const Cc = Components.classes;
const Ci = Components.interfaces;

const NS_ERROR_ILLEGAL_VALUE = Components.results.NS_ERROR_ILLEGAL_VALUE;

var BIS, BOS, _Pipe, COS, FIS, _SS, CIS;

var dataDir;

function run_test()
{
  BIS = Components.Constructor("@mozilla.org/binaryinputstream;1",
                               "nsIBinaryInputStream",
                               "setInputStream");
  BOS = Components.Constructor("@mozilla.org/binaryoutputstream;1",
                               "nsIBinaryOutputStream",
                               "setOutputStream");
  _Pipe = Components.Constructor("@mozilla.org/pipe;1",
                                 "nsIPipe",
                                 "init");
  COS = Components.Constructor("@mozilla.org/intl/converter-output-stream;1",
                               "nsIConverterOutputStream",
                               "init");
  FIS = Components.Constructor("@mozilla.org/network/file-input-stream;1",
                               "nsIFileInputStream",
                               "init");
  _SS = Components.Constructor("@mozilla.org/storagestream;1",
                               "nsIStorageStream",
                               "init");
  CIS = Components.Constructor("@mozilla.org/intl/converter-input-stream;1",
                               "nsIConverterInputStream",
                               "init");

  dataDir = do_get_file("data/");

  test_utf8_1();
  test_utf16_1();
  test_utf16_2();
  test_utf16_3();
  test_cross_conversion();
}

const UNICODE_STRINGS =
  [
    '\u00BD + \u00BE == \u00BD\u00B2 + \u00BC + \u00BE',

    'AZaz09 \u007F ' +               // U+000000 to U+00007F
    '\u0080 \u0398 \u03BB \u0725 ' + // U+000080 to U+0007FF
    '\u0964 \u0F5F \u20AC \uFFFB'    // U+000800 to U+00FFFF

    // there would be strings containing non-BMP code points here, but
    // unfortunately JS strings are UCS-2 (and worse yet are treated as
    // 16-bit values by the spec), so we have to do gymnastics to work
    // with non-BMP -- manual surrogate decoding doesn't work because
    // String.prototype.charCodeAt() ignores surrogate pairs and only
    // returns 16-bit values
  ];

// test conversion equality -- keys are names of files containing equivalent
// Unicode data, values are the encoding of the file in the format expected by
// nsIConverter(In|Out)putStream.init
const UNICODE_FILES =
  {
    "unicode-conversion.utf8.txt":            "UTF-8",
    "unicode-conversion.utf16.txt":           "UTF-16",
    "unicode-conversion.utf16le.txt":         "UTF-16LE",
    "unicode-conversion.utf16be.txt":         "UTF-16BE"
  };

function test_utf8_1()
{
  for (var i = 0; i < UNICODE_STRINGS.length; i++)
  {
    var pipe = Pipe();
    var conv = new COS(pipe.outputStream, "UTF-8", 1024, 0x0);
    do_check_true(conv.writeString(UNICODE_STRINGS[i]));
    conv.close();

    if (!equal(new UTF8(pipe.inputStream),
               stringToCodePoints(UNICODE_STRINGS[i])))
      do_throw("UNICODE_STRINGS[" + i + "] not handled correctly");
  }
}

function test_utf16_1()
{
  for (var i = 0; i < UNICODE_STRINGS.length; i++)
  {
    var pipe = Pipe();
    var conv = new COS(pipe.outputStream, "UTF-16", 1024, 0x0);
    do_check_true(conv.writeString(UNICODE_STRINGS[i]));
    conv.close();

    if (!equal(new UTF16(pipe.inputStream),
               stringToCodePoints(UNICODE_STRINGS[i])))
      do_throw("UNICODE_STRINGS[" + i + "] not handled correctly");
  }
}

function test_utf16_2()
{
  for (var i = 0; i < UNICODE_STRINGS.length; i++)
  {
    var pipe = Pipe();
    var conv = new COS(pipe.outputStream, "UTF-16LE", 1024, 0x0);
    do_check_true(conv.writeString(UNICODE_STRINGS[i]));
    conv.close();

    if (!equal(new UTF16(pipe.inputStream, false),
               stringToCodePoints(UNICODE_STRINGS[i])))
      do_throw("UNICODE_STRINGS[" + i + "] not handled correctly");
  }
}

