scripts/get_maintainer.pl

   1 #!/usr/bin/perl -w
   2 # (c) 2007, Joe Perches <joe@perches.com>
   3 #           created from checkpatch.pl
   4 #
   5 # Print selected MAINTAINERS information for
   6 # the files modified in a patch or for a file
   7 #
   8 # usage: perl scripts/get_maintainer.pl [OPTIONS] <patch>
   9 #        perl scripts/get_maintainer.pl [OPTIONS] -f <file>
  10 #
  11 # Licensed under the terms of the GNU GPL License version 2
  12
  13 use strict;
  14
  15 my $P = $0;
  16 my $V = '0.23';
  17
  18 use Getopt::Long qw(:config no_auto_abbrev);
  19
  20 my $lk_path = "./";
  21 my $email = 1;
  22 my $email_usename = 1;
  23 my $email_maintainer = 1;
  24 my $email_list = 1;
  25 my $email_subscriber_list = 0;
  26 my $email_git_penguin_chiefs = 0;
  27 my $email_git = 1;
  28 my $email_git_blame = 0;
  29 my $email_git_min_signatures = 1;
  30 my $email_git_max_maintainers = 5;
  31 my $email_git_min_percent = 5;
  32 my $email_git_since = "1-year-ago";
  33 my $email_hg_since = "-365";
  34 my $email_remove_duplicates = 1;
  35 my $output_multiline = 1;
  36 my $output_separator = ", ";
  37 my $output_roles = 0;
  38 my $output_rolestats = 0;
  39 my $scm = 0;
  40 my $web = 0;
  41 my $subsystem = 0;
  42 my $status = 0;
  43 my $keywords = 1;
  44 my $sections = 0;
  45 my $file_emails = 0;
  46 my $from_filename = 0;
  47 my $pattern_depth = 0;
  48 my $version = 0;
  49 my $help = 0;
  50
  51 my $exit = 0;
  52
  53 my @penguin_chief = ();
  54 push(@penguin_chief,"Linus Torvalds:torvalds\@linux-foundation.org");
  55 #Andrew wants in on most everything - 2009/01/14
  56 #push(@penguin_chief,"Andrew Morton:akpm\@linux-foundation.org");
  57
  58 my @penguin_chief_names = ();
  59 foreach my $chief (@penguin_chief) {
  60     if ($chief =~ m/^(.*):(.*)/) {
  61         my $chief_name = $1;
  62         my $chief_addr = $2;
  63         push(@penguin_chief_names, $chief_name);
  64     }
  65 }
  66 my $penguin_chiefs = "\(" . join("|",@penguin_chief_names) . "\)";
  67
  68 # rfc822 email address - preloaded methods go here.
  69 my $rfc822_lwsp = "(?:(?:\\r\\n)?[ \\t])";
  70 my $rfc822_char = '[\\000-\\377]';
  71
  72 # VCS command support: class-like functions and strings
  73
  74 my %VCS_cmds;
  75
  76 my %VCS_cmds_git = (
  77     "execute_cmd" => \&git_execute_cmd,
  78     "available" => '(which("git") ne "") && (-d ".git")',
  79     "find_signers_cmd" => "git log --no-color --since=\$email_git_since -- \$file",
  80     "find_commit_signers_cmd" => "git log --no-color -1 \$commit",
  81     "blame_range_cmd" => "git blame -l -L \$diff_start,+\$diff_length \$file",
  82     "blame_file_cmd" => "git blame -l \$file",
  83     "commit_pattern" => "^commit [0-9a-f]{40,40}",
  84     "blame_commit_pattern" => "^([0-9a-f]+) "
  85 );
  86
  87 my %VCS_cmds_hg = (
  88     "execute_cmd" => \&hg_execute_cmd,
  89     "available" => '(which("hg") ne "") && (-d ".hg")',
  90     "find_signers_cmd" =>
  91         "hg log --date=\$email_hg_since" .
  92                 " --template='commit {node}\\n{desc}\\n' -- \$file",
  93     "find_commit_signers_cmd" => "hg log --template='{desc}\\n' -r \$commit",
  94     "blame_range_cmd" => "",            # not supported
  95     "blame_file_cmd" => "hg blame -c \$file",
  96     "commit_pattern" => "^commit [0-9a-f]{40,40}",
  97     "blame_commit_pattern" => "^([0-9a-f]+):"
  98 );
  99
 100 if (!GetOptions(
 101                 'email!' => \$email,
 102                 'git!' => \$email_git,
 103                 'git-blame!' => \$email_git_blame,
 104                 'git-chief-penguins!' => \$email_git_penguin_chiefs,
 105                 'git-min-signatures=i' => \$email_git_min_signatures,
 106                 'git-max-maintainers=i' => \$email_git_max_maintainers,
 107                 'git-min-percent=i' => \$email_git_min_percent,
 108                 'git-since=s' => \$email_git_since,
 109                 'hg-since=s' => \$email_hg_since,
 110                 'remove-duplicates!' => \$email_remove_duplicates,
 111                 'm!' => \$email_maintainer,
 112                 'n!' => \$email_usename,
 113                 'l!' => \$email_list,
 114                 's!' => \$email_subscriber_list,
 115                 'multiline!' => \$output_multiline,
 116                 'roles!' => \$output_roles,
 117                 'rolestats!' => \$output_rolestats,
 118                 'separator=s' => \$output_separator,
 119                 'subsystem!' => \$subsystem,
 120                 'status!' => \$status,
 121                 'scm!' => \$scm,
 122                 'web!' => \$web,
 123                 'pattern-depth=i' => \$pattern_depth,
 124                 'k|keywords!' => \$keywords,
 125                 'sections!' => \$sections,
 126                 'fe|file-emails!' => \$file_emails,
 127                 'f|file' => \$from_filename,
 128                 'v|version' => \$version,
 129                 'h|help|usage' => \$help,
 130                 )) {
 131     die "$P: invalid argument - use --help if necessary\n";
 132 }
 133
 134 if ($help != 0) {
 135     usage();
 136     exit 0;
 137 }
 138
 139 if ($version != 0) {
 140     print("${P} ${V}\n");
 141     exit 0;
 142 }
 143
 144 if (-t STDIN && !@ARGV) {
 145     # We're talking to a terminal, but have no command line arguments.
