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 | ||
bcde44ed JP |
113 | my $conf = which_conf(".get_maintainer.conf"); |
114 | if (-f $conf) { | |
368669da | 115 | my @conf_args; |
bcde44ed JP |
116 | open(my $conffile, '<', "$conf") |
117 | or warn "$P: Can't find a readable .get_maintainer.conf file $!\n"; | |
118 | ||
368669da JP |
119 | while (<$conffile>) { |
120 | my $line = $_; | |
121 | ||
122 | $line =~ s/\s*\n?$//g; | |
123 | $line =~ s/^\s*//g; | |
124 | $line =~ s/\s+/ /g; | |
125 | ||
126 | next if ($line =~ m/^\s*#/); | |
127 | next if ($line =~ m/^\s*$/); | |
128 | ||
129 | my @words = split(" ", $line); | |
130 | foreach my $word (@words) { | |
131 | last if ($word =~ m/^#/); | |
132 | push (@conf_args, $word); | |
133 | } | |
134 | } | |
135 | close($conffile); | |
136 | unshift(@ARGV, @conf_args) if @conf_args; | |
137 | } | |
138 | ||
cb7301c7 JP |
139 | if (!GetOptions( |
140 | 'email!' => \$email, | |
141 | 'git!' => \$email_git, | |
e4d26b02 | 142 | 'git-all-signature-types!' => \$email_git_all_signature_types, |
60db31ac | 143 | 'git-blame!' => \$email_git_blame, |
e3e9d114 | 144 | 'git-fallback!' => \$email_git_fallback, |
cb7301c7 JP |
145 | 'git-chief-penguins!' => \$email_git_penguin_chiefs, |
146 | 'git-min-signatures=i' => \$email_git_min_signatures, | |
147 | 'git-max-maintainers=i' => \$email_git_max_maintainers, | |
afa81ee1 | 148 | 'git-min-percent=i' => \$email_git_min_percent, |
cb7301c7 | 149 | 'git-since=s' => \$email_git_since, |
60db31ac | 150 | 'hg-since=s' => \$email_hg_since, |
11ecf53c | 151 | 'remove-duplicates!' => \$email_remove_duplicates, |
cb7301c7 JP |
152 | 'm!' => \$email_maintainer, |
153 | 'n!' => \$email_usename, | |
154 | 'l!' => \$email_list, | |
155 | 's!' => \$email_subscriber_list, | |
156 | 'multiline!' => \$output_multiline, | |
3c7385b8 JP |
157 | 'roles!' => \$output_roles, |
158 | 'rolestats!' => \$output_rolestats, | |
cb7301c7 JP |
159 | 'separator=s' => \$output_separator, |
160 | 'subsystem!' => \$subsystem, | |
161 | 'status!' => \$status, | |
162 | 'scm!' => \$scm, | |
163 | 'web!' => \$web, | |
3fb55652 | 164 | 'pattern-depth=i' => \$pattern_depth, |
dcf36a92 | 165 | 'k|keywords!' => \$keywords, |
4b76c9da | 166 | 'sections!' => \$sections, |
03372dbb | 167 | 'fe|file-emails!' => \$file_emails, |
4a7fdb5f | 168 | 'f|file' => \$from_filename, |
cb7301c7 | 169 | 'v|version' => \$version, |
64f77f31 | 170 | 'h|help|usage' => \$help, |
cb7301c7 | 171 | )) { |
3c7385b8 | 172 | die "$P: invalid argument - use --help if necessary\n"; |
cb7301c7 JP |
173 | } |
174 | ||
175 | if ($help != 0) { | |
176 | usage(); | |
177 | exit 0; | |
178 | } | |
179 | ||
180 | if ($version != 0) { | |
181 | print("${P} ${V}\n"); | |
182 | exit 0; | |
183 | } | |
184 | ||
64f77f31 JP |
185 | if (-t STDIN && !@ARGV) { |
186 | # We're talking to a terminal, but have no command line arguments. | |
187 | die "$P: missing patchfile or -f file - use --help if necessary\n"; | |
cb7301c7 JP |
188 | } |
189 | ||
42498316 JP |
190 | if ($output_separator ne ", ") { |
191 | $output_multiline = 0; | |
192 | } | |
193 | ||
3c7385b8 JP |
194 | if ($output_rolestats) { |
195 | $output_roles = 1; | |
196 | } | |
197 | ||
4b76c9da JP |
198 | if ($sections) { |
199 | $email = 0; | |
200 | $email_list = 0; | |
201 | $scm = 0; | |
202 | $status = 0; | |
203 | $subsystem = 0; | |
204 | $web = 0; | |
205 | $keywords = 0; | |
206 | } else { | |
207 | my $selections = $email + $scm + $status + $subsystem + $web; | |
208 | if ($selections == 0) { | |
4b76c9da JP |
209 | die "$P: Missing required option: email, scm, status, subsystem or web\n"; |
210 | } | |
cb7301c7 JP |
211 | } |
212 | ||
f5492666 JP |
213 | if ($email && |
214 | ($email_maintainer + $email_list + $email_subscriber_list + | |
215 | $email_git + $email_git_penguin_chiefs + $email_git_blame) == 0) { | |
cb7301c7 JP |
216 | die "$P: Please select at least 1 email option\n"; |
217 | } | |
218 | ||
219 | if (!top_of_kernel_tree($lk_path)) { | |
220 | die "$P: The current directory does not appear to be " | |
221 | . "a linux kernel source tree.\n"; | |
222 | } | |
223 | ||
e4d26b02 JP |
224 | if ($email_git_all_signature_types) { |
225 | $signaturePattern = "(.+?)[Bb][Yy]:"; | |
226 | } | |
227 | ||
cb7301c7 JP |
228 | ## Read MAINTAINERS for type/value pairs |
229 | ||
230 | my @typevalue = (); | |
dcf36a92 JP |
231 | my %keyword_hash; |
232 | ||
22dd5b0c SH |
233 | open (my $maint, '<', "${lk_path}MAINTAINERS") |
234 | or die "$P: Can't open MAINTAINERS: $!\n"; | |
235 | while (<$maint>) { | |
cb7301c7 JP |
236 | my $line = $_; |
237 | ||
238 | if ($line =~ m/^(\C):\s*(.*)/) { | |
239 | my $type = $1; | |
240 | my $value = $2; | |
241 | ||
242 | ##Filename pattern matching | |
243 | if ($type eq "F" || $type eq "X") { | |
244 | $value =~ s@\.@\\\.@g; ##Convert . to \. | |
245 | $value =~ s/\*/\.\*/g; ##Convert * to .* | |
246 | $value =~ s/\?/\./g; ##Convert ? to . | |
870020f9 JP |
247 | ##if pattern is a directory and it lacks a trailing slash, add one |
248 | if ((-d $value)) { | |
249 | $value =~ s@([^/])$@$1/@; | |
250 | } | |
dcf36a92 JP |
251 | } elsif ($type eq "K") { |
252 | $keyword_hash{@typevalue} = $value; | |
cb7301c7 JP |
253 | } |
254 | push(@typevalue, "$type:$value"); | |
255 | } elsif (!/^(\s)*$/) { | |
256 | $line =~ s/\n$//g; | |
257 | push(@typevalue, $line); | |
258 | } | |
259 | } | |
22dd5b0c | 260 | close($maint); |
cb7301c7 | 261 | |
8cbb3a77 JP |
262 | my %mailmap; |
263 | ||
11ecf53c | 264 | if ($email_remove_duplicates) { |
22dd5b0c SH |
265 | open(my $mailmap, '<', "${lk_path}.mailmap") |
266 | or warn "$P: Can't open .mailmap: $!\n"; | |
267 | while (<$mailmap>) { | |
11ecf53c | 268 | my $line = $_; |
8cbb3a77 | 269 | |
11ecf53c JP |
270 | next if ($line =~ m/^\s*#/); |
271 | next if ($line =~ m/^\s*$/); | |
8cbb3a77 | 272 | |
11ecf53c | 273 | my ($name, $address) = parse_email($line); |
a8af2430 | 274 | $line = format_email($name, $address, $email_usename); |
