Write logs based on conventional commits and templates
Install
$ npm install --save conventional-changelog-writer
Usage
var conventionalChangelogWriter = require('conventional-changelog-writer');
conventionalChangelogWriter(context, options);
It returns a transform stream.
It expects an object mode upstream that looks something like this:
{ hash: '9b1aff905b638aa274a5fc8f88662df446d374bd',
header: 'feat(scope): broadcast $destroy event on scope destruction',
type: 'feat',
scope: 'scope',
subject: 'broadcast $destroy event on scope destruction',
body: null,
footer: 'Closes #1',
notes: [],
references: [ { action: 'Closes', owner: null, repository: null, issue: '1', raw: '#1' } ] }
{ hash: '13f31602f396bc269076ab4d389cfd8ca94b20ba',
header: 'feat(ng-list): Allow custom separator',
type: 'feat',
scope: 'ng-list',
subject: 'Allow custom separator',
body: 'bla bla bla',
footer: 'BREAKING CHANGE: some breaking change',
notes: [ { title: 'BREAKING CHANGE', text: 'some breaking change' } ],
references: [] }
Each chunk should be a commit. Json object is also valid. Parts of the objects will be formatted and combined into a log based on the handlebars context, templates and options.
The downstream might look something like this:
<a name="0.0.1"></a>
## 0.0.1 "this is a title" (2015-05-29)
### Features
* **ng-list:** Allow custom separator ([13f3160](https:
* **scope:** broadcast $destroy event on scope destruction ([9b1aff9](https:
### BREAKING CHANGES
* some breaking change
API
conventionalChangelogWriter([context, [options]])
Returns a transform stream.
context
Variables that will be interpolated to the template. This object contains, but not limits to the following fields.
version
Type: string
Version number of the up-coming release. If version
is found in the last commit before generating logs, it will be overwritten.
title
Type: string
isPatch
Type: boolean
Default: semver.patch(context.version) !== 0
By default, this value is true if version
's patch is 0
.
host
Type: string
The hosting website. Eg: 'https://github.com'
or 'https://bitbucket.org'
owner
Type: string
The owner of the repository. Eg: 'stevemao'
.
repository
Type: string
The repository name on host
. Eg: 'conventional-changelog-writer'
.
repoUrl
Type: string
The whole repository url. Eg: 'https://github.com/conventional-changelog/conventional-changelog-writer'
.
The should be used as a fallback when context.repository
doesn't exist.
linkReferences
Type: boolean
Default: true
if (context.repository
or context.repoUrl
), context.commit
and context.issue
are truthy
Should all references be linked?
commit
Type: string
Default: 'commits'
Commit keyword in the url if context.linkReferences === true
.
issue
Type: string
Default: 'issues'
Issue or pull request keyword in the url if context.linkReferences === true
.
date
Type: string
Default: dateFormat(new Date(), 'yyyy-mm-dd', true)
Default to formatted ('yyyy-mm-dd'
) today's date. dateformat is used for formatting the date. If version
is found in the last commit, committerDate
will overwrite this.
options
Type: object
transform
Type: object
or function
Default: get the first 7 digits of hash, and committerDate
will be formatted as 'yyyy-mm-dd'
.
Replace with new values in each commit.
If this is an object, the keys are paths to a nested object property. the values can be a string (static) and a function (dynamic) with the old value and path passed as arguments. This value is merged with your own transform object.
If this is a function, the commit chunk will be passed as the argument and the returned value would be the new commit object. This is a handy function if you can't provide a transform stream as an upstream of this one. If returns a falsy value this commit is ignored.
a raw
object that is originally poured form upstream is attached to commit
.
groupBy
Type: string
Default: 'type'
How to group the commits. EG: based on the same type. If this value is falsy, commits are not grouped.
commitGroupsSort
Type: function
, string
or array
A compare function used to sort commit groups. If it's a string or array, it sorts on the property(ies) by localeCompare
. Will not sort if this is a falsy value.
The string can be a dot path to a nested object property.
commitsSort
Type: function
, string
or array
Default: 'header'
A compare function used to sort commits. If it's a string or array, it sorts on the property(ies) by localeCompare
. Will not sort if this is a falsy value.
The string can be a dot path to a nested object property.
noteGroupsSort
Type: function
, string
or array
Default: 'title'
A compare function used to sort note groups. If it's a string or array, it sorts on the property(ies) by localeCompare
. Will not sort if this is a falsy value.
The string can be a dot path to a nested object property.
notesSort
Type: function
, string
or array
Default: 'text'
A compare function used to sort note groups. If it's a string or array, it sorts on the property(ies) by localeCompare
. Will not sort if this is a falsy value.
