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_all_signature_types = 1;
  29 my $email_git_blame = 0;
  30 my $email_git_min_signatures = 1;
  31 my $email_git_max_maintainers = 5;
  32 my $email_git_min_percent = 5;
  33 my $email_git_since = "1-year-ago";
  34 my $email_hg_since = "-365";
  35 my $email_remove_duplicates = 1;
  36 my $output_multiline = 1;
  37 my $output_separator = ", ";
  38 my $output_roles = 0;
  39 my $output_rolestats = 0;
  40 my $scm = 0;
  41 my $web = 0;
  42 my $subsystem = 0;
  43 my $status = 0;
  44 my $keywords = 1;
  45 my $sections = 0;
  46 my $file_emails = 0;
  47 my $from_filename = 0;
  48 my $pattern_depth = 0;
  49 my $version = 0;
  50 my $help = 0;
  51
  52 my $exit = 0;
  53
  54 my @penguin_chief = ();
  55 push(@penguin_chief, "Linus Torvalds:torvalds\@linux-foundation.org");
  56 #Andrew wants in on most everything - 2009/01/14
  57 #push(@penguin_chief, "Andrew Morton:akpm\@linux-foundation.org");
  58
  59 my @penguin_chief_names = ();
  60 foreach my $chief (@penguin_chief) {
  61     if ($chief =~ m/^(.*):(.*)/) {
  62         my $chief_name = $1;
  63         my $chief_addr = $2;
  64         push(@penguin_chief_names, $chief_name);
  65     }
  66 }
  67 my $penguin_chiefs = "\(" . join("|", @penguin_chief_names) . "\)";
  68
  69 # Signature types of people who are either
  70 #       a) responsible for the code in question, or
  71 #       b) familiar enough with it to give relevant feedback
  72 my @signature_tags = ();
  73 push(@signature_tags, "Signed-off-by:");
  74 push(@signature_tags, "Reviewed-by:");
  75 push(@signature_tags, "Acked-by:");
  76 my $signaturePattern = "\(" . join("|", @signature_tags) . "\)";
  77
  78 # rfc822 email address - preloaded methods go here.
  79 my $rfc822_lwsp = "(?:(?:\\r\\n)?[ \\t])";
  80 my $rfc822_char = '[\\000-\\377]';
  81
  82 # VCS command support: class-like functions and strings
  83
  84 my %VCS_cmds;
  85
  86 my %VCS_cmds_git = (
  87     "execute_cmd" => \&git_execute_cmd,
  88     "available" => '(which("git") ne "") && (-d ".git")',
  89     "find_signers_cmd" => "git log --no-color --since=\$email_git_since -- \$file",
  90     "find_commit_signers_cmd" => "git log --no-color -1 \$commit",
  91     "blame_range_cmd" => "git blame -l -L \$diff_start,+\$diff_length \$file",
  92     "blame_file_cmd" => "git blame -l \$file",
  93     "commit_pattern" => "^commit [0-9a-f]{40,40}",
  94     "blame_commit_pattern" => "^([0-9a-f]+) "
  95 );
  96
  97 my %VCS_cmds_hg = (
  98     "execute_cmd" => \&hg_execute_cmd,
  99     "available" => '(which("hg") ne "") && (-d ".hg")',
 100     "find_signers_cmd" =>
 101         "hg log --date=\$email_hg_since" .
 102                 " --template='commit {node}\\n{desc}\\n' -- \$file",
 103     "find_commit_signers_cmd" => "hg log --template='{desc}\\n' -r \$commit",
 104     "blame_range_cmd" => "",            # not supported
 105     "blame_file_cmd" => "hg blame -c \$file",
 106     "commit_pattern" => "^commit [0-9a-f]{40,40}",
 107     "blame_commit_pattern" => "^([0-9a-f]+):"
 108 );
 109
 110 if (!GetOptions(
 111                 'email!' => \$email,
 112                 'git!' => \$email_git,
 113                 'git-all-signature-types!' => \$email_git_all_signature_types,
 114                 'git-blame!' => \$email_git_blame,
 115                 'git-chief-penguins!' => \$email_git_penguin_chiefs,
 116                 'git-min-signatures=i' => \$email_git_min_signatures,
 117                 'git-max-maintainers=i' => \$email_git_max_maintainers,
 118                 'git-min-percent=i' => \$email_git_min_percent,
 119                 'git-since=s' => \$email_git_since,
 120                 'hg-since=s' => \$email_hg_since,
 121                 'remove-duplicates!' => \$email_remove_duplicates,
 122                 'm!' => \$email_maintainer,
 123                 'n!' => \$email_usename,
 124                 'l!' => \$email_list,
 125                 's!' => \$email_subscriber_list,
 126                 'multiline!' => \$output_multiline,
 127                 'roles!' => \$output_roles,
 128                 'rolestats!' => \$output_rolestats,
 129                 'separator=s' => \$output_separator,
 130                 'subsystem!' => \$subsystem,
 131                 'status!' => \$status,
 132                 'scm!' => \$scm,
 133                 'web!' => \$web,
 134                 'pattern-depth=i' => \$pattern_depth,
 135                 'k|keywords!' => \$keywords,
 136                 'sections!' => \$sections,
 137                 'fe|file-emails!' => \$file_emails,
 138                 'f|file' => \$from_filename,
 139                 'v|version' => \$version,
 140                 'h|help|usage' => \$help,
 141                 )) {
 142     die "$P: invalid argument - use --help if necessary\n";
 143 }
 144
 145 if ($help != 0) {
 146     usage();
 147     exit 0;
 148 }
 149
 150 if ($version != 0) {
 151     print("${P} ${V}\n");
 152     exit 0;
 153 }
 154
 155 if (-t STDIN && !@ARGV) {
 156     # We're talking to a terminal, but have no command line arguments.
