Room Vaults

About this tutorial

This tutorial is free and open source, and all code uses the MIT license - so you are free to do with it as you like. My hope is that you will enjoy the tutorial, and make great games!

If you enjoy this and would like me to keep writing, please consider supporting my Patreon.

Hands-On Rust

The last chapter was getting overly long, so it was broken into two. In the previous chapter, we learned how to load prefabricated maps and map sections, modified the spawn system so that meta-builders could affect the spawn patterns from the previous builder, and demonstrated integration of whole map chunks into levels. In this chapter, we'll explode room vaults - prefabricated content that integrates itself into your level. So you might hand-craft some rooms, and have them seamlessly fit into your existing map.

Designing a room: Totally Not A Trap

The life of a roguelike developer is part programmer, part interior decorator (in a weirdly Gnome Mad Scientist fashion). We've already designed whole levels and level sections, so it isn't a huge leap to designing rooms. Lets go ahead and build a few pre-designed rooms.

We'll make a new file in map_builders/prefab_builders called prefab_rooms.rs. We'll insert a relatively iconic map feature into it:


#![allow(unused)]
fn main() {
#[allow(dead_code)]
#[derive(PartialEq, Copy, Clone)]
pub struct PrefabRoom {
    pub template : &'static str,
    pub width : usize,
    pub height: usize,
    pub first_depth: i32,
    pub last_depth: i32
}

#[allow(dead_code)]
pub const TOTALLY_NOT_A_TRAP : PrefabRoom = PrefabRoom{
    template : TOTALLY_NOT_A_TRAP_MAP,
    width: 5,
    height: 5,
    first_depth: 0,
    last_depth: 100
};

#[allow(dead_code)]
const TOTALLY_NOT_A_TRAP_MAP : &str = "
     
 ^^^ 
 ^!^ 
 ^^^ 
     
";
}

If you look at the ASCII, you'll see a classic piece of map design: a health potion completely surrounded by traps. Since the traps are hidden by default, we're relying on the player to think "well, that doesn't look suspicious at all"! Not that there are spaces all around the content - there's a 1-tile gutter all around it. This ensures that any 5x5 room into which the vault is placed will still be traversable. We're also introducing first_depth and last_depth - these are the levels at which the vault might be applied; for the sake of introduction, we'll pick 0..100 - which should be every level, unless you are a really dedicated play-tester!

Placing the not-a-trap room

We'll start by adding another mode to the PrefabBuiler system:


#![allow(unused)]
fn main() {
#[derive(PartialEq, Copy, Clone)]
#[allow(dead_code)]
pub enum PrefabMode { 
    RexLevel{ template : &'static str },
    Constant{ level : prefab_levels::PrefabLevel },
    Sectional{ section : prefab_sections::PrefabSection },
    RoomVaults
}
}

We're not going to add any parameters yet - by the end of the chapter, we'll have it integrated into a broader system for placing vaults. We'll update our constructor to use this type of placement:


#![allow(unused)]
fn main() {
impl PrefabBuilder {
    #[allow(dead_code)]
    pub fn new(new_depth : i32, previous_builder : Option<Box<dyn MapBuilder>>) -> PrefabBuilder {
        PrefabBuilder{
            map : Map::new(new_depth),
            starting_position : Position{ x: 0, y : 0 },
            depth : new_depth,
            history : Vec::new(),
            mode : PrefabMode::RoomVaults,
            previous_builder,
            spawn_list : Vec::new()
        }
    }
    ...
}

And we'll teach our match function in build to use it:


#![allow(unused)]
fn main() {
fn build(&mut self) {
    match self.mode {
        PrefabMode::RexLevel{template} => self.load_rex_map(&template),
        PrefabMode::Constant{level} => self.load_ascii_map(&level),
        PrefabMode::Sectional{section} => self.apply_sectional(&section),
        PrefabMode::RoomVaults => self.apply_room_vaults()
    }
    self.take_snapshot();
    ...
}

That leaves the next logical step being to write apply_room_vaults. Our objective is to scan the incoming map (from a different builder, even a previous iteration of this one!) for appropriate places into which we can place a vault, and add it to the map. We'll also want to remove any spawned creatures from the vault area - so the vaults remain hand-crafted and aren't interfered with by random spawning.

