Pivot

Pivot is a library used to access, query, and aggregate data across a variety of database systems, written in Golang.

Where: Packages and Usage

Pivot is organized into multiple sub-packages that perform various functions:

Package	What it does...
`pivot`	Entry point for the package. Connect to a database from here.
`pivot/dal`	Data Abstraction Layer; provides a database-agnostic view of collections (i.e: tables), records (i.e.: rows), and the fields that make up those records (i.e.: columns).
`pivot/filter`	A database-agnostic representation of queries that return some subset of the data from a collection.
`pivot/mapper`	A more user-friendly data mapping layer that provides more traditional ODM/ORM semantics for working with collections.
`pivot/backends`	Where all the database and search index adapters live.
`pivot/utils`	Utilities largely used within the rest of this library.

Supported Backends

Below is a table describing which data systems are currently supported. For systems with Backend support, Pivot can create and delete collections, and create/retrieve/update/delete records by their primary key / ID. If a system has Indexer support, Pivot can perform arbitrary queries against collections using a standard filter syntax, and return the results as a set of standard Record objects.

Product	Backend	Indexer
MySQL / MariaDB	X	X
PostgreSQL	X	X
SQLite 3.x	X	X
Filesystem	X	X
MongoDB	X	X
Amazon DynamoDB	X
Elasticsearch		X

How: Examples

Example 1: Basic CRUD operations using the `mapper.Mapper` interface

Here is a simple example for connecting to a SQLite database, creating a table; and inserting, retrieving, and deleting a record.

package main

import (
    "fmt"
    "time"

    "github.com/ghetzel/pivot"
    "github.com/ghetzel/pivot/dal"
    "github.com/ghetzel/pivot/mapper"
)

var Widgets mapper.Mapper

var WidgetsSchema = &dal.Collection{
    Name:                   `widgets`,
    IdentityFieldType:      dal.StringType,
    IdentityFieldFormatter: dal.GenerateUUID,
    Fields: []dal.Field{
        {
            Name:        `type`,
            Description: `The type of widget.`,
            Type:        dal.StringType,
            Validator:   dal.ValidateIsOneOf(`foo`, `bar`, `baz`),
            Required:    true,
        }, {
            Name:        `usage`,
            Description: `Short description on how to use this widget.`,
            Type:        dal.StringType,
        }, {
            Name:        `created_at`,
            Description: `When the widget was created.`,
            Type:        dal.TimeType,
            Formatter:   dal.CurrentTimeIfUnset,
        }, {
            Name:        `updated_at`,
            Description: `Last time the widget was updated.`,
            Type:        dal.TimeType,
            Formatter:   dal.CurrentTime,
        },
    },
}

type Widget struct {
    ID        string    `pivot:"id,identity"`
    Type      string    `pivot:"type"`
    Usage     string    `pivot:"usage"`
    CreatedAt time.Time `pivot:"created_at"`
    UpdatedAt time.Time `pivot:"updated_at"`
}

func main() {
    // setup a new backend instance based on the supplied connection string
    if backend, err := pivot.NewDatabase(`sqlite:///./test.db`); err == nil {

        // initialize the backend (connect to/open it)
        if err := backend.Initialize(); err == nil {

            // register models to this database backend
            Widgets = mapper.NewModel(backend, WidgetsSchema)

            // create the model tables if they don't exist
            if err := Widgets.Migrate(); err != nil {
                fmt.Printf("failed to create widget table: %v\n", err)
                return
            }

            // make a new Widget instance, containing the data we want to see
            // the ID field will be populated after creation with the auto-
            // generated UUID.
            newWidget := Widget{
                Type:  `foo`,
                Usage: `A fooable widget.`,
            }

            // insert a widget (ID will be auto-generated because of dal.GenerateUUID)
            if err := Widgets.Create(&newWidget); err != nil {
                fmt.Printf("failed to insert widget: %v\n", err)
                return
            }

            // retrieve the widget using the ID we just got back
            var gotWidget Widget

            if err := Widgets.Get(newWidget.ID, &gotWidget); err != nil {
                fmt.Printf("failed to retrieve widget: %v\n", err)
                return
            }

            fmt.Printf("Got Widget: %#+v", gotWidget)

            // delete the widget
            if err := Widgets.Delete(newWidget.ID); err != nil {
                fmt.Printf("failed to delete widget: %v\n", err)
                return
            }
        } else {
            fmt.Printf("failed to initialize backend: %v\n", err)
            return
        }
    } else {
        fmt.Printf("failed to create backend: %v\n", err)
        return
    }

}

Why: Why Use This?

The ability to mix and match persistent structured data storage and retrieval mechanisms with various indexing strategies is a powerful one. The idea here is to provide a common interface for systems to integrate with in a way that doesn't tightly couple those systems to specific databases, query languages, and infrastructures. It's an attempt to deliver on the promises of traditional ORM/ODM libraries in a platform- and language-agnostic way.