ppp_generic.c severly whitespace damanged by 9c705260feea6ae329bc6b6d5f6d2ef0227eda0a
[linux-2.6/libata-dev.git] / scripts / get_maintainer.pl
blob445e8845f0a4c06ee5109e63471bab7769d2f3d5
1 #!/usr/bin/perl -w
2 # (c) 2007, Joe Perches <joe@perches.com>
3 # created from checkpatch.pl
5 # Print selected MAINTAINERS information for
6 # the files modified in a patch or for a file
8 # usage: perl scripts/get_maintainer.pl [OPTIONS] <patch>
9 # perl scripts/get_maintainer.pl [OPTIONS] -f <file>
11 # Licensed under the terms of the GNU GPL License version 2
13 use strict;
15 my $P = $0;
16 my $V = '0.23';
18 use Getopt::Long qw(:config no_auto_abbrev);
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 $from_filename = 0;
45 my $pattern_depth = 0;
46 my $version = 0;
47 my $help = 0;
49 my $exit = 0;
51 my @penguin_chief = ();
52 push(@penguin_chief,"Linus Torvalds:torvalds\@linux-foundation.org");
53 #Andrew wants in on most everything - 2009/01/14
54 #push(@penguin_chief,"Andrew Morton:akpm\@linux-foundation.org");
56 my @penguin_chief_names = ();
57 foreach my $chief (@penguin_chief) {
58 if ($chief =~ m/^(.*):(.*)/) {
59 my $chief_name = $1;
60 my $chief_addr = $2;
61 push(@penguin_chief_names, $chief_name);
64 my $penguin_chiefs = "\(" . join("|",@penguin_chief_names) . "\)";
66 # rfc822 email address - preloaded methods go here.
67 my $rfc822_lwsp = "(?:(?:\\r\\n)?[ \\t])";
68 my $rfc822_char = '[\\000-\\377]';
70 # VCS command support: class-like functions and strings
72 my %VCS_cmds;
74 my %VCS_cmds_git = (
75 "execute_cmd" => \&git_execute_cmd,
76 "available" => '(which("git") ne "") && (-d ".git")',
77 "find_signers_cmd" => "git log --since=\$email_git_since -- \$file",
78 "find_commit_signers_cmd" => "git log -1 \$commit",
79 "blame_range_cmd" => "git blame -l -L \$diff_start,+\$diff_length \$file",
80 "blame_file_cmd" => "git blame -l \$file",
81 "commit_pattern" => "^commit [0-9a-f]{40,40}",
82 "blame_commit_pattern" => "^([0-9a-f]+) "
85 my %VCS_cmds_hg = (
86 "execute_cmd" => \&hg_execute_cmd,
87 "available" => '(which("hg") ne "") && (-d ".hg")',
88 "find_signers_cmd" =>
89 "hg log --date=\$email_hg_since" .
90 " --template='commit {node}\\n{desc}\\n' -- \$file",
91 "find_commit_signers_cmd" => "hg log --template='{desc}\\n' -r \$commit",
92 "blame_range_cmd" => "", # not supported
93 "blame_file_cmd" => "hg blame -c \$file",
94 "commit_pattern" => "^commit [0-9a-f]{40,40}",
95 "blame_commit_pattern" => "^([0-9a-f]+):"
98 if (!GetOptions(
99 'email!' => \$email,
100 'git!' => \$email_git,
101 'git-blame!' => \$email_git_blame,
102 'git-chief-penguins!' => \$email_git_penguin_chiefs,
103 'git-min-signatures=i' => \$email_git_min_signatures,
104 'git-max-maintainers=i' => \$email_git_max_maintainers,
105 'git-min-percent=i' => \$email_git_min_percent,
106 'git-since=s' => \$email_git_since,
107 'hg-since=s' => \$email_hg_since,
108 'remove-duplicates!' => \$email_remove_duplicates,
109 'm!' => \$email_maintainer,
110 'n!' => \$email_usename,
111 'l!' => \$email_list,
112 's!' => \$email_subscriber_list,
113 'multiline!' => \$output_multiline,
114 'roles!' => \$output_roles,
115 'rolestats!' => \$output_rolestats,
116 'separator=s' => \$output_separator,
117 'subsystem!' => \$subsystem,
118 'status!' => \$status,
119 'scm!' => \$scm,
120 'web!' => \$web,
121 'pattern-depth=i' => \$pattern_depth,
122 'k|keywords!' => \$keywords,
123 'f|file' => \$from_filename,
124 'v|version' => \$version,
125 'h|help' => \$help,
126 )) {
127 die "$P: invalid argument - use --help if necessary\n";
130 if ($help != 0) {
