svn import: Add a loop limit option
[git.git] / git-svnimport.perl
1 #!/usr/bin/perl -w
2
3 # This tool is copyright (c) 2005, Matthias Urlichs.
4 # It is released under the Gnu Public License, version 2.
5 #
6 # The basic idea is to pull and analyze SVN changes.
7 #
8 # Checking out the files is done by a single long-running CVS connection
9 # / server process.
10 #
11 # The head revision is on branch "origin" by default.
12 # You can change that with the '-o' option.
13
14 require v5.8.0; # for shell-safe open("-|",LIST)
15 use strict;
16 use warnings;
17 use Getopt::Std;
18 use File::Spec;
19 use File::Temp qw(tempfile);
20 use File::Path qw(mkpath);
21 use File::Basename qw(basename dirname);
22 use Time::Local;
23 use IO::Pipe;
24 use POSIX qw(strftime dup2);
25 use IPC::Open2;
26 use SVN::Core;
27 use SVN::Ra;
28
29 die "Need CVN:COre 1.2.1 or better" if $SVN::Core::VERSION lt "1.2.1";
30
31 $SIG{'PIPE'}="IGNORE";
32 $ENV{'TZ'}="UTC";
33
34 our($opt_h,$opt_o,$opt_v,$opt_u,$opt_C,$opt_i,$opt_m,$opt_M,$opt_t,$opt_T,$opt_b,$opt_s,$opt_l);
35
36 sub usage() {
37         print STDERR <<END;
38 Usage: ${\basename $0}     # fetch/update GIT from CVS
39        [-o branch-for-HEAD] [-h] [-v] [-l max_num_changes]
40        [-C GIT_repository] [-t tagname] [-T trunkname] [-b branchname]
41        [-i] [-u] [-s start_chg] [-m] [-M regex] [SVN_URL]
42 END
43         exit(1);
44 }
45
46 getopts("b:C:hil:mM:o:s:t:T:uv") or usage();
47 usage if $opt_h;
48
49 my $tag_name = $opt_t || "tags";
50 my $trunk_name = $opt_T || "trunk";
51 my $branch_name = $opt_b || "branches";
52
53 @ARGV <= 1 or usage();
54
55 $opt_o ||= "origin";
56 $opt_l = 100 unless defined $opt_l;
57 my $git_tree = $opt_C;
58 $git_tree ||= ".";
59
60 my $cvs_tree;
61 if ($#ARGV == 0) {
62         $cvs_tree = $ARGV[0];
63 } elsif (-f 'CVS/Repository') {
64         open my $f, '<', 'CVS/Repository' or 
65             die 'Failed to open CVS/Repository';
66         $cvs_tree = <$f>;
67         chomp $cvs_tree;
68         close $f;
69 } else {
70         usage();
71 }
72
73 our @mergerx = ();
74 if ($opt_m) {
75         @mergerx = ( qr/\W(?:from|of|merge|merging|merged) (\w+)/i );
76 }
77 if ($opt_M) {
78         push (@mergerx, qr/$opt_M/);
79 }
80
81 select(STDERR); $|=1; select(STDOUT);
82
83
84 package SVNconn;
85 # Basic SVN connection.
86 # We're only interested in connecting and downloading, so ...
