]> git.ipfire.org Git - thirdparty/u-boot.git/blob - scripts/get_maintainer.pl
Merge patch series "ufs: Add a PCI UFS controller support"
[thirdparty/u-boot.git] / scripts / get_maintainer.pl
1 #!/usr/bin/env perl
2 # SPDX-License-Identifier: GPL-2.0
3 #
4 # (c) 2007, Joe Perches <joe@perches.com>
5 # created from checkpatch.pl
6 #
7 # Print selected MAINTAINERS information for
8 # the files modified in a patch or for a file
9 #
10 # usage: perl scripts/get_maintainer.pl [OPTIONS] <patch>
11 # perl scripts/get_maintainer.pl [OPTIONS] -f <file>
12
13 use warnings;
14 use strict;
15
16 my $P = $0;
17 my $V = '0.26';
18
19 use Getopt::Long qw(:config no_auto_abbrev);
20 use Cwd;
21 use File::Find;
22 use File::Spec::Functions;
23
24 my $cur_path = fastgetcwd() . '/';
25 my $lk_path = "./";
26 my $email = 1;
27 my $email_usename = 1;
28 my $email_maintainer = 1;
29 my $email_reviewer = 1;
30 my $email_fixes = 1;
31 my $email_list = 1;
32 my $email_moderated_list = 1;
33 my $email_subscriber_list = 0;
34 my $email_git_penguin_chiefs = 0;
35 my $email_git = 0;
36 my $email_git_all_signature_types = 0;
37 my $email_git_blame = 0;
38 my $email_git_blame_signatures = 1;
39 my $email_git_fallback = 1;
40 my $email_git_min_signatures = 1;
41 my $email_git_max_maintainers = 5;
42 my $email_git_min_percent = 5;
43 my $email_git_since = "1-year-ago";
44 my $email_hg_since = "-365";
45 my $interactive = 0;
46 my $email_remove_duplicates = 1;
47 my $email_use_mailmap = 1;
48 my $output_multiline = 1;
49 my $output_separator = ", ";
50 my $output_roles = 0;
51 my $output_rolestats = 1;
52 my $output_section_maxlen = 50;
53 my $scm = 0;
54 my $tree = 1;
55 my $web = 0;
56 my $subsystem = 0;
57 my $status = 0;
58 my $letters = "";
59 my $keywords = 1;
60 my $sections = 0;
61 my $email_file_emails = 0;
62 my $from_filename = 0;
63 my $pattern_depth = 0;
64 my $self_test = undef;
65 my $version = 0;
66 my $help = 0;
67 my $find_maintainer_files = 0;
68 my $maintainer_path;
69 my $vcs_used = 0;
70
71 my $exit = 0;
72
73 my @files = ();
74 my @fixes = (); # If a patch description includes Fixes: lines
75 my @range = ();
76 my @keyword_tvi = ();
77 my @file_emails = ();
78
79 my %commit_author_hash;
80 my %commit_signer_hash;
81
82 my @penguin_chief = ();
83 push(@penguin_chief, "Tom Rini:trini\@konsulko.com");
84
85 my @penguin_chief_names = ();
86 foreach my $chief (@penguin_chief) {
87 if ($chief =~ m/^(.*):(.*)/) {
88 my $chief_name = $1;
89 my $chief_addr = $2;
90 push(@penguin_chief_names, $chief_name);
91 }
92 }
93 my $penguin_chiefs = "\(" . join("|", @penguin_chief_names) . "\)";
94
95 # Signature types of people who are either
96 # a) responsible for the code in question, or
97 # b) familiar enough with it to give relevant feedback
98 my @signature_tags = ();
99 push(@signature_tags, "Signed-off-by:");
100 push(@signature_tags, "Reviewed-by:");
101 push(@signature_tags, "Acked-by:");
102
103 my $signature_pattern = "\(" . join("|", @signature_tags) . "\)";
104
105 # rfc822 email address - preloaded methods go here.
106 my $rfc822_lwsp = "(?:(?:\\r\\n)?[ \\t])";
107 my $rfc822_char = '[\\000-\\377]';
108
109 # VCS command support: class-like functions and strings
110
111 my %VCS_cmds;
112
113 my %VCS_cmds_git = (
114 "execute_cmd" => \&git_execute_cmd,
115 "available" => '(which("git") ne "") && (-e ".git")',
116 "find_signers_cmd" =>
117 "git log --no-color --follow --since=\$email_git_since " .
118 '--numstat --no-merges ' .
119 '--format="GitCommit: %H%n' .
120 'GitAuthor: %an <%ae>%n' .
121 'GitDate: %aD%n' .
122 'GitSubject: %s%n' .
123 '%b%n"' .
124 " -- \$file",
125 "find_commit_signers_cmd" =>
126 "git log --no-color " .
127 '--numstat ' .
128 '--format="GitCommit: %H%n' .
129 'GitAuthor: %an <%ae>%n' .
130 'GitDate: %aD%n' .
131 'GitSubject: %s%n' .
132 '%b%n"' .
133 " -1 \$commit",
134 "find_commit_author_cmd" =>
135 "git log --no-color " .
136 '--numstat ' .
137 '--format="GitCommit: %H%n' .
138 'GitAuthor: %an <%ae>%n' .
139 'GitDate: %aD%n' .
140 'GitSubject: %s%n"' .
141 " -1 \$commit",
142 "blame_range_cmd" => "git blame -l -L \$diff_start,+\$diff_length \$file",
143 "blame_file_cmd" => "git blame -l \$file",
144 "commit_pattern" => "^GitCommit: ([0-9a-f]{40,40})",
145 "blame_commit_pattern" => "^([0-9a-f]+) ",
146 "author_pattern" => "^GitAuthor: (.*)",
147 "subject_pattern" => "^GitSubject: (.*)",
148 "stat_pattern" => "^(\\d+)\\t(\\d+)\\t\$file\$",
149 "file_exists_cmd" => "git ls-files \$file",
150 "list_files_cmd" => "git ls-files \$file",
151 );
152
153 my %VCS_cmds_hg = (
154 "execute_cmd" => \&hg_execute_cmd,
155 "available" => '(which("hg") ne "") && (-d ".hg")',
156 "find_signers_cmd" =>
157 "hg log --date=\$email_hg_since " .
158 "--template='HgCommit: {node}\\n" .
159 "HgAuthor: {author}\\n" .
160 "HgSubject: {desc}\\n'" .
161 " -- \$file",
162 "find_commit_signers_cmd" =>
163 "hg log " .
164 "--template='HgSubject: {desc}\\n'" .
165 " -r \$commit",
166 "find_commit_author_cmd" =>
167 "hg log " .
168 "--template='HgCommit: {node}\\n" .
169 "HgAuthor: {author}\\n" .
170 "HgSubject: {desc|firstline}\\n'" .
171 " -r \$commit",
172 "blame_range_cmd" => "", # not supported
173 "blame_file_cmd" => "hg blame -n \$file",
174 "commit_pattern" => "^HgCommit: ([0-9a-f]{40,40})",
175 "blame_commit_pattern" => "^([ 0-9a-f]+):",
176 "author_pattern" => "^HgAuthor: (.*)",
177 "subject_pattern" => "^HgSubject: (.*)",
178 "stat_pattern" => "^(\\d+)\t(\\d+)\t\$file\$",
179 "file_exists_cmd" => "hg files \$file",
180 "list_files_cmd" => "hg manifest -R \$file",
181 );
182
183 my $conf = which_conf(".get_maintainer.conf");
184 if (-f $conf) {
185 my @conf_args;
186 open(my $conffile, '<', "$conf")
187 or warn "$P: Can't find a readable .get_maintainer.conf file $!\n";
188
189 while (<$conffile>) {
190 my $line = $_;
191
192 $line =~ s/\s*\n?$//g;
193 $line =~ s/^\s*//g;
194 $line =~ s/\s+/ /g;
195
196 next if ($line =~ m/^\s*#/);
197 next if ($line =~ m/^\s*$/);
198
199 my @words = split(" ", $line);
200 foreach my $word (@words) {
201 last if ($word =~ m/^#/);
202 push (@conf_args, $word);
203 }
204 }
205 close($conffile);
206 unshift(@ARGV, @conf_args) if @conf_args;
207 }
208
209 my @ignore_emails = ();
210 my $ignore_file = which_conf(".get_maintainer.ignore");
211 if (-f $ignore_file) {
212 open(my $ignore, '<', "$ignore_file")
213 or warn "$P: Can't find a readable .get_maintainer.ignore file $!\n";
214 while (<$ignore>) {
215 my $line = $_;
216
217 $line =~ s/\s*\n?$//;
218 $line =~ s/^\s*//;
219 $line =~ s/\s+$//;
220 $line =~ s/#.*$//;
221
222 next if ($line =~ m/^\s*$/);
223 if (rfc822_valid($line)) {
224 push(@ignore_emails, $line);
225 }
226 }
227 close($ignore);
228 }
229
230 if ($#ARGV > 0) {
231 foreach (@ARGV) {
232 if ($_ =~ /^-{1,2}self-test(?:=|$)/) {
233 die "$P: using --self-test does not allow any other option or argument\n";
234 }
235 }
236 }
237
238 if (!GetOptions(
239 'email!' => \$email,
240 'git!' => \$email_git,
241 'git-all-signature-types!' => \$email_git_all_signature_types,
242 'git-blame!' => \$email_git_blame,
243 'git-blame-signatures!' => \$email_git_blame_signatures,
244 'git-fallback!' => \$email_git_fallback,
245 'git-chief-penguins!' => \$email_git_penguin_chiefs,
246 'git-min-signatures=i' => \$email_git_min_signatures,
247 'git-max-maintainers=i' => \$email_git_max_maintainers,
248 'git-min-percent=i' => \$email_git_min_percent,
249 'git-since=s' => \$email_git_since,
250 'hg-since=s' => \$email_hg_since,
251 'i|interactive!' => \$interactive,
252 'remove-duplicates!' => \$email_remove_duplicates,
253 'mailmap!' => \$email_use_mailmap,
254 'm!' => \$email_maintainer,
255 'r!' => \$email_reviewer,
256 'n!' => \$email_usename,
257 'l!' => \$email_list,
258 'fixes!' => \$email_fixes,
259 'moderated!' => \$email_moderated_list,
260 's!' => \$email_subscriber_list,
261 'multiline!' => \$output_multiline,
262 'roles!' => \$output_roles,
263 'rolestats!' => \$output_rolestats,
264 'separator=s' => \$output_separator,
265 'subsystem!' => \$subsystem,
266 'status!' => \$status,
267 'scm!' => \$scm,
268 'tree!' => \$tree,
269 'web!' => \$web,
270 'letters=s' => \$letters,
271 'pattern-depth=i' => \$pattern_depth,
272 'k|keywords!' => \$keywords,
273 'sections!' => \$sections,
274 'fe|file-emails!' => \$email_file_emails,
275 'f|file' => \$from_filename,
276 'find-maintainer-files' => \$find_maintainer_files,
277 'mpath|maintainer-path=s' => \$maintainer_path,
278 'self-test:s' => \$self_test,
279 'v|version' => \$version,
280 'h|help|usage' => \$help,
281 )) {
282 die "$P: invalid argument - use --help if necessary\n";
283 }
284
285 if ($help != 0) {
286 usage();
287 exit 0;
288 }
289
290 if ($version != 0) {
291 print("${P} ${V}\n");
292 exit 0;
293 }
294
295 if (defined $self_test) {
296 read_all_maintainer_files();
297 self_test();
298 exit 0;
299 }
300
301 if (-t STDIN && !@ARGV) {
302 # We're talking to a terminal, but have no command line arguments.