8cbb3a77 | 275 | |
11ecf53c | 276 | next if ($line =~ m/^\s*$/); |
8cbb3a77 | 277 | |
11ecf53c JP |
278 | if (exists($mailmap{$name})) { |
279 | my $obj = $mailmap{$name}; | |
280 | push(@$obj, $address); | |
281 | } else { | |
282 | my @arr = ($address); | |
283 | $mailmap{$name} = \@arr; | |
284 | } | |
8cbb3a77 | 285 | } |
22dd5b0c | 286 | close($mailmap); |
8cbb3a77 JP |
287 | } |
288 | ||
4a7fdb5f | 289 | ## use the filenames on the command line or find the filenames in the patchfiles |
cb7301c7 JP |
290 | |
291 | my @files = (); | |
f5492666 | 292 | my @range = (); |
dcf36a92 | 293 | my @keyword_tvi = (); |
03372dbb | 294 | my @file_emails = (); |
cb7301c7 | 295 | |
64f77f31 JP |
296 | if (!@ARGV) { |
297 | push(@ARGV, "&STDIN"); | |
298 | } | |
299 | ||
4a7fdb5f | 300 | foreach my $file (@ARGV) { |
64f77f31 JP |
301 | if ($file ne "&STDIN") { |
302 | ##if $file is a directory and it lacks a trailing slash, add one | |
303 | if ((-d $file)) { | |
304 | $file =~ s@([^/])$@$1/@; | |
305 | } elsif (!(-f $file)) { | |
306 | die "$P: file '${file}' not found\n"; | |
307 | } | |
cb7301c7 | 308 | } |
4a7fdb5f JP |
309 | if ($from_filename) { |
310 | push(@files, $file); | |
fab9ed12 | 311 | if ($file ne "MAINTAINERS" && -f $file && ($keywords || $file_emails)) { |
22dd5b0c SH |
312 | open(my $f, '<', $file) |
313 | or die "$P: Can't open $file: $!\n"; | |
314 | my $text = do { local($/) ; <$f> }; | |
315 | close($f); | |
03372dbb JP |
316 | if ($keywords) { |
317 | foreach my $line (keys %keyword_hash) { | |
318 | if ($text =~ m/$keyword_hash{$line}/x) { | |
319 | push(@keyword_tvi, $line); | |
320 | } | |
dcf36a92 JP |
321 | } |
322 | } | |
03372dbb JP |
323 | if ($file_emails) { |
324 | 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; | |
325 | push(@file_emails, clean_file_emails(@poss_addr)); | |
326 | } | |
dcf36a92 | 327 | } |
4a7fdb5f JP |
328 | } else { |
329 | my $file_cnt = @files; | |
f5492666 | 330 | my $lastfile; |
22dd5b0c | 331 | |
3a4df13d | 332 | open(my $patch, "< $file") |
22dd5b0c SH |
333 | or die "$P: Can't open $file: $!\n"; |
334 | while (<$patch>) { | |
dcf36a92 | 335 | my $patch_line = $_; |
4a7fdb5f JP |
336 | if (m/^\+\+\+\s+(\S+)/) { |
337 | my $filename = $1; | |
338 | $filename =~ s@^[^/]*/@@; | |
339 | $filename =~ s@\n@@; | |
f5492666 | 340 | $lastfile = $filename; |
4a7fdb5f | 341 | push(@files, $filename); |
f5492666 JP |
342 | } elsif (m/^\@\@ -(\d+),(\d+)/) { |
343 | if ($email_git_blame) { | |
344 | push(@range, "$lastfile:$1:$2"); | |
345 | } | |
dcf36a92 JP |
346 | } elsif ($keywords) { |
347 | foreach my $line (keys %keyword_hash) { | |
348 | if ($patch_line =~ m/^[+-].*$keyword_hash{$line}/x) { | |
349 | push(@keyword_tvi, $line); | |
350 | } | |
351 | } | |
4a7fdb5f | 352 | } |
cb7301c7 | 353 | } |
22dd5b0c SH |
354 | close($patch); |
355 | ||
4a7fdb5f | 356 | if ($file_cnt == @files) { |
7f29fd27 | 357 | warn "$P: file '${file}' doesn't appear to be a patch. " |
4a7fdb5f JP |
358 | . "Add -f to options?\n"; |
359 | } | |
360 | @files = sort_and_uniq(@files); | |
cb7301c7 | 361 | } |
cb7301c7 JP |
362 | } |
363 | ||
03372dbb JP |
364 | @file_emails = uniq(@file_emails); |
365 | ||
cb7301c7 | 366 | my @email_to = (); |
290603c1 | 367 | my @list_to = (); |
cb7301c7 JP |
368 | my @scm = (); |
369 | my @web = (); | |
370 | my @subsystem = (); | |
371 | my @status = (); | |
372 | ||
373 | # Find responsible parties | |
374 | ||
375 | foreach my $file (@files) { | |
376 | ||
272a8979 | 377 | my %hash; |
e3e9d114 | 378 | my $exact_pattern_match = 0; |
272a8979 JP |
379 | my $tvi = find_first_section(); |
380 | while ($tvi < @typevalue) { | |
381 | my $start = find_starting_index($tvi); | |
382 | my $end = find_ending_index($tvi); | |
383 | my $exclude = 0; | |
384 | my $i; | |
385 | ||
386 | #Do not match excluded file patterns | |
387 | ||
388 | for ($i = $start; $i < $end; $i++) { | |
389 | my $line = $typevalue[$i]; | |
290603c1 | 390 | if ($line =~ m/^(\C):\s*(.*)/) { |
cb7301c7 JP |
391 | my $type = $1; |
392 | my $value = $2; | |
272a8979 | 393 | if ($type eq 'X') { |
cb7301c7 | 394 | if (file_match_pattern($file, $value)) { |
272a8979 | 395 | $exclude = 1; |
3c840c18 | 396 | last; |
cb7301c7 JP |
397 | } |
398 | } | |
399 | } | |
cb7301c7 | 400 | } |
272a8979 JP |
401 | |
402 | if (!$exclude) { | |
403 | for ($i = $start; $i < $end; $i++) { | |
404 | my $line = $typevalue[$i]; | |
405 | if ($line =~ m/^(\C):\s*(.*)/) { | |
406 | my $type = $1; | |
407 | my $value = $2; | |
408 | if ($type eq 'F') { | |
409 | if (file_match_pattern($file, $value)) { | |
410 | my $value_pd = ($value =~ tr@/@@); | |
411 | my $file_pd = ($file =~ tr@/@@); | |
412 | $value_pd++ if (substr($value,-1,1) ne "/"); | |
e3e9d114 JP |
413 | $value_pd = -1 if ($value =~ /^\.\*/); |
414 | $exact_pattern_match = 1 if ($value_pd >= $file_pd); | |
272a8979 JP |
415 | if ($pattern_depth == 0 || |
416 | (($file_pd - $value_pd) < $pattern_depth)) { | |
417 | $hash{$tvi} = $value_pd; | |
418 | } | |
419 | } | |
420 | } | |
421 | } | |
422 | } | |
1d606b4e | 423 | } |
272a8979 | 424 | |
3c840c18 | 425 | $tvi = $end + 1; |
272a8979 JP |
426 | } |
427 | ||
428 | foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) { | |
429 | add_categories($line); | |
6ffd9485 JP |
430 | if ($sections) { |
431 | my $i; | |
432 | my $start = find_starting_index($line); | |
433 | my $end = find_ending_index($line); | |
434 | for ($i = $start; $i < $end; $i++) { | |
435 | my $line = $typevalue[$i]; | |
436 | if ($line =~ /^[FX]:/) { ##Restore file patterns | |
437 | $line =~ s/([^\\])\.([^\*])/$1\?$2/g; | |
438 | $line =~ s/([^\\])\.$/$1\?/g; ##Convert . back to ? | |
439 | $line =~ s/\\\./\./g; ##Convert \. to . | |
440 | $line =~ s/\.\*/\*/g; ##Convert .* to * | |
4b76c9da | 441 | } |
6ffd9485 JP |
442 | $line =~ s/^([A-Z]):/$1:\t/g; |
443 | print("$line\n"); | |
4b76c9da | 444 | } |
6ffd9485 JP |
445 | print("\n"); |
446 | } | |
cb7301c7 JP |
447 | } |
448 | ||
