1 |
dpavlin |
2 |
#!/usr/bin/perl -w |
2 |
|
|
|
3 |
|
|
package MWS; |
4 |
|
|
|
5 |
|
|
use strict; |
6 |
|
|
use warnings; |
7 |
|
|
use Carp; |
8 |
|
|
|
9 |
dpavlin |
3 |
use Mail::Box::Manager; |
10 |
dpavlin |
4 |
use Config::IniFiles; |
11 |
dpavlin |
13 |
use POSIX qw(strftime); |
12 |
dpavlin |
14 |
use Text::Autoformat; |
13 |
dpavlin |
16 |
use Text::Iconv; |
14 |
dpavlin |
19 |
use Text::Unaccent; |
15 |
dpavlin |
20 |
use Date::Parse; |
16 |
|
|
use POSIX qw(strftime); |
17 |
dpavlin |
21 |
use MIME::Base64; |
18 |
dpavlin |
14 |
|
19 |
dpavlin |
12 |
#use MWS_plucene; |
20 |
|
|
use MWS_swish; |
21 |
dpavlin |
3 |
|
22 |
dpavlin |
2 |
require Exporter; |
23 |
|
|
|
24 |
|
|
our @ISA = qw(Exporter); |
25 |
|
|
|
26 |
|
|
our %EXPORT_TAGS = (); |
27 |
|
|
our @EXPORT_OK; |
28 |
|
|
our @EXPORT; |
29 |
|
|
|
30 |
|
|
our $VERSION = '1.00'; |
31 |
|
|
|
32 |
|
|
my $folder; # placeholder for folders |
33 |
|
|
|
34 |
dpavlin |
13 |
my $debug = 2; |
35 |
dpavlin |
2 |
|
36 |
|
|
sub new { |
37 |
|
|
my $class = shift; |
38 |
|
|
my $self = {}; |
39 |
|
|
bless($self, $class); |
40 |
|
|
|
41 |
dpavlin |
4 |
my $config_file = shift || die "need index file"; |
42 |
dpavlin |
2 |
|
43 |
dpavlin |
4 |
$self->{config} = new Config::IniFiles( -file => $config_file ); |
44 |
dpavlin |
2 |
|
45 |
dpavlin |
4 |
my $index_file = $self->{config}->val('global', 'index') || croak "can't find [index] section in config file with path of index"; |
46 |
|
|
|
47 |
dpavlin |
12 |
$self->{mgr} = Mail::Box::Manager->new(access => 'r'); |
48 |
|
|
$self->{index_file} = $index_file; |
49 |
dpavlin |
4 |
|
50 |
dpavlin |
2 |
# placeholder for opened folders |
51 |
|
|
$self->{folder} = {}; |
52 |
|
|
|
53 |
dpavlin |
14 |
$self->{wrap_margin} = $self->{config}->val('global', 'wrap_margin'); |
54 |
dpavlin |
17 |
$self->{max_results} = $self->{config}->val('global', 'max_results') || 100; |
55 |
dpavlin |
19 |
$self->reset_counters; |
56 |
dpavlin |
14 |
|
57 |
dpavlin |
2 |
return $self; |
58 |
|
|
} |
59 |
|
|
|
60 |
dpavlin |
19 |
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 |
dpavlin |
20 |
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 |
dpavlin |
19 |
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 |
dpavlin |
14 |
sub mbox_name2path { |
142 |
|
|
my $self = shift; |
143 |
|
|
|
144 |
|
|
my $mbox = shift || croak "folder_name2path needs mbox name"; |
145 |
|
|
|
146 |
|
|
return $self->{config}->val('folders', $mbox) || croak "comeone removed folder $mbox from config?"; |
147 |
|
|
} |
148 |
|
|
|
149 |
dpavlin |
7 |
sub open_folder { |
150 |
dpavlin |
2 |
my $self = shift; |
151 |
