src/Pure/General/symbol.scala
author wenzelm
Thu, 07 Jul 2011 13:48:30 +0200
changeset 44569 5130dfe1b7be
parent 44550 8252d51d70e2
child 44570 58bb7ca5c7a2
permissions -rw-r--r--
simplified Symbol based on lazy Symbol.Interpretation -- reduced odd "functorial style";
tuned implicit build/init messages;
wenzelm@27901
     1
/*  Title:      Pure/General/symbol.scala
wenzelm@27901
     2
    Author:     Makarius
wenzelm@27901
     3
wenzelm@27924
     4
Detecting and recoding Isabelle symbols.
wenzelm@27901
     5
*/
wenzelm@27901
     6
wenzelm@27901
     7
package isabelle
wenzelm@27901
     8
wenzelm@27918
     9
import scala.io.Source
wenzelm@36035
    10
import scala.collection.mutable
wenzelm@31537
    11
import scala.util.matching.Regex
wenzelm@27901
    12
wenzelm@27901
    13
wenzelm@31537
    14
object Symbol
wenzelm@31537
    15
{
wenzelm@36772
    16
  /* spaces */
wenzelm@36772
    17
wenzelm@36850
    18
  val spc = ' '
wenzelm@36850
    19
  val space = " "
wenzelm@36850
    20
wenzelm@36850
    21
  private val static_spaces = space * 4000
wenzelm@36772
    22
wenzelm@36772
    23
  def spaces(k: Int): String =
wenzelm@36772
    24
  {
wenzelm@36772
    25
    require(k >= 0)
wenzelm@36772
    26
    if (k < static_spaces.length) static_spaces.substring(0, k)
wenzelm@36850
    27
    else space * k
wenzelm@36772
    28
  }
wenzelm@36772
    29
wenzelm@36772
    30
wenzelm@44292
    31
  /* ASCII characters */
wenzelm@44292
    32
wenzelm@44292
    33
  def is_ascii_letter(c: Char): Boolean = 'A' <= c && c <= 'Z' || 'a' <= c && c <= 'z'
wenzelm@44292
    34
  def is_ascii_digit(c: Char): Boolean = '0' <= c && c <= '9'
wenzelm@44292
    35
  def is_ascii_quasi(c: Char): Boolean = c == '_' || c == '\''
wenzelm@44292
    36
wenzelm@44292
    37
  def is_ascii_letdig(c: Char): Boolean =
wenzelm@44292
    38
    is_ascii_letter(c) || is_ascii_digit(c) || is_ascii_quasi(c)
wenzelm@44292
    39
wenzelm@44292
    40
  def is_ascii_identifier(s: String): Boolean =
wenzelm@44292
    41
    s.length > 0 && is_ascii_letter(s(0)) && s.substring(1).forall(is_ascii_letdig)
wenzelm@44292
    42
wenzelm@44292
    43
wenzelm@34007
    44
  /* Symbol regexps */
wenzelm@27901
    45
wenzelm@31537
    46
  private val plain = new Regex("""(?xs)
wenzelm@40770
    47
      [^\r\\\ud800-\udfff\ufffd] | [\ud800-\udbff][\udc00-\udfff] """)
wenzelm@37563
    48
wenzelm@40768
    49
  private val physical_newline = new Regex("""(?xs) \n | \r\n | \r """)
wenzelm@27901
    50
wenzelm@31537
    51
  private val symbol = new Regex("""(?xs)
wenzelm@31548
    52
      \\ < (?:
wenzelm@27924
    53
      \^? [A-Za-z][A-Za-z0-9_']* |
wenzelm@27924
    54
      \^raw: [\x20-\x7e\u0100-\uffff && [^.>]]* ) >""")
wenzelm@27923
    55
wenzelm@40769
    56
  private val malformed_symbol = new Regex("(?xs) (?!" + symbol + ")" +
wenzelm@40775
    57
    """ [\ud800-\udbff\ufffd] | \\<\^? """)
wenzelm@27923
    58
wenzelm@40769
    59
  val regex_total =
wenzelm@40769
    60
    new Regex(plain + "|" + physical_newline + "|" + symbol + "|" + malformed_symbol + "| .")
