[Fix](Serde) Support hive compatible output format #49036
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Contributor
|
Thank you for your contribution to Apache Doris. Please clearly describe your PR:
|
morningman
pushed a commit
that referenced
this pull request
Mar 20, 2025
There was a problem hiding this comment.
Pull Request Overview
This PR adds Hive support for the serde dialect, allowing users to obtain a Hive-compatible output for complex data types. Key changes include:
- Adding a new case for "hive" in the NereidsPlanner to set format options.
- Extending the allowed values and return mapping for serde dialect in SessionVariable.
Reviewed Changes
Copilot reviewed 2 out of 9 changed files in this pull request and generated no comments.
| File | Description |
|---|---|
| fe/fe-core/src/main/java/org/apache/doris/nereids/NereidsPlanner.java | Added hive case in switch block to set format options for hive. |
| fe/fe-core/src/main/java/org/apache/doris/qe/SessionVariable.java | Updated serde dialect validation to include hive and return enum. |
Files not reviewed (7)
- be/src/vec/data_types/serde/data_type_array_serde.cpp: Language not supported
- be/src/vec/data_types/serde/data_type_map_serde.cpp: Language not supported
- be/src/vec/data_types/serde/data_type_number_serde.cpp: Language not supported
- be/src/vec/data_types/serde/data_type_serde.h: Language not supported
- be/src/vec/data_types/serde/data_type_struct_serde.cpp: Language not supported
- be/src/vec/sink/vmysql_result_writer.cpp: Language not supported
- gensrc/thrift/PaloInternalService.thrift: Language not supported
Comments suppressed due to low confidence (2)
fe/fe-core/src/main/java/org/apache/doris/nereids/NereidsPlanner.java:804
- Ensure that using FormatOptions.getDefault() for the hive dialect produces Hive-compatible output as expected; if additional formatting adjustments are needed for Hive, consider using or creating a dedicated FormatOptions method.
case "hive":
fe/fe-core/src/main/java/org/apache/doris/qe/SessionVariable.java:4519
- The error message mistakenly refers to 'sqlDialect' instead of 'serdeDialect', which could confuse users; please update it for accuracy.
throw new UnsupportedOperationException("sqlDialect value is invalid, the invalid value is " + serdeDialect);
morningman
reviewed
Mar 24, 2025
BePPPower
force-pushed
the
supportHiveComplexFormat
branch
from
81cedc1 to
eb5ed02
Compare
github-actions
bot
added
the
approved
Contributor
|
PR approved by at least one committer and no changes requested. |
Contributor
|
PR approved by anyone and no changes requested. |
github-actions
bot
removed
the
approved
Contributor
Author
|
run buildall |
|
TeamCity cloud ut coverage result: |
TPC-H: Total hot run time: 34258 ms |
TPC-DS: Total hot run time: 186192 ms |
ClickBench: Total hot run time: 31.1 s |
Contributor
Author
|
run buildall |
|
TeamCity cloud ut coverage result: |
TPC-H: Total hot run time: 35167 ms |
TPC-DS: Total hot run time: 193735 ms |
ClickBench: Total hot run time: 30.73 s |
github-actions bot
pushed a commit
that referenced
this pull request
Apr 7, 2025
Problem Summary:
The output format of complex data types are different between Hive and
Doris, such as array, map and struct.
When user migrate from Hive to Doris, they expect the same format so
that they don't need to modify their business code.
This PR mainly changes:
Add a new option to session variable `serde_dialect`: If set to hive,
the output format returned to MySQL client of some datatypes will be
changed:
Array
Doris: ["abc", "def", "", null, 1]
Hive: ["abc","def","",null,true]
Map
Doris: {"k1":null, "k2":"v3"}
Hive: {"k1":null,"k2":"v3"}
Struct
Doris: {"s_id":100, "s_name":"abc , "", "s_address":null}
Hive: {"s_id":100,"s_name":"abc ,"","s_address":null}
Related #37039
github-actions bot
pushed a commit
that referenced
this pull request
Apr 7, 2025
Problem Summary:
The output format of complex data types are different between Hive and
Doris, such as array, map and struct.
