refactor: replace time pre-parsing with lazy parsing (#629)

Cargo.toml

@@ -4,7 +4,7 @@ members = [".", "crate/encstr"]
 [workspace.package]
 repository = "https://github.com/pamburus/hl"
 authors = ["Pavel Ivanov <[email protected]>"]
-version = "0.30.1-alpha.1"
+version = "0.30.1-alpha.2"
 edition = "2021"
 license = "MIT"
 

benches/parse-and-format.rs

@@ -23,7 +23,7 @@ fn benchmark(c: &mut Criterion) {
         for theme in ["universal", "classic"] {
             c.bench_function(format!("{}/{}", name, theme), |b| {
                 let settings = Settings::default();
-                let parser = Parser::new(ParserSettings::new(&settings.fields.predefined, empty(), false, None));
+                let parser = Parser::new(ParserSettings::new(&settings.fields.predefined, empty(), None));
                 let formatter = RecordFormatter::new(
                     Arc::new(Theme::embedded(theme).unwrap()),
                     DateTimeFormatter::new(

src/app.rs

@@ -664,7 +664,6 @@ impl App {
         Parser::new(ParserSettings::new(
             &self.options.fields.settings.predefined,
             &self.options.fields.settings.ignore,
-            self.options.filter.since.is_some() || self.options.filter.until.is_some() || self.options.follow,
             self.options.unix_ts_unit,
         ))
     }

src/index.rs

@@ -268,12 +268,7 @@ where
             buffer_size: settings.buffer_size.into(),
             max_message_size: settings.max_message_size.into(),
             dir,
-            parser: Parser::new(ParserSettings::new(
-                &settings.fields,
-                empty(),
-                false,
-                settings.unix_ts_unit,
-            )),
+            parser: Parser::new(ParserSettings::new(&settings.fields, empty(), settings.unix_ts_unit)),
             delimiter: settings.delimiter,
             allow_prefix: settings.allow_prefix,
             format: settings.format,

src/model.rs

@@ -493,7 +493,6 @@ impl RecordFilter for RecordFilterNone {
 // ---
 
 pub struct ParserSettings {
-    pre_parse_time: bool,
     unix_ts_unit: Option<UnixTimestampUnit>,
     level: Vec<(HashMap<String, Level>, Option<Level>)>,
     blocks: Vec<ParserSettingsBlock>,
@@ -504,11 +503,9 @@ impl ParserSettings {
     pub fn new<'a, I: IntoIterator<Item = &'a String>>(
         predefined: &PredefinedFields,
         ignore: I,
-        pre_parse_time: bool,
         unix_ts_unit: Option<UnixTimestampUnit>,
     ) -> Self {
         let mut result = Self {
-            pre_parse_time,
             unix_ts_unit,
             level: Vec::new(),
             blocks: vec![ParserSettingsBlock::default()],
@@ -618,7 +615,7 @@ impl ParserSettings {
 impl Default for ParserSettings {
     #[inline]
     fn default() -> Self {
-        Self::new(&PredefinedFields::default(), Vec::new(), false, None)
+        Self::new(&PredefinedFields::default(), Vec::new(), None)
     }
 }
 
@@ -743,11 +740,7 @@ impl FieldSettings {
                 let s = value.raw_str();
                 let s = if s.as_bytes()[0] == b'"' { &s[1..s.len() - 1] } else { s };
                 let ts = Timestamp::new(s).with_unix_unit(ps.unix_ts_unit);
-                if ps.pre_parse_time {
-                    to.ts = Some(ts.preparsed())
-                } else {
-                    to.ts = Some(ts);
-                }
+                to.ts = Some(ts);
                 true
             }
             Self::Level(i) => {

src/timestamp.rs

@@ -1,3 +1,6 @@
+// stdlib imports
+use std::cell::OnceCell;
+
 // third-party imports
 use chrono::{DateTime, Duration, FixedOffset, NaiveDateTime};
 