131 usage();
132 exit 0;
135 if ($version != 0) {
136 print("${P} ${V}\n");
137 exit 0;
140 if ($#ARGV < 0) {
141 usage();
142 die "$P: argument missing: patchfile or -f file please\n";
145 if ($output_separator ne ", ") {
146 $output_multiline = 0;
149 if ($output_rolestats) {
150 $output_roles = 1;
153 my $selections = $email + $scm + $status + $subsystem + $web;
154 if ($selections == 0) {
155 usage();
156 die "$P: Missing required option: email, scm, status, subsystem or web\n";
159 if ($email &&
160 ($email_maintainer + $email_list + $email_subscriber_list +
161 $email_git + $email_git_penguin_chiefs + $email_git_blame) == 0) {
162 usage();
163 die "$P: Please select at least 1 email option\n";
166 if (!top_of_kernel_tree($lk_path)) {
167 die "$P: The current directory does not appear to be "
168 . "a linux kernel source tree.\n";
171 ## Read MAINTAINERS for type/value pairs
173 my @typevalue = ();
174 my %keyword_hash;
176 open(MAINT, "<${lk_path}MAINTAINERS") || die "$P: Can't open MAINTAINERS\n";
177 while (<MAINT>) {
178 my $line = $_;
180 if ($line =~ m/^(\C):\s*(.*)/) {
181 my $type = $1;
182 my $value = $2;
184 ##Filename pattern matching
185 if ($type eq "F" || $type eq "X") {
186 $value =~ s@\.@\\\.@g; ##Convert . to \.
187 $value =~ s/\*/\.\*/g; ##Convert * to .*
188 $value =~ s/\?/\./g; ##Convert ? to .
189 ##if pattern is a directory and it lacks a trailing slash, add one
190 if ((-d $value)) {
191 $value =~ s@([^/])$@$1/@;
193 } elsif ($type eq "K") {
194 $keyword_hash{@typevalue} = $value;
196 push(@typevalue, "$type:$value");
197 } elsif (!/^(\s)*$/) {
198 $line =~ s/\n$//g;
199 push(@typevalue, $line);
202 close(MAINT);
204 my %mailmap;
206 if ($email_remove_duplicates) {
207 open(MAILMAP, "<${lk_path}.mailmap") || warn "$P: Can't open .mailmap\n";
208 while (<MAILMAP>) {
209 my $line = $_;
211 next if ($line =~ m/^\s*#/);
212 next if ($line =~ m/^\s*$/);
214 my ($name, $address) = parse_email($line);
215 $line = format_email($name, $address, $email_usename);
217 next if ($line =~ m/^\s*$/);
219 if (exists($mailmap{$name})) {
220 my $obj = $mailmap{$name};
221 push(@$obj, $address);
222 } else {
223 my @arr = ($address);
224 $mailmap{$name} = \@arr;
227 close(MAILMAP);
230 ## use the filenames on the command line or find the filenames in the patchfiles
232 my @files = ();
233 my @range = ();
234 my @keyword_tvi = ();
236 foreach my $file (@ARGV) {
237 ##if $file is a directory and it lacks a trailing slash, add one
238 if ((-d $file)) {
239 $file =~ s@([^/])$@$1/@;
240 } elsif (!(-f $file)) {
241 die "$P: file '${file}' not found\n";
243 if ($from_filename) {
244 push(@files, $file);
245 if (-f $file && $keywords) {
246 open(FILE, "<$file") or die "$P: Can't open ${file}\n";
247 my $text = do { local($/) ; <FILE> };
248 foreach my $line (keys %keyword_hash) {
249 if ($text =~ m/$keyword_hash{$line}/x) {
250 push(@keyword_tvi, $line);
253 close(FILE);
255 } else {
256 my $file_cnt = @files;
257 my $lastfile;
258 open(PATCH, "<$file") or die "$P: Can't open ${file}\n";
259 while (<PATCH>) {
260 my $patch_line = $_;
261 if (m/^\+\+\+\s+(\S+)/) {
262 my $filename = $1;
263 $filename =~ s@^[^/]*/@@;
264 $filename =~ s@\n@@;
265 $lastfile = $filename;
266 push(@files, $filename);
267 } elsif (m/^\@\@ -(\d+),(\d+)/) {
268 if ($email_git_blame) {
269 push(@range, "$lastfile:$1:$2");
271 } elsif ($keywords) {
272 foreach my $line (keys %keyword_hash) {
273 if ($patch_line =~ m/^[+-].*$keyword_hash{$line}/x) {
274 push(@keyword_tvi, $line);
279 close(PATCH);
280 if ($file_cnt == @files) {
