12 |
use Text::Autoformat; |
use Text::Autoformat; |
13 |
use Text::Iconv; |
use Text::Iconv; |
14 |
use Text::Unaccent; |
use Text::Unaccent; |
15 |
|
use Date::Parse; |
16 |
|
use POSIX qw(strftime); |
17 |
|
use MIME::Base64; |
18 |
|
|
19 |
#use MWS_plucene; |
#use MWS_plucene; |
20 |
use MWS_swish; |
use MWS_swish; |
66 |
$v = join('',sort split(/\s+/,$v)); |
$v = join('',sort split(/\s+/,$v)); |
67 |
$v =~ s/\W+//g; |
$v =~ s/\W+//g; |
68 |
|
|
69 |
return $v; |
return lc($v); |
70 |
} |
} |
71 |
|
|
72 |
# reset tables for search results |
# reset tables for search results |
91 |
return $self->{counter}->{$c}->{$k}->{usage}++; |
return $self->{counter}->{$c}->{$k}->{usage}++; |
92 |
} |
} |
93 |
|
|
94 |
|
sub yyyymmdd { |
95 |
|
my $self = shift; |
96 |
|
|
97 |
|
my $t = shift || time; |
98 |
|
|
99 |
|
my (undef,undef,undef,$dd,$mm,$yyyy) = localtime($t); |
100 |
|
$mm++; |
101 |
|
$yyyy+=1900; |
102 |
|
return ($yyyy,$mm,$dd); |
103 |
|
} |
104 |
|
|
105 |
|
sub fmtdate { |
106 |
|
my $self = shift; |
107 |
|
|
108 |
|
my @out; |
109 |
|
my @formats = qw(%04d %02d %02d); |
110 |
|
while (my $v = shift) { |
111 |
|
my $f = shift @formats; |
112 |
|
push @out, sprintf($f, $v); |
113 |
|
} |
114 |
|
|
115 |
|
print STDERR "fmtdate: ",join('|',@out),"\n"; |
116 |
|
|
117 |
|
return (wantarray ? @out : join("-",@out)); |
118 |
|
} |
119 |
|
|
120 |
|
sub add_counter_calendar($) { |
121 |
|
my $self = shift; |
122 |
|
|
123 |
|
my $t = shift || croak "add_counter_calendar without argument!"; |
124 |
|
|
125 |
|
my ($yyyy,$mm,$dd) = $self->fmtdate($self->yyyymmdd($t)); |
126 |
|
|
127 |
|
return $self->{counter}->{calendar}->{"$yyyy-$mm"}->{$dd}++; |
128 |
|
} |
129 |
|
|
130 |
|
|
131 |
sub counter { |
sub counter { |
132 |
my $self = shift; |
my $self = shift; |
133 |
|
|
206 |
sub search { |
sub search { |
207 |
my $self = shift; |
my $self = shift; |
208 |
|
|
209 |
my $s = shift || carp "search called without argument!"; |
carp "search called without argument!" if (! @_); |
210 |
|
|
211 |
$self->reset_counters; |
$self->reset_counters; |
212 |
|
|
213 |
print STDERR "search_index($s)\n" if ($debug == 2); |
print STDERR "search(",join(" ",@_),")\n" if ($debug == 2); |
214 |
my @index_ids = $self->search_index($s); |
my @index_ids = $self->search_index(@_); |
215 |
|
|
216 |
$self->{'index_ids'} = \@index_ids; |
$self->{'index_ids'} = \@index_ids; |
217 |
|
|
222 |
|
|
223 |
$self->{'curr_result'} = 0; |
$self->{'curr_result'} = 0; |
224 |
|
|
225 |
|
$self->reset_counters; |
226 |
|
|
227 |
print STDERR "$results results\n" if ($debug == 2); |
print STDERR "$results results\n" if ($debug == 2); |
228 |
|
|
229 |
return $results || 'error'; |
return $results || 'error'; |
234 |
|
|
235 |
my $tmp = shift || return; |
my $tmp = shift || return; |
236 |
|
|
237 |
sub decode($$) { |
sub decode($$$) { |
238 |
my ($cp,$qp) = @_; |
my ($cp,$enc,$qp) = @_; |
239 |
|
|
240 |
|
print STDERR "decode($cp,$qp) -> " if ($debug == 2); |
241 |
|
|
242 |
|
if (uc($enc) eq "Q") { |
243 |
|
$qp =~ s/=([a-f0-9][a-f0-9])/chr(hex($1))/ieg; |
244 |
|
$qp =~ s/_/ /g; |
245 |
|
} elsif (uc($enc) eq "B") { |
246 |
|
$qp = decode_base64($qp); |
247 |
|
} else { |
248 |
|
croak "unsupported encoding '$enc' in decode_qp\n"; |
249 |
|
