node package manager


A collection of dgeni packages for generating documentation from source code

Dgeni Packages

This repository contains a collection of Dgeni Packages that can be used by the Dgeni documentation generator to create documentation from source code.

Out of the box there are the following packages:

  • base - The minimal set of processors to get started with Dgeni
  • git - Provides some git and version information
  • jsdoc - Tag parsing and extracting
  • nunjucks - The nunjucks template rendering engine. No longer in jsdoc - you must add this explicitly to your config or you will get Error: No provider for "templateEngine"! (Resolving: templateEngine)
  • ngdoc - The angular.js specific tag-defs, processors and templates. This loads the jsdoc and nunjucks packages for you.
  • examples - Processors to support the runnable examples feature in the angular.js docs site.
  • dgeni - Support for documenting Dgeni packages (incomplete)
  • typescript - Tag parsing and extracting for TypeScript modules.

base Package


  • computeIdsProcessor - Computes the id and aliases for documents using templates or helper functions, on a per docType basis.
  • computePathsProcessor - Computes the path and outputPath for documents using templates or helper functions, on a per docType basis.
  • debugDumpProcessor - dump the current state of the docs array to a file (disabled by default)
  • readFilesProcessor - used to load up documents from files. This processor can be configured to use a set of file readers. There are file readers in the jsdoc and ngdoc packages.
  • renderDocsProcessor - render the documents into a property (doc.renderedContent) using a templateEngine, which must be provided separately - see nunjucks package.
  • unescapeCommentsProcessor - unescape comment markers that would break the jsdoc comment style, e.g. */
  • writeFilesProcessor - write the docs that have an outputPath to disk


  • aliasMap - A map of ids/aliases to docs. This is used for matching references to documents in links and relations such as modules and object members.
  • createDocMessage - a helper for creating nice messages about documents (useful in logging and errors)
  • encodeDocBlock - convert a block of code into HTML
  • templateFinder - search folders using patterns to find a template that matches a given document.
  • trimIndentation - "intelligently" trim whitespace indentation from the start of each line of a block of text.
  • writeFile - Write some contents to a file, ensuring the path to the file exists.

Template Finding

The template used to render a doc is computed by the templateFinder, which uses the first match from a set of patterns in a set of folders, provided in the configuration. This allows a lot of control to provide generic templates for most situations and specific templates for exceptional cases.

Here is an example of some standard template patterns:

templateFinder.templatePatterns = [
  '${ doc.template }',
  '${doc.area}/${ }.${ doc.docType }.template.html',
  '${doc.area}/${ }.template.html',
  '${doc.area}/${ doc.docType }.template.html',
  '${ }.${ doc.docType }.template.html',
  '${ }.template.html',
  '${ doc.docType }.template.html'

git Package

This package provides some git and version information to the renderDocsPocessor that is available in the templates. This code as it is was made for the angular.js document generation, including some custom logic for special versions. However, any of the services can be overridden with custom behavior.

The git information is made available to templates via the extraData.git property. See the section below to see an example usage.


  • decorateVersion - all semvers are passed through this function so that additional data can before added to them.
  • getPreviousVersions - pulls versions from git tags of the repository.
  • gitData - the additional information that is added to the extraData of renderDocsPocessor.
  • gitRepoInfo - the owner and repo of the local git repository.
  • packageInfo - the contents of the package.json.
  • versionInfo - aggregated version and git information.

Using extraData.git

An example as used in git/templates/api/api.template.html

<a href='{$ $}/{$ $}/tree/{$ git.version.isSnapshot and 'master' or git.version.raw $}/{$ doc.fileInfo.projectRelativePath $}#L{$ doc.startingLine $}' class='view-source pull-right btn btn-primary'>
  <i class="glyphicon glyphicon-zoom-in">&nbsp;</i>View Source

nunjucks Package

This package provides a nunjucks driven implementation of the templateEngine required by the base package renderDocsPocessor. The "nunjucks" JavaScript template tool-kit to generates HTML based on the data in each document. We have nunjucks templates, tags and filters that can render links and text as markdown and will highlight code.


  • nunjucks-template-engine - provide a templateEngine that uses the Nunjucks template library to render the documents into text, such as HTML or JS, based on templates.

jsdoc Package

File Readers:

  • jsdoc - can read documents from jsdoc style comments in source code files.