wenzelm@27918
    61
wenzelm@34146
    62
wenzelm@34146
    63
  /* basic matching */
wenzelm@34146
    64
wenzelm@37563
    65
  def is_plain(c: Char): Boolean = !(c == '\r' || c == '\\' || '\ud800' <= c && c <= '\udfff')
wenzelm@34146
    66
wenzelm@44550
    67
  def is_physical_newline(s: String): Boolean =
wenzelm@44550
    68
    s == "\n" || s == "\r" || s == "\r\n"
wenzelm@39203
    69
wenzelm@44550
    70
  def is_malformed(s: String): Boolean =
wenzelm@44550
    71
    !(s.length == 1 && is_plain(s(0))) && malformed_symbol.pattern.matcher(s).matches
wenzelm@34146
    72
wenzelm@34146
    73
  class Matcher(text: CharSequence)
wenzelm@34146
    74
  {
wenzelm@40769
    75
    private val matcher = regex_total.pattern.matcher(text)
wenzelm@34146
    76
    def apply(start: Int, end: Int): Int =
wenzelm@34146
    77
    {
wenzelm@34146
    78
      require(0 <= start && start < end && end <= text.length)
wenzelm@34319
    79
      if (is_plain(text.charAt(start))) 1
wenzelm@34147
    80
      else {
wenzelm@34146
    81
        matcher.region(start, end).lookingAt
wenzelm@34146
    82
        matcher.group.length
wenzelm@34146
    83
      }
wenzelm@34146
    84
    }
wenzelm@31537
    85
  }
wenzelm@27901
    86
wenzelm@27901
    87
wenzelm@44569
    88
  /* iterator */
wenzelm@31939
    89
wenzelm@44550
    90
  private val char_symbols: Array[String] =
wenzelm@44550
    91
    (0 until 256).iterator.map(i => new String(Array(i.toChar))).toArray
wenzelm@44550
    92
wenzelm@44550
    93
  def iterator(text: CharSequence): Iterator[String] =
wenzelm@44550
    94
    new Iterator[String]
wenzelm@40768
    95
    {
wenzelm@44360
    96
      private val matcher = new Matcher(text)
wenzelm@44360
    97
      private var i = 0
wenzelm@44360
    98
      def hasNext = i < text.length
wenzelm@44360
    99
      def next =
wenzelm@44360
   100
      {
wenzelm@44360
   101
        val n = matcher(i, text.length)
wenzelm@44550
   102
        val s =
wenzelm@44550
   103
          if (n == 0) ""
wenzelm@44550
   104
          else if (n == 1) {
wenzelm@44550
   105
            val c = text.charAt(i)
wenzelm@44550
   106
            if (c < char_symbols.length) char_symbols(c)
wenzelm@44550
   107
            else text.subSequence(i, i + n).toString
wenzelm@44550
   108
          }
wenzelm@44550
   109
          else text.subSequence(i, i + n).toString
wenzelm@44360
   110
        i += n
wenzelm@44360
   111
        s
wenzelm@44360
   112
      }
wenzelm@34007
   113
    }
wenzelm@44360
   114
wenzelm@34007
   115
wenzelm@34007
   116
  /* decoding offsets */
wenzelm@34007
   117
wenzelm@34007
   118
  class Index(text: CharSequence)
wenzelm@31939
   119
  {
wenzelm@31939
   120
    case class Entry(chr: Int, sym: Int)
wenzelm@31939
   121
    val index: Array[Entry] =
wenzelm@31939
   122
    {
wenzelm@34146
   123
      val matcher = new Matcher(text)
wenzelm@31939
   124
      val buf = new mutable.ArrayBuffer[Entry]
wenzelm@31939
   125
      var chr = 0
wenzelm@31939
   126
      var sym = 0
wenzelm@34007
   127