281 warn "$P: file '${file}' doesn't appear to be a patch. "
282 . "Add -f to options?\n";
284 @files = sort_and_uniq(@files);
288 my @email_to = ();
289 my @list_to = ();
290 my @scm = ();
291 my @web = ();
292 my @subsystem = ();
293 my @status = ();
295 # Find responsible parties
297 foreach my $file (@files) {
299 #Do not match excluded file patterns
301 my $exclude = 0;
302 foreach my $line (@typevalue) {
303 if ($line =~ m/^(\C):\s*(.*)/) {
304 my $type = $1;
305 my $value = $2;
306 if ($type eq 'X') {
307 if (file_match_pattern($file, $value)) {
308 $exclude = 1;
309 last;
315 if (!$exclude) {
316 my $tvi = 0;
317 my %hash;
318 foreach my $line (@typevalue) {
319 if ($line =~ m/^(\C):\s*(.*)/) {
320 my $type = $1;
321 my $value = $2;
322 if ($type eq 'F') {
323 if (file_match_pattern($file, $value)) {
324 my $value_pd = ($value =~ tr@/@@);
325 my $file_pd = ($file =~ tr@/@@);
326 $value_pd++ if (substr($value,-1,1) ne "/");
327 if ($pattern_depth == 0 ||
328 (($file_pd - $value_pd) < $pattern_depth)) {
329 $hash{$tvi} = $value_pd;
334 $tvi++;
336 foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) {
337 add_categories($line);
341 if ($email && $email_git) {
342 vcs_file_signoffs($file);
345 if ($email && $email_git_blame) {
346 vcs_file_blame($file);
350 if ($keywords) {
351 @keyword_tvi = sort_and_uniq(@keyword_tvi);
352 foreach my $line (@keyword_tvi) {
353 add_categories($line);
357 if ($email) {
358 foreach my $chief (@penguin_chief) {
359 if ($chief =~ m/^(.*):(.*)/) {
360 my $email_address;
362 $email_address = format_email($1, $2, $email_usename);
363 if ($email_git_penguin_chiefs) {
364 push(@email_to, [$email_address, 'chief penguin']);
365 } else {
366 @email_to = grep($_->[0] !~ /${email_address}/, @email_to);
372 if ($email || $email_list) {
373 my @to = ();
374 if ($email) {
375 @to = (@to, @email_to);
377 if ($email_list) {
378 @to = (@to, @list_to);
380 output(merge_email(@to));
383 if ($scm) {
384 @scm = uniq(@scm);
385 output(@scm);
388 if ($status) {
389 @status = uniq(@status);
390 output(@status);
393 if ($subsystem) {
394 @subsystem = uniq(@subsystem);
395 output(@subsystem);
398 if ($web) {
399 @web = uniq(@web);
400 output(@web);
403 exit($exit);
405 sub file_match_pattern {
406 my ($file, $pattern) = @_;
407 if (substr($pattern, -1) eq "/") {
408 if ($file =~ m@^$pattern@) {
409 return 1;
411 } else {
412 if ($file =~ m@^$pattern@) {
413 my $s1 = ($file =~ tr@/@@);
414 my $s2 = ($pattern =~ tr@/@@);
415 if ($s1 == $s2) {
416 return 1;
420 return 0;
423 sub usage {
424 print <<EOT;
425 usage: $P [options] patchfile
426 $P [options] -f file|directory
427 version: $V
429 MAINTAINER field selection options:
430 --email => print email address(es) if any
431 --git => include recent git \*-by: signers
432 --git-chief-penguins => include ${penguin_chiefs}
433 --git-min-signatures => number of signatures required (default: 1)
434 --git-max-maintainers => maximum maintainers to add (default: 5)
435 --git-min-percent => minimum percentage of commits required (default: 5)
436 --git-blame => use git blame to find modified commits for patch or file
437 --git-since => git history to use (default: 1-year-ago)
438 --hg-since => hg history to use (default: -365)
439 --m => include maintainer(s) if any
440 --n => include name 'Full Name <addr\@domain.tld>'
441 --l => include list(s) if any
442 --s => include subscriber only list(s) if any
443 --remove-duplicates => minimize duplicate email names/addresses
444 --roles => show roles (status:subsystem, git-signer, list, etc...)
