GH-17211: [C++] Add hash_64 scalar compute function (old)

.asf.yaml

@@ -19,13 +19,13 @@ github:
   description: "Apache Arrow is a multi-language toolbox for accelerated data interchange and in-memory processing"
   homepage: https://arrow.apache.org/
   collaborators:
+    - amoeba
     - anjakefala
     - benibus
     - danepitkin
     - davisusanibar
-    - felipecrv
     - js8544
-    - amoeba
+    - vibhatha
 
 notifications:
   commits:      commits@arrow.apache.org

.github/workflows/cpp.yml

@@ -101,7 +101,7 @@ jobs:
           fetch-depth: 0
           submodules: recursive
       - name: Cache Docker Volumes
-        uses: actions/cache@88522ab9f39a2ea568f7027eddc7d8d8bc9d59c8 # v3.3.1
+        uses: actions/cache@13aacd865c20de90d75de3b17ebe84f7a17d57d2 # v4.0.0
         with:
           path: .docker
           key: ${{ matrix.image }}-${{ hashFiles('cpp/**') }}
@@ -214,7 +214,7 @@ jobs:
         run: |
           echo "cache-dir=$(ccache --get-config cache_dir)" >> $GITHUB_OUTPUT
       - name: Cache ccache
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: ${{ steps.ccache-info.outputs.cache-dir }}
           key: cpp-ccache-macos-${{ hashFiles('cpp/**') }}
@@ -310,7 +310,7 @@ jobs:
         run: |
           echo "cache-dir=$(ccache --get-config cache_dir)" >> $GITHUB_OUTPUT
       - name: Cache ccache
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: ${{ steps.ccache-info.outputs.cache-dir }}
           key: cpp-ccache-windows-${{ env.CACHE_VERSION }}-${{ hashFiles('cpp/**') }}
@@ -402,7 +402,7 @@ jobs:
         shell: msys2 {0}
         run: ci/scripts/msys2_setup.sh cpp
       - name: Cache ccache
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: ccache
           key: cpp-ccache-${{ matrix.msystem_lower}}-${{ hashFiles('cpp/**') }}

.github/workflows/docs.yml

@@ -45,7 +45,7 @@ jobs:
         run: |
           ci/scripts/util_free_space.sh
       - name: Cache Docker Volumes
-        uses: actions/cache@88522ab9f39a2ea568f7027eddc7d8d8bc9d59c8 # v3.3.1
+        uses: actions/cache@13aacd865c20de90d75de3b17ebe84f7a17d57d2 # v4.0.0
         with:
           path: .docker
           key: ubuntu-docs-${{ hashFiles('cpp/**') }}

.github/workflows/docs_light.yml

@@ -51,7 +51,7 @@ jobs:
         with:
           fetch-depth: 0
       - name: Cache Docker Volumes
-        uses: actions/cache@88522ab9f39a2ea568f7027eddc7d8d8bc9d59c8 # v3.3.1
+        uses: actions/cache@13aacd865c20de90d75de3b17ebe84f7a17d57d2 # v4.0.0
         with:
           path: .docker
           key: conda-docs-${{ hashFiles('cpp/**') }}

.github/workflows/integration.yml

@@ -75,7 +75,7 @@ jobs:
         run: |
           ci/scripts/util_free_space.sh
       - name: Cache Docker Volumes
-        uses: actions/cache@88522ab9f39a2ea568f7027eddc7d8d8bc9d59c8 # v3.3.1
+        uses: actions/cache@13aacd865c20de90d75de3b17ebe84f7a17d57d2 # v4.0.0
         with:
           path: .docker
           key: conda-${{ hashFiles('cpp/**') }}

.github/workflows/java.yml

@@ -69,7 +69,7 @@ jobs:
           fetch-depth: 0
           submodules: recursive
       - name: Cache Docker Volumes
-        uses: actions/cache@88522ab9f39a2ea568f7027eddc7d8d8bc9d59c8 # v3.3.1
+        uses: actions/cache@13aacd865c20de90d75de3b17ebe84f7a17d57d2 # v4.0.0
         with:
           path: .docker
           key: maven-${{ hashFiles('java/**') }}

