Store per-region biome list in IndexSet

Index into the biome list instead of duplicating the biome data for each
coordinate. This reduces the size of the processed data and speeds up
encoding/decoding.

Cargo.lock

@@ -220,6 +220,12 @@ dependencies = [
  "syn",
 ]
 
+[[package]]
+name = "equivalent"
+version = "1.0.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5443807d6dff69373d433ab9ef5378ad8df50ca6298caf15de6e52e24aaf54d5"
+
 [[package]]
 name = "errno"
 version = "0.3.2"
@@ -279,6 +285,12 @@ version = "0.24.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "42218cb640844e3872cc3c153dc975229e080a6c4733b34709ef445610550226"
 
+[[package]]
+name = "hashbrown"
+version = "0.14.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2c6201b9ff9fd90a5a3bac2e56a830d0caa509576f0e503818ee82c181b3437a"
+
 [[package]]
 name = "heck"
 version = "0.4.1"
@@ -305,6 +317,17 @@ dependencies = [
  "png",
 ]
 
+[[package]]
+name = "indexmap"
+version = "2.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d5477fe2230a79769d8dc68e0eabf5437907c0457a5614a9e8dddb67f65eb65d"
+dependencies = [
+ "equivalent",
+ "hashbrown",
+ "serde",
+]
+
 [[package]]
 name = "is-terminal"
 version = "0.4.9"
@@ -375,6 +398,7 @@ dependencies = [
  "flate2",
  "glam",
  "image",
+ "indexmap",
  "itertools",
  "num-integer",
  "serde",

Cargo.toml

@@ -17,6 +17,7 @@ fastnbt = "2.3.2"
 flate2 = "1.0.25"
 glam = "0.24.0"
 image = { version = "0.24.5", default-features = false, features = ["png"] }
+indexmap = { version = "2.0.0", features = ["serde"] }
 itertools = "0.11.0"
 num-integer = "0.1.45"
 serde = "1.0.152"

src/bin/minedmap/common.rs

@@ -4,9 +4,10 @@ use std::{
 	path::{Path, PathBuf},
 };
 
+use indexmap::IndexSet;
 use serde::{Deserialize, Serialize};
 
-use minedmap::{io::fs::FileMetaVersion, types::*, world::layer};
+use minedmap::{io::fs::FileMetaVersion, resource::Biome, types::*, world::layer};
 
 // Increase to force regeneration of all output files
 pub const FILE_META_VERSION: FileMetaVersion = FileMetaVersion(0);
@@ -43,7 +44,12 @@ pub struct ProcessedChunk {
 	pub biomes: Box<layer::BiomeArray>,
 	pub depths: Box<layer::DepthArray>,
 }
-pub type ProcessedRegion = ChunkArray<Option<ProcessedChunk>>;
+
+#[derive(Debug, Clone, Default, Serialize, Deserialize)]
+pub struct ProcessedRegion {
+	pub biome_list: IndexSet<Biome>,
+	pub chunks: ChunkArray<Option<ProcessedChunk>>,
+}
 
 pub struct Config {
 	pub region_dir: PathBuf,

src/bin/minedmap/region_processor.rs

@@ -2,9 +2,10 @@ use std::{path::Path, time::SystemTime};
 
 use anyhow::{Context, Result};
 
+use indexmap::IndexSet;
 use minedmap::{
 	io::{fs, storage},
-	resource,
+	resource::{self, Biome},
 	types::*,
 	world::{
 		self,
@@ -45,9 +46,13 @@ impl<'a> RegionProcessor<'a> {
 	}
 
 	/// Processes a single chunk
-	fn process_chunk(&self, data: world::de::Chunk) -> Result<Option<LayerData>> {
+	fn process_chunk(
+		&self,
+		biome_list: &mut IndexSet<Biome>,
+		data: world::de::Chunk,
+	) -> Result<Option<LayerData>> {
 		let chunk = world::chunk::Chunk::new(&data, &self.block_types, &self.biome_types)?;
-		world::layer::top_layer(&chunk)
+		world::layer::top_layer(biome_list, &chunk)
 	}
 
 	fn render_chunk_lightmap(
@@ -110,12 +115,12 @@ impl<'a> RegionProcessor<'a> {
 		minedmap::io::region::from_file(path)?.foreach_chunk(
 			|chunk_coords, data: world::de::Chunk| {
 				let Some(layer::LayerData{ blocks, biomes, block_light, depths }) = self
-					.process_chunk(data)
+					.process_chunk(&mut processed_region.biome_list, data)
 					.with_context(|| format!("Failed to process chunk {:?}", chunk_coords))?
 				else {
 					return Ok(());
 				};
-				processed_region[chunk_coords] = Some(ProcessedChunk {
+				processed_region.chunks[chunk_coords] = Some(ProcessedChunk {
 					blocks,
 					biomes,
 					depths,

