Compress and combine emoji data (#5229)

app/javascript/mastodon/features/emoji/emoji.js

@@ -0,0 +1,72 @@
+import unicodeMapping from './emoji_unicode_mapping_light';
+import Trie from 'substring-trie';
+
+const trie = new Trie(Object.keys(unicodeMapping));
+
+const assetHost = process.env.CDN_HOST || '';
+
+const emojify = (str, customEmojis = {}) => {
+  let rtn = '';
+  for (;;) {
+    let match, i = 0, tag;
+    while (i < str.length && (tag = '<&:'.indexOf(str[i])) === -1 && !(match = trie.search(str.slice(i)))) {
+      i += str.codePointAt(i) < 65536 ? 1 : 2;
+    }
+    let rend, replacement = '';
+    if (i === str.length) {
+      break;
+    } else if (str[i] === ':') {
+      if (!(() => {
+        rend = str.indexOf(':', i + 1) + 1;
+        if (!rend) return false; // no pair of ':'
+        const lt = str.indexOf('<', i + 1);
+        if (!(lt === -1 || lt >= rend)) return false; // tag appeared before closing ':'
+        const shortname = str.slice(i, rend);
+        // now got a replacee as ':shortname:'
+        // if you want additional emoji handler, add statements below which set replacement and return true.
+        if (shortname in customEmojis) {
+          replacement = `<img draggable="false" class="emojione" alt="${shortname}" title="${shortname}" src="${customEmojis[shortname]}" />`;
+          return true;
+        }
+        return false;
+      })()) rend = ++i;
+    } else if (tag >= 0) { // <, &
+      rend = str.indexOf('>;'[tag], i + 1) + 1;
+      if (!rend) break;
+      i = rend;
+    } else { // matched to unicode emoji
+      const { filename, shortCode } = unicodeMapping[match];
+      const title = shortCode ? `:${shortCode}:` : '';
+      replacement = `<img draggable="false" class="emojione" alt="${match}" title="${title}" src="${assetHost}/emoji/${filename}.svg" />`;
+      rend = i + match.length;
+    }
+    rtn += str.slice(0, i) + replacement;
+    str = str.slice(rend);
+  }
+  return rtn + str;
+};
+
+export default emojify;
+
+export const buildCustomEmojis = customEmojis => {
+  const emojis = [];
+
+  customEmojis.forEach(emoji => {
+    const shortcode = emoji.get('shortcode');
+    const url       = emoji.get('static_url');
+    const name      = shortcode.replace(':', '');
+
+    emojis.push({
+      id: name,
+      name,
+      short_names: [name],
+      text: '',
+      emoticons: [],
+      keywords: [name],
+      imageUrl: url,
+      custom: true,
+    });
+  });
+
+  return emojis;
+};

app/javascript/mastodon/features/emoji/emoji_compressed.js

@@ -0,0 +1,90 @@
+// @preval
+// http://www.unicode.org/Public/emoji/5.0/emoji-test.txt
+// This file contains the compressed version of the emoji data from
+// both emoji_map.json and from emoji-mart's emojiIndex and data objects.
+// It's designed to be emitted in an array format to take up less space
+// over the wire.
+
+const { unicodeToFilename } = require('./unicode_to_filename');
+const { unicodeToUnifiedName } = require('./unicode_to_unified_name');
+const emojiMap         = require('./emoji_map.json');
+const { emojiIndex } = require('emoji-mart');
+const emojiMartData = require('emoji-mart/dist/data').default;
+const excluded       = ['®', '©', '™'];
+const skins          = ['🏻', '🏼', '🏽', '🏾', '🏿'];
+const shortcodeMap   = {};
+
+const shortCodesToEmojiData = {};
+const emojisWithoutShortCodes = [];
+
+Object.keys(emojiIndex.emojis).forEach(key => {
+  shortcodeMap[emojiIndex.emojis[key].native] = emojiIndex.emojis[key].id;
+});
+
+const stripModifiers = unicode => {
+  skins.forEach(tone => {
+    unicode = unicode.replace(tone, '');
+  });
+
+  return unicode;
+};
+
+Object.keys(emojiMap).forEach(key => {
+  if (excluded.includes(key)) {
+    delete emojiMap[key];
+    return;
+  }
+
+  const normalizedKey = stripModifiers(key);
+  let shortcode       = shortcodeMap[normalizedKey];
+
+  if (!shortcode) {
+    shortcode = shortcodeMap[normalizedKey + '\uFE0F'];
+  }
+
+  const filename = emojiMap[key];
+
+  const filenameData = [key];
+
+  if (unicodeToFilename(key) !== filename) {
+    // filename can't be derived using unicodeToFilename
+    filenameData.push(filename);
+  }
+
+  if (typeof shortcode === 'undefined') {
+    emojisWithoutShortCodes.push(filenameData);
+  } else {
+    shortCodesToEmojiData[shortcode] = shortCodesToEmojiData[shortcode] || [[]];
+    shortCodesToEmojiData[shortcode][0].push(filenameData);
+  }
+});
+
+Object.keys(emojiIndex.emojis).forEach(key => {
+  const { native } = emojiIndex.emojis[key];
+  const { short_names, search, unified } = emojiMartData.emojis[key];
+  if (short_names[0] !== key) {
+    throw new Error('The compresser expects the first short_code to be the ' +
+      'key. It may need to be rewritten if the emoji change such that this ' +
+      'is no longer the case.');
+  }
+
+  short_names.splice(0, 1); // first short name can be inferred from the key
+
+  const searchData = [native, short_names, search];
+  if (unicodeToUnifiedName(native) !== unified) {
+    // unified name can't be derived from unicodeToUnifiedName
+    searchData.push(unified);
+  }
+
+  shortCodesToEmojiData[key].push(searchData);
+});
+
+// JSON.parse/stringify is to emulate what @preval is doing and avoid any
+// inconsistent behavior in dev mode
+module.exports = JSON.parse(JSON.stringify([
+  shortCodesToEmojiData,
+  emojiMartData.skins,
+  emojiMartData.categories,
+  emojiMartData.short_names,
+  emojisWithoutShortCodes,
+]));

