Subversion Repositories LCARS

Rev

Rev 30 | Rev 33 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
8 PointedEar 1
#!/usr/bin/env perl
5 PointedEar 2
use strict;
6 PointedEar 3
use warnings;
23 PointedEar 4
require 5.004;
22 PointedEar 5
 
6
#use diagnostics;
6 PointedEar 7
use utf8;
8
 
30 PointedEar 9
## NOTE:
10
## Enable and remove binmode when utf8::all has actually become lexically scoped
11
# use utf8:all;
12
 
22 PointedEar 13
use constant DEBUG => 0;
14
 
24 PointedEar 15
## newsstat.pl
16
## Copyright (C) 2011, 2012  Thomas Lahn <startrek@PointedEars.de>
17
## Based on work by Garry Knight et al.
31 PointedEar 18
##
24 PointedEar 19
## This program is free software: you can redistribute it and/or modify
20
## it under the terms of the GNU General Public License as published by
21
## the Free Software Foundation, either version 3 of the License, or
22
## (at your option) any later version.
31 PointedEar 23
##
24 PointedEar 24
## This program is distributed in the hope that it will be useful,
25
## but WITHOUT ANY WARRANTY; without even the implied warranty of
26
## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
27
## GNU General Public License for more details.
31 PointedEar 28
##
24 PointedEar 29
## You should have received a copy of the GNU General Public License
30
## along with this program.  If not, see <http://www.gnu.org/licenses/>.
31
 
13 PointedEar 32
## Print out all text to STDOUT UTF-8 encoded
33
binmode STDOUT, ':encoding(UTF-8)';
22 PointedEar 34
binmode STDERR, ':encoding(UTF-8)';
5 PointedEar 35
 
23 PointedEar 36
## L10n
37
use locale ':not_characters';
38
 
39
# setlocale( LC_MESSAGES, '' );
40
require Number::Format;
41
 
42
## i18n
43
## FIXME: Automatically include resolved '.' in @INC
22 PointedEar 44
# print join "\n", @INC;
5 PointedEar 45
 
22 PointedEar 46
use Locale::TextDomain ('de.pointedears.newsstat');
23 PointedEar 47
use POSIX              ('locale_h');
22 PointedEar 48
use Locale::Messages qw (bind_textdomain_filter
23 PointedEar 49
  bind_textdomain_codeset
50
  turn_utf_8_on);
51
 
52
bind_textdomain_filter 'de.pointedears.newsstat',  \&turn_utf_8_on;
22 PointedEar 53
bind_textdomain_codeset 'de.pointedears.newsstat', 'utf-8';
23 PointedEar 54
 
22 PointedEar 55
require Mail::Message;
56
require DateTime;
57
require DateTime::Format::Mail;
23 PointedEar 58
 
59
# See comments in previous example
60
my ( $thousands_sep, $mon_thousands_sep, $grouping, $decimal_point ) =
61
  @{ localeconv() }{ 'thousands_sep', 'mon_thousands_sep', 'grouping',
62
  'decimal_point' };
63
 
64
# Apply defaults if values are missing
65
$thousands_sep = $mon_thousands_sep unless $thousands_sep;
66
$thousands_sep = ' ' unless $thousands_sep;
67
 
68
# grouping and mon_grouping are packed lists
69
# of small integers (characters) telling the
70
# grouping (thousand_seps and mon_thousand_seps
71
# being the group dividers) of numbers and
72
# monetary quantities.  The integers' meanings:
73
# 255 means no more grouping, 0 means repeat
74
# the previous grouping, 1-254 means use that
75
# as the current grouping.  Grouping goes from
76
# right to left (low to high digits).  In the
77
# below we cheat slightly by never using anything
78
# else than the first grouping (whatever that is).
79
my @grouping;
80
if ($grouping)
81
{
82
  @grouping = unpack( "C*", $grouping );
83
}
84
else
85
{
86
  @grouping = (3);
87
}
88
 
89
## FIXME: Why don't the defaults work already?
90
my $formatter = new Number::Format(
91
  -decimal_point => $decimal_point,
92
  -thousands_sep => $thousands_sep,
93
 
94
  # -grouping      => $grouping[0]
95
);
96
 
5 PointedEar 97
###################### USER CONFIGURATIONS ############################
98
 
13 PointedEar 99
## The name of the group to do stats for
5 PointedEar 100
my $newsgroup_name = $ARGV[0];
31 PointedEar 101
$newsgroup_name // usage();
5 PointedEar 102
 
13 PointedEar 103
## Check for removal flags
5 PointedEar 104
my $ix;
105
my $j;
106
my %skipSec;
107
my @skiplist;
108
my $args = @ARGV;
7 PointedEar 109
for ( $ix = 1 ; $ix < $args ; $ix++ )
110
{
111
  $j = $ix + 1;
112
  if ( $ARGV[$ix] eq "-x" )
113
  {
114
    @skiplist = split( ",", $ARGV[$j] );
115
  }
116
  elsif ( $ARGV[$ix] =~ /-x(\d.*)/ )
117
  {
118
    @skiplist = split( ",", $1 );
119
  }
5 PointedEar 120
}
7 PointedEar 121
foreach (@skiplist)
122
{
5 PointedEar 123
  $skipSec{$_} = 1;
124
}
125
 
13 PointedEar 126
## Leafnode users will want /var/spool/news for this variable.
5 PointedEar 127
my $news = "/var/spool/news/";
128
 
