Commit 4df5ca45 authored by Michael Hanus's avatar Michael Hanus
Browse files

currypp / SQL added

parent cab3040d
......@@ -16,7 +16,7 @@ currycheck/CurryCheck
......@@ -93,6 +93,13 @@ concatPR (Errors p1) (Errors p2) = Errors (p1 ++ p2)
concatPR (Errors p1) (OK _) = Errors p1
concatPR (OK _) (Errors p2) = Errors p2
--- Combines two PRs by a given functions
combinePRs :: (a -> b -> c) -> PR a -> PR b -> PR c
combinePRs f (OK x) (OK y) = okPR (f x y)
combinePRs _ (Errors p1) (Errors p2) = Errors (p1 ++ p2)
combinePRs _ (Errors p1) (OK _) = Errors p1
combinePRs _ (OK _) (Errors p2) = Errors p2
--- Join multiple Error Monads into one
sequencePR :: [PR a] -> PR [a]
sequencePR [] = okPR []
......@@ -39,6 +39,10 @@ warnOKPM x = warnPM (okPR x)
throwPM :: Pos -> String -> PM _
throwPM p s = cleanWM (throwPMsg p s)
throwMultiPM :: Pos -> [String] -> PM _
throwMultiPM p strs = cleanWM (throwPR (map (\s -> (PError p s)) strs))
--- Return without Errors but with one Warning
singlePM :: a -> Warning -> PM a
singlePM x w = warnOKPM x [w]
......@@ -76,3 +80,12 @@ fstPM = liftPM fst
--- snd defined on PM
sndPM :: PM (a,b) -> PM b
sndPM = liftPM snd
--- combines two PMs by function f, throws error if at least one of
--- the two carries an error
combinePMs :: (a -> b -> c) -> PM a -> PM b -> PM c
combinePMs f p1 p2 = warnPM (combinePRs f (discardWarningsPM p1)
(discardWarningsPM p2))
(concatWarns p1 p2)
concatWarns (WM _ w1) (WM _ w2) = w1 ++ w2
--- This module defines the AST structure generated by the parser
--- and used for the translation of SQL statements.
--- @author Julia Krone
--- @version 0.1
-- ---------------------------------------------------------------
module SQLAst where
import Time(CalendarTime)
data Statement = Select SelectHead Order (Maybe Int)
| Update Table [Assign] Condition
| UpdateEntity Table Value
| Delete Table Condition
| Insert Table [ColumnRef] [[Value]]
| Transaction
| InTransaction [Statement]
| Commit
| Rollback
--contains the table (Unique for a given alias
-- Def (default) in case no alias is given),
-- name of column, type of column, nullable, CDBI-alias
data ColumnRef = Column Tab String Type Bool Int
data Tab = Unique String | Def [String]
data Type = I | B | F | C | S | D | Key String | Entity String | Unknown
data Assign = Assign ColumnRef Value
data Value = Emb String Type
| IntExp Int
| KeyExp String Int -- referenced table (name) and value
| FloatExp Float
| StringExp String
| DateExp CalendarTime
| BoolExp Bool
| CharExp Char
| AbsNull
data SelectHead = Set ASetOp SelectHead SelectHead
| Query SelectClause TableRef Condition (Maybe Group)
data SelectClause = SelAll ASpecifier
| SelColumns ASpecifier [SelElement]
data SelElement = Aggregation AFun ASpecifier ColumnRef
| Col ColumnRef
| Case Condition Operand Operand
data Table = Table String String Int -- name alias and CDBI-alias
data TableRef = TableRef Table (Maybe JoinClause)
data JoinClause = CrossJoin Table (Maybe JoinClause)
| InnerJoin Table JoinCond (Maybe JoinClause)
data JoinCond = JC Condition
data ASpecifier = AAll | ADistinct
data Condition = FK (String,Int) AbsRel (String,Int)--(Table, Alias) Relation (Table,Alias)
| Cmp ALogOp Condition Condition
| Not Condition
| Exists Statement
| IsNull Operand
| NotNull Operand
| AIn Operand [Value]
| ABinOp AstOp Operand Operand
| ABetween Operand Operand Operand
| NoCond
data AbsRel = AMToN String
| ANToOne String
| AOneToN String
| NotSpec String
type Operand = Either ColumnRef Value
data AstOp = ALth | ALe | AGth | AGe | AEq | AUnEq | ALike
data ALogOp = AAnd | AOr
data ASetOp = AUnion | AExcept | AIntersect
data Group = GroupBy [ColumnRef] Having
data Having = SimpleHave Condition
| AggrHave AFun ASpecifier ColumnRef AstOp Operand
| Neg Having
| CmpHave ALogOp Having Having
| NoHave
data AFun = ASum | ACount | AAvg | AMin | AMax
data Order = OrderBy [(ColumnRef, Dir)]
data Dir = Asc | Desc
\ No newline at end of file
This diff is collapsed.