 146     die "$P: missing patchfile or -f file - use --help if necessary\n";
 147 }
 148
 149 if ($output_separator ne ", ") {
 150     $output_multiline = 0;
 151 }
 152
 153 if ($output_rolestats) {
 154     $output_roles = 1;
 155 }
 156
 157 if ($sections) {
 158     $email = 0;
 159     $email_list = 0;
 160     $scm = 0;
 161     $status = 0;
 162     $subsystem = 0;
 163     $web = 0;
 164     $keywords = 0;
 165 } else {
 166     my $selections = $email + $scm + $status + $subsystem + $web;
 167     if ($selections == 0) {
 168         die "$P:  Missing required option: email, scm, status, subsystem or web\n";
 169     }
 170 }
 171
 172 if ($email &&
 173     ($email_maintainer + $email_list + $email_subscriber_list +
 174      $email_git + $email_git_penguin_chiefs + $email_git_blame) == 0) {
 175     die "$P: Please select at least 1 email option\n";
 176 }
 177
 178 if (!top_of_kernel_tree($lk_path)) {
 179     die "$P: The current directory does not appear to be "
 180         . "a linux kernel source tree.\n";
 181 }
 182
 183 ## Read MAINTAINERS for type/value pairs
 184
 185 my @typevalue = ();
 186 my %keyword_hash;
 187
 188 open(MAINT, "<${lk_path}MAINTAINERS") || die "$P: Can't open MAINTAINERS\n";
 189 while (<MAINT>) {
 190     my $line = $_;
 191
 192     if ($line =~ m/^(\C):\s*(.*)/) {
 193         my $type = $1;
 194         my $value = $2;
 195
 196         ##Filename pattern matching
 197         if ($type eq "F" || $type eq "X") {
 198             $value =~ s@\.@\\\.@g;       ##Convert . to \.
 199             $value =~ s/\*/\.\*/g;       ##Convert * to .*
 200             $value =~ s/\?/\./g;         ##Convert ? to .
 201             ##if pattern is a directory and it lacks a trailing slash, add one
 202             if ((-d $value)) {
 203                 $value =~ s@([^/])$@$1/@;
 204             }
 205         } elsif ($type eq "K") {
 206             $keyword_hash{@typevalue} = $value;
 207         }
 208         push(@typevalue, "$type:$value");
 209     } elsif (!/^(\s)*$/) {
 210         $line =~ s/\n$//g;
 211         push(@typevalue, $line);
 212     }
 213 }
 214 close(MAINT);
 215
 216 my %mailmap;
 217
 218 if ($email_remove_duplicates) {
 219     open(MAILMAP, "<${lk_path}.mailmap") || warn "$P: Can't open .mailmap\n";
 220     while (<MAILMAP>) {
 221         my $line = $_;
 222
 223         next if ($line =~ m/^\s*#/);
 224         next if ($line =~ m/^\s*$/);
 225
 226         my ($name, $address) = parse_email($line);
 227         $line = format_email($name, $address, $email_usename);
 228
 229         next if ($line =~ m/^\s*$/);
 230
 231         if (exists($mailmap{$name})) {
 232             my $obj = $mailmap{$name};
 233             push(@$obj, $address);
 234         } else {
 235             my @arr = ($address);
 236             $mailmap{$name} = \@arr;
 237         }
 238     }
 239     close(MAILMAP);
 240 }
 241
 242 ## use the filenames on the command line or find the filenames in the patchfiles
 243
 244 my @files = ();
 245 my @range = ();
 246 my @keyword_tvi = ();
 247 my @file_emails = ();
 248
 249 if (!@ARGV) {
 250     push(@ARGV, "&STDIN");
 251 }
 252
 253 foreach my $file (@ARGV) {
 254     if ($file ne "&STDIN") {
 255         ##if $file is a directory and it lacks a trailing slash, add one
 256         if ((-d $file)) {
 257             $file =~ s@([^/])$@$1/@;
 258         } elsif (!(-f $file)) {
 259             die "$P: file '${file}' not found\n";
 260         }
 261     }
 262     if ($from_filename) {
 263         push(@files, $file);
 264         if (-f $file && ($keywords || $file_emails)) {
 265             open(FILE, "<$file") or die "$P: Can't open ${file}\n";
 266             my $text = do { local($/) ; <FILE> };
 267             close(FILE);
 268             if ($keywords) {
 269                 foreach my $line (keys %keyword_hash) {
 270                     if ($text =~ m/$keyword_hash{$line}/x) {
 271                         push(@keyword_tvi, $line);
 272                     }
 273                 }
 274             }
 275             if ($file_emails) {
 276                 my @poss_addr = $text =~ m$[A-Za-zÀ-ÿ\"\' \,\.\+-]*\s*[\,]*\s*[\(\<\{]{0,1}[A-Za-z0-9_\.\+-]+\@[A-Za-z0-9\.-]+\.[A-Za-z0-9]+[\)\>\}]{0,1}$g;
 277                 push(@file_emails, clean_file_emails(@poss_addr));
 278             }
 279         }
 280     } else {
 281         my $file_cnt = @files;
 282         my $lastfile;
 283         open(PATCH, "<$file") or die "$P: Can't open ${file}\n";
 284         while (<PATCH>) {
 285             my $patch_line = $_;
 286             if (m/^\+\+\+\s+(\S+)/) {
 287                 my $filename = $1;
 288                 $filename =~ s@^[^/]*/@@;
 289                 $filename =~ s@\n@@;
 290                 $lastfile = $filename;
 291                 push(@files, $filename);
 292             } elsif (m/^\@\@ -(\d+),(\d+)/) {
 293                 if ($email_git_blame) {
 294                     push(@range, "$lastfile:$1:$2");
 295                 }
 296             } elsif ($keywords) {
 297                 foreach my $line (keys %keyword_hash) {
 298                     if ($patch_line =~ m/^[+-].*$keyword_hash{$line}/x) {
 299                         push(@keyword_tvi, $line);
 300                     }
 301                 }
 302             }
 303         }
 304         close(PATCH);
 305         if ($file_cnt == @files) {
 306             warn "$P: file '${file}' doesn't appear to be a patch.  "
 307                 . "Add -f to options?\n";
 308         }
 309         @files = sort_and_uniq(@files);
 310     }
 311 }
 312
 313 @file_emails = uniq(@file_emails);
 314
 315 my @email_to = ();
 316 my @list_to = ();
 317 my @scm = ();
 318 my @web = ();
 319 my @subsystem = ();
 320 my @status = ();
 321
 322 # Find responsible parties
 323
 324 foreach my $file (@files) {
 325
 326     my %hash;
 327     my $tvi = find_first_section();
 328     while ($tvi < @typevalue) {
 329         my $start = find_starting_index($tvi);
