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

support format g4 (#232)

Browse Source 
* chore: install antlr-format-cli to format g4 files

* style: format all g4 files

* chore: apply antlr-format to lint-staged
This commit is contained in:
Hayden
2023-12-12 20:05:21 +08:00
committed by GitHub
parent 17eba1f2aa
commit 14a5214e82
20 changed files with 16763 additions and 13690 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1117
src/grammar/flinksql/FlinkSqlLexer.g4
View File

File diff suppressed because it is too large Load Diff

356
src/grammar/flinksql/FlinkSqlParser.g4
View File

@ -1,8 +1,16 @@
// $antlr-format alignTrailingComments true, columnLimit 150, minEmptyLines 1, maxEmptyLinesToKeep 1, reflowComments false, useTab false
// $antlr-format allowShortRulesOnASingleLine false, allowShortBlocksOnASingleLine true, alignSemicolons hanging, alignColons hanging
// $antlr-format spaceBeforeAssignmentOperators false, keepEmptyLinesAtTheStartOfBlocks true
parser grammar FlinkSqlParser;
options { tokenVocab=FlinkSqlLexer; }
options {
tokenVocab=FlinkSqlLexer;
}
program: singleStatement* EOF;
program
: singleStatement* EOF
;
singleStatement
: sqlStatement SEMICOLON?
@ -10,10 +18,18 @@ singleStatement
;
sqlStatement
: ddlStatement | dmlStatement | describeStatement
| explainStatement | useStatement | showStatememt
| loadStatement | unloadStatememt | setStatememt
| resetStatememt | jarStatememt | dtAddStatement
: ddlStatement
| dmlStatement
| describeStatement
| explainStatement
| useStatement
| showStatememt
| loadStatement
| unloadStatememt
| setStatememt
| resetStatememt
| jarStatememt
| dtAddStatement
;
emptyStatement
@ -21,13 +37,25 @@ emptyStatement
;
ddlStatement
: createTable | createDatabase | createView | createFunction | createCatalog
| alterTable | alertView | alterDatabase | alterFunction
| dropCatalog | dropTable | dropDatabase | dropView | dropFunction
: createTable
| createDatabase
| createView
| createFunction
| createCatalog
| alterTable
| alertView
| alterDatabase
| alterFunction
| dropCatalog
| dropTable
| dropDatabase
| dropView
| dropFunction
;
dmlStatement
: queryStatement | insertStatement
: queryStatement
| insertStatement
;
// some statemen
@ -36,7 +64,11 @@ describeStatement
;
explainStatement
: KW_EXPLAIN (explainDetails | KW_PLAN KW_FOR)? (dmlStatement | insertSimpleStatement | insertMulStatement)
: KW_EXPLAIN (explainDetails | KW_PLAN KW_FOR)? (
dmlStatement
| insertSimpleStatement
| insertMulStatement
)
;
explainDetails
@ -44,11 +76,13 @@ explainDetails
;
explainDetail
: KW_CHANGELOG_MODE | KW_JSON_EXECUTION_PLAN | KW_ESTIMATED_COST
: KW_CHANGELOG_MODE
| KW_JSON_EXECUTION_PLAN
| KW_ESTIMATED_COST
;
useStatement
: KW_USE KW_CATALOG catalogPath
: KW_USE KW_CATALOG catalogPath
| KW_USE databasePath
| useModuleStatement
;
@ -60,8 +94,8 @@ useModuleStatement
showStatememt
: KW_SHOW (KW_CATALOGS | KW_DATABASES | KW_VIEWS | KW_JARS)
| KW_SHOW KW_CURRENT (KW_CATALOG | KW_DATABASE)
| KW_SHOW KW_TABLES (( KW_FROM | KW_IN ) databasePath)? likePredicate?
| KW_SHOW KW_COLUMNS ( KW_FROM | KW_IN ) (viewPath| tablePath) likePredicate?
| KW_SHOW KW_TABLES (( KW_FROM | KW_IN) databasePath)? likePredicate?
| KW_SHOW KW_COLUMNS ( KW_FROM | KW_IN) (viewPath | tablePath) likePredicate?
| KW_SHOW KW_CREATE (KW_TABLE tablePath | KW_VIEW viewPath)
| KW_SHOW KW_USER? KW_FUNCTIONS
| KW_SHOW KW_FULL? KW_MODULES
@ -70,7 +104,7 @@ showStatememt
loadStatement
: KW_LOAD KW_MODULE uid (KW_WITH tablePropertyList)?
;
unloadStatememt
: KW_UNLOAD KW_MODULE uid
;
@ -82,7 +116,7 @@ setStatememt
resetStatememt
: KW_RESET tablePropertyKey?
;
jarStatememt
: (KW_ADD | KW_REMOVE) KW_JAR jarFileName
;
@ -91,7 +125,13 @@ jarStatememt
dtAddStatement
: KW_ADD KW_JAR KW_WITH dtFilePath (KW_AS uid)?
| KW_ADD KW_FILE KW_WITH dtFilePath (KW_AS uid)? (KW_RENAME uid)?
| KW_ADD (KW_PYTHON_FILES | KW_PYTHON_REQUIREMENTS | KW_PYTHON_DEPENDENCIES | KW_PYTHON_JAR | KW_PYTHON_ARCHIVES) KW_WITH dtFilePath KW_RENAME uid
| KW_ADD (
KW_PYTHON_FILES
| KW_PYTHON_REQUIREMENTS
| KW_PYTHON_DEPENDENCIES
| KW_PYTHON_JAR
| KW_PYTHON_ARCHIVES
) KW_WITH dtFilePath KW_RENAME uid
| KW_ADD KW_PYTHON_PARAMETER dtFilePath
| KW_ADD KW_ENGINE KW_FILE KW_WITH dtFilePath KW_RENAME uid KW_KEY uid
| KW_ADD KW_CONFIG KW_FILE KW_WITH dtFilePath KW_FOR uid KW_AS uid
@ -106,19 +146,12 @@ dtFilePath
createTable
: (simpleCreateTable | createTableAsSelect)
;
simpleCreateTable
: KW_CREATE KW_TEMPORARY? KW_TABLE ifNotExists? tablePathCreate
LR_BRACKET
columnOptionDefinition (COMMA columnOptionDefinition)*
(COMMA watermarkDefinition)?
(COMMA tableConstraint)?
(COMMA selfDefinitionClause)?
RR_BRACKET
commentSpec?
partitionDefinition?
withOption
likeDefinition?
: KW_CREATE KW_TEMPORARY? KW_TABLE ifNotExists? tablePathCreate LR_BRACKET columnOptionDefinition (
COMMA columnOptionDefinition
)* (COMMA watermarkDefinition)? (COMMA tableConstraint)? (COMMA selfDefinitionClause)? RR_BRACKET commentSpec? partitionDefinition? withOption
likeDefinition?
;
/*
@ -141,10 +174,12 @@ physicalColumnDefinition
columnNameCreate
: uid
| expression;
| expression
;
columnName
: uid | expression
: uid
| expression
;
columnNameList
@ -153,9 +188,21 @@ columnNameList
columnType
: typeName=(KW_DATE | KW_BOOLEAN | KW_NULL)
| typeName=(KW_CHAR | KW_VARCHAR | KW_STRING | KW_BINARY | KW_VARBINARY | KW_BYTES
| KW_TINYINT | KW_SMALLINT |KW_INT | KW_INTEGER | KW_BIGINT
| KW_TIME | KW_TIMESTAMP_LTZ | KW_DATETIME
| typeName=(
KW_CHAR
| KW_VARCHAR
| KW_STRING
| KW_BINARY
| KW_VARBINARY
| KW_BYTES
| KW_TINYINT
| KW_SMALLINT
| KW_INT
| KW_INTEGER
| KW_BIGINT
| KW_TIME
| KW_TIMESTAMP_LTZ
| KW_DATETIME
) lengthOneDimension?
| typeName=KW_TIMESTAMP lengthOneDimension? ((KW_WITHOUT | KW_WITH) KW_LOCAL? KW_TIME KW_ZONE)?
| typeName=(KW_DECIMAL | KW_DEC | KW_NUMERIC | KW_FLOAT | KW_DOUBLE) lengthTwoOptionalDimension?
@ -190,7 +237,8 @@ rowTypeDimension
;
columnConstraint
:(KW_CONSTRAINT constraintName)? KW_PRIMARY KW_KEY (KW_NOT KW_ENFORCED)? | KW_NOT? KW_NULL
: (KW_CONSTRAINT constraintName)? KW_PRIMARY KW_KEY (KW_NOT KW_ENFORCED)?
| KW_NOT? KW_NULL
;
commentSpec
@ -239,9 +287,9 @@ transformList
;
transform
: columnName #identityTransform
| qualifiedName #columnTransform
| LR_BRACKET transformArgument (COMMA transformArgument)* RR_BRACKET #applyTransform
: columnName # identityTransform
| qualifiedName # columnTransform
| LR_BRACKET transformArgument (COMMA transformArgument)* RR_BRACKET # applyTransform
;
transformArgument
@ -271,11 +319,13 @@ createView