We'll be re-using our "create previous iteration" code from apply_sectional - so lets rewrite it into a more generic form:


#![allow(unused)]
fn main() {
fn apply_previous_iteration<F>(&mut self, mut filter: F) 
    where F : FnMut(i32, i32, &(usize, String)) -> bool
{
    // Build the map
    let prev_builder = self.previous_builder.as_mut().unwrap();
    prev_builder.build_map();
    self.starting_position = prev_builder.get_starting_position();
    self.map = prev_builder.get_map().clone();   
    for e in prev_builder.get_spawn_list().iter() {
        let idx = e.0;
        let x = idx as i32 % self.map.width;
        let y = idx as i32 / self.map.width;
        if filter(x, y, e) {
            self.spawn_list.push(
                (idx, e.1.to_string())
            )
        }
    }        
    self.take_snapshot(); 
}
}

There's a lot of new Rust here! Lets walk through it:

  1. You'll notice that we've added a template type to the function. fn apply_previous_iteration<F>. This specifies that we don't know exactly what F is when we write the function.
  2. The second parameter (mut filter: F) is also of type F. So we're telling the function signature to accept the template type as the parameter.
  3. Before the opening curly bracket, we've added a where clause. This type of clause can be used to limit what it accepted by the generic type. In this case, we're saying that F must be an FnMut. An FnMut is a function pointer that is allowed to change state (mutable; if it were immutable it'd be an Fn). We then specify the parameters of the function, and its return type. Inside the function, we can now treat filter like a function - even though we haven't actually written one. We're requiring that function accept two i32 (integers), and a tuple of (usize, String). The latter should look familiar - its our spawn list format. The first two are the x and y coordinates of the spawn - we're passing that to save the caller from doing the math each time.
  4. We then run the prev_builder code we wrote in the previous chapter - it builds the map and obtains the map data itself, along with the spawn_list from the previous algorithm.
  5. We then iterate through the spawn list, and calculate the x/y coordinates and map index for each entity. We call filter with this information, and if it returns true we add it to our spawn_list.
  6. Lastly, we take a snapshot of the map so you can see the step in action.

That sounds really complicated, but most of what it as done is allow us to replace the following code in apply_sectional:


#![allow(unused)]
fn main() {
// Build the map
let prev_builder = self.previous_builder.as_mut().unwrap();
prev_builder.build_map();
self.starting_position = prev_builder.get_starting_position();
self.map = prev_builder.get_map().clone();        
for e in prev_builder.get_spawn_list().iter() {
    let idx = e.0;
    let x = idx as i32 % self.map.width;
    let y = idx as i32 / self.map.width;
    if x < chunk_x || x > (chunk_x + section.width as i32) ||
        y < chunk_y || y > (chunk_y + section.height as i32) {
            self.spawn_list.push(
                (idx, e.1.to_string())
            )
        }
}        
self.take_snapshot();
}

We can replace it with a more generic call:


#![allow(unused)]
fn main() {
// Build the map
self.apply_previous_iteration(|x,y,e| {
    x < chunk_x || x > (chunk_x + section.width as i32) || y < chunk_y || y > (chunk_y + section.height as i32)
}); 
}

This is interesting: we're passing in a closure - a lambda function to the filter. It receives x, y, and e from the previous map's spawn_list for each entity. In this case, we're checking against chunk_x, chunk_y, section.width and section.height to see if the entity is inside our sectional. You've probably noticed that we didn't declare these anywhere in the lambda function; we are relying on capture - you can call a lambda and reference other variables that are in its scope - and it can reference them as if they were its own. This is a very powerful feature, and you can learn about it here.

Room Vaults

Let's start building apply_room_vaults. We'll take it step-by-step, and work our way through. We'll start with the function signature:


#![allow(unused)]
fn main() {
fn apply_room_vaults(&mut self) {
    use prefab_rooms::*;
    let mut rng = RandomNumberGenerator::new();
}

Simple enough: no parameters other than mutable membership of the builder. It is going to be referring to types in prefab_rooms, so rather than type that every time an in-function using statement imports the names to the local namespace to save your fingers. We'll also need a random number generator, so we make one as we have before. Next up:


#![allow(unused)]
fn main() {
// Apply the previous builder, and keep all entities it spawns (for now)
self.apply_previous_iteration(|_x,_y,_e| true);
}

