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; | |
60db31ac | 16 | my $V = '0.23'; |
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; |
60db31ac JP |
27 | my $email_git = 1; |
28 | my $email_git_blame = 0; | |
cb7301c7 JP |
29 | my $email_git_min_signatures = 1; |
30 | my $email_git_max_maintainers = 5; | |
afa81ee1 | 31 | my $email_git_min_percent = 5; |
cb7301c7 | 32 | my $email_git_since = "1-year-ago"; |
60db31ac | 33 | my $email_hg_since = "-365"; |
11ecf53c | 34 | my $email_remove_duplicates = 1; |
cb7301c7 JP |
35 | my $output_multiline = 1; |
36 | my $output_separator = ", "; | |
3c7385b8 JP |
37 | my $output_roles = 0; |
38 | my $output_rolestats = 0; | |
cb7301c7 JP |
39 | my $scm = 0; |
40 | my $web = 0; | |
41 | my $subsystem = 0; | |
42 | my $status = 0; | |
dcf36a92 | 43 | my $keywords = 1; |
03372dbb | 44 | my $file_emails = 0; |
4a7fdb5f | 45 | my $from_filename = 0; |
3fb55652 | 46 | my $pattern_depth = 0; |
cb7301c7 JP |
47 | my $version = 0; |
48 | my $help = 0; | |
49 | ||
50 | my $exit = 0; | |
51 | ||
52 | my @penguin_chief = (); | |
53 | push(@penguin_chief,"Linus Torvalds:torvalds\@linux-foundation.org"); | |
54 | #Andrew wants in on most everything - 2009/01/14 | |
55 | #push(@penguin_chief,"Andrew Morton:akpm\@linux-foundation.org"); | |
56 | ||
57 | my @penguin_chief_names = (); | |
58 | foreach my $chief (@penguin_chief) { | |
59 | if ($chief =~ m/^(.*):(.*)/) { | |
60 | my $chief_name = $1; | |
61 | my $chief_addr = $2; | |
62 | push(@penguin_chief_names, $chief_name); | |
63 | } | |
64 | } | |
65 | my $penguin_chiefs = "\(" . join("|",@penguin_chief_names) . "\)"; | |
66 | ||
5f2441e9 | 67 | # rfc822 email address - preloaded methods go here. |
1b5e1cf6 | 68 | my $rfc822_lwsp = "(?:(?:\\r\\n)?[ \\t])"; |
df4cc036 | 69 | my $rfc822_char = '[\\000-\\377]'; |
1b5e1cf6 | 70 | |
60db31ac JP |
71 | # VCS command support: class-like functions and strings |
72 | ||
73 | my %VCS_cmds; | |
74 | ||
75 | my %VCS_cmds_git = ( | |
76 | "execute_cmd" => \&git_execute_cmd, | |
77 | "available" => '(which("git") ne "") && (-d ".git")', | |
99cf6116 RK |
78 | "find_signers_cmd" => "git log --no-color --since=\$email_git_since -- \$file", |
79 | "find_commit_signers_cmd" => "git log --no-color -1 \$commit", | |
60db31ac JP |
80 | "blame_range_cmd" => "git blame -l -L \$diff_start,+\$diff_length \$file", |
81 | "blame_file_cmd" => "git blame -l \$file", | |
82 | "commit_pattern" => "^commit [0-9a-f]{40,40}", | |
83 | "blame_commit_pattern" => "^([0-9a-f]+) " | |
84 | ); | |
85 | ||
86 | my %VCS_cmds_hg = ( | |
87 | "execute_cmd" => \&hg_execute_cmd, | |
88 | "available" => '(which("hg") ne "") && (-d ".hg")', | |
89 | "find_signers_cmd" => | |
90 | "hg log --date=\$email_hg_since" . | |
91 | " --template='commit {node}\\n{desc}\\n' -- \$file", | |
92 | "find_commit_signers_cmd" => "hg log --template='{desc}\\n' -r \$commit", | |
93 | "blame_range_cmd" => "", # not supported | |
94 | "blame_file_cmd" => "hg blame -c \$file", | |
95 | "commit_pattern" => "^commit [0-9a-f]{40,40}", | |
96 | "blame_commit_pattern" => "^([0-9a-f]+):" | |
97 | ); | |
98 | ||
cb7301c7 JP |
99 | if (!GetOptions( |
100 | 'email!' => \$email, | |
101 | 'git!' => \$email_git, | |
60db31ac | 102 | 'git-blame!' => \$email_git_blame, |
cb7301c7 JP |
103 | 'git-chief-penguins!' => \$email_git_penguin_chiefs, |
104 | 'git-min-signatures=i' => \$email_git_min_signatures, | |
105 | 'git-max-maintainers=i' => \$email_git_max_maintainers, | |
afa81ee1 | 106 | 'git-min-percent=i' => \$email_git_min_percent, |
cb7301c7 | 107 | 'git-since=s' => \$email_git_since, |
60db31ac | 108 | 'hg-since=s' => \$email_hg_since, |
11ecf53c | 109 | 'remove-duplicates!' => \$email_remove_duplicates, |
cb7301c7 JP |
110 | 'm!' => \$email_maintainer, |
111 | 'n!' => \$email_usename, | |
112 | 'l!' => \$email_list, | |
113 | 's!' => \$email_subscriber_list, | |
114 | 'multiline!' => \$output_multiline, | |
3c7385b8 JP |
115 | 'roles!' => \$output_roles, |
116 | 'rolestats!' => \$output_rolestats, | |
cb7301c7 JP |
117 | 'separator=s' => \$output_separator, |
118 | 'subsystem!' => \$subsystem, | |
119 | 'status!' => \$status, | |
120 | 'scm!' => \$scm, | |
121 | 'web!' => \$web, | |
3fb55652 | 122 | 'pattern-depth=i' => \$pattern_depth, |
dcf36a92 | 123 | 'k|keywords!' => \$keywords, |
03372dbb | 124 | 'fe|file-emails!' => \$file_emails, |
4a7fdb5f | 125 | 'f|file' => \$from_filename, |
cb7301c7 JP |
126 | 'v|version' => \$version, |
127 | 'h|help' => \$help, | |
128 | )) { | |
3c7385b8 | 129 | die "$P: invalid argument - use --help if necessary\n"; |
cb7301c7 JP |
130 | } |
131 | ||
132 | if ($help != 0) { | |
133 | usage(); | |
134 | exit 0; | |
135 | } | |
136 | ||
137 | if ($version != 0) { | |
138 | print("${P} ${V}\n"); | |
139 | exit 0; | |
140 | } | |
141 | ||
cb7301c7 JP |
142 | if ($#ARGV < 0) { |
143 | usage(); | |
144 | die "$P: argument missing: patchfile or -f file please\n"; | |
145 | } | |
146 | ||
42498316 JP |
147 | if ($output_separator ne ", ") { |
148 | $output_multiline = 0; | |
149 | } | |
150 | ||
3c7385b8 JP |
151 | if ($output_rolestats) { |
152 | $output_roles = 1; | |
153 | } | |
154 | ||
cb7301c7 JP |
155 | my $selections = $email + $scm + $status + $subsystem + $web; |
156 | if ($selections == 0) { | |
157 | usage(); | |
158 | die "$P: Missing required option: email, scm, status, subsystem or web\n"; | |
159 | } | |
160 | ||
f5492666 JP |
161 | if ($email && |
162 | ($email_maintainer + $email_list + $email_subscriber_list + | |
163 | $email_git + $email_git_penguin_chiefs + $email_git_blame) == 0) { | |
cb7301c7 JP |
164 | usage(); |
