1 |
dpavlin |
32 |
#!/usr/bin/perl -w |
2 |
|
|
# |
3 |
|
|
# create configuration for my version of swish crawler and web interface |
4 |
|
|
# |
5 |
|
|
# 2003-04-26 Dobrica Pavlinusic <dpavlin@rot13.org> |
6 |
|
|
# |
7 |
|
|
|
8 |
|
|
use strict; |
9 |
|
|
|
10 |
|
|
my ($name,$url,$stripurl) = @ARGV; |
11 |
|
|
|
12 |
|
|
if (!$name || !$url) { |
13 |
|
|
print "Usage: $0 name URL\n"; |
14 |
|
|
exit; |
15 |
|
|
} |
16 |
|
|
|
17 |
|
|
my $dir=$0; $dir=~s/\/[^\/]+$//; |
18 |
|
|
|
19 |
|
|
# try to deduce working directory if script is not called with absolute path |
20 |
|
|
if ($dir !~ m#^/#) { |
21 |
|
|
chomp($dir = `pwd`); |
22 |
|
|
} |
23 |
|
|
|
24 |
|
|
my ($host,$urlpath); |
25 |
|
|
|
26 |
|
|
if ($url =~ m#(http://[^/]+)(/.+)$#) { |
27 |
|
|
($host,$urlpath) = ($1,$2); |
28 |
|
|
if ($stripurl && $urlpath =~ m/^(.*$stripurl)/) { |
29 |
|
|
$stripurl = $1; |
30 |
|
|
} else { |
31 |
|
|
$stripurl = ""; |
32 |
|
|
} |
33 |
|
|
} else { |
34 |
|
|
print "Can't parse URL $url into host and path!\n"; |
35 |
|
|
exit 1; |
36 |
|
|
} |
37 |
|
|
|
38 |
|
|
my $baseurl = $url; |
39 |
|
|
$baseurl =~ s#/[^/]+$#/#g; |
40 |
|
|
|
41 |
|
|
print "Config name: $name\nCrawling url: $url [strip $stripurl]\n"; |
42 |
|
|
|
43 |
|
|
|
44 |
|
|
my $config = <<"EOF"; |
45 |
|
|
|
46 |
|
|
################################################### |
47 |
|
|
IncludeConfigFile $dir/common.config |
48 |
|
|
|
49 |
|
|
# this is a cludge to implement no parent URL feature in swish indexer |
50 |
|
|
IndexDir "$url $baseurl" |
51 |
|
|
ReplaceRules replace " $baseurl" "" |
52 |
|
|
ReplaceRules replace "${host}${stripurl}" "" |
53 |
|
|
|
54 |
|
|
IndexFile $dir/index/$name |
55 |
|
|
|
56 |
|
|
StoreDescription HTML <body> 500 |
57 |
|
|
|
58 |
|
|
# not very usefull, but... |
59 |
|
|
MetaNames keywords description |
60 |
|
|
|
61 |
|
|
# store <title> |
62 |
|
|
PropertyNameAlias swishtitle title |
63 |
|
|
|
64 |
|
|
EOF |
65 |
|
|
|
66 |
|
|
my $xml = << "EOF"; |
67 |
|
|
<config |
68 |
|
|
max_hits="1000" |
69 |
|
|
prog="/usr/bin/swish-e" |
70 |
|
|
index="$dir/index/$name" |
71 |
|
|
charset="iso-8859-2" |
72 |
|
|
title="$name search" |
73 |
|
|
affix="/usr/lib/ispell/american.aff" |
74 |
|
|
url="$stripurl" |
75 |
|
|
> |
76 |
|
|
<labels> |
77 |
|
|
<label value="10"> 10</label> |
78 |
|
|
<label value="100"> 100</label> |
79 |
|
|
<label value="0">unlimited</label> |
80 |
|
|
</labels> |
81 |
|
|
<text> |
82 |
|
|
<search>Search for </search> |
83 |
|
|
<documents> documents containing words: </documents> |
84 |
|
|
<submit>Search</submit> |
85 |
|
|
<no_spell>don't use alternative spellings </no_spell> |
86 |
|
|
<no_properties>don't display results details</no_properties> |
87 |
|
|
<hits>Showing %s documents (of maximum %d)... (%s)</hits> |
88 |
|
|
<no_hits>Can't find any documents (%s, %s)</no_hits> |
89 |
|
|
<footer>Searcher will try to create different variations of words using spelling dictionary. If you enter a word with minus (-) before it, it will exclude documents with this word and plus (+) will do the opposite (e.g. -work +play)</footer> |
90 |
|
|
</text> |
91 |
|
|
<!-- additional data to show summary --> |
92 |
|
|
<properties>swishdescription swishdocsize</properties> |
93 |
|
|
<hit><![CDATA[ <a href="%s">%s</a> [%s]<br> |
94 |
|
|
<font color=gray size=-2>%s ... <i>%s bytes</i></font><br> |
95 |
|
|
]]></hit> |
96 |
|
|
|
97 |
|
|
</config> |
98 |
|
|
EOF |
99 |
|
|
|
100 |
|
|
print "Creating $dir/$name.config\n"; |
101 |
|
|
open(C,"> $dir/$name.config") || die "can't open $dir/$name.config: $!"; |
102 |
|
|
print C $config; |
103 |
|
|
close(C); |
104 |
|
|
|
105 |
|
|
print "Creating $dir/html/$name.xml\n"; |
106 |
|
|
open(C, "> $dir/html/$name.xml") || die "can't open $dir/html/$name.xml: $!"; |
107 |
|
|
print C $xml; |
108 |
|
|
close(C); |
109 |
|
|
|
110 |
|
|
print "Creating symlink to $dir/html/$name.cgi\n"; |
111 |
|
|
symlink "$dir/html/swish.cgi","$dir/html/$name.cgi" || die "can't create symlink $dir/html/$name.cgi: $!"; |
112 |
|
|
|
113 |
|
|
#print "Index this with:\nswish-e -S http -c $dir/$name.config\n"; |
114 |
|
|
#print "Search using $name.cgi\n"; |