mws/trunk/MWS.pm

#!/usr/bin/perl -w

package MWS;

use strict;
use warnings;

use lib '.';

use Carp;
use Mail::Box::Manager;
use Config::IniFiles;
use POSIX qw(strftime);
use Text::Autoformat;
use Text::Iconv;
use Text::Unaccent;
use Date::Parse;
use POSIX qw(strftime);
use MIME::Base64;

our $VERSION = '1.00';

my $debug = 1;

my @counters = qw(from to cc bcc folder);

sub new {
        my $class = shift;
        my $self = {@_};
        bless($self, $class);


        my $config_name = $self->{config_file} || croak "need config_file";
        $config_name =~ s/\.conf.*$//;
        $self->{config_name} = $config_name;

        $self->{config} = new Config::IniFiles( -file => $self->{config_file} );

        my $index_dir = $self->{config}->val('global', 'index') || croak "can't find [index] section in config file with path of index";

        $self->{mgr} = Mail::Box::Manager->new(access => 'r');
        $self->{index_dir} = $index_dir;

        # placeholder for opened folders
        $self->{folder} = {};

        $self->{wrap_margin} = $self->{config}->val('global', 'wrap_margin');
        $self->{max_results} = $self->{config}->val('global', 'max_results') || 100;
        $self->reset_counters;

        return $self;
}

sub normalize_string {
        my $self = shift;

        my $v = shift || return;

        $v = unac_string('ISO-8859-2', $v);
        $v = join('',sort split(/\s+/,$v));
        $v =~ s/\W+//g;

        return lc($v);
}

# reset tables for search results
sub reset_counters {
        my $self = shift;

        $self->{counter} = {};

#       foreach my $c (qw(thread from to cc bcc lists links att)) {
#               $self->{counter}->{$c} = {};
#       }

}

sub add_counter($$) {
        my $self = shift;

        my ($c,$v) = @_;
        my $k = $self->normalize_string($v);

        $self->{counter}->{$c}->{$k}->{name} = $v || return;
        return $self->{counter}->{$c}->{$k}->{usage}++;
}

sub yyyymmdd {
        my $self = shift;

        my $t = shift || time;

        my (undef,undef,undef,$dd,$mm,$yyyy) = localtime($t);
        $mm++;
        $yyyy+=1900;
        return ($yyyy,$mm,$dd);
}

sub fmtdate {
        my $self = shift;

        my @out;
        my @formats = qw(%04d %02d %02d);
        while (my $v = shift) {
                my $f = shift @formats;
                push @out, sprintf($f, $v);
        }

        print STDERR "fmtdate: ",join('|',@out),"\n" if ($debug == 2);

        return (wantarray ? @out : join("-",@out));
}

sub add_counter_calendar($) {
        my $self = shift;

        my $t = shift || croak "add_counter_calendar without argument!";
        
        my ($yyyy,$mm,$dd) = $self->fmtdate($self->yyyymmdd($t));

        return $self->{counter}->{calendar}->{"$yyyy-$mm"}->{$dd}++;
}


sub counter {
        my $self = shift;

        my $c = shift || return;

        return if (! $self->{counter}->{$c});

        return $self->{counter}->{$c};
}

sub mbox_name2path {
        my $self = shift;

        my $mbox = shift || croak "folder_name2path needs mbox name";

        return $self->{config}->val('folders', $mbox) || croak "comeone removed folder $mbox from config?";
}

sub open_folder {
        my $self = shift;

        my $mbox = shift || croak "open_folder needs mbox name";

        if (! $self->{folder}->{$mbox}) {
                my $mbox_path = $self->mbox_name2path($mbox);

                print STDERR "about to open_folder($mbox)\n" if ($debug == 2);

                $self->{folder}->{$mbox} = $self->{mgr}->open($mbox_path) || croak "can't open folder $mbox at '$mbox_path': $!";

                print STDERR "open_folder($mbox) ok\n" if ($debug);
        }

        $self->{fetch_count} = 0;

        return $self->{folder}->{$mbox};

}

sub close_folder {
        my $self = shift;

        my $mbox = shift || croak "open_folder needs mbox name";

        $self->{folder}->{$mbox}->close(write => 'NEVER') || croak "can't close folder $mbox";

        # XXX this is rather agressive!!!
        $self->{folder} = {};
        return 
}

sub fetch_message {
        my $self = shift;

        my $mbox_id = shift || die "need mbox_id!";
        my ($mbox,$id) = split(/ /,$mbox_id);

