A caveat and then an argument.
Language is the symbolic representation of ideas, and exceptions can be imported into any symbolic representation. For example: suppose we had no word that meant "rape". What is rape? To rape is "to have sex, except for sex with consent". So the absolute statement, "never rape" is exactly identical to the conditional statement, "never have sex, except for sex with consent".
You can redefine conditionals to be absolutes simply defining a new term. "Never steal, except to preserve a life," can be "Never florbleblarp," where florbleblarp is the act of taking something that is not yours in a situation in which lives would not be saved by the taking.
You can also reframe conditionals as absolutes by appropriately thinking them through. We could replace "Never lie except..." with "never deprive a person of a truth that they might use virtuously". This simultaneously solves the problem of lying-by-omission to circumvent the absolute prohibition on lying, and the problem of "when the secret police knock on the door and demand to know if you have any persecuted minorities hiding in your attic."
In short: a deontological system is one which consists of rules, whether or not the symbolic representations of those rules include words like "if" or "except".
A consequentialist argument for being a deontologist:
A system of morality is judged on its consequences. Therefore if deontological ethics generates, on average, better consequences than consequentialist ethics, one ought to act as though deontological ethics were correct, whether one is a consequentialist or a deontological ethicist.
Why might we expect an absolute morality to generate better consequences than a consequentialist morality?
Humans lack total knowledge of outcomes. Humans are extremely vulnerable to cognitive bias when it comes to predicting outcomes. Humans are extremely bad at doing accurate consequentialist calculus rapidly in their heads under stress; any attempt to try will usually lead to stalling until the chance to do the right thing has already come and gone. Even humans with plenty of time to think about something are pretty bad at not convincing themselves through specious arguments to do whatever provides the most short-term personal gratification and the least social stress.
Rule-based morality is resilient to cognitive biases specifically because it is indifferent to arguments about consequences. Absolute morality exports all the calculus to the level of society: an individual need not assess or even know the priors that led to the imperative, he or she has direct access to the conclusion in the form of the imperative. Absolute morality is accessible immediately under stress. Absolute morality, given time to think, is resilient against arguments to do whatever provides the most short-term personal gratification and even tends to have a prohibition against doing exactly that.