2 package IkiWiki::Plugin::git;
9 use URI::Escape q{uri_escape_utf8};
10 use open qw{:utf8 :std};
12 my $sha1_pattern = qr/[0-9a-fA-F]{40}/; # pattern to validate Git sha1sums
13 my $dummy_commit_msg = 'dummy commit'; # message to skip in recent changes
16 hook(type => "checkconfig", id => "git", call => \&checkconfig);
17 hook(type => "getsetup", id => "git", call => \&getsetup);
18 hook(type => "genwrapper", id => "git", call => \&genwrapper);
19 hook(type => "rcs", id => "rcs_update", call => \&rcs_update);
20 hook(type => "rcs", id => "rcs_prepedit", call => \&rcs_prepedit);
21 hook(type => "rcs", id => "rcs_commit", call => \&rcs_commit);
22 hook(type => "rcs", id => "rcs_commit_staged", call => \&rcs_commit_staged);
23 hook(type => "rcs", id => "rcs_add", call => \&rcs_add);
24 hook(type => "rcs", id => "rcs_remove", call => \&rcs_remove);
25 hook(type => "rcs", id => "rcs_rename", call => \&rcs_rename);
26 hook(type => "rcs", id => "rcs_recentchanges", call => \&rcs_recentchanges);
27 hook(type => "rcs", id => "rcs_diff", call => \&rcs_diff);
28 hook(type => "rcs", id => "rcs_getctime", call => \&rcs_getctime);
29 hook(type => "rcs", id => "rcs_getmtime", call => \&rcs_getmtime);
30 hook(type => "rcs", id => "rcs_receive", call => \&rcs_receive);
31 hook(type => "rcs", id => "rcs_preprevert", call => \&rcs_preprevert);
32 hook(type => "rcs", id => "rcs_revert", call => \&rcs_revert);
33 hook(type => "rcs", id => "rcs_find_changes", call => \&rcs_find_changes);
34 hook(type => "rcs", id => "rcs_get_current_rev", call => \&rcs_get_current_rev);
38 if (! defined $config{gitorigin_branch}) {
39 $config{gitorigin_branch}="origin";
41 if (! defined $config{gitmaster_branch}) {
42 $config{gitmaster_branch}="master";
44 if (defined $config{git_wrapper} &&
45 length $config{git_wrapper}) {
46 push @{$config{wrappers}}, {
47 wrapper => $config{git_wrapper},
48 wrappermode => (defined $config{git_wrappermode} ? $config{git_wrappermode} : "06755"),
49 wrapper_background_command => $config{git_wrapper_background_command},
53 if (defined $config{git_test_receive_wrapper} &&
54 length $config{git_test_receive_wrapper} &&
55 defined $config{untrusted_committers} &&
56 @{$config{untrusted_committers}}) {
57 push @{$config{wrappers}}, {
59 wrapper => $config{git_test_receive_wrapper},
60 wrappermode => (defined $config{git_wrappermode} ? $config{git_wrappermode} : "06755"),
64 # Avoid notes, parser does not handle and they only slow things down.
65 $ENV{GIT_NOTES_REF}="";
67 # Run receive test only if being called by the wrapper, and not
68 # when generating same.
69 if ($config{test_receive} && ! exists $config{wrapper}) {
70 require IkiWiki::Receive;
71 IkiWiki::Receive::test();
78 safe => 0, # rcs plugin
84 example => "/git/wiki.git/hooks/post-update",
85 description => "git hook to generate",
89 git_wrapper_background_command => {
91 example => "git push github",
92 description => "shell command for git_wrapper to run, in the background",
99 description => "mode for git_wrapper (can safely be made suid)",
103 git_test_receive_wrapper => {
105 example => "/git/wiki.git/hooks/pre-receive",
106 description => "git pre-receive hook to generate",
110 untrusted_committers => {
113 description => "unix users whose commits should be checked by the pre-receive hook",
119 example => "http://git.example.com/gitweb.cgi?p=wiki.git;a=history;f=[[file]];hb=HEAD",
120 description => "gitweb url to show file history ([[file]] substituted)",
126 example => "http://git.example.com/gitweb.cgi?p=wiki.git;a=blobdiff;f=[[file]];h=[[sha1_to]];hp=[[sha1_from]];hb=[[sha1_commit]];hpb=[[sha1_parent]]",
127 description => "gitweb url to show a diff ([[file]], [[sha1_to]], [[sha1_from]], [[sha1_commit]], and [[sha1_parent]] substituted)",
131 gitorigin_branch => {
134 description => "where to pull and push changes (set to empty string to disable)",
135 safe => 0, # paranoia
138 gitmaster_branch => {
141 description => "branch that the wiki is stored in",
142 safe => 0, # paranoia
148 if ($config{test_receive}) {
149 require IkiWiki::Receive;
150 return IkiWiki::Receive::genwrapper();
160 sub in_git_dir ($$) {
169 # Start a child process safely without resorting to /bin/sh.
