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