445 --rolestats => show roles and statistics (commits/total_commits, %)
446 --scm => print SCM tree(s) if any
447 --status => print status if any
448 --subsystem => print subsystem name if any
449 --web => print website(s) if any
451 Output type options:
452 --separator [, ] => separator for multiple entries on 1 line
453 using --separator also sets --nomultiline if --separator is not [, ]
454 --multiline => print 1 entry per line
456 Other options:
457 --pattern-depth => Number of pattern directory traversals (default: 0 (all))
458 --keywords => scan patch for keywords (default: 1 (on))
459 --version => show version
460 --help => show this help information
462 Default options:
463 [--email --git --m --n --l --multiline --pattern-depth=0 --remove-duplicates]
465 Notes:
466 Using "-f directory" may give unexpected results:
467 Used with "--git", git signators for _all_ files in and below
468 directory are examined as git recurses directories.
469 Any specified X: (exclude) pattern matches are _not_ ignored.
470 Used with "--nogit", directory is used as a pattern match,
471 no individual file within the directory or subdirectory
472 is matched.
473 Used with "--git-blame", does not iterate all files in directory
474 Using "--git-blame" is slow and may add old committers and authors
475 that are no longer active maintainers to the output.
476 Using "--roles" or "--rolestats" with git send-email --cc-cmd or any
477 other automated tools that expect only ["name"] <email address>
478 may not work because of additional output after <email address>.
479 Using "--rolestats" and "--git-blame" shows the #/total=% commits,
480 not the percentage of the entire file authored. # of commits is
481 not a good measure of amount of code authored. 1 major commit may
482 contain a thousand lines, 5 trivial commits may modify a single line.
483 If git is not installed, but mercurial (hg) is installed and an .hg
484 repository exists, the following options apply to mercurial:
485 --git,
486 --git-min-signatures, --git-max-maintainers, --git-min-percent, and
487 --git-blame
488 Use --hg-since not --git-since to control date selection
492 sub top_of_kernel_tree {
493 my ($lk_path) = @_;
495 if ($lk_path ne "" && substr($lk_path,length($lk_path)-1,1) ne "/") {
496 $lk_path .= "/";
498 if ( (-f "${lk_path}COPYING")
499 && (-f "${lk_path}CREDITS")
500 && (-f "${lk_path}Kbuild")
501 && (-f "${lk_path}MAINTAINERS")
502 && (-f "${lk_path}Makefile")
503 && (-f "${lk_path}README")
504 && (-d "${lk_path}Documentation")
505 && (-d "${lk_path}arch")
506 && (-d "${lk_path}include")
507 && (-d "${lk_path}drivers")
508 && (-d "${lk_path}fs")
509 && (-d "${lk_path}init")
510 && (-d "${lk_path}ipc")
511 && (-d "${lk_path}kernel")
512 && (-d "${lk_path}lib")
513 && (-d "${lk_path}scripts")) {
514 return 1;
516 return 0;
519 sub parse_email {
520 my ($formatted_email) = @_;
522 my $name = "";
523 my $address = "";
525 if ($formatted_email =~ /^([^<]+)<(.+\@.*)>.*$/) {
526 $name = $1;
527 $address = $2;
528 } elsif ($formatted_email =~ /^\s*<(.+\@\S*)>.*$/) {
529 $address = $1;
530 } elsif ($formatted_email =~ /^(.+\@\S*).*$/) {
531 $address = $1;
534 $name =~ s/^\s+|\s+$//g;
535 $name =~ s/^\"|\"$//g;
536 $address =~ s/^\s+|\s+$//g;
538 if ($name =~ /[^a-z0-9 \.\-]/i) { ##has "must quote" chars
539 $name =~ s/(?<!\\)"/\\"/g; ##escape quotes
540 $name = "\"$name\"";
543 return ($name, $address);
546 sub format_email {
547 my ($name, $address, $usename) = @_;
549 my $formatted_email;
551 $name =~ s/^\s+|\s+$//g;
552 $name =~ s/^\"|\"$//g;
553 $address =~ s/^\s+|\s+$//g;
555 if ($name =~ /[^a-z0-9 \.\-]/i) { ##has "must quote" chars
556 $name =~ s/(?<!\\)"/\\"/g; ##escape quotes
557 $name = "\"$name\"";
560 if ($usename) {
561 if ("$name" eq "") {
562 $formatted_email = "$address";
563 } else {
564 $formatted_email = "$name <$address>";
566 } else {
567 $formatted_email = $address;
570 return $formatted_email;
