Metadoc generates documentation metadata for Object Oriented (Class) JavaScript libraries. Running the utility will produce a JSON file describing the code. This can be used as a data source for creating custom HTML documentation (or any other output format), or for further processing.
Metadoc was designed as command line utility, but can also be used programmatically. It is a custom extension of the productionline build utility (from the same authors of this tool). It was originally designed to document the NGN and Chassis libraries.
Project Update: We are re-engineering this libary. There will still be support for OOP-style JS, but we want to add support for ES modules and other new development patterns. There is no ETA at this time, since our priority is on launching the debut of metadoc.io.
Metadoc was designed to support a "code and comment" workflow. It will parse JavaScript code and extract as much metadata as possible from the code's Abstract Syntax Tree.
AST parsing creates a significant amount of information, but isn't always sufficient for creating detailed documentation, such as class inheritance chains.
To support greater detail, Metadoc reads inline comment blocks, written directly in the code. Comments can be used to supplement and/or override AST parsing. Comment parsing follows a style similar to JSDoc. Using a familiar @tag
syntax, Metadoc provides powerful capabilities for creating fine detail in documentation.
Input Files: Event.js & Meetup.js
Output: api.json
// Install metadoc
npm install -g @author.io/metadoc
// Run metadoc
metadoc --source "/path/to/source_directory" --output "/path/to/output_directory"
If you want to use metadoc programatically (i.e. require('@author.io/metadoc')
), take a look at the cli.js file as an example (which includes the metadoc generator). The metadoc generator is an extension of productionline.
It is possible to ignore files and/or directories using glob syntax.
For example:
--ignore "./node_modules"
ignores the entirenode_modules
directory.--ignore /path/to/**/.*
ignores all files in any directory whose name starts with an dot (ex:.testfile.js
).
It is possible to use the --ignore
flag multiple times.
Metadoc is capable of warning developers about some common code issues/concerns:
-
--warnOnNoCode
triggers a warning whenever a code comment triggers an action for which no related code can be found. This is most useful for identifying comments that shouldn't actually be in the code base. -
--warnOnSkippedEvents
triggers a warning whenever an event is detected but not documented. This is most commonly used to identify events that are considered "internal" to a class. -
--warnOnSkippedTags
triggers a warning whenever a tag is skipped. This is the default behavior, but this tag will allow this feature to be turned off (i.e.--warnOnSkippedTags false
) -
--errorOnCommentFailure
throws an error when a comment cannot be processed. This is the default behavior, but this tag will allow this feature to be turned off (i.e.--errorOnCommentFailure false
)
The code will be automatically documented based on the JavaScript AST (Abstract Syntax Tree). However; this doesn't always reflect the true nature of how a library should be used. To accommodate customizations, the generator parses comments within the code, allowing developers to override the AST documentation with custom comment blocks.
Tags can be used to modify documentation snippets. Tags use the following format unless otherwise defined:
/**
* @tag {[type]} <name>
* <description>
*/
The following tags are available:
Identifies a specific person/organization recognized as the author of a snippet of code.
Identifies a write-only configuration property.
Aliases: config
, configuration
Identifies a configuration property (write-only) that also has a corresponding readable/writable property.
Identifies a class.
Marks a method as the constructor of a class.
Identifies a custom NGN Exception.
Identifies which class is being extended.
Identifies an event. See "Documenting Events" below for additional detail.
Aliases: triggers
, trigger
, event
@hidden
Indicates the section should be included in the documentation but hidden from view. This differs from the @ignore tag, which prevents the documentation from being generated at all.
Indicates a section should be ignored from the documentation (i.e. prevents generation of a segment of code documentation).
Keep information separated from descriptions, such has help comments or tooltips. Use of multiple @info tags are supported.
This tag also supports content titles:
/**
* @info title goes here
* primary content goes here.
*/
Identifies a method.
Identifies a namespace. Namespaces identify class structure/hierarchy and cannot be ignored or hidden using @ignore
or @hidden
.
Identifies an argument/paramenter. See "Documenting Parameters" for details.
Aliases: arg
, argument
, parameter
Indicates the snippet is private/not explicitly accessible as a developer interface (internal).
Identifies a property of a class.
Aliases: prop
Indicates a snippet is read-only. This applies to properties.
Identifies the data returned by a method.
Aliases: returns
This is a special tag that annotates the documentation with a known task that requires completion (a developer to-do task).
Format: @todo Describe the task here
This is a special tag that defines a simple custom type. JavaScript does not enforce types (a weakly typed language). This tag allows developers to document general guidelines for arbitrary simplistic data structures. This is useful when the same type of data is used/expected repeatedly within a code base.
Format:
/**
* @typedef {<type>} <name> (<options>)
* <description>
*/
The <type>
can be any valid JavaScript primitive, such as object
, string
, number
, etc. The <name>
should describe the data type uniquely throughout the entire code base. (<options>)
is an optional list of possible values (enumeration). <description>
is a custom description of the data type.
