src/Pure/Isar/outer_syntax.scala
author wenzelm
Sun Jul 10 11:18:35 2016 +0200 (2016-07-10)
changeset 63429 baedd4724f08
parent 63424 e4e15bbfb3e2
child 63441 4c3fa4dba79f
permissions -rw-r--r--
tuned signature: more uniform Keyword.spec;
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@43411
    10
import scala.collection.mutable
wenzelm@58706
    11
import scala.annotation.tailrec
wenzelm@34166
    12
wenzelm@34166
    13
wenzelm@43774
    14
object Outer_Syntax
wenzelm@43774
    15
{
wenzelm@58706
    16
  /* syntax */
wenzelm@58706
    17
wenzelm@58706
    18
  val empty: Outer_Syntax = new Outer_Syntax()
wenzelm@58706
    19
wenzelm@58706
    20
  def init(): Outer_Syntax = new Outer_Syntax(completion = Completion.init())
wenzelm@58706
    21
wenzelm@58706
    22
wenzelm@58706
    23
  /* string literals */
wenzelm@58706
    24
wenzelm@43774
    25
  def quote_string(str: String): String =
wenzelm@43774
    26
  {
wenzelm@43774
    27
    val result = new StringBuilder(str.length + 10)
wenzelm@43774
    28
    result += '"'
wenzelm@43774
    29
    for (s <- Symbol.iterator(str)) {
wenzelm@43774
    30
      if (s.length == 1) {
wenzelm@43774
    31
        val c = s(0)
wenzelm@43774
    32
        if (c < 32 && c != YXML.X && c != YXML.Y || c == '\\' || c == '"') {
wenzelm@43774
    33
          result += '\\'
wenzelm@43774
    34
          if (c < 10) result += '0'
wenzelm@43774
    35
          if (c < 100) result += '0'
wenzelm@60215
    36
          result ++= c.asInstanceOf[Int].toString
wenzelm@43774
    37
        }
wenzelm@43774
    38
        else result += c
wenzelm@43774
    39
      }
wenzelm@43774
    40
      else result ++= s
wenzelm@43774
    41
    }
wenzelm@43774
    42
    result += '"'
wenzelm@43774
    43
    result.toString
wenzelm@43774
    44
  }
wenzelm@46626
    45
wenzelm@58696
    46
wenzelm@58697
    47
  /* line-oriented structure */
wenzelm@58696
    48
wenzelm@58697
    49
  object Line_Structure
wenzelm@58696
    50
  {
wenzelm@58700
    51
    val init = Line_Structure()
wenzelm@58696
    52
  }
wenzelm@58696
    53
wenzelm@58700
    54
  sealed case class Line_Structure(
wenzelm@58700
    55
    improper: Boolean = true,
wenzelm@58700
    56
    command: Boolean = false,
wenzelm@58700
    57
    depth: Int = 0,
wenzelm@58700
    58
    span_depth: Int = 0,
wenzelm@58700
    59
    after_span_depth: Int = 0)
wenzelm@58706
    60
wenzelm@58706
    61
wenzelm@58706
    62
  /* overall document structure */
wenzelm@58706
    63
wenzelm@58706
    64
  sealed abstract class Document { def length: Int }
wenzelm@58747
    65
  case class Document_Block(name: String, text: String, body: List[Document]) extends Document
wenzelm@58706
    66
  {
wenzelm@58706
    67
    val length: Int = (0 /: body)(_ + _.length)
wenzelm@58706
    68
  }
wenzelm@58747
    69
  case class Document_Atom(command: Command) extends Document
wenzelm@58706
    70
  {
wenzelm@58706
    71
    def length: Int = command.length
wenzelm@58706
    72
  }
wenzelm@43774
    73
}
wenzelm@43774
    74
wenzelm@46712
    75
final class Outer_Syntax private(
wenzelm@58900
    76
  val keywords: Keyword.Keywords = Keyword.Keywords.empty,
wenzelm@53280
    77
  val completion: Completion = Completion.empty,
wenzelm@55749
    78
  val language_context: Completion.Language_Context = Completion.Language_Context.outer,
wenzelm@56393
    79
  val has_tokens: Boolean = true) extends Prover.Syntax
wenzelm@34166
    80
{
wenzelm@58706
    81
  /** syntax content **/
wenzelm@58706
    82
wenzelm@58900
    83
  override def toString: String = keywords.toString