.github/workflows/java_jni.yml

@@ -63,7 +63,7 @@ jobs:
         run: |
           ci/scripts/util_free_space.sh
       - name: Cache Docker Volumes
-        uses: actions/cache@88522ab9f39a2ea568f7027eddc7d8d8bc9d59c8 # v3.3.1
+        uses: actions/cache@13aacd865c20de90d75de3b17ebe84f7a17d57d2 # v4.0.0
         with:
           path: .docker
           key: java-jni-manylinux-2014-${{ hashFiles('cpp/**', 'java/**') }}
@@ -103,7 +103,7 @@ jobs:
           fetch-depth: 0
           submodules: recursive
       - name: Cache Docker Volumes
-        uses: actions/cache@88522ab9f39a2ea568f7027eddc7d8d8bc9d59c8 # v3.3.1
+        uses: actions/cache@13aacd865c20de90d75de3b17ebe84f7a17d57d2 # v4.0.0
         with:
           path: .docker
           key: maven-${{ hashFiles('java/**') }}

.github/workflows/js.yml

@@ -91,7 +91,7 @@ jobs:
         with:
           fetch-depth: 0
       - name: Jest Cache
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: js/.jest-cache
           key: js-jest-cache-${{ runner.os }}-${{ hashFiles('js/src/**/*.ts', 'js/test/**/*.ts', 'js/yarn.lock') }}
@@ -121,7 +121,7 @@ jobs:
         with:
           fetch-depth: 0
       - name: Jest Cache
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: js/.jest-cache
           key: js-jest-cache-${{ runner.os }}-${{ hashFiles('js/src/**/*.ts', 'js/test/**/*.ts', 'js/yarn.lock') }}

.github/workflows/matlab.yml

@@ -65,7 +65,7 @@ jobs:
         shell: bash
         run: echo "cache-dir=$(ccache --get-config cache_dir)" >> $GITHUB_OUTPUT
       - name: Cache ccache
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: ${{ steps.ccache-info.outputs.cache-dir }}
           key: matlab-ccache-ubuntu-${{ hashFiles('cpp/**', 'matlab/**') }}
@@ -113,7 +113,7 @@ jobs:
         shell: bash
         run: echo "cache-dir=$(ccache --get-config cache_dir)" >> $GITHUB_OUTPUT
       - name: Cache ccache
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: ${{ steps.ccache-info.outputs.cache-dir }}
           key: matlab-ccache-macos-${{ hashFiles('cpp/**', 'matlab/**') }}
@@ -155,7 +155,7 @@ jobs:
         shell: bash
         run: echo "cache-dir=$(ccache --get-config cache_dir)" >> $GITHUB_OUTPUT
       - name: Cache ccache
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: |
             ${{ steps.ccache-info.outputs.cache-dir }}

.github/workflows/python.yml

@@ -94,7 +94,7 @@ jobs:
           fetch-depth: 0
           submodules: recursive
       - name: Cache Docker Volumes
-        uses: actions/cache@88522ab9f39a2ea568f7027eddc7d8d8bc9d59c8 # v3.3.1
+        uses: actions/cache@13aacd865c20de90d75de3b17ebe84f7a17d57d2 # v4.0.0
         with:
           path: .docker
           key: ${{ matrix.cache }}-${{ hashFiles('cpp/**') }}

.github/workflows/r.yml

@@ -73,7 +73,7 @@ jobs:
           fetch-depth: 0
           submodules: recursive
       - name: Cache Docker Volumes
-        uses: actions/cache@88522ab9f39a2ea568f7027eddc7d8d8bc9d59c8 # v3.3.1
+        uses: actions/cache@13aacd865c20de90d75de3b17ebe84f7a17d57d2 # v4.0.0
         with:
           path: .docker
           # As this key is identical on both matrix builds only one will be able to successfully cache,
@@ -206,7 +206,7 @@ jobs:
           ci/scripts/ccache_setup.sh
           echo "CCACHE_DIR=$(cygpath --absolute --windows ccache)" >> $GITHUB_ENV
       - name: Cache ccache
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: ccache
           key: r-${{ matrix.config.rtools }}-ccache-mingw-${{ matrix.config.arch }}-${{ hashFiles('cpp/src/**/*.cc','cpp/src/**/*.h)') }}-${{ github.run_id }}

.github/workflows/r_nightly.yml

@@ -86,7 +86,7 @@ jobs:
             exit 1
           fi
       - name: Cache Repo
-        uses: actions/cache@88522ab9f39a2ea568f7027eddc7d8d8bc9d59c8 # v3.3.1
+        uses: actions/cache@13aacd865c20de90d75de3b17ebe84f7a17d57d2 # v4.0.0
         with:
           path: repo
           key: r-nightly-${{ github.run_id }}

