A parser combinator SPARQL grammar for Scala (work-in-progress).

Grammar.scala

/*
 * A parser combinator SPARQL grammar for Scala.
 * Written in November 2009 by Arto Bendiken <http://ar.to/>
 *
 * This is free and unencumbered software released into the public domain.
 * For more information, please refer to <http://unlicense.org/>
 */

package org.datagraph.sparql
import java.io.FileReader
import scala.util.parsing.combinator._
import scala.collection.mutable.ListBuffer

/**
 * @see http://www.w3.org/TR/rdf-sparql-query/#grammar
 */
object Grammar extends RegexParsers {
  def main(args: Array[String]) {
    for (arg <- args) {
      println(parseAll(Query, new FileReader(arg)))
    }
  }

  // [1]
  def Query: Parser[List[Any]] =
    Prologue ~ (SelectQuery | ConstructQuery | DescribeQuery | AskQuery) ^^ {
      case List() ~ query => query
      case prolog ~ query => prolog ++ query
    }

  // [2]
  def Prologue: Parser[List[Any]] =
    (BaseDecl?) ~ (PrefixDecl*) ^^ {
      case None ~ List()   => List()
      case base ~ List()   => List(base)
      case None ~ prefixes => List(prefixes)
      case base ~ prefixes => List(base, prefixes)
    }

  // [3]
  def BaseDecl: Parser[List[Any]] =
    "BASE" ~ IRI_REF ^^ {
      case _ ~ iri => List('BASE, iri)
    }

  // [4]
  def PrefixDecl: Parser[List[Any]] =
    "PREFIX" ~ PNAME_NS ~ IRI_REF ^^ {
      case _ ~ ns ~ iri => List('PREFIX, ns, iri)
    }

  // [5]
  def SelectQuery: Parser[List[Any]] =
    "SELECT" ~ (("DISTINCT" | "REDUCED")?) ~ ("*" | (Var+)) ~ (DatasetClause*) ~ WhereClause ~ SolutionModifier ^^ {
      case _ ~ None ~ vs ~ List() ~ w ~ None => List('SELECT, vs, w)
      case _ ~ None ~ vs ~ List() ~ w ~ s    => List('SELECT, vs, w, s)
      case _ ~ dr ~ vs ~ d ~ w ~ s           => List('SELECT, dr, vs, d, w, s)
    }

  // [6]
  def ConstructQuery: Parser[List[Any]] =
    "CONSTRUCT" ~ ConstructTemplate ~ (DatasetClause*) ~ WhereClause ~ SolutionModifier ^^ {
      case _ => List('CONSTRUCT) // TODO
    }

  // [7]
  def DescribeQuery: Parser[List[Any]] =
    "DESCRIBE" ~ ("*" | (VarOrIRIref+)) ~ (DatasetClause*) ~ (WhereClause?) ~ SolutionModifier ^^ {
      case _ => List('DESCRIBE) // TODO
    }

  // [8]
  def AskQuery: Parser[List[Any]] =
    "ASK" ~ (DatasetClause*) ~ WhereClause ^^ {
      case _ => List('ASK) // TODO
    }

  // [9]
  def DatasetClause: Parser[List[Any]] =
    "FROM" ~> (DefaultGraphClause | NamedGraphClause) ^^ (List('FROM, _))

  // [10]
  def DefaultGraphClause: Parser[Any] =
    SourceSelector

  // [11]
  def NamedGraphClause: Parser[List[Any]] =
    "NAMED" ~> SourceSelector ^^ (List('NAMED, _))

  // [12]
  def SourceSelector: Parser[Any] =
    IRIref

  // [13]
  def WhereClause: Parser[List[Any]] =
    ("WHERE"?) ~> GroupGraphPattern ^^ (List('WHERE, _))

  // [14]
  def SolutionModifier: Parser[Any] =
    (OrderClause?) ~ (LimitOffsetClauses?) ^^ {
      case None ~ None => None
      case xy          => xy
    }

  // [15]
  def LimitOffsetClauses: Parser[Any] =
    (LimitClause ~ (OffsetClause?)) | (OffsetClause ~ (LimitClause?))

