work around for haddock refusal to parse literal comment lines with *

in first character position
get rid of code_units since these are not in sql2011
implement next value for
parse the nullary functions with reserved names
updates to the sql2003 file

Language/SQL/SimpleSQL/Parser.lhs

@@ -38,9 +38,9 @@ lexers, this isn't 100% complete at the moment and needs fixing.
 The parsing code is aggressively left factored, and try is avoided as
 much as possible. Try is avoided because:
 
-* when it is overused it makes the code hard to follow
+ * when it is overused it makes the code hard to follow
-* when it is overused it makes the parsing code harder to debug
+ * when it is overused it makes the parsing code harder to debug
-* it makes the parser error messages much worse
+ * it makes the parser error messages much worse
 
 The code could be made a bit simpler with a few extra 'trys', but this
 isn't done because of the impact on the parser error
@@ -74,9 +74,9 @@ syntax.
 
 There are three big areas which are tricky to left factor:
 
-* typenames
+ * typenames
-* value expressions which can start with an identifier
+ * value expressions which can start with an identifier
-* infix and suffix operators
+ * infix and suffix operators
 
 === typenames
 
@@ -97,12 +97,12 @@ error messages really bad.
 
 Here is a list of these nodes:
 
-* identifiers
+ * identifiers
-* function application
+ * function application
-* aggregate application
+ * aggregate application
-* window application
+ * window application
-* typed literal: typename 'literal string'
+ * typed literal: typename 'literal string'
-* interval literal which is like the typed literal with some extras
+ * interval literal which is like the typed literal with some extras
 
 There is further ambiguity e.g. with typed literals with precision,
 functions, aggregates, etc. - these are an identifier, followed by
@@ -113,12 +113,12 @@ is.
 There is also a set of nodes which start with an identifier/keyword
 but can commit since no other syntax can start the same way:
 
-* case
+ * case
-* cast
+ * cast
-* exists, unique subquery
+ * exists, unique subquery
-* array constructor
+ * array constructor
-* multiset constructor
+ * multiset constructor
-* all the special syntax functions: extract, position, substring,
+ * all the special syntax functions: extract, position, substring,
   convert, translate, overlay, trim, etc.
 
 The interval literal mentioned above is treated in this group at the
@@ -143,10 +143,10 @@ standard which is able to eliminate a number of possibilities just in
 the grammar, which this parser allows. This is done for a number of
 reasons:
 
-* it makes the parser simple - less variations
+ * it makes the parser simple - less variations
-* it should allow for dialects and extensibility more easily in the
+ * it should allow for dialects and extensibility more easily in the
   future (e.g. new infix binary operators with custom precedence)
-* many things which are effectively checked in the grammar in the
+ * many things which are effectively checked in the grammar in the
   standard, can be checked using a typechecker or other simple static
   analysis
 
@@ -481,7 +481,6 @@ TODO: this code needs heavy refactoring
 >                     ,return Nothing]
 >         return (p,x)
 >     lobUnits = choice [LobCharacters <$ keyword_ "characters"
->                       ,LobCodeUnits <$ keyword_ "code_units"
 >                       ,LobOctets <$ keyword_ "octets"]
 >     -- deal with multiset and array suffixes
 >     tnSuffix x =
@@ -657,6 +656,10 @@ multiset(query expr). It must be there for compatibility or something.
 >     ,keyword_ "table" >>
 >      MultisetQueryCtor <$> parens queryExpr]
 
+> nextValueFor :: Parser ValueExpr
+> nextValueFor = keywords_ ["next","value","for"] >>
+>     NextValueFor <$> names
+
 === interval
 