wenzelm@56393
    84
wenzelm@58695
    85
wenzelm@58695
    86
  /* add keywords */
wenzelm@58695
    87
wenzelm@63429
    88
  def + (name: String, kind: String = "", tags: List[String] = Nil, replace: Option[String] = None)
wenzelm@58901
    89
    : Outer_Syntax =
wenzelm@53280
    90
  {
wenzelm@63429
    91
    val keywords1 = keywords + (name, kind, tags)
wenzelm@53280
    92
    val completion1 =
wenzelm@58853
    93
      if (replace == Some("")) completion
wenzelm@53280
    94
      else completion + (name, replace getOrElse name)
wenzelm@58900
    95
    new Outer_Syntax(keywords1, completion1, language_context, true)
wenzelm@53280
    96
  }
wenzelm@48706
    97
wenzelm@48873
    98
  def add_keywords(keywords: Thy_Header.Keywords): Outer_Syntax =
wenzelm@48873
    99
    (this /: keywords) {
wenzelm@63429
   100
      case (syntax, (name, ((kind, tags), _), replace)) =>
wenzelm@50128
   101
        syntax +
wenzelm@63429
   102
          (Symbol.decode(name), kind, tags, replace) +
wenzelm@63429
   103
          (Symbol.encode(name), kind, tags, replace)
wenzelm@46940
   104
    }
wenzelm@34166
   105
wenzelm@58695
   106
wenzelm@59073
   107
  /* merge */
wenzelm@59073
   108
wenzelm@59077
   109
  def ++ (other: Prover.Syntax): Prover.Syntax =
wenzelm@59073
   110
    if (this eq other) this
wenzelm@59073
   111
    else {
wenzelm@59077
   112
      val keywords1 = keywords ++ other.asInstanceOf[Outer_Syntax].keywords
wenzelm@59077
   113
      val completion1 = completion ++ other.asInstanceOf[Outer_Syntax].completion
wenzelm@59077
   114
      if ((keywords eq keywords1) && (completion eq completion1)) this
wenzelm@59077
   115
      else new Outer_Syntax(keywords1, completion1, language_context, has_tokens)
wenzelm@59073
   116
    }
wenzelm@59073
   117
wenzelm@59073
   118
wenzelm@59735
   119
  /* load commands */
wenzelm@58900
   120
wenzelm@58900
   121
  def load_command(name: String): Option[List[String]] = keywords.load_command(name)
wenzelm@58900
   122
  def load_commands_in(text: String): Boolean = keywords.load_commands_in(text)
wenzelm@58900
   123
wenzelm@58900
   124
wenzelm@58706
   125
  /* language context */
wenzelm@34166
   126
wenzelm@58706
   127
  def set_language_context(context: Completion.Language_Context): Outer_Syntax =
wenzelm@58900
   128
    new Outer_Syntax(keywords, completion, context, has_tokens)
wenzelm@58706
   129
wenzelm@58706
   130
  def no_tokens: Outer_Syntax =
wenzelm@46969
   131
  {
wenzelm@58900
   132
    require(keywords.is_empty)
wenzelm@58706
   133
    new Outer_Syntax(
wenzelm@58706
   134
      completion = completion,
wenzelm@58706
   135
      language_context = language_context,
wenzelm@58706
   136
      has_tokens = false)
wenzelm@46969
   137
  }
wenzelm@40454
   138
wenzelm@58706
   139
wenzelm@40454
   140
wenzelm@58706
   141
  /** parsing **/
wenzelm@34166
   142
wenzelm@58697
   143
  /* line-oriented structure */
wenzelm@58696
   144
wenzelm@59924
   145
  def line_structure(tokens: List[Token], structure: Outer_Syntax.Line_Structure)
wenzelm@58700
   146
    : Outer_Syntax.Line_Structure =
wenzelm@58696
   147
  {
wenzelm@58700
   148
    val improper1 = tokens.forall(_.is_improper)
wenzelm@58700
   149
    val command1 = tokens.exists(_.is_command)
wenzelm@58700
   150
wenzelm@58696
   151
    val depth1 =
wenzelm@63424
   152
      if (tokens.exists(keywords.is_command(_, Keyword.theory))) 0
wenzelm@59924
   153
      else if (command1) structure.after_span_depth
wenzelm@59924
   154
      else structure.span_depth
wenzelm@58700
   155
wenzelm@58700
   156
