1 |
dpavlin |
5 |
#!/usr/bin/perl |
2 |
|
|
|
3 |
|
|
# based on post |
4 |
|
|
# http://www.mail-archive.com/libwww@perl.org/msg04750.html |
5 |
|
|
|
6 |
dpavlin |
43 |
BEGIN { |
7 |
|
|
my $basedir = readlink($0) || $0; $basedir =~ s#/[^/]+$##; |
8 |
|
|
unshift(@INC, $basedir); |
9 |
|
|
} |
10 |
|
|
|
11 |
dpavlin |
5 |
use strict; |
12 |
|
|
use warnings; |
13 |
dpavlin |
41 |
use MWS::SWISH; |
14 |
|
|
#use MWS::Plucene; |
15 |
dpavlin |
5 |
use HTTP::Daemon; |
16 |
|
|
use HTTP::Status; |
17 |
|
|
use IO::String; |
18 |
dpavlin |
6 |
use CGI::Lite; |
19 |
|
|
use Template; |
20 |
dpavlin |
24 |
use URI::Escape; |
21 |
dpavlin |
5 |
|
22 |
dpavlin |
6 |
use Data::Dumper; |
23 |
|
|
|
24 |
dpavlin |
17 |
my $debug = 1; |
25 |
|
|
|
26 |
dpavlin |
27 |
my $config_file = shift @ARGV || 'global.conf'; |
27 |
|
|
|
28 |
|
|
if (! -f $config_file) { |
29 |
|
|
print qq{Usage: $0 [/path/to/local.conf] |
30 |
|
|
|
31 |
|
|
If local.conf is not specified, global.conf in current directory will |
32 |
|
|
be used. |
33 |
|
|
}; |
34 |
|
|
exit 1; |
35 |
|
|
} |
36 |
|
|
|
37 |
dpavlin |
41 |
my $mws = MWS::SWISH->new(config_file => $config_file); |
38 |
|
|
#my $mws = MWS::Plucene->new(config_file => $config_file, debug => $debug); |
39 |
dpavlin |
39 |
|
40 |
|
|
my ($local_addr,$local_port) = ('127.0.0.1',6969); |
41 |
|
|
|
42 |
|
|
my $listen = $mws->{config}->val('global', 'listen'); |
43 |
dpavlin |
41 |
|
44 |
|
|
print STDERR "using listen $listen\n" if ($listen); |
45 |
|
|
|
46 |
|
|
if ($listen && $listen =~ m/:/) { |
47 |
|
|
($local_addr,$local_port) = split(/:/,$listen,2); |
48 |
dpavlin |
39 |
} elsif ($listen) { |
49 |
|
|
$local_addr = $listen; |
50 |
|
|
} |
51 |
|
|
|
52 |
|
|
my $d = HTTP::Daemon->new( |
53 |
|
|
Reuse => 1, |
54 |
|
|
LocalAddr => $local_addr, |
55 |
|
|
LocalPort => $local_port, |
56 |
dpavlin |
41 |
) || die "can't create HTTP::Daemon on $local_addr:$local_port: $!"; |
57 |
dpavlin |
39 |
|
58 |
dpavlin |
6 |
my $cgi = new CGI::Lite; |
59 |
|
|
my $tt = Template->new({ |
60 |
|
|
INCLUDE_PATH => $mws->{config}->val('global', 'templates'), |
61 |
|
|
FILTERS => { |
62 |
|
|
'body5' => \&body5_filter, |
63 |
dpavlin |
28 |
'body' => \&body_filter, |
64 |
dpavlin |
6 |
}, |
65 |
dpavlin |
20 |
EVAL_PERL => 1, |
66 |
dpavlin |
6 |
}); |
67 |
|
|
|
68 |
dpavlin |
25 |
my $static_html = $mws->{config}->val('global', 'static_html'); |
69 |
|
|
|
70 |
dpavlin |
13 |
print "Web server ready at: ", $d->url, "\n"; |
