4 # Jérémie Nikaes <jeremie.nikaes@ensimag.imag.fr>
5 # Arnaud Lacurie <arnaud.lacurie@ensimag.imag.fr>
6 # Claire Fousse <claire.fousse@ensimag.imag.fr>
7 # David Amouyal <david.amouyal@ensimag.imag.fr>
8 # Matthieu Moy <matthieu.moy@grenoble-inp.fr>
9 # License: GPL v2 or later
11 # Gateway between Git and MediaWiki.
12 # https://github.com/Bibzball/Git-Mediawiki/wiki
16 # - Several strategies are provided to fetch modifications from the
17 # wiki, but no automatic heuristics is provided, the user has
18 # to understand and chose which strategy is appropriate for him.
20 # - Git renames could be turned into MediaWiki renames (see TODO
23 # - No way to import "one page, and all pages included in it"
25 # - Multiple remote MediaWikis have not been very well tested.
29 use DateTime::Format::ISO8601;
31 # By default, use UTF-8 to communicate with Git and the user
32 binmode STDERR, ":utf8";
33 binmode STDOUT, ":utf8";
40 # Mediawiki filenames can contain forward slashes. This variable decides by which pattern they should be replaced
41 use constant SLASH_REPLACEMENT => "%2F";
43 # It's not always possible to delete pages (may require some
44 # priviledges). Deleted pages are replaced with this content.
45 use constant DELETED_CONTENT => "[[Category:Deleted]]\n";
47 # It's not possible to create empty pages. New empty files in Git are
48 # sent with this content instead.
49 use constant EMPTY_CONTENT => "<!-- empty page -->\n";
51 # used to reflect file creation or deletion in diff.
52 use constant NULL_SHA1 => "0000000000000000000000000000000000000000";
54 # Used on Git's side to reflect empty edit messages on the wiki
55 use constant EMPTY_MESSAGE => '*Empty MediaWiki Message*';
57 my $remotename = $ARGV[0];
60 # Accept both space-separated and multiple keys in config file.
61 # Spaces should be written as _ anyway because we'll use chomp.
62 my @tracked_pages = split(/[ \n]/, run_git("config --get-all remote.". $remotename .".pages"));
63 chomp(@tracked_pages);
65 # Just like @tracked_pages, but for MediaWiki categories.
66 my @tracked_categories = split(/[ \n]/, run_git("config --get-all remote.". $remotename .".categories"));
67 chomp(@tracked_categories);
69 # Import media files on pull
70 my $import_media = run_git("config --get --bool remote.". $remotename .".mediaimport");
72 $import_media = ($import_media eq "true");
74 # Export media files on push
75 my $export_media = run_git("config --get --bool remote.". $remotename .".mediaexport");
77 $export_media = !($export_media eq "false");
79 my $wiki_login = run_git("config --get remote.". $remotename .".mwLogin");
80 # Note: mwPassword is discourraged. Use the credential system instead.
81 my $wiki_passwd = run_git("config --get remote.". $remotename .".mwPassword");
82 my $wiki_domain = run_git("config --get remote.". $remotename .".mwDomain");
87 # Import only last revisions (both for clone and fetch)
88 my $shallow_import = run_git("config --get --bool remote.". $remotename .".shallow");
89 chomp($shallow_import);
90 $shallow_import = ($shallow_import eq "true");
92 # Fetch (clone and pull) by revisions instead of by pages. This behavior
93 # is more efficient when we have a wiki with lots of pages and we fetch
94 # the revisions quite often so that they concern only few pages.
96 # - by_rev: perform one query per new revision on the remote wiki
97 # - by_page: query each tracked page for new revision
98 my $fetch_strategy = run_git("config --get remote.$remotename.fetchStrategy");
99 unless ($fetch_strategy) {
100 $fetch_strategy = run_git("config --get mediawiki.fetchStrategy");
102 chomp($fetch_strategy);
103 unless ($fetch_strategy) {
104 $fetch_strategy = "by_page";
107 # Dumb push: don't update notes and mediawiki ref to reflect the last push.
109 # Configurable with mediawiki.dumbPush, or per-remote with
110 # remote.<remotename>.dumbPush.
112 # This means the user will have to re-import the just-pushed
113 # revisions. On the other hand, this means that the Git revisions
114 # corresponding to MediaWiki revisions are all imported from the wiki,
115 # regardless of whether they were initially created in Git or from the
116 # web interface, hence all users will get the same history (i.e. if
117 # the push from Git to MediaWiki loses some information, everybody
118 # will get the history with information lost). If the import is
119 # deterministic, this means everybody gets the same sha1 for each
120 # MediaWiki revision.
