Merge a1849184cb into c0c83236b6

docs/en/operations/server-configuration-parameters/settings.md

@@ -2567,6 +2567,18 @@ The time zone is necessary for conversions between String and DateTime formats w
 
 - [session_timezone](../settings/settings.md#session_timezone)
 
+## prefer_system_tzdata {#prefer_system_tzdata}
+
+If set to `true`, the system tzdata library will be preferred over the one embedded in ClickHouse.
+
+The default value is `false`.
+
+**Example**
+
+``` xml
+<prefer_system_tzdata>true</prefer_system_tzdata>
+```
+
 ## tcp_port {#tcp_port}
 
 Port for communicating with clients over the TCP protocol.

docs/ru/operations/server-configuration-parameters/settings.md

@@ -1597,6 +1597,18 @@ ClickHouse использует потоки из глобального пул
 
 - [session_timezone](../settings/settings.md#session_timezone)
 
+## prefer_system_tzdata {#prefer_system_tzdata}
+
+Если данный параметр имеет значение `true`, то в первую очередь будет использована системная библиотека tzdata, а не встроенная в ClickHouse.
+
+Значение по умолчанию: `false`.
+
+**Пример**
+
+``` xml
+<prefer_system_tzdata>true</prefer_system_tzdata>
+```
+
 ## tcp_port {#server_configuration_parameters-tcp_port}
 
 Порт для взаимодействия с клиентами по протоколу TCP.

programs/server/config.xml

@@ -816,6 +816,11 @@
     -->
     <!-- <timezone>UTC</timezone> -->
 
+    <!-- Allows to choose between ClickHouse built-in timezone data and system-wide tzdata (usually located at /usr/share/zoneinfo/)
+        By default, ClickHouse built-in tzdata is used.
+    -->
+    <!-- <prefer_system_tzdata>true</prefer_system_tzdata>-->
+
     <!-- You can specify umask here (see "man umask"). Server will apply it on startup.
          Number is always parsed as octal. Default umask is 027 (other users cannot read logs, data files, etc; group can only read).
     -->

src/Common/DateLUT.cpp

@@ -13,6 +13,10 @@
 #include <filesystem>
 #include <fstream>
 
+#include <cctz/zone_info_source.h>
+
+/// Embedded timezones.
+std::string_view getTimeZone(const char * name);
 
 namespace
 {
@@ -205,3 +209,81 @@ std::string DateLUT::extractTimezoneFromContext(DB::ContextPtr query_context)
 {
     return query_context->getSettingsRef().session_timezone.value;
 }
+
+/// By default prefer to load timezones from blobs linked to the binary.
+/// The blobs are provided by "tzdata" library.
+/// This allows to avoid dependency on system tzdata.
+namespace cctz_extension
+{
+    namespace
+    {
+        class Source : public cctz::ZoneInfoSource
+        {
+        public:
+            Source(const char * data_, size_t size_) : data(data_), size(size_) { }
+
+            size_t Read(void * buf, size_t bytes) override
+            {
+                bytes = std::min(bytes, size);
+                memcpy(buf, data, bytes);
+                data += bytes;
+                size -= bytes;
+                return bytes;
+            }
+
+            int Skip(size_t offset) override
+            {
+                if (offset <= size)
+                {
+                    data += offset;
+                    size -= offset;
+                    return 0;
+                }
+                else
+                {
+                    errno = EINVAL;
+                    return -1;
+                }
+            }
+
+        private:
+            const char * data;
+            size_t size;
+        };
+
+        std::unique_ptr<cctz::ZoneInfoSource>
+        custom_factory(const std::string & name, const std::function<std::unique_ptr<cctz::ZoneInfoSource>(const std::string & name)> & fallback)
+        {
+            std::string_view tz_file = getTimeZone(name.data());
+
+            if (!tz_file.empty())
+                return std::make_unique<Source>(tz_file.data(), tz_file.size());
+
+            return fallback(name);
+        }
+    }
+    cctz_extension::ZoneInfoSourceFactory zone_info_source_factory = custom_factory;
+}
+
+/// If `prefer_system_tzdata` is turned on in config, redefine tzdata lookup order:
+/// First, try to use system tzdata, then use built-in.
+void DateLUT::setPreferSystemTZData()
+{
+    cctz_extension::zone_info_source_factory = [] (
+                                                   const std::string & name,
+                                                   const std::function<std::unique_ptr<cctz::ZoneInfoSource>(const std::string & name)> & fallback
+                                                   ) -> std::unique_ptr<cctz::ZoneInfoSource>
+    {
+	    auto system_tz_source = fallback(name);
+        if (system_tz_source)
+            return system_tz_source;
+
+        std::string_view tz_file = ::getTimeZone(name.data());
+
+        if (!tz_file.empty())
+            return std::make_unique<cctz_extension::Source>(tz_file.data(), tz_file.size());
+
+        /// If not found in system AND in built-in, let fallback() handle this.
+        return system_tz_source;
+    };
+}

src/Common/DateLUT.h

@@ -55,6 +55,9 @@ public:
 protected:
     DateLUT();
 
