What mode does the command `query-replace' activate?

Question

I'm editing my keybinds mode-by-mode, and I've now come to the query-replace-map.

I want to define my keybinds in a mode-hook to keep my files logically consistent, but I cannot determine which mode the command query-replace activates.

I've tried M-x describe-mode, but activating the minibuffer immediately disables query-replace, thus I cannot look at the list of modes the buffer is currently in.

Or are query-replace/isearch-query-replace even modes?
If not, what sort of functionality is this, and how should I handle the rebinding of keys in these peculiarities (using mode-hooks or any other method that I might not know of)?

Also are there any other caveats when dealing with these pseudo-modes?

Answer 1

query-replace is just a command, it is not a major or minor mode. Before the phase in which it prompts for what to do with each match, it has no special bindings and uses minibuffer-local-map (this is where the usual keys for input history, M-r, M-p and M-n are bound, for example). The second phase uses query-replace-map.

If there is a particular keybinding you want to change, in recent versions of Emacs you can find out which keymap contains it by using describe-key (bound by default to C-h k). You can usually use that even while a command is prompting you in the minibuffer, which is how I found that M-r is bound in minibuffer-local-map

Unfortunately, that doesn't work for keys in the second phase of query-replace! So how did I find out what map is used then? Well, since it's name starts with query-replace-map I could have simply guessed, but what I actually did was read the source code of query-replace (which led me to perform-replace which is where query-replace-map is used).