303 die "$P: missing patchfile or -f file - use --help if necessary\n";
304 }
305
306 $output_multiline = 0 if ($output_separator ne ", ");
307 $output_rolestats = 1 if ($interactive);
308 $output_roles = 1 if ($output_rolestats);
309
310 if ($sections || $letters ne "") {
311 $sections = 1;
312 $email = 0;
313 $email_list = 0;
314 $scm = 0;
315 $status = 0;
316 $subsystem = 0;
317 $web = 0;
318 $keywords = 0;
319 $interactive = 0;
320 } else {
321 my $selections = $email + $scm + $status + $subsystem + $web;
322 if ($selections == 0) {
323 die "$P: Missing required option: email, scm, status, subsystem or web\n";
324 }
325 }
326
327 if ($email &&
328 ($email_maintainer + $email_reviewer +
329 $email_list + $email_subscriber_list +
330 $email_git + $email_git_penguin_chiefs + $email_git_blame) == 0) {
331 die "$P: Please select at least 1 email option\n";
332 }
333
334 if ($tree && !top_of_kernel_tree($lk_path)) {
335 die "$P: The current directory does not appear to be "
336 . "a U-Boot source tree.\n";
337 }
338
339 ## Read MAINTAINERS for type/value pairs
340
341 my @typevalue = ();
342 my %keyword_hash;
343 my @mfiles = ();
344 my @self_test_info = ();
345
346 sub read_maintainer_file {
347 my ($file) = @_;
348
349 open (my $maint, '<', "$file")
350 or die "$P: Can't open MAINTAINERS file '$file': $!\n";
351 my $i = 1;
352 while (<$maint>) {
353 my $line = $_;
354 chomp $line;
355
356 if ($line =~ m/^([A-Z]):\s*(.*)/) {
357 my $type = $1;
358 my $value = $2;
359
360 ##Filename pattern matching
361 if ($type eq "F" || $type eq "X") {
362 $value =~ s@\.@\\\.@g; ##Convert . to \.
363 $value =~ s/\*/\.\*/g; ##Convert * to .*
364 $value =~ s/\?/\./g; ##Convert ? to .
365 ##if pattern is a directory and it lacks a trailing slash, add one
366 if ((-d $value)) {
367 $value =~ s@([^/])$@$1/@;
368 }
369 } elsif ($type eq "K") {
370 $keyword_hash{@typevalue} = $value;
371 }
372 push(@typevalue, "$type:$value");
373 } elsif (!(/^\s*$/ || /^\s*\#/)) {
374 push(@typevalue, $line);
375 }
376 if (defined $self_test) {
377 push(@self_test_info, {file=>$file, linenr=>$i, line=>$line});
378 }
379 $i++;
380 }
381 close($maint);
382 }
383
384 sub find_is_maintainer_file {
385 my ($file) = $_;
386 return if ($file !~ m@/MAINTAINERS$@);
387 $file = $File::Find::name;
388 return if (! -f $file);
389 push(@mfiles, $file);
390 }
391
392 sub find_ignore_git {
393 return grep { $_ !~ /^\.git$/; } @_;
394 }
395
396 read_all_maintainer_files();
397
398 sub read_all_maintainer_files {
399 my $path = "${lk_path}MAINTAINERS";
400 if (defined $maintainer_path) {
401 $path = $maintainer_path;
402 # Perl Cookbook tilde expansion if necessary
403 $path =~ s@^~([^/]*)@ $1 ? (getpwnam($1))[7] : ( $ENV{HOME} || $ENV{LOGDIR} || (getpwuid($<))[7])@ex;
404 }
405
406 if (-d $path) {
407 $path .= '/' if ($path !~ m@/$@);
408 if ($find_maintainer_files) {
409 find( { wanted => \&find_is_maintainer_file,
410 preprocess => \&find_ignore_git,
411 no_chdir => 1,
412 }, "$path");
413 } else {
414 opendir(DIR, "$path") or die $!;
415 my @files = readdir(DIR);
416 closedir(DIR);
417 foreach my $file (@files) {
418 push(@mfiles, "$path$file") if ($file !~ /^\./);
419 }
420 }
421 } elsif (-f "$path") {
422 push(@mfiles, "$path");
423 } else {
424 die "$P: MAINTAINER file not found '$path'\n";
425 }
426 die "$P: No MAINTAINER files found in '$path'\n" if (scalar(@mfiles) == 0);
427 foreach my $file (@mfiles) {
428 read_maintainer_file("$file");
429 }
430 }
431
432 sub maintainers_in_file {
433 my ($file) = @_;
434
435 return if ($file =~ m@\bMAINTAINERS$@);
436
437 if (-f $file && ($email_file_emails || $file =~ /\.yaml$/)) {
438 open(my $f, '<', $file)
439 or die "$P: Can't open $file: $!\n";
440 my $text = do { local($/) ; <$f> };
441 close($f);
442
443 my @poss_addr = $text =~ m$[A-Za--ÿ\"\' \,\.\+-]*\s*[\,]*\s*[\(\<\{]{0,1}[A-Za-z0-9_\.\+-]+\@[A-Za-z0-9\.-]+\.[A-Za-z0-9]+[\)\>\}]{0,1}$g;
444 push(@file_emails, clean_file_emails(@poss_addr));
445 }
446 }
447
448 #
449 # Read mail address map
450 #
451
452 my $mailmap;
453
454 read_mailmap();
455
456 sub read_mailmap {
457 $mailmap = {
458 names => {},
459 addresses => {}
460 };
461
462 return if (!$email_use_mailmap || !(-f "${lk_path}.mailmap"));
463
464 open(my $mailmap_file, '<', "${lk_path}.mailmap")
465 or warn "$P: Can't open .mailmap: $!\n";
466
467 while (<$mailmap_file>) {
468 s/#.*$//; #strip comments
469 s/^\s+|\s+$//g; #trim
470
471 next if (/^\s*$/); #skip empty lines
472 #entries have one of the following formats:
473 # name1 <mail1>
474 # <mail1> <mail2>
475 # name1 <mail1> <mail2>
476 # name1 <mail1> name2 <mail2>
477 # (see man git-shortlog)
478
479 if (/^([^<]+)<([^>]+)>$/) {
480 my $real_name = $1;
481 my $address = $2;
482
483 $real_name =~ s/\s+$//;
484 ($real_name, $address) = parse_email("$real_name <$address>");
485 $mailmap->{names}->{$address} = $real_name;
486
487 } elsif (/^<([^>]+)>\s*<([^>]+)>$/) {
488 my $real_address = $1;
489 my $wrong_address = $2;
490
491 $mailmap->{addresses}->{$wrong_address} = $real_address;
492
493 } elsif (/^(.+)<([^>]+)>\s*<([^>]+)>$/) {
494 my $real_name = $1;
495 my $real_address = $2;
496 my $wrong_address = $3;
497
498 $real_name =~ s/\s+$//;
499 ($real_name, $real_address) =
500 parse_email("$real_name <$real_address>");
501 $mailmap->{names}->{$wrong_address} = $real_name;
502 $mailmap->{addresses}->{$wrong_address} = $real_address;
503
504 } elsif (/^(.+)<([^>]+)>\s*(.+)\s*<([^>]+)>$/) {
505 my $real_name = $1;
506 my $real_address = $2;
507 my $wrong_name = $3;
508 my $wrong_address = $4;
509
510 $real_name =~ s/\s+$//;
511 ($real_name, $real_address) =
512 parse_email("$real_name <$real_address>");
513
514 $wrong_name =~ s/\s+$//;
515 ($wrong_name, $wrong_address) =
516 parse_email("$wrong_name <$wrong_address>");
517
518 my $wrong_email = format_email($wrong_name, $wrong_address, 1);
519 $mailmap->{names}->{$wrong_email} = $real_name;
520 $mailmap->{addresses}->{$wrong_email} = $real_address;
521 }
522 }
523 close($mailmap_file);
524 }
525
526 ## use the filenames on the command line or find the filenames in the patchfiles
527
528 if (!@ARGV) {
529 push(@ARGV, "&STDIN");
530 }
531
532 foreach my $file (@ARGV) {
533 if ($file ne "&STDIN") {
534 $file = canonpath($file);
535 ##if $file is a directory and it lacks a trailing slash, add one
536 if ((-d $file)) {
537 $file =~ s@([^/])$@$1/@;
538 } elsif (!(-f $file)) {
539 die "$P: file '${file}' not found\n";
540 }
541 }
542 if ($from_filename && (vcs_exists() && !vcs_file_exists($file))) {
543 warn "$P: file '$file' not found in version control $!\n";
544 }
545 if ($from_filename || ($file ne "&STDIN" && vcs_file_exists($file))) {
546 $file =~ s/^\Q${cur_path}\E//; #strip any absolute path
547 $file =~ s/^\Q${lk_path}\E//; #or the path to the lk tree
548 push(@files, $file);
549 if ($file ne "MAINTAINERS" && -f $file && $keywords) {
550 open(my $f, '<', $file)
551 or die "$P: Can't open $file: $!\n";
552 my $text = do { local($/) ; <$f> };
553 close($f);
554 if ($keywords) {
555 foreach my $line (keys %keyword_hash) {
556 if ($text =~ m/$keyword_hash{$line}/x) {
557 push(@keyword_tvi, $line);
558 }
559 }
560 }
561 }
562 } else {
563 my $file_cnt = @files;
564 my $lastfile;
565
566 open(my $patch, "< $file")
567 or die "$P: Can't open $file: $!\n";
568
569 # We can check arbitrary information before the patch
570 # like the commit message, mail headers, etc...
571 # This allows us to match arbitrary keywords against any part
572 # of a git format-patch generated file (subject tags, etc...)