165 | die "$P: Please select at least 1 email option\n"; | |
166 | } | |
167 | ||
168 | if (!top_of_kernel_tree($lk_path)) { | |
169 | die "$P: The current directory does not appear to be " | |
170 | . "a linux kernel source tree.\n"; | |
171 | } | |
172 | ||
173 | ## Read MAINTAINERS for type/value pairs | |
174 | ||
175 | my @typevalue = (); | |
dcf36a92 JP |
176 | my %keyword_hash; |
177 | ||
cb7301c7 JP |
178 | open(MAINT, "<${lk_path}MAINTAINERS") || die "$P: Can't open MAINTAINERS\n"; |
179 | while (<MAINT>) { | |
180 | my $line = $_; | |
181 | ||
182 | if ($line =~ m/^(\C):\s*(.*)/) { | |
183 | my $type = $1; | |
184 | my $value = $2; | |
185 | ||
186 | ##Filename pattern matching | |
187 | if ($type eq "F" || $type eq "X") { | |
188 | $value =~ s@\.@\\\.@g; ##Convert . to \. | |
189 | $value =~ s/\*/\.\*/g; ##Convert * to .* | |
190 | $value =~ s/\?/\./g; ##Convert ? to . | |
870020f9 JP |
191 | ##if pattern is a directory and it lacks a trailing slash, add one |
192 | if ((-d $value)) { | |
193 | $value =~ s@([^/])$@$1/@; | |
194 | } | |
dcf36a92 JP |
195 | } elsif ($type eq "K") { |
196 | $keyword_hash{@typevalue} = $value; | |
cb7301c7 JP |
197 | } |
198 | push(@typevalue, "$type:$value"); | |
199 | } elsif (!/^(\s)*$/) { | |
200 | $line =~ s/\n$//g; | |
201 | push(@typevalue, $line); | |
202 | } | |
203 | } | |
204 | close(MAINT); | |
205 | ||
8cbb3a77 JP |
206 | my %mailmap; |
207 | ||
11ecf53c JP |
208 | if ($email_remove_duplicates) { |
209 | open(MAILMAP, "<${lk_path}.mailmap") || warn "$P: Can't open .mailmap\n"; | |
210 | while (<MAILMAP>) { | |
211 | my $line = $_; | |
8cbb3a77 | 212 | |
11ecf53c JP |
213 | next if ($line =~ m/^\s*#/); |
214 | next if ($line =~ m/^\s*$/); | |
8cbb3a77 | 215 | |
11ecf53c | 216 | my ($name, $address) = parse_email($line); |
a8af2430 | 217 | $line = format_email($name, $address, $email_usename); |
8cbb3a77 | 218 | |
11ecf53c | 219 | next if ($line =~ m/^\s*$/); |
8cbb3a77 | 220 | |
11ecf53c JP |
221 | if (exists($mailmap{$name})) { |
222 | my $obj = $mailmap{$name}; | |
223 | push(@$obj, $address); | |
224 | } else { | |
225 | my @arr = ($address); | |
226 | $mailmap{$name} = \@arr; | |
227 | } | |
8cbb3a77 | 228 | } |
11ecf53c | 229 | close(MAILMAP); |
8cbb3a77 JP |
230 | } |
231 | ||
4a7fdb5f | 232 | ## use the filenames on the command line or find the filenames in the patchfiles |
cb7301c7 JP |
233 | |
234 | my @files = (); | |
f5492666 | 235 | my @range = (); |
dcf36a92 | 236 | my @keyword_tvi = (); |
03372dbb | 237 | my @file_emails = (); |
cb7301c7 | 238 | |
4a7fdb5f | 239 | foreach my $file (@ARGV) { |
870020f9 JP |
240 | ##if $file is a directory and it lacks a trailing slash, add one |
241 | if ((-d $file)) { | |
242 | $file =~ s@([^/])$@$1/@; | |
243 | } elsif (!(-f $file)) { | |
4a7fdb5f | 244 | die "$P: file '${file}' not found\n"; |
cb7301c7 | 245 | } |
4a7fdb5f JP |
246 | if ($from_filename) { |
247 | push(@files, $file); | |
03372dbb | 248 | if (-f $file && ($keywords || $file_emails)) { |
dcf36a92 | 249 | open(FILE, "<$file") or die "$P: Can't open ${file}\n"; |
a8af2430 | 250 | my $text = do { local($/) ; <FILE> }; |
03372dbb JP |
251 | close(FILE); |
252 | if ($keywords) { | |
253 | foreach my $line (keys %keyword_hash) { | |
254 | if ($text =~ m/$keyword_hash{$line}/x) { | |
255 | push(@keyword_tvi, $line); | |
256 | } | |
dcf36a92 JP |
257 | } |
258 | } | |
03372dbb JP |
259 | if ($file_emails) { |
260 | 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; | |
261 | push(@file_emails, clean_file_emails(@poss_addr)); | |
262 | } | |
dcf36a92 | 263 | } |
4a7fdb5f JP |
264 | } else { |
265 | my $file_cnt = @files; | |
f5492666 | 266 | my $lastfile; |
4a7fdb5f JP |
267 | open(PATCH, "<$file") or die "$P: Can't open ${file}\n"; |
268 | while (<PATCH>) { | |
dcf36a92 | 269 | my $patch_line = $_; |
4a7fdb5f JP |
270 | if (m/^\+\+\+\s+(\S+)/) { |
271 | my $filename = $1; | |
272 | $filename =~ s@^[^/]*/@@; | |
273 | $filename =~ s@\n@@; | |
f5492666 | 274 | $lastfile = $filename; |
4a7fdb5f | 275 | push(@files, $filename); |
f5492666 JP |
276 | } elsif (m/^\@\@ -(\d+),(\d+)/) { |
277 | if ($email_git_blame) { | |
278 | push(@range, "$lastfile:$1:$2"); | |
279 | } | |
dcf36a92 JP |
280 | } elsif ($keywords) { |
281 | foreach my $line (keys %keyword_hash) { | |
282 | if ($patch_line =~ m/^[+-].*$keyword_hash{$line}/x) { | |
283 | push(@keyword_tvi, $line); | |
284 | } | |
285 | } | |
4a7fdb5f | 286 | } |
cb7301c7 | 287 | } |
4a7fdb5f JP |
288 | close(PATCH); |
289 | if ($file_cnt == @files) { | |
7f29fd27 | 290 | warn "$P: file '${file}' doesn't appear to be a patch. " |
4a7fdb5f JP |
291 | . "Add -f to options?\n"; |
292 | } | |
293 | @files = sort_and_uniq(@files); | |
cb7301c7 | 294 | } |
cb7301c7 JP |
295 | } |
296 | ||
03372dbb JP |
297 | @file_emails = uniq(@file_emails); |
298 | ||
cb7301c7 | 299 | my @email_to = (); |
290603c1 | 300 | my @list_to = (); |
cb7301c7 JP |
301 | my @scm = (); |
302 | my @web = (); | |
303 | my @subsystem = (); | |
304 | my @status = (); | |
305 | ||
306 | # Find responsible parties | |
307 | ||
308 | foreach my $file (@files) { | |
309 | ||
272a8979 JP |
310 | my %hash; |
311 | my $tvi = find_first_section(); | |
312 | while ($tvi < @typevalue) { | |
313 | my $start = find_starting_index($tvi); | |
314 | my $end = find_ending_index($tvi); | |
315 | my $exclude = 0; | |
316 | my $i; | |
317 | ||
318 | #Do not match excluded file patterns | |
319 | ||
320 | for ($i = $start; $i < $end; $i++) { | |
321 | my $line = $typevalue[$i]; | |
290603c1 | 322 | if ($line =~ m/^(\C):\s*(.*)/) { |
cb7301c7 JP |
323 | my $type = $1; |
324 | my $value = $2; | |