22 PointedEar 129
## Number of top or bottom posters to show
130
my $topposters = 20;
5 PointedEar 131
 
13 PointedEar 132
## Number of threads we want to know about
5 PointedEar 133
my $topthreads = 20;
134
 
13 PointedEar 135
## Number of cross-posted threads to show
5 PointedEar 136
my $topcrossposts = 10;
137
 
22 PointedEar 138
## Number of agents we list
139
my $topagents = 10;
140
 
13 PointedEar 141
## Number of time zones to show
5 PointedEar 142
my $toptz = 10;
143
 
144
###################### DATA STRUCTURES ######################
145
my $group = $newsgroup_name;
146
$group =~ s!\.!/!g;
22 PointedEar 147
my %data;    # name, count, agent, total, orig, quoted
148
my $totsize = 0;    # holds total sizes of all files
149
my %crossposts;     # group, count
150
my %threads;        # subject, count
151
my $replies   = 0;  # total no. of replies
152
my $origposts = 0;  # total no. of original posts
153
my %tz;             # timezones by count
154
my $earliest;       # earliest article we have found
155
my $latest;         # latest article we have found
156
my $totheader = 0;  # total size of header material
157
my $totbody   = 0;  # total size of body material
158
my $totsig    = 0;  # total size of sig material
159
my $totquoted = 0;  # total size of quoted material
160
my $totorig   = 0;  # total size of original material
161
my $totalposts;     # total no. of posts considered
5 PointedEar 162
my %distinct_agent;
13 PointedEar 163
 
164
## Used to hold counts of User Agents used
165
my %agents = (
166
  "Compuserver"               => 0,
167
  "Foorum"                    => 0,
7 PointedEar 168
  "Forte Agent"               => 0,
169
  "Forte Free Agent"          => 0,
13 PointedEar 170
  "Gnus"                      => 0,
171
  "KNode"                     => 0,
172
  "MacSOUP"                   => 0,
173
  "MT-NewsWatcher"            => 0,
7 PointedEar 174
  "MicroPlanet Gravity"       => 0,
175
  "Microsoft Outlook Express" => 0,
13 PointedEar 176
  "Microsoft Windows Mail"    => 0,
177
  "Mozilla"                   => 0,
178
  "News Rover"                => 0,
179
  "NN"                        => 0,
180
  "Pan"                       => 0,
181
  "rn"                        => 0,
7 PointedEar 182
  "slrn"                      => 0,
13 PointedEar 183
  "Sylpheed"                  => 0,
7 PointedEar 184
  "tin"                       => 0,
13 PointedEar 185
  "VSoup"                     => 0,
7 PointedEar 186
  "WebTV"                     => 0,
22 PointedEar 187
  "Xnews"                     => 0,
13 PointedEar 188
);
5 PointedEar 189
 
22 PointedEar 190
my $datetime_parser = DateTime::Format::Mail->new();
191
$datetime_parser->loose();
5 PointedEar 192
 
22 PointedEar 193
my $today = DateTime->today( time_zone => 'UTC' );
194
my $prev_month = $today->clone()->subtract( months => 1 )->set_day(1);
195
my $start      = int $prev_month->strftime('%s');
196
my $numdays    = int DateTime->last_day_of_month(
197
  year      => $prev_month->year(),
198
  month     => $prev_month->month(),
199
  time_zone => $prev_month->time_zone(),
200
)->day();
201
my $end = int $today->clone()->set_day(1)->strftime('%s');
202
 
203
dmsg( $start, " to ", $end ) if DEBUG;
204
 
205
chdir("$news$group")
206
  or die __x(
207
  "Can't cd to {newsgroup}: {error}\n",
208
  newsgroup => "$news$group",
209
  error     => $!
210
  );
211
opendir( DIR, "." )
212
  or die __x(
213
  "Can't open {newsgroup}: {error}\n",
214
  newsgroup => "$news$group",
215
  error     => $!
216
  );
217
 
218
while ( defined( my $filename = readdir(DIR) ) )
7 PointedEar 219
{
22 PointedEar 220
  next unless -f $filename;    # only want real files
221
  next if ( $filename eq ".overview" );    # real articles only
222
 
223
  get_article($filename);                  # read in the article
5 PointedEar 224
}
22 PointedEar 225
closedir(DIR);                             # finished with the directory
7 PointedEar 226
 
22 PointedEar 227
dmsg("\nearliest: $earliest\nlatest:   $latest") if DEBUG;
228
 
13 PointedEar 229
## Post-processing
22 PointedEar 230
count_agents();                            # count agents, collapsing versions
231
fix_percent();
5 PointedEar 232
 
22 PointedEar 233
write_data();
234
display_results();
5 PointedEar 235
 
13 PointedEar 236
########################################
22 PointedEar 237
## Get current article's header and body
13 PointedEar 238
########################################
22 PointedEar 239
sub get_article
7 PointedEar 240
{
22 PointedEar 241
  my $filename = shift;
5 PointedEar 242
 
22 PointedEar 243
  open( my $FILE, '<', $filename )
244
    or
245
    die __x( "Can't open {file}: {error}\n", file => $filename, error => $! );
246
  my $msg       = Mail::Message->read($FILE);
247
  my $timestamp = $msg->timestamp();
248
  my $date      = $msg->study('Date');
5 PointedEar 249
 
22 PointedEar 250
  ## Disregard article if timestamp is not in range
251
  dmsg($timestamp) if DEBUG;
31 PointedEar 252
  if ( $timestamp < $start || $timestamp >= $end )
7 PointedEar 253
  {
22 PointedEar 254
    dmsg("Posting on $date ignored.") if DEBUG;
255
    return;
7 PointedEar 256
  }
5 PointedEar 257
 