We use the code we just wrote to apply the previous map. The filter we're passing in this time always returns true: keep all the entities for now. Next:


#![allow(unused)]
fn main() {
// Note that this is a place-holder and will be moved out of this function
let master_vault_list = vec![TOTALLY_NOT_A_TRAP];

// Filter the vault list down to ones that are applicable to the current depth
let possible_vaults : Vec<&PrefabRoom> = master_vault_list
    .iter()
    .filter(|v| { self.depth >= v.first_depth && self.depth <= v.last_depth })
    .collect();

if possible_vaults.is_empty() { return; } // Bail out if there's nothing to build

let vault_index = if possible_vaults.len() == 1 { 0 } else { (rng.roll_dice(1, possible_vaults.len() as i32)-1) as usize };
let vault = possible_vaults[vault_index];
}

We make a vector of all possible vault types - there's currently only one, but when we have more they go in here. This isn't really ideal, but we'll worry about making it a global resource in a future chapter. We then make a possible_vaults list by taking the master_vault_list and filtering it to only include those whose first_depth and last_depth line up with the requested dungeon depth. The iter().filter(...).collect() pattern has been described before, and it's a very powerful way to quickly extract what you need from a vector. If there are no possible vaults, we return out of the function - nothing to do here! Finally, we use another pattern we've used before: we pick a vault to create by selecting a random member of the possible_vaults vector.

Next up:


#![allow(unused)]
fn main() {
// We'll make a list of places in which the vault could fit
let mut vault_positions : Vec<Position> = Vec::new();

let mut idx = 0usize;
loop {
    let x = (idx % self.map.width as usize) as i32;
    let y = (idx / self.map.width as usize) as i32;

    // Check that we won't overflow the map
    if x > 1 
        && (x+vault.width as i32) < self.map.width-2
        && y > 1 
        && (y+vault.height as i32) < self.map.height-2
    {

        let mut possible = true;
        for ty in 0..vault.height as i32 {
            for tx in 0..vault.width as i32 {

                let idx = self.map.xy_idx(tx + x, ty + y);
                if self.map.tiles[idx] != TileType::Floor {
                    possible = false;
                }
            }
        }

        if possible {
            vault_positions.push(Position{ x,y });
            break;
        }

    }

    idx += 1;
    if idx >= self.map.tiles.len()-1 { break; }
}
}

There's quite a bit of code in this section (which determines all the places a the vault might fit). Lets walk through it:

  1. We make a new vector of Positions. This will contain all the possible places in which we could spawn our vault.
  2. We set idx to 0 - we plan to iterate through the whole map.
  3. We start a loop - the Rust loop type that doesn't exit until you call break.
    1. We calculate x and y to know where we are on the map.
    2. We do an overflow check; x needs to be greater than 1, and x+1 needs to be less than the map width. We check the same with y and the map height. If we're within the bounds:
      1. We set possible to true.
      2. We iterate every tile on the map in the range (x .. x+vault width), (y .. y + vault height) - if any tile isn't a floor, we set possible to false.
      3. If it is possible to place the vault here, we add the position to our vault_positions vector from step 1.
    3. We increment idx by 1.
    4. If we've run out of map, we break out of the loop.

In other words, we quickly scan the whole map for everywhere we could put the vault - and make a list of possible placements. We then:


#![allow(unused)]
fn main() {
if !vault_positions.is_empty() {
    let pos_idx = if vault_positions.len()==1 { 0 } else { (rng.roll_dice(1, vault_positions.len() as i32)-1) as usize };
    let pos = &vault_positions[pos_idx];

    let chunk_x = pos.x;
    let chunk_y = pos.y;

    let string_vec = PrefabBuilder::read_ascii_to_vec(vault.template);
    let mut i = 0;
    for ty in 0..vault.height {
        for tx in 0..vault.width {
            let idx = self.map.xy_idx(tx as i32 + chunk_x, ty as i32 + chunk_y);
            self.char_to_map(string_vec[i], idx);
            i += 1;
        }
    }
    self.take_snapshot();
}
}

So if there are any valid positions for the vault, we:

  1. Pick a random entry in the vault_positions vector - this is where we will place the vault.
  2. Use read_ascii_to_vec to read in the ASCII, just like we did in prefabs and sectionals.
  3. Iterate the vault data and use char_to_map to place it - just like we did before.