        # return message with ID
        print STDERR "fetch $id from $mbox\n" if ($debug);

        if ($self->{fetch_count}++ > 100) {
                $self->close_folder($mbox);
                print STDERR "close_folder($mbox) forced on ",$self->{fetch_count},"iteration\n";
        }
        
        my $msg = $self->open_folder($mbox)->find($id);
        if ($msg) {
                return $msg;
        } else {
                print STDERR "can't find message $id in $mbox. Time to re-index?\n";
                return;
        }
}


sub search {
        my $self = shift;

        carp "search called without argument!" if (! @_);

        $self->reset_counters;

        print STDERR "search(",join(" ",@_),")\n" if ($debug == 2);
        my @index_ids = $self->search_index(@_);

        $self->{'index_ids'} = \@index_ids;

        #my $results = $#index_ids + 1;
        #$self->{'results'} = $results;
        
        my $results = $self->{'total_hits'} || ($#index_ids + 1);

        $self->{'curr_result'} = 0;

        $self->reset_counters;

        print STDERR "$results results\n" if ($debug == 2);

        return $results || 'error';
}

sub decode_qp($) {
        my $self = shift;

        my $tmp = shift || return;

        sub decode($$$) {
                my ($cp,$enc,$qp) = @_;

                print STDERR "decode($cp,$qp) -> " if ($debug == 2);

                if (uc($enc) eq "Q") {
                        $qp =~ s/=([a-f0-9][a-f0-9])/chr(hex($1))/ieg;
                        $qp =~ s/_/ /g;
                } elsif (uc($enc) eq "B") {
                        $qp = decode_base64($qp);
                } else {
                        croak "unsupported encoding '$enc' in decode_qp\n";
                        return $qp;
                }

                print STDERR "$qp\n" if ($debug == 2);

                my $iconv = Text::Iconv->new($cp,'ISO-8859-2');
                return $iconv->convert($qp) || '';
        }

        $tmp =~ s/=\?([^\?]+)\?([QB])\?(.+?)\?=/decode($1,$2,$3)/ige;
        $tmp =~ s/^\s*["']+(.*?)["']+\s*$/$1/g;
        #print STDERR "$tmp\n" if ($debug == 2);
        return $tmp;
}

sub unroll($$$) {
        my $self = shift;

        my ($message,$part,$sub) = @_;

        my @arr;

        return if (! $message->$part);

        foreach my $from ($message->$part) {
                my $tmp = $from->$sub || next;

                $tmp = $self->decode_qp($tmp);
                push @arr, $tmp;
        }

        return @arr;
}

sub fetch_all_results {
        my $self = shift;

        croak "results called before search!" if (! $self->{'index_ids'});

        print STDERR "fetch_all_results [",scalar @{$self->{'index_ids'}},"]\n" if ($debug == 2);

        my @arr;

        foreach my $id (@{$self->{'index_ids'}}) {
                push @arr, $self->fetch_result_by_id($id);
        }


        return @arr;
}

sub fetch_result {
        my $self = shift;

        my $args = {@_};

        croak "results called before search!" if (! $self->{'index_ids'});

        my $curr = $self->{'curr_result'}++;

        my $id = $self->{'index_ids'}->[$curr];
        
        print STDERR "fetch_result: $curr = $id\n" if ($debug == 2);

        return $self->fetch_result_by_id($id);
}

sub plain_text_body {
        my $self = shift;
        my $message = shift || croak "plain_text_body needs message!";

        my $body;

        if (! $message->isMultipart) {
                $body = $message->decoded->string;
        } else {
                foreach my $part ($message->parts) {
                        if ($part->body->mimeType eq 'text/plain') {
                                $body = $part->decoded->string;
                                last;
                        }
                }
        }

        if (! $body) {
                $body = "[plain/text body not found]" if ($debug == 2);
                print STDERR "plain/text body not found\n" if ($debug);
                return;
        }

        # reformat with Text::Autoformat
        my $wrap = $self->{wrap_margin};
        if ($wrap && $body && $body =~ m/^.{$wrap}..*$/m) {
                $body = autoformat($body, {right=>$wrap, all=>1});
                $body .="\n[reformated using autoformat, margin at $wrap]" if ($debug == 2);
        }

        return $body;
}


sub fetch_result_by_id {
        my $self = shift;

        my $id = shift || return;

        my $row = $self->{cache}->{$id};

        if (! $row) {

                print STDERR "fetch_result_by_id($id) not in cache, hitting disk\n" if ($debug == 2);

                my $message = $self->fetch_message($id) || return;

