Searchlight - Frequently Asked Questions

I wouldn't ay that the following is a list of frequent questions. They are, however, genuine.

Question

I am trying to install the 'Searchlight' on a client's web site and am a little confused - the code calls for 'Search.class' however of the files required for the applet, there only appears to be 'Search.exe'.

Answer

There's no file called Search.exe on my web site. I can only think that your browser renamed the file when you copied it down. Try again, this time refusing any attempt to rename the file. If the filename does change then rename it. You will have to copy down all of the class files and make sure that they are named correctly on the web site otherwise it won't work.

If you're using Windows 3.1 or DOS then it won't let you use 5 character file extensions. This shouldn't be a problem as long as the full file name is used on the web site.

Question

I searched for "Rankin!" but it did not find any instances although there are many. People come to my site specifically looking for Rankin!'s material. How do I get the module to find it?

Answer

I've not encountered anyone with punctuation in their name before, Searching for "Rankin" works where "Rankin!" fails. Searchlight considers "!" to be the end of a word in the search files but not in the search words. I'll consider the implications of ignoring the punctuation in the search words as well. Then your search should work. In the meantime, perhaps you could put a note on your search page saying don't use punctuation in the search strings.

Question

The searching seems a might slow. I'll try taking extra spaces and letters out of the index file. Any other tips?

Answer

The problem is that Searchlight has to read your files across the network. So, the smaller the better.

If your clients only want to search for, say, names, song titles or book titles then you could use the INDEX File facility. Create a file called PAGE.IDX and put the key words or phrases in it. Then use a search.idx entry like:

PAGE.HTML HTML INDEX PAGE.IDX My page

Searchlight will search PAGE.IDX, which should be quick, but display PAGE.HTML.

If you can assume that your clients know which state they wish to search then you can provide a search page for each state. Each page will have its own configuration file, referencing its own set of pages. When a search is performed, only the relevant state will be searched, not the whole country.

The best option for searching web-sites is to do so on the servers. Unfortunately, your web service provide (and mine) don't provide this facility. I hope that as more people use client searching (like Searchlight) then this will encourage the web service provides to provide us with server side searching. It will reduce the loading on their network links and improve web performance.

If you find a good server side search utility then please let me know so that I can recommend it to others.

Question

I get IO error message. What does this mean and how can I change this message to another?

Answer

This usually means that the file you are searching for is missing. You can change the "IO Error" message to, for example: "Something has gone wrong" with the following bit of HTML:

<applet code="Search.class" ....> 
<param name=badIOState value="Something has gone wrong"> 
: 
: 
<applet>

Question

... I have a problem. It keeps telling me that search.class file is not found when I know that it is in there.

Answer

The problem is that you have spelt "search.class", it should be "Search.class". It is conventional to name Java classes starting with a capital letter.

Question

I see that only the class files are available for download, is it possible to have a copy of the source code that made the class files?

Answer

In general, I'm not willing to give away the source. However, if an official of your university makes a formal request, assuring me that there will be no commercial gain then I may be willing to make it available.

Question

Please let me know whether and how I might use Searchlight properly on a CD-ROM.

Answer

I'm happy for you to use Searchlight on CDROM. You must keep me informed of your use of Searchlight and I will add your name to the user list and link to your web site if you have one. You must also put the logo and link on your search page.

There are a couple of organisations already using Searchlight on CDROM but I havn't seen them. I'm told that it can take some time to search a whole disc so you must be careful how you use it. If you have a lot of text to search you may want to implement several search pages, each dealing with a subset of the data. The configuration and use of Searchlight on CDROM is pretty much as described for network use. I can't think of anything else to add.

Question

The applet is unable to recognise &xxx as a character. What should I do?

Answer

Searchlight does not recognise this method of encoding charaters. I have added this to this list of possible enhancements. In the meantime, you may like to consider this tip from Sister Janice:

At the very end of my document, I wrote a series of key words, consisting of the accented words in the document, but without accents, very small and in white letters, (same color as background) so they are invisible. That way, Searchlight sees them and opens the file. It works fine.

Question

When Searchlight runs, the browser crashes. What should I do?

Answer

IE4 and IE5 sometimes crashes when running Searchlight. The problem seems to have been due to an error in the browser. However, the problem seems to have been fixed in Searchlight 5.0, where a new compiler was used.

If you do see this problem the please let me know.

For now I will leave the following text in this FAQ. It shows an earlier answer. It may help you workaround the problem if it should occur.

Question

Originally I had all the Searchlight files at the top level of the site, but I want to put it all into a single folder. I've restructured my .idx file as follows:

../index.html HTML FILE Home Page 
../site_contents.html HTML FILE Site Contents 
../news/index.html HTML FILE Latest News 
../news//archive/mac.html HTML FILE Archived Mac News

but all is not well. Does Searchlight have to be at the root level of the directory?

Answer

As applets may be loaded over a network, they are usually considered to be untrusted code. Web browsers restrict what untrusted applets are allowed to do. For example, they are not allowed to read any of the files on client machines.

An applet is not allowed to access files from any system other than the one from which it was loaded. In practice, this seems to include files accessed via directories higher than the one from which it was loaded.

This means that Searchlight has to be installed in the directory at the top of the search tree.

Question

The .idx file (search.idx) can get quite long. Is there a way to truncate similar filenames, or even bundle files which are under one directory?

Answer

No, each file has to be explicitly stated.

This was a deliberate decision. It means that the webmaster has to think about what he searches, so searching becomes more efficient. For example, if he could specify all the files in a directory with an asterisk. Then Searchlight could end up searching all sorts of files, e.g.: *.gif, *.jpg, robots.txt, *.bak. This is what I hope to avoid.

Question

I was wondering if and how I could use Searchlight to search material on other sites. For the filename, I just added the address. It comes out with a Java error saying that it couldn't connect.

Answer

No, Searchlight can't search other sites. Java security as implemented in browsers does not allow applets to read files from sites other than the one from which they are loaded.

Question

... when you click on a found link and then go back to click on another, the list is empty. Any suggestions?

Answer

You might consider using the "target" parameter to load the selected page into another window or frame. The applet will then stay on the screen and not be reset.