wetsuite.helpers.split.Fragments

Hand the document bytestring into this. Nothing happens yet; you call accepts(), then suitableness(), then possibly fragments() -- see example use in decide().

whether we would consider parsing that at all. Often, "is this the right file type".

yields a tuple for each fragment

e.g.

• 5: I recognize that's PDF, from OP, and specifically Stcrt so I probably know how to fetch out the text fairly well
• 50: I recognize that's PDF, from OP, so I may do better than entirely generic
• 500: I recognize that's PDF, I will do something generic (because I am a fallback for PDFs)
• 5000: I recognize that's PDF, but I'm specific and it's probably a bad idea if I do something generic The idea is that with multiple of these, we can find the thing that (says) is most specific to this document.