11 |
use POSIX qw(strftime); |
use POSIX qw(strftime); |
12 |
use Text::Autoformat; |
use Text::Autoformat; |
13 |
use Text::Iconv; |
use Text::Iconv; |
14 |
|
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; |
51 |
$self->{folder} = {}; |
$self->{folder} = {}; |
52 |
|
|
53 |
$self->{wrap_margin} = $self->{config}->val('global', 'wrap_margin'); |
$self->{wrap_margin} = $self->{config}->val('global', 'wrap_margin'); |
54 |
|
$self->{max_results} = $self->{config}->val('global', 'max_results') || 100; |
55 |
|
$self->reset_counters; |
56 |
|
|
57 |
return $self; |
return $self; |
58 |
} |
} |
59 |
|
|
60 |
|
sub normalize_string { |
61 |
|
my $self = shift; |
62 |
|
|
63 |
|
my $v = shift || return; |
64 |
|
|
65 |
|
$v = unac_string('ISO-8859-2', $v); |
66 |
|
$v = join('',sort split(/\s+/,$v)); |
67 |
|
$v =~ s/\W+//g; |
68 |
|
|
69 |
|
return $v; |
70 |
|
} |
71 |
|
|
72 |
|
# reset tables for search results |
73 |
|
sub reset_counters { |
74 |
|
my $self = shift; |
75 |
|
|
76 |
|
$self->{counter} = {}; |
77 |
|
|
78 |
|
# foreach my $c (qw(thread from to cc bcc lists links att)) { |
79 |
|
# $self->{counter}->{$c} = {}; |
80 |
|
# } |
81 |
|
|
82 |
|
} |
83 |
|
|
84 |
|
sub add_counter($$) { |
85 |
|
my $self = shift; |
86 |
|
|
87 |
|
my ($c,$v) = @_; |
88 |
|
my $k = $self->normalize_string($v); |
89 |
|
|
90 |
|
$self->{counter}->{$c}->{$k}->{name} = $v; |
91 |
|
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 { |
132 |
|
my $self = shift; |
133 |
|
|
134 |
|
my $c = shift || return; |
135 |
|
|
136 |
|
return if (! $self->{counter}->{$c}); |
137 |
|
|
138 |
|
return $self->{counter}->{$c}; |
139 |
|
} |
140 |
|
|
141 |
sub mbox_name2path { |
sub mbox_name2path { |
142 |
my $self = shift; |
my $self = shift; |
143 |
|
|
193 |
print STDERR "close_folder($mbox) forced on ",$self->{fetch_count},"iteration\n"; |
print STDERR "close_folder($mbox) forced on ",$self->{fetch_count},"iteration\n"; |
194 |
} |
} |
195 |
|
|
196 |
return $self->open_folder($mbox)->find($id) || |
my $msg = $self->open_folder($mbox)->find($id); |
197 |
|
if ($msg) { |
198 |
|
return $msg; |
199 |
|
} else { |
200 |
print STDERR "can't find message $id in $mbox. Time to re-index?\n"; |
print STDERR "can't find message $id in $mbox. Time to re-index?\n"; |
201 |
|
return; |
202 |
|
} |
203 |
} |
} |
204 |
|
|
205 |
|
|
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 |
print STDERR "search_index($s)\n" if ($debug == 2); |
$self->reset_counters; |
212 |
my @index_ids = $self->search_index($s); |
|
213 |
|
print STDERR "search(",join(" ",@_),")\n" if ($debug == 2); |
214 |
|
my @index_ids = $self->search_index(@_); |
215 |
|
|
216 |
$self->{'index_ids'} = \@index_ids; |
$self->{'index_ids'} = \@index_ids; |
217 |
|
|
218 |
my $results = $#index_ids + 1; |
#my $results = $#index_ids + 1; |
219 |
$self->{'results'} = $results; |
#$self->{'results'} = $results; |
220 |
|
|
221 |
|
my $results = $self->{'total_hits'} || ($#index_ids + 1); |
222 |
|
|
223 |
$self->{'curr_result'} = 0; |
$self->{'curr_result'} = 0; |
224 |
|
|
232 |
|
|
233 |
my $tmp = shift || return; |
my $tmp = shift || return; |
234 |
|
|
235 |
sub decode($$) { |
sub decode($$$) { |
236 |
my ($cp,$qp) = @_; |
my ($cp,$enc,$qp) = @_; |
237 |
|
|
238 |
|
print STDERR "decode($cp,$qp) -> " if ($debug == 2); |
239 |
|
|
240 |
|
if (uc($enc) eq "Q") { |
241 |
|
$qp =~ s/=([a-f0-9][a-f0-9])/chr(hex($1))/ieg; |
