Porting/checkAUTHORS.pl

   1 #!/usr/bin/perl -w
   2 use strict;
   3 use Text::Wrap;
   4 $Text::Wrap::columns = 80;
   5 my ($committer, $patch, $log);
   6 use Getopt::Long;
   7
   8 my ($rank, $percentage, $cumulative, $reverse, $ta, @authors, %authors,
   9     %untraced, %patchers, %committers);
  10 my $result = GetOptions ("rank" => \$rank,                  # rank authors
  11                          "thanks-applied" => \$ta,          # ranks committers
  12                          "acknowledged=s"   => \@authors ,  # authors files
  13                          "percentage" => \$percentage,      # show as %age
  14                          "cumulative" => \$cumulative,
  15                          "reverse" => \$reverse,
  16                         );
  17
  18 if (!$result or (($rank||0) + ($ta||0) + (@authors ? 1 : 0) != 1) or !@ARGV) {
  19   die <<"EOS";
  20 $0 --rank Changelogs                        # rank authors by patches
  21 $0 --acknowledged <authors file> Changelogs # Display unacknowledged authors
  22 $0 --thanks-applied Changelogs              # ranks committers
  23 $0 --percentage ...                         # show rankings as percentages
  24 $0 --cumulative ...                         # show rankings cumulatively
  25 $0 --reverse ...                            # show rankings in reverse
  26 Specify stdin as - if needs be. Remember that option names can be abbreviated.
  27 EOS
  28 }
  29
  30
  31 my $prev = "";
  32 my %map;
  33
  34 while (<DATA>) {
  35     chomp;
  36     s/\\100/\@/g;
  37     $_ = lc;
  38     if (my ($correct, $alias) = /^\s*([^#\s]\S*)\s+(.*\S)/) {
  39         $correct =~ s/^\\043/#/;
  40         if ($correct eq '+') {$correct = $prev} else {$prev = $correct}
  41         $map {$alias} = $correct;
  42     }
  43 }
  44
  45 #
  46 # Email addresses for we do not have names.
  47 #
  48 $map {$_} = "?" for
  49     "agrow\100thegotonerd.com",
  50     "alexander_bluhm\100genua.de",
  51     "alexander_gernler\100genua.de",
  52     "ammon\100rhythm.com",
  53     "andrew\100sundale.net",
  54     "bah\100longitude.com",
  55     "bbucklan\100jpl-devvax.jpl.nasa.gov",
  56     "ben\100linuxgazette.net",
  57     "bilbo\100ua.fm",
  58     "blino\100mandriva.com",
  59     "bob\100starlabs.net",
  60     "bonefish\100cs.tu-berlin.de",
  61     "bstrand\100switchmanagement.com",
  62     "cygwin\100cygwin.com",
  63     "david\100dhaller.de",
  64     "dformosa\100dformosa.zeta.org.au",
  65     "dgay\100acm.org",
  66     "erik\100cs.uni-jena.de",
  67     "glasser\100tang-eleven-seventy-nine.mit.edu",
  68     "gml4410\100ggr.co.uk",
  69     "grommel\100sears.com",
  70     "ilmari\100vesla.ilmari.org",
  71     "ilya\100juil.nonet",
  72     "info\100lingo.kiev.ua",
  73     "jms\100mathras.comcast.net",
  74     "johnh\100isi.edu",
  75     "kan\100dcit.cz",
  76     "kaminsky\100math.huji.ac.il",
  77     "knew-p5p\100pimb.org",
  78     "kvr\100centrum.cz",
  79     "lemkemch\100t-online.de",
  80     "mauzo\100csv.warwick.ac.uk",
  81     "merijnb\100ms.com",
  82     "mlelstv\100serpens.de",
  83     "p.boven\100sara.nl",
  84     "padre\100elte.hu",
  85     "pancho\100pancho.name",
  86     "perlbug\100veggiechinese.net",
  87     "pm\100capmon.dk",
  88     "premchai21\100yahoo.com",
  89     "pxm\100nubz.org",
  90     "raf\100tradingpost.com.au",
  91     "scott\100perlcode.org",
  92     "smoketst\100hp46t243.cup.hp.com",
  93     "yath-perlbug\100yath.de",
  94 ;
  95
  96 #
  97 # Email addresses for people that don't have an email address in AUTHORS
  98 # Presumably deliberately?
