This repository contains the logic for building the emoji data set, sprite sheets, and optimized images used by the official FrankerFaceZ client.

1. Clone this repository.
2. Initialize the git submodules in sources/
3. npm install

npm install --save emojibase-data@latest emoji-datasource@latest

The following command runs all the scripts necessary to generate emoji.json and emoji-compact.json.

npm run data

Before building images, you need several utilities available within your PATH that will be used for processing images:

• montage - Part of ImageMagick. Used to build sprite sheets.

• mogrify - Part of ImageMagick. Used to resize PNG images in bulk.

• pngcrush, optipng, pngout, advpng, zopflipng - Used during image optimization.

Note: Image optimization programs are not required, however images will obviously not be optimized if none are installed.

The official FrankerFaceZ client consumes the emoji-compact.json data format. This format uses arrays to minimize the amount of data the client must download, and is designed to be rehydrated on the client.

Each emoji entry is formatted as follows:

EMOJI = [group, order, SHORTCODES, name || 0, key, POS, IMAGES, [list, of, VERSION, ...] || 0, type];

SHORTCODES = String || [list, of, String, ...];
POS = [Number, Number];
IMAGES = (
	0b1000 + // Has Noto Image
	0b0100 + // Has Blob Image
	0b0010 + // Has Twemoji Image
	0b0001   // Has OpenMoji Image
);
VERSION = [
	key,
	POS,
	IMAGES,
	tone,
	type,
	SHORTCODES
];

• group is the numeric id of the emoji category. Categories are included in the data file as an associative array.
• order is the sorting order for the emoji. Emoji within each category should be sorted using this number.
• SHORTCODES is either a String if there is a single short code, or an array of strings if there are multiple short codes.
• name is zero if the name is not distinct from the first short code. Otherwise, name is a String. The name is checked against the short code by converting it to lower case and replacing spaces with underscores.
• key is the hexadecimal code for the emoji, with code points separated by - characters.
• POS is the sprite sheet position for the emoji or emoji version.
• IMAGES is a bitfield that lets the client know if each emoji style has an image for a specific emoji.
• VERSION is an array of versions for an emoji. This is used for emoji that have variations with different skin tones. If there are no versions for an emoji, this will be 0.
• type indicates whether the emoji is represented as text by default (0) or graphically (1). If an emoji is presented as text by default, then clients should only display this emoji if it's marked with a graphical presentation variation. (So, followed with a -fe0f.) Additionally, when inserting the emoji into text after being selected from a menu or completed from a short code, text emoji should also have an additional -fe0f inserted to force them to render graphically.
• tone indicates the skin tone for emoji variants.

The following command runs the commands to:

1. copy all image files into the directory structure that the FrankerFaceZ client expects
2. resize the PNG images to the desired size (72x72)
3. Construct sprite sheets
4. Optimize all images.

npm run images

Note: Optimizing images is a very slow operation. If you want to do everything except optimization, please run the command: npm run images:basic instead.

Downloads Noto emoji that are missing from the Noto repository. Notably, this includes all national flag emoji.

Resizes Noto and Blob emoji to 72x72, as the repositories only include 128x128.

Download a list of short codes from the GitHub API and merge it with extra short codes from the emoji-data package.

Build the main emoji data files. This checks for the existence of image files in each style and merges in short codes from the shortcodes command.

Copy all image files into the images/ subdirectory using the expected naming scheme. Populates missing Blob emoji images from the Noto image sources.

Build sprite sheets for each of the emoji styles in 18x18, 36x36, and 72x72 sizes.

For each individual emoji PNG, as well as each sprite sheet, try to optimize the image using pngcrush, optipng, pngout, advpng, and zopflipng. The smallest file will be taken. This script takes a long time to execute.

Note: The scripts related to image processing are based off of build scripts from the emoji-data library, without any extra dependency on PHP.

The scripts in this repository use the MIT license. The data, however, inherits licenses from several different sources as indicated in the Sources section below:

• Emojibase - This is the primary data source we use as the truth to determine which emoji exist.
  • License: MIT
• GitHub - We use GitHub's emojis API endpoint to look up additional short codes.
  • License: GitHub ToS
• emoji-data - We used to use emoji-data as our main source of emoji data, but it isn't updated frequently. Now, we use the module only to continue supporting short codes that we used before.
  • License: MIT

• Twemoji - The default emoji style for FrankerFaceZ. Used by and provided by Twitter.
  • Data License: Apache 2
  • Image License: CC BY 4.0
• Google Noto - The style used by Google projects, including Android.
  • License: Apache 2
• Blobmoji - A fork of an old version of Noto emoji, popular with many people due to the charming 'blob' look of the smileys.
  • License: Apache 2
• OpenMoji - An open source, community driven emoji style.
  • License: CC BY-SA 4.0