<font color="#800000">Note: the changes below are built in to Glimpse 4.14 and Webglimpse 2.1.01 and above; no manual changes are necessary with those versions</font>

How to configure Glimpse with International Charsets

Instructions written by Reuven Lerner, <A HREF="http://lerner.co.il">http://lerner.co.il</A>
Thanks to Reuven and to Didi Melchior for making this available!

(1) Modify the file Makefile.linux, such that Glimpse will be able to
    index international characters.  The line that currently reads

        ISO_CHAR_SET = 0

    should be changed to read

        ISO_CHAR_SET = 1


(2) Compile Glimpse by saying

        make -f Makefile.linux

    or the appropriate makefile for your system.

    If you have already compiled Glimpse in the past, then you will
    first want to "clean" out the previous compilations with:

        make -f Makefile.linux clean

(3) Download and install Webglimpse.  (http://webglimpse.net/download.html)

(4) You will now have to tell Glimpse to pay attention to the
    "locale," which is the Unix term for "what country you're in, and
    thus what characters should be considered letters."  The locale
    for Hebrew is "iw_IL", and is described in the directory
    /usr/share/local/eiw_IL.  The contents of this directory are as
    follows:

        LC_COLLATE   LC_CTYPE    LC_MESSAGES
        LC_MONETARY  LC_NUMERIC  LC_TIME

    If LC_CTYPE and LC_COLLATE aren't there, you will probably have
    problems.

(5) In Webglimpse, you will have to edit the files "confarc",
    "makecron", "makehn", and "webglimpse".  (The latter is a program
    in the cgi-bin directory.)

    Each of these files must have the following two lines added at the
    top:

        $ENV{'LANG'} = 'iw_IL';
        $ENV{'LC_ALL'} = 'iw_IL';

(6) Remove the existing index with rmarc, and build it again with
    confarc.  Specify all subdirectories to enable search by
    directory.  So you can specify /home/kalir/www,
    /home/kalir/www/Articles/, and /home/kalir/www/Articles-Havot/.

(7) To enable searching in Hebrew, you will have to modify the
    "webglimpse" program itself, commenting out the following two
    lines by adding a # character at the beginning of each line:

       #$highlight =~ s/^\W+//;
       #$highlight = join("|",split(/\W+/,$highlight));

(8) The $ind_dir variable in "webglimpse" should contain the name
    of the HTML root directory.  For example, this might be
    "/usr/local/httpd/htdocs"

(9) There are several important files to keep track of, and which you
    may want to modify:

    (a) wgindex.html, which is the form for searching through *all*
        indexed files.
    (b) wgall.html, an HTML form which allows you to search by
        directories.
    (c) archive.cfg, configuration information about indexed
        directories.  You probably won't have to change this.

(10) Every time you index a directory, check the file .glimpse_index
     (with an underscore, not a hyphen) to ensure that words --
     particularly Hebrew words -- were included.  If Hebrew words
     weren't included in the index, then searches won't return any
     results.  (This file will be in your archive directory, currently by default 
     may be the same directory you indexed.)

(11) In the same directory as .glimpse_index is .wgfilter-index (with
     a hyphen, not an underscore) which removes certain file
     extensions from the indexing.  For example, you probably don't
     want to index Perl programs, so you can remove everything ending
     with .pl.  You probably won't need to change this, but it exists
     if you need it.


(12) Another important thing is to use:
	<INPUT NAME=nonascii TYPE=hidden value=on>
     in the HTML search form.