 330         my $end = find_ending_index($tvi);
 331         my $exclude = 0;
 332         my $i;
 333
 334         #Do not match excluded file patterns
 335
 336         for ($i = $start; $i < $end; $i++) {
 337             my $line = $typevalue[$i];
 338             if ($line =~ m/^(\C):\s*(.*)/) {
 339                 my $type = $1;
 340                 my $value = $2;
 341                 if ($type eq 'X') {
 342                     if (file_match_pattern($file, $value)) {
 343                         $exclude = 1;
 344                     }
 345                 }
 346             }
 347         }
 348
 349         if (!$exclude) {
 350             for ($i = $start; $i < $end; $i++) {
 351                 my $line = $typevalue[$i];
 352                 if ($line =~ m/^(\C):\s*(.*)/) {
 353                     my $type = $1;
 354                     my $value = $2;
 355                     if ($type eq 'F') {
 356                         if (file_match_pattern($file, $value)) {
 357                             my $value_pd = ($value =~ tr@/@@);
 358                             my $file_pd = ($file  =~ tr@/@@);
 359                             $value_pd++ if (substr($value,-1,1) ne "/");
 360                             if ($pattern_depth == 0 ||
 361                                 (($file_pd - $value_pd) < $pattern_depth)) {
 362                                 $hash{$tvi} = $value_pd;
 363                             }
 364                         }
 365                     }
 366                 }
 367             }
 368         }
 369
 370         $tvi += ($end - $start);
 371
 372     }
 373
 374     foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) {
 375         add_categories($line);
 376             if ($sections) {
 377                 my $i;
 378                 my $start = find_starting_index($line);
 379                 my $end = find_ending_index($line);
 380                 for ($i = $start; $i < $end; $i++) {
 381                     my $line = $typevalue[$i];
 382                     if ($line =~ /^[FX]:/) {            ##Restore file patterns
 383                         $line =~ s/([^\\])\.([^\*])/$1\?$2/g;
 384                         $line =~ s/([^\\])\.$/$1\?/g;   ##Convert . back to ?
 385                         $line =~ s/\\\./\./g;           ##Convert \. to .
 386                         $line =~ s/\.\*/\*/g;           ##Convert .* to *
 387                     }
 388                     $line =~ s/^([A-Z]):/$1:\t/g;
 389                     print("$line\n");
 390                 }
 391                 print("\n");
 392             }
 393     }
 394
 395     if ($email && $email_git) {
 396         vcs_file_signoffs($file);
 397     }
 398
 399     if ($email && $email_git_blame) {
 400         vcs_file_blame($file);
 401     }
 402 }
 403
 404 if ($keywords) {
 405     @keyword_tvi = sort_and_uniq(@keyword_tvi);
 406     foreach my $line (@keyword_tvi) {
 407         add_categories($line);
 408     }
 409 }
 410
 411 if ($email) {
 412     foreach my $chief (@penguin_chief) {
 413         if ($chief =~ m/^(.*):(.*)/) {
 414             my $email_address;
 415
 416             $email_address = format_email($1, $2, $email_usename);
 417             if ($email_git_penguin_chiefs) {
 418                 push(@email_to, [$email_address, 'chief penguin']);
 419             } else {
 420                 @email_to = grep($_->[0] !~ /${email_address}/, @email_to);
 421             }
 422         }
 423     }
 424
 425     foreach my $email (@file_emails) {
 426         my ($name, $address) = parse_email($email);
 427
 428         my $tmp_email = format_email($name, $address, $email_usename);
 429         push_email_address($tmp_email, '');
 430         add_role($tmp_email, 'in file');
 431     }
 432 }
 433
 434 if ($email || $email_list) {
 435     my @to = ();
 436     if ($email) {
 437         @to = (@to, @email_to);
 438     }
 439     if ($email_list) {
 440         @to = (@to, @list_to);
 441     }
 442     output(merge_email(@to));
 443 }
 444
 445 if ($scm) {
 446     @scm = uniq(@scm);
 447     output(@scm);
 448 }
 449
 450 if ($status) {
 451     @status = uniq(@status);
 452     output(@status);
 453 }
 454
 455 if ($subsystem) {
 456     @subsystem = uniq(@subsystem);
 457     output(@subsystem);
 458 }
 459
 460 if ($web) {
 461     @web = uniq(@web);
 462     output(@web);
 463 }
 464
 465 exit($exit);
 466
 467 sub file_match_pattern {
 468     my ($file, $pattern) = @_;
 469     if (substr($pattern, -1) eq "/") {
 470         if ($file =~ m@^$pattern@) {
 471             return 1;
 472         }
 473     } else {
 474         if ($file =~ m@^$pattern@) {
 475             my $s1 = ($file =~ tr@/@@);
 476             my $s2 = ($pattern =~ tr@/@@);
 477             if ($s1 == $s2) {
 478                 return 1;
 479             }
 480         }
 481     }
 482     return 0;
 483 }
 484
 485 sub usage {
 486     print <<EOT;
 487 usage: $P [options] patchfile
 488        $P [options] -f file|directory
 489 version: $V
 490
 491 MAINTAINER field selection options:
 492   --email => print email address(es) if any
 493     --git => include recent git \*-by: signers
 494     --git-chief-penguins => include ${penguin_chiefs}
 495     --git-min-signatures => number of signatures required (default: 1)
 496     --git-max-maintainers => maximum maintainers to add (default: 5)
 497     --git-min-percent => minimum percentage of commits required (default: 5)
 498     --git-blame => use git blame to find modified commits for patch or file
 499     --git-since => git history to use (default: 1-year-ago)
 500     --hg-since => hg history to use (default: -365)
 501     --m => include maintainer(s) if any
 502     --n => include name 'Full Name <addr\@domain.tld>'
 503     --l => include list(s) if any
 504     --s => include subscriber only list(s) if any
 505     --remove-duplicates => minimize duplicate email names/addresses
 506     --roles => show roles (status:subsystem, git-signer, list, etc...)