|
|
|
152 |
dpavlin |
7 |
my $mbox = shift || croak "open_folder needs mbox name"; |
153 |
dpavlin |
2 |
|
154 |
|
|
if (! $self->{folder}->{$mbox}) { |
155 |
dpavlin |
14 |
my $mbox_path = $self->mbox_name2path($mbox); |
156 |
|
|
|
157 |
dpavlin |
13 |
print STDERR "about to open_folder($mbox)\n" if ($debug == 2); |
158 |
dpavlin |
14 |
|
159 |
dpavlin |
4 |
$self->{folder}->{$mbox} = $self->{mgr}->open($mbox_path) || croak "can't open folder $mbox at '$mbox_path': $!"; |
160 |
dpavlin |
14 |
|
161 |
|
|
print STDERR "open_folder($mbox) ok\n" if ($debug); |
162 |
dpavlin |
2 |
} |
163 |
|
|
|
164 |
dpavlin |
16 |
$self->{fetch_count} = 0; |
165 |
|
|
|
166 |
dpavlin |
7 |
return $self->{folder}->{$mbox}; |
167 |
|
|
|
168 |
|
|
} |
169 |
|
|
|
170 |
dpavlin |
14 |
sub close_folder { |
171 |
|
|
my $self = shift; |
172 |
|
|
|
173 |
|
|
my $mbox = shift || croak "open_folder needs mbox name"; |
174 |
|
|
|
175 |
dpavlin |
16 |
$self->{folder}->{$mbox}->close(write => 'NEVER') || croak "can't close folder $mbox"; |
176 |
|
|
|
177 |
|
|
# XXX this is rather agressive!!! |
178 |
|
|
$self->{folder} = {}; |
179 |
|
|
return |
180 |
dpavlin |
14 |
} |
181 |
|
|
|
182 |
dpavlin |
7 |
sub fetch_message { |
183 |
|
|
my $self = shift; |
184 |
|
|
|
185 |
|
|
my $mbox_id = shift || die "need mbox_id!"; |
186 |
|
|
my ($mbox,$id) = split(/ /,$mbox_id); |
187 |
|
|
|
188 |
|
|
# return message with ID |
189 |
dpavlin |
13 |
print STDERR "fetch $id from $mbox\n" if ($debug); |
190 |
dpavlin |
16 |
|
191 |
|
|
if ($self->{fetch_count}++ > 100) { |
192 |
|
|
$self->close_folder($mbox); |
193 |
|
|
print STDERR "close_folder($mbox) forced on ",$self->{fetch_count},"iteration\n"; |
194 |
|
|
} |
195 |
|
|
|
196 |
dpavlin |
17 |
my $msg = $self->open_folder($mbox)->find($id); |
197 |
|
|
if ($msg) { |
198 |
|
|
return $msg; |
199 |
|
|
} else { |
200 |
dpavlin |
2 |
print STDERR "can't find message $id in $mbox. Time to re-index?\n"; |
201 |
dpavlin |
17 |
return; |
202 |
|
|
} |
203 |
dpavlin |
2 |
} |
204 |
|
|
|
205 |
|
|
|
206 |
|
|
sub search { |
207 |
|
|
my $self = shift; |
208 |
|
|
|
209 |
dpavlin |
21 |
carp "search called without argument!" if (! @_); |
210 |
dpavlin |
2 |
|
211 |
dpavlin |
19 |
$self->reset_counters; |
212 |
|
|
|
213 |
dpavlin |
21 |
print STDERR "search(",join(" ",@_),")\n" if ($debug == 2); |
214 |
|
|
my @index_ids = $self->search_index(@_); |
215 |
dpavlin |
2 |
|
216 |
|
|
$self->{'index_ids'} = \@index_ids; |
217 |
|
|
|
218 |
dpavlin |
17 |
#my $results = $#index_ids + 1; |
219 |
|
|
#$self->{'results'} = $results; |
220 |
|
|
|
221 |
|
|
my $results = $self->{'total_hits'} || ($#index_ids + 1); |
