From 8f2102665949c7b87bd49ea82854c1d3306ae0cc Mon Sep 17 00:00:00 2001
From: Adrien Guillo <adrien@quickwit.io>
Date: Fri, 29 Mar 2024 05:36:09 -0700
Subject: [PATCH] Handle missing year in date time parser (#4811)

---
 .../quickwit-datetime/src/date_time_format.rs | 48 ++++++++++++++++++-
 .../src/date_time_parsing.rs                  |  8 ++++
 2 files changed, 55 insertions(+), 1 deletion(-)

diff --git a/quickwit/quickwit-datetime/src/date_time_format.rs b/quickwit/quickwit-datetime/src/date_time_format.rs
index 2b04afa79e6..42b282ef6db 100644
--- a/quickwit/quickwit-datetime/src/date_time_format.rs
+++ b/quickwit/quickwit-datetime/src/date_time_format.rs
@@ -28,7 +28,7 @@ use time::error::Format;
 use time::format_description::well_known::{Iso8601, Rfc2822, Rfc3339};
 use time::format_description::FormatItem;
 use time::parsing::Parsed;
-use time::{OffsetDateTime, PrimitiveDateTime};
+use time::{Month, OffsetDateTime, PrimitiveDateTime};
 use time_fmt::parse::time_format_item::parse_to_format_item;
 
 use crate::TantivyDateTime;
@@ -84,6 +84,11 @@ impl StrptimeParser {
             parsed.set_minute(0u8);
             parsed.set_second(0u8);
         }
+        if parsed.year().is_none() {
+            let now = OffsetDateTime::now_utc();
+            let year = infer_year(parsed.month(), now.month(), now.year());
+            parsed.set_year(year);
+        }
         let date_time = parsed.try_into()?;
         Ok(date_time)
     }
@@ -318,9 +323,26 @@ impl<'de> Deserialize<'de> for DateTimeOutputFormat {
     }
 }
 
+/// Infers the year of a parsed date time. It assumes that events appear more often delayed than in
+/// the future and, as a result, skews towards the past year.
+pub(super) fn infer_year(
+    parsed_month_opt: Option<Month>,
+    this_month: Month,
+    this_year: i32,
+) -> i32 {
+    let Some(parsed_month) = parsed_month_opt else {
+        return this_year;
+    };
+    if parsed_month as u8 > this_month as u8 + 3 {
+        return this_year - 1;
+    }
+    this_year
+}
+
 #[cfg(test)]
 mod tests {
     use time::macros::datetime;
+    use time::Month;
 
     use super::*;
 
@@ -453,4 +475,28 @@ mod tests {
             "datetime string `2021-01-01TABC` does not match strptime format `%Y-%m-%d`"
         );
     }
+
+    #[test]
+    fn test_infer_year() {
+        let inferred_year = infer_year(None, Month::January, 2024);
+        assert_eq!(inferred_year, 2024);
+
+        let inferred_year = infer_year(Some(Month::December), Month::January, 2024);
+        assert_eq!(inferred_year, 2023);
+
+        let inferred_year = infer_year(Some(Month::January), Month::January, 2024);
+        assert_eq!(inferred_year, 2024);
+
+        let inferred_year = infer_year(Some(Month::February), Month::January, 2024);
+        assert_eq!(inferred_year, 2024);
+
+        let inferred_year = infer_year(Some(Month::March), Month::January, 2024);
+        assert_eq!(inferred_year, 2024);
+
+        let inferred_year = infer_year(Some(Month::April), Month::January, 2024);
+        assert_eq!(inferred_year, 2024);
+
+        let inferred_year = infer_year(Some(Month::May), Month::January, 2024);
+        assert_eq!(inferred_year, 2023);
+    }
 }
diff --git a/quickwit/quickwit-datetime/src/date_time_parsing.rs b/quickwit/quickwit-datetime/src/date_time_parsing.rs
index b048efcbaaf..34b9b62e6bb 100644
--- a/quickwit/quickwit-datetime/src/date_time_parsing.rs
+++ b/quickwit/quickwit-datetime/src/date_time_parsing.rs
@@ -182,8 +182,10 @@ mod tests {
     use std::str::FromStr;
 
     use time::macros::datetime;
+    use time::Month;
 
     use super::*;
+    use crate::date_time_format::infer_year;
     use crate::StrptimeParser;
 
     #[test]
@@ -247,6 +249,12 @@ mod tests {
                 "2024-01-31 18:40:19.950188123",
                 datetime!(2024-01-31 18:40:19.950188123 UTC),
             ),
+            ("%b %d %H:%M:%S", "Mar  6 17:40:02", {
+                let dt = datetime!(1900-03-06 17:40:02 UTC);
+                let now = OffsetDateTime::now_utc();
+                let year = infer_year(Some(Month::March), now.month(), now.year());
+                dt.replace_year(year).unwrap()
+            }),
         ];
         for (fmt, date_time_str, expected) in test_data {
             let parser = StrptimeParser::from_str(fmt).unwrap();