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 | # | |
8 | # usage: perl scripts/get_maintainers.pl [OPTIONS] <patch> | |
9 | # perl scripts/get_maintainers.pl [OPTIONS] -f <file> | |
10 | # | |
11 | # Licensed under the terms of the GNU GPL License version 2 | |
12 | ||
13 | use strict; | |
14 | ||
15 | my $P = $0; | |
0e70e83d | 16 | my $V = '0.20'; |
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; | |
26 | my $email_git = 1; | |
27 | my $email_git_penguin_chiefs = 0; | |
28 | my $email_git_min_signatures = 1; | |
29 | my $email_git_max_maintainers = 5; | |
afa81ee1 | 30 | my $email_git_min_percent = 5; |
cb7301c7 | 31 | my $email_git_since = "1-year-ago"; |
f5492666 | 32 | my $email_git_blame = 0; |
cb7301c7 JP |
33 | my $output_multiline = 1; |
34 | my $output_separator = ", "; | |
35 | my $scm = 0; | |
36 | my $web = 0; | |
37 | my $subsystem = 0; | |
38 | my $status = 0; | |
4a7fdb5f | 39 | my $from_filename = 0; |
3fb55652 | 40 | my $pattern_depth = 0; |
cb7301c7 JP |
41 | my $version = 0; |
42 | my $help = 0; | |
43 | ||
44 | my $exit = 0; | |
45 | ||
46 | my @penguin_chief = (); | |
47 | push(@penguin_chief,"Linus Torvalds:torvalds\@linux-foundation.org"); | |
48 | #Andrew wants in on most everything - 2009/01/14 | |
49 | #push(@penguin_chief,"Andrew Morton:akpm\@linux-foundation.org"); | |
50 | ||
51 | my @penguin_chief_names = (); | |
52 | foreach my $chief (@penguin_chief) { | |
53 | if ($chief =~ m/^(.*):(.*)/) { | |
54 | my $chief_name = $1; | |
55 | my $chief_addr = $2; | |
56 | push(@penguin_chief_names, $chief_name); | |
57 | } | |
58 | } | |
59 | my $penguin_chiefs = "\(" . join("|",@penguin_chief_names) . "\)"; | |
60 | ||
5f2441e9 | 61 | # rfc822 email address - preloaded methods go here. |
1b5e1cf6 | 62 | my $rfc822_lwsp = "(?:(?:\\r\\n)?[ \\t])"; |
df4cc036 | 63 | my $rfc822_char = '[\\000-\\377]'; |
1b5e1cf6 | 64 | |
cb7301c7 JP |
65 | if (!GetOptions( |
66 | 'email!' => \$email, | |
67 | 'git!' => \$email_git, | |
68 | 'git-chief-penguins!' => \$email_git_penguin_chiefs, | |
69 | 'git-min-signatures=i' => \$email_git_min_signatures, | |
70 | 'git-max-maintainers=i' => \$email_git_max_maintainers, | |
afa81ee1 | 71 | 'git-min-percent=i' => \$email_git_min_percent, |
cb7301c7 | 72 | 'git-since=s' => \$email_git_since, |
f5492666 | 73 | 'git-blame!' => \$email_git_blame, |
cb7301c7 JP |
74 | 'm!' => \$email_maintainer, |
75 | 'n!' => \$email_usename, | |
76 | 'l!' => \$email_list, | |
77 | 's!' => \$email_subscriber_list, | |
78 | 'multiline!' => \$output_multiline, | |
79 | 'separator=s' => \$output_separator, | |
80 | 'subsystem!' => \$subsystem, | |
81 | 'status!' => \$status, | |
82 | 'scm!' => \$scm, | |
83 | 'web!' => \$web, | |
3fb55652 | 84 | 'pattern-depth=i' => \$pattern_depth, |
4a7fdb5f | 85 | 'f|file' => \$from_filename, |
cb7301c7 JP |
86 | 'v|version' => \$version, |
87 | 'h|help' => \$help, | |
88 | )) { | |
89 | usage(); | |
90 | die "$P: invalid argument\n"; | |
91 | } | |
92 | ||
93 | if ($help != 0) { | |
94 | usage(); | |
95 | exit 0; | |
96 | } | |
97 | ||
98 | if ($version != 0) { | |
99 | print("${P} ${V}\n"); | |
100 | exit 0; | |
101 | } | |
102 | ||
cb7301c7 JP |
103 | if ($#ARGV < 0) { |
104 | usage(); | |
105 | die "$P: argument missing: patchfile or -f file please\n"; | |
106 | } | |
107 | ||
108 | my $selections = $email + $scm + $status + $subsystem + $web; | |
109 | if ($selections == 0) { | |