--- This module implements the Interface for the CurryPP Translator.
--- To provide the tranformation of SQL-statements it calls in this order:
--- Scanner
--- Parser
--- Consistency check
--- Type check
--- Translator (to functions of CDBI interface)
--- Reads the .info file which contains information about the data model and
--- passes the information to the corresponding subroutines.
--- Aborts compilation process in case a stage returns with an error.
---@author Julia Krone
---@version 0.1
-- ---------------------------------------------------------------------------
module SQLConverter(parse, readParserInfo, ParserInfo) where
import IO(openFile, IOMode(..), hGetContents)
import ReadShowTerm(readsQTerm)
import ParseTypes
import SQLConsistency
import SQLNamer
import SQLParser
import SQLParserInfoType
import SQLScanner
import SQLTranslator
import SQLTyper
--- Converts a string representing SQL-requests into functions defined
--- in the CDBI interface by calling the different stage of transformation.
---@parserInfo - either the parser information or an error message
---@param pos - Position of the integrated SQL-String in the orginal file
---@param code - the SQL-request as string
---@return A String in Curry-Syntax (CDBI-functions).
parse :: Either String ParserInfo -> LangParser
parse parserInfo pos code =
case parserInfo of
Left err -> return (throwPM pos err)
Right pi -> processCompilation pi pos code
--- Reader for parser information file.
---@param filename - path/name of the .info file
---@return either an error message or the parser
--- information
readParserInfo :: String -> IO (Either String ParserInfo)
readParserInfo filename =
handle <- openFile filename ReadMode
contents <- (hGetContents handle)
case (readsQTerm contents) of
[] -> return (Left ("ParserInfo-file was not found"++
" or is corrupted."))
((a,_):_) -> return (Right a)
-- auxiliary function to check Result after each stage
checkResult :: PM a -> Either (PM String) (PM a)
checkResult (WM (Errors err) ws) = Left (WM (throwPR err) ws)
checkResult pm@(WM (OK _ ) _) = Right pm
-- Calls the stages of parsing process in the right order
-- and passes corresponding parts of the parser information.
-- Aborts process in case a stage returns with an error.
processCompilation :: ParserInfo -> LangParser
processCompilation parserInfo pos code =
let parsedCode = parseTkLs pos (scan code)
in case checkResult parsedCode of
Left pm -> return pm
Right _ -> callNamer parsedCode
where callNamer res =
let namedCode = nameStatements res pos
in case checkResult namedCode of
Left pm -> return pm
Right _ -> callConsistency namedCode
callConsistency res =
let consCheckedCode = checkConsistency res
in case checkResult consCheckedCode of
Left pm -> return pm
Right _ -> callTyper consCheckedCode
callTyper res =
let typeCheckedCode = checkTypes res
(getTypes parserInfo)
in case checkResult typeCheckedCode of
Left pm -> return pm
Right _ -> return (translate typeCheckedCode
(dbName parserInfo)
(cdbiModule parserInfo)
\ No newline at end of file
--- This module resolves the pseudonym-tablename-binding for all columns
--- and all relationship constraints in each parsed statement
--- with the help of a symboltable.
--- Errors are thrown whenever an alias can not be resolved,
--- because it was not defined or is not visible or it was defined
--- more than once. An error is also thrown for an alias that was
--- defined but not used.