  99 #
 100
 101 $map {$_} = '!' for
 102      # Nick Ing-Simmons has passed away (2006-09-25).
 103      "nick\100ing-simmons.net",
 104      "nik\100tiuk.ti.com",
 105      "nick.ing-simmons\100elixent.com",
 106
 107      # Iain Truskett has passed away (2003-12-29).
 108      "perl\100dellah.anu.edu.au",
 109      "spoon\100dellah.org",
 110      "spoon\100cpan.org",
 111
 112      # Ton Hospel
 113      "me-02\100ton.iguana.be",
 114      "perl-5.8.0\100ton.iguana.be",
 115      "perl5-porters\100ton.iguana.be",
 116
 117      # Beau Cox
 118      "beau\100beaucox.com",
 119
 120      # Randy W. Sims
 121      "ml-perl\100thepierianspring.org",
 122
 123      # Yuval Kogman
 124      "nothingmuch\100woobling.org",
 125
 126 ;
 127
 128
 129 if (@authors) {
 130   my %raw;
 131   foreach my $filename (@authors) {
 132     open FH, "<$filename" or die "Can't open $filename: $!";
 133     while (<FH>) {
 134       next if /^\#/;
 135       next if /^-- /;
 136       if (/<([^>]+)>/) {
 137         # Easy line.
 138         $raw{$1}++;
 139       } elsif (/^([-A-Za-z0-9 .\'À-ÖØöø-ÿ]+)[\t\n]/) {
 140         # Name only
 141         $untraced{$1}++;
 142       } else {
 143         chomp;
 144         warn "Can't parse line '$_'";
 145       }
 146     }
 147   }
 148   foreach (keys %raw) {
 149     print "E-mail $_ occurs $raw{$_} times\n" if $raw{$_} > 1;
 150     $_ = lc $_;
 151     $authors{$map{$_} || $_}++;
 152   }
 153   ++$authors{'!'};
 154   ++$authors{'?'};
 155 }
 156
 157 while (<>) {
 158   next if /^-+/;
 159   if (m!^\[\s+(\d+)\]\s+By:\s+(\S+)\s+on!) {
 160     # new patch
 161     my @new = ($1, $2);
 162     &process ($committer, $patch, $log);
 163     ($patch, $committer) = @new;
 164     undef $log;
 165   } elsif (s/^(\s+Log: )//) {
 166     die "Duplicate Log:" if $log;
 167     $log = $_;
 168     my $prefix = " " x length $1;
 169     LOG: while (<>) {
 170       next if /^$/;
 171       s/^\t/        /;
 172       if (s/^$prefix//) {
 173         $log .= $_;
 174       } elsif (/^\s+Branch:/) {
 175         last LOG;
 176       } else {
 177         chomp;
 178         die "Malformed log end with '$_'";
 179       }
 180     }
 181   }
 182 }
 183
 184 &process ($committer, $patch, $log);
 185
 186 if ($rank) {
 187   &display_ordered(\%patchers);
 188 } elsif ($ta) {
 189   &display_ordered(\%committers);
 190 } elsif (%authors) {
 191   my %missing;
 192   foreach (sort keys %patchers) {
 193     next if $authors{$_};
 194     # Sort by number of patches, then name.