71 |
dpavlin |
6 |
|
72 |
dpavlin |
13 |
|
73 |
dpavlin |
5 |
while ( my $c = $d->accept ) { |
74 |
|
|
while ( my $r = $c->get_request ) { |
75 |
|
|
|
76 |
|
|
# environs that a webserver should set. |
77 |
|
|
$ENV{'REQUEST_METHOD'} = $r->method; |
78 |
|
|
$ENV{'GATEWAY_INTERFACE'} = "CGI/1.0"; |
79 |
|
|
$ENV{'SERVER_PROTOCOL'} = $r->protocol; |
80 |
|
|
$ENV{'CONTENT_TYPE'} = $r->content_type; |
81 |
|
|
|
82 |
dpavlin |
6 |
# this part is based on CGI::Lite |
83 |
dpavlin |
5 |
|
84 |
dpavlin |
6 |
$cgi->close_all_files(); |
85 |
|
|
$cgi->{web_data} = {}; |
86 |
|
|
$cgi->{ordered_keys} = []; |
87 |
|
|
$cgi->{all_handles} = []; |
88 |
|
|
$cgi->{error_status} = 0; |
89 |
|
|
$cgi->{error_message} = undef; |
90 |
|
|
|
91 |
dpavlin |
5 |
if ( $r->method eq 'GET' || $r->uri =~ /\?/ ) { |
92 |
dpavlin |
6 |
my $query_string = $r->uri; |
93 |
|
|
$query_string =~ s/[^\?]+\?(.*)/$1/; |
94 |
|
|
$cgi->_decode_url_encoded_data (\$query_string, 'form'); |
95 |
|
|
|
96 |
|
|
} elsif ( $r->method eq 'POST' ) { |
97 |
|
|
|
98 |
|
|
if ($r->content_type eq 'application/x-www-form-urlencoded') { |
99 |
|
|
# local $^W = 0; |
100 |
|
|
$cgi->_decode_url_encoded_data (\$r->content, 'form'); |
101 |
|
|
} elsif ($r->content_type =~ /multipart\/form-data/) { |
102 |
|
|
my ($boundary) = $r->content_type =~ /boundary=(\S+)$/; |
103 |
|
|
$cgi->_parse_multipart_data ($r->content_length, $boundary); |
104 |
|
|
} |
105 |
|
|
} else { |
106 |
|
|
$c->send_error(RC_FORBIDDEN); |
107 |
dpavlin |
5 |
} |
108 |
|
|
|
109 |
dpavlin |
6 |
my $param = $cgi->{web_data}; |
110 |
|
|
my $url = $r->url->path; |
111 |
dpavlin |
5 |
|
112 |
dpavlin |
6 |
# XXX LOG |
113 |
dpavlin |
30 |
print $r->method," ",$url,"\n"; |
114 |
dpavlin |
42 |
print Dumper($param,$mws->{counter}),"\n" if ($debug); |
115 |
dpavlin |
5 |
|
116 |
dpavlin |
25 |
# is this static page? |
117 |
|
|
if ($static_html && -f "$static_html/$url") { |
118 |
|
|
print "static file: $static_html/$url\n" if ($debug); |
119 |
|
|
$c->send_file_response("$static_html/$url"); |
120 |
|
|
$c->close; |
121 |
|
|
next; |
122 |
|
|
} |
123 |
|
|
|
124 |
dpavlin |
7 |
# template file name (use ?format=html as default) |
125 |
|
|
my $tpl_file = 'master.'; |
126 |
|
|
$tpl_file .= $param->{'format'} || 'html'; |
127 |
|
|
|
128 |
dpavlin |
20 |
# parse date from url |
129 |
|
|
my ($yyyy,$mm,$dd) = $mws->yyyymmdd; |
130 |
|
|
|
131 |
|
|
my $yyyymm; |
132 |
|
|
|
