GH-50674: Add support for recovering Yum repositories

.env

@@ -54,7 +54,7 @@ ALMALINUX=8
 ALPINE_LINUX=3.22
 DEBIAN=13
 FEDORA=42
-UBUNTU=22.04
+UBUNTU=24.04
 
 # Default versions for various dependencies
 CLANG_TOOLS=18

.github/workflows/verify_rc.yml

@@ -153,7 +153,6 @@ jobs:
         distro:
           - almalinux-8
           - conda
-          - ubuntu-22.04
           - ubuntu-24.04
     env:
       RC: ${{ needs.target.outputs.rc }}

compose.yaml

@@ -2084,7 +2084,7 @@ services:
     #   docker compose build ubuntu-verify-rc
     #   docker compose run -e VERIFY_VERSION=6.0.1 -e VERIFY_RC=1 ubuntu-verify-rc
     # Parameters:
-    #   UBUNTU: 22.04, 24.04
+    #   UBUNTU: 24.04
     image: ${REPO}:${ARCH_SHORT}-ubuntu-${UBUNTU}-verify-rc
     build:
       context: .

cpp/gdb_arrow.py

@@ -47,8 +47,11 @@
     'EXTENSION', 'FIXED_SIZE_LIST', 'DURATION', 'LARGE_STRING',
     'LARGE_BINARY', 'LARGE_LIST', 'INTERVAL_MONTH_DAY_NANO']
 
+_type_id_tuples = [(name, i) for i, name in enumerate(_type_ids)]
+_type_id_tuples.extend([('STRING_VIEW', 39), ('BINARY_VIEW', 40)])
+
 # Mirror the C++ Type::type enum
-Type = enum.IntEnum('Type', _type_ids, start=0)
+Type = enum.IntEnum('Type', _type_id_tuples)
 
 # Mirror the C++ TimeUnit::type enum
 TimeUnit = enum.IntEnum('TimeUnit', ['SECOND', 'MILLI', 'MICRO', 'NANO'],
@@ -1041,8 +1044,10 @@ def num_rows(self):
     'Decimal256Type': 'decimal256',
     'StringType': 'utf8',
     'LargeStringType': 'large_utf8',
+    'StringViewType': 'utf8_view',
     'BinaryType': 'binary',
     'LargeBinaryType': 'large_binary',
+    'BinaryViewType': 'binary_view',
     'FixedSizeBinaryType': 'fixed_size_binary',
     'ListType': 'list',
     'LargeListType': 'large_list',
@@ -2042,6 +2047,8 @@ class ExtensionTypeClass(DataTypeClass):
     Type.BINARY: DataTypeTraits(BaseBinaryTypeClass, 'BinaryType'),
     Type.LARGE_STRING: DataTypeTraits(BaseBinaryTypeClass, 'LargeStringType'),
     Type.LARGE_BINARY: DataTypeTraits(BaseBinaryTypeClass, 'LargeBinaryType'),
+    Type.STRING_VIEW: DataTypeTraits(BaseBinaryTypeClass, 'StringViewType'),
+    Type.BINARY_VIEW: DataTypeTraits(BaseBinaryTypeClass, 'BinaryViewType'),
 
     Type.FIXED_SIZE_BINARY: DataTypeTraits(FixedSizeBinaryTypeClass,
                                            'FixedSizeBinaryType'),

dev/release/binary-recover.sh

@@ -45,9 +45,13 @@ fi
 : "${RECOVER_DEFAULT:=1}"
 : "${RECOVER_DEBIAN:=${RECOVER_DEFAULT}}"
 : "${RECOVER_UBUNTU:=${RECOVER_DEFAULT}}"
+: "${RECOVER_ALMALINUX:=${RECOVER_DEFAULT}}"
+: "${RECOVER_AMAZON_LINUX:=${RECOVER_DEFAULT}}"
+: "${RECOVER_CENTOS:=${RECOVER_DEFAULT}}"
 