22 PointedEar 258
  $totalposts++;    # bump count of articles considered
5 PointedEar 259
 
22 PointedEar 260
  ## DEBUG
261
  dmsg($date) if DEBUG;
13 PointedEar 262
 
22 PointedEar 263
  ## get stats about the file itself
264
  my $filesize = -s $filename;    # get total size of file
265
  $totsize += $filesize;          # bump total sizes of all files
5 PointedEar 266
 
31 PointedEar 267
  if ( ( not defined $earliest ) || $timestamp < $earliest )
7 PointedEar 268
  {
22 PointedEar 269
    $earliest = $timestamp;
7 PointedEar 270
  }
31 PointedEar 271
  elsif ( ( not defined $latest ) || $timestamp > $latest )
7 PointedEar 272
  {
22 PointedEar 273
    $latest = $timestamp;
7 PointedEar 274
  }
13 PointedEar 275
 
22 PointedEar 276
  #print "timestamp: $timestamp\n";
5 PointedEar 277
 
22 PointedEar 278
  ## count header size
279
  $totheader += $msg->head()->size();
5 PointedEar 280
 
22 PointedEar 281
  ## get the poster's name (MIME-decoded, in UTF-8)
282
  my $poster = $msg->study('From');
283
  if ( defined $poster )
7 PointedEar 284
  {
22 PointedEar 285
    ## Convert old to new format
286
    $poster =~ s/^\s*(.+?\@.+?)\s*\((.+?)\)\s*$/$2 <$1>/;
5 PointedEar 287
 
22 PointedEar 288
    ## Collapse whitespace
289
    $poster =~ s/\s+/ /g;
5 PointedEar 290
 
22 PointedEar 291
    ## Remove outer quotes; TODO: observe RFC 5322 strictly
292
    $poster =~ s/^ " (.+ ) " \s+ (.*)/$1 $2/x;
5 PointedEar 293
 
22 PointedEar 294
    ## DEBUG
295
    dmsg($poster) if DEBUG;
7 PointedEar 296
 
22 PointedEar 297
    ## seen this one before?
298
    if ( !defined( $data{$poster} ) )
299
    {
300
      $data{$poster}{'agent'}  = __ 'unknown';    # comes after For: field
301
      $data{$poster}{'orig'}   = 0;
302
      $data{$poster}{'quoted'} = 0;
303
    }
304
    $data{$poster}{'count'}++;                    # bump count for this poster
305
    $data{$poster}{'size'} += $filesize;          # total size of file
5 PointedEar 306
 
22 PointedEar 307
    ## The User-Agent and/or X-Newsreader fields
308
    ## for User-Agent by poster
31 PointedEar 309
    my $ua = $msg->study('User-Agent') // $msg->study('X-Newsreader');
22 PointedEar 310
    if ( defined $ua )
7 PointedEar 311
    {
22 PointedEar 312
      $data{$poster}{'agent'} = $ua;
313
 
314
      ## DEBUG
315
      dmsg($ua) if DEBUG;
7 PointedEar 316
    }
5 PointedEar 317
 
22 PointedEar 318
    ## The User Agent for User-Agent by number of posts
319
    get_agent($msg);
5 PointedEar 320
 
22 PointedEar 321
    ## Get all cross-posted newsgroups
322
    for ( split( /,/, $msg->study('Newsgroups') ) )
323
    {
324
      $crossposts{$_}++;    # bump count for each
325
    }
5 PointedEar 326
 
22 PointedEar 327
    ## Get threads
328
    my $thread = $msg->study('Subject');
329
    $thread =~ s/^re:\s+//i;    # Remove Re: or re: at start
330
    $thread =~ s/\s+/ /g;       # collapse whitespace
331
    $threads{$thread}{'count'}++;    # bump count of this subject
332
    $threads{$thread}{'size'} += $filesize;    # bump bytes for this thread
5 PointedEar 333
 
22 PointedEar 334
    ## Is this an original post or a reply?
335
    if ( defined $msg->study('References') )
7 PointedEar 336
    {
22 PointedEar 337
      $replies++;
7 PointedEar 338
    }
22 PointedEar 339
    else
7 PointedEar 340
    {
22 PointedEar 341
      $origposts++;
7 PointedEar 342
    }
22 PointedEar 343
 
344
    ## Get the time zone
345
    my $datetime = $datetime_parser->parse_datetime($date);
346
    my $tz       = $datetime->strftime('%z');
347
    $tz = "UTC" if $tz =~ m{^(?:GMT|0000)$}o;
348
    $tz{$tz}++;
349
 
350
    ## DEBUG
351
    dmsg($tz) if DEBUG;
352
 
353
#### Now analyse the body text ####
354
    my $body = $msg->body();
355
 
356
    my $insig = 0;
357
    my @body  = $body->lines;
358
    for (@body)
7 PointedEar 359
    {
22 PointedEar 360
      $totbody += length($_);    # bump total body size
361
      next if (m{^$>}o);         # don't count blank lines in body
362
      if ( $insig == 1 )
363
      {
364
 