 interval literals are a special case and we follow the grammar less
@@ -1161,6 +1164,7 @@ fragile and could at least do with some heavy explanation.
 >               ,cast
 >               ,arrayCtor
 >               ,multisetCtor
+>               ,nextValueFor
 >               ,subquery
 >               ,intervalLit
 >               ,specialOpKs
@@ -1740,7 +1744,7 @@ means).
 >     ,"class_origin"
 >     ,"coalesce"
 >     ,"cobol"
->     ,"code_units"
+>     --,"code_units"
 >     ,"collation"
 >     ,"collation_catalog"
 >     ,"collation_name"
@@ -2002,13 +2006,13 @@ means).
 >     ,"cube"
 >     ,"current"
 >     --,"current_date"
->     ,"current_default_transform_group"
+>     --,"current_default_transform_group"
->     ,"current_path"
+>     --,"current_path"
->     ,"current_role"
+>     --,"current_role"
 >     ,"current_time"
 >     ,"current_timestamp"
 >     ,"current_transform_group_for_type"
->     ,"current_user"
+>     --,"current_user"
 >     ,"cursor"
 >     ,"cycle"
 >     ,"date"
@@ -2052,7 +2056,7 @@ means).
 >     ,"global"
 >     ,"grant"
 >     ,"group"
->     ,"grouping"
+>     --,"grouping"
 >     ,"having"
 >     ,"hold"
 >     --,"hour"
@@ -2088,7 +2092,7 @@ means).
 >     ,"method"
 >     --,"minute"
 >     ,"modifies"
->     ,"module"
+>     --,"module"
 >     --,"month"
 >     ,"multiset"
 >     ,"national"
@@ -2151,7 +2155,7 @@ means).
 >     --,"second"
 >     ,"select"
 >     ,"sensitive"
->     ,"session_user"
+>     --,"session_user"
 >     --,"set"
 >     ,"similar"
 >     ,"smallint"
@@ -2167,7 +2171,7 @@ means).
 >     ,"submultiset"
 >     ,"symmetric"
 >     ,"system"
->     ,"system_user"
+>     --,"system_user"
 >     ,"table"
 >     ,"then"
 >     ,"time"
@@ -2187,7 +2191,7 @@ means).
 >     ,"unnest"
 >     ,"update"
 >     ,"upper"
->     ,"user"
+>     --,"user"
 >     ,"using"
 >     --,"value"
 >     ,"values"

Language/SQL/SimpleSQL/Pretty.lhs

@@ -218,6 +218,9 @@ which have been changed to try to improve the layout of the output.
 > valueExpr (Collate v c) =
 >     valueExpr v <+> text "collate" <+> names c
 
+> valueExpr (NextValueFor ns) =
+>     text "next value for" <+> names ns
+
 
 > doubleUpQuotes :: String -> String
 > doubleUpQuotes [] = []
@@ -263,7 +266,6 @@ which have been changed to try to improve the layout of the output.
 >                            LobG -> text "G") m
 >                <+> me (\x -> case x of
 >                        LobCharacters -> text "CHARACTERS"
->                        LobCodeUnits -> text "CODE_UNITS"
 >                        LobOctets -> text "OCTETS") u)
 > typeName (CharTypeName t i cs col) =
 >     names t

Language/SQL/SimpleSQL/Syntax.lhs

@@ -158,6 +158,7 @@
 >     | MultisetBinOp ValueExpr CombineOp SetQuantifier ValueExpr
 >     | MultisetCtor [ValueExpr]
 >     | MultisetQueryCtor QueryExpr
+>     | NextValueFor [Name]
 >       deriving (Eq,Show,Read,Data,Typeable)
 
 > -- | Represents an identifier name, which can be quoted or unquoted.
@@ -190,7 +191,6 @@ TODO: add ref and scope, any others?
 > data LobMultiplier = LobK | LobM | LobG
 >                      deriving (Eq,Show,Read,Data,Typeable)
 > data LobUnits = LobCharacters
->               | LobCodeUnits
 >               | LobOctets
 >                 deriving (Eq,Show,Read,Data,Typeable)
 

TODO

@@ -1,5 +1,11 @@
 continue 2003 review and tests
 
+1. start replacing the 2003 stuff with 2011
+2. create an error message document for the website
+   - base of error messages but add some more variations
+3. start thinking about tests for invalid syntax
+
+
 touch up the expr hack as best as can
 
 careful review of token parses wrt trailing delimiters/junk
@@ -56,6 +62,8 @@ rules for changing the multi keyword parsing:
 
 rough SQL 2003 todo, including tests to write:
 
