concepts-parser/lib/parsers/base.js

Concepts Extracting from text

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
const utils = require("../utils");
const concept_1 = require("../concept");
class BaseParser {
    constructor(options) {
        options = options || {};
        options = Object.assign({ acceptConceptWords: [], acceptConnectChars: ["&", "-", "'", ".", "’", "`"], acceptStartQuotes: ['"', "“", "”", "„", "«"], acceptEndQuotes: ['"', "“", "”", "„", "»"], acceptPrefixes: [] }, options);
        if (options.acceptConceptWords) {
            options.acceptConceptWords.sort(function (a, b) {
                return b.length - a.length;
            });
        }
        options.acceptConceptWordsRegex = new RegExp("^(" + options.acceptConceptWords.join("|") + ") ");
        options.acceptConceptWordsRegex2 = new RegExp("^[ ](" + options.acceptConceptWords.join("|") + ")[ ]$");
        this.options = options;
    }
    isIn(name, value) {
        var options = this.options;
        return options[name].indexOf(value) >= 0;
    }
    isInConnectChars(value) {
        return this.isIn("acceptConnectChars", value);
    }
    isInStartQuotes(value) {
        return this.isIn("acceptStartQuotes", value);
    }
    isInEndQuotes(value) {
        return this.isIn("acceptEndQuotes", value);
    }
    isInPrefixes(value) {
        return this.isIn("acceptPrefixes", value);
    }
    isInConceptWords(value) {
        return this.isIn("acceptConceptWords", value);
    }
    getStartConceptWord(value) {
        let result = this.options.acceptConceptWordsRegex.exec(value);
        if (result) {
            return result[1];
        }
        return null;
    }
    formatConcept(context, input, i, start) {
        let text = input.substr(start, i - start - 1);
        return new concept_1.Concept({ value: text, index: start, lang: context.lang });
    }
    isValidWordChar(c) {
        if (utils.isLetterOrDigit(c)) {
            return true;
        }
        return this.options.acceptConnectChars.indexOf(c) > -1;
    }
    isValidStartWordChar(c) {
        return utils.isLetterOrDigit(c);
    }
    isWordsSeparatorChar(c) {
        return /[\s]/.test(c);
    }
    isLowerStartUpperWord(text, index) {
        let t = text.substr(index).trim();
        for (let i = 0; i < t.length; i++) {
            let c = t[i];
            if (!this.isValidWordChar(c)) {
                return false;
            }
            if (utils.isUpper(c)) {
                return true;
            }
        }
        return false;
    }
}
exports.BaseParser = BaseParser;