You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hive.apache.org by mm...@apache.org on 2018/01/29 11:57:25 UTC
hive git commit: HIVE-18562: Vectorization: CHAR/VARCHAR conversion
in VectorDeserializeRow is broken (Matt McCline, reviewed by Teddy Choi)
Repository: hive
Updated Branches:
refs/heads/master a631e6e53 -> 5efd3660f
HIVE-18562: Vectorization: CHAR/VARCHAR conversion in VectorDeserializeRow is broken (Matt McCline, reviewed by Teddy Choi)
Project: http://git-wip-us.apache.org/repos/asf/hive/repo
Commit: http://git-wip-us.apache.org/repos/asf/hive/commit/5efd3660
Tree: http://git-wip-us.apache.org/repos/asf/hive/tree/5efd3660
Diff: http://git-wip-us.apache.org/repos/asf/hive/diff/5efd3660
Branch: refs/heads/master
Commit: 5efd3660f3107c6d6688893a19c84aa5a25fd620
Parents: a631e6e
Author: Matt McCline <mm...@hortonworks.com>
Authored: Mon Jan 29 05:57:14 2018 -0600
Committer: Matt McCline <mm...@hortonworks.com>
Committed: Mon Jan 29 05:57:14 2018 -0600
----------------------------------------------------------------------
.../test/resources/testconfiguration.properties | 1 +
.../ql/exec/vector/VectorDeserializeRow.java | 20 +-
.../clientpositive/vector_char_varchar_1.q | 30 ++
.../llap/vector_char_varchar_1.q.out | 296 +++++++++++++++++++
4 files changed, 345 insertions(+), 2 deletions(-)
----------------------------------------------------------------------
http://git-wip-us.apache.org/repos/asf/hive/blob/5efd3660/itests/src/test/resources/testconfiguration.properties
----------------------------------------------------------------------
diff --git a/itests/src/test/resources/testconfiguration.properties b/itests/src/test/resources/testconfiguration.properties
index 1ce3ba6..f7ed4eb 100644
--- a/itests/src/test/resources/testconfiguration.properties
+++ b/itests/src/test/resources/testconfiguration.properties
@@ -705,6 +705,7 @@ minillaplocal.query.files=\
union_remove_26.q,\
union_top_level.q,\
vector_auto_smb_mapjoin_14.q,\
+ vector_char_varchar_1.q,\
vector_complex_all.q,\
vector_decimal_2.q,\
vector_decimal_udf.q,\
http://git-wip-us.apache.org/repos/asf/hive/blob/5efd3660/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/VectorDeserializeRow.java
----------------------------------------------------------------------
diff --git a/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/VectorDeserializeRow.java b/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/VectorDeserializeRow.java
index 470e761..8ea625e 100644
--- a/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/VectorDeserializeRow.java
+++ b/ql/src/java/org/apache/hadoop/hive/ql/exec/vector/VectorDeserializeRow.java
@@ -179,6 +179,10 @@ public final class VectorDeserializeRow<T extends DeserializeRead> {
return dataTypePhysicalVariation;
}
+ public void setMaxLength(int maxLength) {
+ this.maxLength = maxLength;
+ }
+
public int getMaxLength() {
return maxLength;
}
@@ -339,13 +343,25 @@ public final class VectorDeserializeRow<T extends DeserializeRead> {
/*
* Initialize the conversion related arrays. Assumes initTopLevelField has already been called.
*/
- private void addTopLevelConversion(int logicalColumnIndex) {
+ private void addTopLevelConversion(int logicalColumnIndex, TypeInfo targetTypeInfo) {
final Field field = topLevelFields[logicalColumnIndex];
field.setIsConvert(true);
if (field.getIsPrimitive()) {
+ PrimitiveTypeInfo targetPrimitiveTypeInfo = (PrimitiveTypeInfo) targetTypeInfo;
+ switch (targetPrimitiveTypeInfo.getPrimitiveCategory()) {
+ case CHAR:
+ field.setMaxLength(((CharTypeInfo) targetPrimitiveTypeInfo).getLength());
+ break;
+ case VARCHAR:
+ field.setMaxLength(((VarcharTypeInfo) targetPrimitiveTypeInfo).getLength());
+ break;
+ default:
+ // No additional data type specific setting.