When user migrate from Hive to Doris, they expect the same format so
that they don't need to modify their business code.
This PR mainly changes:
Add a new option to session variable `serde_dialect`: If set to hive,
the output format returned to MySQL client of some datatypes will be
changed:
Array
Doris: ["abc", "def", "", null, 1]
Hive: ["abc","def","",null,true]
Map
Doris: {"k1":null, "k2":"v3"}
Hive: {"k1":null,"k2":"v3"}
Struct
Doris: {"s_id":100, "s_name":"abc , "", "s_address":null}
Hive: {"s_id":100,"s_name":"abc ,"","s_address":null}
Related #37039
morningman
added a commit
that referenced
this pull request
Apr 10, 2025
morningman
added a commit
that referenced
this pull request
Apr 10, 2025
16 tasks
morningman
added a commit
that referenced
this pull request
Apr 10, 2025
yiguolei
pushed a commit
that referenced
this pull request
Apr 11, 2025
yiguolei
pushed a commit
that referenced
this pull request
Apr 11, 2025
morningman
added a commit
that referenced
this pull request
Apr 16, 2025
### What problem does this PR solve? 1. In #49036, we only support hive serde dialect in BE side. But some constant expr will be evaluated and output in FE side, need to support it too. 2. Refactor the method of getting string format value for all type of literals in FE side. There are 2 kind of string format value for literal. One is for Query, the other is for Stream Load. Here is some difference: - NullLiteral For query, it should be `null`. For load, it should be `\N`. - StructLiteral For query, it should be `{"k1":"v1", "k2":null, "k3":"", "k4":"a"}`. For load, it should be `{"v1", null, "", "a"}` So we need 2 different methods to distinguish them: `getStringValueForQuery` and `getStringValueForStreamLoad`. And I removed or renamed some old and messy methods. **Exmples** - `Doris/Hive/Presto` means when setting `serde_dialect` to these types, the format of query result for different column types. - `Stream Load ` means what format should be like in csv format when loading to the table | Type | Doris | Hive | Presto | Stream Load | Comment | | --- | --- | --- | --- | --- | --- | | Bool | `1`, `0` | `1`, `0` | `1`, `0` | `1|0`, `true|false` || | Integer | `1`, `1000` | `1`, `1000` | `1`, `1000` | `1|1000` | | | Float/Decimal | `1.2`, `3.00` | `1.2`, `3.00` | `1.2`, `3.00` | `1.2|3.00` | | | Date/Datetime | `2025-01-01`， `2025-01-01 10:11:11` | `2025-01-01`， `2025-01-01 10:11:11` | `2025-01-01`， `2025-01-01 10:11:11` | `2025-01-01|2025-01-01 10:11:11` | | | String | `abc`, `中国` | `abc`, `中国` | `abc`, `中国` | `abc,中国` | | | Null | `null` | `null` | `NULL` | `\N` || | Array<bool> | `[1, 0]` | `[true,false]` | `[1, 0]` | `[1, 0]`, `[true, false]` || | Array<int> | `[1, 1000]` | `[1,1000]` | `[1, 1000]` | `[1, 1000]` || | Array<string> | `["abc", "中国"]` | `["abc","中国"]` | `["abc", "中国"]` | `["abc", "中国"]` | | | Array<date/datetime> | `["2025-01-01", "2025-01-01 10:11:11"]` | `["2025-01-01","2025-01-01 10:11:11"]` | `["2025-01-01", "2025-01-01 10:11:11"]` | `["2025-01-01", "2025-01-01 10:11:11"]` || | Array<null> | `[null]` | `[null]` | `[NULL]` | `[null]` | | | Map<int, string> | `{1:"abc", 2:"中国"}` |`{1:"abc",2:"中国"}` |`{1=abc, 2=中国}` | `{1:"abc", 2:"中国"}` | | | Map<string, date/datetime> | `{"k1":"2022-10-01", "k2":"2022-10-01 10:10:10"}` | `{"k1":"2022-10-01","k2":"2022-10-01 10:10:10"}` | `{k1=2022-10-01, k2=2022-10-01 10:10:10}` | `{"k1":"2022-10-01", "k2":"2022-10-01 10:10:10"}` | | | Map<int, null> | `{1:null, 2:null}` | `{1:null,2:null}` | `{1=NULL, 2=NULL}` | `{1:null, 2:null}` | | | Struct<> | Same as map | Same as map | Same as map | Same as map | | 3. Fix a bug that for batch insert transaction, the `trim_double_quotas` should be set to false
seawinde
