From 27450699d7f2705a575c487d2aa001714f328f96 Mon Sep 17 00:00:00 2001
From: Pedro Eugenio Rocha Pedreira <pedroerp@meta.com>
Date: Thu, 29 Feb 2024 14:51:44 -0800
Subject: [PATCH] parse_datetime() should allow UCT|UCT|GMT|GMT0 as 'Z'

Summary:
The Joda library accepts UCT|GMT|UCT and GMT0 as input for the 'Z'
(capital case) identifier, even though capital Z means timezone offset. Adding
support for compatibility with Joda, which is used in Presto java.

Reviewed By: mbasmanova

Differential Revision: D54346046

fbshipit-source-id: 82af2fea7eee47bd0eaf854b16c693e3d43d059a
---
 velox/docs/functions/presto/datetime.rst         |  7 ++++++-
 velox/functions/lib/DateTimeFormatter.cpp        | 16 +++++++++++++++-
 .../prestosql/tests/DateTimeFunctionsTest.cpp    | 15 +++++++++++++++
 3 files changed, 36 insertions(+), 2 deletions(-)

diff --git a/velox/docs/functions/presto/datetime.rst b/velox/docs/functions/presto/datetime.rst
index be66ebb51e70..a46c9d2b1b87 100644
--- a/velox/docs/functions/presto/datetime.rst
+++ b/velox/docs/functions/presto/datetime.rst
@@ -227,7 +227,12 @@ The functions in this section leverage a native cpp implementation that follows
 a format string compatible with JodaTime’s `DateTimeFormat
 <http://joda-time.sourceforge.net/apidocs/org/joda/time/format/DateTimeFormat.html>`_
 pattern format. The symbols currently supported are ``y``, ``Y``, ``M`` , ``d``,
-``H``, ``m``, ``s``, ``S``, and ``Z``.
+``H``, ``m``, ``s``, ``S``, ``z`` and ``Z``.
+
+``z`` represents a timezone name (3-letter format), and ``Z`` a timezone offset
+specified using the format ``+00``, ``+00:00`` or ``+0000`` (or ``-``). ``Z``
+also accepts ``UTC``,  ``UCT``, ``GMT``, and ``GMT0`` as valid representations
+of GMT.
 
 .. function:: parse_datetime(string, format) -> timestamp with time zone
 
diff --git a/velox/functions/lib/DateTimeFormatter.cpp b/velox/functions/lib/DateTimeFormatter.cpp
index e732862ad306..e15b4228b4d8 100644
--- a/velox/functions/lib/DateTimeFormatter.cpp
+++ b/velox/functions/lib/DateTimeFormatter.cpp
@@ -410,11 +410,25 @@ int64_t parseTimezoneOffset(const char* cur, const char* end, Date& date) {
         return 3;
       }
     }
-    // Single 'Z' character maps to GMT
+    // Single 'Z' character maps to GMT.
     else if (*cur == 'Z') {
       date.timezoneId = 0;
       return 1;
     }
+    // "UTC", "UCT", "GMT" and "GMT0" are also acceptable by joda.
+    else if ((end - cur) >= 3) {
+      if (std::strncmp(cur, "UTC", 3) == 0 ||
+          std::strncmp(cur, "UCT", 3) == 0) {
+        date.timezoneId = 0;
+        return 3;
+      } else if (std::strncmp(cur, "GMT", 3) == 0) {
+        date.timezoneId = 0;
+        if ((end - cur) >= 4 && *(cur + 3) == '0') {
+          return 4;
+        }
+        return 3;
+      }
+    }
   }
   return -1;
 }
diff --git a/velox/functions/prestosql/tests/DateTimeFunctionsTest.cpp b/velox/functions/prestosql/tests/DateTimeFunctionsTest.cpp
index f8e0581c8728..7c0f6f97955a 100644
--- a/velox/functions/prestosql/tests/DateTimeFunctionsTest.cpp
+++ b/velox/functions/prestosql/tests/DateTimeFunctionsTest.cpp
@@ -2638,6 +2638,21 @@ TEST_F(DateTimeFunctionsTest, parseDatetime) {
   EXPECT_EQ(
       TimestampWithTimezone(-66600000, util::getTimeZoneID("+02:00")),
       parseDatetime("1969-12-31+07:30+02:00", "YYYY-MM-dd+HH:mmZZ"));
+
+  // Joda also lets 'Z' to be UTC|UCT|GMT|GMT0.
+  auto ts = TimestampWithTimezone(1708840800000, util::getTimeZoneID("GMT"));
+  EXPECT_EQ(
+      ts, parseDatetime("2024-02-25+06:00:99 GMT", "yyyy-MM-dd+HH:mm:99 ZZZ"));
+  EXPECT_EQ(
+      ts, parseDatetime("2024-02-25+06:00:99 GMT0", "yyyy-MM-dd+HH:mm:99 ZZZ"));
+  EXPECT_EQ(
+      ts, parseDatetime("2024-02-25+06:00:99 UTC", "yyyy-MM-dd+HH:mm:99 ZZZ"));
+  EXPECT_EQ(
+      ts, parseDatetime("2024-02-25+06:00:99 UTC", "yyyy-MM-dd+HH:mm:99 ZZZ"));
+
+  VELOX_ASSERT_THROW(
+      parseDatetime("2024-02-25+06:00:99 PST", "yyyy-MM-dd+HH:mm:99 ZZZ"),
+      "Invalid format: \"2024-02-25+06:00:99 PST\" is malformed at \"PST\"");
 }
 
 TEST_F(DateTimeFunctionsTest, formatDateTime) {