  // [16]
  def OrderClause: Parser[List[Any]] =
    "ORDER" ~ "BY" ~ (OrderCondition+) ^^ {
      case _ ~ _ ~ condition => List('ORDERBY, condition)
    }

  // [17]
  def OrderCondition: Parser[Any] =
    (("ASC" | "DESC") ~ BrackettedExpression) | (Constraint | Var)

  // [18]
  def LimitClause: Parser[List[Any]] =
    "LIMIT" ~> INTEGER ^^ (List('LIMIT, _))

  // [19]
  def OffsetClause: Parser[List[Any]] =
    "OFFSET" ~> INTEGER ^^ (List('OFFSET, _))

  // [20]
  def GroupGraphPattern: Parser[List[Any]] =
    "{" ~> (TriplesBlock?) ~ (((GraphPatternNotTriples | Filter) <~ ("."?) ~> (TriplesBlock?))*) <~ "}" ^^ {
      case Some(a) ~ List() => a
      case a ~ b            => List(a, b) // FIXME
    }

  // [21]
  def TriplesBlock: Parser[List[Any]] =
    TriplesSameSubject ~ (("." ~ (TriplesBlock?))?) ^^ {
      case a ~ None => List('BGP) ++ a
      case a ~ b    => List('BGP, a, b) // FIXME
    }

  // [22]
  def GraphPatternNotTriples: Parser[Any] =
    OptionalGraphPattern | GroupOrUnionGraphPattern | GraphGraphPattern

  // [23]
  def OptionalGraphPattern: Parser[List[Any]] =
    "OPTIONAL" ~> GroupGraphPattern ^^ (List('OPTIONAL, _))

  // [24]
  def GraphGraphPattern: Parser[List[Any]] =
    "GRAPH" ~ VarOrIRIref ~ GroupGraphPattern ^^ {
      case _ ~ varOrIRI ~ bgp => List('GRAPH, varOrIRI, bgp)
    }

  // [25]
  def GroupOrUnionGraphPattern: Parser[Any] =
    GroupGraphPattern ~ (("UNION" ~ GroupGraphPattern)*)

  // [26]
  def Filter: Parser[List[Any]] =
    "FILTER" ~> Constraint ^^ (List('FILTER, _))

  // [27]
  def Constraint: Parser[Any] =
    BrackettedExpression | BuiltInCall | FunctionCall

  // [28]
  def FunctionCall: Parser[Any] =
    IRIref ~ ArgList

  // [29]
  def ArgList: Parser[Any] =
    NIL | ("(" ~> repsep(Expression, ",") <~ ")")

  // [30]
  def ConstructTemplate: Parser[Any] =
    "{" ~> (ConstructTriples?) <~ "}"

  // [31]
  def ConstructTriples: Parser[Any] =
    TriplesSameSubject ~ (("." ~> (ConstructTriples?))?)

  // [32]
  def TriplesSameSubject: Parser[List[Any]] =
    TriplesSameSubject1 | TriplesSameSubject2

  def TriplesSameSubject1: Parser[List[(Any, Any, Any)]] =
    (VarOrTerm ~ PropertyListNotEmpty) ^^ {
      case s ~ polist =>
        var out = new ListBuffer[(Any, Any, Any)]
        for ((p, os) <- polist) {
          for (o <- os) out += (s, p, o)
        }
        out.toList
    }

  def TriplesSameSubject2: Parser[List[Any]] =
    (TriplesNode ~ PropertyList) ^^ {
      case a ~ b => List(a, b) // FIXME
    }

  // [33]
  def PropertyListNotEmpty: Parser[List[(Any, List[Any])]] =
    repsep(Verb ~ ObjectList ^^ { case p ~ olist => (p, olist) }, ";") <~ (";"?)

  // [34]
  def PropertyList: Parser[Any] =
    PropertyListNotEmpty?

  // [35]
  def ObjectList: Parser[List[Any]] =
    repsep(Object, ",")

  // [36]
  def Object: Parser[Any] =
    GraphNode

  // [37]
  def Verb: Parser[Any] =
    VarOrIRIref | "a" // TODO: "a" => rdf:type

  // [38]
  def TriplesNode: Parser[Any] =
    Collection | BlankNodePropertyList

  // [39]
  def BlankNodePropertyList: Parser[Any] =
    "[" ~> PropertyListNotEmpty <~ "]"

  // [40]
  def Collection: Parser[Any] =
    "(" ~> (GraphNode+) <~ ")"

  // [41]
  def GraphNode: Parser[Any] =
    VarOrTerm | TriplesNode

  // [42]
  def VarOrTerm: Parser[Any] =
    Var | GraphTerm

  // [43]
  def VarOrIRIref: Parser[Any] =
    Var | IRIref

  // [44]
  def Var: Parser[Any] = //Parser[Symbol] =
    VAR1 | VAR2 ^^ { case x => println(x); ('VAR, x) }

