I recently took a trip from my office to Singapore’s Ministry of Manpower using Bus no. 2. As you see on the picture, there is a small lcd screen that indicates the number of seats available on the top-deck. It’s quite neat to save yourself from wasting time looking for available seats by peeking at the top-deck.

From the design of the lower-deck which provides wide spaces, I think this bus is specially designed to cater those on the wheelchair. Which was later confirmed when I saw the wheelchair sign in entrance door. This type of bus is mostly available in routes that serving City area (Chinatown, Orchard, Raffles City). I haven’t seen one in my ulu place (a.k.a. Jurong West):D

Note: ulu is a singlish adjective, very likely taken from Malay language, to describe a place that is so far away from ‘human civilization’ 😀 Well actually, any place in Singapore which is far from Orchard/City area usually labelled as Ulu. 😀

This post is copying the style that is used by Stephen Ostermiller in his article HERE.

Motivation
I am currently creating a RichTextBox control that will format and highlight the SQL string. I notice that the current regex that I used is not match correctly to a certain combination of SQL-style string.

Consider we have the following SQL statement:

First Try
‘.*’

This is incorrect. The regex should only match those within two single-quotes, not expand to the last single-quote.

(After n-Try)
‘.*?[^]’

This should be the correct one.

For those who want to practice their Regular Expression skill, try to download THIS software, it’s free and it’s really good for Regex practice.

Credit goes to Buddie for pointing out this PAGE.

UPDATE (22 Sep 2007)

As it turns out, I totally forgot that SQL-style string doesn’t use ‘ (backslash-quote) but uses ” (quote-quote). So things getting more complicated now.
So we need to change the string into something like this:

Another Try
‘[a-zA-Z ]*(”)*[a-zA-Z ]*’

But if we change the input string into

It becomes

Which is wrong.

Finally
‘([a-zA-Z ]*(”)*)*’

This should be the correct one.

Don’t ask why I come up with this post 🙂 Let’s just say it has something to do with Regular Expression.

The first method that immediately come up to mind (and usually the worst 🙂 ) is as follows:

struct Res
{
   //To Record the Result
   public String words;
   public int wordcount;
}

public static Res WorstWordByLengthSort(string raw)
{
   Res myresult = new Res();
   myresult.wordcount = 0;
   string[] words = raw.Split("n".ToCharArray());
   raw = "";
   foreach (string word in words)
      raw += word + " ";

   words = raw.Split(" ".ToCharArray());
   ArrayList ar = new ArrayList();

   foreach (string word in words)
   {
      string tstr = word.Trim();
      if (tstr == "")
         continue;
      myresult.wordcount++;
      if (ar.Count == 0)
      {
         ar.Add(tstr);
      }
      else
      {
          int count = ar.Count;
          bool inserted = false;
          for (int i = 0; i < count; i++)
          {
              if (ar[i].ToString().Length <= tstr.Length)
              {
                  if (!ar.Contains(tstr))
                  {
                      ar.Insert(i, tstr);
                      inserted = true;
                   }
              }
          }
          if (!inserted && !ar.Contains(tstr))
             ar.Add(tstr);
      }
   }
   StringBuilder sb = new StringBuilder();
   foreach (object o in ar)
       sb.AppendLine(o.ToString());
    myresult.words= sb.ToString();
    return myresult;
}

It will work flawlessly (sort of .. 🙂 ), but it will not allow duplication of word. After tinkering for a while, I came up with an idea to improve its performance. A better solution would be to use a dictionary where the length of the word becomes the key. If a key is already exist in the dictionary, we just simply append the word into the value of that particular key. The idea is implemented as follows:

struct Res
{
   //To Record the Result
   public String words;
   public int wordcount;
}

public static Res BetterWordByLengthSort(string raw)
{
   Res myresult = new Res();
   myresult.wordcount = 0;
   myresult.words = "";
   StringBuilder result = new StringBuilder();
   Dictionary<int, string> myDict = new Dictionary<int, string>();         
   ArrayList keys = new ArrayList();
   string[] words = raw.Split("n".ToCharArray());
   
   raw = "";
   foreach (string word in words)
   {
      string tword = word.Trim();
      raw += tword + " ";
   }
   words = null;
   words = raw.Split(" ".ToCharArray());
         
   foreach (string word in words)
   {
      string tempWord = word.Trim();
      if (tempWord == "")
         continue;
      int tlength = tempWord.Length;
      if (myDict.ContainsKey(tlength))
      {
         myDict[tlength] = myDict[tlength] + "n" + tempWord;
      }
      else
      {
          keys.Add(tlength);
          myDict.Add(tlength, tempWord);
      }
      myresult.wordcount++;
   }
   //Sort the keys ASC
   keys.Sort();
   for (int i=keys.Count-1; i>=0; i--)
   {
      result.AppendLine(myDict[(int)keys[i]]);
   }
   myresult.words = result.ToString();
   return myresult;
}

I created a GUI project to compare their performance. With same input of 1443 words, the Worst method took 734 ms, while the Better method took only 15 ms. And yes, if you remember your Big O complexity, the Better method is definitely much more efficient compared to the Worst method 🙂