@owsas/sentiment-multilang/out/index.js

Multilanguage AFINN-based sentiment analysis for Node.js

"use strict";
/*
* Multilanguage AFINN-based sentiment analysis for Node.js
*/
exports.__esModule = true;
// Dependencies
var latinize = require("latinize");
var stemmer = require("stemmer");
var lexicon_1 = require("./lib/lexicon");
/**
 * Tokenizes a string into an array of strings
 * @param input
 */
function tokenize(input) {
    return input
        .toLowerCase()
        // .replace(/^\s+|^0-9+|[^a-z-úñäâàáéèëêïîíìöôùüûœç\- ]+/g, '')
        .replace(/\r?\n|\r/g, ' ') // line breaks changed to space https://stackoverflow.com/a/10805292
        .replace(/n\'t/g, ' not') // n't changed to not
        .replace(/'s/g, ' is') // 's changed to is
        .replace(/['’]/g, ' ') // apos changed to space
        .replace(/[.,\/#!$%\^&\*;:{}=_`\"~()]/g, '') // remove punctuation
        .replace(/\s{2,}/g, ' ') // remove extra spaces https://stackoverflow.com/a/4328722
        .split(' ');
}
exports.tokenize = tokenize;
;
/**
 * Performs sentiment analysis on the provided input 'phrase'
 * @param phrase
 * @param lang
 * @param callback
 */
function sentiment(phrase, lang, callback) {
    // Parse arguments
    if (typeof phrase === 'undefined')
        phrase = '';
    if ((typeof (lang) === 'undefined') || !lexicon_1["default"]["langs"][lang])
        lang = 'unknown';
    if (typeof callback === 'undefined')
        callback = null;
    // Storage objects
    var tokens = tokenize(phrase), words = [], positive = [], negative = [];
    var score = 0;
    // Iterate over tokens if language is knowed
    var len = tokens.length;
    if (lang !== 'unknown') {
        while (len--) {
            // var prevobj = (len > 0) ? String(tokens[len-1]): "";
            var negation = (lexicon_1["default"]["negations"][lang] && lexicon_1["default"]["negations"][lang][tokens[len - 1]]) ? -1 : 1;
            var stringToken = lexicon_1["default"]["truncated"][lang] ? tokens[len].replace(/[aeiouúäâàáéèëêïîíìöôùüû]$/, "") : String(tokens[len]);
            var punctuation = 0;
            // Extract the value using the input word, it's stemmed or it's latinized (no accents) version 
            var tokenValue = lexicon_1["default"][lang] && (lexicon_1["default"][lang][stringToken] || lexicon_1["default"][lang][stemmer(stringToken)] || lexicon_1["default"][lang][latinize(stringToken)]);
            if (tokenValue === undefined) {
                // Search on the emojis data
                if (!lexicon_1["default"]['emoji'][stringToken]) {
                    continue; // continue the while loop with the next iteration
                }
                // It's an emoji
                punctuation = Number(lexicon_1["default"]['emoji'][stringToken]);
            }
            else {
                // It's a word
                punctuation = tokenValue;
            }
            words.push(stringToken);
            if (punctuation > 0)
                positive.push(stringToken);
            if (punctuation < 0)
                negative.push(stringToken);
            score += punctuation * negation;
        }
    }
    // Handle optional async interface
    var result = {
        score: score,
        comparative: score / tokens.length,
        vote: 'neutral',
        tokens: tokens,
        words: words,
        positive: positive,
        negative: negative,
        language: lang
    };
    // Classify text as positive, negative or neutral.
    if (result.score > 0) {
        result.vote = 'positive';
    }
    else if (result.score < 0) {
        result.vote = 'negative';
    }
    if (!callback) {
        return result;
    }
    else {
        callback(null, result);
    }
}
exports.sentiment = sentiment;
;