;
createFunction
: KW_CREATE (KW_TEMPORARY|KW_TEMPORARY KW_SYSTEM)? KW_FUNCTION ifNotExists? functionNameCreate KW_AS identifier (KW_LANGUAGE (KW_JAVA|KW_SCALA|KW_PYTHON))? usingClause?
: KW_CREATE (KW_TEMPORARY | KW_TEMPORARY KW_SYSTEM)? KW_FUNCTION ifNotExists? functionNameCreate KW_AS identifier (
KW_LANGUAGE (KW_JAVA | KW_SCALA | KW_PYTHON)
)? usingClause?
;
usingClause
: KW_USING KW_JAR jarFileName (COMMA KW_JAR jarFileName)*
: KW_USING KW_JAR jarFileName (COMMA KW_JAR jarFileName)*
;
jarFileName
@ -287,7 +337,13 @@ jarFileName
// it only includes rename, set key, add constraint, drop constraint, add unique
alterTable
: KW_ALTER KW_TABLE ifExists? tablePath (renameDefinition | setKeyValueDefinition | addConstraint | dropConstraint | addUnique)
: KW_ALTER KW_TABLE ifExists? tablePath (
renameDefinition
| setKeyValueDefinition
| addConstraint
| dropConstraint
| addUnique
)
;
renameDefinition
@ -323,10 +379,11 @@ alterDatabase
;
alterFunction
: KW_ALTER (KW_TEMPORARY|KW_TEMPORARY KW_SYSTEM)? KW_FUNCTION ifExists? functionName KW_AS identifier (KW_LANGUAGE (KW_JAVA|KW_SCALA|KW_PYTHON))? // TODO
: KW_ALTER (KW_TEMPORARY | KW_TEMPORARY KW_SYSTEM)? KW_FUNCTION ifExists? functionName KW_AS identifier (
KW_LANGUAGE (KW_JAVA | KW_SCALA | KW_PYTHON)
)? // TODO
;
// Drop statements
dropCatalog
@ -346,20 +403,19 @@ dropView
;
dropFunction
: KW_DROP (KW_TEMPORARY|KW_TEMPORARY KW_SYSTEM)? KW_FUNCTION ifExists? functionName
: KW_DROP (KW_TEMPORARY | KW_TEMPORARY KW_SYSTEM)? KW_FUNCTION ifExists? functionName
;
// Insert statements
insertStatement
: (KW_EXECUTE? insertSimpleStatement)
| insertMulStatementCompatibility | (KW_EXECUTE insertMulStatement)
: (KW_EXECUTE? insertSimpleStatement)
| insertMulStatementCompatibility
| (KW_EXECUTE insertMulStatement)
;
insertSimpleStatement
: KW_INSERT (KW_INTO | KW_OVERWRITE) tablePath
(
: KW_INSERT (KW_INTO | KW_OVERWRITE) tablePath (
insertPartitionDefinition? columnNameList? queryStatement
| valuesDefinition
)
@ -374,13 +430,11 @@ valuesDefinition
;
valuesRowDefinition
: LR_BRACKET
constant (COMMA constant)*
RR_BRACKET
: LR_BRACKET constant (COMMA constant)* RR_BRACKET
;
insertMulStatementCompatibility
: KW_BEGIN KW_STATEMENT KW_SET SEMICOLON (insertSimpleStatement SEMICOLON)+ KW_END
: KW_BEGIN KW_STATEMENT KW_SET SEMICOLON (insertSimpleStatement SEMICOLON)+ KW_END
;
insertMulStatement
@ -399,7 +453,7 @@ queryStatement
;
valuesCaluse
: KW_VALUES expression (COMMA expression )*
: KW_VALUES expression (COMMA expression)*
;
withClause
@ -420,12 +474,15 @@ selectStatement
;
selectClause
: KW_SELECT setQuantifier? (ASTERISK_SIGN | projectItemDefinition (COMMA projectItemDefinition)*)
: KW_SELECT setQuantifier? (
ASTERISK_SIGN
| projectItemDefinition (COMMA projectItemDefinition)*
)
;
projectItemDefinition
: overWindowItem
| columnName (KW_AS? expression )?
| columnName (KW_AS? expression)?
;
overWindowItem
@ -481,7 +538,7 @@ windoTVFName
: KW_TUMBLE
| KW_HOP
| KW_CUMULATE
;
;
windowTVFParam
: KW_TABLE timeAttrColumn
@ -563,25 +620,12 @@ namedWindow
;
windowSpec
: name=errorCapturingIdentifier?
LR_BRACKET
partitionByClause?
orderByCaluse?
windowFrame?
RR_BRACKET
: name=errorCapturingIdentifier? LR_BRACKET partitionByClause? orderByCaluse? windowFrame? RR_BRACKET
;
matchRecognizeClause
: KW_MATCH_RECOGNIZE
LR_BRACKET
partitionByClause?
orderByCaluse?
measuresClause?
outputMode?
afterMatchStrategy?
patternDefination?
patternVariablesDefination
RR_BRACKET ( KW_AS? identifier )?
: KW_MATCH_RECOGNIZE LR_BRACKET partitionByClause? orderByCaluse? measuresClause? outputMode? afterMatchStrategy? patternDefination?
patternVariablesDefination RR_BRACKET (KW_AS? identifier)?
;
orderByCaluse
@ -605,7 +649,7 @@ quantifiers
| (ADD_SIGN)
| (QUESTION_MARK_SIGN)
| (LB_BRACKET DIG_LITERAL COMMA DIG_LITERAL RB_BRACKET)
| (LB_BRACKET DIG_LITERAL COMMA RB_BRACKET)
| (LB_BRACKET DIG_LITERAL COMMA RB_BRACKET)
| (LB_BRACKET COMMA DIG_LITERAL RB_BRACKET)
;
@ -614,11 +658,7 @@ measuresClause
;
patternDefination
: KW_PATTERN
LR_BRACKET
patternVariable+
RR_BRACKET
withinClause?
: KW_PATTERN LR_BRACKET patternVariable+ RR_BRACKET withinClause?
;
patternVariable
@ -631,7 +671,7 @@ outputMode
;
afterMatchStrategy
: KW_AFTER KW_MATCH KW_SKIP KW_PAST KW_LAST KW_ROW
: KW_AFTER KW_MATCH KW_SKIP KW_PAST KW_LAST KW_ROW
| KW_AFTER KW_MATCH KW_SKIP KW_TO KW_NEXT KW_ROW
| KW_AFTER KW_MATCH KW_SKIP KW_TO KW_LAST unquotedIdentifier
| KW_AFTER KW_MATCH KW_SKIP KW_TO KW_FIRST unquotedIdentifier
@ -654,7 +694,6 @@ withinClause
: KW_WITHIN timeIntervalExpression
;
// expression
expression
@ -662,19 +701,16 @@ expression
;
booleanExpression
: KW_NOT booleanExpression #logicalNot
| KW_EXISTS LR_BRACKET queryStatement RR_BRACKET #exists
| valueExpression predicate? #predicated
| left=booleanExpression operator=KW_AND right=booleanExpression #logicalBinary
| left=booleanExpression operator=KW_OR right=booleanExpression #logicalBinary
| booleanExpression KW_IS KW_NOT? kind=(KW_TRUE | KW_FALSE | KW_UNKNOWN | KW_NULL) #logicalNested
: KW_NOT booleanExpression # logicalNot
| KW_EXISTS LR_BRACKET queryStatement RR_BRACKET # exists
| valueExpression predicate? # predicated
| left=booleanExpression operator=KW_AND right=booleanExpression # logicalBinary
| left=booleanExpression operator=KW_OR right=booleanExpression # logicalBinary
| booleanExpression KW_IS KW_NOT? kind=(KW_TRUE | KW_FALSE | KW_UNKNOWN | KW_NULL) # logicalNested
;
predicate
: KW_NOT?
kind=KW_BETWEEN (KW_ASYMMETRIC | KW_SYMMETRIC)?
lower=valueExpression KW_AND
upper=valueExpression
: KW_NOT? kind=KW_BETWEEN (KW_ASYMMETRIC | KW_SYMMETRIC)? lower=valueExpression KW_AND upper=valueExpression
| KW_NOT? kind=KW_IN LR_BRACKET expression (COMMA expression)* RR_BRACKET
| KW_NOT? kind=KW_IN LR_BRACKET queryStatement RR_BRACKET
| kind=KW_EXISTS LR_BRACKET queryStatement RR_BRACKET
@ -686,42 +722,45 @@ predicate
;
likePredicate
: KW_NOT? kind=KW_LIKE quantifier=(KW_ANY | KW_ALL) (LR_BRACKET RR_BRACKET | LR_BRACKET expression (COMMA expression)* RR_BRACKET)
: KW_NOT? kind=KW_LIKE quantifier=(KW_ANY | KW_ALL) (
LR_BRACKET RR_BRACKET
| LR_BRACKET expression (COMMA expression)* RR_BRACKET
)
| KW_NOT? kind=KW_LIKE pattern=valueExpression (KW_ESCAPE stringLiteral)?
;
valueExpression
: primaryExpression #valueExpressionDefault
| operator=(HYPNEN_SIGN | ADD_SIGN | BIT_NOT_OP) valueExpression #arithmeticUnary
| left=valueExpression operator=(ASTERISK_SIGN | SLASH_SIGN | PENCENT_SIGN | KW_DIV) right=valueExpression #arithmeticBinary