 157     die "$P: missing patchfile or -f file - use --help if necessary\n";
 158 }
 159
 160 if ($output_separator ne ", ") {
 161     $output_multiline = 0;
 162 }
 163
 164 if ($output_rolestats) {
 165     $output_roles = 1;
 166 }
 167
 168 if ($sections) {
 169     $email = 0;
 170     $email_list = 0;
 171     $scm = 0;
 172     $status = 0;
 173     $subsystem = 0;
 174     $web = 0;
 175     $keywords = 0;
 176 } else {
 177     my $selections = $email + $scm + $status + $subsystem + $web;
 178     if ($selections == 0) {
 179         die "$P:  Missing required option: email, scm, status, subsystem or web\n";
 180     }
 181 }
 182
 183 if ($email &&
 184     ($email_maintainer + $email_list + $email_subscriber_list +
 185      $email_git + $email_git_penguin_chiefs + $email_git_blame) == 0) {
 186     die "$P: Please select at least 1 email option\n";
 187 }
 188
 189 if (!top_of_kernel_tree($lk_path)) {
 190     die "$P: The current directory does not appear to be "
 191         . "a linux kernel source tree.\n";
 192 }
 193
 194 if ($email_git_all_signature_types) {
 195     $signaturePattern = "(.+?)[Bb][Yy]:";
 196 }
 197
 198 ## Read MAINTAINERS for type/value pairs
 199
 200 my @typevalue = ();
 201 my %keyword_hash;
 202
 203 open (my $maint, '<', "${lk_path}MAINTAINERS")
 204     or die "$P: Can't open MAINTAINERS: $!\n";
 205 while (<$maint>) {
 206     my $line = $_;
 207
 208     if ($line =~ m/^(\C):\s*(.*)/) {
 209         my $type = $1;
 210         my $value = $2;
 211
 212         ##Filename pattern matching
 213         if ($type eq "F" || $type eq "X") {
 214             $value =~ s@\.@\\\.@g;       ##Convert . to \.
 215             $value =~ s/\*/\.\*/g;       ##Convert * to .*
 216             $value =~ s/\?/\./g;         ##Convert ? to .
 217             ##if pattern is a directory and it lacks a trailing slash, add one
 218             if ((-d $value)) {
 219                 $value =~ s@([^/])$@$1/@;
 220             }
 221         } elsif ($type eq "K") {
 222             $keyword_hash{@typevalue} = $value;
 223         }
 224         push(@typevalue, "$type:$value");
 225     } elsif (!/^(\s)*$/) {
 226         $line =~ s/\n$//g;
 227         push(@typevalue, $line);
 228     }
 229 }
 230 close($maint);
 231
 232 my %mailmap;
 233
 234 if ($email_remove_duplicates) {
 235     open(my $mailmap, '<', "${lk_path}.mailmap")
 236         or warn "$P: Can't open .mailmap: $!\n";
 237     while (<$mailmap>) {
 238         my $line = $_;
 239
 240         next if ($line =~ m/^\s*#/);
 241         next if ($line =~ m/^\s*$/);
 242
 243         my ($name, $address) = parse_email($line);
 244         $line = format_email($name, $address, $email_usename);
 245
 246         next if ($line =~ m/^\s*$/);
 247
 248         if (exists($mailmap{$name})) {
 249             my $obj = $mailmap{$name};
 250             push(@$obj, $address);
 251         } else {
 252             my @arr = ($address);
 253             $mailmap{$name} = \@arr;
 254         }
 255     }
 256     close($mailmap);
 257 }
 258
 259 ## use the filenames on the command line or find the filenames in the patchfiles
 260
 261 my @files = ();
 262 my @range = ();
 263 my @keyword_tvi = ();
 264 my @file_emails = ();
 265
 266 if (!@ARGV) {
 267     push(@ARGV, "&STDIN");
 268 }
 269
 270 foreach my $file (@ARGV) {
 271     if ($file ne "&STDIN") {
 272         ##if $file is a directory and it lacks a trailing slash, add one
 273         if ((-d $file)) {
 274             $file =~ s@([^/])$@$1/@;
 275         } elsif (!(-f $file)) {
 276             die "$P: file '${file}' not found\n";
 277         }
 278     }
 279     if ($from_filename) {
 280         push(@files, $file);
 281         if (-f $file && ($keywords || $file_emails)) {
 282             open(my $f, '<', $file)
 283                 or die "$P: Can't open $file: $!\n";
 284             my $text = do { local($/) ; <$f> };
 285             close($f);
 286             if ($keywords) {
 287                 foreach my $line (keys %keyword_hash) {
 288                     if ($text =~ m/$keyword_hash{$line}/x) {
 289                         push(@keyword_tvi, $line);
 290                     }
 291                 }
 292             }
 293             if ($file_emails) {
 294                 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;
 295                 push(@file_emails, clean_file_emails(@poss_addr));
 296             }
 297         }
 298     } else {
 299         my $file_cnt = @files;
 300         my $lastfile;
 301
 302         open(my $patch, "< $file")
 303             or die "$P: Can't open $file: $!\n";
 304         while (<$patch>) {
 305             my $patch_line = $_;
 306             if (m/^\+\+\+\s+(\S+)/) {
 307                 my $filename = $1;
 308                 $filename =~ s@^[^/]*/@@;
 309                 $filename =~ s@\n@@;
 310                 $lastfile = $filename;
 311                 push(@files, $filename);
 312             } elsif (m/^\@\@ -(\d+),(\d+)/) {
 313                 if ($email_git_blame) {
 314                     push(@range, "$lastfile:$1:$2");
 315                 }
 316             } elsif ($keywords) {
 317                 foreach my $line (keys %keyword_hash) {
 318                     if ($patch_line =~ m/^[+-].*$keyword_hash{$line}/x) {
 319                         push(@keyword_tvi, $line);
 320                     }
 321                 }
 322             }
 323         }
 324         close($patch);
 325
 326         if ($file_cnt == @files) {
 327             warn "$P: file '${file}' doesn't appear to be a patch.  "
 328                 . "Add -f to options?\n";
 329         }
 330         @files = sort_and_uniq(@files);
 331     }
 332 }
 333
 334 @file_emails = uniq(@file_emails);
 335
 336 my @email_to = ();
 337 my @list_to = ();
 338 my @scm = ();
 339 my @web = ();
 340 my @subsystem = ();
 341 my @status = ();
 342
 343 # Find responsible parties
 344
 345 foreach my $file (@files) {
 346
 347     my %hash;
 348     my $tvi = find_first_section();
 349     while ($tvi < @typevalue) {
 350         my $start = find_starting_index($tvi);
 351         my $end = find_ending_index($tvi);
 352         my $exclude = 0;