Putting it all together, you have the following function:


#![allow(unused)]
fn main() {
fn apply_room_vaults(&mut self) {
    use prefab_rooms::*;
    let mut rng = RandomNumberGenerator::new();

    // Apply the previous builder, and keep all entities it spawns (for now)
    self.apply_previous_iteration(|_x,_y,_e| true);

    // Note that this is a place-holder and will be moved out of this function
    let master_vault_list = vec![TOTALLY_NOT_A_TRAP];

    // Filter the vault list down to ones that are applicable to the current depth
    let possible_vaults : Vec<&PrefabRoom> = master_vault_list
        .iter()
        .filter(|v| { self.depth >= v.first_depth && self.depth <= v.last_depth })
        .collect();

    if possible_vaults.is_empty() { return; } // Bail out if there's nothing to build

    let vault_index = if possible_vaults.len() == 1 { 0 } else { (rng.roll_dice(1, possible_vaults.len() as i32)-1) as usize };
    let vault = possible_vaults[vault_index];

    // We'll make a list of places in which the vault could fit
    let mut vault_positions : Vec<Position> = Vec::new();

    let mut idx = 0usize;
    loop {
        let x = (idx % self.map.width as usize) as i32;
        let y = (idx / self.map.width as usize) as i32;

        // Check that we won't overflow the map
        if x > 1 
            && (x+vault.width as i32) < self.map.width-2
            && y > 1 
            && (y+vault.height as i32) < self.map.height-2
        {

            let mut possible = true;
            for ty in 0..vault.height as i32 {
                for tx in 0..vault.width as i32 {

                    let idx = self.map.xy_idx(tx + x, ty + y);
                    if self.map.tiles[idx] != TileType::Floor {
                        possible = false;
                    }
                }
            }

            if possible {
                vault_positions.push(Position{ x,y });
                break;
            }

        }

        idx += 1;
        if idx >= self.map.tiles.len()-1 { break; }
    }

    if !vault_positions.is_empty() {
        let pos_idx = if vault_positions.len()==1 { 0 } else { (rng.roll_dice(1, vault_positions.len() as i32)-1) as usize };
        let pos = &vault_positions[pos_idx];

        let chunk_x = pos.x;
        let chunk_y = pos.y;

        let string_vec = PrefabBuilder::read_ascii_to_vec(vault.template);
        let mut i = 0;
        for ty in 0..vault.height {
            for tx in 0..vault.width {
                let idx = self.map.xy_idx(tx as i32 + chunk_x, ty as i32 + chunk_y);
                self.char_to_map(string_vec[i], idx);
                i += 1;
            }
        }
        self.take_snapshot();
    }
}
}

It's more likely that a square vault will fit in rectangular rooms, so we'll pop over to map_builders/mod.rs and slightly adjust the random_builder to use the original simple map algorithm for the base map:


#![allow(unused)]
fn main() {
Box::new(
    PrefabBuilder::new(
        new_depth, 
        Some(
            Box::new(
                SimpleMapBuilder::new(new_depth)
            )
        )
    )
)
}

If you cargo run now, the vault will probably be placed on your map. Here's a screenshot of a run in which I found it:

Screenshot.

Filtering out entities

We probably don't want to keep entities that are inside our new vault from the previous map iteration. You might have a cunningly placed trap and spawn a goblin on top of it! (While fun, probably not what you had in mind). So we'll extend apply_room_vaults to do some filtering when it places the vault. We want to filter before we spawn new stuff, and then spawn more stuff with the room. Enter the retain feature:


#![allow(unused)]
fn main() {
...
let chunk_y = pos.y;

let width = self.map.width; // The borrow checker really doesn't like it
let height = self.map.height; // when we access `self` inside the `retain`
self.spawn_list.retain(|e| {
    let idx = e.0 as i32;
    let x = idx % width;
    let y = idx / height;
    x < chunk_x || x > chunk_x + vault.width as i32 || y < chunk_y || y > chunk_y + vault.height as i32
});
...
}

Calling retain on a vector iterates through every entry, and calls the passed closure/lambda function. If it returns true, then the element is retained (kept) - otherwise it is removed. So here we're catching width and height (to avoid borrowing self), and then calculate the location for each entry. If it is outside of the new vault - we keep it.