e3e9d114 JP |
449 | if ($email && |
450 | ($email_git || ($email_git_fallback && !$exact_pattern_match))) { | |
60db31ac | 451 | vcs_file_signoffs($file); |
cb7301c7 JP |
452 | } |
453 | ||
f5492666 | 454 | if ($email && $email_git_blame) { |
60db31ac | 455 | vcs_file_blame($file); |
f5492666 | 456 | } |
cb7301c7 JP |
457 | } |
458 | ||
dcf36a92 JP |
459 | if ($keywords) { |
460 | @keyword_tvi = sort_and_uniq(@keyword_tvi); | |
461 | foreach my $line (@keyword_tvi) { | |
462 | add_categories($line); | |
463 | } | |
464 | } | |
465 | ||
f5f5078d | 466 | if ($email) { |
cb7301c7 JP |
467 | foreach my $chief (@penguin_chief) { |
468 | if ($chief =~ m/^(.*):(.*)/) { | |
f5f5078d | 469 | my $email_address; |
0e70e83d | 470 | |
a8af2430 | 471 | $email_address = format_email($1, $2, $email_usename); |
f5f5078d | 472 | if ($email_git_penguin_chiefs) { |
3c7385b8 | 473 | push(@email_to, [$email_address, 'chief penguin']); |
f5f5078d | 474 | } else { |
3c7385b8 | 475 | @email_to = grep($_->[0] !~ /${email_address}/, @email_to); |
cb7301c7 JP |
476 | } |
477 | } | |
478 | } | |
03372dbb JP |
479 | |
480 | foreach my $email (@file_emails) { | |
481 | my ($name, $address) = parse_email($email); | |
482 | ||
483 | my $tmp_email = format_email($name, $address, $email_usename); | |
484 | push_email_address($tmp_email, ''); | |
485 | add_role($tmp_email, 'in file'); | |
486 | } | |
cb7301c7 JP |
487 | } |
488 | ||
290603c1 JP |
489 | if ($email || $email_list) { |
490 | my @to = (); | |
491 | if ($email) { | |
492 | @to = (@to, @email_to); | |
cb7301c7 | 493 | } |
290603c1 | 494 | if ($email_list) { |
290603c1 | 495 | @to = (@to, @list_to); |
290603c1 | 496 | } |
3c7385b8 | 497 | output(merge_email(@to)); |
cb7301c7 JP |
498 | } |
499 | ||
500 | if ($scm) { | |
b781655a | 501 | @scm = uniq(@scm); |
cb7301c7 JP |
502 | output(@scm); |
503 | } | |
504 | ||
505 | if ($status) { | |
b781655a | 506 | @status = uniq(@status); |
cb7301c7 JP |
507 | output(@status); |
508 | } | |
509 | ||
510 | if ($subsystem) { | |
b781655a | 511 | @subsystem = uniq(@subsystem); |
cb7301c7 JP |
512 | output(@subsystem); |
513 | } | |
514 | ||
515 | if ($web) { | |
b781655a | 516 | @web = uniq(@web); |
cb7301c7 JP |
517 | output(@web); |
518 | } | |
519 | ||
520 | exit($exit); | |
521 | ||
522 | sub file_match_pattern { | |
523 | my ($file, $pattern) = @_; | |
524 | if (substr($pattern, -1) eq "/") { | |
525 | if ($file =~ m@^$pattern@) { | |
526 | return 1; | |
527 | } | |
528 | } else { | |
529 | if ($file =~ m@^$pattern@) { | |
530 | my $s1 = ($file =~ tr@/@@); | |
531 | my $s2 = ($pattern =~ tr@/@@); | |
532 | if ($s1 == $s2) { | |
533 | return 1; | |
534 | } | |
535 | } | |
536 | } | |
537 | return 0; | |
538 | } | |
539 | ||
540 | sub usage { | |
541 | print <<EOT; | |
542 | usage: $P [options] patchfile | |
870020f9 | 543 | $P [options] -f file|directory |
cb7301c7 JP |
544 | version: $V |
545 | ||
546 | MAINTAINER field selection options: | |
547 | --email => print email address(es) if any | |
548 | --git => include recent git \*-by: signers | |
e4d26b02 JP |
549 | --git-all-signature-types => include signers regardless of signature type |
550 | or use only ${signaturePattern} signers (default: $email_git_all_signature_types) | |
e3e9d114 | 551 | --git-fallback => use git when no exact MAINTAINERS pattern (default: $email_git_fallback) |
cb7301c7 | 552 | --git-chief-penguins => include ${penguin_chiefs} |
e4d26b02 JP |
553 | --git-min-signatures => number of signatures required (default: $email_git_min_signatures) |
554 | --git-max-maintainers => maximum maintainers to add (default: $email_git_max_maintainers) | |
555 | --git-min-percent => minimum percentage of commits required (default: $email_git_min_percent) | |
f5492666 | 556 | --git-blame => use git blame to find modified commits for patch or file |
e4d26b02 JP |
557 | --git-since => git history to use (default: $email_git_since) |
558 | --hg-since => hg history to use (default: $email_hg_since) | |
cb7301c7 JP |
559 | --m => include maintainer(s) if any |
560 | --n => include name 'Full Name <addr\@domain.tld>' | |
561 | --l => include list(s) if any | |
562 | --s => include subscriber only list(s) if any | |
11ecf53c | 563 | --remove-duplicates => minimize duplicate email names/addresses |
3c7385b8 JP |
564 | --roles => show roles (status:subsystem, git-signer, list, etc...) |
565 | --rolestats => show roles and statistics (commits/total_commits, %) | |
03372dbb | 566 | --file-emails => add email addresses found in -f file (default: 0 (off)) |
cb7301c7 JP |
567 | --scm => print SCM tree(s) if any |
568 | --status => print status if any | |
569 | --subsystem => print subsystem name if any | |
570 | --web => print website(s) if any | |
571 | ||
572 | Output type options: | |
573 | --separator [, ] => separator for multiple entries on 1 line | |
42498316 | 574 | using --separator also sets --nomultiline if --separator is not [, ] |
cb7301c7 JP |
575 | --multiline => print 1 entry per line |
576 | ||
cb7301c7 | 577 | Other options: |
3fb55652 | 578 | --pattern-depth => Number of pattern directory traversals (default: 0 (all)) |
dcf36a92 | 579 | --keywords => scan patch for keywords (default: 1 (on)) |
4b76c9da | 580 | --sections => print the entire subsystem sections with pattern matches |
f5f5078d | 581 | --version => show version |
cb7301c7 JP |
582 | --help => show this help information |
583 | ||
3fb55652 | 584 | Default options: |
11ecf53c | 585 | [--email --git --m --n --l --multiline --pattern-depth=0 --remove-duplicates] |
3fb55652 | 586 | |
870020f9 JP |
587 | Notes: |
588 | Using "-f directory" may give unexpected results: | |
f5492666 JP |
589 | Used with "--git", git signators for _all_ files in and below |
590 | directory are examined as git recurses directories. | |
591 | Any specified X: (exclude) pattern matches are _not_ ignored. | |
592 | Used with "--nogit", directory is used as a pattern match, | |
60db31ac JP |
