conventional-commits-parser

Parse raw conventional commits.

Install • Usage • Message format • API • CLI

Install

# pnpm
pnpm add conventional-commits-parser
# yarn
yarn add conventional-commits-parser
# npm
npm i conventional-commits-parser

Usage

import {
  CommitParser,
  parseCommits,
  parseCommitsStream
} from 'conventional-commits-parser'
import { pipeline } from 'stream/promises'
import { Readable } from 'stream'

const rawCommitMessage = 'feat(scope): broadcast $destroy event on scope destruction\nCloses #1'

// to parse raw commit message manually:
const parser = new CommitParser(options)

console.log(parser.parse(rawCommitMessage))

// to parse raw commit messages async iterables:
await pipeline(
  [rawCommitMessage],
  parseCommits(options),
  async function* (parsedCommits) {
    for await (const commit of parsedCommits) {
      console.log(commit)
    }
  }
)

// to parse raw commit messages streams:
Readable.from([rawCommitMessage])
  .pipe(parseCommitsStream(options))
  .on('data', commit => console.log(commit))

Parser expects raw commit message. Examples:

'feat(scope): broadcast $destroy event on scope destruction\nCloses #1'
'feat(ng-list): Allow custom separator\nbla bla bla\n\nBREAKING CHANGE: some breaking change.\nThanks @stevemao\n'

It will return parsed commit object. Examples:

{
  type: 'feat',
  scope: 'scope',
  subject: 'broadcast $destroy event on scope destruction',
  merge: null,
  header: 'feat(scope): broadcast $destroy event on scope destruction',
  body: null,
  footer: 'Closes #1',
  notes: [],
  references:
   [{
     action: 'Closes',
     owner: null,
     repository: null,
     issue: '1',
     raw: '#1',
     prefix: '#'
   }],
  mentions: [],
  revert: null
}
{
  type: 'feat',
  scope: 'ng-list',
  subject: 'Allow custom separator',
  merge: null,
  header: 'feat(ng-list): Allow custom separator',
  body: 'bla bla bla',
  footer: 'BREAKING CHANGE: some breaking change.\nThanks @stevemao',
  notes:
   [{
     title: 'BREAKING CHANGE',
     text: 'some breaking change.\nThanks @stevemao'
   }],
  references: [],
  mentions: ['stevemao'],
  revert: null
}

Conventional Commit Message Format

A minimum input should contain a raw message.

Each commit message consists of a merge header, a header (mandatory), a body and a footer. Mention (optional) someone using the @ notation.

<merge>
<header>
<body>
<footer>

merge

The merge header may optionally have a special format that includes other parts, such as branch, issueId or source.

Merge branch <branch>
Merge pull request <issue-id> from <source>

header

The header may optionally have a special format that includes other parts, such as type, scope and subject. You could reference (optional) issues here.

<type>(<scope>): <subject>

footer

The footer should contain any information about Important Notes (optional) and is also the place to reference (optional) issues.

<important note>
<references>

other parts

This module will only parse the message body. However, it is possible to include other fields such as hash, committer or date.

My commit message
-sideNotes-
It should warn the correct unfound file names.
Also it should continue if one file cannot be found.
Tests are added for these

Then sideNotes will be It should warn the correct unfound file names.\nAlso it should continue if one file cannot be found.\nTests are added for these. You can customize the fieldPattern.

API

`new CommitParser(options?: ParserOptions)`

Creates parser instance with parse method.

`parseCommits(options?: ParserOptions)`

Create async generator function to parse async iterable of raw commits. If there is any malformed commits it will be gracefully ignored (an empty data will be emitted so down async iterable can notice).

`parseCommitsStream(options?: ParserOptions)`

Creates an transform stream. If there is any malformed commits it will be gracefully ignored (an empty data will be emitted so down stream can notice).

ParserOptions

mergePattern

Type: RegExp Default: null

Pattern to match merge headers. EG: branch merge, GitHub or GitLab like pull requests headers. When a merge header is parsed, the next line is used for conventional header parsing.

