I came across a document that was using a null char 0x00 as array items separator and it wasn't handled correctly. According to chapter 7.2.2 The White-space characters shown in Table 1 separate syntactic constructs such as names and numbers from each other. and Table 1 contains NUL, HT, LF, FF, CR, SP.
BaseParser::isEndOfName is missing NUL and FF, as a result, if you parse the array
where /DeviceRGB and 255 are separated by NUL, and you write it back you get