Commit | Line | Data |
---|---|---|
cb7301c7 JP |
1 | #!/usr/bin/perl -w |
2 | # (c) 2007, Joe Perches <joe@perches.com> | |
3 | # created from checkpatch.pl | |
4 | # | |
5 | # Print selected MAINTAINERS information for | |
6 | # the files modified in a patch or for a file | |
7 | # | |
3bd7bf5f RK |
8 | # usage: perl scripts/get_maintainer.pl [OPTIONS] <patch> |
9 | # perl scripts/get_maintainer.pl [OPTIONS] -f <file> | |
cb7301c7 JP |
10 | # |
11 | # Licensed under the terms of the GNU GPL License version 2 | |
12 | ||
13 | use strict; | |
14 | ||
15 | my $P = $0; | |
63ab52db | 16 | my $V = '0.25'; |
cb7301c7 JP |
17 | |
18 | use Getopt::Long qw(:config no_auto_abbrev); | |
19 | ||
20 | my $lk_path = "./"; | |
21 | my $email = 1; | |
22 | my $email_usename = 1; | |
23 | my $email_maintainer = 1; | |
24 | my $email_list = 1; | |
25 | my $email_subscriber_list = 0; | |
cb7301c7 | 26 | my $email_git_penguin_chiefs = 0; |
e3e9d114 | 27 | my $email_git = 0; |
0fa05599 | 28 | my $email_git_all_signature_types = 0; |
60db31ac | 29 | my $email_git_blame = 0; |
e3e9d114 | 30 | my $email_git_fallback = 1; |
cb7301c7 JP |
31 | my $email_git_min_signatures = 1; |
32 | my $email_git_max_maintainers = 5; | |
afa81ee1 | 33 | my $email_git_min_percent = 5; |
cb7301c7 | 34 | my $email_git_since = "1-year-ago"; |
60db31ac | 35 | my $email_hg_since = "-365"; |
11ecf53c | 36 | my $email_remove_duplicates = 1; |
cb7301c7 JP |
37 | my $output_multiline = 1; |
38 | my $output_separator = ", "; | |
3c7385b8 JP |
39 | my $output_roles = 0; |
40 | my $output_rolestats = 0; | |
cb7301c7 JP |
41 | my $scm = 0; |
42 | my $web = 0; | |
43 | my $subsystem = 0; | |
44 | my $status = 0; | |
dcf36a92 | 45 | my $keywords = 1; |
4b76c9da | 46 | my $sections = 0; |
03372dbb | 47 | my $file_emails = 0; |
4a7fdb5f | 48 | my $from_filename = 0; |
3fb55652 | 49 | my $pattern_depth = 0; |
cb7301c7 JP |
50 | my $version = 0; |
51 | my $help = 0; | |
52 | ||
53 | my $exit = 0; | |
54 | ||
55 | my @penguin_chief = (); | |
e4d26b02 | 56 | push(@penguin_chief, "Linus Torvalds:torvalds\@linux-foundation.org"); |
cb7301c7 | 57 | #Andrew wants in on most everything - 2009/01/14 |
e4d26b02 | 58 | #push(@penguin_chief, "Andrew Morton:akpm\@linux-foundation.org"); |
cb7301c7 JP |
59 | |
60 | my @penguin_chief_names = (); | |
61 | foreach my $chief (@penguin_chief) { | |
62 | if ($chief =~ m/^(.*):(.*)/) { | |
63 | my $chief_name = $1; | |
64 | my $chief_addr = $2; | |
65 | push(@penguin_chief_names, $chief_name); | |
66 | } | |
67 | } | |
e4d26b02 JP |
68 | my $penguin_chiefs = "\(" . join("|", @penguin_chief_names) . "\)"; |
69 | ||
70 | # Signature types of people who are either | |
71 | # a) responsible for the code in question, or | |
72 | # b) familiar enough with it to give relevant feedback | |
73 | my @signature_tags = (); | |
74 | push(@signature_tags, "Signed-off-by:"); | |
75 | push(@signature_tags, "Reviewed-by:"); | |
76 | push(@signature_tags, "Acked-by:"); | |
77 | my $signaturePattern = "\(" . join("|", @signature_tags) . "\)"; | |
cb7301c7 | 78 | |
5f2441e9 | 79 | # rfc822 email address - preloaded methods go here. |
1b5e1cf6 | 80 | my $rfc822_lwsp = "(?:(?:\\r\\n)?[ \\t])"; |
df4cc036 | 81 | my $rfc822_char = '[\\000-\\377]'; |
1b5e1cf6 | 82 | |
60db31ac JP |
83 | # VCS command support: class-like functions and strings |
84 | ||
85 | my %VCS_cmds; | |
86 | ||
87 | my %VCS_cmds_git = ( | |
88 | "execute_cmd" => \&git_execute_cmd, | |
89 | "available" => '(which("git") ne "") && (-d ".git")', | |
99cf6116 RK |
90 | "find_signers_cmd" => "git log --no-color --since=\$email_git_since -- \$file", |
91 | "find_commit_signers_cmd" => "git log --no-color -1 \$commit", | |
63ab52db | 92 | "find_commit_author_cmd" => "git log -1 --format=\"%an <%ae>\" \$commit", |
60db31ac JP |
93 | "blame_range_cmd" => "git blame -l -L \$diff_start,+\$diff_length \$file", |
94 | "blame_file_cmd" => "git blame -l \$file", | |
95 | "commit_pattern" => "^commit [0-9a-f]{40,40}", | |
96 | "blame_commit_pattern" => "^([0-9a-f]+) " | |
97 | ); | |
98 | ||
99 | my %VCS_cmds_hg = ( | |
100 | "execute_cmd" => \&hg_execute_cmd, | |
101 | "available" => '(which("hg") ne "") && (-d ".hg")', | |
102 | "find_signers_cmd" => | |
103 | "hg log --date=\$email_hg_since" . | |
104 | " --template='commit {node}\\n{desc}\\n' -- \$file", | |
105 | "find_commit_signers_cmd" => "hg log --template='{desc}\\n' -r \$commit", | |
63ab52db | 106 | "find_commit_author_cmd" => "hg log -l 1 --template='{author}\\n' -r \$commit", |
60db31ac JP |
107 | "blame_range_cmd" => "", # not supported |
108 | "blame_file_cmd" => "hg blame -c \$file", | |
109 | "commit_pattern" => "^commit [0-9a-f]{40,40}", | |
110 | "blame_commit_pattern" => "^([0-9a-f]+):" | |
111 | ); | |
112 | ||
368669da JP |
113 | if (-f "${lk_path}.get_maintainer.conf") { |
114 | my @conf_args; | |
115 | open(my $conffile, '<', "${lk_path}.get_maintainer.conf") | |
116 | or warn "$P: Can't open .get_maintainer.conf: $!\n"; | |
117 | while (<$conffile>) { | |
118 | my $line = $_; | |
119 | ||
120 | $line =~ s/\s*\n?$//g; | |
121 | $line =~ s/^\s*//g; | |
122 | $line =~ s/\s+/ /g; | |
123 | ||
124 | next if ($line =~ m/^\s*#/); | |
125 | next if ($line =~ m/^\s*$/); | |
126 | ||
127 | my @words = split(" ", $line); | |
128 | foreach my $word (@words) { | |
129 | last if ($word =~ m/^#/); | |
130 | push (@conf_args, $word); | |
131 | } | |
132 | } | |
133 | close($conffile); | |
134 | unshift(@ARGV, @conf_args) if @conf_args; | |
135 | } | |
136 | ||
cb7301c7 JP |
137 | if (!GetOptions( |
138 | 'email!' => \$email, | |
139 | 'git!' => \$email_git, | |
e4d26b02 | 140 | 'git-all-signature-types!' => \$email_git_all_signature_types, |
60db31ac | 141 | 'git-blame!' => \$email_git_blame, |
e3e9d114 | 142 | 'git-fallback!' => \$email_git_fallback, |
cb7301c7 JP |
143 | 'git-chief-penguins!' => \$email_git_penguin_chiefs, |
144 | 'git-min-signatures=i' => \$email_git_min_signatures, | |
145 | 'git-max-maintainers=i' => \$email_git_max_maintainers, | |
afa81ee1 | 146 | 'git-min-percent=i' => \$email_git_min_percent, |
cb7301c7 | 147 | 'git-since=s' => \$email_git_since, |
60db31ac | 148 | 'hg-since=s' => \$email_hg_since, |
11ecf53c | 149 | 'remove-duplicates!' => \$email_remove_duplicates, |
cb7301c7 JP |
150 | 'm!' => \$email_maintainer, |
151 | 'n!' => \$email_usename, | |
152 | 'l!' => \$email_list, | |
153 | 's!' => \$email_subscriber_list, | |
154 | 'multiline!' => \$output_multiline, | |
3c7385b8 JP |
155 | 'roles!' => \$output_roles, |
