2 # Git backend for IkiWiki.
3 # Copyright 2006 Recai Oktaş <roktas@debian.org>
5 # Licensed under the same terms as IkiWiki.
13 my $origin_branch = 'origin'; # Git ref for main repository
14 my $master_branch = 'master'; # Working branch
15 my $sha1_pattern = qr/[0-9a-fA-F]{40}/; # pattern to validate Git sha1sums
16 my $dummy_commit_msg = 'dummy commit'; # will be used in all dummy commits
17 # and skipped in recent changes list
18 my $web_commit_msg = qr/^web commit by (\w+):?(.*)/; # pattern for web commits
20 sub _safe_git (&@) { #{{{
21 # Start a child process safely without resorting /bin/sh.
22 # Return command output or success state (in scalar context).
24 my ($error_handler, @cmdline) = @_;
26 my $pid = open my $OUT, "-|";
28 error("Cannot fork: $!") if !defined $pid;
32 open STDERR, ">&STDOUT"
33 or error("Cannot dup STDOUT: $!");
34 # Git commands want to be in wc.
36 or error("Cannot chdir to $config{srcdir}: $!");
37 exec @cmdline or error("Cannot exec '@cmdline': $!");
49 ($error_handler || sub { })->("'@cmdline' failed: $!") if $?;
51 return wantarray ? @lines : ($? == 0);
53 # Convenient wrappers.
54 sub run_or_die ($@) { _safe_git(\&IkiWiki::error, @_) }
55 sub run_or_cry ($@) { _safe_git(sub { warn @_ }, @_) }
56 sub run_or_non ($@) { _safe_git(undef, @_) }
59 sub _merge_past ($$$) { #{{{
60 # Unlike with Subversion, Git cannot make a 'svn merge -rN:M file'.
61 # Git merge commands work with the committed changes, except in the
62 # implicit case of '-m' of git-checkout(1). So we should invent a
63 # kludge here. In principle, we need to create a throw-away branch
64 # in preparing for the merge itself. Since branches are cheap (and
65 # branching is fast), this shouldn't cost high.
67 # The main problem is the presence of _uncommitted_ local changes. One
68 # possible approach to get rid of this situation could be that we first
69 # make a temporary commit in the master branch and later restore the
70 # initial state (this is possible since Git has the ability to undo a
71 # commit, i.e. 'git-reset --soft HEAD^'). The method can be summarized
74 # - create a diff of HEAD:current-sha1
76 # - create a dummy branch and switch to it
77 # - rewind to past (reset --hard to the current-sha1)
78 # - apply the diff and commit
79 # - switch to master and do the merge with the dummy branch
80 # - make a soft reset (undo the last commit of master)
82 # The above method has some drawbacks: (1) it needs a redundant commit
83 # just to get rid of local changes, (2) somewhat slow because of the
84 # required system forks. Until someone points a more straight method
85 # (which I would be grateful) I have implemented an alternative method.
86 # In this approach, we hide all the modified files from Git by renaming
87 # them (using the 'rename' builtin) and later restore those files in
88 # the throw-away branch (that is, we put the files themselves instead
89 # of applying a patch).
91 my ($sha1, $file, $message) = @_;
93 my @undo; # undo stack for cleanup in case of an error
94 my $conflict; # file content with conflict markers
97 # Hide local changes from Git by renaming the modified file.
98 # Relative paths must be converted to absolute for renaming.
99 my ($target, $hidden) = (
100 "$config{srcdir}/${file}", "$config{srcdir}/${file}.${sha1}"
102 rename($target, $hidden)
103 or error("rename '$target' to '$hidden' failed: $!");
104 # Ensure to restore the renamed file on error.
106 return if ! -e "$hidden"; # already renamed
107 rename($hidden, $target)
108 or debug("rename '$hidden' to '$target' failed: $!");
111 my $branch = "throw_away_${sha1}"; # supposed to be unique
113 # Create a throw-away branch and rewind backward.
114 push @undo, sub { run_or_cry('git-branch', '-D', $branch) };
115 run_or_die('git-branch', $branch, $sha1);
117 # Switch to throw-away branch for the merge operation.
119 if (!run_or_cry('git-checkout', $master_branch)) {
120 run_or_cry('git-checkout','-f',$master_branch);
123 run_or_die('git-checkout', $branch);
125 # Put the modified file in _this_ branch.
126 rename($hidden, $target)
127 or error("rename '$hidden' to '$target' failed: $!");
129 # _Silently_ commit all modifications in the current branch.
130 run_or_non('git-commit', '-m', $message, '-a');
131 # ... and re-switch to master.
132 run_or_die('git-checkout', $master_branch);
134 # Attempt to merge without complaining.
135 if (!run_or_non('git-pull', '--no-commit', '.', $branch)) {
136 $conflict = readfile($target);
137 run_or_die('git-reset', '--hard');
142 # Process undo stack (in reverse order). By policy cleanup
143 # actions should normally print a warning on failure.
