Skip to the content.

Wrapd is a SQL amplifier that makes SQL easier to use and more powerful in Java.

Rather than hiding SQL behind complexity or difficulty, it exposes SQL where it helps, hides SQL where it should, and makes SQL fast, safe and easy to integrate into Java while staying light, lean, and loveable.

See https://github.com/DaveVoorhis/Wrapd-demo for a Wrapd-ly written demo application.

“Wrapd” is pronounced “wrapped”, “rapid” and “rapt.”

Wrapd is for Java 11 and above.

The project integration tests explicitly verify that Wrapd works with MySQL, MS SQL Server, PostgreSQL and SQLite, but it should work with almost any JDBC-compliant SQL DBMS.

Table of contents


What is Wrapd?

An annotation-free, pure Java library that:

  1. Lets you define SQL queries in your Java code or YAML files, tests them, and generates Java code to invoke your queries using conventional, Streams-compatible, SQL-injection-avoiding, high performance, statically-type-checked method invocations in your Java applications.
  2. Makes it easy to manage and migrate schema changes from your Java applications.

It automatically turns simple, straightforward, readable, tested SQL query definitions you write like this in YAML…

# Define a query called 'JoinABCXYZWhere'
defineQuery:
    JoinABCXYZWhere:
        - SELECT *
            FROM ABC, XYZ
            WHERE x = a AND x > {lower} AND x < {higher}
        - [2, 5]

…or like this in Java…

// Define a query called 'JoinABCXYZWhere'.
defineQuery("JoinABCXYZWhere", 
  "SELECT * FROM ABC, XYZ WHERE x = a AND x > {lower} AND x < {higher}", 2, 5);

…into re-usable, testable, type-safe, SQL-injection-free, Streams-compatible Java methods you can invoke like this…

// Invoke 'JoinABCXYZWhere' query via joinABCXYZWhere(...) method.
joinABCXYZWhere(1002, 1008)
        .forEach(row -> System.out.println("Row:" + 
          " a = " + row.a + " b = " + row.b + " c = " + row.c +
          " x = " + row.x + " y = " + row.y + " z = " + row.z));

…whilst keeping your code clean and the database schema in sync.


Why Wrapd?

Features and Benefits

  1. SQL queries are defined in Java and/or YAML, unit tested, and invoked with conventional type-safe Java methods. Database access is just Java, defined with straightforward SQL; no need for complex ORMs or frameworks.
  2. Code generation is part of the Wrapd library and integrates into your build pipeline. No external tools are required.
  3. Database access is kept in database abstraction layers to increase cohesion, reduce coupling, and promote separation of concerns.
  4. SQL text is exposed in query definitions but hidden in query invocations, making them safe, secure, and SQL-injection-free.
  5. SELECT queries generate Streams-friendly results. Java integration is easy and straightforward and practically eliminates the classic object-relational impedance mismatch.
  6. Inserting and updating table rows is easy.
  7. If needed, table names can be automatically prefixed to avoid table name collisions in shared databases.
  8. Easy, automatic schema migration keeps your code in sync with your database schema and vice versa.

Wrapd is SQL amplified, not obscured!

ORM or SQL Amplifier?

Need help deciding whether Wrapd is right for you?

See ORM or SQL Amplifier?

Details

SQL queries are defined in Java and/or YAML, unit tested, and invoked with conventional type-safe Java methods.

Each SQL query definition in Wrapd acts as:

Predefine SQL queries like this, which tests the queries and automatically generates a database access layer:

...

public class Definitions extends Definer {

    public Definitions(Database database, String codeDirectory, String packageSpec) {
        super(database, codeDirectory, packageSpec);
    }

    void generate() throws Throwable {
        purgeTarget();

        defineTable("$$ABC");
        defineQuery("JoinABCXYZWhere", 
          "SELECT * FROM $$ABC, $$XYZ WHERE x = a AND x > {lower} AND x < {higher}", 2, 5);
        defineValueOf("ValueOfXYZz", 
          "SELECT z FROM $$XYZ WHERE x = {xValue}", 33);

        emitDatabaseAbstractionLayer("DatabaseAbstractionLayer");
    }

    public static void main(String[] args) throws Throwable {
        ...
        var sqlDefinitions = new Definitions(database, codeDirectory, codePackage);
        sqlDefinitions.generate();
        System.out.println("OK: Queries are ready.");
    }
}

Note the example arguments, which are used to test parametric queries and determine their parameter and result types.

Note the ‘$$’ prefixes on table names, which are used to automatically prefix table names with a specified string. This is an optional feature used to avoid table name collisions in environments where multiple disjoint applications need to share a database.