 353         my $i;
 354
 355         #Do not match excluded file patterns
 356
 357         for ($i = $start; $i < $end; $i++) {
 358             my $line = $typevalue[$i];
 359             if ($line =~ m/^(\C):\s*(.*)/) {
 360                 my $type = $1;
 361                 my $value = $2;
 362                 if ($type eq 'X') {
 363                     if (file_match_pattern($file, $value)) {
 364                         $exclude = 1;
 365                         last;
 366                     }
 367                 }
 368             }
 369         }
 370
 371         if (!$exclude) {
 372             for ($i = $start; $i < $end; $i++) {
 373                 my $line = $typevalue[$i];
 374                 if ($line =~ m/^(\C):\s*(.*)/) {
 375                     my $type = $1;
 376                     my $value = $2;
 377                     if ($type eq 'F') {
 378                         if (file_match_pattern($file, $value)) {
 379                             my $value_pd = ($value =~ tr@/@@);
 380                             my $file_pd = ($file  =~ tr@/@@);
 381                             $value_pd++ if (substr($value,-1,1) ne "/");
 382                             if ($pattern_depth == 0 ||
 383                                 (($file_pd - $value_pd) < $pattern_depth)) {
 384                                 $hash{$tvi} = $value_pd;
 385                             }
 386                         }
 387                     }
 388                 }
 389             }
 390         }
 391
 392         $tvi = $end + 1;
 393     }
 394
 395     foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) {
 396         add_categories($line);
 397             if ($sections) {
 398                 my $i;
 399                 my $start = find_starting_index($line);
 400                 my $end = find_ending_index($line);
 401                 for ($i = $start; $i < $end; $i++) {
 402                     my $line = $typevalue[$i];
 403                     if ($line =~ /^[FX]:/) {            ##Restore file patterns
 404                         $line =~ s/([^\\])\.([^\*])/$1\?$2/g;
 405                         $line =~ s/([^\\])\.$/$1\?/g;   ##Convert . back to ?
 406                         $line =~ s/\\\./\./g;           ##Convert \. to .
 407                         $line =~ s/\.\*/\*/g;           ##Convert .* to *
 408                     }
 409                     $line =~ s/^([A-Z]):/$1:\t/g;
 410                     print("$line\n");
 411                 }
 412                 print("\n");
 413             }
 414     }
 415
 416     if ($email && $email_git) {
 417         vcs_file_signoffs($file);
 418     }
 419
 420     if ($email && $email_git_blame) {
 421         vcs_file_blame($file);
 422     }
 423 }
 424
 425 if ($keywords) {
 426     @keyword_tvi = sort_and_uniq(@keyword_tvi);
 427     foreach my $line (@keyword_tvi) {
 428         add_categories($line);
 429     }
 430 }
 431
 432 if ($email) {
 433     foreach my $chief (@penguin_chief) {
 434         if ($chief =~ m/^(.*):(.*)/) {
 435             my $email_address;
 436
 437             $email_address = format_email($1, $2, $email_usename);
 438             if ($email_git_penguin_chiefs) {
 439                 push(@email_to, [$email_address, 'chief penguin']);
 440             } else {
 441                 @email_to = grep($_->[0] !~ /${email_address}/, @email_to);
 442             }
 443         }
 444     }
 445
 446     foreach my $email (@file_emails) {
 447         my ($name, $address) = parse_email($email);
 448
 449         my $tmp_email = format_email($name, $address, $email_usename);
 450         push_email_address($tmp_email, '');
 451         add_role($tmp_email, 'in file');
 452     }
 453 }
 454
 455 if ($email || $email_list) {
 456     my @to = ();
 457     if ($email) {
 458         @to = (@to, @email_to);
 459     }
 460     if ($email_list) {
 461         @to = (@to, @list_to);
 462     }
 463     output(merge_email(@to));
 464 }
 465
 466 if ($scm) {
 467     @scm = uniq(@scm);
 468     output(@scm);
 469 }
 470
 471 if ($status) {
 472     @status = uniq(@status);
 473     output(@status);
 474 }
 475
 476 if ($subsystem) {
 477     @subsystem = uniq(@subsystem);
 478     output(@subsystem);
 479 }
 480
 481 if ($web) {
 482     @web = uniq(@web);
 483     output(@web);
 484 }
 485
 486 exit($exit);
 487
 488 sub file_match_pattern {
 489     my ($file, $pattern) = @_;
 490     if (substr($pattern, -1) eq "/") {
 491         if ($file =~ m@^$pattern@) {
 492             return 1;
 493         }
 494     } else {
 495         if ($file =~ m@^$pattern@) {
 496             my $s1 = ($file =~ tr@/@@);
 497             my $s2 = ($pattern =~ tr@/@@);
 498             if ($s1 == $s2) {
 499                 return 1;
 500             }
 501         }
 502     }
 503     return 0;
 504 }
 505
 506 sub usage {
 507     print <<EOT;
 508 usage: $P [options] patchfile
 509        $P [options] -f file|directory
 510 version: $V
 511
 512 MAINTAINER field selection options:
 513   --email => print email address(es) if any
 514     --git => include recent git \*-by: signers
 515     --git-all-signature-types => include signers regardless of signature type
 516         or use only ${signaturePattern} signers (default: $email_git_all_signature_types)
 517     --git-chief-penguins => include ${penguin_chiefs}
 518     --git-min-signatures => number of signatures required (default: $email_git_min_signatures)
 519     --git-max-maintainers => maximum maintainers to add (default: $email_git_max_maintainers)
 520     --git-min-percent => minimum percentage of commits required (default: $email_git_min_percent)
 521     --git-blame => use git blame to find modified commits for patch or file
 522     --git-since => git history to use (default: $email_git_since)
 523     --hg-since => hg history to use (default: $email_hg_since)
 524     --m => include maintainer(s) if any
 525     --n => include name 'Full Name <addr\@domain.tld>'
 526     --l => include list(s) if any
 527     --s => include subscriber only list(s) if any
 528     --remove-duplicates => minimize duplicate email names/addresses
 529     --roles => show roles (status:subsystem, git-signer, list, etc...)
