src/Pure/Isar/outer_syntax.scala
author wenzelm
Wed Nov 05 21:59:21 2014 +0100 (2014-11-05)
changeset 58907 0ee3563803c9
parent 58901 47809a811eba
child 58938 0c45680b7d9d
permissions -rw-r--r--
more uniform header_keywords in ML/Scala;
tuned signature;
wenzelm@34166
     1
/*  Title:      Pure/Isar/outer_syntax.scala
wenzelm@34166
     2
    Author:     Makarius
wenzelm@34166
     3
wenzelm@34166
     4
Isabelle/Isar outer syntax.
wenzelm@34166
     5
*/
wenzelm@34166
     6
wenzelm@34166
     7
package isabelle
wenzelm@34166
     8
wenzelm@34166
     9
wenzelm@34166
    10
import scala.util.parsing.input.{Reader, CharSequenceReader}
wenzelm@43411
    11
import scala.collection.mutable
wenzelm@58706
    12
import scala.annotation.tailrec
wenzelm@34166
    13
wenzelm@34166
    14
wenzelm@43774
    15
object Outer_Syntax
wenzelm@43774
    16
{
wenzelm@58706
    17
  /* syntax */
wenzelm@58706
    18
wenzelm@58706
    19
  val empty: Outer_Syntax = new Outer_Syntax()
wenzelm@58706
    20
wenzelm@58706
    21
  def init(): Outer_Syntax = new Outer_Syntax(completion = Completion.init())
wenzelm@58706
    22
wenzelm@58706
    23
wenzelm@58706
    24
  /* string literals */
wenzelm@58706
    25
wenzelm@43774
    26
  def quote_string(str: String): String =
wenzelm@43774
    27
  {
wenzelm@43774
    28
    val result = new StringBuilder(str.length + 10)
wenzelm@43774
    29
    result += '"'
wenzelm@43774
    30
    for (s <- Symbol.iterator(str)) {
wenzelm@43774
    31
      if (s.length == 1) {
wenzelm@43774
    32
        val c = s(0)
wenzelm@43774
    33
        if (c < 32 && c != YXML.X && c != YXML.Y || c == '\\' || c == '"') {
wenzelm@43774
    34
          result += '\\'
wenzelm@43774
    35
          if (c < 10) result += '0'
wenzelm@43774
    36
          if (c < 100) result += '0'
wenzelm@43774
    37
          result ++= (c.asInstanceOf[Int].toString)
wenzelm@43774
    38
        }
wenzelm@43774
    39
        else result += c
wenzelm@43774
    40
      }
wenzelm@43774
    41
      else result ++= s
wenzelm@43774
    42
    }
wenzelm@43774
    43
    result += '"'
wenzelm@43774
    44
    result.toString
wenzelm@43774
    45
  }
wenzelm@46626
    46
wenzelm@58696
    47
wenzelm@58697
    48
  /* line-oriented structure */
wenzelm@58696
    49
wenzelm@58697
    50
  object Line_Structure
wenzelm@58696
    51
  {
wenzelm@58700
    52
    val init = Line_Structure()
wenzelm@58696
    53
  }
wenzelm@58696
    54
wenzelm@58700
    55
  sealed case class Line_Structure(
wenzelm@58700
    56
    improper: Boolean = true,
wenzelm@58700
    57
    command: Boolean = false,
wenzelm@58700
    58
    depth: Int = 0,
wenzelm@58700
    59
    span_depth: Int = 0,
wenzelm@58700
    60
    after_span_depth: Int = 0)
wenzelm@58706
    61
wenzelm@58706
    62
wenzelm@58706
    63
  /* overall document structure */
wenzelm@58706
    64
wenzelm@58706
    65
  sealed abstract class Document { def length: Int }
wenzelm@58747
    66
  case class Document_Block(name: String, text: String, body: List[Document]) extends Document
wenzelm@58706
    67
  {
wenzelm@58706
    68
    val length: Int = (0 /: body)(_ + _.length)
wenzelm@58706
    69
  }
wenzelm@58747
    70
  case class Document_Atom(command: Command) extends Document
wenzelm@58706
    71
  {
wenzelm@58706
    72
    def length: Int = command.length
wenzelm@58706
    73
  }
wenzelm@43774
    74
}
wenzelm@43774
    75
wenzelm@46712
    76
final class Outer_Syntax private(
wenzelm@58900
    77
  val keywords: Keyword.Keywords = Keyword.Keywords.empty,
wenzelm@53280
    78
  val completion: Completion = Completion.empty,
wenzelm@55749
    79
  val language_context: Completion.Language_Context = Completion.Language_Context.outer,
wenzelm@56393
    80
  val has_tokens: Boolean = true) extends Prover.Syntax
