src/library_strings.js

/**
 * @license
 * Copyright 2020 The Emscripten Authors
 * SPDX-License-Identifier: MIT
 */

#include "arrayUtils.js"

mergeInto(LibraryManager.library, {
  $intArrayFromString__docs: '/** @type {function(string, boolean=, number=)} */',
  $intArrayFromString: intArrayFromString,
  $intArrayToString: intArrayToString,

  // Given a pointer 'ptr' to a null-terminated ASCII-encoded string in the
  // emscripten HEAP, returns a copy of that string as a Javascript String
  // object.
  $AsciiToString: function(ptr) {
#if CAN_ADDRESS_2GB
    ptr >>>= 0;
#endif
    var str = '';
    while (1) {
      var ch = {{{ makeGetValue('ptr++', 0, 'u8') }}};
      if (!ch) return str;
      str += String.fromCharCode(ch);
    }
  },

  // Copies the given Javascript String object 'str' to the emscripten HEAP at
  // address 'outPtr', null-terminated and encoded in ASCII form. The copy will
  // require at most str.length+1 bytes of space in the HEAP.
  $stringToAscii__deps: ['$writeAsciiToMemory'],
  $stringToAscii: function(str, outPtr) {
    return writeAsciiToMemory(str, outPtr, false);
  },

#if TEXTDECODER == 2
  $UTF16Decoder: "new TextDecoder('utf-16le');",
#elif TEXTDECODER == 1
  $UTF16Decoder: "typeof TextDecoder != 'undefined' ? new TextDecoder('utf-16le') : undefined;",
#endif

  // Given a pointer 'ptr' to a null-terminated UTF16LE-encoded string in the
  // emscripten HEAP, returns a copy of that string as a Javascript String
  // object.
#if TEXTDECODER
  $UTF16ToString__deps: ['$UTF16Decoder'],
#endif
  $UTF16ToString: function(ptr, maxBytesToRead) {
#if ASSERTIONS
    assert(ptr % 2 == 0, 'Pointer passed to UTF16ToString must be aligned to two bytes!');
#endif
#if TEXTDECODER
    var endPtr = ptr;
    // TextDecoder needs to know the byte length in advance, it doesn't stop on
    // null terminator by itself.
    // Also, use the length info to avoid running tiny strings through
    // TextDecoder, since .subarray() allocates garbage.
    var idx = endPtr >> 1;
    var maxIdx = idx + maxBytesToRead / 2;
    // If maxBytesToRead is not passed explicitly, it will be undefined, and this
    // will always evaluate to true. This saves on code size.
    while (!(idx >= maxIdx) && HEAPU16[idx]) ++idx;
    endPtr = idx << 1;

#if TEXTDECODER != 2
    if (endPtr - ptr > 32 && UTF16Decoder)
#endif // TEXTDECODER != 2
      return UTF16Decoder.decode({{{ getUnsharedTextDecoderView('HEAPU8', 'ptr', 'endPtr') }}});
#endif // TEXTDECODER

#if TEXTDECODER != 2
    // Fallback: decode without UTF16Decoder
    var str = '';

    // If maxBytesToRead is not passed explicitly, it will be undefined, and the
    // for-loop's condition will always evaluate to true. The loop is then
    // terminated on the first null char.
    for (var i = 0; !(i >= maxBytesToRead / 2); ++i) {
      var codeUnit = {{{ makeGetValue('ptr', 'i*2', 'i16') }}};
      if (codeUnit == 0) break;
      // fromCharCode constructs a character from a UTF-16 code unit, so we can
      // pass the UTF16 string right through.
      str += String.fromCharCode(codeUnit);
    }