593 | no individual file within the directory or subdirectory |
594 | is matched. | |
f5492666 JP |
595 | Used with "--git-blame", does not iterate all files in directory |
596 | Using "--git-blame" is slow and may add old committers and authors | |
597 | that are no longer active maintainers to the output. | |
3c7385b8 JP |
598 | Using "--roles" or "--rolestats" with git send-email --cc-cmd or any |
599 | other automated tools that expect only ["name"] <email address> | |
600 | may not work because of additional output after <email address>. | |
601 | Using "--rolestats" and "--git-blame" shows the #/total=% commits, | |
602 | not the percentage of the entire file authored. # of commits is | |
603 | not a good measure of amount of code authored. 1 major commit may | |
604 | contain a thousand lines, 5 trivial commits may modify a single line. | |
60db31ac JP |
605 | If git is not installed, but mercurial (hg) is installed and an .hg |
606 | repository exists, the following options apply to mercurial: | |
607 | --git, | |
608 | --git-min-signatures, --git-max-maintainers, --git-min-percent, and | |
609 | --git-blame | |
610 | Use --hg-since not --git-since to control date selection | |
368669da JP |
611 | File ".get_maintainer.conf", if it exists in the linux kernel source root |
612 | directory, can change whatever get_maintainer defaults are desired. | |
613 | Entries in this file can be any command line argument. | |
614 | This file is prepended to any additional command line arguments. | |
615 | Multiple lines and # comments are allowed. | |
cb7301c7 JP |
616 | EOT |
617 | } | |
618 | ||
619 | sub top_of_kernel_tree { | |
620 | my ($lk_path) = @_; | |
621 | ||
622 | if ($lk_path ne "" && substr($lk_path,length($lk_path)-1,1) ne "/") { | |
623 | $lk_path .= "/"; | |
624 | } | |
625 | if ( (-f "${lk_path}COPYING") | |
626 | && (-f "${lk_path}CREDITS") | |
627 | && (-f "${lk_path}Kbuild") | |
628 | && (-f "${lk_path}MAINTAINERS") | |
629 | && (-f "${lk_path}Makefile") | |
630 | && (-f "${lk_path}README") | |
631 | && (-d "${lk_path}Documentation") | |
632 | && (-d "${lk_path}arch") | |
633 | && (-d "${lk_path}include") | |
634 | && (-d "${lk_path}drivers") | |
635 | && (-d "${lk_path}fs") | |
636 | && (-d "${lk_path}init") | |
637 | && (-d "${lk_path}ipc") | |
638 | && (-d "${lk_path}kernel") | |
639 | && (-d "${lk_path}lib") | |
640 | && (-d "${lk_path}scripts")) { | |
641 | return 1; | |
642 | } | |
643 | return 0; | |
644 | } | |
645 | ||
0e70e83d JP |
646 | sub parse_email { |
647 | my ($formatted_email) = @_; | |
648 | ||
649 | my $name = ""; | |
650 | my $address = ""; | |
651 | ||
11ecf53c | 652 | if ($formatted_email =~ /^([^<]+)<(.+\@.*)>.*$/) { |
0e70e83d JP |
653 | $name = $1; |
654 | $address = $2; | |
11ecf53c | 655 | } elsif ($formatted_email =~ /^\s*<(.+\@\S*)>.*$/) { |
0e70e83d | 656 | $address = $1; |
b781655a | 657 | } elsif ($formatted_email =~ /^(.+\@\S*).*$/) { |
0e70e83d JP |
658 | $address = $1; |
659 | } | |
cb7301c7 JP |
660 | |
661 | $name =~ s/^\s+|\s+$//g; | |
d789504a | 662 | $name =~ s/^\"|\"$//g; |
0e70e83d | 663 | $address =~ s/^\s+|\s+$//g; |
cb7301c7 | 664 | |
a63ceb4c | 665 | if ($name =~ /[^\w \-]/i) { ##has "must quote" chars |
0e70e83d JP |
666 | $name =~ s/(?<!\\)"/\\"/g; ##escape quotes |
667 | $name = "\"$name\""; | |
668 | } | |
669 | ||
670 | return ($name, $address); | |
671 | } | |
672 | ||
673 | sub format_email { | |
a8af2430 | 674 | my ($name, $address, $usename) = @_; |
0e70e83d JP |
675 | |
676 | my $formatted_email; | |
677 | ||
678 | $name =~ s/^\s+|\s+$//g; | |
679 | $name =~ s/^\"|\"$//g; | |
680 | $address =~ s/^\s+|\s+$//g; | |
cb7301c7 | 681 | |
a63ceb4c | 682 | if ($name =~ /[^\w \-]/i) { ##has "must quote" chars |
cb7301c7 | 683 | $name =~ s/(?<!\\)"/\\"/g; ##escape quotes |
0e70e83d JP |
684 | $name = "\"$name\""; |
685 | } | |
686 | ||
a8af2430 | 687 | if ($usename) { |
0e70e83d JP |
688 | if ("$name" eq "") { |
689 | $formatted_email = "$address"; | |
690 | } else { | |
a8af2430 | 691 | $formatted_email = "$name <$address>"; |
0e70e83d | 692 | } |
cb7301c7 | 693 | } else { |
0e70e83d | 694 | $formatted_email = $address; |
cb7301c7 | 695 | } |
0e70e83d | 696 | |
cb7301c7 JP |
697 | return $formatted_email; |
698 | } | |
699 | ||
272a8979 JP |
700 | sub find_first_section { |
701 | my $index = 0; | |
702 | ||
703 | while ($index < @typevalue) { | |
704 | my $tv = $typevalue[$index]; | |
705 | if (($tv =~ m/^(\C):\s*(.*)/)) { | |
706 | last; | |
707 | } | |
708 | $index++; | |
709 | } | |
710 | ||
711 | return $index; | |
712 | } | |
713 | ||
b781655a | 714 | sub find_starting_index { |
b781655a JP |
715 | my ($index) = @_; |
716 | ||
717 | while ($index > 0) { | |
718 | my $tv = $typevalue[$index]; | |
719 | if (!($tv =~ m/^(\C):\s*(.*)/)) { | |
720 | last; | |
721 | } | |
722 | $index--; | |
723 | } | |
724 | ||
725 | return $index; | |
726 | } | |
727 | ||
728 | sub find_ending_index { | |
cb7301c7 JP |
729 | my ($index) = @_; |
730 | ||
b781655a | 731 | while ($index < @typevalue) { |
cb7301c7 | 732 | my $tv = $typevalue[$index]; |
b781655a JP |
733 | if (!($tv =~ m/^(\C):\s*(.*)/)) { |
734 | last; | |
735 | } | |
736 | $index++; | |
737 | } | |
738 | ||
739 | return $index; | |
740 | } | |
741 | ||
3c7385b8 JP |
742 | sub get_maintainer_role { |
743 | my ($index) = @_; | |
744 | ||
745 | my $i; | |
746 | my $start = find_starting_index($index); | |
747 | my $end = find_ending_index($index); | |
748 | ||
749 | my $role; | |
750 | my $subsystem = $typevalue[$start]; | |
751 | if (length($subsystem) > 20) { | |
752 | $subsystem = substr($subsystem, 0, 17); | |
753 | $subsystem =~ s/\s*$//; | |
754 | $subsystem = $subsystem . "..."; | |
755 | } | |
756 | ||
757 | for ($i = $start + 1; $i < $end; $i++) { | |
758 | my $tv = $typevalue[$i]; | |
759 | if ($tv =~ m/^(\C):\s*(.*)/) { | |
760 | my $ptype = $1; | |
761 | my $pvalue = $2; | |
762 | if ($ptype eq "S") { | |
763 | $role = $pvalue; | |
764 | } | |
765 | } | |
766 | } | |