 rake_tasks=()
 apt_targets=()
+yum_targets=()
 if [ "${RECOVER_DEBIAN}" -gt 0 ]; then
   rake_tasks+=(apt:recover)
   apt_targets+=(debian)
@@ -56,6 +60,18 @@ if [ "${RECOVER_UBUNTU}" -gt 0 ]; then
   rake_tasks+=(apt:recover)
   apt_targets+=(ubuntu)
 fi
+if [ "${RECOVER_ALMALINUX}" -gt 0 ]; then
+  rake_tasks+=(yum:recover)
+  yum_targets+=(almalinux)
+fi
+if [ "${RECOVER_AMAZON_LINUX}" -gt 0 ]; then
+  rake_tasks+=(yum:recover)
+  yum_targets+=(amazon-linux)
+fi
+if [ "${RECOVER_CENTOS}" -gt 0 ]; then
+  rake_tasks+=(yum:recover)
+  yum_targets+=(centos)
+fi
 
 tmp_dir=binary/tmp
 mkdir -p "${tmp_dir}"
@@ -69,6 +85,10 @@ docker_run \
     IFS=,
     echo "${apt_targets[*]}"
   )" \
+  YUM_TARGETS="$(
+    IFS=,
+    echo "${yum_targets[*]}"
+  )" \
   ARTIFACTORY_API_KEY="${ARTIFACTORY_API_KEY}" \
   ARTIFACTS_DIR="${tmp_dir}/artifacts" \
   RC="" \

dev/release/binary-task.rb

@@ -825,7 +825,7 @@ def download
           if @pattern
             next unless @pattern.match?(path)
           end
-          if dynamic_paths
+          if dynamic_paths and path.include?("/repodata/")
             next unless dynamic_paths.include?(path)
           end
           output_path = "#{@destination}/#{path}"
@@ -1854,6 +1854,10 @@ def yum_release_repositories_dir
     "#{release_dir}/yum/repositories"
   end
 
+  def yum_recover_repositories_dir
+    "#{recover_dir}/yum/repositories"
+  end
+
   def available_yum_targets
     [
       ["almalinux", "10"],
@@ -2189,10 +2193,76 @@ def define_yum_release_tasks
     end
   end
 
+  def define_yum_recover_tasks
+    namespace :yum do
+      namespace :recover do
+        desc "Download repositories"
+        task :download do
+          yum_targets.each do |distribution, distribution_version|
+            not_checksum_pattern = /.+(?<!\.asc|\.sha512)\z/
+            target_dir =
+              "#{yum_recover_repositories_dir}/#{distribution}/#{distribution_version}"
+            download_distribution(:artifactory,
+                                  distribution,
+                                  target_dir,
+                                  :base,
+                                  pattern: not_checksum_pattern,
+                                  prefix: distribution_version)
+          end
+        end
+
+        desc "Update repositories"
+        task :update do
+          yum_update(nil, yum_recover_repositories_dir)
+          yum_targets.each do |distribution, distribution_version|
+            target_dir = [
+              yum_recover_repositories_dir,
+              distribution,
+              distribution_version,
+            ].join("/")
+            target_dir = Pathname(target_dir)
+            next unless target_dir.directory?
+            target_dir.glob("*") do |arch_dir|
+              next unless arch_dir.directory?
+              sign_label =
+                "#{distribution}-#{distribution_version} #{arch_dir.basename}"
+              sign_dir(sign_label,
+                       arch_dir.to_s)
+            end
+          end
+        end
+
+        desc "Upload repositories"
+        task :upload do
+          yum_distributions.each do |distribution|
+            distribution_dir =
+              "#{yum_recover_repositories_dir}/#{distribution}"
+            uploader = ArtifactoryUploader.new(api_key: artifactory_api_key,
+                                               distribution: distribution,
+                                               source: distribution_dir,
+                                               staging: staging?,
+                                               sync: false,
+                                               sync_pattern: /\/repodata\//)
+            uploader.upload
+          end
+        end
+      end
+
+      desc "Recover Yum repositories"
+      yum_recover_tasks = [
+        "yum:recover:download",
+        "yum:recover:update",
+        "yum:recover:upload",
+      ]
+      task :recover => yum_recover_tasks
+    end
+  end
+
   def define_yum_tasks
     define_yum_staging_tasks
     define_yum_rc_tasks
     define_yum_release_tasks
+    define_yum_recover_tasks
   end
 
   def define_summary_tasks

dev/tasks/python-wheels/github.linux.yml

@@ -104,12 +104,12 @@ jobs:
             -e TEST_WHEELS=1 \
             almalinux-verify-rc
 