    return str;
#endif // TEXTDECODER != 2
  },

  // Copies the given Javascript String object 'str' to the emscripten HEAP at
  // address 'outPtr', null-terminated and encoded in UTF16 form. The copy will
  // require at most str.length*4+2 bytes of space in the HEAP.  Use the
  // function lengthBytesUTF16() to compute the exact number of bytes (excluding
  // null terminator) that this function will write.
  // Parameters:
  //   str: the Javascript string to copy.
  //   outPtr: Byte address in Emscripten HEAP where to write the string to.
  //   maxBytesToWrite: The maximum number of bytes this function can write to
  //                    the array. This count should include the null
  //                    terminator, i.e. if maxBytesToWrite=2, only the null
  //                    terminator will be written and nothing else.
  //                    maxBytesToWrite<2 does not write any bytes to the
  //                    output, not even the null terminator.
  // Returns the number of bytes written, EXCLUDING the null terminator.
  $stringToUTF16: function(str, outPtr, maxBytesToWrite) {
#if ASSERTIONS
    assert(outPtr % 2 == 0, 'Pointer passed to stringToUTF16 must be aligned to two bytes!');
#endif
#if ASSERTIONS
    assert(typeof maxBytesToWrite == 'number', 'stringToUTF16(str, outPtr, maxBytesToWrite) is missing the third parameter that specifies the length of the output buffer!');
#endif
    // Backwards compatibility: if max bytes is not specified, assume unsafe unbounded write is allowed.
    if (maxBytesToWrite === undefined) {
      maxBytesToWrite = 0x7FFFFFFF;
    }
    if (maxBytesToWrite < 2) return 0;
    maxBytesToWrite -= 2; // Null terminator.
    var startPtr = outPtr;
    var numCharsToWrite = (maxBytesToWrite < str.length*2) ? (maxBytesToWrite / 2) : str.length;
    for (var i = 0; i < numCharsToWrite; ++i) {
      // charCodeAt returns a UTF-16 encoded code unit, so it can be directly written to the HEAP.
      var codeUnit = str.charCodeAt(i); // possibly a lead surrogate
      {{{ makeSetValue('outPtr', 0, 'codeUnit', 'i16') }}};
      outPtr += 2;
    }
    // Null-terminate the pointer to the HEAP.
    {{{ makeSetValue('outPtr', 0, 0, 'i16') }}};
    return outPtr - startPtr;
  },

  // Returns the number of bytes the given Javascript string takes if encoded as
  // a UTF16 byte array, EXCLUDING the null terminator byte.
  $lengthBytesUTF16: function(str) {
    return str.length*2;
  },

  $UTF32ToString: function(ptr, maxBytesToRead) {
#if ASSERTIONS
    assert(ptr % 4 == 0, 'Pointer passed to UTF32ToString must be aligned to four bytes!');
#endif
    var i = 0;

    var str = '';
    // If maxBytesToRead is not passed explicitly, it will be undefined, and this
    // will always evaluate to true. This saves on code size.
    while (!(i >= maxBytesToRead / 4)) {
      var utf32 = {{{ makeGetValue('ptr', 'i*4', 'i32') }}};
      if (utf32 == 0) break;
      ++i;
      // Gotcha: fromCharCode constructs a character from a UTF-16 encoded code (pair), not from a Unicode code point! So encode the code point to UTF-16 for constructing.
      // See http://unicode.org/faq/utf_bom.html#utf16-3
      if (utf32 >= 0x10000) {
        var ch = utf32 - 0x10000;
        str += String.fromCharCode(0xD800 | (ch >> 10), 0xDC00 | (ch & 0x3FF));
      } else {
        str += String.fromCharCode(utf32);
      }
    }
    return str;
  },

  // Copies the given Javascript String object 'str' to the emscripten HEAP at
  // address 'outPtr', null-terminated and encoded in UTF32 form. The copy will
  // require at most str.length*4+4 bytes of space in the HEAP.
  // Use the function lengthBytesUTF32() to compute the exact number of bytes
  // (excluding null terminator) that this function will write.
  // Parameters:
  //   str: the Javascript string to copy.
  //   outPtr: Byte address in Emscripten HEAP where to write the string to.
  //   maxBytesToWrite: The maximum number of bytes this function can write to
  //                    the array. This count should include the null
  //                    terminator, i.e. if maxBytesToWrite=4, only the null
  //                    terminator will be written and nothing else.
  //                    maxBytesToWrite<4 does not write any bytes to the
  //                    output, not even the null terminator.
  // Returns the number of bytes written, EXCLUDING the null terminator.
  $stringToUTF32: function(str, outPtr, maxBytesToWrite) {
#if CAN_ADDRESS_2GB
    outPtr >>>= 0;
#endif
#if ASSERTIONS
    assert(outPtr % 4 == 0, 'Pointer passed to stringToUTF32 must be aligned to four bytes!');
#endif
#if ASSERTIONS
    assert(typeof maxBytesToWrite == 'number', 'stringToUTF32(str, outPtr, maxBytesToWrite) is missing the third parameter that specifies the length of the output buffer!');
#endif
    // Backwards compatibility: if max bytes is not specified, assume unsafe unbounded write is allowed.
    if (maxBytesToWrite === undefined) {
      maxBytesToWrite = 0x7FFFFFFF;
    }
    if (maxBytesToWrite < 4) return 0;
    var startPtr = outPtr;
    var endPtr = startPtr + maxBytesToWrite - 4;
    for (var i = 0; i < str.length; ++i) {
      // Gotcha: charCodeAt returns a 16-bit word that is a UTF-16 encoded code unit, not a Unicode code point of the character! We must decode the string to UTF-32 to the heap.
      // See http://unicode.org/faq/utf_bom.html#utf16-3
      var codeUnit = str.charCodeAt(i); // possibly a lead surrogate
      if (codeUnit >= 0xD800 && codeUnit <= 0xDFFF) {
        var trailSurrogate = str.charCodeAt(++i);
        codeUnit = 0x10000 + ((codeUnit & 0x3FF) << 10) | (trailSurrogate & 0x3FF);
      }
      {{{ makeSetValue('outPtr', 0, 'codeUnit', 'i32') }}};
      outPtr += 4;
      if (outPtr + 4 > endPtr) break;
    }
    // Null-terminate the pointer to the HEAP.
    {{{ makeSetValue('outPtr', 0, 0, 'i32') }}};
    return outPtr - startPtr;
  },