767 | ||
768 | $role = lc($role); | |
769 | if ($role eq "supported") { | |
770 | $role = "supporter"; | |
771 | } elsif ($role eq "maintained") { | |
772 | $role = "maintainer"; | |
773 | } elsif ($role eq "odd fixes") { | |
774 | $role = "odd fixer"; | |
775 | } elsif ($role eq "orphan") { | |
776 | $role = "orphan minder"; | |
777 | } elsif ($role eq "obsolete") { | |
778 | $role = "obsolete minder"; | |
779 | } elsif ($role eq "buried alive in reporters") { | |
780 | $role = "chief penguin"; | |
781 | } | |
782 | ||
783 | return $role . ":" . $subsystem; | |
784 | } | |
785 | ||
786 | sub get_list_role { | |
787 | my ($index) = @_; | |
788 | ||
789 | my $i; | |
790 | my $start = find_starting_index($index); | |
791 | my $end = find_ending_index($index); | |
792 | ||
793 | my $subsystem = $typevalue[$start]; | |
794 | if (length($subsystem) > 20) { | |
795 | $subsystem = substr($subsystem, 0, 17); | |
796 | $subsystem =~ s/\s*$//; | |
797 | $subsystem = $subsystem . "..."; | |
798 | } | |
799 | ||
800 | if ($subsystem eq "THE REST") { | |
801 | $subsystem = ""; | |
802 | } | |
803 | ||
804 | return $subsystem; | |
805 | } | |
806 | ||
b781655a JP |
807 | sub add_categories { |
808 | my ($index) = @_; | |
809 | ||
810 | my $i; | |
811 | my $start = find_starting_index($index); | |
812 | my $end = find_ending_index($index); | |
813 | ||
814 | push(@subsystem, $typevalue[$start]); | |
815 | ||
816 | for ($i = $start + 1; $i < $end; $i++) { | |
817 | my $tv = $typevalue[$i]; | |
290603c1 | 818 | if ($tv =~ m/^(\C):\s*(.*)/) { |
cb7301c7 JP |
819 | my $ptype = $1; |
820 | my $pvalue = $2; | |
821 | if ($ptype eq "L") { | |
290603c1 JP |
822 | my $list_address = $pvalue; |
823 | my $list_additional = ""; | |
3c7385b8 JP |
824 | my $list_role = get_list_role($i); |
825 | ||
826 | if ($list_role ne "") { | |
827 | $list_role = ":" . $list_role; | |
828 | } | |
290603c1 JP |
829 | if ($list_address =~ m/([^\s]+)\s+(.*)$/) { |
830 | $list_address = $1; | |
831 | $list_additional = $2; | |
832 | } | |
bdf7c685 | 833 | if ($list_additional =~ m/subscribers-only/) { |
cb7301c7 | 834 | if ($email_subscriber_list) { |
3c7385b8 | 835 | push(@list_to, [$list_address, "subscriber list${list_role}"]); |
cb7301c7 JP |
836 | } |
837 | } else { | |
838 | if ($email_list) { | |
3c7385b8 | 839 | push(@list_to, [$list_address, "open list${list_role}"]); |
cb7301c7 JP |
840 | } |
841 | } | |
842 | } elsif ($ptype eq "M") { | |
0e70e83d JP |
843 | my ($name, $address) = parse_email($pvalue); |
844 | if ($name eq "") { | |
b781655a JP |
845 | if ($i > 0) { |
846 | my $tv = $typevalue[$i - 1]; | |
0e70e83d JP |
847 | if ($tv =~ m/^(\C):\s*(.*)/) { |
848 | if ($1 eq "P") { | |
849 | $name = $2; | |
a8af2430 | 850 | $pvalue = format_email($name, $address, $email_usename); |
5f2441e9 JP |
851 | } |
852 | } | |
853 | } | |
854 | } | |
0e70e83d | 855 | if ($email_maintainer) { |
3c7385b8 JP |
856 | my $role = get_maintainer_role($i); |
857 | push_email_addresses($pvalue, $role); | |
cb7301c7 JP |
858 | } |
859 | } elsif ($ptype eq "T") { | |
860 | push(@scm, $pvalue); | |
861 | } elsif ($ptype eq "W") { | |
862 | push(@web, $pvalue); | |
863 | } elsif ($ptype eq "S") { | |
864 | push(@status, $pvalue); | |
865 | } | |
cb7301c7 JP |
866 | } |
867 | } | |
868 | } | |
869 | ||
11ecf53c JP |
870 | my %email_hash_name; |
871 | my %email_hash_address; | |
0e70e83d | 872 | |
11ecf53c JP |
873 | sub email_inuse { |
874 | my ($name, $address) = @_; | |
875 | ||
876 | return 1 if (($name eq "") && ($address eq "")); | |
877 | return 1 if (($name ne "") && exists($email_hash_name{$name})); | |
878 | return 1 if (($address ne "") && exists($email_hash_address{$address})); | |
0e70e83d | 879 | |
0e70e83d JP |
880 | return 0; |
881 | } | |
882 | ||
1b5e1cf6 | 883 | sub push_email_address { |
3c7385b8 | 884 | my ($line, $role) = @_; |
1b5e1cf6 | 885 | |
0e70e83d | 886 | my ($name, $address) = parse_email($line); |
1b5e1cf6 | 887 | |
b781655a JP |
888 | if ($address eq "") { |
889 | return 0; | |
890 | } | |
891 | ||
11ecf53c | 892 | if (!$email_remove_duplicates) { |
a8af2430 | 893 | push(@email_to, [format_email($name, $address, $email_usename), $role]); |
11ecf53c | 894 | } elsif (!email_inuse($name, $address)) { |
a8af2430 | 895 | push(@email_to, [format_email($name, $address, $email_usename), $role]); |
11ecf53c JP |
896 | $email_hash_name{$name}++; |
897 | $email_hash_address{$address}++; | |
1b5e1cf6 | 898 | } |
b781655a JP |
899 | |
900 | return 1; | |
1b5e1cf6 JP |
901 | } |
902 | ||
903 | sub push_email_addresses { | |
3c7385b8 | 904 | my ($address, $role) = @_; |
1b5e1cf6 JP |
905 | |
906 | my @address_list = (); | |
907 | ||
5f2441e9 | 908 | if (rfc822_valid($address)) { |
3c7385b8 | 909 | push_email_address($address, $role); |
5f2441e9 | 910 | } elsif (@address_list = rfc822_validlist($address)) { |
1b5e1cf6 JP |
911 | my $array_count = shift(@address_list); |
912 | while (my $entry = shift(@address_list)) { | |
3c7385b8 | 913 | push_email_address($entry, $role); |
1b5e1cf6 | 914 | } |
5f2441e9 | 915 | } else { |
3c7385b8 | 916 | if (!push_email_address($address, $role)) { |
b781655a JP |
917 | warn("Invalid MAINTAINERS address: '" . $address . "'\n"); |
918 | } | |
1b5e1cf6 | 919 | } |
1b5e1cf6 JP |
920 | } |
921 | ||
3c7385b8 JP |
922 | sub add_role { |
923 | my ($line, $role) = @_; | |
924 | ||
925 | my ($name, $address) = parse_email($line); | |
a8af2430 | 926 | my $email = format_email($name, $address, $email_usename); |
3c7385b8 JP |
927 | |
928 | foreach my $entry (@email_to) { | |
929 | if ($email_remove_duplicates) { | |
930 | my ($entry_name, $entry_address) = parse_email($entry->[0]); | |
03372dbb JP |
931 | if (($name eq $entry_name || $address eq $entry_address) |
932 | && ($role eq "" || !($entry->[1] =~ m/$role/)) | |
933 | ) { | |
3c7385b8 JP |
934 | if ($entry->[1] eq "") { |
935 | $entry->[1] = "$role"; | |
936 | } else { | |
937 | $entry->[1] = "$entry->[1],$role"; | |
938 | } | |
