Auto CRUD

zero can scaffold REST handlers for a struct in one line, using the AddRESTHandlers fn.

See examples/zero-autocrud for a runnable example.

Generate the handlers

const std = @import("std");
const zero = @import("zero");

const App = zero.App;
const Context = zero.Context;
const utils = zero.utils;

const User = struct { id: i64, name: []const u8, email: []const u8 };

pub fn main(init: std.process.Init) !void {
    var gpa: std.heap.DebugAllocator(.{}) = .init;
    const allocator = gpa.allocator();

    const app = try App.new(allocator, init.io, init.environ_map);

    // One line wires up list / get / create / update / delete for `User`.
    try app.addRestHandlers(User, .{ .resource = "users" });

    try app.run();
}

This registers:

GET    /users        # list   (LIMIT 100)
GET    /users/:id    # get one
POST   /users        # create (body -> struct, 201)
PUT    /users/:id    # update (re-selects and returns the row)
DELETE /users/:id    # delete (204-style {deleted: n})

The generated SQL is emitted for Postgres ($N placeholders), SQLite (?), and DuckDB and dispatched at runtime on ctx.SQL.dialect, so the same struct works against any of those backends (selected via DB_DIALECT / DUCKDB_PATH).

Rules

• resource is the URL segment. table defaults to resource (override via opts.table).

• The primary key is auto-detected as the field named id; override with opts.id_field. The struct must have that field or it fails to compile.

• Struct field names map to column names exactly (the pgz mapper is reused), so name your columns to match. POST/PUT bind the request JSON into the struct.

• The primary key is taken from the request body on create (supply it) and from the :id path param on get/update/delete.

Auto CRUD does not create the table — run your migration (or ctx.SQL.exec) first, as the examples/zero-autocrud demo does with a /init handler.