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