 195     $missing{$patchers{$_}}->{$_}++;
 196   }
 197   foreach my $patches (sort {$b <=> $a} keys %missing) {
 198     print "$patches patch(es)\n";
 199     foreach my $author (sort keys %{$missing{$patches}}) {
 200       print "  $author\n";
 201     }
 202   }
 203 }
 204
 205 sub display_ordered {
 206   my $what = shift;
 207   my @sorted;
 208   my $total;
 209   while (my ($name, $count) = each %$what) {
 210     push @{$sorted[$count]}, $name;
 211     $total += $count;
 212   }
 213
 214   my $i = @sorted;
 215   return unless @sorted;
 216   my $sum = 0;
 217   foreach my $i ($reverse ? 0 .. $#sorted : reverse 0 .. $#sorted) {
 218     next unless $sorted[$i];
 219     my $prefix;
 220     $sum += $i * @{$sorted[$i]};
 221     # Value to display is either this one, or the cumulative sum.
 222     my $value = $cumulative ? $sum : $i;
 223     if ($percentage) {
 224         $prefix = sprintf "%6.2f:\t", 100 * $value / $total;
 225     } else {
 226         $prefix = "$value:\t";
 227     }
 228     print wrap ($prefix, "\t", join (" ", sort @{$sorted[$i]}), "\n");
 229   }
 230 }
 231
 232 sub process {
 233   my ($committer, $patch, $log) = @_;
 234   return unless $committer;
 235   my @authors = $log =~ /From:\s+.*?([^"\@ \t\n<>]+\@[^"\@ \t\n<>]+)/gm;
 236
 237   if (@authors) {
 238     foreach (@authors) {
 239       s/^<//;
 240       s/>$//;
 241       $_ = lc $_;
 242       $patchers{$map{$_} || $_}++;
 243     }
 244     # print "$patch: @authors\n";
 245     ++$committers{$committer};
 246   } else {
 247     # print "$patch: $committer\n";
 248     # Not entirely fair as this means that the maint pumpking scores for
 249     # everything intergrated that wasn't a third party patch in blead
 250     $patchers{$committer}++;
 251   }
 252 }
 253
 254
 255 __DATA__
 256
 257 #
 258 # List of mappings. First entry the "correct" email address, as appears
 259 # in the AUTHORS file. Second is any "alias" mapped to it.
 260 #
 261 # If the "correct" email address is a '+', the entry above is reused;
 262 # this for addresses with more than one alias.
 263 #
 264 # Note that all entries are in lowercase. Further, no '@' signs should
 265 # appear; use \100 instead.
 266 #
 267 #
 268 #  Committers.
 269 #
 270 adi                                     enache\100rdslink.ro
 271 alanbur                                 alan.burlison\100sun.com
 272 +                                       alan.burlison\100uk.sun.com
 273 ams                                     ams\100wiw.org
 274 chip                                    chip\100pobox.com
 275 craigb                                  craig.berry\100psinetcs.com
 276 +                                       craig.berry\100metamorgs.com
 277 +                                       craig.berry\100signaltreesolutions.com
 278 +                                       craigberry\100mac.com
 279 +                                       craig.a.berry\100gmail.com
 280 davem                                   davem\100fdgroup.com
 281 +                                       davem\100iabyn.nospamdeletethisbit.com
 282 +                                       davem\100iabyn.com
 283 +                                       davem\100fdgroup.co.uk
 284 +                                       davem\100fdisolutions.com
 285 +                                       davem\100iabyn.com
 286 demerphq                                demerphq\100gmail.com
 287 +                                       yves.orton\100de.mci.com
 288 +                                       yves.orton\100mciworldcom.de
 289 doughera                                doughera\100lafayette.edu
 290 gbarr                                   gbarr\100pobox.com
 291 gisle                                   gisle\100activestate.com
 292 +                                       gisle\100aas.no
 293 gsar                                    gsar\100activestate.com
 294 +                                       gsar\100cpan.org
 295 hv                                      hv\100crypt.compulink.co.uk
 296 +                                       hv\100crypt.org
 297 jhi                                     jhi\100iki.fi