pushed a commit
to seawinde/doris
that referenced
this pull request
Apr 17, 2025
### What problem does this PR solve? 1. In apache#49036, we only support hive serde dialect in BE side. But some constant expr will be evaluated and output in FE side, need to support it too. 2. Refactor the method of getting string format value for all type of literals in FE side. There are 2 kind of string format value for literal. One is for Query, the other is for Stream Load. Here is some difference: - NullLiteral For query, it should be `null`. For load, it should be `\N`. - StructLiteral For query, it should be `{"k1":"v1", "k2":null, "k3":"", "k4":"a"}`. For load, it should be `{"v1", null, "", "a"}` So we need 2 different methods to distinguish them: `getStringValueForQuery` and `getStringValueForStreamLoad`. And I removed or renamed some old and messy methods. **Exmples** - `Doris/Hive/Presto` means when setting `serde_dialect` to these types, the format of query result for different column types. - `Stream Load ` means what format should be like in csv format when loading to the table | Type | Doris | Hive | Presto | Stream Load | Comment | | --- | --- | --- | --- | --- | --- | | Bool | `1`, `0` | `1`, `0` | `1`, `0` | `1|0`, `true|false` || | Integer | `1`, `1000` | `1`, `1000` | `1`, `1000` | `1|1000` | | | Float/Decimal | `1.2`, `3.00` | `1.2`, `3.00` | `1.2`, `3.00` | `1.2|3.00` | | | Date/Datetime | `2025-01-01`， `2025-01-01 10:11:11` | `2025-01-01`， `2025-01-01 10:11:11` | `2025-01-01`， `2025-01-01 10:11:11` | `2025-01-01|2025-01-01 10:11:11` | | | String | `abc`, `中国` | `abc`, `中国` | `abc`, `中国` | `abc,中国` | | | Null | `null` | `null` | `NULL` | `\N` || | Array<bool> | `[1, 0]` | `[true,false]` | `[1, 0]` | `[1, 0]`, `[true, false]` || | Array<int> | `[1, 1000]` | `[1,1000]` | `[1, 1000]` | `[1, 1000]` || | Array<string> | `["abc", "中国"]` | `["abc","中国"]` | `["abc", "中国"]` | `["abc", "中国"]` | | | Array<date/datetime> | `["2025-01-01", "2025-01-01 10:11:11"]` | `["2025-01-01","2025-01-01 10:11:11"]` | `["2025-01-01", "2025-01-01 10:11:11"]` | `["2025-01-01", "2025-01-01 10:11:11"]` || | Array<null> | `[null]` | `[null]` | `[NULL]` | `[null]` | | | Map<int, string> | `{1:"abc", 2:"中国"}` |`{1:"abc",2:"中国"}` |`{1=abc, 2=中国}` | `{1:"abc", 2:"中国"}` | | | Map<string, date/datetime> | `{"k1":"2022-10-01", "k2":"2022-10-01 10:10:10"}` | `{"k1":"2022-10-01","k2":"2022-10-01 10:10:10"}` | `{k1=2022-10-01, k2=2022-10-01 10:10:10}` | `{"k1":"2022-10-01", "k2":"2022-10-01 10:10:10"}` | | | Map<int, null> | `{1:null, 2:null}` | `{1:null,2:null}` | `{1=NULL, 2=NULL}` | `{1:null, 2:null}` | | | Struct<> | Same as map | Same as map | Same as map | Same as map | | 3. Fix a bug that for batch insert transaction, the `trim_double_quotas` should be set to false
BePPPower
added a commit
to BePPPower/doris
that referenced
this pull request
Apr 27, 2025
Problem Summary:
The output format of complex data types are different between Hive and
Doris, such as array, map and struct.