                $row->{'id'} = $id;

                foreach my $p (@counters) {
                        foreach my $v ($self->unroll($message,$p,'phrase')) {
                                push @{$row->{$p}},$v;
                                $self->add_counter($p,$v);
                        }
                }

                $self->add_counter('folder', $1) if ($id =~ m/^(\S+)\s/);

                $row->{'subject'} = $self->decode_qp($message->subject);
                $row->{'body'} = $self->plain_text_body($message);
                my $utime = str2time($message->date);

                $row->{'date_utime'} = $utime;

                $row->{'date'} = strftime("%Y-%m-%d %H:%M:%S", localtime($utime));
                $self->add_counter_calendar($utime);

                # XXX store in cache?
                $self->{cache}->{$id} = $row;
                print STDERR "$id stored in cache\n" if ($debug == 2);
        } else {
                print STDERR "fetch_result_by_id($id) in cache\n" if ($debug == 2);
                foreach my $p (@counters) {
                        foreach my $v (@{$row->{$p}}) {
                                $self->add_counter($p,$v);
                        }
                }

                $self->add_counter('folder', $1) if ($id =~ m/^(\S+)\s/);

                $self->add_counter_calendar($row->{date_utime});
        }

        return $row;
        
}
1	#!/usr/bin/perl -w
2
3	package MWS;
4
5	use strict;
6	use warnings;
7
8	use lib '.';
9
10	use Carp;
11	use Mail::Box::Manager;
12	use Config::IniFiles;
13	use POSIX qw(strftime);
14	use Text::Autoformat;
15	use Text::Iconv;
16	use Text::Unaccent;
17	use Date::Parse;
18	use POSIX qw(strftime);
19	use MIME::Base64;
20
21	our $VERSION = '1.00';
22
23	my $debug = 1;
24
25	my @counters = qw(from to cc bcc folder);
26
27	sub new {
28	my $class = shift;
29	my $self = {@_};
30	bless($self, $class);
31
32
33	my $config_name = $self->{config_file} \|\| croak "need config_file";
34	$config_name =~ s/\.conf.*$//;
35	$self->{config_name} = $config_name;
36
37	$self->{config} = new Config::IniFiles( -file => $self->{config_file} );
38
39	my $index_dir = $self->{config}->val('global', 'index') \|\| croak "can't find [index] section in config file with path of index";
40
41	$self->{mgr} = Mail::Box::Manager->new(access => 'r');
42	$self->{index_dir} = $index_dir;
43
44	# placeholder for opened folders
45	$self->{folder} = {};
46
47	$self->{wrap_margin} = $self->{config}->val('global', 'wrap_margin');
48	$self->{max_results} = $self->{config}->val('global', 'max_results') \|\| 100;
49	$self->reset_counters;
50
51	return $self;
52	}
53
54	sub normalize_string {
55	my $self = shift;
56
57	my $v = shift \|\| return;
58
59	$v = unac_string('ISO-8859-2', $v);
60	$v = join('',sort split(/\s+/,$v));
61	$v =~ s/\W+//g;
62
63	return lc($v);
64	}
65
66	# reset tables for search results
67	sub reset_counters {
68	my $self = shift;
69
70	$self->{counter} = {};
71
72	# foreach my $c (qw(thread from to cc bcc lists links att)) {
73	# $self->{counter}->{$c} = {};
74	# }
75
76	}
77
78	sub add_counter($$) {
79	my $self = shift;
80
81	my ($c,$v) = @_;
82	my $k = $self->normalize_string($v);
83
84	$self->{counter}->{$c}->{$k}->{name} = $v \|\| return;
85	return $self->{counter}->{$c}->{$k}->{usage}++;
86	}
87
88	sub yyyymmdd {
89	my $self = shift;
90
91	my $t = shift \|\| time;
92
93	my (undef,undef,undef,$dd,$mm,$yyyy) = localtime($t);
94	$mm++;
95	$yyyy+=1900;
96	return ($yyyy,$mm,$dd);
97	}
98
99	sub fmtdate {
100	my $self = shift;