121 my $dumb_push = run_git("config --get --bool remote.$remotename.dumbPush");
122 unless ($dumb_push) {
123 $dumb_push = run_git("config --get --bool mediawiki.dumbPush");
126 $dumb_push = ($dumb_push eq "true");
128 my $wiki_name = $url;
129 $wiki_name =~ s/[^\/]*:\/\///;
130 # If URL is like http://user:password@example.com/, we clearly don't
131 # want the password in $wiki_name. While we're there, also remove user
132 # and '@' sign, to avoid author like MWUser@HTTPUser@host.com
133 $wiki_name =~ s/^.*@//;
141 if (defined($cmd[0])) {
143 if ($cmd[0] eq "capabilities") {
144 die("Too many arguments for capabilities") unless (!defined($cmd[1]));
146 } elsif ($cmd[0] eq "list") {
147 die("Too many arguments for list") unless (!defined($cmd[2]));
149 } elsif ($cmd[0] eq "import") {
150 die("Invalid arguments for import") unless ($cmd[1] ne "" && !defined($cmd[2]));
152 } elsif ($cmd[0] eq "option") {
153 die("Too many arguments for option") unless ($cmd[1] ne "" && $cmd[2] ne "" && !defined($cmd[3]));
154 mw_option($cmd[1],$cmd[2]);
155 } elsif ($cmd[0] eq "push") {
158 print STDERR "Unknown command. Aborting...\n";
162 # blank line: we should terminate
166 BEGIN { $| = 1 } # flush STDOUT, to make sure the previous
167 # command is fully processed.
170 ########################## Functions ##############################
172 ## credential API management (generic functions)
174 sub credential_from_url {
176 my $parsed = URI->new($url);
179 if ($parsed->scheme) {
180 $credential{protocol} = $parsed->scheme;
183 $credential{host} = $parsed->host;
186 $credential{path} = $parsed->path;
188 if ($parsed->userinfo) {
189 if ($parsed->userinfo =~ /([^:]*):(.*)/) {
190 $credential{username} = $1;
191 $credential{password} = $2;
193 $credential{username} = $parsed->userinfo;
200 sub credential_read {
205 my ($key, $value) = /([^=]*)=(.*)/;
206 if (not defined $key) {
207 die "ERROR receiving response from git credential $op:\n$_\n";
209 $credential{$key} = $value;
214 sub credential_write {
215 my $credential = shift;
217 while (my ($key, $value) = each(%$credential) ) {
219 print $writer "$key=$value\n";
226 my $credential = shift;
227 my $pid = open2(my $reader, my $writer, "git credential $op");
228 credential_write($credential, $writer);
233 %$credential = credential_read($reader, $op);
236 die "ERROR while running git credential $op:\n$_";
241 my $child_exit_status = $? >> 8;
242 if ($child_exit_status != 0) {
243 die "'git credential $op' failed with code $child_exit_status.";
247 # MediaWiki API instance, created lazily.
250 sub mw_connect_maybe {
254 $mediawiki = MediaWiki::API->new;
255 $mediawiki->{config}->{api_url} = "$url/api.php";
257 my %credential = credential_from_url($url);
258 $credential{username} = $wiki_login;
259 $credential{password} = $wiki_passwd;
260 credential_run("fill", \%credential);
261 my $request = {lgname => $credential{username},
262 lgpassword => $credential{password},
263 lgdomain => $wiki_domain};
264 if ($mediawiki->login($request)) {
265 credential_run("approve", \%credential);
266 print STDERR "Logged in mediawiki user \"$credential{username}\".\n";
268 print STDERR "Failed to log in mediawiki user \"$credential{username}\" on $url\n";
269 print STDERR " (error " .
270 $mediawiki->{error}->{code} . ': ' .
271 $mediawiki->{error}->{details} . ")\n";
272 credential_run("reject", \%credential);
278 ## Functions for listing pages on the remote wiki
279 sub get_mw_tracked_pages {
281 get_mw_page_list(\@tracked_pages, $pages);
284 sub get_mw_page_list {
285 my $page_list = shift;
287 my @some_pages = @$page_list;
288 while (@some_pages) {
290 if ($#some_pages < $last) {
291 $last = $#some_pages;
293 my @slice = @some_pages[0..$last];
294 get_mw_first_pages(\@slice, $pages);
295 @some_pages = @some_pages[51..$#some_pages];
299 sub get_mw_tracked_categories {
301 foreach my $category (@tracked_categories) {
302 if (index($category, ':') < 0) {
303 # Mediawiki requires the Category
304 # prefix, but let's not force the user
306 $category = "Category:" . $category;
308 my $mw_pages = $mediawiki->list( {
310 list => 'categorymembers',
311 cmtitle => $category,
313 || die $mediawiki->{error}->{code} . ': '
314 . $mediawiki->{error}->{details};
315 foreach my $page (@{$mw_pages}) {
316 $pages->{$page->{title}} = $page;
321 sub get_mw_all_pages {
323 # No user-provided list, get the list of pages from the API.
324 my $mw_pages = $mediawiki->list({
329 if (!defined($mw_pages)) {
330 print STDERR "fatal: could not get the list of wiki pages.\n";
331 print STDERR "fatal: '$url' does not appear to be a mediawiki\n";
332 print STDERR "fatal: make sure '$url/api.php' is a valid page.\n";
335 foreach my $page (@{$mw_pages}) {
336 $pages->{$page->{title}} = $page;
340 # queries the wiki for a set of pages. Meant to be used within a loop
341 # querying the wiki for slices of page list.
342 sub get_mw_first_pages {
343 my $some_pages = shift;
344 my @some_pages = @{$some_pages};
348 # pattern 'page1|page2|...' required by the API
349 my $titles = join('|', @some_pages);
351 my $mw_pages = $mediawiki->api({
355 if (!defined($mw_pages)) {
356 print STDERR "fatal: could not query the list of wiki pages.\n";
357 print STDERR "fatal: '$url' does not appear to be a mediawiki\n";
358 print STDERR "fatal: make sure '$url/api.php' is a valid page.\n";
361 while (my ($id, $page) = each(%{$mw_pages->{query}->{pages}})) {
363 print STDERR "Warning: page $page->{title} not found on wiki\n";
365 $pages->{$page->{title}} = $page;
370 # Get the list of pages to be fetched according to configuration.