573
574 my $patch_prefix = ""; #Parsing the intro
575
576 while (<$patch>) {
577 my $patch_line = $_;
578 if (m/^ mode change [0-7]+ => [0-7]+ (\S+)\s*$/) {
579 my $filename = $1;
580 push(@files, $filename);
581 } elsif (m/^rename (?:from|to) (\S+)\s*$/) {
582 my $filename = $1;
583 push(@files, $filename);
584 } elsif (m/^diff --git a\/(\S+) b\/(\S+)\s*$/) {
585 my $filename1 = $1;
586 my $filename2 = $2;
587 push(@files, $filename1);
588 push(@files, $filename2);
589 } elsif (m/^Fixes:\s+([0-9a-fA-F]{6,40})/) {
590 push(@fixes, $1) if ($email_fixes);
591 } elsif (m/^\+\+\+\s+(\S+)/ or m/^---\s+(\S+)/) {
592 my $filename = $1;
593 $filename =~ s@^[^/]*/@@;
594 $filename =~ s@\n@@;
595 $lastfile = $filename;
596 push(@files, $filename);
597 $patch_prefix = "^[+-].*"; #Now parsing the actual patch
598 } elsif (m/^\@\@ -(\d+),(\d+)/) {
599 if ($email_git_blame) {
600 push(@range, "$lastfile:$1:$2");
601 }
602 } elsif ($keywords) {
603 foreach my $line (keys %keyword_hash) {
604 if ($patch_line =~ m/${patch_prefix}$keyword_hash{$line}/x) {
605 push(@keyword_tvi, $line);
606 }
607 }
608 }
609 }
610 close($patch);
611
612 if ($file_cnt == @files) {
613 warn "$P: file '${file}' doesn't appear to be a patch. "
614 . "Add -f to options?\n";
615 }
616 @files = sort_and_uniq(@files);
617 }
618 }
619
620 @file_emails = uniq(@file_emails);
621 @fixes = uniq(@fixes);
622
623 my %email_hash_name;
624 my %email_hash_address;
625 my @email_to = ();
626 my %hash_list_to;
627 my @list_to = ();
628 my @scm = ();
629 my @web = ();
630 my @subsystem = ();
631 my @status = ();
632 my %deduplicate_name_hash = ();
633 my %deduplicate_address_hash = ();
634
635 my @maintainers = get_maintainers();
636 if (@maintainers) {
637 @maintainers = merge_email(@maintainers);
638 output(@maintainers);
639 }
640
641 if ($scm) {
642 @scm = uniq(@scm);
643 output(@scm);
644 }
645
646 if ($status) {
647 @status = uniq(@status);
648 output(@status);
649 }
650
651 if ($subsystem) {
652 @subsystem = uniq(@subsystem);
653 output(@subsystem);
654 }
655
656 if ($web) {
657 @web = uniq(@web);
658 output(@web);
659 }
660
661 exit($exit);
662
663 sub self_test {
664 my @lsfiles = ();
665 my @good_links = ();
666 my @bad_links = ();
667 my @section_headers = ();
668 my $index = 0;
669
670 @lsfiles = vcs_list_files($lk_path);
671
672 for my $x (@self_test_info) {
673 $index++;
674
675 ## Section header duplication and missing section content
676 if (($self_test eq "" || $self_test =~ /\bsections\b/) &&
677 $x->{line} =~ /^\S[^:]/ &&
678 defined $self_test_info[$index] &&
679 $self_test_info[$index]->{line} =~ /^([A-Z]):\s*\S/) {
680 my $has_S = 0;
681 my $has_F = 0;
682 my $has_ML = 0;
683 my $status = "";
684 if (grep(m@^\Q$x->{line}\E@, @section_headers)) {
685 print("$x->{file}:$x->{linenr}: warning: duplicate section header\t$x->{line}\n");
686 } else {
687 push(@section_headers, $x->{line});
688 }
689 my $nextline = $index;
690 while (defined $self_test_info[$nextline] &&
691 $self_test_info[$nextline]->{line} =~ /^([A-Z]):\s*(\S.*)/) {
692 my $type = $1;
693 my $value = $2;
694 if ($type eq "S") {
695 $has_S = 1;
696 $status = $value;
697 } elsif ($type eq "F" || $type eq "N") {
698 $has_F = 1;
699 } elsif ($type eq "M" || $type eq "R" || $type eq "L") {
700 $has_ML = 1;
701 }
702 $nextline++;
703 }
704 if (!$has_ML && $status !~ /orphan|obsolete/i) {
705 print("$x->{file}:$x->{linenr}: warning: section without email address\t$x->{line}\n");
706 }
707 if (!$has_S) {
708 print("$x->{file}:$x->{linenr}: warning: section without status \t$x->{line}\n");
709 }
710 if (!$has_F) {
711 print("$x->{file}:$x->{linenr}: warning: section without file pattern\t$x->{line}\n");
712 }
713 }
714
715 next if ($x->{line} !~ /^([A-Z]):\s*(.*)/);
716
717 my $type = $1;
718 my $value = $2;
719
720 ## Filename pattern matching
721 if (($type eq "F" || $type eq "X") &&
722 ($self_test eq "" || $self_test =~ /\bpatterns\b/)) {
723 $value =~ s@\.@\\\.@g; ##Convert . to \.
724 $value =~ s/\*/\.\*/g; ##Convert * to .*
725 $value =~ s/\?/\./g; ##Convert ? to .
726 ##if pattern is a directory and it lacks a trailing slash, add one
727 if ((-d $value)) {
728 $value =~ s@([^/])$@$1/@;
729 }
730 if (!grep(m@^$value@, @lsfiles)) {
731 print("$x->{file}:$x->{linenr}: warning: no file matches\t$x->{line}\n");
732 }
733
734 ## Link reachability
735 } elsif (($type eq "W" || $type eq "Q" || $type eq "B") &&
736 $value =~ /^https?:/ &&
737 ($self_test eq "" || $self_test =~ /\blinks\b/)) {
738 next if (grep(m@^\Q$value\E$@, @good_links));
739 my $isbad = 0;
740 if (grep(m@^\Q$value\E$@, @bad_links)) {
741 $isbad = 1;
742 } else {
743 my $output = `wget --spider -q --no-check-certificate --timeout 10 --tries 1 $value`;
744 if ($? == 0) {
745 push(@good_links, $value);
746 } else {
747 push(@bad_links, $value);
748 $isbad = 1;
749 }
750 }
751 if ($isbad) {
752 print("$x->{file}:$x->{linenr}: warning: possible bad link\t$x->{line}\n");
753 }
754
755 ## SCM reachability
756 } elsif ($type eq "T" &&
757 ($self_test eq "" || $self_test =~ /\bscm\b/)) {
758 next if (grep(m@^\Q$value\E$@, @good_links));
759 my $isbad = 0;
760 if (grep(m@^\Q$value\E$@, @bad_links)) {
761 $isbad = 1;
762 } elsif ($value !~ /^(?:git|quilt|hg)\s+\S/) {
763 print("$x->{file}:$x->{linenr}: warning: malformed entry\t$x->{line}\n");
764 } elsif ($value =~ /^git\s+(\S+)(\s+([^\(]+\S+))?/) {
765 my $url = $1;
766 my $branch = "";
767 $branch = $3 if $3;
768 my $output = `git ls-remote --exit-code -h "$url" $branch > /dev/null 2>&1`;
769 if ($? == 0) {
770 push(@good_links, $value);
771 } else {
772 push(@bad_links, $value);
773 $isbad = 1;
774 }
775 } elsif ($value =~ /^(?:quilt|hg)\s+(https?:\S+)/) {
776 my $url = $1;
777 my $output = `wget --spider -q --no-check-certificate --timeout 10 --tries 1 $url`;
778 if ($? == 0) {
779 push(@good_links, $value);
780 } else {
781 push(@bad_links, $value);
782 $isbad = 1;
783 }
784 }
785 if ($isbad) {
786 print("$x->{file}:$x->{linenr}: warning: possible bad link\t$x->{line}\n");
787 }
788 }
789 }
790 }
791
792 sub ignore_email_address {
793 my ($address) = @_;
794
795 foreach my $ignore (@ignore_emails) {
796 return 1 if ($ignore eq $address);
797 }
798
799 return 0;
800 }
801
802 sub range_is_maintained {
803 my ($start, $end) = @_;
804
805 for (my $i = $start; $i < $end; $i++) {
806 my $line = $typevalue[$i];
807 if ($line =~ m/^([A-Z]):\s*(.*)/) {
808 my $type = $1;
809 my $value = $2;
810 if ($type eq 'S') {
811 if ($value =~ /(maintain|support)/i) {
812 return 1;
813 }
814 }
815 }
816 }
817 return 0;
818 }
819
820 sub range_has_maintainer {
821 my ($start, $end) = @_;
822
823 for (my $i = $start; $i < $end; $i++) {
824 my $line = $typevalue[$i];
825 if ($line =~ m/^([A-Z]):\s*(.*)/) {
826 my $type = $1;
827 my $value = $2;
828 if ($type eq 'M') {
829 return 1;
830 }
831 }
832 }
833 return 0;
834 }
835
836 sub get_maintainers {
837 %email_hash_name = ();
838 %email_hash_address = ();
839 %commit_author_hash = ();
840 %commit_signer_hash = ();
841 @email_to = ();
842 %hash_list_to = ();
843 @list_to = ();
844 @scm = ();
845 @web = ();
846 @subsystem = ();
847 @status = ();
848 %deduplicate_name_hash = ();
849 %deduplicate_address_hash = ();
850 if ($email_git_all_signature_types) {
851 $signature_pattern = "(.+?)[Bb][Yy]:";
852 } else {
853 $signature_pattern = "\(" . join("|", @signature_tags) . "\)";
854 }
855
856 # Find responsible parties
857
858 my %exact_pattern_match_hash = ();
859
860 foreach my $file (@files) {
861
862 my %hash;
863 my $tvi = find_first_section();
864 while ($tvi < @typevalue) {
865 my $start = find_starting_index($tvi);
866 my $end = find_ending_index($tvi);
867 my $exclude = 0;
868 my $i;
869
870 #Do not match excluded file patterns
871
872 for ($i = $start; $i < $end; $i++) {
873 my $line = $typevalue[$i];
874 if ($line =~ m/^([A-Z]):\s*(.*)/) {
875 my $type = $1;
876 my $value = $2;
877 if ($type eq 'X') {
878 if (file_match_pattern($file, $value)) {
879 $exclude = 1;
880 last;
881 }
882 }
883 }
884 }
885
886 if (!$exclude) {
887 for ($i = $start; $i < $end; $i++) {
888 my $line = $typevalue[$i];
889 if ($line =~ m/^([A-Z]):\s*(.*)/) {
890 my $type = $1;
891 my $value = $2;
892 if ($type eq 'F') {
893 if (file_match_pattern($file, $value)) {
894 my $value_pd = ($value =~ tr@/@@);
895 my $file_pd = ($file =~ tr@/@@);
896 $value_pd++ if (substr($value,-1,1) ne "/");
897 $value_pd = -1 if ($value =~ /^\.\*/);
898 if ($value_pd >= $file_pd &&
899 range_is_maintained($start, $end) &&
900 range_has_maintainer($start, $end)) {
901 $exact_pattern_match_hash{$file} = 1;
902 }
903 if ($pattern_depth == 0 ||
904 (($file_pd - $value_pd) < $pattern_depth)) {
905 $hash{$tvi} = $value_pd;
906 }
907 }
908 } elsif ($type eq 'N') {
909 if ($file =~ m/$value/x) {
910 $hash{$tvi} = 0;
911 }
912 }
913 }
914 }
915 }
916 $tvi = $end + 1;
917 }
918
919 foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) {
920 add_categories($line);
921 if ($sections) {
922 my $i;
923 my $start = find_starting_index($line);
924 my $end = find_ending_index($line);
925 for ($i = $start; $i < $end; $i++) {
926 my $line = $typevalue[$i];
927 if ($line =~ /^[FX]:/) { ##Restore file patterns
928 $line =~ s/([^\\])\.([^\*])/$1\?$2/g;
929 $line =~ s/([^\\])\.$/$1\?/g; ##Convert . back to ?
930 $line =~ s/\\\./\./g; ##Convert \. to .