| left=valueExpression operator=(ADD_SIGN | HYPNEN_SIGN | DOUBLE_VERTICAL_SIGN) right=valueExpression #arithmeticBinary
| left=valueExpression operator=BIT_AND_OP right=valueExpression #arithmeticBinary
| left=valueExpression operator=BIT_XOR_OP right=valueExpression #arithmeticBinary
| left=valueExpression operator=BIT_OR_OP right=valueExpression #arithmeticBinary
| left=valueExpression comparisonOperator right=valueExpression #comparison
: primaryExpression # valueExpressionDefault
| operator=(HYPNEN_SIGN | ADD_SIGN | BIT_NOT_OP) valueExpression # arithmeticUnary
| left=valueExpression operator=(ASTERISK_SIGN | SLASH_SIGN | PENCENT_SIGN | KW_DIV) right=valueExpression # arithmeticBinary
| left=valueExpression operator=(ADD_SIGN | HYPNEN_SIGN | DOUBLE_VERTICAL_SIGN) right=valueExpression # arithmeticBinary
| left=valueExpression operator=BIT_AND_OP right=valueExpression # arithmeticBinary
| left=valueExpression operator=BIT_XOR_OP right=valueExpression # arithmeticBinary
| left=valueExpression operator=BIT_OR_OP right=valueExpression # arithmeticBinary
| left=valueExpression comparisonOperator right=valueExpression # comparison
;
primaryExpression
: KW_CASE whenClause+ (KW_ELSE elseExpression=expression)? KW_END #searchedCase
| KW_CASE value=expression whenClause+ (KW_ELSE elseExpression=expression)? KW_END #simpleCase
| KW_CAST LR_BRACKET expression KW_AS columnType RR_BRACKET #cast
: KW_CASE whenClause+ (KW_ELSE elseExpression=expression)? KW_END # searchedCase
| KW_CASE value=expression whenClause+ (KW_ELSE elseExpression=expression)? KW_END # simpleCase
| KW_CAST LR_BRACKET expression KW_AS columnType RR_BRACKET # cast
// | STRUCT LR_BRACKET (argument+=namedExpression (COMMA argument+=namedExpression)*)? RR_BRACKET #struct
| KW_FIRST LR_BRACKET expression (KW_IGNORE KW_NULLS)? RR_BRACKET #first
| KW_LAST LR_BRACKET expression (KW_IGNORE KW_NULLS)? RR_BRACKET #last
| KW_POSITION LR_BRACKET substr=valueExpression KW_IN str=valueExpression RR_BRACKET #position
| constant #constantDefault
| ASTERISK_SIGN #star
| uid DOT ASTERISK_SIGN #star
| KW_FIRST LR_BRACKET expression (KW_IGNORE KW_NULLS)? RR_BRACKET # first
| KW_LAST LR_BRACKET expression (KW_IGNORE KW_NULLS)? RR_BRACKET # last
| KW_POSITION LR_BRACKET substr=valueExpression KW_IN str=valueExpression RR_BRACKET # position
| constant # constantDefault
| ASTERISK_SIGN # star
| uid DOT ASTERISK_SIGN # star
// | LR_BRACKET namedExpression (COMMA namedExpression)+ RR_BRACKET #rowConstructor
| LR_BRACKET queryStatement RR_BRACKET #subqueryExpression
| functionName LR_BRACKET (setQuantifier? functionParam (COMMA functionParam)*)? RR_BRACKET #functionCall
| LR_BRACKET queryStatement RR_BRACKET # subqueryExpression
| functionName LR_BRACKET (setQuantifier? functionParam (COMMA functionParam)*)? RR_BRACKET # functionCall
// | identifier '->' expression #lambda
// | '(' identifier (',' identifier)+ ')' '->' expression #lambda
| value=primaryExpression LS_BRACKET index=valueExpression RS_BRACKET #subscript
| identifier #columnReference
| dereferenceDefinition #dereference
| LR_BRACKET expression RR_BRACKET #parenthesizedExpression
| KW_CURRENT_TIMESTAMP #dateFunctionExpression
| value=primaryExpression LS_BRACKET index=valueExpression RS_BRACKET # subscript
| identifier # columnReference
| dereferenceDefinition # dereference
| LR_BRACKET expression RR_BRACKET # parenthesizedExpression
| KW_CURRENT_TIMESTAMP # dateFunctionExpression
// | EXTRACT LR_BRACKET field=identifier KW_FROM source=valueExpression RR_BRACKET #extract
// | (SUBSTR | SUBSTRING) LR_BRACKET str=valueExpression (KW_FROM | COMMA) pos=valueExpression
// ((KW_FOR | COMMA) len=valueExpression)? RR_BRACKET #substring
@ -758,7 +797,8 @@ correlationName
;
qualifiedName
: identifier | dereferenceDefinition
: identifier
| dereferenceDefinition
;
timeIntervalExpression
@ -795,8 +835,8 @@ errorCapturingIdentifier
;
errorCapturingIdentifierExtra
: (KW_MINUS identifier)+ #errorIdent
| #realIdent
: (KW_MINUS identifier)+ # errorIdent
| # realIdent
;
identifierList
@ -808,13 +848,14 @@ identifierSeq
;
identifier
: unquotedIdentifier #unquotedIdentifierAlternative
| quotedIdentifier #quotedIdentifierAlternative
| nonReservedKeywords #nonReservedKeywordsAlternative
: unquotedIdentifier # unquotedIdentifierAlternative
| quotedIdentifier # quotedIdentifierAlternative
| nonReservedKeywords # nonReservedKeywordsAlternative
;
unquotedIdentifier
: DIG_LITERAL | ID_LITERAL
: DIG_LITERAL
| ID_LITERAL
;
quotedIdentifier
@ -870,10 +911,12 @@ withOption
;
ifNotExists
: KW_IF KW_NOT KW_EXISTS;
: KW_IF KW_NOT KW_EXISTS
;
ifExists
: KW_IF KW_EXISTS;
: KW_IF KW_EXISTS
;
tablePropertyList
: LR_BRACKET tableProperty (COMMA tableProperty)* RR_BRACKET
@ -884,7 +927,8 @@ tableProperty
;
tablePropertyKey
: identifier | dereferenceDefinition
: identifier
| dereferenceDefinition
| STRING_LITERAL
;
@ -898,56 +942,56 @@ tablePropertyValue
logicalOperator
: KW_AND
| BIT_AND_OP BIT_AND_OP
| KW_OR
| KW_OR
| BIT_OR_OP BIT_OR_OP
;
comparisonOperator
: EQUAL_SYMBOL
| GREATER_SYMBOL
| LESS_SYMBOL
| LESS_SYMBOL EQUAL_SYMBOL
: EQUAL_SYMBOL
| GREATER_SYMBOL
| LESS_SYMBOL
| LESS_SYMBOL EQUAL_SYMBOL
| GREATER_SYMBOL EQUAL_SYMBOL
| LESS_SYMBOL GREATER_SYMBOL
| EXCLAMATION_SYMBOL EQUAL_SYMBOL
| LESS_SYMBOL GREATER_SYMBOL
| EXCLAMATION_SYMBOL EQUAL_SYMBOL
| LESS_SYMBOL EQUAL_SYMBOL GREATER_SYMBOL
;
bitOperator
: LESS_SYMBOL LESS_SYMBOL
| GREATER_SYMBOL GREATER_SYMBOL
| BIT_AND_OP
| BIT_XOR_OP
: LESS_SYMBOL LESS_SYMBOL
| GREATER_SYMBOL GREATER_SYMBOL
| BIT_AND_OP
| BIT_XOR_OP
| BIT_OR_OP
;
mathOperator
: ASTERISK_SIGN
| SLASH_SIGN
| PENCENT_SIGN
| KW_DIV
| ADD_SIGN
| HYPNEN_SIGN
: ASTERISK_SIGN
| SLASH_SIGN
| PENCENT_SIGN
| KW_DIV
| ADD_SIGN
| HYPNEN_SIGN
| DOUBLE_HYPNEN_SIGN
;
unaryOperator
: EXCLAMATION_SYMBOL
| BIT_NOT_OP
| ADD_SIGN
| HYPNEN_SIGN
: EXCLAMATION_SYMBOL
| BIT_NOT_OP
| ADD_SIGN
| HYPNEN_SIGN
| KW_NOT
;
constant
: timeIntervalExpression
| timePointLiteral
| stringLiteral // 引号包含的字符串
| HYPNEN_SIGN? decimalLiteral // 正/负整数
| booleanLiteral // 布尔值
| REAL_LITERAL // 小数
| stringLiteral // 引号包含的字符串
| HYPNEN_SIGN? decimalLiteral // 正/负整数
| booleanLiteral // 布尔值
| REAL_LITERAL // 小数
| BIT_STRING
| KW_NOT? KW_NULL // 空 | 非空
| KW_NOT? KW_NULL // 空 | 非空
;
timePointLiteral
@ -963,7 +1007,9 @@ decimalLiteral
;
booleanLiteral
: KW_TRUE | KW_FALSE;
: KW_TRUE
| KW_FALSE
;
setQuantifier
: KW_DISTINCT
@ -1159,5 +1205,5 @@ nonReservedKeywords
| KW_VIEW
| KW_WEEK
| KW_YEARS
| KW_ZONE
| KW_ZONE
;

