oushu/lava-oushudb-dt-sql-parser
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

feat: collect entity (#265)

Browse Source 
* feat: add text and word utils

* feat: add entity collector class

* refactor: rename SyntaxContextType to EntityContextType

* refactor: improve EntityCollector

* feat: improve mysql parser grammar

* feat: add mysql entity collector

* test: mysql entity collector tests

* feat: remove useless method

* feat: improve spark grammar file

* feat: add spark entity collector

* test: spark entity collector unit tests

* feat: remove useless code

* feat: add queryStatement label

* feat: add crateDatabaseStmt

* feat: add trino entity collector

* feat: rename trinosql to trino

* test: trino collect entity unit tests

* test: fix spark test

* feat(impala): support impale entity collector (#256)

* Feat/collect entity hive (#263)

* feat(hive): support hive collect entity

* feat(hive): update tableAllColumns

* feat: replace antlr4ts with antlr4ng

* feat(pgsql): pgsql collect entity (#268)

* feat(pgsql): pgsql collect entity

* feat(pgsql): optimize some name

---------

Co-authored-by: zhaoge <>

* feat: get word text by token.text

* feat: supprt collect db/function and add splitListener (#270)

* feat: supprt collect db/function and add splitListner

* feat: remove SplitListener interface in baseParser to use SplitListener in root

* fix(mysql): fix show create xxx not celloct as createXXXEntity type

* test: fix pgsql unit tests

* Feat/error recover predicate (#274)

* feat: optimize pgsql grammar

* feat: add sql parser base

* feat: apply SQLParserBase

* feat: add geAllEntities method

* test: test collect table when missing column

* feat: compose collect and suggestion (#276)

* feat: mark stmt which contain caret

* test: correct name of getAllEntities

* test: remove misscolumn unit tests

* test: add suggestionWithEntity tests

* feat: flink collect entity (#277)

* feat: improve flink sql parser

* feat: support flink entity collector

* test: flink entity collect unit test

* feat: move combine entities to parent class

---------

Co-authored-by: 霜序 <976060700@qq.com>
Co-authored-by: XCynthia <942884029@qq.com>
This commit is contained in:
Hayden
2024-03-26 14:28:27 +08:00
committed by GitHub
parent 3f62ad0d32
commit a99721162b
230 changed files with 56908 additions and 46672 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
test/parser/flinksql/suggestion/fixtures/suggestionWithEntity.sql Normal file
View File

@ -0,0 +1,11 @@
SELECT FROM tb1;
SELECT col1, col2, FROM tb;
INSERT INTO insert_tb PARTITION (country, state) SELECT col1, col2, country, state FROM ( SELECT FROM inside_tb ) subquery;
INSERT INTO insert_tb PARTITION (country, state) SELECT col1, col2, country, state FROM ( SELECT id, FROM inside_tb ) subquery;
CREATE TABLE IF NOT EXISTS derived_table WITH ('connector' = 'kafka') AS SELECT FROM origin_table;
CREATE TABLE IF NOT EXISTS derived_table WITH ('connector' = 'kafka') AS SELECT id, FROM origin_table;

2
test/parser/flinksql/suggestion/fixtures/syntaxSuggestion.sql
View File

@ -32,7 +32,7 @@ SELECT col, FROM tb1;
SELECT * FROM tb ORDER BY ;
SELECT * FROM tb GROUP BY tb. ;
SELECT * FROM tb GROUP BY ;
INSERT INTO tb (col, tb.c );