931 $line =~ s/\.\*/\*/g; ##Convert .* to *
932 }
933 my $count = $line =~ s/^([A-Z]):/$1:\t/g;
934 if ($letters eq "" || (!$count || $letters =~ /$1/i)) {
935 print("$line\n");
936 }
937 }
938 print("\n");
939 }
940 }
941
942 maintainers_in_file($file);
943 }
944
945 if ($keywords) {
946 @keyword_tvi = sort_and_uniq(@keyword_tvi);
947 foreach my $line (@keyword_tvi) {
948 add_categories($line);
949 }
950 }
951
952 foreach my $email (@email_to, @list_to) {
953 $email->[0] = deduplicate_email($email->[0]);
954 }
955
956 foreach my $file (@files) {
957 if ($email &&
958 ($email_git ||
959 ($email_git_fallback &&
960 $file !~ /MAINTAINERS$/ &&
961 !$exact_pattern_match_hash{$file}))) {
962 vcs_file_signoffs($file);
963 }
964 if ($email && $email_git_blame) {
965 vcs_file_blame($file);
966 }
967 }
968
969 if ($email) {
970 foreach my $chief (@penguin_chief) {
971 if ($chief =~ m/^(.*):(.*)/) {
972 my $email_address;
973
974 $email_address = format_email($1, $2, $email_usename);
975 if ($email_git_penguin_chiefs) {
976 push(@email_to, [$email_address, 'chief penguin']);
977 } else {
978 @email_to = grep($_->[0] !~ /${email_address}/, @email_to);
979 }
980 }
981 }
982
983 foreach my $email (@file_emails) {
984 $email = mailmap_email($email);
985 my ($name, $address) = parse_email($email);
986
987 my $tmp_email = format_email($name, $address, $email_usename);
988 push_email_address($tmp_email, '');
989 add_role($tmp_email, 'in file');
990 }
991 }
992
993 foreach my $fix (@fixes) {
994 vcs_add_commit_signers($fix, "blamed_fixes");
995 }
996
997 my @to = ();
998 if ($email || $email_list) {
999 if ($email) {
1000 @to = (@to, @email_to);
1001 }
1002 if ($email_list) {
1003 @to = (@to, @list_to);
1004 }
1005 }
1006
1007 if ($interactive) {
1008 @to = interactive_get_maintainers(\@to);
1009 }
1010
1011 return @to;
1012 }
1013
1014 sub file_match_pattern {
1015 my ($file, $pattern) = @_;
1016 if (substr($pattern, -1) eq "/") {
1017 if ($file =~ m@^$pattern@) {
1018 return 1;
1019 }
1020 } else {
1021 if ($file =~ m@^$pattern@) {
1022 my $s1 = ($file =~ tr@/@@);
1023 my $s2 = ($pattern =~ tr@/@@);
1024 if ($s1 == $s2) {
1025 return 1;
1026 }
1027 }
1028 }
1029 return 0;
1030 }
1031
1032 sub usage {
1033 print <<EOT;
1034 usage: $P [options] patchfile
1035 $P [options] -f file|directory
1036 version: $V
1037
1038 MAINTAINER field selection options:
1039 --email => print email address(es) if any
1040 --git => include recent git \*-by: signers
1041 --git-all-signature-types => include signers regardless of signature type
1042 or use only ${signature_pattern} signers (default: $email_git_all_signature_types)
1043 --git-fallback => use git when no exact MAINTAINERS pattern (default: $email_git_fallback)
1044 --git-chief-penguins => include ${penguin_chiefs}
1045 --git-min-signatures => number of signatures required (default: $email_git_min_signatures)
1046 --git-max-maintainers => maximum maintainers to add (default: $email_git_max_maintainers)
1047 --git-min-percent => minimum percentage of commits required (default: $email_git_min_percent)
1048 --git-blame => use git blame to find modified commits for patch or file
1049 --git-blame-signatures => when used with --git-blame, also include all commit signers
1050 --git-since => git history to use (default: $email_git_since)
1051 --hg-since => hg history to use (default: $email_hg_since)
1052 --interactive => display a menu (mostly useful if used with the --git option)
1053 --m => include maintainer(s) if any
1054 --r => include reviewer(s) if any
1055 --n => include name 'Full Name <addr\@domain.tld>'
1056 --l => include list(s) if any
1057 --moderated => include moderated lists(s) if any (default: true)
1058 --s => include subscriber only list(s) if any (default: false)
1059 --remove-duplicates => minimize duplicate email names/addresses
1060 --roles => show roles (status:subsystem, git-signer, list, etc...)
1061 --rolestats => show roles and statistics (commits/total_commits, %)
1062 --file-emails => add email addresses found in -f file (default: 0 (off))
1063 --fixes => for patches, add signatures of commits with 'Fixes: <commit>' (default: 1 (on))
1064 --scm => print SCM tree(s) if any
1065 --status => print status if any
1066 --subsystem => print subsystem name if any
1067 --web => print website(s) if any
1068
1069 Output type options:
1070 --separator [, ] => separator for multiple entries on 1 line
1071 using --separator also sets --nomultiline if --separator is not [, ]
1072 --multiline => print 1 entry per line
1073
1074 Other options:
1075 --pattern-depth => Number of pattern directory traversals (default: 0 (all))
1076 --keywords => scan patch for keywords (default: $keywords)
1077 --sections => print all of the subsystem sections with pattern matches
1078 --letters => print all matching 'letter' types from all matching sections
1079 --mailmap => use .mailmap file (default: $email_use_mailmap)
1080 --no-tree => run without a kernel tree
1081 --self-test => show potential issues with MAINTAINERS file content
1082 --version => show version
1083 --help => show this help information
1084
1085 Default options:
1086 [--email --tree --nogit --git-fallback --m --r --n --l --multiline
1087 --pattern-depth=0 --remove-duplicates --rolestats]
1088
1089 Notes:
1090 Using "-f directory" may give unexpected results:
1091 Used with "--git", git signators for _all_ files in and below
1092 directory are examined as git recurses directories.
1093 Any specified X: (exclude) pattern matches are _not_ ignored.
1094 Used with "--nogit", directory is used as a pattern match,
1095 no individual file within the directory or subdirectory
1096 is matched.
1097 Used with "--git-blame", does not iterate all files in directory
1098 Using "--git-blame" is slow and may add old committers and authors
1099 that are no longer active maintainers to the output.
1100 Using "--roles" or "--rolestats" with git send-email --cc-cmd or any
1101 other automated tools that expect only ["name"] <email address>
1102 may not work because of additional output after <email address>.
1103 Using "--rolestats" and "--git-blame" shows the #/total=% commits,
1104 not the percentage of the entire file authored. # of commits is
1105 not a good measure of amount of code authored. 1 major commit may
1106 contain a thousand lines, 5 trivial commits may modify a single line.
1107 If git is not installed, but mercurial (hg) is installed and an .hg
1108 repository exists, the following options apply to mercurial:
1109 --git,
1110 --git-min-signatures, --git-max-maintainers, --git-min-percent, and
1111 --git-blame
1112 Use --hg-since not --git-since to control date selection
1113 File ".get_maintainer.conf", if it exists in the linux kernel source root
1114 directory, can change whatever get_maintainer defaults are desired.
1115 Entries in this file can be any command line argument.
1116 This file is prepended to any additional command line arguments.
1117 Multiple lines and # comments are allowed.
1118 Most options have both positive and negative forms.
1119 The negative forms for --<foo> are --no<foo> and --no-<foo>.
1120
1121 EOT
1122 }
1123
1124 sub top_of_kernel_tree {
1125 my ($lk_path) = @_;
1126
1127 if ($lk_path ne "" && substr($lk_path,length($lk_path)-1,1) ne "/") {
1128 $lk_path .= "/";
1129 }
1130 if ( (-f "${lk_path}Kbuild")
1131 && (-e "${lk_path}MAINTAINERS")
1132 && (-f "${lk_path}Makefile")
1133 && (-f "${lk_path}README")
1134 && (-d "${lk_path}arch")
1135 && (-d "${lk_path}board")
1136 && (-d "${lk_path}common")
1137 && (-d "${lk_path}doc")
1138 && (-d "${lk_path}drivers")
1139 && (-d "${lk_path}dts")
1140 && (-d "${lk_path}fs")
1141 && (-d "${lk_path}lib")
1142 && (-d "${lk_path}include")
1143 && (-d "${lk_path}net")
1144 && (-d "${lk_path}post")
1145 && (-d "${lk_path}scripts")
1146 && (-d "${lk_path}test")
1147 && (-d "${lk_path}tools")) {
1148 return 1;
1149 }
1150 return 0;
1151 }
1152
1153 sub parse_email {
1154 my ($formatted_email) = @_;
1155
1156 my $name = "";
1157 my $address = "";
1158
1159 if ($formatted_email =~ /^([^<]+)<(.+\@.*)>.*$/) {
1160 $name = $1;
1161 $address = $2;
1162 } elsif ($formatted_email =~ /^\s*<(.+\@\S*)>.*$/) {
1163 $address = $1;
1164 } elsif ($formatted_email =~ /^(.+\@\S*).*$/) {
1165 $address = $1;
1166 }
1167
1168 $name =~ s/^\s+|\s+$//g;
1169 $name =~ s/^\"|\"$//g;
1170 $address =~ s/^\s+|\s+$//g;
1171
1172 if ($name =~ /[^\w \-]/i) { ##has "must quote" chars
1173 $name =~ s/(?<!\\)"/\\"/g; ##escape quotes
1174 $name = "\"$name\"";
1175 }
1176
1177 return ($name, $address);
1178 }
1179
1180 sub format_email {
1181 my ($name, $address, $usename) = @_;
1182
1183 my $formatted_email;
1184
1185 $name =~ s/^\s+|\s+$//g;
1186 $name =~ s/^\"|\"$//g;
1187 $address =~ s/^\s+|\s+$//g;
1188
1189 if ($name =~ /[^\w \-]/i) { ##has "must quote" chars
1190 $name =~ s/(?<!\\)"/\\"/g; ##escape quotes
1191 $name = "\"$name\"";
1192 }
1193
1194 if ($usename) {
1195 if ("$name" eq "") {
1196 $formatted_email = "$address";
1197 } else {
1198 $formatted_email = "$name <$address>";
1199 }
1200 } else {
1201 $formatted_email = $address;
1202 }
1203
1204 return $formatted_email;
1205 }
1206
1207 sub find_first_section {
1208 my $index = 0;
1209
1210 while ($index < @typevalue) {
1211 my $tv = $typevalue[$index];
1212 if (($tv =~ m/^([A-Z]):\s*(.*)/)) {
1213 last;
1214 }
1215 $index++;
1216 }
1217
1218 return $index;
1219 }
1220
1221 sub find_starting_index {
1222 my ($index) = @_;
1223
1224 while ($index > 0) {
1225 my $tv = $typevalue[$index];
1226 if (!($tv =~ m/^([A-Z]):\s*(.*)/)) {
1227 last;
1228 }
1229 $index--;
1230 }
1231
1232 return $index;
1233 }
1234
1235 sub find_ending_index {
1236 my ($index) = @_;
1237
1238 while ($index < @typevalue) {
1239 my $tv = $typevalue[$index];
1240 if (!($tv =~ m/^([A-Z]):\s*(.*)/)) {
1241 last;
1242 }
1243 $index++;
1244 }
1245
1246 return $index;
1247 }
1248
1249 sub get_subsystem_name {
1250 my ($index) = @_;
1251
1252 my $start = find_starting_index($index);
1253
1254 my $subsystem = $typevalue[$start];
1255 if ($output_section_maxlen && length($subsystem) > $output_section_maxlen) {
