pyrites/haskell-experiments/CLAUDE.md

# CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

## Build Commands

```bash
cabal build          # Build the project
cabal test           # Run all tests
cabal repl           # Start interactive REPL
```

To run a single test file or specific test:
```bash
cabal test --test-option=--match="/DatalogParser/"
```

## Architecture

This is an experimental Datalog implementation in Haskell with the following components:

### Parsing Layer (`src/Datalog/DatalogParser.hs`)
Megaparsec-based parser that converts Datalog text into an AST:
- `Term` - Variables (`Var`), symbols (`Sym`), numbers (`Num`)
- `Literal` - Predicate with arguments, optionally negated
- `Statement` - Fact, Rule (with `Head` and body literals), or Query
- Entry points: `parseDatalog`, `parseDatalogFile`

### Evaluation Engine (`src/Datalog/InMemoryDB.hs`)
In-memory database storing relations and rules:
- `InMemoryDB` - Contains `relations` map and `constants` set (Herbrand universe)
- `Relation` - Name, arity, tuples (facts), and associated rules
- `RelationRule` - Head variables and body constraints
- Key functions: `withFacts`, `withFactsAndRules` for building databases
- Query evaluation is not yet implemented

### Data Flow
1. Text parsed into `Statement` (Fact/Rule/Query)
2. Facts populate relations with tuples; rules encode derivation logic via `RuleContext` and `BodyConstraint`
3. Variables are indexed during rule processing for efficient constraint matching

### Other Modules
- `SimpleParser` / `ArithmeticParser` - Experimental arithmetic expression parsers
- `Ologs` - Categorical data structures (dots and arcs representing objects and morphisms)

## Development Environment

Nix support via `shell.nix` provides HLS, cabal-install, and Hoogle. Use `direnv allow` for automatic environment loading.