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 |
|
my $UNAC_FILTER =$cfg_global->val('global', 'unac_filter'); |
32 |
|
my $BASE_PATH =$cfg_global->val('webpac', 'base_path'); |
33 |
|
|
34 |
|
if ($UNAC_FILTER) { |
35 |
|
require $UNAC_FILTER; |
36 |
|
} |
37 |
|
|
38 |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
39 |
|
|
40 |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
41 |
|
|
42 |
|
# use path from cgi script to support templates in subdirs |
43 |
|
sub url_ex { |
44 |
|
my $q = shift || die "suff2file needs CGI object!"; |
45 |
|
my $tpl = shift || die "url_ex needs template name!"; |
46 |
|
return suff2file($BASE_PATH, $q->url(-absolute => 1,-path => 1),$TEMPLATE_PATH,$tpl); |
47 |
|
} |
48 |
|
|
49 |
|
sub suff2file($$$$) { |
50 |
|
my ($base_path, $p, $path, $tpl) = @_; |
51 |
|
|
52 |
|
return $tpl if (! $base_path); |
53 |
|
|
54 |
|
# strip everything to and including base path, leaving only |
55 |
|
# additional (virtual) path |
56 |
|
if ($base_path eq "/") { |
57 |
|
$p =~ s,/*,,g; |
58 |
|
my ($name,$ext) = split(/\./,$tpl); |
59 |
|
$p = $name . "-" . $p . "." . $ext; |
60 |
|
} elsif ($p =~ s,^.*?$base_path,,) { |
61 |
|
$p =~ s,/*,,g; |
62 |
|
my ($name,$ext) = split(/\./,$tpl); |
63 |
|
$p = $name . $p . "." . $ext; |
64 |
|
} else { |
65 |
|
# if unable reset it! |
66 |
|
$p = $tpl; |
67 |
|
} |
68 |
|
|
69 |
|
if ( -e "$path/$p") { |
70 |
|
return $p; |
71 |
|
} else { |
72 |
|
return $tpl; |
73 |
|
} |
74 |
|
|
75 |
|
} |
76 |
|
|
77 |
sub setup { |
sub setup { |
78 |
my $self = shift; |
my $self = shift; |
79 |
$self->tmpl_path($TEMPLATE_PATH); |
$self->tmpl_path($TEMPLATE_PATH); |
89 |
$self->header_props(-charset=>$CHARSET); |
$self->header_props(-charset=>$CHARSET); |
90 |
} |
} |
91 |
|
|
92 |
|
sub in_template { |
93 |
|
my $q = shift || die "need CGI object!"; |
94 |
|
my $html = shift || die "This page is left unintentionally blank"; |
95 |
|
return $html if (! defined($TEMPLATE)); |
96 |
|
|
97 |
|
my ($dir,$tpl); |
98 |
|
if ($TEMPLATE =~ m,^(.*?/*)([^/]+)$,) { |
99 |
|
($dir,$tpl) = ($1,$2); |
100 |
|
} else { |
101 |
|
die "can't parse TEMPLATE path"; |
102 |
|
} |
103 |
|
|
104 |
|
my $master_tpl = suff2file($BASE_PATH, $q->url(-absolute => 1, -path => 1),$dir,$tpl); |
105 |
|
if (open(T, $master_tpl)) { |
106 |
|
my $template_html = join("\n",<T>); |
107 |
|
close(T); |
108 |
|
$template_html =~ s/##webpac##/$html/gsi; |
109 |
|
return $template_html; |
110 |
|
} else { |
111 |
|
return "Can't read template '$master_tpl'"; |
112 |
|
} |
113 |
|
} |
114 |
|
|
115 |
sub show_search_form { |
sub show_search_form { |
116 |
my $self = shift; |
my $self = shift; |
117 |
|
|
118 |
# Get the CGI.pm query object |
# Get the CGI.pm query object |
119 |
my $q = $self->query(); |
my $q = $self->query(); |
120 |
|
|
121 |
my $tmpl = $self->load_tmpl('search.html'); |
my $tmpl = $self->load_tmpl(url_ex($q,'search.html')); |
122 |
my $html = $tmpl->output; |
my $html = $tmpl->output; |
123 |
|
|
124 |
my $fif = new HTML::FillInForm; |
my $fif = new HTML::FillInForm; |
125 |
|
|
126 |