+switch to SQL 2011
+
 now:
 review the commented out reserved keyword entries and work out how to
    fix
@@ -64,7 +72,6 @@ go through almost all the predicates
 window functions missing bits, window clauses
 from: more tests, review missing
   tablesample, unnest, etc.
-aggregates: where, filter + review
 rows review
 match missing bit
 between symmetric

changelog

@@ -1,4 +1,4 @@
-0.4.0-dev (updated to 7a847045163feb2339ab40ebe93afe2f1c9ad813)
+0.4.0-dev (updated to 705724197463cd19dd8749dfd51e2eb8f1d02b8e)
 	completely remove dependency on haskell-src-exts
 	improve the error messages a great deal
 	fix some trailing whitespace issues in the keyword style functions,
@@ -52,6 +52,11 @@
 	  quote character in the identifier
 	implement complete interval literals (fixed the handling of the
 	  interval qualifier)
+	make most of the standard reserved words actually reserved (still
+	  some gaps)
+	change the natural in join abstract syntax to match the concrete
+	  syntax instead of combining natural, on and using into one field
+	support filter and within group for aggregates
 0.3.1 (commit 5cba9a1cac19d66166aed2876d809aef892ff59f)
 	update to work with ghc 7.8.1
 0.3.0 (commit 9e75fa93650b4f1a08d94f4225a243bcc50445ae)

simple-sql-parser.cabal

@@ -28,7 +28,7 @@ library
                        Language.SQL.SimpleSQL.Parser,
                        Language.SQL.SimpleSQL.Syntax
   other-extensions:    TupleSections
-  build-depends:       base >=4.6 && <4.7,
+  build-depends:       base >=4.6 && <4.8,
                        parsec >=3.1 && <3.2,
                        mtl >=2.1 && <2.2,
                        pretty >= 1.1 && < 1.2
@@ -40,7 +40,7 @@ Test-Suite Tests
   type:                exitcode-stdio-1.0
   main-is:             RunTests.lhs
   hs-source-dirs:      .,tools
-  Build-Depends:       base >=4.6 && <4.7,
+  Build-Depends:       base >=4.6 && <4.8,
                        parsec >=3.1 && <3.2,
                        mtl >=2.1 && <2.2,
                        pretty >= 1.1 && < 1.2,
@@ -71,7 +71,7 @@ Test-Suite Tests
 executable SQLIndent
   main-is:             SQLIndent.lhs
   hs-source-dirs:      .,tools
-  Build-Depends:       base >=4.6 && <4.7,
+  Build-Depends:       base >=4.6 && <4.8,
                        parsec >=3.1 && <3.2,
                        mtl >=2.1 && <2.2,
                        pretty >= 1.1 && < 1.2

tools/Language/SQL/SimpleSQL/SQL2003.lhs

@@ -15,7 +15,8 @@ large amount of the SQL.
 