101
102	my @out;
103	my @formats = qw(%04d %02d %02d);
104	while (my $v = shift) {
105	my $f = shift @formats;
106	push @out, sprintf($f, $v);
107	}
108
109	print STDERR "fmtdate: ",join('\|',@out),"\n" if ($debug == 2);
110
111	return (wantarray ? @out : join("-",@out));
112	}
113
114	sub add_counter_calendar($) {
115	my $self = shift;
116
117	my $t = shift \|\| croak "add_counter_calendar without argument!";
118
119	my ($yyyy,$mm,$dd) = $self->fmtdate($self->yyyymmdd($t));
120
121	return $self->{counter}->{calendar}->{"$yyyy-$mm"}->{$dd}++;
122	}
123
124
125	sub counter {
126	my $self = shift;
127
128	my $c = shift \|\| return;
129
130	return if (! $self->{counter}->{$c});
131
132	return $self->{counter}->{$c};
133	}
134
135	sub mbox_name2path {
136	my $self = shift;
137
138	my $mbox = shift \|\| croak "folder_name2path needs mbox name";
139
140	return $self->{config}->val('folders', $mbox) \|\| croak "comeone removed folder $mbox from config?";
141	}
142
143	sub open_folder {
144	my $self = shift;
145
146	my $mbox = shift \|\| croak "open_folder needs mbox name";
147
148	if (! $self->{folder}->{$mbox}) {
149	my $mbox_path = $self->mbox_name2path($mbox);
150
151	print STDERR "about to open_folder($mbox)\n" if ($debug == 2);
152
153	$self->{folder}->{$mbox} = $self->{mgr}->open($mbox_path) \|\| croak "can't open folder $mbox at '$mbox_path': $!";
154
155	print STDERR "open_folder($mbox) ok\n" if ($debug);
156	}
157
158	$self->{fetch_count} = 0;
159
160	return $self->{folder}->{$mbox};
161
162	}
163
164	sub close_folder {
165	my $self = shift;
166
167	my $mbox = shift \|\| croak "open_folder needs mbox name";
168
169	$self->{folder}->{$mbox}->close(write => 'NEVER') \|\| croak "can't close folder $mbox";
170
171	# XXX this is rather agressive!!!
172	$self->{folder} = {};
173	return
174	}
175
176	sub fetch_message {
177	my $self = shift;
178
179	my $mbox_id = shift \|\| die "need mbox_id!";
180	my ($mbox,$id) = split(/ /,$mbox_id);
181
182	# return message with ID
183	print STDERR "fetch $id from $mbox\n" if ($debug);
184
185	if ($self->{fetch_count}++ > 100) {
186	$self->close_folder($mbox);
187	print STDERR "close_folder($mbox) forced on ",$self->{fetch_count},"iteration\n";
188	}
189
190	my $msg = $self->open_folder($mbox)->find($id);
191	if ($msg) {
192	return $msg;
193	} else {
194	print STDERR "can't find message $id in $mbox. Time to re-index?\n";
195	return;
196	}
197	}
198
199
200	sub search {
201	my $self = shift;
202
203	carp "search called without argument!" if (! @_);
204
205	$self->reset_counters;
206
207	print STDERR "search(",join(" ",@_),")\n" if ($debug == 2);
208	my @index_ids = $self->search_index(@_);
209
210	$self->{'index_ids'} = \@index_ids;
211
212	#my $results = $#index_ids + 1;
213	#$self->{'results'} = $results;
214
215	my $results = $self->{'total_hits'} \|\| ($#index_ids + 1);
216
217	$self->{'curr_result'} = 0;
218
219	$self->reset_counters;
220
221	print STDERR "$results results\n" if ($debug == 2);
222
223	return $results \|\| 'error';
224	}
225
226	sub decode_qp($) {
227	my $self = shift;
228
229	my $tmp = shift \|\| return;
230
231	sub decode($$$) {
232	my ($cp,$enc,$qp) = @_;
233
234	print STDERR "decode($cp,$qp) -> " if ($debug == 2);
235
236	if (uc($enc) eq "Q") {
237	$qp =~ s/=([a-f0-9][a-f0-9])/chr(hex($1))/ieg;
238	$qp =~ s/_/ /g;
239	} elsif (uc($enc) eq "B") {
240	$qp = decode_base64($qp);
241	} else {
242	croak "unsupported encoding '$enc' in decode_qp\n";
243	return $qp;
244	}
245
246	print STDERR "$qp\n" if ($debug == 2);
247
