Before answering an inappropriate email from a PI or collaborator, think the following.

Is this person worth replying to? For it is known that many academics, especially the ones that pride themselves to be “on top”, match criteria for personality disorders that predict they will not read your reply, and no matter how serious the circumstances you faced were or how extensively illustrated and explained they are, they will still be dismissed as “excuses”.
It is also known that although a person may not fit such criteria, they may have absorbed such behavior as a side effect of their work environment, which means the net result is still that your reply will not be read.

Consider first if the issue is due to the other site in reality not being able to read your information, if you have made it available for them in the past. Even if you did send them your work, it may have been misplaced, corrupted, or otherwise lost. If the request concerns information that has been produced in the past, but it is not immediately available to you (because a long time has past, because you left the project, or you have archived it for whatever reason), your situation is the same as having a request for incomplete work, because now you have to go back and retrieve it and what I say next applies all the same.

If this person is requesting data that is not completely processed, or not yet in conditions to be presented, two things can happen:
a) they will get the attachment, and put your data in a folder that will not be touched for months, and sleep soundly because they broke you to the point you gave them what you could and what you couldn’t, or
b) they will look at it, and dismiss your work (and possibly you) for doing substandard work.
In either situation, a reply is therefore a waste of time. To some, they will send substandard information as in A, regardless, and this is not without consequence to a third party: as the unrealistic request has been done to you, it has likely been done to other co-workers too, and if you are told something is “ready” and you “just have to go and use it” in a very large project, you may be caught by surprise by the fact that it is not: people will have told this person things were ready because they did not want to deal with a temper tantrum, and knew the data would not be vouched.

If, due to personal commitment, you feel you should send any information that is still not available, let it be for yourself and your own peace of mind, and not because there is someone pressing you. For while some people think things will only be produced as a result of aggressiveness, you do not want to reinforce this thought. Notwithstanding your effort, it is very unlikely that the person will change their belief, though, and therefore you have no control over it. One does have control over first, the impact of such a request on your own health (and priority queue), and second, whether you should spend time questioning the fairness of such request, for people who send aggressive requests with unrealistic deadlines do not set those same standards for themselves.

If a request is made for raw data that can be answered in a few days more with a journal-formatted attachment, it is better to go for the second, since in the vision of such a person, they often are unable to put pieces together and see a paper is ready when it is not in journal format. By formatting your own work, you ensure your remaining in the authorship line (in a fair world).

As a last resource, consider also that some difficulties may be so common they may even be acknowledged by HR , and therefore there are official policies in place which may prevent you from falling victim to retaliation. You should consider that because collaborators that make such requests often engage in unprofessional behavior, and will not hold themselves to the same standards they expect from you.