---@author: Julia Krone
---@version: 0.1
-- ------------------------------------------------------------------------
module SQLNamer(nameStatements) where
import Char (toLower)
import ParseTypes
import SQLAst
import Symboltab
--- Takes a list of completely parsed statements (wrapped in a PM),
--- inserts the name of the corresponding table (resolving the
--- pseudonym if given) and sets the right alias (int) for CDBI
--- in each AST-node representing a column. The correct CDBI-alias
--- is also set for each table-node and relationship-constraint-node.
--- An error is thrown, if the referenced alias was not defined
--- in the corresponding statement or if it is defined more than once.
--- Consistence with the data model is not checked at this stage.
---@param pm - the PM containing list of parsed statements
---@return unchanged PM if it contained errors
--- PM with named columns otherwise
nameStatements :: PM [Statement] -> Pos -> PM [Statement]
nameStatements (WM (Errors err) ws) _ = WM (throwPR err) ws
nameStatements (WM (OK ast) ws) p =
let (WM resPR warns) = sequencePM (map (nameStatement p emptyTable) ast)
in (WM resPR (ws ++ warns))
-- Symboltable structure used for the naming process.
-- The first finite map saves the notation given by the user,
-- the pseudonym and the CDBI-alias. The second FM keeps track of how often
-- a table name was referenced to calculate the correct CDBI-alias.
type AliasSymTab = Symboltable [(String,String, Int)] Int
--naming-function for statement node
nameStatement :: Pos -> AliasSymTab -> Statement -> PM Statement
nameStatement p st (Select selhead order lim) =
let (nSelHeadPm, st1) = nameSelHead p st selhead
in combinePMs (\nSelHead nOrder -> Select nSelHead nOrder lim)
(nameOrder p st1 order)
nameStatement p st (Update tab assigns cond) = nameUpdate p st tab assigns cond
nameStatement _ _ (UpdateEntity tab val) = cleanPM (UpdateEntity tab val)
nameStatement p st (Delete tab cond) = nameDelete p tab st cond
nameStatement p st (Insert tab cols valss) = nameInsert p tab st cols valss
nameStatement _ _ Transaction = cleanPM Transaction
nameStatement p _ (InTransaction stats) =
liftPM (\nStats -> (InTransaction nStats))
(sequencePM (map (nameStatement p emptyTable) stats))
nameStatement _ _ Commit = cleanPM Commit
nameStatement _ _ Rollback = cleanPM Rollback
-- naming-funtion for selectHead-node
nameSelHead :: Pos -> AliasSymTab -> SelectHead -> (PM SelectHead, AliasSymTab)
nameSelHead p st (Query selclause tabs cond group) =
let symtab = insertTabs tabs st
in case symtab of
Right st1 -> ((combinePMs (\(nsc, ntabs) (nCond, nGroup)
-> (Query nsc ntabs nCond nGroup))
(combinePMs (,)(nameSelClause p st1 selclause)
(nameTableRefs p st1 tabs))
(combinePMs (,)(nameCond p st1 cond)
(nameGroup p st1 group))), st1)
Left err -> ((throwPM p err), st)
nameSelHead p st (Set op h1 h2) =
let (nhead1, st1) = nameSelHead p st h1
(nhead2, st2) = nameSelHead p emptyTable h2
in ((combinePMs (\nh1 nh2 -> (Set op nh1 nh2))
,(combineST st1 st2))
-- naming-function for SelectClause-node
nameSelClause :: Pos -> AliasSymTab -> SelectClause -> PM SelectClause
nameSelClause _ _ sc@(SelAll _) = cleanPM sc
nameSelClause p st (SelColumns sp elems) =
liftPM (\nElems -> (SelColumns sp nElems))
(nameElems p st elems)
-- naming-function for TableRef-node
nameTableRefs :: Pos -> AliasSymTab -> TableRef -> PM TableRef