156 | 'rolestats!' => \$output_rolestats, | |
cb7301c7 JP |
157 | 'separator=s' => \$output_separator, |
158 | 'subsystem!' => \$subsystem, | |
159 | 'status!' => \$status, | |
160 | 'scm!' => \$scm, | |
161 | 'web!' => \$web, | |
3fb55652 | 162 | 'pattern-depth=i' => \$pattern_depth, |
dcf36a92 | 163 | 'k|keywords!' => \$keywords, |
4b76c9da | 164 | 'sections!' => \$sections, |
03372dbb | 165 | 'fe|file-emails!' => \$file_emails, |
4a7fdb5f | 166 | 'f|file' => \$from_filename, |
cb7301c7 | 167 | 'v|version' => \$version, |
64f77f31 | 168 | 'h|help|usage' => \$help, |
cb7301c7 | 169 | )) { |
3c7385b8 | 170 | die "$P: invalid argument - use --help if necessary\n"; |
cb7301c7 JP |
171 | } |
172 | ||
173 | if ($help != 0) { | |
174 | usage(); | |
175 | exit 0; | |
176 | } | |
177 | ||
178 | if ($version != 0) { | |
179 | print("${P} ${V}\n"); | |
180 | exit 0; | |
181 | } | |
182 | ||
64f77f31 JP |
183 | if (-t STDIN && !@ARGV) { |
184 | # We're talking to a terminal, but have no command line arguments. | |
185 | die "$P: missing patchfile or -f file - use --help if necessary\n"; | |
cb7301c7 JP |
186 | } |
187 | ||
42498316 JP |
188 | if ($output_separator ne ", ") { |
189 | $output_multiline = 0; | |
190 | } | |
191 | ||
3c7385b8 JP |
192 | if ($output_rolestats) { |
193 | $output_roles = 1; | |
194 | } | |
195 | ||
4b76c9da JP |
196 | if ($sections) { |
197 | $email = 0; | |
198 | $email_list = 0; | |
199 | $scm = 0; | |
200 | $status = 0; | |
201 | $subsystem = 0; | |
202 | $web = 0; | |
203 | $keywords = 0; | |
204 | } else { | |
205 | my $selections = $email + $scm + $status + $subsystem + $web; | |
206 | if ($selections == 0) { | |
4b76c9da JP |
207 | die "$P: Missing required option: email, scm, status, subsystem or web\n"; |
208 | } | |
cb7301c7 JP |
209 | } |
210 | ||
f5492666 JP |
211 | if ($email && |
212 | ($email_maintainer + $email_list + $email_subscriber_list + | |
213 | $email_git + $email_git_penguin_chiefs + $email_git_blame) == 0) { | |
cb7301c7 JP |
214 | die "$P: Please select at least 1 email option\n"; |
215 | } | |
216 | ||
217 | if (!top_of_kernel_tree($lk_path)) { | |
218 | die "$P: The current directory does not appear to be " | |
219 | . "a linux kernel source tree.\n"; | |
220 | } | |
221 | ||
e4d26b02 JP |
222 | if ($email_git_all_signature_types) { |
223 | $signaturePattern = "(.+?)[Bb][Yy]:"; | |
224 | } | |
225 | ||
cb7301c7 JP |
226 | ## Read MAINTAINERS for type/value pairs |
227 | ||
228 | my @typevalue = (); | |
dcf36a92 JP |
229 | my %keyword_hash; |
230 | ||
22dd5b0c SH |
231 | open (my $maint, '<', "${lk_path}MAINTAINERS") |
232 | or die "$P: Can't open MAINTAINERS: $!\n"; | |
233 | while (<$maint>) { | |
cb7301c7 JP |
234 | my $line = $_; |
235 | ||
236 | if ($line =~ m/^(\C):\s*(.*)/) { | |
237 | my $type = $1; | |
238 | my $value = $2; | |
239 | ||
240 | ##Filename pattern matching | |
241 | if ($type eq "F" || $type eq "X") { | |
242 | $value =~ s@\.@\\\.@g; ##Convert . to \. | |
243 | $value =~ s/\*/\.\*/g; ##Convert * to .* | |
244 | $value =~ s/\?/\./g; ##Convert ? to . | |
870020f9 JP |
245 | ##if pattern is a directory and it lacks a trailing slash, add one |
246 | if ((-d $value)) { | |
247 | $value =~ s@([^/])$@$1/@; | |
248 | } | |
dcf36a92 JP |
249 | } elsif ($type eq "K") { |
250 | $keyword_hash{@typevalue} = $value; | |
cb7301c7 JP |
251 | } |
252 | push(@typevalue, "$type:$value"); | |
253 | } elsif (!/^(\s)*$/) { | |
254 | $line =~ s/\n$//g; | |
255 | push(@typevalue, $line); | |
256 | } | |
257 | } | |
22dd5b0c | 258 | close($maint); |
cb7301c7 | 259 | |
8cbb3a77 JP |
260 | my %mailmap; |
261 | ||
11ecf53c | 262 | if ($email_remove_duplicates) { |
22dd5b0c SH |
263 | open(my $mailmap, '<', "${lk_path}.mailmap") |
264 | or warn "$P: Can't open .mailmap: $!\n"; | |
265 | while (<$mailmap>) { | |
11ecf53c | 266 | my $line = $_; |
8cbb3a77 | 267 | |
11ecf53c JP |
268 | next if ($line =~ m/^\s*#/); |
269 | next if ($line =~ m/^\s*$/); | |
8cbb3a77 | 270 | |
11ecf53c | 271 | my ($name, $address) = parse_email($line); |
a8af2430 | 272 | $line = format_email($name, $address, $email_usename); |
8cbb3a77 | 273 | |
11ecf53c | 274 | next if ($line =~ m/^\s*$/); |
8cbb3a77 | 275 | |
11ecf53c JP |
276 | if (exists($mailmap{$name})) { |
277 | my $obj = $mailmap{$name}; | |
278 | push(@$obj, $address); | |
279 | } else { | |
280 | my @arr = ($address); | |
281 | $mailmap{$name} = \@arr; | |
282 | } | |
8cbb3a77 | 283 | } |
22dd5b0c | 284 | close($mailmap); |
8cbb3a77 JP |
285 | } |
286 | ||
4a7fdb5f | 287 | ## use the filenames on the command line or find the filenames in the patchfiles |
cb7301c7 JP |
288 | |
289 | my @files = (); | |
f5492666 | 290 | my @range = (); |
dcf36a92 | 291 | my @keyword_tvi = (); |
03372dbb | 292 | my @file_emails = (); |
cb7301c7 | 293 | |
64f77f31 JP |
294 | if (!@ARGV) { |
295 | push(@ARGV, "&STDIN"); | |
296 | } | |
297 | ||
4a7fdb5f | 298 | foreach my $file (@ARGV) { |
64f77f31 JP |
299 | if ($file ne "&STDIN") { |
300 | ##if $file is a directory and it lacks a trailing slash, add one | |
301 | if ((-d $file)) { | |
302 | $file =~ s@([^/])$@$1/@; | |
303 | } elsif (!(-f $file)) { | |
304 | die "$P: file '${file}' not found\n"; | |
305 | } | |
cb7301c7 | 306 | } |
4a7fdb5f JP |
307 | if ($from_filename) { |
308 | push(@files, $file); | |
fab9ed12 | 309 | if ($file ne "MAINTAINERS" && -f $file && ($keywords || $file_emails)) { |
22dd5b0c SH |
310 | open(my $f, '<', $file) |
311 | or die "$P: Can't open $file: $!\n"; | |
312 | my $text = do { local($/) ; <$f> }; | |
313 | close($f); | |
03372dbb JP |
314 | if ($keywords) { |
315 | foreach my $line (keys %keyword_hash) { | |
316 | if ($text =~ m/$keyword_hash{$line}/x) { | |
317 | push(@keyword_tvi, $line); | |
318 | } | |
dcf36a92 JP |
319 | } |
320 | } | |
03372dbb JP |
321 | if ($file_emails) { |
322 | my @poss_addr = $text =~ m$[A-Za-zÀ-ÿ\"\' \,\.\+-]*\s*[\,]*\s*[\(\<\{]{0,1}[A-Za-z0-9_\.\+-]+\@[A-Za-z0-9\.-]+\.[A-Za-z0-9]+[\)\>\}]{0,1}$g; | |
323 | push(@file_emails, clean_file_emails(@poss_addr)); | |
324 | } | |
dcf36a92 | 325 | } |
4a7fdb5f JP |
326 | } else { |
327 | my $file_cnt = @files; | |
f5492666 | 328 | my $lastfile; |
22dd5b0c | 329 | |
3a4df13d | 330 | open(my $patch, "< $file") |
22dd5b0c SH |
331 | or die "$P: Can't open $file: $!\n"; |
332 | while (<$patch>) { | |
dcf36a92 | 333 | my $patch_line = $_; |
4a7fdb5f JP |
334 | if (m/^\+\+\+\s+(\S+)/) { |
335 | my $filename = $1; | |
336 | $filename =~ s@^[^/]*/@@; | |
337 | $filename =~ s@\n@@; | |
f5492666 | 338 | $lastfile = $filename; |
