Method _consumeEntity

packages/compiler/src/ml_parser/lexer.ts:705–753

(textTokenType: TokenType)

  }

  private _consumeEntity(textTokenType: TokenType): void {
    this._beginToken(TokenType.ENCODED_ENTITY);
    const start = this._cursor.clone();
    this._cursor.advance();
    if (this._attemptCharCode(chars.$HASH)) {
      const isHex = this._attemptCharCode(chars.$x) || this._attemptCharCode(chars.$X);
      const codeStart = this._cursor.clone();
      this._attemptCharCodeUntilFn(isDigitEntityEnd);
      if (this._cursor.peek() != chars.$SEMICOLON) {
        // Advance cursor to include the peeked character in the string provided to the error
        // message.
        this._cursor.advance();
        const entityType = isHex ? CharacterReferenceType.HEX : CharacterReferenceType.DEC;
        throw this._createError(
          _unparsableEntityErrorMsg(entityType, this._cursor.getChars(start)),
          this._cursor.getSpan(),
        );
      }
      const strNum = this._cursor.getChars(codeStart);
      this._cursor.advance();
      try {
        const charCode = parseInt(strNum, isHex ? 16 : 10);
        this._endToken([String.fromCodePoint(charCode), this._cursor.getChars(start)]);
      } catch {
        throw this._createError(
          _unknownEntityErrorMsg(this._cursor.getChars(start)),
          this._cursor.getSpan(),
        );
      }
    } else {
      const nameStart = this._cursor.clone();
      this._attemptCharCodeUntilFn(isNamedEntityEnd);
      if (this._cursor.peek() != chars.$SEMICOLON) {
        // No semicolon was found so abort the encoded entity token that was in progress, and treat
        // this as a text token
        this._beginToken(textTokenType, start);
        this._cursor = nameStart;
        this._endToken(['&']);
      } else {
        const name = this._cursor.getChars(nameStart);
        this._cursor.advance();
        const char = Object.hasOwn(NAMED_ENTITIES, name) && NAMED_ENTITIES[name];
        if (!char) {
          throw this._createError(_unknownEntityErrorMsg(name), this._cursor.getSpan(start));
        }
        this._endToken([char, `&${name};`]);
      }
    }
  }

  private _consumeRawText(consumeEntities: boolean, endMarkerPredicate: () => boolean): void {
    this._beginToken(consumeEntities ? TokenType.ESCAPABLE_RAW_TEXT : TokenType.RAW_TEXT);