[#5071] Improvement (test): Add integration tests for Trino cascading queries

.github/workflows/trino-integration-test.yml

@@ -76,7 +76,7 @@ jobs:
 
       - name: Package Gravitino
         run: |
-          ./gradlew compileDistribution -x test -PjdkVersion=${{ matrix.java-version }}
+          ./gradlew compileDistribution compileTrinoConnector -x test -PjdkVersion=${{ matrix.java-version }}
 
       - name: Free up disk space
         run: |
@@ -87,6 +87,7 @@ jobs:
         run: |
           ./gradlew -PskipTests -PtestMode=embedded -PjdkVersion=${{ matrix.java-version }} -PskipDockerTests=false :trino-connector:integration-test:test
           ./gradlew -PskipTests -PtestMode=deploy -PjdkVersion=${{ matrix.java-version }} -PskipDockerTests=false :trino-connector:integration-test:test
+          trino-connector/integration-test/trino-test-tools/run_test.sh
 
       - name: Upload integrate tests reports
         uses: actions/upload-artifact@v3
@@ -95,10 +96,11 @@ jobs:
           name: trino-connector-integrate-test-reports-${{ matrix.java-version }}
           path: |
             build/reports
-            trino-connector/integrate-test/build/*.log
-            trino-connector/integrate-test/build/*.tar
+            trino-connector/integration-test/build/*.log
+            trino-connector/integration-test/build/*.tar
+            trino-connector/integration-test/build/trino-cascading-env
             integration-test-common/build/trino-ci-container-log
             distribution/package/logs/gravitino-server.out
             distribution/package/logs/gravitino-server.log
             catalogs/**/*.log
-            catalogs/**/*.tar
+            catalogs/**/*.tar

build.gradle.kts

@@ -488,11 +488,9 @@ tasks.rat {
     "dev/docker/kerberos-hive/kadm5.acl",
     "**/*.log",
     "**/*.out",
-    "**/testsets",
+    "**/trino-ci-testset",
     "**/licenses/*.txt",
     "**/licenses/*.md",
-    "integration-test/**/*.sql",
-    "integration-test/**/*.txt",
     "docs/**/*.md",
     "spark-connector/spark-common/src/test/resources/**",
     "web/web/.**",

integration-test-common/docker-script/shutdown.sh

@@ -21,7 +21,9 @@
 cd "$(dirname "$0")"
 
 LOG_DIR=../build/trino-ci-container-log
-docker cp trino-ci-hive:/usr/local/hadoop/logs $LOG_DIR/hdfs
-docker cp trino-ci-hive:/tmp/root $LOG_DIR/hive
+if [ -d $LOG_DIR ]; then
+  docker cp trino-ci-hive:/usr/local/hadoop/logs $LOG_DIR/hdfs
+  docker cp trino-ci-hive:/tmp/root $LOG_DIR/hive
+fi
 
 docker compose down

trino-connector/integration-test/build.gradle.kts

@@ -114,7 +114,7 @@ tasks.test {
 
 tasks.register<JavaExec>("TrinoTest") {
   classpath = sourceSets["test"].runtimeClasspath
-  mainClass.set("org.apache.gravitino.integration.test.trino.TrinoQueryTestTool")
+  mainClass.set("org.apache.gravitino.trino.connector.integration.test.TrinoQueryTestTool")
 
   if (JavaVersion.current() > JavaVersion.VERSION_1_8) {
     jvmArgs = listOf(

trino-connector/integration-test/src/test/java/org/apache/gravitino/trino/connector/integration/test/TrinoQueryIT.java

@@ -70,12 +70,6 @@ public class TrinoQueryIT extends TrinoQueryITBase {
   static {
     testsetsDir = TrinoQueryIT.class.getClassLoader().getResource("trino-ci-testset").getPath();
     testsetsDir = ITUtils.joinPath(testsetsDir, "testsets");
-
-    ciTestsets.add("hive");
-    ciTestsets.add("lakehouse-iceberg");
-    ciTestsets.add("jdbc-mysql");
-    ciTestsets.add("jdbc-postgresql");
-    ciTestsets.add("tpch");
   }
 
   @BeforeAll

trino-connector/integration-test/src/test/resources/trino-ci-testset/trino-cascading-testsets/tpcds/00000.sql