110 | usage(); | |
111 | die "$P: Missing required option: email, scm, status, subsystem or web\n"; | |
112 | } | |
113 | ||
f5492666 JP |
114 | if ($email && |
115 | ($email_maintainer + $email_list + $email_subscriber_list + | |
116 | $email_git + $email_git_penguin_chiefs + $email_git_blame) == 0) { | |
cb7301c7 JP |
117 | usage(); |
118 | die "$P: Please select at least 1 email option\n"; | |
119 | } | |
120 | ||
121 | if (!top_of_kernel_tree($lk_path)) { | |
122 | die "$P: The current directory does not appear to be " | |
123 | . "a linux kernel source tree.\n"; | |
124 | } | |
125 | ||
126 | ## Read MAINTAINERS for type/value pairs | |
127 | ||
128 | my @typevalue = (); | |
129 | open(MAINT, "<${lk_path}MAINTAINERS") || die "$P: Can't open MAINTAINERS\n"; | |
130 | while (<MAINT>) { | |
131 | my $line = $_; | |
132 | ||
133 | if ($line =~ m/^(\C):\s*(.*)/) { | |
134 | my $type = $1; | |
135 | my $value = $2; | |
136 | ||
137 | ##Filename pattern matching | |
138 | if ($type eq "F" || $type eq "X") { | |
139 | $value =~ s@\.@\\\.@g; ##Convert . to \. | |
140 | $value =~ s/\*/\.\*/g; ##Convert * to .* | |
141 | $value =~ s/\?/\./g; ##Convert ? to . | |
870020f9 JP |
142 | ##if pattern is a directory and it lacks a trailing slash, add one |
143 | if ((-d $value)) { | |
144 | $value =~ s@([^/])$@$1/@; | |
145 | } | |
cb7301c7 JP |
146 | } |
147 | push(@typevalue, "$type:$value"); | |
148 | } elsif (!/^(\s)*$/) { | |
149 | $line =~ s/\n$//g; | |
150 | push(@typevalue, $line); | |
151 | } | |
152 | } | |
153 | close(MAINT); | |
154 | ||
4a7fdb5f | 155 | ## use the filenames on the command line or find the filenames in the patchfiles |
cb7301c7 JP |
156 | |
157 | my @files = (); | |
f5492666 | 158 | my @range = (); |
cb7301c7 | 159 | |
4a7fdb5f | 160 | foreach my $file (@ARGV) { |
870020f9 JP |
161 | ##if $file is a directory and it lacks a trailing slash, add one |
162 | if ((-d $file)) { | |
163 | $file =~ s@([^/])$@$1/@; | |
164 | } elsif (!(-f $file)) { | |
4a7fdb5f | 165 | die "$P: file '${file}' not found\n"; |
cb7301c7 | 166 | } |
4a7fdb5f JP |
167 | if ($from_filename) { |
168 | push(@files, $file); | |
169 | } else { | |
170 | my $file_cnt = @files; | |
f5492666 | 171 | my $lastfile; |
4a7fdb5f JP |
172 | open(PATCH, "<$file") or die "$P: Can't open ${file}\n"; |
173 | while (<PATCH>) { | |
174 | if (m/^\+\+\+\s+(\S+)/) { | |
175 | my $filename = $1; | |
176 | $filename =~ s@^[^/]*/@@; | |
177 | $filename =~ s@\n@@; | |
f5492666 | 178 | $lastfile = $filename; |
4a7fdb5f | 179 | push(@files, $filename); |
f5492666 JP |
180 | } elsif (m/^\@\@ -(\d+),(\d+)/) { |
181 | if ($email_git_blame) { | |
182 | push(@range, "$lastfile:$1:$2"); | |
183 | } | |
4a7fdb5f | 184 | } |
cb7301c7 | 185 | } |
4a7fdb5f JP |
186 | close(PATCH); |
187 | if ($file_cnt == @files) { | |
7f29fd27 | 188 | warn "$P: file '${file}' doesn't appear to be a patch. " |
4a7fdb5f JP |
189 | . "Add -f to options?\n"; |
190 | } | |
191 | @files = sort_and_uniq(@files); | |
cb7301c7 | 192 | } |
cb7301c7 JP |
193 | } |
194 | ||
195 | my @email_to = (); | |
290603c1 | 196 | my @list_to = (); |
cb7301c7 JP |
197 | my @scm = (); |
198 | my @web = (); | |
199 | my @subsystem = (); | |
200 | my @status = (); | |
201 | ||
202 | # Find responsible parties | |
203 | ||
204 | foreach my $file (@files) { | |
205 | ||
206 | #Do not match excluded file patterns | |
207 | ||
