Skip to content

Latest commit

 

History

History
197 lines (144 loc) · 8.96 KB

README.md

File metadata and controls

197 lines (144 loc) · 8.96 KB
Raw

mention-bot Build Status

Do you have a GitHub project that is too big for people to subscribe to all the notifications? The mention bot will automatically mention potential reviewers on pull requests. It helps getting faster turnaround on pull requests by involving the right people early on.

How To Use?

  • Go to
  • your project on GitHub > Settings > Webhooks & services > Add Webhook or
  • your organization on GitHub > Settings > Webhooks > Add Webhook
  • Payload URL: (https://mention-bot.herokuapp.com/)
  • Let me select individual events > Check Pull Request
  • Add Webhook

And you are done. Next time a pull request is opened, you should see the bot add a comment ;)

Configuration

The bot can be configured by adding a .mention-bot file to the base directory of the repo. Here's a list of the possible options:

{
  "maxReviewers": 5, // Maximum  number of people to ping in the PR message, default is 3
  "numFilesToCheck": 10, // Number of files to check against, default is 5
  "message": "@pullRequester, thanks! @reviewers, please review this.",
             // custom message using @pullRequester and @reviewers
  "alwaysNotifyForPaths": [
    {
      "name": "ghuser", // The user's Github username
      "files": ["src/js/**/*.js"] // The array of file globs associated with the user
    }
  ], // users will always be mentioned based on file glob
  "fallbackNotifyForPaths": [
    {
      "name": "ghuser", // The user's Github username
      "files": ["src/js/**/*.js"] // The array of file globs associated with the user
    }
  ], // users will be mentioned based on file glob if no other user was found
  "findPotentialReviewers": true, // mention-bot will try to find potential reviewers based on files history, if disabled, `alwaysNotifyForPaths` is used instead
  "fileBlacklist": ["*.md"], // mention-bot will ignore any files that match these file globs
  "userBlacklist": [], // users in this list will never be mentioned by mention-bot
  "userBlacklistForPR": [], // PR made by users in this list will be ignored
  "requiredOrgs": [], // mention-bot will only mention user who are a member of one of these organizations
  "actions": ["opened"], // List of PR actions that mention-bot will listen to, default is "opened"
  "skipAlreadyAssignedPR": false, // mention-bot will ignore already assigned PR's
  "skipAlreadyMentionedPR": false, // mention-bot will ignore if there is already existing an exact mention
  "assignToReviewer": false, // mention-bot assigns the most appropriate reviewer for PR
  "skipTitle": "", // mention-bot will ignore PR that includes text in the title,
  "withLabel": "", // mention-bot will only consider PR's with this label. Must set actions to ["labeled"].
  "delayed": false, // mention-bot will wait to comment until specified time in `delayedUntil` value
  "delayedUntil": "3d", // Used if delayed is equal true, permitted values are: minutes, hours, or days, e.g.: '3 days', '40 minutes', '1 hour', '3d', '1h', '10m'
  "skipCollaboratorPR": false, // mention-bot will ignore if PR is made by collaborator
}

The glob matching is an extended form of glob syntax performed by minimatch, with the default options; read the minimatch README for more details.

How Does It Work?

Every time there's a new pull request, GitHub wakes up the mention bot using Webhooks.

Once awakened, the bot will download the diff of the pull request and figure out which files and lines have been touched.

For these, it will download the associated blame to figure out who last touched that line, as they may be a good reviewer.

After running the algorithm described in the next section, it will comment on the pull request notifying those people and go back to sleep.

Algorithm

The problem of finding who the best reviewers are is really hard and I don't think that any algorithm will achieve perfection. Instead, what we want here is to be best effort. We want to notify people that are likely going to be interested and be good reviewers. If we ping a few too many people that's not the end of the world neither if we don't ping the exact right person.

We use two heuristics:

  • If a line was deleted or modified, the person that last touched that line is likely going to care about this pull request.
  • If a person last touched many lines in the file where the change was made, they will want to be notified.

Initialization

Create two empty hash map:

  • DeletedLines for authors of deleted lines.
  • AllLines for authors of lines in the changed files.

Filling the data structures

  • for each deleted line, find the author in the blame and increase its count by one in the DeletedLines map.
  • for each line in every file that was changed, find the author in the blame and increase its count by one in the AllLines map.

Since getting the blame information is sending an http request to GitHub it is pretty expensive. We first sort the files by number of deleted lines and only pick the top 5. Since we're only looking for 3 names and the algorithm is best effort, this greatly speeds up the algorithm in case of large pull requests for little loss in precision.

Putting it all together

  • delete from AllLines all the names that appear in both maps. We don't want to mention the same person twice.
  • sort each map by count
  • concat DeletedLines with AllLines
  • take the first three names

How To Contribute or Run Your Own Bot?

If you want to use a different account for the bot, change the message or extend it with more functionalities, we've tried to make it super easy:

git clone https://github.com/facebook/mention-bot.git
cd mention-bot
npm install
npm start
# Follow the instructions there

Alternatively, click the button below:

Deploy

If you would like the mention-bot to function on private repositories, set the GITHUB_USER and GITHUB_PASSWORD environment variables. You must disable two-factor authentication or you will receive a console log like this: Login to ${USERNAME} failed.

You can also set a REQUIRED_ORG environment variable, so you don't have to configure it in each repository of your organization.

You can also build deploy it as a Docker image:

docker build -t mention-bot .
docker run -e GITHUB_USER="a" -p 5000:5000  mention-bot

Configuring a custom message

If you want to change the default message, you can write your custom logic in message.js, or add 'message' in the .mention-bot configuration file.

How to run the bot on GitHub Enterprise

If you want to run the bot on your GitHub Enterprise instance, add the GHE host and path prefix to the config section of package.json

"config": {
  "gheHost": "github.my-GHE-enabled-company.com",
  "ghePathPrefix": "/api/v3"
}

If you use http protocol, the config section like this:

"config": {
  "gheHost": "github.my-GHE-enabled-company.com",
  "ghePathPrefix": "/api/v3",
  "gheProtocol": "http",
  "ghePort": 80
}

Programmatic API

When you require mention-bot you will get all the functions exposed by mention-bot.js module. You are expected to manage your own server and also connection to the github repository.

npm install mention-bot github

API can be used like this:

var mentionBot = require('mention-bot');
var GitHubApi = require('github');

var github = new GitHubApi({ version: '3.0.0' });
github.authenticate({
  type: 'oauth',
  token: '...token...'
});

mentionBot
  .guessOwnersForPullRequest(
    'https://github.com/fbsamples/bot-testing', // repo
    65, // pull request number
    'mention-bot', // user that created the pull request
    'master', // branch
    { maxReviewers: 3 }, // config
    github
  )
  .then(function(users) {
    // array with user names which should be included in review
    console.log(users);
  })
  .catch(function(err) {
    console.error(err);
  });

License

mention-bot is BSD-licensed. We also provide an additional patent grant.