When user migrate from Hive to Doris, they expect the same format so
that they don't need to modify their business code.
This PR mainly changes:
Add a new option to session variable `serde_dialect`: If set to hive,
the output format returned to MySQL client of some datatypes will be
changed:
Array
Doris: ["abc", "def", "", null, 1]
Hive: ["abc","def","",null,true]
Map
Doris: {"k1":null, "k2":"v3"}
Hive: {"k1":null,"k2":"v3"}
Struct
Doris: {"s_id":100, "s_name":"abc , "", "s_address":null}
Hive: {"s_id":100,"s_name":"abc ,"","s_address":null}
Related apache#37039
dataroaring
pushed a commit
that referenced
this pull request
Apr 30, 2025
Problem Summary:
The output format of complex data types are different between Hive and
Doris, such as array, map and struct.
When user migrate from Hive to Doris, they expect the same format so
that they don't need to modify their business code.
This PR mainly changes:
Add a new option to session variable `serde_dialect`: If set to hive,
the output format returned to MySQL client of some datatypes will be
changed:
Array
Doris: ["abc", "def", "", null, 1]
Hive: ["abc","def","",null,true]
Map
Doris: {"k1":null, "k2":"v3"}
Hive: {"k1":null,"k2":"v3"}
Struct
Doris: {"s_id":100, "s_name":"abc , "", "s_address":null}
Hive: {"s_id":100,"s_name":"abc ,"","s_address":null}
Related #37039
dataroaring
pushed a commit
that referenced
this pull request
May 6, 2025
morningman
added a commit
to morningman/doris
that referenced
this pull request
May 14, 2025
1. In apache#49036, we only support hive serde dialect in BE side. But some constant expr will be evaluated and output in FE side, need to support it too. 2. Refactor the method of getting string format value for all type of literals in FE side. There are 2 kind of string format value for literal. One is for Query, the other is for Stream Load. Here is some difference: - NullLiteral For query, it should be `null`. For load, it should be `\N`. - StructLiteral For query, it should be `{"k1":"v1", "k2":null, "k3":"", "k4":"a"}`. For load, it should be `{"v1", null, "", "a"}` So we need 2 different methods to distinguish them: `getStringValueForQuery` and `getStringValueForStreamLoad`. And I removed or renamed some old and messy methods. **Exmples** - `Doris/Hive/Presto` means when setting `serde_dialect` to these types, the format of query result for different column types. - `Stream Load ` means what format should be like in csv format when loading to the table | Type | Doris | Hive | Presto | Stream Load | Comment | | --- | --- | --- | --- | --- | --- | | Bool | `1`, `0` | `1`, `0` | `1`, `0` | `1|0`, `true|false` || | Integer | `1`, `1000` | `1`, `1000` | `1`, `1000` | `1|1000` | | | Float/Decimal | `1.2`, `3.00` | `1.2`, `3.00` | `1.2`, `3.00` | `1.2|3.00` | | | Date/Datetime | `2025-01-01`， `2025-01-01 10:11:11` | `2025-01-01`， `2025-01-01 10:11:11` | `2025-01-01`， `2025-01-01 10:11:11` | `2025-01-01|2025-01-01 10:11:11` | | | String | `abc`, `中国` | `abc`, `中国` | `abc`, `中国` | `abc,中国` | | | Null | `null` | `null` | `NULL` | `\N` || | Array<bool> | `[1, 0]` | `[true,false]` | `[1, 0]` | `[1, 0]`, `[true, false]` || | Array<int> | `[1, 1000]` | `[1,1000]` | `[1, 1000]` | `[1, 