365
        # bump total sig size
366
        $totsig += length($_);
367
      }
368
      ## are we in a quote line?
369
      ## Bill Unruh uses ] quotes, and another poster uses ::
31 PointedEar 370
      elsif ( m{^\s*[>\]]}o || m{^\s*::}o )
22 PointedEar 371
      {
372
        ## bump count of quoted chrs
373
        $data{$poster}{'quoted'} += length($_);
374
        $totquoted += length($_);
375
      }
376
      elsif (/^-- $/)
377
      {
378
        $insig = 1;
379
      }
380
      else
381
      {
382
        ## We must be processing an original line
383
        $data{$poster}{'orig'} += length($_);    # bump count of original chrs
384
        $totorig += length($_);
385
      }
7 PointedEar 386
    }
22 PointedEar 387
 
388
    # end for (@body)
389
  }
390
 
391
  close($FILE);
392
}
393
 
394
sub get_agent
395
{
396
  my $msg = shift;
397
 
31 PointedEar 398
  my $ua = $msg->study('User-Agent') // $msg->study('X-Newsreader')
399
    // $msg->study('X-Mailer');
400
 
22 PointedEar 401
  if ( not defined $ua )
402
  {
403
    my $org = $msg->study('Organization');
404
    if ( defined $org
405
      and $org =~ /groups\.google|AOL|Supernews|WebTV|compuserve/ )
7 PointedEar 406
    {
22 PointedEar 407
      $ua = $org;
7 PointedEar 408
    }
22 PointedEar 409
    elsif ( $msg->study('Message-ID') =~ /pine/i )
7 PointedEar 410
    {
22 PointedEar 411
      $ua = "Pine";
412
    }
7 PointedEar 413
  }
5 PointedEar 414
 
22 PointedEar 415
  ## Hopefully found UA, else set to unknown
416
  if ( not defined $ua )
417
  {
418
    $ua = __ "unknown";
419
  }
5 PointedEar 420
 
22 PointedEar 421
  $ua = clean($ua);
422
 
423
  my $raw   = $ua;
424
  my $agent = $raw;
425
 
426
  ## strip http
427
  if ( $raw =~ /.*http.*/ )
7 PointedEar 428
  {
22 PointedEar 429
    $raw =~ s!posted via!!i;
430
    $raw =~ s!http://!!g;
431
    $raw =~ s!/!!g;
432
    $raw =~ s! !!g;
433
  }
5 PointedEar 434
 
22 PointedEar 435
  ## Fix Outlook from Mac
436
  if ( $raw =~ /^microsoft/i )
437
  {
438
    $raw =~ s/-/ /g;
439
  }
5 PointedEar 440
 
22 PointedEar 441
  ## Pick out the popular agents
442
  if ( $raw =~ /(outlook express)/i
443
    || $raw =~ /(windows mail)/i
444
    || $raw =~ /(microplanet gravity)/i
445
    || $raw =~ /(news rover)/i
446
    || $raw =~ /(forte agent)/i
447
    || $raw =~ /(forte free agent)/i )
448
  {
449
    $agent = $1;
450
  }
451
  elsif (
452
    $raw =~ /^(
5 PointedEar 453
        pan
454
       |sylpheed
455
       |slrn
456
       |mozilla
457
       |knode
458
       |tin
459
       |hamster
460
       |xrn
461
       |xnews
462
       |aol
463
       |gnus
464
       |krn
465
       |macsoup
466
       |messenger
467
       |openxp
468
       |pine
469
       |thoth
470
       |turnpike
471
       |winvn
472
       |vsoup
473
       |google
474
       |supernews
475
       |nn
476
       |rn
477
       |007
478
       |webtv
479
       |compuserve
7 PointedEar 480
       )/ix
22 PointedEar 481
    )
482
  {
483
    $agent = $1;
484
  }
485
  else
486
  {
487
    ## Clean up unknown agents
488
    if ( $raw =~ m!^(.*?)/! )
7 PointedEar 489
    {
490
      $agent = $1;
491
    }
22 PointedEar 492
    elsif ( $raw =~ /^(\w*)\d.*/ )
7 PointedEar 493
    {
22 PointedEar 494
      $agent = $1;
495
    }
496
  }
497
 
498
  $distinct_agent{$agent}++;
499
  return $agent;
500
}
501
## get_agent
502
 
503
#########################################
504
## Count the User-Agents used, collapsing
505
## different versions into one per agent.
506
#########################################
507
sub count_agents
508
{
509
POSTER:
510
  foreach my $poster ( keys %data )
511
  {
512
    foreach my $agent_name ( keys %distinct_agent )
513
    {    # check against known ones
514
      if ( $data{$poster}{'agent'} =~ /\Q$agent_name\E/ )
7 PointedEar 515
      {
22 PointedEar 516
        $agents{$agent_name}++;
517
        next POSTER;
7 PointedEar 518
      }
519
    }
22 PointedEar 520
    $agents{ $data{$poster}{'agent'} }++;
521
  }
522
}    # count_agents
7 PointedEar 523
 
22 PointedEar 524
#############################################
525
## Set orig/total percentages for all posters
526
#############################################
527
sub fix_percent
528
{
529
  foreach my $poster ( keys %data )
530
  {
531
    my $percent = 100;
532
    if ( ( $data{$poster}{'orig'} != 0 ) and ( $data{$poster}{'quoted'} != 0 ) )
533
    {
534
      $percent =
535
        $data{$poster}{'orig'} * 100 /
536
        ( $data{$poster}{'quoted'} + $data{$poster}{'orig'} );    #/
537
    }
538
    elsif ( $data{$poster}{'orig'} == 0 )
539
    {
540
      $percent = 0;
541
    }
542
    $data{$poster}{'percent'} = $percent;
5 PointedEar 543
  }
22 PointedEar 544
}
545
## fix_percent
5 PointedEar 546
 