 530     --rolestats => show roles and statistics (commits/total_commits, %)
 531     --file-emails => add email addresses found in -f file (default: 0 (off))
 532   --scm => print SCM tree(s) if any
 533   --status => print status if any
 534   --subsystem => print subsystem name if any
 535   --web => print website(s) if any
 536
 537 Output type options:
 538   --separator [, ] => separator for multiple entries on 1 line
 539     using --separator also sets --nomultiline if --separator is not [, ]
 540   --multiline => print 1 entry per line
 541
 542 Other options:
 543   --pattern-depth => Number of pattern directory traversals (default: 0 (all))
 544   --keywords => scan patch for keywords (default: 1 (on))
 545   --sections => print the entire subsystem sections with pattern matches
 546   --version => show version
 547   --help => show this help information
 548
 549 Default options:
 550   [--email --git --m --n --l --multiline --pattern-depth=0 --remove-duplicates]
 551
 552 Notes:
 553   Using "-f directory" may give unexpected results:
 554       Used with "--git", git signators for _all_ files in and below
 555           directory are examined as git recurses directories.
 556           Any specified X: (exclude) pattern matches are _not_ ignored.
 557       Used with "--nogit", directory is used as a pattern match,
 558           no individual file within the directory or subdirectory
 559           is matched.
 560       Used with "--git-blame", does not iterate all files in directory
 561   Using "--git-blame" is slow and may add old committers and authors
 562       that are no longer active maintainers to the output.
 563   Using "--roles" or "--rolestats" with git send-email --cc-cmd or any
 564       other automated tools that expect only ["name"] <email address>
 565       may not work because of additional output after <email address>.
 566   Using "--rolestats" and "--git-blame" shows the #/total=% commits,
 567       not the percentage of the entire file authored.  # of commits is
 568       not a good measure of amount of code authored.  1 major commit may
 569       contain a thousand lines, 5 trivial commits may modify a single line.
 570   If git is not installed, but mercurial (hg) is installed and an .hg
 571       repository exists, the following options apply to mercurial:
 572           --git,
 573           --git-min-signatures, --git-max-maintainers, --git-min-percent, and
 574           --git-blame
 575       Use --hg-since not --git-since to control date selection
 576 EOT
 577 }
 578
 579 sub top_of_kernel_tree {
 580         my ($lk_path) = @_;
 581
 582         if ($lk_path ne "" && substr($lk_path,length($lk_path)-1,1) ne "/") {
 583             $lk_path .= "/";
 584         }
 585         if (   (-f "${lk_path}COPYING")
 586             && (-f "${lk_path}CREDITS")
 587             && (-f "${lk_path}Kbuild")
 588             && (-f "${lk_path}MAINTAINERS")
 589             && (-f "${lk_path}Makefile")
 590             && (-f "${lk_path}README")
 591             && (-d "${lk_path}Documentation")
 592             && (-d "${lk_path}arch")
 593             && (-d "${lk_path}include")
 594             && (-d "${lk_path}drivers")
 595             && (-d "${lk_path}fs")
 596             && (-d "${lk_path}init")
 597             && (-d "${lk_path}ipc")
 598             && (-d "${lk_path}kernel")
 599             && (-d "${lk_path}lib")
 600             && (-d "${lk_path}scripts")) {
 601                 return 1;
 602         }
 603         return 0;
 604 }
 605
 606 sub parse_email {
 607     my ($formatted_email) = @_;
 608
 609     my $name = "";
 610     my $address = "";
 611
 612     if ($formatted_email =~ /^([^<]+)<(.+\@.*)>.*$/) {
 613         $name = $1;
 614         $address = $2;
 615     } elsif ($formatted_email =~ /^\s*<(.+\@\S*)>.*$/) {
 616         $address = $1;
 617     } elsif ($formatted_email =~ /^(.+\@\S*).*$/) {
 618         $address = $1;
 619     }
 620
 621     $name =~ s/^\s+|\s+$//g;
 622     $name =~ s/^\"|\"$//g;
 623     $address =~ s/^\s+|\s+$//g;
 624
 625     if ($name =~ /[^\w \-]/i) {          ##has "must quote" chars
 626         $name =~ s/(?<!\\)"/\\"/g;       ##escape quotes
 627         $name = "\"$name\"";
 628     }
 629
 630     return ($name, $address);
 631 }
 632
 633 sub format_email {
 634     my ($name, $address, $usename) = @_;
 635
 636     my $formatted_email;
 637
 638     $name =~ s/^\s+|\s+$//g;
 639     $name =~ s/^\"|\"$//g;
 640     $address =~ s/^\s+|\s+$//g;
 641
 642     if ($name =~ /[^\w \-]/i) {          ##has "must quote" chars
 643         $name =~ s/(?<!\\)"/\\"/g;       ##escape quotes
 644         $name = "\"$name\"";
 645     }
 646
 647     if ($usename) {
 648         if ("$name" eq "") {
 649             $formatted_email = "$address";
 650         } else {
 651             $formatted_email = "$name <$address>";
 652         }
 653     } else {
 654         $formatted_email = $address;
 655     }
 656
 657     return $formatted_email;
 658 }
 659
 660 sub find_first_section {
 661     my $index = 0;
 662
 663     while ($index < @typevalue) {
 664         my $tv = $typevalue[$index];
 665         if (($tv =~ m/^(\C):\s*(.*)/)) {
 666             last;
 667         }
 668         $index++;
 669     }
 670
 671     return $index;
 672 }
 673
 674 sub find_starting_index {
 675     my ($index) = @_;
 676
 677     while ($index > 0) {
 678         my $tv = $typevalue[$index];
 679         if (!($tv =~ m/^(\C):\s*(.*)/)) {
 680             last;
 681         }
 682         $index--;
 683     }
 684
 685     return $index;
 686 }
 687
 688 sub find_ending_index {
 689     my ($index) = @_;
 690
 691     while ($index < @typevalue) {
 692         my $tv = $typevalue[$index];
 693         if (!($tv =~ m/^(\C):\s*(.*)/)) {
 694             last;
 695         }
 696         $index++;
 697     }
 698
 699     return $index;
