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