nlcst-emoji-modifier

nlcst-emoji-modifier

Classify unicode emoji and Github emoji (gemoji) as EmoticonNodes.

Implemented by retext-emoji, but separated for use by standalone (non-retext) parsers.

Note: this project is useful in combination with natural language parsers like parse-latin, parse-dutch, and parse-english.

Installation

npm:

$ npm install nlcst-emoji-modifier

Component.js:

$ component install wooorm/nlcst-emoji-modifier

Bower:

$ bower install nlcst-emoji-modifier

Usage

var modifier = require('nlcst-emoji-modifier');
var ParseEnglish = require('parse-english');
var english = new ParseEnglish();

/* Attach the modifier. */
modifier(english);

english.parse('Who doesn’t like Gemoji? :+1: You? 💩').children[0].children;

Yields:

[
  {
    "type": "SentenceNode",
    "children": [
      {
        "type": "WordNode",
        "children": [
          {
            "type": "TextNode",
            "value": "Who"
          }
        ]
      },
      {
        "type": "WhiteSpaceNode",
        "value": " "
      },
      {
        "type": "WordNode",
        "children": [
          {
            "type": "TextNode",
            "value": "doesn"
          },
          {
            "type": "PunctuationNode",
            "value": "’"
          },
          {
            "type": "TextNode",
            "value": "t"
          }
        ]
      },
      {
        "type": "WhiteSpaceNode",
        "value": " "
      },
      {
        "type": "WordNode",
        "children": [
          {
            "type": "TextNode",
            "value": "like"
          }
        ]
      },
      {
        "type": "WhiteSpaceNode",
        "value": " "
      },
      {
        "type": "WordNode",
        "children": [
          {
            "type": "TextNode",
            "value": "Gemoji"
          }
        ]
      },
      {
        "type": "PunctuationNode",
        "value": "?"
      },
      {
        "type": "WhiteSpaceNode",
        "value": " "
      },
      {
        "type": "EmoticonNode",
        "value": ":+1:"
      }
    ]
  },
  {
    "type": "WhiteSpaceNode",
    "value": " "
  },
  {
    "type": "SentenceNode",
    "children": [
      {
        "type": "WordNode",
        "children": [
          {
            "type": "TextNode",
            "value": "You"
          }
        ]
      },
      {
        "type": "PunctuationNode",
        "value": "?"
      },
      {
        "type": "WhiteSpaceNode",
        "value": " "
      },
      {
        "type": "EmoticonNode",
        "value": "💩"
      }
    ]
  }
]

Performance

On a MacBook Air, parse-english performs about 27% slower on content filled with (g)emoji, and a 18% slower on content without (g)emoji, when using this modifier.

             parse w/ modifier
  1,303 op/s » A paragraph (5 sentences, 100 words, 5 emoji, 5 gemoji)
  1,653 op/s » A paragraph (5 sentences, 100 words, no emoji, no gemoji)

             parse w/o modifier
  1,784 op/s » A paragraph (5 sentences, 100 words, 5 emoji, 5 gemoji)
  2,038 op/s » A paragraph (5 sentences, 100 words, no emoji, no gemoji)

Related

• nlcst
• parse-latin
• parse-dutch
• parse-english
• retext
• textom

License

MIT © Titus Wormer