diff options
Diffstat (limited to 'docs-developer/src/doc/docs/developer_guide/architecture')
8 files changed, 947 insertions, 0 deletions
diff --git a/docs-developer/src/doc/docs/developer_guide/architecture/architecture_overview.md b/docs-developer/src/doc/docs/developer_guide/architecture/architecture_overview.md new file mode 100644 index 00000000..d72dda91 --- /dev/null +++ b/docs-developer/src/doc/docs/developer_guide/architecture/architecture_overview.md @@ -0,0 +1,123 @@ +# Architecture overview + +Normally, you would think that a tool like Dokka simply parses some programming language sources and generates +HTML pages for whatever it sees along the way, with little to no abstractions. That would be the simplest and +the most straightforward way to implement an API documentation engine. + +However, it was clear that Dokka may need to generate documentation from various sources (not only Kotlin), that users +might request additional output formats (like Markdown), that users might need additional features like supporting +custom KDoc tags or rendering [mermaid.js](https://mermaid.js.org/) diagrams - all these things would require changing +a lot of code inside Dokka itself if all solutions were hardcoded. + +For this reason, Dokka was built from the ground up to be easily extensible and customizable by adding several layers +of abstractions to the data model, and by providing pluggable extension points, giving you the ability to introduce +selective changes on a given level. + +## Overview of data model + +Generating API documentation begins with input source files (`.kt`, `.java`, etc) and ends with some output files +(`.html`/`.md`, etc). However, to allow for extensibility and customization, several input and output independent +abstractions have been added to the data model. + +Below you can find the general pipeline of processing data gathered from sources and the explanation for each stage. + +```mermaid +flowchart TD + Input --> Documentables --> Pages --> Output +``` + +* `Input` - generalization of sources, by default Kotlin / Java sources, but could be virtually anything +* [`Documentables`](data_model/documentable_model.md) - unified data model that represents _any_ parsed sources as a + tree, independent of the source language. Examples of a `Documentable`: class, function, package, property, etc +* [`Pages`](data_model/page_content.md) - universal model that represents output pages (e.g a function/property page) + and the content it's composed of (lists, text, code blocks) that the users needs to see. Not to be confused with + `.html` pages. Goes hand in hand with the so-called [Content model](data_model/page_content.md#content-model). +* `Output` - specific output formats like HTML / Markdown / Javadoc and so on. This is a mapping of the pages/content + model to a human-readable and visual representation. For instance: + * `PageNode` is mapped as + * `.html` file for the HTML format + * `.md` file for the Markdown format + * `ContentList` is mapped as + * `<li>` / `<ul>` for the HTML format + * `1.` / `*` for the Markdown format + * `ContentCodeBlock` is mapped as + * `<code>` or `<pre>` with some CSS styles in the HTML format + * Text wrapped in triple backticks for the Markdown format + + +You, as a Dokka developer or a plugin writer, can use extension points to introduce selective changes to the +model on one particular level without altering everything else. + +For instance, if you wanted to make an annotation / function / class invisible in the final documentation, you would only +need to modify the `Documentables` level by filtering undesirable declarations out. If you wanted to display all overloaded +methods on the same page instead of on separate ones, you would only need to modify the `Pages` layer by merging multiple +pages into one, and so on. + +For a deeper dive into Dokka's model with more examples and details, +see sections about [Documentables](data_model/documentable_model.md) and [Page/Content](data_model/page_content.md) + +For an overview of existing extension points that let you transform Dokka's models, see +[Core extension points](extension_points/core_extension_points.md) and [Base extensions](extension_points/base_plugin.md). + +## Overview of extension points + +An _extension point_ usually represents a pluggable interface that performs an action during one of the stages of +generating documentation. An _extension_ is, therefore, an implementation of the interface which is extending the +extension point. + +You can create extension points, provide your own implementations (extensions) and configure them. All of +this is possible with Dokka's plugin / extension point API. + +Here's a sneak peek of the DSL: + +```kotlin +// declare your own plugin +class MyPlugin : DokkaPlugin() { + // create an extension point for developers to use + val signatureProvider by extensionPoint<SignatureProvider>() + + // provide