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; |
31 |
|
|
32 |
my $folder; # placeholder for folders |
my $folder; # placeholder for folders |
33 |
|
|
34 |
my $debug = 2; |
my $debug = 1; |
35 |
|
|
36 |
sub new { |
sub new { |
37 |
my $class = shift; |
my $class = shift; |
42 |
|
|
43 |
$self->{config} = new Config::IniFiles( -file => $config_file ); |
$self->{config} = new Config::IniFiles( -file => $config_file ); |
44 |
|
|
45 |
my $index_file = $self->{config}->val('global', 'index') || croak "can't find [index] section in config file with path of index"; |
$self->{config_file} = $config_file; |
46 |
|
$config_file =~ s/\.conf.*$//; |
47 |
|
$self->{config_name} = $config_file; |
48 |
|
|
49 |
|
my $index_dir = $self->{config}->val('global', 'index') || croak "can't find [index] section in config file with path of index"; |
50 |
|
|
51 |
$self->{mgr} = Mail::Box::Manager->new(access => 'r'); |
$self->{mgr} = Mail::Box::Manager->new(access => 'r'); |
52 |
$self->{index_file} = $index_file; |
$self->{index_dir} = $index_dir; |
53 |
|
|
54 |
# placeholder for opened folders |
# placeholder for opened folders |
55 |
$self->{folder} = {}; |
$self->{folder} = {}; |
56 |
|
|
57 |
$self->{wrap_margin} = $self->{config}->val('global', 'wrap_margin'); |
$self->{wrap_margin} = $self->{config}->val('global', 'wrap_margin'); |
58 |
$self->{max_results} = $self->{config}->val('global', 'max_results') || 100; |
$self->{max_results} = $self->{config}->val('global', 'max_results') || 100; |
59 |
|
$self->reset_counters; |
60 |
|
|
61 |
return $self; |
return $self; |
62 |
} |
} |
63 |
|
|
64 |
|
sub normalize_string { |
65 |
|
my $self = shift; |
66 |
|
|
67 |
|
my $v = shift || return; |
68 |
|
|
69 |
|
$v = unac_string('ISO-8859-2', $v); |
70 |
|
$v = join('',sort split(/\s+/,$v)); |
71 |
|
$v =~ s/\W+//g; |
72 |
|
|
73 |
|
return lc($v); |
74 |
|
} |
75 |
|
|
76 |
|
# reset tables for search results |
77 |
|
sub reset_counters { |
78 |
|
my $self = shift; |
79 |
|
|
80 |
|
$self->{counter} = {}; |
81 |
|
|
82 |
|
# foreach my $c (qw(thread from to cc bcc lists links att)) { |
83 |
|
# $self->{counter}->{$c} = {}; |
84 |
|
# } |
85 |
|
|
86 |
|
} |
87 |
|
|
88 |
|
sub add_counter($$) { |
89 |
|
my $self = shift; |
90 |
|
|
91 |
|
my ($c,$v) = @_; |
92 |
|
my $k = $self->normalize_string($v); |
93 |
|
|
94 |
|
$self->{counter}->{$c}->{$k}->{name} = $v; |
95 |
|
return $self->{counter}->{$c}->{$k}->{usage}++; |
96 |
|
} |
97 |
|
|
98 |
|
sub yyyymmdd { |
99 |
|
my $self = shift; |
100 |
|
|
101 |
|
my $t = shift || time; |
102 |
|
|
103 |
|
my (undef,undef,undef,$dd,$mm,$yyyy) = localtime($t); |
104 |
|
$mm++; |
105 |
|
$yyyy+=1900; |
106 |
|
return ($yyyy,$mm,$dd); |
107 |
|
} |
108 |
|
|
109 |
|
sub fmtdate { |
110 |
|
my $self = shift; |
111 |
|
|
112 |
|
my @out; |
113 |
|
my @formats = qw(%04d %02d %02d); |
114 |
|
while (my $v = shift) { |
