From 679b57b28c49d1bec9f4e57707eef48092677d07 Mon Sep 17 00:00:00 2001
From: Mihnea Iancu <m-iancu@users.noreply.github.com>
Date: Sun, 11 Feb 2018 15:37:38 +0200
Subject: [PATCH] [ENG-2684] Fix bug timezone parsing for timestamp literals

Summary:
Minor fix in the timestamp regex to avoid the bad parses.
Enhancing the timezone support will be done in a followup commit.

Test Plan: TestTimestampDataType

Reviewers: neil, robert

Reviewed By: robert

Subscribers: yql

Differential Revision: https://phabricator.dev.yugabyte.com/D4110
---
 .../src/test/java/org/yb/cql/TestTimestampDataType.java   | 4 ++++
 src/yb/util/date_time.cc                                  | 8 ++++++--
 2 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/java/yb-cql/src/test/java/org/yb/cql/TestTimestampDataType.java b/java/yb-cql/src/test/java/org/yb/cql/TestTimestampDataType.java
index 522e540debff..a37e7264e07a 100644
--- a/java/yb-cql/src/test/java/org/yb/cql/TestTimestampDataType.java
+++ b/java/yb-cql/src/test/java/org/yb/cql/TestTimestampDataType.java
@@ -118,6 +118,10 @@ public void testInvalidInsert() throws Exception {
     runInvalidInsert(tableName, "1992-13-12");
     runInvalidInsert(tableName, "1992-12-12 14:23:30:31");
     runInvalidInsert(tableName, "1992-12-12 14:23:30.12.32");
+    runInvalidInsert(tableName, "2017-12-21 00:00:01+0000");
+    runInvalidInsert(tableName, "2017-12-21 00:00:01.000+123:30");
+    runInvalidInsert(tableName, "2017-12-21 00:00:01.0000");
+    runInvalidInsert(tableName, "2017-12-21 00:00:01.000000+0000");
   }
 
   @Test
diff --git a/src/yb/util/date_time.cc b/src/yb/util/date_time.cc
index 136c9fe61e2e..a888fccc7c85 100644
--- a/src/yb/util/date_time.cc
+++ b/src/yb/util/date_time.cc
@@ -173,8 +173,12 @@ DateTimeInputFormat DateTime::CqlDateTimeInputFormat = []() -> DateTimeInputForm
   string time_fmt_no_sec = "(\\d{1,2}):(\\d{1,2})" + fmt_empty;
   string time_empty = fmt_empty + fmt_empty + fmt_empty;
   string frac_fmt = "\\.(\\d{1,3})";
-  string tzX_fmt = "((?:\\+|-).+)";
-  string tzZ_fmt = " (.+)";
+  // Offset, i.e. +/-xx:xx
+  string tzX_fmt = "((?:\\+|-)\\d{2}:\\d{2})";
+  // Timezone name, abbreviation, or offset (preceded by space), e.g. PDT, UDT+/-xx:xx, etc..
+  // At this point this allows anything that starts with a letter or '+' (after space), and leaves
+  // further processing to the timezone parser.
+  string tzZ_fmt = " ([a-zA-Z\\+].+)";
 
   // These cases match the valid Cassandra input formats
   vector<std::regex> regexes {