tiempo-rs/src/timeparse.rs

use chrono::{DateTime, Utc, Local, TimeZone, LocalResult, FixedOffset};
use regex::Regex;

use crate::error::{Result, Error};

pub fn parse_time(input: &str) -> Result<DateTime<Utc>> {
    // first try to parse as a full datetime with optional timezone
    let re = Regex::new(r"(?xi)
        (?P<year>\d{4})  # the year, mandatory
        .
        (?P<month>\d{2}) # the month, mandatory
        .
        (?P<day>\d{2})   # the day, mandatory
        (. # a separator
        (?P<hour>\d{2}) # the hour, optional
        (. # a separator
        (?P<minute>\d{2})? # the minute, optional
        (. # a separator
        (?P<second>\d{2}))?)?)? # the second, optional, implies minute
        (?P<offset>
            (?P<utc>Z)|((?P<sign>\+|-)(?P<ohour>\d{1,2}):(?P<omin>\d{2}))
        )? # the offset, optional
    ").unwrap();

    if let Some(caps) = re.captures(input) {
        return if let Some(_) = caps.name("offset") {
            // start with a specific offset or utc
            if let Some(_) = caps.name("utc") {
                // start with utc
                let try_date = Utc.ymd_opt(
                    (&caps["year"]).parse().unwrap(),
                    (&caps["month"]).parse().unwrap(),
                    (&caps["day"]).parse().unwrap(),
                ).and_hms_opt(
                    caps.name("hour").map(|m| m.as_str().parse().unwrap()).unwrap_or(0),
                    caps.name("minute").map(|m| m.as_str().parse().unwrap()).unwrap_or(0),
                    caps.name("second").map(|m| m.as_str().parse().unwrap()).unwrap_or(0),
                );

                match try_date {
                    LocalResult::None => Err(Error::NoneLocalTime(input.into())),
                    LocalResult::Single(t) => Ok(t),
                    LocalResult::Ambiguous(t1, t2) => Err(Error::AmbiguousLocalTime {
                        orig: input.into(),
                        t1: t1.naive_utc(),
                        t2: t2.naive_utc(),
                    }),
                }
            } else {
                let mut offset: i32 = (&caps["ohour"]).parse::<i32>().unwrap() * 60 * 60;
                offset += (&caps["omin"]).parse::<i32>().unwrap() * 60;

                let fo = if &caps["sign"] == "+" {
                    FixedOffset::east(offset)
                } else {
                    FixedOffset::west(offset)
                };

                let try_date = fo.ymd_opt(
                    (&caps["year"]).parse().unwrap(),
                    (&caps["month"]).parse().unwrap(),
                    (&caps["day"]).parse().unwrap(),
                ).and_hms_opt(
                    caps.name("hour").map(|m| m.as_str().parse().unwrap()).unwrap_or(0),
                    caps.name("minute").map(|m| m.as_str().parse().unwrap()).unwrap_or(0),
                    caps.name("second").map(|m| m.as_str().parse().unwrap()).unwrap_or(0),
                );

                match try_date {
                    LocalResult::None => Err(Error::NoneLocalTime(input.into())),
                    LocalResult::Single(t) => Ok(Utc.from_utc_datetime(&t.naive_utc())),
                    LocalResult::Ambiguous(t1, t2) => Err(Error::AmbiguousLocalTime {
                        orig: input.into(),
                        t1: t1.naive_utc(),
                        t2: t2.naive_utc(),
                    }),
                }
            }
        } else {
            // start with a local time
            let try_date = Local.ymd_opt(
                (&caps["year"]).parse().unwrap(),
                (&caps["month"]).parse().unwrap(),
                (&caps["day"]).parse().unwrap(),
            ).and_hms_opt(
                caps.name("hour").map(|m| m.as_str().parse().unwrap()).unwrap_or(0),
                caps.name("minute").map(|m| m.as_str().parse().unwrap()).unwrap_or(0),
                caps.name("second").map(|m| m.as_str().parse().unwrap()).unwrap_or(0),
            );

            match try_date {
                LocalResult::None => Err(Error::NoneLocalTime(input.into())),
                LocalResult::Single(t) => Ok(Utc.from_utc_datetime(&t.naive_utc())),
                LocalResult::Ambiguous(t1, t2) => Err(Error::AmbiguousLocalTime {
                    orig: input.into(),
                    t1: t1.naive_utc(),
                    t2: t2.naive_utc(),
                }),
            }
        };
    }

