 ... from the developers of Ignite |
| ... from the developers of Ignite |
Fluffy Search must be configured for your own web site. This is quite simple. You only need to edit two files.
After you have unpacked the Fluffy Search archive into a directory of it's own, you need to alter the .config file. The archive contains two starting points, fluffysearch.config.unix and fluffysearch.config.nt, for UNIX and NT based systems respectively. Choose the correct one for your web server, rename it fluffysearch.config and delete the other.
Load fluffysearch.config into your text editor. This is a perl script, but don't let that worry you. It's really quite simple.
Each configurable option is defined as a perl variable. These are always in the form:
$name = 'value';
where name is the name of the config option, and value is what you want to set it to. (Some options are set in slightly different ways, but we'll discuss them as we come to them. And as an exception, if you set an option to a number, you don't need the quotes, shown in red above.)
Note that all directory names do not have ending '/' or '\'s. If you put them in, things may not work.
We'll now go through each option in turn.
Set this to the full directory name of the directory on the disc in which your documents start. This is the directory that you FTP your documents too.
However, what you see when you FTP in might be misleading, since the average web server doesn't place you in the root directory of the server when you log in, but instead in your 'home directory'.
So, for example, you might see your document root as public_html but in fact it's really /home/user/public_html on a UNIX system. You need to use this fully qualified path, which starts with '/'. Typing pwd at an FTP prompt might give you the answer you need, but if you're not sure, ask your system administrator.
On an NT based server, it's a little different. You might have a root which looks like c:/inetpub/wwwroot. (NT usually uses '\' as a directory separator, but you can use '/' instead in the config file. It helps to avoid confusion, since the web based paths must use '/' whatever platform the scripts are running on). Again, ask your administrator if you're not sure.
Set this to the name of the above directory, as seen on the web server. For example, if it's seen as http://www.example.com/subsite/ set this option to /subsite. If it's just http://www.example.com/ set it to nothing (ie just have $docroot_web = '';).
docroot_disc and docroot_web are the most important options, as they specify where the indexer can find the documents, and where the documents appear on the web server respectively. If they aren't set correctly, then it simply won't work, either because references to documents won't work, or no files will be found to index.
If you are searching your entire web site -- you want to seach all document from http://www.example.com/ -- then you should set docroot_disc to the directory you upload files to for the root of the web server, and set docroot_web to nothing ('').
However, if you're searching in and below a subdirectory, let's say that you only want to search a product catalogue which is located at http://www.example.com/products/ and below, then you need to set docroot_disc to the location of the products directory on the web server (the directory you FTP files to). You must then set docroot_web to /products (note there's not final /) otherwise Fluffy Search will think that the files it's searching are located at http://www.example.com/, which will obviously cause problems.
The directory to store the index files in. Make sure you have permissions to write files inside it, and that it's not inside the document root. Warning All other files within this directory will be deleted by the indexer, so don't use a directory with other files in it!
The URL of the search script as seem from a web browser, ie what you'd type in after the server name to retrieve the page. For example, if the URL of the script is
http://www.example.com/cgi-bin/fcs/fluffysearch.pl
you need to enter /cgi-bin/fcs/fluffysearch.pl for this option.
Similarly for the fcp.pl script. For example, the script might be
http://www.example.com/cgi-bin/fcs/fcp.pl
and you will need to enter /cgi-bin/fcs/fcp.pl.
Used to set the target for the links. If you're not using frames, leave this blank. Otherwise, set it to
' target="_top"'
(note there's a space before the word target) replacing _top with the name of the frame whose contents you want replace with the generated frameset.
If you're using frames, set this to the name of the frame that contains the content in your framesets. If you're not using frames, set it to nothing to disable this feature.
The generic rule which tells Fluffy Search how to locate the file containing the appropriate frameset, given a HTML file. Set this to the href you'd use if you were linking to it from the HTML files using a relative link. For example, to specify the file index.html in the directory above, use ../index.html
You can specify exceptions to the rules specified by the above two options using this option. You can also use it to specify all the locations of every frameset if you require -- but in this case, make sure frames_framename is set to something other than nothing.
