path: root/src/filters/log.rs

//! This module contains specific filters that operate on log files.
//!
//! This is the "base unit", as each file corresponds to one log. Filters on other items (such as
//! players) have to be lifted into log filters first.
use super::{
    super::{EarlyLogResult, FightOutcome, LogResult},
    Filter, Inclusion,
};

use std::{collections::HashSet, ffi::OsStr};

use evtclib::Boss;

use chrono::{DateTime, Datelike, Local, TimeZone, Utc, Weekday};
use num_traits::FromPrimitive as _;
use once_cell::sync::Lazy;
use regex::Regex;

/// The regular expression used to extract datetimes from filenames.
static DATE_REGEX: Lazy<Regex> = Lazy::new(|| Regex::new(r"\d{8}-\d{6}").unwrap());

/// Filter trait used for filters that operate on complete logs.
pub trait LogFilter = Filter<EarlyLogResult, LogResult>;

#[derive(Debug, Clone)]
struct BossFilter(HashSet<Boss>);

impl Filter<EarlyLogResult, LogResult> for BossFilter {
    fn filter_early(&self, early_log: &EarlyLogResult) -> Inclusion {
        let boss = Boss::from_u16(early_log.evtc.header.combat_id);
        boss.map(|b| self.0.contains(&b).into())
            .unwrap_or(Inclusion::Exclude)
    }

    fn filter(&self, log: &LogResult) -> bool {
        let boss = Boss::from_u16(log.boss_id);
        boss.map(|b| self.0.contains(&b)).unwrap_or(false)
    }
}

/// A `LogFilter` that only accepts logs with one of the given bosses.
pub fn boss(bosses: HashSet<Boss>) -> Box<dyn LogFilter> {
    Box::new(BossFilter(bosses))
}

#[derive(Debug, Clone)]
struct OutcomeFilter(HashSet<FightOutcome>);

impl Filter<EarlyLogResult, LogResult> for OutcomeFilter {
    fn filter(&self, log: &LogResult) -> bool {
        self.0.contains(&log.outcome)
    }
}

/// A `LogFilter` that only accepts logs with one of the given outcomes.
///
/// See also [`success`][success] and [`wipe`][wipe].
pub fn outcome(outcomes: HashSet<FightOutcome>) -> Box<dyn LogFilter> {
    Box::new(OutcomeFilter(outcomes))
}

/// A `LogFilter` that only accepts successful logs.
///
/// See also [`outcome`][outcome] and [`wipe`][wipe].
pub fn success() -> Box<dyn LogFilter> {
    let mut outcomes = HashSet::new();
    outcomes.insert(FightOutcome::Success);
    outcome(outcomes)
}

/// A `LogFilter` that only accepts failed logs.
///
/// See also [`outcome`][outcome] and [`success`][wipe].
pub fn wipe() -> Box<dyn LogFilter> {
    let mut outcomes = HashSet::new();
    outcomes.insert(FightOutcome::Wipe);
    outcome(outcomes)
}

#[derive(Debug, Clone)]
struct WeekdayFilter(HashSet<Weekday>);

impl Filter<EarlyLogResult, LogResult> for WeekdayFilter {
    fn filter(&self, log: &LogResult) -> bool {
        self.0.contains(&log.time.weekday())
    }
}

/// A `LogFilter` that only accepts logs if they were done on one of the given weekdays.
pub fn weekday(weekdays: HashSet<Weekday>) -> Box<dyn LogFilter> {
    Box::new(WeekdayFilter(weekdays))
}

#[derive(Debug, Clone)]
struct TimeFilter(Option<DateTime<Utc>>, Option<DateTime<Utc>>, bool);

impl Filter<EarlyLogResult, LogResult> for TimeFilter {
    fn filter_early(&self, early_log: &EarlyLogResult) -> Inclusion {
        // Ignore the filename heuristic if the user wishes so.
        if !self.2 {
            return Inclusion::Unknown;
        }
        early_log
            .log_file
            .file_name()
            .and_then(datetime_from_filename)
            .map(|d| time_is_between(d, self.0, self.1))
            .map(Into::into)
            .unwrap_or(Inclusion::Unknown)
    }

    fn filter(&self, log: &LogResult) -> bool {
        time_is_between(log.time, self.0, self.1)
    }
}

