There is no hindsight effect about the notion of identifying classes of text in a string of text. I fully and completely anticipate that patent being invalidated.
"If something is blindingly obvious at the time, it probably won't get a patent. The USPTO and other such agencies are not perfect but they are not total imbeciles."
The USPTO has shown, time and time again, to be willing imebeciles. They have habitually approved virtually all patents, opening them for review only when contested. I have zero expectation that they are doing any serious review of these patents, and they most certainly are not consulting any industry experts to gauge the true novelty of a patent.