Method decode

src/common/input/TextDecoder.ts:139–345

* Decodes UTF8 byte sequences in `input` to UTF32 codepoints in `target`. * The methods assumes stream input and will store partly transmitted bytes * and decode them with the next data chunk. * Note: The method does no bound checks for target, therefore make sure * the provided data chu

(input: Uint8Array, target: Uint32Array)

   * Returns the number of written codepoints in `target`.
   */
  public decode(input: Uint8Array, target: Uint32Array): number {
    const length = input.length;

    if (!length) {
      return 0;
    }

    let size = 0;
    let byte1: number;
    let byte2: number;
    let byte3: number;
    let byte4: number;
    let codepoint = 0;
    let startPos = 0;

    // handle leftover bytes
    if (this.interim[0]) {
      let discardInterim = false;
      let cp = this.interim[0];
      cp &= ((((cp & 0xE0) === 0xC0)) ? 0x1F : (((cp & 0xF0) === 0xE0)) ? 0x0F : 0x07);
      let pos = 0;
      let tmp: number;
      while ((tmp = this.interim[++pos] & 0x3F) && pos < 4) {
        cp <<= 6;
        cp |= tmp;
      }
      // missing bytes - read ahead from input
      const type = (((this.interim[0] & 0xE0) === 0xC0)) ? 2 : (((this.interim[0] & 0xF0) === 0xE0)) ? 3 : 4;
      const missing = type - pos;
      while (startPos < missing) {
        if (startPos >= length) {
          return 0;
        }
        tmp = input[startPos++];
        if ((tmp & 0xC0) !== 0x80) {
          // wrong continuation, discard interim bytes completely
          startPos--;
          discardInterim = true;
          break;
        } else {
          // need to save so we can continue short inputs in next call
          this.interim[pos++] = tmp;
          cp <<= 6;
          cp |= tmp & 0x3F;
        }
      }
      if (!discardInterim) {
        // final test is type dependent
        if (type === 2) {
          if (cp < 0x80) {
            // wrong starter byte
            startPos--;
          } else {
            target[size++] = cp;
          }
        } else if (type === 3) {
          if (cp < 0x0800 || (cp >= 0xD800 && cp <= 0xDFFF) || cp === 0xFEFF) {
            // illegal codepoint or BOM