208 | my $exclude = 0; | |
209 | foreach my $line (@typevalue) { | |
290603c1 | 210 | if ($line =~ m/^(\C):\s*(.*)/) { |
cb7301c7 JP |
211 | my $type = $1; |
212 | my $value = $2; | |
213 | if ($type eq 'X') { | |
214 | if (file_match_pattern($file, $value)) { | |
215 | $exclude = 1; | |
1d606b4e | 216 | last; |
cb7301c7 JP |
217 | } |
218 | } | |
219 | } | |
220 | } | |
221 | ||
222 | if (!$exclude) { | |
223 | my $tvi = 0; | |
1d606b4e | 224 | my %hash; |
cb7301c7 | 225 | foreach my $line (@typevalue) { |
290603c1 | 226 | if ($line =~ m/^(\C):\s*(.*)/) { |
cb7301c7 JP |
227 | my $type = $1; |
228 | my $value = $2; | |
229 | if ($type eq 'F') { | |
230 | if (file_match_pattern($file, $value)) { | |
3fb55652 JP |
231 | my $value_pd = ($value =~ tr@/@@); |
232 | my $file_pd = ($file =~ tr@/@@); | |
233 | $value_pd++ if (substr($value,-1,1) ne "/"); | |
234 | if ($pattern_depth == 0 || | |
235 | (($file_pd - $value_pd) < $pattern_depth)) { | |
236 | $hash{$tvi} = $value_pd; | |
237 | } | |
cb7301c7 JP |
238 | } |
239 | } | |
240 | } | |
241 | $tvi++; | |
242 | } | |
1d606b4e JP |
243 | foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) { |
244 | add_categories($line); | |
245 | } | |
cb7301c7 JP |
246 | } |
247 | ||
4a7fdb5f | 248 | if ($email && $email_git) { |
cb7301c7 JP |
249 | recent_git_signoffs($file); |
250 | } | |
251 | ||
f5492666 JP |
252 | if ($email && $email_git_blame) { |
253 | git_assign_blame($file); | |
254 | } | |
cb7301c7 JP |
255 | } |
256 | ||
f5f5078d | 257 | if ($email) { |
cb7301c7 JP |
258 | foreach my $chief (@penguin_chief) { |
259 | if ($chief =~ m/^(.*):(.*)/) { | |
f5f5078d | 260 | my $email_address; |
0e70e83d JP |
261 | |
262 | $email_address = format_email($1, $2); | |
f5f5078d JP |
263 | if ($email_git_penguin_chiefs) { |
264 | push(@email_to, $email_address); | |
265 | } else { | |
266 | @email_to = grep(!/${email_address}/, @email_to); | |
cb7301c7 JP |
267 | } |
268 | } | |
269 | } | |
270 | } | |
271 | ||
290603c1 JP |
272 | if ($email || $email_list) { |
273 | my @to = (); | |
274 | if ($email) { | |
275 | @to = (@to, @email_to); | |
cb7301c7 | 276 | } |
290603c1 | 277 | if ($email_list) { |
290603c1 | 278 | @to = (@to, @list_to); |
290603c1 JP |
279 | } |
280 | output(uniq(@to)); | |
cb7301c7 JP |
281 | } |
282 | ||
283 | if ($scm) { | |
4a7fdb5f | 284 | @scm = sort_and_uniq(@scm); |
cb7301c7 JP |
285 | output(@scm); |
286 | } | |
287 | ||
288 | if ($status) { | |
4a7fdb5f | 289 | @status = sort_and_uniq(@status); |
cb7301c7 JP |
290 | output(@status); |
291 | } | |
292 | ||
293 | if ($subsystem) { | |
4a7fdb5f | 294 | @subsystem = sort_and_uniq(@subsystem); |
cb7301c7 JP |
295 | output(@subsystem); |
296 | } | |
297 | ||
298 | if ($web) { | |
4a7fdb5f | 299 | @web = sort_and_uniq(@web); |
cb7301c7 JP |
300 | output(@web); |
301 | } | |
302 | ||
303 | exit($exit); | |
304 | ||
305 | sub file_match_pattern { | |
306 | my ($file, $pattern) = @_; | |
307 | if (substr($pattern, -1) eq "/") { | |
308 | if ($file =~ m@^$pattern@) { | |
309 | return 1; | |
310 | } | |
311 | } else { | |
312 | if ($file =~ m@^$pattern@) { | |
313 | my $s1 = ($file =~ tr@/@@); | |
314 | my $s2 = ($pattern =~ tr@/@@); | |
315 | if ($s1 == $s2) { | |
316 | return 1; | |
317 | } | |
318 | } | |
319 | } | |
320 | return 0; | |
321 | } | |
322 | ||
323 | sub usage { | |
324 | print <<EOT; | |
325 | usage: $P [options] patchfile | |