272a8979 | 325 | if ($type eq 'X') { |
cb7301c7 | 326 | if (file_match_pattern($file, $value)) { |
272a8979 | 327 | $exclude = 1; |
cb7301c7 JP |
328 | } |
329 | } | |
330 | } | |
cb7301c7 | 331 | } |
272a8979 JP |
332 | |
333 | if (!$exclude) { | |
334 | for ($i = $start; $i < $end; $i++) { | |
335 | my $line = $typevalue[$i]; | |
336 | if ($line =~ m/^(\C):\s*(.*)/) { | |
337 | my $type = $1; | |
338 | my $value = $2; | |
339 | if ($type eq 'F') { | |
340 | if (file_match_pattern($file, $value)) { | |
341 | my $value_pd = ($value =~ tr@/@@); | |
342 | my $file_pd = ($file =~ tr@/@@); | |
343 | $value_pd++ if (substr($value,-1,1) ne "/"); | |
344 | if ($pattern_depth == 0 || | |
345 | (($file_pd - $value_pd) < $pattern_depth)) { | |
346 | $hash{$tvi} = $value_pd; | |
347 | } | |
348 | } | |
349 | } | |
350 | } | |
351 | } | |
1d606b4e | 352 | } |
272a8979 JP |
353 | |
354 | $tvi += ($end - $start); | |
355 | ||
356 | } | |
357 | ||
358 | foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) { | |
359 | add_categories($line); | |
cb7301c7 JP |
360 | } |
361 | ||
4a7fdb5f | 362 | if ($email && $email_git) { |
60db31ac | 363 | vcs_file_signoffs($file); |
cb7301c7 JP |
364 | } |
365 | ||
f5492666 | 366 | if ($email && $email_git_blame) { |
60db31ac | 367 | vcs_file_blame($file); |
f5492666 | 368 | } |
cb7301c7 JP |
369 | } |
370 | ||
dcf36a92 JP |
371 | if ($keywords) { |
372 | @keyword_tvi = sort_and_uniq(@keyword_tvi); | |
373 | foreach my $line (@keyword_tvi) { | |
374 | add_categories($line); | |
375 | } | |
376 | } | |
377 | ||
f5f5078d | 378 | if ($email) { |
cb7301c7 JP |
379 | foreach my $chief (@penguin_chief) { |
380 | if ($chief =~ m/^(.*):(.*)/) { | |
f5f5078d | 381 | my $email_address; |
0e70e83d | 382 | |
a8af2430 | 383 | $email_address = format_email($1, $2, $email_usename); |
f5f5078d | 384 | if ($email_git_penguin_chiefs) { |
3c7385b8 | 385 | push(@email_to, [$email_address, 'chief penguin']); |
f5f5078d | 386 | } else { |
3c7385b8 | 387 | @email_to = grep($_->[0] !~ /${email_address}/, @email_to); |
cb7301c7 JP |
388 | } |
389 | } | |
390 | } | |
03372dbb JP |
391 | |
392 | foreach my $email (@file_emails) { | |
393 | my ($name, $address) = parse_email($email); | |
394 | ||
395 | my $tmp_email = format_email($name, $address, $email_usename); | |
396 | push_email_address($tmp_email, ''); | |
397 | add_role($tmp_email, 'in file'); | |
398 | } | |
cb7301c7 JP |
399 | } |
400 | ||
290603c1 JP |
401 | if ($email || $email_list) { |
402 | my @to = (); | |
403 | if ($email) { | |
404 | @to = (@to, @email_to); | |
cb7301c7 | 405 | } |
290603c1 | 406 | if ($email_list) { |
290603c1 | 407 | @to = (@to, @list_to); |
290603c1 | 408 | } |
3c7385b8 | 409 | output(merge_email(@to)); |
cb7301c7 JP |
410 | } |
411 | ||
412 | if ($scm) { | |
b781655a | 413 | @scm = uniq(@scm); |
cb7301c7 JP |
414 | output(@scm); |
415 | } | |
416 | ||
417 | if ($status) { | |
b781655a | 418 | @status = uniq(@status); |
cb7301c7 JP |
419 | output(@status); |
420 | } | |
421 | ||
422 | if ($subsystem) { | |
b781655a | 423 | @subsystem = uniq(@subsystem); |
cb7301c7 JP |
424 | output(@subsystem); |
425 | } | |
426 | ||
427 | if ($web) { | |
b781655a | 428 | @web = uniq(@web); |
cb7301c7 JP |
429 | output(@web); |
430 | } | |
431 | ||
432 | exit($exit); | |
433 | ||
434 | sub file_match_pattern { | |
435 | my ($file, $pattern) = @_; | |
436 | if (substr($pattern, -1) eq "/") { | |
437 | if ($file =~ m@^$pattern@) { | |
438 | return 1; | |
439 | } | |
440 | } else { | |
441 | if ($file =~ m@^$pattern@) { | |
442 | my $s1 = ($file =~ tr@/@@); | |
443 | my $s2 = ($pattern =~ tr@/@@); | |
444 | if ($s1 == $s2) { | |
445 | return 1; | |
446 | } | |
447 | } | |
448 | } | |
449 | return 0; | |
450 | } | |
451 | ||
452 | sub usage { | |
453 | print <<EOT; | |
454 | usage: $P [options] patchfile | |
870020f9 | 455 | $P [options] -f file|directory |
cb7301c7 JP |
456 | version: $V |
457 | ||
458 | MAINTAINER field selection options: | |
459 | --email => print email address(es) if any | |
460 | --git => include recent git \*-by: signers | |
461 | --git-chief-penguins => include ${penguin_chiefs} | |
462 | --git-min-signatures => number of signatures required (default: 1) | |
463 | --git-max-maintainers => maximum maintainers to add (default: 5) | |
3d202aeb | 464 | --git-min-percent => minimum percentage of commits required (default: 5) |
f5492666 | 465 | --git-blame => use git blame to find modified commits for patch or file |
60db31ac JP |
466 | --git-since => git history to use (default: 1-year-ago) |
467 | --hg-since => hg history to use (default: -365) | |
cb7301c7 JP |
468 | --m => include maintainer(s) if any |
469 | --n => include name 'Full Name <addr\@domain.tld>' | |
470 | --l => include list(s) if any | |
471 | --s => include subscriber only list(s) if any | |
11ecf53c | 472 | --remove-duplicates => minimize duplicate email names/addresses |
3c7385b8 JP |
473 | --roles => show roles (status:subsystem, git-signer, list, etc...) |
474 | --rolestats => show roles and statistics (commits/total_commits, %) | |
03372dbb | 475 | --file-emails => add email addresses found in -f file (default: 0 (off)) |
cb7301c7 JP |
476 | --scm => print SCM tree(s) if any |
477 | --status => print status if any | |
478 | --subsystem => print subsystem name if any | |
479 | --web => print website(s) if any | |
480 | ||
481 | Output type options: | |
482 | --separator [, ] => separator for multiple entries on 1 line | |
42498316 | 483 | using --separator also sets --nomultiline if --separator is not [, ] |
cb7301c7 JP |
484 | --multiline => print 1 entry per line |
485 | ||