nameTableRefs p st (TableRef tab Nothing) =
liftPM (\nTab -> (TableRef nTab Nothing))
(nameTable p st tab)
nameTableRefs p st (TableRef tab (Just (CrossJoin tab2 join))) =
combinePMs (\(nTab, nTab2) nJoin ->
(TableRef nTab (Just (CrossJoin nTab2 nJoin))))
(combinePMs (,)
(nameTable p st tab)
(nameTable p st tab2))
(nameJoins p st join)
nameTableRefs p st (TableRef tab (Just (InnerJoin tab2 cond join))) =
combinePMs (\(nTab, nTab2) (nJoinCond, nJoin) ->
(TableRef nTab (Just (InnerJoin nTab2 nJoinCond nJoin))))
(combinePMs (,) (nameTable p st tab)
(nameTable p st tab2))
(combinePMs (,) (nameJoinCond p st cond)
(nameJoins p st join))
-- naming-function for JoinClause-node
nameJoins :: Pos -> AliasSymTab -> Maybe JoinClause -> PM (Maybe JoinClause)
nameJoins _ _ Nothing = cleanPM Nothing
nameJoins p st (Just (CrossJoin tab join)) =
combinePMs (\nTab nJoin -> (Just (CrossJoin nTab nJoin)))
(nameTable p st tab)
(nameJoins p st join)
nameJoins p st (Just (InnerJoin tab cond join)) =
combinePMs (\nTab (nCond, nJoin) -> (Just (InnerJoin nTab nCond nJoin)))
(nameTable p st tab)
(combinePMs (,)(nameJoinCond p st cond)
(nameJoins p st join))
-- naming-function for JoinCond-node
nameJoinCond :: Pos -> AliasSymTab -> JoinCond -> PM JoinCond
nameJoinCond p st (JC cond)=
liftPM (\nCond -> (JC nCond))
(nameCond p st cond)
-- naming-function for Condition-node
nameCond :: Pos -> AliasSymTab -> Condition -> PM Condition
nameCond p st (FK (ps1,_) rel (ps2,_)) =
combinePMs (\nTab1 nTab2 -> (FK nTab1 rel nTab2))
(getTable p ps1 st)
(getTable p ps2 st)
nameCond p st (Cmp op cond1 cond2) =
combinePMs (\nCond1 nCond2 ->(Cmp op nCond1 nCond2))
(nameCond p st cond1)
(nameCond p st cond2)
nameCond p st (Not cond) = liftPM (\nCond -> (Not nCond))
(nameCond p st cond)
nameCond p st (Exists stat) = liftPM (\nStat -> (Exists nStat))
(nameStatement p (enterScope st) stat)
nameCond p st (IsNull op) = liftPM (\nOp -> (IsNull nOp))
(nameOperand p st op)
nameCond p st (NotNull op) = liftPM (\nOp -> (NotNull nOp))
(nameOperand p st op)
nameCond p st (AIn op vals) = liftPM (\nOp -> (AIn nOp vals))
(nameOperand p st op)
nameCond p st (ABinOp bop op1 op2) =
combinePMs (\nOp1 nOp2 -> (ABinOp bop nOp1 nOp2))
(nameOperand p st op1)
(nameOperand p st op2)
nameCond p st (ABetween op1 op2 op3) =
combinePMs (\nOp1 (nOp2,nOp3) -> (ABetween nOp1 nOp2 nOp3))
(nameOperand p st op1)
(combinePMs (,) (nameOperand p st op2)
(nameOperand p st op3))
nameCond _ _ NoCond = cleanPM NoCond
-- naming-function for Group-node
nameGroup :: Pos -> AliasSymTab -> Maybe Group -> PM (Maybe Group)
nameGroup p st (Just (GroupBy cols hav)) =
combinePMs (\nCols nHave -> (Just (GroupBy nCols nHave)))
(nameColumns p st cols)
(nameHaving p st hav)
nameGroup _ _ Nothing = cleanPM Nothing
nameElems :: Pos -> AliasSymTab -> [SelElement] -> PM [SelElement]
nameElems p st elems = sequencePM (map (nameSingleElem p st) elems)
-- naming-function for an Element-node in SelectClause
nameSingleElem :: Pos -> AliasSymTab -> SelElement -> PM SelElement
nameSingleElem p st (Aggregation fun sp col) =
liftPM (\nCol -> (Aggregation fun sp nCol))
(nameSingleColumn p st col)
nameSingleElem p st (Col col) = liftPM (\nCol -> (Col nCol))
(nameSingleColumn p st col)
nameSingleElem p st (Case cond op1 op2) =
combinePMs (\nCond (nOp1, nOp2) -> (Case nCond nOp1 nOp2))
(nameCond p st cond)
