neat
TypeScript icon, indicating that this package has built-in type declarations

2.1.0 • Public • Published

neat

Neat is a CLI tool and a collection of the neatest repository templates to boost your repos.

Table of Contents

💾 Installation

Install with:

  • NPM

    npm install -g neat
  • Yarn

    yarn global add neat
  • Or use directly with NPX

    npx neat

🔥 CLI usage

Neat is very powerful, as it allows you to enhance existing repos or create new repos by:

  • downloading files
  • asking questions
  • replacing strings with other strings, files or command outputs
  • running pre-defined commands or scripts

The default behaviour is to just download files, but Neat repos can specify a .neat.yml which can virtually do anything when a user "neats" a repo.

Neat is safe to run in an existing directory because it will never change existing files unless you use the force flags. That said, always inspect the configuration file before neating a repo to spot anything that could be harmful to your system or your files (potential threats will be highlighted).

There is a collection of neat repos in neat-repos.json (See contributing if you want to submit yours). Although we look at the repo at the time of adding it to the list, we cannot vouch for any changes committed after that.

⚠️ As a general rule (not just for Neat), you should never execute a remote file without prior verification because it could have been tampered with malicious code. As such, it is always recommended to execute remote files in a controlled environment such as a remote CI environment or a local docker container to contain eventual damage.

Use a "registered" repo

Use a repo name from the neat-repos.json

Download files in the current working directory (without overwriting existing files):

neat repo

By default the master branch is used, if you prefer another branch/tag, you can use the @ notation:

neat repo@emoji

Use any GitHub repo

Download files in the current working directory (without overwriting existing files):

neat your/repo

By default the master branch is used, if you prefer another branch/tag, you can use the @ notation:

neat your/repo@v2

Some repos may have a .neat.yml config file to change their behaviour when they are "neated" but this file is optional.

Specify a target folder

Download files in my-project (without overwriting existing files):

neat repo my-project

This is usually used for creating a new repo using a neat template

Inspect a repo configuration

You can easily inspect a configuration file before neating repo by running:

neat inspect repo

Options

--force-download

The default behaviour is to skip processing remote files that already exist locally.

Using this flag will overwrite all local files with their remote counterparts.

neat repo --force-download

--force-inject

The default behaviour is to skip injecting local files that existed already before running Neat.

Using this flag will force injecting those files.

neat repo --force-inject

This flag is quite commonly used, its default behavior is a safety measure to prevent modification of existing files and is inline with the philosophy of Neat not being invasive on an existing repo unless specifically told.

-f, --force

If you wish to use both --force-download and --force-inject, you can use the alias --forceor -f

Note on future versions of Neat: As new features will be added to Neat, using --force will always be the same as combining all the --force-* flags. Keep that in mind if you use Neat in automation or if you're a heavy neat user.

So, in the current version, running this:

neat repo --force

Is the same as running this:

neat repo --force-download --force-inject

-s, --silent

Don't ask for any user input. If this flag is used Neat will skip any questions asked by a template and will use answers from NEAT_ASK_* environment variables

It is best used in CI environments.

export NEAT_ASK_PROJECT_NAME="My project"
export NEAT_ASK_SECURITY_CONTACT="me@example.com"
export NEAT_ASK_SUPPORT_URL="https://stackoverflow.com"
 
neat repo --silent

-e, --except

Filter out remote files from processing by passing a regular expression.

Example: Process all but markdown files

neat repo --except "\.md$"

The regular expression is run as case insensitive.

-o, --only

Filter remote files to process by passing a regular expression.

Example: Process only markdown files

neat repo --only "\.md$"

The regular expression is run as case insensitive.