    val (span_depth1, after_span_depth1) =
wenzelm@59924
   157
      ((structure.span_depth, structure.after_span_depth) /: tokens) {
wenzelm@58703
   158
        case ((x, y), tok) =>
wenzelm@58703
   159
          if (tok.is_command) {
wenzelm@63424
   160
            if (keywords.is_command(tok, Keyword.theory_goal)) (2, 1)
wenzelm@63424
   161
            else if (keywords.is_command(tok, Keyword.theory)) (1, 0)
wenzelm@63424
   162
            else if (keywords.is_command(tok, Keyword.proof_open)) (y + 2, y + 1)
wenzelm@63424
   163
            else if (keywords.is_command(tok, Keyword.PRF_BLOCK == _)) (y + 2, y + 1)
wenzelm@63424
   164
            else if (keywords.is_command(tok, Keyword.QED_BLOCK == _)) (y + 1, y - 2)
wenzelm@63424
   165
            else if (keywords.is_command(tok, Keyword.proof_close)) (y + 1, y - 1)
wenzelm@63424
   166
            else if (keywords.is_command(tok, Keyword.qed_global)) (1, 0)
wenzelm@58703
   167
            else (x, y)
wenzelm@58703
   168
          }
wenzelm@58703
   169
          else (x, y)
wenzelm@58696
   170
      }
wenzelm@58700
   171
wenzelm@58700
   172
    Outer_Syntax.Line_Structure(improper1, command1, depth1, span_depth1, after_span_depth1)
wenzelm@58696
   173
  }
wenzelm@58696
   174
wenzelm@58696
   175
wenzelm@58706
   176
  /* command spans */
wenzelm@57905
   177
wenzelm@57905
   178
  def parse_spans(toks: List[Token]): List[Command_Span.Span] =
wenzelm@57905
   179
  {
wenzelm@57905
   180
    val result = new mutable.ListBuffer[Command_Span.Span]
wenzelm@57905
   181
    val content = new mutable.ListBuffer[Token]
wenzelm@57905
   182
    val improper = new mutable.ListBuffer[Token]
wenzelm@57905
   183
wenzelm@57905
   184
    def ship(span: List[Token])
wenzelm@57905
   185
    {
wenzelm@57905
   186
      val kind =
wenzelm@59924
   187
        if (span.forall(_.is_improper)) Command_Span.Ignored_Span
wenzelm@59924
   188
        else if (span.exists(_.is_error)) Command_Span.Malformed_Span
wenzelm@59924
   189
        else
wenzelm@59924
   190
          span.find(_.is_command) match {
wenzelm@59924
   191
            case None => Command_Span.Malformed_Span
wenzelm@59924
   192
            case Some(cmd) =>
wenzelm@59924
   193
              val name = cmd.source
wenzelm@59924
   194
              val offset =
wenzelm@59924
   195
                (0 /: span.takeWhile(_ != cmd)) {
wenzelm@59924
   196
                  case (i, tok) => i + Symbol.iterator(tok.source).length }
wenzelm@59924
   197
              val end_offset = offset + Symbol.iterator(name).length
wenzelm@59924
   198
              val pos = Position.Range(Text.Range(offset, end_offset) + 1)
wenzelm@59924
   199
              Command_Span.Command_Span(name, pos)
wenzelm@59924
   200
          }
wenzelm@57905
   201
      result += Command_Span.Span(kind, span)
wenzelm@57905
   202
    }
wenzelm@57905
   203
wenzelm@57905
   204
    def flush()
wenzelm@57905
   205
    {
wenzelm@59319
   206
      if (content.nonEmpty) { ship(content.toList); content.clear }
wenzelm@59319
   207
      if (improper.nonEmpty) { ship(improper.toList); improper.clear }
wenzelm@57905
   208
    }
wenzelm@57905
   209
wenzelm@57905
   210
    for (tok <- toks) {
wenzelm@59924
   211
      if (tok.is_improper) improper += tok
wenzelm@59939
   212
      else if (tok.is_command_modifier ||
wenzelm@59939
   213
        tok.is_command && (!content.exists(_.is_command_modifier) || content.exists(_.is_command)))
wenzelm@59924
   214
      { flush(); content += tok }
wenzelm@57905
   215
      else { content ++= improper; improper.clear; content += tok }
