/[webpac]/trunk/WebPac.pm
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /trunk/WebPac.pm

Parent Directory Parent Directory | Revision Log Revision Log


Revision 198 - (hide annotations)
Sun Dec 21 06:35:43 2003 UTC (20 years, 3 months ago) by dpavlin
File size: 9083 byte(s)
Implement different interfaces to webpac (for books, articles etc, or in
different languages) easily: just add base_path with relative or
absolute URL to root of your webpac and you can then create multiple
aliases (using Alias directive in httpd.conf) which have same beginning
as base_path. Everything added after base_path will be added after filename
for templates, but before extension. Easy, eh?

1 dpavlin 7 package WebPac;
2    
3     use base 'CGI::Application';
4     use strict;
5    
6     use HTML::Pager;
7     use HTML::FillInForm;
8     use SWISH;
9 dpavlin 14 use Text::Iconv;
10 dpavlin 9 use DBI;
11 dpavlin 53 use Config::IniFiles;
12 dpavlin 73 use Text::Unaccent;
13 dpavlin 7
14 dpavlin 11 use lib '..';
15 dpavlin 128 use index_DBI_cache;
16 dpavlin 13 use back2html;
17 dpavlin 11
18 dpavlin 7
19 dpavlin 76 # read global.conf configuration
20     my $cfg_global = new Config::IniFiles( -file => '../global.conf' ) || die "can't open 'global.conf'";
21    
22     # configuration options from global.conf
23     my $TEMPLATE_PATH = $cfg_global->val('webpac', 'template_html') || die "need template_html in global.conf, section webpac";
24     my $CHARSET = $cfg_global->val('webpac', 'charset') || 'ISO-8859-1';
25     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 dpavlin 120 my $MIN_WILDCARD =$cfg_global->val('webpac', 'min_wildcard') || 1;
30 dpavlin 147 my $TEMPLATE =$cfg_global->val('webpac', 'template');
31 dpavlin 164 my $UNAC_FILTER =$cfg_global->val('global', 'unac_filter');
32 dpavlin 198 my $BASE_PATH =$cfg_global->val('webpac', 'base_path');
33 dpavlin 76
34 dpavlin 164 if ($UNAC_FILTER) {
35     require $UNAC_FILTER;
36     }
37 dpavlin 76
38 dpavlin 30 Text::Iconv->raise_error(0); # Conversion errors raise exceptions
39 dpavlin 7
40 dpavlin 14 my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET);
41    
42 dpavlin 198 # 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, $q->url(-path => 1),$TEMPLATE_PATH,$tpl);
47     }
48 dpavlin 53
49 dpavlin 198 sub suff2file($$$$) {
50     my ($base_path, $p, $path, $tpl) = @_;
51    
52     return $tpl if (! $base_path);
53    
54     # strip everything to and including base path, leaving only
55     # additional (virtual) path
56     if ($p =~ s,^.*?$base_path,,) {
57     $p =~ s,/*,,g;
58     my ($name,$ext) = split(/\./,$tpl);
59     $p = $name . $p . "." . $ext;
60     } else {
61     # if unable reset it!