170 # Returns command output (in list content) or success state
171 # (in scalar context), or runs the specified data handler.
173 my ($error_handler, $data_handler, @cmdline) = @_;
175 my $pid = open my $OUT, "-|";
177 error("Cannot fork: $!") if !defined $pid;
181 # Git commands want to be in wc.
182 if (! defined $git_dir) {
183 chdir $config{srcdir}
184 or error("cannot chdir to $config{srcdir}: $!");
188 or error("cannot chdir to $git_dir: $!");
190 exec @cmdline or error("Cannot exec '@cmdline': $!");
194 # git output is probably utf-8 encoded, but may contain
195 # other encodings or invalidly encoded stuff. So do not rely
196 # on the normal utf-8 IO layer, decode it by hand.
201 $_=decode_utf8($_, 0);
205 if (! defined $data_handler) {
209 last unless $data_handler->($_);
215 $error_handler->("'@cmdline' failed: $!") if $? && $error_handler;
217 return wantarray ? @lines : ($? == 0);
219 # Convenient wrappers.
220 sub run_or_die ($@) { safe_git(\&error, undef, @_) }
221 sub run_or_cry ($@) { safe_git(sub { warn @_ }, undef, @_) }
222 sub run_or_non ($@) { safe_git(undef, undef, @_) }
225 sub merge_past ($$$) {
226 # Unlike with Subversion, Git cannot make a 'svn merge -rN:M file'.
227 # Git merge commands work with the committed changes, except in the
228 # implicit case of '-m' of git checkout(1). So we should invent a
229 # kludge here. In principle, we need to create a throw-away branch
230 # in preparing for the merge itself. Since branches are cheap (and
231 # branching is fast), this shouldn't cost high.
233 # The main problem is the presence of _uncommitted_ local changes. One
234 # possible approach to get rid of this situation could be that we first
235 # make a temporary commit in the master branch and later restore the
236 # initial state (this is possible since Git has the ability to undo a
237 # commit, i.e. 'git reset --soft HEAD^'). The method can be summarized
240 # - create a diff of HEAD:current-sha1
242 # - create a dummy branch and switch to it
243 # - rewind to past (reset --hard to the current-sha1)
244 # - apply the diff and commit
245 # - switch to master and do the merge with the dummy branch
246 # - make a soft reset (undo the last commit of master)
248 # The above method has some drawbacks: (1) it needs a redundant commit
249 # just to get rid of local changes, (2) somewhat slow because of the
250 # required system forks. Until someone points a more straight method
251 # (which I would be grateful) I have implemented an alternative method.
252 # In this approach, we hide all the modified files from Git by renaming
253 # them (using the 'rename' builtin) and later restore those files in
254 # the throw-away branch (that is, we put the files themselves instead
255 # of applying a patch).
257 my ($sha1, $file, $message) = @_;
259 my @undo; # undo stack for cleanup in case of an error
260 my $conflict; # file content with conflict markers
263 # Hide local changes from Git by renaming the modified file.
264 # Relative paths must be converted to absolute for renaming.
265 my ($target, $hidden) = (
266 "$config{srcdir}/${file}", "$config{srcdir}/${file}.${sha1}"
268 rename($target, $hidden)
269 or error("rename '$target' to '$hidden' failed: $!");
270 # Ensure to restore the renamed file on error.
272 return if ! -e "$hidden"; # already renamed
273 rename($hidden, $target)
274 or warn "rename '$hidden' to '$target' failed: $!";
277 my $branch = "throw_away_${sha1}"; # supposed to be unique
279 # Create a throw-away branch and rewind backward.