I want more than one vault!

Having only one vault is pretty dull - albeit a good start in terms of proving the functionality works. In prefab_rooms.rs we'll go ahead and write a couple more. These aren't intended to be seminal examples of level design, but they illustrate the process. We'll add some more room prefabs:


#![allow(unused)]
fn main() {
#[allow(dead_code)]
#[derive(PartialEq, Copy, Clone)]
pub struct PrefabRoom {
    pub template : &'static str,
    pub width : usize,
    pub height: usize,
    pub first_depth: i32,
    pub last_depth: i32
}

#[allow(dead_code)]
pub const TOTALLY_NOT_A_TRAP : PrefabRoom = PrefabRoom{
    template : TOTALLY_NOT_A_TRAP_MAP,
    width: 5,
    height: 5,
    first_depth: 0,
    last_depth: 100
};

#[allow(dead_code)]
const TOTALLY_NOT_A_TRAP_MAP : &str = "
     
 ^^^ 
 ^!^ 
 ^^^ 
     
";

#[allow(dead_code)]
pub const SILLY_SMILE : PrefabRoom = PrefabRoom{
    template : SILLY_SMILE_MAP,
    width: 6,
    height: 6,
    first_depth: 0,
    last_depth: 100
};

#[allow(dead_code)]
const SILLY_SMILE_MAP : &str = "
      
 ^  ^ 
  #  
      
 ### 
      
";

#[allow(dead_code)]
pub const CHECKERBOARD : PrefabRoom = PrefabRoom{
    template : CHECKERBOARD_MAP,
    width: 6,
    height: 6,
    first_depth: 0,
    last_depth: 100
};

#[allow(dead_code)]
const CHECKERBOARD_MAP : &str = "
      
 ^#  
 g#%# 
 #!#  
 ^# # 
      
";
}

We've added CHECKERBOARD (a grid of walls and spaces with traps, a goblin and goodies in it), and SILLY_SMILE which just looks like a silly wall feature. Now open up apply_room_vaults in map_builders/prefab_builder/mod.rs and add these to the master vector:


#![allow(unused)]
fn main() {
// Note that this is a place-holder and will be moved out of this function
let master_vault_list = vec![TOTALLY_NOT_A_TRAP, CHECKERBOARD, SILLY_SMILE];
}

If you cargo run now, you'll most likely encounter one of the three vaults. Each time you advance a depth, you will probably encounter one of the three. My test ran into the checkerboard almost immediately:

Screenshot.

That's a great start, and gives a bit of flair to maps as you descend - but it may not be quite what you were asking for when you said you wanted more than one vault! How about more than one vault on a level? Back to apply_room_vaults! It's easy enough to come up with a number of vaults to spawn:


#![allow(unused)]
fn main() {
let n_vaults = i32::min(rng.roll_dice(1, 3), possible_vaults.len() as i32);
}

This sets n_vaults to the minimum value of a dice roll (1d3) and the number of possible vaults - so it'll never exceed the number of options, but can vary a bit. It's also pretty easy to wrap the creation function in a for loop:


#![allow(unused)]
fn main() {
if possible_vaults.is_empty() { return; } // Bail out if there's nothing to build

        let n_vaults = i32::min(rng.roll_dice(1, 3), possible_vaults.len() as i32);

        for _i in 0..n_vaults {

            let vault_index = if possible_vaults.len() == 1 { 0 } else { (rng.roll_dice(1, possible_vaults.len() as i32)-1) as usize };
            let vault = possible_vaults[vault_index];

            ...

                self.take_snapshot();

                possible_vaults.remove(vault_index);
            }
        }
}

Notice that at the end of the loop, we're removing the vault we added from possible_vaults. We have to change the declaration to be able to do that: let mut possible_vaults : Vec<&PrefabRoom> = ... - we add the mut to allow us to change the vector. This way, we won't keep adding the same vault - they only get spawned once.

