the problem is, this is an ancient PDF feature from the 90s that nobody has time to write and debug

so they just took an open source parser and slapped it in there

it’s hard to rewrite PDF parsing to a new language so that everything still works. Especially those weird features that people forgot about.