573 sub find_starting_index {
574 my ($index) = @_;
576 while ($index > 0) {
577 my $tv = $typevalue[$index];
578 if (!($tv =~ m/^(\C):\s*(.*)/)) {
579 last;
581 $index--;
584 return $index;
587 sub find_ending_index {
588 my ($index) = @_;
590 while ($index < @typevalue) {
591 my $tv = $typevalue[$index];
592 if (!($tv =~ m/^(\C):\s*(.*)/)) {
593 last;
595 $index++;
598 return $index;
601 sub get_maintainer_role {
602 my ($index) = @_;
604 my $i;
605 my $start = find_starting_index($index);
606 my $end = find_ending_index($index);
608 my $role;
609 my $subsystem = $typevalue[$start];
610 if (length($subsystem) > 20) {
611 $subsystem = substr($subsystem, 0, 17);
612 $subsystem =~ s/\s*$//;
613 $subsystem = $subsystem . "...";
616 for ($i = $start + 1; $i < $end; $i++) {
617 my $tv = $typevalue[$i];
618 if ($tv =~ m/^(\C):\s*(.*)/) {
619 my $ptype = $1;
620 my $pvalue = $2;
621 if ($ptype eq "S") {
622 $role = $pvalue;
627 $role = lc($role);
628 if ($role eq "supported") {
629 $role = "supporter";
630 } elsif ($role eq "maintained") {
631 $role = "maintainer";
632 } elsif ($role eq "odd fixes") {
633 $role = "odd fixer";
634 } elsif ($role eq "orphan") {
635 $role = "orphan minder";
636 } elsif ($role eq "obsolete") {
637 $role = "obsolete minder";
638 } elsif ($role eq "buried alive in reporters") {
639 $role = "chief penguin";
642 return $role . ":" . $subsystem;
645 sub get_list_role {
646 my ($index) = @_;
648 my $i;
649 my $start = find_starting_index($index);
650 my $end = find_ending_index($index);
652 my $subsystem = $typevalue[$start];
653 if (length($subsystem) > 20) {
654 $subsystem = substr($subsystem, 0, 17);
655 $subsystem =~ s/\s*$//;
656 $subsystem = $subsystem . "...";
659 if ($subsystem eq "THE REST") {
660 $subsystem = "";
663 return $subsystem;
666 sub add_categories {
667 my ($index) = @_;
669 my $i;
670 my $start = find_starting_index($index);
671 my $end = find_ending_index($index);
673 push(@subsystem, $typevalue[$start]);
675 for ($i = $start + 1; $i < $end; $i++) {
676 my $tv = $typevalue[$i];
677 if ($tv =~ m/^(\C):\s*(.*)/) {
678 my $ptype = $1;
679 my $pvalue = $2;
680 if ($ptype eq "L") {
681 my $list_address = $pvalue;
682 my $list_additional = "";
683 my $list_role = get_list_role($i);
685 if ($list_role ne "") {
686 $list_role = ":" . $list_role;
688 if ($list_address =~ m/([^\s]+)\s+(.*)$/) {
689 $list_address = $1;
690 $list_additional = $2;
692 if ($list_additional =~ m/subscribers-only/) {
693 if ($email_subscriber_list) {
694 push(@list_to, [$list_address, "subscriber list${list_role}"]);
696 } else {
697 if ($email_list) {
698 push(@list_to, [$list_address, "open list${list_role}"]);
701 } elsif ($ptype eq "M") {
702 my ($name, $address) = parse_email($pvalue);
703 if ($name eq "") {
704 if ($i > 0) {
705 my $tv = $typevalue[$i - 1];
706 if ($tv =~ m/^(\C):\s*(.*)/) {
707 if ($1 eq "P") {
708 $name = $2;
709 $pvalue = format_email($name, $address, $email_usename);
714 if ($email_maintainer) {
715 my $role = get_maintainer_role($i);
716 push_email_addresses($pvalue, $role);
718 } elsif ($ptype eq "T") {
719 push(@scm, $pvalue);
720 } elsif ($ptype eq "W") {
721 push(@web, $pvalue);
722 } elsif ($ptype eq "S") {
723 push(@status, $pvalue);
729 my %email_hash_name;
730 my %email_hash_address;
732 sub email_inuse {
733 my ($name, $address) = @_;
735 return 1 if (($name eq "") && ($address eq ""));
736 return 1 if (($name ne "") && exists($email_hash_name{$name}));
737 return 1 if (($address ne "") && exists($email_hash_address{$address}));
739 return 0;
742 sub push_email_address {
743 my ($line, $role) = @_;
745 my ($name, $address) = parse_email($line);
747 if ($address eq "") {
748 return 0;
751 if (!$email_remove_duplicates) {
752 push(@email_to, [format_email($name, $address, $email_usename), $role]);
753 } elsif (!email_inuse($name, $address)) {
754 push(@email_to, [format_email($name, $address, $email_usename), $role]);