app/javascript/mastodon/features/emoji/emoji_mart_data_light.js

@@ -0,0 +1,41 @@
+// The output of this module is designed to mimic emoji-mart's
+// "data" object, such that we can use it for a light version of emoji-mart's
+// emojiIndex.search functionality.
+const { unicodeToUnifiedName } = require('./unicode_to_unified_name');
+const [ shortCodesToEmojiData, skins, categories, short_names ] = require('./emoji_compressed');
+
+const emojis = {};
+
+// decompress
+Object.keys(shortCodesToEmojiData).forEach((shortCode) => {
+  let [
+    filenameData, // eslint-disable-line no-unused-vars
+    searchData,
+  ] = shortCodesToEmojiData[shortCode];
+  let [
+    native,
+    short_names,
+    search,
+    unified,
+  ] = searchData;
+
+  if (!unified) {
+    // unified name can be derived from unicodeToUnifiedName
+    unified = unicodeToUnifiedName(native);
+  }
+
+  short_names = [shortCode].concat(short_names);
+  emojis[shortCode] = {
+    native,
+    search,
+    short_names,
+    unified,
+  };
+});
+
+module.exports = {
+  emojis,
+  skins,
+  categories,
+  short_names,
+};

app/javascript/mastodon/features/emoji/emoji_mart_search_light.js

@@ -0,0 +1,154 @@
+// This code is largely borrowed from:
+// https://github.com/missive/emoji-mart/blob/bbd4fbe/src/utils/emoji-index.js
+
+import data from './emoji_mart_data_light';
+import { getData, getSanitizedData, intersect } from './emoji_utils';
+
+let index = {};
+let emojisList = {};
+let emoticonsList = {};
+let previousInclude = [];
+let previousExclude = [];
+
+for (let emoji in data.emojis) {
+  let emojiData = data.emojis[emoji],
+    { short_names, emoticons } = emojiData,
+    id = short_names[0];
+
+  for (let emoticon of (emoticons || [])) {
+    if (!emoticonsList[emoticon]) {
+      emoticonsList[emoticon] = id;
+    }
+  }
+
+  emojisList[id] = getSanitizedData(id);
+}
+
+function search(value, { emojisToShowFilter, maxResults, include, exclude, custom = [] } = {}) {
+  maxResults = maxResults || 75;
+  include = include || [];
+  exclude = exclude || [];
+
+  if (custom.length) {
+    for (const emoji of custom) {
+      data.emojis[emoji.id] = getData(emoji);
+      emojisList[emoji.id] = getSanitizedData(emoji);
+    }
+
+    data.categories.push({
+      name: 'Custom',
+      emojis: custom.map(emoji => emoji.id),
+    });
+  }
+
+  let results = null;
+  let pool = data.emojis;
+
+  if (value.length) {
+    if (value === '-' || value === '-1') {
+      return [emojisList['-1']];
+    }
+
+    let values = value.toLowerCase().split(/[\s|,|\-|_]+/);
+
+    if (values.length > 2) {
+      values = [values[0], values[1]];
+    }
+
+    if (include.length || exclude.length) {
+      pool = {};
+
+      if (previousInclude !== include.sort().join(',') || previousExclude !== exclude.sort().join(',')) {
+        previousInclude = include.sort().join(',');
+        previousExclude = exclude.sort().join(',');
+        index = {};
+      }
+
+      for (let category of data.categories) {
+        let isIncluded = include && include.length ? include.indexOf(category.name.toLowerCase()) > -1 : true;
+        let isExcluded = exclude && exclude.length ? exclude.indexOf(category.name.toLowerCase()) > -1 : false;
+        if (!isIncluded || isExcluded) {