939 | } | |
940 | } else { | |
03372dbb JP |
941 | if ($email eq $entry->[0] |
942 | && ($role eq "" || !($entry->[1] =~ m/$role/)) | |
943 | ) { | |
3c7385b8 JP |
944 | if ($entry->[1] eq "") { |
945 | $entry->[1] = "$role"; | |
946 | } else { | |
947 | $entry->[1] = "$entry->[1],$role"; | |
948 | } | |
949 | } | |
950 | } | |
951 | } | |
952 | } | |
953 | ||
cb7301c7 JP |
954 | sub which { |
955 | my ($bin) = @_; | |
956 | ||
f5f5078d | 957 | foreach my $path (split(/:/, $ENV{PATH})) { |
cb7301c7 JP |
958 | if (-e "$path/$bin") { |
959 | return "$path/$bin"; | |
960 | } | |
961 | } | |
962 | ||
963 | return ""; | |
964 | } | |
965 | ||
bcde44ed JP |
966 | sub which_conf { |
967 | my ($conf) = @_; | |
968 | ||
969 | foreach my $path (split(/:/, ".:$ENV{HOME}:.scripts")) { | |
970 | if (-e "$path/$conf") { | |
971 | return "$path/$conf"; | |
972 | } | |
973 | } | |
974 | ||
975 | return ""; | |
976 | } | |
977 | ||
8cbb3a77 | 978 | sub mailmap { |
a8af2430 | 979 | my (@lines) = @_; |
8cbb3a77 JP |
980 | my %hash; |
981 | ||
982 | foreach my $line (@lines) { | |
983 | my ($name, $address) = parse_email($line); | |
984 | if (!exists($hash{$name})) { | |
985 | $hash{$name} = $address; | |
11ecf53c JP |
986 | } elsif ($address ne $hash{$name}) { |
987 | $address = $hash{$name}; | |
a8af2430 | 988 | $line = format_email($name, $address, $email_usename); |
8cbb3a77 JP |
989 | } |
990 | if (exists($mailmap{$name})) { | |
991 | my $obj = $mailmap{$name}; | |
992 | foreach my $map_address (@$obj) { | |
993 | if (($map_address eq $address) && | |
994 | ($map_address ne $hash{$name})) { | |
a8af2430 | 995 | $line = format_email($name, $hash{$name}, $email_usename); |
8cbb3a77 JP |
996 | } |
997 | } | |
998 | } | |
999 | } | |
1000 | ||
1001 | return @lines; | |
1002 | } | |
1003 | ||
60db31ac JP |
1004 | sub git_execute_cmd { |
1005 | my ($cmd) = @_; | |
1006 | my @lines = (); | |
cb7301c7 | 1007 | |
60db31ac JP |
1008 | my $output = `$cmd`; |
1009 | $output =~ s/^\s*//gm; | |
1010 | @lines = split("\n", $output); | |
1011 | ||
1012 | return @lines; | |
a8af2430 JP |
1013 | } |
1014 | ||
60db31ac | 1015 | sub hg_execute_cmd { |
a8af2430 | 1016 | my ($cmd) = @_; |
60db31ac JP |
1017 | my @lines = (); |
1018 | ||
1019 | my $output = `$cmd`; | |
1020 | @lines = split("\n", $output); | |
a8af2430 | 1021 | |
60db31ac JP |
1022 | return @lines; |
1023 | } | |
1024 | ||
1025 | sub vcs_find_signers { | |
1026 | my ($cmd) = @_; | |
a8af2430 JP |
1027 | my @lines = (); |
1028 | my $commits; | |
1029 | ||
60db31ac | 1030 | @lines = &{$VCS_cmds{"execute_cmd"}}($cmd); |
cb7301c7 | 1031 | |
60db31ac | 1032 | my $pattern = $VCS_cmds{"commit_pattern"}; |
cb7301c7 | 1033 | |
60db31ac | 1034 | $commits = grep(/$pattern/, @lines); # of commits |
afa81ee1 | 1035 | |
e4d26b02 | 1036 | @lines = grep(/^[ \t]*${signaturePattern}.*\@.*$/, @lines); |
0e70e83d JP |
1037 | if (!$email_git_penguin_chiefs) { |
1038 | @lines = grep(!/${penguin_chiefs}/i, @lines); | |
1039 | } | |
63ab52db JP |
1040 | |
1041 | return (0, @lines) if !@lines; | |
1042 | ||
0e70e83d JP |
1043 | # cut -f2- -d":" |
1044 | s/.*:\s*(.+)\s*/$1/ for (@lines); | |
1045 | ||
a8af2430 JP |
1046 | ## Reformat email addresses (with names) to avoid badly written signatures |
1047 | ||
3c7385b8 JP |
1048 | foreach my $line (@lines) { |
1049 | my ($name, $address) = parse_email($line); | |
a8af2430 JP |
1050 | $line = format_email($name, $address, 1); |
1051 | } | |
1052 | ||
1053 | return ($commits, @lines); | |
1054 | } | |
1055 | ||
63ab52db JP |
1056 | sub vcs_find_author { |
1057 | my ($cmd) = @_; | |
1058 | my @lines = (); | |
1059 | ||
1060 | @lines = &{$VCS_cmds{"execute_cmd"}}($cmd); | |
1061 | ||
1062 | if (!$email_git_penguin_chiefs) { | |
1063 | @lines = grep(!/${penguin_chiefs}/i, @lines); | |
1064 | } | |
1065 | ||
1066 | return @lines if !@lines; | |
1067 | ||
1068 | ## Reformat email addresses (with names) to avoid badly written signatures | |
1069 | ||
1070 | foreach my $line (@lines) { | |
1071 | my ($name, $address) = parse_email($line); | |
1072 | $line = format_email($name, $address, 1); | |
1073 | } | |
1074 | ||
1075 | return @lines; | |
1076 | } | |
1077 | ||
60db31ac JP |
1078 | sub vcs_save_commits { |
1079 | my ($cmd) = @_; | |
1080 | my @lines = (); | |
1081 | my @commits = (); | |
1082 | ||
1083 | @lines = &{$VCS_cmds{"execute_cmd"}}($cmd); | |
1084 | ||
1085 | foreach my $line (@lines) { | |
1086 | if ($line =~ m/$VCS_cmds{"blame_commit_pattern"}/) { | |
1087 | push(@commits, $1); | |
1088 | } | |
1089 | } | |
1090 | ||
1091 | return @commits; | |
1092 | } | |
1093 | ||
1094 | sub vcs_blame { | |
1095 | my ($file) = @_; | |
1096 | my $cmd; | |
1097 | my @commits = (); | |
1098 | ||
1099 | return @commits if (!(-f $file)); | |
1100 | ||
1101 | if (@range && $VCS_cmds{"blame_range_cmd"} eq "") { | |
1102 | my @all_commits = (); | |
1103 | ||
1104 | $cmd = $VCS_cmds{"blame_file_cmd"}; | |
1105 | $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd | |
1106 | @all_commits = vcs_save_commits($cmd); | |
1107 | ||
1108 | foreach my $file_range_diff (@range) { | |
1109 | next if (!($file_range_diff =~ m/(.+):(.+):(.+)/)); | |
1110 | my $diff_file = $1; | |
1111 | my $diff_start = $2; | |
1112 | my $diff_length = $3; | |
1113 | next if ("$file" ne "$diff_file"); | |
1114 | for (my $i = $diff_start; $i < $diff_start + $diff_length; $i++) { | |
1115 | push(@commits, $all_commits[$i]); | |
1116 | } | |
1117 | } | |
1118 | } elsif (@range) { | |
1119 | foreach my $file_range_diff (@range) { | |
1120 | next if (!($file_range_diff =~ m/(.+):(.+):(.+)/)); | |
1121 | my $diff_file = $1; | |
1122 | my $diff_start = $2; | |
1123 | my $diff_length = $3; | |
1124 | next if ("$file" ne "$diff_file"); | |
1125 | $cmd = $VCS_cmds{"blame_range_cmd"}; | |
1126 | $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd | |
1127 | push(@commits, vcs_save_commits($cmd)); | |
1128 | } | |
1129 | } else { | |
1130 | $cmd = $VCS_cmds{"blame_file_cmd"}; | |