      while (chr < text.length) {
wenzelm@34146
   128
        val n = matcher(chr, text.length)
wenzelm@34146
   129
        chr += n
wenzelm@31939
   130
        sym += 1
wenzelm@34146
   131
        if (n > 1) buf += Entry(chr, sym)
wenzelm@31939
   132
      }
wenzelm@31939
   133
      buf.toArray
wenzelm@31939
   134
    }
wenzelm@38797
   135
    def decode(sym1: Int): Int =
wenzelm@31939
   136
    {
wenzelm@38797
   137
      val sym = sym1 - 1
wenzelm@31939
   138
      val end = index.length
wenzelm@31939
   139
      def bisect(a: Int, b: Int): Int =
wenzelm@31939
   140
      {
wenzelm@31939
   141
        if (a < b) {
wenzelm@31939
   142
          val c = (a + b) / 2
wenzelm@31939
   143
          if (sym < index(c).sym) bisect(a, c)
wenzelm@31939
   144
          else if (c + 1 == end || sym < index(c + 1).sym) c
wenzelm@31939
   145
          else bisect(c + 1, b)
wenzelm@31939
   146
        }
wenzelm@31939
   147
        else -1
wenzelm@31939
   148
      }
wenzelm@31939
   149
      val i = bisect(0, end)
wenzelm@31939
   150
      if (i < 0) sym
wenzelm@31939
   151
      else index(i).chr + sym - index(i).sym
wenzelm@31939
   152
    }
wenzelm@38797
   153
    def decode(range: Text.Range): Text.Range = range.map(decode(_))
wenzelm@31939
   154
  }
wenzelm@31939
   155
wenzelm@31939
   156
wenzelm@34007
   157
  /* recoding text */
wenzelm@27924
   158
wenzelm@31537
   159
  private class Recoder(list: List[(String, String)])
wenzelm@31537
   160
  {
wenzelm@31537
   161
    private val (min, max) =
wenzelm@31537
   162
    {
wenzelm@27937
   163
      var min = '\uffff'
wenzelm@27937
   164
      var max = '\u0000'
wenzelm@27937
   165
      for ((x, _) <- list) {
wenzelm@27937
   166
        val c = x(0)
wenzelm@27937
   167
        if (c < min) min = c
wenzelm@27937
   168
        if (c > max) max = c
wenzelm@27937
   169
      }
wenzelm@27937
   170
      (min, max)
wenzelm@27937
   171
    }
wenzelm@40690
   172
    private val table =
wenzelm@40690
   173
    {
wenzelm@40690
   174
      var tab = Map[String, String]()
wenzelm@40690
   175
      for ((x, y) <- list) {
wenzelm@40690
   176
        tab.get(x) match {
wenzelm@40690
   177
          case None => tab += (x -> y)
wenzelm@40690
   178
          case Some(z) =>
wenzelm@40690
   179
            error("Duplicate mapping of \"" + x + "\" to \"" + y + "\" vs. \"" + z + "\"")
wenzelm@40690
   180
        }
wenzelm@40690
   181
      }
wenzelm@40690
   182
      tab
wenzelm@40690
   183
    }
wenzelm@31537
   184
    def recode(text: String): String =
wenzelm@31537
   185
    {
wenzelm@27937
   186
      val len = text.length
wenzelm@40769
   187
      val matcher = regex_total.pattern.matcher(text)
wenzelm@27937
   188
      val result = new StringBuilder(len)
wenzelm@27937
   189
      var i = 0
wenzelm@27937
   190
      while (i < len) {
wenzelm@27937
   191
        val c = text(i)
wenzelm@27937
   192
        if (min <= c && c <= max) {
wenzelm@31939
   193
          matcher.region(i, len).lookingAt
wenzelm@27938
   194
