# <img src="terrain_forge_logo.png" alt="TerrainForge Logo" width="32" height="32" style="vertical-align: middle;"> TerrainForge
<div align="center">
<img src="demo/output/showcase/hires/pipeline_epic_scaled.png" alt="Epic Pipeline Showcase">
<p><em>BSP → Cellular pipeline generating structured caves (240×180, 85 regions, 24 markers)</em></p>
</div>
A modular procedural generation engine for terrain, dungeons, and maps in Rust.
**🎯 Now with Semantic Layers in v0.3.0!**
## Features
- **🎯 Semantic Layers**: Game-agnostic metadata for entity spawning and region analysis
- **14 Generation Algorithms**: BSP, Cellular Automata, DLA, Drunkard Walk, Maze, Rooms, Voronoi, WFC, Percolation, Diamond Square, Fractal, Agent-based, Glass Seam, Room Accretion
- **🔗 Advanced Connectivity**: Region-aware connectors with spanning tree analysis
- **🎨 Enhanced Demo Framework**: Semantic visualization with color-coded markers
- **Noise Generation**: Perlin, Simplex, Value, Worley with FBM, Ridged, and modifiers
- **Effects**: Morphology, spatial analysis, filters, connectivity
- **Composition**: Pipeline chaining and layered generation
- **Prefab System**: Rotatable prefabs with 90°/180°/270° variants
- **Deterministic**: Seeded RNG for reproducible results
- **Generic**: Works with custom cell types via traits
## Quick Start
```rust
use terrain_forge::{Grid, Tile, algorithms, Algorithm, Rng, SemanticExtractor};
fn main() {
let mut grid = Grid::new(80, 60);
let algo = algorithms::get("bsp").unwrap();
algo.generate(&mut grid, 12345);
println!("Generated {} floor tiles", grid.count(|t| t.is_floor()));
}
```
### NEW: Semantic Generation
```rust
use terrain_forge::{SemanticExtractor, Rng};
use terrain_forge::algorithms::Bsp;
fn main() {
let mut grid = Grid::new(80, 60);
let mut rng = Rng::new(12345);
let algo = Bsp::default();
// Generate terrain
algo.generate(&mut grid, 12345);
// Extract semantic layers
let semantic = SemanticExtractor::for_rooms().extract(&grid, &mut rng);
println!("Generated {} markers", semantic.markers.len());
println!("Found {} room centers", semantic.masks.room_centers.len());
// Access entity spawn points
for (x, y, marker) in &semantic.markers {
println!("Marker {:?} at ({}, {})", marker, x, y);
}
}
```
## Installation
```toml
[dependencies]
terrain-forge = "0.3"
```
## Algorithms
| `bsp` | Binary Space Partitioning - structured rooms | ✅ `for_rooms()` |
| `cellular` | Cellular Automata - organic caves | ✅ `for_caves()` |
| `drunkard` | Drunkard's Walk - winding corridors | ✅ `default()` |
| `maze` | Perfect maze generation | ✅ `for_mazes()` |
| `rooms` | Simple rectangular rooms | ✅ `for_rooms()` |
| `voronoi` | Voronoi-based regions | ✅ `default()` |
| `dla` | Diffusion-Limited Aggregation | ✅ `default()` |
| `wfc` | Wave Function Collapse | ✅ `default()` |
| `percolation` | Connected cluster generation | ✅ `default()` |
| `diamond_square` | Heightmap terrain | ✅ `default()` |
| `fractal` | Fractal terrain | ✅ `default()` |
| `agent` | Multi-agent carving | ✅ `default()` |
| `glass_seam` | Region connector | ✅ `default()` |
| `room_accretion` | **NEW**: Brogue-style organic dungeons | ✅ `for_rooms()` |
**Note**: All algorithms support semantic analysis through `SemanticExtractor`. Algorithm-specific extractors (`for_caves()`, `for_rooms()`, `for_mazes()`) provide optimized analysis, while `default()` works with any terrain type.
