swish/trunk/make_config.pl

#!/usr/bin/perl -w
#
# create configuration for my version of swish crawler and web interface
#
# 2003-04-26 Dobrica Pavlinusic <dpavlin@rot13.org>
#

use strict;

my $name = shift @ARGV;
my $url = shift @ARGV;
#my $stripurl = shift @ARGV;
my $stripurl;
my $title = join(" ",@ARGV);

if (!$name || !$url) {
        print "Usage: $0 name URL [strip from url] [optional title]\n";
        exit;
}

my $dir=$0; $dir=~s/\/[^\/]+$//;

# try to deduce working directory if script is not called with absolute path
if ($dir !~ m#^/#) {
        chomp($dir = `pwd`);
}

my ($host,$urlpath);

if ($url =~ m#(http://[^/]+)(/.*)$#) {
        ($host,$urlpath) = ($1,$2);
        if ($stripurl && $urlpath =~ m/^(.*$stripurl)/) {
                $stripurl = $1;
        } else {
                $stripurl = "";
        }
} else {
        print "Can't parse URL $url into host and path!\n";
        exit 1;
}

my $baseurl = $url;
$baseurl =~ s#/[^/]+$#/#g;

print "Config name: $name\nCrawling url: $url [strip $stripurl]\n";
print "Title: $title\n" if ($title);


my $config = <<"EOF";

###################################################
IncludeConfigFile $dir/common.config

# this is a cludge to implement no parent URL feature in swish indexer
IndexDir "$url $baseurl"
ReplaceRules replace " $baseurl" ""
ReplaceRules replace "${host}${stripurl}" ""

IndexFile $dir/index/$name

StoreDescription HTML <body> 500

# not very usefull, but...
MetaNames keywords description

# store <title>
PropertyNameAlias swishtitle title

EOF

my $xml = << "EOF";
<config
        max_hits="1000"
        prog="/usr/bin/swish-e"
        index="$dir/index/$name"
        charset="iso-8859-2"
        affix="/usr/lib/ispell/american.aff"
EOF
if ($title) {
        $xml .= "\ttitle=\"$title\"\n";
} else {
        $xml .= "\ttitle=\"$name search\"\n";
}
$xml .= "\turl=\"$stripurl\"\n" if ($stripurl);
$xml .= << "EOF";
        >
        <labels>
                <label value="10"> 10</label>
                <label value="100"> 100</label>
                <label value="0">unlimited</label>
        </labels>
        <text>
                <search>Search for </search>
                <documents> documents containing words: </documents>
                <submit>Search</submit>
                <no_spell>don't use alternative spellings </no_spell>
                <no_properties>don't display results details</no_properties>
                <hits>Showing %s documents (of maximum %d)... (%s)</hits>
                <no_hits>Can't find any documents (%s, %s)</no_hits>
                <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>
        </text>
        <!-- additional data to show summary -->
        <properties>swishdescription swishdocsize</properties>
        <hit><![CDATA[ <a href="%s">%s</a> [%s]<br>
        <font color=gray size=-2>%s ... <i>%s bytes</i></font><br>
        ]]></hit>

</config>
EOF

print "Creating $dir/$name.config\n";
open(C,"> $dir/$name.config") || die "can't open $dir/$name.config: $!";
print C $config;
close(C);

print "Creating $dir/html/$name.xml\n";
open(C, "> $dir/html/$name.xml") || die "can't open $dir/html/$name.xml: $!";
print C $xml;
close(C);

print "Creating symlink to $dir/html/$name.cgi\n";
symlink "$dir/html/swish.cgi","$dir/html/$name.cgi" || die "can't create symlink $dir/html/$name.cgi: $!";