115 |
|
my $f = shift @formats; |
116 |
|
push @out, sprintf($f, $v); |
117 |
|
} |
118 |
|
|
119 |
|
print STDERR "fmtdate: ",join('|',@out),"\n" if ($debug == 2); |
120 |
|
|
121 |
|
return (wantarray ? @out : join("-",@out)); |
122 |
|
} |
123 |
|
|
124 |
|
sub add_counter_calendar($) { |
125 |
|
my $self = shift; |
126 |
|
|
127 |
|
my $t = shift || croak "add_counter_calendar without argument!"; |
128 |
|
|
129 |
|
my ($yyyy,$mm,$dd) = $self->fmtdate($self->yyyymmdd($t)); |
130 |
|
|
131 |
|
return $self->{counter}->{calendar}->{"$yyyy-$mm"}->{$dd}++; |
132 |
|
} |
133 |
|
|
134 |
|
|
135 |
|
sub counter { |
136 |
|
my $self = shift; |
137 |
|
|
138 |
|
my $c = shift || return; |
139 |
|
|
140 |
|
return if (! $self->{counter}->{$c}); |
141 |
|
|
142 |
|
return $self->{counter}->{$c}; |
143 |
|
} |
144 |
|
|
145 |
sub mbox_name2path { |
sub mbox_name2path { |
146 |
my $self = shift; |
my $self = shift; |
147 |
|
|
210 |
sub search { |
sub search { |
211 |
my $self = shift; |
my $self = shift; |
212 |
|
|
213 |
my $s = shift || carp "search called without argument!"; |
carp "search called without argument!" if (! @_); |
214 |
|
|
215 |
print STDERR "search_index($s)\n" if ($debug == 2); |
$self->reset_counters; |
216 |
my @index_ids = $self->search_index($s); |
|
217 |
|
print STDERR "search(",join(" ",@_),")\n" if ($debug == 2); |
218 |
|
my @index_ids = $self->search_index(@_); |
219 |
|
|
220 |
$self->{'index_ids'} = \@index_ids; |
$self->{'index_ids'} = \@index_ids; |
221 |
|
|
226 |
|
|
227 |
$self->{'curr_result'} = 0; |
$self->{'curr_result'} = 0; |
228 |
|
|
229 |
|
$self->reset_counters; |
230 |
|
|
231 |
print STDERR "$results results\n" if ($debug == 2); |
print STDERR "$results results\n" if ($debug == 2); |
232 |
|
|
233 |
return $results || 'error'; |
return $results || 'error'; |
238 |
|
|
239 |
my $tmp = shift || return; |
my $tmp = shift || return; |
240 |
|
|
241 |
sub decode($$) { |
sub decode($$$) { |
242 |
my ($cp,$qp) = @_; |
my ($cp,$enc,$qp) = @_; |
243 |
|
|
244 |
|
print STDERR "decode($cp,$qp) -> " if ($debug == 2); |
245 |
|
|
246 |
|
if (uc($enc) eq "Q") { |
247 |
|
$qp =~ s/=([a-f0-9][a-f0-9])/chr(hex($1))/ieg; |
248 |
|
$qp =~ s/_/ /g; |
249 |
|
} elsif (uc($enc) eq "B") { |
250 |
|
$qp = decode_base64($qp); |
251 |
|
} else { |
252 |
|
croak "unsupported encoding '$enc' in decode_qp\n"; |
253 |
|
return $qp; |
254 |
|
} |
255 |
|
|
256 |
|
print STDERR "$qp\n" if ($debug == 2); |
257 |
|
|
258 |
my $iconv = Text::Iconv->new($cp,'ISO-8859-2'); |
my $iconv = Text::Iconv->new($cp,'ISO-8859-2'); |
259 |
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; |
|
260 |
} |
} |
261 |
|
|
262 |
$tmp =~ s/=\?([^\?]+)\?Q\?(.+?)\?=/decode($1,$2)/ex; |
$tmp =~ s/=\?([^\?]+)\?([QB])\?(.+?)\?=/decode($1,$2,$3)/ige; |
263 |
$tmp =~ s/^\s*["']+(.*?)["']+\s*$/$1/g; |
$tmp =~ s/^\s*["']+(.*?)["']+\s*$/$1/g; |
264 |