133 |
|
|
my $date_limit; |
134 |
|
|
|
135 |
|
|
if ($url =~ m,^/(\d{4})[/-](\d+)[/-](\d+),) { |
136 |
|
|
($yyyy, $mm, $dd) = $mws->fmtdate($1,$2,$3); |
137 |
|
|
$date_limit = "$yyyy-$mm-$dd"; |
138 |
|
|
} elsif ($url =~ m,^/(\d{4})[/-](\d+),) { |
139 |
|
|
($yyyy,$mm) = $mws->fmtdate($1,$2); |
140 |
|
|
$date_limit = "$yyyy-$mm"; |
141 |
|
|
} elsif ($url =~ m,^/(\d{4}),) { |
142 |
|
|
$date_limit = $mws->fmtdate($1); |
143 |
|
|
} |
144 |
|
|
|
145 |
dpavlin |
7 |
# |
146 |
|
|
# implement functionality and generate HTML |
147 |
|
|
# |
148 |
dpavlin |
6 |
my $html; |
149 |
dpavlin |
5 |
|
150 |
dpavlin |
12 |
if ($param->{'search_val'} && $param->{'search_fld'} && !$param->{'search'}) { |
151 |
|
|
$param->{'search'} = $param->{'search_fld'}.":".$param->{'search_val'}; |
152 |
dpavlin |
16 |
} elsif ($param->{'search'}) { |
153 |
|
|
($param->{'search_fld'}, $param->{'search_val'}) = split(/:/,$param->{'search'},2); |
154 |
dpavlin |
12 |
} |
155 |
|
|
|
156 |
dpavlin |
13 |
my $tpl_var = { |
157 |
dpavlin |
20 |
param => $param, |
158 |
|
|
yyyy => $yyyy, |
159 |
|
|
mm => $mm, |
160 |
|
|
dd => $dd, |
161 |
dpavlin |
21 |
date_limit => $date_limit, |
162 |
dpavlin |
13 |
}; |
163 |
|
|
|
164 |
dpavlin |
27 |
# is this access to root of web server? |
165 |
|
|
if ($url eq "/" && !$param->{'search'}) { |
166 |
|
|
# if first access, go to current year |
167 |
|
|
$date_limit = $mws->fmtdate($yyyy); |
168 |
dpavlin |
28 |
$param->{sort_by} = "date desc"; |
169 |
dpavlin |
27 |
} |
170 |
|
|
|
171 |
dpavlin |
21 |
# ?show_id=XXXXxxxx___message_id___xxxxXXXX |
172 |
|
|
if ($param->{'show_id'}) { |
173 |
dpavlin |
20 |
|
174 |
dpavlin |
21 |
$mws->reset_counters; |
175 |
|
|
my $row = $mws->fetch_result_by_id($param->{'show_id'}); |
176 |
|
|
$tpl_var->{message} = $row; |
177 |
|
|
} elsif ($param->{'search'} || $date_limit) { |
178 |
dpavlin |
5 |
|
179 |
dpavlin |
21 |
# show search results |
180 |
|
|
# ?search=foo:bar |
181 |
dpavlin |
5 |
|
182 |
dpavlin |
27 |
my @search; |
183 |
|
|
push @search, $param->{'search'} if ($param->{'search'}); |
184 |
dpavlin |
21 |
|
185 |
|
|
if ($date_limit) { |
186 |
|
|
push @search, "and" if (@search); |
187 |
|
|
push @search, "date:\"$date_limit\""; |
188 |
|
|
} |
189 |
|
|
|
190 |
dpavlin |
24 |
if ($param->{sort_by}) { |
191 |
|
|
push @search, "sort:".$param->{sort_by}; |
192 |
|
|
} |
193 |
|
|
|
194 |
dpavlin |
21 |
print STDERR "search: ",join(" ",@search),"\n"; |
