9 |
use Text::Iconv; |
use Text::Iconv; |
10 |
use DBI; |
use DBI; |
11 |
use Config::IniFiles; |
use Config::IniFiles; |
12 |
|
use Text::Unaccent; |
13 |
|
|
14 |
use lib '..'; |
use lib '..'; |
15 |
use index_DBI; |
use index_DBI; |
16 |
use back2html; |
use back2html; |
17 |
|
|
18 |
# configuration options |
|
19 |
# FIXME they really should go in configuration file! |
# read global.conf configuration |
20 |
my $TEMPLATE_PATH = '/data/webpac/template_html'; |
my $cfg_global = new Config::IniFiles( -file => '../global.conf' ) || die "can't open 'global.conf'"; |
21 |
my $CHARSET = 'ISO-8859-2'; |
|
22 |
my $SWISH = '/usr/bin/swish-e'; |
# configuration options from global.conf |
23 |
my $INDEX = '/data/webpac/index/isis.index'; |
my $TEMPLATE_PATH = $cfg_global->val('webpac', 'template_html') || die "need template_html in global.conf, section webpac"; |
24 |
my $MAX_HITS = 500; |
my $CHARSET = $cfg_global->val('webpac', 'charset') || 'ISO-8859-1'; |
25 |
my $ON_PAGE = 10; |
my $SWISH = $cfg_global->val('webpac', 'swish') || '/usr/bin/swish-e'; |
26 |
|
my $INDEX = $cfg_global->val('webpac', 'index') || die "need index in global.conf, section webpac"; |
27 |
|
my $MAX_HITS = $cfg_global->val('webpac', 'max_hits') || 0; |
28 |
|
my $ON_PAGE =$cfg_global->val('webpac', 'on_page') || 10; |
29 |
|
my $MIN_WILDCARD =$cfg_global->val('webpac', 'min_wildcard') || 1; |
30 |
|
|
31 |
|
|
32 |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
33 |
|
|
34 |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
35 |
|
|
|
# read global.conf configuration |
|
|
my $cfg_global = new Config::IniFiles( -file => '../global.conf' ) || die "can't open 'global.conf'"; |
|
|
|
|
36 |
|
|
37 |
sub setup { |
sub setup { |
38 |
my $self = shift; |
my $self = shift; |
78 |
my @path_arr = $q->param('path'); |
my @path_arr = $q->param('path'); |
79 |
my $full = $q->param('full'); |
my $full = $q->param('full'); |
80 |
|
|
81 |
for(my $i = 1; $i <=10; $i++) { |
my @persist_vars = ( 'rm' ); |
82 |
|
|
83 |
|
for(my $i = 1; $i <=30; $i++) { |
84 |
|
|
85 |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
86 |
next if (! $q->param("f$i")); |
|
87 |
next if (! $q->param("v$i")); |
next if (! $q->param("v$i")); |
88 |
|
next if (! $q->param("f$i")); |
89 |
|
|
90 |
|
push @persist_vars, "f$i"; |
91 |
|
push @persist_vars, "v$i"; |
92 |
|
|
93 |
# re-write query from +/- to and/and not |
# re-write query from +/- to and/and not |
94 |
my $s; |
my @param_vals = $q->param("v$i"); |
95 |
my $search = $q->param("v$i"); |
my @swish_q; |
96 |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
while (my $search = shift @param_vals) { |
97 |
$s .= "$1 "; |
my $s; |
98 |
} |
# remove accents |
99 |
$search =~ s/^\s+//; |
$search = unac_string($CHARSET,$search); |
100 |
$search =~ s/\s+$//; |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
101 |
|
$s .= "$1 "; |
102 |
|
} |
103 |
|
$search =~ s/^\s+//; |
104 |
|
$search =~ s/\s+$//; |
105 |
|
|
106 |
foreach (split(/\s+/,$search)) { |
foreach (split(/\s+/,$search)) { |
107 |
if (m/^([+-])(\S+)/) { |
if (m/^([+-])(\S+)/) { |
108 |
$s.= ($s) ? "and " : ""; |
$s.= ($s) ? "and " : ""; |
109 |
$s.="not " if ($1 eq "-"); |
$s.="not " if ($1 eq "-"); |
110 |
$s.="$2* "; |
$s.="$2* "; |
111 |