-d, --debug

Outputs debug information to help debug something gone wrong. This is very helpful to debug a Neat config file that doesn't work as expected.

neat repo --debug

Example use case

Your organization, maintains a "default" repo which contains files to be used when creating other repositories:

docs/SECURITY.md
LICENSE
README.md

When you create a new repo you can use

neat organization/default new-repo

Or if you already worked in a repo and you forgot to create it from the default repo, you could run from within your repo folder, it will just add new files that are not present in your local folder and will not overwrite any files

neat organization/default

Now, let's say you want to:

  • Make sure your repo's security policy is always up to date with your organization's latest security policy
  • Add files from a repo template you created for your favourite framework on your personal GitHub account
  • Add to your repo any new files added in the organization's default repo
  • Add any generic files that you didn't create already from the neatest repo

You could run the following (or add it in your CI pipeline, package.json, pre-commit hook, etc.)

neat organization/default -f -o docs/SECURITY.md
neat mygithub/favourite-framework
neat organization/default
neat oss

This is non invasive: it will not overwrite your files except for docs/SECURITY.md

🤘 Creating a neat repo

Each Neat repo can contain a .neat.yml configuration file which specifies what to do when someone "neats" the repo.

There is a neat template to help you make a neat template.

Tips and advice

Examples

  • You can find configuration examples in the examples folder
  • For more real-world examples, you can look at the configuration files of the neat-repos.json

Debugging

You can easily inspect a local configuration file by providing a local directory instead of a repo:

neat inspect ./localrepo

The path must start with "." to be recognized as a local path.

When your Neat repo doesn't work as expected, you can use the --debug command if inspect is not of enough help, however you have to push your repo to GitHub first.

neat your/repo --debug

Note that inspect is meant to help final users to inspect a repo. Commands are always highlighted in red and scripts in yellow, so the colors DO NOT indicate errors in your configuration.

If there is an error in your configuration:

  • You will see an error if the YAML file is malformed
  • If the YAML is wellformed, but Neat cannot understand some configuration, it will ignore it and this configuration will not appear when inspecting.

Composability

In order to improve composability between neat templates, respect these guidelines:

  • Make use of injections and before/after hooks.
  • Choose your injection IDs/patterns wisely so that you don't overwrite another template's content. For example, if two templates both use the ID 'description', the only description remaining will be the last one that was run. A good practice is to precede all your IDs with your template name.
  • If your repo is composing well with another repo, say it in your Readme.

Pre-run

Pre-run commands are run on the local machine before any files are processed.

They can be commands to be run as is on the local machine or JavaScript (using Node.js).

pre-run:
  - script: console.log('I am javascript')
  - echo 'I am ran at the begining, before any file is downloaded'

The variable fs is available on all Javascript commands so you don't have to require the filesystem module.

Javascript is always preferred because of its cross-os compatibility.

If you plan for other people to neat your repo, you should make sure these commands can run on any OS, or tell otherwise in your README.

Symbolic links

Sometimes, it is necessary for one file's content to be another file's content. Any specified symbolic links are created before downloading any files.

For example, the most common use case is that you have a readme explaining how to use your Neat template. But you don't want this readme to be the default readme once your user has neated your template. Instead, you prepared another readme.

symlink:
  - README.md: README.tpl.md

Essentially, in this example:

  • The content of README.md will be replaced with the content of README.tpl.md
  • README.tpl.md will be ignored from processing
  • Neat will process README.md like any other file, including replacements, injections or even ignoring it.

Ask questions

When someone neats your repository, you can ask him/her some questions and then use those values to replace strings or run any other arbitrary command.

A question is structured as follows:

ask:
  - id: project_name
    description: What is your project name?
    default: My project
  • id is the only required field
  • description is the actual question asked to the user. If it is not set, Neat will use the id by replacing underscores by spaces (eg. Project name)
  • default is used to provide default values and determine the question type

Neat supports three question types: input, choice and multiple choice that are deducted based on the provided value for default

Input

If no default value is specified or if the default value is a string

ask:
  - id: project_name
    description: What is your project name?

input type

Choice

If the default value is a list of strings

ask:
  - id: ci
    description: What is your preferred CI?
    default: [Travis, Circle CI, Github Actions]

choice type

Multiple choice

If the default value is a list of key/value pairs

  • true means this choice is checked by default
  • false means this choice is unchecked by default
ask:
  - id: options
    description: What options do you want?
    default:
      - "Code Coverage": true
      - "PR template": false
      - "Issue templates": true

multiple choice type

Replacements

Neat can search the added files and replace certain strings with answers to questions in the added files.

