GH-47081: [Release] Verify reproducible source build explicitly

dev/release/verify-release-candidate.sh

@@ -789,14 +789,6 @@ ensure_source_directory() {
     if [ ! -d "${ARROW_SOURCE_DIR}" ]; then
       pushd $ARROW_TMPDIR
       fetch_archive ${dist_name}
-      git clone https://github.com/${GITHUB_REPOSITORY}.git arrow
-      pushd arrow
-      dev/release/utils-create-release-tarball.sh ${VERSION} ${RC_NUMBER}
-      if ! cmp ${dist_name}.tar.gz ../${dist_name}.tar.gz; then
-        echo "Source archive isn't reproducible"
-        return 1
-      fi
-      popd
       tar xf ${dist_name}.tar.gz
       popd
     fi
@@ -845,6 +837,27 @@ test_source_distribution() {
 
   pushd $ARROW_SOURCE_DIR
 
+  if [ "${SOURCE_KIND}" = "tarball" ] && [ "${TEST_SOURCE_REPRODUCIBLE}" -gt 0 ]; then
+    pushd ..
+    git clone "https://github.com/${GITHUB_REPOSITORY}.git" arrow
+    pushd arrow
+    dev/release/utils-create-release-tarball.sh "${VERSION}" "${RC_NUMBER}"
+    tarball="apache-arrow-${VERSION}.tar.gz"
+    if ! cmp "${tarball}" "../${tarball}"; then
+      echo "Source archive isn't reproducible"
+      if ! tar --version | grep --quiet --fixed GNU && \
+          ! gtar --version | grep --quiet --fixed GNU; then
+        echo "We need GNU tar to verify reproducible build"
+      fi
+      if ! gzip --version | grep --quiet --fixed GNU; then
+        echo "We need GNU gzip to verify reproducible build"
+      fi
+      return 1
+    fi
+    popd
+    popd
+  fi
+
   if [ ${TEST_CSHARP} -gt 0 ]; then
     test_csharp
   fi
@@ -1033,6 +1046,7 @@ test_wheels() {
 : ${TEST_YUM:=${TEST_BINARIES}}
 
 # Source verification tasks
+: ${TEST_SOURCE_REPRODUCIBLE:=0}
 : ${TEST_CPP:=${TEST_SOURCE}}
 : ${TEST_CSHARP:=${TEST_SOURCE}}
 : ${TEST_GLIB:=${TEST_SOURCE}}

dev/tasks/verify-rc/github.linux.amd64.docker.yml

@@ -52,6 +52,9 @@ jobs:
             {% endif %}
             -e VERIFY_RC="{{ rc|default("") }}" \
             -e TEST_DEFAULT=0 \
+            {% if target == "cpp" %}
+            -e TEST_SOURCE_REPRODUCIBLE=1 \
+            {% endif %}
             -e TEST_{{ target|upper }}=1 \
             {{ distro }}-verify-rc
 

dev/tasks/verify-rc/github.macos.yml

@@ -59,6 +59,9 @@ jobs:
           if [ -x "${pkgconf}" ]; then
             echo "PKG_CONFIG=${pkgconf}" >> $GITHUB_ENV
           fi
+
+          # For reproducible source archive verification
+          brew install gzip
       {% endif %}
 
       - uses: actions/setup-java@v2
@@ -91,6 +94,18 @@ jobs:
           USE_CONDA: 1
         {% else %}
           GTest_SOURCE: SYSTEM
+          {% if target == "cpp" %}
+          TEST_SOURCE_REPRODUCIBLE: 1
+          {% endif %}
         {% endif %}
         run: |
-          arrow/dev/release/verify-release-candidate.sh {{ release|default("") }} {{ rc|default("") }}
+          version={{ release|default("") }}
+          rc={{ rc|default("") }}
+          if [ -n "${version}" ] && [ -n "${rc}" ]; then
+            args=("${version}" "${rc}")
+            GITHUB_REPOSITORY=apache/arrow
+          else
+            args=()
+            GITHUB_REPOSITORY={{ arrow.github_repo }}
+          fi
+          arrow/dev/release/verify-release-candidate.sh "${args[@]}"