62     $p = $tpl;
63     }
64    
65     if ( -e "$path/$p") {
66     return $p;
67     } else {
68     return $tpl;
69     }
70    
71     }
72    
73 dpavlin 7 sub setup {
74     my $self = shift;
75     $self->tmpl_path($TEMPLATE_PATH);
76     $self->run_modes(
77     'search' => 'show_search_form',
78     'results' => 'show_results_list',
79 dpavlin 9 # 'user' => 'show_user_detail',
80     'index' => 'show_index',
81 dpavlin 7 );
82     $self->start_mode('search');
83     $self->mode_param('rm');
84    
85     $self->header_props(-charset=>$CHARSET);
86     }
87    
88 dpavlin 147 sub in_template {
89 dpavlin 198 my $q = shift || die "need CGI object!";
90     my $html = shift || die "This page is left unintentionally blank";
91 dpavlin 147 return $html if (! defined($TEMPLATE));
92 dpavlin 198
93     my ($dir,$tpl);
94     if ($TEMPLATE =~ m,^(.*?/*)([^/]+)$,) {
95     ($dir,$tpl) = ($1,$2);
96     } else {
97     die "can't parse TEMPLATE path";
98     }
99    
100     my $master_tpl = suff2file($BASE_PATH, $q->url(-path => 1),$dir,$tpl);
101     if (open(T, $master_tpl)) {
102 dpavlin 147 my $template_html = join("\n",<T>);
103     close(T);
104     $template_html =~ s/##webpac##/$html/gsi;
105     return $template_html;
106     } else {
107 dpavlin 198 return "Can't read template '$master_tpl'";
108 dpavlin 147 }
109     }
110    
111 dpavlin 7 sub show_search_form {
112     my $self = shift;
113    
114     # Get the CGI.pm query object
115     my $q = $self->query();
116    
117 dpavlin 198 my $tmpl = $self->load_tmpl(url_ex($q,'search.html'));
118 dpavlin 7 my $html = $tmpl->output;
119    
120     my $fif = new HTML::FillInForm;
121    
122 dpavlin 198 return in_template($q,$fif->fill(scalarref => \$html, fobject => $q,
123 dpavlin 147 target => 'search'));
124 dpavlin 7 }
125    
126     sub show_results_list {
127     my $self = shift;
128    
129     my $q = $self->query();
130    
131     my @swish_results; # results from swish
132    
133     # load template for this page
134    
135     my @s_arr; # all queries are located here
136    
137 dpavlin 47 my @path_arr = $q->param('path');
138     my $full = $q->param('full');
139    
140 dpavlin 112 my @persist_vars = ( 'rm' );
141 dpavlin 126 my @url_params = ( 'rm=results', 'show_full=1', 'last_PAGER_offset='.$q->param('PAGER_offset') || 0 );
142 dpavlin 112
143 dpavlin 150 # support parametars "f" and "v" for start
144     for(my $i = ""; $i <=30; $i++) {
145 dpavlin 7
146 dpavlin 9 return show_index($self, $i) if ($q->param("f".$i."_index"));
147 dpavlin 112
148 dpavlin 71 next if (! $q->param("v$i"));
149 dpavlin 9 next if (! $q->param("f$i"));
150 dpavlin 7
151 dpavlin 112 push @persist_vars, "f$i";
152     push @persist_vars, "v$i";
153 dpavlin 186 push @persist_vars, "e$i" if ($q->param("e$i"));
154 dpavlin 112
155 dpavlin 126 push @url_params,"f$i=".$q->url_param("f$i");
156 dpavlin 158 foreach my $v ($q->url_param("v$i")) {
157     push @url_params,"v$i=$v";
158     }
159 dpavlin 183 push @url_params,"e$i=".$q->url_param("e$i");
160 dpavlin 126
161 dpavlin 150 my $wc="*"; # swish wildcard
162     $wc="" if ($i eq ""); # don't apply wildcard on field 0
163    
164 dpavlin 7 # re-write query from +/- to and/and not
165 dpavlin 73 my @param_vals = $q->param("v$i");
166     my @swish_q;
167 dpavlin 163 my ($pre,$post,$exact) = ('','','');
168 dpavlin 73 while (my $search = shift @param_vals) {
169     my $s;
170     # remove accents
171     $search = unac_string($CHARSET,$search);
172     while ($search =~ s/\s*("[^"]+")\s*/ /) {
173     $s .= "$1 ";
174     }
175     $search =~ s/^\s+//;
176     $search =~ s/\s+$//;
177 dpavlin 7
178 dpavlin 163 # filed e[nr] is exact match bitmask
179     # 1 = beginning, 2=end, 3=both
180     $pre = '"xxbxx ' if ($q->param("e$i") & 1);
181     $post = ' xxexx"' if ($q->param("e$i") & 2);
182     # add qotes on other side
183     if ($q->param("e$i")) {
184     $pre = '"' if (! $pre);
185     $post = '"' if (! $post);
186 dpavlin 190 # what about wildcards?