4a7fdb5f | 339 | push(@files, $filename); |
f5492666 JP |
340 | } elsif (m/^\@\@ -(\d+),(\d+)/) { |
341 | if ($email_git_blame) { | |
342 | push(@range, "$lastfile:$1:$2"); | |
343 | } | |
dcf36a92 JP |
344 | } elsif ($keywords) { |
345 | foreach my $line (keys %keyword_hash) { | |
346 | if ($patch_line =~ m/^[+-].*$keyword_hash{$line}/x) { | |
347 | push(@keyword_tvi, $line); | |
348 | } | |
349 | } | |
4a7fdb5f | 350 | } |
cb7301c7 | 351 | } |
22dd5b0c SH |
352 | close($patch); |
353 | ||
4a7fdb5f | 354 | if ($file_cnt == @files) { |
7f29fd27 | 355 | warn "$P: file '${file}' doesn't appear to be a patch. " |
4a7fdb5f JP |
356 | . "Add -f to options?\n"; |
357 | } | |
358 | @files = sort_and_uniq(@files); | |
cb7301c7 | 359 | } |
cb7301c7 JP |
360 | } |
361 | ||
03372dbb JP |
362 | @file_emails = uniq(@file_emails); |
363 | ||
cb7301c7 | 364 | my @email_to = (); |
290603c1 | 365 | my @list_to = (); |
cb7301c7 JP |
366 | my @scm = (); |
367 | my @web = (); | |
368 | my @subsystem = (); | |
369 | my @status = (); | |
370 | ||
371 | # Find responsible parties | |
372 | ||
373 | foreach my $file (@files) { | |
374 | ||
272a8979 | 375 | my %hash; |
e3e9d114 | 376 | my $exact_pattern_match = 0; |
272a8979 JP |
377 | my $tvi = find_first_section(); |
378 | while ($tvi < @typevalue) { | |
379 | my $start = find_starting_index($tvi); | |
380 | my $end = find_ending_index($tvi); | |
381 | my $exclude = 0; | |
382 | my $i; | |
383 | ||
384 | #Do not match excluded file patterns | |
385 | ||
386 | for ($i = $start; $i < $end; $i++) { | |
387 | my $line = $typevalue[$i]; | |
290603c1 | 388 | if ($line =~ m/^(\C):\s*(.*)/) { |
cb7301c7 JP |
389 | my $type = $1; |
390 | my $value = $2; | |
272a8979 | 391 | if ($type eq 'X') { |
cb7301c7 | 392 | if (file_match_pattern($file, $value)) { |
272a8979 | 393 | $exclude = 1; |
3c840c18 | 394 | last; |
cb7301c7 JP |
395 | } |
396 | } | |
397 | } | |
cb7301c7 | 398 | } |
272a8979 JP |
399 | |
400 | if (!$exclude) { | |
401 | for ($i = $start; $i < $end; $i++) { | |
402 | my $line = $typevalue[$i]; | |
403 | if ($line =~ m/^(\C):\s*(.*)/) { | |
404 | my $type = $1; | |
405 | my $value = $2; | |
406 | if ($type eq 'F') { | |
407 | if (file_match_pattern($file, $value)) { | |
408 | my $value_pd = ($value =~ tr@/@@); | |
409 | my $file_pd = ($file =~ tr@/@@); | |
410 | $value_pd++ if (substr($value,-1,1) ne "/"); | |
e3e9d114 JP |
411 | $value_pd = -1 if ($value =~ /^\.\*/); |
412 | $exact_pattern_match = 1 if ($value_pd >= $file_pd); | |
272a8979 JP |
413 | if ($pattern_depth == 0 || |
414 | (($file_pd - $value_pd) < $pattern_depth)) { | |
415 | $hash{$tvi} = $value_pd; | |
416 | } | |
417 | } | |
418 | } | |
419 | } | |
420 | } | |
1d606b4e | 421 | } |
272a8979 | 422 | |
3c840c18 | 423 | $tvi = $end + 1; |
272a8979 JP |
424 | } |
425 | ||
426 | foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) { | |
427 | add_categories($line); | |
6ffd9485 JP |
428 | if ($sections) { |
429 | my $i; | |
430 | my $start = find_starting_index($line); | |
431 | my $end = find_ending_index($line); | |
432 | for ($i = $start; $i < $end; $i++) { | |
433 | my $line = $typevalue[$i]; | |
434 | if ($line =~ /^[FX]:/) { ##Restore file patterns | |
435 | $line =~ s/([^\\])\.([^\*])/$1\?$2/g; | |
436 | $line =~ s/([^\\])\.$/$1\?/g; ##Convert . back to ? | |
437 | $line =~ s/\\\./\./g; ##Convert \. to . | |
438 | $line =~ s/\.\*/\*/g; ##Convert .* to * | |
4b76c9da | 439 | } |
6ffd9485 JP |
440 | $line =~ s/^([A-Z]):/$1:\t/g; |
441 | print("$line\n"); | |
4b76c9da | 442 | } |
6ffd9485 JP |
443 | print("\n"); |
444 | } | |
cb7301c7 JP |
445 | } |
446 | ||
e3e9d114 JP |
447 | if ($email && |
448 | ($email_git || ($email_git_fallback && !$exact_pattern_match))) { | |
60db31ac | 449 | vcs_file_signoffs($file); |
cb7301c7 JP |
450 | } |
451 | ||
f5492666 | 452 | if ($email && $email_git_blame) { |
60db31ac | 453 | vcs_file_blame($file); |
f5492666 | 454 | } |
cb7301c7 JP |
455 | } |
456 | ||
dcf36a92 JP |
457 | if ($keywords) { |
458 | @keyword_tvi = sort_and_uniq(@keyword_tvi); | |
459 | foreach my $line (@keyword_tvi) { | |
460 | add_categories($line); | |
461 | } | |
462 | } | |
463 | ||
f5f5078d | 464 | if ($email) { |
cb7301c7 JP |
465 | foreach my $chief (@penguin_chief) { |
466 | if ($chief =~ m/^(.*):(.*)/) { | |
f5f5078d | 467 | my $email_address; |
0e70e83d | 468 | |
a8af2430 | 469 | $email_address = format_email($1, $2, $email_usename); |
f5f5078d | 470 | if ($email_git_penguin_chiefs) { |
3c7385b8 | 471 | push(@email_to, [$email_address, 'chief penguin']); |
f5f5078d | 472 | } else { |
3c7385b8 | 473 | @email_to = grep($_->[0] !~ /${email_address}/, @email_to); |
cb7301c7 JP |
474 | } |
475 | } | |
476 | } | |
03372dbb JP |
477 | |
478 | foreach my $email (@file_emails) { | |
479 | my ($name, $address) = parse_email($email); | |
480 | ||
481 | my $tmp_email = format_email($name, $address, $email_usename); | |
482 | push_email_address($tmp_email, ''); | |
483 | add_role($tmp_email, 'in file'); | |
484 | } | |
cb7301c7 JP |
485 | } |
486 | ||
290603c1 JP |
487 | if ($email || $email_list) { |
488 | my @to = (); | |
489 | if ($email) { | |
490 | @to = (@to, @email_to); | |
cb7301c7 | 491 | } |
290603c1 | 492 | if ($email_list) { |
290603c1 | 493 | @to = (@to, @list_to); |
290603c1 | 494 | } |
3c7385b8 | 495 | output(merge_email(@to)); |
cb7301c7 JP |
496 | } |
497 | ||
498 | if ($scm) { | |
b781655a | 499 | @scm = uniq(@scm); |
cb7301c7 JP |
500 | output(@scm); |
501 | } | |
502 | ||
503 | if ($status) { | |
b781655a | 504 | @status = uniq(@status); |
cb7301c7 JP |
505 | output(@status); |
506 | } | |
507 | ||
508 | if ($subsystem) { | |
b781655a | 509 | @subsystem = uniq(@subsystem); |
cb7301c7 JP |
510 | output(@subsystem); |
511 | } | |
512 | ||
513 | if ($web) { | |
b781655a | 514 | @web = uniq(@web); |
cb7301c7 JP |
515 | output(@web); |
516 | } | |
517 | ||
518 | exit($exit); | |
519 | ||
520 | sub file_match_pattern { | |
521 | my ($file, $pattern) = @_; | |
522 | if (substr($pattern, -1) eq "/") { | |
523 | if ($file =~ m@^$pattern@) { | |
524 | return 1; | |
525 | } | |
526 | } else { | |
527 | if ($file =~ m@^$pattern@) { | |
528 | my $s1 = ($file =~ tr@/@@); | |
529 | my $s2 = ($pattern =~ tr@/@@); | |
530 | if ($s1 == $s2) { | |
531 | return 1; | |
532 | } | |
533 | } | |
534 | } | |
535 | return 0; | |
536 | } | |
537 | ||
538 | sub usage { | |
539 | print <<EOT; | |
540 | usage: $P [options] patchfile | |
870020f9 | 541 | $P [options] -f file|directory |
cb7301c7 JP |
542 | version: $V |