242 |
|
$qp =~ s/_/ /g; |
243 |
|
} elsif (uc($enc) eq "B") { |
244 |
|
$qp = decode_base64($qp); |
245 |
|
} else { |
246 |
|
croak "unsupported encoding '$enc' in decode_qp\n"; |
247 |
|
return $qp; |
248 |
|
} |
249 |
|
|
250 |
|
print STDERR "$qp -> " if ($debug == 2); |
251 |
|
|
252 |
my $iconv = Text::Iconv->new($cp,'ISO-8859-2'); |
my $iconv = Text::Iconv->new($cp,'ISO-8859-2'); |
253 |
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); |
|
254 |
} |
} |
255 |
|
|
256 |
$tmp =~ s/=\?([^\?]+)\?Q\?(.+)\?=/decode($1,$2)/ex; |
$tmp =~ s/=\?([^\?]+)\?([QB])\?(.+?)\?=/decode($1,$2,$3)/ige; |
257 |
|
$tmp =~ s/^\s*["']+(.*?)["']+\s*$/$1/g; |
258 |
|
#print STDERR "$tmp\n" if ($debug == 2); |
259 |
return $tmp; |
return $tmp; |
260 |
} |
} |
261 |
|
|
266 |
|
|
267 |
my @arr; |
my @arr; |
268 |
|
|
269 |
|
return if (! $message->$part); |
270 |
|
|
271 |
foreach my $from ($message->$part) { |
foreach my $from ($message->$part) { |
272 |
my $tmp = $from->$sub || next; |
my $tmp = $from->$sub || next; |
273 |
|
|
274 |
$tmp = $self->decode_qp($tmp); |
$tmp = $self->decode_qp($tmp); |
|
$tmp =~ s/^\s*["'](.*)["']\s*$/$1/; |
|
275 |
push @arr, $tmp; |
push @arr, $tmp; |
276 |
} |
} |
277 |
|
|
291 |
push @arr, $self->fetch_result_by_id($id); |
push @arr, $self->fetch_result_by_id($id); |
292 |
} |
} |
293 |
|
|
294 |
|
|
295 |
return @arr; |
return @arr; |
296 |
} |
} |
297 |
|
|
332 |
my $wrap = $self->{wrap_margin}; |
my $wrap = $self->{wrap_margin}; |
333 |
if ($wrap && $body && $body =~ m/^.{$wrap}..*$/m) { |
if ($wrap && $body && $body =~ m/^.{$wrap}..*$/m) { |
334 |
$body =~ s/[\r\n]/\n/gs; |
$body =~ s/[\r\n]/\n/gs; |
335 |
$body = autoformat($body, {right=>$wrap}); |
$body = autoformat($body, {right=>$wrap, all=>1}); |
336 |
$body .="\n[reformated using autoformat, margin at $wrap]" if ($debug == 2); |
$body .="\n[reformated using autoformat, margin at $wrap]" if ($debug == 2); |
337 |
} |
} |
338 |
|
|
351 |
|
|
352 |
print STDERR "fetch_result_by_id($id) not in cache, hitting disk\n" if ($debug == 2); |
print STDERR "fetch_result_by_id($id) not in cache, hitting disk\n" if ($debug == 2); |
353 |
|
|
354 |
my $message = $self->fetch_message($id) || print STDERR "can't fetch message '$id'"; |
my $message = $self->fetch_message($id) || return; |
355 |
|
|
356 |
$row->{'id'} = $id; |
$row->{'id'} = $id; |
357 |
@{$row->{'from'}} = $self->unroll($message,'from','phrase'); |
|
358 |
@{$row->{'to'}} = $self->unroll($message,'to','phrase'); |
foreach my $p (qw(from to cc bcc)) { |
359 |
@{$row->{'cc'}} = $self->unroll($message,'cc','phrase'); |
foreach my $v ($self->unroll($message,'from','phrase')) { |
360 |
|
push @{$row->{$p}},$v; |
361 |
|
$self->add_counter($p,$v); |
362 |
|
} |
363 |
|
} |
364 |
$row->{'subject'} = $self->decode_qp($message->subject); |
$row->{'subject'} = $self->decode_qp($message->subject); |
365 |
$row->{'body'} = $self->plain_text_body($message); |
$row->{'body'} = $self->plain_text_body($message); |
366 |
$row->{'date'} = $message->date; |
my $utime = str2time($message->date); |
367 |
|
|
368 |
|
$row->{'date_utime'} = $utime; |
369 |
|
|
370 |
|
$row->{'date'} = strftime("%Y-%m-%d %H:%M:%S", localtime($utime)); |
371 |
|
$self->add_counter_calendar($utime); |
372 |
|
|
373 |
# XXX store in cache? |
# XXX store in cache? |
374 |
$self->{cache}->{$id} = $row; |
$self->{cache}->{$id} = $row; |