 298 +                                       jhietaniemi\100gmail.com
 299 +                                       jhi\100kosh.hut.fi
 300 +                                       jhi\100cc.hut.fi
 301 +                                       jarkko.hietaniemi\100nokia.com
 302 merijn                                  h.m.brand\100xs4all.nl
 303 +                                       h.m.brand\100hccnet.nl
 304 +                                       merijn\100l1.procura.nl
 305 mhx                                     mhx-perl\100gmx.net
 306 nicholas                                nick\100unfortu.net
 307 +                                       nick\100ccl4.org
 308 +                                       nick\100talking.bollo.cx
 309 +                                       nick\100plum.flirble.org
 310 +                                       nick\100babyhippo.co.uk
 311 +                                       nick\100bagpuss.unfortu.net
 312 pudge                                   pudge\100pobox.com
 313 rgs                                     rgarciasuarez\100free.fr
 314 +                                       rgarciasuarez\100mandrakesoft.com
 315 +                                       rgarciasuarez\100mandriva.com
 316 +                                       rgarciasuarez\100gmail.com
 317 +                                       raphel.garcia-suarez\100hexaflux.com
 318 sky                                     sky\100nanisky.com
 319 +                                       artur\100contiller.se
 320 +                                       arthur\100contiller.se
 321 steveh                                  stevehay\100planit.com
 322 +                                       steve.hay\100uk.radan.com
 323 stevep                                  steve\100fisharerojo.org
 324 +                                       steve.peters\100gmail.com
 325
 326 #
 327 # Mere mortals.
 328 #
 329 \043####\100juerd.nl                    juerd\100cpan.org
 330 +                                       juerd\100convolution.nl
 331 7k8lrvf02\100sneakemail.com             kjx9zthh3001\100sneakemail.com
 332 +                                       dtr8sin02\100sneakemail.com
 333 +                                       rt8363b02\100sneakemail.com
 334 +                                       o6hhmk002\100sneakemail.com
 335 +                                       smueller\100cpan.org
 336 +                                       l2ot9pa02\100sneakemail.com
 337 +                                       wyp3rlx02\100sneakemail.com
 338
 339 a.r.ferreira\100gmail.com               aferreira\100shopzilla.com
 340 abe\100ztreet.demon.nl                  abeltje\100cpan.org
 341 abigail\100abigail.be                   abigail\100foad.org
 342 +                                       abigail\100abigail.nl
 343 ajohnson\100nvidia.com                  ajohnson\100wischip.com
 344 alexm\100netli.com                      alexm\100w-m.ru
 345 alex-p5p\100earth.li                    alex\100rcon.rog
 346 alian\100cpan.org                       alian\100alianwebserver.com
 347 allens\100cpan.org                      easmith\100beatrice.rutgers.edu
 348 andreas.koenig\100anima.de              andreas.koenig.gmwojprw\100franz.ak.mind.de
 349 +                                       andreas.koenig.7os6vvqr\100franz.ak.mind.de
 350 +                                       a.koenig\100mind.de
 351 anno4000\100lublin.zrz.tu-berlin.de     anno4000\100mailbox.tu-berlin.de
 352 +                                       siegel\100zrz.tu-berlin.de
 353 ash\100cpan.org                         ash_cpan\100firemirror.com
 354 avarab\100gmail.com                     avar\100cpan.org
 355
 356 bah\100ecnvantage.com                   bholzman\100longitude.com
 357 bcarter\100gumdrop.flyinganvil.org      q.eibcartereio.=~m-b.{6}-cgimosx\100gumdrop.flyinganvil.org
 358 ben_tilly\100operamail.com              btilly\100gmail.com
 359
 360 chromatic\100wgz.org                    chromatic\100rmci.net
 361 clkao\100clkao.org                      clkao\100bestpractical.com
 362 corion\100corion.net                    corion\100cpan.org
 363 cp\100onsitetech.com                    publiustemp-p5p\100yahoo.com
