@sap/cds-lsp

--- # shorty: Definition Language synopsis: > Specification of the definition language used to model data models and services in an easy and user-centric syntax. Includes a reference and overview of all CDS concepts and features with compact examples. #permalink: /cds/cdl/ status: released uacp: Used as link target from Help Portal at https://help.sap.com/products/BTP/65de2977205c403bbc107264b8eccf4b/855e00bd559742a3b8276fbed4af1008.html --- # Conceptual Definition Language (CDL) The *Conceptual Definition Language (CDL)* is a human-readable language for defining CDS models. Sources are commonly provided in files with`.cds` extensions and get compiled into [CSN representations](csn). Following sections provide a reference of all language constructs in CDL, which also serves as a reference of all corresponding CDS concepts and features. [[toc]] ## Language Preliminaries - [Keywords & Identifiers](#keywords-identifiers) - [Built-in Types](#built-in-types) - [Literals](#literals) - [Model Imports](#model-imports) - [Namespaces](#namespaces) - [Comments](#comments) ### Keywords & Identifiers *Keywords* in CDL are used to prelude statements, such as imports and namespace directives as well as entity and type declarations. *Identifiers* are used to refer to definitions. ```cds namespace capire.bookshop; using { managed } from `@sap/cds/common`; aspect entity : managed { key ID: Integer } entity Books : entity { title : String; author : Association to Authors; } entity Authors : entity { name : String; } ``` ::: details Noteworthy... In the example above `entity` shows up as a keyword, as well as an identifier of an aspect declaration and references to that. ::: Keywords are *case-insensitive*, but are most commonly used in lowercase notation. Identifiers are *case-significant*, that is, `Foo` and `foo` would identify different things. Identifiers have to comply to `/[$A-Za-z_]\w*/` or be enclosed in `![`...`]` like that: ```cds type ![Delimited Identifier] : String; ``` ::: warning Avoid using delimited identifiers Delimited identifiers in general, but in particular non-ASCII characters, or keywords as identifiers should be avoided as much as possible, for reasons of interoperability. ::: ### Built-in Types  ### Literals The following literals can be used in CDL (mostly as in JavaScript, Java, and SQL):  ```cds true , false , null // as in all common languages 11 , 2.4 , 1e3, 1.23e-11 // for numbers 'A string''s literal' // for strings `A string\n paragraph` // for strings with escape sequences { foo:'boo', bar:'car' } // for records [ 1, 'two', {three:4} ] // for arrays ``` [Learn more about literals and their representation in CSN.](./csn#literals) {.learn-more} #### Date & Time Literals In addition, type-keyword-prefixed strings can be used for date & time literals:  ```cds date'2016-11-24' time'16:11:32' timestamp'2016-11-24T12:34:56.789Z' ``` #### Multiline String Literals {#multiline-literals} Use string literals enclosed in **single or triple backticks** for multiline strings: ```cds @escaped: `OK Emoji: \u{1f197}` @multiline: ``` This is a CDS multiline string. - The indentation is stripped. - \u{0055}nicode escape sequences are possible, just like common escapes from JavaScript such as \r \t \n and more! ``` @data: ```xml <main> The tag is ignored by the core-compiler but may be used for syntax highlighting, similar to markdown. </main> ``` entity DocumentedEntity { // ... } ``` Within those strings, escape sequences from JavaScript, such as `\t` or `\u0020`, are supported. Line endings are normalized. If you don't want a line ending at that position, end a line with a backslash (`\`). For string literals inside triple backticks, indentation is stripped and tagging is possible. ### Model Imports #### The `using` Directive {#using} Using directives allows to import definitions from other CDS models. As shown in line three below you can specify aliases to be used subsequently. You can import single definitions as well as several ones with a common namespace prefix. Optional: Choose a local alias. ::: code-group ```cds [using-from.cds] using foo.bar.scoped.Bar from './contexts'; using foo.bar.scoped.nested from './contexts'; using foo.bar.scoped.nested as specified from './contexts'; entity Car : Bar {} //> : foo.bar.scoped.Bar entity Moo : nested.Zoo {} //> : foo.bar.scoped.nested.Zoo entity Zoo : specified.Zoo {} //> : foo.bar.scoped.nested.Zoo ``` ::: Multiple named imports through ES6-like deconstructors: ```cds using { Foo as Moo, sub.Bar } from './base-model'; entity Boo : Moo { /*...*/ } entity Car : Bar { /*...*/ } ``` > Also in the deconstructor variant of `using` shown in the previous example, specify fully qualified names. #### Model Resolution Imports in `cds` work very much like [`require` in Node.js](https://nodejs.org/api/modules.html#requireid) and `import`s in [ES6](https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Statements/import). In fact, we reuse **[Node's module loading mechanisms](https://nodejs.org/api/modules.html#modules_all_together)**. Hence, the same rules apply: - Relative path resolution<br> Names starting with `./` or `../` are resolved relative to the current model. - Resolving absolute references<br> Names starting with `/` are resolved absolute to the file system. - Resolving module references<br> Names starting with neither `.` nor `/` such as `@sap/cds/common` are fetched for in `node_modules` folders: - Files having _.cds_, _.csn_, or _.json_ as suffixes, appended in order - Folders, from either the file set in `cds.main` in the folder's _package.json_ or `index.