Javascript parseFloat '1.23e-7' gives 1.23

问题:

parseFloat(1.51e-6);
// returns 0.00000151

parseFloat(1.23e-7);
// returns 1.23e-7
// required 0.000000123

I am sorting table columns containing a wide range of floating-point numbers, some represented in scientific notation.

I am using the jQuery tablesorter2.0 plugin which is using 'parseFloat' for cells that start with a digit. The issue is that parseFloat returns very small numbers represented as 1.23e-7 as a string and is not expanding this to 0.000000123. As a result tablesorter sorts the content of the column as text instead of numerics.

**Column To Sort**
2.34
1.01
13.56
1.23e-7

**After Sort Now**
1.01
1.23e-7
13.56
2.34

**Expect**
1.23e-7
1.01
2.34
13.56

Is there an efficient way of representing very small scientific notation numbers as expanded floating-point numbers?

Solution:

tablesorter determines how to sort a column based on the first of tablesorters automatic parsers to return true for the content of a cell in that column. If the cell contained 1.23e-7 than it defaulted to sort by text because the 'digit' parser does not interpret this as a number.

So to workaround, the following code represents the scientific notation number as a string that tablesorter can interpret/parse as a digit and so ensures numerical sorting on the column. @bitplitter - thanks for the toFixed() tip.

var s = "1.23e-7";
// Handle exponential numbers.
if (s.match(/^[-+]?[1-9]\.[0-9]+e[-]?[1-9][0-9]*$/)) {
  s = (+s).toFixed(getPrecision(s));
}
//returns 0.000000123

// Get a nice decimal place precision for the scientific notation number.
// e.g. 1.23e-7 yields 7+2 places after the decimal point
// e.g. 4.5678e-11 yields 11+4 places after the decimal point
function getPrecision(scinum) {
  var arr = new Array();
  // Get the exponent after 'e', make it absolute.  
  arr = scinum.split('e');
  var exponent = Math.abs(arr[1]);

  // Add to it the number of digits between the '.' and the 'e'
  // to give our required precision.
  var precision = new Number(exponent);
  arr = arr[0].split('.');
  precision += arr[1].length;

  return precision;
}

回答1:

Even though the OP has posted his solution, I'd like to share a rather simpler solution I stumbled upon, which is based on the parsers in the tablesorter source code and the regex given by JasonS on another question.

// add parser through the tablesorter addParser method 
$.tablesorter.addParser({ 
// set a unique id
id: 'scinot', 
is: function(s) { 
    return /[+\-]?(?:0|[1-9]\d*)(?:\.\d*)?(?:[eE][+\-]?\d+)?/.test(s); 
}, 
format: function(s) { 
    return $.tablesorter.formatFloat(s);
}, 
type: 'numeric' 
});

It works on my tables with pretty much all values given in scientific notation. It auto-detects (the is: part) and correctly sorts multiple fields. Hope it helps others who might stumble upon this question.

回答2:

You can use toFixed() instead of parseFloat() to format the numbers the way you want. For example (1.23e-7).toFixed(9) will render as 0.000000123

To be able to sort these with sorts default string comparison, make sure to prefix all of them with zeroes and make them all the same size so the decimal dots wil line up.

You can extend the string object with padLeft like this:

String.prototype.padLeft = function(len, char){
var prefix = '';
var length=this.length;
if(len>length)
 for(var i=0;i < len-length;i++) prefix += char;
return prefix + this;
}

Now you can call ((1.23e-7).toFixed(9)).padLeft(15,'0')

回答3:

the problem is not parseFloat, but sort that uses string comparison by default. Try enforcing numeric comparison:

a.sort(function(x, y) { return x - y })