1000]` || | Array<string> | `["abc", "中国"]` | `["abc","中国"]` | `["abc", "中国"]` | `["abc", "中国"]` | | | Array<date/datetime> | `["2025-01-01", "2025-01-01 10:11:11"]` | `["2025-01-01","2025-01-01 10:11:11"]` | `["2025-01-01", "2025-01-01 10:11:11"]` | `["2025-01-01", "2025-01-01 10:11:11"]` || | Array<null> | `[null]` | `[null]` | `[NULL]` | `[null]` | | | Map<int, string> | `{1:"abc", 2:"中国"}` |`{1:"abc",2:"中国"}` |`{1=abc, 2=中国}` | `{1:"abc", 2:"中国"}` | | | Map<string, date/datetime> | `{"k1":"2022-10-01", "k2":"2022-10-01 10:10:10"}` | `{"k1":"2022-10-01","k2":"2022-10-01 10:10:10"}` | `{k1=2022-10-01, k2=2022-10-01 10:10:10}` | `{"k1":"2022-10-01", "k2":"2022-10-01 10:10:10"}` | | | Map<int, null> | `{1:null, 2:null}` | `{1:null,2:null}` | `{1=NULL, 2=NULL}` | `{1:null, 2:null}` | | | Struct<> | Same as map | Same as map | Same as map | Same as map | | 3. Fix a bug that for batch insert transaction, the `trim_double_quotas` should be set to false
Merged
16 tasks
BiteTheDDDDt
pushed a commit
that referenced
this pull request
May 21, 2025
…and use _nesting_level. (#50977) We don't need to maintain a separate level; we can achieve the functionality of this #49036 by directly using _nesting_level. ```C++ // This parameter indicates what level the serde belongs to and is mainly used for complex types // The default level is 1, and each time you nest, the level increases by 1, // for example: struct<string> // The _nesting_level of StructSerde is 1 // The _nesting_level of StringSerde is 2 int _nesting_level = 1; ```
koarz
pushed a commit
to koarz/doris
that referenced
this pull request
Jun 4, 2025
Problem Summary:
The output format of complex data types are different between Hive and
Doris, such as array, map and struct.
When user migrate from Hive to Doris, they expect the same format so
that they don't need to modify their business code.
This PR mainly changes:
Add a new option to session variable `serde_dialect`: If set to hive,
the output format returned to MySQL client of some datatypes will be
changed:
Array
Doris: ["abc", "def", "", null, 1]
Hive: ["abc","def","",null,true]
Map
Doris: {"k1":null, "k2":"v3"}
Hive: {"k1":null,"k2":"v3"}
Struct
Doris: {"s_id":100, "s_name":"abc , "", "s_address":null}
Hive: {"s_id":100,"s_name":"abc ,"","s_address":null}
Related apache#37039
koarz
pushed a commit
to koarz/doris
that referenced
this pull request
Jun 4, 2025
### What problem does this PR solve? 1. In apache#49036, we only support hive serde dialect in BE side. But some constant expr will be evaluated and output in FE side, need to support it too. 2. Refactor the method of getting string format value for all type of literals in FE side. There are 2 kind of string format value for literal. One is for Query, the other is for Stream Load. Here is some difference: - NullLiteral For query, it should be `null`. For load, it should be `\N`. - StructLiteral For query, it should be `{"k1":"v1", "k2":null, "k3":"", "k4":"a"}`. For load, it should be `{"v1", null, "", "a"}` So we need 2 different methods to distinguish them: `getStringValueForQuery` and `getStringValueForStreamLoad`. And I removed or renamed some old and messy methods. **Exmples** - `Doris/Hive/Presto` means when setting `serde_dialect` to these types, the format of query result for different column types. - `Stream Load ` means