8 |
use SWISH; |
use SWISH; |
9 |
use Text::Iconv; |
use Text::Iconv; |
10 |
use DBI; |
use DBI; |
11 |
|
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 |
# configuration options |
19 |
# FIX: they really should go in configuration file! |
# FIXME they really should go in configuration file! |
20 |
my $TEMPLATE_PATH = '/data/webpac/template_html'; |
my $TEMPLATE_PATH = '/data/webpac/template_html'; |
21 |
my $CHARSET = 'ISO-8859-2'; |
my $CHARSET = 'ISO-8859-2'; |
22 |
my $SWISH = '/usr/bin/swish-e'; |
my $SWISH = '/usr/bin/swish-e'; |
23 |
my $INDEX = '/data/webpac/index/isis.index'; |
my $INDEX = '/data/webpac/index/isis.index'; |
24 |
my $MAX_HITS = 500; |
my $MAX_HITS = 0; |
25 |
my $ON_PAGE = 10; |
my $ON_PAGE = 10; |
26 |
|
|
27 |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
28 |
|
|
29 |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
30 |
|
|
31 |
|
# read global.conf configuration |
32 |
|
my $cfg_global = new Config::IniFiles( -file => '../global.conf' ) || die "can't open 'global.conf'"; |
33 |
|
|
34 |
|
|
35 |
sub setup { |
sub setup { |
36 |
my $self = shift; |
my $self = shift; |
37 |
$self->tmpl_path($TEMPLATE_PATH); |
$self->tmpl_path($TEMPLATE_PATH); |
73 |
|
|
74 |
my @s_arr; # all queries are located here |
my @s_arr; # all queries are located here |
75 |
|
|
76 |
for(my $i = 1; $i <=10; $i++) { |
my @path_arr = $q->param('path'); |
77 |
|
my $full = $q->param('full'); |
78 |
|
|
79 |
|
for(my $i = 1; $i <=30; $i++) { |
80 |
|
|
81 |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
|
next if (! $q->param("f$i")); |
|
82 |
next if (! $q->param("v$i")); |
next if (! $q->param("v$i")); |
83 |
|
next if (! $q->param("f$i")); |
84 |
|
|
85 |
# re-write query from +/- to and/and not |
# re-write query from +/- to and/and not |
86 |
my $s; |
my @param_vals = $q->param("v$i"); |
87 |
my $search = $q->param("v$i"); |
my @swish_q; |
88 |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
while (my $search = shift @param_vals) { |
89 |
$s .= "$1 "; |
my $s; |
90 |
} |
# remove accents |
91 |
$search =~ s/^\s+//; |
$search = unac_string($CHARSET,$search); |
92 |
$search =~ s/\s+$//; |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
93 |
|
$s .= "$1 "; |
94 |
|
} |
95 |
|
$search =~ s/^\s+//; |
96 |
|
$search =~ s/\s+$//; |
97 |
|
|
98 |
foreach (split(/\s+/,$search)) { |
foreach (split(/\s+/,$search)) { |
99 |
if (m/^([+-])(\S+)/) { |
if (m/^([+-])(\S+)/) { |
100 |
$s.= ($s) ? "and " : ""; |
$s.= ($s) ? "and " : ""; |
101 |
$s.="not " if ($1 eq "-"); |
$s.="not " if ($1 eq "-"); |
102 |
$s.="$2* "; |
$s.="$2* "; |
103 |
} else { |
} elsif (m/(and|or|not)/i) { |
104 |
$s.="$_* "; |
$s.="$_ "; |
105 |
|
} else { |
106 |
|
$s.="$_* "; |
107 |
|
} |
108 |
} |
} |
109 |
|
$s =~ s/\*+/*/g; |
110 |
|
push @swish_q,$s; |
111 |
} |
} |
112 |
$s =~ s/\*+/*/g; |
# FIXME default operator for multi-value fields is or. There is |
113 |
|
# no way to change it, except here for now. Is there need? |
114 |
push @s_arr,$q->param("f$i")."_swish=($s)"; |
push @s_arr, $q->param("f$i")."_swish=(".join(" or ",@swish_q).")"; |
115 |
} |
} |
116 |
|
|
117 |
my $tmpl = $self->load_tmpl('results.html'); |
my $tmpl = $self->load_tmpl('results.html'); |
127 |
push @swish_results, { |
push @swish_results, { |
128 |
nr => ($#swish_results + 2), |
nr => ($#swish_results + 2), |
129 |
path => $hit->swishdocpath, |
path => $hit->swishdocpath, |
130 |
# headline => $from_utf8->convert($hit->headline), |
headline => $from_utf8->convert($hit->headline), |
131 |
# html => back2html($from_utf8->convert($hit->html)), |
html => back2html($from_utf8->convert($hit->html)), |
|
headline => $hit->headline, |
|
|
html => back2html($hit->html), |
|
132 |
rank => $hit->swishrank }; |
rank => $hit->swishrank }; |
133 |
|
|
134 |
}, |
}, |
135 |
#startnum => 0, |
#startnum => 0, |
136 |
maxhits => $MAX_HITS, |
maxhits => $MAX_HITS |
137 |
); |
); |
138 |
|
|
139 |
die $SWISH::errstr unless $sh; |
die $SWISH::errstr unless $sh; |
140 |
|
# construct swish query |
141 |
|
my $sw_q = join(" and ",@s_arr); |
142 |
|
if (@path_arr) { |
143 |
|
$sw_q .= "and (swishdocpath=\""; |
144 |
|
$sw_q .= join("\" or swishdocpath=\"",@path_arr); |
145 |
|
$sw_q .= "\")"; |
146 |
|
$tmpl->param('full',1); # show full records |
147 |
|
} |
148 |
|
|
149 |
my $hits = $sh->query(join(" and ",@s_arr)) || 0; # FIX: and/or |
my $hits = $sh->query($sw_q); |
150 |
|
|
151 |
$tmpl->param('hits',$hits); |
$tmpl->param('hits',$hits); |
152 |
$tmpl->param('search',join(" and ",@s_arr)); |
$tmpl->param('search',$sw_q); |
153 |
|
|
154 |
|
$tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 1); |
155 |
|
$tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 1); |
156 |
|
|
157 |
# create a Pager object |
# create a Pager object |
158 |
my $pager = HTML::Pager->new( |
my $pager = HTML::Pager->new( |
206 |
|
|
207 |
my $html; |
my $html; |
208 |
|
|
209 |
my $index = new index_DBI(); |
my $index = new index_DBI( |
210 |
|
$cfg_global->val('global', 'dbi_dbd'), |
211 |
|
$cfg_global->val('global', 'dbi_dsn'), |
212 |
|
$cfg_global->val('global', 'dbi_user'), |
213 |
|
$cfg_global->val('global', 'dbi_passwd') || '' |
214 |
|
); |
215 |
|
|
216 |
my $total = $index->check($field); |
my $total = $index->check($field); |
217 |
if (! $total) { |
if (! $total) { |
226 |
$tmpl->param('limit',$limit); |
$tmpl->param('limit',$limit); |
227 |
$tmpl->param('total',$total); |
$tmpl->param('total',$total); |
228 |
|
|
229 |
# FIX: I should set offset and leave out limit from fetch!! |
# FIXME I should set offset and leave out limit from fetch!! |
230 |
# if (! $q->param("PAGER_offset") { |
# if (! $q->param("PAGER_offset") { |
231 |
# $q->param("Pager_offet) |
# $q->param("Pager_offet) |
232 |
# } |
# } |