755 $email_hash_name{$name}++;
756 $email_hash_address{$address}++;
759 return 1;
762 sub push_email_addresses {
763 my ($address, $role) = @_;
765 my @address_list = ();
767 if (rfc822_valid($address)) {
768 push_email_address($address, $role);
769 } elsif (@address_list = rfc822_validlist($address)) {
770 my $array_count = shift(@address_list);
771 while (my $entry = shift(@address_list)) {
772 push_email_address($entry, $role);
774 } else {
775 if (!push_email_address($address, $role)) {
776 warn("Invalid MAINTAINERS address: '" . $address . "'\n");
781 sub add_role {
782 my ($line, $role) = @_;
784 my ($name, $address) = parse_email($line);
785 my $email = format_email($name, $address, $email_usename);
787 foreach my $entry (@email_to) {
788 if ($email_remove_duplicates) {
789 my ($entry_name, $entry_address) = parse_email($entry->[0]);
790 if ($name eq $entry_name || $address eq $entry_address) {
791 if ($entry->[1] eq "") {
792 $entry->[1] = "$role";
793 } else {
794 $entry->[1] = "$entry->[1],$role";
797 } else {
798 if ($email eq $entry->[0]) {
799 if ($entry->[1] eq "") {
800 $entry->[1] = "$role";
801 } else {
802 $entry->[1] = "$entry->[1],$role";
809 sub which {
810 my ($bin) = @_;
812 foreach my $path (split(/:/, $ENV{PATH})) {
813 if (-e "$path/$bin") {
814 return "$path/$bin";
818 return "";
821 sub mailmap {
822 my (@lines) = @_;
823 my %hash;
825 foreach my $line (@lines) {
826 my ($name, $address) = parse_email($line);
827 if (!exists($hash{$name})) {
828 $hash{$name} = $address;
829 } elsif ($address ne $hash{$name}) {
830 $address = $hash{$name};
831 $line = format_email($name, $address, $email_usename);
833 if (exists($mailmap{$name})) {
834 my $obj = $mailmap{$name};
835 foreach my $map_address (@$obj) {
836 if (($map_address eq $address) &&
837 ($map_address ne $hash{$name})) {
838 $line = format_email($name, $hash{$name}, $email_usename);
844 return @lines;
847 sub git_execute_cmd {
848 my ($cmd) = @_;
849 my @lines = ();
851 my $output = `$cmd`;
852 $output =~ s/^\s*//gm;
853 @lines = split("\n", $output);
855 return @lines;
858 sub hg_execute_cmd {
859 my ($cmd) = @_;
860 my @lines = ();
862 my $output = `$cmd`;
863 @lines = split("\n", $output);
865 return @lines;
868 sub vcs_find_signers {
869 my ($cmd) = @_;
870 my @lines = ();
871 my $commits;
873 @lines = &{$VCS_cmds{"execute_cmd"}}($cmd);
875 my $pattern = $VCS_cmds{"commit_pattern"};
877 $commits = grep(/$pattern/, @lines); # of commits
879 @lines = grep(/^[-_ a-z]+by:.*\@.*$/i, @lines);
880 if (!$email_git_penguin_chiefs) {
881 @lines = grep(!/${penguin_chiefs}/i, @lines);
883 # cut -f2- -d":"
884 s/.*:\s*(.+)\s*/$1/ for (@lines);
886 ## Reformat email addresses (with names) to avoid badly written signatures
888 foreach my $line (@lines) {
889 my ($name, $address) = parse_email($line);
890 $line = format_email($name, $address, 1);
893 return ($commits, @lines);
896 sub vcs_save_commits {
897 my ($cmd) = @_;
898 my @lines = ();
899 my @commits = ();
901 @lines = &{$VCS_cmds{"execute_cmd"}}($cmd);
903 foreach my $line (@lines) {
904 if ($line =~ m/$VCS_cmds{"blame_commit_pattern"}/) {
905 push(@commits, $1);
909 return @commits;
912 sub vcs_blame {
913 my ($file) = @_;
914 my $cmd;
915 my @commits = ();
917 return @commits if (!(-f $file));
919 if (@range && $VCS_cmds{"blame_range_cmd"} eq "") {
920 my @all_commits = ();
922 $cmd = $VCS_cmds{"blame_file_cmd"};
923 $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd
924 @all_commits = vcs_save_commits($cmd);
926 foreach my $file_range_diff (@range) {
927 next if (!($file_range_diff =~ m/(.+):(.+):(.+)/));
928 my $diff_file = $1;
929 my $diff_start = $2;
930 my $diff_length = $3;
931 next if ("$file" ne "$diff_file");
932 for (my $i = $diff_start; $i < $diff_start + $diff_length; $i++) {
933 push(@commits, $all_commits[$i]);
936 } elsif (@range) {
937 foreach my $file_range_diff (@range) {
938 next if (!($file_range_diff =~ m/(.+):(.+):(.+)/));