  // [45]
  def GraphTerm: Parser[Any] =
    IRIref | RDFLiteral | NumericLiteral | BooleanLiteral | BlankNode | NIL

  // [46]
  def Expression: Parser[Any] =
    ConditionalOrExpression

  // [47]
  def ConditionalOrExpression: Parser[Any] =
    ConditionalAndExpression ~ (("||" ~ ConditionalAndExpression)*)

  // [48]
  def ConditionalAndExpression: Parser[Any] =
    ValueLogical ~ (("&&" ~ ValueLogical)*)

  // [49]
  def ValueLogical: Parser[Any] =
    RelationalExpression

  // [50]
  def RelationalExpression: Parser[Any] =
    NumericExpression ~
    ((("=" ~ NumericExpression) |
      ("!=" ~ NumericExpression) |
      ("<" ~ NumericExpression) |
      (">" ~ NumericExpression) |
      ("<=" ~ NumericExpression) |
      (">=" ~ NumericExpression))?)

  // [51]
  def NumericExpression: Parser[Any] =
    AdditiveExpression

  // [52]
  def AdditiveExpression: Parser[Any] =
    MultiplicativeExpression ~
    ((("+" ~ MultiplicativeExpression) |
      ("-" ~ MultiplicativeExpression) |
      NumericLiteralPositive |
      NumericLiteralNegative)*)

  // [53]
  def MultiplicativeExpression: Parser[Any] =
    UnaryExpression ~ ((("*" ~ UnaryExpression) | ("/" ~ UnaryExpression))*)

  // [54]
  def UnaryExpression: Parser[Any] =
    "!" ~ PrimaryExpression |
    "+" ~ PrimaryExpression |
    "-" ~ PrimaryExpression |
    PrimaryExpression

  // [55]
  def PrimaryExpression: Parser[Any] =
    BrackettedExpression | BuiltInCall | IRIrefOrFunction | RDFLiteral | NumericLiteral | BooleanLiteral | Var

  // [56]
  def BrackettedExpression: Parser[Any] =
    "(" ~> Expression <~ ")"

  // [57]
  def BuiltInCall: Parser[Any] =
    "STR" ~ "(" ~> Expression <~ ")" |
    "LANG" ~ "(" ~> Expression <~ ")" |
    "LANGMATCHES" ~ "(" ~> Expression <~ "," ~> Expression <~ ")" |
    "DATATYPE" ~ "(" ~> Expression <~ ")" |
    "BOUND" ~ "(" ~> Var <~ ")" |
    "sameTerm" ~ "(" ~> Expression <~ "," ~> Expression <~ ")" |
    "isIRI" ~ "(" ~> Expression <~ ")" |
    "isURI" ~ "(" ~> Expression <~ ")" |
    "isBLANK" ~ "(" ~> Expression <~ ")" |
    "isLITERAL" ~ "(" ~> Expression <~ ")" |
    RegexExpression

  // [58]
  def RegexExpression: Parser[Any] =
    "REGEX" ~ "(" ~ Expression <~ "," ~> Expression ~ (("," ~> Expression)?) ~ ")"

  // [59]
  def IRIrefOrFunction: Parser[Any] =
    IRIref ~ (ArgList?)

  // [60]
  def RDFLiteral: Parser[Any] =
    String ~ ((LANGTAG | ("^^" ~> IRIref))?)

  // [61]
  def NumericLiteral: Parser[Any] =
    NumericLiteralUnsigned | NumericLiteralPositive | NumericLiteralNegative

  // [62]
  def NumericLiteralUnsigned: Parser[Any] =
    INTEGER | DECIMAL | DOUBLE

  // [63]
  def NumericLiteralPositive: Parser[Any] =
    INTEGER_POSITIVE | DECIMAL_POSITIVE | DOUBLE_POSITIVE

  // [64]
  def NumericLiteralNegative: Parser[Any] =
    INTEGER_NEGATIVE | DECIMAL_NEGATIVE | DOUBLE_NEGATIVE

  // [65]
  def BooleanLiteral: Parser[Any] =
    "true" | "false"

