Query langauges provide a set of features to querying, projecting and retrieving data (usually relational data). But how to introduces these standard, easily-learned patterns for querying data?

In this article we will explore Go LINQ packages that bridges the gap between the world of objects and the world of data.

LINQ

For first time is introduced by Microsoft in their programming language C#. Its purpose is to bridge the gap between query languages such as SQL and programming languages.

Go LINQ is a query package for Go. Essentially it has ability to apply queries to slices and collections using SQL-like methods.

Installation

As any other go package we should execute go get command:

$ go get ahmetalpbalkan.github.io/go-linq

Usage

The package consists two query structs that have a set of functions for querying, projection, grouping, filtering, sorting, aggregation and many more which will explore in detail. Queries are processed synchronously and asynchronously as well. It does not evaluate the data lazily. No deferred execution due to lack of enumeration abstraction. The package works only with slices.

Lets work with a slice of Company struct:

type Company struct {
	Name    string
	Country string
	City    string
}
companies := []company.Company{
	company.Company{Name: "Microsoft", Country: "USA", City: "Redmond"},
	company.Company{Name: "Google", Country: "USA", City: "Palo Alto"},
	company.Company{Name: "Facebook", Country: "USA", City: "Palo Alto"},
	company.Company{Name: "Uber", Country: "USA", City: "San Francisco"},
	company.Company{Name: "Tweeter", Country: "USA", City: "San Francisco"},
	company.Company{Name: "SoundCloud", Country: "Germany", City: "Berlin"},
}

The package itself uses reflection to work any slice or collection of data. It declares linq.T interface that used by most of the package functions. In order to work with a concrete type, it must be casted:

var value linq.T
obj := value.(yourType)
Examples

The example are pretty similar to my previous blog post where I am using Gen. The difference between both libraries is that Gen relies on code generation while LINQ works by using reflection. I haven’t done any performance comparisions to evaluate how slow LINQ is.

Most of the clojure functions receive linq.T object as arguments.

// selects all companies located at USA
allUSCompanies := From(companies).Where(func(c T) (bool, error) {
	return c.(company.Company).Country == "USA", nil
})
// distincts the companies by their country of origin
uniqueCompanies := From(companies).DistinctBy(func(compA T, compB T) (bool, error) {
	return compA.(company.Company).Country == compB.(company.Company).Country, nil
})
// sorts the companies by their name
sortedCompanies := From(companies).OrderBy(func(compA T, compB T) bool {
	return strings.Compare(compA.(company.Company).Name, compB.(company.Company).Name) == -1
})

Grouping a slice is processed by GroupBy function which accepts as argument two selector functions. The first clojure selects the group key, while the second returns the object for that key. The result is again a map[T]T.

groupedCompanies, err := From(companies).GroupBy(func(comp T) T {
	return comp.(company.Company).Country
}, func(c T) T {
	return c
})

if err != nil {
	panic(err)
}

fmt.Println("US Companies: ", groupedCompanies["USA"])
fmt.Println("German Companies: ", groupedCompanies["Germany"])
// projects a slice of companies into a slice of company names
companyNames := From(companies).Select(func(comp T) (T, error) {
	return comp.(company.Company).Name, nil
})

Advanced Samples

The LINQ package provides some advanced features that are very first citizen in the query langauges.

You can intersect two slices by using the following code snippet:

intersectedCompanies := From(companies).Intersect([]company.Company{
	company.Company{Name: "Microsoft",
		Country: "USA",
		City:    "Redmond"},
})

If you want to combine two slice into one, you should use the Union function:

unionCompanies := From(companies).Union([]company.Company{
	company.Company{Name: "Skyp",
		Country: "Latvia",
		City:    "Talin"},
})

Lets check whether the slice has at least one company that is in Germany. For that purpose we can use AnyOf function:

hasGermanCompany, err := From(companies).AnyWith(func(comp T) (bool, error) {
	return strings.Compare(comp.(company.Company).Country, "Germany") == 0, nil
})

If you want to get all companies that are different than Microsoft, you should consider using Except function:

openSourceCompanies := From(companies).Except([]company.Company{
	company.Company{Name: "Microsoft",
		Country: "USA",
		City:    "Redmond"},
})

The package provides us with a join function that correlates the elements of collections based on their equality. The first two clojure functions extract the key for every item in each slice. The third functions extracts the result that we want.

In the example below the inner join is between companies and countries slice based on company.Counutry and country.Name properties. Objects that have the same property value are correlated.

countries := []company.Country{
	company.Country{Name: "USA",
		Wikipedia: "https://en.wikipedia.org/wiki/United_States"},
	company.Country{Name: "Germany",
		Wikipedia: "https://en.wikipedia.org/wiki/Germany"},
}

// The join function produces a slice of struct that has two properties
// Company name and Countr Info
companiesWithCountryInfo := From(companies).Join(countries, func(comp T) T {
	return comp.(company.Company).Country
}, func(cntry T) T {
	return cntry.(company.Country).Name
}, func(outer, inner T) T {
	var result struct {
		Company     string
		CountryInfo string
	}

	result.Company = outer.(company.Company).Name
	result.CountryInfo = inner.(company.Country).Wikipedia
	return result
})

The code snippet above produces a slice of objects that have Company and CountryInfo properties:

{[{Microsoft https://en.wikipedia.org/wiki/United_States}
{Google https://en.wikipedia.org/wiki/United_States}
{Facebook https://en.wikipedia.org/wiki/United_States}
{Uber https://en.wikipedia.org/wiki/United_States}
{Tweeter https://en.wikipedia.org/wiki/United_States}
{SoundCloud https://en.wikipedia.org/wiki/Germany}] <nil>}

If you want to get top 3 companies in the slice you can use Take function:

top3comapnies := From(companies).Take(3)

If you want to get all companies except the first 3 you can use Skip function:

restOfComapnies := From(companies).Skip(3)

These functions are very handy in implementing paging:

pageNumber := 1
pageItemCount := 20

pagedCompanies := From(companies).
	Skip(pageNumber * pageItemCount).
	Take(pageItemCount)

You can read more about the rest of features in the official documentation.

Verdict

LINQ package is not very Go idiomatic due to its reflection. However, it provides us with great set of features which does not require any code generation and can be use out of the box.