  • codeNameProcessor - infer the name of the document from the code following the document in the source file.
  • extractTagsProcessor - use a tagExtractor to extract information from the parsed tags.
  • inlineTagsProcessor - Search the docs for inline tags that need to have content injected
  • parseTagsProcessor - use a tagParser to parses the jsdoc tags in the document content.

Tag Definitions

The jsdoc package contains definitions for a number of standard jsdoc tags including: name, memberof, param, property, returns, module, description, usage, animations, constructor, class, classdesc, global, namespace, method, type, kind, access, public, private and protected.

Services (Tag Transformations)

This package provides a number of Transform services that are used in Tag Definitions to transform the value of the tag from the string in the tag description to something more meaningful in the doc.

  • extractAccessTransform - extract an access level (e.g. public, protected, private) from tags You can configure this transform to register access tags and set the property where access info is written.
    • extractAccessTransform.allowedTags.set('tagName', [propertValue]) - register a tag that can act as as an alias to set an access level. The propertyValue is optional and if not undefined will return this value from the transform that will be written to the property. (defaults to public:undefined, private:undefined, protected:undefined)
    • extractAccessTransformImpl.allowedDocTypes.set('docType') - register a docType that can contain access type tags (defaults to "property" and "method")
    • extractAccessTransformImpl.accessProperty - specify the property to which to write the access value (defaults to "access")
    • extractAccessTransformImpl.accessTagName - specify the name of the tag that can hold access values (defaults to "access")
  • extractNameTransform - extract a name from a tag
  • extractTypeTransform - extract a type from a tag
  • trimWhitespaceTransform - trim whitespace from before and after the tag value
  • unknownTagTransform - add an error to the tag if it is unknown
  • wholeTagTransform - Use the whole tag as the value rather than using a tag property
  • codeNameService - helper service for codeNameProcessor, registers code name matchers and performs actual matches against AST tree


This package does not provide any templates nor a templateEngine to render templates (use the nunjucks package to add this).

Tag Definitions

This package provides a minimal implementation of tags from the JSDoc project. They extract the name and type from the tag description accordingly but do not fully implement all the JSDoc tag functionality.

Code Name Matchers

Matcher performs a search for a suitable code name at the given jsdoc code point (AST node). codeNameService matches AST node name against matcher name and if suitable matcher is found, executes it.

Matcher name consists of <AstNodeName> and NodeMatcher substrings, i.e. FunctionExpressionNodeMatcher then latter is stripped and matcher is used by the former part, i.e. FunctionExpression.

Matcher should accept single argument - node and return either string with name or literal null.


  • ArrayExpression
  • ArrowFunctionExpression
  • AssignmentExpression
  • CallExpression
  • ClassDeclaration
  • ExportDefaultDeclaration
  • ExpressionStatement
  • FunctionDeclaration
  • FunctionExpression
  • Identifier
  • Literal
  • MemberExpression
  • MethodDefinition
  • NewExpression
  • ObjectExpression
  • Program
  • Property
  • ReturnStatement
  • ThrowStatement
  • VariableDeclaration
  • VariableDeclarator

ngdoc Package

The ngdoc Package depends upon the jsdoc and nunjucks packages. It provides additional support for non-API documents written in files with .ngdoc extension; it also computes additional properties specific to Angular related code.

File Readers

  • ngdoc - can pull a single document from an ngdoc content file.


  • filterNgdocsProcessor - For AngularJS we are only interested in documents that contain the @ngdoc tag. This processor removes docs that do not contain this tag.

  • generateComponentGroupsProcessor - Generate documents for each group of components (by type) within a module

  • memberDocsProcessor - This processor connects docs that are members (properties, methods and events) to their container docs, removing them from the main docs collection.

  • moduleDocsProcessor - This processor computes properties for module docs such as packageName and packageFileName; it adds modules to the moduleMap service and connects all the docs that are in a module to the module doc in the components property

  • providerDocsProcessor - This processor relates documents about angular services to their corresponding provider document.

Tag Definitions

This package modifies and adds new tag definitions on top of those provided by the jsdoc package: area, element, eventType, example, fullName, id, module, name, ngdoc, packageName, parent, priority, restrict, scope and title.

