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_cache; |
16 |
use back2html; |
use back2html; |
17 |
|
|
18 |
# configuration options |
|
19 |
# FIX: 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 |
|
my $TEMPLATE =$cfg_global->val('webpac', 'template'); |
31 |
|
|
32 |
|
|
33 |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
34 |
|
|
35 |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
36 |
|
|
37 |
|
|
38 |
sub setup { |
sub setup { |
39 |
my $self = shift; |
my $self = shift; |
40 |
$self->tmpl_path($TEMPLATE_PATH); |
$self->tmpl_path($TEMPLATE_PATH); |
50 |
$self->header_props(-charset=>$CHARSET); |
$self->header_props(-charset=>$CHARSET); |
51 |
} |
} |
52 |
|
|
53 |
|
sub in_template { |
54 |
|
my $html = shift || "This page is left unintentionally blank"; |
55 |
|
return $html if (! defined($TEMPLATE)); |
56 |
|
if (open(T, $TEMPLATE)) { |
57 |
|
my $template_html = join("\n",<T>); |
58 |
|
close(T); |
59 |
|
$template_html =~ s/##webpac##/$html/gsi; |
60 |
|
return $template_html; |
61 |
|
} else { |
62 |
|
return "Can't read template '$TEMPLATE'"; |
63 |
|
} |
64 |
|
} |
65 |
|
|
66 |
sub show_search_form { |
sub show_search_form { |
67 |
my $self = shift; |
my $self = shift; |
68 |
|
|
74 |
|
|
75 |
my $fif = new HTML::FillInForm; |
my $fif = new HTML::FillInForm; |
76 |
|
|
77 |
return $fif->fill(scalarref => \$html, fobject => $q, |
return in_template($fif->fill(scalarref => \$html, fobject => $q, |
78 |
target => 'search'); |
target => 'search')); |
79 |
} |
} |
80 |
|
|
81 |
sub show_results_list { |
sub show_results_list { |
89 |
|
|
90 |
my @s_arr; # all queries are located here |
my @s_arr; # all queries are located here |
91 |
|
|
92 |
for(my $i = 1; $i <=10; $i++) { |
my @path_arr = $q->param('path'); |
93 |
|
my $full = $q->param('full'); |
94 |
|
|
95 |
|
my @persist_vars = ( 'rm' ); |
96 |
|
my @url_params = ( 'rm=results', 'show_full=1', 'last_PAGER_offset='.$q->param('PAGER_offset') || 0 ); |
97 |
|
|
98 |
|
# support parametars "f" and "v" for start |
99 |
|
for(my $i = ""; $i <=30; $i++) { |
100 |
|
|
101 |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
102 |
next if (! $q->param("f$i")); |
|
103 |
next if (! $q->param("v$i")); |
next if (! $q->param("v$i")); |
104 |
|
next if (! $q->param("f$i")); |
105 |
|
|
106 |
# re-write query from +/- to and/and not |
push @persist_vars, "f$i"; |
107 |
my $s; |
push @persist_vars, "v$i"; |
108 |
my $search = $q->param("v$i"); |
|
109 |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
push @url_params,"f$i=".$q->url_param("f$i"); |
110 |
$s .= "$1 "; |
foreach my $v ($q->url_param("v$i")) { |
111 |
|
push @url_params,"v$i=$v"; |
112 |
} |
} |
|
$search =~ s/^\s+//; |
|
|
$search =~ s/\s+$//; |
|
113 |
|
|
114 |
foreach (split(/\s+/,$search)) { |
my $wc="*"; # swish wildcard |
115 |
if (m/^([+-])(\S+)/) { |
$wc="" if ($i eq ""); # don't apply wildcard on field 0 |
116 |
$s.= ($s) ? "and " : ""; |
|
117 |
$s.="not " if ($1 eq "-"); |
# re-write query from +/- to and/and not |
118 |
$s.="$2* "; |
my @param_vals = $q->param("v$i"); |
119 |
} else { |
my @swish_q; |
120 |
$s.="$_* "; |
my ($pre,$post,$exact) = ('','',''); |
121 |
|
while (my $search = shift @param_vals) { |
122 |
|
my $s; |
123 |
|
# remove accents |
124 |
|
$search = unac_string($CHARSET,$search); |
125 |
|
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
126 |
|
$s .= "$1 "; |
127 |