 507     --rolestats => show roles and statistics (commits/total_commits, %)
 508     --file-emails => add email addresses found in -f file (default: 0 (off))
 509   --scm => print SCM tree(s) if any
 510   --status => print status if any
 511   --subsystem => print subsystem name if any
 512   --web => print website(s) if any
 513
 514 Output type options:
 515   --separator [, ] => separator for multiple entries on 1 line
 516     using --separator also sets --nomultiline if --separator is not [, ]
 517   --multiline => print 1 entry per line
 518
 519 Other options:
 520   --pattern-depth => Number of pattern directory traversals (default: 0 (all))
 521   --keywords => scan patch for keywords (default: 1 (on))
 522   --sections => print the entire subsystem sections with pattern matches
 523   --version => show version
 524   --help => show this help information
 525
 526 Default options:
 527   [--email --git --m --n --l --multiline --pattern-depth=0 --remove-duplicates]
 528
 529 Notes:
 530   Using "-f directory" may give unexpected results:
 531       Used with "--git", git signators for _all_ files in and below
 532           directory are examined as git recurses directories.
 533           Any specified X: (exclude) pattern matches are _not_ ignored.
 534       Used with "--nogit", directory is used as a pattern match,
 535           no individual file within the directory or subdirectory
 536           is matched.
 537       Used with "--git-blame", does not iterate all files in directory
 538   Using "--git-blame" is slow and may add old committers and authors
 539       that are no longer active maintainers to the output.
 540   Using "--roles" or "--rolestats" with git send-email --cc-cmd or any
 541       other automated tools that expect only ["name"] <email address>
 542       may not work because of additional output after <email address>.
 543   Using "--rolestats" and "--git-blame" shows the #/total=% commits,
 544       not the percentage of the entire file authored.  # of commits is
 545       not a good measure of amount of code authored.  1 major commit may
 546       contain a thousand lines, 5 trivial commits may modify a single line.
 547   If git is not installed, but mercurial (hg) is installed and an .hg
 548       repository exists, the following options apply to mercurial:
 549           --git,
 550           --git-min-signatures, --git-max-maintainers, --git-min-percent, and
 551           --git-blame
 552       Use --hg-since not --git-since to control date selection
 553 EOT
 554 }
 555
 556 sub top_of_kernel_tree {
 557         my ($lk_path) = @_;
 558
 559         if ($lk_path ne "" && substr($lk_path,length($lk_path)-1,1) ne "/") {
 560             $lk_path .= "/";
 561         }
 562         if (   (-f "${lk_path}COPYING")
 563             && (-f "${lk_path}CREDITS")
 564             && (-f "${lk_path}Kbuild")
 565             && (-f "${lk_path}MAINTAINERS")
 566             && (-f "${lk_path}Makefile")
 567             && (-f "${lk_path}README")
 568             && (-d "${lk_path}Documentation")
 569             && (-d "${lk_path}arch")
 570             && (-d "${lk_path}include")
 571             && (-d "${lk_path}drivers")
 572             && (-d "${lk_path}fs")
 573             && (-d "${lk_path}init")
 574             && (-d "${lk_path}ipc")
 575             && (-d "${lk_path}kernel")
 576             && (-d "${lk_path}lib")
 577             && (-d "${lk_path}scripts")) {
 578                 return 1;
 579         }
 580         return 0;
 581 }
 582
 583 sub parse_email {
 584     my ($formatted_email) = @_;
 585
 586     my $name = "";
 587     my $address = "";
 588
 589     if ($formatted_email =~ /^([^<]+)<(.+\@.*)>.*$/) {
 590         $name = $1;
 591         $address = $2;
 592     } elsif ($formatted_email =~ /^\s*<(.+\@\S*)>.*$/) {
 593         $address = $1;
 594     } elsif ($formatted_email =~ /^(.+\@\S*).*$/) {
 595         $address = $1;
 596     }
 597
 598     $name =~ s/^\s+|\s+$//g;
 599     $name =~ s/^\"|\"$//g;
 600     $address =~ s/^\s+|\s+$//g;
 601
 602     if ($name =~ /[^a-z0-9 \.\-]/i) {    ##has "must quote" chars
 603         $name =~ s/(?<!\\)"/\\"/g;       ##escape quotes
 604         $name = "\"$name\"";
 605     }
 606
 607     return ($name, $address);
 608 }
 609
 610 sub format_email {
 611     my ($name, $address, $usename) = @_;
 612
 613     my $formatted_email;
 614
 615     $name =~ s/^\s+|\s+$//g;
 616     $name =~ s/^\"|\"$//g;
 617     $address =~ s/^\s+|\s+$//g;
 618
 619     if ($name =~ /[^a-z0-9 \.\-]/i) {    ##has "must quote" chars
 620         $name =~ s/(?<!\\)"/\\"/g;       ##escape quotes
 621         $name = "\"$name\"";
 622     }
 623
 624     if ($usename) {
 625         if ("$name" eq "") {
 626             $formatted_email = "$address";
 627         } else {
 628             $formatted_email = "$name <$address>";
 629         }
 630     } else {
 631         $formatted_email = $address;
 632     }
 633
 634     return $formatted_email;
 635 }
 636
 637 sub find_first_section {
 638     my $index = 0;
 639
 640     while ($index < @typevalue) {
 641         my $tv = $typevalue[$index];
 642         if (($tv =~ m/^(\C):\s*(.*)/)) {
 643             last;
 644         }
 645         $index++;
 646     }
 647
 648     return $index;
 649 }
 650
 651 sub find_starting_index {
 652     my ($index) = @_;
 653
 654     while ($index > 0) {