return $qp; |
250 |
|
} |
251 |
|
|
252 |
|
print STDERR "$qp\n" if ($debug == 2); |
253 |
|
|
254 |
my $iconv = Text::Iconv->new($cp,'ISO-8859-2'); |
my $iconv = Text::Iconv->new($cp,'ISO-8859-2'); |
255 |
print STDERR "decode($cp,$qp) -> " if ($debug == 2); |
return $iconv->convert($qp) || ''; |
|
$qp =~ s/=([a-f0-9][a-f0-9])/chr(hex($1))/ieg; |
|
|
$qp =~ s/_/ /g; |
|
|
print STDERR "$qp\n" if ($debug == 2); |
|
|
return $iconv->convert($qp) || $qp; |
|
256 |
} |
} |
257 |
|
|
258 |
$tmp =~ s/=\?([^\?]+)\?Q\?(.+?)\?=/decode($1,$2)/ex; |
$tmp =~ s/=\?([^\?]+)\?([QB])\?(.+?)\?=/decode($1,$2,$3)/ige; |
259 |
$tmp =~ s/^\s*["']+(.*?)["']+\s*$/$1/g; |
$tmp =~ s/^\s*["']+(.*?)["']+\s*$/$1/g; |
260 |
|
#print STDERR "$tmp\n" if ($debug == 2); |
261 |
return $tmp; |
return $tmp; |
262 |
} |
} |
263 |
|
|
330 |
} |
} |
331 |
} |
} |
332 |
|
|
333 |
|
if (! $body) { |
334 |
|
$body = "[body of this message not found]\n" if ($debug == 2); |
335 |
|
$body .= $message->decoded->string; |
336 |
|
} |
337 |
|
|
338 |
# reformat with Text::Autoformat |
# reformat with Text::Autoformat |
339 |
my $wrap = $self->{wrap_margin}; |
my $wrap = $self->{wrap_margin}; |
340 |
if ($wrap && $body && $body =~ m/^.{$wrap}..*$/m) { |
if ($wrap && $body && $body =~ m/^.{$wrap}..*$/m) { |
341 |
$body =~ s/[\r\n]/\n/gs; |
$body = autoformat($body, {right=>$wrap, all=>1}); |
342 |
$body = autoformat($body, {right=>$wrap}); |
print "reformatted [$wrap]:\n$body\n" if ($debug == 2); |
343 |
$body .="\n[reformated using autoformat, margin at $wrap]" if ($debug == 2); |
$body .="\n[reformated using autoformat, margin at $wrap]" if ($debug == 2); |
344 |
} |
} |
345 |
|
|
363 |
$row->{'id'} = $id; |
$row->{'id'} = $id; |
364 |
|
|
365 |
foreach my $p (qw(from to cc bcc)) { |
foreach my $p (qw(from to cc bcc)) { |
366 |
foreach my $v ($self->unroll($message,'from','phrase')) { |
foreach my $v ($self->unroll($message,$p,'phrase')) { |
367 |
push @{$row->{$p}},$v; |
push @{$row->{$p}},$v; |
368 |
$self->add_counter($p,$v); |
$self->add_counter($p,$v); |
369 |
} |
} |
370 |
} |
} |
371 |
$row->{'subject'} = $self->decode_qp($message->subject); |
$row->{'subject'} = $self->decode_qp($message->subject); |
372 |
$row->{'body'} = $self->plain_text_body($message); |
$row->{'body'} = $self->plain_text_body($message); |
373 |
$row->{'date'} = $message->date; |
my $utime = str2time($message->date); |
374 |
|
|
375 |
|
$row->{'date_utime'} = $utime; |
376 |
|
|
377 |
|
$row->{'date'} = strftime("%Y-%m-%d %H:%M:%S", localtime($utime)); |
378 |
|
$self->add_counter_calendar($utime); |
379 |
|
|
380 |
# XXX store in cache? |
# XXX store in cache? |
381 |
$self->{cache}->{$id} = $row; |
$self->{cache}->{$id} = $row; |
382 |
print STDERR "$id stored in cache\n" if ($debug == 2); |
print STDERR "$id stored in cache\n" if ($debug == 2); |
383 |
} else { |
} else { |
384 |
print STDERR "fetch_result_by_id($id) in cache\n" if ($debug == 2); |
print STDERR "fetch_result_by_id($id) in cache\n" if ($debug == 2); |
385 |
|
foreach my $p (qw(from to cc bcc)) { |
386 |
|
foreach my $v (@{$row->{$p}}) { |
387 |
|
$self->add_counter($p,$v); |
388 |
|
} |
389 |
|
} |
390 |
|
|
391 |
|
$self->add_counter_calendar($row->{date_utime}); |
392 |
} |
} |
393 |
|
|
394 |
return $row; |
return $row; |