374 print STDERR "Listing pages on remote wiki...\n";
376 my %pages; # hash on page titles to avoid duplicates
378 if (@tracked_pages) {
380 # The user provided a list of pages titles, but we
381 # still need to query the API to get the page IDs.
382 get_mw_tracked_pages(\%pages);
384 if (@tracked_categories) {
386 get_mw_tracked_categories(\%pages);
388 if (!$user_defined) {
389 get_mw_all_pages(\%pages);
392 print STDERR "Getting media files for selected pages...\n";
394 get_linked_mediafiles(\%pages);
396 get_all_mediafiles(\%pages);
399 print STDERR (scalar keys %pages) . " pages found.\n";
403 # usage: $out = run_git("command args");
404 # $out = run_git("command args", "raw"); # don't interpret output as UTF-8.
407 my $encoding = (shift || "encoding(UTF-8)");
408 open(my $git, "-|:$encoding", "git " . $args);
409 my $res = do { local $/; <$git> };
416 sub get_all_mediafiles {
418 # Attach list of all pages for media files from the API,
419 # they are in a different namespace, only one namespace
420 # can be queried at the same moment
421 my $mw_pages = $mediawiki->list({
424 apnamespace => get_mw_namespace_id("File"),
427 if (!defined($mw_pages)) {
428 print STDERR "fatal: could not get the list of pages for media files.\n";
429 print STDERR "fatal: '$url' does not appear to be a mediawiki\n";
430 print STDERR "fatal: make sure '$url/api.php' is a valid page.\n";
433 foreach my $page (@{$mw_pages}) {
434 $pages->{$page->{title}} = $page;
438 sub get_linked_mediafiles {
440 my @titles = map $_->{title}, values(%{$pages});
442 # The query is split in small batches because of the MW API limit of
443 # the number of links to be returned (500 links max).
446 if ($#titles < $batch) {
449 my @slice = @titles[0..$batch];
451 # pattern 'page1|page2|...' required by the API
452 my $mw_titles = join('|', @slice);
454 # Media files could be included or linked from
455 # a page, get all related
458 prop => 'links|images',
459 titles => $mw_titles,
460 plnamespace => get_mw_namespace_id("File"),
463 my $result = $mediawiki->api($query);
465 while (my ($id, $page) = each(%{$result->{query}->{pages}})) {
467 if (defined($page->{links})) {
468 my @link_titles = map $_->{title}, @{$page->{links}};
469 push(@media_titles, @link_titles);
471 if (defined($page->{images})) {
472 my @image_titles = map $_->{title}, @{$page->{images}};
473 push(@media_titles, @image_titles);
476 get_mw_page_list(\@media_titles, $pages);
480 @titles = @titles[($batch+1)..$#titles];
484 sub get_mw_mediafile_for_page_revision {
485 # Name of the file on Wiki, with the prefix.
486 my $filename = shift;
487 my $timestamp = shift;
490 # Search if on a media file with given timestamp exists on
491 # MediaWiki. In that case download the file.
495 titles => "File:" . $filename,
496 iistart => $timestamp,
498 iiprop => 'timestamp|archivename|url',
501 my $result = $mediawiki->api($query);
503 my ($fileid, $file) = each( %{$result->{query}->{pages}} );
504 # If not defined it means there is no revision of the file for
506 if (defined($file->{imageinfo})) {
507 $mediafile{title} = $filename;
509 my $fileinfo = pop(@{$file->{imageinfo}});
510 $mediafile{timestamp} = $fileinfo->{timestamp};
511 # Mediawiki::API's download function doesn't support https URLs
512 # and can't download old versions of files.
513 print STDERR "\tDownloading file $mediafile{title}, version $mediafile{timestamp}\n";
514 $mediafile{content} = download_mw_mediafile($fileinfo->{url});
519 sub download_mw_mediafile {
522 my $response = $mediawiki->{ua}->get($url);
523 if ($response->code == 200) {
524 return $response->decoded_content;
526 print STDERR "Error downloading mediafile from :\n";
527 print STDERR "URL: $url\n";
528 print STDERR "Server response: " . $response->code . " " . $response->message . "\n";
533 sub get_last_local_revision {
534 # Get note regarding last mediawiki revision
535 my $note = run_git("notes --ref=$remotename/mediawiki show refs/mediawiki/$remotename/master 2>/dev/null");
536 my @note_info = split(/ /, $note);
538 my $lastrevision_number;
539 if (!(defined($note_info[0]) && $note_info[0] eq "mediawiki_revision:")) {
540 print STDERR "No previous mediawiki revision found";
541 $lastrevision_number = 0;
543 # Notes are formatted : mediawiki_revision: #number
544 $lastrevision_number = $note_info[1];
545 chomp($lastrevision_number);
546 print STDERR "Last local mediawiki revision found is $lastrevision_number";
548 return $lastrevision_number;
551 # Remember the timestamp corresponding to a revision id.