          val x = matcher.group
wenzelm@31537
   195
          result.append(table.get(x) getOrElse x)
wenzelm@27937
   196
          i = matcher.end
wenzelm@27937
   197
        }
wenzelm@27937
   198
        else { result.append(c); i += 1 }
wenzelm@27937
   199
      }
wenzelm@27937
   200
      result.toString
wenzelm@27924
   201
    }
wenzelm@27924
   202
  }
wenzelm@27918
   203
wenzelm@27918
   204
wenzelm@27937
   205
wenzelm@44569
   206
  /** symbol interpretation **/
wenzelm@27937
   207
wenzelm@44569
   208
  private lazy val symbols =
wenzelm@44569
   209
    new Interpretation(
wenzelm@44569
   210
      Isabelle_System.try_read(Path.split(Isabelle_System.getenv_strict("ISABELLE_SYMBOLS"))))
wenzelm@44569
   211
wenzelm@44569
   212
  private class Interpretation(symbols_spec: String)
wenzelm@29569
   213
  {
wenzelm@31537
   214
    /* read symbols */
wenzelm@31537
   215
wenzelm@31537
   216
    private val empty = new Regex("""(?xs) ^\s* (?: \#.* )? $ """)
wenzelm@31537
   217
    private val key = new Regex("""(?xs) (.+): """)
wenzelm@31537
   218
wenzelm@31537
   219
    private def read_decl(decl: String): (String, Map[String, String]) =
wenzelm@31537
   220
    {
wenzelm@31537
   221
      def err() = error("Bad symbol declaration: " + decl)
wenzelm@31537
   222
wenzelm@31537
   223
      def read_props(props: List[String]): Map[String, String] =
wenzelm@31537
   224
      {
wenzelm@31537
   225
        props match {
wenzelm@31537
   226
          case Nil => Map()
wenzelm@31537
   227
          case _ :: Nil => err()
wenzelm@31537
   228
          case key(x) :: y :: rest => read_props(rest) + (x -> y)
wenzelm@31537
   229
          case _ => err()
wenzelm@31537
   230
        }
wenzelm@31537
   231
      }
wenzelm@31537
   232
      decl.split("\\s+").toList match {
wenzelm@40769
   233
        case sym :: props if sym.length > 1 && !is_malformed(sym) => (sym, read_props(props))
wenzelm@34193
   234
        case _ => err()
wenzelm@31537
   235
      }
wenzelm@31537
   236
    }
wenzelm@31537
   237
wenzelm@31537
   238
    private val symbols: List[(String, Map[String, String])] =
wenzelm@40690
   239
      Map((
wenzelm@44569
   240
        for (decl <- split_lines(symbols_spec) if !empty.pattern.matcher(decl).matches)
wenzelm@40690
   241
          yield read_decl(decl)): _*) toList
wenzelm@31537
   242
wenzelm@31537
   243
wenzelm@31651
   244
    /* misc properties */
wenzelm@31651
   245
wenzelm@34143
   246
    val names: Map[String, String] =
wenzelm@34143
   247
    {
wenzelm@44331
   248
      val name = new Regex("""\\<\^?([A-Za-z][A-Za-z0-9_']*)>""")
wenzelm@31651
   249
      Map((for ((sym @ name(a), _) <- symbols) yield (sym -> a)): _*)
wenzelm@31651
   250
    }
wenzelm@31651
   251
wenzelm@44359
   252
    val abbrevs: Map[String, String] =
wenzelm@44359
   253
      Map((
wenzelm@44359
   254
        for ((sym, props) <- symbols if props.isDefinedAt("abbrev"))
wenzelm@44359
   255
          yield (sym -> props("abbrev"))): _*)
wenzelm@44359
   256
wenzelm@44359
   257