 655         my $tv = $typevalue[$index];
 656         if (!($tv =~ m/^(\C):\s*(.*)/)) {
 657             last;
 658         }
 659         $index--;
 660     }
 661
 662     return $index;
 663 }
 664
 665 sub find_ending_index {
 666     my ($index) = @_;
 667
 668     while ($index < @typevalue) {
 669         my $tv = $typevalue[$index];
 670         if (!($tv =~ m/^(\C):\s*(.*)/)) {
 671             last;
 672         }
 673         $index++;
 674     }
 675
 676     return $index;
 677 }
 678
 679 sub get_maintainer_role {
 680     my ($index) = @_;
 681
 682     my $i;
 683     my $start = find_starting_index($index);
 684     my $end = find_ending_index($index);
 685
 686     my $role;
 687     my $subsystem = $typevalue[$start];
 688     if (length($subsystem) > 20) {
 689         $subsystem = substr($subsystem, 0, 17);
 690         $subsystem =~ s/\s*$//;
 691         $subsystem = $subsystem . "...";
 692     }
 693
 694     for ($i = $start + 1; $i < $end; $i++) {
 695         my $tv = $typevalue[$i];
 696         if ($tv =~ m/^(\C):\s*(.*)/) {
 697             my $ptype = $1;
 698             my $pvalue = $2;
 699             if ($ptype eq "S") {
 700                 $role = $pvalue;
 701             }
 702         }
 703     }
 704
 705     $role = lc($role);
 706     if      ($role eq "supported") {
 707         $role = "supporter";
 708     } elsif ($role eq "maintained") {
 709         $role = "maintainer";
 710     } elsif ($role eq "odd fixes") {
 711         $role = "odd fixer";
 712     } elsif ($role eq "orphan") {
 713         $role = "orphan minder";
 714     } elsif ($role eq "obsolete") {
 715         $role = "obsolete minder";
 716     } elsif ($role eq "buried alive in reporters") {
 717         $role = "chief penguin";
 718     }
 719
 720     return $role . ":" . $subsystem;
 721 }
 722
 723 sub get_list_role {
 724     my ($index) = @_;
 725
 726     my $i;
 727     my $start = find_starting_index($index);
 728     my $end = find_ending_index($index);
 729
 730     my $subsystem = $typevalue[$start];
 731     if (length($subsystem) > 20) {
 732         $subsystem = substr($subsystem, 0, 17);
 733         $subsystem =~ s/\s*$//;
 734         $subsystem = $subsystem . "...";
 735     }
 736
 737     if ($subsystem eq "THE REST") {
 738         $subsystem = "";
 739     }
 740
 741     return $subsystem;
 742 }
 743
 744 sub add_categories {
 745     my ($index) = @_;
 746
 747     my $i;
 748     my $start = find_starting_index($index);
 749     my $end = find_ending_index($index);
 750
 751     push(@subsystem, $typevalue[$start]);
 752
 753     for ($i = $start + 1; $i < $end; $i++) {
 754         my $tv = $typevalue[$i];
 755         if ($tv =~ m/^(\C):\s*(.*)/) {
 756             my $ptype = $1;
 757             my $pvalue = $2;
 758             if ($ptype eq "L") {
 759                 my $list_address = $pvalue;
 760                 my $list_additional = "";
 761                 my $list_role = get_list_role($i);
 762
 763                 if ($list_role ne "") {
 764                     $list_role = ":" . $list_role;
 765                 }
 766                 if ($list_address =~ m/([^\s]+)\s+(.*)$/) {
 767                     $list_address = $1;
 768                     $list_additional = $2;
 769                 }
 770                 if ($list_additional =~ m/subscribers-only/) {
 771                     if ($email_subscriber_list) {
 772                         push(@list_to, [$list_address, "subscriber list${list_role}"]);
 773                     }
 774                 } else {
 775                     if ($email_list) {
 776                         push(@list_to, [$list_address, "open list${list_role}"]);
 777                     }
 778                 }
 779             } elsif ($ptype eq "M") {
 780                 my ($name, $address) = parse_email($pvalue);
 781                 if ($name eq "") {
 782                     if ($i > 0) {
 783                         my $tv = $typevalue[$i - 1];
 784                         if ($tv =~ m/^(\C):\s*(.*)/) {
 785                             if ($1 eq "P") {
 786                                 $name = $2;
 787                                 $pvalue = format_email($name, $address, $email_usename);
 788                             }
 789                         }
 790                     }
 791                 }
 792                 if ($email_maintainer) {
 793                     my $role = get_maintainer_role($i);
 794                     push_email_addresses($pvalue, $role);
 795                 }
 796             } elsif ($ptype eq "T") {
 797                 push(@scm, $pvalue);
 798             } elsif ($ptype eq "W") {
 799                 push(@web, $pvalue);
 800             } elsif ($ptype eq "S") {
 801                 push(@status, $pvalue);
 802             }
 803         }
 804     }
 805 }
 806
 807 my %email_hash_name;
 808 my %email_hash_address;
 809
 810 sub email_inuse {
 811     my ($name, $address) = @_;
 812
 813     return 1 if (($name eq "") && ($address eq ""));
 814     return 1 if (($name ne "") && exists($email_hash_name{$name}));
 815     return 1 if (($address ne "") && exists($email_hash_address{$address}));
 816
 817     return 0;
 818 }
 819
 820 sub push_email_address {
 821     my ($line, $role) = @_;
 822
 823     my ($name, $address) = parse_email($line);
 824
 825     if ($address eq "") {
