/[mws]/trunk/MWS.pm
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /trunk/MWS.pm

Parent Directory Parent Directory | Revision Log Revision Log


Revision 26 - (hide annotations)
Sat May 8 16:24:46 2004 UTC (20 years ago) by dpavlin
File size: 8029 byte(s)
small improvements

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     $v =~ s/\W+//g;
68    
69 dpavlin 23 return lc($v);
70 dpavlin 19 }
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 23 $self->reset_counters;
226    
227 dpavlin 13 print STDERR "$results results\n" if ($debug == 2);
228    
229 dpavlin 2 return $results || 'error';
230     }
231    
232 dpavlin 16 sub decode_qp($) {
233     my $self = shift;
234    
235     my $tmp = shift || return;
236    
237 dpavlin 21 sub decode($$$) {
238     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 $qp;
250     }
251    
252 dpavlin 22 print STDERR "$qp\n" if ($debug == 2);
253 dpavlin 21
254 dpavlin 20 my $iconv = Text::Iconv->new($cp,'ISO-8859-2');
255     return $iconv->convert($qp) || '';
256 dpavlin 16 }
257    
258 dpavlin 21 $tmp =~ s/=\?([^\?]+)\?([QB])\?(.+?)\?=/decode($1,$2,$3)/ige;
259 dpavlin 17 $tmp =~ s/^\s*["']+(.*?)["']+\s*$/$1/g;
260 dpavlin 20 #print STDERR "$tmp\n" if ($debug == 2);
261 dpavlin 16 return $tmp;
262     }
263    
264 dpavlin 2 sub unroll($$$) {
265 dpavlin 13 my $self = shift;
266    
267 dpavlin 2 my ($message,$part,$sub) = @_;
268    
269     my @arr;
270    
271 dpavlin 17 return if (! $message->$part);
272    
273 dpavlin 2 foreach my $from ($message->$part) {
274 dpavlin 16 my $tmp = $from->$sub || next;
275    
276     $tmp = $self->decode_qp($tmp);
277     push @arr, $tmp;
278 dpavlin 2 }
279 dpavlin 14
280     return @arr;
281 dpavlin 2 }
282 dpavlin 4
283     sub fetch_all_results {
284     my $self = shift;
285    
286     croak "results called before search!" if (! $self->{'index_ids'});
287    
288 dpavlin 14 print STDERR "fetch_all_results [",scalar @{$self->{'index_ids'}},"]\n" if ($debug == 2);
289 dpavlin 13
290 dpavlin 4 my @arr;
291    
292     foreach my $id (@{$self->{'index_ids'}}) {
293     push @arr, $self->fetch_result_by_id($id);
294     }
295    
296 dpavlin 19
297 dpavlin 4 return @arr;
298     }
299    
300 dpavlin 2 sub fetch_result {
301     my $self = shift;
302    
303 dpavlin 3 my $args = {@_};
304    
305 dpavlin 2 croak "results called before search!" if (! $self->{'index_ids'});
306    
307     my $curr = $self->{'curr_result'}++;
308    
309     my $id = $self->{'index_ids'}->[$curr];
310 dpavlin 14
311     print STDERR "fetch_result: $curr = $id\n" if ($debug == 2);
312 dpavlin 2
313 dpavlin 4 return $self->fetch_result_by_id($id);
314     }
315 dpavlin 2
316 dpavlin 7 sub plain_text_body {
317     my $self = shift;
318     my $message = shift || croak "plain_text_body needs message!";
319    
320 dpavlin 14 my $body;
321    
322 dpavlin 7 if (! $message->isMultipart) {
323 dpavlin 14 $body = $message->decoded->string;
324 dpavlin 7 } else {
325     foreach my $part ($message->parts) {
326     if ($part->body->mimeType eq 'text/plain') {
327 dpavlin 14 $body = $part->decoded->string;
328     last;
329 dpavlin 7 }
330     }
331     }
332 dpavlin 14
333 dpavlin 25 if (! $body) {
334 dpavlin 26 $body = "[plain/text body not found]" if ($debug == 2);
335     print STDERR "plain/text body not found\n" if ($debug);
336     return;
337 dpavlin 25 }
338    
339 dpavlin 14 # reformat with Text::Autoformat
340     my $wrap = $self->{wrap_margin};
341     if ($wrap && $body && $body =~ m/^.{$wrap}..*$/m) {
342 dpavlin 21 $body = autoformat($body, {right=>$wrap, all=>1});
343 dpavlin 14 $body .="\n[reformated using autoformat, margin at $wrap]" if ($debug == 2);
344     }
345    
346     return $body;
347 dpavlin 7 }
348    
349    
350 dpavlin 4 sub fetch_result_by_id {
351     my $self = shift;
352    
353     my $id = shift || return;
354    
355 dpavlin 13 my $row = $self->{cache}->{$id};
356 dpavlin 2
357 dpavlin 13 if (! $row) {
358 dpavlin 2
359 dpavlin 13 print STDERR "fetch_result_by_id($id) not in cache, hitting disk\n" if ($debug == 2);
360 dpavlin 2
361 dpavlin 17 my $message = $self->fetch_message($id) || return;
362 dpavlin 13
363     $row->{'id'} = $id;
364 dpavlin 19
365     foreach my $p (qw(from to cc bcc)) {
366 dpavlin 22 foreach my $v ($self->unroll($message,$p,'phrase')) {
367 dpavlin 19 push @{$row->{$p}},$v;
368     $self->add_counter($p,$v);
369     }
370     }
371 dpavlin 16 $row->{'subject'} = $self->decode_qp($message->subject);
372 dpavlin 13 $row->{'body'} = $self->plain_text_body($message);
373 dpavlin 20 my $utime = str2time($message->date);
374 dpavlin 13
375 dpavlin 20 $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 dpavlin 13 # XXX store in cache?
381     $self->{cache}->{$id} = $row;
382     print STDERR "$id stored in cache\n" if ($debug == 2);
383     } else {
384     print STDERR "fetch_result_by_id($id) in cache\n" if ($debug == 2);
385 dpavlin 23 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 dpavlin 13 }
393    
394 dpavlin 2 return $row;
395    
396     }

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26