wenzelm@44361
   258
    /* recoding */
wenzelm@31537
   259
wenzelm@31537
   260
    private val (decoder, encoder) =
wenzelm@31537
   261
    {
wenzelm@31537
   262
      val mapping =
wenzelm@31537
   263
        for {
wenzelm@31537
   264
          (sym, props) <- symbols
wenzelm@31537
   265
          val code =
wenzelm@31537
   266
            try { Integer.decode(props("code")).intValue }
wenzelm@31537
   267
            catch {
wenzelm@31537
   268
              case _: NoSuchElementException => error("Missing code for symbol " + sym)
wenzelm@31537
   269
              case _: NumberFormatException => error("Bad code for symbol " + sym)
wenzelm@31537
   270
            }
wenzelm@31537
   271
          val ch = new String(Character.toChars(code))
wenzelm@34193
   272
        } yield {
wenzelm@34193
   273
          if (code < 128) error("Illegal ASCII code for symbol " + sym)
wenzelm@34193
   274
          else (sym, ch)
wenzelm@34193
   275
        }
wenzelm@31548
   276
      (new Recoder(mapping),
wenzelm@31551
   277
       new Recoder(mapping map { case (x, y) => (y, x) }))
wenzelm@31537
   278
    }
wenzelm@27918
   279
wenzelm@34104
   280
    def decode(text: String): String = decoder.recode(text)
wenzelm@34104
   281
    def encode(text: String): String = encoder.recode(text)
wenzelm@34143
   282
wenzelm@44361
   283
    private def recode_set(elems: String*): Set[String] =
wenzelm@44361
   284
    {
wenzelm@44361
   285
      val content = elems.toList
wenzelm@44361
   286
      Set((content ::: content.map(decode)): _*)
wenzelm@44361
   287
    }
wenzelm@44361
   288
wenzelm@44361
   289
    private def recode_map[A](elems: (String, A)*): Map[String, A] =
wenzelm@44361
   290
    {
wenzelm@44361
   291
      val content = elems.toList
wenzelm@44361
   292
      Map((content ::: content.map({ case (sym, a) => (decode(sym), a) })): _*)
wenzelm@44361
   293
    }
wenzelm@44361
   294
wenzelm@44361
   295
wenzelm@44361
   296
    /* user fonts */
wenzelm@44361
   297
wenzelm@44361
   298
    val fonts: Map[String, String] =
wenzelm@44361
   299
      recode_map((
wenzelm@44361
   300
        for ((sym, props) <- symbols if props.isDefinedAt("font"))
wenzelm@44361
   301
          yield (sym -> props("font"))): _*)
wenzelm@44361
   302
wenzelm@44361
   303
    val font_names: List[String] = Set(fonts.toList.map(_._2): _*).toList
wenzelm@44361
   304
    val font_index: Map[String, Int] = Map((font_names zip (0 until font_names.length).toList): _*)
wenzelm@44361
   305
wenzelm@34143
   306
wenzelm@34143
   307
    /* classification */
wenzelm@34143
   308
wenzelm@44569
   309
    val letters = recode_set(
wenzelm@34143
   310
      "A", "B", "C", "D", "E", "F", "G", "H", "I", "J", "K", "L", "M",
wenzelm@34143
   311
      "N", "O", "P", "Q", "R", "S", "T", "U", "V", "W", "X", "Y", "Z",
wenzelm@34143
   312
      "a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "k", "l", "m",
wenzelm@34143
   313
      "n", "o", "p", "q", "r", "s", "t", "u", "v", "w", "x", "y", "z",
wenzelm@34143
   314
wenzelm@34143
