This is a parser for SIGMET which is a format for weather information. See also: http://d.hatena.ne.jp/hideshi_o/20130105/1357367954

sigmetparser.scala

abstract sealed class Bulletin

case class SigmetBulletin (
	 header:Header
	,firstLine:FirstLine
	,mainBody:MainBody
	) extends Bulletin

case class Header(
	 identificationMessage:String = ""
	,icao:String = ""
	,dateTimeGroup:String = ""
	,correction:String = ""
	) extends Bulletin

case class FirstLine(
	 icaoLocationIndicatorOfATS:String
	,messageIdentifier:String
	,dailySequenceNumber:String
	,validityIndicator:String
	,beginAndEndDateTimeGroup:String
	,icaoLocationIndicatorOfMWO:String
	) extends Bulletin

case class MainBody(
	 firuircta:FIRUIRCTA
	,phenomenon:String
	,obsfcst:OBSFCST
	,locationOfThePhenomenon:String
	,flightLevelAndExtent:FlightLevel
	,movement:Movement
	,expectedChanges:String
	) extends Bulletin

case class FIRUIRCTA (
	 icao:String
	,name:String
	,kind:String 
	) extends Bulletin

case class OBSFCST (
	 observedOrFcst:String
	,optionalTimeGroup:String = ""
	) extends Bulletin

case class FlightLevel (
	 indication:String = ""
	,level:String = ""
	) extends Bulletin

case class Movement (
	 indication:String = ""
	,direction:String = ""
	,speed:String = ""
	) extends Bulletin

import scala.util.parsing.combinator.RegexParsers

class SigmetParser extends RegexParsers {
	def parse(data:String) = parseAll(all, data)

	def all:Parser[Any] = {
		opt(header)~firstline~mainBody ^^ {
			case Some(header)~firstLine~mainBody =>
				SigmetBulletin(header, firstLine, mainBody)
			case None~firstLine~mainBody =>
				SigmetBulletin(Header(), firstLine, mainBody)
		}
	}
	def header:Parser[Header] = {
		("(WC|WS|WV)[A-Z]{2}[0-9]{2}".r)~("[A-Z]{4}".r)~("[0-9]{6}".r)~("(CC[AB])?".r) ^^ {
			case identificationMessage~icao~dateTimeGroup~correction =>
				Header(identificationMessage,icao,dateTimeGroup,correction)
		}
	}
	def firstline:Parser[FirstLine] = {
		("[A-Z]{4}".r)~"SIGMET"~("(([A-Z]{1,2})?[0-9]?[0-9])".r)~"VALID"~("[0-9]{6}/[0-9]{6}".r)~("[A-Z]{4}-".r) ^^ {
			case icaoLocationIndicatorOfATS~messageIdentifier~dailySequenceNumber~validityIndicator~beginAndEndDateTimeGroup~icaoLocationIndicatorOfMWO =>
				FirstLine(icaoLocationIndicatorOfATS,messageIdentifier,dailySequenceNumber,validityIndicator,beginAndEndDateTimeGroup,icaoLocationIndicatorOfMWO)
		}
	}
	def mainBody:Parser[MainBody] = {
		firuircta~phenomenon~obsfcst~location~opt(flightLevel)~opt(movement)~opt(expectedChanges)<~(""".*""".r) ^^ {
			case firuircta~phenomenon~obsfcst~location~Some(flightLevel)~Some(movement)~Some(expectedChanges) =>
				MainBody(firuircta, phenomenon, obsfcst, location, flightLevel, movement, expectedChanges)
			case firuircta~phenomenon~obsfcst~location~Some(flightLevel)~None~Some(expectedChanges) =>
				MainBody(firuircta, phenomenon, obsfcst, location, flightLevel, Movement(), expectedChanges)
			case firuircta~phenomenon~obsfcst~location~Some(flightLevel)~Some(movement)~None =>
				MainBody(firuircta, phenomenon, obsfcst, location, flightLevel, movement, "")
			case firuircta~phenomenon~obsfcst~location~Some(flightLevel)~None~None =>
				MainBody(firuircta, phenomenon, obsfcst, location, flightLevel, Movement(), "")
			case firuircta~phenomenon~obsfcst~location~None~Some(movement)~Some(expectedChanges) =>
				MainBody(firuircta, phenomenon, obsfcst, location, FlightLevel(), movement, expectedChanges)
			case firuircta~phenomenon~obsfcst~location~None~None~Some(expectedChanges) =>
				MainBody(firuircta, phenomenon, obsfcst, location, FlightLevel(), Movement(), expectedChanges)
			case firuircta~phenomenon~obsfcst~location~None~Some(movement)~None =>
				MainBody(firuircta, phenomenon, obsfcst, location, FlightLevel(), movement, "")
			case firuircta~phenomenon~obsfcst~location~None~None~None =>
				MainBody(firuircta, phenomenon, obsfcst, location, FlightLevel(), Movement(), "")
		}
	}
	def firuircta:Parser[FIRUIRCTA] = {
		("""[A-Z]{4}""".r)~("""[A-Z\s]+?(FIR(/UIR)?|UIR|CTA)""".r) ^^ {
			case icao~name =>
				val xs = name.split("""\s""")
				FIRUIRCTA(icao, xs.init.reduceLeft{_ + " " + _}, xs.last)
		}
	}
	def phenomenon:Parser[String] = {
		"""(OBSC|EMBD|FRQ|SQL|SEV|FZRA|HVY|VA CLD|RDOACT CLD)\s?(TSGR|TS|TURB|ICE|MTW)?""".r
	}
	def obsfcst:Parser[OBSFCST] = {
		("""(OBS|FCST)""".r)~opt("""AT [0-9]{4,6}Z""".r) ^^ {
			case observedOrFcst~Some(optionalTimeGroup) => OBSFCST(observedOrFcst, optionalTimeGroup)
			case observedOrFcst~None => OBSFCST(observedOrFcst)
		}
	}
	def location:Parser[String] = {
		rep("""([NSEW]{1,3}\s|(WI(THIN)?\s)?(AREA\s)?([0-9]{2}\s?NM\s)?([NSEW]{1,3}\s)?)?(OF\s)?((LINE|ETNA)\s)?\(?([NSEW][0-9]{2,5}|-|AND)\)?""".r) ^^ {
			case (xs:List[String]) if !(xs.isEmpty) =>
				println(xs)
				xs.reduceLeft{_ + " " + _}
			case _ => ""
		}
	}
	def flightLevel:Parser[FlightLevel] = {
		("""(TOPS?\s)?(TOPS?\s|ABV\s|BLW\s|SFC\/)?FL[0-9]{3}((\/|-)[0-9]{3})?""".r) ^^ {
			case flightLevel =>
				val indication = flightLevel.split("FL")(0) + "FL"
				val level = flightLevel.split("FL")(1)
				FlightLevel(indication, level)
		}
	}
	def movement:Parser[Movement] = {
		("""MOV"""~("""[NSEW]{1,3}""".r)~opt("""[0-9]{2}\s?(KMH|KT)""".r) | """STNR""") ^^ {
	 		case indication~direction~Some(speed) => Movement(indication.toString, direction.toString, speed.toString)
	 		case indication~direction~None => Movement(indication.toString, direction.toString)
	 		case indication => Movement(indication.toString)
		}
	}
	def expectedChanges:Parser[String] = {
		("""(INTSF|WKN|NC)""".r)<~("""\s?=""".r)
	}
}