Wrapd will generate classes and type-checked methods to conveniently invoke your queries. Run the queries defined above like this:

package org.reldb.wrapd.demo;

import org.reldb.wrapd.demo.generated.*;
import org.reldb.wrapd.demo.mysql.GetDatabase;

import java.sql.SQLException;

public class Application {

    private static class Demo extends DatabaseAbstractionLayer {

        public Demo() throws Exception {
            super(GetDatabase.getDatabase());
        }

        void run() throws Exception {
            System.out.println("== ABC ==");
            aBC().forEach(row -> System.out.println("Row:" + 
                    " a = " + row.a + " b = " + row.b + " c = " + row.c));
                    
            System.out.println("== JoinABCXYZWhere (1002, 1008) ==");
            joinABCXYZWhere(1002, 1008)
                    .forEach(row -> System.out.println("Row:" +
                            " a = " + row.a + " b = " + row.b + " c = " + row.c +
                            " x = " + row.x + " y = " + row.y + " z = " + row.z));
                            
            System.out.println("== ValueOfXYZz ==");
            System.out.println(valueOfXYZz(1007).orElse("?"));
        }
    }

    public static void main(String[] args) throws Exception {
        new Demo().run();
    }
}

Code generation is part of the Wrapd library and integrates into your build pipeline.

Code generation steps are invoked with conventional Java, and can be easily integrated into your build pipeline without external tools or special plugins.

In your project Java source, code like this runs the SQL query definitions to test them and generate Java code to invoke them:

 // Generate SQL-invocation methods
 public static void main(String[] args) throws Throwable {
     var database = GetDatabase.getDatabase();
     var codeDirectory = "../application/src/main/java";
     var codePackage = "org.reldb.wrapd.demo.generated";
     var sqlDefinitions = new Definitions(database, codeDirectory, codePackage);
     sqlDefinitions.generate();
     System.out.println("OK: Queries are ready.");
 }

In your project build.gradle (assuming Gradle build pipeline), this runs the above:

task runQueryBuild(type: JavaExec) {
    group = "Wrapd"
    description = "Generate code from query definitions."
    classpath = sourceSets.main.runtimeClasspath
    mainClass = "org.reldb.wrapd.demo.Definitions"
}

That exposes a Gradle task called runQueryBuild. Run it to turn the SQL query definitions into invocable Java methods.

Database access is kept in database abstraction layers.

The emitDatabaseAbstractionLayer method emits a class definition that includes all the previously-defined queries as methods.

...
defineQuery("JoinABCXYZWhere", 
   "SELECT * FROM ABC, XYZ WHERE x = a AND x > {lower} AND x < {higher}", 2, 5);
defineValueOf("ValueOfXYZz", "SELECT z FROM $$XYZ WHERE x = {xValue}", 33);

emitDatabaseAbstractionLayer("DatabaseAbstractionLayer");

Invoke them like this:

 private static class Demo extends DatabaseAbstractionLayer {

     public Demo(Database database) {
         super(database);
     }

     public void run() throws Exception {
         joinABCXYZWhere(1002, 1008)
            .forEach(row -> System.out.println("Row:" + 
                " a = " + row.a + " b = " + row.b + " c = " + row.c +
                " x = " + row.x + " y = " + row.y + " z = " + row.z));
         System.out.println(valueOfXYZz(1007).orElse("?"));
     }
 }

This increases cohesion between related queries whilst reducing coupling between Java application code and SQL text, and promotes separation of concerns.

SQL text is exposed in query definitions but hidden in query invocations.

At run-time, queries are safe from inadvertent modification and safe from SQL injection.

Define SQL queries in Java like this:

defineQuery("JoinABCXYZWhere", 
  "SELECT * FROM ABC, XYZ WHERE x = a AND x > {lower} AND x < {higher}", 2, 5);

…or in YAML like this…

JoinABCXYZWhere:
    - SELECT *
        FROM ABC, XYZ
        WHERE x = a AND x > {lower} AND x < {higher}
    - [2, 5]

What you run is:

joinABCXYZWhere(1002, 1008)
       .forEach(row -> System.out.println("Row:" +
          " a = " + row.a + " b = " + row.b + " c = " + row.c +
          " x = " + row.x + " y = " + row.y + " z = " + row.z));

SQL text is confined to definitions, not exposed in invocations, tested and statically type-safe. Internally, parametric queries are implemented as prepared statements. SQL text is not dynamically generated.

SELECT queries generate Streams-friendly results.

Result columns are referenced as native instance variables.

joinABCXYZWhere(1002, 1008)
       .forEach(row -> System.out.println("Row:" + 
          " a = " + row.a + " b = " + row.b + " c = " + row.c +
          " x = " + row.x + " y = " + row.y + " z = " + row.z));

Note how attributes of the result set are accessed as native Java instance variables in a statically-compiled, type-safe, Streams-compatible manner.

Inserting and updating table rows is easy.

Insert a row:

var row = new XYZTuple(database);
row.x = i;
row.y = i * 2;
row.z = Integer.toString(i * 100);
row.insert();

Update rows:

aBCForUpdate().forEach(row -> {
    row.b += 100;
    try {
        row.update();
    } catch (SQLException e) {
        System.out.println("Row update failed due to: " + e);
    }
});

The latter example updates rows on an iterative, row-by-row basis. You can also write update queries.