function test_utf16_3()
{
  for (var i = 0; i < UNICODE_STRINGS.length; i++)
  {
    var pipe = Pipe();
    var conv = new COS(pipe.outputStream, "UTF-16BE", 1024, 0x0);
    do_check_true(conv.writeString(UNICODE_STRINGS[i]));
    conv.close();

    if (!equal(new UTF16(pipe.inputStream, true),
               stringToCodePoints(UNICODE_STRINGS[i])))
      do_throw("UNICODE_STRINGS[" + i + "] not handled correctly");
  }
}


function test_cross_conversion()
{
  for (var fn1 in UNICODE_FILES)
  {
    var fin = getBinaryInputStream(fn1);
    var ss = StorageStream();

    var bos = new BOS(ss.getOutputStream(0));
    var av;
    while ((av = fin.available()) > 0)
    {
      var data = fin.readByteArray(av);
      bos.writeByteArray(data, data.length);
    }
    fin.close();
    bos.close();

    for (var fn2 in UNICODE_FILES)
    {
      var fin2 = getUnicharInputStream(fn2, UNICODE_FILES[fn2]);
      var unichar = new CIS(ss.newInputStream(0),
                            UNICODE_FILES[fn1], 8192, 0x0);

      if (!equalUnicharStreams(unichar, fin2))
        do_throw("unequal streams: " +
                 UNICODE_FILES[fn1] + ", " +
                 UNICODE_FILES[fn2]);
    }
  }
}


// utility functions

function StorageStream()
{
  return new _SS(8192, Math.pow(2, 32) - 1, null);
}

function getUnicharInputStream(filename, encoding)
{
  var file = dataDir.clone();
  file.append(filename);

  const PR_RDONLY = 0x1;
  var fis = new FIS(file, PR_RDONLY, 0644, Ci.nsIFileInputStream.CLOSE_ON_EOF);
  return new CIS(fis, encoding, 8192, 0x0);
}

function getBinaryInputStream(filename, encoding)
{
  var file = dataDir.clone();
  file.append(filename);

  const PR_RDONLY = 0x1;
  var fis = new FIS(file, PR_RDONLY, 0644, Ci.nsIFileInputStream.CLOSE_ON_EOF);
  return new BIS(fis);
}

function equal(stream, codePoints)
{
  var sz, currIndex = 0;
  while (true)
  {
    var unit = stream.readUnit();
    if (unit < 0)
      return currIndex == codePoints.length;
    if (unit !== codePoints[currIndex++])
      return false;
  }

  do_throw("not reached");
  return false;
}

function equalUnicharStreams(s1, s2)
{
  var r1, r2;
  var str1 = {}, str2 = {};
  while (true)
  {
    r1 = s1.readString(1024, str1);
    r2 = s2.readString(1024, str2);

    if (r1 != r2 || str1.value != str2.value)
    {
      print("r1: " + r1 + ", r2: " + r2);
      print(str1.value.length);
      print(str2.value.length);
      return false;
    }
    if (r1 == 0 && r2 == 0)
      return true;
  }

  // not reached
  return false;
}

function stringToCodePoints(str)
{
  return str.split('').map(function(v){ return v.charCodeAt(0); });
}

function lowbits(n)
{
  return Math.pow(2, n) - 1;
}

function Pipe()
{
  return new _Pipe(false, false, 1024, 10, null);
}


// complex charset readers

/**
 * Wraps a UTF-8 stream to allow access to the Unicode code points in it.
 *
 * @param stream
 *   the stream to wrap
 */
function UTF8(stream)
{
  this._stream = new BIS(stream);
}
UTF8.prototype =
  {
    // returns numeric code point at front of stream encoded in UTF-8, -1 if at
    // end of stream, or throws if valid (and properly encoded!) code point not
    // found
    readUnit: function()
    {
      var str = this._stream;

      var c, c2, c3, c4, rv;