|
} |
128 |
|
$search =~ s/^\s+//; |
129 |
|
$search =~ s/\s+$//; |
130 |
|
|
131 |
|
# filed e[nr] is exact match bitmask |
132 |
|
# 1 = beginning, 2=end, 3=both |
133 |
|
$pre = '"xxbxx ' if ($q->param("e$i") & 1); |
134 |
|
$post = ' xxexx"' if ($q->param("e$i") & 2); |
135 |
|
# add qotes on other side |
136 |
|
if ($q->param("e$i")) { |
137 |
|
$pre = '"' if (! $pre); |
138 |
|
$post = '"' if (! $post); |
139 |
|
$wc = ''; # don't use windcard in exact |
140 |
|
$exact = '_exact'; |
141 |
} |
} |
|
} |
|
|
$s =~ s/\*+/*/g; |
|
142 |
|
|
143 |
push @s_arr,$q->param("f$i")."_swish=($s)"; |
foreach (split(/\s+/,$search)) { |
144 |
|
if (m/^([+-])(\S+)/) { |
145 |
|
$s.= ($s) ? "and " : ""; |
146 |
|
$s.="not " if ($1 eq "-"); |
147 |
|
$s.=$2.$wc." "; |
148 |
|
} elsif (m/^\s*(and|or|not)\s*$/i) { |
149 |
|
$s.=$_." "; |
150 |
|
# don't add * to words with less than x chars |
151 |
|
} elsif (length($_) <= $MIN_WILDCARD) { |
152 |
|
$s.=$_." "; |
153 |
|
} else { |
154 |
|
$s.=$_.$wc." "; |
155 |
|
} |
156 |
|
} |
157 |
|
$s =~ s/\*+/*/g; |
158 |
|
$s = $pre.$s.$post if ($q->param("e$i")); |
159 |
|
push @swish_q,$s; |
160 |
|
} |
161 |
|
# FIXME default operator for multi-value fields is or. There is |
162 |
|
# no way to change it, except here for now. Is there need? |
163 |
|
push @s_arr, $q->param("f$i")."_swish".$exact."=(".join(" or ",@swish_q).")"; |
164 |
} |
} |
165 |
|
|
166 |
my $tmpl = $self->load_tmpl('results.html'); |
my $tmpl = $self->load_tmpl('results.html', global_vars => 1); |
167 |
|
|
168 |
|
sub esc_html { |
169 |
|
my $html = shift; |
170 |
|
$html =~ s/</</g; |
171 |
|
$html =~ s/>/>/g; |
172 |
|
return $html; |
173 |
|
} |
174 |
|
|
175 |
# call swish |
# call swish |
176 |
my $sh = SWISH->connect('Fork', |
my $sh = SWISH->connect('Fork', |
183 |
push @swish_results, { |
push @swish_results, { |
184 |
nr => ($#swish_results + 2), |
nr => ($#swish_results + 2), |
185 |
path => $hit->swishdocpath, |
path => $hit->swishdocpath, |
186 |
# headline => $from_utf8->convert($hit->headline), |
headline => esc_html($from_utf8->convert($hit->headline)), |
187 |
# html => back2html($from_utf8->convert($hit->html)), |
html => back2html($from_utf8->convert($hit->html)), |
|
headline => $hit->headline, |
|
|
html => back2html($hit->html), |
|
188 |
rank => $hit->swishrank }; |
rank => $hit->swishrank }; |
189 |
|
|
190 |
}, |
}, |
191 |
#startnum => 0, |
#startnum => 0, |
192 |
maxhits => $MAX_HITS, |
maxhits => $MAX_HITS |
193 |
); |
); |
194 |
|
|
195 |
die $SWISH::errstr unless $sh; |
die $SWISH::errstr unless $sh; |
196 |
|
# construct swish query |
197 |
|
my $sw_q = join(" and ",@s_arr); |
198 |
|
if (@path_arr && $q->param('show_full')) { |
199 |
|
$sw_q .= "and (swishdocpath=\""; |
200 |
|
$sw_q .= join("\" or swishdocpath=\"",@path_arr); |
201 |
|
$sw_q .= "\")"; |
202 |
|
$tmpl->param('full',1); # show full records |
203 |
|
} elsif ($q->param('show_full')) { |
204 |
|
# just show full path, no path defined |
205 |
|
$tmpl->param('full',1); |
206 |
|
} else { |
207 |
|
$tmpl->param('full',0); |
208 |
|
} |
209 |
|
|
210 |
my $hits = $sh->query(join(" and ",@s_arr)) || 0; # FIX: and/or |
my $hits = $sh->query($sw_q); |
211 |
|
|
212 |
$tmpl->param('hits',$hits); |
$tmpl->param('hits',$hits); |
213 |
$tmpl->param('search',join(" and ",@s_arr)); |
$tmpl->param('search',$sw_q); |
214 |
|
|
215 |
|
$tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0); |
216 |
|
$tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0); |
217 |
|
|
218 |
|
$tmpl->param('url_params',"?".join("&",@url_params)); |
219 |
|
|
220 |
# create a Pager object |
# create a Pager object |
221 |
my $pager = HTML::Pager->new( |
my $pager = HTML::Pager->new( |
226 |
|
|
227 |
my @result; |
my @result; |
228 |
for (my $i=0; $i<$rows; $i++) { |
for (my $i=0; $i<$rows; $i++) { |
229 |
push @result, $swish_results[$offset+$i] if $swish_results[$offset+$i]; |
my $r = $swish_results[$offset+$i]; |
230 |
|
if ($r && $tmpl->param('full')) { |
231 |
|
push @result, $r; |
232 |
|
} elsif ($r) { |
233 |
|
# if not full output, skip html |
234 |
|
delete $r->{html}; |
235 |
|
push @result, $r; |
236 |
|
} |
237 |
} |
} |
238 |
return \@result; |
return \@result; |
239 |
}, |
}, |
240 |
rows => $hits, |
rows => $hits, |
241 |
page_size => $ON_PAGE, |
page_size => $ON_PAGE, |
242 |
# some optional parameters |
# some optional parameters |
243 |
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', |
|
|
], |
|
244 |
#cell_space_color => '#000000', |
#cell_space_color => '#000000', |
245 |
#cell_background_color => '#ffffff', |
#cell_background_color => '#ffffff', |
246 |
#nav_background_color => '#dddddd', |
#nav_background_color => '#dddddd', |
251 |
|
|
252 |
my $html = $pager->output; |
my $html = $pager->output; |
253 |
|
|
254 |
return $html; |
return in_template($html); |
255 |
} |
} |
256 |
|
|
257 |
sub show_index { |
sub show_index { |
265 |
|
|
266 |
my $html; |
my $html; |
267 |
|
|
268 |
my $index = new index_DBI(); |
my $index = new index_DBI( |
269 |
|
$cfg_global->val('global', 'dbi_dbd'), |
270 |
|
$cfg_global->val('global', 'dbi_dsn'), |
271 |
|
$cfg_global->val('global', 'dbi_user'), |
272 |
|
$cfg_global->val('global', 'dbi_passwd') || '' |
273 |
|
); |
274 |
|
|
275 |
my $total = $index->check($field); |
my $total = $index->count($field,$limit); |
276 |
if (! $total) { |
if (! $total) { |
277 |
my $tmpl = $self->load_tmpl('no_index.html'); |
my $tmpl = $self->load_tmpl('no_index.html'); |
278 |
$tmpl->param('field',$field); |
$tmpl->param('field',$field); |
280 |
return $html; |
return $html; |
281 |
} |
} |
282 |
|
|
283 |
my $tmpl = $self->load_tmpl('index_res.html'); |
my $tmpl = $self->load_tmpl('index_res.html', global_vars => 1); |
284 |
$tmpl->param('field',$field); |
$tmpl->param('field',$field); |
285 |
$tmpl->param('limit',$limit); |
$tmpl->param('limit',$limit); |
286 |
$tmpl->param('total',$total); |
$tmpl->param('total',$total); |
287 |
|
|
288 |
# FIX: I should set offset and leave out limit from fetch!! |
# FIXME I should set offset and leave out limit from fetch!! |
289 |
# if (! $q->param("PAGER_offset") { |
# if (! $q->param("PAGER_offset") { |
290 |
# $q->param("Pager_offet) |
# $q->param("Pager_offet) |
291 |
# } |
# } |
295 |
get_data_callback => sub { |
get_data_callback => sub { |
296 |
my ($offset, $rows) = @_; |
my ($offset, $rows) = @_; |
297 |
|
|
298 |
my @result = $index->fetch($field,'item',$limit, $offset, $rows); |
my @result = $index->fetch($field,$limit, $offset, $rows); |
299 |
return \@result; |
return \@result; |
300 |
}, |
}, |
301 |
rows => $total, |
rows => $total, |
309 |
template => $tmpl, |
template => $tmpl, |
310 |
); |
); |
311 |
|
|
312 |
return $pager->output; |
return in_template($pager->output); |
313 |
} |
} |
314 |
|
|
315 |
1; |
1; |