,
             offset: Some(Offset {
                 negative: true,
                 hours: 5,