<cds|csn|json>` file. ::: tip To allow for loading from precompiled _.json_ files it's recommended to **omit _.cds_ suffixes** in import statements, as shown in the provided examples. ::: ### Namespaces #### The `namespace` Directive To prefix the names of all subsequent definitions, place a `namespace` directive at the top of a model. This is comparable to other languages, like Java. ::: code-group ```cds[namespace.cds] namespace foo.bar; entity Foo {} //> foo.bar.Foo entity Bar : Foo {} //> foo.bar.Bar ``` ::: A namespace is not an object of its own. There is no corresponding definition in CSN. #### The `context` Directive {#context} Use `contexts` for nested namespace sections. ::: code-group ```cds[contexts.cds] namespace foo.bar; entity Foo {} //> foo.bar.Foo context scoped { entity Bar : Foo {} //> foo.bar.scoped.Bar context nested { entity Zoo {} //> foo.bar.scoped.nested.Zoo } } ``` ::: #### Scoped Definitions {#scoped-names} You can define types and entities with other definitions' names as prefixes: ```cds namespace foo.bar; entity Foo {} //> foo.bar.Foo entity Foo.Bar {} //> foo.bar.Foo.Bar type Foo.Bar.Car {} //> foo.bar.Foo.Bar.Car ``` #### Fully Qualified Names A model ultimately is a collection of definitions with unique, fully qualified names. For example, the second model above would compile to this [CSN](./csn): ::: code-group ```json [contexts.json] {"definitions":{ "foo.bar.Foo": { "kind": "entity" }, "foo.bar.scoped": { "kind": "context" }, "foo.bar.scoped.Bar": { "kind": "entity", "includes": [ "foo.bar.Foo" ] }, "foo.bar.scoped.nested": { "kind": "context" }, "foo.bar.scoped.nested.Zoo": { "kind": "entity" } }} ``` ::: ### Comments CDL supports line-end, block comments, and *doc* comments as in Java and JavaScript: ```cds // line-end comment /* block comment */ /** doc comment */ ``` #### Doc Comments {#doc-comment} A multi-line comment of the form `/** … */` at an [annotation position](#annotation-targets) is considered a *doc comment*: ```cds /** * I am the description for "Employee" */ entity Employees { key ID : Integer; /** * I am the description for "name" */ name : String; } ``` The text of a doc comment is stored in CSN in the property `doc`. When generating OData EDM(X), it appears as value for the annotation `@Core.Description`. When generating output for deployment to SAP HANA, the first paragraph of a doc comment is translated to the HANA `COMMENT` feature for tables, table columns, and for views (but not for view columns): ```sql CREATE TABLE Employees ( ID INTEGER, name NVARCHAR(...) COMMENT 'I am the description for "name"' ) COMMENT 'I am the description for "Employee"' ``` ::: tip Propagation of doc comments can be stopped via an empty one: `/** */`. ::: In CAP Node.js, doc comments need to be switched on when calling the compiler: ::: code-group ```sh [CLI] cds compile foo.cds --docs ``` ```js [JavaScript] cds.compile(..., { docs: true }) ``` ::: ::: tip Doc comments are enabled by default in CAP Java. In CAP Java, doc comments are automatically enabled by the [CDS Maven Plugin](../java/developing-applications/building#cds-maven-plugin). In generated interfaces they are [converted to corresponding Javadoc comments](../java/assets/cds-maven-plugin-site/generate-mojo.html#documentation). ::: ## Entities & Type Definitions - [Entity Definitions](#entity-definitions) - [Type Definitions](#type-definitions) - [Structured Types](#structured-types) - [Arrayed Types](#arrayed-types) - [Virtual Elements](#virtual-elements) - [Calculated elements](#calculated-elements) - [Default Values](#default-values) - [Type References](#type-references) - [Constraints](#constraints) - [Enums](#enums) ### Entity Definitions {#entities} Entities are structured types with named and typed elements, representing sets of (persisted) data that can be read and manipulated using usual CRUD operations. They usually contain one or more designated primary key elements:  ```cds define entity Employees { key ID : Integer; name : String; jobTitle : String; } ``` > The `define` keyword is optional, that means `define entity Foo` is equal to `entity Foo`. ### Type Definitions {#types} You can declare custom types to reuse later on, for example, for elements in entity definitions. Custom-defined types can be simple, that is derived from one of the predefined types, structure types or [Associations](#associations). ```cds define type User : String(111); define type Amount { value : Decimal(10,3); currency : Currency; } define type Currency : Association to Currencies; ``` > The `define` keyword is optional, that means `define type Foo` is equal to `type Foo`. [Learn more about **Definitions of Named Aspects**.](#aspects){.learn-more} ### Structured Types You can declare and use custom struct types as follows: ```cds type Amount { value : Decimal(10,3); currency : Currency; } entity Books { price : Amount; } ``` Elements can also be specified with anonymous inline struct types. For example, the following is equivalent to the definition of `Books` above: ```cds entity Books { price : { value : Decimal(10,3); currency : Currency; }; } ``` You can declare structured types based on other definitions using the `projection on` syntax. You can use nested projections or aliases as known from entity projections. Only the effective signature of the projection is relevant.  ```cds type CustomerData : projection on Customer { name.firstName, // select from structures name.lastName, address as customerAddress, // aliases } ``` ### Arrayed Types Prefix a type specification with `array of` or `many` to signify array types. ```cds entity Foo { emails: many String; } entity Bar { emails: many { kind:String; address:String; }; } entity Car { emails: many EmailAddress; } entity Car { emails: EmailAddresses; } type EmailAddresses : many { kind:String; address:String; } type EmailAddress : { kind:String; address:String; } ``` > Keywords `many` and `array of` are mere syntax variants with identical semantics and implementations. When deployed to SQL databases, such fields are mapped to [LargeString](types) columns and the data is stored denormalized as JSON array. With OData V4, arrayed types are rendered as `Collection` in the EDM(X). ::: warning Filter expressions, [instance-based authorization](../guides/security/authorization#instance-based-auth) and [search](../guides/providing-services#searching-data) are not supported on arrayed elements. ::: #### Null Values For arrayed types the `null` and `not null` constraints apply to the _members_ of the collections. The default is `not null` indicating that the collections can't hold `null` values. ::: warning An empty collection is represented by an empty JSON array. A `null` value is invalid for an element with arrayed type. ::: In the following example the collection `emails` may hold members that are `null`. It may also hold a member where the element `kind` is `null`. The collection `emails` itself must not be `null`! ```cds entity Bar { emails : many { kind : String null; address : String not null; } null; // -> collection emails may hold null values, overwriting default } ``` ### Virtual Elements An element definition can be prefixed with modifier keyword `virtual`. This keyword indicates that this element isn't added to persistent artifacts, that is, tables or views in SQL databases. Virtual elements are part of OData metadata. By default, virtual elements are annotated with `@Core.Computed: true`, not writable for the client and will be [silently ignored](../guides/providing-services#readonly). This means also, that they are not accessible in custom event handlers. If you want to make virtual elements writable for the client, you explicitly need to annotate these elements with `@Core.Computed: false`. Still those elements are not persisted and therefore, for example, not sortable or filterable. ```cds entity Employees { [...] virtual something : String(11); } ``` <span id="calculated-fields"/> ### Calculated Elements Elements of entities and aspects can be specified with a calculation expression, in which you can refer to other elements of the same entity/aspect. This can be either a value expression or an expression that resolves to an association. Calculated elements with a value expression are read-only, no value must be provided for them in a WRITE operation. When reading such a calculated element, the result of the expression is returned. They come in two variants: "on-read" and "on-write". The difference between them is the point in time when the expression is evaluated. #### On-read ```cds entity Employees { firstName : String; lastName : String; name : String = firstName || ' ' || lastName; name_upper = upper(name); addresses : Association to many Addresses; city = addresses[kind='home'].city; } ``` For a calculated element with "on-read" semantics, the calculation expression is evaluated when reading an entry from the entity. Using such a calculated element in a query or view definition is equivalent to writing the expression directly into the query, both with respect to semantics and to performance. In CAP, it is implemented by replacing each occurrence of a calculated element in a query by the respective expression. Entity using calculated elements: ```cds entity EmployeeView as select from Employees { name, city }; ``` Equivalent entity: ```cds entity EmployeeView as select from Employees { firstName || ' ' || lastName as name : String, addresses[kind='home'].city as city }; ``` Calculated elements "on-read" are a pure convenience feature. Instead of having to write the same expression several times in queries, you can define a calculated element **once** and then simply refer to it. In the _definition_ of a calculated element "on-read", you can use almost all expressions that are allowed in queries. Some restrictions apply: * Subqueries are not allowed. * Nested projections (inline/expand) are not allowed. * A calculated element can't be key. Like for views, the expressions are sent unchanged to the database, so you need to ensure that they work on your respective database system(s). A calculated element can be *used* in every location where an expression can occur. A calculated element can't be used in the following cases: * in the ON condition of an unmanaged association * as the foreign key of a managed association * in a query together with nested projections (inline/expand) ::: warning For the Node.js runtime, only the new database services under the _@cap-js_ scope support this feature. ::: #### On-write Calculated elements "on-write" (also referred to as "stored" calculated elements) are defined by adding the keyword `stored`. A type specification is mandatory. ```cds entity Employees { firstName : String; lastName : String; name : String = (firstName || ' ' || lastName) stored; } ``` For a calculated element "on-write", the expression is already evaluated when an entry is written into the database. The resulting value is then stored/persisted like a regular field, and when reading from the entity, it behaves like a regular field as well. Using a stored calculated element can improve performance, in particular when it's used for sorting or filtering. This is paid for by higher memory consumption. While calculated elements "on-read" are handled entirely by CAP, the "on-write" variant is implemented by using the corresponding feature for database tables. The previous entity definition results in the following table definition: ```sql -- SAP HANA syntax -- CREATE TABLE Employees ( firstName NVARCHAR, lastName NVARCHAR, name NVARCHAR GENERATED ALWAYS AS (firstName || ' ' || lastName) ); ``` For the definition of calculated elements on-write, all the on-read variant's restrictions apply and referencing localized elements isn't allowed. In addition, there are restrictions that depend on the particular database. Currently all databases supported by CAP have a common restriction: The calculation expression may only refer to fields of the same table row. Therefore, such an expression must not contain subqueries, aggregate functions, or paths with associations. No restrictions apply for reading a calculated element on-write. #### Association-like calculated elements {#association-like-calculated-elements} A calculated element can also define a filtered association/composition using infix