939 my $diff_file = $1;
940 my $diff_start = $2;
941 my $diff_length = $3;
942 next if ("$file" ne "$diff_file");
943 $cmd = $VCS_cmds{"blame_range_cmd"};
944 $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd
945 push(@commits, vcs_save_commits($cmd));
947 } else {
948 $cmd = $VCS_cmds{"blame_file_cmd"};
949 $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd
950 @commits = vcs_save_commits($cmd);
953 return @commits;
956 my $printed_novcs = 0;
957 sub vcs_exists {
958 %VCS_cmds = %VCS_cmds_git;
959 return 1 if eval $VCS_cmds{"available"};
960 %VCS_cmds = %VCS_cmds_hg;
961 return 1 if eval $VCS_cmds{"available"};
962 %VCS_cmds = ();
963 if (!$printed_novcs) {
964 warn("$P: No supported VCS found. Add --nogit to options?\n");
965 warn("Using a git repository produces better results.\n");
966 warn("Try Linus Torvalds' latest git repository using:\n");
967 warn("git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git\n");
968 $printed_novcs = 1;
970 return 0;
973 sub vcs_assign {
974 my ($role, $divisor, @lines) = @_;
976 my %hash;
977 my $count = 0;
979 return if (@lines <= 0);
981 if ($divisor <= 0) {
982 warn("Bad divisor in " . (caller(0))[3] . ": $divisor\n");
983 $divisor = 1;
986 if ($email_remove_duplicates) {
987 @lines = mailmap(@lines);
990 @lines = sort(@lines);
992 # uniq -c
993 $hash{$_}++ for @lines;
995 # sort -rn
996 foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) {
997 my $sign_offs = $hash{$line};
998 my $percent = $sign_offs * 100 / $divisor;
1000 $percent = 100 if ($percent > 100);
1001 $count++;
1002 last if ($sign_offs < $email_git_min_signatures ||
1003 $count > $email_git_max_maintainers ||
1004 $percent < $email_git_min_percent);
1005 push_email_address($line, '');
1006 if ($output_rolestats) {
1007 my $fmt_percent = sprintf("%.0f", $percent);
1008 add_role($line, "$role:$sign_offs/$divisor=$fmt_percent%");
1009 } else {
1010 add_role($line, $role);
1015 sub vcs_file_signoffs {
1016 my ($file) = @_;
1018 my @signers = ();
1019 my $commits;
1021 return if (!vcs_exists());
1023 my $cmd = $VCS_cmds{"find_signers_cmd"};
1024 $cmd =~ s/(\$\w+)/$1/eeg; # interpolate $cmd
1026 ($commits, @signers) = vcs_find_signers($cmd);
1027 vcs_assign("commit_signer", $commits, @signers);
1030 sub vcs_file_blame {
1031 my ($file) = @_;
1033 my @signers = ();
1034 my @commits = ();
1035 my $total_commits;
1037 return if (!vcs_exists());
1039 @commits = vcs_blame($file);
1040 @commits = uniq(@commits);
1041 $total_commits = @commits;
1043 foreach my $commit (@commits) {
1044 my $commit_count;
1045 my @commit_signers = ();
1047 my $cmd = $VCS_cmds{"find_commit_signers_cmd"};
1048 $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd
1050 ($commit_count, @commit_signers) = vcs_find_signers($cmd);
1051 push(@signers, @commit_signers);
1054 if ($from_filename) {
1055 vcs_assign("commits", $total_commits, @signers);
1056 } else {
1057 vcs_assign("modified commits", $total_commits, @signers);
1061 sub uniq {
1062 my (@parms) = @_;
1064 my %saw;
1065 @parms = grep(!$saw{$_}++, @parms);
1066 return @parms;
1069 sub sort_and_uniq {
1070 my (@parms) = @_;
1072 my %saw;
1073 @parms = sort @parms;
1074 @parms = grep(!$saw{$_}++, @parms);
1075 return @parms;
1078 sub merge_email {
1079 my @lines;
1080 my %saw;
1082 for (@_) {
1083 my ($address, $role) = @$_;
1084 if (!$saw{$address}) {
1085 if ($output_roles) {
1086 push(@lines, "$address ($role)");
1087 } else {
1088 push(@lines, $address);
1090 $saw{$address} = 1;
1094 return @lines;
1097 sub output {
1098 my (@parms) = @_;
1100 if ($output_multiline) {
1101 foreach my $line (@parms) {
1102 print("${line}\n");
1104 } else {
1105 print(join($output_separator, @parms));
1106 print("\n");
1110 my $rfc822re;
1112 sub make_rfc822re {
1113 # Basic lexical tokens are specials, domain_literal, quoted_string, atom, and
1114 # comment. We must allow for rfc822_lwsp (or comments) after each of these.
