Generator API Reference

The generator is built into the nevr package and provides functions for generating Prisma schemas and OpenAPI specifications from your Nevr entity definitions.

Configuration with defineConfig

The nevr.config.ts file is required for the CLI to work:

// nevr.config.ts
import { defineConfig } from "nevr"
import { user, post } from "./entities"
import { auth } from "nevr/plugins/auth"

export default defineConfig({
  database: "postgresql",  // "sqlite" | "postgresql" | "mysql"
  outDir: "./prisma",      // Output directory (default: "./prisma")
  entities: [user, post],
  plugins: [auth()],
  incremental: true,       // Enable caching (default: true)
})

Then use the CLI:

npx nevr generate    # Auto-loads nevr.config.ts
npx nevr db:push     # Push schema to database

Config file locations (auto-discovered in order):

• nevr.config.ts
• nevr.config.js
• nevr.config.mjs
• src/nevr.config.ts
• src/nevr.config.js
• src/nevr.config.mjs
• lib/nevr.config.ts
• lib/nevr.config.js
• lib/nevr.config.mjs

---

Programmatic API

For programmatic use, import directly from nevr/generator:

import { generate, generatePrismaSchema, generateOpenAPI } from "nevr/generator"

Primary Functions

generate

Generate a complete Prisma schema from your entities with full support for incremental caching and schema splitting.

import { generate } from "nevr/generator"
import { entities } from "./entities"

const result = generate(entities, {
  provider: "postgresql",  // sqlite, mysql, postgresql
  outDir: "./prisma",
})

Options:

Option	Type	Default	Description
provider	"sqlite" | "postgresql" | "mysql"	"sqlite"	Database provider
outDir	string	"./prisma"	Output directory
incremental	boolean	true	Enable incremental caching
force	boolean	false	Force full regeneration
splitSchema	boolean	false	Enable schema splitting by namespace

Returns:

{
  files: string[]           // Generated file paths
  schemaPath: string        // Main schema path
  namespaceFiles?: Record<string, string>  // Namespace files (if splitSchema)
  cache?: {                 // Cache info
    wasIncremental: boolean
    changedEntities: string[]
    unchangedEntities: string[]
    skipped: boolean
  }
}

---

generatePrismaSchema

Generate a Prisma schema string directly (useful for programmatic access).

import { generatePrismaSchema } from "nevr/generator"

const schemaString = generatePrismaSchema(entities, {
  provider: "postgresql",
})
// Returns the complete Prisma schema as a string

---

generatePrismaHeader

Generate the header portion of a Prisma schema (generator + datasource blocks).

import { generatePrismaHeader } from "nevr/generator"

const header = generatePrismaHeader({
  provider: "postgresql",
  splitSchema: false,
})
// Returns string with generator and datasource blocks

---

generatePrismaModels

Generate only the model definitions from entities (without header).

import { generatePrismaModels } from "nevr/generator"

const models = generatePrismaModels(entities, allEntities)
// Returns string with all model definitions

---

generateOpenAPI

Generate an OpenAPI 3.0 specification from your entities.

import { generateOpenAPI } from "nevr/generator"

const spec = generateOpenAPI(entities, {
  info: {
    title: "My API",
    version: "1.0.0",
    description: "API generated by Nevr",
  },
  servers: [{ url: "http://localhost:3000" }],
  basePath: "/api",
  security: true,
  outPath: "./docs/openapi.json",  // Optional: write to file
})

See OpenAPI Guide for more details.

---

Incremental Caching

The generator uses an incremental caching system to avoid regenerating unchanged entities.

// Cache is stored in .nevr-cache.json in the output directory
{
  "version": "1.0.0",
  "entities": {
    "user": "abc123...",  // MD5 hash of entity config
    "post": "def456..."
  },
  "lastGenerated": "2024-01-15T10:30:00Z"
}

Disable caching:

generate(entities, { incremental: false })
// or force regeneration
generate(entities, { force: true })

---

Schema Splitting

For large codebases (100+ entities), you can enable schema splitting to organize models by namespace:

// Define entities with namespaces
const user = entity("user", { ... }).namespace("auth")
const post = entity("post", { ... }).namespace("content")

// Generate with split schema
generate(entities, { splitSchema: true })

// Creates:
// prisma/schema/schema.prisma (header + datasource)
// prisma/schema/auth.prisma (auth models)
// prisma/schema/content.prisma (content models)

This uses Prisma's prismaSchemaFolder preview feature.

---

CLI Integration

The generator is typically invoked via the CLI:

npx nevr generate
# or
npx nevr g

See CLI > Code Generation for details.

---

Type Inference

Note: TypeScript types and API clients are no longer generated by the generator. They are automatically inferred from the server through Nevr's E2E type system.

import { createClient } from "nevr/client"

const client = createClient({ baseURL: "http://localhost:3000" })

// Types are automatically synced from server!
const users = await client.users.list()  // TypeScript knows the shape

See Type Inference for details.

---

Next Steps

• CLI Reference
• Type Inference
• Prisma Integration
• OpenAPI Guide