wenzelm@34166
    81
{
wenzelm@58706
    82
  /** syntax content **/
wenzelm@58706
    83
wenzelm@58900
    84
  override def toString: String = keywords.toString
wenzelm@56393
    85
wenzelm@58695
    86
wenzelm@58695
    87
  /* add keywords */
wenzelm@58695
    88
wenzelm@58907
    89
  def + (name: String): Outer_Syntax = this + (name, None, None)
wenzelm@58907
    90
  def + (name: String, kind: String): Outer_Syntax = this + (name, Some((kind, Nil)), None)
wenzelm@58901
    91
  def + (name: String, opt_kind: Option[(String, List[String])], replace: Option[String])
wenzelm@58901
    92
    : Outer_Syntax =
wenzelm@53280
    93
  {
wenzelm@58901
    94
    val keywords1 =
wenzelm@58901
    95
      opt_kind match {
wenzelm@58901
    96
        case None => keywords + name
wenzelm@58901
    97
        case Some(kind) => keywords + (name, kind)
wenzelm@58901
    98
      }
wenzelm@53280
    99
    val completion1 =
wenzelm@58853
   100
      if (replace == Some("")) completion
wenzelm@53280
   101
      else completion + (name, replace getOrElse name)
wenzelm@58900
   102
    new Outer_Syntax(keywords1, completion1, language_context, true)
wenzelm@53280
   103
  }
wenzelm@48706
   104
wenzelm@48873
   105
  def add_keywords(keywords: Thy_Header.Keywords): Outer_Syntax =
wenzelm@48873
   106
    (this /: keywords) {
wenzelm@58901
   107
      case (syntax, (name, opt_spec, replace)) =>
wenzelm@58901
   108
        val opt_kind = opt_spec.map(_._1)
wenzelm@50128
   109
        syntax +
wenzelm@58901
   110
          (Symbol.decode(name), opt_kind, replace) +
wenzelm@58901
   111
          (Symbol.encode(name), opt_kind, replace)
wenzelm@46940
   112
    }
wenzelm@34166
   113
wenzelm@58695
   114
wenzelm@58900
   115
  /* load commands */
wenzelm@58900
   116
wenzelm@58900
   117
  def load_command(name: String): Option[List[String]] = keywords.load_command(name)
wenzelm@58900
   118
  def load_commands_in(text: String): Boolean = keywords.load_commands_in(text)
wenzelm@58900
   119
wenzelm@58900
   120
wenzelm@58706
   121
  /* language context */
wenzelm@34166
   122
wenzelm@58706
   123
  def set_language_context(context: Completion.Language_Context): Outer_Syntax =
wenzelm@58900
   124
    new Outer_Syntax(keywords, completion, context, has_tokens)
wenzelm@58706
   125
wenzelm@58706
   126
  def no_tokens: Outer_Syntax =
wenzelm@46969
   127
  {
wenzelm@58900
   128
    require(keywords.is_empty)
wenzelm@58706
   129
    new Outer_Syntax(
wenzelm@58706
   130
      completion = completion,
wenzelm@58706
   131
      language_context = language_context,
wenzelm@58706
   132
      has_tokens = false)
wenzelm@46969
   133
  }
wenzelm@40454
   134
wenzelm@58706
   135
wenzelm@40454
   136
wenzelm@58706
   137
  /** parsing **/
wenzelm@34166
   138
wenzelm@58697
   139
  /* line-oriented structure */
wenzelm@58696
   140
wenzelm@58700
   141
  def line_structure(tokens: List[Token], struct: Outer_Syntax.Line_Structure)
wenzelm@58700
   142
    : Outer_Syntax.Line_Structure =
wenzelm@58696
   143
  {
wenzelm@58700
   144
    val improper1 = tokens.forall(_.is_improper)
wenzelm@58700
   145
    val command1 = tokens.exists(_.is_command)
wenzelm@58700
   146
wenzelm@58696
   147
    val depth1 =
wenzelm@58901
   148
      if (tokens.exists(tok => keywords.is_command_kind(tok, Keyword.theory))) 0
wenzelm@58700
   149
      else if (command1) struct.after_span_depth
wenzelm@58700
   150
      else struct.span_depth
wenzelm@58700
   151
wenzelm@58700
   152
    val (span_depth1, after_span_depth1) =
wenzelm@58700
   153
      ((struct.span_depth, struct.after_span_depth) /: tokens) {
wenzelm@58703
   154
        case ((x, y), tok) =>
wenzelm@58703
   155
          if (tok.is_command) {
wenzelm@58901
   156
            if (keywords.is_command_kind(tok, Keyword.theory_goal))