10
test/parser/flinksql/suggestion/multipleStatement.test.ts
View File

@ -1,7 +1,7 @@
import fs from 'fs';
import path from 'path';
import FlinkSQL from 'src/parser/flinksql';
import { CaretPosition, SyntaxContextType } from 'src/parser/common/basic-parser-types';
import { CaretPosition, EntityContextType } from 'src/parser/common/basic-parser-types';
const syntaxSql = fs.readFileSync(
path.join(__dirname, 'fixtures', 'multipleStatement.sql'),
@ -18,7 +18,7 @@ describe('FlinkSQL Multiple Statements Syntax Suggestion', () => {
};
const syntaxes = parser.getSuggestionAtCaretPosition(syntaxSql, pos)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.TABLE
(syn) => syn.syntaxContextType === EntityContextType.TABLE
);
expect(suggestion).not.toBeUndefined();
@ -32,7 +32,7 @@ describe('FlinkSQL Multiple Statements Syntax Suggestion', () => {
};
const syntaxes = parser.getSuggestionAtCaretPosition(syntaxSql, pos)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.TABLE_CREATE
(syn) => syn.syntaxContextType === EntityContextType.TABLE_CREATE
);
expect(suggestion).not.toBeUndefined();
@ -46,7 +46,7 @@ describe('FlinkSQL Multiple Statements Syntax Suggestion', () => {
};
const syntaxes = parser.getSuggestionAtCaretPosition(syntaxSql, pos)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.TABLE
(syn) => syn.syntaxContextType === EntityContextType.TABLE
);
expect(suggestion).not.toBeUndefined();
@ -60,7 +60,7 @@ describe('FlinkSQL Multiple Statements Syntax Suggestion', () => {
};
const syntaxes = parser.getSuggestionAtCaretPosition(syntaxSql, pos)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.TABLE
(syn) => syn.syntaxContextType === EntityContextType.TABLE
);
expect(suggestion).not.toBeUndefined();

162
test/parser/flinksql/suggestion/suggestionWithEntity.test.ts Normal file
View File