1256 $subsystem = substr($subsystem, 0, $output_section_maxlen - 3);
1257 $subsystem =~ s/\s*$//;
1258 $subsystem = $subsystem . "...";
1259 }
1260 return $subsystem;
1261 }
1262
1263 sub get_maintainer_role {
1264 my ($index) = @_;
1265
1266 my $i;
1267 my $start = find_starting_index($index);
1268 my $end = find_ending_index($index);
1269
1270 my $role = "unknown";
1271 my $subsystem = get_subsystem_name($index);
1272
1273 for ($i = $start + 1; $i < $end; $i++) {
1274 my $tv = $typevalue[$i];
1275 if ($tv =~ m/^([A-Z]):\s*(.*)/) {
1276 my $ptype = $1;
1277 my $pvalue = $2;
1278 if ($ptype eq "S") {
1279 $role = $pvalue;
1280 }
1281 }
1282 }
1283
1284 $role = lc($role);
1285 if ($role eq "supported") {
1286 $role = "supporter";
1287 } elsif ($role eq "maintained") {
1288 $role = "maintainer";
1289 } elsif ($role eq "odd fixes") {
1290 $role = "odd fixer";
1291 } elsif ($role eq "orphan") {
1292 $role = "orphan minder";
1293 } elsif ($role eq "obsolete") {
1294 $role = "obsolete minder";
1295 } elsif ($role eq "buried alive in reporters") {
1296 $role = "chief penguin";
1297 }
1298
1299 return $role . ":" . $subsystem;
1300 }
1301
1302 sub get_list_role {
1303 my ($index) = @_;
1304
1305 my $subsystem = get_subsystem_name($index);
1306
1307 if ($subsystem eq "THE REST") {
1308 $subsystem = "";
1309 }
1310
1311 return $subsystem;
1312 }
1313
1314 sub add_categories {
1315 my ($index) = @_;
1316
1317 my $i;
1318 my $start = find_starting_index($index);
1319 my $end = find_ending_index($index);
1320
1321 push(@subsystem, $typevalue[$start]);
1322
1323 for ($i = $start + 1; $i < $end; $i++) {
1324 my $tv = $typevalue[$i];
1325 if ($tv =~ m/^([A-Z]):\s*(.*)/) {
1326 my $ptype = $1;
1327 my $pvalue = $2;
1328 if ($ptype eq "L") {
1329 my $list_address = $pvalue;
1330 my $list_additional = "";
1331 my $list_role = get_list_role($i);
1332
1333 if ($list_role ne "") {
1334 $list_role = ":" . $list_role;
1335 }
1336 if ($list_address =~ m/([^\s]+)\s+(.*)$/) {
1337 $list_address = $1;
1338 $list_additional = $2;
1339 }
1340 if ($list_additional =~ m/subscribers-only/) {
1341 if ($email_subscriber_list) {
1342 if (!$hash_list_to{lc($list_address)}) {
1343 $hash_list_to{lc($list_address)} = 1;
1344 push(@list_to, [$list_address,
1345 "subscriber list${list_role}"]);
1346 }
1347 }
1348 } else {
1349 if ($email_list) {
1350 if (!$hash_list_to{lc($list_address)}) {
1351 if ($list_additional =~ m/moderated/) {
1352 if ($email_moderated_list) {
1353 $hash_list_to{lc($list_address)} = 1;
1354 push(@list_to, [$list_address,
1355 "moderated list${list_role}"]);
1356 }
1357 } else {
1358 $hash_list_to{lc($list_address)} = 1;
1359 push(@list_to, [$list_address,
1360 "open list${list_role}"]);
1361 }
1362 }
1363 }
1364 }
1365 } elsif ($ptype eq "M") {
1366 if ($email_maintainer) {
1367 my $role = get_maintainer_role($i);
1368 push_email_addresses($pvalue, $role);
1369 }
1370 } elsif ($ptype eq "R") {
1371 if ($email_reviewer) {
1372 my $subsystem = get_subsystem_name($i);
1373 push_email_addresses($pvalue, "reviewer:$subsystem");
1374 }
1375 } elsif ($ptype eq "T") {
1376 push(@scm, $pvalue);
1377 } elsif ($ptype eq "W") {
1378 push(@web, $pvalue);
1379 } elsif ($ptype eq "S") {
1380 push(@status, $pvalue);
1381 }
1382 }
1383 }
1384 }
1385
1386 sub email_inuse {
1387 my ($name, $address) = @_;
1388
1389 return 1 if (($name eq "") && ($address eq ""));
1390 return 1 if (($name ne "") && exists($email_hash_name{lc($name)}));
1391 return 1 if (($address ne "") && exists($email_hash_address{lc($address)}));
1392
1393 return 0;
1394 }
1395
1396 sub push_email_address {
1397 my ($line, $role) = @_;
1398
1399 my ($name, $address) = parse_email($line);
1400
1401 if ($address eq "") {
1402 return 0;
1403 }
1404
1405 if (!$email_remove_duplicates) {
1406 push(@email_to, [format_email($name, $address, $email_usename), $role]);
1407 } elsif (!email_inuse($name, $address)) {
1408 push(@email_to, [format_email($name, $address, $email_usename), $role]);
1409 $email_hash_name{lc($name)}++ if ($name ne "");
1410 $email_hash_address{lc($address)}++;
1411 }
1412
1413 return 1;
1414 }
1415
1416 sub push_email_addresses {
1417 my ($address, $role) = @_;
1418
1419 my @address_list = ();
1420
1421 if (rfc822_valid($address)) {
1422 push_email_address($address, $role);
1423 } elsif (@address_list = rfc822_validlist($address)) {
1424 my $array_count = shift(@address_list);
1425 while (my $entry = shift(@address_list)) {
1426 push_email_address($entry, $role);
1427 }
1428 } else {
1429 if (!push_email_address($address, $role)) {
1430 warn("Invalid MAINTAINERS address: '" . $address . "'\n");
1431 }
1432 }
1433 }
1434
1435 sub add_role {
1436 my ($line, $role) = @_;
1437
1438 my ($name, $address) = parse_email($line);
1439 my $email = format_email($name, $address, $email_usename);
1440
1441 foreach my $entry (@email_to) {
1442 if ($email_remove_duplicates) {
1443 my ($entry_name, $entry_address) = parse_email($entry->[0]);
1444 if (($name eq $entry_name || $address eq $entry_address)
1445 && ($role eq "" || !($entry->[1] =~ m/$role/))
1446 ) {
1447 if ($entry->[1] eq "") {
1448 $entry->[1] = "$role";
1449 } else {
1450 $entry->[1] = "$entry->[1],$role";
1451 }
1452 }
1453 } else {
1454 if ($email eq $entry->[0]
1455 && ($role eq "" || !($entry->[1] =~ m/$role/))
1456 ) {
1457 if ($entry->[1] eq "") {
1458 $entry->[1] = "$role";
1459 } else {
1460 $entry->[1] = "$entry->[1],$role";
1461 }
1462 }
1463 }
1464 }
1465 }
1466
1467 sub which {
1468 my ($bin) = @_;
1469
1470 foreach my $path (split(/:/, $ENV{PATH})) {
1471 if (-e "$path/$bin") {
1472 return "$path/$bin";
1473 }
1474 }
1475
1476 return "";
1477 }
1478
1479 sub which_conf {
1480 my ($conf) = @_;
1481
1482 foreach my $path (split(/:/, ".:$ENV{HOME}:.scripts")) {
1483 if (-e "$path/$conf") {
1484 return "$path/$conf";
1485 }
1486 }
1487
1488 return "";
1489 }
1490
1491 sub mailmap_email {
1492 my ($line) = @_;
1493
1494 my ($name, $address) = parse_email($line);
1495 my $email = format_email($name, $address, 1);
1496 my $real_name = $name;
1497 my $real_address = $address;
1498
1499 if (exists $mailmap->{names}->{$email} ||
1500 exists $mailmap->{addresses}->{$email}) {
1501 if (exists $mailmap->{names}->{$email}) {
1502 $real_name = $mailmap->{names}->{$email};
1503 }
1504 if (exists $mailmap->{addresses}->{$email}) {
1505 $real_address = $mailmap->{addresses}->{$email};
1506 }
1507 } else {
1508 if (exists $mailmap->{names}->{$address}) {
1509 $real_name = $mailmap->{names}->{$address};
1510 }
1511 if (exists $mailmap->{addresses}->{$address}) {
1512 $real_address = $mailmap->{addresses}->{$address};
1513 }
1514 }
1515 return format_email($real_name, $real_address, 1);
1516 }
1517
1518 sub mailmap {
1519 my (@addresses) = @_;
1520
1521 my @mapped_emails = ();
1522 foreach my $line (@addresses) {
1523 push(@mapped_emails, mailmap_email($line));
1524 }
1525 merge_by_realname(@mapped_emails) if ($email_use_mailmap);
1526 return @mapped_emails;
1527 }
1528
1529 sub merge_by_realname {
1530 my %address_map;
1531 my (@emails) = @_;
1532
1533 foreach my $email (@emails) {
1534 my ($name, $address) = parse_email($email);
1535 if (exists $address_map{$name}) {
1536 $address = $address_map{$name};
1537 $email = format_email($name, $address, 1);
1538 } else {
1539 $address_map{$name} = $address;
1540 }
1541 }
1542 }
1543
1544 sub git_execute_cmd {
1545 my ($cmd) = @_;
1546 my @lines = ();
1547
1548 my $output = `$cmd`;
1549 $output =~ s/^\s*//gm;
1550 @lines = split("\n", $output);
1551
1552 return @lines;
1553 }
1554
1555 sub hg_execute_cmd {
1556 my ($cmd) = @_;
1557 my @lines = ();
1558
1559 my $output = `$cmd`;
1560 @lines = split("\n", $output);
1561
1562 return @lines;
1563 }
1564
1565 sub extract_formatted_signatures {
1566 my (@signature_lines) = @_;
1567
1568 my @type = @signature_lines;
1569
1570 s/\s*(.*):.*/$1/ for (@type);
1571
1572 # cut -f2- -d":"
1573 s/\s*.*:\s*(.+)\s*/$1/ for (@signature_lines);
1574
1575 ## Reformat email addresses (with names) to avoid badly written signatures
1576
1577 foreach my $signer (@signature_lines) {
1578 $signer = deduplicate_email($signer);
1579 }
1580
1581 return (\@type, \@signature_lines);
1582 }
1583
1584 sub vcs_find_signers {
1585 my ($cmd, $file) = @_;
1586 my $commits;
1587 my @lines = ();
1588 my @signatures = ();
1589 my @authors = ();
1590 my @stats = ();
1591
1592 @lines = &{$VCS_cmds{"execute_cmd"}}($cmd);
1593
1594 my $pattern = $VCS_cmds{"commit_pattern"};
1595 my $author_pattern = $VCS_cmds{"author_pattern"};
1596 my $stat_pattern = $VCS_cmds{"stat_pattern"};
1597
1598 $stat_pattern =~ s/(\$\w+)/$1/eeg; #interpolate $stat_pattern
1599
1600 $commits = grep(/$pattern/, @lines); # of commits
1601
1602 @authors = grep(/$author_pattern/, @lines);
1603 @signatures = grep(/^[ \t]*${signature_pattern}.*\@.*$/, @lines);
1604 @stats = grep(/$stat_pattern/, @lines);
1605
1606 # print("stats: <@stats>\n");
1607
1608 return (0, \@signatures, \@authors, \@stats) if !@signatures;
1609
1610 save_commits_by_author(@lines) if ($interactive);
1611 save_commits_by_signer(@lines) if ($interactive);
1612
1613 if (!$email_git_penguin_chiefs) {
1614 @signatures = grep(!/${penguin_chiefs}/i, @signatures);
1615 }
1616
1617 my ($author_ref, $authors_ref) = extract_formatted_signatures(@authors);
1618 my ($types_ref, $signers_ref) = extract_formatted_signatures(@signatures);
1619
1620 return ($commits, $signers_ref, $authors_ref, \@stats);
1621 }
1622
1623 sub vcs_find_author {
1624 my ($cmd) = @_;
1625 my @lines = ();
1626
1627 @lines = &{$VCS_cmds{"execute_cmd"}}($cmd);
1628
1629 if (!$email_git_penguin_chiefs) {
1630 @lines = grep(!/${penguin_chiefs}/i, @lines);
1631 }
1632
1633 return @lines if !@lines;
1634
1635 my @authors = ();
1636 foreach my $line (@lines) {
1637 if ($line =~ m/$VCS_cmds{"author_pattern"}/) {
1638 my $author = $1;
1639 my ($name, $address) = parse_email($author);
1640 $author = format_email($name, $address, 1);
1641 push(@authors, $author);
1642 }
1643 }
1644
1645 save_commits_by_author(@lines) if ($interactive);
1646 save_commits_by_signer(@lines) if ($interactive);
1647
1648 return @authors;
1649 }