554 # Get the last remote revision without taking in account which pages are
555 # tracked or not. This function makes a single request to the wiki thus
556 # avoid a loop onto all tracked pages. This is useful for the fetch-by-rev
558 sub get_last_global_remote_rev {
563 list => 'recentchanges',
568 my $result = $mediawiki->api($query);
569 return $result->{query}->{recentchanges}[0]->{revid};
572 # Get the last remote revision concerning the tracked pages and the tracked
574 sub get_last_remote_revision {
577 my %pages_hash = get_mw_pages();
578 my @pages = values(%pages_hash);
582 print STDERR "Getting last revision id on tracked pages...\n";
584 foreach my $page (@pages) {
585 my $id = $page->{pageid};
590 rvprop => 'ids|timestamp',
594 my $result = $mediawiki->api($query);
596 my $lastrev = pop(@{$result->{query}->{pages}->{$id}->{revisions}});
598 $basetimestamps{$lastrev->{revid}} = $lastrev->{timestamp};
600 $max_rev_num = ($lastrev->{revid} > $max_rev_num ? $lastrev->{revid} : $max_rev_num);
603 print STDERR "Last remote revision found is $max_rev_num.\n";
607 # Clean content before sending it to MediaWiki
608 sub mediawiki_clean {
610 my $page_created = shift;
611 # Mediawiki does not allow blank space at the end of a page and ends with a single \n.
612 # This function right trims a string and adds a \n at the end to follow this rule
614 if ($string eq "" && $page_created) {
615 # Creating empty pages is forbidden.
616 $string = EMPTY_CONTENT;
621 # Filter applied on MediaWiki data before adding them to Git
622 sub mediawiki_smudge {
624 if ($string eq EMPTY_CONTENT) {
627 # This \n is important. This is due to mediawiki's way to handle end of files.
631 sub mediawiki_clean_filename {
632 my $filename = shift;
633 $filename =~ s/@{[SLASH_REPLACEMENT]}/\//g;
634 # [, ], |, {, and } are forbidden by MediaWiki, even URL-encoded.
635 # Do a variant of URL-encoding, i.e. looks like URL-encoding,
636 # but with _ added to prevent MediaWiki from thinking this is
637 # an actual special character.
638 $filename =~ s/[\[\]\{\}\|]/sprintf("_%%_%x", ord($&))/ge;
639 # If we use the uri escape before
640 # we should unescape here, before anything
645 sub mediawiki_smudge_filename {
646 my $filename = shift;
647 $filename =~ s/\//@{[SLASH_REPLACEMENT]}/g;
648 $filename =~ s/ /_/g;
649 # Decode forbidden characters encoded in mediawiki_clean_filename
650 $filename =~ s/_%_([0-9a-fA-F][0-9a-fA-F])/sprintf("%c", hex($1))/ge;
656 print STDOUT "data ", bytes::length($content), "\n", $content;
659 sub literal_data_raw {
660 # Output possibly binary content.
662 # Avoid confusion between size in bytes and in characters
663 utf8::downgrade($content);
664 binmode STDOUT, ":raw";
665 print STDOUT "data ", bytes::length($content), "\n", $content;
666 binmode STDOUT, ":utf8";
669 sub mw_capabilities {
670 # Revisions are imported to the private namespace
671 # refs/mediawiki/$remotename/ by the helper and fetched into
672 # refs/remotes/$remotename later by fetch.
673 print STDOUT "refspec refs/heads/*:refs/mediawiki/$remotename/*\n";
674 print STDOUT "import\n";
675 print STDOUT "list\n";
676 print STDOUT "push\n";
681 # MediaWiki do not have branches, we consider one branch arbitrarily
682 # called master, and HEAD pointing to it.
683 print STDOUT "? refs/heads/master\n";
684 print STDOUT "\@refs/heads/master HEAD\n";
689 print STDERR "remote-helper command 'option $_[0]' not yet implemented\n";
690 print STDOUT "unsupported\n";
693 sub fetch_mw_revisions_for_page {
696 my $fetch_from = shift;
703 rvstartid => $fetch_from,
709 # Get 500 revisions at a time due to the mediawiki api limit
711 my $result = $mediawiki->api($query);
713 # Parse each of those 500 revisions
714 foreach my $revision (@{$result->{query}->{pages}->{$id}->{revisions}}) {
716 $page_rev_ids->{pageid} = $page->{pageid};
717 $page_rev_ids->{revid} = $revision->{revid};
718 push(@page_revs, $page_rev_ids);
721 last unless $result->{'query-continue'};
722 $query->{rvstartid} = $result->{'query-continue'}->{revisions}->{rvstartid};
724 if ($shallow_import && @page_revs) {
725 print STDERR " Found 1 revision (shallow import).\n";
726 @page_revs = sort {$b->{revid} <=> $a->{revid}} (@page_revs);
727 return $page_revs[0];
729 print STDERR " Found ", $revnum, " revision(s).\n";
733 sub fetch_mw_revisions {
734 my $pages = shift; my @pages = @{$pages};
735 my $fetch_from = shift;
739 foreach my $page (@pages) {
740 my $id = $page->{pageid};
742 print STDERR "page $n/", scalar(@pages), ": ". $page->{title} ."\n";
744 my @page_revs = fetch_mw_revisions_for_page($page, $id, $fetch_from);
745 @revisions = (@page_revs, @revisions);