 364 +                                       publiustemp-p5p3\100yahoo.com
 365 cpan\100audreyt.org                     autrijus\100egb.elixus.org
 366 +                                       autrijus\100geb.elixus.org
 367 +                                       autrijus\100gmail.com
 368 +                                       autrijus\100ossf.iis.sinica.edu.tw
 369 +                                       autrijus\100autrijus.org
 370 +                                       audreyt\100audreyt.org
 371
 372 damian\100cs.monash.edu.au              damian\100conway.org
 373 david.dyck\100fluke.com                 dcd\100tc.fluke.com
 374 demerphq\100gmail.com                   demerphq\100hotmail.com
 375 domo\100computer.org                    shouldbedomo\100mac.com
 376
 377 epeschko\100den-mdev1                   esp5\100pge.com
 378
 379 fugazi\100zyx.net                       larrysh\100cpan.org
 380
 381 jns\100integration-house.com            jns\100gellyfish.com
 382 +                                       gellyfish\100gellyfish.com
 383 gp\100familiehaase.de                   gerrit\100familiehaase.de
 384 grazz\100pobox.com                      grazz\100nyc.rr.com
 385
 386 hio\100ymir.co.jp                       hio\100hio.jp
 387
 388 japhy\100pobox.com                      japhy\100pobox.org
 389 +                                       japhy\100perlmonk.org
 390 +                                       japhy\100cpan.org
 391 jari.aalto\100poboxes.com               jari.aalto\100cante.net
 392 jcromie\100divsol.com                   jcromie\100cpan.org
 393 +                                       jim.cromie\100gmail.com
 394 jdhedden\100cpan.org                    jerry\100hedden.us
 395 +                                       jdhedden\1001979.usna.com
 396 +                                       jdhedden\100gmail.com
 397 +                                       jdhedden\100yahoo.com
 398 jfriedl\100yahoo.com                    jfriedl\100yahoo-inc.com
 399 jjore\100cpan.org                       twists\100gmail.com
 400
 401 kane\100dwim.org                        kane\100xs4all.net
 402 +                                       kane\100cpan.org
 403 +                                       kane\100xs4all.nl
 404 +                                       jos\100dwim.org
 405 +                                       jib\100ripe.net
 406 ken\100mathforum.org                    kenahoo\100gmail.com
 407 kroepke\100dolphin-services.de          kay\100dolphin-services.de
 408 kstar\100wolfetech.com                  kstar\100cpan.org
 409 +                                       kurt_starsinic\100ml.com
 410
 411 marcel\100codewerk.com                  gr\100univie.ac.at
 412 mats\100sm6sxl.net                      mats\100sm5sxl.net
 413 mbarbon\100dsi.unive.it                 mattia.barbon\100libero.it
 414 mcmahon\100ibiblio.org                  mcmahon\100metalab.unc.edu
 415 merijnb\100iloquent.nl                  merijnb\100iloquent.com
 416 mgjv\100comdyn.com.au                   mgjv\100tradingpost.com.au
 417 michael.schroeder\100informatik.uni-erlangen.de mls\100suse.de
 418 mike\100stok.co.uk                      mike\100exegenix.com
 419 mjtg\100cam.ac.uk                       mjtg\100cus.cam.ac.uk
 420
 421 nospam-abuse\100bloodgate.com           tels\100bloodgate.com
 422 +                                       perl_dummy\100bloodgate.com
 423 ilya\100math.berkeley.edu               ilya\100math.ohio-state.edu
 424 +                                       nospam-abuse\100ilyaz.org
 425
 426 p5-authors\100crystalflame.net          perl\100crystalflame.net
 427 +                                       rs\100crystalflame.net
 428 paul.green\100stratus.com               paul_greenvos\100vos.stratus.com
 429 paul.marquess\100btinternet.com         paul_marquess\100yahoo.co.uk
 430 +                                       paul.marquess\100ntlworld.com
 431 +                                       paul.marquess\100openwave.com
 432 pcg\100goof.com                         schmorp\100schmorp.de
 433 perl-rt\100wizbit.be                    p5p\100perl.wizbit.be
 434 # Maybe we should special case this to get real names out?