wenzelm@58900
   157
              (2, 1)
wenzelm@58901
   158
            else if (keywords.is_command_kind(tok, Keyword.theory))
wenzelm@58900
   159
              (1, 0)
wenzelm@58901
   160
            else if (keywords.is_command_kind(tok, Keyword.proof_goal) || tok.is_begin_block)
wenzelm@58900
   161
              (y + 2, y + 1)
wenzelm@58901
   162
            else if (keywords.is_command_kind(tok, Keyword.qed) || tok.is_end_block)
wenzelm@58900
   163
              (y + 1, y - 1)
wenzelm@58901
   164
            else if (keywords.is_command_kind(tok, Keyword.qed_global))
wenzelm@58900
   165
              (1, 0)
wenzelm@58703
   166
            else (x, y)
wenzelm@58703
   167
          }
wenzelm@58703
   168
          else (x, y)
wenzelm@58696
   169
      }
wenzelm@58700
   170
wenzelm@58700
   171
    Outer_Syntax.Line_Structure(improper1, command1, depth1, span_depth1, after_span_depth1)
wenzelm@58696
   172
  }
wenzelm@58696
   173
wenzelm@58696
   174
wenzelm@53280
   175
  /* token language */
wenzelm@53280
   176
wenzelm@57907
   177
  def scan(input: CharSequence): List[Token] =
wenzelm@52066
   178
  {
wenzelm@58503
   179
    val in: Reader[Char] = new CharSequenceReader(input)
wenzelm@58900
   180
    Token.Parsers.parseAll(Token.Parsers.rep(Token.Parsers.token(keywords)), in) match {
wenzelm@55494
   181
      case Token.Parsers.Success(tokens, _) => tokens
wenzelm@57907
   182
      case _ => error("Unexpected failure of tokenizing input:\n" + input.toString)
wenzelm@34166
   183
    }
wenzelm@52066
   184
  }
wenzelm@34166
   185
wenzelm@58748
   186
  def scan_line(input: CharSequence, context: Scan.Line_Context): (List[Token], Scan.Line_Context) =
wenzelm@52066
   187
  {
wenzelm@52066
   188
    var in: Reader[Char] = new CharSequenceReader(input)
wenzelm@52066
   189
    val toks = new mutable.ListBuffer[Token]
wenzelm@52066
   190
    var ctxt = context
wenzelm@52066
   191
    while (!in.atEnd) {
wenzelm@58900
   192
      Token.Parsers.parse(Token.Parsers.token_line(keywords, ctxt), in) match {
wenzelm@55494
   193
        case Token.Parsers.Success((x, c), rest) => { toks += x; ctxt = c; in = rest }
wenzelm@55494
   194
        case Token.Parsers.NoSuccess(_, rest) =>
wenzelm@52066
   195
          error("Unexpected failure of tokenizing input:\n" + rest.source.toString)
wenzelm@43411
   196
      }
wenzelm@43411
   197
    }
wenzelm@58748
   198
    (toks.toList, ctxt)
wenzelm@52066
   199
  }
wenzelm@55616
   200
wenzelm@55616
   201
wenzelm@58706
   202
  /* command spans */
wenzelm@57905
   203
wenzelm@57905
   204
  def parse_spans(toks: List[Token]): List[Command_Span.Span] =
wenzelm@57905
   205
  {
wenzelm@57905
   206
    val result = new mutable.ListBuffer[Command_Span.Span]
wenzelm@57905
   207
    val content = new mutable.ListBuffer[Token]
wenzelm@57905
   208
    val improper = new mutable.ListBuffer[Token]
wenzelm@57905
   209
wenzelm@57905
   210
    def ship(span: List[Token])
wenzelm@57905
   211
    {
wenzelm@57905
   212
      val kind =
wenzelm@57910
   213
        if (!span.isEmpty && span.head.is_command && !span.exists(_.is_error)) {
wenzelm@57910
   214
          val name = span.head.source
wenzelm@57911
   215
          val pos = Position.Range(Text.Range(0, Symbol.iterator(name).length) + 1)
wenzelm@57910
   216
          Command_Span.Command_Span(name, pos)
wenzelm@57910
   217
        }
wenzelm@57905
   218
        else if (span.forall(_.is_improper)) Command_Span.Ignored_Span
wenzelm@57905
   219
        else Command_Span.Malformed_Span
wenzelm@57905
   220
      result += Command_Span.Span(kind, span)
wenzelm@57905
   221
    }
wenzelm@57905
   222
wenzelm@57905
   223
    def flush()
wenzelm@57905
   224
    {
wenzelm@57905
   225