For example:
/**
* @typedef {Error} MyError
* This is my custom error.
*/
The example above defines a data type called MyError
, which is a custom error.
An example using options:
/**
* @typedef {String} MyLetter (a, b, c)
* Identifies my favorite letter.
*/
This example recognizes a type called MyLetter
, a string, which can have a
, b
, or c
as valid values.
Aliases: @type
Indicates a property is only writable.
In addition to tags, there are a number of recognized flags that can be used to annotate a documentation snippet.
@protected
Identifies a protected method/attribute.@deprecated
Indicates the feature will no longer be available in a future version.@experimental
Indicates the feature is not considered "production ready".@warning
Provides a warning message.@hidden
Indicates the feature should be hidden but not removed from the documentation.@singleton
Indicates a class is a singleton.@interface
Indicates a class is an interface.@static
Indicates a method is static.@since
* Identifies the version and/or date when the feature is generally available. This is typically used to identify new features that have been added to the original platform.
It is also possible to create a custom flag using @flag <flag_name>
.
While parameters (function arguments) in JavaScript can have default values, there are still several cases where it is necessary to provide greater detail about parameters. For example, some methods only accept a parameter value from a predetermined set (enumeration).
Parameters can be documented with additional detail using the following format:
/**
* @param {type} [<parameter_name>=<default>] (<enumerable_list)
* <description>
*/
The type
indicates the data type, while the [
and ]
indicate the parameter is optional. A default value may be supplied, as well as a description.
For example:
/**
* @param {String} [myParameter=example] (example,a,b)
* This is an example parameter.
*/
The example above describes a string parameter named myParameter
. Acceptable (enumerable) values are example
, a
, and b
. The default value is example
. The description is This is an example parameter.
.
Callback functions are a unique type of parameter. These parameters may have their own arguments/parameters. Metadoc supports them using a dot notation syntax:
/**
* @param {function} callback
* This is an example callback.
* @param {boolean} callback.a
* The first element is a.
* @param {string} callback.b (possible,values)
* The next element is b.
*/
The comment above indicates a parameter is a callback method that receives two arguments: a
and b
. The first argument (a
) is a boolean
value. The second (b
) is a string
whose value will be either possible
or values
.
Metadoc was built to document the NGN and Chassis libraries. NGN ships with an event emitter class (works with Node.js events.EventEmitter
). This class is commonly extended, meaning many classes within the library fire events. As a result, metadoc supports documenting the most common event emitter styles, plus those found in NGN.
The following syntax provides a powerful way to generate event documentation overrides:
/**
* @fires {<arg1_name>:<arg1_type>} <event_name>
* <description>
*/
- @fires is the tag. This is required.
<arg_name>
is the optional descriptive name of a callback argument passed to an event handler.<arg_type>
is the data type of the argument passed to an event handler.<event_name>
is the name of the the event that gets fired.<description>
Example:
- Basic Event
/**
* @fires {Object} myEvent
* myEvent is fired from time to time.
*/
this.on('myEvent', function (obj) {
console.log(obj) // Outputs { data: 'abc' }
})
this.emit('myEvent', { data: 'abc' })
This event is called "myEvent", and it sends an object to event handlers.
- Basic Event: Named Arguments
/**
* @fires {myName:Object} myEvent
* myEvent is fired from time to time.
*/
this.on('myEvent', function (obj) {
console.log(obj) // Outputs { data: 'abc' }
})
this.emit('myEvent', { data: 'abc' })
This is the exact same event as the basic event in #1, but the @fires {myName:Object}
will produce a label called "myName", which represents { data: 'abc' }
(payload), a known Object
.
- Complex Event: Multiple Callback Arguments
/**
* @fires {Object,String} myEvent
* myEvent is fired from time to time.
*/
this.on('myEvent', function (obj, label) {
console.log(obj) // Outputs { data: 'abc' }
console.log(label) // Outputs 'event fired'
})
this.emit('myEvent', { data: 'abc' }, 'event fired')
The major difference is the comma separated data types ({Object,String}
), which tells the documentation generator that the event will send two arguments to event handlers. The first is an Object
and the second is String
.
It is possible to document multiple name:type callback arguments by separating with a comma.
@fires {a:Object,b:String}
would generate a label called a
for the Object
argument and a label called b
for the String
object.
It is also possible for an argument to have more than one valid data type by separating types with the pipe |
character. For example, @fires {a:Object|Boolean,b:String}
states that the first argument (labeled a
) can be an Object
or Boolean
value.
- metadoc-md: Convert markdown, mermaid, and mathjax descriptions to HTML.
- metadoc-api: Generate a static JSON API (splits metadoc up into individual JSON files for serving over HTTP).