280 push @undo, sub { run_or_cry('git', 'branch', '-D', $branch) };
281 run_or_die('git', 'branch', $branch, $sha1);
283 # Switch to throw-away branch for the merge operation.
285 if (!run_or_cry('git', 'checkout', $config{gitmaster_branch})) {
286 run_or_cry('git', 'checkout','-f',$config{gitmaster_branch});
289 run_or_die('git', 'checkout', $branch);
291 # Put the modified file in _this_ branch.
292 rename($hidden, $target)
293 or error("rename '$hidden' to '$target' failed: $!");
295 # _Silently_ commit all modifications in the current branch.
296 run_or_non('git', 'commit', '-m', $message, '-a');
297 # ... and re-switch to master.
298 run_or_die('git', 'checkout', $config{gitmaster_branch});
300 # Attempt to merge without complaining.
301 if (!run_or_non('git', 'pull', '--no-commit', '.', $branch)) {
302 $conflict = readfile($target);
303 run_or_die('git', 'reset', '--hard');
308 # Process undo stack (in reverse order). By policy cleanup
309 # actions should normally print a warning on failure.
310 while (my $handle = pop @undo) {
314 error("Git merge failed!\n$failure\n") if $failure;
319 sub decode_git_file ($) {
322 # git does not output utf-8 filenames, but instead
323 # double-quotes them with the utf-8 characters
324 # escaped as \nnn\nnn.
325 if ($file =~ m/^"(.*)"$/) {
326 ($file=$1) =~ s/\\([0-7]{1,3})/chr(oct($1))/eg;
329 # strip prefix if in a subdir
330 if (! defined $prefix) {
331 ($prefix) = run_or_die('git', 'rev-parse', '--show-prefix');
332 if (! defined $prefix) {
336 $file =~ s/^\Q$prefix\E//;
338 return decode("utf8", $file);
341 sub parse_diff_tree ($) {
342 # Parse the raw diff tree chunk and return the info hash.
343 # See git-diff-tree(1) for the syntax.
347 return if ! @{ $dt_ref } ||
348 !defined $dt_ref->[0] || !length $dt_ref->[0];
352 while (my $line = shift @{ $dt_ref }) {
353 return if $line !~ m/^(.+) ($sha1_pattern)/;
360 # Identification lines for the commit.
361 while (my $line = shift @{ $dt_ref }) {
362 # Regexps are semi-stolen from gitweb.cgi.
363 if ($line =~ m/^tree ([0-9a-fA-F]{40})$/) {
366 elsif ($line =~ m/^parent ([0-9a-fA-F]{40})$/) {
367 # XXX: collecting in reverse order
368 push @{ $ci{'parents'} }, $1;
370 elsif ($line =~ m/^(author|committer) (.*) ([0-9]+) (.*)$/) {
371 my ($who, $name, $epoch, $tz) =
375 $ci{ "${who}_epoch" } = $epoch;
376 $ci{ "${who}_tz" } = $tz;
378 if ($name =~ m/^([^<]+)\s+<([^@>]+)/) {
379 $ci{"${who}_name"} = $1;
380 $ci{"${who}_username"} = $2;
382 elsif ($name =~ m/^([^<]+)\s+<>$/) {
383 $ci{"${who}_username"} = $1;
386 $ci{"${who}_username"} = $name;
389 elsif ($line =~ m/^$/) {
390 # Trailing empty line signals next section.
395 debug("No 'tree' seen in diff-tree output") if !defined $ci{'tree'};
397 if (defined $ci{'parents'}) {
398 $ci{'parent'} = @{ $ci{'parents'} }[0];
401 $ci{'parent'} = 0 x 40;
404 # Commit message (optional).
405 while ($dt_ref->[0] =~ /^ /) {
406 my $line = shift @{ $dt_ref };
408 push @{ $ci{'comment'} }, $line;
410 shift @{ $dt_ref } if $dt_ref->[0] =~ /^$/;
413 while (my $line = shift @{ $dt_ref }) {
415 (:+) # number of parents
416 ([^\t]+)\t # modes, sha1, status
419 my $num_parents = length $1;
420 my @tmp = split(" ", $2);
421 my ($file, $file_to) = split("\t", $3);
422 my @mode_from = splice(@tmp, 0, $num_parents);
423 my $mode_to = shift(@tmp);
424 my @sha1_from = splice(@tmp, 0, $num_parents);
425 my $sha1_to = shift(@tmp);
426 my $status = shift(@tmp);
429 push @{ $ci{'details'} }, {
430 'file' => decode_git_file($file),
431 'sha1_from' => $sha1_from[0],
432 'sha1_to' => $sha1_to,
433 'mode_from' => $mode_from[0],
434 'mode_to' => $mode_to,
446 sub git_commit_info ($;$) {
447 # Return an array of commit info hashes of num commits
448 # starting from the given sha1sum.