957
src/grammar/hive/HiveSqlLexer.g4
View File

File diff suppressed because it is too large Load Diff

602
src/grammar/hive/HiveSqlParser.g4
View File

File diff suppressed because it is too large Load Diff

592
src/grammar/impala/ImpalaSqlLexer.g4
View File

@ -19,351 +19,317 @@ OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
THE SOFTWARE.
*/
// $antlr-format alignTrailingComments true, columnLimit 150, maxEmptyLinesToKeep 1, reflowComments false, useTab false
// $antlr-format allowShortRulesOnASingleLine true, allowShortBlocksOnASingleLine true, minEmptyLines 0, alignSemicolons none, alignColons trailing
// $antlr-format singleLineOverrulesHangingColon true, alignLexerCommands true, alignLabels true, alignTrailers true
// $antlr-format spaceBeforeAssignmentOperators false, groupedAlignments true
lexer grammar ImpalaSqlLexer;
options { caseInsensitive = true; }
options {
caseInsensitive=true;
}
KW_ADD : 'ADD';
KW_ALL : 'ALL';
KW_ANALYTIC : 'ANALYTIC';
KW_ALTER : 'ALTER';
KW_AND : 'AND';
KW_ANY : 'ANY';
KW_ANTI : 'ANTI';
KW_ARRAY : 'ARRAY';
KW_AS : 'AS';
KW_ASC : 'ASC';
KW_AT : 'AT';
KW_AGGREGATE : 'AGGREGATE';
KW_AUTHORIZATION : 'AUTHORIZATION';
KW_BERNOULLI : 'BERNOULLI';
KW_BETWEEN : 'BETWEEN';
KW_BLOCK_SIZE : 'BLOCK_SIZE';
KW_PARTITIONED : 'PARTITIONED';
KW_PREPARE_FN : 'PREPARE_FN';
KW_EXTERNAL : 'EXTERNAL';
KW_CLOSEFN : 'CLOSEFN';
KW_SORT : 'SORT';
KW_PURGE : 'PURGE';
KW_STORED : 'STORED';
KW_LOCATION : 'LOCATION';
KW_TBLPROPERTIES : 'TBLPROPERTIES';
KW_BY : 'BY';
KW_CASCADE : 'CASCADE';
KW_CASE : 'CASE';
KW_CAST : 'CAST';
KW_CACHED : 'CACHED';
KW_CHANGE : 'CHANGE';
KW_COLUMN : 'COLUMN';
KW_COLUMNS : 'COLUMNS';
KW_COMMENT : 'COMMENT';
KW_COMPRESSION : 'COMPRESSION';
KW_COMPUTE : 'COMPUTE';
KW_CREATE : 'CREATE';
KW_CROSS : 'CROSS';
KW_CURRENT : 'CURRENT';
KW_CURRENT_DATE : 'CURRENT_DATE';
KW_CURRENT_PATH : 'CURRENT_PATH';
KW_CURRENT_TIME : 'CURRENT_TIME';
KW_CURRENT_TIMESTAMP : 'CURRENT_TIMESTAMP';
KW_CURRENT_USER : 'CURRENT_USER';
KW_DATA : 'DATA';
KW_DATABASE : 'DATABASE';
KW_DATABASES : 'DATABASES';
KW_DAY : 'DAY';
KW_DAYS : 'DAYS';
KW_DELETE : 'DELETE';
KW_DEFAULT : 'DEFAULT';
KW_DELIMITED : 'DELIMITED';
KW_DISABLE : 'DISABLE';
KW_UPDATE : 'UPDATE';
KW_DESC : 'DESC';
KW_DESCRIBE : 'DESCRIBE';
KW_DISTINCT : 'DISTINCT';
KW_DROP : 'DROP';
KW_ELSE : 'ELSE';
KW_ENCODING : 'ENCODING';
KW_END : 'END';
KW_ESCAPE : 'ESCAPE';
KW_ESCAPED : 'ESCAPED';
KW_EXCEPT : 'EXCEPT';
KW_EXCLUDING : 'EXCLUDING';
KW_EXISTS : 'EXISTS';
KW_EXPLAIN : 'EXPLAIN';
KW_EXTRACT : 'EXTRACT';
KW_EXTENDED : 'EXTENDED';
KW_FALSE : 'FALSE';
KW_FIELDS : 'FIELDS';
KW_FILEFORMAT : 'FILEFORMAT';
KW_FILES : 'FILES';
KW_FILTER : 'FILTER';
KW_FIRST : 'FIRST';
KW_FINALIZE_FN : 'FINALIZE_FN';
KW_FOLLOWING : 'FOLLOWING';
KW_FOR : 'FOR';
KW_FORMAT : 'FORMAT';
KW_FORMATTED : 'FORMATTED';
KW_FOREIGN : 'FOREIGN';
KW_FROM : 'FROM';
KW_FULL : 'FULL';
KW_FUNCTION : 'FUNCTION';
KW_FUNCTIONS : 'FUNCTIONS';
KW_GRANT : 'GRANT';
KW_GROUP : 'GROUP';
KW_GROUPING : 'GROUPING';
KW_HASH : 'HASH';
KW_HAVING : 'HAVING';
KW_HOUR : 'HOUR';
KW_HOURS : 'HOURS';
KW_IF : 'IF';
KW_IN : 'IN';
KW_INCLUDING : 'INCLUDING';
KW_INCREMENTAL : 'INCREMENTAL';
KW_INNER : 'INNER';
KW_INPATH : 'INPATH';
KW_INSERT : 'INSERT';
KW_INTERSECT : 'INTERSECT';
KW_INTERVAL : 'INTERVAL';
KW_INTERMEDIATE : 'INTERMEDIATE';
KW_INTO : 'INTO';
KW_INIT_FN : 'INIT_FN';
KW_IREGEXP : 'IREGEXP';
KW_ILIKE : 'ILIKE';
KW_INVALIDATE : 'INVALIDATE';
KW_IS : 'IS';
KW_JOIN : 'JOIN';
KW_KEY : 'KEY';
KW_KUDU : 'KUDU';
KW_LAST : 'LAST';
KW_LATERAL : 'LATERAL';
KW_LEFT : 'LEFT';
KW_LIKE : 'LIKE';
KW_LIMIT : 'LIMIT';
KW_LINES : 'LINES';
KW_LOAD : 'LOAD';
KW_LOCALTIME : 'LOCALTIME';
KW_LOCALTIMESTAMP : 'LOCALTIMESTAMP';
KW_METADATA : 'METADATA';
KW_MAP : 'MAP';
KW_MINUTE : 'MINUTE';
KW_MINUTES : 'MINUTES';
KW_MONTH : 'MONTH';
KW_MONTHS : 'MONTHS';
KW_MERGE_FN : 'MERGE_FN';
KW_NFC : 'NFC';
KW_NFD : 'NFD';
KW_NFKC : 'NFKC';
KW_NFKD : 'NFKD';
KW_NORMALIZE : 'NORMALIZE';
KW_NOT : 'NOT';
KW_NULL : 'NULL';
KW_NULLS : 'NULLS';
KW_OFFSET : 'OFFSET';
KW_ON : 'ON';
KW_OPTION : 'OPTION';
KW_OR : 'OR';
KW_ORDER : 'ORDER';
KW_ORDINALITY : 'ORDINALITY';
KW_OUTER : 'OUTER';
KW_OWNER : 'OWNER';
KW_OVER : 'OVER';
KW_OVERWRITE : 'OVERWRITE';
KW_PARTITION : 'PARTITION';
KW_PARTITIONS : 'PARTITIONS';
KW_PARQUET : 'PARQUET';
KW_POSITION : 'POSITION';
KW_PRECEDING : 'PRECEDING';
KW_PRIMARY : 'PRIMARY';
KW_REPLICATION : 'REPLICATION';
KW_PRIVILEGES : 'PRIVILEGES';
KW_PROPERTIES : 'PROPERTIES';
KW_RANGE : 'RANGE';
KW_RECOVER : 'RECOVER';
KW_RENAME : 'RENAME';
KW_REPEATABLE : 'REPEATABLE';
KW_REPLACE : 'REPLACE';
KW_RESTRICT : 'RESTRICT';
KW_RETURNS : 'RETURNS';
KW_REVOKE : 'REVOKE';
KW_REFRESH : 'REFRESH';
KW_REGEXP : 'REGEXP';
KW_RLIKE : 'RLIKE';
KW_RIGHT : 'RIGHT';
KW_ROLE : 'ROLE';
KW_ROLES : 'ROLES';
KW_ROW : 'ROW';
KW_ROWS : 'ROWS';
KW_SCHEMA : 'SCHEMA';
KW_SCHEMAS : 'SCHEMAS';
KW_SECOND : 'SECOND';
KW_SECONDS : 'SECONDS';
KW_SELECT : 'SELECT';
KW_SERDEPROPERTIES : 'SERDEPROPERTIES';
KW_SET : 'SET';
KW_SEMI : 'SEMI';
KW_SERVER : 'SERVER';
KW_SHOW : 'SHOW';
KW_SHUTDOWN : 'SHUTDOWN';
KW_SOME : 'SOME';
KW_STATS : 'STATS';
KW_STRUCT : 'STRUCT';
KW_STRAIGHT_JOIN : 'STRAIGHT_JOIN';
KW_SUBSTRING : 'SUBSTRING';
KW_SYSTEM : 'SYSTEM';
KW_SYMBOL : 'SYMBOL';
KW_SERIALIZE_FN : 'SERIALIZE_FN';
KW_TABLE : 'TABLE';
KW_TABLES : 'TABLES';
KW_TABLESAMPLE : 'TABLESAMPLE';
KW_TERMINATED : 'TERMINATED ';
KW_THEN : 'THEN';
KW_TO : 'TO';
KW_TRUE : 'TRUE';
KW_TRY_CAST : 'TRY_CAST';
KW_TRUNCATE : 'TRUNCATE';
KW_UNCACHED : 'UNCACHED';
KW_UESCAPE : 'UESCAPE';
KW_UNBOUNDED : 'UNBOUNDED';
KW_UNION : 'UNION';
KW_UNNEST : 'UNNEST';
KW_UNSET : 'UNSET';
KW_USE : 'USE';
KW_USER : 'USER';
KW_USING : 'USING';
KW_UPDATE_FN : 'UPDATE_FN';
KW_UPSERT : 'UPSERT';
KW_UNKNOWN : 'UNKNOWN';
KW_URI : 'URI';
KW_VALUE : 'VALUE';
KW_VALUES : 'VALUES';
KW_VIEW : 'VIEW';
KW_VIEWS : 'VIEWS';
KW_WHEN : 'WHEN';
KW_WHERE : 'WHERE';
KW_WITH : 'WITH';
KW_YEAR : 'YEAR';
KW_YEARS : 'YEARS';
KW_TEXTFILE : 'TEXTFILE';
KW_ORC : 'ORC';
KW_AVRO : 'AVRO';
KW_SEQUENCEFILE : 'SEQUENCEFILE';
KW_RCFILE : 'RCFILE';
KW_REFERENCES : 'REFERENCES';
KW_NOVALIDATE : 'NOVALIDATE';
KW_RELY : 'RELY';
KW_ADD : 'ADD';
KW_ALL : 'ALL';
KW_ANALYTIC : 'ANALYTIC';
KW_ALTER : 'ALTER';
KW_AND : 'AND';
KW_ANY : 'ANY';
KW_ANTI : 'ANTI';
KW_ARRAY : 'ARRAY';
KW_AS : 'AS';
KW_ASC : 'ASC';
KW_AT : 'AT';
KW_AGGREGATE : 'AGGREGATE';
KW_AUTHORIZATION : 'AUTHORIZATION';
KW_BERNOULLI : 'BERNOULLI';
KW_BETWEEN : 'BETWEEN';
KW_BLOCK_SIZE : 'BLOCK_SIZE';
KW_PARTITIONED : 'PARTITIONED';
KW_PREPARE_FN : 'PREPARE_FN';
KW_EXTERNAL : 'EXTERNAL';
KW_CLOSEFN : 'CLOSEFN';
KW_SORT : 'SORT';
KW_PURGE : 'PURGE';
KW_STORED : 'STORED';
KW_LOCATION : 'LOCATION';
KW_TBLPROPERTIES : 'TBLPROPERTIES';
KW_BY : 'BY';
KW_CASCADE : 'CASCADE';
KW_CASE : 'CASE';
KW_CAST : 'CAST';
KW_CACHED : 'CACHED';
KW_CHANGE : 'CHANGE';
KW_COLUMN : 'COLUMN';
KW_COLUMNS : 'COLUMNS';
KW_COMMENT : 'COMMENT';
KW_COMPRESSION : 'COMPRESSION';
KW_COMPUTE : 'COMPUTE';
KW_CREATE : 'CREATE';
KW_CROSS : 'CROSS';
KW_CURRENT : 'CURRENT';
KW_CURRENT_DATE : 'CURRENT_DATE';
KW_CURRENT_PATH : 'CURRENT_PATH';
KW_CURRENT_TIME : 'CURRENT_TIME';