/// Check if the given time is after `after` but before `before`.
///
/// If one of the bounds is `None`, the time is always in bounds w.r.t. that bound.
fn time_is_between(
    time: DateTime<Utc>,
    after: Option<DateTime<Utc>>,
    before: Option<DateTime<Utc>>,
) -> bool {
    let after_ok = match after {
        Some(after) => after <= time,
        None => true,
    };
    let before_ok = match before {
        Some(before) => before >= time,
        None => true,
    };

    after_ok && before_ok
}

/// Try to extract the log time from the filename.
///
/// This expects the filename to have the datetime in the pattern `YYYYmmdd-HHMMSS` somewhere in
/// it.
fn datetime_from_filename(name: &OsStr) -> Option<DateTime<Utc>> {
    let date_match = DATE_REGEX.find(name.to_str()?)?;
    let local_time = Local
        .datetime_from_str(date_match.as_str(), "%Y%m%d-%H%M%S")
        .ok()?;
    Some(local_time.with_timezone(&Utc))
}

/// A `LogFilter` that only accepts logs in the given time frame.
///
/// If a bound is not given, -Infinity is assumed for the lower bound, and Infinity for the upper
/// bound.
pub fn time(lower: Option<DateTime<Utc>>, upper: Option<DateTime<Utc>>) -> Box<dyn LogFilter> {
    Box::new(TimeFilter(lower, upper, true))
}

/// A `LogFilter` that only accepts logs after the given date.
///
/// Also see [`time`][time] and [`before`][before].
pub fn after(when: DateTime<Utc>) -> Box<dyn LogFilter> {
    time(Some(when), None)
}

/// A `LogFilter` that only accepts logs before the given date.
///
/// Also see [`time`][time] and [`after`][after].
pub fn before(when: DateTime<Utc>) -> Box<dyn LogFilter> {
    time(None, Some(when))
}

/// A `LogFilter` that only accepts logs in the given time frame.
///
/// Compared to [`time`][time], this filter ignores the file name. This can result in more accurate
/// results if you renamed logs, but if also leads to a worse runtime.
pub fn log_time(lower: Option<DateTime<Utc>>, upper: Option<DateTime<Utc>>) -> Box<dyn LogFilter> {
    Box::new(TimeFilter(lower, upper, false))
}

/// Like [`after`][after], but ignores the file name for date calculations.
pub fn log_after(when: DateTime<Utc>) -> Box<dyn LogFilter> {
    log_time(Some(when), None)
}

/// Like [`before`][before], but ignores the file name for date calculations.
pub fn log_before(when: DateTime<Utc>) -> Box<dyn LogFilter> {
    log_time(None, Some(when))
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_time_is_between() {
        assert!(time_is_between(
            Utc.ymd(1955, 11, 5).and_hms(6, 15, 0),
            None,
            None,
        ));
        assert!(time_is_between(
            Utc.ymd(1955, 11, 5).and_hms(6, 15, 0),
            Some(Utc.ymd(1955, 11, 5).and_hms(5, 0, 0)),
            None,
        ));
        assert!(time_is_between(
            Utc.ymd(1955, 11, 5).and_hms(6, 15, 0),
            None,
            Some(Utc.ymd(1955, 11, 5).and_hms(7, 0, 0)),
        ));
        assert!(time_is_between(
            Utc.ymd(1955, 11, 5).and_hms(6, 15, 0),
            Some(Utc.ymd(1955, 11, 5).and_hms(5, 0, 0)),
            Some(Utc.ymd(1955, 11, 5).and_hms(7, 0, 0)),
        ));

        assert!(!time_is_between(
            Utc.ymd(1955, 11, 5).and_hms(6, 15, 0),
            Some(Utc.ymd(1955, 11, 5).and_hms(7, 0, 0)),
            None,
        ));
        assert!(!time_is_between(
            Utc.ymd(1955, 11, 5).and_hms(6, 15, 0),
            None,
            Some(Utc.ymd(1955, 11, 5).and_hms(5, 0, 0)),
        ));
        assert!(!time_is_between(
            Utc.ymd(1955, 11, 5).and_hms(6, 15, 0),
            Some(Utc.ymd(1955, 11, 5).and_hms(5, 0, 0)),
            Some(Utc.ymd(1955, 11, 5).and_hms(6, 0, 0)),
        ));
    }
}