87
88 use File::Spec;
89 use File::Temp qw(tempfile);
90 use POSIX qw(strftime dup2);
91
92 sub new {
93         my($what,$repo) = @_;
94         $what=ref($what) if ref($what);
95
96         my $self = {};
97         $self->{'buffer'} = "";
98         bless($self,$what);
99
100         $repo =~ s#/+$##;
101         $self->{'fullrep'} = $repo;
102         $self->conn();
103
104         return $self;
105 }
106
107 sub conn {
108         my $self = shift;
109         my $repo = $self->{'fullrep'};
110         my $s = SVN::Ra->new($repo);
111
112         die "SVN connection to $repo: $!\n" unless defined $s;
113         $self->{'svn'} = $s;
114         $self->{'repo'} = $repo;
115         $self->{'maxrev'} = $s->get_latest_revnum();
116 }
117
118 sub file {
119         my($self,$path,$rev) = @_;
120         my $res;
121
122         my ($fh, $name) = tempfile('gitsvn.XXXXXX', 
123                     DIR => File::Spec->tmpdir(), UNLINK => 1);
124
125         print "... $rev $path ...\n" if $opt_v;
126         eval { $self->{'svn'}->get_file($path,$rev,$fh); };
127         if ($@ and $@ !~ /Attempted to get checksum/) {
128             # retry
129             $self->conn();
130                 eval { $self->{'svn'}->get_file($path,$rev,$fh); };
131         };
132         return () if $@ and $@ !~ /Attempted to get checksum/;
133         die $@ if $@;
134         close ($fh);
135
136         return ($name, $res);
137 }
138
139
140 package main;
141
142 my $svn = SVNconn->new($cvs_tree);
143
144
145 sub pdate($) {
146         my($d) = @_;
147         $d =~ m#(\d\d\d\d)-(\d\d)-(\d\d)T(\d\d):(\d\d):(\d\d)#
148                 or die "Unparseable date: $d\n";
149         my $y=$1; $y-=1900 if $y>1900;
150         return timegm($6||0,$5,$4,$3,$2-1,$y);
151 }
152
153 sub getwd() {
154         my $pwd = `pwd`;
155         chomp $pwd;
156         return $pwd;
157 }
158
159
160 sub get_headref($$) {
161     my $name    = shift;
162     my $git_dir = shift; 
163     my $sha;
164     
165     if (open(C,"$git_dir/refs/heads/$name")) {
166         chomp($sha = <C>);
167         close(C);
168         length($sha) == 40
169             or die "Cannot get head id for $name ($sha): $!\n";
170     }
171     return $sha;
172 }
173
174
175 -d $git_tree
176         or mkdir($git_tree,0777)
177         or die "Could not create $git_tree: $!";
178 chdir($git_tree);
179
180 my $orig_branch = "";
181 my $forward_master = 0;
182 my %branches;
183
184 my $git_dir = $ENV{"GIT_DIR"} || ".git";
185 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
186 $ENV{"GIT_DIR"} = $git_dir;
187 my $orig_git_index;
188 $orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
189 my ($git_ih, $git_index) = tempfile('gitXXXXXX', SUFFIX => '.idx',
190                                     DIR => File::Spec->tmpdir());
191 close ($git_ih);
192 $ENV{GIT_INDEX_FILE} = $git_index;
193 my $maxnum = 0;
194 my $last_rev = "";
195 my $last_branch;
196 my $current_rev = $opt_s ? ($opt_s-1) : 0;
197 unless(-d $git_dir) {
198         system("git-init-db");
199         die "Cannot init the GIT db at $git_tree: $?\n" if $?;
200         system("git-read-tree");
201         die "Cannot init an empty tree: $?\n" if $?;
202
203         $last_branch = $opt_o;
204         $orig_branch = "";
205 } else {
206         -f "$git_dir/refs/heads/$opt_o"
207                 or die "Branch '$opt_o' does not exist.\n".
208                        "Either use the correct '-o branch' option,\n".
209                        "or import to a new repository.\n";
210
211         -f "$git_dir/svn2git"
212                 or die "'$git_dir/svn2git' does not exist.\n".
213                        "You need that file for incremental imports.\n";
214         $last_branch = basename(readlink("$git_dir/HEAD"));
215         unless($last_branch) {
216                 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
217                 $last_branch = "master";
218         }
219         $orig_branch = $last_branch;
220         $last_rev = get_headref($orig_branch, $git_dir);
221         if (-f "$git_dir/SVN2GIT_HEAD") {
222                 die <<EOM;