 700 }
 701
 702 sub get_maintainer_role {
 703     my ($index) = @_;
 704
 705     my $i;
 706     my $start = find_starting_index($index);
 707     my $end = find_ending_index($index);
 708
 709     my $role;
 710     my $subsystem = $typevalue[$start];
 711     if (length($subsystem) > 20) {
 712         $subsystem = substr($subsystem, 0, 17);
 713         $subsystem =~ s/\s*$//;
 714         $subsystem = $subsystem . "...";
 715     }
 716
 717     for ($i = $start + 1; $i < $end; $i++) {
 718         my $tv = $typevalue[$i];
 719         if ($tv =~ m/^(\C):\s*(.*)/) {
 720             my $ptype = $1;
 721             my $pvalue = $2;
 722             if ($ptype eq "S") {
 723                 $role = $pvalue;
 724             }
 725         }
 726     }
 727
 728     $role = lc($role);
 729     if      ($role eq "supported") {
 730         $role = "supporter";
 731     } elsif ($role eq "maintained") {
 732         $role = "maintainer";
 733     } elsif ($role eq "odd fixes") {
 734         $role = "odd fixer";
 735     } elsif ($role eq "orphan") {
 736         $role = "orphan minder";
 737     } elsif ($role eq "obsolete") {
 738         $role = "obsolete minder";
 739     } elsif ($role eq "buried alive in reporters") {
 740         $role = "chief penguin";
 741     }
 742
 743     return $role . ":" . $subsystem;
 744 }
 745
 746 sub get_list_role {
 747     my ($index) = @_;
 748
 749     my $i;
 750     my $start = find_starting_index($index);
 751     my $end = find_ending_index($index);
 752
 753     my $subsystem = $typevalue[$start];
 754     if (length($subsystem) > 20) {
 755         $subsystem = substr($subsystem, 0, 17);
 756         $subsystem =~ s/\s*$//;
 757         $subsystem = $subsystem . "...";
 758     }
 759
 760     if ($subsystem eq "THE REST") {
 761         $subsystem = "";
 762     }
 763
 764     return $subsystem;
 765 }
 766
 767 sub add_categories {
 768     my ($index) = @_;
 769
 770     my $i;
 771     my $start = find_starting_index($index);
 772     my $end = find_ending_index($index);
 773
 774     push(@subsystem, $typevalue[$start]);
 775
 776     for ($i = $start + 1; $i < $end; $i++) {
 777         my $tv = $typevalue[$i];
 778         if ($tv =~ m/^(\C):\s*(.*)/) {
 779             my $ptype = $1;
 780             my $pvalue = $2;
 781             if ($ptype eq "L") {
 782                 my $list_address = $pvalue;
 783                 my $list_additional = "";
 784                 my $list_role = get_list_role($i);
 785
 786                 if ($list_role ne "") {
 787                     $list_role = ":" . $list_role;
 788                 }
 789                 if ($list_address =~ m/([^\s]+)\s+(.*)$/) {
 790                     $list_address = $1;
 791                     $list_additional = $2;
 792                 }
 793                 if ($list_additional =~ m/subscribers-only/) {
 794                     if ($email_subscriber_list) {
 795                         push(@list_to, [$list_address, "subscriber list${list_role}"]);
 796                     }
 797                 } else {
 798                     if ($email_list) {
 799                         push(@list_to, [$list_address, "open list${list_role}"]);
 800                     }
 801                 }
 802             } elsif ($ptype eq "M") {
 803                 my ($name, $address) = parse_email($pvalue);
 804                 if ($name eq "") {
 805                     if ($i > 0) {
 806                         my $tv = $typevalue[$i - 1];
 807                         if ($tv =~ m/^(\C):\s*(.*)/) {
 808                             if ($1 eq "P") {
 809                                 $name = $2;
 810                                 $pvalue = format_email($name, $address, $email_usename);
 811                             }
 812                         }
 813                     }
 814                 }
 815                 if ($email_maintainer) {
 816                     my $role = get_maintainer_role($i);
 817                     push_email_addresses($pvalue, $role);
 818                 }
 819             } elsif ($ptype eq "T") {
 820                 push(@scm, $pvalue);
 821             } elsif ($ptype eq "W") {
 822                 push(@web, $pvalue);
 823             } elsif ($ptype eq "S") {
 824                 push(@status, $pvalue);
 825             }
 826         }
 827     }
 828 }
 829
 830 my %email_hash_name;
 831 my %email_hash_address;
 832
 833 sub email_inuse {
 834     my ($name, $address) = @_;
 835
 836     return 1 if (($name eq "") && ($address eq ""));
 837     return 1 if (($name ne "") && exists($email_hash_name{$name}));
 838     return 1 if (($address ne "") && exists($email_hash_address{$address}));
 839
 840     return 0;
 841 }
 842
 843 sub push_email_address {
 844     my ($line, $role) = @_;
 845
 846     my ($name, $address) = parse_email($line);
 847
 848     if ($address eq "") {
 849         return 0;
 850     }
 851
 852     if (!$email_remove_duplicates) {
 853         push(@email_to, [format_email($name, $address, $email_usename), $role]);
 854     } elsif (!email_inuse($name, $address)) {
 855         push(@email_to, [format_email($name, $address, $email_usename), $role]);
 856         $email_hash_name{$name}++;
 857         $email_hash_address{$address}++;
 858     }
 859
 860     return 1;
 861 }
 862
 863 sub push_email_addresses {
 864     my ($address, $role) = @_;
 865
 866     my @address_list = ();
 867
 868     if (rfc822_valid($address)) {
 869         push_email_address($address, $role);
 870     } elsif (@address_list = rfc822_validlist($address)) {
 871         my $array_count = shift(@address_list);
 872         while (my $entry = shift(@address_list)) {
 873             push_email_address($entry, $role);
 874         }
 875     } else {
 876         if (!push_email_address($address, $role)) {
 877             warn("Invalid MAINTAINERS address: '" . $address . "'\n");