870020f9 | 326 | $P [options] -f file|directory |
cb7301c7 JP |
327 | version: $V |
328 | ||
329 | MAINTAINER field selection options: | |
330 | --email => print email address(es) if any | |
331 | --git => include recent git \*-by: signers | |
332 | --git-chief-penguins => include ${penguin_chiefs} | |
333 | --git-min-signatures => number of signatures required (default: 1) | |
334 | --git-max-maintainers => maximum maintainers to add (default: 5) | |
3d202aeb | 335 | --git-min-percent => minimum percentage of commits required (default: 5) |
cb7301c7 | 336 | --git-since => git history to use (default: 1-year-ago) |
f5492666 | 337 | --git-blame => use git blame to find modified commits for patch or file |
cb7301c7 JP |
338 | --m => include maintainer(s) if any |
339 | --n => include name 'Full Name <addr\@domain.tld>' | |
340 | --l => include list(s) if any | |
341 | --s => include subscriber only list(s) if any | |
342 | --scm => print SCM tree(s) if any | |
343 | --status => print status if any | |
344 | --subsystem => print subsystem name if any | |
345 | --web => print website(s) if any | |
346 | ||
347 | Output type options: | |
348 | --separator [, ] => separator for multiple entries on 1 line | |
349 | --multiline => print 1 entry per line | |
350 | ||
cb7301c7 | 351 | Other options: |
3fb55652 | 352 | --pattern-depth => Number of pattern directory traversals (default: 0 (all)) |
f5f5078d | 353 | --version => show version |
cb7301c7 JP |
354 | --help => show this help information |
355 | ||
3fb55652 JP |
356 | Default options: |
357 | [--email --git --m --n --l --multiline --pattern-depth=0] | |
358 | ||
870020f9 JP |
359 | Notes: |
360 | Using "-f directory" may give unexpected results: | |
f5492666 JP |
361 | Used with "--git", git signators for _all_ files in and below |
362 | directory are examined as git recurses directories. | |
363 | Any specified X: (exclude) pattern matches are _not_ ignored. | |
364 | Used with "--nogit", directory is used as a pattern match, | |
365 | no individual file within the directory or subdirectory | |
366 | is matched. | |
367 | Used with "--git-blame", does not iterate all files in directory | |
368 | Using "--git-blame" is slow and may add old committers and authors | |
369 | that are no longer active maintainers to the output. | |
cb7301c7 JP |
370 | EOT |
371 | } | |
372 | ||
373 | sub top_of_kernel_tree { | |
374 | my ($lk_path) = @_; | |
375 | ||
376 | if ($lk_path ne "" && substr($lk_path,length($lk_path)-1,1) ne "/") { | |
377 | $lk_path .= "/"; | |
378 | } | |
379 | if ( (-f "${lk_path}COPYING") | |
380 | && (-f "${lk_path}CREDITS") | |
381 | && (-f "${lk_path}Kbuild") | |
382 | && (-f "${lk_path}MAINTAINERS") | |
383 | && (-f "${lk_path}Makefile") | |
384 | && (-f "${lk_path}README") | |
385 | && (-d "${lk_path}Documentation") | |
386 | && (-d "${lk_path}arch") | |
387 | && (-d "${lk_path}include") | |
388 | && (-d "${lk_path}drivers") | |
389 | && (-d "${lk_path}fs") | |
390 | && (-d "${lk_path}init") | |
391 | && (-d "${lk_path}ipc") | |
392 | && (-d "${lk_path}kernel") | |
393 | && (-d "${lk_path}lib") | |
394 | && (-d "${lk_path}scripts")) { | |
395 | return 1; | |
396 | } | |
397 | return 0; | |
398 | } | |
399 | ||
0e70e83d JP |
400 | sub parse_email { |
401 | my ($formatted_email) = @_; | |
402 | ||
403 | my $name = ""; | |
404 | my $address = ""; | |
405 | ||