223 SVN2GIT_HEAD exists.
224 Make sure your working directory corresponds to HEAD and remove SVN2GIT_HEAD.
225 You may need to run
226
227     git-read-tree -m -u SVN2GIT_HEAD HEAD
228 EOM
229         }
230         system('cp', "$git_dir/HEAD", "$git_dir/SVN2GIT_HEAD");
231
232         $forward_master =
233             $opt_o ne 'master' && -f "$git_dir/refs/heads/master" &&
234             system('cmp', '-s', "$git_dir/refs/heads/master", 
235                                 "$git_dir/refs/heads/$opt_o") == 0;
236
237         # populate index
238         system('git-read-tree', $last_rev);
239         die "read-tree failed: $?\n" if $?;
240
241         # Get the last import timestamps
242         open my $B,"<", "$git_dir/svn2git";
243         while(<$B>) {
244                 chomp;
245                 my($num,$branch,$ref) = split;
246                 $branches{$branch}{$num} = $ref;
247                 $branches{$branch}{"LAST"} = $ref;
248                 $current_rev = $num if $current_rev < $num;
249         }
250         close($B);
251 }
252 -d $git_dir
253         or die "Could not create git subdir ($git_dir).\n";
254
255 open BRANCHES,">>", "$git_dir/svn2git";
256
257
258 ## cvsps output:
259 #---------------------
260 #PatchSet 314
261 #Date: 1999/09/18 13:03:59
262 #Author: wkoch
263 #Branch: STABLE-BRANCH-1-0
264 #Ancestor branch: HEAD
265 #Tag: (none)
266 #Log:
267 #    See ChangeLog: Sat Sep 18 13:03:28 CEST 1999  Werner Koch
268 #Members:
269 #       README:1.57->1.57.2.1
270 #       VERSION:1.96->1.96.2.1
271 #
272 #---------------------
273
274 my $state = 0;
275
276 sub get_file($$$) {
277         my($rev,$branch,$path) = @_;
278
279         # revert split_path(), below
280         my $svnpath;
281         $path = "" if $path eq "/"; # this should not happen, but ...
282         if($branch eq "/") {
283                 $svnpath = "/$trunk_name/$path";
284         } elsif($branch =~ m#^/#) {
285                 $svnpath = "/$tag_name$branch/$path";
286         } else {
287                 $svnpath = "/$branch_name/$branch/$path";
288         }
289
290         # now get it
291         my ($name, $res) = eval { $svn->file($svnpath,$rev); };
292         return () unless defined $name;
293
294         open my $F, '-|', "git-hash-object", "-w", $name
295                 or die "Cannot create object: $!\n";
296         my $sha = <$F>;
297         chomp $sha;
298         close $F;
299         my $mode = "0644"; # SV does not seem to store any file modes
300         return [$mode, $sha, $path];
301 }
302
303 sub split_path($$) {
304         my($rev,$path) = @_;
305         my $branch;
306
307         if($path =~ s#^/\Q$tag_name\E/([^/]+)/?##) {
308                 $branch = "/$1";
309         } elsif($path =~ s#^/\Q$trunk_name\E/?##) {
310                 $branch = "/";
311         } elsif($path =~ s#^/\Q$branch_name\E/([^/]+)/?##) {
312                 $branch = $1;
313         } else {
314                 print STDERR "$rev: Unrecognized path: $path\n";
315                 return ()
316         }
317         $path = "/" if $path eq "";
318         return ($branch,$path);
319 }
320
321 sub commit {
322         my($branch, $changed_paths, $revision, $author, $date, $message) = @_;
323         my($author_name,$author_email,$dest);
324         my(@old,@new);
325
326         if (not defined $author) {
327                 $author_name = $author_email = "unknown";
328         } elsif ($author =~ /^(.*?)\s+<(.*)>$/) {
329                 ($author_name, $author_email) = ($1, $2);
330         } else {
331                 $author =~ s/^<(.*)>$/$1/;
332                 $author_name = $author_email = $author;
333         }
334         $date = pdate($date);
335
336         my $tag;
337         my $parent;
338         if($branch eq "/") { # trunk
339                 $parent = $opt_o;
340         } elsif($branch =~ m#^/(.+)#) { # tag
341                 $tag = 1;
342                 $parent = $1;
343         } else { # "normal" branch
344                 # nothing to do
345                 $parent = $branch;
346         }
347         $dest = $parent;
348
349         my $prev = $changed_paths->{"/"};
350         if($prev and $prev->[0] eq "A") {
351                 delete $changed_paths->{"/"};
352                 my $oldpath = $prev->[1];
353                 my $rev;
354                 if(defined $oldpath) {
355                         my $p;
356                         ($parent,$p) = split_path($revision,$oldpath);
357                         if($parent eq "/") {
358                                 $parent = $opt_o;
359                         } else {
360                                 $parent =~ s#^/##; # if it's a tag
361                         }
362                 } else {
363                         $parent = undef;
364                 }
365         }
366
367         my $rev;