+ break;
+ }
field.setConversionWritable(
VectorizedBatchUtil.getPrimitiveWritable(field.getPrimitiveCategory()));
}
@@ -482,7 +498,7 @@ public final class VectorDeserializeRow<T extends DeserializeRead> {
initTopLevelField(i, i, sourceTypeInfo, dataTypePhysicalVariations[i]);
// UNDONE: No for List and Map; Yes for Struct and Union when field count different...
- addTopLevelConversion(i);
+ addTopLevelConversion(i, targetTypeInfo);
atLeastOneConvert = true;
}
http://git-wip-us.apache.org/repos/asf/hive/blob/5efd3660/ql/src/test/queries/clientpositive/vector_char_varchar_1.q
----------------------------------------------------------------------
diff --git a/ql/src/test/queries/clientpositive/vector_char_varchar_1.q b/ql/src/test/queries/clientpositive/vector_char_varchar_1.q
new file mode 100644
index 0000000..58ec7a7
--- /dev/null
+++ b/ql/src/test/queries/clientpositive/vector_char_varchar_1.q
@@ -0,0 +1,30 @@
+set hive.cli.print.header=true;
+set hive.mapred.mode=nonstrict;
+set hive.explain.user=false;
+set hive.vectorized.execution.enabled=true;
+
+create table varchar_table(vs varchar(50));
+insert into table varchar_table
+ values ("blue"), ("a long sentence"), ("more sunshine and rain"), ("exactly 10"), ("longer 10 "), ("tells the truth");
+alter table varchar_table change column vs vs varchar(10);
+
+
+explain vectorization detail
+create table varchar_ctas_1 as select length(vs),reverse(vs) from varchar_table;
+
+create table varchar_ctas_1 as select length(vs),reverse(vs) from varchar_table;
+
+select * from varchar_ctas_1 order by 2;
+
+
+create table char_table(vs char(50));
+insert into table char_table
+ values (" yellow"), ("some words and spaces "), ("less sunshine and rain"), ("exactly 10"), ("sunlight moon"), ("begs the truth");
+alter table char_table change column vs vs char(10);
+
+explain vectorization detail
+create table char_ctas_1 as select length(vs),reverse(vs) from char_table;
+
+create table char_ctas_1 as select length(vs),reverse(vs) from char_table;
+
+select * from char_ctas_1 order by 2;
\ No newline at end of file
http://git-wip-us.apache.org/repos/asf/hive/blob/5efd3660/ql/src/test/results/clientpositive/llap/vector_char_varchar_1.q.out
----------------------------------------------------------------------
diff --git a/ql/src/test/results/clientpositive/llap/vector_char_varchar_1.q.out b/ql/src/test/results/clientpositive/llap/vector_char_varchar_1.q.out
new file mode 100644
index 0000000..83473ad
--- /dev/null
+++ b/ql/src/test/results/clientpositive/llap/vector_char_varchar_1.q.out
@@ -0,0 +1,296 @@
+PREHOOK: query: create table varchar_table(vs varchar(50))
+PREHOOK: type: CREATETABLE
+PREHOOK: Output: database:default
+PREHOOK: Output: default@varchar_table
+POSTHOOK: query: create table varchar_table(vs varchar(50))
+POSTHOOK: type: CREATETABLE
+POSTHOOK: Output: database:default
+POSTHOOK: Output: default@varchar_table
+PREHOOK: query: insert into table varchar_table
+ values ("blue"), ("a long sentence"), ("more sunshine and rain"), ("exactly 10"), ("longer 10 "), ("tells the truth")
+PREHOOK: type: QUERY
+PREHOOK: Input: _dummy_database@_dummy_table
+PREHOOK: Output: default@varchar_table
+POSTHOOK: query: insert into table varchar_table
+ values ("blue"), ("a long sentence"), ("more sunshine and rain"), ("exactly 10"), ("longer 10 "), ("tells the truth")
+POSTHOOK: type: QUERY