222 |
dpavlin |
2 |
|
223 |
|
|
$self->{'curr_result'} = 0; |
224 |
|
|
|
225 |
dpavlin |
13 |
print STDERR "$results results\n" if ($debug == 2); |
226 |
|
|
|
227 |
dpavlin |
2 |
return $results || 'error'; |
228 |
|
|
} |
229 |
|
|
|
230 |
dpavlin |
16 |
sub decode_qp($) { |
231 |
|
|
my $self = shift; |
232 |
|
|
|
233 |
|
|
my $tmp = shift || return; |
234 |
|
|
|
235 |
dpavlin |
21 |
sub decode($$$) { |
236 |
|
|
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 $qp; |
248 |
|
|
} |
249 |
|
|
|
250 |
dpavlin |
22 |
print STDERR "$qp\n" if ($debug == 2); |
251 |
dpavlin |
21 |
|
252 |
dpavlin |
20 |
my $iconv = Text::Iconv->new($cp,'ISO-8859-2'); |
253 |
|
|
return $iconv->convert($qp) || ''; |
254 |
dpavlin |
16 |
} |
255 |
|
|
|
256 |
dpavlin |
21 |
$tmp =~ s/=\?([^\?]+)\?([QB])\?(.+?)\?=/decode($1,$2,$3)/ige; |
257 |
dpavlin |
17 |
$tmp =~ s/^\s*["']+(.*?)["']+\s*$/$1/g; |
258 |
dpavlin |
20 |
#print STDERR "$tmp\n" if ($debug == 2); |
259 |
dpavlin |
16 |
return $tmp; |
260 |
|
|
} |
261 |
|
|
|
262 |
dpavlin |
2 |
sub unroll($$$) { |
263 |
dpavlin |
13 |
my $self = shift; |
264 |
|
|
|
265 |
dpavlin |
2 |
my ($message,$part,$sub) = @_; |
266 |
|
|
|
267 |
|
|
my @arr; |
268 |
|
|
|
269 |
dpavlin |
17 |
return if (! $message->$part); |
270 |
|
|
|
271 |
dpavlin |
2 |
foreach my $from ($message->$part) { |
272 |
dpavlin |
16 |
my $tmp = $from->$sub || next; |
273 |
|
|
|
274 |
|
|
$tmp = $self->decode_qp($tmp); |
275 |
|
|
push @arr, $tmp; |
276 |
dpavlin |
2 |
} |
277 |
dpavlin |
14 |
|
278 |
|
|
return @arr; |
279 |
dpavlin |
2 |
} |
280 |
dpavlin |
4 |
|
281 |
|
|
sub fetch_all_results { |
282 |
|
|
my $self = shift; |
283 |
|
|
|
284 |
|
|
croak "results called before search!" if (! $self->{'index_ids'}); |
285 |
|
|
|
286 |
dpavlin |
14 |
print STDERR "fetch_all_results [",scalar @{$self->{'index_ids'}},"]\n" if ($debug == 2); |
287 |
dpavlin |
13 |
|
288 |
dpavlin |
4 |
my @arr; |
289 |
|
|
|
290 |
|
|
foreach my $id (@{$self->{'index_ids'}}) { |
291 |
|
|
push @arr, $self->fetch_result_by_id($id); |
292 |
|
|
} |
293 |
|
|
|
294 |
dpavlin |
19 |
|
295 |
dpavlin |
4 |
return @arr; |
296 |
|
|
} |
297 |
|
|
|
298 |
dpavlin |
2 |
sub fetch_result { |
299 |
|
|
my $self = shift; |
300 |
|
|
|
301 |
dpavlin |
3 |
my $args = {@_}; |
302 |
|
|
|
303 |
dpavlin |
2 |
croak "results called before search!" if (! $self->{'index_ids'}); |
304 |
|
|
|
305 |
|
|
my $curr = $self->{'curr_result'}++; |
306 |
|
|
|
307 |
|
|
my $id = $self->{'index_ids'}->[$curr]; |
308 |
dpavlin |
14 |
|
309 |
|
|
print STDERR "fetch_result: $curr = $id\n" if ($debug == 2); |
310 |