For example, if we have a commit

Merge pull request #1 from user/feature/feature-name

feat(scope): broadcast $destroy event on scope destruction

We can parse it with these options and the default headerPattern:

{
  mergePattern: /^Merge pull request #(\d+) from (.*)$/,
  mergeCorrespondence: ['id', 'source']
}

mergeCorrespondence

Type: string[], Default: null

Used to define what capturing group of mergePattern.

headerPattern

Type: RegExp, Default: /^(\w*)(?:$([\w\$\.\-\* ]*)$)?\: (.*)$/

Used to match header pattern.

headerCorrespondence

Type: string[], Default ['type', 'scope', 'subject']

Used to define what capturing group of headerPattern captures what header part. The order of the array should correspond to the order of headerPattern's capturing group. If the part is not captured it is null.

referenceActions

Type: string[], Default: ['close', 'closes', 'closed', 'fix', 'fixes', 'fixed', 'resolve', 'resolves', 'resolved']

Keywords to reference an issue. This value is case insensitive.

Set it to null to reference an issue without any action.

issuePrefixes

Type: string[], Default: ['#']

The prefixes of an issue. EG: In gh-123 gh- is the prefix.

issuePrefixesCaseSensitive

Type: boolean, Default: false

Used to define if issuePrefixes should be considered case sensitive.

noteKeywords

Type: string[], Default: ['BREAKING CHANGE', 'BREAKING-CHANGE']

Keywords for important notes. This value is case insensitive.

notesPattern

Type: function, Default: noteKeywordsSelection => ^[\\s|*]*(' + noteKeywordsSelection + ')[:\\s]+(.*) where noteKeywordsSelection is join(noteKeywords, '|')

A function that takes noteKeywordsSelection and returns a RegExp to be matched against the notes.

fieldPattern

Type: RegExp, Default: /^-(.*?)-$/

Pattern to match other fields.

revertPattern

Type: RegExp, Default: /^Revert\s"([\s\S]*)"\s*This reverts commit (\w*)\./

Pattern to match what this commit reverts.

revertCorrespondence

Type: string[], Default: ['header', 'hash']

Used to define what capturing group of revertPattern captures what reverted commit fields. The order of the array should correspond to the order of revertPattern's capturing group.

For example, if we had commit

Revert "throw an error if a callback is passed"

This reverts commit 9bb4d6c.

If configured correctly, the parsed result would be

{
  revert: {
    header: 'throw an error if a callback is passed',
    hash: '9bb4d6c'
  }
}

It implies that this commit reverts a commit with header 'throw an error if a callback is passed' and hash '9bb4d6c'.

commentChar

Type: string or null, Default: null

What commentChar to use. By default it is null, so no comments are stripped. Set to # if you pass the contents of .git/COMMIT_EDITMSG directly.

If you have configured the git commentchar via git config core.commentchar you'll want to pass what you have set there.

warn

Type: function or boolean, Default: () => {}

What warn function to use. For example, console.warn.bind(console). By default, it's a noop. If it is true, it will error if commit cannot be parsed (strict).

CLI

You can use cli to practice writing commit messages or parse messages from files. Note: the sample output might be different. It's just for demonstration purposes.

If you run conventional-commits-parser without any arguments

$ conventional-commits-parser --help # for more details

You will enter an interactive shell. To show your parsed output enter "return" three times (or enter your specified separator).

> fix(title): a title is fixed


{"type":"fix","scope":"title","subject":"a title is fixed","header":"fix(title): a title is fixed","body":null,"footer":null,"notes":[],"references":[],"revert":null}

You can also use cli to parse messages from files.

If you have log.txt

feat(ngMessages): provide support for dynamic message resolution

Prior to this fix it was impossible to apply a binding to a the ngMessage directive to represent the name of the error.

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.

Closes #10036
Closes #9338

And you run

$ conventional-commits-parser log.txt
# or
$ cat log.txt | conventional-commits-parser

An array of json will be printed to stdout.

[
{"type":"feat","scope":"ngMessages","subject":"provide support for dynamic message resolution","header":"feat(ngMessages): 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"}],"revert":null}
]

Commits should be split by at least three newlines (\n\n\n) or you can specify a separator as the second argument.

Eg: in log2.txt


docs(ngMessageExp): split ngMessage docs up to show its alias more clearly
===

fix($animate): applyStyles from options on leave

Closes #10068