22 PointedEar 547
##################################
548
## Write data structures to a file
549
##################################
550
sub write_data
551
{
552
  open( my $OUTF, ">:encoding(UTF-8)", "/tmp/XDATA" )
553
    or die __x( "Can't create XDATA: {error}\n", error => $! );
554
  print $OUTF "Data collected from $newsgroup_name\n\n";
555
  print $OUTF
556
    "Poster Data\nname : agent : count : size: orig : quoted : per cent\n";
557
  foreach my $name ( keys %data )
7 PointedEar 558
  {
22 PointedEar 559
    print $OUTF
560
"$name : $data{$name}{'agent'} : $data{$name}{'count'} : $data{$name}{'size'} : $data{$name}{'orig'} : $data{$name}{'quoted'} : $data{$name}{'percent'}\n";
7 PointedEar 561
  }
22 PointedEar 562
  print $OUTF
563
"============================================================================\n";
564
  print $OUTF "Thread subjects\n";
565
  print $OUTF
566
"----------------------------------------------------------------------------\n";
567
  foreach my $thread ( sort { "\L$a" cmp "\L$b" } keys %threads )
568
  {
569
    print $OUTF
570
      "$thread : $threads{$thread}{'count'} : $threads{$thread}{'size'}\n";
571
  }
572
  print $OUTF
573
"============================================================================\n";
574
  print $OUTF "Cross-posts\n";
575
  print $OUTF
576
"----------------------------------------------------------------------------\n";
577
  foreach my $name ( sort keys %crossposts )
578
  {
579
    print $OUTF "$name : $crossposts{$name}\n";
580
  }
581
  print $OUTF
582
"============================================================================\n";
583
  print $OUTF "User agents\n";
584
  print $OUTF
585
"----------------------------------------------------------------------------\n";
586
  foreach my $name ( sort keys %agents )
587
  {
588
    print $OUTF "$name : $agents{$name}\n";
589
  }
590
  print $OUTF
591
"============================================================================\n";
592
  print $OUTF "Time zones\n";
593
  print $OUTF
594
"----------------------------------------------------------------------------\n";
595
  foreach my $name ( sort keys %tz )
596
  {
597
    print $OUTF "$name : $tz{$name}\n";
598
  }
599
  close $OUTF;
600
}    # write_data
5 PointedEar 601
 
22 PointedEar 602
sub display_results
603
{
604
  #################### DISPLAY RESULTS #####################
23 PointedEar 605
  println( "=" x 76 );
22 PointedEar 606
  printf "%s\n",
607
    centred(
608
    __x( "Analysis of posts to {newsgroup}", newsgroup => $newsgroup_name ),
609
    76 );
23 PointedEar 610
  println( "=" x 76 );
22 PointedEar 611
  printf "%s\n",
612
    centred(
23 PointedEar 613
    __(
614
"(compiled with a script by Thomas 'PointedEars' Lahn, based on work by\nGarry Knight et al.)"
615
    ),
22 PointedEar 616
    76
617
    );
618
  print "\n\n";
23 PointedEar 619
  printf __"Total posts considered: %s over %d days\n",
620
    $formatter->format_number($totalposts),
621
    $formatter->format_number($numdays);
622
  my $time_locale       = setlocale(LC_TIME);
22 PointedEar 623
  my $earliest_datetime = DateTime->from_epoch(
23 PointedEar 624
    epoch     => $earliest,
625
    locale    => $time_locale,
22 PointedEar 626
    time_zone => 'UTC',
627
  );
628
  my $latest_datetime = DateTime->from_epoch(
23 PointedEar 629
    epoch     => $latest,
630
    locale    => $time_locale,
22 PointedEar 631
    time_zone => 'UTC',
632
  );
633
  my $datetime_format = '%a, %Y-%m-%dT%H:%M:%S %Z';
23 PointedEar 634
  printf __"Earliest article: %s\n",
635
    $earliest_datetime->strftime($datetime_format);
636
  printf __"Latest article:   %s\n",
637
    $latest_datetime->strftime($datetime_format);
638
  printf __"Original articles: %s; replies: %s\n",
639
    $formatter->format_number($origposts),
640
    $formatter->format_number($replies);
641
  printf __"Total size of posts: %s bytes (%s)" . "\n",
642
    $formatter->format_number($totsize),
643
    $formatter->format_bytes( $totsize, ( 'precision' => 1, 'mode' => 'iec' ) );
644
  printf __"Average %s articles per day, %s per day, %s bytes per article\n",
645
    $formatter->format_number( int( $totalposts / $numdays ) ),
646
    $formatter->format_bytes( $totsize / $numdays, ( 'mode' => 'iec' ) ),
647
    $formatter->format_number( int( $totsize / $totalposts ) );
648
 