For each question, you can specify if Neat has to make a replacement by adding replace: true

ask:
  - id: project_name
    description: What is your project name?
    replace: true

This will have the effect of searching all added files and replacing the question ID in mustache style {{project_name}} with the value of the answer (eg. My project)

For the multiple choice question type, the answer is a string of comma-space-separated values (eg. PR template, Issue templates)

You can change the pattern format or filter which files to search and replace.

If these replacement options are not enough for your use case, you can make use of the post-run commands to do pretty much anything you like

Replacement pattern

You can specify which pattern to replace. By default, it will search and replace mustache variables: {{%s}}.

Example: Replace HTML comments

ask:
  - id: project_name
    description: What is your project name?
    replace: true
replace_pattern: "<!-- %s -->"

Replacement filter

You can specify which files to run replacements on. By default, it will search and replace in all added files.

Example: Make replacements only in markdown and text files

ask:
  - id: project_name
    description: What is your project name?
    replace: true
replace_filter: \.(md|txt)$

Pre-download

Pre-download commands are run on the local machine before any files are processed but just after asking questions, so you can have access to the answers environment variables.

ask:
  - id: project_name
pre-download:
  - script: console.log(process.env.NEAT_ASK_PROJECT_NAME)

Just like pre-run commands, they can accept system commands or Javascript.

Inject files

You can specify a list of files or command outputs to be "injected" into specific files.

This could be used for example to ensure certain chunks of text are included in a readme, even if the local folder of a user already has a readme when he neats your repo.

You can specify either a file, a command (system command) or an url as the source.

This is an example config for the xyz template:

inject:
  - id: xyz-hello
    command: echo "hello world"
    target: README.md
  - id: xyz-support
    file: readme/support.md
    target: docs/CONTRIBUTING.md
  - id: xyz-google
    url: https://google.com
    target: google.html

If the target file does not exist, it will be created.

Any replacements will also be applied to each injected chunk (including replacement filter).

Be sure to read our note on cross-OS compatibility if you plan to use a command as the source.

Injection pattern

The default pattern is <!-- id --> with id being the value of the corresponding id. For example, the id xyz-hello in the example above would inject:

<!-- xyz-hello -->
 
hello world
 
<!-- xyz-hello -->

Neat will find either:

  • Two occurences of the pattern with or without any text in between
  • One occurence of the pattern

If it cannot find this pattern in the target file, Neat will add it at the bottom of the file (unless you specified a before/after pattern).

You can customize the replacement pattern:

inject:
  - id: xyz-hello
    command: echo "hello world"
    target: README.md
    pattern: "<!-- hello-docs -->"

In this example, the following will be injected in README.md

<!-- hello-docs -->
 
hello world
 
<!-- hello-docs -->

Several targets

This example will find the pattern <!-- xyz-hello --> and replace it with hello world in both targets

inject:
  - id: xyz-hello
    command: echo "hello world"
    target: [docs/CONTRIBUTING.md, README.md]

if and if not

You can tweak what you inject based on these conditional parameters:

  • the target file does not exist (no-file)
  • the target file exists and no occurence of the pattern is found (no-pattern)
  • the target file exists and one occurence of the pattern is found (single-pattern)
  • the target file exists and two occurences of the pattern are found (double-pattern)

You can include an if or ifnot declaration per injection. If both are specified, ifnot is ignored. The if/ifnot declaration can either be one conditional parameter or an array of them. If if/ifnot are not specified, it is the same as adding all of the conditional parameters in if if: [no-file, no-pattern, single-pattern, double-pattern]

The example below will inject the content of Google.com in the file search.html only if the search.html file doesn't exist. If search.html exists already, it will inject the content of Bing.com

inject:
  - id: google
    if: no-file
    url: https://google.com
    target: search.html
    pattern: "<!-- search-engine -->"
  - id: bing
    ifnot: [no-file]
    url: https://bing.com
    target: search.html
    pattern: "<!-- search-engine -->"

This other example will produce exactly the same result:

inject:
  - id: google
    ifnot: [no-pattern, single-pattern, double-pattern]
    url: https://google.com
    target: search.html
    pattern: "<!-- search-engine -->"
  - id: bing
    if: [no-pattern, single-pattern, double-pattern]
    url: https://bing.com
    target: search.html
    pattern: "<!-- search-engine -->"

The example below will inject the content of Google.com in the file search.html only if the search.html file doesn't exist, otherwise it will inject the content of Bing.com except if search.html already includes two occurences of the pattern. So, essentially, no injection will happen if a double pattern is found.

inject:
  - id: google
    if: no-file
    url: https://google.com
    target: search.html
    pattern: "<!-- search-engine -->"
  - id: bing
    if: [no-pattern, single-pattern]
    url: https://bing.com
    target: search.html
    pattern: "<!-- search-engine -->"

Position

If the pattern is not found in the target file, the injection is inserted at the end of the file by default.

