From f3af2cf6966b6033563d315492fc6b84f433f780 Mon Sep 17 00:00:00 2001
From: Daniel <kingdread@gmx.de>
Date: Fri, 1 May 2020 14:09:30 +0200
Subject: enable date parsing from filenames

This allows the date-based filters to work much faster.
---
 src/filters/log.rs | 107 +++++++++++++++++++++++++++++++++++++++++++++++------
 1 file changed, 95 insertions(+), 12 deletions(-)

(limited to 'src')

diff --git a/src/filters/log.rs b/src/filters/log.rs
index 60e6912..1728d4b 100644
--- a/src/filters/log.rs
+++ b/src/filters/log.rs
@@ -7,12 +7,17 @@ use super::{
     Filter, Inclusion,
 };
 
-use std::collections::HashSet;
+use std::{collections::HashSet, ffi::OsStr};
 
 use evtclib::Boss;
 
-use chrono::{DateTime, Datelike, Utc, Weekday};
+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>;
@@ -90,20 +95,53 @@ pub fn weekday(weekdays: HashSet<Weekday>) -> Box<dyn LogFilter> {
 struct TimeFilter(Option<DateTime<Utc>>, Option<DateTime<Utc>>);
 
 impl Filter<EarlyLogResult, LogResult> for TimeFilter {
+    fn filter_early(&self, early_log: &EarlyLogResult) -> Inclusion {
+        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 {
-        let after_ok = match self.0 {
-            Some(time) => time <= log.time,
-            None => true,
-        };
-        let before_ok = match self.1 {
-            Some(time) => time >= log.time,
-            None => true,
-        };
-
-        after_ok && before_ok
+        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
@@ -125,3 +163,48 @@ pub fn after(when: DateTime<Utc>) -> Box<dyn LogFilter> {
 pub fn before(when: DateTime<Utc>) -> Box<dyn LogFilter> {
     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)),
+        ));
+    }
+}
-- 
cgit v1.2.3