Our solution is based principally on two of VBScript's objects: The FileSystemObject, in charge of retrieving the target pages' text, and the RegExp object, to do the actual search and to extract the document's title. We encapsulated the search functionality within two self-contained procedures to allow us flexibility in the search page design. This means that you can change the search page to match the look and feel of your site without requiring major changes in the code.

We begin by creating our starting procedure. This is the procedure we call to start the search process. It takes a single parameter, SearchString, that will hold our search criteria. First, we do a standard instantiation of the objects. Second, we set up the RegExp objects, and here's where the magic begins. By setting the RegExp's Global property we instruct the object to find every match of our search pattern. If we set this to False, as is the case of the GetTitle object, the search stops at the first match found. The IgnoreCase property should be self-explanatory, this simply instructs the object to do case insensitive searches. The Pattern property is where we state the search expression. Note the difference between Regex.Pattern and GetTitle.Pattern below. In the former we just feed the content of the SearchString parameter as it came from the user. In the later we construct a special pattern to match text enclosed in <title> tags. Observe in the code window below the special metacharacters right between <title> and </title>. We use parenthesis to change the order of precedence, the . match any single character except the new line character (In VBScript this would be vbCrLf). The \n match the new line character. The pipe character | in between indicates an or, and the asterisk * indicates to match zero or more of the preceding characters. In summary, this pattern will match anything (any amount of characters or new lines) between <title> and </title>. Third, we make sure that our paths variables contain their trailing slashes as we will be using these as the base path for our matched documents. Fourth, we start the actual search process by calling the SearchFiles procedure. And fifth and last, we display a message if no matches were found and we do some object cleaning. Find below the code for our starting procedure.

~~*~~

The next part of our project is the search engine itself. This engine is in the form of a self calling procedure, otherwise known as recursive. We decided to implement the engine as a recursive procedure to simplify the process of traversing a directory tree. Note that in a recursive procedure, a new and independent set of variables and objects are created each time it is called. First, we get the current 'root' folder where files and other folders may exist. Then we iterate thru each file in the folder. We then compare each file's extension to a global variable (not shown) holding a list of extensions for valid files (e.g. html, asp, txt, etc.). If a match is found, the file is opened to get the text contained inside, and the RegExp search is applied. If the search returned one or more matches we then proceed to try and get hold of the document's title by executing the GetTitle RegExp search. This, of course, will only return something for HTML and some ASP files. If we find a title, we use this as our results entry text, otherwise we use the file name. Note that we need to strip out the <title> tags. In version 5.5 of the scripting engine (as found in Windows 2000) a SubMatches object is available, returning what's inside the entities called captured matches, a pattern enclosed in parenthesis, avoiding the need to prepare the match manually. Unfortunately, there's no SubMatches object in the more popular versions 4 or 5 of the scripting engine. Anyway, once we got our entry's name, we proceed to construct the line that will be displayed on our results page. We add some miscellaneous (also known as fancy or mostly useless) information to the entry, and do some html-formatting as we go. Check out the somewhat commented code to the recursive procedure below.

As you can see, it is very easy to create a simple search engine without expending big bucks on third-party solutions. Bear in mind that this is a very simplistic approach to the search engine problem. Aside from the fact of the absent-minded nature of this engine (it will match text inside code procedures or text inside html tags, something not always desirable), a robust solution would index each file in a separate process and store the information in a database for fast retrieval. Even thought, the solution presented here is sure to satisfy many web developers in need of a simple search facility, and it sure demonstrate what can be done with the sometimes neglected tools available in every ASP developer's toolbox.

Feel free to send your comments and suggestions to sixto@tecnoapoyo.com (threat mail is strongly discouraged).