 826         return 0;
 827     }
 828
 829     if (!$email_remove_duplicates) {
 830         push(@email_to, [format_email($name, $address, $email_usename), $role]);
 831     } elsif (!email_inuse($name, $address)) {
 832         push(@email_to, [format_email($name, $address, $email_usename), $role]);
 833         $email_hash_name{$name}++;
 834         $email_hash_address{$address}++;
 835     }
 836
 837     return 1;
 838 }
 839
 840 sub push_email_addresses {
 841     my ($address, $role) = @_;
 842
 843     my @address_list = ();
 844
 845     if (rfc822_valid($address)) {
 846         push_email_address($address, $role);
 847     } elsif (@address_list = rfc822_validlist($address)) {
 848         my $array_count = shift(@address_list);
 849         while (my $entry = shift(@address_list)) {
 850             push_email_address($entry, $role);
 851         }
 852     } else {
 853         if (!push_email_address($address, $role)) {
 854             warn("Invalid MAINTAINERS address: '" . $address . "'\n");
 855         }
 856     }
 857 }
 858
 859 sub add_role {
 860     my ($line, $role) = @_;
 861
 862     my ($name, $address) = parse_email($line);
 863     my $email = format_email($name, $address, $email_usename);
 864
 865     foreach my $entry (@email_to) {
 866         if ($email_remove_duplicates) {
 867             my ($entry_name, $entry_address) = parse_email($entry->[0]);
 868             if (($name eq $entry_name || $address eq $entry_address)
 869                 && ($role eq "" || !($entry->[1] =~ m/$role/))
 870             ) {
 871                 if ($entry->[1] eq "") {
 872                     $entry->[1] = "$role";
 873                 } else {
 874                     $entry->[1] = "$entry->[1],$role";
 875                 }
 876             }
 877         } else {
 878             if ($email eq $entry->[0]
 879                 && ($role eq "" || !($entry->[1] =~ m/$role/))
 880             ) {
 881                 if ($entry->[1] eq "") {
 882                     $entry->[1] = "$role";
 883                 } else {
 884                     $entry->[1] = "$entry->[1],$role";
 885                 }
 886             }
 887         }
 888     }
 889 }
 890
 891 sub which {
 892     my ($bin) = @_;
 893
 894     foreach my $path (split(/:/, $ENV{PATH})) {
 895         if (-e "$path/$bin") {
 896             return "$path/$bin";
 897         }
 898     }
 899
 900     return "";
 901 }
 902
 903 sub mailmap {
 904     my (@lines) = @_;
 905     my %hash;
 906
 907     foreach my $line (@lines) {
 908         my ($name, $address) = parse_email($line);
 909         if (!exists($hash{$name})) {
 910             $hash{$name} = $address;
 911         } elsif ($address ne $hash{$name}) {
 912             $address = $hash{$name};
 913             $line = format_email($name, $address, $email_usename);
 914         }
 915         if (exists($mailmap{$name})) {
 916             my $obj = $mailmap{$name};
 917             foreach my $map_address (@$obj) {
 918                 if (($map_address eq $address) &&
 919                     ($map_address ne $hash{$name})) {
 920                     $line = format_email($name, $hash{$name}, $email_usename);
 921                 }
 922             }
 923         }
 924     }
 925
 926     return @lines;
 927 }
 928
 929 sub git_execute_cmd {
 930     my ($cmd) = @_;
 931     my @lines = ();
 932
 933     my $output = `$cmd`;
 934     $output =~ s/^\s*//gm;
 935     @lines = split("\n", $output);
 936
 937     return @lines;
 938 }
 939
 940 sub hg_execute_cmd {
 941     my ($cmd) = @_;
 942     my @lines = ();
 943
 944     my $output = `$cmd`;
 945     @lines = split("\n", $output);
 946
 947     return @lines;
 948 }
 949
 950 sub vcs_find_signers {
 951     my ($cmd) = @_;
 952     my @lines = ();
 953     my $commits;
 954
 955     @lines = &{$VCS_cmds{"execute_cmd"}}($cmd);
 956
 957     my $pattern = $VCS_cmds{"commit_pattern"};
 958
 959     $commits = grep(/$pattern/, @lines);        # of commits
 960
 961     @lines = grep(/^[-_         a-z]+by:.*\@.*$/i, @lines);
 962     if (!$email_git_penguin_chiefs) {
 963         @lines = grep(!/${penguin_chiefs}/i, @lines);
 964     }
 965     # cut -f2- -d":"
 966     s/.*:\s*(.+)\s*/$1/ for (@lines);
 967
 968 ## Reformat email addresses (with names) to avoid badly written signatures
 969
 970     foreach my $line (@lines) {
 971         my ($name, $address) = parse_email($line);
 972         $line = format_email($name, $address, 1);
 973     }
 974
 975     return ($commits, @lines);
 976 }
 977
 978 sub vcs_save_commits {
 979     my ($cmd) = @_;
 980     my @lines = ();
 981     my @commits = ();
 982
 983     @lines = &{$VCS_cmds{"execute_cmd"}}($cmd);
 984
 985     foreach my $line (@lines) {
 986         if ($line =~ m/$VCS_cmds{"blame_commit_pattern"}/) {
 987             push(@commits, $1);
 988         }
 989     }
 990
 991     return @commits;
 992 }
 993
 994 sub vcs_blame {
 995     my ($file) = @_;
 996     my $cmd;
 997     my @commits = ();
 998
 999     return @commits if (!(-f $file));
1000
1001     if (@range && $VCS_cmds{"blame_range_cmd"} eq "") {
1002         my @all_commits = ();
1003
1004         $cmd = $VCS_cmds{"blame_file_cmd"};
1005         $cmd =~ s/(\$\w+)/$1/eeg;               #interpolate $cmd
1006         @all_commits = vcs_save_commits($cmd);
1007
1008         foreach my $file_range_diff (@range) {
