wenzelm@34166: /*  Title:      Pure/Isar/outer_syntax.scala
wenzelm@34166:     Author:     Makarius
wenzelm@34166: 
wenzelm@34166: Isabelle/Isar outer syntax.
wenzelm@34166: */
wenzelm@34166: 
wenzelm@34166: package isabelle
wenzelm@34166: 
wenzelm@34166: 
wenzelm@34166: import scala.util.parsing.input.{Reader, CharSequenceReader}
wenzelm@43411: import scala.collection.mutable
wenzelm@34166: 
wenzelm@34166: 
wenzelm@43774: object Outer_Syntax
wenzelm@43774: {
wenzelm@43774:   def quote_string(str: String): String =
wenzelm@43774:   {
wenzelm@43774:     val result = new StringBuilder(str.length + 10)
wenzelm@43774:     result += '"'
wenzelm@43774:     for (s <- Symbol.iterator(str)) {
wenzelm@43774:       if (s.length == 1) {
wenzelm@43774:         val c = s(0)
wenzelm@43774:         if (c < 32 && c != YXML.X && c != YXML.Y || c == '\\' || c == '"') {
wenzelm@43774:           result += '\\'
wenzelm@43774:           if (c < 10) result += '0'
wenzelm@43774:           if (c < 100) result += '0'
wenzelm@43774:           result ++= (c.asInstanceOf[Int].toString)
wenzelm@43774:         }
wenzelm@43774:         else result += c
wenzelm@43774:       }
wenzelm@43774:       else result ++= s
wenzelm@43774:     }
wenzelm@43774:     result += '"'
wenzelm@43774:     result.toString
wenzelm@43774:   }
wenzelm@46626: 
wenzelm@46941:   val empty: Outer_Syntax = new Outer_Syntax()
wenzelm@48870: 
wenzelm@46941:   def init(): Outer_Syntax = new Outer_Syntax(completion = Completion.init())
wenzelm@48870: 
wenzelm@48870:   def init_pure(): Outer_Syntax =
wenzelm@48870:     init() + ("theory", Keyword.THY_BEGIN) + ("ML_file", Keyword.THY_LOAD)
wenzelm@43774: }
wenzelm@43774: 
wenzelm@46712: final class Outer_Syntax private(
wenzelm@48864:   keywords: Map[String, (String, List[String])] = Map.empty,
wenzelm@46626:   lexicon: Scan.Lexicon = Scan.Lexicon.empty,
wenzelm@46941:   val completion: Completion = Completion.empty)
wenzelm@34166: {
wenzelm@48660:   override def toString: String =
wenzelm@48864:     (for ((name, (kind, files)) <- keywords) yield {
wenzelm@48660:       if (kind == Keyword.MINOR) quote(name)
wenzelm@48864:       else
wenzelm@48864:         quote(name) + " :: " + quote(kind) +
wenzelm@48864:         (if (files.isEmpty) "" else " (" + commas_quote(files) + ")")
wenzelm@48671:     }).toList.sorted.mkString("keywords\n  ", " and\n  ", "")
wenzelm@48660: 
wenzelm@48864:   def keyword_kind_files(name: String): Option[(String, List[String])] = keywords.get(name)
wenzelm@48864:   def keyword_kind(name: String): Option[String] = keyword_kind_files(name).map(_._1)
wenzelm@38471: 
wenzelm@48872:   def thy_load_commands: List[String] =
wenzelm@48872:     (for ((name, (Keyword.THY_LOAD, _)) <- keywords.iterator) yield name).toList
wenzelm@48872: 
wenzelm@48864:   def + (name: String, kind: (String, List[String]), replace: String): Outer_Syntax =
wenzelm@46626:     new Outer_Syntax(
wenzelm@46626:       keywords + (name -> kind),
wenzelm@46626:       lexicon + name,
wenzelm@48864:       if (Keyword.control(kind._1)) completion else completion + (name, replace))
wenzelm@34166: 
wenzelm@48864:   def + (name: String, kind: (String, List[String])): Outer_Syntax = this + (name, kind, name)