(combinePMs (,) (nameOperand p st op1)
(nameOperand p st op2))
-- naming-function for an operand-node
nameOperand :: Pos -> AliasSymTab-> Operand -> PM Operand
nameOperand p st (Left col) = liftPM (\nCol -> (Left nCol))
(nameSingleColumn p st col)
nameOperand _ _ (Right val) = cleanPM (Right val)
-- naming-function for Having-node
nameHaving :: Pos -> AliasSymTab -> Having -> PM Having
nameHaving p st (SimpleHave cond) = liftPM (\nCond -> (SimpleHave nCond))
(nameCond p st cond)
nameHaving p st (AggrHave fun sp col bop op) =
combinePMs (\nCol nOp -> (AggrHave fun sp nCol bop nOp))
(nameSingleColumn p st col )
(nameOperand p st op)
nameHaving p st (Neg have) = liftPM (\nHave -> (Neg nHave))
(nameHaving p st have)
nameHaving p st (CmpHave lop have1 have2) =
combinePMs (\nH1 nH2 -> (CmpHave lop nH1 nH2))
(nameHaving p st have1)
(nameHaving p st have2)
nameHaving _ _ NoHave = cleanPM NoHave
-- naming-function for Order-node
nameOrder :: Pos -> AliasSymTab -> Order -> PM Order
nameOrder p st (OrderBy colDirs) =
liftPM (\nOrds -> OrderBy nOrds)
(sequencePM (map (nameSingleOrder p st) colDirs))
-- naming-function for a single column-direction-pair in Order-clause
nameSingleOrder :: Pos -> AliasSymTab -> (ColumnRef, Dir) -> PM (ColumnRef, Dir)
nameSingleOrder p st (col, dir) = liftPM (\nCol -> (nCol,dir))
(nameSingleColumn p st col)
nameUpdate :: Pos ->
AliasSymTab ->
Table ->
[Assign] ->
Condition ->
PM Statement
nameUpdate p st tab assigns cond =
let symtab = insertTab tab st
in case symtab of
Right st1 -> combinePMs (\nAssigns nCond -> (Update tab nAssigns nCond))
(nameAssignments p st1 assigns)
(nameCond p st1 cond)
-- this should not happen in update statements
Left err -> throwPM p err
nameAssignments :: Pos -> AliasSymTab -> [Assign] -> PM [Assign]
nameAssignments p st assigns = sequencePM (map (nameAssign p st) assigns)
-- naming-function for Assign-node
nameAssign :: Pos -> AliasSymTab -> Assign -> PM Assign
nameAssign p st (Assign col val) =
liftPM (\nCol -> (Assign nCol val))
(nameSingleColumn p st col)
nameInsert :: Pos ->
Table ->
AliasSymTab ->
[ColumnRef] ->
[[Value]] ->
PM Statement
nameInsert p tab st cols valss =
let symtab = insertTab tab st
in case symtab of
Right st1 -> liftPM (\nCols -> (Insert tab nCols valss))
(nameColumns p st1 cols)
--this should not happen in isert statements
Left err -> throwPM p err
nameDelete :: Pos -> Table -> AliasSymTab -> Condition -> PM Statement
nameDelete p tab st cond =
let symtab = insertTab tab st
in case symtab of
Right st1 -> liftPM (\nCond -> (Delete tab nCond))
(nameCond p st1 cond)
--this should not happen in delete statements
Left err -> throwPM p err
--insert of one or more wrapped in TableRef-node
insertTabs :: TableRef -> AliasSymTab -> Either String AliasSymTab
insertTabs (TableRef tab join) st =
checkJoinForTabs join (insertTab tab st)
checkJoinForTabs :: Maybe JoinClause ->
Either String AliasSymTab ->
Either String AliasSymTab
checkJoinForTabs Nothing st = st
checkJoinForTabs (Just (CrossJoin tab join)) (Right st) =
checkJoinForTabs join (insertTab tab st)
checkJoinForTabs (Just (InnerJoin tab _ join)) (Right st) =
checkJoinForTabs join (insertTab tab st)
checkJoinForTabs (Just _) (Left tab) = (Left tab)
-- inserts a single table into the symbol table, returns modified symbol table
-- Increases count for CDBI-alias and also inserts it.