@ -0,0 +1,162 @@
import fs from 'fs';
import path from 'path';
import FlinkSQL from 'src/parser/flinksql';
import { CaretPosition, EntityContextType } from 'src/parser/common/basic-parser-types';
import { commentOtherLine } from 'test/helper';
const syntaxSql = fs.readFileSync(
path.join(__dirname, 'fixtures', 'suggestionWithEntity.sql'),
'utf-8'
);
describe('Flink SQL Syntax Suggestion with collect entity', () => {
const flink = new FlinkSQL();
test('Validate Syntax SQL', () => {
expect(flink.validate(syntaxSql).length).not.toBe(0);
});
test('select with no columns', () => {
const pos: CaretPosition = {
lineNumber: 1,
column: 8,
};
const sql = commentOtherLine(syntaxSql, pos.lineNumber);
const parseTree = flink.parse(sql);
const syntaxes = flink.getSuggestionAtCaretPosition(sql, pos)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === EntityContextType.COLUMN
);
expect(suggestion).not.toBeUndefined();
expect(suggestion?.wordRanges.map((token) => token.text)).toEqual([]);
const entities = flink.getAllEntities(sql, pos);
expect(entities.length).toBe(1);
expect(entities[0].text).toBe('tb1');
expect(entities[0].entityContextType).toBe(EntityContextType.TABLE);
expect(entities[0].belongStmt.isContainCaret).toBeTruthy();
});
test('select with columns with columns and trailing comma', () => {
const pos: CaretPosition = {
lineNumber: 3,
column: 20,
};
const sql = commentOtherLine(syntaxSql, pos.lineNumber);
const syntaxes = flink.getSuggestionAtCaretPosition(sql, pos)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === EntityContextType.COLUMN
);
expect(suggestion).not.toBeUndefined();
expect(suggestion?.wordRanges.map((token) => token.text)).toEqual([]);
const entities = flink.getAllEntities(sql, pos);
expect(entities.length).toBe(1);
expect(entities[0].text).toBe('tb');
expect(entities[0].entityContextType).toBe(EntityContextType.TABLE);
expect(entities[0].belongStmt.isContainCaret).toBeTruthy();
});
test('insert into from nested query with no column', () => {
const pos: CaretPosition = {
lineNumber: 5,
column: 98,
};
const sql = commentOtherLine(syntaxSql, pos.lineNumber);
const syntaxes = flink.getSuggestionAtCaretPosition(sql, pos)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === EntityContextType.COLUMN
);
expect(suggestion).not.toBeUndefined();
expect(suggestion?.wordRanges.map((token) => token.text)).toEqual([]);
const entities = flink.getAllEntities(sql, pos);
expect(entities.length).toBe(2);
expect(entities[0].text).toBe('insert_tb');
expect(entities[0].entityContextType).toBe(EntityContextType.TABLE);
expect(entities[0].belongStmt.isContainCaret).toBeTruthy();
expect(entities[1].text).toBe('inside_tb');
expect(entities[1].entityContextType).toBe(EntityContextType.TABLE);
expect(entities[1].belongStmt.isContainCaret).toBeTruthy();
});
test('insert into from nested query with columns and trailing comma', () => {
const pos: CaretPosition = {
lineNumber: 7,
column: 102,
};
const sql = commentOtherLine(syntaxSql, pos.lineNumber);
const syntaxes = flink.getSuggestionAtCaretPosition(sql, pos)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === EntityContextType.COLUMN
);
expect(suggestion).not.toBeUndefined();
expect(suggestion?.wordRanges.map((token) => token.text)).toEqual([]);
const entities = flink.getAllEntities(sql, pos);
expect(entities.length).toBe(2);
expect(entities[0].text).toBe('insert_tb');
expect(entities[0].entityContextType).toBe(EntityContextType.TABLE);
expect(entities[0].belongStmt.isContainCaret).toBeTruthy();
expect(entities[1].text).toBe('inside_tb');
expect(entities[1].entityContextType).toBe(EntityContextType.TABLE);
expect(entities[1].belongStmt.isContainCaret).toBeTruthy();
});
test('create table as select with no column', () => {
const pos: CaretPosition = {
lineNumber: 9,
column: 82,
};
const sql = commentOtherLine(syntaxSql, pos.lineNumber);
const syntaxes = flink.getSuggestionAtCaretPosition(sql, pos)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === EntityContextType.COLUMN
);
expect(suggestion).not.toBeUndefined();
expect(suggestion?.wordRanges.map((token) => token.text)).toEqual([]);
const entities = flink.getAllEntities(sql, pos);
expect(entities.length).toBe(2);
expect(entities[0].text).toBe('derived_table');
expect(entities[0].entityContextType).toBe(EntityContextType.TABLE_CREATE);
expect(entities[0].belongStmt.isContainCaret).toBeTruthy();
expect(entities[1].text).toBe('origin_table');
expect(entities[1].entityContextType).toBe(EntityContextType.TABLE);
expect(entities[1].belongStmt.isContainCaret).toBeTruthy();
});
test('create table as select with columns and trailing comma', () => {
const pos: CaretPosition = {
lineNumber: 11,
column: 86,
};
const sql = commentOtherLine(syntaxSql, pos.lineNumber);
const syntaxes = flink.getSuggestionAtCaretPosition(sql, pos)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === EntityContextType.COLUMN
);
expect(suggestion).not.toBeUndefined();
expect(suggestion?.wordRanges.map((token) => token.text)).toEqual([]);
const entities = flink.getAllEntities(sql, pos);
expect(entities.length).toBe(2);
expect(entities[0].text).toBe('derived_table');
expect(entities[0].entityContextType).toBe(EntityContextType.TABLE_CREATE);
expect(entities[0].belongStmt.isContainCaret).toBeTruthy();
expect(entities[1].text).toBe('origin_table');
expect(entities[1].entityContextType).toBe(EntityContextType.TABLE);
expect(entities[1].belongStmt.isContainCaret).toBeTruthy();
});
});