If you want to insert it in another position, you can use before or after.

For example, the hooks provided by the neat repo can be used as values for before/after.

This example config for the xyz template will find <!-- project-usage --> and inject just after:

inject:
  - id: xyz-hello
    command: echo "hello world"
    target: README.md
    after: <!-- project-usage -->

As a result, when a user runs:

neat repo
neat xyz

The project usage part of the Readme will become:

<!-- project-usage -->
 
<!-- xyz-hello -->
 
hello world
 
<!-- xyz-hello -->

Note that before/after are only used when no occurence of the pattern (in this case <!-- xyz-hello -->) are found.

If the before/after pattern is not found neither, the injection will be appended at the end of the file.

Wrap

The default behaviour is to wrap the injected content with the pattern. You can change this by specifying the strings added before and after the content to inject.

For example, the following only changes the behaviour of what is added before the injected content

inject:
  - id: xyz-hello
    command: echo "hello world"
    target: README.md
    wrap:
      before: <!-- before-hello -->
<!-- before-hello -->
 
hello world
 
<!-- xyz-hello -->

The following does not add the pattern after the injected content

inject:
  - id: xyz-hello
    command: echo "hello world"
    target: README.md
    wrap:
      before: <!-- before-hello -->
      after: false
<!-- before-hello -->
 
hello world

The following changes the whole wrapping

inject:
  - id: xyz-hello
    command: echo "hello world"
    target: README.md
    wrap: <!-- wrap-hello -->
<!-- wrap-hello -->
 
hello world
 
<!-- wrap-hello -->

The following disables wrapping

inject:
  - id: xyz-hello
    command: echo "hello world"
    target: README.md
    wrap: false
hello world

Inject used in conjuction with ignore

For example, if your neat repo is using a README.md to describe what it can do and how to use it but you don't want this README.md to be downloaded when someone neats it, you can ignore it.

  • Using ignore in conjunction with inject:

    ignore: [README.md]
    inject:
      - id: xyz-hello
        command: echo "hello world"
        target: README.md

    As expected, README.md will not be downloaded. However:

    • If Neat is run in a folder that doesn't contain a README.md already, it will create it and inject the xyz-hello section.
    • If Neat is run in a folder that already contains a README.md, it will either replace or append the pattern (as described in Injection pattern)
  • If the source file for the injection is ignored:

    ignore: [readme/support.md]
    inject:
      - id: xyz-support
        file: readme/support.md
        target: README.md

    As expected, readme/support.md will not be downloaded. However:

    • If Neat is run in a folder that doesn't contain a readme/support.md already, it will inject into README.md the content of readme/support.md that will be fetched remotely from your neat repo.
    • If Neat is run in a folder that already contains a readme/support.md, it will use that content as the content to inject into README.md

Ignore files

You can specify a list of relative paths to files that should be ignored (aka never downloaded)

You could use it to provide a documentation for your template in a README.md but you don't want this to be included when people neat your template:

ignore: [README.md]

Note: .neat.yml is always ignored by default, the only way not to include it is to use a symbolic link

Post-run

Post-run commands are run on the local machine after files are processed.

post-run:
  - script: console.log('I am run at the end')

Just like pre-run commands, they can accept system commands or Javascript.

Files environment variables

Post-run commands can access several environment variables. These variables will never include files and directories filtered out using the --only or --except flags because they are simply not processed by Neat.

The most used environment variable is NEAT_ADDED_FILES because it contains a lit of files that were effectively added in the repo.