543 | ||
544 | MAINTAINER field selection options: | |
545 | --email => print email address(es) if any | |
546 | --git => include recent git \*-by: signers | |
e4d26b02 JP |
547 | --git-all-signature-types => include signers regardless of signature type |
548 | or use only ${signaturePattern} signers (default: $email_git_all_signature_types) | |
e3e9d114 | 549 | --git-fallback => use git when no exact MAINTAINERS pattern (default: $email_git_fallback) |
cb7301c7 | 550 | --git-chief-penguins => include ${penguin_chiefs} |
e4d26b02 JP |
551 | --git-min-signatures => number of signatures required (default: $email_git_min_signatures) |
552 | --git-max-maintainers => maximum maintainers to add (default: $email_git_max_maintainers) | |
553 | --git-min-percent => minimum percentage of commits required (default: $email_git_min_percent) | |
f5492666 | 554 | --git-blame => use git blame to find modified commits for patch or file |
e4d26b02 JP |
555 | --git-since => git history to use (default: $email_git_since) |
556 | --hg-since => hg history to use (default: $email_hg_since) | |
cb7301c7 JP |
557 | --m => include maintainer(s) if any |
558 | --n => include name 'Full Name <addr\@domain.tld>' | |
559 | --l => include list(s) if any | |
560 | --s => include subscriber only list(s) if any | |
11ecf53c | 561 | --remove-duplicates => minimize duplicate email names/addresses |
3c7385b8 JP |
562 | --roles => show roles (status:subsystem, git-signer, list, etc...) |
563 | --rolestats => show roles and statistics (commits/total_commits, %) | |
03372dbb | 564 | --file-emails => add email addresses found in -f file (default: 0 (off)) |
cb7301c7 JP |
565 | --scm => print SCM tree(s) if any |
566 | --status => print status if any | |
567 | --subsystem => print subsystem name if any | |
568 | --web => print website(s) if any | |
569 | ||
570 | Output type options: | |
571 | --separator [, ] => separator for multiple entries on 1 line | |
42498316 | 572 | using --separator also sets --nomultiline if --separator is not [, ] |
cb7301c7 JP |
573 | --multiline => print 1 entry per line |
574 | ||
cb7301c7 | 575 | Other options: |
3fb55652 | 576 | --pattern-depth => Number of pattern directory traversals (default: 0 (all)) |
dcf36a92 | 577 | --keywords => scan patch for keywords (default: 1 (on)) |
4b76c9da | 578 | --sections => print the entire subsystem sections with pattern matches |
f5f5078d | 579 | --version => show version |
cb7301c7 JP |
580 | --help => show this help information |
581 | ||
3fb55652 | 582 | Default options: |
11ecf53c | 583 | [--email --git --m --n --l --multiline --pattern-depth=0 --remove-duplicates] |
3fb55652 | 584 | |
870020f9 JP |
585 | Notes: |
586 | Using "-f directory" may give unexpected results: | |
f5492666 JP |
587 | Used with "--git", git signators for _all_ files in and below |
588 | directory are examined as git recurses directories. | |
589 | Any specified X: (exclude) pattern matches are _not_ ignored. | |
590 | Used with "--nogit", directory is used as a pattern match, | |
60db31ac JP |
591 | no individual file within the directory or subdirectory |
592 | is matched. | |
f5492666 JP |
593 | Used with "--git-blame", does not iterate all files in directory |
594 | Using "--git-blame" is slow and may add old committers and authors | |
595 | that are no longer active maintainers to the output. | |
3c7385b8 JP |
596 | Using "--roles" or "--rolestats" with git send-email --cc-cmd or any |
597 | other automated tools that expect only ["name"] <email address> | |
598 | may not work because of additional output after <email address>. | |
599 | Using "--rolestats" and "--git-blame" shows the #/total=% commits, | |
600 | not the percentage of the entire file authored. # of commits is | |
601 | not a good measure of amount of code authored. 1 major commit may | |
602 | contain a thousand lines, 5 trivial commits may modify a single line. | |
60db31ac JP |
603 | If git is not installed, but mercurial (hg) is installed and an .hg |
604 | repository exists, the following options apply to mercurial: | |
605 | --git, | |
606 | --git-min-signatures, --git-max-maintainers, --git-min-percent, and | |
607 | --git-blame | |
608 | Use --hg-since not --git-since to control date selection | |
368669da JP |
609 | File ".get_maintainer.conf", if it exists in the linux kernel source root |
610 | directory, can change whatever get_maintainer defaults are desired. | |
611 | Entries in this file can be any command line argument. | |
612 | This file is prepended to any additional command line arguments. | |
613 | Multiple lines and # comments are allowed. | |
cb7301c7 JP |
614 | EOT |
615 | } | |
616 | ||
617 | sub top_of_kernel_tree { | |
618 | my ($lk_path) = @_; | |
619 | ||
620 | if ($lk_path ne "" && substr($lk_path,length($lk_path)-1,1) ne "/") { | |
621 | $lk_path .= "/"; | |
622 | } | |
623 | if ( (-f "${lk_path}COPYING") | |
624 | && (-f "${lk_path}CREDITS") | |
625 | && (-f "${lk_path}Kbuild") | |
626 | && (-f "${lk_path}MAINTAINERS") | |
627 | && (-f "${lk_path}Makefile") | |
628 | && (-f "${lk_path}README") | |
629 | && (-d "${lk_path}Documentation") | |
630 | && (-d "${lk_path}arch") | |
631 | && (-d "${lk_path}include") | |
632 | && (-d "${lk_path}drivers") | |
633 | && (-d "${lk_path}fs") | |
634 | && (-d "${lk_path}init") | |
635 | && (-d "${lk_path}ipc") | |
636 | && (-d "${lk_path}kernel") | |
637 | && (-d "${lk_path}lib") | |
638 | && (-d "${lk_path}scripts")) { | |
639 | return 1; | |
640 | } | |
641 | return 0; | |
642 | } | |
643 | ||
0e70e83d JP |
644 | sub parse_email { |
645 | my ($formatted_email) = @_; | |
646 | ||
647 | my $name = ""; | |
648 | my $address = ""; | |
649 | ||
11ecf53c | 650 | if ($formatted_email =~ /^([^<]+)<(.+\@.*)>.*$/) { |
0e70e83d JP |
651 | $name = $1; |
652 | $address = $2; | |
11ecf53c | 653 | } elsif ($formatted_email =~ /^\s*<(.+\@\S*)>.*$/) { |
0e70e83d | 654 | $address = $1; |
b781655a | 655 | } elsif ($formatted_email =~ /^(.+\@\S*).*$/) { |
0e70e83d JP |
656 | $address = $1; |
657 | } | |
cb7301c7 JP |
658 | |
659 | $name =~ s/^\s+|\s+$//g; | |
d789504a | 660 | $name =~ s/^\"|\"$//g; |
0e70e83d | 661 | $address =~ s/^\s+|\s+$//g; |
cb7301c7 | 662 | |
a63ceb4c | 663 | if ($name =~ /[^\w \-]/i) { ##has "must quote" chars |
0e70e83d JP |
664 | $name =~ s/(?<!\\)"/\\"/g; ##escape quotes |
665 | $name = "\"$name\""; | |
666 | } | |
667 | ||
668 | return ($name, $address); | |
669 | } | |
670 | ||
671 | sub format_email { | |
a8af2430 | 672 | my ($name, $address, $usename) = @_; |