144 while (my $handle = pop @undo) {
148 error("Git merge failed!\n$failure\n") if $failure;
153 sub _parse_diff_tree (@) { #{{{
154 # Parse the raw diff tree chunk and return the info hash.
155 # See git-diff-tree(1) for the syntax.
160 return if !defined @{ $dt_ref } || !length @{ $dt_ref }[0];
165 HEADER: while (my $line = shift @{ $dt_ref }) {
166 return if $line !~ m/^diff-tree (\S+)/;
173 # Identification lines for the commit.
174 IDENT: while (my $line = shift @{ $dt_ref }) {
175 # Regexps are semi-stolen from gitweb.cgi.
176 if ($line =~ m/^tree ([0-9a-fA-F]{40})$/) {
178 } elsif ($line =~ m/^parent ([0-9a-fA-F]{40})$/) {
179 # XXX: collecting in reverse order
180 push @{ $ci{'parents'} }, $1;
181 } elsif ($line =~ m/^(author|committer) (.*) ([0-9]+) (.*)$/) {
182 my ($who, $name, $epoch, $tz) =
186 $ci{ "${who}_epoch" } = $epoch;
187 $ci{ "${who}_tz" } = $tz;
189 if ($name =~ m/^([^<]+) <([^@]+)/) {
190 my ($fullname, $username) = ($1, $2);
191 $ci{"${who}_fullname"} = $fullname;
192 $ci{"${who}_username"} = $username;
194 $ci{"${who}_fullname"} =
195 $ci{"${who}_username"} = $name;
197 } elsif ($line =~ m/^$/) {
198 # Trailing empty line signals next section.
203 error("No 'tree' or 'parents' seen in diff-tree output")
204 if !defined $ci{'tree'} || !defined $ci{'parents'};
206 $ci{'parent'} = @{ $ci{'parents'} }[0];
209 COMMENT: while (my $line = shift @{ $dt_ref }) {
210 if ($line =~ m/^$/) {
211 # Trailing empty line signals next section.
215 push @{ $ci{'comment'} }, $line;
219 FILE: while (my $line = shift @{ $dt_ref }) {
221 :([0-7]{6})[ ] # from mode
222 ([0-7]{6})[ ] # to mode
223 ([0-9a-fA-F]{40})[ ] # from sha1
224 ([0-9a-fA-F]{40})[ ] # to sha1
226 ([0-9]{0,3})\t # similarity
229 my ($sha1_from, $sha1_to, $file) =
232 if ($file =~ m/^"(.*)"$/) {
233 ($file=$1) =~ s/\\([0-7]{1,3})/chr(oct($1))/eg;
236 push @{ $ci{'details'} }, {
238 'sha1_from' => $sha1_from,
239 'sha1_to' => $sha1_to,
247 error("No detail in diff-tree output") if !defined $ci{'details'};
252 sub git_commit_info (;$$) { #{{{
253 # Return an array of commit info hashes of num commits (default: 1)
254 # starting from the given sha1sum (default: HEAD).
256 my ($sha1, $num) = @_;
261 run_or_die(qq{git-rev-list --max-count=$num $sha1 |
262 git-diff-tree --stdin --pretty=raw -c -M -r});
265 while (my $parsed = _parse_diff_tree(\@raw_lines)) {
269 return wantarray ? @ci : $ci[0];
272 sub git_sha1 (;$) { #{{{
273 # Return head sha1sum (of given file).
275 my $file = shift || q{--};
277 my ($sha1) = run_or_die('git-rev-list', '--max-count=1', 'HEAD', $file);
278 ($sha1) = $sha1 =~ m/($sha1_pattern)/; # sha1sum is untainted now
279 debug("Empty sha1sum for '$file'.") if !length $sha1;
283 sub rcs_update () { #{{{
284 # Update working directory.
286 run_or_cry('git-pull', $origin_branch);
289 sub rcs_prepedit ($) { #{{{
290 # Return the commit sha1sum of the file when editing begins.
291 # This will be later used in rcs_commit if a merge is required.
295 my $sha1 = git_sha1($file);
296 return defined $sha1 ? $sha1 : q{};
299 sub rcs_commit ($$$) { #{{{
300 # Try to commit the page; returns undef on _success_ and
301 # a version of the page with the rcs's conflict markers on
304 my ($file, $message, $rcstoken) = @_;
306 # XXX: Wiki directory is in the unlocked state when starting this
307 # action. But it takes time for a Git process to finish its job
308 # (especially if a merge required), so we must re-lock to prevent
309 # race conditions. Only when the time of the real commit action
310 # (i.e. git-push(1)) comes, we'll unlock the directory.
313 # Check to see if the page has been changed by someone else since
314 # rcs_prepedit was called.