449 my ($sha1, $num) = @_;
452 push @opts, "--max-count=$num" if defined $num;
454 my @raw_lines = run_or_die('git', 'log', @opts,
455 '--pretty=raw', '--raw', '--abbrev=40', '--always', '-c',
456 '-r', $sha1, '--', '.');
459 while (my $parsed = parse_diff_tree(\@raw_lines)) {
463 warn "Cannot parse commit info for '$sha1' commit" if !@ci;
465 return wantarray ? @ci : $ci[0];
468 sub rcs_find_changes ($) {
471 my @raw_lines = run_or_die('git', 'log',
472 '--pretty=raw', '--raw', '--abbrev=40', '--always', '-c',
473 '--no-renames', , '--reverse',
474 '-r', "$oldrev..HEAD", '--', '.');
476 # Due to --reverse, we see changes in chronological order.
479 my $nullsha = 0 x 40;
481 while (my $ci = parse_diff_tree(\@raw_lines)) {
483 foreach my $i (@{$ci->{details}}) {
485 if ($i->{sha1_to} eq $nullsha) {
486 delete $changed{$file};
490 delete $deleted{$file};
496 return (\%changed, \%deleted, $newrev);
499 sub git_sha1_file ($) {
501 git_sha1("--", $file);
505 # Ignore error since a non-existing file might be given.
506 my ($sha1) = run_or_non('git', 'rev-list', '--max-count=1', 'HEAD',
509 ($sha1) = $sha1 =~ m/($sha1_pattern)/; # sha1 is untainted now
511 return defined $sha1 ? $sha1 : '';
514 sub rcs_get_current_rev () {
519 # Update working directory.
521 if (length $config{gitorigin_branch}) {
522 run_or_cry('git', 'pull', '--prune', $config{gitorigin_branch});
526 sub rcs_prepedit ($) {
527 # Return the commit sha1sum of the file when editing begins.
528 # This will be later used in rcs_commit if a merge is required.
531 return git_sha1_file($file);
535 # Try to commit the page; returns undef on _success_ and
536 # a version of the page with the rcs's conflict markers on
540 # Check to see if the page has been changed by someone else since
541 # rcs_prepedit was called.
542 my $cur = git_sha1_file($params{file});
543 my ($prev) = $params{token} =~ /^($sha1_pattern)$/; # untaint
545 if (defined $cur && defined $prev && $cur ne $prev) {
546 my $conflict = merge_past($prev, $params{file}, $dummy_commit_msg);
547 return $conflict if defined $conflict;
550 return rcs_commit_helper(@_);
553 sub rcs_commit_staged (@) {
554 # Commits all staged changes. Changes can be staged using rcs_add,
555 # rcs_remove, and rcs_rename.
556 return rcs_commit_helper(@_);
559 sub rcs_commit_helper (@) {
564 if (defined $params{session}) {
565 # Set the commit author and email based on web session info.
567 if (defined $params{session}->param("name")) {
568 $u=$params{session}->param("name");
570 elsif (defined $params{session}->remote_addr()) {
571 $u=$params{session}->remote_addr();
575 # MITLOGIN This algorithm could be improved
576 $ENV{GIT_AUTHOR_NAME}=IkiWiki::userinfo_get($u, "realname");
578 if (defined $params{session}->param("nickname")) {
579 $u=encode_utf8($params{session}->param("nickname"));
581 $u=~s/[^-_0-9[:alnum:]]+//g;
584 $ENV{GIT_AUTHOR_EMAIL}="$u\@mit.edu";
588 $params{message} = IkiWiki::possibly_foolish_untaint($params{message});
590 if ($params{message} !~ /\S/) {
591 # Force git to allow empty commit messages.
