src/Pure/System/standard_system.scala
author wenzelm
Fri, 20 Jul 2012 23:16:54 +0200
changeset 49426 5b3440850d36
parent 49424 0d2114eb412a
child 51218 00d8ad713e32
permissions -rw-r--r--
more abstract file system operations in Scala, corresponding to ML version;
wenzelm@34201
     1
/*  Title:      Pure/System/standard_system.scala
wenzelm@46548
     2
    Module:     PIDE
wenzelm@34201
     3
    Author:     Makarius
wenzelm@34201
     4
wenzelm@34201
     5
Standard system operations, with basic Cygwin/Posix compatibility.
wenzelm@34201
     6
*/
wenzelm@34201
     7
wenzelm@34201
     8
package isabelle
wenzelm@34201
     9
wenzelm@44400
    10
import java.lang.System
wenzelm@34201
    11
import java.util.regex.Pattern
wenzelm@34201
    12
import java.util.Locale
wenzelm@39970
    13
import java.net.URL
wenzelm@49426
    14
import java.io.{File => JFile}
wenzelm@44396
    15
import java.nio.charset.Charset
wenzelm@34201
    16
wenzelm@49292
    17
import scala.io.Codec
wenzelm@34201
    18
import scala.util.matching.Regex
wenzelm@34201
    19
wenzelm@34201
    20
wenzelm@34201
    21
object Standard_System
wenzelm@34201
    22
{
wenzelm@38562
    23
  /* UTF-8 charset */
wenzelm@38562
    24
wenzelm@44396
    25
  val charset_name: String = "UTF-8"
wenzelm@44396
    26
  val charset: Charset = Charset.forName(charset_name)
wenzelm@36039
    27
  def codec(): Codec = Codec(charset)
wenzelm@34201
    28
wenzelm@38562
    29
  def string_bytes(s: String): Array[Byte] = s.getBytes(charset)
wenzelm@38562
    30
wenzelm@34201
    31
wenzelm@34201
    32
  /* permissive UTF-8 decoding */
wenzelm@34201
    33
wenzelm@34201
    34
  // see also http://en.wikipedia.org/wiki/UTF-8#Description
wenzelm@34201
    35
  // overlong encodings enable byte-stuffing
wenzelm@34201
    36
wenzelm@34201
    37
  def decode_permissive_utf8(text: CharSequence): String =
wenzelm@34201
    38
  {
wenzelm@34201
    39
    val buf = new java.lang.StringBuilder(text.length)
wenzelm@34201
    40
    var code = -1
wenzelm@34201
    41
    var rest = 0
wenzelm@34201
    42
    def flush()
wenzelm@34201
    43
    {
wenzelm@34201
    44
      if (code != -1) {
wenzelm@34201
    45
        if (rest == 0 && Character.isValidCodePoint(code))
wenzelm@34201
    46
          buf.appendCodePoint(code)
wenzelm@34201
    47
        else buf.append('\uFFFD')
wenzelm@34201
    48
        code = -1
wenzelm@34201
    49
        rest = 0
wenzelm@34201
    50
      }
wenzelm@34201
    51
    }
wenzelm@34201
    52
    def init(x: Int, n: Int)
wenzelm@34201
    53
    {
wenzelm@34201
    54
      flush()
wenzelm@34201
    55
      code = x
wenzelm@34201
    56
      rest = n
wenzelm@34201
    57
    }
wenzelm@34201
    58
    def push(x: Int)
wenzelm@34201
    59
    {
wenzelm@34201
    60
      if (rest <= 0) init(x, -1)
wenzelm@34201
    61
      else {
wenzelm@34201
    62
        code <<= 6
wenzelm@34201
    63
        code += x
wenzelm@34201
    64
        rest -= 1
wenzelm@34201
    65
      }
wenzelm@34201
    66
    }
wenzelm@34201
    67
    for (i <- 0 until text.length) {
wenzelm@34201
    68
      val c = text.charAt(i)
wenzelm@34201
    69
      if (c < 128) { flush(); buf.append(c) }
wenzelm@34201
    70
      else if ((c & 0xC0) == 0x80) push(c & 0x3F)
wenzelm@34201
    71
      else if ((c & 0xE0) == 0xC0) init(c & 0x1F, 1)
wenzelm@34201
    72
      else if ((c & 0xF0) == 0xE0) init(c & 0x0F, 2)
wenzelm@34201
    73
      else if ((c & 0xF8) == 0xF0) init(c & 0x07, 3)
wenzelm@34201
    74
    }
wenzelm@34201
    75
    flush()
wenzelm@34201
    76
    buf.toString
wenzelm@34201
    77
  }
wenzelm@34201
    78
wenzelm@44622
    79
  private class Decode_Chars(decode: String => String,
wenzelm@44622
    80
    buffer: Array[Byte], start: Int, end: Int) extends CharSequence
