Skip to main content

@prism-lang/validator

The Prism Validator package provides comprehensive validation, linting, and type-checking capabilities for Prism code. It's designed to help both developers and LLMs write correct, idiomatic Prism programs.

Installation​

npm install @prism-lang/validator
# or
pnpm add @prism-lang/validator

Quick Start​

import { createValidator } from '@prism-lang/validator';

const validator = createValidator();

const code = `
const result = llm("Analyze this") ~> 0.8
uncertain if (result.confidence > 0.7) {
  high { print("High confidence:", result) }
  low { print("Low confidence") }
}
`;

const validation = validator.validateAll(code);
if (validation.valid) {
  console.log("Code is valid!");
} else {
  console.log("Errors:", validation.formattedErrors);
}

Features​

1. Syntax Validation​

  • Complete AST validation
  • Detailed error messages with line/column info
  • Recovery suggestions for common mistakes

2. Type Checking​

  • Runtime type validation
  • Confidence propagation checking
  • Function signature validation

3. Linting​

  • Configurable rules
  • Best practice enforcement
  • Performance suggestions

4. Confidence Analysis​

  • Flow tracking
  • Completeness checking
  • Threshold validation

5. Streaming Support​

  • Validate code as it's generated
  • Provide completions for partial code
  • Ideal for LLM integration

Architecture​

The validator consists of several specialized modules:

@prism-lang/validator
├── PrismValidator      # Core syntax validation
├── TypeChecker         # Runtime type checking
├── ConfidenceChecker   # Confidence flow analysis
├── PrismLinter         # Configurable linting rules
└── StreamingValidator  # Real-time validation

Use Cases​

Development Time​

// Configure strict validation for development
const validator = createValidator({
  rules: {
    'no-unused-variables': true,
    'require-confidence-in-uncertain': true,
    'no-infinite-loops': true
  }
});

LLM Integration​

// Use streaming validation for AI code generation
const validator = createValidator();
validator.resetStreaming();

// Validate chunks as they're generated
const chunk1 = validator.validateStreaming("const x = ");
const chunk2 = validator.validateStreaming("llm(");
const completions = validator.getStreamingCompletions(); // ["llm(\""]

CI/CD Pipeline​

// Run comprehensive checks in CI
const validator = createValidator({ strict: true });
const result = validator.validateAll(sourceCode);

if (!result.valid) {
  console.error("Validation failed:");
  result.errors.forEach(err => {
    console.error(`${err.file}:${err.line}:${err.column} - ${err.message}`);
  });
  process.exit(1);
}

Next Steps​