+    friend class BaseDaemon;
+    static void setPreferSystemTZData();
+
 private:
     static DateLUT & getInstance();
 

src/Common/DateLUTImpl.cpp

@@ -21,10 +21,6 @@ namespace ErrorCodes
 }
 }
 
-/// Embedded timezones.
-std::string_view getTimeZone(const char * name);
-
-
 namespace
 {
 
@@ -236,60 +232,3 @@ unsigned int DateLUTImpl::toMillisecond(const DB::DateTime64 & datetime, Int64 s
     UInt16 millisecond = static_cast<UInt16>(fractional / divider);
     return millisecond;
 }
-
-
-/// Prefer to load timezones from blobs linked to the binary.
-/// The blobs are provided by "tzdata" library.
-/// This allows to avoid dependency on system tzdata.
-namespace cctz_extension
-{
-    namespace
-    {
-        class Source : public cctz::ZoneInfoSource
-        {
-        public:
-            Source(const char * data_, size_t size_) : data(data_), size(size_) {}
-
-            size_t Read(void * buf, size_t bytes) override
-            {
-                bytes = std::min(bytes, size);
-                memcpy(buf, data, bytes);
-                data += bytes;
-                size -= bytes;
-                return bytes;
-            }
-
-            int Skip(size_t offset) override
-            {
-                if (offset <= size)
-                {
-                    data += offset;
-                    size -= offset;
-                    return 0;
-                }
-                else
-                {
-                    errno = EINVAL;
-                    return -1;
-                }
-            }
-        private:
-            const char * data;
-            size_t size;
-        };
-
-        std::unique_ptr<cctz::ZoneInfoSource> custom_factory(
-            const std::string & name,
-            const std::function<std::unique_ptr<cctz::ZoneInfoSource>(const std::string & name)> & fallback)
-        {
-            std::string_view tz_file = getTimeZone(name.data());
-
-            if (!tz_file.empty())
-                return std::make_unique<Source>(tz_file.data(), tz_file.size());
-
-            return fallback(name);
-        }
-    }
-
-    ZoneInfoSourceFactory zone_info_source_factory = custom_factory;
-}

src/Daemon/BaseDaemon.cpp

@@ -280,7 +280,11 @@ void BaseDaemon::initialize(Application & self)
         }
     }
 
-    /// This must be done before any usage of DateLUT. In particular, before any logging.
+    /// `prefer_system_tzdata` and `timezone` parameters must be processed before any usage of DateLUT.
+    /// In particular, before any logging.
+    if (config().has("prefer_system_tzdata") && config().getBool("prefer_system_tzdata"))
+        DateLUT::setPreferSystemTZData();
+
     if (config().has("timezone"))
     {
         const std::string config_timezone = config().getString("timezone");

tests/integration/test_prefer_system_tzdata/config.d/prefer_system_tz.xml

@@ -0,0 +1,4 @@
+<clickhouse>
+     <prefer_system_tzdata>true</prefer_system_tzdata>
+     <timezone>UTC</timezone>
+</clickhouse>

tests/integration/test_prefer_system_tzdata/test.py

@@ -0,0 +1,31 @@
+import pytest
+from helpers.cluster import ClickHouseCluster
+
+cluster = ClickHouseCluster(__file__)
+
+node = cluster.add_instance("node", main_configs=["config.d/prefer_system_tz.xml",], stay_alive=True)
+
+
+@pytest.fixture(scope="module")
+def start_cluster():
+    try:
+        cluster.start()
+
+        yield cluster
+    finally:
+        cluster.shutdown()
+
+
+def test_prefer_system_tzdata(start_cluster):
+    node.exec_in_container(
+        [
+            "bash",
+            "-c",
+            f"rm /usr/share/zoneinfo/Africa/Tunis && ln -s /usr/share/zoneinfo/America/New_York /usr/share/zoneinfo/Africa/Tunis",
+        ],
+        privileged=True,
+    )
+
+    node.restart_clickhouse()
+
+    assert node.exec_in_container([f"bash", "-c", f"echo \"SELECT toDateTime(toDateTime('2024-05-01 12:12:12', 'UTC'), 'Africa/Tunis')\" | curl -s {node.hostname}:8123/ --data-binary @-"]) == "2024-05-01 08:12:12\n"

utils/check-style/aspell-ignore/en/aspell-dict.txt

@@ -2894,6 +2894,7 @@ tupleToNameValuePairs
 turbostat
 txt
 typename
+tzdata
 ubuntu
 uint
 ulid