dpavlin |
2 |
|
311 |
dpavlin |
4 |
return $self->fetch_result_by_id($id); |
312 |
|
|
} |
313 |
dpavlin |
2 |
|
314 |
dpavlin |
7 |
sub plain_text_body { |
315 |
|
|
my $self = shift; |
316 |
|
|
my $message = shift || croak "plain_text_body needs message!"; |
317 |
|
|
|
318 |
dpavlin |
14 |
my $body; |
319 |
|
|
|
320 |
dpavlin |
7 |
if (! $message->isMultipart) { |
321 |
dpavlin |
14 |
$body = $message->decoded->string; |
322 |
dpavlin |
7 |
} else { |
323 |
|
|
foreach my $part ($message->parts) { |
324 |
|
|
if ($part->body->mimeType eq 'text/plain') { |
325 |
dpavlin |
14 |
$body = $part->decoded->string; |
326 |
|
|
last; |
327 |
dpavlin |
7 |
} |
328 |
|
|
} |
329 |
|
|
} |
330 |
dpavlin |
14 |
|
331 |
|
|
# reformat with Text::Autoformat |
332 |
|
|
my $wrap = $self->{wrap_margin}; |
333 |
|
|
if ($wrap && $body && $body =~ m/^.{$wrap}..*$/m) { |
334 |
dpavlin |
16 |
$body =~ s/[\r\n]/\n/gs; |
335 |
dpavlin |
21 |
$body = autoformat($body, {right=>$wrap, all=>1}); |
336 |
dpavlin |
14 |
$body .="\n[reformated using autoformat, margin at $wrap]" if ($debug == 2); |
337 |
|
|
} |
338 |
|
|
|
339 |
|
|
return $body; |
340 |
dpavlin |
7 |
} |
341 |
|
|
|
342 |
|
|
|
343 |
dpavlin |
4 |
sub fetch_result_by_id { |
344 |
|
|
my $self = shift; |
345 |
|
|
|
346 |
|
|
my $id = shift || return; |
347 |
|
|
|
348 |
dpavlin |
13 |
my $row = $self->{cache}->{$id}; |
349 |
dpavlin |
2 |
|
350 |
dpavlin |
13 |
if (! $row) { |
351 |
dpavlin |
2 |
|
352 |
dpavlin |
13 |
print STDERR "fetch_result_by_id($id) not in cache, hitting disk\n" if ($debug == 2); |
353 |
dpavlin |
2 |
|
354 |
dpavlin |
17 |
my $message = $self->fetch_message($id) || return; |
355 |
dpavlin |
13 |
|
356 |
|
|
$row->{'id'} = $id; |
357 |
dpavlin |
19 |
|
358 |
|
|
foreach my $p (qw(from to cc bcc)) { |
359 |
dpavlin |
22 |
foreach my $v ($self->unroll($message,$p,'phrase')) { |
360 |
dpavlin |
19 |
push @{$row->{$p}},$v; |
361 |
|
|
$self->add_counter($p,$v); |
362 |
|
|
} |
363 |
|
|
} |
364 |
dpavlin |
16 |
$row->{'subject'} = $self->decode_qp($message->subject); |
365 |
dpavlin |
13 |
$row->{'body'} = $self->plain_text_body($message); |
366 |
dpavlin |
20 |
my $utime = str2time($message->date); |
367 |
dpavlin |
13 |
|
368 |
dpavlin |
20 |
$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 |
dpavlin |
13 |
# XXX store in cache? |
374 |
|
|
$self->{cache}->{$id} = $row; |
375 |
|
|
print STDERR "$id stored in cache\n" if ($debug == 2); |
376 |
|
|
} else { |
377 |
|
|
print STDERR "fetch_result_by_id($id) in cache\n" if ($debug == 2); |
378 |
|
|
} |
379 |
|
|
|
380 |
dpavlin |
2 |
return $row; |
381 |
|
|
|
382 |
|
|
} |