compromise/src/terms/match/lib/syntax.js

natural language processing in the browser

'use strict';
// parse a search lookup term find the regex-like syntax in this term
const fns = require('./paths').fns;

//turn 'regex-like' search string into parsed json
const parse_term = function (term, i) {
  term = term || '';
  term = term.trim();
  let reg = {};
  //order matters..

  //negation ! flag
  if (fns.startsWith(term, '!')) {
    term = term.substr(1, term.length);
    reg.negative = true;
  }
  //leading ^ flag
  if (fns.startsWith(term, '^')) {
    term = term.substr(1, term.length);
    reg.starting = true;
  }
  //trailing $ flag means ending
  if (fns.endsWith(term, '$')) {
    term = term.replace(/\$$/, '');
    reg.ending = true;
  }
  //optional flag
  if (fns.endsWith(term, '?')) {
    term = term.replace(/\?$/, '');
    reg.optional = true;
  }
  //atleast-one-but-greedy flag
  if (fns.endsWith(term, '+')) {
    term = term.replace(/\+$/, '');
    reg.optional = false;
    reg.consecutive = true;
  }

  //pos flag
  if (fns.startsWith(term, '#')) {
    term = term.replace(/^\#/, '');
    reg.tag = [fns.titleCase(term)]
    term = null;
  }
  //one_of options flag
  if (fns.startsWith(term, '(') && fns.endsWith(term, ')')) {
    term = term.replace(/\)$/, '');
    term = term.replace(/^\(/, '');
    reg.oneOf = term.split(/\|/g)
    term = null;
  }
  //min/max any '{1,3}'
  if (fns.startsWith(term, '{') && fns.endsWith(term, '}')) {
    let m = term.match(/\{([0-9]+), ?([0-9]+)\}/);
    reg.minMax = {
      min: parseInt(m[1], 10),
      max: parseInt(m[2], 10)
    };
    term = null;
  }
  //a period means any one term
  if (term === '.') {
    reg.anyOne = true;
    term = null;
  }
  //a * means anything until sentence end
  if (term === '*') {
    reg.astrix = true;
    term = null;
  }
  reg.normal = term;
  if (reg.normal) {
    reg.normal = reg.normal.toLowerCase();
  }
  return reg;
};

//turn a match string into an array of objects
const parse_all = function (reg) {
  reg = reg || ''
  reg = reg.split(/ +/);
  return reg.map(parse_term);
};
// console.log(parse_all(''));

module.exports = parse_all;