what format should be like in csv format when loading to the table | Type | Doris | Hive | Presto | Stream Load | Comment | | --- | --- | --- | --- | --- | --- | | Bool | `1`, `0` | `1`, `0` | `1`, `0` | `1|0`, `true|false` || | Integer | `1`, `1000` | `1`, `1000` | `1`, `1000` | `1|1000` | | | Float/Decimal | `1.2`, `3.00` | `1.2`, `3.00` | `1.2`, `3.00` | `1.2|3.00` | | | Date/Datetime | `2025-01-01`， `2025-01-01 10:11:11` | `2025-01-01`， `2025-01-01 10:11:11` | `2025-01-01`， `2025-01-01 10:11:11` | `2025-01-01|2025-01-01 10:11:11` | | | String | `abc`, `中国` | `abc`, `中国` | `abc`, `中国` | `abc,中国` | | | Null | `null` | `null` | `NULL` | `\N` || | Array<bool> | `[1, 0]` | `[true,false]` | `[1, 0]` | `[1, 0]`, `[true, false]` || | Array<int> | `[1, 1000]` | `[1,1000]` | `[1, 1000]` | `[1, 1000]` || | Array<string> | `["abc", "中国"]` | `["abc","中国"]` | `["abc", "中国"]` | `["abc", "中国"]` | | | Array<date/datetime> | `["2025-01-01", "2025-01-01 10:11:11"]` | `["2025-01-01","2025-01-01 10:11:11"]` | `["2025-01-01", "2025-01-01 10:11:11"]` | `["2025-01-01", "2025-01-01 10:11:11"]` || | Array<null> | `[null]` | `[null]` | `[NULL]` | `[null]` | | | Map<int, string> | `{1:"abc", 2:"中国"}` |`{1:"abc",2:"中国"}` |`{1=abc, 2=中国}` | `{1:"abc", 2:"中国"}` | | | Map<string, date/datetime> | `{"k1":"2022-10-01", "k2":"2022-10-01 10:10:10"}` | `{"k1":"2022-10-01","k2":"2022-10-01 10:10:10"}` | `{k1=2022-10-01, k2=2022-10-01 10:10:10}` | `{"k1":"2022-10-01", "k2":"2022-10-01 10:10:10"}` | | | Map<int, null> | `{1:null, 2:null}` | `{1:null,2:null}` | `{1=NULL, 2=NULL}` | `{1:null, 2:null}` | | | Struct<> | Same as map | Same as map | Same as map | Same as map | | 3. Fix a bug that for batch insert transaction, the `trim_double_quotas` should be set to false
koarz
pushed a commit
to koarz/doris
that referenced
this pull request
Jun 4, 2025
…and use _nesting_level. (apache#50977) We don't need to maintain a separate level; we can achieve the functionality of this apache#49036 by directly using _nesting_level. ```C++ // This parameter indicates what level the serde belongs to and is mainly used for complex types // The default level is 1, and each time you nest, the level increases by 1, // for example: struct<string> // The _nesting_level of StructSerde is 1 // The _nesting_level of StringSerde is 2 int _nesting_level = 1; ```
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
9 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Problem Summary:
The output format of complex data types are different between Hive and Doris, such as array, map and struct.
When user migrate from Hive to Doris, they expect the same format so that they don't need to modify their business code.
This PR mainly changes:
Add a new option to session variable
serde_dialect: If set to hive,the output format returned to MySQL client of some datatypes will be changed:
Array
Doris: ["abc", "def", "", null, 1]
Hive: ["abc","def","",null,true]
Map
Doris: {"k1":null, "k2":"v3"}
Hive: {"k1":null,"k2":"v3"}
Struct
Doris: {"s_id":100, "s_name":"abc , "", "s_address":null}
Hive: {"s_id":100,"s_name":"abc ,"","s_address":null}
Related #37039
Release note
None
Check List (For Author)
Test
Behavior changed:
Does this need documentation?
Check List (For Reviewer who merge this PR)