a default implementation + val defaultSignatureProvider by extending { + signatureProvider with KotlinSignatureProvider() + } + + // register our own extension in Dokka's Base plugin by overriding its default implementation + val dokkaBasePlugin by lazy { plugin<DokkaBase>() } + val multimoduleLocationProvider by extending { + (dokkaBasePlugin.locationProviderFactory + providing MultimoduleLocationProvider::Factory + override dokkaBasePlugin.locationProvider) + } +} + +class MyExtension(val context: DokkaContext) { + + // use an existing extension + val signatureProvider: SignatureProvider = context.plugin<MyPlugin>().querySingle { signatureProvider } + + fun doSomething() { + signatureProvider.signature(..) + } +} + +interface SignatureProvider { + fun signature(documentable: Documentable): List<ContentNode> +} + +class KotlinSignatureProvider : SignatureProvider { + override fun signature(documentable: Documentable): List<ContentNode> = listOf() +} +``` + +For a deeper dive into extensions and extension points, see [Introduction to Extensions](extension_points/extension_points.md). + +For an overview of existing extension points, see [Core extension points](extension_points/core_extension_points.md) and +[Base extensions](extension_points/base_plugin.md). + +## Historical context + +This is a second iteration of Dokka that was built from scratch. + +If you want to learn more about why Dokka was redesigned this way, watch this great talk by Paweł Marks: +[New Dokka - Designed for Fearless Creativity](https://www.youtube.com/watch?v=OvFoTRhqaKg). The general principles +and general architecture are the same, although it may be outdated in some areas, so please double-check. diff --git a/docs-developer/src/doc/docs/developer_guide/architecture/data_model/documentable_model.md b/docs-developer/src/doc/docs/developer_guide/architecture/data_model/documentable_model.md new file mode 100644 index 00000000..b30780fc --- /dev/null +++ b/docs-developer/src/doc/docs/developer_guide/architecture/data_model/documentable_model.md @@ -0,0 +1,251 @@ +# Documentable Model + +The Documentable model represents the data that is parsed from some programming language sources. Think of this data as +of something that could be seen or produced by a compiler frontend, it's not far off from the truth. + +By default, the documentables are created from: + +* Descriptors (Kotlin's K1 compiler) +* Symbols (Kotlin's K2 compiler) +* [PSI](https://plugins.jetbrains.com/docs/intellij/psi.html) (Java's model). + +Code-wise, you can have a look at following classes: + +* `DefaultDescriptorToDocumentableTranslator` - responsible for Kotlin -> `Documentable` mapping +* `DefaultPsiToDocumentableTranslator` - responsible for Java -> `Documentable` mapping + +Upon creation, the documentable model represents a collection of trees, each with `DModule` as root. + +Take some arbitrary Kotlin source code that is located within the same module: + +```kotlin +// Package 1 +class Clazz(val property: String) { + fun function(parameter: String) {} +} + +fun topLevelFunction() {} + +// Package 2 +enum class Enum { } + +val topLevelProperty: String +``` + +This would be represented roughly as the following Documentable tree: + +```mermaid +flowchart TD + DModule --> firstPackage[DPackage] + firstPackage --> DClass + firstPackage --> toplevelfunction[DFunction] + DClass --> DProperty + DClass --> DFunction + DFunction --> DParameter + DModule --> secondPackage[DPackage] + secondPackage --> DEnum + secondPackage --> secondPackageProperty[DProperty] +``` + +At later stages of transformation, all trees are folded into one by +[DocumentableMerger](../extension_points/core_extension_points.md#documentablemerger). + +## Documentable + +The main building block of the documentable model is the `Documentable` class. It is the base class for all more specific +types. All implementations represent elements of source code with mostly self-explanatory names: `DFunction`, +`DPackage`, `DProperty`, and so on. + +`DClasslike` is the base class for all class-like documentables, such as `DClass`, `DEnum`, `DAnnotation` and others. + +The contents of each documentable normally represent what you would see in the source code. + +For example, if you open +`DClass`, you should find that it contains references to functions, properties, companion objects, constructors and so +on. `DEnum` should have references to its entries, and `DPackage` can have references to both classlikes and top-level +functions and properties (Kotlin-specific). + +Here's an example of a documentable: + +```kotlin +data class DClass( + val dri: DRI, + val name: String, + val constructors: List<DFunction>, + val functions: List<DFunction>, + val properties: List<DProperty>, + val classlikes: List<DClasslike>, + val sources: SourceSetDependent<DocumentableSource>, + val visibility: SourceSetDependent<Visibility>, + val