748 return ($n, @revisions);
751 sub import_file_revision {
753 my %commit = %{$commit};
754 my $full_import = shift;
756 my $mediafile = shift;
759 %mediafile = %{$mediafile};
762 my $title = $commit{title};
763 my $comment = $commit{comment};
764 my $content = $commit{content};
765 my $author = $commit{author};
766 my $date = $commit{date};
768 print STDOUT "commit refs/mediawiki/$remotename/master\n";
769 print STDOUT "mark :$n\n";
770 print STDOUT "committer $author <$author\@$wiki_name> ", $date->epoch, " +0000\n";
771 literal_data($comment);
773 # If it's not a clone, we need to know where to start from
774 if (!$full_import && $n == 1) {
775 print STDOUT "from refs/mediawiki/$remotename/master^0\n";
777 if ($content ne DELETED_CONTENT) {
778 print STDOUT "M 644 inline $title.mw\n";
779 literal_data($content);
781 print STDOUT "M 644 inline $mediafile{title}\n";
782 literal_data_raw($mediafile{content});
786 print STDOUT "D $title.mw\n";
789 # mediawiki revision number in the git note
790 if ($full_import && $n == 1) {
791 print STDOUT "reset refs/notes/$remotename/mediawiki\n";
793 print STDOUT "commit refs/notes/$remotename/mediawiki\n";
794 print STDOUT "committer $author <$author\@$wiki_name> ", $date->epoch, " +0000\n";
795 literal_data("Note added by git-mediawiki during import");
796 if (!$full_import && $n == 1) {
797 print STDOUT "from refs/notes/$remotename/mediawiki^0\n";
799 print STDOUT "N inline :$n\n";
800 literal_data("mediawiki_revision: " . $commit{mw_revision});
804 # parse a sequence of
808 # (like batch sequence of import and sequence of push statements)
814 if ($line =~ m/^$cmd (.*)$/) {
816 } elsif ($line eq "\n") {
819 die("Invalid command in a '$cmd' batch: ". $_);
825 # multiple import commands can follow each other.
826 my @refs = (shift, get_more_refs("import"));
827 foreach my $ref (@refs) {
830 print STDOUT "done\n";
835 # The remote helper will call "import HEAD" and
836 # "import refs/heads/master".
837 # Since HEAD is a symbolic ref to master (by convention,
838 # followed by the output of the command "list" that we gave),
839 # we don't need to do anything in this case.
840 if ($ref eq "HEAD") {
846 print STDERR "Searching revisions...\n";
847 my $last_local = get_last_local_revision();
848 my $fetch_from = $last_local + 1;
849 if ($fetch_from == 1) {
850 print STDERR ", fetching from beginning.\n";
852 print STDERR ", fetching from here.\n";
856 if ($fetch_strategy eq "by_rev") {
857 print STDERR "Fetching & writing export data by revs...\n";
858 $n = mw_import_ref_by_revs($fetch_from);
859 } elsif ($fetch_strategy eq "by_page") {
860 print STDERR "Fetching & writing export data by pages...\n";
861 $n = mw_import_ref_by_pages($fetch_from);
863 print STDERR "fatal: invalid fetch strategy \"$fetch_strategy\".\n";
864 print STDERR "Check your configuration variables remote.$remotename.fetchStrategy and mediawiki.fetchStrategy\n";
868 if ($fetch_from == 1 && $n == 0) {
869 print STDERR "You appear to have cloned an empty MediaWiki.\n";
870 # Something has to be done remote-helper side. If nothing is done, an error is
871 # thrown saying that HEAD is refering to unknown object 0000000000000000000
872 # and the clone fails.
876 sub mw_import_ref_by_pages {
878 my $fetch_from = shift;
879 my %pages_hash = get_mw_pages();
880 my @pages = values(%pages_hash);
882 my ($n, @revisions) = fetch_mw_revisions(\@pages, $fetch_from);
884 @revisions = sort {$a->{revid} <=> $b->{revid}} @revisions;
885 my @revision_ids = map $_->{revid}, @revisions;
887 return mw_import_revids($fetch_from, \@revision_ids, \%pages_hash);
890 sub mw_import_ref_by_revs {
892 my $fetch_from = shift;
893 my %pages_hash = get_mw_pages();
895 my $last_remote = get_last_global_remote_rev();
896 my @revision_ids = $fetch_from..$last_remote;
897 return mw_import_revids($fetch_from, \@revision_ids, \%pages_hash);
900 # Import revisions given in second argument (array of integers).
901 # Only pages appearing in the third argument (hash indexed by page titles)
903 sub mw_import_revids {
904 my $fetch_from = shift;
905 my $revision_ids = shift;
910 my $last_timestamp = 0; # Placeholer in case $rev->timestamp is undefined
912 foreach my $pagerevid (@$revision_ids) {
913 # Count page even if we skip it, since we display
914 # $n/$total and $total includes skipped pages.
917 # fetch the content of the pages
921 rvprop => 'content|timestamp|comment|user|ids',
922 revids => $pagerevid,
925 my $result = $mediawiki->api($query);
928 die "Failed to retrieve modified page for revision $pagerevid";
931 if (defined($result->{query}->{badrevids}->{$pagerevid})) {
932 # The revision id does not exist on the remote wiki.