1009             next if (!($file_range_diff =~ m/(.+):(.+):(.+)/));
1010             my $diff_file = $1;
1011             my $diff_start = $2;
1012             my $diff_length = $3;
1013             next if ("$file" ne "$diff_file");
1014             for (my $i = $diff_start; $i < $diff_start + $diff_length; $i++) {
1015                 push(@commits, $all_commits[$i]);
1016             }
1017         }
1018     } elsif (@range) {
1019         foreach my $file_range_diff (@range) {
1020             next if (!($file_range_diff =~ m/(.+):(.+):(.+)/));
1021             my $diff_file = $1;
1022             my $diff_start = $2;
1023             my $diff_length = $3;
1024             next if ("$file" ne "$diff_file");
1025             $cmd = $VCS_cmds{"blame_range_cmd"};
1026             $cmd =~ s/(\$\w+)/$1/eeg;           #interpolate $cmd
1027             push(@commits, vcs_save_commits($cmd));
1028         }
1029     } else {
1030         $cmd = $VCS_cmds{"blame_file_cmd"};
1031         $cmd =~ s/(\$\w+)/$1/eeg;               #interpolate $cmd
1032         @commits = vcs_save_commits($cmd);
1033     }
1034
1035     return @commits;
1036 }
1037
1038 my $printed_novcs = 0;
1039 sub vcs_exists {
1040     %VCS_cmds = %VCS_cmds_git;
1041     return 1 if eval $VCS_cmds{"available"};
1042     %VCS_cmds = %VCS_cmds_hg;
1043     return 1 if eval $VCS_cmds{"available"};
1044     %VCS_cmds = ();
1045     if (!$printed_novcs) {
1046         warn("$P: No supported VCS found.  Add --nogit to options?\n");
1047         warn("Using a git repository produces better results.\n");
1048         warn("Try Linus Torvalds' latest git repository using:\n");
1049         warn("git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git\n");
1050         $printed_novcs = 1;
1051     }
1052     return 0;
1053 }
1054
1055 sub vcs_assign {
1056     my ($role, $divisor, @lines) = @_;
1057
1058     my %hash;
1059     my $count = 0;
1060
1061     return if (@lines <= 0);
1062
1063     if ($divisor <= 0) {
1064         warn("Bad divisor in " . (caller(0))[3] . ": $divisor\n");
1065         $divisor = 1;
1066     }
1067
1068     if ($email_remove_duplicates) {
1069         @lines = mailmap(@lines);
1070     }
1071
1072     @lines = sort(@lines);
1073
1074     # uniq -c
1075     $hash{$_}++ for @lines;
1076
1077     # sort -rn
1078     foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) {
1079         my $sign_offs = $hash{$line};
1080         my $percent = $sign_offs * 100 / $divisor;
1081
1082         $percent = 100 if ($percent > 100);
1083         $count++;
1084         last if ($sign_offs < $email_git_min_signatures ||
1085                  $count > $email_git_max_maintainers ||
1086                  $percent < $email_git_min_percent);
1087         push_email_address($line, '');
1088         if ($output_rolestats) {
1089             my $fmt_percent = sprintf("%.0f", $percent);
1090             add_role($line, "$role:$sign_offs/$divisor=$fmt_percent%");
1091         } else {
1092             add_role($line, $role);
1093         }
1094     }
1095 }
1096
1097 sub vcs_file_signoffs {
1098     my ($file) = @_;
1099
1100     my @signers = ();
1101     my $commits;
1102
1103     return if (!vcs_exists());
1104
1105     my $cmd = $VCS_cmds{"find_signers_cmd"};
1106     $cmd =~ s/(\$\w+)/$1/eeg;           # interpolate $cmd
1107
1108     ($commits, @signers) = vcs_find_signers($cmd);
1109     vcs_assign("commit_signer", $commits, @signers);
1110 }
1111
1112 sub vcs_file_blame {
1113     my ($file) = @_;
1114
1115     my @signers = ();
1116     my @commits = ();
1117     my $total_commits;
1118
1119     return if (!vcs_exists());
1120
1121     @commits = vcs_blame($file);
1122     @commits = uniq(@commits);
1123     $total_commits = @commits;
1124
1125     foreach my $commit (@commits) {
1126         my $commit_count;
1127         my @commit_signers = ();
1128
1129         my $cmd = $VCS_cmds{"find_commit_signers_cmd"};
1130         $cmd =~ s/(\$\w+)/$1/eeg;       #interpolate $cmd
1131
1132         ($commit_count, @commit_signers) = vcs_find_signers($cmd);
1133         push(@signers, @commit_signers);
1134     }
1135
1136     if ($from_filename) {
1137         vcs_assign("commits", $total_commits, @signers);
1138     } else {
1139         vcs_assign("modified commits", $total_commits, @signers);
1140     }
1141 }
1142
1143 sub uniq {
1144     my (@parms) = @_;
1145
1146     my %saw;
1147     @parms = grep(!$saw{$_}++, @parms);
1148     return @parms;
1149 }
1150
1151 sub sort_and_uniq {
1152     my (@parms) = @_;
1153
1154     my %saw;
1155     @parms = sort @parms;
1156     @parms = grep(!$saw{$_}++, @parms);
1157     return @parms;
1158 }
1159
1160 sub clean_file_emails {
1161     my (@file_emails) = @_;
1162     my @fmt_emails = ();
1163
1164     foreach my $email (@file_emails) {
1165         $email =~ s/[\(\<\{]{0,1}([A-Za-z0-9_\.\+-]+\@[A-Za-z0-9\.-]+)[\)\>\}]{0,1}/\<$1\>/g;
1166         my ($name, $address) = parse_email($email);
1167         if ($name eq '"[,\.]"') {
1168             $name = "";
1169         }
1170
1171         my @nw = split(/[^A-Za-zÀ-ÿ\'\,\.\+-]/, $name);
1172         if (@nw > 2) {
1173             my $first = $nw[@nw - 3];
1174             my $middle = $nw[@nw - 2];
1175             my $last = $nw[@nw - 1];
1176
1177             if (((length($first) == 1 && $first =~ m/[A-Za-z]/) ||