   315
      "\\<A>", "\\<B>", "\\<C>", "\\<D>", "\\<E>", "\\<F>", "\\<G>",
wenzelm@34143
   316
      "\\<H>", "\\<I>", "\\<J>", "\\<K>", "\\<L>", "\\<M>", "\\<N>",
wenzelm@34143
   317
      "\\<O>", "\\<P>", "\\<Q>", "\\<R>", "\\<S>", "\\<T>", "\\<U>",
wenzelm@34143
   318
      "\\<V>", "\\<W>", "\\<X>", "\\<Y>", "\\<Z>", "\\<a>", "\\<b>",
wenzelm@34143
   319
      "\\<c>", "\\<d>", "\\<e>", "\\<f>", "\\<g>", "\\<h>", "\\<i>",
wenzelm@34143
   320
      "\\<j>", "\\<k>", "\\<l>", "\\<m>", "\\<n>", "\\<o>", "\\<p>",
wenzelm@34143
   321
      "\\<q>", "\\<r>", "\\<s>", "\\<t>", "\\<u>", "\\<v>", "\\<w>",
wenzelm@34143
   322
      "\\<x>", "\\<y>", "\\<z>",
wenzelm@34143
   323
wenzelm@34143
   324
      "\\<AA>", "\\<BB>", "\\<CC>", "\\<DD>", "\\<EE>", "\\<FF>",
wenzelm@34143
   325
      "\\<GG>", "\\<HH>", "\\<II>", "\\<JJ>", "\\<KK>", "\\<LL>",
wenzelm@34143
   326
      "\\<MM>", "\\<NN>", "\\<OO>", "\\<PP>", "\\<QQ>", "\\<RR>",
wenzelm@34143
   327
      "\\<SS>", "\\<TT>", "\\<UU>", "\\<VV>", "\\<WW>", "\\<XX>",
wenzelm@34143
   328
      "\\<YY>", "\\<ZZ>", "\\<aa>", "\\<bb>", "\\<cc>", "\\<dd>",
wenzelm@34143
   329
      "\\<ee>", "\\<ff>", "\\<gg>", "\\<hh>", "\\<ii>", "\\<jj>",
wenzelm@34143
   330
      "\\<kk>", "\\<ll>", "\\<mm>", "\\<nn>", "\\<oo>", "\\<pp>",
wenzelm@34143
   331
      "\\<qq>", "\\<rr>", "\\<ss>", "\\<tt>", "\\<uu>", "\\<vv>",
wenzelm@34143
   332
      "\\<ww>", "\\<xx>", "\\<yy>", "\\<zz>",
wenzelm@34143
   333
wenzelm@34143
   334
      "\\<alpha>", "\\<beta>", "\\<gamma>", "\\<delta>", "\\<epsilon>",
wenzelm@34143
   335
      "\\<zeta>", "\\<eta>", "\\<theta>", "\\<iota>", "\\<kappa>",
wenzelm@34143
   336
      "\\<mu>", "\\<nu>", "\\<xi>", "\\<pi>", "\\<rho>", "\\<sigma>",
wenzelm@34143
   337
      "\\<tau>", "\\<upsilon>", "\\<phi>", "\\<chi>", "\\<psi>",
wenzelm@34143
   338
      "\\<omega>", "\\<Gamma>", "\\<Delta>", "\\<Theta>", "\\<Lambda>",
wenzelm@34143
   339
      "\\<Xi>", "\\<Pi>", "\\<Sigma>", "\\<Upsilon>", "\\<Phi>",
wenzelm@34143
   340
      "\\<Psi>", "\\<Omega>",
wenzelm@34143
   341
wenzelm@34143
   342
      "\\<^isub>", "\\<^isup>")
wenzelm@34143
   343
wenzelm@44569
   344
    val blanks =
wenzelm@44361
   345
      recode_set(space, "\t", "\n", "\u000B", "\f", "\r", "\\<spacespace>", "\\<^newline>")
wenzelm@34147
   346
wenzelm@44569
   347
    val sym_chars =
wenzelm@34147
   348
      Set("!", "#", "$", "%", "&", "*", "+", "-", "/", "<", "=", ">", "?", "@", "^", "_", "|", "~")
wenzelm@34143
   349
wenzelm@44330
   350
wenzelm@44359
   351
    /* control symbols */
wenzelm@44359
   352
wenzelm@44569
   353
    val ctrl_decoded: Set[String] =
wenzelm@44359
   354
      Set((for ((sym, _) <- symbols if sym.startsWith("\\<^")) yield decode(sym)): _*)
wenzelm@44359
   355
wenzelm@44569
   356
    val subscript_decoded = Set(decode("\\<^sub>"), decode("\\<^isub>"))
wenzelm@44569
   357
    val superscript_decoded = Set(decode("\\<^sup>"), decode("\\<^isup>"))
wenzelm@44382
   358
wenzelm@44382
   359
    val bold_decoded = decode("\\<^bold>")