406 | if ($formatted_email =~ /^([^<]+)<(.*\@.*)>$/) { | |
407 | $name = $1; | |
408 | $address = $2; | |
409 | } elsif ($formatted_email =~ /^<(.*\@.*)>$/) { | |
410 | $address = $1; | |
411 | } elsif ($formatted_email =~ /^(.*\@.*)$/) { | |
412 | $address = $1; | |
413 | } | |
cb7301c7 JP |
414 | |
415 | $name =~ s/^\s+|\s+$//g; | |
d789504a | 416 | $name =~ s/^\"|\"$//g; |
0e70e83d | 417 | $address =~ s/^\s+|\s+$//g; |
cb7301c7 | 418 | |
0e70e83d JP |
419 | if ($name =~ /[^a-z0-9 \.\-]/i) { ##has "must quote" chars |
420 | $name =~ s/(?<!\\)"/\\"/g; ##escape quotes | |
421 | $name = "\"$name\""; | |
422 | } | |
423 | ||
424 | return ($name, $address); | |
425 | } | |
426 | ||
427 | sub format_email { | |
428 | my ($name, $address) = @_; | |
429 | ||
430 | my $formatted_email; | |
431 | ||
432 | $name =~ s/^\s+|\s+$//g; | |
433 | $name =~ s/^\"|\"$//g; | |
434 | $address =~ s/^\s+|\s+$//g; | |
cb7301c7 JP |
435 | |
436 | if ($name =~ /[^a-z0-9 \.\-]/i) { ##has "must quote" chars | |
437 | $name =~ s/(?<!\\)"/\\"/g; ##escape quotes | |
0e70e83d JP |
438 | $name = "\"$name\""; |
439 | } | |
440 | ||
441 | if ($email_usename) { | |
442 | if ("$name" eq "") { | |
443 | $formatted_email = "$address"; | |
444 | } else { | |
445 | $formatted_email = "$name <${address}>"; | |
446 | } | |
cb7301c7 | 447 | } else { |
0e70e83d | 448 | $formatted_email = $address; |
cb7301c7 | 449 | } |
0e70e83d | 450 | |
cb7301c7 JP |
451 | return $formatted_email; |
452 | } | |
453 | ||
454 | sub add_categories { | |
455 | my ($index) = @_; | |
456 | ||
457 | $index = $index - 1; | |
458 | while ($index >= 0) { | |
459 | my $tv = $typevalue[$index]; | |
290603c1 | 460 | if ($tv =~ m/^(\C):\s*(.*)/) { |
cb7301c7 JP |
461 | my $ptype = $1; |
462 | my $pvalue = $2; | |
463 | if ($ptype eq "L") { | |
290603c1 JP |
464 | my $list_address = $pvalue; |
465 | my $list_additional = ""; | |
466 | if ($list_address =~ m/([^\s]+)\s+(.*)$/) { | |
467 | $list_address = $1; | |
468 | $list_additional = $2; | |
469 | } | |
bdf7c685 | 470 | if ($list_additional =~ m/subscribers-only/) { |
cb7301c7 | 471 | if ($email_subscriber_list) { |
290603c1 | 472 | push(@list_to, $list_address); |
cb7301c7 JP |
473 | } |
474 | } else { | |
475 | if ($email_list) { | |
290603c1 | 476 | push(@list_to, $list_address); |
cb7301c7 JP |
477 | } |
478 | } | |
479 | } elsif ($ptype eq "M") { | |
0e70e83d JP |
480 | my ($name, $address) = parse_email($pvalue); |
481 | if ($name eq "") { | |
482 | if ($index >= 0) { | |
483 | my $tv = $typevalue[$index - 1]; | |
484 | if ($tv =~ m/^(\C):\s*(.*)/) { | |
485 | if ($1 eq "P") { | |
486 | $name = $2; | |
5f2441e9 JP |
487 | } |
488 | } | |
489 | } | |
490 | } | |
0e70e83d | 491 | if ($email_maintainer) { |
1b5e1cf6 | 492 | push_email_addresses($pvalue); |
cb7301c7 JP |
493 | } |
494 | } elsif ($ptype eq "T") { | |
495 | push(@scm, $pvalue); | |
496 | } elsif ($ptype eq "W") { | |
497 | push(@web, $pvalue); | |
498 | } elsif ($ptype eq "S") { | |
499 | push(@status, $pvalue); | |
500 | } | |
501 | ||
502 | $index--; | |
503 | } else { | |
504 | push(@subsystem,$tv); | |
505 | $index = -1; | |
506 | } | |
507 | } | |
508 | } | |
509 | ||
0e70e83d JP |
510 | sub email_address_inuse { |
511 | my ($test_address) = @_; | |
512 | ||
513 | foreach my $line (@email_to) { | |
514 | my ($name, $address) = parse_email($line); | |
515 | ||
516 | return 1 if ($address eq $test_address); | |
517 | } | |
