Fix format_datetime to used linked TimeZone names

.github/workflows/linux-build.yml

@@ -177,6 +177,7 @@ jobs:
       - name: Make Debug Build
         env:
           VELOX_DEPENDENCY_SOURCE: BUNDLED
+          ICU_SOURCE: SYSTEM
           MAKEFLAGS: "NUM_THREADS=8 MAX_HIGH_MEM_JOBS=4 MAX_LINK_JOBS=3"
           EXTRA_CMAKE_FLAGS: "-DVELOX_ENABLE_ARROW=ON -DVELOX_ENABLE_PARQUET=ON"
         run: |

CMake/resolve_dependency_modules/boost.cmake

@@ -13,30 +13,13 @@
 # limitations under the License.
 include_guard(GLOBAL)
 
-if(CMAKE_SYSTEM_NAME MATCHES "Darwin")
-  if(ON_APPLE_M1)
-    list(APPEND CMAKE_PREFIX_PATH "/opt/homebrew/opt/icu4c")
-  else()
-    list(APPEND CMAKE_PREFIX_PATH "/usr/local/opt/icu4c")
-  endif()
-endif()
-
-# ICU is only needed with Boost build from source
-set_source(ICU)
-resolve_dependency(
-  ICU
-  COMPONENTS
-  data
-  i18n
-  io
-  uc
-  tu
-  test)
-
 add_subdirectory(${CMAKE_CURRENT_LIST_DIR}/boost)
-if(${ICU_SOURCE} STREQUAL "BUNDLED")
-  # ensure ICU is built before Boost
-  add_dependencies(boost_regex ICU ICU::i18n)
+
+if(ICU_SOURCE)
+  if(${ICU_SOURCE} STREQUAL "BUNDLED")
+    # ensure ICU is built before Boost
+    add_dependencies(boost_regex ICU ICU::i18n)
+  endif()
 endif()
 
 # This prevents system boost from leaking in

CMakeLists.txt

@@ -385,6 +385,25 @@ endif()
 
 set(CMAKE_EXPORT_COMPILE_COMMANDS ON)
 
+if(CMAKE_SYSTEM_NAME MATCHES "Darwin")
+  if(ON_APPLE_M1)
+    list(APPEND CMAKE_PREFIX_PATH "/opt/homebrew/opt/icu4c")
+  else()
+    list(APPEND CMAKE_PREFIX_PATH "/usr/local/opt/icu4c")
+  endif()
+endif()
+
+set_source(ICU)
+resolve_dependency(
+  ICU
+  COMPONENTS
+  data
+  i18n
+  io
+  uc
+  tu
+  test)
+
 set(BOOST_INCLUDE_LIBRARIES
     atomic
     context

scripts/setup-ubuntu.sh

@@ -297,4 +297,3 @@ function install_apt_deps {
     fi
   fi
 )
-

velox/external/date/patches/0006-add_get_time_zone_names.patch

@@ -0,0 +1,30 @@
+diff --git a/velox/external/date/tz.cpp b/velox/external/date/tz.cpp
+--- a/velox/external/date/tz.cpp
++++ b/velox/external/date/tz.cpp
+@@ -3538,6 +3538,14 @@
+     return get_tzdb_list().front();
+ }
+
++std::vector<std::string> get_time_zone_names() {
++  std::vector<std::string> result;
++  for (const auto& z : get_tzdb().zones) {
++    result.push_back(z.name());
++  }
++  return result;
++}
++
+ const time_zone*
+ #if HAS_STRING_VIEW
+ tzdb::locate_zone(std::string_view tz_name) const
+diff --git a/velox/external/date/tz.h b/velox/external/date/tz.h
+--- a/velox/external/date/tz.h
++++ b/velox/external/date/tz.h
+@@ -1258,6 +1258,8 @@
+
+ DATE_API const tzdb& get_tzdb();
+
++std::vector<std::string> get_time_zone_names();
++
+ class tzdb_list
+ {
+     std::atomic<tzdb*> head_{nullptr};

velox/external/date/tz.cpp

@@ -3538,6 +3538,14 @@ get_tzdb()
     return get_tzdb_list().front();
 }
 
+std::vector<std::string> get_time_zone_names() {
+  std::vector<std::string> result;
+  for (const auto& z : get_tzdb().zones) {
+    result.push_back(z.name());
+  }
+  return result;
+}
+
 const time_zone*
 #if HAS_STRING_VIEW
 tzdb::locate_zone(std::string_view tz_name) const

velox/external/date/tz.h

@@ -1258,6 +1258,8 @@ operator<<(std::ostream& os, const tzdb& db);
 
 DATE_API const tzdb& get_tzdb();
 
+std::vector<std::string> get_time_zone_names();
+
 class tzdb_list
 {
     std::atomic<tzdb*> head_{nullptr};

velox/functions/lib/CMakeLists.txt

@@ -24,7 +24,8 @@ velox_link_libraries(velox_functions_util velox_vector velox_common_base)
 velox_add_library(velox_functions_lib_date_time_formatter DateTimeFormatter.cpp
                   DateTimeFormatterBuilder.cpp)
 
-velox_link_libraries(velox_functions_lib_date_time_formatter velox_type_tz)
+velox_link_libraries(velox_functions_lib_date_time_formatter velox_type_tz
+                     ICU::i18n ICU::uc)
 
 velox_add_library(
   velox_functions_lib