+          continue;
+        }
+
+        for (let emojiId of category.emojis) {
+          pool[emojiId] = data.emojis[emojiId];
+        }
+      }
+    } else if (previousInclude.length || previousExclude.length) {
+      index = {};
+    }
+
+    let allResults = values.map((value) => {
+      let aPool = pool;
+      let aIndex = index;
+      let length = 0;
+
+      for (let char of value.split('')) {
+        length++;
+
+        aIndex[char] = aIndex[char] || {};
+        aIndex = aIndex[char];
+
+        if (!aIndex.results) {
+          let scores = {};
+
+          aIndex.results = [];
+          aIndex.pool = {};
+
+          for (let id in aPool) {
+            let emoji = aPool[id],
+              { search } = emoji,
+              sub = value.substr(0, length),
+              subIndex = search.indexOf(sub);
+
+            if (subIndex !== -1) {
+              let score = subIndex + 1;
+              if (sub === id) {
+                score = 0;
+              }
+
+              aIndex.results.push(emojisList[id]);
+              aIndex.pool[id] = emoji;
+
+              scores[id] = score;
+            }
+          }
+
+          aIndex.results.sort((a, b) => {
+            let aScore = scores[a.id],
+              bScore = scores[b.id];
+
+            return aScore - bScore;
+          });
+        }
+
+        aPool = aIndex.pool;
+      }
+
+      return aIndex.results;
+    }).filter(a => a);
+
+    if (allResults.length > 1) {
+      results = intersect(...allResults);
+    } else if (allResults.length) {
+      results = allResults[0];
+    } else {
+      results = [];
+    }
+  }
+
+  if (results) {
+    if (emojisToShowFilter) {
+      results = results.filter((result) => emojisToShowFilter(data.emojis[result.id].unified));
+    }
+
+    if (results && results.length > maxResults) {
+      results = results.slice(0, maxResults);
+    }
+  }
+
+  return results;
+}
+
+export { search };

app/javascript/mastodon/features/emoji/emoji_unicode_mapping_light.js

@@ -0,0 +1,35 @@
+// A mapping of unicode strings to an object containing the filename
+// (i.e. the svg filename) and a shortCode intended to be shown
+// as a "title" attribute in an HTML element (aka tooltip).
+
+const [
+  shortCodesToEmojiData,
+  skins, // eslint-disable-line no-unused-vars
+  categories, // eslint-disable-line no-unused-vars
+  short_names, // eslint-disable-line no-unused-vars
+  emojisWithoutShortCodes,
+] = require('./emoji_compressed');
+const { unicodeToFilename } = require('./unicode_to_filename');
+
+// decompress
+const unicodeMapping = {};
+
+function processEmojiMapData(emojiMapData, shortCode) {
+  let [ native, filename ] = emojiMapData;
+  if (!filename) {
+    // filename name can be derived from unicodeToFilename
+    filename = unicodeToFilename(native);
+  }
+  unicodeMapping[native] = {
+    shortCode: shortCode,
+    filename: filename,
+  };
+}
+
+Object.keys(shortCodesToEmojiData).forEach((shortCode) => {
+  let [ filenameData ] = shortCodesToEmojiData[shortCode];
+  filenameData.forEach(emojiMapData => processEmojiMapData(emojiMapData, shortCode));
+});
+emojisWithoutShortCodes.forEach(emojiMapData => processEmojiMapData(emojiMapData));
+
+module.exports = unicodeMapping;