22 PointedEar 649
  my $count = keys %data;
23 PointedEar 650
  printf __"Total headers: %s; bodies: %s\n",
651
    $formatter->format_bytes(
652
    $totheader, ( 'precision' => 1, 'mode' => 'iec' )
653
    ),
654
    $formatter->format_bytes( $totbody, ( 'precision' => 1, 'mode' => 'iec' ) );
22 PointedEar 655
  printf __
23 PointedEar 656
    "Body text - quoted: %s; original: %s = %s%%; sigs: %s\n",
657
    $formatter->format_bytes(
658
    $totquoted, ( 'precision' => 1, 'mode' => 'iec' )
659
    ),
660
    $formatter->format_bytes( $totorig, ( 'precision' => 1, 'mode' => 'iec' ) ),
661
    $formatter->format_number( ( $totorig * 100 ) / ( $totorig + $totquoted ) ),
662
    $formatter->format_bytes( $totsig, ( 'precision' => 1, 'mode' => 'iec' ) );
663
  printf __"Total number of posters: %s, average %s per poster\n",
664
    $formatter->format_number($count),
665
    $formatter->format_bytes( $totsize / $count,
666
    ( 'precision' => 1, 'mode' => 'iec' ) );
22 PointedEar 667
  $count = keys %threads;
23 PointedEar 668
  printf __"Total number of threads: %s, average %s per thread\n",
669
    $formatter->format_number($count),
670
    $formatter->format_bytes( $totsize / $count,
671
    ( 'precision' => 1, 'mode' => 'iec' ) );
672
  printf __"Total number of user agents: %d\n",
673
    $formatter->format_number( scalar keys %agents );
31 PointedEar 674
  print "\n", "=" x 76, "\n";
22 PointedEar 675
  ########################################
676
  ## Show posters by article count  Sec 1;
677
  ########################################
678
  unless ( $skipSec{1} )
7 PointedEar 679
  {
22 PointedEar 680
    if ( keys %data < $topposters )
681
    {
682
      $count = keys %data;
683
    }
684
    else
685
    {
686
      $count = $topposters;
687
    }
688
    printf "%s\n",
689
      centred(
23 PointedEar 690
      __x( "Top {count} posters by number of articles", count => $topposters ),
691
      76
692
      );
22 PointedEar 693
    print "=" x 76, "\n";
694
    my $i = 0;
695
    foreach
696
      my $poster ( sort { $data{$b}{count} <=> $data{$a}{count} } keys %data )
697
    {
698
      my $name = substr( $poster, 0, 65 );
25 PointedEar 699
      printf "%2d. %-63s : %6d\n", $i + 1, rpad( $poster, 63, "." ),
22 PointedEar 700
        $data{$poster}{count};
701
      last if ( ++$i == $count );
702
    }
703
    print "\n", "=" x 76, "\n";
7 PointedEar 704
  }
22 PointedEar 705
 
706
  ######################################
707
  ## Show posters by size in KiB  Sec 2;
708
  ######################################
709
  unless ( $skipSec{2} )
7 PointedEar 710
  {
22 PointedEar 711
    if ( keys %data < $topposters )
712
    {
713
      $count = keys %data;
714
    }
715
    else
716
    {
717
      $count = $topposters;
718
    }
719
    printf "%s\n",
720
      centred(
721
      __x( "Top {count} posters by article size in KiB", count => $topposters ),
23 PointedEar 722
      76
723
      );
22 PointedEar 724
    print "=" x 76, "\n";
725
    my $i = 0;
726
    foreach
727
      my $poster ( sort { $data{$b}{size} <=> $data{$a}{size} } keys %data )
728
    {
729
      my $name = substr( $poster, 0, 62 );
25 PointedEar 730
      printf "%2d. %-63s : %6d\n", $i + 1, rpad( $poster, 63, "." ),
22 PointedEar 731
        $data{$poster}{size} / 1024;    #/
732
      last if ( ++$i == $count );
733
    }
734
    print "\n", "=" x 76, "\n";
7 PointedEar 735
  }
5 PointedEar 736
 
22 PointedEar 737
  #####################################
738
  ## Show top posters for original text
739
  #####################################
31 PointedEar 740
  my $topposters_real = 0;
741
 
22 PointedEar 742
  unless ( $skipSec{3} )
7 PointedEar 743
  {
22 PointedEar 744
    if ( keys %data < $topposters )
745
    {
746
      $count = keys %data;
747
    }
748
    else
749
    {
750
      $count = $topposters;
751
    }
31 PointedEar 752
 
22 PointedEar 753
    printf "%s\n",
754
      centred(
755
      __x(
756
        "Top {count} responders by original text (> 5 posts)",
757
        count => $topposters
758
      ),
759
      76
760
      );
761
    print "=" x 76, "\n";
762
    foreach my $poster (
763
      sort { $data{$b}{percent} <=> $data{$a}{percent} }
764
      keys %data
765
      )
766
    {
767
      next if $data{$poster}{quoted} == 0;
768
      next if $data{$poster}{count} < 5;
769
      my $name = substr( $poster, 0, 63 );
31 PointedEar 770
      printf "%2d. %-63s : %02.2f%%\n", $topposters_real + 1,
771
        rpad( $poster, 63, "." ),
22 PointedEar 772
        $data{$poster}{percent};
31 PointedEar 773
      last if ( ++$topposters_real == $count );
22 PointedEar 774
    }
775
    print "\n", "=" x 76, "\n";
7 PointedEar 776
  }
5 PointedEar 777
 
22 PointedEar 778
  ########################################
779
  ## Show bottom posters for original text
780
  ########################################
31 PointedEar 781
 
782
  $skipSec{4} = ( $topposters_real <= $topposters ) unless defined $skipSec{4};
783
 
22 PointedEar 784
  unless ( $skipSec{4} )
7 PointedEar 785
  {
22 PointedEar 786
    if ( keys %data < $topposters )
7 PointedEar 787
    {
22 PointedEar 788
      $count = keys %data;
789
    }
790
    else
791
    {
792
      $count = $topposters;
793
    }
31 PointedEar 794
 