315 my $cur = git_sha1($file);
316 my ($prev) = $rcstoken =~ m/^$sha1_pattern$/; # untaint
318 if (defined $cur && defined $prev && $cur ne $prev) {
319 my $conflict = _merge_past($prev, $file, $dummy_commit_msg);
320 return $conflict if defined $conflict;
323 # git-commit(1) returns non-zero if file has not been really changed.
324 # so we should ignore its exit status (hence run_or_non).
325 if (run_or_non('git-commit', '-m', $message, '-i', $file)) {
327 run_or_cry('git-push', $origin_branch);
330 return undef; # success
333 sub rcs_add ($) { # {{{
334 # Add file to archive.
338 run_or_cry('git-add', $file);
341 sub rcs_recentchanges ($) { #{{{
342 # List of recent changes.
346 eval q{use CGI 'escapeHTML'};
347 eval q{use Date::Parse};
348 eval q{use Time::Duration};
350 my ($sha1, $type, $when, $diffurl, $user, @pages, @message, @rets);
351 INFO: foreach my $ci (git_commit_info('HEAD', $num)) {
352 my $title = @{ $ci->{'comment'} }[0];
354 # Skip redundant commits.
355 next INFO if ($title eq $dummy_commit_msg);
357 $sha1 = $ci->{'sha1'};
359 $when = concise(ago(time - $ci->{'author_epoch'}));
361 foreach my $bit (@{ $ci->{'details'} }) {
362 my $diffurl = $config{'diffurl'};
363 my $file = $bit->{'file'};
365 $diffurl =~ s/\[\[file\]\]/$file/go;
366 $diffurl =~ s/\[\[sha1_parent\]\]/$ci->{'parent'}/go;
367 $diffurl =~ s/\[\[sha1_from\]\]/$bit->{'sha1_from'}/go;
368 $diffurl =~ s/\[\[sha1_to\]\]/$bit->{'sha1_to'}/go;
371 link => htmllink("", pagename($file), 1),
376 push @message, { line => escapeHTML($title) };
378 if (defined $message[0] &&
379 $message[0]->{line} =~ m/$web_commit_msg/) {
381 $message[0]->{line} = $2;
384 $user = $ci->{'author_username'};
389 user => htmllink("", $user, 1),
392 message => [@message],
396 $sha1 = $type = $when = $diffurl = $user = undef;
397 @pages = @message = ();
403 sub rcs_notify () { #{{{
404 # Send notification mail to subscribed users.
406 # In usual Git usage, hooks/update script is presumed to send
407 # notification mails (see git-receive-pack(1)). But we prefer
408 # hooks/post-update to support IkiWiki commits coming from a
409 # cloned repository (through command line) because post-update
410 # is called _after_ each ref in repository is updated (update
411 # hook is called _before_ the repository is updated). Since
412 # post-update hook does not accept command line arguments, we
413 # don't have an $ENV variable in this function.
415 # Here, we rely on a simple fact: we can extract all parts of the
416 # notification content by parsing the "HEAD" commit (which also
417 # triggers a refresh of IkiWiki pages) and we can obtain the diff
418 # by comparing HEAD and HEAD^ (the previous commit).
420 my $sha1 = 'HEAD'; # the commit which triggers this action
422 my $ci = git_commit_info($sha1);
424 warn "Cannot parse info for '$sha1' commit";
428 my @changed_pages = map { $_->{'file'} } @{ $ci->{'details'} };
430 my ($user, $message);
431 if (@{ $ci->{'comment'} }[0] =~ m/$web_commit_msg/) {
435 $user = $ci->{'author_username'};
436 $message = join "\n", @{ $ci->{'comment'} };
439 require IkiWiki::UserInfo;
440 my @email_recipients = commit_notify_list($user, @changed_pages);
441 return if !@email_recipients;
443 # TODO: if a commit spans multiple pages, this will send
444 # subscribers a diff that might contain pages they did not
445 # sign up for. Should separate the diff per page and
446 # reassemble into one mail with just the pages subscribed to.
447 my $diff = join "\n", run_or_die('git-diff', "${sha1}^", $sha1);
449 my $subject = "$config{wikiname} update of ";
450 if (@changed_pages > 2) {
451 $subject .= "$changed_pages[0] $changed_pages[1] etc";
453 $subject .= join " ", @changed_pages;
455 $subject .= " by $user";
457 my $template = HTML::Template->new(
458 filename => "$config{templatedir}/notifymail.tmpl"
461 wikiname => $config{wikiname},
467 eval q{use Mail::Sendmail};
468 foreach my $email (@email_recipients) {
471 From => "$config{wikiname} <$config{adminemail}>",
473 Message => $template->output,
474 ) or error("Failed to send update notification mail: $!");
478 sub rcs_getctime ($) { #{{{
479 # Get the ctime of file.
483 my $sha1 = git_sha1($file);
484 my $ci = git_commit_info($sha1);
485 my $ctime = $ci->{'author_epoch'};
486 debug("ctime for '$file': ". localtime($ctime) . "\n");