This option is specified slightly differently to the others. You place the entries, one per line, between $frames_exceptions = <<__ENDEXCEPTIONS;> and __ENDEXCEPTIONS. They are formatted as
source_directory_name frameset_name frame_name
where source_directory_name is the name of the directory we're specifying the frameset for, frameset_name is the filename of the file containing the frameset, and frame_name is the name of the frame containing the content. For example:
/shopping/socialcare /shopping/socialcare/index.html main
All directory and filenames are relative to the search root. You can have as many entries as you need.
This is specified in a similar way to the frames exceptions, with one perl pattern per line. You don't have to worry about the way things are specified if you don't understand perl patterns. The files come set up to index .htm, .html and .shtml files. If you need to change this, delete the entries you don't need, and add extra ones for extra file types. To include a new file type, add an entry like:
\.ext\Z
where ext is the extension you want to index.
Using this option, you can specify exceptions to the above rule. For example, use \Atoc\.html\Z to exclude all files called toc.html.
When a page is returned with highlighting, the web server won't expand SHTML includes for security reasons, so the page script needs to perform this operation. You need to set this option to the root directory for shtml includes. This will probably be the same as docroot_disc if you are indexing from the root level of your web server.
The password to start indexing from a web browser.
If you aren't using the <fcs_ni> tags to stop the indexing of certain areas of the pages, set this to 1 for slightly faster indexing.
Highlighting simply inserts HTML tags at the beginning and end of the text to be indexed. These two options specify what is inserted to create the highlighted text. You can use anything you want here.
How many results to return in each page of found pages.
The command to run the indexer using the fluffyindex.pl script from a web browser. You may not need to change this, but it is the system command used to run the script fluffymkindex.pl. It may be that on a UNIX system you might need to specify the full path to this script, or use something like ./fluffymkindex.pl depending on how you system is set up.
Under NT, you might need to use <full-path-to-perl.exe> <full-path-to-fluffymkindex.pl> to get this to run. If you are running IIS 4, you may need to make some configuration alterations. See this info on using perl to run external commands for details. If you're running IIS 3, this'll be more help.
If you site is especially large, increase this number to 4, 8 or 13 to create a more granular index, that is increase the number of files written. This should speed up searching on larger sites. Only change if you think there is a performance problem.
A word is located in the index by looking at its first and second letters. Taken together, they indicate which file within the index directory the references to the keyword are found in, giving a filename like a.3. This configuration parameter determines how many different sections (the granularity of the index) the second letters are split up into -- the higher the value the more files are created, and the less data stored in each. This means that the search script has to search through less data to find the keyword it needs, and so the search becomes faster. Increasing this parameter on small sites may decrease performance, due to the relative inefficiency of storing lots of small files in one directory.
For ranking the pages, if a word is found in the meta keywords tag, it counts this number of times more than if it was found in normal text.
This file defines the header and footer for the search page. Within it, ##### is replaced by the search form and/or results. All references to files (for images and links etc) must be absolute (that is, start with '/' to refer to the root of the web server).
If you need to change the look for the form itself, look at the bottom of fluffysearch.pl -- but be careful what you change.
At the top of all the .pl files, there is a line which reads #!/usr/bin/perl. This specifies the location of perl on UNIX machines (and some Windows machines running Apache). It must be set correctly for the scripts to run.
If the scripts won't run at all (maybe getting a 500 Internal Configuration Error) these lines may need altering -- although on most UNIX systems this default is correct, and NT IIS servers ignore it. Change the line to #![location of perl interpreter]
Fluffy Search is unsupported. However, if you email fluffysearch@fluffy.co.uk we'll try and help you out, although we can't promise an immediate response. You must include a copy of your .config file in any request for help. Emails without this will be ignored, as it is difficult to diagnose problems without seeing this file.
 |
Need blazingly fast downloads? Existing tools a bit clumsy?
Use Ignite to create
better, faster, web graphics in less time. Download your free 30 day trial now! |
(c) Fluffy Clouds Ltd 1999 | home page