KW_CURRENT_TIMESTAMP : 'CURRENT_TIMESTAMP';
KW_CURRENT_USER : 'CURRENT_USER';
KW_DATA : 'DATA';
KW_DATABASE : 'DATABASE';
KW_DATABASES : 'DATABASES';
KW_DAY : 'DAY';
KW_DAYS : 'DAYS';
KW_DELETE : 'DELETE';
KW_DEFAULT : 'DEFAULT';
KW_DELIMITED : 'DELIMITED';
KW_DISABLE : 'DISABLE';
KW_UPDATE : 'UPDATE';
KW_DESC : 'DESC';
KW_DESCRIBE : 'DESCRIBE';
KW_DISTINCT : 'DISTINCT';
KW_DROP : 'DROP';
KW_ELSE : 'ELSE';
KW_ENCODING : 'ENCODING';
KW_END : 'END';
KW_ESCAPE : 'ESCAPE';
KW_ESCAPED : 'ESCAPED';
KW_EXCEPT : 'EXCEPT';
KW_EXCLUDING : 'EXCLUDING';
KW_EXISTS : 'EXISTS';
KW_EXPLAIN : 'EXPLAIN';
KW_EXTRACT : 'EXTRACT';
KW_EXTENDED : 'EXTENDED';
KW_FALSE : 'FALSE';
KW_FIELDS : 'FIELDS';
KW_FILEFORMAT : 'FILEFORMAT';
KW_FILES : 'FILES';
KW_FILTER : 'FILTER';
KW_FIRST : 'FIRST';
KW_FINALIZE_FN : 'FINALIZE_FN';
KW_FOLLOWING : 'FOLLOWING';
KW_FOR : 'FOR';
KW_FORMAT : 'FORMAT';
KW_FORMATTED : 'FORMATTED';
KW_FOREIGN : 'FOREIGN';
KW_FROM : 'FROM';
KW_FULL : 'FULL';
KW_FUNCTION : 'FUNCTION';
KW_FUNCTIONS : 'FUNCTIONS';
KW_GRANT : 'GRANT';
KW_GROUP : 'GROUP';
KW_GROUPING : 'GROUPING';
KW_HASH : 'HASH';
KW_HAVING : 'HAVING';
KW_HOUR : 'HOUR';
KW_HOURS : 'HOURS';
KW_IF : 'IF';
KW_IN : 'IN';
KW_INCLUDING : 'INCLUDING';
KW_INCREMENTAL : 'INCREMENTAL';
KW_INNER : 'INNER';
KW_INPATH : 'INPATH';
KW_INSERT : 'INSERT';
KW_INTERSECT : 'INTERSECT';
KW_INTERVAL : 'INTERVAL';
KW_INTERMEDIATE : 'INTERMEDIATE';
KW_INTO : 'INTO';
KW_INIT_FN : 'INIT_FN';
KW_IREGEXP : 'IREGEXP';
KW_ILIKE : 'ILIKE';
KW_INVALIDATE : 'INVALIDATE';
KW_IS : 'IS';
KW_JOIN : 'JOIN';
KW_KEY : 'KEY';
KW_KUDU : 'KUDU';
KW_LAST : 'LAST';
KW_LATERAL : 'LATERAL';
KW_LEFT : 'LEFT';
KW_LIKE : 'LIKE';
KW_LIMIT : 'LIMIT';
KW_LINES : 'LINES';
KW_LOAD : 'LOAD';
KW_LOCALTIME : 'LOCALTIME';
KW_LOCALTIMESTAMP : 'LOCALTIMESTAMP';
KW_METADATA : 'METADATA';
KW_MAP : 'MAP';
KW_MINUTE : 'MINUTE';
KW_MINUTES : 'MINUTES';
KW_MONTH : 'MONTH';
KW_MONTHS : 'MONTHS';
KW_MERGE_FN : 'MERGE_FN';
KW_NFC : 'NFC';
KW_NFD : 'NFD';
KW_NFKC : 'NFKC';
KW_NFKD : 'NFKD';
KW_NORMALIZE : 'NORMALIZE';
KW_NOT : 'NOT';
KW_NULL : 'NULL';
KW_NULLS : 'NULLS';
KW_OFFSET : 'OFFSET';
KW_ON : 'ON';
KW_OPTION : 'OPTION';
KW_OR : 'OR';
KW_ORDER : 'ORDER';
KW_ORDINALITY : 'ORDINALITY';
KW_OUTER : 'OUTER';
KW_OWNER : 'OWNER';
KW_OVER : 'OVER';
KW_OVERWRITE : 'OVERWRITE';
KW_PARTITION : 'PARTITION';
KW_PARTITIONS : 'PARTITIONS';
KW_PARQUET : 'PARQUET';
KW_POSITION : 'POSITION';
KW_PRECEDING : 'PRECEDING';
KW_PRIMARY : 'PRIMARY';
KW_REPLICATION : 'REPLICATION';
KW_PRIVILEGES : 'PRIVILEGES';
KW_PROPERTIES : 'PROPERTIES';
KW_RANGE : 'RANGE';
KW_RECOVER : 'RECOVER';
KW_RENAME : 'RENAME';
KW_REPEATABLE : 'REPEATABLE';
KW_REPLACE : 'REPLACE';
KW_RESTRICT : 'RESTRICT';
KW_RETURNS : 'RETURNS';
KW_REVOKE : 'REVOKE';
KW_REFRESH : 'REFRESH';
KW_REGEXP : 'REGEXP';
KW_RLIKE : 'RLIKE';
KW_RIGHT : 'RIGHT';
KW_ROLE : 'ROLE';
KW_ROLES : 'ROLES';
KW_ROW : 'ROW';
KW_ROWS : 'ROWS';
KW_SCHEMA : 'SCHEMA';
KW_SCHEMAS : 'SCHEMAS';
KW_SECOND : 'SECOND';
KW_SECONDS : 'SECONDS';
KW_SELECT : 'SELECT';
KW_SERDEPROPERTIES : 'SERDEPROPERTIES';
KW_SET : 'SET';
KW_SEMI : 'SEMI';
KW_SERVER : 'SERVER';
KW_SHOW : 'SHOW';
KW_SHUTDOWN : 'SHUTDOWN';
KW_SOME : 'SOME';
KW_STATS : 'STATS';
KW_STRUCT : 'STRUCT';
KW_STRAIGHT_JOIN : 'STRAIGHT_JOIN';
KW_SUBSTRING : 'SUBSTRING';
KW_SYSTEM : 'SYSTEM';
KW_SYMBOL : 'SYMBOL';
KW_SERIALIZE_FN : 'SERIALIZE_FN';
KW_TABLE : 'TABLE';
KW_TABLES : 'TABLES';
KW_TABLESAMPLE : 'TABLESAMPLE';
KW_TERMINATED : 'TERMINATED ';
KW_THEN : 'THEN';
KW_TO : 'TO';
KW_TRUE : 'TRUE';
KW_TRY_CAST : 'TRY_CAST';
KW_TRUNCATE : 'TRUNCATE';
KW_UNCACHED : 'UNCACHED';
KW_UESCAPE : 'UESCAPE';
KW_UNBOUNDED : 'UNBOUNDED';
KW_UNION : 'UNION';
KW_UNNEST : 'UNNEST';
KW_UNSET : 'UNSET';
KW_USE : 'USE';
KW_USER : 'USER';
KW_USING : 'USING';
KW_UPDATE_FN : 'UPDATE_FN';
KW_UPSERT : 'UPSERT';
KW_UNKNOWN : 'UNKNOWN';
KW_URI : 'URI';
KW_VALUE : 'VALUE';
KW_VALUES : 'VALUES';
KW_VIEW : 'VIEW';
KW_VIEWS : 'VIEWS';
KW_WHEN : 'WHEN';
KW_WHERE : 'WHERE';
KW_WITH : 'WITH';
KW_YEAR : 'YEAR';
KW_YEARS : 'YEARS';
KW_TEXTFILE : 'TEXTFILE';
KW_ORC : 'ORC';
KW_AVRO : 'AVRO';
KW_SEQUENCEFILE : 'SEQUENCEFILE';
KW_RCFILE : 'RCFILE';
KW_REFERENCES : 'REFERENCES';
KW_NOVALIDATE : 'NOVALIDATE';
KW_RELY : 'RELY';
STATS_NUMDVS : '\'NUMDVS\'';
STATS_NUMNULLS : '\'NUMNULLS\'';
STATS_AVGSIZE : '\'AVGSIZE\'';
STATS_MAXSIZE : '\'MAXSIZE\'';
EQ : '=';
NEQ : '<>' | '!=';
LT : '<';
LTE : '<=';
GT : '>';
GTE : '>=';
PLUS: '+';
MINUS: '-';
ASTERISK: '*';
SLASH: '/';
PERCENT: '%';
CONCAT: '||';
EQ : '=';
NEQ : '<>' | '!=';
LT : '<';
LTE : '<=';
GT : '>';
GTE : '>=';
PLUS : '+';
MINUS : '-';
ASTERISK : '*';
SLASH : '/';
PERCENT : '%';
CONCAT : '||';
DOT : '.';
SEMICOLON: ';';
COMMA: ',';
COLON : ':' ;
LPAREN : '(' ;
RPAREN : ')' ;
LSQUARE : '[' ;
RSQUARE : ']' ;
LCURLY : '{';
RCURLY : '}';
DOT : '.';
SEMICOLON : ';';
COMMA : ',';
COLON : ':';
LPAREN : '(';
RPAREN : ')';
LSQUARE : '[';
RSQUARE : ']';
LCURLY : '{';
RCURLY : '}';
BITWISEOR : '|';
QUESTION : '?';
QUESTION : '?';
RIGHT_ARROW : '->';
RIGHT_ARROW: '->';
STRING
: '\'' ( ~'\'' | '\'\'' )* '\''
| '"' ( ~'"' | '""' )* '"'
;
STRING: '\'' ( ~'\'' | '\'\'')* '\'' | '"' ( ~'"' | '""')* '"';
UNICODE_STRING
: 'U&\'' ( ~'\'' | '\'\'' )* '\''
;
UNICODE_STRING: 'U&\'' ( ~'\'' | '\'\'')* '\'';
// Note: we allow any character inside the binary literal and validate
// its a correct literal when the AST is being constructed. This
// allows us to provide more meaningful error messages to the user
BINARY_LITERAL
: 'X\'' (~'\'')* '\''
;
BINARY_LITERAL: 'X\'' (~'\'')* '\'';
INTEGER_VALUE
: DIGIT+
;
INTEGER_VALUE: DIGIT+;
DECIMAL_VALUE
: DIGIT+ '.' DIGIT*
| '.' DIGIT+
;
DECIMAL_VALUE: DIGIT+ '.' DIGIT* | '.' DIGIT+;
DOUBLE_VALUE
: DIGIT+ ('.' DIGIT*)? EXPONENT
| '.' DIGIT+ EXPONENT
;
DOUBLE_VALUE: DIGIT+ ('.' DIGIT*)? EXPONENT | '.' DIGIT+ EXPONENT;
IDENTIFIER
: (LETTER | '_') (LETTER | DIGIT | '_' | '@' | ':')*
;
IDENTIFIER: (LETTER | '_') (LETTER | DIGIT | '_' | '@' | ':')*;
DIGIT_IDENTIFIER
: DIGIT (LETTER | DIGIT | '_' | '@' | ':')+
;
DIGIT_IDENTIFIER: DIGIT (LETTER | DIGIT | '_' | '@' | ':')+;
QUOTED_IDENTIFIER
: '"' ( ~'"' | '""' )* '"'
;
QUOTED_IDENTIFIER: '"' ( ~'"' | '""')* '"';
BACKQUOTED_IDENTIFIER
: '`' ( ~'`' | '``' )* '`'
;
BACKQUOTED_IDENTIFIER: '`' ( ~'`' | '``')* '`';
TIME_WITH_TIME_ZONE
: 'TIME' WS 'WITH' WS 'TIME' WS 'ZONE'
;
TIME_WITH_TIME_ZONE: 'TIME' WS 'WITH' WS 'TIME' WS 'ZONE';
TIMESTAMP_WITH_TIME_ZONE
: 'TIMESTAMP' WS 'WITH' WS 'TIME' WS 'ZONE'
;
TIMESTAMP_WITH_TIME_ZONE: 'TIMESTAMP' WS 'WITH' WS 'TIME' WS 'ZONE';
DOUBLE_PRECISION
: 'DOUBLE' WS 'PRECISION'
;
DOUBLE_PRECISION: 'DOUBLE' WS 'PRECISION';
fragment EXPONENT
: 'E' [+-]? DIGIT+
;
fragment EXPONENT: 'E' [+-]? DIGIT+;
fragment DIGIT
: [0-9]
;
fragment DIGIT: [0-9];
fragment LETTER
: [A-Z]
;
fragment LETTER: [A-Z];
SIMPLE_COMMENT
: '--' ~[\r\n]* '\r'? '\n'? -> channel(HIDDEN)
;
SIMPLE_COMMENT: '--' ~[\r\n]* '\r'? '\n'? -> channel(HIDDEN);
BRACKETED_COMMENT
: '/*' .*? '*/' -> channel(HIDDEN)
;
BRACKETED_COMMENT: '/*' .*? '*/' -> channel(HIDDEN);
WS
: [ \r\n\t]+ -> channel(HIDDEN)
;
WS: [ \r\n\t]+ -> channel(HIDDEN);