Now for the more difficult part: making sure that our new vaults don't overlap the previously spawned ones. We'll create a new HashSet of tiles we've consumed:


#![allow(unused)]
fn main() {
let mut used_tiles : HashSet<usize> = HashSet::new();
}

Hash sets have the advantage of offering a quick way to say if they contain a value, so they are ideal for what we need. We'll insert the tile idx into the set when we add a tile:


#![allow(unused)]
fn main() {
for ty in 0..vault.height {
    for tx in 0..vault.width {
        let idx = self.map.xy_idx(tx as i32 + chunk_x, ty as i32 + chunk_y);
        self.char_to_map(string_vec[i], idx);
        used_tiles.insert(idx);
        i += 1;
    }
}
}

Lastly, in our possibility checking we want to do a check against used_tiles to ensure we aren't overlapping:


#![allow(unused)]
fn main() {
let idx = self.map.xy_idx(tx + x, ty + y);
if self.map.tiles[idx] != TileType::Floor {
    possible = false;
}
if used_tiles.contains(&idx) {
    possible = false;
}
}

Now if you cargo run your project, you might encounter several vaults. Here's a case where we encountered two vaults:

Screenshot.

I don't always want a vault!

If you offer all of your vaults on every level, the game will be a bit more predictable than you probably want (unless you make a lot of vaults!). We'll modify apply_room_vaults to only sometimes have any vaults, with an increasing probability as you descend into the dungeon:


#![allow(unused)]
fn main() {
// Apply the previous builder, and keep all entities it spawns (for now)
self.apply_previous_iteration(|_x,_y,_e| true);

// Do we want a vault at all?
let vault_roll = rng.roll_dice(1, 6) + self.depth;
if vault_roll < 4 { return; }
}

This is very simple: we roll a six-sided dice and add the current depth. If we rolled less than 4, we bail out and just provide the previously generated map. If you cargo run your project now, you'll sometimes encounter vaults - and sometimes you won't.

Finishing up: offering some constructors other than just new

We should offer some more friendly ways to build our PrefabBuilder, so it's obvious what we're doing when we construct our builder chain. Add the following constructors to prefab_builder/mod.rs:


#![allow(unused)]
fn main() {
#[allow(dead_code)]
pub fn rex_level(new_depth : i32, template : &'static str) -> PrefabBuilder {
    PrefabBuilder{
        map : Map::new(new_depth),
        starting_position : Position{ x: 0, y : 0 },
        depth : new_depth,
        history : Vec::new(),
        mode : PrefabMode::RexLevel{ template },
        previous_builder : None,
        spawn_list : Vec::new()
    }
}

#[allow(dead_code)]
pub fn constant(new_depth : i32, level : prefab_levels::PrefabLevel) -> PrefabBuilder {
    PrefabBuilder{
        map : Map::new(new_depth),
        starting_position : Position{ x: 0, y : 0 },
        depth : new_depth,
        history : Vec::new(),
        mode : PrefabMode::Constant{ level },
        previous_builder : None,
        spawn_list : Vec::new()
    }
}

#[allow(dead_code)]
pub fn sectional(new_depth : i32, section : prefab_sections::PrefabSection, previous_builder : Box<dyn MapBuilder>) -> PrefabBuilder {
    PrefabBuilder{
        map : Map::new(new_depth),
        starting_position : Position{ x: 0, y : 0 },
        depth : new_depth,
        history : Vec::new(),
        mode : PrefabMode::Sectional{ section },
        previous_builder : Some(previous_builder),
        spawn_list : Vec::new()
    }
}

#[allow(dead_code)]
pub fn vaults(new_depth : i32, previous_builder : Box<dyn MapBuilder>) -> PrefabBuilder {
    PrefabBuilder{
        map : Map::new(new_depth),
        starting_position : Position{ x: 0, y : 0 },
        depth : new_depth,
        history : Vec::new(),
        mode : PrefabMode::RoomVaults,
        previous_builder : Some(previous_builder),
        spawn_list : Vec::new()
    }
}
}

We now have a decent interface for creating our meta-builder!

It's Turtles (Or Meta-Builders) All The Way Down

The last few chapters have all created meta builders - they aren't really builders in that they don't create an entirely new map, they modify the results of another algorithm. The really interesting thing here is that you can keep chaining them together to achieve the results you want. For example, lets make a map by starting with a Cellular Automata map, feeding it through Wave Function Collapse, possibly adding a castle wall, and then searching for vaults!