  // [66]
  def String: Parser[Any] =
    STRING_LITERAL1 | STRING_LITERAL2 | STRING_LITERAL_LONG1 | STRING_LITERAL_LONG2

  // [67]
  def IRIref: Parser[Any] =
    IRI_REF | PrefixedName

  // [68]
  def PrefixedName: Parser[Any] =
    PNAME_LN | PNAME_NS

  // [69]
  def BlankNode: Parser[Any] =
    BLANK_NODE_LABEL | ANON

  // [70]
  def IRI_REF: Parser[Any] =
    "<" ~> """([^<>"{}|^`\])*""".r <~ ">" // FIXME

  // [71]
  def PNAME_NS: Parser[Any] =
    (PN_PREFIX?) <~ ":"

  // [72]
  def PNAME_LN: Parser[Any] =
    PNAME_NS ~ PN_LOCAL

  // [73]
  def BLANK_NODE_LABEL: Parser[Any] =
    "_:" ~> PN_LOCAL

  // [74]
  def VAR1: Parser[Symbol] =
    "?" ~> VARNAME

  // [75]
  def VAR2: Parser[Symbol] =
    "$" ~> VARNAME

  // [76]
  def LANGTAG: Parser[Any] =
    "@" ~> "[a-zA-Z]+".r ~ (("-" ~ "[a-zA-Z0-9]+".r)*)

  // [77]
  def INTEGER: Parser[Int] =
    "[0-9]+".r ^^ (_.toInt)

  // [78]
  def DECIMAL: Parser[Any] =
    ("[0-9]+".r ~ "." ~ "[0-9]*".r) | ("." ~ "[0-9]+".r)

  // [79]
  def DOUBLE: Parser[Any] =
    ("[0-9]+".r ~ "." ~ "[0-9]*".r ~ EXPONENT) | ("." ~ "([0-9])+".r ~ EXPONENT) | ("([0-9])+".r ~ EXPONENT)

  // [80]
  def INTEGER_POSITIVE: Parser[Int] =
    "+" ~> INTEGER

  // [81]
  def DECIMAL_POSITIVE: Parser[Any] =
    "+" ~> DECIMAL

  // [82]
  def DOUBLE_POSITIVE: Parser[Any] =
    "+" ~> DOUBLE

  // [83]
  def INTEGER_NEGATIVE: Parser[Int] =
    "-" ~> INTEGER ^^ (_ * -1)

  // [84]
  def DECIMAL_NEGATIVE: Parser[Any] =
    "-" ~ DECIMAL // TODO

  // [85]
  def DOUBLE_NEGATIVE: Parser[Any] =
    "-" ~ DOUBLE // TODO

  // [86]
  def EXPONENT: Parser[Any] =
    "[eE]".r ~ "[+-]?".r ~ "[0-9]+".r

  // [87]
  def STRING_LITERAL1: Parser[String] =
    "'" ~> TODO <~ "'"

  // [88]
  def STRING_LITERAL2: Parser[String] =
    "\"" ~> TODO <~ "\""

  // [89]
  def STRING_LITERAL_LONG1: Parser[String] =
    "'''" ~> TODO <~ "'''"

  // [90]
  def STRING_LITERAL_LONG2: Parser[String] =
    "\"\"\"" ~> TODO <~ "\"\"\""

  // [91]
  def ECHAR: Parser[Any] =
    "\\" ~ """[tbnrf\"']""".r

  // [92]
  def NIL: Parser[Any] =
    "(" ~ (WS*) ~ ")"

  // [93]
  def WS: Parser[Any] =
    """\s+""".r // FIXME?

  // [94]
  def ANON: Parser[Any] =
    "[" ~ (WS*) ~ "]"

  // [95]
  def PN_CHARS_BASE: Parser[Any] =
    "[A-Z]".r | "[a-z]".r | TODO

  // [96]
  def PN_CHARS_U: Parser[Any] =
    PN_CHARS_BASE | "_"

  // [97]
  def VARNAME: Parser[Symbol] =
    """[a-zA-Z_]\w*""".r ^^ (Symbol(_)) // FIXME

  // [98]
  def PN_CHARS: Parser[Any] =
    PN_CHARS_U | "-" | "[0-9]".r | TODO

  // [99]
  def PN_PREFIX: Parser[Any] =
    PN_CHARS_BASE ~ ((((PN_CHARS | ".")*) ~ PN_CHARS)?)

  // [100]
  def PN_LOCAL: Parser[Any] =
    (PN_CHARS_U | "[0-9]".r) ~ ((((PN_CHARS | ".")*) ~ PN_CHARS)?)

  def TODO: Parser[String] = "TODO".r
}

Nice work! Wish you'd finish this and document its usage.