|
#print STDERR "$tmp\n" if ($debug == 2); |
265 |
return $tmp; |
return $tmp; |
266 |
} |
} |
267 |
|
|
297 |
push @arr, $self->fetch_result_by_id($id); |
push @arr, $self->fetch_result_by_id($id); |
298 |
} |
} |
299 |
|
|
300 |
|
|
301 |
return @arr; |
return @arr; |
302 |
} |
} |
303 |
|
|
334 |
} |
} |
335 |
} |
} |
336 |
|
|
337 |
|
if (! $body) { |
338 |
|
$body = "[plain/text body not found]" if ($debug == 2); |
339 |
|
print STDERR "plain/text body not found\n" if ($debug); |
340 |
|
return; |
341 |
|
} |
342 |
|
|
343 |
# reformat with Text::Autoformat |
# reformat with Text::Autoformat |
344 |
my $wrap = $self->{wrap_margin}; |
my $wrap = $self->{wrap_margin}; |
345 |
if ($wrap && $body && $body =~ m/^.{$wrap}..*$/m) { |
if ($wrap && $body && $body =~ m/^.{$wrap}..*$/m) { |
346 |
$body =~ s/[\r\n]/\n/gs; |
$body = autoformat($body, {right=>$wrap, all=>1}); |
|
$body = autoformat($body, {right=>$wrap}); |
|
347 |
$body .="\n[reformated using autoformat, margin at $wrap]" if ($debug == 2); |
$body .="\n[reformated using autoformat, margin at $wrap]" if ($debug == 2); |
348 |
} |
} |
349 |
|
|
365 |
my $message = $self->fetch_message($id) || return; |
my $message = $self->fetch_message($id) || return; |
366 |
|
|
367 |
$row->{'id'} = $id; |
$row->{'id'} = $id; |
368 |
@{$row->{'from'}} = $self->unroll($message,'from','phrase'); |
|
369 |
@{$row->{'to'}} = $self->unroll($message,'to','phrase'); |
foreach my $p (qw(from to cc bcc)) { |
370 |
@{$row->{'cc'}} = $self->unroll($message,'cc','phrase'); |
foreach my $v ($self->unroll($message,$p,'phrase')) { |
371 |
|
push @{$row->{$p}},$v; |
372 |
|
$self->add_counter($p,$v); |
373 |
|
} |
374 |
|
} |
375 |
$row->{'subject'} = $self->decode_qp($message->subject); |
$row->{'subject'} = $self->decode_qp($message->subject); |
376 |
$row->{'body'} = $self->plain_text_body($message); |
$row->{'body'} = $self->plain_text_body($message); |
377 |
$row->{'date'} = $message->date; |
my $utime = str2time($message->date); |
378 |
|
|
379 |
|
$row->{'date_utime'} = $utime; |
380 |
|
|
381 |
|
$row->{'date'} = strftime("%Y-%m-%d %H:%M:%S", localtime($utime)); |
382 |
|
$self->add_counter_calendar($utime); |
383 |
|
|
384 |
# XXX store in cache? |
# XXX store in cache? |
385 |
$self->{cache}->{$id} = $row; |
$self->{cache}->{$id} = $row; |
386 |
print STDERR "$id stored in cache\n" if ($debug == 2); |
print STDERR "$id stored in cache\n" if ($debug == 2); |
387 |
} else { |
} else { |
388 |
print STDERR "fetch_result_by_id($id) in cache\n" if ($debug == 2); |
print STDERR "fetch_result_by_id($id) in cache\n" if ($debug == 2); |
389 |
|
foreach my $p (qw(from to cc bcc)) { |
390 |
|
foreach my $v (@{$row->{$p}}) { |
391 |
|
$self->add_counter($p,$v); |
392 |
|
} |
393 |
|
} |
394 |
|
|
395 |
|
$self->add_counter_calendar($row->{date_utime}); |
396 |
} |
} |
397 |
|
|
398 |
return $row; |
return $row; |