0e70e83d JP |
673 | |
674 | my $formatted_email; | |
675 | ||
676 | $name =~ s/^\s+|\s+$//g; | |
677 | $name =~ s/^\"|\"$//g; | |
678 | $address =~ s/^\s+|\s+$//g; | |
cb7301c7 | 679 | |
a63ceb4c | 680 | if ($name =~ /[^\w \-]/i) { ##has "must quote" chars |
cb7301c7 | 681 | $name =~ s/(?<!\\)"/\\"/g; ##escape quotes |
0e70e83d JP |
682 | $name = "\"$name\""; |
683 | } | |
684 | ||
a8af2430 | 685 | if ($usename) { |
0e70e83d JP |
686 | if ("$name" eq "") { |
687 | $formatted_email = "$address"; | |
688 | } else { | |
a8af2430 | 689 | $formatted_email = "$name <$address>"; |
0e70e83d | 690 | } |
cb7301c7 | 691 | } else { |
0e70e83d | 692 | $formatted_email = $address; |
cb7301c7 | 693 | } |
0e70e83d | 694 | |
cb7301c7 JP |
695 | return $formatted_email; |
696 | } | |
697 | ||
272a8979 JP |
698 | sub find_first_section { |
699 | my $index = 0; | |
700 | ||
701 | while ($index < @typevalue) { | |
702 | my $tv = $typevalue[$index]; | |
703 | if (($tv =~ m/^(\C):\s*(.*)/)) { | |
704 | last; | |
705 | } | |
706 | $index++; | |
707 | } | |
708 | ||
709 | return $index; | |
710 | } | |
711 | ||
b781655a | 712 | sub find_starting_index { |
b781655a JP |
713 | my ($index) = @_; |
714 | ||
715 | while ($index > 0) { | |
716 | my $tv = $typevalue[$index]; | |
717 | if (!($tv =~ m/^(\C):\s*(.*)/)) { | |
718 | last; | |
719 | } | |
720 | $index--; | |
721 | } | |
722 | ||
723 | return $index; | |
724 | } | |
725 | ||
726 | sub find_ending_index { | |
cb7301c7 JP |
727 | my ($index) = @_; |
728 | ||
b781655a | 729 | while ($index < @typevalue) { |
cb7301c7 | 730 | my $tv = $typevalue[$index]; |
b781655a JP |
731 | if (!($tv =~ m/^(\C):\s*(.*)/)) { |
732 | last; | |
733 | } | |
734 | $index++; | |
735 | } | |
736 | ||
737 | return $index; | |
738 | } | |
739 | ||
3c7385b8 JP |
740 | sub get_maintainer_role { |
741 | my ($index) = @_; | |
742 | ||
743 | my $i; | |
744 | my $start = find_starting_index($index); | |
745 | my $end = find_ending_index($index); | |
746 | ||
747 | my $role; | |
748 | my $subsystem = $typevalue[$start]; | |
749 | if (length($subsystem) > 20) { | |
750 | $subsystem = substr($subsystem, 0, 17); | |
751 | $subsystem =~ s/\s*$//; | |
752 | $subsystem = $subsystem . "..."; | |
753 | } | |
754 | ||
755 | for ($i = $start + 1; $i < $end; $i++) { | |
756 | my $tv = $typevalue[$i]; | |
757 | if ($tv =~ m/^(\C):\s*(.*)/) { | |
758 | my $ptype = $1; | |
759 | my $pvalue = $2; | |
760 | if ($ptype eq "S") { | |
761 | $role = $pvalue; | |
762 | } | |
763 | } | |
764 | } | |
765 | ||
766 | $role = lc($role); | |
767 | if ($role eq "supported") { | |
768 | $role = "supporter"; | |
769 | } elsif ($role eq "maintained") { | |
770 | $role = "maintainer"; | |
771 | } elsif ($role eq "odd fixes") { | |
772 | $role = "odd fixer"; | |
773 | } elsif ($role eq "orphan") { | |
774 | $role = "orphan minder"; | |
775 | } elsif ($role eq "obsolete") { | |
776 | $role = "obsolete minder"; | |
777 | } elsif ($role eq "buried alive in reporters") { | |
778 | $role = "chief penguin"; | |
779 | } | |
780 | ||
781 | return $role . ":" . $subsystem; | |
782 | } | |
783 | ||
784 | sub get_list_role { | |
785 | my ($index) = @_; | |
786 | ||
787 | my $i; | |
788 | my $start = find_starting_index($index); | |
789 | my $end = find_ending_index($index); | |
790 | ||
791 | my $subsystem = $typevalue[$start]; | |
792 | if (length($subsystem) > 20) { | |
793 | $subsystem = substr($subsystem, 0, 17); | |
794 | $subsystem =~ s/\s*$//; | |
795 | $subsystem = $subsystem . "..."; | |
796 | } | |
797 | ||
798 | if ($subsystem eq "THE REST") { | |
799 | $subsystem = ""; | |
800 | } | |
801 | ||
802 | return $subsystem; | |
803 | } | |
804 | ||
b781655a JP |
805 | sub add_categories { |
806 | my ($index) = @_; | |
807 | ||
808 | my $i; | |
809 | my $start = find_starting_index($index); | |
810 | my $end = find_ending_index($index); | |
811 | ||
812 | push(@subsystem, $typevalue[$start]); | |
813 | ||
814 | for ($i = $start + 1; $i < $end; $i++) { | |
815 | my $tv = $typevalue[$i]; | |
290603c1 | 816 | if ($tv =~ m/^(\C):\s*(.*)/) { |
cb7301c7 JP |
817 | my $ptype = $1; |
818 | my $pvalue = $2; | |
819 | if ($ptype eq "L") { | |
290603c1 JP |
820 | my $list_address = $pvalue; |
821 | my $list_additional = ""; | |
3c7385b8 JP |
822 | my $list_role = get_list_role($i); |
823 | ||
824 | if ($list_role ne "") { | |
825 | $list_role = ":" . $list_role; | |
826 | } | |
290603c1 JP |
827 | if ($list_address =~ m/([^\s]+)\s+(.*)$/) { |
828 | $list_address = $1; | |
829 | $list_additional = $2; | |
830 | } | |
bdf7c685 | 831 | if ($list_additional =~ m/subscribers-only/) { |
cb7301c7 | 832 | if ($email_subscriber_list) { |
3c7385b8 | 833 | push(@list_to, [$list_address, "subscriber list${list_role}"]); |
cb7301c7 JP |
834 | } |
835 | } else { | |
836 | if ($email_list) { | |
3c7385b8 | 837 | push(@list_to, [$list_address, "open list${list_role}"]); |
cb7301c7 JP |
838 | } |
839 | } | |
840 | } elsif ($ptype eq "M") { | |
0e70e83d JP |
841 | my ($name, $address) = parse_email($pvalue); |
842 | if ($name eq "") { | |
b781655a JP |
843 | if ($i > 0) { |
844 | my $tv = $typevalue[$i - 1]; | |
0e70e83d JP |
845 | if ($tv =~ m/^(\C):\s*(.*)/) { |
846 | if ($1 eq "P") { | |
847 | $name = $2; | |
a8af2430 | 848 | $pvalue = format_email($name, $address, $email_usename); |
5f2441e9 JP |
849 | } |
850 | } | |
851 | } | |
852 | } | |
0e70e83d | 853 | if ($email_maintainer) { |
3c7385b8 JP |
854 | my $role = get_maintainer_role($i); |
855 | push_email_addresses($pvalue, $role); | |
cb7301c7 JP |
856 | } |
857 | } elsif ($ptype eq "T") { | |
858 | push(@scm, $pvalue); | |
859 | } elsif ($ptype eq "W") { | |
860 | push(@web, $pvalue); | |
861 | } elsif ($ptype eq "S") { | |
862 | push(@status, $pvalue); | |
863 | } | |
cb7301c7 JP |
864 | } |
865 | } | |
866 | } | |
867 | ||
11ecf53c JP |
868 | my %email_hash_name; |
869 | my %email_hash_address; | |
0e70e83d | 870 | |
11ecf53c JP |
871 | sub email_inuse { |
872 | my ($name, $address) = @_; | |
873 | ||
874 | return 1 if (($name eq "") && ($address eq "")); | |
875 | return 1 if (($name ne "") && exists($email_hash_name{$name})); | |
876 | return 1 if (($address ne "") && exists($email_hash_address{$address})); | |
0e70e83d | 877 | |
0e70e83d JP |
878 | return 0; |
879 | } | |
880 | ||
1b5e1cf6 | 881 | sub push_email_address { |
3c7385b8 | 882 | my ($line, $role) = @_; |
1b5e1cf6 | 883 | |
0e70e83d | 884 | my ($name, $address) = parse_email($line); |
1b5e1cf6 | 885 | |
b781655a JP |
886 | if ($address eq "") { |
887 | return 0; | |