518 | return 0; | |
519 | } | |
520 | ||
1b5e1cf6 | 521 | sub push_email_address { |
0e70e83d | 522 | my ($line) = @_; |
1b5e1cf6 | 523 | |
0e70e83d | 524 | my ($name, $address) = parse_email($line); |
1b5e1cf6 | 525 | |
0e70e83d JP |
526 | if (!email_address_inuse($address)) { |
527 | push(@email_to, format_email($name, $address)); | |
1b5e1cf6 JP |
528 | } |
529 | } | |
530 | ||
531 | sub push_email_addresses { | |
532 | my ($address) = @_; | |
533 | ||
534 | my @address_list = (); | |
535 | ||
5f2441e9 JP |
536 | if (rfc822_valid($address)) { |
537 | push_email_address($address); | |
538 | } elsif (@address_list = rfc822_validlist($address)) { | |
1b5e1cf6 JP |
539 | my $array_count = shift(@address_list); |
540 | while (my $entry = shift(@address_list)) { | |
541 | push_email_address($entry); | |
542 | } | |
5f2441e9 JP |
543 | } else { |
544 | warn("Invalid MAINTAINERS address: '" . $address . "'\n"); | |
1b5e1cf6 | 545 | } |
1b5e1cf6 JP |
546 | } |
547 | ||
cb7301c7 JP |
548 | sub which { |
549 | my ($bin) = @_; | |
550 | ||
f5f5078d | 551 | foreach my $path (split(/:/, $ENV{PATH})) { |
cb7301c7 JP |
552 | if (-e "$path/$bin") { |
553 | return "$path/$bin"; | |
554 | } | |
555 | } | |
556 | ||
557 | return ""; | |
558 | } | |
559 | ||
560 | sub recent_git_signoffs { | |
561 | my ($file) = @_; | |
562 | ||
563 | my $sign_offs = ""; | |
564 | my $cmd = ""; | |
565 | my $output = ""; | |
566 | my $count = 0; | |
567 | my @lines = (); | |
0e70e83d | 568 | my %hash; |
afa81ee1 | 569 | my $total_sign_offs; |
cb7301c7 JP |
570 | |
571 | if (which("git") eq "") { | |
de2fc492 JP |
572 | warn("$P: git not found. Add --nogit to options?\n"); |
573 | return; | |
574 | } | |
575 | if (!(-d ".git")) { | |
5f2441e9 JP |
576 | warn("$P: .git directory not found. Use a git repository for better results.\n"); |
577 | warn("$P: perhaps 'git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git'\n"); | |
de2fc492 | 578 | return; |
cb7301c7 JP |
579 | } |
580 | ||
581 | $cmd = "git log --since=${email_git_since} -- ${file}"; | |
cb7301c7 JP |
582 | |
583 | $output = `${cmd}`; | |
584 | $output =~ s/^\s*//gm; | |
585 | ||
586 | @lines = split("\n", $output); | |
afa81ee1 | 587 | |
0e70e83d JP |
588 | @lines = grep(/^[-_ a-z]+by:.*\@.*$/i, @lines); |
589 | if (!$email_git_penguin_chiefs) { | |
590 | @lines = grep(!/${penguin_chiefs}/i, @lines); | |
591 | } | |
592 | # cut -f2- -d":" | |
593 | s/.*:\s*(.+)\s*/$1/ for (@lines); | |
594 | ||
595 | @lines = mailmap(@lines); | |
596 | ||
597 | $total_sign_offs = @lines; | |
598 | @lines = sort(@lines); | |
599 | # uniq -c | |
afa81ee1 | 600 | foreach my $line (@lines) { |
0e70e83d JP |
601 | $hash{$line}++; |
602 | } | |
603 | # sort -rn | |
604 | @lines = (); | |
605 | foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) { | |
606 | push(@lines,"$hash{$line} $line"); | |
afa81ee1 JP |
607 | } |
608 | ||
cb7301c7 | 609 | foreach my $line (@lines) { |
4a7fdb5f | 610 | if ($line =~ m/([0-9]+)\s+(.*)/) { |
cb7301c7 | 611 | my $sign_offs = $1; |
4a7fdb5f | 612 | $line = $2; |
cb7301c7 JP |
613 | $count++; |
614 | if ($sign_offs < $email_git_min_signatures || | |
afa81ee1 JP |
615 | $count > $email_git_max_maintainers || |
616 | $sign_offs * 100 / $total_sign_offs < $email_git_min_percent) { | |
cb7301c7 JP |
617 | last; |
618 | } | |
0e70e83d | 619 | push_email_address($line); |
cb7301c7 | 620 | } |
f5492666 JP |