filters: ```cds entity Employees { addresses : Association to many Addresses; homeAddress = addresses [1: kind='home']; } ``` For such a calculated element, no explicit type can be specified. Only a single association or composition can occur in the expression, and a filter must be specified. The effect essentially is like [publishing an association with an infix filter](#publish-associations-with-filter). ### Default Values As in SQL you can specify default values to fill in upon INSERTs if no value is specified for a given element. ```cds entity Foo { bar : String default 'bar'; boo : Integer default 1; } ``` Default values can also be specified in custom type definitions: ```cds type CreatedAt : Timestamp default $now; type Complex { real : Decimal default 0.0; imag : Decimal default 0.0; } ``` If the element has an enum type, you can use the enum symbol instead of a literal value: ```cds type Status : String enum {open; closed;} entity Order { status : Status default #open; } ``` ### Type References If you want to base an element's type on another element of the same structure, you can use the `type of` operator. ```cds entity Author { firstname : String(100); lastname : type of firstname; // has type "String(100)" } ``` For referencing elements of other artifacts, you can use the element access through `:`. Element references with `:` don't require `type of` in front of them. ```cds entity Employees { firstname: Author:firstname; lastname: Author:lastname; } ``` ### Constraints Element definitions can be augmented with constraint `not null` as known from SQL. ```cds entity Employees { name : String(111) not null; } ``` ### Enums You can specify enumeration values for a type as a semicolon-delimited list of symbols. For string types, declaration of actual values is optional; if omitted, the actual values are the string counterparts of the symbols. ```cds type Gender : String enum { male; female; non_binary = 'non-binary'; } entity Order { status : Integer enum { submitted = 1; fulfilled = 2; shipped = 3; canceled = -1; }; } ``` To enforce your _enum_ values during runtime, use the [`@assert.range` annotation](../guides/providing-services#assert-range). For localization of enum values, model them as [code list](./common#adding-own-code-lists). <br> ## Views & Projections {#views} Use `as select from` or `as projection on` to derive new entities from existing ones by projections, very much like views in SQL. When mapped to relational databases, such entities are in fact translated to SQL views but they're frequently also used to declare projections without any SQL views involved. The entity signature is inferred from the projection. - [The `as select from` Variant](#as-select-from) - [The `as projection on` Variant](#as-projection-on) - [Views with Inferred Signatures](#views-with-inferred-signatures) <div id="linkintoc" /> - [Views with Parameters](#views-with-parameters) ### The `as select from` Variant {#as-select-from} Use the `as select from` variant to use all possible features an underlying relational database would support using any valid [CQL](./cql) query including all query clauses. ```cds entity Foo1 as select from Bar; //> implicit {*} entity Foo2 as select from Employees { * }; entity Foo3 as select from Employees LEFT JOIN Bar on Employees.ID=Bar.ID { foo, bar as car, sum(boo) as moo } where exists ( SELECT 1 as anyXY from SomeOtherEntity as soe where soe.x = y ) group by foo, bar order by moo asc; ``` ### The `as projection on` Variant {#as-projection-on} Use the `as projection on` variant instead of `as select from` to indicate that you don't use the full power of SQL in your query. For example, having a restricted query in an entity allows us to serve such an entity from external OData services. ```cds entity Foo as projection on Bar {...} ``` Currently, the restrictions of `as projection on` compared to `as select from` are: - no explicit, manual `JOINs` - no explicit, manual `UNIONs` - no sub selects in from clauses Over time, we can add additional checks depending on specific outbound protocols. ### Views with Inferred Signatures By default views inherit all properties and annotations from their primary underlying base entity. Their [`elements`](./csn#structured-types) signature is **inferred** from the projection on base elements. Each element inherits all properties from the respective base element, except the `key` property. The `key` property is only inherited if all of the following applies: - No explicit `key` is set in the query. - All key elements of the primary base entity are selected (for example, by using `*`). - No path expression with a to-many association is used. - No `union`, `join` or similar query construct is used. For example, the following definition: ```cds entity SomeView as select from Employees { ID, name, job.title as jobTitle }; ``` Might result in this inferred signature: ```cds entity SomeView { key ID: Integer; name: String; jobTitle: String; }; ``` Note: CAP does **not** enforce uniqueness for key elements of a view or projection. Use a CDL cast to set an element's type, if one of the following conditions apply: + You don't want to use the inferred type. + The query column is an expression (no inferred type is computed). ```cds entity SomeView as select from Employees { ID : Integer64, name : LargeString, 'SAP SE' as company : String }; ``` ::: tip By using a cast, annotations and other properties are inherited from the provided type and not the base element, see [Annotation Propagation](#annotation-propagation) ::: <div id="afterinferredsig" /> <div id="beforeviewwithparam" /> ### Views with Parameters You can equip views with parameters that are passed in whenever that view is queried. Default values can be specified. Refer to these parameters in the view's query using the prefix `:`. ```cds entity SomeView ( foo: Integer, bar: Boolean ) as SELECT * from Employees where ID=:foo; ``` When selecting from a view with parameters, the