888 | } | |
889 | ||
11ecf53c | 890 | if (!$email_remove_duplicates) { |
a8af2430 | 891 | push(@email_to, [format_email($name, $address, $email_usename), $role]); |
11ecf53c | 892 | } elsif (!email_inuse($name, $address)) { |
a8af2430 | 893 | push(@email_to, [format_email($name, $address, $email_usename), $role]); |
11ecf53c JP |
894 | $email_hash_name{$name}++; |
895 | $email_hash_address{$address}++; | |
1b5e1cf6 | 896 | } |
b781655a JP |
897 | |
898 | return 1; | |
1b5e1cf6 JP |
899 | } |
900 | ||
901 | sub push_email_addresses { | |
3c7385b8 | 902 | my ($address, $role) = @_; |
1b5e1cf6 JP |
903 | |
904 | my @address_list = (); | |
905 | ||
5f2441e9 | 906 | if (rfc822_valid($address)) { |
3c7385b8 | 907 | push_email_address($address, $role); |
5f2441e9 | 908 | } elsif (@address_list = rfc822_validlist($address)) { |
1b5e1cf6 JP |
909 | my $array_count = shift(@address_list); |
910 | while (my $entry = shift(@address_list)) { | |
3c7385b8 | 911 | push_email_address($entry, $role); |
1b5e1cf6 | 912 | } |
5f2441e9 | 913 | } else { |
3c7385b8 | 914 | if (!push_email_address($address, $role)) { |
b781655a JP |
915 | warn("Invalid MAINTAINERS address: '" . $address . "'\n"); |
916 | } | |
1b5e1cf6 | 917 | } |
1b5e1cf6 JP |
918 | } |
919 | ||
3c7385b8 JP |
920 | sub add_role { |
921 | my ($line, $role) = @_; | |
922 | ||
923 | my ($name, $address) = parse_email($line); | |
a8af2430 | 924 | my $email = format_email($name, $address, $email_usename); |
3c7385b8 JP |
925 | |
926 | foreach my $entry (@email_to) { | |
927 | if ($email_remove_duplicates) { | |
928 | my ($entry_name, $entry_address) = parse_email($entry->[0]); | |
03372dbb JP |
929 | if (($name eq $entry_name || $address eq $entry_address) |
930 | && ($role eq "" || !($entry->[1] =~ m/$role/)) | |
931 | ) { | |
3c7385b8 JP |
932 | if ($entry->[1] eq "") { |
933 | $entry->[1] = "$role"; | |
934 | } else { | |
935 | $entry->[1] = "$entry->[1],$role"; | |
936 | } | |
937 | } | |
938 | } else { | |
03372dbb JP |
939 | if ($email eq $entry->[0] |
940 | && ($role eq "" || !($entry->[1] =~ m/$role/)) | |
941 | ) { | |
3c7385b8 JP |
942 | if ($entry->[1] eq "") { |
943 | $entry->[1] = "$role"; | |
944 | } else { | |
945 | $entry->[1] = "$entry->[1],$role"; | |
946 | } | |
947 | } | |
948 | } | |
949 | } | |
950 | } | |
951 | ||
cb7301c7 JP |
952 | sub which { |
953 | my ($bin) = @_; | |
954 | ||
f5f5078d | 955 | foreach my $path (split(/:/, $ENV{PATH})) { |
cb7301c7 JP |
956 | if (-e "$path/$bin") { |
957 | return "$path/$bin"; | |
958 | } | |
959 | } | |
960 | ||
961 | return ""; | |
962 | } | |
963 | ||
8cbb3a77 | 964 | sub mailmap { |
a8af2430 | 965 | my (@lines) = @_; |
8cbb3a77 JP |
966 | my %hash; |
967 | ||
968 | foreach my $line (@lines) { | |
969 | my ($name, $address) = parse_email($line); | |
970 | if (!exists($hash{$name})) { | |
971 | $hash{$name} = $address; | |
11ecf53c JP |
972 | } elsif ($address ne $hash{$name}) { |
973 | $address = $hash{$name}; | |
a8af2430 | 974 | $line = format_email($name, $address, $email_usename); |
8cbb3a77 JP |
975 | } |
976 | if (exists($mailmap{$name})) { | |
977 | my $obj = $mailmap{$name}; | |
978 | foreach my $map_address (@$obj) { | |
979 | if (($map_address eq $address) && | |
980 | ($map_address ne $hash{$name})) { | |
a8af2430 | 981 | $line = format_email($name, $hash{$name}, $email_usename); |
8cbb3a77 JP |
982 | } |
983 | } | |
984 | } | |
985 | } | |
986 | ||
987 | return @lines; | |
988 | } | |
989 | ||
60db31ac JP |
990 | sub git_execute_cmd { |
991 | my ($cmd) = @_; | |
992 | my @lines = (); | |
cb7301c7 | 993 | |
60db31ac JP |
994 | my $output = `$cmd`; |
995 | $output =~ s/^\s*//gm; | |
996 | @lines = split("\n", $output); | |
997 | ||
998 | return @lines; | |
a8af2430 JP |
999 | } |
1000 | ||
60db31ac | 1001 | sub hg_execute_cmd { |
a8af2430 | 1002 | my ($cmd) = @_; |
60db31ac JP |
1003 | my @lines = (); |
1004 | ||
1005 | my $output = `$cmd`; | |
1006 | @lines = split("\n", $output); | |
a8af2430 | 1007 | |
60db31ac JP |
1008 | return @lines; |
1009 | } | |
1010 | ||
1011 | sub vcs_find_signers { | |
1012 | my ($cmd) = @_; | |
a8af2430 JP |
1013 | my @lines = (); |
1014 | my $commits; | |
1015 | ||
60db31ac | 1016 | @lines = &{$VCS_cmds{"execute_cmd"}}($cmd); |
cb7301c7 | 1017 | |
60db31ac | 1018 | my $pattern = $VCS_cmds{"commit_pattern"}; |
cb7301c7 | 1019 | |
60db31ac | 1020 | $commits = grep(/$pattern/, @lines); # of commits |
afa81ee1 | 1021 | |
e4d26b02 | 1022 | @lines = grep(/^[ \t]*${signaturePattern}.*\@.*$/, @lines); |
0e70e83d JP |
1023 | if (!$email_git_penguin_chiefs) { |
1024 | @lines = grep(!/${penguin_chiefs}/i, @lines); | |
1025 | } | |
63ab52db JP |
1026 | |
1027 | return (0, @lines) if !@lines; | |
1028 | ||
0e70e83d JP |
1029 | # cut -f2- -d":" |
1030 | s/.*:\s*(.+)\s*/$1/ for (@lines); | |
1031 | ||
a8af2430 JP |
1032 | ## Reformat email addresses (with names) to avoid badly written signatures |
1033 | ||
3c7385b8 JP |
1034 | foreach my $line (@lines) { |
1035 | my ($name, $address) = parse_email($line); | |
a8af2430 JP |
1036 | $line = format_email($name, $address, 1); |
1037 | } | |
1038 | ||
1039 | return ($commits, @lines); | |
1040 | } | |
1041 | ||
63ab52db JP |
1042 | sub vcs_find_author { |
1043 | my ($cmd) = @_; | |
1044 | my @lines = (); | |
1045 | ||
1046 | @lines = &{$VCS_cmds{"execute_cmd"}}($cmd); | |
1047 | ||
1048 | if (!$email_git_penguin_chiefs) { | |
1049 | @lines = grep(!/${penguin_chiefs}/i, @lines); | |
1050 | } | |
1051 | ||
1052 | return @lines if !@lines; | |
1053 | ||
1054 | ## Reformat email addresses (with names) to avoid badly written signatures | |
1055 | ||
1056 | foreach my $line (@lines) { | |
1057 | my ($name, $address) = parse_email($line); | |
1058 | $line = format_email($name, $address, 1); | |
1059 | } | |
1060 | ||
1061 | return @lines; | |
1062 | } | |
1063 | ||
60db31ac JP |
1064 | sub vcs_save_commits { |
1065 | my ($cmd) = @_; | |
1066 | my @lines = (); | |
1067 | my @commits = (); | |
1068 | ||
1069 | @lines = &{$VCS_cmds{"execute_cmd"}}($cmd); | |
1070 | ||
1071 | foreach my $line (@lines) { | |
1072 | if ($line =~ m/$VCS_cmds{"blame_commit_pattern"}/) { | |
1073 | push(@commits, $1); | |
1074 | } | |
1075 | } | |
1076 | ||
1077 | return @commits; | |
1078 | } | |
1079 | ||
1080 | sub vcs_blame { | |
1081 | my ($file) = @_; | |
1082 | my $cmd; | |
1083 | my @commits = (); | |
1084 | ||
1085 | return @commits if (!(-f $file)); | |
1086 | ||
1087 | if (@range && $VCS_cmds{"blame_range_cmd"} eq "") { | |
1088 | my @all_commits = (); | |