+POSTHOOK: Input: _dummy_database@_dummy_table
+POSTHOOK: Output: default@varchar_table
+POSTHOOK: Lineage: varchar_table.vs SCRIPT []
+_col0
+PREHOOK: query: alter table varchar_table change column vs vs varchar(10)
+PREHOOK: type: ALTERTABLE_RENAMECOL
+PREHOOK: Input: default@varchar_table
+PREHOOK: Output: default@varchar_table
+POSTHOOK: query: alter table varchar_table change column vs vs varchar(10)
+POSTHOOK: type: ALTERTABLE_RENAMECOL
+POSTHOOK: Input: default@varchar_table
+POSTHOOK: Output: default@varchar_table
+PREHOOK: query: explain vectorization detail
+create table varchar_ctas_1 as select length(vs),reverse(vs) from varchar_table
+PREHOOK: type: CREATETABLE_AS_SELECT
+POSTHOOK: query: explain vectorization detail
+create table varchar_ctas_1 as select length(vs),reverse(vs) from varchar_table
+POSTHOOK: type: CREATETABLE_AS_SELECT
+Explain
+PLAN VECTORIZATION:
+ enabled: true
+ enabledConditionsMet: [hive.vectorized.execution.enabled IS true]
+
+STAGE DEPENDENCIES:
+ Stage-1 is a root stage
+ Stage-2 depends on stages: Stage-1
+ Stage-4 depends on stages: Stage-2, Stage-0
+ Stage-3 depends on stages: Stage-4
+ Stage-0 depends on stages: Stage-1
+
+STAGE PLANS:
+ Stage: Stage-1
+ Tez
+#### A masked pattern was here ####
+ Vertices:
+ Map 1
+ Map Operator Tree:
+ TableScan
+ alias: varchar_table
+ Statistics: Num rows: 6 Data size: 564 Basic stats: COMPLETE Column stats: NONE
+ TableScan Vectorization:
+ native: true
+ vectorizationSchemaColumns: [0:vs:varchar(10), 1:ROW__ID:struct<transactionid:bigint,bucketid:int,rowid:bigint>]
+ Select Operator
+ expressions: length(vs) (type: int), reverse(vs) (type: string)
+ outputColumnNames: _col0, _col1
+ Select Vectorization:
+ className: VectorSelectOperator
+ native: true
+ projectedOutputColumnNums: [2, 3]
+ selectExpressions: StringLength(col 0:varchar(10)) -> 2:int, VectorUDFAdaptor(reverse(vs)) -> 3:string
+ Statistics: Num rows: 6 Data size: 564 Basic stats: COMPLETE Column stats: NONE
+ File Output Operator
+ compressed: false
+ File Sink Vectorization:
+ className: VectorFileSinkOperator
+ native: false
+ Statistics: Num rows: 6 Data size: 564 Basic stats: COMPLETE Column stats: NONE
+ table:
+ input format: org.apache.hadoop.mapred.TextInputFormat
+ output format: org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat
+ serde: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe
+ name: default.varchar_ctas_1
+ Execution mode: vectorized, llap
+ LLAP IO: no inputs
+ Map Vectorization:
+ enabled: true
+ enabledConditionsMet: hive.vectorized.use.vector.serde.deserialize IS true
+ inputFormatFeatureSupport: [DECIMAL_64]
+ vectorizationSupportRemovedReasons: [DECIMAL_64 disabled because LLAP is enabled]
+ featureSupportInUse: []
+ inputFileFormats: org.apache.hadoop.mapred.TextInputFormat
+ allNative: false
+ usesVectorUDFAdaptor: true
+ vectorized: true
+ rowBatchContext:
+ dataColumnCount: 1
+ includeColumns: [0]
+ dataColumns: vs:varchar(10)
+ partitionColumnCount: 0
+ scratchColumnTypeNames: [bigint, string]
+
+ Stage: Stage-2
+ Dependency Collection
+
+ Stage: Stage-4
+ Create Table Operator:
+ Create Table
+ columns: _c0 int, _c1 string
+ input format: org.apache.hadoop.mapred.TextInputFormat
+ output format: org.apache.hadoop.hive.ql.io.IgnoreKeyTextOutputFormat
+ serde name: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe
+ name: default.varchar_ctas_1
+
+ Stage: Stage-3