parameters are passed by name. In the following example, `UsingView` also has a parameter `bar` that is passed down to `SomeView`. ```cds entity UsingView ( bar: Boolean ) as SELECT * from SomeView(foo: 17, bar: :bar); ``` For Node.js, there's no programmatic API yet. You need to provide a [CQN snippet](/cds/cqn#select). In CAP Java, run a select statement against the view with named [parameter values](/java/working-with-cql/query-execution#querying-views): ::: code-group ```js [Node] SELECT.from({ ref: [{ id: 'UsingView', args: { bar: { val: true }}} ]} ) ``` ```Java [Java] var params = Map.of("bar", true); Result result = service.run(Select.from("UsingView"), params); ``` ::: [Learn more about how to expose views with parameters in **Services - Exposed Entities**.](#exposed-entities){ .learn-more} [Learn more about views with parameters for existing HANA artifacts in **Native SAP HANA Artifacts**.](../advanced/hana){ .learn-more} ## Associations Associations capture relationships between entities. They are like forward-declared joins added to a table definition in SQL. - [Unmanaged Associations](#unmanaged-associations) - [Managed Associations](#managed-associations) - [To-many Associations](#to-many-associations) - [Many-to-many Associations](#many-to-many-associations) <span id="linkinassoctoc" /> - [Compositions](#compositions) - [Managed Compositions](#managed-compositions) ### Unmanaged Associations Unmanaged associations specify arbitrary join conditions in their `on` clause, which refer to available foreign key elements. The association's name (`address` in the following example) is used as the alias for the to-be-joined target entity. ```cds entity Employees { address : Association to Addresses on address.ID = address_ID; address_ID : Integer; //> foreign key } ``` ```cds entity Addresses { key ID : Integer; } ``` ### Managed (To-One) Associations {#managed-associations} For to-one associations, CDS can automatically resolve and add requisite foreign key elements from the target's primary keys and implicitly add respective join conditions. ```cds entity Employees { address : Association to Addresses; } ``` This example is equivalent to the [unmanaged example above](#unmanaged-associations), with the foreign key element `address_ID` being added automatically upon activation to a SQL database. The names of the automatically added foreign key elements cannot be changed. > Note: For adding foreign key constraints on database level, see [Database Constraints.](../guides/databases#database-constraints). If the target has a single primary key, a default value can be provided. This default applies to the generated foreign key element `address_ID`: ```cds entity Employees { address : Association to Addresses default 17; } ``` ### To-many Associations For to-many associations specify an `on` condition following the canonical expression pattern `<assoc>.<backlink> = $self` as in this example: ```cds entity Employees { key ID : Integer; addresses : Association to many Addresses on addresses.owner = $self; } ``` ```cds entity Addresses { owner : Association to Employees; //> the backlink } ``` > The backlink can be any managed to-one association on the _many_ side pointing back to the _one_ side. ### Many-to-many Associations For many-to-many association, follow the common practice of resolving logical many-to-many relationships into two one-to-many associations using a link entity to connect both. For example: ```cds entity Employees { [...] addresses : Association to many Emp2Addr on addresses.emp = $self; } entity Emp2Addr { key emp : Association to Employees; key adr : Association to Addresses; } ``` [Learn more about **Managed Compositions for Many-to-many Relationships**.](#for-many-to-many-relationships){.learn-more} <div id="aftermanytomany" /> <div id="inbetweenthings" /> <div id="beforecompo" /> ### Compositions Compositions constitute document structures through _contained-in_ relationships. They frequently show up in to-many header-child scenarios. ```cds entity Orders { key ID: Integer; //... Items : Composition of many Orders.Items on Items.parent = $self; } entity Orders.Items { key pos : Integer; key parent : Association to Orders; product : Association to Products; quantity : Integer; } ``` :::info Contained-in relationship Essentially, Compositions are the same as _[associations](#associations)_, just with the additional information that this association represents a _contained-in_ relationship so the same syntax and rules apply in their base form. ::: ::: warning Limitations of Compositions of one Using of compositions of one for entities is discouraged. There is often no added value of using them as the information can be placed in the root entity. Compositions of one have limitations as follow: - Very limited Draft support. Fiori elements does not support compositions of one unless you take care of their creation in a custom handler. - No extensive support for modifications over paths if compositions of one are involved. You must fill in foreign keys manually in a custom handler. ::: ### Managed Compositions of Aspects {#managed-compositions} Use managed compositions variant to nicely reflect document structures in your domain models, without the need for separate entities, reverse associations, and unmanaged `on` conditions. #### With Inline Targets ```cds entity Orders { key ID: Integer; //... Items : Composition of many { key pos : Integer; product : Association to Products; quantity : Integer; } }; ``` Managed Compositions are mostly syntactical sugar: Behind the scenes, they are unfolded to the [unmanaged equivalent as shown above](#compositions) by automatically adding a new entity, the name of which being constructed as a [scoped name](#scoped-names) from the name of parent entity, followed by the name of the composition element, that is `Orders.Items` in the previous example. You can safely use this name at other places, for example to define an association