 435 perlbug\100perl.org                     perlbug-followup\100perl.org
 436 +                                       bugs-perl5\100bugs6.perl.org
 437 phil\100perkpartners.com                phil\100finchcomputer.com
 438 pimlott\100idiomtech.com                andrew\100pimlott.net
 439 pne\100cpan.org                         philip.newton\100gmx.net
 440 +                                       philip.newton\100datenrevision.de
 441 +                                       pnewton\100gmx.de
 442
 443 radu\100netsoft.ro                      rgreab\100fx.ro
 444 richard.foley\100ubsw.com               richard.foley\100t-online.de
 445 +                                       richard.foley\100ubs.com
 446 +                                       richard.foley\100rfi.net
 447 rick\100consumercontact.com             rick\100bort.ca
 448 +                                       rick.delaney\100rogers.com
 449 +                                       rick\100bort.ca
 450 rjbs\100cpan.org                        rjbs-perl-p5p\100lists.manxome.org
 451 +                                       perl.p5p\100rjbs.manxome.org
 452 rjk\100linguist.dartmouth.edu           rjk\100linguist.thayer.dartmouth.edu
 453 +                                       rjk-perl-p5p\100tamias.net
 454 rmgiroux\100acm.org                     rmgiroux\100hotmail.com
 455 rmbarker\100cpan.org                    rmb1\100cise.npl.co.uk
 456 +                                       robin.barker\100npl.co.uk
 457 robertmay\100cpan.org                   rob\100themayfamily.me.uk
 458 rootbeer\100teleport.com                rootbeer\100redcat.com
 459
 460 schubiger\100cpan.org                   steven\100accognoscere.org
 461 +                                       sts\100accognoscere.org
 462 +                                       schubiger\100gmail.com
 463 schwern\100pobox.com                    schwern\100gmail.com
 464 +                                       schwern\100athens.arena-i.com
 465 +                                       schwern\100blackrider.aocn.com
 466 +                                       schwern\100ool-18b93024.dyn.optonline.net
 467 sebastien\100aperghis.net               maddingue\100free.fr
 468 +                                       saper\100cpan.org
 469 simon\100simon-cozens.org               simon\100pembro4.pmb.ox.ac.uk
 470 +                                       simon\100brecon.co.uk
 471 +                                       simon\100othersideofthe.earth.li
 472 +                                       simon\100cozens.net
 473 +                                       simon\100netthink.co.uk
 474 slaven\100rezic.de                      slaven.rezic\100berlin.de
 475 +                                       srezic\100iconmobile.com
 476 +                                       srezic\100cpan.org
 477 smcc\100mit.edu                         smcc\100ocf.berkeley.edu
 478 +                                       smcc\100csua.berkeley.edu
 479 spider\100orb.nashua.nh.us              spider\100web.zk3.dec.com
 480 +                                       spider\100leggy.zk3.dec.com
 481 +                                       spider-perl\100orb.nashua.nh.us
 482 +                                       spider\100peano.zk3.dec.com
 483 stef\100mongueurs.net                   stef\100payrard.net
 484 +                                       s.payrard\100wanadoo.fr
 485
 486 tassilo.parseval\100post.rwth-aachen.de tassilo.von.parseval\100rwth-aachen.de
 487 thomas.dorner\100start.de               tdorner\100amadeus.net
 488 tjenness\100cpan.org                    t.jenness\100jach.hawaii.edu
 489 +                                       timj\100jach.hawaii.edu
 490 tom.horsley\100mail.ccur.com            tom.horsley\100ccur.com
 491
 492 vkonovalov\100lucent.com                vkonovalov\100peterstar.ru
 493 +                                       konovalo\100mail.wplus.net
 494 +                                       vadim\100vkonovalov.ru
 495 +                                       vkonovalov\100spb.lucent.com
 496 +                                       vkonovalov\100alcatel-lucent.com
 497
 498 whatever\100davidnicol.com              davidnicol\100gmail.com
 499 wolfgang.laun\100alcatel.at             wolfgang.laun\100chello.at
 500 +                                       wolfgang.laun\100thalesgroup.com
 501 +                                       wolfgang.laun\100gmail.com