companion: DObject?, + val generics: List<DTypeParameter>, + val supertypes: SourceSetDependent<List<TypeConstructorWithKind>>, + val documentation: SourceSetDependent<DocumentationNode>, + val expectPresentInSet: DokkaSourceSet?, + val modifier: SourceSetDependent<Modifier>, + val sourceSets: Set<DokkaSourceSet>, + val isExpectActual: Boolean, + val extra: PropertyContainer<DClass> = PropertyContainer.empty() +) : DClasslike(), WithAbstraction, WithCompanion, WithConstructors, + WithGenerics, WithSupertypes, WithExtraProperties<DClass> +``` + +___ + +There are three non-documentable classes that are important for this model: + +* `DRI` +* `SourceSetDependent` +* `ExtraProperty`. + +### DRI + +`DRI` stans for _Dokka Resource Identifier_ - a unique value that identifies a specific `Documentable`. +All references and relations between the documentables (other than direct ownership) are described using `DRI`. + +For example, `DFunction` with a parameter of type `Foo` only has `Foo`'s `DRI`, but not the actual reference +to `Foo`'s `Documentable` object. + +#### Example + +For an example of how a `DRI` can look like, let's take the `limitedParallelism` function from `kotlinx.coroutines`: + +```kotlin +package kotlinx.coroutines + +import ... + +public abstract class MainCoroutineDispatcher : CoroutineDispatcher() { + + override fun limitedParallelism(parallelism: Int): CoroutineDispatcher { + ... + } +} +``` + +If we were to re-create the DRI of this function in code, it would look something like this: + +```kotlin +DRI( + packageName = "kotlinx.coroutines", + classNames = "MainCoroutineDispatcher", + callable = Callable( + name = "limitedParallelism", + receiver = null, + params = listOf( + TypeConstructor( + fullyQualifiedName = "kotlin.Int", + params = emptyList() + ) + ) + ), + target = PointingToDeclaration, + extra = null +) +``` + +If you format it as `String`, it would look like this: + +``` +kotlinx.coroutines/MainCoroutineDispatcher/limitedParallelism/#kotlin.Int/PointingToDeclaration/ +``` + +### SourceSetDependent + +`SourceSetDependent` helps handling multiplatform data by associating platform-specific data (declared with either +`expect` or `actual` modifiers) with particular +[source sets](https://kotlinlang.org/docs/multiplatform-discover-project.html#source-sets). + +This comes in handy if the `expect` / `actual` declarations differ. For example, the default value for `actual` might +differ from that declared in `expect`, or code comments written for `expect` might be different from what's written +for `actual`. + +Under the hood, it's a `typealias` to a `Map`: + +```kotlin +typealias SourceSetDependent<T> = Map<DokkaSourceSet, T> +``` + +### ExtraProperty + +`ExtraProperty` is used to store any additional information that falls outside of the regular model. It is highly +recommended to use extras to provide any additional information when creating custom Dokka plugins. + +This element is a bit more complex, so you can read more about how to use it +[in a separate section](extra.md). + +___ + +## Documentation model + +The Documentation model is used alongside documentables to store data obtained by parsing +code comments (such as KDocs / Javadocs). + +### DocTag + +`DocTag` describes a specific documentation syntax element. + +It's universal across language sources. For example, the DocTag `B` is the same for `**bold**` in Kotlin and +`<b>bold</b>` in Java. + +However, some DocTag elements are specific to one language. There are many such examples for Java, because it allows +HTML tags inside the Javadoc comments, some of which are simply not possible to reproduce with Markdown that KDocs use. + +`DocTag` elements can be deeply nested with other `DocTag` children elements. + +Examples: + +```kotlin +data class H1( + override val children: List<DocTag> = emptyList(), + override val params: Map<String, String> = emptyMap() +) : DocTag() + +data class H2( + override val children: List<DocTag> = emptyList(), + override val params: Map<String, String> = emptyMap() +) : DocTag() + +data class Strikethrough( + override val children: List<DocTag> = emptyList(), + override val params: Map<String, String> = emptyMap() +) : DocTag() + +data class Strong( + override val children: List<DocTag> = emptyList(), + override val params: Map<String, String> = emptyMap() +) : DocTag() + +data class CodeBlock( + override val children: List<DocTag> = emptyList(), + override val params: Map<String, String> = emptyMap() +) : Code() + +``` + +### TagWrapper + +`TagWrapper` describes the whole comment description or a specific comment tag. For example: `@see` / `@author` / `@return`. + +Since each such section may contain formatted text inside it, each `TagWrapper` has `DocTag` children. + +```kotlin +/** + * @author **Ben Affleck* + * @return nothing, except _sometimes_ it may throw an [Error] + */ +fun foo() {} +``` + +### DocumentationNode + +`DocumentationNode` acts as a container for multiple `TagWrapper` elements for a specific `Documentable`, usually +used like this: + +```kotlin +data class DFunction( + ... + val documentation: SourceSetDependent<DocumentationNode>, + ... +) +``` diff --git a/docs-developer/src/doc/docs/developer_guide/architecture/data_model/extra.md b/docs-developer/src/doc/docs/developer_guide/architecture/data_model/extra.md new file mode 100644 index 00000000..d7412e36 --- /dev/null +++ b/docs-developer/src/doc/docs/developer_guide/architecture/data_model/extra.md @@ -0,0 +1,91 @@ +# Extra + +## Introduction + +`ExtraProperty` is used to store any additional information that falls outside of the regular model. It is highly +recommended to use extras to provide any additional information when creating custom Dokka plugins. + +`ExtraProperty` classes are available both in the [Documentable](documentable_model.md) and the [Content](page_content.md#content-model) +models. + +To create a new extra, you need to implement the `ExtraProperty` interface. It is advised to use the following pattern +when declaring new extras: + +```kotlin +data class CustomExtra( + [any data relevant to your extra], + [any data relevant to your extra] +): ExtraProperty<Documentable> { + override val key: CustomExtra.Key<Documentable, *> = CustomExtra + companion object : CustomExtra.Key<Documentable, CustomExtra> +} +``` + +Merge strategy (the `mergeStrategyFor` method) for extras is invoked during the +[merging](../extension_points/core_extension_points.md#documentablemerger) of the documentables from different +[source sets](https://kotlinlang.org/docs/multiplatform-discover-project.html#source-sets), when the documentables being +merged have their own `Extra` of the same type. + +## PropertyContainer + +All extras for `ContentNode` and `Documentable` classes are stored in the `PropertyContainer<C : Any>` class instances. + +```kotlin +data class DFunction( + ... + override val extra: PropertyContainer<DFunction> = PropertyContainer.empty() + ... +) : WithExtraProperties<DFunction> +``` + +`PropertyContainer` has a number of convenient functions for handling extras in a collection-like manner. + +The generic class parameter `C` limits the types of properties that can be stored in the container - it must +match the generic `C` class parameter from the `ExtraProperty` interface. This allows creating extra properties +which can only be stored in a specific `Documentable`. + +## Usage example + +In following example we will create a `DFunction`-only extra property, store it and then retrieve its value: + +```kotlin +// Extra that is applicable only to DFunction +data class CustomExtra(val customExtraValue: String) : ExtraProperty<DFunction> { + override val key: ExtraProperty.Key<Documentable, *> = CustomExtra + companion object: ExtraProperty.Key<Documentable, CustomExtra> +} + +// Storing it inside the documentable +fun DFunction.withCustomExtraProperty(data: String): DFunction { + return this.copy( + extra = extra + CustomExtra(data) + ) +} + +// Retrieveing it from the documentable +fun DFunction.getCustomExtraPropertyValue(): String? { + return this.extra[CustomExtra]?.customExtraValue +} +``` + +___ + +You can also use extras as markers, without storing any data in them: + +```kotlin + +object MarkerExtra : ExtraProperty<Any>, ExtraProperty.Key<Any, MarkerExtra> { + override val key: ExtraProperty.Key<Any, *> = this +} + +fun Documentable.markIfFunction(): Documentable { + return when(this) { + is DFunction -> this.copy(extra = extra + MarkerExtra) + else -> this + } +} + +fun WithExtraProperties<Documentable>.isMarked(): Boolean { + return this.extra[MarkerExtra] != null +} +``` diff --git a/docs-developer/src/doc/docs/developer_guide/architecture/data_model/page_content.md b/docs-developer/src/doc/docs/developer_guide/architecture/data_model/page_content.md new file mode 100644 index 00000000..eb85200f --- /dev/null +++ b/docs-developer/src/doc/docs/developer_guide/architecture/data_model/page_content.md @@ -0,0 +1,144 @@ +# Page / Content Model + +Even though the `Page` and `Content` models reside on the same level (under `Page`), it is easier to view them as two +different models altogether, even though `Content` is only used in conjunction with and inside the `Page` model only. + +## Page + +The Page model represents the structure of documentation pages to be generated. During rendering, each page +is processed separately, so one page corresponds to exactly one output file. + +The Page model is independent of the final output format. In other words, it's universal. Which file extension the pages +should be created as (`.html`, `.md`, etc), and how, is up to the +[Renderer](../extension_points/core_extension_points.md#renderer) extension. + +Subclasses of the `PageNode` class represent the different kinds of pages, such as `ModulePage`, `PackagePage`, +`ClasslikePage`, `MemberPage` and