.github/workflows/ruby.yml

@@ -76,7 +76,7 @@ jobs:
           fetch-depth: 0
           submodules: recursive
       - name: Cache Docker Volumes
-        uses: actions/cache@88522ab9f39a2ea568f7027eddc7d8d8bc9d59c8 # v3.3.1
+        uses: actions/cache@13aacd865c20de90d75de3b17ebe84f7a17d57d2 # v4.0.0
         with:
           path: .docker
           key: ubuntu-${{ matrix.ubuntu }}-ruby-${{ hashFiles('cpp/**') }}
@@ -167,7 +167,7 @@ jobs:
         run: |
           echo "cache-dir=$(ccache --get-config cache_dir)" >> $GITHUB_OUTPUT
       - name: Cache ccache
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: ${{ steps.ccache-info.outputs.cache-dir }}
           key: ruby-ccache-macos-${{ hashFiles('cpp/**') }}
@@ -252,7 +252,7 @@ jobs:
         run: |
           ridk exec bash ci\scripts\msys2_setup.sh ruby
       - name: Cache ccache
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: ccache
           key: ruby-ccache-ucrt${{ matrix.mingw-n-bits }}-${{ hashFiles('cpp/**') }}
@@ -277,7 +277,7 @@ jobs:
           Write-Output "gem-dir=$(ridk exec gem env gemdir)" | `
             Out-File -FilePath $env:GITHUB_OUTPUT -Encoding utf8 -Append
       - name: Cache RubyGems
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: ${{ steps.rubygems-info.outputs.gem-dir }}
           key: ruby-rubygems-ucrt${{ matrix.mingw-n-bits }}-${{ hashFiles('**/Gemfile', 'ruby/*/*.gemspec') }}

c_glib/arrow-glib/basic-data-type.cpp

@@ -125,9 +125,9 @@ G_BEGIN_DECLS
  * data types.
  */
 
-typedef struct GArrowDataTypePrivate_ {
+struct GArrowDataTypePrivate {
   std::shared_ptr<arrow::DataType> data_type;
-} GArrowDataTypePrivate;
+};
 
 enum {
   PROP_DATA_TYPE = 1
@@ -1113,9 +1113,71 @@ garrow_date64_data_type_new(void)
 }
 
 
-G_DEFINE_TYPE(GArrowTimestampDataType,
-              garrow_timestamp_data_type,
-              GARROW_TYPE_TEMPORAL_DATA_TYPE)
+struct GArrowTimestampDataTypePrivate {
+  GTimeZone *time_zone;
+};
+
+enum {
+  PROP_TIME_ZONE = 1
+};
+
+G_DEFINE_TYPE_WITH_PRIVATE(GArrowTimestampDataType,
+                           garrow_timestamp_data_type,
+                           GARROW_TYPE_TEMPORAL_DATA_TYPE)
+
+#define GARROW_TIMESTAMP_DATA_TYPE_GET_PRIVATE(object)  \
+  static_cast<GArrowTimestampDataTypePrivate *>(        \
+    garrow_timestamp_data_type_get_instance_private(    \
+      GARROW_TIMESTAMP_DATA_TYPE(object)))
+
+static void
+garrow_timestamp_data_type_dispose(GObject *object)
+{
+  auto priv = GARROW_TIMESTAMP_DATA_TYPE_GET_PRIVATE(object);
+
+  if (priv->time_zone) {
+    g_time_zone_unref(priv->time_zone);
+    priv->time_zone = nullptr;
+  }
+
+  G_OBJECT_CLASS(garrow_timestamp_data_type_parent_class)->dispose(object);
+}
+
+static void
+garrow_timestamp_data_type_set_property(GObject *object,
+                                        guint prop_id,
+                                        const GValue *value,
+                                        GParamSpec *pspec)
+{
+  auto priv = GARROW_TIMESTAMP_DATA_TYPE_GET_PRIVATE(object);
+
+  switch (prop_id) {
+  case PROP_TIME_ZONE:
+    priv->time_zone = static_cast<GTimeZone *>(g_value_dup_boxed(value));
+    break;
+  default:
+    G_OBJECT_WARN_INVALID_PROPERTY_ID(object, prop_id, pspec);
+    break;
+  }
+}
+
+static void
+garrow_timestamp_data_type_get_property(GObject *object,
+                                        guint prop_id,
+                                        GValue *value,
+                                        GParamSpec *pspec)
+{
+  auto priv = GARROW_TIMESTAMP_DATA_TYPE_GET_PRIVATE(object);
+
+  switch (prop_id) {
+  case PROP_TIME_ZONE:
+    g_value_set_boxed(value, priv->time_zone);
+    break;
+  default:
+    G_OBJECT_WARN_INVALID_PROPERTY_ID(object, prop_id, pspec);
+    break;
+  }
+}
 
