Can you elaborate on what you mean by “it’s almost assumed..”? I’m actively working on datasheet parsing at scale at Zenode.ai, and I haven’t read anything specifically about this…