From e28258010182b56f27cfbd3f9f9a58fd9cd8870d Mon Sep 17 00:00:00 2001 From: Nolan Lawson Date: Mon, 3 Jul 2017 02:02:36 -0700 Subject: Faster emojify() by avoiding str.replace() entirely (#4049) --- app/javascript/mastodon/emoji.js | 69 +++++++++++++++++++--------------------- 1 file changed, 32 insertions(+), 37 deletions(-) (limited to 'app/javascript') diff --git a/app/javascript/mastodon/emoji.js b/app/javascript/mastodon/emoji.js index d0df71ea3..7043d5f3a 100644 --- a/app/javascript/mastodon/emoji.js +++ b/app/javascript/mastodon/emoji.js @@ -1,60 +1,55 @@ import emojione from 'emojione'; +import Trie from 'substring-trie'; -const toImage = str => shortnameToImage(unicodeToImage(str)); +const mappedUnicode = emojione.mapUnicodeToShort(); +const trie = new Trie(Object.keys(emojione.jsEscapeMap)); -const unicodeToImage = str => { - const mappedUnicode = emojione.mapUnicodeToShort(); - - return str.replace(emojione.regUnicode, unicodeChar => { - if (typeof unicodeChar === 'undefined' || unicodeChar === '' || !(unicodeChar in emojione.jsEscapeMap)) { - return unicodeChar; - } - - const unicode = emojione.jsEscapeMap[unicodeChar]; - const short = mappedUnicode[unicode]; - const filename = emojione.emojioneList[short].fname; - const alt = emojione.convert(unicode.toUpperCase()); - - return `${alt}`; - }); -}; - -const shortnameToImage = str => { - // This walks through the string from end to start, ignoring any tags (

,
, etc.) - // and replacing valid shortnames like :smile: and :wink: that _aren't_ within - // tags with an version. - // The goal is to be the same as an emojione.regShortNames replacement, but faster. - // The reason we go backwards is because then we can replace substrings as we go. - let i = str.length; +function emojify(str) { + // This walks through the string from start to end, ignoring any tags (

,
, etc.) + // and replacing valid shortnames like :smile: and :wink: as well as unicode strings + // that _aren't_ within tags with an version. + // The goal is to be the same as an emojione.regShortNames/regUnicode replacement, but faster. + let i = -1; let insideTag = false; let insideShortname = false; - let shortnameEndIndex = -1; - while (i--) { + let shortnameStartIndex = -1; + let match; + while (++i < str.length) { const char = str.charAt(i); if (insideShortname && char === ':') { - const shortname = str.substring(i, shortnameEndIndex + 1); + const shortname = str.substring(shortnameStartIndex, i + 1); if (shortname in emojione.emojioneList) { const unicode = emojione.emojioneList[shortname].unicode[emojione.emojioneList[shortname].unicode.length - 1]; const alt = emojione.convert(unicode.toUpperCase()); const replacement = `${alt}`; - str = str.substring(0, i) + replacement + str.substring(shortnameEndIndex + 1); + str = str.substring(0, shortnameStartIndex) + replacement + str.substring(i + 1); + i += (replacement.length - shortname.length - 1); // jump ahead the length we've added to the string } else { - i++; // stray colon, try again + i--; // stray colon, try again } insideShortname = false; - } else if (insideTag && char === '<') { + } else if (insideTag && char === '>') { insideTag = false; - } else if (char === '>') { + } else if (char === '<') { insideTag = true; insideShortname = false; } else if (!insideTag && char === ':') { insideShortname = true; - shortnameEndIndex = i; + shortnameStartIndex = i; + } else if (!insideTag && (match = trie.search(str.substring(i)))) { + const unicodeStr = match; + if (unicodeStr in emojione.jsEscapeMap) { + const unicode = emojione.jsEscapeMap[unicodeStr]; + const short = mappedUnicode[unicode]; + const filename = emojione.emojioneList[short].fname; + const alt = emojione.convert(unicode.toUpperCase()); + const replacement = `${alt}`; + str = str.substring(0, i) + replacement + str.substring(i + unicodeStr.length); + i += (replacement.length - unicodeStr.length); // jump ahead the length we've added to the string + } } } return str; -}; +} -export default function emojify(text) { - return toImage(text); -}; +export default emojify; -- cgit