The syntax for this is currently quite ugly (that will be a future chapter topic). In map_builders/mod.rs:


#![allow(unused)]
fn main() {
Box::new(
    PrefabBuilder::vaults(
        new_depth,
        Box::new(PrefabBuilder::sectional(
            new_depth,
            prefab_builder::prefab_sections::UNDERGROUND_FORT,
            Box::new(WaveformCollapseBuilder::derived_map(
                new_depth, 
                Box::new(CellularAutomataBuilder::new(new_depth))
            ))
        ))
    )
)
}

Also in map_builders/prefab_builder/mod.rs make sure that you are publicly sharing the map modules:


#![allow(unused)]
fn main() {
pub mod prefab_levels;
pub mod prefab_sections;
pub mod prefab_rooms;
}

If you cargo run this, you get to watch it cycle through the layered building:

Screenshot.

Restoring Randomness

Now that we've completed a two-chapter marathon of prefabricated, layered map building - it's time to restore the random_builder function to provide randomness once more. Here's the new function from map_builders/mod.rs:


#![allow(unused)]
fn main() {
pub fn random_builder(new_depth: i32) -> Box<dyn MapBuilder> {
    let mut rng = rltk::RandomNumberGenerator::new();
    let builder = rng.roll_dice(1, 17);
    let mut result : Box<dyn MapBuilder>;
    match builder {
        1 => { result = Box::new(BspDungeonBuilder::new(new_depth)); }
        2 => { result = Box::new(BspInteriorBuilder::new(new_depth)); }
        3 => { result = Box::new(CellularAutomataBuilder::new(new_depth)); }
        4 => { result = Box::new(DrunkardsWalkBuilder::open_area(new_depth)); }
        5 => { result = Box::new(DrunkardsWalkBuilder::open_halls(new_depth)); }
        6 => { result = Box::new(DrunkardsWalkBuilder::winding_passages(new_depth)); }
        7 => { result = Box::new(DrunkardsWalkBuilder::fat_passages(new_depth)); }
        8 => { result = Box::new(DrunkardsWalkBuilder::fearful_symmetry(new_depth)); }
        9 => { result = Box::new(MazeBuilder::new(new_depth)); }
        10 => { result = Box::new(DLABuilder::walk_inwards(new_depth)); }
        11 => { result = Box::new(DLABuilder::walk_outwards(new_depth)); }
        12 => { result = Box::new(DLABuilder::central_attractor(new_depth)); }
        13 => { result = Box::new(DLABuilder::insectoid(new_depth)); }
        14 => { result = Box::new(VoronoiCellBuilder::pythagoras(new_depth)); }
        15 => { result = Box::new(VoronoiCellBuilder::manhattan(new_depth)); }
        16 => { result = Box::new(PrefabBuilder::constant(new_depth, prefab_builder::prefab_levels::WFC_POPULATED)) },
        _ => { result = Box::new(SimpleMapBuilder::new(new_depth)); }
    }

    if rng.roll_dice(1, 3)==1 {
        result = Box::new(WaveformCollapseBuilder::derived_map(new_depth, result));
    }

    if rng.roll_dice(1, 20)==1 {
        result = Box::new(PrefabBuilder::sectional(new_depth, prefab_builder::prefab_sections::UNDERGROUND_FORT ,result));
    }

    result = Box::new(PrefabBuilder::vaults(new_depth, result));

    result
}
}

We're taking full advantage of the composability of our layers system now! Our random builder now:

  1. In the first layer, we roll 1d17 and pick a map type; we've included our pre-made level as one of the options.
  2. Next, we roll 1d3 - and on a 1, we run the WaveformCollapse algorithm on that builder.
  3. We roll 1d20, and on a 1 - we apply a PrefabBuilder sectional, and add our fortress. That way, you'll only occasionally run into it.
  4. We run whatever builder we came up with against our PrefabBuilder's Room Vault system (the focus of this chapter!), to add premade rooms to the mix.

Wrap-Up

In this chapter, we've gained the ability to prefabricate rooms and include them if they fit into our level design. We've also explored the ability to add algorithms together, giving even more layers of randomness.

...

The source code for this chapter may be found here

Run this chapter's example with web assembly, in your browser (WebGL2 required)

Copyright (C) 2019, Herbert Wolverson.