22 PointedEar 795
    printf "%s\n",
796
      centred(
797
      __x(
798
        "Bottom {count} responders by original text  (> 5 posts)",
799
        count => $topposters
800
      ),
801
      76
802
      );
803
    print "=" x 76, "\n";
804
    my $i = 0;
805
    foreach my $poster (
806
      sort { $data{$a}{percent} <=> $data{$b}{percent} }
807
      keys %data
808
      )
809
    {
810
      next if $data{$poster}{quoted} == 0;
811
      next if $data{$poster}{count} < 5;
812
      my $name = substr( $poster, 0, 63 );
25 PointedEar 813
      printf "%2d. %-63s : %02.2f%%\n", $i + 1, rpad( $poster, 63, "." ),
22 PointedEar 814
        $data{$poster}{percent};
815
      last if ( ++$i == $count );
816
    }
817
    print "\n", "=" x 76, "\n";
818
  }
5 PointedEar 819
 
22 PointedEar 820
  #####################################
821
  ## Show threads by number of articles
822
  #####################################
823
  unless ( $skipSec{5} )
824
  {
825
    if ( keys %threads < $topthreads )
826
    {
827
      $count = keys %threads;
7 PointedEar 828
    }
22 PointedEar 829
    else
830
    {
831
      $count = $topthreads;
7 PointedEar 832
    }
22 PointedEar 833
    printf "%s\n",
23 PointedEar 834
      centred(
835
      __x( "Top {count} threads by no. of articles", count => $topthreads ),
22 PointedEar 836
      76 );
837
    print "=" x 76, "\n";
838
    my $i = 0;
839
    foreach my $thread (
840
      sort { $threads{$b}{'count'} <=> $threads{$a}{'count'} }
841
      keys %threads
842
      )
7 PointedEar 843
    {
22 PointedEar 844
      my $name = substr( $thread, 0, 65 );
25 PointedEar 845
      printf "%2d. %-63s : %6d\n", $i + 1, rpad( $name, 63, "." ),
22 PointedEar 846
        $threads{$thread}{'count'};
847
      last if ( ++$i == $count );
7 PointedEar 848
    }
22 PointedEar 849
    print "\n", "=" x 76, "\n";
850
  }
851
 
852
  ##############################
853
  ## Show threads by size in KiB
854
  ##############################
855
  unless ( $skipSec{6} )
856
  {
857
    if ( keys %threads < $topthreads )
858
    {
859
      $count = keys %threads;
860
    }
7 PointedEar 861
    else
862
    {
22 PointedEar 863
      $count = $topthreads;
864
    }
865
    printf "%s\n",
23 PointedEar 866
      centred(
867
      __x( "Top {count} threads by size in KiB", count => $topthreads ), 76 );
22 PointedEar 868
    print "=" x 76, "\n";
869
    my $i = 0;
870
    foreach my $thread (
871
      sort { $threads{$b}{'size'} <=> $threads{$a}{'size'} }
872
      keys %threads
873
      )
874
    {
875
      my $name = substr( $thread, 0, 65 );
25 PointedEar 876
      printf "%2d. %-63s : %6d\n", $i + 1, rpad( $name, 63, "." ),
22 PointedEar 877
        $threads{$thread}{'size'} / 1024;    #/
878
      last if ( ++$i == $count );
879
    }
880
    print "\n", "=" x 76, "\n";
881
  }
5 PointedEar 882
 
22 PointedEar 883
  ##################################
884
  ## Show top 10 cross-posted groups
885
  ##################################
886
  unless ( $skipSec{7} )
887
  {
888
    delete $crossposts{"$newsgroup_name"};    # don't include ours
889
    if ( keys %crossposts < $topcrossposts )
890
    {
891
      $count = keys %crossposts;
7 PointedEar 892
    }
22 PointedEar 893
    else
894
    {
895
      $count = $topcrossposts;
896
    }
897
    printf "%s\n",
23 PointedEar 898
      centred(
899
      __x( "Top {count} cross-posted groups", count => $topcrossposts ), 76 );
22 PointedEar 900
    print "=" x 76, "\n";
901
    my $i = 0;
902
    foreach
903
      my $name ( sort { $crossposts{$b} <=> $crossposts{$a} } keys %crossposts )
904
    {
25 PointedEar 905
      printf "%2d. %-63s : %6d\n", $i + 1, rpad( $name, 63, "." ),
22 PointedEar 906
        $crossposts{$name};
907
      last if ( ++$i == $count );
908
    }
909
    print "\n", "=" x 76, "\n";
910
  }
5 PointedEar 911
 
22 PointedEar 912
  #########################
913
  ## Show agents and counts
914
  #########################
915
  unless ( $skipSec{8} )
916
  {
917
    if ( keys %agents < $topagents )
918
    {
919
      $count = keys %agents;
920
    }
921
    else
922
    {
923
      $count = $topagents;
924
    }
925
    printf "%s\n",
926
      centred( __x( "Top {count} user agents by poster", count => $topagents ),
927
      76 );
928
    print "=" x 76, "\n";
929
    my $i = 0;
930
    foreach my $agent ( sort { $agents{$b} <=> $agents{$a} } keys %agents )
931
    {
25 PointedEar 932
      printf "%2d. %-63s : %6d\n", $i + 1, rpad( $agent, 63, "." ),
22 PointedEar 933
        $agents{$agent};
934
      last if ( ++$i == $count );
935
    }
936
    print "\n", "=" x 76, "\n";
937
  }
5 PointedEar 938
 