1089 | ||
1090 | $cmd = $VCS_cmds{"blame_file_cmd"}; | |
1091 | $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd | |
1092 | @all_commits = vcs_save_commits($cmd); | |
1093 | ||
1094 | foreach my $file_range_diff (@range) { | |
1095 | next if (!($file_range_diff =~ m/(.+):(.+):(.+)/)); | |
1096 | my $diff_file = $1; | |
1097 | my $diff_start = $2; | |
1098 | my $diff_length = $3; | |
1099 | next if ("$file" ne "$diff_file"); | |
1100 | for (my $i = $diff_start; $i < $diff_start + $diff_length; $i++) { | |
1101 | push(@commits, $all_commits[$i]); | |
1102 | } | |
1103 | } | |
1104 | } elsif (@range) { | |
1105 | foreach my $file_range_diff (@range) { | |
1106 | next if (!($file_range_diff =~ m/(.+):(.+):(.+)/)); | |
1107 | my $diff_file = $1; | |
1108 | my $diff_start = $2; | |
1109 | my $diff_length = $3; | |
1110 | next if ("$file" ne "$diff_file"); | |
1111 | $cmd = $VCS_cmds{"blame_range_cmd"}; | |
1112 | $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd | |
1113 | push(@commits, vcs_save_commits($cmd)); | |
1114 | } | |
1115 | } else { | |
1116 | $cmd = $VCS_cmds{"blame_file_cmd"}; | |
1117 | $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd | |
1118 | @commits = vcs_save_commits($cmd); | |
1119 | } | |
1120 | ||
63ab52db JP |
1121 | foreach my $commit (@commits) { |
1122 | $commit =~ s/^\^//g; | |
1123 | } | |
1124 | ||
60db31ac JP |
1125 | return @commits; |
1126 | } | |
1127 | ||
1128 | my $printed_novcs = 0; | |
1129 | sub vcs_exists { | |
1130 | %VCS_cmds = %VCS_cmds_git; | |
1131 | return 1 if eval $VCS_cmds{"available"}; | |
1132 | %VCS_cmds = %VCS_cmds_hg; | |
1133 | return 1 if eval $VCS_cmds{"available"}; | |
1134 | %VCS_cmds = (); | |
1135 | if (!$printed_novcs) { | |
1136 | warn("$P: No supported VCS found. Add --nogit to options?\n"); | |
1137 | warn("Using a git repository produces better results.\n"); | |
1138 | warn("Try Linus Torvalds' latest git repository using:\n"); | |
1139 | warn("git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git\n"); | |
1140 | $printed_novcs = 1; | |
1141 | } | |
1142 | return 0; | |
1143 | } | |
1144 | ||
1145 | sub vcs_assign { | |
a8af2430 JP |
1146 | my ($role, $divisor, @lines) = @_; |
1147 | ||
1148 | my %hash; | |
1149 | my $count = 0; | |
1150 | ||
a8af2430 JP |
1151 | return if (@lines <= 0); |
1152 | ||
1153 | if ($divisor <= 0) { | |
60db31ac | 1154 | warn("Bad divisor in " . (caller(0))[3] . ": $divisor\n"); |
a8af2430 | 1155 | $divisor = 1; |
3c7385b8 | 1156 | } |
8cbb3a77 | 1157 | |
11ecf53c JP |
1158 | if ($email_remove_duplicates) { |
1159 | @lines = mailmap(@lines); | |
1160 | } | |
0e70e83d | 1161 | |
63ab52db JP |
1162 | return if (@lines <= 0); |
1163 | ||
0e70e83d | 1164 | @lines = sort(@lines); |
11ecf53c | 1165 | |
0e70e83d | 1166 | # uniq -c |
11ecf53c JP |
1167 | $hash{$_}++ for @lines; |
1168 | ||
0e70e83d | 1169 | # sort -rn |
0e70e83d | 1170 | foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) { |
11ecf53c | 1171 | my $sign_offs = $hash{$line}; |
a8af2430 | 1172 | my $percent = $sign_offs * 100 / $divisor; |
3c7385b8 | 1173 | |
a8af2430 | 1174 | $percent = 100 if ($percent > 100); |
11ecf53c JP |
1175 | $count++; |
1176 | last if ($sign_offs < $email_git_min_signatures || | |
1177 | $count > $email_git_max_maintainers || | |
a8af2430 | 1178 | $percent < $email_git_min_percent); |
3c7385b8 | 1179 | push_email_address($line, ''); |
3c7385b8 | 1180 | if ($output_rolestats) { |
a8af2430 JP |
1181 | my $fmt_percent = sprintf("%.0f", $percent); |
1182 | add_role($line, "$role:$sign_offs/$divisor=$fmt_percent%"); | |
1183 | } else { | |
1184 | add_role($line, $role); | |
3c7385b8 | 1185 | } |
f5492666 JP |
1186 | } |
1187 | } | |
1188 | ||
60db31ac | 1189 | sub vcs_file_signoffs { |
a8af2430 JP |
1190 | my ($file) = @_; |
1191 | ||
1192 | my @signers = (); | |
60db31ac | 1193 | my $commits; |
f5492666 | 1194 | |
60db31ac | 1195 | return if (!vcs_exists()); |
a8af2430 | 1196 | |
60db31ac JP |
1197 | my $cmd = $VCS_cmds{"find_signers_cmd"}; |
1198 | $cmd =~ s/(\$\w+)/$1/eeg; # interpolate $cmd | |
f5492666 | 1199 | |
60db31ac JP |
1200 | ($commits, @signers) = vcs_find_signers($cmd); |
1201 | vcs_assign("commit_signer", $commits, @signers); | |
f5492666 JP |
1202 | } |
1203 | ||
60db31ac | 1204 | sub vcs_file_blame { |
f5492666 JP |
1205 | my ($file) = @_; |
1206 | ||
a8af2430 | 1207 | my @signers = (); |
63ab52db | 1208 | my @all_commits = (); |
60db31ac | 1209 | my @commits = (); |
a8af2430 | 1210 | my $total_commits; |
63ab52db | 1211 | my $total_lines; |
f5492666 | 1212 | |
60db31ac | 1213 | return if (!vcs_exists()); |
f5492666 | 1214 | |
63ab52db JP |
1215 | @all_commits = vcs_blame($file); |
1216 | @commits = uniq(@all_commits); | |
a8af2430 | 1217 | $total_commits = @commits; |
63ab52db | 1218 | $total_lines = @all_commits; |
8cbb3a77 | 1219 | |
a8af2430 JP |
1220 | foreach my $commit (@commits) { |
1221 | my $commit_count; | |
1222 | my @commit_signers = (); | |
8cbb3a77 | 1223 | |
60db31ac JP |
1224 | my $cmd = $VCS_cmds{"find_commit_signers_cmd"}; |
1225 | $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd | |
1226 | ||
1227 | ($commit_count, @commit_signers) = vcs_find_signers($cmd); | |
63ab52db | 1228 | |
60db31ac | 1229 | push(@signers, @commit_signers); |
f5492666 JP |
1230 | } |
1231 | ||
a8af2430 | 1232 | if ($from_filename) { |
63ab52db JP |
1233 | if ($output_rolestats) { |
1234 | my @blame_signers; | |
1235 | foreach my $commit (@commits) { | |
1236 | my $i; | |
1237 | my $cmd = $VCS_cmds{"find_commit_author_cmd"}; | |
1238 | $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd | |
1239 | my @author = vcs_find_author($cmd); | |
1240 | next if !@author; | |
1241 | my $count = grep(/$commit/, @all_commits); | |
1242 | for ($i = 0; $i < $count ; $i++) { | |
1243 | push(@blame_signers, $author[0]); | |
1244 | } | |
1245 | } | |
1246 | if (@blame_signers) { | |
1247 | vcs_assign("authored lines", $total_lines, @blame_signers); | |
1248 | } | |
1249 | } | |
60db31ac | 1250 | vcs_assign("commits", $total_commits, @signers); |
a8af2430 | 1251 | } else { |
60db31ac | 1252 | vcs_assign("modified commits", $total_commits, @signers); |
cb7301c7 | 1253 | } |
cb7301c7 JP |
1254 | } |
1255 | ||
1256 | sub uniq { | |
a8af2430 | 1257 | my (@parms) = @_; |
cb7301c7 JP |
1258 | |
1259 | my %saw; | |
1260 | @parms = grep(!$saw{$_}++, @parms); | |
1261 | return @parms; | |
1262 | } | |
1263 | ||
1264 | sub sort_and_uniq { | |
a8af2430 | 1265 | my (@parms) = @_; |
cb7301c7 JP |
1266 | |
1267 | my %saw; | |