wenzelm@44382
   360
    val bsub_decoded = decode("\\<^bsub>")
wenzelm@44382
   361
    val esub_decoded = decode("\\<^esub>")
wenzelm@44382
   362
    val bsup_decoded = decode("\\<^bsup>")
wenzelm@44382
   363
    val esup_decoded = decode("\\<^esup>")
wenzelm@27918
   364
  }
wenzelm@44569
   365
wenzelm@44569
   366
wenzelm@44569
   367
  /* tables */
wenzelm@44569
   368
wenzelm@44569
   369
  def names: Map[String, String] = symbols.names
wenzelm@44569
   370
  def abbrevs: Map[String, String] = symbols.abbrevs
wenzelm@44569
   371
wenzelm@44569
   372
  def decode(text: String): String = symbols.decode(text)
wenzelm@44569
   373
  def encode(text: String): String = symbols.encode(text)
wenzelm@44569
   374
wenzelm@44569
   375
  def fonts: Map[String, String] = symbols.fonts
wenzelm@44569
   376
  def font_names: List[String] = symbols.font_names
wenzelm@44569
   377
  def font_index: Map[String, Int] = symbols.font_index
wenzelm@44569
   378
  def lookup_font(sym: String): Option[Int] = symbols.fonts.get(sym).map(font_index(_))
wenzelm@44569
   379
wenzelm@44569
   380
wenzelm@44569
   381
  /* classification */
wenzelm@44569
   382
wenzelm@44569
   383
  def is_letter(sym: String): Boolean = symbols.letters.contains(sym)
wenzelm@44569
   384
  def is_digit(sym: String): Boolean = sym.length == 1 && '0' <= sym(0) && sym(0) <= '9'
wenzelm@44569
   385
  def is_quasi(sym: String): Boolean = sym == "_" || sym == "'"
wenzelm@44569
   386
  def is_letdig(sym: String): Boolean = is_letter(sym) || is_digit(sym) || is_quasi(sym)
wenzelm@44569
   387
  def is_blank(sym: String): Boolean = symbols.blanks.contains(sym)
wenzelm@44569
   388
  def is_symbolic_char(sym: String): Boolean = symbols.sym_chars.contains(sym)
wenzelm@44569
   389
  def is_symbolic(sym: String): Boolean =
wenzelm@44569
   390
    sym.startsWith("\\<") && sym.endsWith(">") && !sym.startsWith("\\<^")
wenzelm@44569
   391
wenzelm@44569
   392
wenzelm@44569
   393
  /* control symbols */
wenzelm@44569
   394
wenzelm@44569
   395
  def is_ctrl(sym: String): Boolean =
wenzelm@44569
   396
    sym.startsWith("\\<^") || symbols.ctrl_decoded.contains(sym)
wenzelm@44569
   397
wenzelm@44569
   398
  def is_controllable(sym: String): Boolean =
wenzelm@44569
   399
    !is_blank(sym) && !is_ctrl(sym) && !is_malformed(sym)
wenzelm@44569
   400
wenzelm@44569
   401
  def is_subscript_decoded(sym: String): Boolean = symbols.subscript_decoded.contains(sym)
wenzelm@44569
   402
  def is_superscript_decoded(sym: String): Boolean = symbols.superscript_decoded.contains(sym)
wenzelm@44569
   403
  def is_bold_decoded(sym: String): Boolean = sym == symbols.bold_decoded
wenzelm@44569
   404
  def is_bsub_decoded(sym: String): Boolean = sym == symbols.bsub_decoded
wenzelm@44569
   405
  def is_esub_decoded(sym: String): Boolean = sym == symbols.esub_decoded
wenzelm@44569
   406
  def is_bsup_decoded(sym: String): Boolean = sym == symbols.bsup_decoded
wenzelm@44569
   407
  def is_esup_decoded(sym: String): Boolean = sym == symbols.esup_decoded
wenzelm@27901
   408
}