If needed, table names can be automatically prefixed.

Avoid table name collisions when multiple applications share a database.

Note how the table names in this example are prefixed with ‘$$’:

defineQuery("JoinABCXYZWhere", 
  "SELECT * FROM $$ABC, $$XYZ WHERE x = a AND x > {lower} AND x < {higher}", 2, 5);

Wherever $$ appears in SQL query text, it will be replaced with a predefined string.

This is an optional feature. If you don’t need it, simply define queries with explicit table names:

defineQuery("JoinABCXYZWhere", 
  "SELECT * FROM ABC, XYZ WHERE x = a AND x > {lower} AND x < {higher}", 2, 5);

Easy automatic schema migration.

Integrate schema migration into your application or deploy a stand-alone schema migrator.

Create a new ‘version 1’ database:

var schema = new SQLSchema(database) {
    protected AbstractSchema.Update[] getUpdates() {
        return new AbstractSchema.Update[] {
            // version 1
            schema -> {
                database.updateAll("CREATE TABLE mytab01 (x INT NOT NULL PRIMARY KEY)");
                return Result.OK;
            }
        };
    }
};
schema.setup();

When schema.setup() is run:

Later, add a migration from version 1 to version 2:

var schema = new SQLSchema(database) {
    protected AbstractSchema.Update[] getUpdates() {
        return new AbstractSchema.Update[] {
            // version 1
            schema -> {
                database.updateAll("CREATE TABLE mytab01 (x INT NOT NULL PRIMARY KEY)");
                return Result.OK;
            },
            // migration to version 2
            schema -> {
                database.updateAll("CREATE TABLE mytab02 (a INT NOT NULL PRIMARY KEY)");
                return Result.OK;
            }
        };
    }
};
schema.setup();

When schema.setup() is run:

The schema version is stored in the database, so schema migrations can be done without any need for external version tracking.

This means you can:

In short, Wrapd makes it easy to deploy database creation and database upgrades.


How to Use Wrapd

Overview

  1. Create your database. Wrapd is an “SQL first” library, so the assumption is that the database exists, either via Wrapd’s schema migration mechanisms or externally defined.

  2. Define your SQL queries in your project. A collection of query definitions specifies a database abstraction layer.

  3. Run the code generator to test the query definitions and generate Java source code to invoke them.

  4. Use the generated methods to run your previously-defined SQL queries with type-checked parameters and no visible SQL. SELECT queries emit Java Streams with native, statically type-checked attributes.

Creating a Wrapd Project

It is recommended to organise your Wrapd project into (at least) three main subprojects:

  1. A schema subproject to create/update the SQL schema.

    If schema migration is managed elsewhere, you can skip this subproject.

  2. A queries subproject to turn SQL query definitions into Java methods in the application subproject.

    Larger projects may wish to generate the Java methods in their own subproject, which the application(s) will reference as a dependency.

  3. An application subproject to host the application.

    Wrapd uses the queries subproject to write the database access layer(s) for you.

You may want a fourth database subproject to manage database connectivity, as it will be shared by the three main subprojects above.

Dividing a Wrapd project into (at least) these three main subprojects makes it possible to:

  1. Reliably convert SQL query definitions into Java methods without being blocked by compilation failures in the application(s).
  2. Avoid deploying the SQL query definitions, as they’re not needed in production.
  3. Deploy schema updates independently of the application(s), if desired.

You can integrate them all into one project without subprojects, but you will almost certainly encounter difficulties with compilation failures as you migrate schemas and change queries, unless you are careful to delete the generated Java code for obsolete query definitions and (possibly) comment out the application code that uses them.

It’s generally much easier to develop and maintain a Wrapd project if you divide it into subprojects as described above.

The Wrapd-demo Project

The Wrapd-demo demonstration application has been built according the above structure. It consists of three main subprojects…

  1. schema - creates/updates the SQL schema.
  2. queries - turns SQL query definitions into Java methods in the application subproject.
  3. application - the application.

…plus two additional subprojects:

  1. database - database connectivity, including a demonstration of connection pooling because most real-world applications will need it. It’s used by all three of the above subprojects.
  2. buildSrc - Gradle project settings.

Perhaps the easiest way to build a new Wrapd application is to simply copy the Wrapd-demo demonstration application and modify it to suit your requirements, or you can follow the step-by-step tutorial below.

The Wrapd Tutorial

See the Wrapd Tutorial for a step-by-step intro to setting up a Wrapd project using Gradle.


Get Wrapd

Maven Central: https://search.maven.org/artifact/org.reldb/Wrapd

GitHub: https://github.com/DaveVoorhis/Wrapd


Know Wrapd

Javadocs:

https://www.javadoc.io/doc/org.reldb/Wrapd

https://www.javadoc.io/doc/org.reldb/Toolbox


Documentation and examples are a work-in-progress. Watch this space!