195 |
|
|
|
196 |
|
|
my $results = $mws->search(@search); |
197 |
dpavlin |
6 |
my @res = $mws->fetch_all_results(); |
198 |
dpavlin |
5 |
|
199 |
dpavlin |
20 |
$tpl_var->{results} = \@res if (@res); |
200 |
dpavlin |
25 |
$tpl_var->{total_hits} = $mws->{total_hits} || 0; |
201 |
dpavlin |
7 |
|
202 |
dpavlin |
30 |
# no hits, offer suggestions |
203 |
|
|
if (! $tpl_var->{results}) { |
204 |
|
|
@{$tpl_var->{apropos}} = $mws->apropos_index($param->{'search_fld'}, $param->{'search_val'}); |
205 |
|
|
} |
206 |
|
|
|
207 |
dpavlin |
6 |
} |
208 |
dpavlin |
5 |
|
209 |
dpavlin |
19 |
|
210 |
|
|
# push counters to template |
211 |
dpavlin |
42 |
foreach my $f (qw(from to cc bcc folder)) { |
212 |
dpavlin |
19 |
my $h = $mws->counter($f) || next; |
213 |
|
|
my @a; |
214 |
|
|
foreach my $k (sort { $h->{$b}->{usage} <=> $h->{$a}->{usage} } keys %$h) { |
215 |
|
|
push @a, $h->{$k}; |
216 |
|
|
} |
217 |
|
|
$tpl_var->{counters}->{$f} = [ @a ] if (@a); |
218 |
|
|
} |
219 |
|
|
|
220 |
dpavlin |
20 |
# push calendar in template |
221 |
|
|
$tpl_var->{calendar} = $mws->counter('calendar'); |
222 |
|
|
|
223 |
dpavlin |
13 |
$tt->process($tpl_file, $tpl_var, \$html) || die $tt->error(); |
224 |
|
|
|
225 |
dpavlin |
7 |
# |
226 |
|
|
# send HTMLto client |
227 |
|
|
# |
228 |
|
|
|
229 |
dpavlin |
5 |
my $res = HTTP::Response->new(RC_OK); |
230 |
dpavlin |
14 |
$res->header( 'Content-type' => 'text/html; charset=ISO-8859-2' ); |
231 |
dpavlin |
6 |
$res->content($html); |
232 |
dpavlin |
5 |
$c->send_response($res); |
233 |
|
|
|
234 |
|
|
$c->close; |
235 |
|
|
} |
236 |
|
|
undef($c); |
237 |
|
|
} |
238 |
dpavlin |
6 |
|
239 |
|
|
# template toolkit filter |
240 |
|
|
|
241 |
dpavlin |
28 |
sub html_escape($) { |
242 |
dpavlin |
30 |
my $text = shift || return; |
243 |
dpavlin |
28 |
|
244 |
dpavlin |
30 |
# don't re-escape html |
245 |
dpavlin |
37 |
#return $text if ($text =~ /&(?:lt|gt|amp|quot);/); |
246 |
dpavlin |
30 |
|
247 |
dpavlin |
28 |
# Escape <, >, & and ", and to produce valid XML |
248 |
|
|
my %escape = ('<'=>'<', '>'=>'>', '&'=>'&', '"'=>'"'); |
249 |
|
|
my $escape_re = join '|' => keys %escape; |
250 |
|
|
|
251 |
|
|
$text =~ s/($escape_re)/$escape{$1}/gs; |
252 |
dpavlin |
34 |
|
253 |
|
|
while ($text =~ s/#-#(quote|signature)(\d*)##(.+?)##\1\2#-#/<span class="$1">$3<\/span>/gs) { } ; |
254 |
|
|
|
255 |
dpavlin |
28 |
return $text; |
256 |
|
|
} |
257 |
|
|
|
258 |
dpavlin |
12 |
#use Text::Context::EitherSide; |