+ Stats Work
+ Basic Stats Work:
+
+ Stage: Stage-0
+ Move Operator
+ files:
+ hdfs directory: true
+#### A masked pattern was here ####
+
+PREHOOK: query: create table varchar_ctas_1 as select length(vs),reverse(vs) from varchar_table
+PREHOOK: type: CREATETABLE_AS_SELECT
+PREHOOK: Input: default@varchar_table
+PREHOOK: Output: database:default
+PREHOOK: Output: default@varchar_ctas_1
+POSTHOOK: query: create table varchar_ctas_1 as select length(vs),reverse(vs) from varchar_table
+POSTHOOK: type: CREATETABLE_AS_SELECT
+POSTHOOK: Input: default@varchar_table
+POSTHOOK: Output: database:default
+POSTHOOK: Output: default@varchar_ctas_1
+POSTHOOK: Lineage: varchar_ctas_1._c0 EXPRESSION [(varchar_table)varchar_table.FieldSchema(name:vs, type:varchar(10), comment:null), ]
+POSTHOOK: Lineage: varchar_ctas_1._c1 EXPRESSION [(varchar_table)varchar_table.FieldSchema(name:vs, type:varchar(10), comment:null), ]
+_c0 _c1
+PREHOOK: query: select * from varchar_ctas_1 order by 2
+PREHOOK: type: QUERY
+PREHOOK: Input: default@varchar_ctas_1
+#### A masked pattern was here ####
+POSTHOOK: query: select * from varchar_ctas_1 order by 2
+POSTHOOK: type: QUERY
+POSTHOOK: Input: default@varchar_ctas_1
+#### A masked pattern was here ####
+varchar_ctas_1._c0 varchar_ctas_1._c1
+10 01 regnol
+10 eht sllet
+10 01 yltcaxe
+4 eulb
+10 hsnus erom
+10 nes gnol a
+PREHOOK: query: create table char_table(vs char(50))
+PREHOOK: type: CREATETABLE
+PREHOOK: Output: database:default
+PREHOOK: Output: default@char_table
+POSTHOOK: query: create table char_table(vs char(50))
+POSTHOOK: type: CREATETABLE
+POSTHOOK: Output: database:default
+POSTHOOK: Output: default@char_table
+PREHOOK: query: insert into table char_table
+ values (" yellow"), ("some words and spaces "), ("less sunshine and rain"), ("exactly 10"), ("sunlight moon"), ("begs the truth")
+PREHOOK: type: QUERY
+PREHOOK: Input: _dummy_database@_dummy_table
+PREHOOK: Output: default@char_table
+POSTHOOK: query: insert into table char_table
+ values (" yellow"), ("some words and spaces "), ("less sunshine and rain"), ("exactly 10"), ("sunlight moon"), ("begs the truth")
+POSTHOOK: type: QUERY
+POSTHOOK: Input: _dummy_database@_dummy_table
+POSTHOOK: Output: default@char_table
+POSTHOOK: Lineage: char_table.vs SCRIPT []
+_col0
+PREHOOK: query: alter table char_table change column vs vs char(10)
+PREHOOK: type: ALTERTABLE_RENAMECOL
+PREHOOK: Input: default@char_table
+PREHOOK: Output: default@char_table
+POSTHOOK: query: alter table char_table change column vs vs char(10)
+POSTHOOK: type: ALTERTABLE_RENAMECOL
+POSTHOOK: Input: default@char_table
+POSTHOOK: Output: default@char_table
+PREHOOK: query: explain vectorization detail
+create table char_ctas_1 as select length(vs),reverse(vs) from char_table
+PREHOOK: type: CREATETABLE_AS_SELECT
+POSTHOOK: query: explain vectorization detail
+create table char_ctas_1 as select length(vs),reverse(vs) from char_table
+POSTHOOK: type: CREATETABLE_AS_SELECT
+Explain
+PLAN VECTORIZATION:
+ enabled: true
+ enabledConditionsMet: [hive.vectorized.execution.enabled IS true]
+
+STAGE DEPENDENCIES:
+ Stage-1 is a root stage
+ Stage-2 depends on stages: Stage-1
+ Stage-4 depends on stages: Stage-2, Stage-0
+ Stage-3 depends on stages: Stage-4
+ Stage-0 depends on stages: Stage-1
+
+STAGE PLANS:
+ Stage: Stage-1
+ Tez
+#### A masked pattern was here ####
+ Vertices:
+ Map 1
+ Map Operator Tree:
+ TableScan
+ alias: char_table
+ Statistics: Num rows: 6 Data size: 564 Basic stats: COMPLETE Column stats: NONE
+ TableScan Vectorization:
+ native: true
+ vectorizationSchemaColumns: [0:vs:char(10), 1:ROW__ID:struct<transactionid:bigint,bucketid:int,rowid:bigint>]
+ Select Operator
+ expressions: length(vs) (type: int), reverse(vs) (type: string)
+ outputColumnNames: _col0, _col1
+ Select Vectorization:
+ className: VectorSelectOperator
+ native: true
+ projectedOutputColumnNums: [2, 3]
+ selectExpressions: StringLength(col 0:char(10)) -> 2:int, VectorUDFAdaptor(reverse(vs)) -> 3:string
+ Statistics: Num rows: 6 Data size: 564 Basic stats: COMPLETE Column stats: NONE
+ File Output Operator
+ compressed: false
+ File Sink Vectorization:
+ className: VectorFileSinkOperator
+ native: false
+ Statistics: Num rows: 6 Data size: 564 Basic stats: COMPLETE Column stats: NONE
+ table:
+ input format: org.apache.hadoop.mapred.TextInputFormat
+ output format: org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat
+ serde: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe
+ name: default.char_ctas_1
+ Execution mode: vectorized, llap
+ LLAP IO: no inputs
+ Map Vectorization:
+ enabled: true
+ enabledConditionsMet: hive.vectorized.use.vector.serde.deserialize IS true
+ inputFormatFeatureSupport: [DECIMAL_64]
+ vectorizationSupportRemovedReasons: [DECIMAL_64 disabled because LLAP is enabled]
+ featureSupportInUse: []
+ inputFileFormats: org.apache.hadoop.mapred.TextInputFormat
+ allNative: false
+ usesVectorUDFAdaptor: true
+ vectorized: true
+ rowBatchContext:
+ dataColumnCount: 1
+ includeColumns: [0]
+ dataColumns: vs:char(10)
+ partitionColumnCount: 0
+ scratchColumnTypeNames: [bigint, string]
+
+ Stage: Stage-2
+ Dependency Collection
+
+ Stage: Stage-4
+ Create Table Operator:
+ Create Table
+ columns: _c0 int, _c1 string
+ input format: org.apache.hadoop.mapred.TextInputFormat
+ output format: org.apache.hadoop.hive.ql.io.IgnoreKeyTextOutputFormat
+ serde name: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe
+ name: default.char_ctas_1
+
+ Stage: Stage-3
+ Stats Work
+ Basic Stats Work:
+
+ Stage: Stage-0
+ Move Operator
+ files:
+ hdfs directory: true
+#### A masked pattern was here ####
+
+PREHOOK: query: create table char_ctas_1 as select length(vs),reverse(vs) from char_table
+PREHOOK: type: CREATETABLE_AS_SELECT
+PREHOOK: Input: default@char_table
+PREHOOK: Output: database:default
+PREHOOK: Output: default@char_ctas_1
+POSTHOOK: query: create table char_ctas_1 as select length(vs),reverse(vs) from char_table
+POSTHOOK: type: CREATETABLE_AS_SELECT
+POSTHOOK: Input: default@char_table
+POSTHOOK: Output: database:default
+POSTHOOK: Output: default@char_ctas_1
+POSTHOOK: Lineage: char_ctas_1._c0 EXPRESSION [(char_table)char_table.FieldSchema(name:vs, type:char(10), comment:null), ]
+POSTHOOK: Lineage: char_ctas_1._c1 EXPRESSION [(char_table)char_table.FieldSchema(name:vs, type:char(10), comment:null), ]
+_c0 _c1
+PREHOOK: query: select * from char_ctas_1 order by 2
+PREHOOK: type: QUERY
+PREHOOK: Input: default@char_ctas_1
+#### A masked pattern was here ####
+POSTHOOK: query: select * from char_ctas_1 order by 2
+POSTHOOK: type: QUERY
+POSTHOOK: Input: default@char_ctas_1
+#### A masked pattern was here ####
+char_ctas_1._c0 char_ctas_1._c1
+10 01 yltcaxe
+10 hsnus ssel
+10 m thgilnus
+10 sdrow emos
+10 t eht sgeb
+8 wolley