cb7301c7 | 486 | Other options: |
3fb55652 | 487 | --pattern-depth => Number of pattern directory traversals (default: 0 (all)) |
dcf36a92 | 488 | --keywords => scan patch for keywords (default: 1 (on)) |
f5f5078d | 489 | --version => show version |
cb7301c7 JP |
490 | --help => show this help information |
491 | ||
3fb55652 | 492 | Default options: |
11ecf53c | 493 | [--email --git --m --n --l --multiline --pattern-depth=0 --remove-duplicates] |
3fb55652 | 494 | |
870020f9 JP |
495 | Notes: |
496 | Using "-f directory" may give unexpected results: | |
f5492666 JP |
497 | Used with "--git", git signators for _all_ files in and below |
498 | directory are examined as git recurses directories. | |
499 | Any specified X: (exclude) pattern matches are _not_ ignored. | |
500 | Used with "--nogit", directory is used as a pattern match, | |
60db31ac JP |
501 | no individual file within the directory or subdirectory |
502 | is matched. | |
f5492666 JP |
503 | Used with "--git-blame", does not iterate all files in directory |
504 | Using "--git-blame" is slow and may add old committers and authors | |
505 | that are no longer active maintainers to the output. | |
3c7385b8 JP |
506 | Using "--roles" or "--rolestats" with git send-email --cc-cmd or any |
507 | other automated tools that expect only ["name"] <email address> | |
508 | may not work because of additional output after <email address>. | |
509 | Using "--rolestats" and "--git-blame" shows the #/total=% commits, | |
510 | not the percentage of the entire file authored. # of commits is | |
511 | not a good measure of amount of code authored. 1 major commit may | |
512 | contain a thousand lines, 5 trivial commits may modify a single line. | |
60db31ac JP |
513 | If git is not installed, but mercurial (hg) is installed and an .hg |
514 | repository exists, the following options apply to mercurial: | |
515 | --git, | |
516 | --git-min-signatures, --git-max-maintainers, --git-min-percent, and | |
517 | --git-blame | |
518 | Use --hg-since not --git-since to control date selection | |
cb7301c7 JP |
519 | EOT |
520 | } | |
521 | ||
522 | sub top_of_kernel_tree { | |
523 | my ($lk_path) = @_; | |
524 | ||
525 | if ($lk_path ne "" && substr($lk_path,length($lk_path)-1,1) ne "/") { | |
526 | $lk_path .= "/"; | |
527 | } | |
528 | if ( (-f "${lk_path}COPYING") | |
529 | && (-f "${lk_path}CREDITS") | |
530 | && (-f "${lk_path}Kbuild") | |
531 | && (-f "${lk_path}MAINTAINERS") | |
532 | && (-f "${lk_path}Makefile") | |
533 | && (-f "${lk_path}README") | |
534 | && (-d "${lk_path}Documentation") | |
535 | && (-d "${lk_path}arch") | |
536 | && (-d "${lk_path}include") | |
537 | && (-d "${lk_path}drivers") | |
538 | && (-d "${lk_path}fs") | |
539 | && (-d "${lk_path}init") | |
540 | && (-d "${lk_path}ipc") | |
541 | && (-d "${lk_path}kernel") | |
542 | && (-d "${lk_path}lib") | |
543 | && (-d "${lk_path}scripts")) { | |
544 | return 1; | |
545 | } | |
546 | return 0; | |
547 | } | |
548 | ||
0e70e83d JP |
549 | sub parse_email { |
550 | my ($formatted_email) = @_; | |
551 | ||
552 | my $name = ""; | |
553 | my $address = ""; | |
554 | ||
11ecf53c | 555 | if ($formatted_email =~ /^([^<]+)<(.+\@.*)>.*$/) { |
0e70e83d JP |
556 | $name = $1; |
557 | $address = $2; | |
11ecf53c | 558 | } elsif ($formatted_email =~ /^\s*<(.+\@\S*)>.*$/) { |
0e70e83d | 559 | $address = $1; |
b781655a | 560 | } elsif ($formatted_email =~ /^(.+\@\S*).*$/) { |
0e70e83d JP |
561 | $address = $1; |
562 | } | |
cb7301c7 JP |
563 | |
564 | $name =~ s/^\s+|\s+$//g; | |
d789504a | 565 | $name =~ s/^\"|\"$//g; |
0e70e83d | 566 | $address =~ s/^\s+|\s+$//g; |
cb7301c7 | 567 | |
0e70e83d JP |
568 | if ($name =~ /[^a-z0-9 \.\-]/i) { ##has "must quote" chars |
569 | $name =~ s/(?<!\\)"/\\"/g; ##escape quotes | |
570 | $name = "\"$name\""; | |
571 | } | |
572 | ||
573 | return ($name, $address); | |
574 | } | |
575 | ||
576 | sub format_email { | |
a8af2430 | 577 | my ($name, $address, $usename) = @_; |
0e70e83d JP |
578 | |
579 | my $formatted_email; | |
580 | ||
581 | $name =~ s/^\s+|\s+$//g; | |
582 | $name =~ s/^\"|\"$//g; | |
583 | $address =~ s/^\s+|\s+$//g; | |
cb7301c7 JP |
584 | |
585 | if ($name =~ /[^a-z0-9 \.\-]/i) { ##has "must quote" chars | |
586 | $name =~ s/(?<!\\)"/\\"/g; ##escape quotes | |
0e70e83d JP |
587 | $name = "\"$name\""; |
588 | } | |
589 | ||
a8af2430 | 590 | if ($usename) { |
0e70e83d JP |
591 | if ("$name" eq "") { |
592 | $formatted_email = "$address"; | |
593 | } else { | |
a8af2430 | 594 | $formatted_email = "$name <$address>"; |
0e70e83d | 595 | } |
cb7301c7 | 596 | } else { |
0e70e83d | 597 | $formatted_email = $address; |
cb7301c7 | 598 | } |
0e70e83d | 599 | |
cb7301c7 JP |
600 | return $formatted_email; |
601 | } | |
602 | ||
272a8979 JP |
603 | sub find_first_section { |
604 | my $index = 0; | |
605 | ||
606 | while ($index < @typevalue) { | |
607 | my $tv = $typevalue[$index]; | |
608 | if (($tv =~ m/^(\C):\s*(.*)/)) { | |
609 | last; | |
610 | } | |
611 | $index++; | |
612 | } | |
613 | ||
614 | return $index; | |
615 | } | |
616 | ||
b781655a | 617 | sub find_starting_index { |
b781655a JP |
618 | my ($index) = @_; |
619 | ||
620 | while ($index > 0) { | |
621 | my $tv = $typevalue[$index]; | |
622 | if (!($tv =~ m/^(\C):\s*(.*)/)) { | |
623 | last; | |
624 | } | |
625 | $index--; | |
626 | } | |
627 | ||
628 | return $index; | |
629 | } | |
630 | ||
631 | sub find_ending_index { | |
cb7301c7 JP |
632 | my ($index) = @_; |
633 | ||
b781655a | 634 | while ($index < @typevalue) { |
cb7301c7 | 635 | my $tv = $typevalue[$index]; |
b781655a JP |
636 | if (!($tv =~ m/^(\C):\s*(.*)/)) { |
637 | last; | |
638 | } | |