Environment variable Description
NEAT_ALL_FILES_DIRS Space-separated list of all files and directories that were processed, whether they were added or skipped.
NEAT_ADDED_FILES_DIRS Space-separated list of files and directories that were added.
NEAT_SKIPPED_FILES_DIRS Space-separated list of files and directories that were skipped.
NEAT_ALL_FILES Space-separated list of all files that were processed, whether they were added or skipped.
NEAT_ADDED_FILES Space-separated list of files that were added.
NEAT_SKIPPED_FILES Space-separated list of files that were skipped.
NEAT_ALL_DIRS Space-separated list of all directories that were processed, whether they were added or skipped.
NEAT_ADDED_DIRS Space-separated list of directories that were added.
NEAT_SKIPPED_DIRS Space-separated list of directories that were skipped.

Answers environment variables

In addition, if some questions were asked, their answers are available as environment variables constructed with the question ID in uppercase.

Examples:

  • Will produce the environment variable NEAT_ASK_PROJECT_NAME whose value will be a string of the user's answer (eg. My project)

    id: project_name
  • Will produce the environment variable NEAT_ASK_CI whose value will be a string of the user's answer (eg. Circle CI)

    id: ci
      default: [Travis, Circle CI, Github Actions]
  • Will produce the environment variable NEAT_ASK_OPTIONS whose value will be a string of comma-separated answers (eg. PR template, Issue templates)

    id: options
      default:
        "Code Coverage": true
        "PR template": false
        "Issue templates": true

A note on cross-OS compatibility

When running commands on the user machine, you can either use system commands or Javascript (Node.js). We recommend using Javascript as much as possible because it is the most cross-os compatible way to execute commands. Note that the variable fs is available so you could run filesystem commands directly.

pre-run:
  - script: console.log(fs.readFileSync('README.md', 'utf8'))

If your script becomes quite long, you can add it as an external script file:

pre-run:
  - node script.js

If using a script file, be sure to bundle any dependencies (Webpack could help). Also, we recommend cleaning up your script by removing it from the filesystem so you don't leave any configuration artifacts behind.

For example:

const fs = require("fs");
 
// Run commands
console.log("Hello world!");
 
// Remove this script
fs.unlinkSync(__filename);

If you are running other programs or system-specific commands in your Neat configuration file, make sure to tell it in your Readme in a requirements section for example so as to not disappoint other users of your template.

💚 Contributing

Build Codecov

Add your repo to the registered repo list

If you created a neat repo you're proud of, please add it to the registered repos list:

  1. Fork this repo

  2. Add your repo to neat-repos.json

    • The syntax is "name": "repo/path". As a result, running neat name will fetch repo/path
    • Insert your repo in alphabetical order
    • Only use strings, numbers and dashes (-) in the neat name. It must not start or end with a dash
  3. Open a pull request

Top five ways to contribute

⭐ Star this repo: it's quick and goes a long way! 🔝

🗣️ Spread the word

🐞 Report bugs

Resolve issues

📝 Improve the documentation

Please see the docs/CONTRIBUTING.md for more information.

Project specifics

Install dependencies:

yarn install

Run tests:

yarn test

Run neat:

./bin/run

Use your code as the neat version you use on your system:

yarn link

For maintainers

We follow Semantic versioning and make use of Yarn version to manage new versions.

Patch

When you make backwards compatible bug fixes:

yarn version --patch

New feature

When you add functionality in a backwards compatible manner:

yarn version --minor

Major version

When you make incompatible API changes:

yarn version --major

💡 Todo

  • Overwrite remote config with local config (Add a .neat.yml locally with for example remote: [repo: [inject: [...]]] and it will ignore any injections from the remote)
  • GitHub action running on a schedule to perform automated verification of pre/post run commands in list of neatest repos and add the SHA of the latest commit to neatest-repos.json
  • When neating a repo, verify which SHA is used and display a warning if it has not been verified yet
  • Provide a Docker image with Neat already installed to easily run it in a containerized environment

💬 Support

Join Olivr on Keybase 🔐

Or you can use our Reddit community

📜 License

This project is licensed under the Apache 2.0 License - see the LICENSE file for details

⭕ About Olivr

Olivr is an AI co-founder for your startup.

Readme

Keywords

Package Sidebar

Install

npm i neat

Weekly Downloads

416

Version

2.1.0

License

Apache-2.0

Unpacked Size

105 kB

Total Files

17

Last publish

Collaborators

  • iromain