docparse-supplier-nge/lib/parse/otherCosts/processCostOtherText.js

process ngrid electric utility bill data for use in the docparse system

var num = require('num');
var _ = require('underscore');
var inspect = require('eyespect').inspector({maxLength:20000000});
module.exports = function(text, cb) {
  var lines = text.split(/[\r\n]+/);
  var skipPattern = /Other Charges\/Adjustments/;
  var totalPattern = /Total Other (?:(?:Charges)|(?:Credits)|(?:Adjustments))/i;
  var totalLine = null;
  lines = _.filter(lines, function(line) {
    if (totalPattern.test(line)) {
      totalLine = line;
      return false;
    }
    if (skipPattern.test(line)) {
      return false;
    }

    return true;
  });

  var costs = [];
  var total = num(0);
  _.each(lines, function(line) {
    var cost = processLine(line);
    if (cost) {
      costs.push(cost);
      total = total.add(cost.value);
    }
  });

  if (!totalLine) {
    return cb('total other charges line not found in other charges text');
  }
  var printedTotal = getCostOtherTotal(totalLine);
  if (printedTotal === null) {
    return cb('printed total other charges value not found in total other charges line text: ' + totalLine);
  }
  var match = doesPrintedTotalMatchCalculated(printedTotal, total);
  if (!match) {
    var errData = {
      printedTotal: printedTotal,
      calcTotal: total.toString(),
      costs: costs
    };
    inspect(errData, 'printed total other cost does not match calculated total');
    // use the printed charges
    costs = [
      { description: 'Unabled to parse cost description',
        value: printedTotal
      }
    ]
    total = num(printedTotal);
  }
  var output = {
    costs: costs,
    total: total.toString()
  };

  return cb(null, output);
};

function processLine(text) {
  text = text.trim();
  var pattern = /(.*?)(-?\d+\.?\d+)\s*$/;
  var matches = text.match(pattern);
  if (!matches || matches.length !== 3) {
    inspect('wrong number of matches');
    return null;
  }
  var description = matches[1].trim();
  description = description.replace(/\s{2,}/, ' ');
  var value = matches[2];

  var output = {
    description: description,
    value: value
  };
  return output;
}

function getCostOtherTotal(text) {
  text = text.replace(/\s/g, '');
  var pattern = /.*?(-?\d+\.?\d+)$/;
  var matches = text.match(pattern);
  if (matches.length !== 2) {
    inspect('wrong number of matches when gettting printed total other charges');
    return null;
  }
  return matches[1];
}

function doesPrintedTotalMatchCalculated(printed, calc) {
  if (calc.eq(printed)) {
    return true;
  }
  var diff = calc.sub(printed);
  var absoluteDiff = diff.abs();
  var epsilon = num('0.1');
  if (absoluteDiff.lte(epsilon)) {
    return true;
  }
  return false;
}