368         if($revision > $opt_s and defined $parent) {
369                 open(H,"git-rev-parse --verify $parent |");
370                 $rev = <H>;
371                 close(H) or do {
372                         print STDERR "$revision: cannot find commit '$parent'!\n";
373                         return;
374                 };
375                 chop $rev;
376                 if(length($rev) != 40) {
377                         print STDERR "$revision: cannot find commit '$parent'!\n";
378                         return;
379                 }
380                 $rev = $branches{($parent eq $opt_o) ? "/" : $parent}{"LAST"};
381                 if($revision != $opt_s and not $rev) {
382                         print STDERR "$revision: do not know ancestor for '$parent'!\n";
383                         return;
384                 }
385         } else {
386                 $rev = undef;
387         }
388
389 #       if($prev and $prev->[0] eq "A") {
390 #               if(not $tag) {
391 #                       unless(open(H,"> $git_dir/refs/heads/$branch")) {
392 #                               print STDERR "$revision: Could not create branch $branch: $!\n";
393 #                               $state=11;
394 #                               next;
395 #                       }
396 #                       print H "$rev\n"
397 #                               or die "Could not write branch $branch: $!";
398 #                       close(H)
399 #                               or die "Could not write branch $branch: $!";
400 #               }
401 #       }
402         if(not defined $rev) {
403                 unlink($git_index);
404         } elsif ($rev ne $last_rev) {
405                 print "Switching from $last_rev to $rev ($branch)\n" if $opt_v;
406                 system("git-read-tree", $rev);
407                 die "read-tree failed for $rev: $?\n" if $?;
408                 $last_rev = $rev;
409         }
410
411         my $cid;
412         if($tag and not %$changed_paths) {
413                 $cid = $rev;
414         } else {
415                 while(my($path,$action) = each %$changed_paths) {
416                         if ($action->[0] eq "A") {
417                                 my $f = get_file($revision,$branch,$path);
418                                 push(@new,$f) if $f;
419                         } elsif ($action->[0] eq "D") {
420                                 push(@old,$path);
421                         } elsif ($action->[0] eq "M") {
422                                 my $f = get_file($revision,$branch,$path);
423                                 push(@new,$f) if $f;
424                         } elsif ($action->[0] eq "R") {
425                                 # refer to a file/tree in an earlier commit
426                                 push(@old,$path); # remove any old stuff
427
428                                 # ... and add any new stuff
429                                 my($b,$p) = split_path($revision,$action->[1]);
430                                 open my $F,"-|","git-ls-tree","-r","-z", $branches{$b}{$action->[2]}, $p;
431                                 local $/ = '\0';
432                                 while(<$F>) {
433                                         chomp;
434                                         my($m,$p) = split(/\t/,$_,2);
435                                         my($mode,$type,$sha1) = split(/ /,$m);
436                                         next if $type ne "blob";
437                                         push(@new,[$mode,$sha1,$p]);
438                                 }
439                         } else {
440                                 die "$revision: unknown action '".$action->[0]."' for $path\n";
441                         }
442                 }
443
444                 if(@old) {
445                         open my $F, "-|", "git-ls-files", "-z", @old or die $!;
446                         @old = ();
447                         local $/ = '\0';
448                         while(<$F>) {
449                                 chomp;
450                                 push(@old,$_);
451                         }
452                         close($F);
453
454                         while(@old) {
455                                 my @o2;
456                                 if(@old > 55) {
457                                         @o2 = splice(@old,0,50);
458                                 } else {
459                                         @o2 = @old;
460                                         @old = ();
461                                 }
462                                 system("git-update-index","--force-remove","--",@o2);
463                                 die "Cannot remove files: $?\n" if $?;
464                         }
465                 }
466                 while(@new) {