to the generated child entity:  ```cds entity Orders { // … specialItem : Association to Orders.Items; }; ``` #### With Named Targets Instead of anonymous target aspects you can also specify named aspects, which are unfolded the same way as anonymous inner types, as shown in the previous example: ```cds entity Orders { key ID: Integer; //... Items : Composition of many OrderItems; } aspect OrderItems { key pos : Integer; product : Association to Products; quantity : Integer; } ``` #### Default Target Cardinality If not otherwise specified, a managed composition of an aspect has the default target cardinality *to-one*. #### For Many-to-many Relationships Managed Compositions are handy for [many-to-many relationships](#many-to-many-associations), where a link table usually is private to one side. ```cds entity Teams { [...] members : Composition of many { key user: Association to Users; } } entity Users { [...] teams: Association to many Teams.members on teams.user = $self; } ``` And here's an example of an attributed many-to-many relationship: ```cds entity Teams { [...] members : Composition of many { key user : Association to Users; role : String enum { Lead; Member; Collaborator; } } } entity Users { ... } ``` To navigate between _Teams_ and _Users_, you have to follow two associations: `members.user` or `teams.up_`. In OData, to get all users of all teams, use a query like the following: ```cds GET /Teams?$expand=members($expand=user) ``` ### Publish Associations in Projections {#publish-associations} As associations are first class citizens, you can put them into the select list of a view or projection ("publish") like regular elements. A `select *` includes all associations. If you need to rename an association, you can provide an alias. ```cds entity P_Employees as projection on Employees { ID, addresses } ``` The effective signature of the projection contains an association `addresses` with the same properties as association `addresses` of entity `Employees`. #### Publish Associations with Infix Filter {#publish-associations-with-filter} When publishing an unmanaged association in a view or projection, you can add a filter condition. The ON condition of the resulting association is the ON condition of the original association plus the filter condition, combined with `and`. ```cds entity P_Authors as projection on Authors { *, books[stock > 0] as availableBooks }; ``` In this example, in addition to `books` projection `P_Authors` has a new association `availableBooks` that points only to those books where `stock > 0`. If the filter condition effectively reduces the cardinality of the association to one, you should make this explicit in the filter by adding a `1:` before the condition: ```cds entity P_Employees as projection on Employees { *, addresses[1: kind='home'] as homeAddress // homeAddress is to-one } ``` Filters usually are provided only for to-many associations, which usually are unmanaged. Thus publishing with a filter is almost exclusively used for unmanaged associations. Nevertheless you can also publish a managed association with a filter. This will automatically turn the resulting association into an unmanaged one. You must ensure that all foreign key elements needed for the ON condition are explicitly published. ```cds entity P_Books as projection on Books { author.ID as authorID, // needed for ON condition of deadAuthor author[dateOfDeath is not null] as deadAuthor // -> unmanaged association }; ``` Publishing a _composition_ with a filter is similar, with an important difference: in a deep Update, Insert, or Delete statement the respective operation does not cascade to the target entities. Thus the type of the resulting element is set to `cds.Association`. [Learn more about `cds.Association`.](/cds/csn#associations){.learn-more} In [SAP Fiori Draft](../advanced/fiori#draft-support), it behaves like an "enclosed" association, that means, it points to the target draft entity. In the following example, `singleItem` has type `cds.Association`. In draft mode, navigating along `singleItems` doesn't leave the draft tree. ```cds @odata.draft.enabled entity P_orders as projection on Orders { *, Items[quantity = 1] as singleItems } ``` ## Annotations This section describes how to add Annotations to model definitions written in CDL, focused on the common syntax options, and fundamental concepts. Find additional information in the [OData Annotations](../advanced/odata#annotations) guide. - [Annotation Syntax](#annotation-syntax) - [Annotation Targets](#annotation-targets) - [Annotation Values](#annotation-values) - [Expressions as Annotation Values](#expressions-as-annotation-values) - [Records as Syntax Shortcuts](#records-as-syntax-shortcuts) - [Annotation Propagation](#annotation-propagation) - [The `annotate` Directive](#annotate) - [Extend Array Annotations](#extend-array-annotations) ### Annotation Syntax Annotations in CDL are prefixed with an `@` character and can be placed before a definition, after the defined name or at the end of simple definitions. ```cds @before entity Foo @inner { @before simpleElement @inner : String @after; @before structElement @inner { /* elements */ } } ``` Multiple annotations can be placed in each spot separated by whitespaces or enclosed in `@(...)` and separated by comma - like the following are equivalent: ```cds entity Foo @( my.annotation: foo, another.one: 4711 ) { /* elements */ } ``` ```cds @my.annotation:foo @another.one: 4711 entity Foo { /* elements */ } ``` For an `@inner` annotation, only the syntax `@(...)