JSON-ify sleep-affecting mutations, trying to sleep, and sleep comfort

[phi2dao]

Summary

Infrastructure "JSON-ify sleep-affecting mutations"

Purpose of change

Several mutations change where a character can try to sleep. This is hard coded into those mutations. Accordingly, these mutations cannot be changed, nor can new sleep-affecting mutations be added, without editing code.

Describe the solution

Added an optional comfort field to traits, which is a list of comfort_data. comfort_data consists of:

• A list of conditions used to evaluate a location
• The comfort of that location if the conditions are true
• Several modifiers for mutations that take 'normal' comfort into account or can use sleep aids
• A message to print when a character tries to sleep in a location where the conditions are true
• A message to print after the previous message, currently used by post-threshold spiders to suggest spinning a web first.
• A message to print when a character actually falls asleep in that location (not currently used by any mutations)

When a character tries to sleep, all of their traits are checked for comfort data. Within a trait, the first comfort data with true conditions is chosen. If multiple traits choose comfort data, the comfort data with the worst base comfort is chosen. This is so "impossible" comfort levels, such as experienced by a post-threshold spider sleeping without a web, are prioritized.

The resulting comfort is cached so the handful of functions that use comfort and related information can fetch it without having to go back through all the comfort data. This cache is invalidated if time passes or the location changes.

Originally, I used dialogue conditions to evaluate locations, but I found myself using character-focused conditions such as u_underwater instead of location-focused conditions like map_terrain_has_flag. This was a problem, as NPCs evaluate the comfort of multiple locations around them to decide where to sleep, and character-focused conditions would result in not actually evaluating those locations.

Accordingly, I took a page from climbing aids' book and rolled my own conditions. They allow checking terrain, furniture, traps, fields, vehicles, character flags, and traits. With the exception of traps, all are used by at least one of the existing sleep-affecting mutations. I added traps because traps can affect comfort in the form of the bed roll, cot, and similar.

This solution has not changed the behavior of any mutations, with the following exceptions:

• Plants (CHLOROMORPH) can sleep on all PLOWABLE and PLANTABLE terrain, not just t_dirt, t_grass, t_dirtmound, t_pit, and t_pit_shallow. This can be changed back if desired.
• Gastropods (SHELL2 and SHELL3) can use sleep aids while in their shell and in a vehicle. They previously couldn't, but could use sleep aids when in their shell and outside of a vehicle. There was even a comment saying that they could still use sleep aids, so this seemed like a bug.
• Aqueous Repose no longer gives an extra 10 comfort on top of its base 10 comfort. WATERSLEEP was being checked in both Character::base_comfort_value and Character::sleep_spot and added 10 comfort in both places. This seemed unintentional.
• Aquatic Sleeper no longer allows characters to try and sleep in deep water. Previously, all it did was bypass the "You cannot sleep while swimming." message; it did not actually affect comfort, and thus your ability to actually fall asleep, in any way. I'm not actually sure how Aquatic Sleeper is supposed to differ from Aqueous Repose, so I just left it comfort_data-less.
• Sleeping in an uncomfortable location prints a message describing the location. Previously, certain mutation/situation combinations prevented this message from printing. I don't think it's a problem if a plant hears that the rock floor they are sleeping on is uncomfortable.

Describe alternatives you've considered

Adding a WATERSLEEP flag. That's what I did initially, then I made the mistake of thinking that creating a more robust solution "wouldn't be that bad".

Testing

I debug mutated each sleep-affecting trait one at a time and tried sleeping in all conditions that would produce a distinct result. For example, for Aqueous Repose I tried sleeping on land, in deep water, and in swimmable terrain. When I got a result that didn't match the hardcoding, I fixed the code or json until it did. As of my most recent test, all sleep-affecting mutations behave as they used to, save for the above-mentioned exceptions.

Additional context

I somehow missed the "keep your PR small" advice until right now, so, uh, sorry. Almost all the new code is in sleep.h and sleep.cpp if that helps.

Also, I'm pretty sure there's nothing actually stopping a character from trying to fall asleep in mid-air (via the LEVITATION flag or the like). I didn't actually check, and didn't add any code to prevent that, but I thought it was pretty funny. I can fix it with another PR if desired.

There are also a bunch of hardcoded things that occur while a character is sleeping, some of which can probably be moved to enchantments. I didn't touch those either.

[Maleclypse]

I'm going to try and test this tonight or over the weekend and then merge it :)

[Maleclypse]

Thank you for your contribution!