639 | $index++; | |
640 | } | |
641 | ||
642 | return $index; | |
643 | } | |
644 | ||
3c7385b8 JP |
645 | sub get_maintainer_role { |
646 | my ($index) = @_; | |
647 | ||
648 | my $i; | |
649 | my $start = find_starting_index($index); | |
650 | my $end = find_ending_index($index); | |
651 | ||
652 | my $role; | |
653 | my $subsystem = $typevalue[$start]; | |
654 | if (length($subsystem) > 20) { | |
655 | $subsystem = substr($subsystem, 0, 17); | |
656 | $subsystem =~ s/\s*$//; | |
657 | $subsystem = $subsystem . "..."; | |
658 | } | |
659 | ||
660 | for ($i = $start + 1; $i < $end; $i++) { | |
661 | my $tv = $typevalue[$i]; | |
662 | if ($tv =~ m/^(\C):\s*(.*)/) { | |
663 | my $ptype = $1; | |
664 | my $pvalue = $2; | |
665 | if ($ptype eq "S") { | |
666 | $role = $pvalue; | |
667 | } | |
668 | } | |
669 | } | |
670 | ||
671 | $role = lc($role); | |
672 | if ($role eq "supported") { | |
673 | $role = "supporter"; | |
674 | } elsif ($role eq "maintained") { | |
675 | $role = "maintainer"; | |
676 | } elsif ($role eq "odd fixes") { | |
677 | $role = "odd fixer"; | |
678 | } elsif ($role eq "orphan") { | |
679 | $role = "orphan minder"; | |
680 | } elsif ($role eq "obsolete") { | |
681 | $role = "obsolete minder"; | |
682 | } elsif ($role eq "buried alive in reporters") { | |
683 | $role = "chief penguin"; | |
684 | } | |
685 | ||
686 | return $role . ":" . $subsystem; | |
687 | } | |
688 | ||
689 | sub get_list_role { | |
690 | my ($index) = @_; | |
691 | ||
692 | my $i; | |
693 | my $start = find_starting_index($index); | |
694 | my $end = find_ending_index($index); | |
695 | ||
696 | my $subsystem = $typevalue[$start]; | |
697 | if (length($subsystem) > 20) { | |
698 | $subsystem = substr($subsystem, 0, 17); | |
699 | $subsystem =~ s/\s*$//; | |
700 | $subsystem = $subsystem . "..."; | |
701 | } | |
702 | ||
703 | if ($subsystem eq "THE REST") { | |
704 | $subsystem = ""; | |
705 | } | |
706 | ||
707 | return $subsystem; | |
708 | } | |
709 | ||
b781655a JP |
710 | sub add_categories { |
711 | my ($index) = @_; | |
712 | ||
713 | my $i; | |
714 | my $start = find_starting_index($index); | |
715 | my $end = find_ending_index($index); | |
716 | ||
717 | push(@subsystem, $typevalue[$start]); | |
718 | ||
719 | for ($i = $start + 1; $i < $end; $i++) { | |
720 | my $tv = $typevalue[$i]; | |
290603c1 | 721 | if ($tv =~ m/^(\C):\s*(.*)/) { |
cb7301c7 JP |
722 | my $ptype = $1; |
723 | my $pvalue = $2; | |
724 | if ($ptype eq "L") { | |
290603c1 JP |
725 | my $list_address = $pvalue; |
726 | my $list_additional = ""; | |
3c7385b8 JP |
727 | my $list_role = get_list_role($i); |
728 | ||
729 | if ($list_role ne "") { | |
730 | $list_role = ":" . $list_role; | |
731 | } | |
290603c1 JP |
732 | if ($list_address =~ m/([^\s]+)\s+(.*)$/) { |
733 | $list_address = $1; | |
734 | $list_additional = $2; | |
735 | } | |
bdf7c685 | 736 | if ($list_additional =~ m/subscribers-only/) { |
cb7301c7 | 737 | if ($email_subscriber_list) { |
3c7385b8 | 738 | push(@list_to, [$list_address, "subscriber list${list_role}"]); |
cb7301c7 JP |
739 | } |
740 | } else { | |
741 | if ($email_list) { | |
3c7385b8 | 742 | push(@list_to, [$list_address, "open list${list_role}"]); |
cb7301c7 JP |
743 | } |
744 | } | |
745 | } elsif ($ptype eq "M") { | |
0e70e83d JP |
746 | my ($name, $address) = parse_email($pvalue); |
747 | if ($name eq "") { | |
b781655a JP |
748 | if ($i > 0) { |
749 | my $tv = $typevalue[$i - 1]; | |
0e70e83d JP |
750 | if ($tv =~ m/^(\C):\s*(.*)/) { |
751 | if ($1 eq "P") { | |
752 | $name = $2; | |
a8af2430 | 753 | $pvalue = format_email($name, $address, $email_usename); |
5f2441e9 JP |
754 | } |
755 | } | |
756 | } | |
757 | } | |
0e70e83d | 758 | if ($email_maintainer) { |
3c7385b8 JP |
759 | my $role = get_maintainer_role($i); |
760 | push_email_addresses($pvalue, $role); | |
cb7301c7 JP |
761 | } |
762 | } elsif ($ptype eq "T") { | |
763 | push(@scm, $pvalue); | |
764 | } elsif ($ptype eq "W") { | |
765 | push(@web, $pvalue); | |
766 | } elsif ($ptype eq "S") { | |
767 | push(@status, $pvalue); | |
768 | } | |
cb7301c7 JP |
769 | } |
770 | } | |
771 | } | |
772 | ||
11ecf53c JP |
773 | my %email_hash_name; |
774 | my %email_hash_address; | |
0e70e83d | 775 | |
11ecf53c JP |
776 | sub email_inuse { |
777 | my ($name, $address) = @_; | |
778 | ||
779 | return 1 if (($name eq "") && ($address eq "")); | |
780 | return 1 if (($name ne "") && exists($email_hash_name{$name})); | |
781 | return 1 if (($address ne "") && exists($email_hash_address{$address})); | |
0e70e83d | 782 | |
0e70e83d JP |
783 | return 0; |
784 | } | |
785 | ||
1b5e1cf6 | 786 | sub push_email_address { |
3c7385b8 | 787 | my ($line, $role) = @_; |
1b5e1cf6 | 788 | |
0e70e83d | 789 | my ($name, $address) = parse_email($line); |
1b5e1cf6 | 790 | |
b781655a JP |
791 | if ($address eq "") { |
792 | return 0; | |
793 | } | |
794 | ||
11ecf53c | 795 | if (!$email_remove_duplicates) { |
a8af2430 | 796 | push(@email_to, [format_email($name, $address, $email_usename), $role]); |
11ecf53c | 797 | } elsif (!email_inuse($name, $address)) { |
a8af2430 | 798 | push(@email_to, [format_email($name, $address, $email_usename), $role]); |
11ecf53c JP |
799 | $email_hash_name{$name}++; |
800 | $email_hash_address{$address}++; | |
1b5e1cf6 | 801 | } |
b781655a JP |
802 | |
803 | return 1; | |
1b5e1cf6 JP |
804 | } |
805 | ||
806 | sub push_email_addresses { | |
3c7385b8 | 807 | my ($address, $role) = @_; |
1b5e1cf6 JP |
808 | |
809 | my @address_list = (); | |
810 | ||
5f2441e9 | 811 | if (rfc822_valid($address)) { |
3c7385b8 | 812 | push_email_address($address, $role); |
5f2441e9 | 813 | } elsif (@address_list = rfc822_validlist($address)) { |
