Dot Net Perls

Problem

You have a string that contains English words and you need to count the words in it similarly to how Microsoft Word counts words, which is the industry standard. Words are broken by punctuation, a space, or by being at the start or the end of string. Here are some required results.

C# Solution

Here I show two methods. Each way yields fairly similar results to Microsoft Word and Microsoft Office. This article presents algorithms that are fairly accurate and useful in different scenarios. First I show the Regex word count version, and then the loop word count version.

using System;
using System.Text;
using System.Text.RegularExpressions;

/// <summary>
/// This is original code that implements two word-count algorithms
/// that yield very similar results to Microsoft Word.
/// </summary>
static class WordCount
{
    /// <summary>
    /// Count the number of words in the string using regular expression.
    /// This method is more ACCURATE and SIMPLER.
    /// </summary>
    /// <param name="textIn">The string you want a word count of.</param>
    /// <returns>The number of words in the string.</returns>
    static public int CountRegex(string textIn)
    {
        MatchCollection collection = Regex.Matches(textIn, @"[\S]+");
        return collection.Count;
    }

    /// <summary>
    /// Count words (using spaces and character analysis).
    /// </summary>
    /// <param name="str">The string to count the words of.</param>
    /// <returns>The number of words in the document.</returns>
    static public int Count(string str)
    {
        int c = 0;
        for (int i = 1; i < str.Length; i++)
        {
            if (char.IsWhiteSpace(str[i - 1]) == true)
            {
                if (char.IsLetterOrDigit(str[i]) == true ||
                    char.IsPunctuation(str[i]))
                {
                    c++;
                }
            }
        }
        if (str.Length > 2)
        {
            c++;
        }
        return c;
    }
}

• Static class and methods
  This code is best stored in a static class because it doesn't maintain state or any data. It is just a process, not a machine.
• Methods receive strings
  Both approaches above receive a string and return an integer equal to the number of words they detect.
• First method uses Regex
  The first method here is better in every way except perhaps performance. It is shorter and simpler to maintain, and is also more accurate. (Sorry for spoiling the suspense.)
• Uses not-space code
  The backslash-S characters (\S) mean characters that are not spaces. So the first method considers each non-letter character to be part of a word, similar to Microsoft Word.

Microsoft Word Comparison

Microsoft Office dominates the business world, so I will provide some stats about the results of these 2 algorithms versus Microsoft Word 2007. Here are my results for five text documents.

Discussion

View these methods more easily (and copy them) at my source code site. The regular expression method here is better. It is clear to me that unless you have a strict performance requirement, the Regex method is easier to maintain and more accurate. It is probably very fast on large blocks of text. These methods should work with a variety of Romance languages.

Ads by The Lounge