jinqu-odata - Javascript-Linq to Odata

Jinqu-odata lets you write LINQ queries against an odata source. For those who don't know LINQ, the benefits are:

• A unified query language, whether querying local arrays, odata sources, or any other remote data source
• Static typing where Typescript can verify your query is sound

jinqu-odata is dependent on the jinqu package.

Installation

1npm install jinqu-odata

Usage

First, we need classes that map to our odata resources. For example:

1@oDataResource('Books')
2export class Book {
3    Id: number
4    Title: string
5}

We can now query filtered books as follows:

1const service = new ODataService ("https://www.solenya.org/odata")
2
3const books = await service
4    .createQuery(Book)
5    .where(b => b.Price > 60)
6    .toArrayAsync()
7
8for (var b of books)
9    console.log (b)

You can play with the live sample here

The query is translated to the following odata url:

1https://www.solenya.org/odata/Books?$filter=Price gt 60

Inheriting from ODataService

A common pattern is to inherit from ODataService to provide stubs for your odata resources as follows:

1export class CompanyService extends ODataService {
2
3    constructor (provider?: IAjaxProvider) {
4        super('odata')
5    }
6
7    companies() {
8        return this.createQuery(Company)
9    }
10}

Code Generation

Currently we don't have code generators for jinqu-odata. However, we're actively considering this feature and it's tracked by this github issue:

https://github.com/jin-qu/jinqu-odata/issues/5

LINQ to OData Translation

jinqu-odata translates LINQ queries to OData Version 4 query strings. In the quries that follow, translations are shown as comments. You can check the unit tests for more thorough coverage of the translations.

Where

To filter results we use the where operator:

1const result = await query
2    .where(c => c.name.startsWith('Net'))
3    .toArrayAsync()
4
5// odata/Companies?$filter=startsWith(name, "Net")

Supported Operators

Name	TypeScript/JavaScript	OData
Equals	==, ===	eq
Not Equals	!=, !==	ne
Greater Than	>	gt
Greater Than or Equal	>=	ge
Less Than	<	lt
Less Than or Equal	<=	le
Logical And	&&	and
Logical Or	||	or
Logical Not	!	not
Arithmetic Add	+	add
Arithmetic Subtraction	-	sub
Arithmetic Multiplication	*	mul
Arithmetic Division	/	div
Arithmetic Modulo	%	mod
Arithmetic Negation	-	-

Supported Inline Functions

TypeScript/JavaScript	OData
includes	substringof
endsWith	endswith
startsWith	startswith
length	length
indexOf	indexof
replace	replace
substring	substring
toLowerCase	tolower
toUpperCase	toupper
trim	trim
concat	concat
getMonth	month
getDate	day
getHours	hour
getMinutes	minute
getSeconds	second
Math.round	round
Math.floor	floor
Math.ceiling	ceiling

Select

The select operator lets us select only a subset of the fields of a type. It can only occur as the last operator in a query, so must be awaited:

1const result = await query.select("name")
2
3// $select=name

OrderBy

The orderBy operator, optionally followed by some thenBy operators, specifies result order:

1const result = await query
2    .orderBy(c => c.category)
3    .thenByDescending(c => c.created).toArrayAsync()
4
5// $orderby=category,created desc

Count

To get the count of a resource:

1const count = await query.count()
2
3// Companies/$count will be executed

Skip and Take

We can skip a number of items, or limit the number of items, by calling skip and take. Here we query for the 3rd page in a result, by skipping the first 20 results, and then returning the top 10 of the remaining results:

1const result = await query.skip(20).take(10).toArrayAsync()
2
3// $skip=20&$top=10

InlineCount

We can use the inlineCount operator to include the inlineCount property on the results. This will cause query to wrap result.

1const result = await query.inlineCount().toArrayAsync()
2const value = result.value
3const inlineCount = result.inlineCount // only populated if inlineCount operator was called

This is useful in the preceding skip/take scenario, where to implement paging, we'd like the result to include a total non-paged count, without having to write a separate query. Just add the inlineCount operator before calling skip/take.

Expand

jinqu-odata supports expand, which enables you to pull in related entities. In this example, we don't merely want to return books; we also want to return the press associated with each book. We can do this as follows:

1const companies = await service
2    .createQuery(Book)
3    .expand("Press")
4    .toArrayAsync()
5
6// books$expand=Press

Nested Expand

Sometimes we want to drill down more than one level. In this example, our odata source has books, where we want to return all the authors for some books. However, since books can have multiple authors, there's a join table between Authors and Books. Our model will mirror the odata metadata as follows:

1@oDataResource('Books')
2export class Book {
3    Title: string
4    @Type(() => AuthorBook) AuthorBooks: AuthorBook[]
5}
6
7export class AuthorBook {
8    @Type(() => Author) Author: Author
9}
10
11export class Author {
12    Name: string
13}

To query, we first expand the AuthorBooks property, and thenExpand the Book property, as follows:

1const books = await service
2    .createQuery(Book)
3    .expand("AuthorBooks")
4    .thenExpand("Author")
5    .toArrayAsync()
6
7// books?$expand=AuthorBooks($expand=Author)

Filtering Expand by Rows and Columns

For efficiency, we can filter by rows an expand/thenExpand query by providing a predicate:

.thenExpand("Author") // no filter
.thenExpand("Author", a => a.endsWith("Albahari")) // filtered

// books?$expand=AuthorBooks($expand=Author($filter=endswith(Name,'Albahari')))

Similarly, for efficiency, we can filter by columns an expand/thenExpand query by providing an array of column names:

.thenExpand("Author") // no filter
.thenExpand("Author", ["Name"]) // filtered columns

// books?$expand=AuthorBooks($expand=Author($select=Name))

Deserialization

The @Type decorators belong to the class-transformer library that handles deserialization. We need those annotations since the typescript types aren't actually available at runtime. The class-transformer library imposes the small design restriction on us that any constructor arguments to our classes are optional.

GroupBy

groupBy lets you group results by a particular property. Like select, it can only be used as the last operator in a query, and must therefore be awaited:

1// we group resources by "deleted" field
2// and select the count of each group with "deleted" field
3const promise = await query.groupBy(
4    c => ({ deleted: c.deleted }),
5    g => ({ deleted: g.deleted, count: g.count() })
6)
7
8// $apply=groupby((deleted),aggregate(deleted,$count as count))

As you can see in the translation, jinqu-odata supports groupBy with the $apply convention.

Old Browsers

jinqu-odata uses jinqu as a querying platform, if you want to use jinqu features with old browsers, please refer to jinqu documentation.

License

jinqu-odata is licensed under the MIT License.