wenzelm@44622
    81
  {
wenzelm@44622
    82
    def length: Int = end - start
wenzelm@44622
    83
    def charAt(i: Int): Char = (buffer(start + i).asInstanceOf[Int] & 0xFF).asInstanceOf[Char]
wenzelm@44622
    84
    def subSequence(i: Int, j: Int): CharSequence =
wenzelm@44622
    85
      new Decode_Chars(decode, buffer, start + i, start + j)
wenzelm@44622
    86
wenzelm@44622
    87
    // toString with adhoc decoding: abuse of CharSequence interface
wenzelm@44622
    88
    override def toString: String = decode(decode_permissive_utf8(this))
wenzelm@44622
    89
  }
wenzelm@44622
    90
wenzelm@44622
    91
  def decode_chars(decode: String => String,
wenzelm@44622
    92
    buffer: Array[Byte], start: Int, end: Int): CharSequence =
wenzelm@44622
    93
  {
wenzelm@44622
    94
    require(0 <= start && start <= end && end <= buffer.length)
wenzelm@44622
    95
    new Decode_Chars(decode, buffer, start, end)
wenzelm@44622
    96
  }
wenzelm@44622
    97
wenzelm@34201
    98
wenzelm@34201
    99
  /* shell processes */
wenzelm@34201
   100
wenzelm@49424
   101
  def raw_execute(cwd: JFile, env: Map[String, String], redirect: Boolean, args: String*): Process =
wenzelm@34201
   102
  {
wenzelm@34201
   103
    val cmdline = new java.util.LinkedList[String]
wenzelm@34201
   104
    for (s <- args) cmdline.add(s)
wenzelm@34201
   105
wenzelm@34201
   106
    val proc = new ProcessBuilder(cmdline)
wenzelm@34222
   107
    if (cwd != null) proc.directory(cwd)
wenzelm@34202
   108
    if (env != null) {
wenzelm@34202
   109
      proc.environment.clear
wenzelm@34202
   110
      for ((x, y) <- env) proc.environment.put(x, y)
wenzelm@34202
   111
    }
wenzelm@34201
   112
    proc.redirectErrorStream(redirect)
wenzelm@39791
   113
    proc.start
wenzelm@34201
   114
  }
wenzelm@34201
   115
wenzelm@34201
   116
  def process_output(proc: Process): (String, Int) =
wenzelm@34201
   117
  {
wenzelm@34201
   118
    proc.getOutputStream.close
wenzelm@49426
   119
    val output = File.read(proc.getInputStream)
wenzelm@34201
   120
    val rc =
wenzelm@34201
   121
      try { proc.waitFor }
wenzelm@34201
   122
      finally {
wenzelm@34201
   123
        proc.getInputStream.close
wenzelm@34201
   124
        proc.getErrorStream.close
wenzelm@34201
   125
        proc.destroy
wenzelm@34201
   126
        Thread.interrupted
wenzelm@34201
   127
      }
wenzelm@34201
   128
    (output, rc)
wenzelm@34201
   129
  }
wenzelm@34258
   130
wenzelm@49424
   131
  def raw_exec(cwd: JFile, env: Map[String, String], redirect: Boolean, args: String*)
wenzelm@39791
   132
    : (String, Int) = process_output(raw_execute(cwd, env, redirect, args: _*))
wenzelm@39970
   133
wenzelm@39970
   134
wenzelm@49013
   135
  /* cygwin_root */
wenzelm@49013
   136
wenzelm@49013
   137
  def cygwin_root(): String =
wenzelm@49013
   138
  {
wenzelm@49013
   139
    val cygwin_root1 = System.getenv("CYGWIN_ROOT")
wenzelm@49013
   140
    val cygwin_root2 = System.getProperty("cygwin.root")
wenzelm@49013
   141
    val root =
wenzelm@49013
   142
      if (cygwin_root1 != null && cygwin_root1 != "") cygwin_root1
wenzelm@49013
   143
      else if (cygwin_root2 != null && cygwin_root2 != "") cygwin_root2
wenzelm@49013
   144
      else error("Bad Cygwin installation: unknown root")
wenzelm@49013
   145
wenzelm@49424
   146
    val root_file = new JFile(root)
wenzelm@49424
   147
    if (!new JFile(root_file, "bin\\bash.exe").isFile ||
wenzelm@49424
   148
        !new JFile(root_file, "bin\\env.exe").isFile ||
wenzelm@49424
   149
        !new JFile(root_file, "bin\\tar.exe").isFile)
wenzelm@49013
   150
      error("Bad Cygwin installation: " + quote(root))
wenzelm@49013
   151
wenzelm@49013
   152
    root
wenzelm@49013
   153
  }
wenzelm@34201
   154