-- In case of default alias "table" also the tablename itself is
-- inserted as alias.
insertTab :: Table -> AliasSymTab -> Either String AliasSymTab
insertTab (Table name al _) st =
let cnt = lookupSecondTable (toLowerCase name) st
in let n = case cnt of
Nothing -> 0
Just c -> c
in if(al == "table")
then case lookupFirstTable (toLowerCase name) st of
Nothing -> Right (insertDefFirstTab
(toLowerCase al)
[(name, al, n)]
(toLowerCase name)
(insertFirstTable (toLowerCase name)
[(name, name, n)]
(Just _) -> Left ("Ambiguous table reference: "++name)
else case lookupFirstTable (toLowerCase al) st of
Nothing -> Right (insertFirstTable
(toLowerCase al)
[(name, al, n)]
(insertSecondTable (toLowerCase name)
(Just _) -> Left ("Alias "++al++" was defined for more "
++"than one table.")
nameColumns :: Pos -> AliasSymTab -> [ColumnRef] -> PM [ColumnRef]
nameColumns p st cols = sequencePM (map (nameSingleColumn p st) cols)
-- naming-function for ColumnRef-node
-- replaces alias with table name and inserts CDBI-alias into column node
-- If alias is not found an error is thrown.
-- Warning is generated if notation of aliases differs.
nameSingleColumn :: Pos -> AliasSymTab -> ColumnRef -> PM ColumnRef
nameSingleColumn p st (Column (Unique pseudo) column typ nullable _) =
case lookupFirstTable (toLowerCase pseudo) st of
Nothing -> throwPM p ("Table alias "++pseudo++
" was not defined or "++
"is not visible.")
(Just ((tab, orgAl, cnt):_)) ->
if orgAl == pseudo
then cleanPM (Column (Unique tab) column typ nullable cnt)
else warnOKPM (Column (Unique tab) column typ nullable cnt)
[(p, ("Found different notation of same "++
"alias: "++pseudo++" and "++orgAl))]
nameSingleColumn p st (Column (Def pseudo) column typ nullable cnt) =
case lookupFirstTable (toLowerCase (head pseudo)) st of --can just be default name "table"
Nothing -> throwPM p ("No table given for column "
++column++". Maybe alias was"
++" defined but not used.")
(Just tabs) -> let tns = map fstOfTriple tabs
in cleanPM (Column (Def tns) column typ nullable cnt)
-- inserts correct CDBI-Alias into each table-node
-- ensures that alias is unique
nameTable :: Pos -> AliasSymTab -> Table -> PM Table
nameTable p st (Table name al _) =
let key = if(al == "table")
then (toLowerCase name)
else (toLowerCase al)
in case lookupFirstTable key st of
Nothing -> throwPM p ("No Table found for table "++name) --can not happen
(Just ((n, _ , cnt):_)) -> if ((toLowerCase n) == (toLowerCase name))
then cleanPM (Table name al cnt)
else throwPM p ("Alias "++key++
" is defined for" ++
" more than one table.")
-- Lookup-function used by relationship constraint
-- Error is thrown if alias was not defined, a warning
-- is generated if notation differs from defined alias.
getTable :: Pos -> String -> AliasSymTab -> PM (String, Int)
getTable p pseudo st =
case lookupFirstTable (toLowerCase pseudo) st of
Nothing -> throwPM p ("Table alias "++pseudo++
" was not defined.")
(Just ((tab, orgAl, cnt):_)) ->
if orgAl == pseudo
then cleanPM (tab, cnt)
else warnOKPM (tab, cnt)
[(p, ("Found different notation of same "++
"alias or table: "++pseudo++" and "++
toLowerCase :: String -> String
toLowerCase str = map toLower str
fstOfTriple :: (a,b,c) -> a
fstOfTriple (ele, _ , _) = ele
\ No newline at end of file
This diff is collapsed.
--- This module defines the structure of the parser information
--- used to pass information of the data model from an ERD-term to
--- the SQL-Translator which is part of currypp.
--- It consists of a tuple of the path to the corresponding database as string,
--- and the name of the corresponding CDBI-module as String,