-      - name: Test wheel on Ubuntu 22.04
+      - name: Test wheel on Ubuntu 24.04
         shell: bash
         if: |
           '{{ python_version }}' == '3.11' && '{{ linux_wheel_kind }}' == 'manylinux'
         env:
-          UBUNTU: "22.04"
+          UBUNTU: "24.04"
         run: |
           archery docker run \
             -e TEST_DEFAULT=0 \

dev/tasks/tasks.yml

@@ -272,7 +272,6 @@ tasks:
 
 {% for distribution, version in [("conda", "latest"),
                                  ("almalinux", "10"),
-                                 ("ubuntu", "22.04"),
                                  ("ubuntu", "24.04")] %}
   {% for target in ["cpp",
                     "integration",

python/pyarrow/src/arrow/python/gdb.cc

@@ -165,6 +165,10 @@ void TestSession() {
   StringType string_type;
   LargeBinaryType large_binary_type;
   LargeStringType large_string_type;
+  BinaryViewType binary_view_type;
+  StringViewType string_view_type;
+  auto heap_binary_view_type = binary_view();
+  auto heap_string_view_type = utf8_view();
   FixedSizeBinaryType fixed_size_binary_type(10);
   auto heap_fixed_size_binary_type = fixed_size_binary(10);
 
@@ -313,6 +317,10 @@ void TestSession() {
 
   LargeBinaryScalar large_binary_scalar_abc{Buffer::FromString("abc")};
   LargeStringScalar large_string_scalar_hehe{Buffer::FromString("héhé")};
+  BinaryViewScalar binary_view_scalar_null{binary_view()};
+  BinaryViewScalar binary_view_scalar_abc{Buffer::FromString("abc")};
+  StringViewScalar string_view_scalar_null{utf8_view()};
+  StringViewScalar string_view_scalar_hehe{Buffer::FromString("héhé")};
 
   FixedSizeBinaryScalar fixed_size_binary_scalar{Buffer::FromString("abc"),
                                                  fixed_size_binary(3)};

python/pyarrow/tests/test_gdb.py

@@ -368,6 +368,8 @@ def test_types_stack(gdb_arrow):
     check_stack_repr(gdb_arrow, "string_type", "arrow::utf8()")
     check_stack_repr(gdb_arrow, "large_binary_type", "arrow::large_binary()")
     check_stack_repr(gdb_arrow, "large_string_type", "arrow::large_utf8()")
+    check_stack_repr(gdb_arrow, "binary_view_type", "arrow::binary_view()")
+    check_stack_repr(gdb_arrow, "string_view_type", "arrow::utf8_view()")
     check_stack_repr(gdb_arrow, "fixed_size_binary_type",
                      "arrow::fixed_size_binary(10)")
 
@@ -427,6 +429,8 @@ def test_types_heap(gdb_arrow):
 
     check_heap_repr(gdb_arrow, "heap_decimal128_type",
                     "arrow::decimal128(16, 5)")
+    check_heap_repr(gdb_arrow, "heap_binary_view_type", "arrow::binary_view()")
+    check_heap_repr(gdb_arrow, "heap_string_view_type", "arrow::utf8_view()")
 
     check_heap_repr(gdb_arrow, "heap_list_type",
                     "arrow::list(arrow::uint8())")
@@ -625,6 +629,12 @@ def test_scalars_stack(gdb_arrow):
     check_stack_repr(
         gdb_arrow, "large_binary_scalar_abc",
         'arrow::LargeBinaryScalar of size 3, value "abc"')
+    check_stack_repr(
+        gdb_arrow, "binary_view_scalar_null",
+        "arrow::BinaryViewScalar of null value")
+    check_stack_repr(
+        gdb_arrow, "binary_view_scalar_abc",
+        'arrow::BinaryViewScalar of size 3, value "abc"')
 
     check_stack_repr(
         gdb_arrow, "string_scalar_null",
@@ -645,6 +655,12 @@ def test_scalars_stack(gdb_arrow):
     check_stack_repr(
         gdb_arrow, "large_string_scalar_hehe",
         'arrow::LargeStringScalar of size 6, value "héhé"')
+    check_stack_repr(
+        gdb_arrow, "string_view_scalar_null",
+        "arrow::StringViewScalar of null value")
+    check_stack_repr(
+        gdb_arrow, "string_view_scalar_hehe",
+        'arrow::StringViewScalar of size 6, value "héhé"')
 
     check_stack_repr(
         gdb_arrow, "fixed_size_binary_scalar",