From ea68a88cd2a43fa492fff664a271925ff8a1c758 Mon Sep 17 00:00:00 2001
From: Christophe Petitjean <christophe.petitjean@gmail.com>
Date: Fri, 10 Mar 2023 14:42:02 +0100
Subject: [PATCH] Fixed issue with Time32/Time64 datatype in csv reader (#1425)

---
 src/io/csv/read_utils.rs    | 35 ++++++++++++++++++++++++++-
 src/temporal_conversions.rs |  6 ++---
 tests/it/io/csv/read.rs     | 48 +++++++++++++++++++++++++++++++++++++
 3 files changed, 85 insertions(+), 4 deletions(-)

diff --git a/src/io/csv/read_utils.rs b/src/io/csv/read_utils.rs
index 2212ac45acc..d23f6c1197f 100644
--- a/src/io/csv/read_utils.rs
+++ b/src/io/csv/read_utils.rs
@@ -1,4 +1,4 @@
-use chrono::Datelike;
+use chrono::{Datelike, Timelike};
 
 use crate::{
     array::*,
@@ -202,6 +202,29 @@ pub(crate) fn deserialize_column<B: ByteRecordGeneric>(
                 .and_then(|x| x.parse::<chrono::NaiveDateTime>().ok())
                 .map(|x| x.timestamp_millis())
         }),
+        Time32(time_unit) => deserialize_primitive(rows, column, datatype, |bytes| {
+            let factor = get_factor_from_timeunit(time_unit);
+            to_utf8(bytes)
+                .and_then(|x| x.parse::<chrono::NaiveTime>().ok())
+                .map(|x| {
+                    (x.hour() * 3_600 * factor
+                        + x.minute() * 60 * factor
+                        + x.second() * factor
+                        + x.nanosecond() / (1_000_000_000 / factor)) as i32
+                })
+        }),
+        Time64(time_unit) => deserialize_primitive(rows, column, datatype, |bytes| {
+            let factor: u64 = get_factor_from_timeunit(time_unit).into();
+            to_utf8(bytes)
+                .and_then(|x| x.parse::<chrono::NaiveTime>().ok())
+                .map(|x| {
+                    (x.hour() as u64 * 3_600 * factor
+                        + x.minute() as u64 * 60 * factor
+                        + x.second() as u64 * factor
+                        + x.nanosecond() as u64 / (1_000_000_000 / factor))
+                        as i64
+                })
+        }),
         Timestamp(time_unit, None) => deserialize_primitive(rows, column, datatype, |bytes| {
             to_utf8(bytes)
                 .and_then(|x| x.parse::<chrono::NaiveDateTime>().ok())
@@ -274,3 +297,13 @@ where
         .collect::<Result<Vec<_>>>()
         .and_then(Chunk::try_new)
 }
+
+// Return the factor by how small is a time unit compared to seconds
+fn get_factor_from_timeunit(time_unit: TimeUnit) -> u32 {
+    match time_unit {
+        TimeUnit::Second => 1,
+        TimeUnit::Millisecond => 1_000,
+        TimeUnit::Microsecond => 1_000_000,
+        TimeUnit::Nanosecond => 1_000_000_000,
+    }
+}
diff --git a/src/temporal_conversions.rs b/src/temporal_conversions.rs
index 67f896c1154..e26d3d8d1b3 100644
--- a/src/temporal_conversions.rs
+++ b/src/temporal_conversions.rs
@@ -60,7 +60,7 @@ pub fn date64_to_date(milliseconds: i64) -> NaiveDate {
     date64_to_datetime(milliseconds).date()
 }
 
-/// converts a `i32` representing a `time32(s)` to [`NaiveDateTime`]
+/// converts a `i32` representing a `time32(s)` to [`NaiveTime`]
 #[inline]
 pub fn time32s_to_time(v: i32) -> NaiveTime {
     NaiveTime::from_num_seconds_from_midnight_opt(v as u32, 0).expect("invalid time")
@@ -78,7 +78,7 @@ pub fn time32ms_to_time(v: i32) -> NaiveTime {
         .expect("invalid time")
 }
 
-/// converts a `i64` representing a `time64(us)` to [`NaiveDateTime`]
+/// converts a `i64` representing a `time64(us)` to [`NaiveTime`]
 #[inline]
 pub fn time64us_to_time(v: i64) -> NaiveTime {
     NaiveTime::from_num_seconds_from_midnight_opt(
@@ -91,7 +91,7 @@ pub fn time64us_to_time(v: i64) -> NaiveTime {
     .expect("invalid time")
 }
 
-/// converts a `i64` representing a `time64(ns)` to [`NaiveDateTime`]
+/// converts a `i64` representing a `time64(ns)` to [`NaiveTime`]
 #[inline]
 pub fn time64ns_to_time(v: i64) -> NaiveTime {
     NaiveTime::from_num_seconds_from_midnight_opt(
diff --git a/tests/it/io/csv/read.rs b/tests/it/io/csv/read.rs
index c1756a71cf8..ca9d56b1912 100644
--- a/tests/it/io/csv/read.rs
+++ b/tests/it/io/csv/read.rs
@@ -302,6 +302,54 @@ fn date64() -> Result<()> {
     Ok(())
 }
 
+#[test]
+fn time32_s() -> Result<()> {
+    let result = test_deserialize(
+        "00:00:00,\n23:59:59,\n11:00:11,\n",
+        DataType::Time32(TimeUnit::Second),
+    )?;
+    let expected = Int32Array::from(&[Some(0), Some(86399), Some(39611)])
+        .to(DataType::Time32(TimeUnit::Second));
+    assert_eq!(expected, result.as_ref());
+    Ok(())
+}
+
+#[test]
+fn time32_ms() -> Result<()> {
+    let result = test_deserialize(
+        "00:00:00.000,\n23:59:59.999,\n00:00:00.999,\n",
+        DataType::Time32(TimeUnit::Millisecond),
+    )?;
+    let expected = Int32Array::from(&[Some(0), Some(86_399_999), Some(999)])
+        .to(DataType::Time32(TimeUnit::Millisecond));
+    assert_eq!(expected, result.as_ref());
+    Ok(())
+}
+
+#[test]
+fn time64_us() -> Result<()> {
+    let result = test_deserialize(
+        "00:00:00.000000,\n23:59:59.999999,\n00:00:00.000001,\n",
+        DataType::Time64(TimeUnit::Microsecond),
+    )?;
+    let expected = Int64Array::from(&[Some(0), Some(86_399_999_999), Some(1)])
+        .to(DataType::Time64(TimeUnit::Microsecond));
+    assert_eq!(expected, result.as_ref());
+    Ok(())
+}
+
+#[test]
+fn time64_ns() -> Result<()> {
+    let result = test_deserialize(
+        "00:00:00.000000000,\n23:59:59.999999999,\n00:00:00.000000001,\n",
+        DataType::Time64(TimeUnit::Nanosecond),
+    )?;
+    let expected = Int64Array::from(&[Some(0), Some(86_399_999_999_999), Some(1)])
+        .to(DataType::Time64(TimeUnit::Nanosecond));
+    assert_eq!(expected, result.as_ref());
+    Ok(())
+}
+
 #[test]
 fn decimal() -> Result<()> {
     let result = test_deserialize("1.1,\n1.2,\n1.22,\n1.3,\n", DataType::Decimal(2, 1))?;