48
test/parser/flinksql/suggestion/syntaxSuggestion.test.ts
View File

@ -1,7 +1,7 @@
import fs from 'fs';
import path from 'path';
import FlinkSQL from 'src/parser/flinksql';
import { CaretPosition, SyntaxContextType } from 'src/parser/common/basic-parser-types';
import { CaretPosition, EntityContextType } from 'src/parser/common/basic-parser-types';
import { commentOtherLine } from 'test/helper';
const syntaxSql = fs.readFileSync(
@ -28,7 +28,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.CATALOG
(syn) => syn.syntaxContextType === EntityContextType.CATALOG
);
expect(suggestion).not.toBeUndefined();
@ -45,7 +45,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.TABLE
(syn) => syn.syntaxContextType === EntityContextType.TABLE
);
expect(suggestion).not.toBeUndefined();
@ -62,7 +62,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.TABLE_CREATE
(syn) => syn.syntaxContextType === EntityContextType.TABLE_CREATE
);
expect(suggestion).not.toBeUndefined();
@ -79,7 +79,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.DATABASE
(syn) => syn.syntaxContextType === EntityContextType.DATABASE
);
expect(suggestion).not.toBeUndefined();
@ -96,7 +96,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.DATABASE
(syn) => syn.syntaxContextType === EntityContextType.DATABASE
);
expect(suggestion).not.toBeUndefined();
@ -113,7 +113,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.VIEW
(syn) => syn.syntaxContextType === EntityContextType.VIEW
);
expect(suggestion).not.toBeUndefined();
@ -130,7 +130,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.VIEW
(syn) => syn.syntaxContextType === EntityContextType.VIEW
);
expect(suggestion).not.toBeUndefined();
@ -147,7 +147,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.VIEW_CREATE
(syn) => syn.syntaxContextType === EntityContextType.VIEW_CREATE
);
expect(suggestion).not.toBeUndefined();
@ -164,7 +164,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.FUNCTION
(syn) => syn.syntaxContextType === EntityContextType.FUNCTION
);
expect(suggestion).not.toBeUndefined();
@ -181,7 +181,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.FUNCTION_CREATE
(syn) => syn.syntaxContextType === EntityContextType.FUNCTION_CREATE
);
expect(suggestion).not.toBeUndefined();
@ -198,7 +198,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.VIEW
(syn) => syn.syntaxContextType === EntityContextType.VIEW
);
expect(suggestion).not.toBeUndefined();
@ -215,7 +215,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.TABLE
(syn) => syn.syntaxContextType === EntityContextType.TABLE
);
expect(suggestion).not.toBeUndefined();
@ -232,7 +232,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.VIEW
(syn) => syn.syntaxContextType === EntityContextType.VIEW
);
expect(suggestion).not.toBeUndefined();
@ -249,7 +249,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.COLUMN
(syn) => syn.syntaxContextType === EntityContextType.COLUMN
);
expect(suggestion).not.toBeUndefined();
@ -266,7 +266,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.COLUMN
(syn) => syn.syntaxContextType === EntityContextType.COLUMN
);
expect(suggestion).not.toBeUndefined();
@ -283,7 +283,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.COLUMN
(syn) => syn.syntaxContextType === EntityContextType.COLUMN
);
expect(suggestion).not.toBeUndefined();
@ -300,7 +300,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.COLUMN
(syn) => syn.syntaxContextType === EntityContextType.COLUMN
);
expect(suggestion).not.toBeUndefined();
@ -310,18 +310,18 @@ describe('Flink SQL Syntax Suggestion', () => {
test('Select group by column', () => {
const pos: CaretPosition = {
lineNumber: 35,
column: 30,
column: 27,
};
const syntaxes = parser.getSuggestionAtCaretPosition(
commentOtherLine(syntaxSql, pos.lineNumber),
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.COLUMN
(syn) => syn.syntaxContextType === EntityContextType.COLUMN
);
expect(suggestion).not.toBeUndefined();
expect(suggestion?.wordRanges.map((token) => token.text)).toEqual(['tb', '.']);
expect(suggestion?.wordRanges.map((token) => token.text)).toEqual([]);
});
test('Insert into spec columns', () => {
@ -334,7 +334,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.COLUMN
(syn) => syn.syntaxContextType === EntityContextType.COLUMN
);
expect(suggestion).not.toBeUndefined();
@ -351,7 +351,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.COLUMN
(syn) => syn.syntaxContextType === EntityContextType.COLUMN
);
expect(suggestion).not.toBeUndefined();
@ -368,7 +368,7 @@ describe('Flink SQL Syntax Suggestion', () => {
pos
)?.syntax;
const suggestion = syntaxes?.find(
(syn) => syn.syntaxContextType === SyntaxContextType.COLUMN_CREATE
(syn) => syn.syntaxContextType === EntityContextType.COLUMN_CREATE
);
expect(suggestion).not.toBeUndefined();