wenzelm@48864:   def + (name: String, kind: String): Outer_Syntax = this + (name, (kind, Nil), name)
wenzelm@36947:   def + (name: String): Outer_Syntax = this + (name, Keyword.MINOR)
wenzelm@48706: 
wenzelm@48707:   def add_keywords(header: Document.Node.Header): Outer_Syntax =
wenzelm@48707:     (this /: header.keywords) {
wenzelm@48708:       case (syntax, ((name, Some((kind, _))))) =>
wenzelm@48708:         syntax + (Symbol.decode(name), kind) + (Symbol.encode(name), kind)
wenzelm@48708:       case (syntax, ((name, None))) =>
wenzelm@48708:         syntax + Symbol.decode(name) + Symbol.encode(name)
wenzelm@46940:     }
wenzelm@34166: 
wenzelm@34166:   def is_command(name: String): Boolean =
wenzelm@40458:     keyword_kind(name) match {
wenzelm@36947:       case Some(kind) => kind != Keyword.MINOR
wenzelm@34166:       case None => false
wenzelm@34166:     }
wenzelm@34166: 
wenzelm@40454:   def heading_level(name: String): Option[Int] =
wenzelm@46969:   {
wenzelm@46969:     keyword_kind(name) match {
wenzelm@46969:       case _ if name == "header" => Some(0)
wenzelm@46969:       case Some(Keyword.THY_HEADING1) => Some(1)
wenzelm@46969:       case Some(Keyword.THY_HEADING2) | Some(Keyword.PRF_HEADING2) => Some(2)
wenzelm@46969:       case Some(Keyword.THY_HEADING3) | Some(Keyword.PRF_HEADING3) => Some(3)
wenzelm@46969:       case Some(Keyword.THY_HEADING4) | Some(Keyword.PRF_HEADING4) => Some(4)
wenzelm@46969:       case Some(kind) if Keyword.theory(kind) => Some(5)
wenzelm@46969:       case _ => None
wenzelm@40454:     }
wenzelm@46969:   }
wenzelm@40454: 
wenzelm@40454:   def heading_level(command: Command): Option[Int] =
wenzelm@40454:     heading_level(command.name)
wenzelm@40454: 
wenzelm@34166: 
wenzelm@34166:   /* tokenize */
wenzelm@34166: 
wenzelm@36956:   def scan(input: Reader[Char]): List[Token] =
wenzelm@34166:   {
wenzelm@34166:     import lexicon._
wenzelm@34166: 
wenzelm@43695:     parseAll(rep(token(is_command)), input) match {
wenzelm@34166:       case Success(tokens, _) => tokens
wenzelm@34264:       case _ => error("Unexpected failure of tokenizing input:\n" + input.source.toString)
wenzelm@34166:     }
wenzelm@34166:   }
wenzelm@34166: 
wenzelm@36956:   def scan(input: CharSequence): List[Token] =
wenzelm@34166:     scan(new CharSequenceReader(input))
wenzelm@43411: 
wenzelm@43411:   def scan_context(input: CharSequence, context: Scan.Context): (List[Token], Scan.Context) =
wenzelm@43411:   {
wenzelm@43411:     import lexicon._
wenzelm@43411: 
wenzelm@43411:     var in: Reader[Char] = new CharSequenceReader(input)
wenzelm@43411:     val toks = new mutable.ListBuffer[Token]
wenzelm@43411:     var ctxt = context
wenzelm@43411:     while (!in.atEnd) {
wenzelm@43695:       parse(token_context(is_command, ctxt), in) match {
wenzelm@43411:         case Success((x, c), rest) => { toks += x; ctxt = c; in = rest }
wenzelm@43411:         case NoSuccess(_, rest) =>
wenzelm@43411:           error("Unexpected failure of tokenizing input:\n" + rest.source.toString)
wenzelm@43411:       }
wenzelm@43411:     }
wenzelm@43411:     (toks.toList, ctxt)
wenzelm@43411:   }
wenzelm@34166: }