strings.js

// String encode/decode helpers
'use strict';


var utils = require('./common');


// Quick check if we can use fast array to bin string conversion
//
// - apply(Array) can fail on Android 2.2
// - apply(Uint8Array) can fail on iOS 5.1 Safary
//
var STR_APPLY_OK = true;
var STR_APPLY_UIA_OK = true;

try { String.fromCharCode.apply(null, [0]); } catch(__) { STR_APPLY_OK = false; }
try { String.fromCharCode.apply(null, new Uint8Array(1)); } catch(__) { STR_APPLY_UIA_OK = false; }


// Table with utf8 lengths (calculated by first byte of sequence)
// Note, that 5 & 6-byte values and some 4-byte values can not be represented in JS,
// because max possible codepoint is 0x10ffff
var _utf8len = new utils.Buf8(256);
for (var i=0; i<256; i++) {
  _utf8len[i] = (i >= 252 ? 6 : i >= 248 ? 5 : i >= 240 ? 4 : i >= 224 ? 3 : i >= 192 ? 2 : 1);
}
_utf8len[254]=_utf8len[254]=1; // Invalid sequence start


// convert string to array (typed, when possible)
exports.string2buf = function (str) {
  var buf, c, c2, m_pos, i, str_len = str.length, buf_len = 0;

  // count binary size
  for (m_pos = 0; m_pos < str_len; m_pos++) {
    c = str.charCodeAt(m_pos);
    if ((c & 0xfc00) === 0xd800 && (m_pos+1 < str_len)) {
      c2 = str.charCodeAt(m_pos+1);
      if ((c2 & 0xfc00) === 0xdc00) {
        c = 0x10000 + ((c - 0xd800) << 10) + (c2 - 0xdc00);
        m_pos++;
      }
    }
    buf_len += c < 0x80 ? 1 : c < 0x800 ? 2 : c < 0x10000 ? 3 : 4;
  }

  // allocate buffer
  buf = new utils.Buf8(buf_len);

  // convert
  for (i=0, m_pos = 0; i < buf_len; m_pos++) {
    c = str.charCodeAt(m_pos);
    if ((c & 0xfc00) === 0xd800 && (m_pos+1 < str_len)) {
      c2 = str.charCodeAt(m_pos+1);
      if ((c2 & 0xfc00) === 0xdc00) {
        c = 0x10000 + ((c - 0xd800) << 10) + (c2 - 0xdc00);
        m_pos++;
      }
    }
    if (c < 0x80) {
      /* one byte */
      buf[i++] = c;
    } else if (c < 0x800) {
      /* two bytes */
      buf[i++] = 0xC0 | (c >>> 6);
      buf[i++] = 0x80 | (c & 0x3f);
    } else if (c < 0x10000) {
      /* three bytes */
      buf[i++] = 0xE0 | (c >>> 12);
      buf[i++] = 0x80 | (c >>> 6 & 0x3f);
      buf[i++] = 0x80 | (c & 0x3f);
    } else {
      /* four bytes */
      buf[i++] = 0xf0 | (c >>> 18);
      buf[i++] = 0x80 | (c >>> 12 & 0x3f);
      buf[i++] = 0x80 | (c >>> 6 & 0x3f);
      buf[i++] = 0x80 | (c & 0x3f);
    }
  }

  return buf;
};

// Helper (used in 2 places)
function buf2binstring(buf, len) {
  // use fallback for big arrays to avoid stack overflow
  if (len < 65537) {
    if ((buf.subarray && STR_APPLY_UIA_OK) || (!buf.subarray && STR_APPLY_OK)) {
      return String.fromCharCode.apply(null, utils.shrinkBuf(buf, len));
    }
  }

  var result = '';
  for(var i=0; i < len; i++) {
    result += String.fromCharCode(buf[i]);
  }
  return result;
}


// Convert byte array to binary string
exports.buf2binstring = function(buf) {
  return buf2binstring(buf, buf.length);
};


// Convert binary string (typed, when possible)
exports.binstring2buf = function(str) {
  var buf = new utils.Buf8(str.length);
  for(var i=0, len=buf.length; i < len; i++) {
    buf[i] = str.charCodeAt(i);
  }
  return buf;
};


// convert array to string
exports.buf2string = function (buf, max) {
  var i, out, c, c_len;
  var len = max || buf.length;

  // Reserve max possible length (2 words per char)
  // NB: by unknown reasons, Array is significantly faster for
  //     String.fromCharCode.apply than Uint16Array.
  var utf16buf = new Array(len*2);

  for (out=0, i=0; i<len;) {
    c = buf[i++];
    // quick process ascii
    if (c < 0x80) { utf16buf[out++] = c; continue; }

    c_len = _utf8len[c];
    // skip 5 & 6 byte codes
    if (c_len > 4) { utf16buf[out++] = 0xfffd; i += c_len-1; continue; }

    // apply mask on first byte
    c &= c_len === 2 ? 0x1f : c_len === 3 ? 0x0f : 0x07;
    // join the rest
    while (c_len > 1 && i < len) {
      c = (c << 6) | (buf[i++] & 0x3f);
      c_len--;
    }

    // terminated by end of string?
    if (c_len > 1) { utf16buf[out++] = 0xfffd; continue; }

    if (c < 0x10000) {
      utf16buf[out++] = c;
    } else {
      c -= 0x10000;
      utf16buf[out++] = 0xd800 | ((c >> 10) & 0x3ff);
      utf16buf[out++] = 0xdc00 | (c & 0x3ff);
    }
  }

  return buf2binstring(utf16buf, out);
};


// Calculate max possible position in utf8 buffer,
// that will not break sequence. If that's not possible
// - (very small limits) return max size as is.
//
// buf[] - utf8 bytes array
// max   - length limit (mandatory);
exports.utf8border = function(buf, max) {
  var pos;

  max = max || buf.length;
  if (max > buf.length) { max = buf.length; }

  // go back from last position, until start of sequence found
  pos = max-1;
  while (pos >= 0 && (buf[pos] & 0xC0) === 0x80) { pos--; }

  // Fuckup - very small and broken sequence,
  // return max, because we should return something anyway.
  if (pos < 0) { return max; }

  // If we came to start of buffer - that means vuffer is too small,
  // return max too.
  if (pos === 0) { return max; }

  return (pos + _utf8len[buf[pos]] > max) ? pos : max;
};