 878         }
 879     }
 880 }
 881
 882 sub add_role {
 883     my ($line, $role) = @_;
 884
 885     my ($name, $address) = parse_email($line);
 886     my $email = format_email($name, $address, $email_usename);
 887
 888     foreach my $entry (@email_to) {
 889         if ($email_remove_duplicates) {
 890             my ($entry_name, $entry_address) = parse_email($entry->[0]);
 891             if (($name eq $entry_name || $address eq $entry_address)
 892                 && ($role eq "" || !($entry->[1] =~ m/$role/))
 893             ) {
 894                 if ($entry->[1] eq "") {
 895                     $entry->[1] = "$role";
 896                 } else {
 897                     $entry->[1] = "$entry->[1],$role";
 898                 }
 899             }
 900         } else {
 901             if ($email eq $entry->[0]
 902                 && ($role eq "" || !($entry->[1] =~ m/$role/))
 903             ) {
 904                 if ($entry->[1] eq "") {
 905                     $entry->[1] = "$role";
 906                 } else {
 907                     $entry->[1] = "$entry->[1],$role";
 908                 }
 909             }
 910         }
 911     }
 912 }
 913
 914 sub which {
 915     my ($bin) = @_;
 916
 917     foreach my $path (split(/:/, $ENV{PATH})) {
 918         if (-e "$path/$bin") {
 919             return "$path/$bin";
 920         }
 921     }
 922
 923     return "";
 924 }
 925
 926 sub mailmap {
 927     my (@lines) = @_;
 928     my %hash;
 929
 930     foreach my $line (@lines) {
 931         my ($name, $address) = parse_email($line);
 932         if (!exists($hash{$name})) {
 933             $hash{$name} = $address;
 934         } elsif ($address ne $hash{$name}) {
 935             $address = $hash{$name};
 936             $line = format_email($name, $address, $email_usename);
 937         }
 938         if (exists($mailmap{$name})) {
 939             my $obj = $mailmap{$name};
 940             foreach my $map_address (@$obj) {
 941                 if (($map_address eq $address) &&
 942                     ($map_address ne $hash{$name})) {
 943                     $line = format_email($name, $hash{$name}, $email_usename);
 944                 }
 945             }
 946         }
 947     }
 948
 949     return @lines;
 950 }
 951
 952 sub git_execute_cmd {
 953     my ($cmd) = @_;
 954     my @lines = ();
 955
 956     my $output = `$cmd`;
 957     $output =~ s/^\s*//gm;
 958     @lines = split("\n", $output);
 959
 960     return @lines;
 961 }
 962
 963 sub hg_execute_cmd {
 964     my ($cmd) = @_;
 965     my @lines = ();
 966
 967     my $output = `$cmd`;
 968     @lines = split("\n", $output);
 969
 970     return @lines;
 971 }
 972
 973 sub vcs_find_signers {
 974     my ($cmd) = @_;
 975     my @lines = ();
 976     my $commits;
 977
 978     @lines = &{$VCS_cmds{"execute_cmd"}}($cmd);
 979
 980     my $pattern = $VCS_cmds{"commit_pattern"};
 981
 982     $commits = grep(/$pattern/, @lines);        # of commits
 983
 984     @lines = grep(/^[ \t]*${signaturePattern}.*\@.*$/, @lines);
 985     if (!$email_git_penguin_chiefs) {
 986         @lines = grep(!/${penguin_chiefs}/i, @lines);
 987     }
 988     # cut -f2- -d":"
 989     s/.*:\s*(.+)\s*/$1/ for (@lines);
 990
 991 ## Reformat email addresses (with names) to avoid badly written signatures
 992
 993     foreach my $line (@lines) {
 994         my ($name, $address) = parse_email($line);
 995         $line = format_email($name, $address, 1);
 996     }
 997
 998     return ($commits, @lines);
 999 }
1000
1001 sub vcs_save_commits {
1002     my ($cmd) = @_;
1003     my @lines = ();
1004     my @commits = ();
1005
1006     @lines = &{$VCS_cmds{"execute_cmd"}}($cmd);
1007
1008     foreach my $line (@lines) {
1009         if ($line =~ m/$VCS_cmds{"blame_commit_pattern"}/) {
1010             push(@commits, $1);
1011         }
1012     }
1013
1014     return @commits;
1015 }
1016
1017 sub vcs_blame {
1018     my ($file) = @_;
1019     my $cmd;
1020     my @commits = ();
1021
1022     return @commits if (!(-f $file));
1023
1024     if (@range && $VCS_cmds{"blame_range_cmd"} eq "") {
1025         my @all_commits = ();
1026
1027         $cmd = $VCS_cmds{"blame_file_cmd"};
1028         $cmd =~ s/(\$\w+)/$1/eeg;               #interpolate $cmd
1029         @all_commits = vcs_save_commits($cmd);
1030
1031         foreach my $file_range_diff (@range) {
1032             next if (!($file_range_diff =~ m/(.+):(.+):(.+)/));
1033             my $diff_file = $1;
1034             my $diff_start = $2;
1035             my $diff_length = $3;
1036             next if ("$file" ne "$diff_file");
1037             for (my $i = $diff_start; $i < $diff_start + $diff_length; $i++) {
1038                 push(@commits, $all_commits[$i]);
1039             }
1040         }
1041     } elsif (@range) {
1042         foreach my $file_range_diff (@range) {
1043             next if (!($file_range_diff =~ m/(.+):(.+):(.+)/));
1044             my $diff_file = $1;
1045             my $diff_start = $2;
1046             my $diff_length = $3;
1047             next if ("$file" ne "$diff_file");
1048             $cmd = $VCS_cmds{"blame_range_cmd"};
1049             $cmd =~ s/(\$\w+)/$1/eeg;           #interpolate $cmd
1050             push(@commits, vcs_save_commits($cmd));
1051         }
1052     } else {
1053         $cmd = $VCS_cmds{"blame_file_cmd"};
1054         $cmd =~ s/(\$\w+)/$1/eeg;               #interpolate $cmd
1055         @commits = vcs_save_commits($cmd);
1056     }
1057
1058     return @commits;
1059 }
1060
1061 my $printed_novcs = 0;
1062 sub vcs_exists {
1063     %VCS_cmds = %VCS_cmds_git;
1064     return 1 if eval $VCS_cmds{"available"};
1065     %VCS_cmds = %VCS_cmds_hg;
1066     return 1 if eval $VCS_cmds{"available"};
1067     %VCS_cmds = ();
1068     if (!$printed_novcs) {