1b5e1cf6 JP |
814 | my $array_count = shift(@address_list); |
815 | while (my $entry = shift(@address_list)) { | |
3c7385b8 | 816 | push_email_address($entry, $role); |
1b5e1cf6 | 817 | } |
5f2441e9 | 818 | } else { |
3c7385b8 | 819 | if (!push_email_address($address, $role)) { |
b781655a JP |
820 | warn("Invalid MAINTAINERS address: '" . $address . "'\n"); |
821 | } | |
1b5e1cf6 | 822 | } |
1b5e1cf6 JP |
823 | } |
824 | ||
3c7385b8 JP |
825 | sub add_role { |
826 | my ($line, $role) = @_; | |
827 | ||
828 | my ($name, $address) = parse_email($line); | |
a8af2430 | 829 | my $email = format_email($name, $address, $email_usename); |
3c7385b8 JP |
830 | |
831 | foreach my $entry (@email_to) { | |
832 | if ($email_remove_duplicates) { | |
833 | my ($entry_name, $entry_address) = parse_email($entry->[0]); | |
03372dbb JP |
834 | if (($name eq $entry_name || $address eq $entry_address) |
835 | && ($role eq "" || !($entry->[1] =~ m/$role/)) | |
836 | ) { | |
3c7385b8 JP |
837 | if ($entry->[1] eq "") { |
838 | $entry->[1] = "$role"; | |
839 | } else { | |
840 | $entry->[1] = "$entry->[1],$role"; | |
841 | } | |
842 | } | |
843 | } else { | |
03372dbb JP |
844 | if ($email eq $entry->[0] |
845 | && ($role eq "" || !($entry->[1] =~ m/$role/)) | |
846 | ) { | |
3c7385b8 JP |
847 | if ($entry->[1] eq "") { |
848 | $entry->[1] = "$role"; | |
849 | } else { | |
850 | $entry->[1] = "$entry->[1],$role"; | |
851 | } | |
852 | } | |
853 | } | |
854 | } | |
855 | } | |
856 | ||
cb7301c7 JP |
857 | sub which { |
858 | my ($bin) = @_; | |
859 | ||
f5f5078d | 860 | foreach my $path (split(/:/, $ENV{PATH})) { |
cb7301c7 JP |
861 | if (-e "$path/$bin") { |
862 | return "$path/$bin"; | |
863 | } | |
864 | } | |
865 | ||
866 | return ""; | |
867 | } | |
868 | ||
8cbb3a77 | 869 | sub mailmap { |
a8af2430 | 870 | my (@lines) = @_; |
8cbb3a77 JP |
871 | my %hash; |
872 | ||
873 | foreach my $line (@lines) { | |
874 | my ($name, $address) = parse_email($line); | |
875 | if (!exists($hash{$name})) { | |
876 | $hash{$name} = $address; | |
11ecf53c JP |
877 | } elsif ($address ne $hash{$name}) { |
878 | $address = $hash{$name}; | |
a8af2430 | 879 | $line = format_email($name, $address, $email_usename); |
8cbb3a77 JP |
880 | } |
881 | if (exists($mailmap{$name})) { | |
882 | my $obj = $mailmap{$name}; | |
883 | foreach my $map_address (@$obj) { | |
884 | if (($map_address eq $address) && | |
885 | ($map_address ne $hash{$name})) { | |
a8af2430 | 886 | $line = format_email($name, $hash{$name}, $email_usename); |
8cbb3a77 JP |
887 | } |
888 | } | |
889 | } | |
890 | } | |
891 | ||
892 | return @lines; | |
893 | } | |
894 | ||
60db31ac JP |
895 | sub git_execute_cmd { |
896 | my ($cmd) = @_; | |
897 | my @lines = (); | |
cb7301c7 | 898 | |
60db31ac JP |
899 | my $output = `$cmd`; |
900 | $output =~ s/^\s*//gm; | |
901 | @lines = split("\n", $output); | |
902 | ||
903 | return @lines; | |
a8af2430 JP |
904 | } |
905 | ||
60db31ac | 906 | sub hg_execute_cmd { |
a8af2430 | 907 | my ($cmd) = @_; |
60db31ac JP |
908 | my @lines = (); |
909 | ||
910 | my $output = `$cmd`; | |
911 | @lines = split("\n", $output); | |
a8af2430 | 912 | |
60db31ac JP |
913 | return @lines; |
914 | } | |
915 | ||
916 | sub vcs_find_signers { | |
917 | my ($cmd) = @_; | |
a8af2430 JP |
918 | my @lines = (); |
919 | my $commits; | |
920 | ||
60db31ac | 921 | @lines = &{$VCS_cmds{"execute_cmd"}}($cmd); |
cb7301c7 | 922 | |
60db31ac | 923 | my $pattern = $VCS_cmds{"commit_pattern"}; |
cb7301c7 | 924 | |
60db31ac | 925 | $commits = grep(/$pattern/, @lines); # of commits |
afa81ee1 | 926 | |
0e70e83d JP |
927 | @lines = grep(/^[-_ a-z]+by:.*\@.*$/i, @lines); |
928 | if (!$email_git_penguin_chiefs) { | |
929 | @lines = grep(!/${penguin_chiefs}/i, @lines); | |
930 | } | |
931 | # cut -f2- -d":" | |
932 | s/.*:\s*(.+)\s*/$1/ for (@lines); | |
933 | ||
a8af2430 JP |
934 | ## Reformat email addresses (with names) to avoid badly written signatures |
935 | ||
3c7385b8 JP |
936 | foreach my $line (@lines) { |
937 | my ($name, $address) = parse_email($line); | |
a8af2430 JP |
938 | $line = format_email($name, $address, 1); |
939 | } | |
940 | ||
941 | return ($commits, @lines); | |
942 | } | |
943 | ||
60db31ac JP |
944 | sub vcs_save_commits { |
945 | my ($cmd) = @_; | |
946 | my @lines = (); | |
947 | my @commits = (); | |
948 | ||
949 | @lines = &{$VCS_cmds{"execute_cmd"}}($cmd); | |
950 | ||
951 | foreach my $line (@lines) { | |
952 | if ($line =~ m/$VCS_cmds{"blame_commit_pattern"}/) { | |
953 | push(@commits, $1); | |
954 | } | |
955 | } | |
956 | ||
957 | return @commits; | |
958 | } | |
959 | ||
960 | sub vcs_blame { | |
961 | my ($file) = @_; | |
962 | my $cmd; | |
963 | my @commits = (); | |
964 | ||
965 | return @commits if (!(-f $file)); | |
966 | ||
967 | if (@range && $VCS_cmds{"blame_range_cmd"} eq "") { | |
968 | my @all_commits = (); | |
969 | ||
970 | $cmd = $VCS_cmds{"blame_file_cmd"}; | |
971 | $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd | |
972 | @all_commits = vcs_save_commits($cmd); | |
973 | ||
974 | foreach my $file_range_diff (@range) { | |
975 | next if (!($file_range_diff =~ m/(.+):(.+):(.+)/)); | |
976 | my $diff_file = $1; | |
977 | my $diff_start = $2; | |
978 | my $diff_length = $3; | |
979 | next if ("$file" ne "$diff_file"); | |
980 | for (my $i = $diff_start; $i < $diff_start + $diff_length; $i++) { | |
981 | push(@commits, $all_commits[$i]); | |
982 | } | |
983 | } | |
984 | } elsif (@range) { | |
985 | foreach my $file_range_diff (@range) { | |
986 | next if (!($file_range_diff =~ m/(.+):(.+):(.+)/)); | |
987 | my $diff_file = $1; | |