so on. + +The Page model can be represented as a tree, with `RootPageNode` at the root. + +Here's an example of how an arbitrary project's `Page` tree might look like, if the project consists of a module with +3 packages, one of which contains a top level function, a top level property and a class, inside which there's a function +and a property: + +```mermaid +flowchart TD + RootPageNode --> firstPackage[PackagePageNode] + RootPageNode --> secondPackage[PackagePageNode] + RootPageNode --> thirdPackage[PackagePageNode] + firstPackage --> firstPackageFirstMember[MemberPageNode - Function] + firstPackage --> firstPackageSecondMember[MemberPageNode - Property] + firstPackage ---> firstPackageClasslike[ClasslikePageNode - Class] + firstPackageClasslike --> firstPackageClasslikeFirstMember[MemberPageNode - Function] + firstPackageClasslike --> firstPackageClasslikeSecondMember[MemberPageNode - Property] + secondPackage --> etcOne[...] + thirdPackage --> etcTwo[...] +``` + +Almost all pages are derivatives of `ContentPage` - it's the type of a page that has user-visible content on it. + +## Content Model + +The Content model describes what the pages consist of. It is essentially a set of building blocks that you can put +together to represent some content. It is also output-format independent and universal. + +For an example, have a look at the subclasses of `ContentNode`: `ContentText`, `ContentList`, `ContentTable`, +`ContentCodeBlock`, `ContentHeader` and so on -- all self-explanatory. You can group chunks of content together with +`ContentGroup` - for example, to wrap all children with a style. + +```kotlin +// real example of composing content using the `DocumentableContentBuilder` DSL +orderedList { + item { + text("This list contains a nested table:") + table { + header { + text("Col1") + text("Col2") + } + row { + text("Text1") + text("Text2") + } + } + } + item { + group(styles = setOf(TextStyle.Bold)) { + text("This is bald") + text("This is also bald") + } + } +} +``` + +It is the responsibility of the `Renderer` (i.e a specific output format) to render it in a way the user can process it, +be it visually (html pages) or otherwise (json). + +For instance, `HtmlRenderer` might render `ContentCodeBlock` as `<code>text</code>`, but `CommonmarkRenderer` might +render it using backticks. + +### DCI + +Each node is identified by a unique `DCI`, which stands for _Dokka Content Identifier_. + +`DCI` aggregates `DRI`s of all documentables that are used by the given `ContentNode`. + +```kotlin +data class DCI(val dri: Set<DRI>, val kind: Kind) +``` + +All references to other nodes (other than direct ownership) are described using `DCI`. + +### ContentKind + +`ContentKind` represents a grouping of content of one kind that can be rendered as part of a composite +page, like a single one tab or a block within a class's page. + +For example, on the same page that describes a class you can have multiple sections (== `ContentKind`s). +One to describe functions, one to describe properties, another one to describe the constructors, and so on. + +### Styles + +Each `ContentNode` has a `styles` property in case you want to indicate to the `Renderer` that this content needs to be +rendered in a certain way. + +```kotlin +group(styles = setOf(TextStyle.Paragraph)) { + text("Text1", styles = setOf(TextStyle.Bold)) + text("Text2", styles = setOf(TextStyle.Italic)) +} +``` + +It is responsibility of the `Renderer` (i.e a specific output format) to render it in a way the user can process it. +For instance, `HtmlRenderer` might render `TextStyle.Bold` as `<b>text</b>`, but `CommonmarkRenderer` might render it +as `**text**`. + +There's a number of existing styles that you can use, most of them are supported by the `HtmlRenderer` extension out of +the box: + +```kotlin +// for code highlighting +enum class TokenStyle : Style { + Keyword, Punctuation, Function, Operator, Annotation, + Number, String, Boolean, Constant, Builtin, ... +} + +enum class TextStyle : Style { + Bold, Italic, Strong, Strikethrough, Paragraph, ... +} + +enum class ContentStyle : Style { + TabbedContent, RunnableSample, Wrapped, Indented, ... +} +``` + +### Extra + +`ExtraProperty` is used to store any additional information that falls outside of the regular model. + +It is highly recommended to use extras to provide any additional information when creating custom Dokka plugins. + +All `ExtraProperty` elements from the `Documentable` model are propagated into the `Content` model, and are available +in the `Renderer` extensions. + +This element is a bit complex, so you can read more about how to use it [in a separate section](extra.md). diff --git a/docs-developer/src/doc/docs/developer_guide/architecture/extension_points/base_plugin.md b/docs-developer/src/doc/docs/developer_guide/architecture/extension_points/base_plugin.md new file mode 100644 index 