 > sql2003Tests :: TestItem
 > sql2003Tests = Group "sql2003Tests"
->     [stringLiterals
+>     [Group "literals" [
+>      stringLiterals
 >     ,nationalCharacterStringLiterals
 >     ,unicodeStringLiterals
 >     ,binaryStringLiterals
@@ -23,32 +24,72 @@ large amount of the SQL.
 >     ,intervalLiterals
 >     ,booleanLiterals
 >     ,identifiers
->     ,typeNameTests
+>      ],Group "value expressions"
+>     [typeNameTests
 >     ,parenthesizedValueExpression
+>     ,someGeneralValues
 >     ,targetSpecification
 >     ,contextuallyTypeValueSpec
->     --,nextValueExpression
+>     ,moduleColumnRef
+>     ,groupingOperation
+>     --,windowFunction
+>     --,caseExpression
+>     --,castSpecification
+>     ,nextValueExpression
+>     -- subtype treatment, method invoc, static m i, new spec, attrib/method ref, deref, method ref, ref res
 >     ,arrayElementReference
 >     ,multisetElementReference
->     --,numericValueExpression
+>     ,numericValueExpression
+>     --,numericValueFunction
+>     --,stringValueExpression
+>     --,stringValueFunction
+>     --,datetimeValueExpression
+>     --,datetimeValueFunction
+>     --,intervalValueExpression
+>     --,intervalValueFunction
 >     --,booleanValueExpression
+>     --arrayValueExpression
 >     ,arrayValueConstructor
 >     ,multisetValueExpression
 >     ,multisetValueFunction
 >     ,multisetValueConstructor
+>     ],Group "query expressions"
+>     [
+>     -- rowValueConstructor
+>     --,rowValueExpression
 >     --,tableValueConstructor
 >     --,fromClause
+>     --,joinedTable
 >     --,whereClause
->     ,groupbyClause
+>      groupbyClause
+>     --,havingClause
+>     --,windowClause
 >     --,querySpecification
->     --,queryExpressions
+>     --,querySpecifications
->     ,quantifiedComparisonPredicate
+>     --,setOperations
+>     --,withExpressions
+>     ],Group "predicates"
+>     [--comparisonPredicate
+>     --,betweenPredicate
+>     --,inPredicate
+>     --,likePredicate
+>     --,similarPredicae
+>     --,nullPredicate
+>      quantifiedComparisonPredicate
+>     --,existsPredicate
 >     ,uniquePredicate
+>     --,normalizedPredicate
 >     ,matchPredicate
+>     --,overlapsPredicate
+>     --,distinctPredicate
+>     --,memberPredicate
+>     --,submultisetPredicate
+>     --,setPredicate
 >     ,collateClause
 >     ,aggregateFunctions
 >     ,sortSpecificationList
 >     ]
+>     ]
 
 = 5 Lexical Elements
 
@@ -1001,8 +1042,6 @@ create a list of type name variations:
 >          ,("blob(3M)", LobTypeName [Name "blob"] 3 (Just LobM) Nothing)
 >          ,("blob(4M characters) "
 >           ,LobTypeName [Name "blob"] 4 (Just LobM) (Just LobCharacters))
->          ,("blob(5 code_units) "
->           ,LobTypeName [Name "blob"] 5 Nothing (Just LobCodeUnits))
 >          ,("blob(6G octets) "
 >           ,LobTypeName [Name "blob"] 6 (Just LobG) (Just LobOctets))
 >          ,("national character large object(7K) "
@@ -1172,7 +1211,19 @@ This is used in row type names.
      |     USER 
      |     VALUE
 
-TODO: review how the special keywords are parsed and add tests for these
+
+> someGeneralValues :: TestItem
+> someGeneralValues = Group "some general values" $ map (uncurry TestValueExpr) $
+>     map mkIden ["CURRENT_DEFAULT_TRANSFORM_GROUP"
+>                ,"CURRENT_PATH"
+>                ,"CURRENT_ROLE"
+>                ,"CURRENT_USER"
+>                ,"SESSION_USER"
+>                ,"SYSTEM_USER"
+>                ,"USER"
+>                ,"VALUE"]
+>   where
+>     mkIden nm = (nm,Iden [Name nm])
 
 <simple value specification>    ::= 
          <literal>
@@ -1269,8 +1320,11 @@ already covered above in the identifiers and names section
          <basic identifier chain>
      |     MODULE <period> <qualified identifier> <period> <column name>
 
-TODO: work out the exact syntax and add
+> moduleColumnRef :: TestItem
-
+> moduleColumnRef = Group "module column ref" $ map (uncurry TestValueExpr)
+>     [("MODULE.something.something", Iden [Name "MODULE"
+>                                          ,Name "something"
+>                                          ,Name "something"])]
 
 
 == 6.8 <SQL parameter reference> (p190)
@@ -1304,7 +1358,19 @@ ORDER BY department, job, "Total Empl", "Average Sal";
 