Inline Tag Definitions

  • link - Process inline link tags (of the form {@link some/uri Some Title}), replacing them with HTML anchors


  • getAliases() - Get a list of all the aliases that can be made from the provided doc
  • getDocFromAliases() - Find a document from the aliasMap that matches the given alias
  • getLinkInfo() - Get link information to a document that matches the given url
  • getTypeClass() - Get a CSS class string for the given type string
  • moduleMap - A collection of modules keyed on the module id


This package provides a set of templates for generating an HTML file for each document: api, directive, error, filter function, input, module, object, overview, provider, service, type and a number to support rendering of the runnable examples.

You should be aware that because of the overlap in syntax between Nunjucks bindings and AngularJS bindings, the ngdoc package changes the default Nunjucks binding tags:

templateEngine.config.tags = {
  variableStart: '{$',
  variableEnd: '$}'

Rendering Filters

  • code - Render a span of text as code
  • link - Render a HTML anchor link
  • typeClass - Render a CSS class for a given type

Rendering Tags

  • code - Render a block of code

examples Package

This package is a mix-in that provides functionality for working with examples in the docs.

Inside your docs you can markup inline-examples such as:

Some text before the example
<example name="example-name">
  <file name="index.html">
    <div>The main HTML for the example</div>
  <file name="app.js">
    // Some JavaScript code to be included in the example
Some text after the example


  • generateExamplesProcessor - Add new docs to the docs collection for each example in the examples service that will be rendered as files that can be run in the browser, for example as live in-place demos of the examples or for e2e testing. This processor must be configured with a collection of deployments that tell it what versions of each example to generate. See the section of Deployment Configuration below.
  • parseExamplesProcessor - Parse the <example> tags from the content and add them to the examples service
  • generateProtractorTestsProcessor - Generate a protractor test files from the e2e tests in the examples. This processor must be configured with a collection of deployments that tell versions of the protractor tests to generate. See the section of Deployment Configuration below.

Deployment Configuration

The generateExamplesProcessor and generateProtractorTestsProcessor processors have a required property called deployments. This property should be an array of deployment information objects telling the processor what files to generate.

For instance you might have a "debug" deployment that loads angular.js into the example, and also a "default" deployment that loads angular.min.js into the example. Equally you might have deployments that use JQuery and some that only use Angular's jqLite.

You can configure this in your package like so:

.config(function(generateExamplesProcessor, generateProtractorTestsProcessor) {
  var deployments = [
    { name: 'debug', ... },
    { name: 'default', ... }
  generateExamplesProcessor.deployments = deployments;
  generateProtractorTestsProcessor.deployments = deployments;

A deployment can must have a name property and can also include an examples property that contains information about paths and extra files to inject into runtime examples. Further a protractor test is generated for each deployment and it uses the deployment name to find the path to the associated example for that deployment.

  name: 'default',
  examples: {
    commonFiles: {
      scripts: [ '../../../angular.js' ]
    dependencyPath: '../../../'

Here you can see we have a default deployment that injects the angular.js file into all examples, plus any dependencies referenced in the example itself are made relative to the given dependencyPath.

Inline Tag Definitions

  • runnableExample - Inject the specified runnable example into the doc


  • exampleMap - a hash map holding each example by id, which is a unique id generated from the name of the example

typescript Package

File Readers:

  • at the moment we are not using a filereader but the readTypeScriptModules processor to read our modules.


  • readTypeScriptModules - parse the sourceFiles with the help of the tsParser service and return a doc for each exported member. You can either pass an array of strings or an array of objects with include and exclude globbing patterns. A mix of both is possible as well. The processor can be configured to export private members (marked as /** @internal */ as well as members starting with an underscore (_)) by setting the property hidePrivateMembers to false. Set sortClassMembers to true to sort instance and static members by name (defaults to order of appearence). You can ignore special exports by adding strings or regexes to the ignoreExportsMatching property (defaults to ___esModule.


  • convertPrivateClassesToInterfaces - pass this service a list of exported docs and if it represents a class that is marked as /** @internal */ the doc will be converted to represent an interface.
  • tsParser - uses the typescript compiler and a host created by createCompilerHost to actually read and compile the source files. The docs are created from the symbols read by the typescript program.
  • createCompilerHost - creates a new compiler host which can, among other things, resolve file paths and check if files exist
  • getContent - retrieves the file contents and comments.


This package does not provide any templates nor a templateEngine to render templates (use the nunjucks package to add this).

Tag Definitions

Please note that at the moment the @param documentation is ignored.