src/bin/minedmap/tile_renderer.rs

@@ -56,7 +56,8 @@ fn biome_at(
 		z: block_z,
 	};
 	let region = region_group.get(region_x, region_z)?;
-	region[chunk].as_ref()?.biomes[block].as_ref()
+	let index = region.chunks[chunk].as_ref()?.biomes[block]?.get() - 1;
+	region.biome_list.get_index(index.into())
 }
 
 pub struct TileRenderer<'a> {
@@ -144,7 +145,7 @@ impl<'a> TileRenderer<'a> {
 	}
 
 	fn render_region(image: &mut image::RgbaImage, region_group: &RegionGroup<ProcessedRegion>) {
-		for (coords, chunk) in region_group.center().iter() {
+		for (coords, chunk) in region_group.center().chunks.iter() {
 			let Some(chunk) = chunk else {
 				continue;
 			};

src/resource/biomes.rs

@@ -2,13 +2,13 @@ use serde::{Deserialize, Serialize};
 
 use super::Color;
 
-#[derive(Debug, Clone, Copy, Serialize, Deserialize)]
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, Serialize, Deserialize)]
 pub enum BiomeGrassColorModifier {
 	DarkForest,
 	Swamp,
 }
 
-#[derive(Debug, Clone, Copy, Serialize, Deserialize)]
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, Serialize, Deserialize)]
 pub struct Biome {
 	pub temp: i8,
 	pub downfall: i8,

src/resource/mod.rs

@@ -35,7 +35,7 @@ where
 	BitFlags::<BlockFlag>::from_bits(bits).map_err(de::Error::custom)
 }
 
-#[derive(Debug, Clone, Copy, Serialize, Deserialize)]
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, Serialize, Deserialize)]
 pub struct Color(pub [u8; 3]);
 
 #[derive(Debug, Clone, Copy, Serialize, Deserialize)]

src/world/layer.rs

@@ -1,4 +1,7 @@
+use std::num::NonZeroU16;
+
 use anyhow::{Context, Result};
+use indexmap::IndexSet;
 use serde::{Deserialize, Serialize};
 
 use super::chunk::{Chunk, SectionIterItem};
@@ -26,13 +29,13 @@ impl BlockHeight {
 }
 
 pub type BlockArray = LayerBlockArray<Option<BlockType>>;
-pub type BiomeArray = LayerBlockArray<Option<Biome>>;
+pub type BiomeArray = LayerBlockArray<Option<NonZeroU16>>;
 pub type BlockLightArray = LayerBlockArray<u8>;
 pub type DepthArray = LayerBlockArray<Option<BlockHeight>>;
 
 struct LayerEntry<'a> {
 	block: &'a mut Option<BlockType>,
-	biome: &'a mut Option<Biome>,
+	biome: &'a mut Option<NonZeroU16>,
 	block_light: &'a mut u8,
 	depth: &'a mut Option<BlockHeight>,
 }
@@ -46,7 +49,12 @@ impl<'a> LayerEntry<'a> {
 		self.depth.is_some()
 	}
 
-	fn fill(&mut self, section: SectionIterItem, coords: SectionBlockCoords) -> Result<bool> {
+	fn fill(
+		&mut self,
+		biome_list: &mut IndexSet<Biome>,
+		section: SectionIterItem,
+		coords: SectionBlockCoords,
+	) -> Result<bool> {
 		let Some(block_type) = section.section.block_at(coords)?
 			.filter(|block_type| block_type.is(BlockFlag::Opaque))
 		else {
@@ -59,7 +67,14 @@ impl<'a> LayerEntry<'a> {
 
 		if self.is_empty() {
 			*self.block = Some(block_type);
-			*self.biome = section.biomes.biome_at(section.y, coords)?.copied();
+			if let Some(biome) = section.biomes.biome_at(section.y, coords)? {
+				let (biome_index, _) = biome_list.insert_full(*biome);
+				*self.biome = NonZeroU16::new(
+					(biome_index + 1)
+						.try_into()
+						.expect("biome index not in range"),
+				);
+			}
 		}
 
 		if block_type.is(BlockFlag::Water) {
@@ -99,7 +114,7 @@ impl LayerData {
 /// determined as the block that should be visible on the rendered
 /// map. For water blocks, the height of the first non-water block
 /// is additionally filled in as the water depth.
-pub fn top_layer(chunk: &Chunk) -> Result<Option<LayerData>> {
+pub fn top_layer(biome_list: &mut IndexSet<Biome>, chunk: &Chunk) -> Result<Option<LayerData>> {
 	use BLOCKS_PER_CHUNK as N;
 
 	if chunk.is_empty() {
@@ -121,7 +136,7 @@ pub fn top_layer(chunk: &Chunk) -> Result<Option<LayerData>> {
 					}
 
 					let coords = SectionBlockCoords { xz, y };
-					if !entry.fill(section, coords)? {
+					if !entry.fill(biome_list, section, coords)? {
 						continue;
 					}