` is available. #### Using `annotate` Directives Instead of interspersing annotations with definitions, you can also use the `annotate` directive to add annotations to existing definitions. ```cds annotate Foo with @( my.annotation: foo, another.one: 4711 ); ``` [Learn more about the `annotate` directive in the _Aspects_ chapter below.](#annotate){.learn-more} ### Annotation Targets You can basically annotate any named thing in a CDS model, such as: Contexts and services:  ```cds @before context foo.bar @inner { ... } @before service Sue @inner { ... } ``` Definitions and elements with simple or struct types:  ```cds @before type Foo @inner : String @after; @before entity Foo @inner { @before key ID @inner : String @after; @before title @inner : String @after; @before struct @inner { ...elements... }; } ``` Enums:  ```cds … status : String @inner enum { open @after; closed @after; cancelled @after; accepted @after; rejected @after; } ``` Columns in a view definition's query:  ```cds … as select from Foo { @before expr as alias @inner : String, … } ``` Parameters in view definitions:  ```cds … with parameters ( @before param @(inner) : String @after ) … ``` Actions/functions including their parameters and result: ```cds @before action doSomething @inner ( @before param @(inner) : String @after ) returns @before resultType; ``` Or in case of a structured result: ```cds action doSomething() returns @before { @before resultElem @inner : String @after; }; ``` ### Annotation Values Values can be literals, references, or expressions. Expressions are explained in more detail in the next section. If no value is given, the default value is `true` as for `@aFlag` in the following example: ```cds @aFlag //= true, if no value is given @aBoolean: false @aString: 'foo' @anInteger: 11 @aDecimal: 11.1 @aSymbol: #foo @aReference: foo.bar @anArray: [ /* can contain any kind of value */ ] @anExpression: ( foo.bar * 17 ) // expression, see next section ``` As described in the [CSN spec](./csn#literals), the previously mentioned annotations would compile to CSN as follows: ```jsonc { "@aFlag": true, "@aBoolean": false, "@aString": "foo", "@anInteger": 11, "@aDecimal": 11.1, "@aSymbol": {"#":"foo"}, "@aReference": {"=":"foo.bar"}, "@anArray": [ /* … */ ], "@anExpression": { /* see next section */ } } ``` ::: tip In contrast to references in [expressions](#expressions-as-annotation-values), plain references aren't checked, resolved, or rewritten by CDS parsers or linkers. They're interpreted and evaluated only on consumption-specific modules. For example, for SAP Fiori models, it's the _4odata_ and _2edm(x)_ processors. ::: ### Records as Syntax Shortcuts Annotations in CDS are flat lists of key-value pairs assigned to a target. The record syntax - that is, `{key:<value>, ...}` - is a shortcut notation that applies a common prefix to nested annotations. For example, the following are equivalent: ```cds @Common.foo.bar @Common.foo.car: 'wheels' ``` ```cds @Common: { foo.bar, foo.car: 'wheels' } ``` ```cds @Common.foo: { bar } @Common.foo.car: 'wheels' ``` ```cds @Common.foo: { bar, car: 'wheels' } ``` and they would show up as follows in a parsed model (→ see [CSN](./csn)): ```json { "@Common.foo.bar": true, "@Common.foo.car": "wheels" } ``` ### Annotation Propagation {#annotation-propagation} Annotations are inherited from types and base types to derived types, entities, and elements as well as from elements of underlying entities in case of views. For example, given this view definition: ```cds using Books from './bookshop-model'; entity BooksList as select from Books { ID, genre : Genre, title, author.name as author }; ``` * `BooksList` would inherit annotations from `Books` * `BooksList:ID` would inherit from `Books:ID` * `BooksList:author` would inherit from `Books:author.name` * `BooksList.genre` would inherit from type `Genre` The rules are: 1. Entity-level properties and annotations are inherited from the **primary** underlying source entity — here `Books`. 2. Each element that can **unambiguously** be traced back to a single source element, inherits that element's properties. 3. An explicit **cast** in the select clause cuts off the inheritance, for example, as for `genre` in our previous example. ::: tip Propagation of annotations can be stopped via value `null`, for example, `@anno: null`. ::: ### Expressions as Annotation Values <Beta /> {#expressions-as-annotation-values} In order to use an expression as an annotation value, it must be enclosed in parentheses: ```cds @anExpression: ( foo.bar * 11 ) ``` Syntactically, the same expressions are supported as in a select item or in the where clause of a query, except subqueries. The expression can of course also be a single reference or a simple value: ```cds @aRefExpr: ( foo.bar ) @aValueExpr: ( 11 ) ``` Some advantages of using expressions as "first class" annotation values are: * syntax and references are checked by the compiler * code completion * [automatic path rewriting in propagated annotations](#propagation) * [automatic translation of expressions in OData annotations](#odata-annotations) ::: info Limitations Elements that are not available to the compiler, for example the OData draft decoration, can't be used in annotation expressions. ::: #### Name resolution Each path in the expression is checked: * For an annotation assigned to an entity, the first path step is resolved as element of the entity. * For an annotation assigned to an entity element, the first path step is resolved as the annotated element or its siblings. * If the annotation is assigned to a subelement of a structured element, the top level elements of the entity can be accessed via `$self`. * A parameter `par` can be accessed via `:par`, just like parameters of a parametrized entity in queries. * For an annotation assigned to a bound action or function, elements of the respective entity can be accessed via `$self`. * The draft-specific elements `IsActiveEntity`, `HasActiveEntity`, and `HasDraftEntity` can be referred to with respective magic variables `$draft.IsActiveEntity`, `$draft.HasActiveEntity`, and `$draft.HasDraftEntity`. During draft augmentation, `$draft.<...>` is rewritten to `$self.<...