1650
1651 sub vcs_save_commits {
1652 my ($cmd) = @_;
1653 my @lines = ();
1654 my @commits = ();
1655
1656 @lines = &{$VCS_cmds{"execute_cmd"}}($cmd);
1657
1658 foreach my $line (@lines) {
1659 if ($line =~ m/$VCS_cmds{"blame_commit_pattern"}/) {
1660 push(@commits, $1);
1661 }
1662 }
1663
1664 return @commits;
1665 }
1666
1667 sub vcs_blame {
1668 my ($file) = @_;
1669 my $cmd;
1670 my @commits = ();
1671
1672 return @commits if (!(-f $file));
1673
1674 if (@range && $VCS_cmds{"blame_range_cmd"} eq "") {
1675 my @all_commits = ();
1676
1677 $cmd = $VCS_cmds{"blame_file_cmd"};
1678 $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd
1679 @all_commits = vcs_save_commits($cmd);
1680
1681 foreach my $file_range_diff (@range) {
1682 next if (!($file_range_diff =~ m/(.+):(.+):(.+)/));
1683 my $diff_file = $1;
1684 my $diff_start = $2;
1685 my $diff_length = $3;
1686 next if ("$file" ne "$diff_file");
1687 for (my $i = $diff_start; $i < $diff_start + $diff_length; $i++) {
1688 push(@commits, $all_commits[$i]);
1689 }
1690 }
1691 } elsif (@range) {
1692 foreach my $file_range_diff (@range) {
1693 next if (!($file_range_diff =~ m/(.+):(.+):(.+)/));
1694 my $diff_file = $1;
1695 my $diff_start = $2;
1696 my $diff_length = $3;
1697 next if ("$file" ne "$diff_file");
1698 $cmd = $VCS_cmds{"blame_range_cmd"};
1699 $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd
1700 push(@commits, vcs_save_commits($cmd));
1701 }
1702 } else {
1703 $cmd = $VCS_cmds{"blame_file_cmd"};
1704 $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd
1705 @commits = vcs_save_commits($cmd);
1706 }
1707
1708 foreach my $commit (@commits) {
1709 $commit =~ s/^\^//g;
1710 }
1711
1712 return @commits;
1713 }
1714
1715 my $printed_novcs = 0;
1716 sub vcs_exists {
1717 %VCS_cmds = %VCS_cmds_git;
1718 return 1 if eval $VCS_cmds{"available"};
1719 %VCS_cmds = %VCS_cmds_hg;
1720 return 2 if eval $VCS_cmds{"available"};
1721 %VCS_cmds = ();
1722 if (!$printed_novcs && $email_git) {
1723 warn("$P: No supported VCS found. Add --nogit to options?\n");
1724 warn("Using a git repository produces better results.\n");
1725 warn("Try Linus Torvalds' latest git repository using:\n");
1726 warn("git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git\n");
1727 $printed_novcs = 1;
1728 }
1729 return 0;
1730 }
1731
1732 sub vcs_is_git {
1733 vcs_exists();
1734 return $vcs_used == 1;
1735 }
1736
1737 sub vcs_is_hg {
1738 return $vcs_used == 2;
1739 }
1740
1741 sub vcs_add_commit_signers {
1742 return if (!vcs_exists());
1743
1744 my ($commit, $desc) = @_;
1745 my $commit_count = 0;
1746 my $commit_authors_ref;
1747 my $commit_signers_ref;
1748 my $stats_ref;
1749 my @commit_authors = ();
1750 my @commit_signers = ();
1751 my $cmd;
1752
1753 $cmd = $VCS_cmds{"find_commit_signers_cmd"};
1754 $cmd =~ s/(\$\w+)/$1/eeg; #substitute variables in $cmd
1755
1756 ($commit_count, $commit_signers_ref, $commit_authors_ref, $stats_ref) = vcs_find_signers($cmd, "");
1757 @commit_authors = @{$commit_authors_ref} if defined $commit_authors_ref;
1758 @commit_signers = @{$commit_signers_ref} if defined $commit_signers_ref;
1759
1760 foreach my $signer (@commit_signers) {
1761 $signer = deduplicate_email($signer);
1762 }
1763
1764 vcs_assign($desc, 1, @commit_signers);
1765 }
1766
1767 sub interactive_get_maintainers {
1768 my ($list_ref) = @_;
1769 my @list = @$list_ref;
1770
1771 vcs_exists();
1772
1773 my %selected;
1774 my %authored;
1775 my %signed;
1776 my $count = 0;
1777 my $maintained = 0;
1778 foreach my $entry (@list) {
1779 $maintained = 1 if ($entry->[1] =~ /^(maintainer|supporter)/i);
1780 $selected{$count} = 1;
1781 $authored{$count} = 0;
1782 $signed{$count} = 0;
1783 $count++;
1784 }
1785
1786 #menu loop
1787 my $done = 0;
1788 my $print_options = 0;
1789 my $redraw = 1;
1790 while (!$done) {
1791 $count = 0;
1792 if ($redraw) {
1793 printf STDERR "\n%1s %2s %-65s",
1794 "*", "#", "email/list and role:stats";
1795 if ($email_git ||
1796 ($email_git_fallback && !$maintained) ||
1797 $email_git_blame) {
1798 print STDERR "auth sign";
1799 }
1800 print STDERR "\n";
1801 foreach my $entry (@list) {
1802 my $email = $entry->[0];
1803 my $role = $entry->[1];
1804 my $sel = "";
1805 $sel = "*" if ($selected{$count});
1806 my $commit_author = $commit_author_hash{$email};
1807 my $commit_signer = $commit_signer_hash{$email};
1808 my $authored = 0;
1809 my $signed = 0;
1810 $authored++ for (@{$commit_author});
1811 $signed++ for (@{$commit_signer});
1812 printf STDERR "%1s %2d %-65s", $sel, $count + 1, $email;
1813 printf STDERR "%4d %4d", $authored, $signed
1814 if ($authored > 0 || $signed > 0);
1815 printf STDERR "\n %s\n", $role;
1816 if ($authored{$count}) {
1817 my $commit_author = $commit_author_hash{$email};
1818 foreach my $ref (@{$commit_author}) {
1819 print STDERR " Author: @{$ref}[1]\n";
1820 }
1821 }
1822 if ($signed{$count}) {
1823 my $commit_signer = $commit_signer_hash{$email};
1824 foreach my $ref (@{$commit_signer}) {
1825 print STDERR " @{$ref}[2]: @{$ref}[1]\n";
1826 }
1827 }
1828
1829 $count++;
1830 }
1831 }
1832 my $date_ref = \$email_git_since;
1833 $date_ref = \$email_hg_since if (vcs_is_hg());
1834 if ($print_options) {
1835 $print_options = 0;
1836 if (vcs_exists()) {
1837 print STDERR <<EOT
1838
1839 Version Control options:
1840 g use git history [$email_git]
1841 gf use git-fallback [$email_git_fallback]
1842 b use git blame [$email_git_blame]
1843 bs use blame signatures [$email_git_blame_signatures]
1844 c# minimum commits [$email_git_min_signatures]
1845 %# min percent [$email_git_min_percent]
1846 d# history to use [$$date_ref]
1847 x# max maintainers [$email_git_max_maintainers]
1848 t all signature types [$email_git_all_signature_types]
1849 m use .mailmap [$email_use_mailmap]
1850 EOT
1851 }
1852 print STDERR <<EOT
1853
1854 Additional options:
1855 0 toggle all
1856 tm toggle maintainers
1857 tg toggle git entries
1858 tl toggle open list entries
1859 ts toggle subscriber list entries
1860 f emails in file [$email_file_emails]
1861 k keywords in file [$keywords]
1862 r remove duplicates [$email_remove_duplicates]
1863 p# pattern match depth [$pattern_depth]
1864 EOT
1865 }
1866 print STDERR
1867 "\n#(toggle), A#(author), S#(signed) *(all), ^(none), O(options), Y(approve): ";
1868
1869 my $input = <STDIN>;
1870 chomp($input);
1871
1872 $redraw = 1;
1873 my $rerun = 0;
1874 my @wish = split(/[, ]+/, $input);
1875 foreach my $nr (@wish) {
1876 $nr = lc($nr);
1877 my $sel = substr($nr, 0, 1);
1878 my $str = substr($nr, 1);
1879 my $val = 0;
1880 $val = $1 if $str =~ /^(\d+)$/;
1881
1882 if ($sel eq "y") {
1883 $interactive = 0;
1884 $done = 1;
1885 $output_rolestats = 0;
1886 $output_roles = 0;
1887 last;
1888 } elsif ($nr =~ /^\d+$/ && $nr > 0 && $nr <= $count) {
1889 $selected{$nr - 1} = !$selected{$nr - 1};
1890 } elsif ($sel eq "*" || $sel eq '^') {
1891 my $toggle = 0;
1892 $toggle = 1 if ($sel eq '*');
1893 for (my $i = 0; $i < $count; $i++) {
1894 $selected{$i} = $toggle;
1895 }
1896 } elsif ($sel eq "0") {
1897 for (my $i = 0; $i < $count; $i++) {
1898 $selected{$i} = !$selected{$i};
1899 }
1900 } elsif ($sel eq "t") {
1901 if (lc($str) eq "m") {
1902 for (my $i = 0; $i < $count; $i++) {
1903 $selected{$i} = !$selected{$i}
1904 if ($list[$i]->[1] =~ /^(maintainer|supporter)/i);
1905 }
1906 } elsif (lc($str) eq "g") {
1907 for (my $i = 0; $i < $count; $i++) {
1908 $selected{$i} = !$selected{$i}
1909 if ($list[$i]->[1] =~ /^(author|commit|signer)/i);
1910 }
1911 } elsif (lc($str) eq "l") {
1912 for (my $i = 0; $i < $count; $i++) {
1913 $selected{$i} = !$selected{$i}
1914 if ($list[$i]->[1] =~ /^(open list)/i);
1915 }
1916 } elsif (lc($str) eq "s") {
1917 for (my $i = 0; $i < $count; $i++) {
1918 $selected{$i} = !$selected{$i}
1919 if ($list[$i]->[1] =~ /^(subscriber list)/i);
1920 }
1921 }
1922 } elsif ($sel eq "a") {
1923 if ($val > 0 && $val <= $count) {
1924 $authored{$val - 1} = !$authored{$val - 1};
1925 } elsif ($str eq '*' || $str eq '^') {
1926 my $toggle = 0;
1927 $toggle = 1 if ($str eq '*');
1928 for (my $i = 0; $i < $count; $i++) {
1929 $authored{$i} = $toggle;
1930 }
1931 }
1932 } elsif ($sel eq "s") {
1933 if ($val > 0 && $val <= $count) {
1934 $signed{$val - 1} = !$signed{$val - 1};
1935 } elsif ($str eq '*' || $str eq '^') {
1936 my $toggle = 0;
1937 $toggle = 1 if ($str eq '*');
1938 for (my $i = 0; $i < $count; $i++) {
1939 $signed{$i} = $toggle;
1940 }
1941 }
1942 } elsif ($sel eq "o") {
1943 $print_options = 1;
1944 $redraw = 1;
1945 } elsif ($sel eq "g") {
1946 if ($str eq "f") {
1947 bool_invert(\$email_git_fallback);
1948 } else {
1949 bool_invert(\$email_git);
1950 }
1951 $rerun = 1;
1952 } elsif ($sel eq "b") {
1953 if ($str eq "s") {
1954 bool_invert(\$email_git_blame_signatures);
1955 } else {
1956 bool_invert(\$email_git_blame);
1957 }
1958 $rerun = 1;
1959 } elsif ($sel eq "c") {
1960 if ($val > 0) {
1961 $email_git_min_signatures = $val;
1962 $rerun = 1;
1963 }
1964 } elsif ($sel eq "x") {
1965 if ($val > 0) {
1966 $email_git_max_maintainers = $val;
1967 $rerun = 1;
1968 }
1969 } elsif ($sel eq "%") {
1970 if ($str ne "" && $val >= 0) {
1971 $email_git_min_percent = $val;
1972 $rerun = 1;
1973 }
1974 } elsif ($sel eq "d") {
1975 if (vcs_is_git()) {
1976 $email_git_since = $str;
1977 } elsif (vcs_is_hg()) {
1978 $email_hg_since = $str;
1979 }
1980 $rerun = 1;
1981 } elsif ($sel eq "t") {
1982 bool_invert(\$email_git_all_signature_types);
1983 $rerun = 1;
1984 } elsif ($sel eq "f") {
1985 bool_invert(\$email_file_emails);
1986 $rerun = 1;
1987 } elsif ($sel eq "r") {
1988 bool_invert(\$email_remove_duplicates);
1989 $rerun = 1;
1990 } elsif ($sel eq "m") {
1991 bool_invert(\$email_use_mailmap);
1992 read_mailmap();
1993 $rerun = 1;
1994 } elsif ($sel eq "k") {
1995 bool_invert(\$keywords);
1996 $rerun = 1;
1997 } elsif ($sel eq "p") {
1998 if ($str ne "" && $val >= 0) {
1999 $pattern_depth = $val;
2000 $rerun = 1;
2001 }
2002 } elsif ($sel eq "h" || $sel eq "?") {
2003 print STDERR <<EOT
2004
2005 Interactive mode allows you to select the various maintainers, submitters,
2006 commit signers and mailing lists that could be CC'd on a patch.
