Initial work on Scala parser combinators for reading sdp (session description protocol) files.

SdpParser.scala

package org.kafecho.learning.parser

import scala.util.parsing.combinator.RegexParsers
import java.net.URI

sealed trait SdpValue

sealed trait Username
case class UserLogin(login: String) extends Username
case object UserIdsNotSupported extends Username

sealed trait Media
case object Audio extends Media
case object Video extends Media
case object Text extends Media
case object Application extends Media

sealed trait Attribute
case class BinaryAttribute(value: String) extends Attribute
case class KeyValueAttribute(key: String, value: String) extends Attribute

case class ProtocolVersion(version: Long) extends SdpValue
case class Origin(username: Username, sessionId: Long, sessionVersion: Long, netType: String, addrType: String, unicastAddress: Option[String]) extends SdpValue
case class SessionName(name: String) extends SdpValue
case class Information(information: String) extends SdpValue
case class Protocol(protocol: String)

case class MediaField(media: Media, port: Long, protocol: Protocol, payloadType: Long)

/** Prototype parser for the Session Description Protocol.
  * See http://tools.ietf.org/html/rfc4566.html#page-39
  * Also see http://www.regular-expressions.info/posixbrackets.html for some help on regexes.
  */
class SdpParser extends RegexParsers {
  def nonWsString: Parser[String] = """\p{Graph}+""".r

  def number: Parser[Long] = """\p{Digit}+""".r ^^ { _.toLong }

  def text: Parser[String] = """\p{Print}+""".r

  def username: Parser[Username] = ("-" | nonWsString) ^^ {
    case "-" => UserIdsNotSupported
    case s: String => UserLogin(s)
  }

  def netType: Parser[String] = "IN"

  def addrType: Parser[String] = "IP4|IP6".r

  def media: Parser[Media] = ("video" | "audio" | "text" | "application") ^^ {
    case "video" => Video
    case "audio" => Audio
    case "text" => Text
    case "application" => Application
  }
  
  val token = """( \x21 | [\x23-\x27]|[\x2A-\x2B]|[\x2D-\x2E]|[\x30-\x39]|[\x41-\x5A]|[\x5E-\x7E])+""".r

  def protocolVersionField: Parser[ProtocolVersion] = "v=" ~ number ^^ { case _ ~ n => ProtocolVersion(n) }

  def originField: Parser[Origin] = ("o=" ~ username ~ number ~ number ~ netType ~ addrType ~ opt(nonWsString)) ^^ {
    case _ ~ username ~ sessionId ~ sessionVersion ~ netType ~ addrType ~ unicastAddress =>
    Origin(username, sessionId, sessionVersion, netType, addrType, unicastAddress)
  }

  def sessionNameField: Parser[SessionName] = ("s=" ~ text) ^^ { case _ ~ text => SessionName(text) }
  
  def informationField: Parser[Information] = ("i=" ~ text) ^^ { case _ ~ text => Information(text) }

  def uriField : Parser[URI] = ("u=" ~ text) ^^ { case _ ~ s =>  new URI(s)}

  def protocol : Parser[Protocol] = ("""RTP/AVP""" | "udp") ^^ { s => Protocol(s) }
  
  def mediaField : Parser[MediaField] = ("m=" ~ media ~ number ~ protocol ~ number) ^^ { case _ ~ media ~ port ~ protocol ~ payloadType =>
  	MediaField(media, port, protocol, payloadType)
  }
  
  def keyValueAttribute : Parser[Attribute]= ( token ~ ":" ~ nonWsString ) ^^ { case key ~ ":" ~ value => KeyValueAttribute(key, value)}
  
  def binaryAttribute: Parser[Attribute]= nonWsString ^^ { s => BinaryAttribute(s)}
  
  def attributeField : Parser[List[Attribute]] = "a=" ~ rep(keyValueAttribute | binaryAttribute) ^^ { case _ ~ l => l }

  def mediaDescriptions: Parser[Any] = rep(mediaField ~ rep (attributeField))
}

object SdpParser extends App{

  val src = """
m=audio 49170 RTP/AVP 0
m=audio 49170 RTP/AVP 0
a=rtpmap:99 h263-1998/90000"""

  val parser = new SdpParser
  val parseResult = parser.parseAll(parser.mediaDescriptions, src)
  println (parseResult)

}