I want to go to grad school in theoretical computer science. I don't like programming. I am in France.
Should I pursue a bachelors in math or in computer science?
I would say there is no "one best/most appropriate route" to becoming a theoretical computer scientist, so it is more a matter of your personal tastes, and even what part of theoretical computer science you are most interested in, which you may not even know yet. I know of great TCS researchers who started out in pure math, and great TCS researchers who started out in CS or (even!) EE.
(And by the way, to the commenters: I know plenty of good TCS researchers who more or less abhor programming, and I also know plenty who are not only great programmers, but great software engineers.)
Whatever route you take, if you think you are really interested in TCS then you should immerse yourself in it, especially to try to find out which parts of it you are really interested in. Introductory classes on algorithms (many of which are purely theoretical and involve no programming, if you so desire) or the theory of computation are good places to start, and beyond that it is a matter of reading, asking professors for references and pointers, sitting in on introductory TCS grad classes if you have the opportunity, etc. From that point of view, your particular degree doesn't matter so much (unless e.g. if you do a mathematics degree your university would perversely prevent you from taking a course on algorithms or theory of computation).
However, it also sounds like it is still so early in your potential research career that it would be healthy to question why you are interested in TCS (even asking this question can lead you on a path to discovering exactly what parts of TCS you are really interested in, and give you better motivation for your future work), and to expose yourself to plenty of other topics to see if your true interests end up being elsewhere.