@@ -0,0 +1,72 @@
+select count(*) from call_center;
+select count(*) from catalog_page;
+select count(*) from catalog_returns;
+select count(*) from catalog_sales;
+select count(*) from customer;
+select count(*) from customer_address;
+select count(*) from customer_demographics;
+select count(*) from date_dim;
+select count(*) from household_demographics;
+select count(*) from income_band;
+select count(*) from inventory;
+select count(*) from item;
+select count(*) from promotion;
+select count(*) from reason;
+select count(*) from ship_mode;
+select count(*) from store;
+select count(*) from store_returns;
+select count(*) from store_sales;
+select count(*) from time_dim;
+select count(*) from warehouse;
+select count(*) from web_page;
+select count(*) from web_returns;
+select count(*) from web_sales;
+select count(*) from web_site;
+
+SELECT * FROM call_center ORDER BY cc_call_center_sk, cc_call_center_id, cc_rec_start_date LIMIT 10;
+
+SELECT * FROM catalog_page ORDER BY cp_catalog_page_sk, cp_catalog_page_id, cp_start_date_sk LIMIT 10;
+
+SELECT * FROM catalog_returns ORDER BY cr_returned_date_sk, cr_returned_time_sk, cr_item_sk LIMIT 10;
+
+SELECT * FROM catalog_sales ORDER BY cs_sold_date_sk, cs_sold_time_sk, cs_ship_date_sk LIMIT 10;
+
+SELECT * FROM customer ORDER BY c_customer_sk, c_customer_id, c_current_cdemo_sk LIMIT 10;
+
+SELECT * FROM customer_address ORDER BY ca_address_sk, ca_address_id, ca_street_number LIMIT 10;
+
+SELECT * FROM customer_demographics ORDER BY cd_demo_sk, cd_gender, cd_marital_status LIMIT 10;
+
+SELECT * FROM date_dim ORDER BY d_date_sk, d_date_id, d_date LIMIT 10;
+
+SELECT * FROM household_demographics ORDER BY hd_demo_sk, hd_income_band_sk, hd_buy_potential LIMIT 10;
+
+SELECT * FROM income_band ORDER BY ib_income_band_sk, ib_lower_bound, ib_upper_bound LIMIT 10;
+
+SELECT * FROM inventory ORDER BY inv_date_sk, inv_item_sk, inv_warehouse_sk LIMIT 10;
+
+SELECT * FROM item ORDER BY i_item_sk, i_item_id, i_rec_start_date LIMIT 10;
+
+SELECT * FROM promotion ORDER BY p_promo_sk, p_promo_id, p_start_date_sk LIMIT 10;
+
+SELECT * FROM reason ORDER BY r_reason_sk, r_reason_id, r_reason_desc LIMIT 10;
+
+SELECT * FROM ship_mode ORDER BY sm_ship_mode_sk, sm_ship_mode_id, sm_type LIMIT 10;
+
+SELECT * FROM store ORDER BY s_store_sk, s_store_id, s_rec_start_date LIMIT 10;
+
+SELECT * FROM store_returns ORDER BY sr_returned_date_sk, sr_return_time_sk, sr_item_sk LIMIT 10;
+
+SELECT * FROM store_sales ORDER BY ss_sold_date_sk, ss_sold_time_sk, ss_item_sk LIMIT 10;
+
+SELECT * FROM time_dim ORDER BY t_time_sk, t_time_id, t_time LIMIT 10;
+
+SELECT * FROM warehouse ORDER BY w_warehouse_sk, w_warehouse_id, w_warehouse_name LIMIT 10;
+
+SELECT * FROM web_page ORDER BY wp_web_page_sk, wp_web_page_id, wp_rec_start_date LIMIT 10;
+
+SELECT * FROM web_returns ORDER BY wr_returned_date_sk, wr_returned_time_sk, wr_item_sk LIMIT 10;
+
+SELECT * FROM web_sales ORDER BY ws_sold_date_sk, ws_sold_time_sk, ws_ship_date_sk LIMIT 10;
+
+SELECT * FROM web_site ORDER BY web_site_sk, web_site_id, web_rec_start_date LIMIT 10;