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.
computeIdsProcessor
- Computes theid
andaliases
for documents using templates or helper functions, on a per docType basis.computePathsProcessor
- Computes thepath
andoutputPath
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 thejsdoc
andngdoc
packages.renderDocsProcessor
- render the documents into a property (doc.renderedContent
) using atemplateEngine
, which must be provided separately - seenunjucks
package.unescapeCommentsProcessor
- unescape comment markers that would break the jsdoc comment style, e.g.*/
writeFilesProcessor
- write the docs that have anoutputPath
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 HTMLtemplateFinder
- 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.
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 }/${ doc.id }.${ doc.docType }.template.html',
'${ doc.area }/${ doc.id }.template.html',
'${ doc.area }/${ doc.docType }.template.html',
'${ doc.id }.${ doc.docType }.template.html',
'${ doc.id }.template.html',
'${ doc.docType }.template.html'
]
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 ofrenderDocsPocessor
.gitRepoInfo
- the owner and repo of the local git repository.packageInfo
- the contents of the package.json.versionInfo
- aggregated version and git information.
An example as used in git/templates/api/api.template.html
<a href='https://github.com/{$ git.info.owner $}/{$ git.info.repo $}/tree/{$ 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"> </i>View Source
</a>
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 atemplateEngine
that uses the Nunjucks template library to render the documents into text, such as HTML or JS, based on templates.
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 atagExtractor
to extract information from the parsed tags.inlineTagsProcessor
- Search the docs for inline tags that need to have content injectedparseTagsProcessor
- use atagParser
to parses the jsdoc tags in the document content.
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
.
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 topublic: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 tagextractTypeTransform
- extract a type from a tagtrimWhitespaceTransform
- trim whitespace from before and after the tag valueunknownTagTransform
- add an error to the tag if it is unknownwholeTagTransform
- Use the whole tag as the value rather than using a tag propertycodeNameService
- helper service forcodeNameProcessor
, 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).
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.
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
.
Matchers:
ArrayExpression
ArrowFunctionExpression
AssignmentExpression
CallExpression
ClassDeclaration
ExportDefaultDeclaration
ExpressionStatement
FunctionDeclaration
FunctionExpression
Identifier
Literal
MemberExpression
MethodDefinition
NewExpression
ObjectExpression
Program
Property
ReturnStatement
ThrowStatement
VariableDeclaration
VariableDeclarator
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.
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 aspackageName
andpackageFileName
; it adds modules to themoduleMap
service and connects all the docs that are in a module to the module doc in thecomponents
property -
providerDocsProcessor
- This processor relates documents about angular services to their corresponding provider document.
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
.
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 docgetDocFromAliases()
- Find a document from thealiasMap
that matches the given aliasgetLinkInfo()
- Get link information to a document that matches the given urlgetTypeClass()
- Get a CSS class string for the given type stringmoduleMap
- 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: '$}'
};
code
- Render a span of text as codelink
- Render a HTML anchor linktypeClass
- Render a CSS class for a given type
code
- Render a block of code
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>
<file name="app.js">
// Some JavaScript code to be included in the example
</file>
</example>
Some text after the example
generateExamplesProcessor
- Add new docs to the docs collection for each example in theexamples
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 theexamples
servicegenerateProtractorTestsProcessor
- 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.
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
.
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
This package provides a HTML post process manager powered by rehype
. It uses the rehype processing engine to manipulate the renderedContent
HTML via rehype "plugins" that work with HTML ASTs (HASTs).
Read more https://github.com/wooorm/rehype
post-process-html
- Use the rehype processing engine to manipulate therenderedContent
HTML via rehype "plugins" that work with HTML ASTs (HASTs). Each plugin is a factory function that will be called with the "rehype" engine asthis
. The factory should return atransform
function that takes a HAST and returns aboolean
orundefined
. The HAST can be mutated by the "transform" function. Iffalse
is returned then the processing stops with that plugin.
- at the moment we are not using a filereader but the
readTypeScriptModules
processor to read our modules.
readTypeScriptModules
- parse thesourceFiles
with the help of thetsParser
service and return a doc for each exported member. You can either pass an array of strings or an array of objects withinclude
andexclude
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 propertyhidePrivateMembers
tofalse
. SetsortClassMembers
totrue
to sort instance and static members by name (defaults to order of appearence). You can ignore special exports by adding strings or regexes to theignoreExportsMatching
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 bycreateCompilerHost
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 existgetContent
- 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).
Please note that at the moment the @param
documentation is ignored.