881
src/grammar/impala/ImpalaSqlParser.g4
View File

File diff suppressed because it is too large Load Diff

1918
src/grammar/mysql/MySqlLexer.g4
View File

File diff suppressed because it is too large Load Diff

2983
src/grammar/mysql/MySqlParser.g4
View File

File diff suppressed because it is too large Load Diff

1295
src/grammar/pgsql/PostgreSQLLexer.g4
View File

File diff suppressed because it is too large Load Diff

8892
src/grammar/pgsql/PostgreSQLParser.g4
View File

File diff suppressed because it is too large Load Diff

4752
src/grammar/plsql/PlSqlLexer.g4
View File

File diff suppressed because it is too large Load Diff

1963
src/grammar/plsql/PlSqlParser.g4
View File

File diff suppressed because it is too large Load Diff

833
src/grammar/spark/SparkSqlLexer.g4
View File

@ -18,6 +18,11 @@
* Reference: https://github.com/apache/spark/blob/master/sql/api/src/main/antlr4/org/apache/spark/sql/catalyst/parser/SqlBaseLexer.g4
*/
// $antlr-format alignTrailingComments true, columnLimit 150, maxEmptyLinesToKeep 1, reflowComments false, useTab false
// $antlr-format allowShortRulesOnASingleLine true, allowShortBlocksOnASingleLine true, minEmptyLines 0, alignSemicolons none, alignColons trailing
// $antlr-format singleLineOverrulesHangingColon true, alignLexerCommands true, alignLabels true, alignTrailers true
// $antlr-format spaceBeforeAssignmentOperators false, groupedAlignments true
lexer grammar SparkSqlLexer;
@members {
@ -39,12 +44,12 @@ lexer grammar SparkSqlLexer;
SEMICOLON: ';';
LEFT_PAREN: '(';
RIGHT_PAREN: ')';
COMMA: ',';
DOT: '.';
LEFT_BRACKET: '[';
RIGHT_BRACKET: ']';
LEFT_PAREN : '(';
RIGHT_PAREN : ')';
COMMA : ',';
DOT : '.';
LEFT_BRACKET : '[';
RIGHT_BRACKET : ']';
// NOTE: If you add a new token in the list below, you should update the list of keywords
// and reserved tag in `docs/sql-ref-ansi-compliance.md#sql-keywords`, and
@ -54,468 +59,420 @@ RIGHT_BRACKET: ']';
// Start of the keywords list
//============================
//--SPARK-KEYWORD-LIST-START
KW_ADD: 'ADD';
KW_AFTER: 'AFTER';
KW_ALL: 'ALL';
KW_ALTER: 'ALTER';
KW_ALWAYS: 'ALWAYS';
KW_ANALYZE: 'ANALYZE';
KW_AND: 'AND';
KW_ANTI: 'ANTI';
KW_ANY: 'ANY';
KW_ANY_VALUE: 'ANY_VALUE';
KW_ARCHIVE: 'ARCHIVE';
KW_ARRAY: 'ARRAY';
KW_AS: 'AS';
KW_ASC: 'ASC';
KW_AT: 'AT';
KW_AUTHORIZATION: 'AUTHORIZATION';
KW_BETWEEN: 'BETWEEN';
KW_BIGINT: 'BIGINT';
KW_BINARY: 'BINARY';
KW_BOOLEAN: 'BOOLEAN';
KW_BOTH: 'BOTH';
KW_BUCKET: 'BUCKET';
KW_BUCKETS: 'BUCKETS';
KW_BY: 'BY';
KW_BYTE: 'BYTE';
KW_CACHE: 'CACHE';
KW_CASCADE: 'CASCADE';
KW_CASE: 'CASE';
KW_CAST: 'CAST';
KW_CATALOG: 'CATALOG';
KW_CATALOGS: 'CATALOGS';
KW_CHANGE: 'CHANGE';
KW_CHAR: 'CHAR';
KW_CHARACTER: 'CHARACTER';
KW_CHECK: 'CHECK';
KW_CLEAR: 'CLEAR';
KW_CLUSTER: 'CLUSTER';
KW_CLUSTERED: 'CLUSTERED';
KW_CODEGEN: 'CODEGEN';
KW_COLLATE: 'COLLATE';
KW_COLLECTION: 'COLLECTION';
KW_COLUMN: 'COLUMN';
KW_COLUMNS: 'COLUMNS';
KW_COMMENT: 'COMMENT';
KW_COMMIT: 'COMMIT';
KW_COMPACT: 'COMPACT';
KW_COMPACTIONS: 'COMPACTIONS';
KW_COMPUTE: 'COMPUTE';
KW_CONCATENATE: 'CONCATENATE';
KW_CONSTRAINT: 'CONSTRAINT';
KW_COST: 'COST';
KW_CREATE: 'CREATE';
KW_CROSS: 'CROSS';
KW_CUBE: 'CUBE';
KW_CURRENT: 'CURRENT';
KW_CURRENT_DATE: 'CURRENT_DATE';
KW_CURRENT_TIME: 'CURRENT_TIME';
KW_CURRENT_TIMESTAMP: 'CURRENT_TIMESTAMP';
KW_CURRENT_USER: 'CURRENT_USER';
KW_DAY: 'DAY';
KW_DAYS: 'DAYS';
KW_DAYOFYEAR: 'DAYOFYEAR';
KW_DATA: 'DATA';
KW_DATE: 'DATE';
KW_DATABASE: 'DATABASE';
KW_DATABASES: 'DATABASES';
KW_DATEADD: 'DATEADD';
KW_DATE_ADD: 'DATE_ADD';
KW_DATEDIFF: 'DATEDIFF';
KW_DATE_DIFF: 'DATE_DIFF';
KW_DBPROPERTIES: 'DBPROPERTIES';
KW_DEC: 'DEC';
KW_DECIMAL: 'DECIMAL';
KW_DECLARE: 'DECLARE';
KW_DEFAULT: 'DEFAULT';
KW_DEFINED: 'DEFINED';
KW_DELETE: 'DELETE';
KW_DELIMITED: 'DELIMITED';
KW_DESC: 'DESC';
KW_DESCRIBE: 'DESCRIBE';
KW_DFS: 'DFS';
KW_DIRECTORIES: 'DIRECTORIES';
KW_DIRECTORY: 'DIRECTORY';
KW_DISTINCT: 'DISTINCT';
KW_DISTRIBUTE: 'DISTRIBUTE';
KW_DIV: 'DIV';
KW_DOUBLE: 'DOUBLE';
KW_DROP: 'DROP';
KW_ELSE: 'ELSE';
KW_END: 'END';
KW_ESCAPE: 'ESCAPE';
KW_ESCAPED: 'ESCAPED';
KW_EXCEPT: 'EXCEPT';
KW_EXCHANGE: 'EXCHANGE';
KW_EXCLUDE: 'EXCLUDE';
KW_EXISTS: 'EXISTS';
KW_EXPLAIN: 'EXPLAIN';
KW_EXPORT: 'EXPORT';
KW_EXTENDED: 'EXTENDED';
KW_EXTERNAL: 'EXTERNAL';
KW_EXTRACT: 'EXTRACT';
KW_FALSE: 'FALSE';
KW_FETCH: 'FETCH';
KW_FIELDS: 'FIELDS';
KW_FILTER: 'FILTER';
KW_FILEFORMAT: 'FILEFORMAT';
KW_FIRST: 'FIRST';
KW_FLOAT: 'FLOAT';
KW_FOLLOWING: 'FOLLOWING';
KW_FOR: 'FOR';
KW_FOREIGN: 'FOREIGN';
KW_FORMAT: 'FORMAT';
KW_FORMATTED: 'FORMATTED';
KW_FROM: 'FROM';
KW_FULL: 'FULL';
KW_FUNCTION: 'FUNCTION';
KW_FUNCTIONS: 'FUNCTIONS';
KW_GENERATED: 'GENERATED';
KW_GLOBAL: 'GLOBAL';
KW_GRANT: 'GRANT';
KW_GROUP: 'GROUP';
KW_GROUPING: 'GROUPING';
KW_HAVING: 'HAVING';
KW_BINARY_HEX: 'X';
KW_HOUR: 'HOUR';
KW_HOURS: 'HOURS';
KW_IDENTIFIER: 'IDENTIFIER';
KW_IF: 'IF';
KW_IGNORE: 'IGNORE';
KW_IMPORT: 'IMPORT';
KW_IN: 'IN';
KW_INCLUDE: 'INCLUDE';
KW_INDEX: 'INDEX';
KW_INDEXES: 'INDEXES';
KW_INNER: 'INNER';
KW_INPATH: 'INPATH';
KW_INPUTFORMAT: 'INPUTFORMAT';
KW_INSERT: 'INSERT';
KW_INTERSECT: 'INTERSECT';
KW_INTERVAL: 'INTERVAL';
KW_INT: 'INT';
KW_INTEGER: 'INTEGER';
KW_INTO: 'INTO';
KW_IS: 'IS';
KW_ITEMS: 'ITEMS';
KW_JOIN: 'JOIN';
KW_KEYS: 'KEYS';
KW_LAST: 'LAST';
KW_LATERAL: 'LATERAL';
KW_LAZY: 'LAZY';
KW_LEADING: 'LEADING';
KW_LEFT: 'LEFT';
KW_LIKE: 'LIKE';
KW_ILIKE: 'ILIKE';
KW_LIMIT: 'LIMIT';
KW_LINES: 'LINES';
KW_LIST: 'LIST';
KW_LOAD: 'LOAD';
KW_LOCAL: 'LOCAL';
KW_LOCATION: 'LOCATION';
KW_LOCK: 'LOCK';
KW_LOCKS: 'LOCKS';
KW_LOGICAL: 'LOGICAL';
KW_LONG: 'LONG';
KW_MACRO: 'MACRO';
KW_MAP: 'MAP';
KW_MATCHED: 'MATCHED';
KW_MERGE: 'MERGE';
KW_MICROSECOND: 'MICROSECOND';
KW_MICROSECONDS: 'MICROSECONDS';
KW_MILLISECOND: 'MILLISECOND';
KW_MILLISECONDS: 'MILLISECONDS';
KW_MINUTE: 'MINUTE';
KW_MINUTES: 'MINUTES';
KW_MONTH: 'MONTH';
KW_MONTHS: 'MONTHS';
KW_MSCK: 'MSCK';
KW_NAME: 'NAME';
KW_NAMESPACE: 'NAMESPACE';
KW_NAMESPACES: 'NAMESPACES';
KW_NANOSECOND: 'NANOSECOND';
KW_NANOSECONDS: 'NANOSECONDS';
KW_NATURAL: 'NATURAL';
KW_NO: 'NO';
KW_NOSCAN: 'NOSCAN';
KW_NOT: 'NOT';
KW_NULL: 'NULL';
KW_NULLS: 'NULLS';
KW_NUMERIC: 'NUMERIC';
KW_OF: 'OF';
KW_OFFSET: 'OFFSET';
KW_ON: 'ON';
KW_ONLY: 'ONLY';
KW_OPTION: 'OPTION';
KW_OPTIONS: 'OPTIONS';
KW_OR: 'OR';
KW_ORDER: 'ORDER';
KW_OUT: 'OUT';
KW_OUTER: 'OUTER';
KW_OUTPUTFORMAT: 'OUTPUTFORMAT';
KW_OVER: 'OVER';
KW_OVERLAPS: 'OVERLAPS';
KW_OVERLAY: 'OVERLAY';
KW_OVERWRITE: 'OVERWRITE';
KW_PARTITION: 'PARTITION';
KW_PARTITIONED: 'PARTITIONED';
KW_PARTITIONS: 'PARTITIONS';
KW_PERCENTILE_CONT: 'PERCENTILE_CONT';
KW_PERCENTILE_DISC: 'PERCENTILE_DISC';
KW_PERCENTLIT: 'PERCENT';
KW_PIVOT: 'PIVOT';
KW_PLACING: 'PLACING';
KW_POSITION: 'POSITION';
KW_PRECEDING: 'PRECEDING';
KW_PRIMARY: 'PRIMARY';
KW_PRINCIPALS: 'PRINCIPALS';
KW_PROPERTIES: 'PROPERTIES';
KW_PURGE: 'PURGE';
KW_QUARTER: 'QUARTER';
KW_QUERY: 'QUERY';
KW_RANGE: 'RANGE';
KW_REAL: 'REAL';
KW_RECORDREADER: 'RECORDREADER';