#print "Index this with:\nswish-e -S http -c $dir/$name.config\n";
#print "Search using $name.cgi\n";
1	#!/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 = shift @ARGV;
11	my $url = shift @ARGV;
12	#my $stripurl = shift @ARGV;
13	my $stripurl;
14	my $title = join(" ",@ARGV);
15
16	if (!$name \|\| !$url) {
17	print "Usage: $0 name URL [strip from url] [optional title]\n";
18	exit;
19	}
20
21	my $dir=$0; $dir=~s/\/[^\/]+$//;
22
23	# try to deduce working directory if script is not called with absolute path
24	if ($dir !~ m#^/#) {
25	chomp($dir = `pwd`);
26	}
27
28	my ($host,$urlpath);
29
30	if ($url =~ m#(http://[^/]+)(/.*)$#) {
31	($host,$urlpath) = ($1,$2);
32	if ($stripurl && $urlpath =~ m/^(.*$stripurl)/) {
33	$stripurl = $1;
34	} else {
35	$stripurl = "";
36	}
37	} else {
38	print "Can't parse URL $url into host and path!\n";
39	exit 1;
40	}
41
42	my $baseurl = $url;
43	$baseurl =~ s#/[^/]+$#/#g;
44
45	print "Config name: $name\nCrawling url: $url [strip $stripurl]\n";
46	print "Title: $title\n" if ($title);
47
48
49	my $config = <<"EOF";
50
51	###################################################
52	IncludeConfigFile $dir/common.config
53
54	# this is a cludge to implement no parent URL feature in swish indexer
55	IndexDir "$url $baseurl"
56	ReplaceRules replace " $baseurl" ""
57	ReplaceRules replace "${host}${stripurl}" ""
58
59	IndexFile $dir/index/$name
60
61	StoreDescription HTML <body> 500
62
63	# not very usefull, but...
64	MetaNames keywords description
65
66	# store <title>
67	PropertyNameAlias swishtitle title
68
69	EOF
70
71	my $xml = << "EOF";
72	<config
73	max_hits="1000"
74	prog="/usr/bin/swish-e"
75	index="$dir/index/$name"
76	charset="iso-8859-2"
77	affix="/usr/lib/ispell/american.aff"
78	EOF
79	if ($title) {
80	$xml .= "\ttitle=\"$title\"\n";
81	} else {
82	$xml .= "\ttitle=\"$name search\"\n";
83	}
84	$xml .= "\turl=\"$stripurl\"\n" if ($stripurl);
85	$xml .= << "EOF";
86	>
87	<labels>
88	<label value="10"> 10</label>
89	<label value="100"> 100</label>
90	<label value="0">unlimited</label>
91	</labels>
92	<text>
93	<search>Search for </search>
94	<documents> documents containing words: </documents>
95	<submit>Search</submit>
96	<no_spell>don't use alternative spellings </no_spell>
97	<no_properties>don't display results details</no_properties>
98	<hits>Showing %s documents (of maximum %d)... (%s)</hits>
99	<no_hits>Can't find any documents (%s, %s)</no_hits>
100	<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>
101	</text>
102	<!-- additional data to show summary -->
103	<properties>swishdescription swishdocsize</properties>
104	<hit><![CDATA[ <a href="%s">%s</a> [%s]<br>
105	<font color=gray size=-2>%s ... <i>%s bytes</i></font><br>
106	]]></hit>
107
108	</config>
109	EOF
110
111	print "Creating $dir/$name.config\n";
112	open(C,"> $dir/$name.config") \|\| die "can't open $dir/$name.config: $!";
113	print C $config;
114	close(C);
115
116	print "Creating $dir/html/$name.xml\n";
117	open(C, "> $dir/html/$name.xml") \|\| die "can't open $dir/html/$name.xml: $!";
118	print C $xml;
119	close(C);
120
121	print "Creating symlink to $dir/html/$name.cgi\n";
122	symlink "$dir/html/swish.cgi","$dir/html/$name.cgi" \|\| die "can't create symlink $dir/html/$name.cgi: $!";
123
124	#print "Index this with:\nswish-e -S http -c $dir/$name.config\n";
125	#print "Search using $name.cgi\n";