00000000..88579be7 --- /dev/null +++ b/docs-developer/src/doc/docs/developer_guide/architecture/extension_points/base_plugin.md @@ -0,0 +1,33 @@ +# Base plugin + +`DokkaBase` represents Dokka's _Base_ plugin, which provides a number of sensible default implementations for +`CoreExtensions`, as well as declares its own, more high-level abstractions and extension points to be used from other +plugins and output formats. + +If you want to develop a simple plugin that only changes a few details, it is very convenient to rely on +default implementations and use extension points defined in `DokkaBase`, as it reduces the scope of changes you need to make. + +`DokkaBase` is used extensively in Dokka's own output formats. + +You can learn how to add, use, override and configure extensions and extension points in +[Introduction to Extensions](extension_points.md) - all of that information is applicable to the `DokkaBase` plugin as well. + +## Extension points + +Some notable extension points defined in Dokka's Base plugin. + +### PreMergeDocumentableTransformer + +`PreMergeDocumentableTransformer` is very similar to the +[DocumentableTransformer](core_extension_points.md#documentabletransformer) core extension point, but it is used during +an earlier stage by the [Single module generation](generation_implementations.md#singlemodulegeneration). + +This extension point allows you to apply any transformations to the [Documentables model](../data_model/documentable_model.md) +before the project's [source sets](https://kotlinlang.org/docs/multiplatform-discover-project.html#source-sets) are merged. + +It is useful if you want to filter/map existing documentables. For example, if you want to exclude members annotated with +`@Internal`, you most likely need an implementation of `PreMergeDocumentableTransformer`. + +For simple condition-based filtering of documentables, consider extending +`SuppressedByConditionDocumentableFilterTransformer` - it implements `PreMergeDocumentableTransformer` and only +requires one function to be overridden, whereas the rest is taken care of. diff --git a/docs-developer/src/doc/docs/developer_guide/architecture/extension_points/core_extension_points.md b/docs-developer/src/doc/docs/developer_guide/architecture/extension_points/core_extension_points.md new file mode 100644 index 00000000..fc0088c9 --- /dev/null +++ b/docs-developer/src/doc/docs/developer_guide/architecture/extension_points/core_extension_points.md @@ -0,0 +1,103 @@ +# Core extension points + +Core extension points represent the main stages of generating documentation. + +These extension points are plugin and output format independent, meaning it's the very core functionality and as +low-level as can get in Dokka. + +For higher-level extension functions that can be used in different output formats, have a look at the +[Base plugin](base_plugin.md). + +You can find all core extensions in the `CoreExtensions` class: + +```kotlin +object CoreExtensions { + val preGenerationCheck by coreExtensionPoint<PreGenerationChecker>() + val generation by coreExtensionPoint<Generation>() + val sourceToDocumentableTranslator by coreExtensionPoint<SourceToDocumentableTranslator>() + val documentableMerger by coreExtensionPoint<DocumentableMerger>() + val documentableTransformer by coreExtensionPoint<DocumentableTransformer>() + val documentableToPageTranslator by coreExtensionPoint<DocumentableToPageTranslator>() + val pageTransformer by coreExtensionPoint<PageTransformer>() + val renderer by coreExtensionPoint<Renderer>() + val postActions by coreExtensionPoint<PostAction>() +} +``` + +On this page, we'll go over each extension point individually. + +## PreGenerationChecker + +`PreGenerationChecker` can be used to run some checks and constraints. + +For example, Dokka's Javadoc plugin does not support generating documentation for multi-platform projects, so it uses +`PreGenerationChecker` to check for multi-platform +[source sets](https://kotlinlang.org/docs/multiplatform-discover-project.html#source-sets), and fails if it finds any. + +## Generation + +`Generation` is responsible for generating documentation as a whole, utilizing higher-level extensions and extension +points where applicable. + +See [Generation implementations](generation_implementations.md) to learn about the default implementations. + +## SourceToDocumentableTranslator + +`SourceToDocumentableTranslator` translates any given sources into the Documentable model. + +Kotlin and Java sources are supported by default by the [Base plugin](base_plugin.md), but you can analyze any language +as long as you can map it to the [Documentable](../data_model/documentable_model.md) model. + +For reference, see + +* `DefaultDescriptorToDocumentableTranslator` for Kotlin sources translation +* `DefaultPsiToDocumentableTranslator` for Java sources translation + +## DocumentableMerger + +`DocumentableMerger` merges all `DModule` instances