936 if (!defined($result->{query}->{pages})) {
937 die "Invalid revision $pagerevid.";
940 my @result_pages = values(%{$result->{query}->{pages}});
941 my $result_page = $result_pages[0];
942 my $rev = $result_pages[0]->{revisions}->[0];
944 my $page_title = $result_page->{title};
946 if (!exists($pages->{$page_title})) {
947 print STDERR "$n/", scalar(@$revision_ids),
948 ": Skipping revision #$rev->{revid} of $page_title\n";
955 $commit{author} = $rev->{user} || 'Anonymous';
956 $commit{comment} = $rev->{comment} || EMPTY_MESSAGE;
957 $commit{title} = mediawiki_smudge_filename($page_title);
958 $commit{mw_revision} = $rev->{revid};
959 $commit{content} = mediawiki_smudge($rev->{'*'});
961 if (!defined($rev->{timestamp})) {
964 $last_timestamp = $rev->{timestamp};
966 $commit{date} = DateTime::Format::ISO8601->parse_datetime($last_timestamp);
968 # Differentiates classic pages and media files.
969 my ($namespace, $filename) = $page_title =~ /^([^:]*):(.*)$/;
972 my $id = get_mw_namespace_id($namespace);
973 if ($id && $id == get_mw_namespace_id("File")) {
974 %mediafile = get_mw_mediafile_for_page_revision($filename, $rev->{timestamp});
977 # If this is a revision of the media page for new version
978 # of a file do one common commit for both file and media page.
979 # Else do commit only for that page.
980 print STDERR "$n/", scalar(@$revision_ids), ": Revision #$rev->{revid} of $commit{title}\n";
981 import_file_revision(\%commit, ($fetch_from == 1), $n_actual, \%mediafile);
987 sub error_non_fast_forward {
988 my $advice = run_git("config --bool advice.pushNonFastForward");
990 if ($advice ne "false") {
991 # Native git-push would show this after the summary.
992 # We can't ask it to display it cleanly, so print it
994 print STDERR "To prevent you from losing history, non-fast-forward updates were rejected\n";
995 print STDERR "Merge the remote changes (e.g. 'git pull') before pushing again. See the\n";
996 print STDERR "'Note about fast-forwards' section of 'git push --help' for details.\n";
998 print STDOUT "error $_[0] \"non-fast-forward\"\n";
1002 sub mw_upload_file {
1003 my $complete_file_name = shift;
1004 my $new_sha1 = shift;
1005 my $extension = shift;
1006 my $file_deleted = shift;
1007 my $summary = shift;
1009 my $path = "File:" . $complete_file_name;
1010 my %hashFiles = get_allowed_file_extensions();
1011 if (!exists($hashFiles{$extension})) {
1012 print STDERR "$complete_file_name is not a permitted file on this wiki.\n";
1013 print STDERR "Check the configuration of file uploads in your mediawiki.\n";
1016 # Deleting and uploading a file requires a priviledged user
1017 if ($file_deleted) {
1024 if (!$mediawiki->edit($query)) {
1025 print STDERR "Failed to delete file on remote wiki\n";
1026 print STDERR "Check your permissions on the remote site. Error code:\n";
1027 print STDERR $mediawiki->{error}->{code} . ':' . $mediawiki->{error}->{details};
1031 # Don't let perl try to interpret file content as UTF-8 => use "raw"
1032 my $content = run_git("cat-file blob $new_sha1", "raw");
1033 if ($content ne "") {
1035 $mediawiki->{config}->{upload_url} =
1036 "$url/index.php/Special:Upload";
1039 filename => $complete_file_name,
1040 comment => $summary,
1042 $complete_file_name,
1043 Content => $content],
1044 ignorewarnings => 1,
1047 } ) || die $mediawiki->{error}->{code} . ':'
1048 . $mediawiki->{error}->{details};
1049 my $last_file_page = $mediawiki->get_page({title => $path});
1050 $newrevid = $last_file_page->{revid};
1051 print STDERR "Pushed file: $new_sha1 - $complete_file_name.\n";
1053 print STDERR "Empty file $complete_file_name not pushed.\n";
1060 my $diff_info = shift;
1061 # $diff_info contains a string in this format:
1062 # 100644 100644 <sha1_of_blob_before_commit> <sha1_of_blob_now> <status>
1063 my @diff_info_split = split(/[ \t]/, $diff_info);
1065 # Filename, including .mw extension
1066 my $complete_file_name = shift;
1068 my $summary = shift;
1069 # MediaWiki revision number. Keep the previous one by default,
1070 # in case there's no edit to perform.