      // if at end of stream, must distinguish failure to read any bytes
      // (correct behavior) from failure to read some byte after the first
      // in the character
      try
      {
        c = str.read8();
      }
      catch (e)
      {
        return -1;
      }

      if (c < 0x80)
        return c;

      if (c < 0xC0) // c < 11000000
      {
        // byte doesn't have enough leading ones (must be at least two)
        throw NS_ERROR_ILLEGAL_VALUE;
      }


      c2 = str.read8();
      if (c2 >= 0xC0 || c2 < 0x80)
        throw NS_ERROR_ILLEGAL_VALUE; // not 10xxxxxx

      if (c < 0xE0) // c < 11100000
      {
        // two-byte between U+000080 and U+0007FF
        rv = ((lowbits(5) & c) << 6) +
              (lowbits(6) & c2);
        // no upper bounds-check needed, by previous lines
        if (rv >= 0x80)
          return rv;
        throw NS_ERROR_ILLEGAL_VALUE;
      }


      c3 = str.read8();
      if (c3 >= 0xC0 || c3 < 0x80)
        throw NS_ERROR_ILLEGAL_VALUE; // not 10xxxxxx

      if (c < 0xF0) // c < 11110000
      {
        // three-byte between U+000800 and U+00FFFF
        rv = ((lowbits(4) & c)  << 12) +
             ((lowbits(6) & c2) <<  6) +
              (lowbits(6) & c3);
        // no upper bounds-check needed, by previous lines
        if (rv >= 0xE000 ||
            (rv >= 0x800 && rv <= 0xD7FF))
          return rv;
        throw NS_ERROR_ILLEGAL_VALUE;
      }


      c4 = str.read8();
      if (c4 >= 0xC0 || c4 < 0x80)
        throw NS_ERROR_ILLEGAL_VALUE; // not 10xxxxxx

      if (c < 0xF8) // c < 11111000
      {
        // four-byte between U+010000 and U+10FFFF
        rv = ((lowbits(3) & c)  << 18) +
             ((lowbits(6) & c2) << 12) +
             ((lowbits(6) & c3) <<  6) +
              (lowbits(6) & c4);
        // need an upper bounds-check since 0x10FFFF isn't (2**n - 1)
        if (rv >= 0x10000 && rv <= 0x10FFFF)
          return rv;
        throw NS_ERROR_ILLEGAL_VALUE;
      }

      // 11111000 or greater -- no UTF-8 mapping
      throw NS_ERROR_ILLEGAL_VALUE;
    }
  };

/**
 * Wraps a UTF-16 stream to allow access to the Unicode code points in it.
 *
 * @param stream
 *   the stream to wrap
 * @param bigEndian
 *   true for UTF-16BE, false for UTF-16LE, not present at all for UTF-16 with
 *   a byte-order mark
 */
function UTF16(stream, bigEndian)
{
  this._stream = new BIS(stream);
  if (arguments.length > 1)
  {
    this._bigEndian = bigEndian;
  }
  else
  {
    var bom = this._stream.read16();
    if (bom == 0xFEFF)
      this._bigEndian = true;
    else if (bom == 0xFFFE)
      this._bigEndian = false;
    else
      do_throw("missing BOM: " + bom.toString(16).toUpperCase());
  }
}
UTF16.prototype =
  {
    // returns numeric code point at front of stream encoded in UTF-16,
    // -1 if at end of stream, or throws if UTF-16 code point not found
    readUnit: function()
    {
      var str = this._stream;

      // if at end of stream, must distinguish failure to read any bytes
      // (correct behavior) from failure to read some byte after the first
      // in the character
      try
      {
        var b1 = str.read8();
      }
      catch (e)
      {
        return -1;
      }

      var b2 = str.read8();

      var w1 = this._bigEndian
             ? (b1 << 8) + b2
             : (b2 << 8) + b1;

      if (w1 > 0xDBFF && w1 < 0xE000)
      {
        // second surrogate, but expecting none or first
        throw NS_ERROR_ILLEGAL_VALUE;
      }

      if (w1 > 0xD7FF && w1 < 0xDC00)
      {
        // non-BMP, use surrogate pair
        b1 = str.read8();
        b2 = str.read8();
        var w2 = this._bigEndian
               ? (b1 << 8) + b2
               : (b2 << 8) + b1;
        if (w2 < 0xDC00 || w2 > 0xDFFF)
          throw NS_ERROR_ILLEGAL_VALUE;

        var rv = 0x100000 +
                 ((lowbits(10) & w2) << 10) +
                  (lowbits(10) & w1);
        if (rv <= 0x10FFFF)
          return rv;
        throw NS_ERROR_ILLEGAL_VALUE;
      }

      // non-surrogate
      return w1;
    }
  };