into one. Only one extension of this type is expected to be +registered. + +## DocumentableTransformer + +`DocumentableTransformer` performs the same function as `PreMergeDocumentableTransformer`, but after merging source +sets. + +Notable example is `InheritorsExtractorTransformer`, it extracts inheritance information from +[source sets](https://kotlinlang.org/docs/multiplatform-discover-project.html#source-sets) and creates an inheritance +map. + +## DocumentableToPageTranslator + +`DocumentableToPageTranslator` is responsible for creating pages and their content. See +[Page / Content model](../data_model/page_content.md) page for more information and examples. + +Output formats can either use the same page structure or define their own. + +Only a single extension of this type is expected to be registered. + +## PageTransformer + +`PageTransformer` is useful if you need to add, remove or modify generated pages or their content. + +Using this extension point, plugins like `org.jetbrains.dokka:mathjax-pligin` can add `.js` scripts to the HTML pages. + +If you want all overloaded functions to be rendered on the same page instead of separate ones, +you can use `PageTransformer` to combine the pages into a single one. + +## Renderer + +`Renderer` - defines the rules on how to render pages and their content: which files to create and how to display +the content properly. + +Custom output format plugins should use the `Renderer` extension point. Notable examples are `HtmlRenderer` +and `CommonmarkRenderer`. + +## PostAction + +`PostAction` can be used for when you want to run some actions after the documentation has been generated - for example, +if you want to move some files around or log some informational messages. + +Dokka's [Versioning plugin](https://github.com/Kotlin/dokka/tree/master/plugins/versioning) utilizes `PostAction` +to move generated documentation to the versioned directories. + diff --git a/docs-developer/src/doc/docs/developer_guide/architecture/extension_points/extension_points.md b/docs-developer/src/doc/docs/developer_guide/architecture/extension_points/extension_points.md new file mode 100644 index 00000000..97b02a7d --- /dev/null +++ b/docs-developer/src/doc/docs/developer_guide/architecture/extension_points/extension_points.md @@ -0,0 +1,162 @@ +# Extension points + +In this section you can learn how to create new extension points, how to configure existing ones, and how to query for +registered extensions when generating documentation. + +## Declaring extension points + +If you are writing a plugin, you can create your own extension points that other developers (or you) can use in other +plugins / parts of code. + +```kotlin +class MyPlugin : DokkaPlugin() { + val sampleExtensionPoint by extensionPoint<SampleExtensionPointInterface>() +} + +interface SampleExtensionPointInterface { + fun doSomething(input: Input): List<Output> +} + +class Input +class Output +``` + +Usually, you would want to provide some default implementations for your extension points. You can do that +within the same plugin class by extending an extension point you've just created. See +[Extending from extension points](#extending-from-extension-points) for examples. + +## Extending from extension points + +You can use extension points to provide your own implementations in order to customize a plugin's behaviour. + +If you want to provide an implementation for an extension point declared in an external plugin (including `DokkaBase`), +you can use plugin querying API to do that. + +The example below shows how to extend `MyPlugin` (that was created above) with an implementation of +`SampleExtensionPointInterface`. + +```kotlin +class MyExtendedPlugin : DokkaPlugin() { + + val mySampleExtensionImplementation by extending { + plugin<MyPlugin>().sampleExtensionPoint with SampleExtensionImpl() + } +} + +class SampleExtensionImpl : SampleExtensionPointInterface { + override fun doSomething(input: Input): List<Output> = listOf() +} +``` + +Alternatively, if it is your own plugin, you can do that within the same class as the extension point itself: + +```kotlin +open class MyPlugin : DokkaPlugin() { + val sampleExtensionPoint by extensionPoint<SampleExtensionPointInterface>() + + val defaultSampleExtension by extending { + sampleExtensionPoint with DefaultSampleExtension() + } +} + +class DefaultSampleExtension : SampleExtensionPointInterface { + override fun doSomething(input: Input): List<Output> = listOf() +} +``` + +### Providing + +If you need to have access to `DokkaContext` when creating an extension, you can use the `providing` keyword instead. + +```kotlin +val defaultSampleExtension by extending { + sampleExtensionPoint providing { context -> + // can use context to query other extensions or get configuration + DefaultSampleExtension() + } +} +``` + +You can read more on what you can do with `context` in [Obtaining extension