>` for all draft enabled entities (root and sub nodes, but not for named types or entity parameters). * If a path can't be resolved successfully, compilation fails with an error. In contrast to `@aReference: foo.bar`, a single reference written as expression `@aRefExpr: ( foo.bar )` is checked by the compiler. ```cds @MyAnno: (a) // reference to element entity Foo (par: Integer) { key ID : Integer; @MyAnno: (:par) // reference to entity parameter a : Integer; @MyAnno: (a) // reference to sibling element b : Integer; s { @MyAnno: (y) // reference to sibling element x : Integer; @MyAnno: ($self.a) // reference to top level element y : Integer; } } actions { @MyAnno: ($self.a) action A () } ``` #### CSN Representation In CSN, the expression is represented as a record with two properties: * A string representation of the expression is stored in property `=`. * A tokenized representation of the expression is stored in one of the properties `xpr`, `ref`, `val`, `func`, etc. (like if the expression was written in a query). ```json { "@anExpression": { "=": "foo.bar * 11", "xpr": [ {"ref": ["foo", "bar"]}, "*", {"value": 11} ] }, "@aRefExpr": { "=": "foo.bar", "ref": ["foo", "bar"] }, "@aValueExpr": { "=": "11", "val": 11 } } ``` Note the different CSN representations for a [plain value](#annotation-values) `"@anInteger": 11` and a value written as expression `@aValueExpr: ( 11 )`, respectively. #### Propagation [Annotations are propagated](#annotation-propagation) in views/projections, via includes, and along type references. If the annotation value is an expression, it is sometimes necessary to adapt references inside the expression during propagation, for example, when a referenced element is renamed in a projection. The compiler automatically takes care of the necessary rewriting. When a reference in an annotation expression is rewritten, the `=` property is set to `true`. Example: ```cds entity E { @Common.Text: (text) code : Integer; text : String; } entity P as projection on E { code, text as descr } ``` When propagated to element `code` of projection `P`, the annotation is automatically rewritten to `@Common.Text: (descr)`. ::: details Resulting CSN ```jsonc { "definitions": { "E": { // ... "elements": { // ... "code": { // original annotation "@Common.Text": { "=": "text", "ref": ["text"] }, "type": "cds.Integer" }, "text": {"type": "cds.String"} } }, "P": { // ... "elements": { // ... "code": { // propagated annotation, reference adapted "@Common.Text": { "=": true, "ref": ["descr"] }, "type": "cds.Integer" }, "descr": {"type": "cds.String"} } } } } ``` ::: ::: info There are situations where automatic rewriting doesn't work, resulting in the compiler error [`anno-missing-rewrite`](https://cap.cloud.sap/docs/cds/compiler/messages#anno-missing-rewrite). Some of these situations are going to be addressed in upcoming releases. ::: #### CDS Annotations Using an expression as annotation value only makes sense if the evaluator of the annotation is prepared to deal with the new CSN representation. Currently, the CAP runtimes only support expressions in the `where` property of the `@restrict` annotation. ```cds entity Orders @(restrict: [ { grant: 'READ', to: 'Auditor', where: (AuditBy = $user.id) } ]) {/*...*/} ``` More annotations are going to follow in upcoming releases. Of course, you can use this feature also in your custom annotations, where you control the code that evaluates the annotations. #### OData Annotations The OData backend of the CAP CDS compiler supports expression-valued annotations. See [Expressions in OData Annotations](../advanced/odata#expression-annotations). ### Extend Array Annotations {#extend-array-annotations} Usually, the annotation value provided in an `annotate` directive overwrites an already existing annotation value. If the existing value is an array, the *ellipsis* syntax allows to insert new values **before** or **after** the existing entries, instead of overwriting the complete array. The ellipsis represents the already existing array entries. Of course, this works with any kind of array entries. This is a sample of an existing array: ```cds @anArray: [3, 4] entity Foo { /* elements */ } ``` This shows how to extend the array: ```cds annotate Foo with @anArray: [1, 2, ...]; //> prepend new values: [1, 2, 3, 4] annotate Foo with @anArray: [..., 5, 6]; //> append new values: [3, 4, 5, 6] annotate Foo with @anArray: [1, 2, ..., 5, 6]; //> prepend and append ``` It's also possible to insert new entries at **arbitrary positions**. For this, use `... up to` with a *comparator* value that identifies the insertion point.  ```cds [... up to <comparator>, newEntry, ...] ``` `... up to` represents the existing entries of the array from the current position up to and including the first entry that matches the comparator. New entries are then inserted behind the matched entry. If there's no match, new entries are appended at the end of the existing array. This is a sample of an existing array: ```cds @anArray: [1, 2, 3, 4, 5, 6] entity Bar { /* elements */ } ``` This shows how to insert values after `2` and `4`: ```cds annotate Bar with @anArray: [ ... up to 2, // existing entries 1, 2 2.1, 2.2, // insert new entries 2.1, 2.2 ... up to 4, // existing entries 3, 4 4.1, 4.2, // insert new entries 4.1, 4.2 ... // remaining existing entries 5, 6 ]; ``` The resulting array is: ```js [1, 2, 2.1, 2.2, 3, 4, 4.1, 4.2, 5, 6] ``` If your array entries are objects, you have to provide a *comparator object*. It matches an existing entry, if all attributes provided in the comparator match the corresponding attributes in an existing entry. The comparator object doesn't have to contain all attributes that the existing array entries have, simply choose those attributes that sufficiently characterize the array entry after which you want to insert. Only simple values are allowed for the comparator attributes. Example: Insert a new entry after `BeginDate`. ```cds @UI.LineItem: [ { $Type: 'UI.DataFieldForAction', Action: 'TravelService.acceptTravel', Label: '{i18n>AcceptTravel}' }, { Value: TravelID, Label: 'ID' }, { Value: BeginDate, Label: 'Begin' }, { Value: EndDate, Label: 'End