return $fif->fill(scalarref => \$html, fobject => $q, |
return in_template($q,$fif->fill(scalarref => \$html, fobject => $q, |
127 |
target => 'search'); |
target => 'search')); |
128 |
} |
} |
129 |
|
|
130 |
sub show_results_list { |
sub show_results_list { |
138 |
|
|
139 |
my @s_arr; # all queries are located here |
my @s_arr; # all queries are located here |
140 |
|
|
141 |
for(my $i = 1; $i <=10; $i++) { |
my @path_arr = $q->param('path'); |
142 |
|
my $full = $q->param('full'); |
143 |
|
|
144 |
|
my @persist_vars = ( 'rm' ); |
145 |
|
my @url_params = ( 'rm=results', 'show_full=1', 'last_PAGER_offset='.$q->param('PAGER_offset') || 0 ); |
146 |
|
|
147 |
|
# support parametars "f" and "v" for start |
148 |
|
for(my $i = ""; $i <=30; $i++) { |
149 |
|
|
150 |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
151 |
next if (! $q->param("f$i")); |
|
152 |
next if (! $q->param("v$i")); |
next if (! $q->param("v$i")); |
153 |
|
next if (! $q->param("f$i")); |
154 |
|
|
155 |
# re-write query from +/- to and/and not |
push @persist_vars, "f$i"; |
156 |
my $s; |
push @persist_vars, "v$i"; |
157 |
my $search = $q->param("v$i"); |
push @persist_vars, "e$i" if ($q->param("e$i")); |
158 |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
|
159 |
$s .= "$1 "; |
push @url_params,"f$i=".$q->url_param("f$i"); |
160 |
|
foreach my $v ($q->url_param("v$i")) { |
161 |
|
push @url_params,"v$i=$v"; |
162 |
} |
} |
163 |
$search =~ s/^\s+//; |
push @url_params,"e$i=".$q->url_param("e$i"); |
164 |
$search =~ s/\s+$//; |
|
165 |
|
my $wc="*"; # swish wildcard |
166 |
|
$wc="" if ($i eq ""); # don't apply wildcard on field 0 |
167 |
|
|
168 |
|
# re-write query from +/- to and/and not |
169 |
|
my @param_vals = $q->param("v$i"); |
170 |
|
my @swish_q; |
171 |
|
my ($pre,$post,$exact) = ('','',''); |
172 |
|
while (my $search = shift @param_vals) { |
173 |
|
my $s; |
174 |
|
# remove accents |
175 |
|
$search = unac_string($CHARSET,$search); |
176 |
|
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
177 |
|
$s .= "$1 "; |
178 |
|
} |
179 |
|
$search =~ s/^\s+//; |
180 |
|
$search =~ s/\s+$//; |
181 |
|
|
182 |
|
# filed e[nr] is exact match bitmask |
183 |
|
# 1 = beginning, 2=end, 3=both |
184 |
|
$pre = '"xxbxx ' if ($q->param("e$i") & 1); |
185 |
|
$post = ' xxexx"' if ($q->param("e$i") & 2); |
186 |
|
# add qotes on other side |
187 |
|
if ($q->param("e$i")) { |
188 |
|
$pre = '"' if (! $pre); |
189 |
|
$post = '"' if (! $post); |
190 |
|
# what about wildcards? |
191 |
|
$wc = ''; |
192 |
|
$wc = '*' if ($q->param("e$i") & 4); |
193 |
|
$exact = '_exact'; |
194 |
|
} |
195 |
|
|
196 |
foreach (split(/\s+/,$search)) { |
foreach (split(/\s+/,$search)) { |
197 |
if (m/^([+-])(\S+)/) { |
if (m/^([+-])(\S+)/) { |
198 |
$s.= ($s) ? "and " : ""; |
$s.= ($s) ? "and " : ""; |
199 |
$s.="not " if ($1 eq "-"); |
$s.="not " if ($1 eq "-"); |
200 |
$s.="$2* "; |
$s.=$2.$wc." "; |
201 |
} else { |
} elsif (m/^\s*(and|or|not)\s*$/i) { |
202 |
$s.="$_* "; |
$s.=$_." "; |
203 |
|
# don't add * to words with less than x chars |
204 |
|
} elsif (length($_) <= $MIN_WILDCARD) { |
205 |
|