}
wenzelm@34201
   155
wenzelm@34201
   156
wenzelm@34201
   157
class Standard_System
wenzelm@34201
   158
{
wenzelm@44539
   159
  /* platform_root */
wenzelm@44539
   160
wenzelm@49013
   161
  val platform_root = if (Platform.is_windows) Standard_System.cygwin_root() else "/"
wenzelm@34201
   162
wenzelm@34201
   163
wenzelm@34201
   164
  /* jvm_path */
wenzelm@34201
   165
wenzelm@34201
   166
  private val Cygdrive = new Regex("/cygdrive/([a-zA-Z])($|/.*)")
wenzelm@36144
   167
  private val Named_Root = new Regex("//+([^/]*)(.*)")
wenzelm@34201
   168
wenzelm@34201
   169
  def jvm_path(posix_path: String): String =
wenzelm@34201
   170
    if (Platform.is_windows) {
wenzelm@34201
   171
      val result_path = new StringBuilder
wenzelm@34201
   172
      val rest =
wenzelm@34201
   173
        posix_path match {
wenzelm@34201
   174
          case Cygdrive(drive, rest) =>
wenzelm@49424
   175
            result_path ++= (drive.toUpperCase(Locale.ENGLISH) + ":" + JFile.separator)
wenzelm@34201
   176
            rest
wenzelm@36144
   177
          case Named_Root(root, rest) =>
wenzelm@49424
   178
            result_path ++= JFile.separator
wenzelm@49424
   179
            result_path ++= JFile.separator
wenzelm@36144
   180
            result_path ++= root
wenzelm@36144
   181
            rest
wenzelm@34201
   182
          case path if path.startsWith("/") =>
wenzelm@34201
   183
            result_path ++= platform_root
wenzelm@34201
   184
            path
wenzelm@34201
   185
          case path => path
wenzelm@34201
   186
        }
wenzelm@44545
   187
      for (p <- space_explode('/', rest) if p != "") {
wenzelm@34201
   188
        val len = result_path.length
wenzelm@49424
   189
        if (len > 0 && result_path(len - 1) != JFile.separatorChar)
wenzelm@49424
   190
          result_path += JFile.separatorChar
wenzelm@34201
   191
        result_path ++= p
wenzelm@34201
   192
      }
wenzelm@34201
   193
      result_path.toString
wenzelm@34201
   194
    }
wenzelm@34201
   195
    else posix_path
wenzelm@34201
   196
wenzelm@34201
   197
wenzelm@34201
   198
  /* posix_path */
wenzelm@34201
   199
wenzelm@34201
   200
  private val Platform_Root = new Regex("(?i)" +
wenzelm@34201
   201
    Pattern.quote(platform_root) + """(?:\\+|\z)(.*)""")
wenzelm@34201
   202
wenzelm@34201
   203
  private val Drive = new Regex("""([a-zA-Z]):\\*(.*)""")
wenzelm@34201
   204
wenzelm@34201
   205
  def posix_path(jvm_path: String): String =
wenzelm@34201
   206
    if (Platform.is_windows) {
wenzelm@34201
   207
      jvm_path.replace('/', '\\') match {
wenzelm@34201
   208
        case Platform_Root(rest) => "/" + rest.replace('\\', '/')
wenzelm@34201
   209
        case Drive(letter, rest) =>
wenzelm@34201
   210
          "/cygdrive/" + letter.toLowerCase(Locale.ENGLISH) +
wenzelm@34201
   211
            (if (rest == "") "" else "/" + rest.replace('\\', '/'))
wenzelm@34201
   212
        case path => path.replace('\\', '/')
wenzelm@34201
   213
      }
wenzelm@34201
   214
    }
wenzelm@34201
   215
    else jvm_path
wenzelm@36193
   216
wenzelm@36193
   217
wenzelm@47984
   218
  /* JDK home of running JVM */
wenzelm@36193
   219
wenzelm@47984
   220
  def this_jdk_home(): String =
wenzelm@36193
   221
  {
wenzelm@36193
   222
    val java_home = System.getProperty("java.home")
wenzelm@49424
   223
    val home = new JFile(java_home)
wenzelm@47984
   224
    val parent = home.getParent
wenzelm@47984
   225
    val jdk_home =
wenzelm@47984
   226
      if (home.getName == "jre" && parent != null &&
wenzelm@49424
   227
          (new JFile(new JFile(parent, "bin"), "javac")).exists) parent
wenzelm@47984
   228
      else java_home
wenzelm@47984
   229
    posix_path(jdk_home)
wenzelm@36193
   230
  }
wenzelm@34201
   231
}