Skip to content

Commit 4eaea65

Browse files
farhad-celofarhad-celo
authored
[Enhancement] Enabling dictification for physical filter (#71093)
Signed-off-by: Farhad Shahmohammadi <f.shahmohammadi@celonis.com>
1 parent 7c84895 commit 4eaea65

5 files changed

Lines changed: 160 additions & 113 deletions

File tree

‎fe/fe-core/src/main/java/com/starrocks/sql/optimizer/rule/tree/lowcardinality/DecodeCollector.java‎

Lines changed: 1 addition & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -639,12 +639,7 @@ public DecodeInfo visitPhysicalTopN(OptExpression optExpression, DecodeInfo cont
639639

640640
@Override
641641
public DecodeInfo visitPhysicalFilter(OptExpression optExpression, DecodeInfo context) {
642-
if (optExpression.getInputs().get(0).getOp() instanceof PhysicalOlapScanOperator) {
643-
// PhysicalFilter->PhysicalOlapScan is a special pattern, the Filter's predicate is extracted from OlapScan,
644-
// we should keep the DecodeInfo from it's input.
645-
return context.createOutputInfo();
646-
}
647-
return context.createDecodeInfo();
642+
return context.createOutputInfo();
648643
}
649644

650645
private boolean tryHandleJoinEqPredicate(DecodeInfo decodeInfo,

‎fe/fe-core/src/test/java/com/starrocks/sql/plan/LowCardinalityTest2.java‎

Lines changed: 51 additions & 31 deletions
Original file line numberDiff line numberDiff line change
@@ -1184,18 +1184,22 @@ public void testJoin() throws Exception {
11841184
" union select 1,2\n" +
11851185
" ) sys";
11861186
plan = getFragmentPlan(sql);
1187-
assertContains(plan, plan, " 20:AGGREGATE (update serialize)\n"
1188-
+ " | STREAMING\n"
1189-
+ " | group by: 32: cast, 33: cast\n"
1190-
+ " | \n"
1191-
+ " 0:UNION\n"
1192-
+ " | \n"
1193-
+ " |----19:EXCHANGE\n"
1194-
+ " | \n"
1195-
+ " 16:EXCHANGE");
1187+
assertContains(plan, plan, " 20:AGGREGATE (update serialize)\n" +
1188+
" | STREAMING\n" +
1189+
" | group by: 32: cast, 33: cast\n" +
1190+
" | \n" +
1191+
" 0:UNION\n" +
1192+
" | \n" +
1193+
" |----19:EXCHANGE\n" +
1194+
" | \n" +
1195+
" 16:Decode\n" +
1196+
" | <dict id 38> : <string id 25>\n" +
1197+
" | <dict id 39> : <string id 26>\n" +
1198+
" | \n" +
1199+
" 15:EXCHANGE");
11961200
assertContains(plan, plan, "Decode");
11971201
plan = getThriftPlan(sql);
1198-
assertNotContains(plan.split("\n")[1], "query_global_dicts");
1202+
assertContains(plan.split("\n")[1], "query_global_dicts");
11991203
}
12001204

12011205
@Test
@@ -1296,9 +1300,6 @@ public void testAnalytic() throws Exception {
12961300
"select row_number() over( partition by L_COMMENT order by L_PARTKEY) as rm from lineitem" +
12971301
") t where rm < 10";
12981302
plan = getCostExplain(sql);
1299-
assertContains(plan, " 4:Decode\n" +
1300-
" | <dict id 20> : <string id 16>");
1301-
13021303
assertContains(plan, " 3:ANALYTIC\n" +
13031304
" | functions: [, row_number[(); args: ; result: BIGINT; " +
13041305
"args nullable: false; result nullable: true], ]\n" +
@@ -1310,7 +1311,7 @@ public void testAnalytic() throws Exception {
13101311
sql = "select * from (select L_COMMENT,l_quantity, row_number() over " +
13111312
"(partition by L_COMMENT order by l_quantity desc) rn from lineitem )t where rn <= 10;";
13121313
plan = getCostExplain(sql);
1313-
assertContains(plan, " 4:Decode\n" +
1314+
assertContains(plan, " 5:Decode\n" +
13141315
" | <dict id 19> : <string id 16>");
13151316
assertContains(plan, " 3:ANALYTIC\n" +
13161317
" | functions: [, row_number[(); args: ; result: BIGINT; " +
@@ -1323,7 +1324,7 @@ public void testAnalytic() throws Exception {
13231324
sql = "select * from (select L_COMMENT,l_quantity, rank() over " +
13241325
"(partition by L_COMMENT order by l_quantity desc) rn from lineitem )t where rn <= 10;";
13251326
plan = getCostExplain(sql);
1326-
assertContains(plan, " 4:Decode\n" +
1327+
assertContains(plan, " 5:Decode\n" +
13271328
" | <dict id 19> : <string id 16>");
13281329
assertContains(plan, " 3:ANALYTIC\n" +
13291330
" | functions: [, rank[(); args: ; result: BIGINT; " +
@@ -1336,7 +1337,7 @@ public void testAnalytic() throws Exception {
13361337
sql = "select * from (select L_COMMENT,l_quantity, rank() over " +
13371338
"(partition by L_COMMENT, l_shipmode order by l_quantity desc) rn from lineitem )t where rn <= 10;";
13381339
plan = getCostExplain(sql);
1339-
assertContains(plan, " 4:Decode\n" +
1340+
assertContains(plan, " 6:Decode\n" +
13401341
" | <dict id 19> : <string id 16>");
13411342
assertContains(plan, " 3:ANALYTIC\n" +
13421343
" | functions: [, rank[(); args: ; result: BIGINT; " +
@@ -1524,7 +1525,7 @@ public void testSubqueryWithLimit() throws Exception {
15241525
String sql = "select t0.S_ADDRESS from (select S_ADDRESS, S_NATIONKEY from supplier_nullable limit 10) t0" +
15251526
" inner join supplier on t0.S_NATIONKEY = supplier.S_NATIONKEY;";
15261527
String plan = getVerboseExplain(sql);
1527-
assertContains(plan, " 2:Decode\n" +
1528+
assertContains(plan, " 7:Decode\n" +
15281529
" | <dict id 17> : <string id 3>\n");
15291530
}
15301531