988 | my $diff_start = $2; | |
989 | my $diff_length = $3; | |
990 | next if ("$file" ne "$diff_file"); | |
991 | $cmd = $VCS_cmds{"blame_range_cmd"}; | |
992 | $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd | |
993 | push(@commits, vcs_save_commits($cmd)); | |
994 | } | |
995 | } else { | |
996 | $cmd = $VCS_cmds{"blame_file_cmd"}; | |
997 | $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd | |
998 | @commits = vcs_save_commits($cmd); | |
999 | } | |
1000 | ||
1001 | return @commits; | |
1002 | } | |
1003 | ||
1004 | my $printed_novcs = 0; | |
1005 | sub vcs_exists { | |
1006 | %VCS_cmds = %VCS_cmds_git; | |
1007 | return 1 if eval $VCS_cmds{"available"}; | |
1008 | %VCS_cmds = %VCS_cmds_hg; | |
1009 | return 1 if eval $VCS_cmds{"available"}; | |
1010 | %VCS_cmds = (); | |
1011 | if (!$printed_novcs) { | |
1012 | warn("$P: No supported VCS found. Add --nogit to options?\n"); | |
1013 | warn("Using a git repository produces better results.\n"); | |
1014 | warn("Try Linus Torvalds' latest git repository using:\n"); | |
1015 | warn("git clone git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git\n"); | |
1016 | $printed_novcs = 1; | |
1017 | } | |
1018 | return 0; | |
1019 | } | |
1020 | ||
1021 | sub vcs_assign { | |
a8af2430 JP |
1022 | my ($role, $divisor, @lines) = @_; |
1023 | ||
1024 | my %hash; | |
1025 | my $count = 0; | |
1026 | ||
a8af2430 JP |
1027 | return if (@lines <= 0); |
1028 | ||
1029 | if ($divisor <= 0) { | |
60db31ac | 1030 | warn("Bad divisor in " . (caller(0))[3] . ": $divisor\n"); |
a8af2430 | 1031 | $divisor = 1; |
3c7385b8 | 1032 | } |
8cbb3a77 | 1033 | |
11ecf53c JP |
1034 | if ($email_remove_duplicates) { |
1035 | @lines = mailmap(@lines); | |
1036 | } | |
0e70e83d | 1037 | |
0e70e83d | 1038 | @lines = sort(@lines); |
11ecf53c | 1039 | |
0e70e83d | 1040 | # uniq -c |
11ecf53c JP |
1041 | $hash{$_}++ for @lines; |
1042 | ||
0e70e83d | 1043 | # sort -rn |
0e70e83d | 1044 | foreach my $line (sort {$hash{$b} <=> $hash{$a}} keys %hash) { |
11ecf53c | 1045 | my $sign_offs = $hash{$line}; |
a8af2430 | 1046 | my $percent = $sign_offs * 100 / $divisor; |
3c7385b8 | 1047 | |
a8af2430 | 1048 | $percent = 100 if ($percent > 100); |
11ecf53c JP |
1049 | $count++; |
1050 | last if ($sign_offs < $email_git_min_signatures || | |
1051 | $count > $email_git_max_maintainers || | |
a8af2430 | 1052 | $percent < $email_git_min_percent); |
3c7385b8 | 1053 | push_email_address($line, ''); |
3c7385b8 | 1054 | if ($output_rolestats) { |
a8af2430 JP |
1055 | my $fmt_percent = sprintf("%.0f", $percent); |
1056 | add_role($line, "$role:$sign_offs/$divisor=$fmt_percent%"); | |
1057 | } else { | |
1058 | add_role($line, $role); | |
3c7385b8 | 1059 | } |
f5492666 JP |
1060 | } |
1061 | } | |
1062 | ||
60db31ac | 1063 | sub vcs_file_signoffs { |
a8af2430 JP |
1064 | my ($file) = @_; |
1065 | ||
1066 | my @signers = (); | |
60db31ac | 1067 | my $commits; |
f5492666 | 1068 | |
60db31ac | 1069 | return if (!vcs_exists()); |
a8af2430 | 1070 | |
60db31ac JP |
1071 | my $cmd = $VCS_cmds{"find_signers_cmd"}; |
1072 | $cmd =~ s/(\$\w+)/$1/eeg; # interpolate $cmd | |
f5492666 | 1073 | |
60db31ac JP |
1074 | ($commits, @signers) = vcs_find_signers($cmd); |
1075 | vcs_assign("commit_signer", $commits, @signers); | |
f5492666 JP |
1076 | } |
1077 | ||
60db31ac | 1078 | sub vcs_file_blame { |
f5492666 JP |
1079 | my ($file) = @_; |
1080 | ||
a8af2430 | 1081 | my @signers = (); |
60db31ac | 1082 | my @commits = (); |
a8af2430 | 1083 | my $total_commits; |
f5492666 | 1084 | |
60db31ac | 1085 | return if (!vcs_exists()); |
f5492666 | 1086 | |
60db31ac | 1087 | @commits = vcs_blame($file); |
f5492666 | 1088 | @commits = uniq(@commits); |
a8af2430 | 1089 | $total_commits = @commits; |
8cbb3a77 | 1090 | |
a8af2430 JP |
1091 | foreach my $commit (@commits) { |
1092 | my $commit_count; | |
1093 | my @commit_signers = (); | |
8cbb3a77 | 1094 | |
60db31ac JP |
1095 | my $cmd = $VCS_cmds{"find_commit_signers_cmd"}; |
1096 | $cmd =~ s/(\$\w+)/$1/eeg; #interpolate $cmd | |
1097 | ||
1098 | ($commit_count, @commit_signers) = vcs_find_signers($cmd); | |
1099 | push(@signers, @commit_signers); | |
f5492666 JP |
1100 | } |
1101 | ||
a8af2430 | 1102 | if ($from_filename) { |
60db31ac | 1103 | vcs_assign("commits", $total_commits, @signers); |
a8af2430 | 1104 | } else { |
60db31ac | 1105 | vcs_assign("modified commits", $total_commits, @signers); |
cb7301c7 | 1106 | } |
cb7301c7 JP |
1107 | } |
1108 | ||
1109 | sub uniq { | |
a8af2430 | 1110 | my (@parms) = @_; |
cb7301c7 JP |
1111 | |
1112 | my %saw; | |
1113 | @parms = grep(!$saw{$_}++, @parms); | |
1114 | return @parms; | |
1115 | } | |
1116 | ||
1117 | sub sort_and_uniq { | |
a8af2430 | 1118 | my (@parms) = @_; |
cb7301c7 JP |
1119 | |
1120 | my %saw; | |
1121 | @parms = sort @parms; | |
1122 | @parms = grep(!$saw{$_}++, @parms); | |
1123 | return @parms; | |
1124 | } | |
1125 | ||
03372dbb JP |
1126 | sub clean_file_emails { |
1127 | my (@file_emails) = @_; | |
1128 | my @fmt_emails = (); | |
1129 | ||
1130 | foreach my $email (@file_emails) { | |
1131 | $email =~ s/[\(\<\{]{0,1}([A-Za-z0-9_\.\+-]+\@[A-Za-z0-9\.-]+)[\)\>\}]{0,1}/\<$1\>/g; | |
1132 | my ($name, $address) = parse_email($email); | |
1133 | if ($name eq '"[,\.]"') { | |
1134 | $name = ""; | |
1135 | } | |
1136 | ||
1137 | my @nw = split(/[^A-Za-zÀ-ÿ\'\,\.\+-]/, $name); | |
1138 | if (@nw > 2) { | |
1139 | my $first = $nw[@nw - 3]; | |
1140 | my $middle = $nw[@nw - 2]; | |
1141 | my $last = $nw[@nw - 1]; | |
1142 | ||
1143 | if (((length($first) == 1 && $first =~ m/[A-Za-z]/) || | |
1144 | (length($first) == 2 && substr($first, -1) eq ".")) || | |