621 | } |
622 | } | |
623 | ||
624 | sub save_commits { | |
625 | my ($cmd, @commits) = @_; | |
626 | my $output; | |
627 | my @lines = (); | |
628 | ||
629 | $output = `${cmd}`; | |
630 | ||
631 | @lines = split("\n", $output); | |
632 | foreach my $line (@lines) { | |
633 | if ($line =~ m/^(\w+) /) { | |
634 | push (@commits, $1); | |
635 | } | |
636 | } | |
637 | return @commits; | |
638 | } | |
639 | ||
640 | sub git_assign_blame { | |
641 | my ($file) = @_; | |
642 | ||
643 | my @lines = (); | |
644 | my @commits = (); | |
645 | my $cmd; | |
646 | my $output; | |
647 | my %hash; | |
648 | my $total_sign_offs; | |
649 | my $count; | |
650 | ||
651 | if (@range) { | |
652 | foreach my $file_range_diff (@range) { | |
653 | next if (!($file_range_diff =~ m/(.+):(.+):(.+)/)); | |
654 | my $diff_file = $1; | |
655 | my $diff_start = $2; | |
656 | my $diff_length = $3; | |
657 | next if (!("$file" eq "$diff_file")); | |
658 | $cmd = "git blame -l -L $diff_start,+$diff_length $file\n"; | |
659 | @commits = save_commits($cmd, @commits); | |
cb7301c7 | 660 | } |
f5492666 JP |
661 | } else { |
662 | if (-f $file) { | |
663 | $cmd = "git blame -l $file\n"; | |
664 | @commits = save_commits($cmd, @commits); | |
665 | } | |
666 | } | |
667 | ||
668 | $total_sign_offs = 0; | |
669 | @commits = uniq(@commits); | |
670 | foreach my $commit (@commits) { | |
671 | $cmd = "git log -1 ${commit}"; | |
f5492666 JP |
672 | |
673 | $output = `${cmd}`; | |
674 | $output =~ s/^\s*//gm; | |
675 | @lines = split("\n", $output); | |
0e70e83d JP |
676 | |
677 | @lines = grep(/^[-_ a-z]+by:.*\@.*$/i, @lines); | |
678 | if (!$email_git_penguin_chiefs) { | |
679 | @lines = grep(!/${penguin_chiefs}/i, @lines); | |
680 | } | |
681 | # cut -f2- -d":" | |
682 | s/.*:\s*(.+)\s*/$1/ for (@lines); | |
683 | ||
f5492666 JP |
684 | $hash{$_}++ for @lines; |
685 | $total_sign_offs += @lines; | |
686 | } | |
687 | ||
688 | $count = 0; | |
689 | foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) { | |
690 | my $sign_offs = $hash{$line}; | |
691 | $count++; | |
692 | last if ($sign_offs < $email_git_min_signatures || | |
693 | $count > $email_git_max_maintainers || | |
694 | $sign_offs * 100 / $total_sign_offs < $email_git_min_percent); | |
695 | push_email_address($line); | |
cb7301c7 | 696 | } |
cb7301c7 JP |
697 | } |
698 | ||
699 | sub uniq { | |
700 | my @parms = @_; | |
701 | ||
702 | my %saw; | |
703 | @parms = grep(!$saw{$_}++, @parms); | |
704 | return @parms; | |
705 | } | |
706 | ||
707 | sub sort_and_uniq { | |
708 | my @parms = @_; | |
709 | ||
710 | my %saw; | |
711 | @parms = sort @parms; | |
712 | @parms = grep(!$saw{$_}++, @parms); | |
713 | return @parms; | |
714 | } | |
715 | ||
716 | sub output { | |
717 | my @parms = @_; | |
718 | ||
719 | if ($output_multiline) { | |
720 | foreach my $line (@parms) { | |
721 | print("${line}\n"); | |
722 | } | |
723 | } else { | |
724 | print(join($output_separator, @parms)); | |
725 | print("\n"); | |
726 | } | |
727 | } | |
1b5e1cf6 JP |
728 | |
729 | my $rfc822re; | |
730 | ||
731 | sub make_rfc822re { | |
732 | # Basic lexical tokens are specials, domain_literal, quoted_string, atom, and | |
733 | # comment. We must allow for rfc822_lwsp (or comments) after each of these. | |
734 | # This regexp will only work on addresses which have had comments stripped | |
735 | # and replaced with rfc822_lwsp. | |
736 | ||