@@ -1560,16 +1561,17 @@ public void testAssignWrongNullableProperty() throws Exception {
15601561
sql = "select S_ADDRESS, S_COMMENT from (select S_ADDRESS, " +
15611562
"S_COMMENT from supplier_nullable order by S_COMMENT limit 10) tb where S_ADDRESS = 'SS' order by S_ADDRESS ";
15621563
plan = getFragmentPlan(sql);
1563-
assertContains(plan, " 5:SORT\n" +
1564-
" | order by: <slot 3> 3: S_ADDRESS ASC\n" +
1564+
assertContains(plan, " 4:SORT\n" +
1565+
" | order by: <slot 9> 9: S_ADDRESS ASC\n" +
15651566
" | offset: 0\n" +
15661567
" | \n" +
1567-
" 4:SELECT\n" +
1568-
" | predicates: 3: S_ADDRESS = 'SS'\n" +
1569-
" | \n" +
1570-
" 3:Decode\n" +
1568+
" 3:SELECT\n" +
1569+
" | predicates: DictDecode(9: S_ADDRESS, [<place-holder> = 'SS'])");
1570+
assertContains(plan, " 6:Decode\n" +
15711571
" | <dict id 9> : <string id 3>\n" +
1572-
" | <dict id 10> : <string id 7>");
1572+
" | <dict id 10> : <string id 7>\n" +
1573+
" | \n" +
1574+
" 5:MERGING-EXCHANGE");
15731575
}
15741576

15751577
@Test
@@ -2167,7 +2169,7 @@ public void testShuffleJoinOn2() throws Exception {
21672169
"where l1 = 'BJ'";
21682170
String plan = getFragmentPlan(sql);
21692171
// TODO: rewrite physical operator
2170-
assertContains(plan, " 9:Decode\n" +
2172+
assertContains(plan, " 10:Decode\n" +
21712173
" | <dict id 22> : <string id 12>");
21722174
}
21732175

@@ -2351,19 +2353,19 @@ public void testWindowFunction() throws Exception {
23512353
"WHERE\n" +
23522354
" t.row_num = 1;";
23532355
String plan = getVerboseExplain(sql);
2354-
assertContains(plan, " 7:Decode\n" +
2355-
" | <dict id 12> : <string id 3>\n" +
2356-
" | <dict id 13> : <string id 9>\n" +
2357-
" | <dict id 14> : <string id 11>\n" +
2356+
assertContains(plan, " 7:SELECT\n" +
2357+
" | predicates: 10: row_number() = 1\n" +
23582358
" | cardinality: 1\n" +
23592359
" | \n" +
23602360
" 6:ANALYTIC\n" +
2361-
" | functions: [, row_number[(); args: ; result: BIGINT; " +
2362-
"args nullable: false; result nullable: true], ]\n" +
2361+
" | functions: [, row_number[(); args: ; result: BIGINT; args nullable: false; result nullable: true], ]\n" +
23632362
" | partition by: [12: S_ADDRESS, INT, false]\n" +
23642363
" | order by: [12: S_ADDRESS, INT, false] ASC\n" +
23652364
" | window: ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW\n" +
23662365
" | cardinality: 1");
2366+
assertContains(plan, " 9:Decode\n" +
2367+
" | <dict id 12> : <string id 3>\n" +
2368+
" | <dict id 13> : <string id 9>");
23672369
}
23682370

23692371
@Test
@@ -2849,4 +2851,22 @@ public void testCreateDecodeInfoDoesNotReturnSharedMutableSingleton() {
28492851
Assertions.assertTrue(d2.getInputStringColumns().isEmpty());
28502852
Assertions.assertTrue(d2.getDecodeStringColumns().isEmpty());
28512853
}
2854+
2855+
@Test
2856+
public void testPhysicalFilter() throws Exception {
2857+
String sql = """
2858+
SELECT *
2859+
FROM (
2860+
SELECT
2861+
MAX(C_USER) OVER (PARTITION BY C_DEPT) max_user
2862+
FROM
2863+
low_card_t1
2864+
) cte
2865+
WHERE max_user != 'abc'
2866+
""";
2867+
String plan = getVerboseExplain(sql);
2868+
assertContains(plan, " 3:SELECT\n" +
2869+
" | predicates: DictDecode(15: max(2: c_user), [<place-holder> != 'abc'])\n" +
2870+
" | cardinality: 1", plan);
2871+
}
28522872
}

0 commit comments

Comments
 (0)