KW_RECORDWRITER: 'RECORDWRITER';
KW_RECOVER: 'RECOVER';
KW_REDUCE: 'REDUCE';
KW_REFERENCES: 'REFERENCES';
KW_REFRESH: 'REFRESH';
KW_RENAME: 'RENAME';
KW_REPAIR: 'REPAIR';
KW_REPEATABLE: 'REPEATABLE';
KW_REPLACE: 'REPLACE';
KW_RESET: 'RESET';
KW_RESPECT: 'RESPECT';
KW_RESTRICT: 'RESTRICT';
KW_REVOKE: 'REVOKE';
KW_RIGHT: 'RIGHT';
KW_RLIKE: 'RLIKE';
KW_REGEXP: 'REGEXP';
KW_ROLE: 'ROLE';
KW_ROLES: 'ROLES';
KW_ROLLBACK: 'ROLLBACK';
KW_ROLLUP: 'ROLLUP';
KW_ROW: 'ROW';
KW_ROWS: 'ROWS';
KW_SECOND: 'SECOND';
KW_SECONDS: 'SECONDS';
KW_SCHEMA: 'SCHEMA';
KW_SCHEMAS: 'SCHEMAS';
KW_SELECT: 'SELECT';
KW_SEMI: 'SEMI';
KW_SEPARATED: 'SEPARATED';
KW_SERDE: 'SERDE';
KW_SERDEPROPERTIES: 'SERDEPROPERTIES';
KW_SESSION_USER: 'SESSION_USER';
KW_SET: 'SET';
KW_MINUS: 'MINUS';
KW_SETS: 'SETS';
KW_SHORT: 'SHORT';
KW_SHOW: 'SHOW';
KW_SINGLE: 'SINGLE';
KW_SKEWED: 'SKEWED';
KW_SMALLINT: 'SMALLINT';
KW_SOME: 'SOME';
KW_SORT: 'SORT';
KW_SORTED: 'SORTED';
KW_SOURCE: 'SOURCE';
KW_START: 'START';
KW_STATISTICS: 'STATISTICS';
KW_STORED: 'STORED';
KW_STRATIFY: 'STRATIFY';
KW_STRING: 'STRING';
KW_STRUCT: 'STRUCT';
KW_SUBSTR: 'SUBSTR';
KW_SUBSTRING: 'SUBSTRING';
KW_SYNC: 'SYNC';
KW_SYSTEM: 'SYSTEM';
KW_SYSTEM_TIME: 'SYSTEM_TIME';
KW_SYSTEM_VERSION: 'SYSTEM_VERSION';
KW_TABLE: 'TABLE';
KW_TABLES: 'TABLES';
KW_TABLESAMPLE: 'TABLESAMPLE';
KW_TARGET: 'TARGET';
KW_TBLPROPERTIES: 'TBLPROPERTIES';
KW_TEMPORARY: 'TEMPORARY';
KW_TERMINATED: 'TERMINATED';
KW_THEN: 'THEN';
KW_TIME: 'TIME';
KW_TIMEDIFF: 'TIMEDIFF';
KW_TIMESTAMP: 'TIMESTAMP';
KW_TIMESTAMP_LTZ: 'TIMESTAMP_LTZ';
KW_TIMESTAMP_NTZ: 'TIMESTAMP_NTZ';
KW_TIMESTAMPADD: 'TIMESTAMPADD';
KW_TIMESTAMPDIFF: 'TIMESTAMPDIFF';
KW_TINYINT: 'TINYINT';
KW_TO: 'TO';
KW_TOUCH: 'TOUCH';
KW_TRAILING: 'TRAILING';
KW_TRANSACTION: 'TRANSACTION';
KW_TRANSACTIONS: 'TRANSACTIONS';
KW_TRANSFORM: 'TRANSFORM';
KW_TRIM: 'TRIM';
KW_TRUE: 'TRUE';
KW_TRUNCATE: 'TRUNCATE';
KW_TRY_CAST: 'TRY_CAST';
KW_TYPE: 'TYPE';
KW_UNARCHIVE: 'UNARCHIVE';
KW_UNBOUNDED: 'UNBOUNDED';
KW_UNCACHE: 'UNCACHE';
KW_UNION: 'UNION';
KW_UNIQUE: 'UNIQUE';
KW_UNKNOWN: 'UNKNOWN';
KW_UNLOCK: 'UNLOCK';
KW_UNPIVOT: 'UNPIVOT';
KW_UNSET: 'UNSET';
KW_UPDATE: 'UPDATE';
KW_USE: 'USE';
KW_USER: 'USER';
KW_USING: 'USING';
KW_VALUES: 'VALUES';
KW_VARCHAR: 'VARCHAR';
KW_VAR: 'VAR';
KW_VARIABLE: 'VARIABLE';
KW_VERSION: 'VERSION';
KW_VIEW: 'VIEW';
KW_VIEWS: 'VIEWS';
KW_VOID: 'VOID';
KW_WEEK: 'WEEK';
KW_WEEKS: 'WEEKS';
KW_WHEN: 'WHEN';
KW_WHERE: 'WHERE';
KW_WINDOW: 'WINDOW';
KW_WITH: 'WITH';
KW_WITHIN: 'WITHIN';
KW_YEAR: 'YEAR';
KW_YEARS: 'YEARS';
KW_ZONE: 'ZONE';
KW_ADD : 'ADD';
KW_AFTER : 'AFTER';
KW_ALL : 'ALL';
KW_ALTER : 'ALTER';
KW_ALWAYS : 'ALWAYS';
KW_ANALYZE : 'ANALYZE';
KW_AND : 'AND';
KW_ANTI : 'ANTI';
KW_ANY : 'ANY';
KW_ANY_VALUE : 'ANY_VALUE';
KW_ARCHIVE : 'ARCHIVE';
KW_ARRAY : 'ARRAY';
KW_AS : 'AS';
KW_ASC : 'ASC';
KW_AT : 'AT';
KW_AUTHORIZATION : 'AUTHORIZATION';
KW_BETWEEN : 'BETWEEN';
KW_BIGINT : 'BIGINT';
KW_BINARY : 'BINARY';
KW_BOOLEAN : 'BOOLEAN';
KW_BOTH : 'BOTH';
KW_BUCKET : 'BUCKET';
KW_BUCKETS : 'BUCKETS';
KW_BY : 'BY';
KW_BYTE : 'BYTE';
KW_CACHE : 'CACHE';
KW_CASCADE : 'CASCADE';
KW_CASE : 'CASE';
KW_CAST : 'CAST';
KW_CATALOG : 'CATALOG';
KW_CATALOGS : 'CATALOGS';
KW_CHANGE : 'CHANGE';
KW_CHAR : 'CHAR';
KW_CHARACTER : 'CHARACTER';
KW_CHECK : 'CHECK';
KW_CLEAR : 'CLEAR';
KW_CLUSTER : 'CLUSTER';
KW_CLUSTERED : 'CLUSTERED';
KW_CODEGEN : 'CODEGEN';
KW_COLLATE : 'COLLATE';
KW_COLLECTION : 'COLLECTION';
KW_COLUMN : 'COLUMN';
KW_COLUMNS : 'COLUMNS';
KW_COMMENT : 'COMMENT';
KW_COMMIT : 'COMMIT';
KW_COMPACT : 'COMPACT';
KW_COMPACTIONS : 'COMPACTIONS';
KW_COMPUTE : 'COMPUTE';
KW_CONCATENATE : 'CONCATENATE';
KW_CONSTRAINT : 'CONSTRAINT';
KW_COST : 'COST';
KW_CREATE : 'CREATE';
KW_CROSS : 'CROSS';
KW_CUBE : 'CUBE';
KW_CURRENT : 'CURRENT';
KW_CURRENT_DATE : 'CURRENT_DATE';
KW_CURRENT_TIME : 'CURRENT_TIME';
KW_CURRENT_TIMESTAMP : 'CURRENT_TIMESTAMP';
KW_CURRENT_USER : 'CURRENT_USER';
KW_DAY : 'DAY';
KW_DAYS : 'DAYS';
KW_DAYOFYEAR : 'DAYOFYEAR';
KW_DATA : 'DATA';
KW_DATE : 'DATE';
KW_DATABASE : 'DATABASE';
KW_DATABASES : 'DATABASES';
KW_DATEADD : 'DATEADD';
KW_DATE_ADD : 'DATE_ADD';
KW_DATEDIFF : 'DATEDIFF';
KW_DATE_DIFF : 'DATE_DIFF';
KW_DBPROPERTIES : 'DBPROPERTIES';
KW_DEC : 'DEC';
KW_DECIMAL : 'DECIMAL';
KW_DECLARE : 'DECLARE';
KW_DEFAULT : 'DEFAULT';
KW_DEFINED : 'DEFINED';
KW_DELETE : 'DELETE';
KW_DELIMITED : 'DELIMITED';
KW_DESC : 'DESC';
KW_DESCRIBE : 'DESCRIBE';
KW_DFS : 'DFS';
KW_DIRECTORIES : 'DIRECTORIES';
KW_DIRECTORY : 'DIRECTORY';
KW_DISTINCT : 'DISTINCT';
KW_DISTRIBUTE : 'DISTRIBUTE';
KW_DIV : 'DIV';
KW_DOUBLE : 'DOUBLE';
KW_DROP : 'DROP';
KW_ELSE : 'ELSE';
KW_END : 'END';
KW_ESCAPE : 'ESCAPE';
KW_ESCAPED : 'ESCAPED';
KW_EXCEPT : 'EXCEPT';
KW_EXCHANGE : 'EXCHANGE';
KW_EXCLUDE : 'EXCLUDE';
KW_EXISTS : 'EXISTS';
KW_EXPLAIN : 'EXPLAIN';
KW_EXPORT : 'EXPORT';
KW_EXTENDED : 'EXTENDED';
KW_EXTERNAL : 'EXTERNAL';
KW_EXTRACT : 'EXTRACT';
KW_FALSE : 'FALSE';
KW_FETCH : 'FETCH';
KW_FIELDS : 'FIELDS';
KW_FILTER : 'FILTER';
KW_FILEFORMAT : 'FILEFORMAT';
KW_FIRST : 'FIRST';
KW_FLOAT : 'FLOAT';
KW_FOLLOWING : 'FOLLOWING';
KW_FOR : 'FOR';
KW_FOREIGN : 'FOREIGN';
KW_FORMAT : 'FORMAT';
KW_FORMATTED : 'FORMATTED';
KW_FROM : 'FROM';
KW_FULL : 'FULL';
KW_FUNCTION : 'FUNCTION';
KW_FUNCTIONS : 'FUNCTIONS';
KW_GENERATED : 'GENERATED';
KW_GLOBAL : 'GLOBAL';
KW_GRANT : 'GRANT';
KW_GROUP : 'GROUP';
KW_GROUPING : 'GROUPING';
KW_HAVING : 'HAVING';
KW_BINARY_HEX : 'X';
KW_HOUR : 'HOUR';
KW_HOURS : 'HOURS';
KW_IDENTIFIER : 'IDENTIFIER';
KW_IF : 'IF';
KW_IGNORE : 'IGNORE';
KW_IMPORT : 'IMPORT';
KW_IN : 'IN';
KW_INCLUDE : 'INCLUDE';
KW_INDEX : 'INDEX';
KW_INDEXES : 'INDEXES';
KW_INNER : 'INNER';
KW_INPATH : 'INPATH';
KW_INPUTFORMAT : 'INPUTFORMAT';
KW_INSERT : 'INSERT';
KW_INTERSECT : 'INTERSECT';
KW_INTERVAL : 'INTERVAL';
KW_INT : 'INT';
KW_INTEGER : 'INTEGER';
KW_INTO : 'INTO';
KW_IS : 'IS';
KW_ITEMS : 'ITEMS';
KW_JOIN : 'JOIN';
KW_KEYS : 'KEYS';
KW_LAST : 'LAST';
KW_LATERAL : 'LATERAL';
KW_LAZY : 'LAZY';
KW_LEADING : 'LEADING';
KW_LEFT : 'LEFT';
KW_LIKE : 'LIKE';
KW_ILIKE : 'ILIKE';
KW_LIMIT : 'LIMIT';
KW_LINES : 'LINES';
KW_LIST : 'LIST';
KW_LOAD : 'LOAD';
KW_LOCAL : 'LOCAL';
KW_LOCATION : 'LOCATION';
KW_LOCK : 'LOCK';
KW_LOCKS : 'LOCKS';
KW_LOGICAL : 'LOGICAL';
KW_LONG : 'LONG';
KW_MACRO : 'MACRO';
KW_MAP : 'MAP';
KW_MATCHED : 'MATCHED';
KW_MERGE : 'MERGE';
KW_MICROSECOND : 'MICROSECOND';
KW_MICROSECONDS : 'MICROSECONDS';
KW_MILLISECOND : 'MILLISECOND';
KW_MILLISECONDS : 'MILLISECONDS';
KW_MINUTE : 'MINUTE';
KW_MINUTES : 'MINUTES';
KW_MONTH : 'MONTH';
KW_MONTHS : 'MONTHS';
KW_MSCK : 'MSCK';
KW_NAME : 'NAME';
KW_NAMESPACE : 'NAMESPACE';
KW_NAMESPACES : 'NAMESPACES';
KW_NANOSECOND : 'NANOSECOND';
KW_NANOSECONDS : 'NANOSECONDS';
KW_NATURAL : 'NATURAL';
KW_NO : 'NO';
KW_NOSCAN : 'NOSCAN';
KW_NOT : 'NOT';
KW_NULL : 'NULL';
KW_NULLS : 'NULLS';
KW_NUMERIC : 'NUMERIC';
KW_OF : 'OF';
KW_OFFSET : 'OFFSET';
KW_ON : 'ON';
KW_ONLY : 'ONLY';