248	my $iconv = Text::Iconv->new($cp,'ISO-8859-2');
249	return $iconv->convert($qp) \|\| '';
250	}
251
252	$tmp =~ s/=\?([^\?]+)\?([QB])\?(.+?)\?=/decode($1,$2,$3)/ige;
253	$tmp =~ s/^\s["']+(.?)["']+\s*$/$1/g;
254	#print STDERR "$tmp\n" if ($debug == 2);
255	return $tmp;
256	}
257
258	sub unroll($$$) {
259	my $self = shift;
260
261	my ($message,$part,$sub) = @_;
262
263	my @arr;
264
265	return if (! $message->$part);
266
267	foreach my $from ($message->$part) {
268	my $tmp = $from->$sub \|\| next;
269
270	$tmp = $self->decode_qp($tmp);
271	push @arr, $tmp;
272	}
273
274	return @arr;
275	}
276
277	sub fetch_all_results {
278	my $self = shift;
279
280	croak "results called before search!" if (! $self->{'index_ids'});
281
282	print STDERR "fetch_all_results [",scalar @{$self->{'index_ids'}},"]\n" if ($debug == 2);
283
284	my @arr;
285
286	foreach my $id (@{$self->{'index_ids'}}) {
287	push @arr, $self->fetch_result_by_id($id);
288	}
289
290
291	return @arr;
292	}
293
294	sub fetch_result {
295	my $self = shift;
296
297	my $args = {@_};
298
299	croak "results called before search!" if (! $self->{'index_ids'});
300
301	my $curr = $self->{'curr_result'}++;
302
303	my $id = $self->{'index_ids'}->[$curr];
304
305	print STDERR "fetch_result: $curr = $id\n" if ($debug == 2);
306
307	return $self->fetch_result_by_id($id);
308	}
309
310	sub plain_text_body {
311	my $self = shift;
312	my $message = shift \|\| croak "plain_text_body needs message!";
313
314	my $body;
315
316	if (! $message->isMultipart) {
317	$body = $message->decoded->string;
318	} else {
319	foreach my $part ($message->parts) {
320	if ($part->body->mimeType eq 'text/plain') {
321	$body = $part->decoded->string;
322	last;
323	}
324	}
325	}
326
327	if (! $body) {
328	$body = "[plain/text body not found]" if ($debug == 2);
329	print STDERR "plain/text body not found\n" if ($debug);
330	return;
331	}
332
333	# reformat with Text::Autoformat
334	my $wrap = $self->{wrap_margin};
335	if ($wrap && $body && $body =~ m/^.{$wrap}..*$/m) {
336	$body = autoformat($body, {right=>$wrap, all=>1});
337	$body .="\n[reformated using autoformat, margin at $wrap]" if ($debug == 2);
338	}
339
340	return $body;
341	}
342
343
344	sub fetch_result_by_id {
345	my $self = shift;
346
347	my $id = shift \|\| return;
348
349	my $row = $self->{cache}->{$id};
350
351	if (! $row) {
352
353	print STDERR "fetch_result_by_id($id) not in cache, hitting disk\n" if ($debug == 2);
354
355	my $message = $self->fetch_message($id) \|\| return;
356
357	$row->{'id'} = $id;
358
359	foreach my $p (@counters) {
360	foreach my $v ($self->unroll($message,$p,'phrase')) {
361	push @{$row->{$p}},$v;
362	$self->add_counter($p,$v);
363	}
364	}
365
366	$self->add_counter('folder', $1) if ($id =~ m/^(\S+)\s/);
367
368	$row->{'subject'} = $self->decode_qp($message->subject);
369	$row->{'body'} = $self->plain_text_body($message);
370	my $utime = str2time($message->date);
371
372	$row->{'date_utime'} = $utime;
373
374	$row->{'date'} = strftime("%Y-%m-%d %H:%M:%S", localtime($utime));
375	$self->add_counter_calendar($utime);
376
377	# XXX store in cache?
378	$self->{cache}->{$id} = $row;
379	print STDERR "$id stored in cache\n" if ($debug == 2);
380	} else {
381	print STDERR "fetch_result_by_id($id) in cache\n" if ($debug == 2);
382	foreach my $p (@counters) {
383	foreach my $v (@{$row->{$p}}) {
384	$self->add_counter($p,$v);
385	}
386	}
387
388	$self->add_counter('folder', $1) if ($id =~ m/^(\S+)\s/);
389
390	$self->add_counter_calendar($row->{date_utime});
391	}
392
393	return $row;
394
395	}