Search This! - Configuring ht://Dig (
Page 2 of 6 )
Before going any further, it would help if you have a clear idea of how ht://Dig
works. There is a very nice explanation on the ht://Dig website (follow the link
to "How it Works" on the left side), but here's the abridged version.
The ht://Dig system performs three major tasks that should be performed in
the following order:
Digging
Before you can search, a database of all the documents that need to be
searched has to be created.
Merging
Once the document database has been created, it has to be converted to
something that can be searched quickly. Also, if you want to only update changed
documents, these changes have to be merged into the searchable database.
Even though this task could be performed at the same time as digging, it is a
separate process for efficiency reasons. It also gives more flexibility to what
actually happens at merge time.
Searching
Finally, the databases that were created in the previous steps can be used to
perform actual searches. Normally, searches will be invoked by a CGI program
which gets its input from the user through an HTML form ... but we'll be doing
it all with PHP3!
So, let's see what the installation created, shall we?
% cd /usr/local/htdig/
% ls
bin common conf db
The bin/ directory contains the executables required by
ht://Dig. Configurations files are in conf/ (believe it or not!)
and I think you can guess where the database files are.
[Note: I'm going to assume from now on the you installed ht://Dig in the
/usr/local/htdig directory as shown above, and that the path to
your website files is /www, probably a symlink to
/usr/local/htdocs.]
If your web server is like mine, you're probably running several web sites
off the same machine. In this case, I've found it useful to have separate
configuration files for each site. ht://Dig installed a sample configuration
file called htdig.conf, so I recommend you copy it to a new
filename and make some changes. Here is the sw98.conf file I used
for the SummerWorks Theatre Festival's website:
database_dir: /usr/local/htdig/db/sw98
start_url: http://www.summerworks.on.ca/
limit_urls_to: http://www.summerworks.on.ca/
exclude_urls: /staff/ /search/ .inc .doc .mcw
search_algorithm: exact:1.0 endings:0.3
matches_per_page: 50
excerpt_length: 200
template_map: sw98 sw98 /www/summerworks/search/results-template.html
search_results_header: /www/summerworks/search/results-header.html
search_results_footer:
nothing_found_file: /www/summerworks/search/results-nomatch.html
syntax_error_file: /www/summerworks/search/results-syntaxerror.html
star_image: http://www.summerworks.on.ca/gifs/x-star.gif
star_blank: http://www.summerworks.on.ca/gifs/x-nostar.gif
max_stars: 5
max_doc_size: 100000
valid_punctuation: .-_/!#$%^&*'«»"
What do all these settings mean? Some are obvious, but the following probably
are not:
database_dir:
Normally all database files are in the
/usr/local/htdig/db/ directory. To keep things a bit cleaner, I've
made sub-directories for all the sites on my server, so this is the
sub-directory for the SummerWorks site. You'll need to make this directory
manually before you can index the site:
% cd /usr/local/htdig/db
% mkdir sw98
exclude_urls:
There are some files we don't want to search. If a
URL contains any of the space separated patterns, it will not be indexed. In
this case, I didn't want the staff pages, nor the search pages indexed. I also
wanted it to ignore .inc files (common PHP code that I include on
various pages), and some Word documents for PC and Mac. On the real SummerWorks
site, there are a couple more exclusions, but you get the idea (I hope!).
search_algorithm:
ht://Dig lets you turn "fuzzy" searching on or
off. When it's on, people who search for "play", for example, will also find
things like "plays", "player", "players", etc.. This slows down the digging
process a fair amount, but can be useful. The different weights applied to each
algorithm mean, in this case, that a search for "play" will turn up "player",
but not as high in the results as an exact match for "play".
template_map:
This makes a new template called "sw98". More on
templates below.
search_results_header, search_results_footer,
nothing_found_file, and syntax_error_file:
These are files used
by the template, and parsed by PHP. More below.
max_doc_size:
This is bigger than usual, since there are some big
pages on the site. If you find that some of your pages aren't being indexed,
setting this to a higher value will often solve the problem.
valid_punctuation:
I basically only added the French quotes and the
single quote to the list of valid punctuation. This is the set of characters
which will be deleted from the document before determining what a word is. This
means that if a document contains something like "Andrew's" the digger will see
this as "Andrews".
The same transformation is performed on the keywords sent to the search
engine.
There are plenty of other setting you can set in the .conf file.
For a definitive list with explanations, check out the ht://Dig website.
Templates
ht://Dig makes extensive use of templates for it's output. These templates
are (usually) plain HTML documents containing variables, which are substituted
with the search results.
There are four "standard" template setting in the .conf file:
search_results_header:
This specifies a filename to be output at
the start of search results.
search_results_footer:
This specifies a filename to be output at
the end of search results.
nothing_found_file:
This specifies the file which contains the text
to display when no matches were found.
syntax_error_file:
This points to the file which will be displayed
if a boolean expression syntax error was found.
There is also the result template file. This is the file referenced in the
template_map attribute of the .conf file. This is where all
the information you want the search to return is displayed.
I said that these templates usually contain a bunch of HTML, with some
variable substitution. When using ht://Dig with PHP, it would be much easier if
a search returned the raw results of the search. You can then use PHP to parse
and display that information however you want.
The easiest way to do this is to not put any HTML in the template files, but
instead just put in the list of variables you want substituted.
So, make your results-header.html file contain only the
following 4 lines:
$(MATCHES)
$(FIRSTDISPLAYED)
$(LASTDISPLAYED)
$(LOGICAL_WORDS)
results-nomatch.html should contain only 1 line (note that it's
not a variable):
NOMATCH
results-template.html consists of 4 lines:
$(TITLE)
$(URL)
$(PERCENT)
$(EXCERPT)
Finally, results-syntaxerror.html consists of 2 lines:
SYNTAXERROR
$(SYNTAXERROR)
It's important that there be no extra line breaks at the beginning or end of
your files, since we're going to rely on the fact that every fourth line of the
results-template.html file (for instance) is the title of the
document.
All of the template files are in a sub-directory of the website called
search/.
If you want to find out what all the variables mean (and what other ones are
available), check the ht://Dig site.
/ 24 