1115 # This regexp will only work on addresses which have had comments stripped
1116 # and replaced with rfc822_lwsp.
1118 my $specials = '()<>@,;:\\\\".\\[\\]';
1119 my $controls = '\\000-\\037\\177';
1121 my $dtext = "[^\\[\\]\\r\\\\]";
1122 my $domain_literal = "\\[(?:$dtext|\\\\.)*\\]$rfc822_lwsp*";
1124 my $quoted_string = "\"(?:[^\\\"\\r\\\\]|\\\\.|$rfc822_lwsp)*\"$rfc822_lwsp*";
1126 # Use zero-width assertion to spot the limit of an atom. A simple
1127 # $rfc822_lwsp* causes the regexp engine to hang occasionally.
1128 my $atom = "[^$specials $controls]+(?:$rfc822_lwsp+|\\Z|(?=[\\[\"$specials]))";
1129 my $word = "(?:$atom|$quoted_string)";
1130 my $localpart = "$word(?:\\.$rfc822_lwsp*$word)*";
1132 my $sub_domain = "(?:$atom|$domain_literal)";
1133 my $domain = "$sub_domain(?:\\.$rfc822_lwsp*$sub_domain)*";
1135 my $addr_spec = "$localpart\@$rfc822_lwsp*$domain";
1137 my $phrase = "$word*";
1138 my $route = "(?:\@$domain(?:,\@$rfc822_lwsp*$domain)*:$rfc822_lwsp*)";
1139 my $route_addr = "\\<$rfc822_lwsp*$route?$addr_spec\\>$rfc822_lwsp*";
1140 my $mailbox = "(?:$addr_spec|$phrase$route_addr)";
1142 my $group = "$phrase:$rfc822_lwsp*(?:$mailbox(?:,\\s*$mailbox)*)?;\\s*";
1143 my $address = "(?:$mailbox|$group)";
1145 return "$rfc822_lwsp*$address";
1148 sub rfc822_strip_comments {
1149 my $s = shift;
1150 # Recursively remove comments, and replace with a single space. The simpler
1151 # regexps in the Email Addressing FAQ are imperfect - they will miss escaped
1152 # chars in atoms, for example.
1154 while ($s =~ s/^((?:[^"\\]|\\.)*
1155 (?:"(?:[^"\\]|\\.)*"(?:[^"\\]|\\.)*)*)
1156 \((?:[^()\\]|\\.)*\)/$1 /osx) {}
1157 return $s;
1160 # valid: returns true if the parameter is an RFC822 valid address
1162 sub rfc822_valid ($) {
1163 my $s = rfc822_strip_comments(shift);
1165 if (!$rfc822re) {
1166 $rfc822re = make_rfc822re();
1169 return $s =~ m/^$rfc822re$/so && $s =~ m/^$rfc822_char*$/;
1172 # validlist: In scalar context, returns true if the parameter is an RFC822
1173 # valid list of addresses.
1175 # In list context, returns an empty list on failure (an invalid
1176 # address was found); otherwise a list whose first element is the
1177 # number of addresses found and whose remaining elements are the
1178 # addresses. This is needed to disambiguate failure (invalid)
1179 # from success with no addresses found, because an empty string is
1180 # a valid list.
1182 sub rfc822_validlist ($) {
1183 my $s = rfc822_strip_comments(shift);
1185 if (!$rfc822re) {
1186 $rfc822re = make_rfc822re();
1188 # * null list items are valid according to the RFC
1189 # * the '1' business is to aid in distinguishing failure from no results
1191 my @r;
1192 if ($s =~ m/^(?:$rfc822re)?(?:,(?:$rfc822re)?)*$/so &&
1193 $s =~ m/^$rfc822_char*$/) {
1194 while ($s =~ m/(?:^|,$rfc822_lwsp*)($rfc822re)/gos) {
1195 push(@r, $1);
1197 return wantarray ? (scalar(@r), @r) : 1;
1199 return wantarray ? () : 0;