259 |
|
|
|
260 |
dpavlin |
6 |
sub body5_filter { |
261 |
|
|
my $text = shift; |
262 |
dpavlin |
28 |
|
263 |
|
|
# remove quote |
264 |
dpavlin |
37 |
$text =~ s/^[\>:\|=]+[^\n\r]*[\n\r]*$/#-q-#/msg; |
265 |
dpavlin |
28 |
# remove quote author |
266 |
dpavlin |
37 |
$text =~ s/[\n\r]+[^\n\r]+:\s*(?:#-q-#[\n\r*])+//gs; |
267 |
|
|
$text =~ s/^[^\n\r]+:\s*(?:#-q-#[\n\r]*)+//gs; |
268 |
dpavlin |
30 |
$text =~ s/#-q-#[\n\r]*//gs; |
269 |
dpavlin |
28 |
# outlook quoting |
270 |
|
|
$text =~ s/(\s*--+\s*Original\s+Message\s*--+.*)$//si; |
271 |
dpavlin |
34 |
$text =~ s/(\s*--+\s*Forwarded\s+message.+\s*--+.*)$//si; |
272 |
dpavlin |
28 |
|
273 |
|
|
# remove signature |
274 |
dpavlin |
44 |
$text =~ s/(?:^|[\n\r]+)*--\s*[\n\r]+.*$//s; |
275 |
|
|
$text =~ s/(?:^|[\n\r]+)*_____+[\n\r]+.*$//s; |
276 |
dpavlin |
28 |
|
277 |
|
|
# compress cr/lf |
278 |
|
|
$text =~ s/[\n\r]+/\n/gs; |
279 |
|
|
|
280 |
|
|
# remove whitespaces |
281 |
|
|
$text =~ s/^\n+//gs; |
282 |
|
|
$text =~ s/[\s\n]+$//gs; |
283 |
|
|
|
284 |
dpavlin |
34 |
if ($text eq "") { |
285 |
|
|
$text="#-#quote##forwarded message##quote#-#"; |
286 |
|
|
} |
287 |
|
|
|
288 |
dpavlin |
28 |
# cut to 5 lines; |
289 |
dpavlin |
14 |
if ($text =~ s,^((?:.*?[\n\r]){5}).*$,$1,s) { |
290 |
|
|
$text =~ s/[\n\r]*$/ .../; |
291 |
|
|
} |
292 |
dpavlin |
12 |
|
293 |
|
|
# my $context = Text::Context::EitherSide->new($text, context => 5); |
294 |
|
|
# return $context->as_string("perl"); |
295 |
|
|
|
296 |
dpavlin |
28 |
return html_escape($text); |
297 |
|
|
} |
298 |
|
|
|
299 |
|
|
sub body_filter { |
300 |
|
|
my $text = shift; |
301 |
|
|
|
302 |
|
|
my $sig = ''; |
303 |
|
|
|
304 |
|
|
# remove signature |
305 |
|
|
if ($text =~ s/([\n\r]+)(--\s*[\n\r]+.*)$//s) { |
306 |
|
|
$sig = "$1#-#signature##$2##signature#-#"; |
307 |
dpavlin |
44 |
} elsif ($text =~s/(^|[\n\r]+)*(_____+[\n\r]+.*)$//s) { |
308 |
|
|
$sig = "$1#-#signature##$2##signature#-#"; |
309 |
dpavlin |
28 |
} |
310 |
|
|
|
311 |
|
|
# find quoted text |
312 |
dpavlin |
37 |
$text =~ s/^([\>:\|=]+[^\n\r]*[\n\r]*)$/#-#quote1##$1##quote1#-#/mg; |
313 |
dpavlin |
30 |
$text =~ s/(--+\s*Original\s+Message\s*--+.*)$/#-#quote2##$1##quote2#-#/si; |
314 |
dpavlin |
34 |
$text =~ s/(--+\s*Forwarded\s+message.+\s*--+.*)$/#-#quote3##$1##quote3#-#/si; |
315 |
dpavlin |
28 |
|
316 |
|
|
$text = html_escape($text . $sig); |
317 |
dpavlin |
6 |
return $text; |
318 |
|
|
} |
319 |
dpavlin |
7 |
|