1131 | $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd | |
1132 | @commits = vcs_save_commits($cmd); | |
1133 | } | |
1134 | ||
63ab52db JP |
1135 | foreach my $commit (@commits) { |
1136 | $commit =~ s/^\^//g; | |
1137 | } | |
1138 | ||
60db31ac JP |
1139 | return @commits; |
1140 | } | |
1141 | ||
1142 | my $printed_novcs = 0; | |
1143 | sub vcs_exists { | |
1144 | %VCS_cmds = %VCS_cmds_git; | |
1145 | return 1 if eval $VCS_cmds{"available"}; | |
1146 | %VCS_cmds = %VCS_cmds_hg; | |
1147 | return 1 if eval $VCS_cmds{"available"}; | |
1148 | %VCS_cmds = (); | |
1149 | if (!$printed_novcs) { | |
1150 | warn("$P: No supported VCS found. Add --nogit to options?\n"); | |
1151 | warn("Using a git repository produces better results.\n"); | |
1152 | warn("Try Linus Torvalds' latest git repository using:\n"); | |
1153 | warn("git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git\n"); | |
1154 | $printed_novcs = 1; | |
1155 | } | |
1156 | return 0; | |
1157 | } | |
1158 | ||
1159 | sub vcs_assign { | |
a8af2430 JP |
1160 | my ($role, $divisor, @lines) = @_; |
1161 | ||
1162 | my %hash; | |
1163 | my $count = 0; | |
1164 | ||
a8af2430 JP |
1165 | return if (@lines <= 0); |
1166 | ||
1167 | if ($divisor <= 0) { | |
60db31ac | 1168 | warn("Bad divisor in " . (caller(0))[3] . ": $divisor\n"); |
a8af2430 | 1169 | $divisor = 1; |
3c7385b8 | 1170 | } |
8cbb3a77 | 1171 | |
11ecf53c JP |
1172 | if ($email_remove_duplicates) { |
1173 | @lines = mailmap(@lines); | |
1174 | } | |
0e70e83d | 1175 | |
63ab52db JP |
1176 | return if (@lines <= 0); |
1177 | ||
0e70e83d | 1178 | @lines = sort(@lines); |
11ecf53c | 1179 | |
0e70e83d | 1180 | # uniq -c |
11ecf53c JP |
1181 | $hash{$_}++ for @lines; |
1182 | ||
0e70e83d | 1183 | # sort -rn |
0e70e83d | 1184 | foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) { |
11ecf53c | 1185 | my $sign_offs = $hash{$line}; |
a8af2430 | 1186 | my $percent = $sign_offs * 100 / $divisor; |
3c7385b8 | 1187 | |
a8af2430 | 1188 | $percent = 100 if ($percent > 100); |
11ecf53c JP |
1189 | $count++; |
1190 | last if ($sign_offs < $email_git_min_signatures || | |
1191 | $count > $email_git_max_maintainers || | |
a8af2430 | 1192 | $percent < $email_git_min_percent); |
3c7385b8 | 1193 | push_email_address($line, ''); |
3c7385b8 | 1194 | if ($output_rolestats) { |
a8af2430 JP |
1195 | my $fmt_percent = sprintf("%.0f", $percent); |
1196 | add_role($line, "$role:$sign_offs/$divisor=$fmt_percent%"); | |
1197 | } else { | |
1198 | add_role($line, $role); | |
3c7385b8 | 1199 | } |
f5492666 JP |
1200 | } |
1201 | } | |
1202 | ||
60db31ac | 1203 | sub vcs_file_signoffs { |
a8af2430 JP |
1204 | my ($file) = @_; |
1205 | ||
1206 | my @signers = (); | |
60db31ac | 1207 | my $commits; |
f5492666 | 1208 | |
60db31ac | 1209 | return if (!vcs_exists()); |
a8af2430 | 1210 | |
60db31ac JP |
1211 | my $cmd = $VCS_cmds{"find_signers_cmd"}; |
1212 | $cmd =~ s/(\$\w+)/$1/eeg; # interpolate $cmd | |
f5492666 | 1213 | |
60db31ac JP |
1214 | ($commits, @signers) = vcs_find_signers($cmd); |
1215 | vcs_assign("commit_signer", $commits, @signers); | |
f5492666 JP |
1216 | } |
1217 | ||
60db31ac | 1218 | sub vcs_file_blame { |
f5492666 JP |
1219 | my ($file) = @_; |
1220 | ||
a8af2430 | 1221 | my @signers = (); |
63ab52db | 1222 | my @all_commits = (); |
60db31ac | 1223 | my @commits = (); |
a8af2430 | 1224 | my $total_commits; |
63ab52db | 1225 | my $total_lines; |
f5492666 | 1226 | |
60db31ac | 1227 | return if (!vcs_exists()); |
f5492666 | 1228 | |
63ab52db JP |
1229 | @all_commits = vcs_blame($file); |
1230 | @commits = uniq(@all_commits); | |
a8af2430 | 1231 | $total_commits = @commits; |
63ab52db | 1232 | $total_lines = @all_commits; |
8cbb3a77 | 1233 | |
a8af2430 JP |
1234 | foreach my $commit (@commits) { |
1235 | my $commit_count; | |
1236 | my @commit_signers = (); | |
8cbb3a77 | 1237 | |
60db31ac JP |
1238 | my $cmd = $VCS_cmds{"find_commit_signers_cmd"}; |
1239 | $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd | |
1240 | ||
1241 | ($commit_count, @commit_signers) = vcs_find_signers($cmd); | |
63ab52db | 1242 | |
60db31ac | 1243 | push(@signers, @commit_signers); |
f5492666 JP |
1244 | } |
1245 | ||
a8af2430 | 1246 | if ($from_filename) { |
63ab52db JP |
1247 | if ($output_rolestats) { |
1248 | my @blame_signers; | |
1249 | foreach my $commit (@commits) { | |
1250 | my $i; | |
1251 | my $cmd = $VCS_cmds{"find_commit_author_cmd"}; | |
1252 | $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd | |
1253 | my @author = vcs_find_author($cmd); | |
1254 | next if !@author; | |
1255 | my $count = grep(/$commit/, @all_commits); | |
1256 | for ($i = 0; $i < $count ; $i++) { | |
1257 | push(@blame_signers, $author[0]); | |
1258 | } | |
1259 | } | |
1260 | if (@blame_signers) { | |
1261 | vcs_assign("authored lines", $total_lines, @blame_signers); | |
1262 | } | |
1263 | } | |
60db31ac | 1264 | vcs_assign("commits", $total_commits, @signers); |
a8af2430 | 1265 | } else { |
60db31ac | 1266 | vcs_assign("modified commits", $total_commits, @signers); |
cb7301c7 | 1267 | } |
cb7301c7 JP |
1268 | } |
1269 | ||
1270 | sub uniq { | |
a8af2430 | 1271 | my (@parms) = @_; |
cb7301c7 JP |
1272 | |
1273 | my %saw; | |
1274 | @parms = grep(!$saw{$_}++, @parms); | |
1275 | return @parms; | |
1276 | } | |
1277 | ||
1278 | sub sort_and_uniq { | |
a8af2430 | 1279 | my (@parms) = @_; |
cb7301c7 JP |
1280 | |
1281 | my %saw; | |
1282 | @parms = sort @parms; | |
1283 | @parms = grep(!$saw{$_}++, @parms); | |
1284 | return @parms; | |
1285 | } | |
1286 | ||
03372dbb JP |
1287 | sub clean_file_emails { |
1288 | my (@file_emails) = @_; | |
1289 | my @fmt_emails = (); | |
1290 | ||
1291 | foreach my $email (@file_emails) { | |
1292 | $email =~ s/[\(\<\{]{0,1}([A-Za-z0-9_\.\+-]+\@[A-Za-z0-9\.-]+)[\)\>\}]{0,1}/\<$1\>/g; | |
1293 | my ($name, $address) = parse_email($email); | |
1294 | if ($name eq '"[,\.]"') { | |
1295 | $name = ""; | |
1296 | } | |
1297 | ||