    Err(Error::DateTimeParseError(input.into()))
}

#[cfg(test)]
mod tests {
    use chrono::TimeZone;

    use super::*;

    #[test]
    fn parse_datetime_string() {
        std::env::set_var("TZ", "America/Mexico_City");

        assert_eq!(parse_time("2021-05-21 11:36").unwrap(), Utc.ymd(2021, 5, 21).and_hms(16, 36, 0));
        assert_eq!(parse_time("2021-05-21 11:36:12").unwrap(), Utc.ymd(2021, 5, 21).and_hms(16, 36, 12));

        assert_eq!(parse_time("2021-05-21T11:36").unwrap(), Utc.ymd(2021, 5, 21).and_hms(16, 36, 0));
        assert_eq!(parse_time("2021-05-21T11:36:12").unwrap(), Utc.ymd(2021, 5, 21).and_hms(16, 36, 12));
    }

    #[test]
    fn parse_date() {
        assert_eq!(parse_time("2021-05-21").unwrap(), Utc.ymd(2021, 5, 21).and_hms(5, 0, 0));
    }

    #[test]
    fn parse_hour() {
        std::env::set_var("TZ", "America/Mexico_City");

        let now = Utc::now();

        assert_eq!(parse_time("11:36").unwrap(), now.date().and_hms(16, 36, 0));
        assert_eq!(parse_time("11:36:35").unwrap(), now.date().and_hms(16, 36, 35));
    }

    #[test]
    fn parse_hour_with_timezone() {
        std::env::set_var("TZ", "America/Mexico_City");

        let now = Utc::now();

        assert_eq!(parse_time("11:36Z").unwrap(), now.date().and_hms(11, 36, 0));
        assert_eq!(parse_time("11:36:35z").unwrap(), now.date().and_hms(11, 36, 35));

        assert_eq!(parse_time("11:36-5:00").unwrap(), now.date().and_hms(16, 36, 0));
        assert_eq!(parse_time("11:36:35+5:00").unwrap(), now.date().and_hms(6, 36, 35));
    }

    #[test]
    fn parse_with_specified_timezone() {
        assert_eq!(parse_time("2021-05-21T11:36:12Z").unwrap(), Utc.ymd(2021, 5, 21).and_hms(11, 36, 12));
        assert_eq!(parse_time("2021-05-21T11:36:12-3:00").unwrap(), Utc.ymd(2021, 5, 21).and_hms(14, 36, 12));
        assert_eq!(parse_time("2021-05-21T11:36:12+3:00").unwrap(), Utc.ymd(2021, 5, 21).and_hms(8, 36, 12));
    }

    #[test]
    fn parse_human_time() {
        assert_eq!(parse_time("an hour ago").unwrap(), Utc.ymd(2021, 5, 21).and_hms(11, 36, 12));
        assert_eq!(parse_time("two hours ago").unwrap(), Utc.ymd(2021, 5, 21).and_hms(11, 36, 12));
        assert_eq!(parse_time("4 hours ago").unwrap(), Utc.ymd(2021, 5, 21).and_hms(11, 36, 12));

        assert_eq!(parse_time("a minute ago").unwrap(), Utc.ymd(2021, 5, 21).and_hms(11, 36, 12));
        assert_eq!(parse_time("two minutes ago").unwrap(), Utc.ymd(2021, 5, 21).and_hms(11, 36, 12));
        assert_eq!(parse_time("4 minutes ago").unwrap(), Utc.ymd(2021, 5, 21).and_hms(11, 36, 12));

        assert_eq!(parse_time("hace una hora").unwrap(), Utc.ymd(2021, 5, 21).and_hms(11, 36, 12));
        assert_eq!(parse_time("hace dos horas").unwrap(), Utc.ymd(2021, 5, 21).and_hms(11, 36, 12));
        assert_eq!(parse_time("hace cuatro horas").unwrap(), Utc.ymd(2021, 5, 21).and_hms(11, 36, 12));

        assert_eq!(parse_time("hace una minuto").unwrap(), Utc.ymd(2021, 5, 21).and_hms(11, 36, 12));
        assert_eq!(parse_time("hace dos minutos").unwrap(), Utc.ymd(2021, 5, 21).and_hms(11, 36, 12));
        assert_eq!(parse_time("hace cuatro minutos").unwrap(), Utc.ymd(2021, 5, 21).and_hms(11, 36, 12));
    }
}