      if (!content.isEmpty) { ship(content.toList); content.clear }
wenzelm@57905
   226
      if (!improper.isEmpty) { ship(improper.toList); improper.clear }
wenzelm@57905
   227
    }
wenzelm@57905
   228
wenzelm@57905
   229
    for (tok <- toks) {
wenzelm@57905
   230
      if (tok.is_command) { flush(); content += tok }
wenzelm@57905
   231
      else if (tok.is_improper) improper += tok
wenzelm@57905
   232
      else { content ++= improper; improper.clear; content += tok }
wenzelm@57905
   233
    }
wenzelm@57905
   234
    flush()
wenzelm@57905
   235
wenzelm@57905
   236
    result.toList
wenzelm@57905
   237
  }
wenzelm@57905
   238
wenzelm@57906
   239
  def parse_spans(input: CharSequence): List[Command_Span.Span] =
wenzelm@57906
   240
    parse_spans(scan(input))
wenzelm@57906
   241
wenzelm@57905
   242
wenzelm@58706
   243
  /* overall document structure */
wenzelm@55616
   244
wenzelm@58706
   245
  def heading_level(command: Command): Option[Int] =
wenzelm@58706
   246
  {
wenzelm@58868
   247
    command.name match {
wenzelm@58868
   248
      case "chapter" => Some(0)
wenzelm@58868
   249
      case "section" | "header" => Some(1)
wenzelm@58868
   250
      case "subsection" => Some(2)
wenzelm@58868
   251
      case "subsubsection" => Some(3)
wenzelm@58868
   252
      case _ =>
wenzelm@58901
   253
        keywords.command_kind(command.name) match {
wenzelm@58868
   254
          case Some(kind) if Keyword.theory(kind) => Some(4)
wenzelm@58868
   255
          case _ => None
wenzelm@58868
   256
        }
wenzelm@58706
   257
    }
wenzelm@58706
   258
  }
wenzelm@58706
   259
wenzelm@58743
   260
  def parse_document(node_name: Document.Node.Name, text: CharSequence):
wenzelm@58743
   261
    List[Outer_Syntax.Document] =
wenzelm@58706
   262
  {
wenzelm@58706
   263
    /* stack operations */
wenzelm@58706
   264
wenzelm@58706
   265
    def buffer(): mutable.ListBuffer[Outer_Syntax.Document] =
wenzelm@58706
   266
      new mutable.ListBuffer[Outer_Syntax.Document]
wenzelm@58706
   267
wenzelm@58747
   268
    var stack: List[(Int, Command, mutable.ListBuffer[Outer_Syntax.Document])] =
wenzelm@58747
   269
      List((0, Command.empty, buffer()))
wenzelm@55616
   270
wenzelm@58706
   271
    @tailrec def close(level: Int => Boolean)
wenzelm@58706
   272
    {
wenzelm@58706
   273
      stack match {
wenzelm@58747
   274
        case (lev, command, body) :: (_, _, body2) :: rest if level(lev) =>
wenzelm@58747
   275
          body2 += Outer_Syntax.Document_Block(command.name, command.source, body.toList)
wenzelm@58706
   276
          stack = stack.tail
wenzelm@58706
   277
          close(level)
wenzelm@58706
   278
        case _ =>
wenzelm@58706
   279
      }
wenzelm@58706
   280
    }
wenzelm@58706
   281
wenzelm@58743
   282
    def result(): List[Outer_Syntax.Document] =
wenzelm@58706
   283
    {
wenzelm@58706
   284
      close(_ => true)
wenzelm@58743
   285
      stack.head._3.toList
wenzelm@58706
   286
    }
wenzelm@58706
   287
wenzelm@58706
   288
    def add(command: Command)
wenzelm@58706
   289
    {
wenzelm@58706
   290
      heading_level(command) match {
wenzelm@58706
   291
        case Some(i) =>
wenzelm@58706
   292
          close(_ > i)
wenzelm@58747
   293
          stack = (i + 1, command, buffer()) :: stack
wenzelm@58706
   294
        case None =>
wenzelm@58706
   295
      }
wenzelm@58706
   296
      stack.head._3 += Outer_Syntax.Document_Atom(command)
wenzelm@58706
   297
    }
wenzelm@58706
   298
wenzelm@58706
   299
wenzelm@58706
   300
    /* result structure */
wenzelm@58706
   301
wenzelm@58706
   302
    val spans = parse_spans(text)
wenzelm@58706
   303
    spans.foreach(span => add(Command(Document_ID.none, node_name, Nil, span)))
wenzelm@58706
   304
    result()
wenzelm@55616
   305
  }
wenzelm@34166
   306
}