1268 | @parms = sort @parms; | |
1269 | @parms = grep(!$saw{$_}++, @parms); | |
1270 | return @parms; | |
1271 | } | |
1272 | ||
03372dbb JP |
1273 | sub clean_file_emails { |
1274 | my (@file_emails) = @_; | |
1275 | my @fmt_emails = (); | |
1276 | ||
1277 | foreach my $email (@file_emails) { | |
1278 | $email =~ s/[\(\<\{]{0,1}([A-Za-z0-9_\.\+-]+\@[A-Za-z0-9\.-]+)[\)\>\}]{0,1}/\<$1\>/g; | |
1279 | my ($name, $address) = parse_email($email); | |
1280 | if ($name eq '"[,\.]"') { | |
1281 | $name = ""; | |
1282 | } | |
1283 | ||
1284 | my @nw = split(/[^A-Za-zÀ-ÿ\'\,\.\+-]/, $name); | |
1285 | if (@nw > 2) { | |
1286 | my $first = $nw[@nw - 3]; | |
1287 | my $middle = $nw[@nw - 2]; | |
1288 | my $last = $nw[@nw - 1]; | |
1289 | ||
1290 | if (((length($first) == 1 && $first =~ m/[A-Za-z]/) || | |
1291 | (length($first) == 2 && substr($first, -1) eq ".")) || | |
1292 | (length($middle) == 1 || | |
1293 | (length($middle) == 2 && substr($middle, -1) eq "."))) { | |
1294 | $name = "$first $middle $last"; | |
1295 | } else { | |
1296 | $name = "$middle $last"; | |
1297 | } | |
1298 | } | |
1299 | ||
1300 | if (substr($name, -1) =~ /[,\.]/) { | |
1301 | $name = substr($name, 0, length($name) - 1); | |
1302 | } elsif (substr($name, -2) =~ /[,\.]"/) { | |
1303 | $name = substr($name, 0, length($name) - 2) . '"'; | |
1304 | } | |
1305 | ||
1306 | if (substr($name, 0, 1) =~ /[,\.]/) { | |
1307 | $name = substr($name, 1, length($name) - 1); | |
1308 | } elsif (substr($name, 0, 2) =~ /"[,\.]/) { | |
1309 | $name = '"' . substr($name, 2, length($name) - 2); | |
1310 | } | |
1311 | ||
1312 | my $fmt_email = format_email($name, $address, $email_usename); | |
1313 | push(@fmt_emails, $fmt_email); | |
1314 | } | |
1315 | return @fmt_emails; | |
1316 | } | |
1317 | ||
3c7385b8 JP |
1318 | sub merge_email { |
1319 | my @lines; | |
1320 | my %saw; | |
1321 | ||
1322 | for (@_) { | |
1323 | my ($address, $role) = @$_; | |
1324 | if (!$saw{$address}) { | |
1325 | if ($output_roles) { | |
60db31ac | 1326 | push(@lines, "$address ($role)"); |
3c7385b8 | 1327 | } else { |
60db31ac | 1328 | push(@lines, $address); |
3c7385b8 JP |
1329 | } |
1330 | $saw{$address} = 1; | |
1331 | } | |
1332 | } | |
1333 | ||
1334 | return @lines; | |
1335 | } | |
1336 | ||
cb7301c7 | 1337 | sub output { |
a8af2430 | 1338 | my (@parms) = @_; |
cb7301c7 JP |
1339 | |
1340 | if ($output_multiline) { | |
1341 | foreach my $line (@parms) { | |
1342 | print("${line}\n"); | |
1343 | } | |
1344 | } else { | |
1345 | print(join($output_separator, @parms)); | |
1346 | print("\n"); | |
1347 | } | |
1348 | } | |
1b5e1cf6 JP |
1349 | |
1350 | my $rfc822re; | |
1351 | ||
1352 | sub make_rfc822re { | |
1353 | # Basic lexical tokens are specials, domain_literal, quoted_string, atom, and | |
1354 | # comment. We must allow for rfc822_lwsp (or comments) after each of these. | |
1355 | # This regexp will only work on addresses which have had comments stripped | |
1356 | # and replaced with rfc822_lwsp. | |
1357 | ||
1358 | my $specials = '()<>@,;:\\\\".\\[\\]'; | |
1359 | my $controls = '\\000-\\037\\177'; | |
1360 | ||
1361 | my $dtext = "[^\\[\\]\\r\\\\]"; | |
1362 | my $domain_literal = "\\[(?:$dtext|\\\\.)*\\]$rfc822_lwsp*"; | |
1363 | ||
1364 | my $quoted_string = "\"(?:[^\\\"\\r\\\\]|\\\\.|$rfc822_lwsp)*\"$rfc822_lwsp*"; | |
1365 | ||
1366 | # Use zero-width assertion to spot the limit of an atom. A simple | |
1367 | # $rfc822_lwsp* causes the regexp engine to hang occasionally. | |
1368 | my $atom = "[^$specials $controls]+(?:$rfc822_lwsp+|\\Z|(?=[\\[\"$specials]))"; | |
1369 | my $word = "(?:$atom|$quoted_string)"; | |
1370 | my $localpart = "$word(?:\\.$rfc822_lwsp*$word)*"; | |
1371 | ||
1372 | my $sub_domain = "(?:$atom|$domain_literal)"; | |
1373 | my $domain = "$sub_domain(?:\\.$rfc822_lwsp*$sub_domain)*"; | |
1374 | ||
1375 | my $addr_spec = "$localpart\@$rfc822_lwsp*$domain"; | |
1376 | ||
1377 | my $phrase = "$word*"; | |
1378 | my $route = "(?:\@$domain(?:,\@$rfc822_lwsp*$domain)*:$rfc822_lwsp*)"; | |
1379 | my $route_addr = "\\<$rfc822_lwsp*$route?$addr_spec\\>$rfc822_lwsp*"; | |
1380 | my $mailbox = "(?:$addr_spec|$phrase$route_addr)"; | |
1381 | ||
1382 | my $group = "$phrase:$rfc822_lwsp*(?:$mailbox(?:,\\s*$mailbox)*)?;\\s*"; | |
1383 | my $address = "(?:$mailbox|$group)"; | |
1384 | ||
1385 | return "$rfc822_lwsp*$address"; | |
1386 | } | |
1387 | ||
1388 | sub rfc822_strip_comments { | |
1389 | my $s = shift; | |
1390 | # Recursively remove comments, and replace with a single space. The simpler | |
1391 | # regexps in the Email Addressing FAQ are imperfect - they will miss escaped | |
1392 | # chars in atoms, for example. | |
1393 | ||
1394 | while ($s =~ s/^((?:[^"\\]|\\.)* | |
1395 | (?:"(?:[^"\\]|\\.)*"(?:[^"\\]|\\.)*)*) | |
1396 | \((?:[^()\\]|\\.)*\)/$1 /osx) {} | |
1397 | return $s; | |
1398 | } | |
1399 | ||
1400 | # valid: returns true if the parameter is an RFC822 valid address | |
1401 | # | |
22dd5b0c | 1402 | sub rfc822_valid { |
1b5e1cf6 JP |
1403 | my $s = rfc822_strip_comments(shift); |
1404 | ||
1405 | if (!$rfc822re) { | |
1406 | $rfc822re = make_rfc822re(); | |
1407 | } | |
1408 | ||
1409 | return $s =~ m/^$rfc822re$/so && $s =~ m/^$rfc822_char*$/; | |
1410 | } | |
1411 | ||
1412 | # validlist: In scalar context, returns true if the parameter is an RFC822 | |
1413 | # valid list of addresses. | |
1414 | # | |
1415 | # In list context, returns an empty list on failure (an invalid | |
1416 | # address was found); otherwise a list whose first element is the | |
1417 | # number of addresses found and whose remaining elements are the | |
1418 | # addresses. This is needed to disambiguate failure (invalid) | |
1419 | # from success with no addresses found, because an empty string is | |
1420 | # a valid list. | |
1421 | ||
22dd5b0c | 1422 | sub rfc822_validlist { |
1b5e1cf6 JP |
1423 | my $s = rfc822_strip_comments(shift); |
1424 | ||
1425 | if (!$rfc822re) { | |
1426 | $rfc822re = make_rfc822re(); | |
1427 | } | |
1428 | # * null list items are valid according to the RFC | |
1429 | # * the '1' business is to aid in distinguishing failure from no results | |
1430 | ||
1431 | my @r; | |
1432 | if ($s =~ m/^(?:$rfc822re)?(?:,(?:$rfc822re)?)*$/so && | |
1433 | $s =~ m/^$rfc822_char*$/) { | |
5f2441e9 | 1434 | while ($s =~ m/(?:^|,$rfc822_lwsp*)($rfc822re)/gos) { |
60db31ac | 1435 | push(@r, $1); |
1b5e1cf6 JP |
1436 | } |
1437 | return wantarray ? (scalar(@r), @r) : 1; | |
1438 | } | |
60db31ac | 1439 | return wantarray ? () : 0; |
1b5e1cf6 | 1440 | } |