app/javascript/mastodon/features/emoji/emoji_utils.js

@@ -0,0 +1,137 @@
+// This code is largely borrowed from:
+// https://github.com/missive/emoji-mart/blob/bbd4fbe/src/utils/index.js
+
+import data from './emoji_mart_data_light';
+
+const COLONS_REGEX = /^(?:\:([^\:]+)\:)(?:\:skin-tone-(\d)\:)?$/;
+
+function buildSearch(thisData) {
+  const search = [];
+
+  let addToSearch = (strings, split) => {
+    if (!strings) {
+      return;
+    }
+
+    (Array.isArray(strings) ? strings : [strings]).forEach((string) => {
+      (split ? string.split(/[-|_|\s]+/) : [string]).forEach((s) => {
+        s = s.toLowerCase();
+
+        if (search.indexOf(s) === -1) {
+          search.push(s);
+        }
+      });
+    });
+  };
+
+  addToSearch(thisData.short_names, true);
+  addToSearch(thisData.name, true);
+  addToSearch(thisData.keywords, false);
+  addToSearch(thisData.emoticons, false);
+
+  return search;
+}
+
+function unifiedToNative(unified) {
+  let unicodes = unified.split('-'),
+    codePoints = unicodes.map((u) => `0x${u}`);
+
+  return String.fromCodePoint(...codePoints);
+}
+
+function sanitize(emoji) {
+  let { name, short_names, skin_tone, skin_variations, emoticons, unified, custom, imageUrl } = emoji,
+    id = emoji.id || short_names[0],
+    colons = `:${id}:`;
+
+  if (custom) {
+    return {
+      id,
+      name,
+      colons,
+      emoticons,
+      custom,
+      imageUrl,
+    };
+  }
+
+  if (skin_tone) {
+    colons += `:skin-tone-${skin_tone}:`;
+  }
+
+  return {
+    id,
+    name,
+    colons,
+    emoticons,
+    unified: unified.toLowerCase(),
+    skin: skin_tone || (skin_variations ? 1 : null),
+    native: unifiedToNative(unified),
+  };
+}
+
+function getSanitizedData(emoji) {
+  return sanitize(getData(emoji));
+}
+
+function getData(emoji) {
+  let emojiData = {};
+
+  if (typeof emoji === 'string') {
+    let matches = emoji.match(COLONS_REGEX);
+
+    if (matches) {
+      emoji = matches[1];
+
+    }
+
+    if (data.short_names.hasOwnProperty(emoji)) {
+      emoji = data.short_names[emoji];
+    }
+
+    if (data.emojis.hasOwnProperty(emoji)) {
+      emojiData = data.emojis[emoji];
+    }
+  } else if (emoji.custom) {
+    emojiData = emoji;
+
+    emojiData.search = buildSearch({
+      short_names: emoji.short_names,
+      name: emoji.name,
+      keywords: emoji.keywords,
+      emoticons: emoji.emoticons,
+    });
+
+    emojiData.search = emojiData.search.join(',');
+  } else if (emoji.id) {
+    if (data.short_names.hasOwnProperty(emoji.id)) {
+      emoji.id = data.short_names[emoji.id];
+    }
+
+    if (data.emojis.hasOwnProperty(emoji.id)) {
+      emojiData = data.emojis[emoji.id];
+    }
+  }
+
+  emojiData.emoticons = emojiData.emoticons || [];
+  emojiData.variations = emojiData.variations || [];
+
+  if (emojiData.variations && emojiData.variations.length) {
+    emojiData = JSON.parse(JSON.stringify(emojiData));
+    emojiData.unified = emojiData.variations.shift();
+  }
+
+  return emojiData;
+}
+
+function intersect(a, b) {
+  let aSet = new Set(a);
+  let bSet = new Set(b);
+  let intersection = new Set(
+    [...aSet].filter(x => bSet.has(x))
+  );
+
+  return Array.from(intersection);
+}
+
+export { getData, getSanitizedData, intersect };

app/javascript/mastodon/features/emoji/unicode_to_filename.js

@@ -0,0 +1,26 @@
+// taken from:
+// https://github.com/twitter/twemoji/blob/47732c7/twemoji-generator.js#L848-L866
+exports.unicodeToFilename = (str) => {
+  let result = '';
+  let charCode = 0;
+  let p = 0;
+  let i = 0;
+  while (i < str.length) {
+    charCode = str.charCodeAt(i++);
+    if (p) {
+      if (result.length > 0) {
+        result += '-';
+      }
+      result += (0x10000 + ((p - 0xD800) << 10) + (charCode - 0xDC00)).toString(16);
+      p = 0;
+    } else if (0xD800 <= charCode && charCode <= 0xDBFF) {
+      p = charCode;
+    } else {
+      if (result.length > 0) {
+        result += '-';
+      }
+      result += charCode.toString(16);
+    }
+  }
+  return result;
+};

app/javascript/mastodon/features/emoji/unicode_to_unified_name.js

@@ -0,0 +1,17 @@
+function padLeft(str, num) {
+  while (str.length < num) {
+    str = '0' + str;
+  }
+  return str;
+}
+
+exports.unicodeToUnifiedName = (str) => {
+  let output = '';
+  for (let i = 0; i < str.length; i += 2) {
+    if (i > 0) {
+      output += '-';
+    }
+    output += padLeft(str.codePointAt(i).toString(16).toUpperCase(), 4);
+  }
+  return output;
+};