 static void
 garrow_timestamp_data_type_init(GArrowTimestampDataType *object)
@@ -1125,11 +1187,32 @@ garrow_timestamp_data_type_init(GArrowTimestampDataType *object)
 static void
 garrow_timestamp_data_type_class_init(GArrowTimestampDataTypeClass *klass)
 {
+  auto gobject_class = G_OBJECT_CLASS(klass);
+  gobject_class->dispose      = garrow_timestamp_data_type_dispose;
+  gobject_class->set_property = garrow_timestamp_data_type_set_property;
+  gobject_class->get_property = garrow_timestamp_data_type_get_property;
+
+  GParamSpec *spec;
+  /**
+   * GArrowTimestampDataType:time-zone:
+   *
+   * The time zone of this data type.
+   *
+   * Since: 16.0.0
+   */
+  spec = g_param_spec_boxed("time-zone",
+                            "Time zone",
+                            "The time zone of this data type",
+                            G_TYPE_TIME_ZONE,
+                            static_cast<GParamFlags>(G_PARAM_READWRITE |
+                                                     G_PARAM_CONSTRUCT_ONLY));
+  g_object_class_install_property(gobject_class, PROP_TIME_ZONE, spec);
 }
 
 /**
  * garrow_timestamp_data_type_new:
  * @unit: The unit of the timestamp data.
+ * @time_zone: (nullable): The time zone of the timestamp data.
  *
  * Returns: A newly created the number of
  *   seconds/milliseconds/microseconds/nanoseconds since UNIX epoch in
@@ -1138,30 +1221,36 @@ garrow_timestamp_data_type_class_init(GArrowTimestampDataTypeClass *klass)
  * Since: 0.7.0
  */
 GArrowTimestampDataType *
-garrow_timestamp_data_type_new(GArrowTimeUnit unit)
+garrow_timestamp_data_type_new(GArrowTimeUnit unit,
+                               GTimeZone *time_zone)
 {
   auto arrow_unit = garrow_time_unit_to_raw(unit);
-  auto arrow_data_type = arrow::timestamp(arrow_unit);
+  std::string arrow_timezone;
+  if (time_zone) {
+    arrow_timezone = g_time_zone_get_identifier(time_zone);
+  }
+  auto arrow_data_type = arrow::timestamp(arrow_unit, arrow_timezone);
   auto data_type =
     GARROW_TIMESTAMP_DATA_TYPE(g_object_new(GARROW_TYPE_TIMESTAMP_DATA_TYPE,
                                             "data-type", &arrow_data_type,
+                                            "time-zone", time_zone,
                                             NULL));
   return data_type;
 }
 
 /**
  * garrow_timestamp_data_type_get_unit:
- * @timestamp_data_type: The #GArrowTimestampDataType.
+ * @data_type: The #GArrowTimestampDataType.
  *
  * Returns: The unit of the timestamp data type.
  *
  * Since: 0.8.0
  */
 GArrowTimeUnit
-garrow_timestamp_data_type_get_unit(GArrowTimestampDataType *timestamp_data_type)
+garrow_timestamp_data_type_get_unit(GArrowTimestampDataType *data_type)
 {
   const auto arrow_data_type =
-    garrow_data_type_get_raw(GARROW_DATA_TYPE(timestamp_data_type));
+    garrow_data_type_get_raw(GARROW_DATA_TYPE(data_type));
   const auto arrow_timestamp_data_type =
     std::static_pointer_cast<arrow::TimestampType>(arrow_data_type);
   return garrow_time_unit_from_raw(arrow_timestamp_data_type->unit());

c_glib/arrow-glib/basic-data-type.h

@@ -425,9 +425,11 @@ struct _GArrowTimestampDataTypeClass
   GArrowTemporalDataTypeClass parent_class;
 };
 
-GArrowTimestampDataType *garrow_timestamp_data_type_new   (GArrowTimeUnit unit);
+GArrowTimestampDataType *
+garrow_timestamp_data_type_new(GArrowTimeUnit unit,
+                               GTimeZone *time_zone);
 GArrowTimeUnit
-garrow_timestamp_data_type_get_unit (GArrowTimestampDataType *timestamp_data_type);
+garrow_timestamp_data_type_get_unit(GArrowTimestampDataType *data_type);
 
 
 #define GARROW_TYPE_TIME_DATA_TYPE (garrow_time_data_type_get_type())