Annotation of /trunk/README

This project is side-effect of WebPac indexer -- I learned so much about
perl and swish-e during working on WebPac so I decided to make smaller
cgi interface to swish for general purpose web crawling and searching.


For this to work, you will have to create symlink from swish.cgi to
each configuration like this:

cd html
ln -s swish.cgi rot13.cgi

Then, rot13.cgi will use rot13.xml file as configuration.


This perl cgi font-end has following interesting features:
1. old (easy to understand :-) swish spider modified to support "no parent"
   URL (url above which it will stop spidering -- useful if you want to
   spider just your personal pages under some URL)
2. separate xml configuration file for one index (all uses same cgi script)
3. no need to design html pages (but, that limits you to one rather ugly
   design with some fill-in words -- someone could change that to use
   templates as I will probably in future)
4. support for searching using Lingua::Spelling::Alternative module which
   uses ispell affix or findaffix data to create variations of entered words
5. support for converting swish UTF-8 output to some other encoding on web
   using iconv.


For example how to create useful configuration file take a look at
included Makefile. In short, make_config.pl can create configuration
file for swish, xml configuration for cgi script and necessary symlink
by just:

$ make_config.pl name_of_config http://host/url_to_index/ [strip from url]

Optional "strip from url" will remove that part of the path when storing
in swish index. That enables you to create indexes which can later be merged
in one combined index easily. It's also useful if you want to index your
whole /doc directory and don't want that prefix in each and every entry in
index (which should save some space on disk too!).


Dobrica Pavlinusic <dpavlin@rot13.org> 2003-04-26
1	dpavlin	20	This project is side-effect of WebPac indexer -- I learned so much about
2			perl and swish-e during working on WebPac so I decided to make smaller
3	dpavlin	12	cgi interface to swish for general purpose web crawling and searching.
4
5
6			For this to work, you will have to create symlink from swish.cgi to
7			each configuration like this:
8
9			cd html
10			ln -s swish.cgi rot13.cgi
11
12			Then, rot13.cgi will use rot13.xml file as configuration.
13
14	dpavlin	20
15			This perl cgi font-end has following interesting features:
16			1. old (easy to understand :-) swish spider modified to support "no parent"
17			URL (url above which it will stop spidering -- useful if you want to
18			spider just your personal pages under some URL)
19			2. separate xml configuration file for one index (all uses same cgi script)
20			3. no need to design html pages (but, that limits you to one rather ugly
21			design with some fill-in words -- someone could change that to use
22			templates as I will probably in future)
23			4. support for searching using Lingua::Spelling::Alternative module which
24			uses ispell affix or findaffix data to create variations of entered words
25			5. support for converting swish UTF-8 output to some other encoding on web
26			using iconv.
27
28
29	dpavlin	32	For example how to create useful configuration file take a look at
30			included Makefile. In short, make_config.pl can create configuration
31			file for swish, xml configuration for cgi script and necessary symlink
32			by just:
33
34			$ make_config.pl name_of_config http://host/url_to_index/ [strip from url]
35
36			Optional "strip from url" will remove that part of the path when storing
37			in swish index. That enables you to create indexes which can later be merged
38			in one combined index easily. It's also useful if you want to index your
39			whole /doc directory and don't want that prefix in each and every entry in
40			index (which should save some space on disk too!).
41
42
43			Dobrica Pavlinusic <dpavlin@rot13.org> 2003-04-26