1069         warn("$P: No supported VCS found.  Add --nogit to options?\n");
1070         warn("Using a git repository produces better results.\n");
1071         warn("Try Linus Torvalds' latest git repository using:\n");
1072         warn("git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git\n");
1073         $printed_novcs = 1;
1074     }
1075     return 0;
1076 }
1077
1078 sub vcs_assign {
1079     my ($role, $divisor, @lines) = @_;
1080
1081     my %hash;
1082     my $count = 0;
1083
1084     return if (@lines <= 0);
1085
1086     if ($divisor <= 0) {
1087         warn("Bad divisor in " . (caller(0))[3] . ": $divisor\n");
1088         $divisor = 1;
1089     }
1090
1091     if ($email_remove_duplicates) {
1092         @lines = mailmap(@lines);
1093     }
1094
1095     @lines = sort(@lines);
1096
1097     # uniq -c
1098     $hash{$_}++ for @lines;
1099
1100     # sort -rn
1101     foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) {
1102         my $sign_offs = $hash{$line};
1103         my $percent = $sign_offs * 100 / $divisor;
1104
1105         $percent = 100 if ($percent > 100);
1106         $count++;
1107         last if ($sign_offs < $email_git_min_signatures ||
1108                  $count > $email_git_max_maintainers ||
1109                  $percent < $email_git_min_percent);
1110         push_email_address($line, '');
1111         if ($output_rolestats) {
1112             my $fmt_percent = sprintf("%.0f", $percent);
1113             add_role($line, "$role:$sign_offs/$divisor=$fmt_percent%");
1114         } else {
1115             add_role($line, $role);
1116         }
1117     }
1118 }
1119
1120 sub vcs_file_signoffs {
1121     my ($file) = @_;
1122
1123     my @signers = ();
1124     my $commits;
1125
1126     return if (!vcs_exists());
1127
1128     my $cmd = $VCS_cmds{"find_signers_cmd"};
1129     $cmd =~ s/(\$\w+)/$1/eeg;           # interpolate $cmd
1130
1131     ($commits, @signers) = vcs_find_signers($cmd);
1132     vcs_assign("commit_signer", $commits, @signers);
1133 }
1134
1135 sub vcs_file_blame {
1136     my ($file) = @_;
1137
1138     my @signers = ();
1139     my @commits = ();
1140     my $total_commits;
1141
1142     return if (!vcs_exists());
1143
1144     @commits = vcs_blame($file);
1145     @commits = uniq(@commits);
1146     $total_commits = @commits;
1147
1148     foreach my $commit (@commits) {
1149         my $commit_count;
1150         my @commit_signers = ();
1151
1152         my $cmd = $VCS_cmds{"find_commit_signers_cmd"};
1153         $cmd =~ s/(\$\w+)/$1/eeg;       #interpolate $cmd
1154
1155         ($commit_count, @commit_signers) = vcs_find_signers($cmd);
1156         push(@signers, @commit_signers);
1157     }
1158
1159     if ($from_filename) {
1160         vcs_assign("commits", $total_commits, @signers);
1161     } else {
1162         vcs_assign("modified commits", $total_commits, @signers);
1163     }
1164 }
1165
1166 sub uniq {
1167     my (@parms) = @_;
1168
1169     my %saw;
1170     @parms = grep(!$saw{$_}++, @parms);
1171     return @parms;
1172 }
1173
1174 sub sort_and_uniq {
1175     my (@parms) = @_;
1176
1177     my %saw;
1178     @parms = sort @parms;
1179     @parms = grep(!$saw{$_}++, @parms);
1180     return @parms;
1181 }
1182
1183 sub clean_file_emails {
1184     my (@file_emails) = @_;
1185     my @fmt_emails = ();
1186
1187     foreach my $email (@file_emails) {
1188         $email =~ s/[\(\<\{]{0,1}([A-Za-z0-9_\.\+-]+\@[A-Za-z0-9\.-]+)[\)\>\}]{0,1}/\<$1\>/g;
1189         my ($name, $address) = parse_email($email);
1190         if ($name eq '"[,\.]"') {
1191             $name = "";
1192         }
1193
1194         my @nw = split(/[^A-Za-zÀ-ÿ\'\,\.\+-]/, $name);
1195         if (@nw > 2) {
1196             my $first = $nw[@nw - 3];
1197             my $middle = $nw[@nw - 2];
1198             my $last = $nw[@nw - 1];
1199
1200             if (((length($first) == 1 && $first =~ m/[A-Za-z]/) ||
1201                  (length($first) == 2 && substr($first, -1) eq ".")) ||
1202                 (length($middle) == 1 ||
1203                  (length($middle) == 2 && substr($middle, -1) eq "."))) {
1204                 $name = "$first $middle $last";
1205             } else {
1206                 $name = "$middle $last";
1207             }
1208         }
1209
1210         if (substr($name, -1) =~ /[,\.]/) {
1211             $name = substr($name, 0, length($name) - 1);
1212         } elsif (substr($name, -2) =~ /[,\.]"/) {
1213             $name = substr($name, 0, length($name) - 2) . '"';
1214         }
1215
1216         if (substr($name, 0, 1) =~ /[,\.]/) {
1217             $name = substr($name, 1, length($name) - 1);
1218         } elsif (substr($name, 0, 2) =~ /"[,\.]/) {
1219             $name = '"' . substr($name, 2, length($name) - 2);
1220         }
1221
1222         my $fmt_email = format_email($name, $address, $email_usename);
1223         push(@fmt_emails, $fmt_email);
1224     }
1225     return @fmt_emails;
1226 }
1227
1228 sub merge_email {
1229     my @lines;
1230     my %saw;
1231
1232     for (@_) {
1233         my ($address, $role) = @$_;
1234         if (!$saw{$address}) {
1235             if ($output_roles) {
1236                 push(@lines, "$address ($role)");
1237             } else {
1238                 push(@lines, $address);
1239             }
1240             $saw{$address} = 1;
1241         }
1242     }
1243
1244     return @lines;
1245 }
1246
1247 sub output {
1248     my (@parms) = @_;
1249
1250     if ($output_multiline) {
1251         foreach my $line (@parms) {
1252             print("${line}\n");
1253         }
1254     } else {
1255         print(join($output_separator, @parms));
1256         print("\n");
1257     }
1258 }
1259
1260 my $rfc822re;
1261
1262 sub make_rfc822re {
1263 #   Basic lexical tokens are specials, domain_literal, quoted_string, atom, and
1264 #   comment.  We must allow for rfc822_lwsp (or comments) after each of these.