1071 my $oldrevid = shift;
1074 if ($summary eq EMPTY_MESSAGE) {
1078 my $new_sha1 = $diff_info_split[3];
1079 my $old_sha1 = $diff_info_split[2];
1080 my $page_created = ($old_sha1 eq NULL_SHA1);
1081 my $page_deleted = ($new_sha1 eq NULL_SHA1);
1082 $complete_file_name = mediawiki_clean_filename($complete_file_name);
1084 my ($title, $extension) = $complete_file_name =~ /^(.*)\.([^\.]*)$/;
1085 if (!defined($extension)) {
1088 if ($extension eq "mw") {
1089 my $ns = get_mw_namespace_id_for_page($complete_file_name);
1090 if ($ns && $ns == get_mw_namespace_id("File") && (!$export_media)) {
1091 print STDERR "Ignoring media file related page: $complete_file_name\n";
1092 return ($oldrevid, "ok");
1095 if ($page_deleted) {
1096 # Deleting a page usually requires
1097 # special priviledges. A common
1098 # convention is to replace the page
1099 # with this content instead:
1100 $file_content = DELETED_CONTENT;
1102 $file_content = run_git("cat-file blob $new_sha1");
1107 my $result = $mediawiki->edit( {
1109 summary => $summary,
1111 basetimestamp => $basetimestamps{$oldrevid},
1112 text => mediawiki_clean($file_content, $page_created),
1114 skip_encoding => 1 # Helps with names with accentuated characters
1117 if ($mediawiki->{error}->{code} == 3) {
1118 # edit conflicts, considered as non-fast-forward
1119 print STDERR 'Warning: Error ' .
1120 $mediawiki->{error}->{code} .
1121 ' from mediwiki: ' . $mediawiki->{error}->{details} .
1123 return ($oldrevid, "non-fast-forward");
1125 # Other errors. Shouldn't happen => just die()
1126 die 'Fatal: Error ' .
1127 $mediawiki->{error}->{code} .
1128 ' from mediwiki: ' . $mediawiki->{error}->{details};
1131 $newrevid = $result->{edit}->{newrevid};
1132 print STDERR "Pushed file: $new_sha1 - $title\n";
1133 } elsif ($export_media) {
1134 $newrevid = mw_upload_file($complete_file_name, $new_sha1,
1135 $extension, $page_deleted,
1138 print STDERR "Ignoring media file $title\n";
1140 $newrevid = ($newrevid or $oldrevid);
1141 return ($newrevid, "ok");
1145 # multiple push statements can follow each other
1146 my @refsspecs = (shift, get_more_refs("push"));
1148 for my $refspec (@refsspecs) {
1149 my ($force, $local, $remote) = $refspec =~ /^(\+)?([^:]*):([^:]*)$/
1150 or die("Invalid refspec for push. Expected <src>:<dst> or +<src>:<dst>");
1152 print STDERR "Warning: forced push not allowed on a MediaWiki.\n";
1155 print STDERR "Cannot delete remote branch on a MediaWiki\n";
1156 print STDOUT "error $remote cannot delete\n";
1159 if ($remote ne "refs/heads/master") {
1160 print STDERR "Only push to the branch 'master' is supported on a MediaWiki\n";
1161 print STDOUT "error $remote only master allowed\n";
1164 if (mw_push_revision($local, $remote)) {
1169 # Notify Git that the push is done
1172 if ($pushed && $dumb_push) {
1173 print STDERR "Just pushed some revisions to MediaWiki.\n";
1174 print STDERR "The pushed revisions now have to be re-imported, and your current branch\n";
1175 print STDERR "needs to be updated with these re-imported commits. You can do this with\n";
1177 print STDERR " git pull --rebase\n";
1182 sub mw_push_revision {
1184 my $remote = shift; # actually, this has to be "refs/heads/master" at this point.
1185 my $last_local_revid = get_last_local_revision();
1186 print STDERR ".\n"; # Finish sentence started by get_last_local_revision()
1187 my $last_remote_revid = get_last_remote_revision();
1188 my $mw_revision = $last_remote_revid;
1190 # Get sha1 of commit pointed by local HEAD
1191 my $HEAD_sha1 = run_git("rev-parse $local 2>/dev/null"); chomp($HEAD_sha1);
1192 # Get sha1 of commit pointed by remotes/$remotename/master
1193 my $remoteorigin_sha1 = run_git("rev-parse refs/remotes/$remotename/master 2>/dev/null");
1194 chomp($remoteorigin_sha1);
1196 if ($last_local_revid > 0 &&
1197 $last_local_revid < $last_remote_revid) {
1198 return error_non_fast_forward($remote);
1201 if ($HEAD_sha1 eq $remoteorigin_sha1) {
1206 # Get every commit in between HEAD and refs/remotes/origin/master,
1207 # including HEAD and refs/remotes/origin/master
1208 my @commit_pairs = ();
1209 if ($last_local_revid > 0) {
1210 my $parsed_sha1 = $remoteorigin_sha1;
1211 # Find a path from last MediaWiki commit to pushed commit
1212 print STDERR "Computing path from local to remote ...\n";
1213 my @local_ancestry = split(/\n/, run_git("rev-list --boundary --parents $local ^$parsed_sha1"));
1215 foreach my $line (@local_ancestry) {
1216 if (my ($child, $parents) = $line =~ m/^-?([a-f0-9]+) ([a-f0-9 ]+)/) {
1217 foreach my $parent (split(' ', $parents)) {
1218 $local_ancestry{$parent} = $child;
1220 } elsif (!$line =~ m/^([a-f0-9]+)/) {
1221 die "Unexpected output from git rev-list: $line";
1224 while ($parsed_sha1 ne $HEAD_sha1) {
1225 my $child = $local_ancestry{$parsed_sha1};
1227 printf STDERR "Cannot find a path in history from remote commit to last commit\n";
1228 return error_non_fast_forward($remote);
1230 push(@commit_pairs, [$parsed_sha1, $child]);
1231 $parsed_sha1 = $child;
1234 # No remote mediawiki revision. Export the whole
1235 # history (linearized with --first-parent)
1236 print STDERR "Warning: no common ancestor, pushing complete history\n";
1237 my $history = run_git("rev-list --first-parent --children $local");
1238 my @history = split('\n', $history);
1239 @history = @history[1..$#history];
1240 foreach my $line (reverse @history) {
1241 my @commit_info_split = split(/ |\n/, $line);
1242 push(@commit_pairs, \@commit_info_split);
1246 foreach my $commit_info_split (@commit_pairs) {
1247 my $sha1_child = @{$commit_info_split}[0];
1248 my $sha1_commit = @{$commit_info_split}[1];
1249 my $diff_infos = run_git("diff-tree -r --raw -z $sha1_child $sha1_commit");
1250 # TODO: we could detect rename, and encode them with a #redirect on the wiki.