## Usage
### Registry API
```rust
use terrain_forge::{Grid, algorithms};
let mut grid = Grid::new(80, 60);
// Get by name
let algo = algorithms::get("cellular").unwrap();
algo.generate(&mut grid, 42);
// List all
for name in algorithms::list() {
println!("{}", name);
}
```
### Direct Instantiation
```rust
use terrain_forge::{Grid, Algorithm};
use terrain_forge::algorithms::{Bsp, BspConfig};
let config = BspConfig {
min_room_size: 6,
max_room_size: 15,
min_depth: 3,
max_depth: 5,
};
let mut grid = Grid::new(80, 60);
Bsp::new(config).generate(&mut grid, 12345);
```
### Noise
```rust
use terrain_forge::noise::{Perlin, Fbm};
let noise = Perlin::new(42);
let value = noise.get(10.5, 20.3); // -1.0 to 1.0
let fbm = Fbm::new(noise, 4, 2.0, 0.5);
let layered = fbm.get(10.5, 20.3);
```
### Semantic Layers
Generate game-agnostic metadata for entity spawning and region analysis:
```rust
use terrain_forge::{SemanticExtractor, Rng};
use terrain_forge::algorithms::{Bsp, CellularAutomata, Maze};
// Different algorithms provide different semantic insights
let mut grid = Grid::new(80, 60);
let mut rng = Rng::new(12345);
// Cave system analysis
let cellular = CellularAutomata::default();
cellular.generate(&mut grid, 12345);
let semantic = SemanticExtractor::for_caves().extract(&grid, &mut rng);
// Maze structure analysis
let maze = Maze::default();
maze.generate(&mut grid, 12345);
let semantic = SemanticExtractor::for_mazes().extract(&grid, &mut rng);
// Entity spawning works the same across all algorithms
for (x, y, marker) in &semantic.markers {
match marker.tag.as_str() {
"PlayerStart" => spawn_player(x, y),
"Exit" => place_exit(x, y),
"Treasure" => place_loot(x, y),
_ => {}
}
}
// Algorithm-specific region analysis
Some("Junction") => println!("Maze junction detected"),
Some("Room") => println!("Rectangular room detected"),
_ => {}
}
```
### Constraints
```rust
use terrain_forge::constraints;
let connectivity = constraints::validate_connectivity(&grid);
let density = constraints::validate_density(&grid);
let border_ok = constraints::validate_border(&grid);
```
## Custom Cell Types
```rust
use terrain_forge::{Grid, Cell};
#[derive(Clone, Default)]
struct MyCell {
terrain: u8,
}
impl Cell for MyCell {
fn is_passable(&self) -> bool {
self.terrain != 0
}
}
let grid = Grid::<MyCell>::new(50, 50);
```
## CLI Tool
The demo framework provides visualization and testing:
```bash
cd demo
# Generate single algorithm
cargo run -- gen bsp -s 12345 -o output.png
# NEW: Generate with semantic layers
cargo run -- gen cellular --semantic --text --png -s 12345
cargo run -- gen maze --semantic --text --png -s 12345
cargo run -- gen rooms --semantic --text --png -s 12345
# NEW: Room accretion (Brogue-style)
cargo run -- gen room_accretion --semantic -s 12345
# Pipeline composition
cargo run -- gen "bsp > cellular" -s 42
# Layer composition
# Run config file
cargo run -- run configs/saltglass_overworld.json
# NEW: Semantic configuration files
cargo run -- run configs/semantic_bsp.json
cargo run -- run configs/semantic_large_rooms.json
cargo run -- run configs/semantic_organic.json
# Compare algorithms
cargo run -- compare bsp cellular maze -s 12345
# List available algorithms
cargo run -- list
```
## What's New
### v0.3.0 - Semantic Layers
- **🎯 Semantic Layers**: Game-agnostic metadata system for entity spawning and region analysis
- **🏗️ Room Accretion Algorithm**: Enhanced with semantic support for diverse marker types
- **🎨 Enhanced Demo Framework**: Semantic visualization with color-coded markers and PNG output
- **🔗 Connectivity Analysis**: Advanced region connectivity with spanning tree generation
- **📊 Spatial Analysis**: Automated detection of rooms, corridors, junctions, and chokepoints
### v0.2.0 - Advanced Features
- **Room Accretion Algorithm**: Brogue-style organic dungeon generation with sliding room placement
- **Advanced Connectivity**: `connect_regions_spanning()` with loop control for better dungeon flow
- **Prefab Rotation**: 90°/180°/270° rotation support for prefabs
- **Public Region Analysis**: `label_regions()` for custom connectivity logic
## Documentation
See [docs/API.md](docs/API.md) for full API reference.
See [docs/ROGUELIKE_GENERATION_ANALYSIS.md](docs/ROGUELIKE_GENERATION_ANALYSIS.md) for advanced techniques.
## License
MIT