@@ -9,15 +12,15 @@ use crate::app::UnixTimestampUnit;
 #[derive(Debug)]
 pub struct Timestamp<'a> {
     raw: &'a str,
-    parsed: Option<Option<DateTime<FixedOffset>>>,
+    parsed: OnceCell<Option<DateTime<FixedOffset>>>,
     unix_unit: Option<UnixTimestampUnit>,
 }
 
 impl<'a> Timestamp<'a> {
     pub fn new(value: &'a str) -> Self {
         Self {
             raw: value,
-            parsed: None,
+            parsed: OnceCell::new(),
             unix_unit: None,
         }
     }
@@ -26,27 +29,27 @@ impl<'a> Timestamp<'a> {
         self.raw
     }
 
-    pub fn preparsed(&self) -> Self {
+    pub fn with_unix_unit(self, unit: Option<UnixTimestampUnit>) -> Self {
         Self {
             raw: self.raw,
-            parsed: Some(self.parse()),
-            unix_unit: self.unix_unit,
+            parsed: if unit == self.unix_unit {
+                self.parsed
+            } else {
+                OnceCell::new()
+            },
+            unix_unit: unit,
         }
     }
 
-    pub fn with_unix_unit(&self, unit: Option<UnixTimestampUnit>) -> Self {
-        Self {
-            raw: self.raw,
-            parsed: if unit == self.unix_unit { self.parsed } else { None },
-            unix_unit: unit,
-        }
+    pub fn parsed(&self) -> &Option<DateTime<FixedOffset>> {
+        self.parsed.get_or_init(|| self.reparse())
     }
 
     pub fn parse(&self) -> Option<DateTime<FixedOffset>> {
-        if let Some(parsed) = self.parsed {
-            return parsed;
-        }
+        *self.parsed()
+    }
 
+    fn reparse(&self) -> Option<DateTime<FixedOffset>> {
         if let Ok(ts) = self.raw.parse() {
             Some(ts)
         } else if let Some(nt) = guess_number_type(self.raw.as_bytes()) {
@@ -101,7 +104,7 @@ impl<'a> Timestamp<'a> {
     }
 
     pub fn unix_utc(&self) -> Option<(i64, u32)> {
-        self.parse()
+        self.parsed()
             .and_then(|ts| Some((ts.timestamp(), ts.timestamp_subsec_nanos())))
     }
 }
@@ -490,12 +493,42 @@ mod tests {
 
     #[test]
     fn test_parse() {
-        let test = |s, unix_timestamp, tz| {
-            let ts = Timestamp::new(s).parse().unwrap();
+        let test = |s, unit, unix_timestamp, nanos, tz| {
+            let ts = Timestamp::new(s).with_unix_unit(unit).parse().unwrap();
             assert_eq!(ts.timestamp(), unix_timestamp);
             assert_eq!(ts.timezone().local_minus_utc(), tz);
+            assert_eq!(ts.timestamp_subsec_nanos(), nanos);
         };
-        test("2020-08-21 07:20:48", 1597994448, 0);
+        test("2020-08-21 07:20:48", None, 1597994448, 0, 0);
+        test("1597994448", None, 1597994448, 0, 0);
+        test("1597994448123", None, 1597994448, 123000000, 0);
+        test("1597994448123456", None, 1597994448, 123456000, 0);
+        test("1597994448123456789", None, 1597994448, 123456789, 0);
+        test("1597994448.123", None, 1597994448, 123000000, 0);
+        test("1597994448.123456", None, 1597994448, 123456000, 0);
+        test("1597994448.123456789", None, 1597994448, 123456789, 0);
+        test("-1.123456789", None, -2, 1000000000 - 123456789, 0);
+        test(
+            "1597994448123.456789",
+            Some(UnixTimestampUnit::Milliseconds),
+            1597994448,
+            123456789,
+            0,
+        );
+        test(
+            "1597994448123456.789",
+            Some(UnixTimestampUnit::Microseconds),
+            1597994448,
+            123456789,
+            0,
+        );
+        test(
+            "1597994448123456789.0",
+            Some(UnixTimestampUnit::Nanoseconds),
+            1597994448,
+            123456789,
+            0,
+        );
     }
 
     #[test]