@@ -11,7 +11,7 @@ import Global._ trait Parsers { private val TAIL_ERROR_PATTERN = "Unexpected trailing characters: '%s'" - implicit def literal(str: String): Parser[String] = new LiteralParser(str) + implicit def literal(str: String) = new LiteralParser(str) def opt[A](p: Parser[A]) = p? @@ -296,11 +296,7 @@ trait Parsers { def +? = (this+)? - /** - * WARNING: may cause non-termination if used recursively. Really - * should be restricted to {@link TerminalParser}. - */ - def \(not: Parser[Any]): Parser[R] = new NonTerminalParser[R] { // TODO + def \(not: TerminalParser[Any]): Parser[R] = new NonTerminalParser[R] { def computeFirst(seen: Set[Parser[Any]]) = self.computeFirst(seen) def queue(t: Trampoline, in: LineStream)(f: Result[R]=>Unit) { @@ -382,7 +378,7 @@ trait Parsers { case other => super.~(other) } - override def \(not: Parser[Any]) = new TerminalParser[R] { + override def \(not: TerminalParser[Any]) = new TerminalParser[R] { def computeFirst(s: Set[Parser[Any]]) = self.computeFirst(s) def parse(in: LineStream) = self.parse(in) match { src/main/scala/edu/uwm/cs/gll/Parsers.scala @@ -8,7 +8,7 @@ trait RegexParsers extends Parsers { protected val whitespace = """\s+"""r protected val skipWhitespace = true - override implicit def literal(str: String): Parser[String] = { + override implicit def literal(str: String): LiteralParser = { if (skipWhitespace) { new LiteralParser(str) { override def computeFirst(seen: Set[Parser[Any]]) = Some(UniversalOptCharSet) // because of whitespace @@ -19,7 +19,7 @@ trait RegexParsers extends Parsers { } else super.literal(str) } - implicit def regex(r: Regex): Parser[String] = { + implicit def regex(r: Regex): RegexParser = { if (skipWhitespace) { new RegexParser(r) { override def computeFirst(seen: Set[Parser[Any]]) = Some(UniversalOptCharSet) @@ -29,7 +29,17 @@ trait RegexParsers extends Parsers { } else new RegexParser(r) } - implicit def disjunctiveRegex(left: Regex) = new RichParser(regex(left)) + override implicit def disjunctiveLiterals(left: String) = new RichParser(literal(left)) { + def |(right: Regex) = new RegexParser(new Regex("(" + escapeRegex(left) + ")|(" + right + ")")) + + def |(right: String) = new RegexParser(new Regex("(" + escapeRegex(left) + ")|(" + escapeRegex(right) + ")")) + } + + implicit def disjunctiveRegex(left: Regex) = new RichParser(regex(left)) { + def |(right: Regex) = new RegexParser(new Regex("(" + left + ")|(" + right + ")")) + + def |(right: String) = new RegexParser(new Regex("(" + left + ")|(" + escapeRegex(right) + ")")) + } implicit def funRegexSyntax(p: Regex) = new RichSyntax1(regex(p)) @@ -42,6 +52,17 @@ trait RegexParsers extends Parsers { super.processTail(newTail) } + private def escapeRegex(str: String) = { + val specialChars = Set('[', ']', '{', '}', '\\', '|', '*', '+', '?', '^', '$', '(', ')') + + str.toCharArray.foldLeft("") { (str, c) => + if (specialChars contains c) + str + '\\' + c + else + str + c + } + } + private def handleWhitespace(s: LineStream) = s.drop(whitespace findPrefixOf s map { _.length } getOrElse 0) src/main/scala/edu/uwm/cs/gll/RegexParsers.scala 73f6950196d72f44b0e2f49035a3c31eef872d7b Daniel Spiewak djspiewak@gmail.com http://github.com/djspiewak/gll-combinators/commit/a0c2e3e0b6dd31e3802a65f93077b43150b28ec9 a0c2e3e0b6dd31e3802a65f93077b43150b28ec9 2009-10-31T09:08:27-07:00 2009-10-31T09:08:27-07:00 Added | combinator override for string parsers within the context of regexps 23fdfae9d872c379aa0a7feb5b329ed2f3439cff Daniel Spiewak djspiewak@gmail.com