2007
2008 Any *'d entry is selected.
2009
2010 If you have git or hg installed, you can choose to summarize the commit
2011 history of files in the patch. Also, each line of the current file can
2012 be matched to its commit author and that commits signers with blame.
2013
2014 Various knobs exist to control the length of time for active commit
2015 tracking, the maximum number of commit authors and signers to add,
2016 and such.
2017
2018 Enter selections at the prompt until you are satisfied that the selected
2019 maintainers are appropriate. You may enter multiple selections separated
2020 by either commas or spaces.
2021
2022 EOT
2023 } else {
2024 print STDERR "invalid option: '$nr'\n";
2025 $redraw = 0;
2026 }
2027 }
2028 if ($rerun) {
2029 print STDERR "git-blame can be very slow, please have patience..."
2030 if ($email_git_blame);
2031 goto &get_maintainers;
2032 }
2033 }
2034
2035 #drop not selected entries
2036 $count = 0;
2037 my @new_emailto = ();
2038 foreach my $entry (@list) {
2039 if ($selected{$count}) {
2040 push(@new_emailto, $list[$count]);
2041 }
2042 $count++;
2043 }
2044 return @new_emailto;
2045 }
2046
2047 sub bool_invert {
2048 my ($bool_ref) = @_;
2049
2050 if ($$bool_ref) {
2051 $$bool_ref = 0;
2052 } else {
2053 $$bool_ref = 1;
2054 }
2055 }
2056
2057 sub deduplicate_email {
2058 my ($email) = @_;
2059
2060 my $matched = 0;
2061 my ($name, $address) = parse_email($email);
2062 $email = format_email($name, $address, 1);
2063 $email = mailmap_email($email);
2064
2065 return $email if (!$email_remove_duplicates);
2066
2067 ($name, $address) = parse_email($email);
2068
2069 if ($name ne "" && $deduplicate_name_hash{lc($name)}) {
2070 $name = $deduplicate_name_hash{lc($name)}->[0];
2071 $address = $deduplicate_name_hash{lc($name)}->[1];
2072 $matched = 1;
2073 } elsif ($deduplicate_address_hash{lc($address)}) {
2074 $name = $deduplicate_address_hash{lc($address)}->[0];
2075 $address = $deduplicate_address_hash{lc($address)}->[1];
2076 $matched = 1;
2077 }
2078 if (!$matched) {
2079 $deduplicate_name_hash{lc($name)} = [ $name, $address ];
2080 $deduplicate_address_hash{lc($address)} = [ $name, $address ];
2081 }
2082 $email = format_email($name, $address, 1);
2083 $email = mailmap_email($email);
2084 return $email;
2085 }
2086
2087 sub save_commits_by_author {
2088 my (@lines) = @_;
2089
2090 my @authors = ();
2091 my @commits = ();
2092 my @subjects = ();
2093
2094 foreach my $line (@lines) {
2095 if ($line =~ m/$VCS_cmds{"author_pattern"}/) {
2096 my $author = $1;
2097 $author = deduplicate_email($author);
2098 push(@authors, $author);
2099 }
2100 push(@commits, $1) if ($line =~ m/$VCS_cmds{"commit_pattern"}/);
2101 push(@subjects, $1) if ($line =~ m/$VCS_cmds{"subject_pattern"}/);
2102 }
2103
2104 for (my $i = 0; $i < @authors; $i++) {
2105 my $exists = 0;
2106 foreach my $ref(@{$commit_author_hash{$authors[$i]}}) {
2107 if (@{$ref}[0] eq $commits[$i] &&
2108 @{$ref}[1] eq $subjects[$i]) {
2109 $exists = 1;
2110 last;
2111 }
2112 }
2113 if (!$exists) {
2114 push(@{$commit_author_hash{$authors[$i]}},
2115 [ ($commits[$i], $subjects[$i]) ]);
2116 }
2117 }
2118 }
2119
2120 sub save_commits_by_signer {
2121 my (@lines) = @_;
2122
2123 my $commit = "";
2124 my $subject = "";
2125
2126 foreach my $line (@lines) {
2127 $commit = $1 if ($line =~ m/$VCS_cmds{"commit_pattern"}/);
2128 $subject = $1 if ($line =~ m/$VCS_cmds{"subject_pattern"}/);
2129 if ($line =~ /^[ \t]*${signature_pattern}.*\@.*$/) {
2130 my @signatures = ($line);
2131 my ($types_ref, $signers_ref) = extract_formatted_signatures(@signatures);
2132 my @types = @$types_ref;
2133 my @signers = @$signers_ref;
2134
2135 my $type = $types[0];
2136 my $signer = $signers[0];
2137
2138 $signer = deduplicate_email($signer);
2139
2140 my $exists = 0;
2141 foreach my $ref(@{$commit_signer_hash{$signer}}) {
2142 if (@{$ref}[0] eq $commit &&
2143 @{$ref}[1] eq $subject &&
2144 @{$ref}[2] eq $type) {
2145 $exists = 1;
2146 last;
2147 }
2148 }
2149 if (!$exists) {
2150 push(@{$commit_signer_hash{$signer}},
2151 [ ($commit, $subject, $type) ]);
2152 }
2153 }
2154 }
2155 }
2156
2157 sub vcs_assign {
2158 my ($role, $divisor, @lines) = @_;
2159
2160 my %hash;
2161 my $count = 0;
2162
2163 return if (@lines <= 0);
2164
2165 if ($divisor <= 0) {
2166 warn("Bad divisor in " . (caller(0))[3] . ": $divisor\n");
2167 $divisor = 1;
2168 }
2169
2170 @lines = mailmap(@lines);
2171
2172 return if (@lines <= 0);
2173
2174 @lines = sort(@lines);
2175
2176 # uniq -c
2177 $hash{$_}++ for @lines;
2178
2179 # sort -rn
2180 foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) {
2181 my $sign_offs = $hash{$line};
2182 my $percent = $sign_offs * 100 / $divisor;
2183
2184 $percent = 100 if ($percent > 100);
2185 next if (ignore_email_address($line));
2186 $count++;
2187 last if ($sign_offs < $email_git_min_signatures ||
2188 $count > $email_git_max_maintainers ||
2189 $percent < $email_git_min_percent);
2190 push_email_address($line, '');
2191 if ($output_rolestats) {
2192 my $fmt_percent = sprintf("%.0f", $percent);
2193 add_role($line, "$role:$sign_offs/$divisor=$fmt_percent%");
2194 } else {
2195 add_role($line, $role);
2196 }
2197 }
2198 }
2199
2200 sub vcs_file_signoffs {
2201 my ($file) = @_;
2202
2203 my $authors_ref;
2204 my $signers_ref;
2205 my $stats_ref;
2206 my @authors = ();
2207 my @signers = ();
2208 my @stats = ();
2209 my $commits;
2210
2211 $vcs_used = vcs_exists();
2212 return if (!$vcs_used);
2213
2214 my $cmd = $VCS_cmds{"find_signers_cmd"};
2215 $cmd =~ s/(\$\w+)/$1/eeg; # interpolate $cmd
2216
2217 ($commits, $signers_ref, $authors_ref, $stats_ref) = vcs_find_signers($cmd, $file);
2218
2219 @signers = @{$signers_ref} if defined $signers_ref;
2220 @authors = @{$authors_ref} if defined $authors_ref;
2221 @stats = @{$stats_ref} if defined $stats_ref;
2222
2223 # print("commits: <$commits>\nsigners:<@signers>\nauthors: <@authors>\nstats: <@stats>\n");
2224
2225 foreach my $signer (@signers) {
2226 $signer = deduplicate_email($signer);
2227 }
2228
2229 vcs_assign("commit_signer", $commits, @signers);
2230 vcs_assign("authored", $commits, @authors);
2231 if ($#authors == $#stats) {
2232 my $stat_pattern = $VCS_cmds{"stat_pattern"};
2233 $stat_pattern =~ s/(\$\w+)/$1/eeg; #interpolate $stat_pattern
2234
2235 my $added = 0;
2236 my $deleted = 0;
2237 for (my $i = 0; $i <= $#stats; $i++) {
2238 if ($stats[$i] =~ /$stat_pattern/) {
2239 $added += $1;
2240 $deleted += $2;
2241 }
2242 }
2243 my @tmp_authors = uniq(@authors);
2244 foreach my $author (@tmp_authors) {
2245 $author = deduplicate_email($author);
2246 }
2247 @tmp_authors = uniq(@tmp_authors);
2248 my @list_added = ();
2249 my @list_deleted = ();
2250 foreach my $author (@tmp_authors) {
2251 my $auth_added = 0;
2252 my $auth_deleted = 0;
2253 for (my $i = 0; $i <= $#stats; $i++) {
2254 if ($author eq deduplicate_email($authors[$i]) &&
2255 $stats[$i] =~ /$stat_pattern/) {
2256 $auth_added += $1;
2257 $auth_deleted += $2;
2258 }
2259 }
2260 for (my $i = 0; $i < $auth_added; $i++) {
2261 push(@list_added, $author);
2262 }
2263 for (my $i = 0; $i < $auth_deleted; $i++) {
2264 push(@list_deleted, $author);
2265 }
2266 }
2267 vcs_assign("added_lines", $added, @list_added);
2268 vcs_assign("removed_lines", $deleted, @list_deleted);
2269 }
2270 }
2271
2272 sub vcs_file_blame {
2273 my ($file) = @_;
2274
2275 my @signers = ();
2276 my @all_commits = ();
2277 my @commits = ();
2278 my $total_commits;
2279 my $total_lines;
2280
2281 $vcs_used = vcs_exists();
2282 return if (!$vcs_used);
2283
2284 @all_commits = vcs_blame($file);
2285 @commits = uniq(@all_commits);
2286 $total_commits = @commits;
2287 $total_lines = @all_commits;
2288
2289 if ($email_git_blame_signatures) {
2290 if (vcs_is_hg()) {
2291 my $commit_count;
2292 my $commit_authors_ref;
2293 my $commit_signers_ref;
2294 my $stats_ref;
2295 my @commit_authors = ();
2296 my @commit_signers = ();
2297 my $commit = join(" -r ", @commits);
2298 my $cmd;
2299
2300 $cmd = $VCS_cmds{"find_commit_signers_cmd"};
2301 $cmd =~ s/(\$\w+)/$1/eeg; #substitute variables in $cmd
2302
2303 ($commit_count, $commit_signers_ref, $commit_authors_ref, $stats_ref) = vcs_find_signers($cmd, $file);
2304 @commit_authors = @{$commit_authors_ref} if defined $commit_authors_ref;
2305 @commit_signers = @{$commit_signers_ref} if defined $commit_signers_ref;
2306
2307 push(@signers, @commit_signers);
2308 } else {
2309 foreach my $commit (@commits) {
2310 my $commit_count;
2311 my $commit_authors_ref;
2312 my $commit_signers_ref;
2313 my $stats_ref;
2314 my @commit_authors = ();
2315 my @commit_signers = ();
2316 my $cmd;
2317