$s.=$_." "; |
206 |
|
} else { |
207 |
|
$s.=$_.$wc." "; |
208 |
|
} |
209 |
} |
} |
210 |
|
$s =~ s/\*+/*/g; |
211 |
|
$s = $pre.$s.$post if ($q->param("e$i")); |
212 |
|
push @swish_q,$s; |
213 |
} |
} |
214 |
$s =~ s/\*+/*/g; |
# FIXME default operator for multi-value fields is or. There is |
215 |
|
# no way to change it, except here for now. Is there need? |
216 |
|
push @s_arr, $q->param("f$i")."_swish".$exact."=(".join(" or ",@swish_q).")"; |
217 |
|
} |
218 |
|
|
219 |
|
my $tmpl = $self->load_tmpl(url_ex($q,'results.html'), global_vars => 1); |
220 |
|
|
221 |
push @s_arr,$q->param("f$i")."_swish=($s)"; |
sub esc_html { |
222 |
|
my $html = shift; |
223 |
|
$html =~ s/</</g; |
224 |
|
$html =~ s/>/>/g; |
225 |
|
return $html; |
226 |
} |
} |
227 |
|
|
228 |
my $tmpl = $self->load_tmpl('results.html'); |
my $sort = 'swishrank'; |
229 |
|
if ($q->param("sort")) { |
230 |
|
$sort = 'headline'; |
231 |
|
push @persist_vars, "sort"; |
232 |
|
} |
233 |
|
|
234 |
# call swish |
# call swish |
235 |
my $sh = SWISH->connect('Fork', |
my $sh = SWISH->connect('Fork', |
242 |
push @swish_results, { |
push @swish_results, { |
243 |
nr => ($#swish_results + 2), |
nr => ($#swish_results + 2), |
244 |
path => $hit->swishdocpath, |
path => $hit->swishdocpath, |
245 |
# headline => $from_utf8->convert($hit->headline), |
headline => esc_html($from_utf8->convert($hit->headline)), |
246 |
# html => back2html($from_utf8->convert($hit->html)), |
html => back2html($from_utf8->convert($hit->html)), |
|
headline => $hit->headline, |
|
|
html => back2html($hit->html), |
|
247 |
rank => $hit->swishrank }; |
rank => $hit->swishrank }; |
248 |
|
|
249 |
}, |
}, |
250 |
#startnum => 0, |
#startnum => 0, |
251 |
maxhits => $MAX_HITS, |
maxhits => $MAX_HITS, |
252 |
|
sortorder => $sort, |
253 |
); |
); |
254 |
|
|
255 |
die $SWISH::errstr unless $sh; |
die $SWISH::errstr unless $sh; |
256 |
|
# construct swish query |
257 |
|
my $sw_q = join(" and ",@s_arr); |
258 |
|
if (@path_arr && $q->param('show_full')) { |
259 |
|
$sw_q .= "and (swishdocpath=\""; |
260 |
|
$sw_q .= join("\" or swishdocpath=\"",@path_arr); |
261 |
|
$sw_q .= "\")"; |
262 |
|
$tmpl->param('full',1); # show full records |
263 |
|
} elsif ($q->param('show_full')) { |
264 |
|
# just show full path, no path defined |
265 |
|
$tmpl->param('full',1); |
266 |
|
} else { |
267 |
|
$tmpl->param('full',0); |
268 |
|
} |
269 |
|
|
270 |
my $hits = $sh->query(join(" and ",@s_arr)) || 0; # FIX: and/or |
my $hits = $sh->query($sw_q); |
271 |
|
|
272 |
$tmpl->param('hits',$hits); |
$tmpl->param('hits',$hits); |
273 |
$tmpl->param('search',join(" and ",@s_arr)); |
$tmpl->param('search',$sw_q); |
274 |
|
|
275 |
|
$tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0); |
276 |
|
$tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0); |
277 |
|
|
278 |
|
$tmpl->param('url_params',"?".join("&",@url_params)); |
279 |
|
|
280 |
# create a Pager object |
# create a Pager object |
281 |
my $pager = HTML::Pager->new( |
my $pager = HTML::Pager->new( |
286 |
|
|
287 |
my @result; |
my @result; |
288 |
for (my $i=0; $i<$rows; $i++) { |