instance](#obtaining-extension-instance). + +### Override + +By extending an extension point, you are registering an _additional_ extension. This behaviour is expected by some +extension points, for example the `Documentable` transformers, because all registered transformer extensions do their own +transformations independently and one after the other. + +However, a plugin can expect only a single extension to be registered for an extension point. In this case, you can use +the `override` keyword to override the existing registered extension: + +```kotlin +class MyExtendedPlugin : DokkaPlugin() { + private val myPlugin by lazy { plugin<MyPlugin>() } + + val mySampleExtensionImplementation by extending { + (myPlugin.sampleExtensionPoint + with SampleExtensionImpl() + override myPlugin.defaultSampleExtension) + } +} +``` + +This is also useful if you wish to override some extension from `DokkaBase`, to disable or alter it. + +### Order + +Sometimes, the order in which extensions are invoked matters. This is something you can control as well using the +`order` construct: + +```kotlin +class MyExtendedPlugin : DokkaPlugin() { + private val myPlugin by lazy { plugin<MyPlugin>() } + + val mySampleExtensionImplementation by extending { + myPlugin.sampleExtensionPoint with SampleExtensionImpl() order { + before(myPlugin.firstExtension) + after(myPlugin.thirdExtension) + } + } +} +``` + +### Conditional apply + +If you want your extension to be registered only if some condition is `true`, you can use the `applyIf` construct: + +```kotlin +class MyExtendedPlugin : DokkaPlugin() { + private val myPlugin by lazy { plugin<MyPlugin>() } + + val mySampleExtensionImplementation by extending { + myPlugin.sampleExtensionPoint with SampleExtensionImpl() applyIf { + Random.Default.nextBoolean() + } + } +} +``` + +## Obtaining extension instance + +After an extension point has been [created](#declaring-extension-points) and some extensions have been +[registered](#extending-from-extension-points), you can use `query` and `querySingle` functions to find all or just a +single implementation. + +```kotlin +class MyExtension(context: DokkaContext) { + // returns all registered extensions for the extension point + val allSampleExtensions = context.plugin<MyPlugin>().query { sampleExtensionPoint } + + // will throw an exception if more than one extension is found. + // use if you expect only a single extension to be registered for the extension point + val singleSampleExtensions = context.plugin<MyPlugin>().querySingle { sampleExtensionPoint } + + fun invoke() { + allSampleExtensions.forEach { it.doSomething(Input()) } + + singleSampleExtensions.doSomething(Input()) + } +} +``` + +In order to have access to `DokkaContext`, you can use the [providing](#providing) keyword when registering an extension. diff --git a/docs-developer/src/doc/docs/developer_guide/architecture/extension_points/generation_implementations.md b/docs-developer/src/doc/docs/developer_guide/architecture/extension_points/generation_implementations.md new file mode 100644 index 00000000..3d857fec --- /dev/null +++ b/docs-developer/src/doc/docs/developer_guide/architecture/extension_points/generation_implementations.md @@ -0,0 +1,40 @@ +# Generation implementations + +There are two main implementations of the [Generation](core_extension_points.md#generation) core extension point: + +* `SingleModuleGeneration` - generates documentation for a single module, for instance when `dokkaHtml` task is invoked +* `AllModulesPageGeneration` - generates multi-module documentation, for instance when `dokkaHtmlMultiModule` task is + invoked. + +## SingleModuleGeneration + +`SingleModuleGeneration` is at the heart of generating documentation. It utilizes [core](core_extension_points.md) and +[base](base_plugin.md) extensions to build the documentation from start to finish. + +Below you can see the flow of how Dokka's [data model](../architecture_overview.md#overview-of-data-model) is transformed +by various core and base extensions. + +```mermaid +flowchart TD + Input -- SourceToDocumentableTranslator --> doc1[Documentables] + subgraph documentables [ ] + doc1 -- PreMergeDocumentableTransformer --> doc2[Documentables] + doc2 -- DocumentableMerger --> doc3[Documentables] + doc3 -- DocumentableTransformer --> doc4[Documentables] + end + doc4 -- DocumentableToPageTranslator --> page1[Pages] + subgraph ide2 [ ] + page1 -- PageTransformer --> page2[Pages] + end + page2 -- Renderer --> Output +``` + +You can read about what each stage does in [Core extension points](core_extension_points.md) and +[Base plugin](base_plugin.md). + +## AllModulesPageGeneration + +`AllModulesPageGeneration` utilizes the output generated by `SingleModuleGeneration`. + +Under the hood, it just collects all of the pages generated for individual modules, and assembles it all together, +creating navigation links between the modules and so on. |