wenzelm@57905
   216
    }
wenzelm@57905
   217
    flush()
wenzelm@57905
   218
wenzelm@57905
   219
    result.toList
wenzelm@57905
   220
  }
wenzelm@57905
   221
wenzelm@57906
   222
  def parse_spans(input: CharSequence): List[Command_Span.Span] =
wenzelm@59083
   223
    parse_spans(Token.explode(keywords, input))
wenzelm@57906
   224
wenzelm@57905
   225
wenzelm@58706
   226
  /* overall document structure */
wenzelm@55616
   227
wenzelm@58706
   228
  def heading_level(command: Command): Option[Int] =
wenzelm@58706
   229
  {
wenzelm@59735
   230
    val name = command.span.name
wenzelm@59735
   231
    name match {
wenzelm@59735
   232
      case Thy_Header.CHAPTER => Some(0)
wenzelm@62453
   233
      case Thy_Header.SECTION => Some(1)
wenzelm@59735
   234
      case Thy_Header.SUBSECTION => Some(2)
wenzelm@59735
   235
      case Thy_Header.SUBSUBSECTION => Some(3)
wenzelm@61463
   236
      case Thy_Header.PARAGRAPH => Some(4)
wenzelm@61463
   237
      case Thy_Header.SUBPARAGRAPH => Some(5)
wenzelm@58868
   238
      case _ =>
wenzelm@59735
   239
        keywords.command_kind(name) match {
wenzelm@61463
   240
          case Some(kind) if Keyword.theory(kind) && !Keyword.theory_end(kind) => Some(6)
wenzelm@58868
   241
          case _ => None
wenzelm@58868
   242
        }
wenzelm@58706
   243
    }
wenzelm@58706
   244
  }
wenzelm@58706
   245
wenzelm@58743
   246
  def parse_document(node_name: Document.Node.Name, text: CharSequence):
wenzelm@58743
   247
    List[Outer_Syntax.Document] =
wenzelm@58706
   248
  {
wenzelm@58706
   249
    /* stack operations */
wenzelm@58706
   250
wenzelm@58706
   251
    def buffer(): mutable.ListBuffer[Outer_Syntax.Document] =
wenzelm@58706
   252
      new mutable.ListBuffer[Outer_Syntax.Document]
wenzelm@58706
   253
wenzelm@58747
   254
    var stack: List[(Int, Command, mutable.ListBuffer[Outer_Syntax.Document])] =
wenzelm@58747
   255
      List((0, Command.empty, buffer()))
wenzelm@55616
   256
wenzelm@58706
   257
    @tailrec def close(level: Int => Boolean)
wenzelm@58706
   258
    {
wenzelm@58706
   259
      stack match {
wenzelm@58747
   260
        case (lev, command, body) :: (_, _, body2) :: rest if level(lev) =>
wenzelm@59735
   261
          body2 += Outer_Syntax.Document_Block(command.span.name, command.source, body.toList)
wenzelm@58706
   262
          stack = stack.tail
wenzelm@58706
   263
          close(level)
wenzelm@58706
   264
        case _ =>
wenzelm@58706
   265
      }
wenzelm@58706
   266
    }
wenzelm@58706
   267
wenzelm@58743
   268
    def result(): List[Outer_Syntax.Document] =
wenzelm@58706
   269
    {
wenzelm@58706
   270
      close(_ => true)
wenzelm@58743
   271
      stack.head._3.toList
wenzelm@58706
   272
    }
wenzelm@58706
   273
wenzelm@58706
   274
    def add(command: Command)
wenzelm@58706
   275
    {
wenzelm@58706
   276
      heading_level(command) match {
wenzelm@58706
   277
        case Some(i) =>
wenzelm@58706
   278
          close(_ > i)
wenzelm@58747
   279
          stack = (i + 1, command, buffer()) :: stack
wenzelm@58706
   280
        case None =>
wenzelm@58706
   281
      }
wenzelm@58706
   282
      stack.head._3 += Outer_Syntax.Document_Atom(command)
wenzelm@58706
   283
    }
wenzelm@58706
   284
wenzelm@58706
   285
wenzelm@58706
   286
    /* result structure */
wenzelm@58706
   287
wenzelm@58706
   288
    val spans = parse_spans(text)
wenzelm@59702
   289
    spans.foreach(span => add(Command(Document_ID.none, node_name, Command.no_blobs, span)))
wenzelm@58706
   290
    result()
wenzelm@55616
   291
  }
wenzelm@34166
   292
}