592 # (If this version of git supports it.)
593 my ($version)=`git --version` =~ /git version (.*)/;
594 if ($version ge "1.7.8") {
595 push @opts, "--allow-empty-message", "--no-edit";
597 if ($version ge "1.7.2") {
598 push @opts, "--allow-empty-message";
600 elsif ($version ge "1.5.4") {
601 push @opts, '--cleanup=verbatim';
604 $params{message}.=".";
607 if (exists $params{file}) {
608 push @opts, '--', $params{file};
610 # git commit returns non-zero if nothing really changed.
611 # So we should ignore its exit status (hence run_or_non).
612 if (run_or_non('git', 'commit', '-m', $params{message}, '-q', @opts)) {
613 if (length $config{gitorigin_branch}) {
614 run_or_cry('git', 'push', $config{gitorigin_branch});
619 return undef; # success
623 # Add file to archive.
627 run_or_cry('git', 'add', $file);
631 # Remove file from archive.
635 run_or_cry('git', 'rm', '-f', $file);
638 sub rcs_rename ($$) {
639 my ($src, $dest) = @_;
641 run_or_cry('git', 'mv', '-f', $src, $dest);
644 sub rcs_recentchanges ($) {
645 # List of recent changes.
649 eval q{use Date::Parse};
653 foreach my $ci (git_commit_info('HEAD', $num || 1)) {
654 # Skip redundant commits.
655 next if ($ci->{'comment'} && @{$ci->{'comment'}}[0] eq $dummy_commit_msg);
657 my ($sha1, $when) = (
659 $ci->{'author_epoch'}
663 foreach my $detail (@{ $ci->{'details'} }) {
664 my $file = $detail->{'file'};
665 my $efile = uri_escape_utf8($file);
667 my $diffurl = defined $config{'diffurl'} ? $config{'diffurl'} : "";
668 $diffurl =~ s/\[\[file\]\]/$efile/go;
669 $diffurl =~ s/\[\[sha1_parent\]\]/$ci->{'parent'}/go;
670 $diffurl =~ s/\[\[sha1_from\]\]/$detail->{'sha1_from'}/go;
671 $diffurl =~ s/\[\[sha1_to\]\]/$detail->{'sha1_to'}/go;
672 $diffurl =~ s/\[\[sha1_commit\]\]/$sha1/go;
675 page => pagename($file),
682 foreach my $line (@{$ci->{'comment'}}) {
683 $pastblank=1 if $line eq '';
684 next if $pastblank && $line=~m/^ *(signed[ \-]off[ \-]by[ :]|acked[ \-]by[ :]|cc[ :])/i;
685 push @messages, { line => $line };
688 my $user=$ci->{'author_username'};
689 my $web_commit = ($ci->{'author'} =~ /\@web>/);
692 # Set nickname only if a non-url author_username is available,
693 # and author_name is an url.
694 if ($user !~ /:\/\// && defined $ci->{'author_name'} &&
695 $ci->{'author_name'} =~ /:\/\//) {
697 $user=$ci->{'author_name'};
700 # compatability code for old web commit messages
702 defined $messages[0] &&
703 $messages[0]->{line} =~ m/$config{web_commit_regexp}/) {
704 $user = defined $2 ? "$2" : "$3";
705 $messages[0]->{line} = $4;
712 nickname => $nickname,
713 committype => $web_commit ? "web" : "git",
715 message => [@messages],
719 last if @rets >= $num;
728 my ($sha1) = $rev =~ /^($sha1_pattern)$/; # untaint
732 return if defined $maxlines && @lines == $maxlines;
733 push @lines, $line."\n"
734 if (@lines || $line=~/^diff --git/);
737 safe_git(undef, $addlines, "git", "show", $sha1);
742 return join("", @lines);
751 my $id=shift; # 0 = mtime ; 1 = ctime
753 if (! keys %time_cache) {
755 foreach my $line (run_or_die('git', 'log',
756 '--pretty=format:%at',
757 '--name-only', '--relative')) {
758 if (! defined $date && $line =~ /^(\d+)$/) {
761 elsif (! length $line) {
765 my $f=decode_git_file($line);
767 if (! $time_cache{$f}) {
768 $time_cache{$f}[0]=$date; # mtime
770 $time_cache{$f}[1]=$date; # ctime
775 return exists $time_cache{$file} ? $time_cache{$file}[$id] : 0;
780 sub rcs_getctime ($) {
783 return findtimes($file, 1);
786 sub rcs_getmtime ($) {
789 return findtimes($file, 0);
795 # The wiki may not be the only thing in the git repo.