1251 # TODO: for now, it's just a delete+add
1252 my @diff_info_list = split(/\0/, $diff_infos);
1253 # Keep the subject line of the commit message as mediawiki comment for the revision
1254 my $commit_msg = run_git("log --no-walk --format=\"%s\" $sha1_commit");
1257 while (@diff_info_list) {
1259 # git diff-tree -z gives an output like
1260 # <metadata>\0<filename1>\0
1261 # <metadata>\0<filename2>\0
1262 # and we've split on \0.
1263 my $info = shift(@diff_info_list);
1264 my $file = shift(@diff_info_list);
1265 ($mw_revision, $status) = mw_push_file($info, $file, $commit_msg, $mw_revision);
1266 if ($status eq "non-fast-forward") {
1267 # we may already have sent part of the
1268 # commit to MediaWiki, but it's too
1269 # late to cancel it. Stop the push in
1270 # the middle, but still give an
1271 # accurate error message.
1272 return error_non_fast_forward($remote);
1274 if ($status ne "ok") {
1275 die("Unknown error from mw_push_file()");
1278 unless ($dumb_push) {
1279 run_git("notes --ref=$remotename/mediawiki add -f -m \"mediawiki_revision: $mw_revision\" $sha1_commit");
1280 run_git("update-ref -m \"Git-MediaWiki push\" refs/mediawiki/$remotename/master $sha1_commit $sha1_child");
1284 print STDOUT "ok $remote\n";
1288 sub get_allowed_file_extensions {
1294 siprop => 'fileextensions'
1296 my $result = $mediawiki->api($query);
1297 my @file_extensions= map $_->{ext},@{$result->{query}->{fileextensions}};
1298 my %hashFile = map {$_ => 1}@file_extensions;
1303 # In memory cache for MediaWiki namespace ids.
1306 # Namespaces whose id is cached in the configuration file
1307 # (to avoid duplicates)
1308 my %cached_mw_namespace_id;
1310 # Return MediaWiki id for a canonical namespace name.
1311 # Ex.: "File", "Project".
1312 sub get_mw_namespace_id {
1316 if (!exists $namespace_id{$name}) {
1317 # Look at configuration file, if the record for that namespace is
1318 # already cached. Namespaces are stored in form:
1319 # "Name_of_namespace:Id_namespace", ex.: "File:6".
1320 my @temp = split(/[\n]/, run_git("config --get-all remote."
1321 . $remotename .".namespaceCache"));
1323 foreach my $ns (@temp) {
1324 my ($n, $id) = split(/:/, $ns);
1325 if ($id eq 'notANameSpace') {
1326 $namespace_id{$n} = {is_namespace => 0};
1328 $namespace_id{$n} = {is_namespace => 1, id => $id};
1330 $cached_mw_namespace_id{$n} = 1;
1334 if (!exists $namespace_id{$name}) {
1335 print STDERR "Namespace $name not found in cache, querying the wiki ...\n";
1336 # NS not found => get namespace id from MW and store it in
1337 # configuration file.
1341 siprop => 'namespaces'
1343 my $result = $mediawiki->api($query);
1345 while (my ($id, $ns) = each(%{$result->{query}->{namespaces}})) {
1346 if (defined($ns->{id}) && defined($ns->{canonical})) {
1347 $namespace_id{$ns->{canonical}} = {is_namespace => 1, id => $ns->{id}};
1349 # alias (e.g. french Fichier: as alias for canonical File:)
1350 $namespace_id{$ns->{'*'}} = {is_namespace => 1, id => $ns->{id}};
1356 my $ns = $namespace_id{$name};
1359 unless (defined $ns) {
1360 print STDERR "No such namespace $name on MediaWiki.\n";
1361 $ns = {is_namespace => 0};
1362 $namespace_id{$name} = $ns;
1365 if ($ns->{is_namespace}) {
1369 # Store "notANameSpace" as special value for inexisting namespaces
1370 my $store_id = ($id || 'notANameSpace');
1372 # Store explicitely requested namespaces on disk
1373 if (!exists $cached_mw_namespace_id{$name}) {
1374 run_git("config --add remote.". $remotename
1375 .".namespaceCache \"". $name .":". $store_id ."\"");
1376 $cached_mw_namespace_id{$name} = 1;
1381 sub get_mw_namespace_id_for_page {
1382 if (my ($namespace) = $_[0] =~ /^([^:]*):/) {
1383 return get_mw_namespace_id($namespace);