  // Returns the number of bytes the given Javascript string takes if encoded as
  // a UTF16 byte array, EXCLUDING the null terminator byte.
  $lengthBytesUTF32: function(str) {
    var len = 0;
    for (var i = 0; i < str.length; ++i) {
      // Gotcha: charCodeAt returns a 16-bit word that is a UTF-16 encoded code unit, not a Unicode code point of the character! We must decode the string to UTF-32 to the heap.
      // See http://unicode.org/faq/utf_bom.html#utf16-3
      var codeUnit = str.charCodeAt(i);
      if (codeUnit >= 0xD800 && codeUnit <= 0xDFFF) ++i; // possibly a lead surrogate, so skip over the tail surrogate.
      len += 4;
    }

    return len;
  },

  // Allocate heap space for a JS string, and write it there.
  // It is the responsibility of the caller to free() that memory.
  $allocateUTF8: function(str) {
    var size = lengthBytesUTF8(str) + 1;
    var ret = {{{ makeMalloc('allocateUTF8', 'size') }}};
    if (ret) stringToUTF8Array(str, HEAP8, ret, size);
    return ret;
  },

  // Allocate stack space for a JS string, and write it there.
  $allocateUTF8OnStack: function(str) {
    var size = lengthBytesUTF8(str) + 1;
    var ret = stackAlloc(size);
    stringToUTF8Array(str, HEAP8, ret, size);
    return ret;
  },

  // Deprecated: This function should not be called because it is unsafe and
  // does not provide a maximum length limit of how many bytes it is allowed to
  // write. Prefer calling the function stringToUTF8Array() instead, which takes
  // in a maximum length that can be used to be secure from out of bounds
  // writes.
  $writeStringToMemory__docs: '/** @deprecated @param {boolean=} dontAddNull */',
  $writeStringToMemory: function(string, buffer, dontAddNull) {
    warnOnce('writeStringToMemory is deprecated and should not be called! Use stringToUTF8() instead!');

    var /** @type {number} */ lastChar, /** @type {number} */ end;
    if (dontAddNull) {
      // stringToUTF8Array always appends null. If we don't want to do that, remember the
      // character that existed at the location where the null will be placed, and restore
      // that after the write (below).
      end = buffer + lengthBytesUTF8(string);
      lastChar = HEAP8[end];
    }
    stringToUTF8(string, buffer, Infinity);
    if (dontAddNull) HEAP8[end] = lastChar; // Restore the value under the null character.
  },

  $writeArrayToMemory: function(array, buffer) {
#if ASSERTIONS
    assert(array.length >= 0, 'writeArrayToMemory array must have a length (should be an array or typed array)')
#endif
    HEAP8.set(array, buffer);
  },

  $writeAsciiToMemory__docs: '/** @param {boolean=} dontAddNull */',
  $writeAsciiToMemory: function(str, buffer, dontAddNull) {
    for (var i = 0; i < str.length; ++i) {
#if ASSERTIONS
      assert(str.charCodeAt(i) === (str.charCodeAt(i) & 0xff));
#endif
      {{{ makeSetValue('buffer++', 0, 'str.charCodeAt(i)', 'i8') }}};
    }
    // Null-terminate the pointer to the HEAP.
    if (!dontAddNull) {{{ makeSetValue('buffer', 0, 0, 'i8') }}};
  },
});