Method #captureText

xml/_tokenizer.ts:799–866

* Capture text content in a tight loop until '<' is found. * Returns true if '<' was found, false if end of buffer reached. * * The "]]>" check (XML 1.0 §2.4) is deferred to #flushText where * a single native `includes` covers both within-chunk and cross-chunk cases. * * Il

(buffer: string, bufferLen: number)

   * regex pre-check in {@link #flushText} for the no-position-tracking path.
   */
  #captureText(buffer: string, bufferLen: number): boolean {
    // Initialize text tracking if this is the start of a new text node
    if (this.#textStartIdx === -1) {
      if (this.#trackPosition) {
        this.#textStartLine = this.#line;
        this.#textStartColumn = this.#column;
        this.#textStartOffset = this.#offset;
      }
      this.#textStartIdx = this.#bufferIndex;
    }

    if (this.#trackPosition) {
      // Scan for '<' while tracking line/column positions.
      // Illegal C0 chars are checked here (we're already per-char).
      let idx = this.#bufferIndex;
      let line = this.#line;
      let column = this.#column;
      let offset = this.#offset;

      while (idx < bufferLen) {
        const code = buffer.charCodeAt(idx);
        if (code === CC_LT) {
          this.#bufferIndex = idx;
          this.#line = line;
          this.#column = column;
          this.#offset = offset;
          return true;
        }

        if (this.#isIllegalLiteralChar(code)) {
          this.#bufferIndex = idx;
          this.#line = line;
          this.#column = column;
          this.#offset = offset;
          this.#error(
            `Illegal XML character U+${
              code.toString(16).toUpperCase().padStart(4, "0")
            }`,
          );
        }

        if (code === CC_LF) {
          line++;
          column = 1;
        } else {
          column++;
        }
        offset++;
        idx++;
      }

      this.#bufferIndex = idx;
      this.#line = line;
      this.#column = column;
      this.#offset = offset;
    } else {
      // Fast path: native indexOf is SIMD-optimized in V8.
      // Illegal C0 chars are checked in #flushText via regex.