1265 #   This regexp will only work on addresses which have had comments stripped
1266 #   and replaced with rfc822_lwsp.
1267
1268     my $specials = '()<>@,;:\\\\".\\[\\]';
1269     my $controls = '\\000-\\037\\177';
1270
1271     my $dtext = "[^\\[\\]\\r\\\\]";
1272     my $domain_literal = "\\[(?:$dtext|\\\\.)*\\]$rfc822_lwsp*";
1273
1274     my $quoted_string = "\"(?:[^\\\"\\r\\\\]|\\\\.|$rfc822_lwsp)*\"$rfc822_lwsp*";
1275
1276 #   Use zero-width assertion to spot the limit of an atom.  A simple
1277 #   $rfc822_lwsp* causes the regexp engine to hang occasionally.
1278     my $atom = "[^$specials $controls]+(?:$rfc822_lwsp+|\\Z|(?=[\\[\"$specials]))";
1279     my $word = "(?:$atom|$quoted_string)";
1280     my $localpart = "$word(?:\\.$rfc822_lwsp*$word)*";
1281
1282     my $sub_domain = "(?:$atom|$domain_literal)";
1283     my $domain = "$sub_domain(?:\\.$rfc822_lwsp*$sub_domain)*";
1284
1285     my $addr_spec = "$localpart\@$rfc822_lwsp*$domain";
1286
1287     my $phrase = "$word*";
1288     my $route = "(?:\@$domain(?:,\@$rfc822_lwsp*$domain)*:$rfc822_lwsp*)";
1289     my $route_addr = "\\<$rfc822_lwsp*$route?$addr_spec\\>$rfc822_lwsp*";
1290     my $mailbox = "(?:$addr_spec|$phrase$route_addr)";
1291
1292     my $group = "$phrase:$rfc822_lwsp*(?:$mailbox(?:,\\s*$mailbox)*)?;\\s*";
1293     my $address = "(?:$mailbox|$group)";
1294
1295     return "$rfc822_lwsp*$address";
1296 }
1297
1298 sub rfc822_strip_comments {
1299     my $s = shift;
1300 #   Recursively remove comments, and replace with a single space.  The simpler
1301 #   regexps in the Email Addressing FAQ are imperfect - they will miss escaped
1302 #   chars in atoms, for example.
1303
1304     while ($s =~ s/^((?:[^"\\]|\\.)*
1305                     (?:"(?:[^"\\]|\\.)*"(?:[^"\\]|\\.)*)*)
1306                     \((?:[^()\\]|\\.)*\)/$1 /osx) {}
1307     return $s;
1308 }
1309
1310 #   valid: returns true if the parameter is an RFC822 valid address
1311 #
1312 sub rfc822_valid {
1313     my $s = rfc822_strip_comments(shift);
1314
1315     if (!$rfc822re) {
1316         $rfc822re = make_rfc822re();
1317     }
1318
1319     return $s =~ m/^$rfc822re$/so && $s =~ m/^$rfc822_char*$/;
1320 }
1321
1322 #   validlist: In scalar context, returns true if the parameter is an RFC822
1323 #              valid list of addresses.
1324 #
1325 #              In list context, returns an empty list on failure (an invalid
1326 #              address was found); otherwise a list whose first element is the
1327 #              number of addresses found and whose remaining elements are the
1328 #              addresses.  This is needed to disambiguate failure (invalid)
1329 #              from success with no addresses found, because an empty string is
1330 #              a valid list.
1331
1332 sub rfc822_validlist {
1333     my $s = rfc822_strip_comments(shift);
1334
1335     if (!$rfc822re) {
1336         $rfc822re = make_rfc822re();
1337     }
1338     # * null list items are valid according to the RFC
1339     # * the '1' business is to aid in distinguishing failure from no results
1340
1341     my @r;
1342     if ($s =~ m/^(?:$rfc822re)?(?:,(?:$rfc822re)?)*$/so &&
1343         $s =~ m/^$rfc822_char*$/) {
1344         while ($s =~ m/(?:^|,$rfc822_lwsp*)($rfc822re)/gos) {
1345             push(@r, $1);
1346         }
1347         return wantarray ? (scalar(@r), @r) : 1;
1348     }
1349     return wantarray ? () : 0;
1350 }