1178                  (length($first) == 2 && substr($first, -1) eq ".")) ||
1179                 (length($middle) == 1 ||
1180                  (length($middle) == 2 && substr($middle, -1) eq "."))) {
1181                 $name = "$first $middle $last";
1182             } else {
1183                 $name = "$middle $last";
1184             }
1185         }
1186
1187         if (substr($name, -1) =~ /[,\.]/) {
1188             $name = substr($name, 0, length($name) - 1);
1189         } elsif (substr($name, -2) =~ /[,\.]"/) {
1190             $name = substr($name, 0, length($name) - 2) . '"';
1191         }
1192
1193         if (substr($name, 0, 1) =~ /[,\.]/) {
1194             $name = substr($name, 1, length($name) - 1);
1195         } elsif (substr($name, 0, 2) =~ /"[,\.]/) {
1196             $name = '"' . substr($name, 2, length($name) - 2);
1197         }
1198
1199         my $fmt_email = format_email($name, $address, $email_usename);
1200         push(@fmt_emails, $fmt_email);
1201     }
1202     return @fmt_emails;
1203 }
1204
1205 sub merge_email {
1206     my @lines;
1207     my %saw;
1208
1209     for (@_) {
1210         my ($address, $role) = @$_;
1211         if (!$saw{$address}) {
1212             if ($output_roles) {
1213                 push(@lines, "$address ($role)");
1214             } else {
1215                 push(@lines, $address);
1216             }
1217             $saw{$address} = 1;
1218         }
1219     }
1220
1221     return @lines;
1222 }
1223
1224 sub output {
1225     my (@parms) = @_;
1226
1227     if ($output_multiline) {
1228         foreach my $line (@parms) {
1229             print("${line}\n");
1230         }
1231     } else {
1232         print(join($output_separator, @parms));
1233         print("\n");
1234     }
1235 }
1236
1237 my $rfc822re;
1238
1239 sub make_rfc822re {
1240 #   Basic lexical tokens are specials, domain_literal, quoted_string, atom, and
1241 #   comment.  We must allow for rfc822_lwsp (or comments) after each of these.
1242 #   This regexp will only work on addresses which have had comments stripped
1243 #   and replaced with rfc822_lwsp.
1244
1245     my $specials = '()<>@,;:\\\\".\\[\\]';
1246     my $controls = '\\000-\\037\\177';
1247
1248     my $dtext = "[^\\[\\]\\r\\\\]";
1249     my $domain_literal = "\\[(?:$dtext|\\\\.)*\\]$rfc822_lwsp*";
1250
1251     my $quoted_string = "\"(?:[^\\\"\\r\\\\]|\\\\.|$rfc822_lwsp)*\"$rfc822_lwsp*";
1252
1253 #   Use zero-width assertion to spot the limit of an atom.  A simple
1254 #   $rfc822_lwsp* causes the regexp engine to hang occasionally.
1255     my $atom = "[^$specials $controls]+(?:$rfc822_lwsp+|\\Z|(?=[\\[\"$specials]))";
1256     my $word = "(?:$atom|$quoted_string)";
1257     my $localpart = "$word(?:\\.$rfc822_lwsp*$word)*";
1258
1259     my $sub_domain = "(?:$atom|$domain_literal)";
1260     my $domain = "$sub_domain(?:\\.$rfc822_lwsp*$sub_domain)*";
1261
1262     my $addr_spec = "$localpart\@$rfc822_lwsp*$domain";
1263
1264     my $phrase = "$word*";
1265     my $route = "(?:\@$domain(?:,\@$rfc822_lwsp*$domain)*:$rfc822_lwsp*)";
1266     my $route_addr = "\\<$rfc822_lwsp*$route?$addr_spec\\>$rfc822_lwsp*";
1267     my $mailbox = "(?:$addr_spec|$phrase$route_addr)";
1268
1269     my $group = "$phrase:$rfc822_lwsp*(?:$mailbox(?:,\\s*$mailbox)*)?;\\s*";
1270     my $address = "(?:$mailbox|$group)";
1271
1272     return "$rfc822_lwsp*$address";
1273 }
1274
1275 sub rfc822_strip_comments {
1276     my $s = shift;
1277 #   Recursively remove comments, and replace with a single space.  The simpler
1278 #   regexps in the Email Addressing FAQ are imperfect - they will miss escaped
1279 #   chars in atoms, for example.
1280
1281     while ($s =~ s/^((?:[^"\\]|\\.)*
1282                     (?:"(?:[^"\\]|\\.)*"(?:[^"\\]|\\.)*)*)
1283                     \((?:[^()\\]|\\.)*\)/$1 /osx) {}
1284     return $s;
1285 }
1286
1287 #   valid: returns true if the parameter is an RFC822 valid address
1288 #
1289 sub rfc822_valid ($) {
1290     my $s = rfc822_strip_comments(shift);
1291
1292     if (!$rfc822re) {
1293         $rfc822re = make_rfc822re();
1294     }
1295
1296     return $s =~ m/^$rfc822re$/so && $s =~ m/^$rfc822_char*$/;
1297 }
1298
1299 #   validlist: In scalar context, returns true if the parameter is an RFC822
1300 #              valid list of addresses.
1301 #
1302 #              In list context, returns an empty list on failure (an invalid
1303 #              address was found); otherwise a list whose first element is the
1304 #              number of addresses found and whose remaining elements are the
1305 #              addresses.  This is needed to disambiguate failure (invalid)
1306 #              from success with no addresses found, because an empty string is
1307 #              a valid list.
1308
1309 sub rfc822_validlist ($) {
1310     my $s = rfc822_strip_comments(shift);
1311
1312     if (!$rfc822re) {
1313         $rfc822re = make_rfc822re();
1314     }
1315     # * null list items are valid according to the RFC
1316     # * the '1' business is to aid in distinguishing failure from no results
1317
1318     my @r;
1319     if ($s =~ m/^(?:$rfc822re)?(?:,(?:$rfc822re)?)*$/so &&
1320         $s =~ m/^$rfc822_char*$/) {
1321         while ($s =~ m/(?:^|,$rfc822_lwsp*)($rfc822re)/gos) {
1322             push(@r, $1);
1323         }
1324         return wantarray ? (scalar(@r), @r) : 1;
1325     }
1326     return wantarray ? () : 0;
1327 }