187     $wc = '';
188     $wc = '*' if ($q->param("e$i") & 4);
189 dpavlin 163 $exact = '_exact';
190     }
191    
192 dpavlin 73 foreach (split(/\s+/,$search)) {
193     if (m/^([+-])(\S+)/) {
194     $s.= ($s) ? "and " : "";
195     $s.="not " if ($1 eq "-");
196 dpavlin 163 $s.=$2.$wc." ";
197 dpavlin 122 } elsif (m/^\s*(and|or|not)\s*$/i) {
198 dpavlin 163 $s.=$_." ";
199 dpavlin 120 # don't add * to words with less than x chars
200     } elsif (length($_) <= $MIN_WILDCARD) {
201 dpavlin 163 $s.=$_." ";
202 dpavlin 73 } else {
203 dpavlin 163 $s.=$_.$wc." ";
204 dpavlin 73 }
205 dpavlin 7 }
206 dpavlin 73 $s =~ s/\*+/*/g;
207 dpavlin 163 $s = $pre.$s.$post if ($q->param("e$i"));
208 dpavlin 73 push @swish_q,$s;
209 dpavlin 7 }
210 dpavlin 73 # FIXME default operator for multi-value fields is or. There is
211     # no way to change it, except here for now. Is there need?
212 dpavlin 163 push @s_arr, $q->param("f$i")."_swish".$exact."=(".join(" or ",@swish_q).")";
213 dpavlin 7 }
214    
215 dpavlin 198 my $tmpl = $self->load_tmpl(url_ex($q,'results.html'), global_vars => 1);
216 dpavlin 9
217 dpavlin 80 sub esc_html {
218     my $html = shift;
219     $html =~ s/</&lt;/g;
220     $html =~ s/>/&gt;/g;
221     return $html;
222     }
223    
224 dpavlin 7 # call swish
225     my $sh = SWISH->connect('Fork',
226     prog => $SWISH,
227     indexes => $INDEX,
228 dpavlin 13 properties => [qw/swishdocpath swishrank swishtitle headline html/],
229 dpavlin 7 results => sub {
230     my ($sh,$hit) = @_;
231    
232     push @swish_results, {
233     nr => ($#swish_results + 2),
234     path => $hit->swishdocpath,
235 dpavlin 80 headline => esc_html($from_utf8->convert($hit->headline)),
236 dpavlin 41 html => back2html($from_utf8->convert($hit->html)),
237 dpavlin 7 rank => $hit->swishrank };
238    
239     },
240     #startnum => 0,
241 dpavlin 47 maxhits => $MAX_HITS
242 dpavlin 7 );
243    
244     die $SWISH::errstr unless $sh;
245 dpavlin 47 # construct swish query
246     my $sw_q = join(" and ",@s_arr);
247 dpavlin 111 if (@path_arr && $q->param('show_full')) {
248 dpavlin 47 $sw_q .= "and (swishdocpath=\"";
249     $sw_q .= join("\" or swishdocpath=\"",@path_arr);
250     $sw_q .= "\")";
251     $tmpl->param('full',1); # show full records
252 dpavlin 149 } elsif ($q->param('show_full')) {
253     # just show full path, no path defined
254     $tmpl->param('full',1);
255 dpavlin 121 } else {
256     $tmpl->param('full',0);
257 dpavlin 47 }
258 dpavlin 7
259 dpavlin 47 my $hits = $sh->query($sw_q);
260    
261 dpavlin 7 $tmpl->param('hits',$hits);
262 dpavlin 47 $tmpl->param('search',$sw_q);
263 dpavlin 7
264 dpavlin 76 $tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0);
265     $tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0);
266 dpavlin 51
267 dpavlin 126 $tmpl->param('url_params',"?".join("&",@url_params));
268    
269 dpavlin 7 # create a Pager object
