Globlist-Packer

CLI and NodeJS utility for packing files based on ignore glob pattern lists, such as .gitignore files.

Demo

Show / Hide Demo

Installation

Use it with / without installing:

npx globlist-packer

For frequent use, you should really install the package:

# Globally
npm i -g globlist-packer

# Local devDependency
npm i -D globlist-packer

Once it is installed, it also exposes two callable names (through .bin)

globlist-packer

# Or

globpack

Usage

🚨 Warning: Ignore glob lists, such as .gitignore, work by either negating, or reversing a previous negation. Meaning, by default all files are included, unless they match a pattern in an ignore list. If you are new to git, or this concept, you might find these docs (Atlassian, git-scm) helpful.

Despite the number of configurable options, this tool is designed to be "zero-config", and can work out-of-the-box with minimal settings, for the majority of use-cases.

For example, if you are working with git and wanted to pack your current repo directory to an archive, while keeping files out that are ignored by .gitignore, it should be possible with just:

npx globlist-packer

Or, if you want a specific globlist to be used:

# This will create `dist.zip` in same directory, based on glob(s) contained in `dist.ignorelist`
npx globlist-packer -i dist.ignorelist

You can also use it via JS / TS, by installing the package and then importing:

// ESM import
import {GloblistPacker} from 'globlist-packer';

GloblistPacker({
	ignoreListFileNames: ['dist.ignorelist'],
	// This will create `distribution.zip`
	archiveName: 'distribution',
	archiveType: 'zip'
});

For more advanced uses, see options below, and examples under "Usage Examples".

As much as possible, I try to make all options and flags available through both the main JS entry-point, as well as through the CLI. Refer to the table below, the source code, or use --help with the CLI.

Option Key	CLI	Description	Type	Default
rootDir	root-dir	Used as the entry point to the filewalker, and used as the base to resolve any relative paths that are passed	string	process.cwd() (working directory)
ignoreListFileNames	ignorelist-files or -i	Files that are formatted like .gitignore - line delimited glob patterns to include or exclude. Warning: Order matters!	string[]	[] (or ['.gitignore'] if useGitIgnoreFiles === true)
useGitIgnoreFiles	use-gitignore-files	Whether or not to check for, and use, .gitignore files as part of the ruleset	boolean	true
includeDefaultIgnores	include-default-ignores	If true, adds some default excludes that should apply to most projects and helps avoid accidental bundling	boolean	true
includeEmpty	include-empty	Include empty directories in the output archive	boolean	false
followSymlink	follow-symlink	Whether or not to follow symlinks when copying files to the archive.	boolean	false
outDir	out-dir or -d	Where to save the generated archive(s). Defaults to the root directory and/or calling directory.	string	The root directory, or calling directory.
copyFilesTo	copy-files-to	Path to directory to copy all matching files to, instead of creating a packed archive. If used, nullifies a lot of other settings, and causes no archive to be generated.	string	NA
archiveName	archive-name or -n	Name for the generated archive. File extension is optional, and will be overwritten anyways, based on archiveType	string	Will default to primary non-gitignore ignore file, and if that is not available, to simply packed.{ext}
archiveType	archive-type or -t	Type of generated archive file. Not the same as file extension.	'tar' or 'zip'	'zip'
archiveRootDirName	archive-root-dir-name	Inject a single folder in the root of the archive, with this name, which will contain all collected files.	string	NA / Not used
maxFileCount	max-files or -m	If you are worried about accidentally including a massive number of files and you want to bail out early of the archiving process if this happens, you can set a hard cap with this option.	number	NA / Infinity
verbose	verbose	Enable extra logging to the console / stdout	boolean	false

The following options are only available when calling the program via JS/TS.

Option Key	Description
archiveOptions	You can extend / override the options that are passed to the archiver package that this program uses.
fileNameTransformer	You can pass a callback function here that will receive the path and name of a file that is being considered for packing. You can return false to stop the file from being packed, a string to rename the file in the archive, or undefined (default return) or true to process the file as-is.
onStepChange	Receive updates when the program has moved between steps. For internal use.

Usage Examples

Example scenario(s):

Scenario: Overriding .gitignore

Files:

.
├── package-lock.json
├── package.json
├── README.md
├── build/
│   ├── index.html
│   ├── index.js
│   └── style.css
├── src/
│   ├── index.js
│   └── style.css
├── vendor/
│   └── vendor-bundle.js
├── node_modules/
│   └── ... (lots of files)
└── tests/
    └── ... (lots of files)

.gitignore

node_modules
build
vendor

Say that we want to share this project with someone else, but not the source code - we want to give them the pre-built project, with as few files as possible. We could create a pack list for this tool that looks something like this:

dist.packlistignore

# Remove all source code, and unneeded files
src
tests
package-lock.json
package.json

# Block recursive packing if ran again
dist.tgz
dist.zip

# OVERRIDE ignores in .gitignore, adding back build files and artifacts
!build
!vendor

Now, we can create our shareable archive file with a single command, as many times as we want:

globlist-packer -i dist.packlistignore

Special Use-Case: Copying Files to Directory

This is semi-experimental, as it is not the main intended use of the program, but for convenience I have added an option that lets you use this to only copy files (based on globlists) to a target directory, and skip the archive / tar / zip generation.

To use this feature, pass a string to the copy-files-to option (or copyFilesTo via JS API) - this will cause the tool to skip all archiver steps and only handle copying files. You can still use all the input control options (such as ignorelist-files, include-default-ignores, etc.).

The truth is that this feature complicates things a little bit. For example, if the target output directory is nested inside the root directory, then I have to block files that live inside the output directory, to prevent recursive copying in case the tool is ran more than once without clearing the output.

Design Decisions

This tool is primarily a wrapper around the ignore-walk package. Due to some limitations in that package, and complexities of resolving glob patterns (remember: you can have negation in glob lists), if the includeDefaultIgnores option is true (which is default), this tool will actually temporarily inject a ignore glob list file in your project root directory.

Multiple considerations are taken around this action:

• The file only exists for as long as it takes to walk the file tree. This can be as short as milliseconds.
• The file is given a long and dynamic filename - highly unlikely to collide with any existing files
  • If, somehow, a file already exists with that name, the program will halt and not overwrite the existing file

Default Ignores

If includeDefaultIgnores is set to true, which is also the default value, then a few default ignore patterns are added to prevent unwanted files from ending up in the packing process.

As of writing this, the default ignore patterns are node_modules and .git, but you can browse ./src to see for yourself.

Change Notes

Version	Date	Notes
v0.1.0	04/25/2021	Initial Release 🚀

Related Projects

Of course, after building this tool I immediately found some that might have fit the bill for what I needed. However, all of these are slightly different from what this tool offers:

• palletjack
• gitzip