2318 $cmd = $VCS_cmds{"find_commit_signers_cmd"};
2319 $cmd =~ s/(\$\w+)/$1/eeg; #substitute variables in $cmd
2320
2321 ($commit_count, $commit_signers_ref, $commit_authors_ref, $stats_ref) = vcs_find_signers($cmd, $file);
2322 @commit_authors = @{$commit_authors_ref} if defined $commit_authors_ref;
2323 @commit_signers = @{$commit_signers_ref} if defined $commit_signers_ref;
2324
2325 push(@signers, @commit_signers);
2326 }
2327 }
2328 }
2329
2330 if ($from_filename) {
2331 if ($output_rolestats) {
2332 my @blame_signers;
2333 if (vcs_is_hg()) {{ # Double brace for last exit
2334 my $commit_count;
2335 my @commit_signers = ();
2336 @commits = uniq(@commits);
2337 @commits = sort(@commits);
2338 my $commit = join(" -r ", @commits);
2339 my $cmd;
2340
2341 $cmd = $VCS_cmds{"find_commit_author_cmd"};
2342 $cmd =~ s/(\$\w+)/$1/eeg; #substitute variables in $cmd
2343
2344 my @lines = ();
2345
2346 @lines = &{$VCS_cmds{"execute_cmd"}}($cmd);
2347
2348 if (!$email_git_penguin_chiefs) {
2349 @lines = grep(!/${penguin_chiefs}/i, @lines);
2350 }
2351
2352 last if !@lines;
2353
2354 my @authors = ();
2355 foreach my $line (@lines) {
2356 if ($line =~ m/$VCS_cmds{"author_pattern"}/) {
2357 my $author = $1;
2358 $author = deduplicate_email($author);
2359 push(@authors, $author);
2360 }
2361 }
2362
2363 save_commits_by_author(@lines) if ($interactive);
2364 save_commits_by_signer(@lines) if ($interactive);
2365
2366 push(@signers, @authors);
2367 }}
2368 else {
2369 foreach my $commit (@commits) {
2370 my $i;
2371 my $cmd = $VCS_cmds{"find_commit_author_cmd"};
2372 $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd
2373 my @author = vcs_find_author($cmd);
2374 next if !@author;
2375
2376 my $formatted_author = deduplicate_email($author[0]);
2377
2378 my $count = grep(/$commit/, @all_commits);
2379 for ($i = 0; $i < $count ; $i++) {
2380 push(@blame_signers, $formatted_author);
2381 }
2382 }
2383 }
2384 if (@blame_signers) {
2385 vcs_assign("authored lines", $total_lines, @blame_signers);
2386 }
2387 }
2388 foreach my $signer (@signers) {
2389 $signer = deduplicate_email($signer);
2390 }
2391 vcs_assign("commits", $total_commits, @signers);
2392 } else {
2393 foreach my $signer (@signers) {
2394 $signer = deduplicate_email($signer);
2395 }
2396 vcs_assign("modified commits", $total_commits, @signers);
2397 }
2398 }
2399
2400 sub vcs_file_exists {
2401 my ($file) = @_;
2402
2403 my $exists;
2404
2405 my $vcs_used = vcs_exists();
2406 return 0 if (!$vcs_used);
2407
2408 my $cmd = $VCS_cmds{"file_exists_cmd"};
2409 $cmd =~ s/(\$\w+)/$1/eeg; # interpolate $cmd
2410 $cmd .= " 2>&1";
2411 $exists = &{$VCS_cmds{"execute_cmd"}}($cmd);
2412
2413 return 0 if ($? != 0);
2414
2415 return $exists;
2416 }
2417
2418 sub vcs_list_files {
2419 my ($file) = @_;
2420
2421 my @lsfiles = ();
2422
2423 my $vcs_used = vcs_exists();
2424 return 0 if (!$vcs_used);
2425
2426 my $cmd = $VCS_cmds{"list_files_cmd"};
2427 $cmd =~ s/(\$\w+)/$1/eeg; # interpolate $cmd
2428 @lsfiles = &{$VCS_cmds{"execute_cmd"}}($cmd);
2429
2430 return () if ($? != 0);
2431
2432 return @lsfiles;
2433 }
2434
2435 sub uniq {
2436 my (@parms) = @_;
2437
2438 my %saw;
2439 @parms = grep(!$saw{$_}++, @parms);
2440 return @parms;
2441 }
2442
2443 sub sort_and_uniq {
2444 my (@parms) = @_;
2445
2446 my %saw;
2447 @parms = sort @parms;
2448 @parms = grep(!$saw{$_}++, @parms);
2449 return @parms;
2450 }
2451
2452 sub clean_file_emails {
2453 my (@file_emails) = @_;
2454 my @fmt_emails = ();
2455
2456 foreach my $email (@file_emails) {
2457 $email =~ s/[\(\<\{]{0,1}([A-Za-z0-9_\.\+-]+\@[A-Za-z0-9\.-]+)[\)\>\}]{0,1}/\<$1\>/g;
2458 my ($name, $address) = parse_email($email);
2459 if ($name eq '"[,\.]"') {
2460 $name = "";
2461 }
2462
2463 my @nw = split(/[^A-Za-zÀ-ÿ\'\,\.\+-]/, $name);
2464 if (@nw > 2) {
2465 my $first = $nw[@nw - 3];
2466 my $middle = $nw[@nw - 2];
2467 my $last = $nw[@nw - 1];
2468
2469 if (((length($first) == 1 && $first =~ m/[A-Za-z]/) ||
2470 (length($first) == 2 && substr($first, -1) eq ".")) ||
2471 (length($middle) == 1 ||
2472 (length($middle) == 2 && substr($middle, -1) eq "."))) {
2473 $name = "$first $middle $last";
2474 } else {
2475 $name = "$middle $last";
2476 }
2477 }
2478
2479 if (substr($name, -1) =~ /[,\.]/) {
2480 $name = substr($name, 0, length($name) - 1);
2481 } elsif (substr($name, -2) =~ /[,\.]"/) {
2482 $name = substr($name, 0, length($name) - 2) . '"';
2483 }
2484
2485 if (substr($name, 0, 1) =~ /[,\.]/) {
2486 $name = substr($name, 1, length($name) - 1);
2487 } elsif (substr($name, 0, 2) =~ /"[,\.]/) {
2488 $name = '"' . substr($name, 2, length($name) - 2);
2489 }
2490
2491 my $fmt_email = format_email($name, $address, $email_usename);
2492 push(@fmt_emails, $fmt_email);
2493 }
2494 return @fmt_emails;
2495 }
2496
2497 sub merge_email {
2498 my @lines;
2499 my %saw;
2500
2501 for (@_) {
2502 my ($address, $role) = @$_;
2503 if (!$saw{$address}) {
2504 if ($output_roles) {
2505 push(@lines, "$address ($role)");
2506 } else {
2507 push(@lines, $address);
2508 }
2509 $saw{$address} = 1;
2510 }
2511 }
2512
2513 return @lines;
2514 }
2515
2516 sub output {
2517 my (@parms) = @_;
2518
2519 if ($output_multiline) {
2520 foreach my $line (@parms) {
2521 print("${line}\n");
2522 }
2523 } else {
2524 print(join($output_separator, @parms));
2525 print("\n");
2526 }
2527 }
2528
2529 my $rfc822re;
2530
2531 sub make_rfc822re {
2532 # Basic lexical tokens are specials, domain_literal, quoted_string, atom, and
2533 # comment. We must allow for rfc822_lwsp (or comments) after each of these.
2534 # This regexp will only work on addresses which have had comments stripped
2535 # and replaced with rfc822_lwsp.
2536
2537 my $specials = '()<>@,;:\\\\".\\[\\]';
2538 my $controls = '\\000-\\037\\177';
2539
2540 my $dtext = "[^\\[\\]\\r\\\\]";
2541 my $domain_literal = "\\[(?:$dtext|\\\\.)*\\]$rfc822_lwsp*";
2542
2543 my $quoted_string = "\"(?:[^\\\"\\r\\\\]|\\\\.|$rfc822_lwsp)*\"$rfc822_lwsp*";
2544
2545 # Use zero-width assertion to spot the limit of an atom. A simple
2546 # $rfc822_lwsp* causes the regexp engine to hang occasionally.
2547 my $atom = "[^$specials $controls]+(?:$rfc822_lwsp+|\\Z|(?=[\\[\"$specials]))";
2548 my $word = "(?:$atom|$quoted_string)";
2549 my $localpart = "$word(?:\\.$rfc822_lwsp*$word)*";
2550
2551 my $sub_domain = "(?:$atom|$domain_literal)";
2552 my $domain = "$sub_domain(?:\\.$rfc822_lwsp*$sub_domain)*";
2553
2554 my $addr_spec = "$localpart\@$rfc822_lwsp*$domain";
2555
2556 my $phrase = "$word*";
2557 my $route = "(?:\@$domain(?:,\@$rfc822_lwsp*$domain)*:$rfc822_lwsp*)";
2558 my $route_addr = "\\<$rfc822_lwsp*$route?$addr_spec\\>$rfc822_lwsp*";
2559 my $mailbox = "(?:$addr_spec|$phrase$route_addr)";
2560
2561 my $group = "$phrase:$rfc822_lwsp*(?:$mailbox(?:,\\s*$mailbox)*)?;\\s*";
2562 my $address = "(?:$mailbox|$group)";
2563
2564 return "$rfc822_lwsp*$address";
2565 }
2566
2567 sub rfc822_strip_comments {
2568 my $s = shift;
2569 # Recursively remove comments, and replace with a single space. The simpler
2570 # regexps in the Email Addressing FAQ are imperfect - they will miss escaped
2571 # chars in atoms, for example.
2572
2573 while ($s =~ s/^((?:[^"\\]|\\.)*
2574 (?:"(?:[^"\\]|\\.)*"(?:[^"\\]|\\.)*)*)
2575 \((?:[^()\\]|\\.)*\)/$1 /osx) {}
2576 return $s;
2577 }
2578
2579 # valid: returns true if the parameter is an RFC822 valid address
2580 #
2581 sub rfc822_valid {
2582 my $s = rfc822_strip_comments(shift);
2583
2584 if (!$rfc822re) {
2585 $rfc822re = make_rfc822re();
2586 }
2587
2588 return $s =~ m/^$rfc822re$/so && $s =~ m/^$rfc822_char*$/;
2589 }
2590
2591 # validlist: In scalar context, returns true if the parameter is an RFC822
2592 # valid list of addresses.
2593 #
2594 # In list context, returns an empty list on failure (an invalid
2595 # address was found); otherwise a list whose first element is the
2596 # number of addresses found and whose remaining elements are the
2597 # addresses. This is needed to disambiguate failure (invalid)
2598 # from success with no addresses found, because an empty string is
2599 # a valid list.
2600
2601 sub rfc822_validlist {
2602 my $s = rfc822_strip_comments(shift);
2603
2604 if (!$rfc822re) {
2605 $rfc822re = make_rfc822re();
2606 }
2607 # * null list items are valid according to the RFC
2608 # * the '1' business is to aid in distinguishing failure from no results
2609
2610 my @r;
2611 if ($s =~ m/^(?:$rfc822re)?(?:,(?:$rfc822re)?)*$/so &&
2612 $s =~ m/^$rfc822_char*$/) {
2613 while ($s =~ m/(?:^|,$rfc822_lwsp*)($rfc822re)/gos) {
2614 push(@r, $1);
2615 }
2616 return wantarray ? (scalar(@r), @r) : 1;
2617 }
2618 return wantarray ? () : 0;
2619 }