270     my $pager = HTML::Pager->new(
271     # required parameters
272     query => $q,
273     get_data_callback => sub {
274     my ($offset, $rows) = @_;
275    
276     my @result;
277     for (my $i=0; $i<$rows; $i++) {
278 dpavlin 114 my $r = $swish_results[$offset+$i];
279 dpavlin 121 if ($r && $tmpl->param('full')) {
280 dpavlin 114 push @result, $r;
281     } elsif ($r) {
282     # if not full output, skip html
283     delete $r->{html};
284     push @result, $r;
285     }
286 dpavlin 7 }
287     return \@result;
288     },
289     rows => $hits,
290     page_size => $ON_PAGE,
291     # some optional parameters
292 dpavlin 112 persist_vars => [ @persist_vars ],
293 dpavlin 7 #cell_space_color => '#000000',
294     #cell_background_color => '#ffffff',
295     #nav_background_color => '#dddddd',
296     #javascript_presubmit => 'last_minute_javascript()',
297     debug => 1,
298     template => $tmpl,
299     );
300    
301     my $html = $pager->output;
302    
303 dpavlin 198 return in_template($q,$html);
304 dpavlin 7 }
305    
306 dpavlin 9 sub show_index {
307     my $self = shift;
308     my $i = shift; # field number
309    
310     my $q = $self->query();
311    
312 dpavlin 11 my $field = $q->param("f$i");
313     my $limit = $q->param("v$i");
314    
315 dpavlin 9 my $html;
316    
317 dpavlin 53 my $index = new index_DBI(
318     $cfg_global->val('global', 'dbi_dbd'),
319     $cfg_global->val('global', 'dbi_dsn'),
320     $cfg_global->val('global', 'dbi_user'),
321     $cfg_global->val('global', 'dbi_passwd') || ''
322     );
323 dpavlin 9
324 dpavlin 140 my $total = $index->count($field,$limit);
325 dpavlin 12 if (! $total) {
326 dpavlin 198 my $tmpl = $self->load_tmpl(url_ex($q,'no_index.html'));
327 dpavlin 11 $tmpl->param('field',$field);
328     $html = $tmpl->output;
329     return $html;
330     }
331 dpavlin 9
332 dpavlin 198 my $tmpl = $self->load_tmpl(url_ex($q,'index_res.html'), global_vars => 1);
333 dpavlin 12 $tmpl->param('field',$field);
334     $tmpl->param('limit',$limit);
335     $tmpl->param('total',$total);
336 dpavlin 11
337 dpavlin 51 # FIXME I should set offset and leave out limit from fetch!!
338 dpavlin 16 # if (! $q->param("PAGER_offset") {
339     # $q->param("Pager_offet)
340     # }
341    
342 dpavlin 12 my $pager = HTML::Pager->new(
343     query => $q,
344     get_data_callback => sub {
345     my ($offset, $rows) = @_;
346 dpavlin 11
347 dpavlin 140 my @result = $index->fetch($field,$limit, $offset, $rows);
348 dpavlin 12 return \@result;
349     },
350     rows => $total,
351     page_size => $ON_PAGE,
352     persist_vars => [
353     'rm',
354     "f$i", "v$i", "f".$i."_index",
355     'offset',
356     ],
357     debug => 1,
358     template => $tmpl,
359     );
360 dpavlin 11
361 dpavlin 198 return in_template($q,$pager->output);
362 dpavlin 9 }
363    
364 dpavlin 7 1;

Properties

Name Value
cvs2svn:cvs-rev 1.39

  ViewVC Help
Powered by ViewVC 1.1.26