796 # Determine if it is in a subdirectory by examining the srcdir,
797 # and its parents, looking for the .git directory.
799 return @$ret if defined $ret;
802 my $dir=$config{srcdir};
803 while (! -d "$dir/.git") {
804 $subdir=IkiWiki::basename($dir)."/".$subdir;
805 $dir=IkiWiki::dirname($dir);
807 error("cannot determine root of git repo");
811 $ret=[$subdir, $dir];
817 sub git_parse_changes {
818 my $reverted = shift;
821 my ($subdir, $rootdir) = git_find_root();
823 foreach my $ci (@changes) {
824 foreach my $detail (@{ $ci->{'details'} }) {
825 my $file = $detail->{'file'};
827 # check that all changed files are in the subdir
828 if (length $subdir &&
829 ! ($file =~ s/^\Q$subdir\E//)) {
830 error sprintf(gettext("you are not allowed to change %s"), $file);
833 my ($action, $mode, $path);
834 if ($detail->{'status'} =~ /^[M]+\d*$/) {
836 $mode=$detail->{'mode_to'};
838 elsif ($detail->{'status'} =~ /^[AM]+\d*$/) {
839 $action= $reverted ? "remove" : "add";
840 $mode=$detail->{'mode_to'};
842 elsif ($detail->{'status'} =~ /^[DAM]+\d*/) {
843 $action= $reverted ? "add" : "remove";
844 $mode=$detail->{'mode_from'};
847 error "unknown status ".$detail->{'status'};
850 # test that the file mode is ok
851 if ($mode !~ /^100[64][64][64]$/) {
852 error sprintf(gettext("you cannot act on a file with mode %s"), $mode);
854 if ($action eq "change") {
855 if ($detail->{'mode_from'} ne $detail->{'mode_to'}) {
856 error gettext("you are not allowed to change file modes");
860 # extract attachment to temp file
861 if (($action eq 'add' || $action eq 'change') &&
863 eval q{use File::Temp};
866 ($fh, $path)=File::Temp::tempfile(undef, UNLINK => 1);
867 my $cmd = "cd $git_dir && ".
868 "git show $detail->{sha1_to} > '$path'";
869 if (system($cmd) != 0) {
870 error("failed writing temp file '$path'.");
889 my ($oldrev, $newrev, $refname) = split(' ', $_, 3);
891 # only allow changes to gitmaster_branch
892 if ($refname !~ /^refs\/heads\/\Q$config{gitmaster_branch}\E$/) {
893 error sprintf(gettext("you are not allowed to change %s"), $refname);
896 # Avoid chdir when running git here, because the changes
897 # are in the master git repo, not the srcdir repo.
898 # (Also, if a subdir is involved, we don't want to chdir to
899 # it and only see changes in it.)
900 # The pre-receive hook already puts us in the right place.
901 in_git_dir(".", sub {
902 push @rets, git_parse_changes(0, git_commit_info($oldrev."..".$newrev));
906 return reverse @rets;
909 sub rcs_preprevert ($) {
911 my ($sha1) = $rev =~ /^($sha1_pattern)$/; # untaint
913 # Examine changes from root of git repo, not from any subdir,
914 # in order to see all changes.
915 my ($subdir, $rootdir) = git_find_root();
916 in_git_dir($rootdir, sub {
917 my @commits=git_commit_info($sha1, 1);
920 error "unknown commit"; # just in case
923 # git revert will fail on merge commits. Add a nice message.
924 if (exists $commits[0]->{parents} &&
925 @{$commits[0]->{parents}} > 1) {
926 error gettext("you are not allowed to revert a merge");
929 git_parse_changes(1, @commits);
934 # Try to revert the given rev; returns undef on _success_.
936 my ($sha1) = $rev =~ /^($sha1_pattern)$/; # untaint
938 if (run_or_non('git', 'revert', '--no-commit', $sha1)) {
942 run_or_die('git', 'reset', '--hard');
943 return sprintf(gettext("Failed to revert commit %s"), $sha1);