for (my $i=0; $i<$rows; $i++) { |
289 |
push @result, $swish_results[$offset+$i] if $swish_results[$offset+$i]; |
my $r = $swish_results[$offset+$i]; |
290 |
|
if ($r && $tmpl->param('full')) { |
291 |
|
push @result, $r; |
292 |
|
} elsif ($r) { |
293 |
|
# if not full output, skip html |
294 |
|
delete $r->{html}; |
295 |
|
push @result, $r; |
296 |
|
} |
297 |
} |
} |
298 |
return \@result; |
return \@result; |
299 |
}, |
}, |
300 |
rows => $hits, |
rows => $hits, |
301 |
page_size => $ON_PAGE, |
page_size => $ON_PAGE, |
302 |
# some optional parameters |
# some optional parameters |
303 |
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', |
|
|
], |
|
304 |
#cell_space_color => '#000000', |
#cell_space_color => '#000000', |
305 |
#cell_background_color => '#ffffff', |
#cell_background_color => '#ffffff', |
306 |
#nav_background_color => '#dddddd', |
#nav_background_color => '#dddddd', |
311 |
|
|
312 |
my $html = $pager->output; |
my $html = $pager->output; |
313 |
|
|
314 |
return $html; |
return in_template($q,$html); |
315 |
} |
} |
316 |
|
|
317 |
sub show_index { |
sub show_index { |
325 |
|
|
326 |
my $html; |
my $html; |
327 |
|
|
328 |
my $index = new index_DBI(); |
my $index = new index_DBI( |
329 |
|
$cfg_global->val('global', 'dbi_dbd'), |
330 |
|
$cfg_global->val('global', 'dbi_dsn'), |
331 |
|
$cfg_global->val('global', 'dbi_user'), |
332 |
|
$cfg_global->val('global', 'dbi_passwd') || '' |
333 |
|
); |
334 |
|
|
335 |
my $total = $index->check($field); |
my $total = $index->count($field,$limit); |
336 |
if (! $total) { |
if (! $total) { |
337 |
my $tmpl = $self->load_tmpl('no_index.html'); |
my $tmpl = $self->load_tmpl(url_ex($q,'no_index.html')); |
338 |
$tmpl->param('field',$field); |
$tmpl->param('field',$field); |
339 |
$html = $tmpl->output; |
$html = $tmpl->output; |
340 |
return $html; |
return $html; |
341 |
} |
} |
342 |
|
|
343 |
my $tmpl = $self->load_tmpl('index_res.html'); |
my $tmpl = $self->load_tmpl(url_ex($q,'index_res.html'), global_vars => 1); |
344 |
$tmpl->param('field',$field); |
$tmpl->param('field',$field); |
345 |
$tmpl->param('limit',$limit); |
$tmpl->param('limit',$limit); |
346 |
$tmpl->param('total',$total); |
$tmpl->param('total',$total); |
347 |
|
|
348 |
# FIX: I should set offset and leave out limit from fetch!! |
# FIXME I should set offset and leave out limit from fetch!! |
349 |
# if (! $q->param("PAGER_offset") { |
# if (! $q->param("PAGER_offset") { |
350 |
# $q->param("Pager_offet) |
# $q->param("Pager_offet) |
351 |
# } |
# } |
355 |
get_data_callback => sub { |
get_data_callback => sub { |
356 |
my ($offset, $rows) = @_; |
my ($offset, $rows) = @_; |
357 |
|
|
358 |
my @result = $index->fetch($field,'item',$limit, $offset, $rows); |
my @result = $index->fetch($field,$limit, $offset, $rows); |
359 |
return \@result; |
return \@result; |
360 |
}, |
}, |
361 |
rows => $total, |
rows => $total, |
369 |
template => $tmpl, |
template => $tmpl, |
370 |
); |
); |
371 |
|
|
372 |
return $pager->output; |
return in_template($q,$pager->output); |
373 |
} |
} |
374 |
|
|
375 |
1; |
1; |