The string can be a dot path to a nested object property.
generateOn
Type: function
, string
or any
Default: if commit.version
is a valid semver.
When the upstream finishes pouring the commits it will generate a block of logs if doFlush
is true
. However, you can generate more than one block based on this criteria (usually a version) even if there are still commits from the upstream.
generateOn(commit, commits, context, options)
####### commit
Current commit.
####### commits
Current collected commits.
####### context
The generated context based on original input context
and options
.
####### options
Normalized options.
NOTE: It checks on the transformed commit chunk instead of the original one (you can check on the original by access the raw
object on the commit
). However, if the transformed commit is ignored it falls back to the original commit.
If this value is a string
, it checks the existence of the field. Set to other type to disable it.
finalizeContext
Type: function
Default: pass through
Last chance to modify your context before generating a changelog.
finalizeContext(context, options, commits, keyCommit)
####### context
The generated context based on original input context
and options
.
####### options
Normalized options.
####### commits
Filtered commits from your git metadata.
####### keyCommit
The commit that triggers to generate the log.
debug
Type: function
Default: function() {}
A function to get debug information.
reverse
Type: boolean
Default: false
The normal order means reverse chronological order. reverse
order means chronological order. Are the commits from upstream in the reverse order? You should only worry about this when generating more than one blocks of logs based on generateOn
. If you find the last commit is in the wrong block inverse this value.
includeDetails
Type: boolean
Default: false
If this value is true
, instead of emitting strings of changelog, it emits objects containing the details the block.
NOTE: The downstream must be in object mode if this is true
.
ignoreReverted
Type: boolean
Default: true
If true
, reverted commits will be ignored.
doFlush
Type: boolean
Default: true
If true
, the stream will flush out the last bit of commits (could be empty) to changelog.
mainTemplate
Type: string
Default: template.hbs
The main handlebars template.
Type: string
Default: header.hbs
commitPartial
Type: string
Default: commit.hbs
Type: string
Default: footer.hbs
partials
Type: object
Partials that used in the main template, if any. The key should be the partial name and the value should be handlebars template strings. If you are using handlebars template files, read files by yourself.
Customization Guide
It is possible to customize this the changelog to suit your needs. Templates are written in handlebars. You can customize all partials or the whole template. Template variables are from either upstream
or context
. The following are a suggested way of defining variables.
upstream
Variables in upstream are commit specific and should be used per commit. Eg: commit date and commit username. You can think of them as "local" or "isolate" variables. A "raw" commit message (original commit poured from upstream) is attached to commit
. transform
can be used to modify a commit.
context
context should be module specific and can be used across the whole log. Thus these variables should not be related to any single commit and should be generic information of the module or all commits. Eg: repository url and author names, etc. You can think of them as "global" or "root" variables.
Basically you can make your own templates and define all your template context. Extra context are based on commits from upstream and options
. For more details, please checkout handlebars and the source code of this module. finalizeContext
can be used at last to modify context before generating a changelog.
CLI
$ npm install --global conventional-changelog-writer
$ conventional-changelog-writer --help
It works with Line Delimited JSON.
If you have commits.ldjson
{"hash":"9b1aff905b638aa274a5fc8f88662df446d374bd","header":"feat(ngMessages): provide support for dynamic message resolution","type":"feat","scope":"ngMessages","subject":"provide support for dynamic message resolution","body":"Prior to this fix it was impossible to apply a binding to a the ngMessage directive to represent the name of the error.","footer":"BREAKING CHANGE: The `ngMessagesInclude` attribute is now its own directive and that must be placed as a **child** element within the element with the ngMessages directive.\nCloses #10036\nCloses #9338","notes":[{"title":"BREAKING CHANGE","text":"The `ngMessagesInclude` attribute is now its own directive and that must be placed as a **child** element within the element with the ngMessages directive."}],"references":[{"action":"Closes","owner",null,"repository":null,"issue":"10036","raw":"#10036"},{"action":"Closes","owner":null,"repository":null,"issue":"9338","raw":"#9338"}]}
And you run
$ conventional-changelog-writer commits.ldjson -o options.js
The output might look something like this
<a name="1.0.0"></a>
# 1.0.0 (2015-04-09)
### Features
* **ngMessages:** provide support for dynamic message resolution 9b1aff9, closes #10036 #9338
### BREAKING CHANGES
* The `ngMessagesInclude` attribute is now its own directive and that must be placed as a **child** element within the element with the ngMessages directive.
It is printed to stdout.
License
MIT © Steve Mao