This is an automated email from the ASF dual-hosted git repository.
kou pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/arrow.git
The following commit(s) were added to refs/heads/main by this push:
new d228ba5a9de GH-54647: [C++] Reject leap seconds in the strptime
timestamp parser (#54659)
d228ba5a9de is described below
commit d228ba5a9def26e67e12b2e475be0b00edb77aea
Author: Pearu Peterson <[email protected]>
AuthorDate: Mon Oct 26 03:38:00 2026 +0200
GH-54647: [C++] Reject leap seconds in the strptime timestamp parser
(#54659)
### Rationale for this change
`strptime` accepts a seconds value of 60 (glibc also 61), and the parser
added it to the time of day, which wrapped it into the next minute. Arrow
cannot represent leap seconds, and the ISO-8601 parser rejects them.
### What changes are included in this PR?
The parser rejects a seconds value of 60 or more. The check applies only
when the time of day is used, so Gandiva's `to_date`, which ignores the time,
still accepts such values.
### Are these changes tested?
Yes, in `TimestampParser.StrptimeParser`, and a Gandiva `to_date` test
checks that a leap second still gives the date when the time is ignored.
### Are there any user-facing changes?
Leap seconds no longer parse with `strptime` formats, except in Gandiva's
`to_date`.
### AI usage disclosure
Developed with AI assistance (Claude Code): the fix, the tests and this
description were AI-generated under my direction and reviewed line by line by
me. I own and can debug these changes. No copyrighted material is included.
* GitHub Issue: #54647
Authored-by: Pearu Peterson <[email protected]>
Signed-off-by: Sutou Kouhei <[email protected]>
---
cpp/src/arrow/util/value_parsing.h | 6 ++++++
cpp/src/arrow/util/value_parsing_test.cc | 25 ++++++++++++++-----------
cpp/src/gandiva/to_date_holder_test.cc | 9 +++++++++
3 files changed, 29 insertions(+), 11 deletions(-)
diff --git a/cpp/src/arrow/util/value_parsing.h
b/cpp/src/arrow/util/value_parsing.h
index 1c2102fe370..d845a71cf1f 100644
--- a/cpp/src/arrow/util/value_parsing.h
+++ b/cpp/src/arrow/util/value_parsing.h
@@ -848,6 +848,12 @@ static inline bool ParseTimestampStrptime(const char* buf,
size_t length,
// ignore the time part
arrow_vendored::date::sys_seconds secs = arrow_vendored::date::sys_days(ymd);
if (!ignore_time_in_day) {
+ // Reject leap seconds, which Arrow cannot represent, instead of adding
them
+ // to the time of day (which wraps them into the next minute); consistent
+ // with the ISO-8601 parser.
+ if (result.tm_sec >= 60) {
+ return false;
+ }
secs += (std::chrono::hours(result.tm_hour) +
std::chrono::minutes(result.tm_min) +
std::chrono::seconds(result.tm_sec));
#if !defined(_WIN32) && !defined(_AIX)
diff --git a/cpp/src/arrow/util/value_parsing_test.cc
b/cpp/src/arrow/util/value_parsing_test.cc
index 875418f3195..2a2d39777bf 100644
--- a/cpp/src/arrow/util/value_parsing_test.cc
+++ b/cpp/src/arrow/util/value_parsing_test.cc
@@ -870,17 +870,20 @@ TEST(TimestampParser, StrptimeParser) {
}
// Unparseable strings
- std::vector<std::string> unparseables = {"foo", "5/1/2000", "5/1/2000
12:34:56:6",
- // Impossible dates are rejected
rather
- // than rolled over into the next
month
- "2/30/2000 12:34:56", "4/31/2000
12:34:56",
- "2/29/2001 12:34:56",
- // A zero day is rejected rather
than
- // taken as the first of the month
- "2/00/2000 12:34:56",
- // A zero month is rejected rather
than
- // taken as December of the
previous year
- "0/15/2000 12:34:56"};
+ std::vector<std::string> unparseables = {
+ "foo", "5/1/2000", "5/1/2000 12:34:56:6",
+ // Impossible dates are rejected rather
+ // than rolled over into the next month
+ "2/30/2000 12:34:56", "4/31/2000 12:34:56", "2/29/2001 12:34:56",
+ // A zero day is rejected rather than
+ // taken as the first of the month
+ "2/00/2000 12:34:56",
+ // A zero month is rejected rather than
+ // taken as December of the previous year
+ "0/15/2000 12:34:56",
+ // Leap seconds are rejected, as by the
+ // ISO-8601 parser
+ "5/31/2000 12:30:60", "5/31/2000 23:59:60", "5/31/2000 23:59:61"};
for (auto& value : unparseables) {
int64_t dummy;
ASSERT_FALSE((*parser)(value.c_str(), value.size(), TimeUnit::SECOND,
&dummy));
diff --git a/cpp/src/gandiva/to_date_holder_test.cc
b/cpp/src/gandiva/to_date_holder_test.cc
index 1612d9b4f44..ada3e55d2fd 100644
--- a/cpp/src/gandiva/to_date_holder_test.cc
+++ b/cpp/src/gandiva/to_date_holder_test.cc
@@ -85,6 +85,15 @@ TEST_F(TestToDateHolder, TestSimpleDateTime) {
millis_since_epoch =
to_date(&execution_context_, s.data(), (int)s.length(), true,
&out_valid);
EXPECT_EQ(millis_since_epoch, 534643200000);
+
+ // A leap second does not make the value invalid, as the time is ignored
+ EXPECT_OK_AND_ASSIGN(auto to_date_holder24,
+ ToDateHolder::Make("YYYY-MM-DD HH24:MI:SS", 1));
+ s = std::string("2020-12-31 23:59:60");
+ millis_since_epoch = (*to_date_holder24)(&execution_context_, s.data(),
(int)s.length(),
+ true, &out_valid);
+ EXPECT_TRUE(out_valid);
+ EXPECT_EQ(millis_since_epoch, 1609372800000);
}
TEST_F(TestToDateHolder, TestSimpleDate) {