val bulletins = List(
	 "WSTH31 VTBS 121200 CCA VHHK SIGMET 4 VALID 031200/031600 VHHH- VHHK HONG KONG FIR EMBD TS FCST S OF N20 E OF E112 TOP FL360 STNR NC="
	,"WSSR20 WSSS 091131 WSJC SIGMET 3 VALID 091140/091540 WSSS- WSJC SINGAPORE FIR EMBD TS OBS AT 1130Z N OF N01 E OF E106 W OF E114 STNR NC="
	,"WSNT03 KKCI 032340 KZNY SIGMET C17 VALID 032345/040345 KKCI- KZNY NEW YORK OCEANIC FIR FRQ TS OBS WI AREA N2400 W05500 - N2300 W04930 - N1845 W05645 - N2100 W05800 - N2400 W05500 TOP FL450 MOV E 15KT INTSF="
	,"WSVS31 VVGL 122305 VVTS SIGMET 9 VALID 122330/130230 VVGL- VVTS HOCHIMINH FIR EMBD TS OBS S OF LINE N1420 E10930 - N1000 E10400 TOP FL280 MOV W 10KMH WKN="
	,"WSUK31 EGGY 121120 EGTT SIGMET 01 VALID 121125/121525 EGRR- EGTT LONDON FIR EMBD TSGR OBS AT 1115Z SE OF LINE N5130 E00200 - N5000 W00400 TOPS FL220 MOV NE 30KT NC="
	,"WSAU21 AMMC 280546 YBBB SIGMET BS02 VALID 280600/281000 YMMC- YBBB BRISBANE FIR SEV TURB FCST AT 0600Z WI S3900 E15100 - S4300 E15100 - S4300 E16000 - S4100 E16300 - S3700 E16300 - S3900 E16000 FL260/370 MOV E 20 KT NC="
	,"EDUU	SIGMET 1 VALID 050200/050400 EDZF- EDUU RHEIN UIR SEV TURB FCST NE OF LINE N5415 E01130 - N5320 E01425 FL245/350 MOV NE NC="
	,"EGTT	SIGMET 04 VALID 042300/050300 EGRR- EGTT LONDON FIR SEV TURB FCST S OF LINE N5000 W00800 - N5030 W00700 - N2030 W00500 - N4930 W00400 BLW FL060 MOV E 45KT INTSF="
	,"EGTT	SIGMET 01 VALID 050300/050700 EGRR- EGTT LONDON FIR SEV TURB OBS S OF LINE N4900 W00800 - N5000 W00400 - N5100 W00300 - N5200 W00100 - N5130 E00100 - N5100 E00130 BLW FL060 MOV ENE 45KT NC="
	,"EKDK	SIGMET 1 VALID 050025/050425 EKMI- EKDK COPENHAGEN FIR SEV TURB FCST AT 0020Z WITHIN 40NM OF LINE N57 E00609 - N5606 E01041 - N5501 E01247 FL240-390 MOV NE 15KT NC="
	,"LFFF	SIGMET 3 VALID 050100/050400 LFPW- LFFF PARIS FIR/UIR SEV TURB FCST N OF N4930 AND W OF E00000 SFC/FL050 MOV E 30KT INTSF="
	,"LFRR	SIGMET 1 VALID 050100/050400 LFPW- LFRR BREST FIR/UIR SEV TURB FCST WI N4930 W00415 - N5000 W00200 - N5000 W00015 - N4930 W00015 - N4845 W00400 SFC/FL050 MOV E 30KT INTSF ="
	,"LIRR	SIGMET 01 VALID 050130/050530 LIMM- LIRR ROMA FIR VA CLD OBS AT 040430Z WI 15 NM ESE OF ETNA (N3745 E1500) FL090/130 MOV SE 15 KT="
	)

val parser = new SigmetParser
bulletins.map(b => (b, parser.parse(b)))./*filter(r => r.successful).*/foreach{x =>
	println()
	println(x._1)
	println(x._2)
}