I don't think either @Buffy or I are being obstructionists here... but only saying that fine-grained attribution is not usually possible... nor interesting to the parties involved!
As in an earlier comment of mine: I think that, in fact, many substantial mathematics papers are not "the thing itself", but are narratives of something, some chain of events, some discussions among the authors, etc. So "the paper" can often be somewhat artificial in format and voice. In particular, with some exceptions, the lemmas and theorems are just a way of formatting and organizing that narrative. That formatting and organization need not much reflect the authors' conceptions... but things have to be written out, and there are strong traditional rules about how things should be written out.
And, again, it seems that many mathematicians are somewhat allergic to the idea of discrimination among authors. Hence, alphabetical order of authors, and definitely no tradition of having the introduction tell who did what. With some exceptions, of course.