22 PointedEar 939
  #######################
940
  ## Show distinct agents
941
  #######################
942
  unless ( $skipSec{9} )
7 PointedEar 943
  {
22 PointedEar 944
    if ( keys %distinct_agent < $topagents )
945
    {
946
      $count = keys %distinct_agent;
7 PointedEar 947
    }
22 PointedEar 948
    else
949
    {
950
      $count = $topagents;
951
    }
952
    printf "%s\n",
953
      centred(
954
      __x( "Top {count} user agents by number of posts", count => $topagents ),
23 PointedEar 955
      76
956
      );
22 PointedEar 957
    print "=" x 76, "\n";
958
    my $i = 0;
959
    foreach my $agent (
960
      sort { $distinct_agent{$b} <=> $distinct_agent{$a} }
961
      keys %distinct_agent
962
      )
963
    {
25 PointedEar 964
      printf "%2d. %-58s : %5d (%2.f%%)\n", $i + 1, rpad( $agent, 58, "." ),
22 PointedEar 965
        $distinct_agent{$agent},
966
        ( ( $distinct_agent{$agent} / $totalposts ) * 100 );
967
      last if ( ++$i == $count );
968
    }
969
    print "\n", "=" x 76, "\n";
7 PointedEar 970
  }
5 PointedEar 971
 
22 PointedEar 972
  ############################
973
  ## Show timezones and counts
974
  ############################
975
  unless ( $skipSec{10} )
7 PointedEar 976
  {
22 PointedEar 977
    if ( keys %tz < $toptz )
7 PointedEar 978
    {
22 PointedEar 979
      $count = keys %tz;
7 PointedEar 980
    }
22 PointedEar 981
    else
7 PointedEar 982
    {
22 PointedEar 983
      $count = $toptz;
7 PointedEar 984
    }
23 PointedEar 985
    printf "%s\n",
986
      centred( __x( "Top {count} time zones", count => $toptz ), 76 );
22 PointedEar 987
    print "=" x 76, "\n";
988
    my $i = 0;
989
    foreach my $zone ( sort { $tz{$b} <=> $tz{$a} } keys %tz )
990
    {
25 PointedEar 991
      printf "%2d. %-63s : %6d\n", $i + 1, rpad( $zone, 63, "." ), $tz{$zone};
22 PointedEar 992
      last if ( ++$i == $count );
993
    }
994
    print "\n", "=" x 76, "\n";
7 PointedEar 995
  }
5 PointedEar 996
}
997
 
22 PointedEar 998
## helper subs
999
 
13 PointedEar 1000
###############################
1001
## Right pad a string with '.'s
1002
###############################
7 PointedEar 1003
sub rpad
1004
{
13 PointedEar 1005
  ## Get text to pad, length to pad, pad chr
7 PointedEar 1006
  my ( $text, $pad_len, $pad_chr ) = @_;
11 PointedEar 1007
 
13 PointedEar 1008
  ## DEBUG
22 PointedEar 1009
  printf( "|%s| = %d\n", $text, length($text) ) if DEBUG > 1;
11 PointedEar 1010
 
7 PointedEar 1011
  if ( length($text) > $pad_len )
1012
  {
1013
    $text = substr( $text, 0, $pad_len );
1014
  }
1015
  my $padded = $text . $pad_chr x ( $pad_len - length($text) );
1016
  return $padded;
5 PointedEar 1017
}
1018
 
13 PointedEar 1019
##################
1020
## Centre a string
1021
##################
7 PointedEar 1022
sub centred
1023
{
1024
  my ( $text, $width ) = @_;    # text to centre, size of field to centre in
1025
  my $pad_len = ( $width - length($text) ) / 2;    #/
1026
  my $centred = " " x $pad_len . $text;
1027
  return $centred;
5 PointedEar 1028
}
1029
 
13 PointedEar 1030
###########################
1031
## Put commas into a number
1032
###########################
7 PointedEar 1033
sub commify
1034
{
13 PointedEar 1035
  local $_ = shift;
23 PointedEar 1036
  my $number = $_;
1037
  $_ = int;                                        # Chop non-integer part
22 PointedEar 1038
  1 while
23 PointedEar 1039
    s/([-+]?\d)(\d{$grouping[0]}($|\Q$thousands_sep\E))/$1$thousands_sep$2/;
1040
  my $int_part  = $_;
1041
  my $real_part = '';
1042
  if ( $number =~ /(\Q$decimal_point\E\d+)$/ )
1043
  {
1044
    $real_part = $1;
1045
  }
1046
  return $int_part . $real_part;
5 PointedEar 1047
}
1048
 
13 PointedEar 1049
################################################################
1050
## Returns a string with leading and trailing whitespace removed
1051
################################################################
7 PointedEar 1052
sub clean
1053
{
1054
  my $dirty = shift;
1055
  my $clean = $dirty;
14 PointedEar 1056
  $clean =~ s/^\s+|\s+$//g;
5 PointedEar 1057
 
7 PointedEar 1058
  return $clean;
5 PointedEar 1059
}
1060
 
7 PointedEar 1061
sub usage
1062
{
23 PointedEar 1063
  println( __ "usage: newsstat.pl NEWS.GROUP" );
7 PointedEar 1064
  exit 1;
5 PointedEar 1065
}
1066
 
22 PointedEar 1067
sub dmsg
7 PointedEar 1068
{
22 PointedEar 1069
  print STDERR @_, "\n";
1070
}
1071
 
1072
sub dmsg2
1073
{
1074
  my ( $level, @msg ) = @_;
1075
  print STDERR @msg, "\n" if $level >= DEBUG;
1076
}
23 PointedEar 1077
 
1078
sub println
1079
{
1080
  print @_, "\n";
1081
}