KW_OPTION : 'OPTION';
KW_OPTIONS : 'OPTIONS';
KW_OR : 'OR';
KW_ORDER : 'ORDER';
KW_OUT : 'OUT';
KW_OUTER : 'OUTER';
KW_OUTPUTFORMAT : 'OUTPUTFORMAT';
KW_OVER : 'OVER';
KW_OVERLAPS : 'OVERLAPS';
KW_OVERLAY : 'OVERLAY';
KW_OVERWRITE : 'OVERWRITE';
KW_PARTITION : 'PARTITION';
KW_PARTITIONED : 'PARTITIONED';
KW_PARTITIONS : 'PARTITIONS';
KW_PERCENTILE_CONT : 'PERCENTILE_CONT';
KW_PERCENTILE_DISC : 'PERCENTILE_DISC';
KW_PERCENTLIT : 'PERCENT';
KW_PIVOT : 'PIVOT';
KW_PLACING : 'PLACING';
KW_POSITION : 'POSITION';
KW_PRECEDING : 'PRECEDING';
KW_PRIMARY : 'PRIMARY';
KW_PRINCIPALS : 'PRINCIPALS';
KW_PROPERTIES : 'PROPERTIES';
KW_PURGE : 'PURGE';
KW_QUARTER : 'QUARTER';
KW_QUERY : 'QUERY';
KW_RANGE : 'RANGE';
KW_REAL : 'REAL';
KW_RECORDREADER : 'RECORDREADER';
KW_RECORDWRITER : 'RECORDWRITER';
KW_RECOVER : 'RECOVER';
KW_REDUCE : 'REDUCE';
KW_REFERENCES : 'REFERENCES';
KW_REFRESH : 'REFRESH';
KW_RENAME : 'RENAME';
KW_REPAIR : 'REPAIR';
KW_REPEATABLE : 'REPEATABLE';
KW_REPLACE : 'REPLACE';
KW_RESET : 'RESET';
KW_RESPECT : 'RESPECT';
KW_RESTRICT : 'RESTRICT';
KW_REVOKE : 'REVOKE';
KW_RIGHT : 'RIGHT';
KW_RLIKE : 'RLIKE';
KW_REGEXP : 'REGEXP';
KW_ROLE : 'ROLE';
KW_ROLES : 'ROLES';
KW_ROLLBACK : 'ROLLBACK';
KW_ROLLUP : 'ROLLUP';
KW_ROW : 'ROW';
KW_ROWS : 'ROWS';
KW_SECOND : 'SECOND';
KW_SECONDS : 'SECONDS';
KW_SCHEMA : 'SCHEMA';
KW_SCHEMAS : 'SCHEMAS';
KW_SELECT : 'SELECT';
KW_SEMI : 'SEMI';
KW_SEPARATED : 'SEPARATED';
KW_SERDE : 'SERDE';
KW_SERDEPROPERTIES : 'SERDEPROPERTIES';
KW_SESSION_USER : 'SESSION_USER';
KW_SET : 'SET';
KW_MINUS : 'MINUS';
KW_SETS : 'SETS';
KW_SHORT : 'SHORT';
KW_SHOW : 'SHOW';
KW_SINGLE : 'SINGLE';
KW_SKEWED : 'SKEWED';
KW_SMALLINT : 'SMALLINT';
KW_SOME : 'SOME';
KW_SORT : 'SORT';
KW_SORTED : 'SORTED';
KW_SOURCE : 'SOURCE';
KW_START : 'START';
KW_STATISTICS : 'STATISTICS';
KW_STORED : 'STORED';
KW_STRATIFY : 'STRATIFY';
KW_STRING : 'STRING';
KW_STRUCT : 'STRUCT';
KW_SUBSTR : 'SUBSTR';
KW_SUBSTRING : 'SUBSTRING';
KW_SYNC : 'SYNC';
KW_SYSTEM : 'SYSTEM';
KW_SYSTEM_TIME : 'SYSTEM_TIME';
KW_SYSTEM_VERSION : 'SYSTEM_VERSION';
KW_TABLE : 'TABLE';
KW_TABLES : 'TABLES';
KW_TABLESAMPLE : 'TABLESAMPLE';
KW_TARGET : 'TARGET';
KW_TBLPROPERTIES : 'TBLPROPERTIES';
KW_TEMPORARY : 'TEMPORARY';
KW_TERMINATED : 'TERMINATED';
KW_THEN : 'THEN';
KW_TIME : 'TIME';
KW_TIMEDIFF : 'TIMEDIFF';
KW_TIMESTAMP : 'TIMESTAMP';
KW_TIMESTAMP_LTZ : 'TIMESTAMP_LTZ';
KW_TIMESTAMP_NTZ : 'TIMESTAMP_NTZ';
KW_TIMESTAMPADD : 'TIMESTAMPADD';
KW_TIMESTAMPDIFF : 'TIMESTAMPDIFF';
KW_TINYINT : 'TINYINT';
KW_TO : 'TO';
KW_TOUCH : 'TOUCH';
KW_TRAILING : 'TRAILING';
KW_TRANSACTION : 'TRANSACTION';
KW_TRANSACTIONS : 'TRANSACTIONS';
KW_TRANSFORM : 'TRANSFORM';
KW_TRIM : 'TRIM';
KW_TRUE : 'TRUE';
KW_TRUNCATE : 'TRUNCATE';
KW_TRY_CAST : 'TRY_CAST';
KW_TYPE : 'TYPE';
KW_UNARCHIVE : 'UNARCHIVE';
KW_UNBOUNDED : 'UNBOUNDED';
KW_UNCACHE : 'UNCACHE';
KW_UNION : 'UNION';
KW_UNIQUE : 'UNIQUE';
KW_UNKNOWN : 'UNKNOWN';
KW_UNLOCK : 'UNLOCK';
KW_UNPIVOT : 'UNPIVOT';
KW_UNSET : 'UNSET';
KW_UPDATE : 'UPDATE';
KW_USE : 'USE';
KW_USER : 'USER';
KW_USING : 'USING';
KW_VALUES : 'VALUES';
KW_VARCHAR : 'VARCHAR';
KW_VAR : 'VAR';
KW_VARIABLE : 'VARIABLE';
KW_VERSION : 'VERSION';
KW_VIEW : 'VIEW';
KW_VIEWS : 'VIEWS';
KW_VOID : 'VOID';
KW_WEEK : 'WEEK';
KW_WEEKS : 'WEEKS';
KW_WHEN : 'WHEN';
KW_WHERE : 'WHERE';
KW_WINDOW : 'WINDOW';
KW_WITH : 'WITH';
KW_WITHIN : 'WITHIN';
KW_YEAR : 'YEAR';
KW_YEARS : 'YEARS';
KW_ZONE : 'ZONE';
//--SPARK-KEYWORD-LIST-END
//============================
// End of the keywords list
//============================
EQ : '=' | '==';
NSEQ: '<=>';
NEQ : '<>';
NEQJ: '!=';
LT : '<';
LTE : '<=' | '!>';
GT : '>';
GTE : '>=' | '!<';
EQ : '=' | '==';
NSEQ : '<=>';
NEQ : '<>';
NEQJ : '!=';
LT : '<';
LTE : '<=' | '!>';
GT : '>';
GTE : '>=' | '!<';
NOT: '!';
PLUS: '+';
MINUS: '-';
ASTERISK: '*';
SLASH: '/';
PERCENT: '%';
TILDE: '~';
AMPERSAND: '&';
PIPE: '|';
CONCAT_PIPE: '||';
HAT: '^';
COLON: ':';
ARROW: '->';
FAT_ARROW : '=>';
HENT_START: '/*+';
HENT_END: '*/';
QUESTION: '?';
NOT : '!';
PLUS : '+';
MINUS : '-';
ASTERISK : '*';
SLASH : '/';
PERCENT : '%';
TILDE : '~';
AMPERSAND : '&';
PIPE : '|';
CONCAT_PIPE : '||';
HAT : '^';
COLON : ':';
ARROW : '->';
FAT_ARROW : '=>';
HENT_START : '/*+';
HENT_END : '*/';
QUESTION : '?';
STRING_LITERAL
: '\'' ( ~('\''|'\\') | ('\\' .) )* '\''
| 'R\'' (~'\'')* '\''
| 'R"'(~'"')* '"'
;
STRING_LITERAL: '\'' ( ~('\'' | '\\') | ('\\' .))* '\'' | 'R\'' (~'\'')* '\'' | 'R"' (~'"')* '"';
DOUBLEQUOTED_STRING
:'"' ( ~('"'|'\\') | ('\\' .) )* '"'
;
DOUBLEQUOTED_STRING: '"' ( ~('"' | '\\') | ('\\' .))* '"';
// NOTE: If you move a numeric literal, you should modify `ParserUtils.toExprAlias()`
// which assumes all numeric literals are between `BIGINT_LITERAL` and `BIGDECIMAL_LITERAL`.
BIGINT_LITERAL
: DIGIT+ 'L'
;
BIGINT_LITERAL: DIGIT+ 'L';
SMALLINT_LITERAL
: DIGIT+ 'S'
;
SMALLINT_LITERAL: DIGIT+ 'S';
TINYINT_LITERAL
: DIGIT+ 'Y'
;
TINYINT_LITERAL: DIGIT+ 'Y';
INTEGER_VALUE
: DIGIT+
;
INTEGER_VALUE: DIGIT+;
EXPONENT_VALUE
: DIGIT+ EXPONENT
| DECIMAL_DIGITS EXPONENT
;
EXPONENT_VALUE: DIGIT+ EXPONENT | DECIMAL_DIGITS EXPONENT;
DECIMAL_VALUE
: DECIMAL_DIGITS
;
DECIMAL_VALUE: DECIMAL_DIGITS;
FLOAT_LITERAL
: DIGIT+ EXPONENT? 'F'
| DECIMAL_DIGITS EXPONENT? 'F'
;
FLOAT_LITERAL: DIGIT+ EXPONENT? 'F' | DECIMAL_DIGITS EXPONENT? 'F';
DOUBLE_LITERAL
: DIGIT+ EXPONENT? 'D'
| DECIMAL_DIGITS EXPONENT? 'D'
;
DOUBLE_LITERAL: DIGIT+ EXPONENT? 'D' | DECIMAL_DIGITS EXPONENT? 'D';
BIGDECIMAL_LITERAL
: DIGIT+ EXPONENT? 'BD'
| DECIMAL_DIGITS EXPONENT? 'BD'
;
BIGDECIMAL_LITERAL: DIGIT+ EXPONENT? 'BD' | DECIMAL_DIGITS EXPONENT? 'BD';
IDENTIFIER
: (LETTER | DIGIT | '_')+
;
IDENTIFIER: (LETTER | DIGIT | '_')+;
BACKQUOTED_IDENTIFIER
: '`' ( ~'`' | '``' )* '`'
;
BACKQUOTED_IDENTIFIER: '`' ( ~'`' | '``')* '`';
fragment DECIMAL_DIGITS
: DIGIT+ '.' DIGIT*
| '.' DIGIT+
;
fragment DECIMAL_DIGITS: DIGIT+ '.' DIGIT* | '.' DIGIT+;
fragment EXPONENT
: 'E' [+-]? DIGIT+
;
fragment EXPONENT: 'E' [+-]? DIGIT+;
fragment DIGIT
: [0-9]
;
fragment DIGIT: [0-9];
fragment LETTER
: [A-Za-z]
;
fragment LETTER: [A-Za-z];
SIMPLE_COMMENT
: '--' ('\\\n' | ~[\r\n])* '\r'? '\n'? -> channel(HIDDEN)
;
SIMPLE_COMMENT: '--' ('\\\n' | ~[\r\n])* '\r'? '\n'? -> channel(HIDDEN);
BRACKETED_COMMENT
: '/*' ( BRACKETED_COMMENT | . )*? ('*/' | {this.markUnclosedComment();} EOF) -> channel(HIDDEN)
;
BRACKETED_COMMENT:
'/*' (BRACKETED_COMMENT | .)*? ('*/' | {this.markUnclosedComment();} EOF) -> channel(HIDDEN);
WS
: [ \r\n\t]+ -> channel(HIDDEN)
;
WS: [ \r\n\t]+ -> channel(HIDDEN);
// Catch-all for anything we can't recognize.
// We use this to be able to ignore and recover all the text
// when splitting statements with DelimiterLexer
UNRECOGNIZED
: .
;
UNRECOGNIZED: .;

669
src/grammar/spark/SparkSqlParser.g4
View File

File diff suppressed because it is too large Load Diff

2208
src/grammar/trinosql/TrinoSql.g4
View File

File diff suppressed because it is too large Load Diff