} elsif (m/(and|or|not)/i) { |
} elsif (m/(and|or|not)/i) { |
112 |
$s.="$_ "; |
$s.="$_ "; |
113 |
} else { |
# don't add * to words with less than x chars |
114 |
$s.="$_* "; |
} elsif (length($_) <= $MIN_WILDCARD) { |
115 |
|
$s.="$_ "; |
116 |
|
} else { |
117 |
|
$s.="$_* "; |
118 |
|
} |
119 |
} |
} |
120 |
|
$s =~ s/\*+/*/g; |
121 |
|
push @swish_q,$s; |
122 |
} |
} |
123 |
$s =~ s/\*+/*/g; |
# FIXME default operator for multi-value fields is or. There is |
124 |
|
# no way to change it, except here for now. Is there need? |
125 |
push @s_arr,$q->param("f$i")."_swish=($s)"; |
push @s_arr, $q->param("f$i")."_swish=(".join(" or ",@swish_q).")"; |
126 |
} |
} |
127 |
|
|
128 |
my $tmpl = $self->load_tmpl('results.html'); |
my $tmpl = $self->load_tmpl('results.html'); |
129 |
|
|
130 |
|
sub esc_html { |
131 |
|
my $html = shift; |
132 |
|
$html =~ s/</</g; |
133 |
|
$html =~ s/>/>/g; |
134 |
|
return $html; |
135 |
|
} |
136 |
|
|
137 |
# call swish |
# call swish |
138 |
my $sh = SWISH->connect('Fork', |
my $sh = SWISH->connect('Fork', |
139 |
prog => $SWISH, |
prog => $SWISH, |
145 |
push @swish_results, { |
push @swish_results, { |
146 |
nr => ($#swish_results + 2), |
nr => ($#swish_results + 2), |
147 |
path => $hit->swishdocpath, |
path => $hit->swishdocpath, |
148 |
headline => $from_utf8->convert($hit->headline), |
headline => esc_html($from_utf8->convert($hit->headline)), |
149 |
html => back2html($from_utf8->convert($hit->html)), |
html => back2html($from_utf8->convert($hit->html)), |
150 |
rank => $hit->swishrank }; |
rank => $hit->swishrank }; |
151 |
|
|
155 |
); |
); |
156 |
|
|
157 |
die $SWISH::errstr unless $sh; |
die $SWISH::errstr unless $sh; |
|
|
|
158 |
# construct swish query |
# construct swish query |
159 |
my $sw_q = join(" and ",@s_arr); |
my $sw_q = join(" and ",@s_arr); |
160 |
if (@path_arr) { |
if (@path_arr && $q->param('show_full')) { |
161 |
$sw_q .= "and (swishdocpath=\""; |
$sw_q .= "and (swishdocpath=\""; |
162 |
$sw_q .= join("\" or swishdocpath=\"",@path_arr); |
$sw_q .= join("\" or swishdocpath=\"",@path_arr); |
163 |
$sw_q .= "\")"; |
$sw_q .= "\")"; |
181 |
|
|
182 |
my @result; |
my @result; |
183 |
for (my $i=0; $i<$rows; $i++) { |
for (my $i=0; $i<$rows; $i++) { |
184 |
push @result, $swish_results[$offset+$i] if $swish_results[$offset+$i]; |
my $r = $swish_results[$offset+$i]; |
185 |
|
if ($r && $q->param('show_full')) { |
186 |
|
push @result, $r; |
187 |
|
} elsif ($r) { |
188 |
|
# if not full output, skip html |
189 |
|
delete $r->{html}; |
190 |
|
push @result, $r; |
191 |
|
} |
192 |
} |
} |
193 |
return \@result; |
return \@result; |
194 |
}, |
}, |
195 |
rows => $hits, |
rows => $hits, |
196 |
page_size => $ON_PAGE, |
page_size => $ON_PAGE, |
197 |
# some optional parameters |
# some optional parameters |
198 |
persist_vars => [ |
persist_vars => [ @persist_vars ], |
|
'rm', |
|
|
'f1', 'v1', |
|
|
'f2', 'v2', |
|
|
'f3', 'v3', |
|
|
'f4', 'v4', |
|
|
'f5', 'v5', |
|
|
'f6', 'v6', |
|
|
'f7', 'v7', |
|
|
'f8', 'v8', |
|
|
'f9', 'v9', |
|
|
], |
|
199 |
#cell_space_color => '#000000', |
#cell_space_color => '#000000', |
200 |
#cell_background_color => '#ffffff', |
#cell_background_color => '#ffffff', |
201 |
#nav_background_color => '#dddddd', |
#nav_background_color => '#dddddd', |