1145 | (length($middle) == 1 || | |
1146 | (length($middle) == 2 && substr($middle, -1) eq "."))) { | |
1147 | $name = "$first $middle $last"; | |
1148 | } else { | |
1149 | $name = "$middle $last"; | |
1150 | } | |
1151 | } | |
1152 | ||
1153 | if (substr($name, -1) =~ /[,\.]/) { | |
1154 | $name = substr($name, 0, length($name) - 1); | |
1155 | } elsif (substr($name, -2) =~ /[,\.]"/) { | |
1156 | $name = substr($name, 0, length($name) - 2) . '"'; | |
1157 | } | |
1158 | ||
1159 | if (substr($name, 0, 1) =~ /[,\.]/) { | |
1160 | $name = substr($name, 1, length($name) - 1); | |
1161 | } elsif (substr($name, 0, 2) =~ /"[,\.]/) { | |
1162 | $name = '"' . substr($name, 2, length($name) - 2); | |
1163 | } | |
1164 | ||
1165 | my $fmt_email = format_email($name, $address, $email_usename); | |
1166 | push(@fmt_emails, $fmt_email); | |
1167 | } | |
1168 | return @fmt_emails; | |
1169 | } | |
1170 | ||
3c7385b8 JP |
1171 | sub merge_email { |
1172 | my @lines; | |
1173 | my %saw; | |
1174 | ||
1175 | for (@_) { | |
1176 | my ($address, $role) = @$_; | |
1177 | if (!$saw{$address}) { | |
1178 | if ($output_roles) { | |
60db31ac | 1179 | push(@lines, "$address ($role)"); |
3c7385b8 | 1180 | } else { |
60db31ac | 1181 | push(@lines, $address); |
3c7385b8 JP |
1182 | } |
1183 | $saw{$address} = 1; | |
1184 | } | |
1185 | } | |
1186 | ||
1187 | return @lines; | |
1188 | } | |
1189 | ||
cb7301c7 | 1190 | sub output { |
a8af2430 | 1191 | my (@parms) = @_; |
cb7301c7 JP |
1192 | |
1193 | if ($output_multiline) { | |
1194 | foreach my $line (@parms) { | |
1195 | print("${line}\n"); | |
1196 | } | |
1197 | } else { | |
1198 | print(join($output_separator, @parms)); | |
1199 | print("\n"); | |
1200 | } | |
1201 | } | |
1b5e1cf6 JP |
1202 | |
1203 | my $rfc822re; | |
1204 | ||
1205 | sub make_rfc822re { | |
1206 | # Basic lexical tokens are specials, domain_literal, quoted_string, atom, and | |
1207 | # comment. We must allow for rfc822_lwsp (or comments) after each of these. | |
1208 | # This regexp will only work on addresses which have had comments stripped | |
1209 | # and replaced with rfc822_lwsp. | |
1210 | ||
1211 | my $specials = '()<>@,;:\\\\".\\[\\]'; | |
1212 | my $controls = '\\000-\\037\\177'; | |
1213 | ||
1214 | my $dtext = "[^\\[\\]\\r\\\\]"; | |
1215 | my $domain_literal = "\\[(?:$dtext|\\\\.)*\\]$rfc822_lwsp*"; | |
1216 | ||
1217 | my $quoted_string = "\"(?:[^\\\"\\r\\\\]|\\\\.|$rfc822_lwsp)*\"$rfc822_lwsp*"; | |
1218 | ||
1219 | # Use zero-width assertion to spot the limit of an atom. A simple | |
1220 | # $rfc822_lwsp* causes the regexp engine to hang occasionally. | |
1221 | my $atom = "[^$specials $controls]+(?:$rfc822_lwsp+|\\Z|(?=[\\[\"$specials]))"; | |
1222 | my $word = "(?:$atom|$quoted_string)"; | |
1223 | my $localpart = "$word(?:\\.$rfc822_lwsp*$word)*"; | |
1224 | ||
1225 | my $sub_domain = "(?:$atom|$domain_literal)"; | |
1226 | my $domain = "$sub_domain(?:\\.$rfc822_lwsp*$sub_domain)*"; | |
1227 | ||
1228 | my $addr_spec = "$localpart\@$rfc822_lwsp*$domain"; | |
1229 | ||
1230 | my $phrase = "$word*"; | |
1231 | my $route = "(?:\@$domain(?:,\@$rfc822_lwsp*$domain)*:$rfc822_lwsp*)"; | |
1232 | my $route_addr = "\\<$rfc822_lwsp*$route?$addr_spec\\>$rfc822_lwsp*"; | |
1233 | my $mailbox = "(?:$addr_spec|$phrase$route_addr)"; | |
1234 | ||
1235 | my $group = "$phrase:$rfc822_lwsp*(?:$mailbox(?:,\\s*$mailbox)*)?;\\s*"; | |
1236 | my $address = "(?:$mailbox|$group)"; | |
1237 | ||
1238 | return "$rfc822_lwsp*$address"; | |
1239 | } | |
1240 | ||
1241 | sub rfc822_strip_comments { | |
1242 | my $s = shift; | |
1243 | # Recursively remove comments, and replace with a single space. The simpler | |
1244 | # regexps in the Email Addressing FAQ are imperfect - they will miss escaped | |
1245 | # chars in atoms, for example. | |
1246 | ||
1247 | while ($s =~ s/^((?:[^"\\]|\\.)* | |
1248 | (?:"(?:[^"\\]|\\.)*"(?:[^"\\]|\\.)*)*) | |
1249 | \((?:[^()\\]|\\.)*\)/$1 /osx) {} | |
1250 | return $s; | |
1251 | } | |
1252 | ||
1253 | # valid: returns true if the parameter is an RFC822 valid address | |
1254 | # | |
1255 | sub rfc822_valid ($) { | |
1256 | my $s = rfc822_strip_comments(shift); | |
1257 | ||
1258 | if (!$rfc822re) { | |
1259 | $rfc822re = make_rfc822re(); | |
1260 | } | |
1261 | ||
1262 | return $s =~ m/^$rfc822re$/so && $s =~ m/^$rfc822_char*$/; | |
1263 | } | |
1264 | ||
1265 | # validlist: In scalar context, returns true if the parameter is an RFC822 | |
1266 | # valid list of addresses. | |
1267 | # | |
1268 | # In list context, returns an empty list on failure (an invalid | |
1269 | # address was found); otherwise a list whose first element is the | |
1270 | # number of addresses found and whose remaining elements are the | |
1271 | # addresses. This is needed to disambiguate failure (invalid) | |
1272 | # from success with no addresses found, because an empty string is | |
1273 | # a valid list. | |
1274 | ||
1275 | sub rfc822_validlist ($) { | |
1276 | my $s = rfc822_strip_comments(shift); | |
1277 | ||
1278 | if (!$rfc822re) { | |
1279 | $rfc822re = make_rfc822re(); | |
1280 | } | |
1281 | # * null list items are valid according to the RFC | |
1282 | # * the '1' business is to aid in distinguishing failure from no results | |
1283 | ||
1284 | my @r; | |
1285 | if ($s =~ m/^(?:$rfc822re)?(?:,(?:$rfc822re)?)*$/so && | |
1286 | $s =~ m/^$rfc822_char*$/) { | |
5f2441e9 | 1287 | while ($s =~ m/(?:^|,$rfc822_lwsp*)($rfc822re)/gos) { |
60db31ac | 1288 | push(@r, $1); |
1b5e1cf6 JP |
1289 | } |
1290 | return wantarray ? (scalar(@r), @r) : 1; | |
1291 | } | |
60db31ac | 1292 | return wantarray ? () : 0; |
1b5e1cf6 | 1293 | } |