 TODO: de-oracle the syntax and add as test case
 
-
+> groupingOperation :: TestItem
+> groupingOperation = Group "grouping operation" $ map (uncurry TestQueryExpr)
+>     [("SELECT SalesQuota, SUM(SalesYTD) TotalSalesYTD,\n\
+>       \   GROUPING(SalesQuota) AS Grouping\n\
+>       \FROM Sales.SalesPerson\n\
+>       \GROUP BY ROLLUP(SalesQuota);"
+>      ,makeSelect
+>       {qeSelectList = [(Iden [Name "SalesQuota"],Nothing)
+>                       ,(App [Name "SUM"] [Iden [Name "SalesYTD"]],Just (Name "TotalSalesYTD"))
+>                       ,(App [Name "GROUPING"] [Iden [Name "SalesQuota"]],Just (Name "Grouping"))]
+>       ,qeFrom = [TRSimple [Name "Sales",Name "SalesPerson"]]
+>       ,qeGroupBy = [Rollup [SimpleGroup (Iden [Name "SalesQuota"])]]})
+>     ]
 
 == 6.10 <window function> (p193)
 
@@ -1323,6 +1389,10 @@ TODO: de-oracle the syntax and add as test case
 
 TODO: window functions
 
+> windowFunctions :: TestItem
+> windowFunctions = Group "window functions" $ map (uncurry TestValueExpr)
+>     [
+>     ]
 
 
 == 6.11 <case expression> (p197)
@@ -1371,7 +1441,10 @@ TODO: window functions
 
 TODO: case expressions plus the 'abbreviations'
 
-
+> caseExpression :: TestItem
+> caseExpression = Group "case expression" $ map (uncurry TestValueExpr)
+>     [
+>     ]
 
 == 6.12 <cast specification> (p200)
 
@@ -1391,7 +1464,7 @@ This is already covered above
 
 > nextValueExpression :: TestItem
 > nextValueExpression = Group "next value expression" $ map (uncurry TestValueExpr)
->     [("next value for a.b", undefined)
+>     [("next value for a.b", NextValueFor [Name "a", Name "b"])
 >     ]
 
 
@@ -1569,13 +1642,16 @@ Specify a numeric value.
 
 > numericValueExpression :: TestItem
 > numericValueExpression = Group "numeric value expression" $ map (uncurry TestValueExpr)
->     [("a + b", undefined)
+>     [("a + b", binOp "+")
->     ,("a - b", undefined)
+>     ,("a - b", binOp "-")
->     ,("a * b", undefined)
+>     ,("a * b", binOp "*")
->     ,("a / b", undefined)
+>     ,("a / b", binOp "/")
->     ,("+a", undefined)
+>     ,("+a", prefOp "+")
->     ,("-a", undefined)
+>     ,("-a", prefOp "-")
 >     ]
+>   where
+>     binOp o = BinOp (Iden [Name "a"]) [Name o] (Iden [Name "b"])
+>     prefOp o = PrefixOp [Name o] (Iden [Name "a"])
 
 == 6.27 <numeric value function> (p242)
 
@@ -1600,12 +1676,22 @@ Specify a function yielding a value of type numeric.
          <string position expression>
      |     <blob position expression>
 
+> numericValueFunction :: TestItem
+> numericValueFunction = Group "numeric value function" $ map (uncurry TestValueExpr)
+>     [
+
+
 <string position expression>    ::= 
          POSITION <left paren> <string value expression> IN <string value expression> [ USING <char length units> ] <right paren>
 
 <blob position expression>    ::= 
          POSITION <left paren> <blob value expression> IN <blob value expression> <right paren>
 
+>      ("position (a in b)",undefined)
+>     ,("position (a in b using characters)",undefined)
+>     ,("position (a in b using octets)",undefined)
+
+
 TODO: position expressions
 
 <length expression>    ::= 
@@ -1663,6 +1749,9 @@ TODO: extract expression
 
 TODO: lots more expressions above
 
+>     ]
+
+
 == 6.28 <string value expression> (p251)
 
 Specify a character string value or a binary string value.
@@ -3160,9 +3249,3 @@ TODO: review sort specifications
 >     qe = makeSelect
 >          {qeSelectList = [(Star,Nothing)]
 >          ,qeFrom = [TRSimple [Name "t"]]}
-
-TODO: what happened to the collation in order by?
-Answer: sort used to be a column reference with an optional
-collate. Since it is now a value expression, the collate doesn't need
-to be mentioned here.
-