1298 | my @nw = split(/[^A-Za-zÀ-ÿ\'\,\.\+-]/, $name); | |
1299 | if (@nw > 2) { | |
1300 | my $first = $nw[@nw - 3]; | |
1301 | my $middle = $nw[@nw - 2]; | |
1302 | my $last = $nw[@nw - 1]; | |
1303 | ||
1304 | if (((length($first) == 1 && $first =~ m/[A-Za-z]/) || | |
1305 | (length($first) == 2 && substr($first, -1) eq ".")) || | |
1306 | (length($middle) == 1 || | |
1307 | (length($middle) == 2 && substr($middle, -1) eq "."))) { | |
1308 | $name = "$first $middle $last"; | |
1309 | } else { | |
1310 | $name = "$middle $last"; | |
1311 | } | |
1312 | } | |
1313 | ||
1314 | if (substr($name, -1) =~ /[,\.]/) { | |
1315 | $name = substr($name, 0, length($name) - 1); | |
1316 | } elsif (substr($name, -2) =~ /[,\.]"/) { | |
1317 | $name = substr($name, 0, length($name) - 2) . '"'; | |
1318 | } | |
1319 | ||
1320 | if (substr($name, 0, 1) =~ /[,\.]/) { | |
1321 | $name = substr($name, 1, length($name) - 1); | |
1322 | } elsif (substr($name, 0, 2) =~ /"[,\.]/) { | |
1323 | $name = '"' . substr($name, 2, length($name) - 2); | |
1324 | } | |
1325 | ||
1326 | my $fmt_email = format_email($name, $address, $email_usename); | |
1327 | push(@fmt_emails, $fmt_email); | |
1328 | } | |
1329 | return @fmt_emails; | |
1330 | } | |
1331 | ||
3c7385b8 JP |
1332 | sub merge_email { |
1333 | my @lines; | |
1334 | my %saw; | |
1335 | ||
1336 | for (@_) { | |
1337 | my ($address, $role) = @$_; | |
1338 | if (!$saw{$address}) { | |
1339 | if ($output_roles) { | |
60db31ac | 1340 | push(@lines, "$address ($role)"); |
3c7385b8 | 1341 | } else { |
60db31ac | 1342 | push(@lines, $address); |
3c7385b8 JP |
1343 | } |
1344 | $saw{$address} = 1; | |
1345 | } | |
1346 | } | |
1347 | ||
1348 | return @lines; | |
1349 | } | |
1350 | ||
cb7301c7 | 1351 | sub output { |
a8af2430 | 1352 | my (@parms) = @_; |
cb7301c7 JP |
1353 | |
1354 | if ($output_multiline) { | |
1355 | foreach my $line (@parms) { | |
1356 | print("${line}\n"); | |
1357 | } | |
1358 | } else { | |
1359 | print(join($output_separator, @parms)); | |
1360 | print("\n"); | |
1361 | } | |
1362 | } | |
1b5e1cf6 JP |
1363 | |
1364 | my $rfc822re; | |
1365 | ||
1366 | sub make_rfc822re { | |
1367 | # Basic lexical tokens are specials, domain_literal, quoted_string, atom, and | |
1368 | # comment. We must allow for rfc822_lwsp (or comments) after each of these. | |
1369 | # This regexp will only work on addresses which have had comments stripped | |
1370 | # and replaced with rfc822_lwsp. | |
1371 | ||
1372 | my $specials = '()<>@,;:\\\\".\\[\\]'; | |
1373 | my $controls = '\\000-\\037\\177'; | |
1374 | ||
1375 | my $dtext = "[^\\[\\]\\r\\\\]"; | |
1376 | my $domain_literal = "\\[(?:$dtext|\\\\.)*\\]$rfc822_lwsp*"; | |
1377 | ||
1378 | my $quoted_string = "\"(?:[^\\\"\\r\\\\]|\\\\.|$rfc822_lwsp)*\"$rfc822_lwsp*"; | |
1379 | ||
1380 | # Use zero-width assertion to spot the limit of an atom. A simple | |
1381 | # $rfc822_lwsp* causes the regexp engine to hang occasionally. | |
1382 | my $atom = "[^$specials $controls]+(?:$rfc822_lwsp+|\\Z|(?=[\\[\"$specials]))"; | |
1383 | my $word = "(?:$atom|$quoted_string)"; | |
1384 | my $localpart = "$word(?:\\.$rfc822_lwsp*$word)*"; | |
1385 | ||
1386 | my $sub_domain = "(?:$atom|$domain_literal)"; | |
1387 | my $domain = "$sub_domain(?:\\.$rfc822_lwsp*$sub_domain)*"; | |
1388 | ||
1389 | my $addr_spec = "$localpart\@$rfc822_lwsp*$domain"; | |
1390 | ||
1391 | my $phrase = "$word*"; | |
1392 | my $route = "(?:\@$domain(?:,\@$rfc822_lwsp*$domain)*:$rfc822_lwsp*)"; | |
1393 | my $route_addr = "\\<$rfc822_lwsp*$route?$addr_spec\\>$rfc822_lwsp*"; | |
1394 | my $mailbox = "(?:$addr_spec|$phrase$route_addr)"; | |
1395 | ||
1396 | my $group = "$phrase:$rfc822_lwsp*(?:$mailbox(?:,\\s*$mailbox)*)?;\\s*"; | |
1397 | my $address = "(?:$mailbox|$group)"; | |
1398 | ||
1399 | return "$rfc822_lwsp*$address"; | |
1400 | } | |
1401 | ||
1402 | sub rfc822_strip_comments { | |
1403 | my $s = shift; | |
1404 | # Recursively remove comments, and replace with a single space. The simpler | |
1405 | # regexps in the Email Addressing FAQ are imperfect - they will miss escaped | |
1406 | # chars in atoms, for example. | |
1407 | ||
1408 | while ($s =~ s/^((?:[^"\\]|\\.)* | |
1409 | (?:"(?:[^"\\]|\\.)*"(?:[^"\\]|\\.)*)*) | |
1410 | \((?:[^()\\]|\\.)*\)/$1 /osx) {} | |
1411 | return $s; | |
1412 | } | |
1413 | ||
1414 | # valid: returns true if the parameter is an RFC822 valid address | |
1415 | # | |
22dd5b0c | 1416 | sub rfc822_valid { |
1b5e1cf6 JP |
1417 | my $s = rfc822_strip_comments(shift); |
1418 | ||
1419 | if (!$rfc822re) { | |
1420 | $rfc822re = make_rfc822re(); | |
1421 | } | |
1422 | ||
1423 | return $s =~ m/^$rfc822re$/so && $s =~ m/^$rfc822_char*$/; | |
1424 | } | |
1425 | ||
1426 | # validlist: In scalar context, returns true if the parameter is an RFC822 | |
1427 | # valid list of addresses. | |
1428 | # | |
1429 | # In list context, returns an empty list on failure (an invalid | |
1430 | # address was found); otherwise a list whose first element is the | |
1431 | # number of addresses found and whose remaining elements are the | |
1432 | # addresses. This is needed to disambiguate failure (invalid) | |
1433 | # from success with no addresses found, because an empty string is | |
1434 | # a valid list. | |
1435 | ||
22dd5b0c | 1436 | sub rfc822_validlist { |
1b5e1cf6 JP |
1437 | my $s = rfc822_strip_comments(shift); |
1438 | ||
1439 | if (!$rfc822re) { | |
1440 | $rfc822re = make_rfc822re(); | |
1441 | } | |
1442 | # * null list items are valid according to the RFC | |
1443 | # * the '1' business is to aid in distinguishing failure from no results | |
1444 | ||
1445 | my @r; | |
1446 | if ($s =~ m/^(?:$rfc822re)?(?:,(?:$rfc822re)?)*$/so && | |
1447 | $s =~ m/^$rfc822_char*$/) { | |
5f2441e9 | 1448 | while ($s =~ m/(?:^|,$rfc822_lwsp*)($rfc822re)/gos) { |
60db31ac | 1449 | push(@r, $1); |
1b5e1cf6 JP |
1450 | } |
1451 | return wantarray ? (scalar(@r), @r) : 1; | |
1452 | } | |
60db31ac | 1453 | return wantarray ? () : 0; |
1b5e1cf6 | 1454 | } |