737 | my $specials = '()<>@,;:\\\\".\\[\\]'; | |
738 | my $controls = '\\000-\\037\\177'; | |
739 | ||
740 | my $dtext = "[^\\[\\]\\r\\\\]"; | |
741 | my $domain_literal = "\\[(?:$dtext|\\\\.)*\\]$rfc822_lwsp*"; | |
742 | ||
743 | my $quoted_string = "\"(?:[^\\\"\\r\\\\]|\\\\.|$rfc822_lwsp)*\"$rfc822_lwsp*"; | |
744 | ||
745 | # Use zero-width assertion to spot the limit of an atom. A simple | |
746 | # $rfc822_lwsp* causes the regexp engine to hang occasionally. | |
747 | my $atom = "[^$specials $controls]+(?:$rfc822_lwsp+|\\Z|(?=[\\[\"$specials]))"; | |
748 | my $word = "(?:$atom|$quoted_string)"; | |
749 | my $localpart = "$word(?:\\.$rfc822_lwsp*$word)*"; | |
750 | ||
751 | my $sub_domain = "(?:$atom|$domain_literal)"; | |
752 | my $domain = "$sub_domain(?:\\.$rfc822_lwsp*$sub_domain)*"; | |
753 | ||
754 | my $addr_spec = "$localpart\@$rfc822_lwsp*$domain"; | |
755 | ||
756 | my $phrase = "$word*"; | |
757 | my $route = "(?:\@$domain(?:,\@$rfc822_lwsp*$domain)*:$rfc822_lwsp*)"; | |
758 | my $route_addr = "\\<$rfc822_lwsp*$route?$addr_spec\\>$rfc822_lwsp*"; | |
759 | my $mailbox = "(?:$addr_spec|$phrase$route_addr)"; | |
760 | ||
761 | my $group = "$phrase:$rfc822_lwsp*(?:$mailbox(?:,\\s*$mailbox)*)?;\\s*"; | |
762 | my $address = "(?:$mailbox|$group)"; | |
763 | ||
764 | return "$rfc822_lwsp*$address"; | |
765 | } | |
766 | ||
767 | sub rfc822_strip_comments { | |
768 | my $s = shift; | |
769 | # Recursively remove comments, and replace with a single space. The simpler | |
770 | # regexps in the Email Addressing FAQ are imperfect - they will miss escaped | |
771 | # chars in atoms, for example. | |
772 | ||
773 | while ($s =~ s/^((?:[^"\\]|\\.)* | |
774 | (?:"(?:[^"\\]|\\.)*"(?:[^"\\]|\\.)*)*) | |
775 | \((?:[^()\\]|\\.)*\)/$1 /osx) {} | |
776 | return $s; | |
777 | } | |
778 | ||
779 | # valid: returns true if the parameter is an RFC822 valid address | |
780 | # | |
781 | sub rfc822_valid ($) { | |
782 | my $s = rfc822_strip_comments(shift); | |
783 | ||
784 | if (!$rfc822re) { | |
785 | $rfc822re = make_rfc822re(); | |
786 | } | |
787 | ||
788 | return $s =~ m/^$rfc822re$/so && $s =~ m/^$rfc822_char*$/; | |
789 | } | |
790 | ||
791 | # validlist: In scalar context, returns true if the parameter is an RFC822 | |
792 | # valid list of addresses. | |
793 | # | |
794 | # In list context, returns an empty list on failure (an invalid | |
795 | # address was found); otherwise a list whose first element is the | |
796 | # number of addresses found and whose remaining elements are the | |
797 | # addresses. This is needed to disambiguate failure (invalid) | |
798 | # from success with no addresses found, because an empty string is | |
799 | # a valid list. | |
800 | ||
801 | sub rfc822_validlist ($) { | |
802 | my $s = rfc822_strip_comments(shift); | |
803 | ||
804 | if (!$rfc822re) { | |
805 | $rfc822re = make_rfc822re(); | |
806 | } | |
807 | # * null list items are valid according to the RFC | |
808 | # * the '1' business is to aid in distinguishing failure from no results | |
809 | ||
810 | my @r; | |
811 | if ($s =~ m/^(?:$rfc822re)?(?:,(?:$rfc822re)?)*$/so && | |
812 | $s =~ m/^$rfc822_char*$/) { | |
5f2441e9 | 813 | while ($s =~ m/(?:^|,$rfc822_lwsp*)($rfc822re)/gos) { |
1b5e1cf6 JP |
814 | push @r, $1; |
815 | } | |
816 | return wantarray ? (scalar(@r), @r) : 1; | |
817 | } | |
818 | else { | |
819 | return wantarray ? () : 0; | |
820 | } | |
821 | } |