467                         my @n2;
468                         if(@new > 12) {
469                                 @n2 = splice(@new,0,10);
470                         } else {
471                                 @n2 = @new;
472                                 @new = ();
473                         }
474                         system("git-update-index","--add",
475                                 (map { ('--cacheinfo', @$_) } @n2));
476                         die "Cannot add files: $?\n" if $?;
477                 }
478
479                 my $pid = open(C,"-|");
480                 die "Cannot fork: $!" unless defined $pid;
481                 unless($pid) {
482                         exec("git-write-tree");
483                         die "Cannot exec git-write-tree: $!\n";
484                 }
485                 chomp(my $tree = <C>);
486                 length($tree) == 40
487                         or die "Cannot get tree id ($tree): $!\n";
488                 close(C)
489                         or die "Error running git-write-tree: $?\n";
490                 print "Tree ID $tree\n" if $opt_v;
491
492                 my $pr = IO::Pipe->new() or die "Cannot open pipe: $!\n";
493                 my $pw = IO::Pipe->new() or die "Cannot open pipe: $!\n";
494                 $pid = fork();
495                 die "Fork: $!\n" unless defined $pid;
496                 unless($pid) {
497                         $pr->writer();
498                         $pw->reader();
499                         open(OUT,">&STDOUT");
500                         dup2($pw->fileno(),0);
501                         dup2($pr->fileno(),1);
502                         $pr->close();
503                         $pw->close();
504
505                         my @par = ();
506                         @par = ("-p",$rev) if defined $rev;
507
508                         # loose detection of merges
509                         # based on the commit msg
510                         foreach my $rx (@mergerx) {
511                                 if ($message =~ $rx) {
512                                         my $mparent = $1;
513                                         if ($mparent eq 'HEAD') { $mparent = $opt_o };
514                                         if ( -e "$git_dir/refs/heads/$mparent") {
515                                                 $mparent = get_headref($mparent, $git_dir);
516                                                 push @par, '-p', $mparent;
517                                                 print OUT "Merge parent branch: $mparent\n" if $opt_v;
518                                         }
519                                 } 
520                         }
521
522                         exec("env",
523                                 "GIT_AUTHOR_NAME=$author_name",
524                                 "GIT_AUTHOR_EMAIL=$author_email",
525                                 "GIT_AUTHOR_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
526                                 "GIT_COMMITTER_NAME=$author_name",
527                                 "GIT_COMMITTER_EMAIL=$author_email",
528                                 "GIT_COMMITTER_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
529                                 "git-commit-tree", $tree,@par);
530                         die "Cannot exec git-commit-tree: $!\n";
531                 }
532                 $pw->writer();
533                 $pr->reader();
534
535                 $message =~ s/[\s\n]+\z//;
536
537                 print $pw "$message\n"
538                         or die "Error writing to git-commit-tree: $!\n";
539                 $pw->close();
540
541                 print "Committed change $revision:$branch ".strftime("%Y-%m-%d %H:%M:%S",gmtime($date)).")\n" if $opt_v;
542                 chomp($cid = <$pr>);
543                 length($cid) == 40
544                         or die "Cannot get commit id ($cid): $!\n";
545                 print "Commit ID $cid\n" if $opt_v;
546                 $pr->close();
547
548                 waitpid($pid,0);
549                 die "Error running git-commit-tree: $?\n" if $?;
550         }
551
552         if(not defined $dest) {
553                 print "... no known parent\n" if $opt_v;
554         } elsif(not $tag) {
555                 print "Writing to refs/heads/$dest\n" if $opt_v;
556                 open(C,">$git_dir/refs/heads/$dest") and 
557                 print C ("$cid\n") and
558                 close(C)
559                         or die "Cannot write branch $dest for update: $!\n";
560         }
561
562         if($tag) {
563                 my($in, $out) = ('','');
564                 $last_rev = "-" if %$changed_paths;
565                 # the tag was 'complex', i.e. did not refer to a "real" revision
566                 
567                 $dest =~ tr/_/\./ if $opt_u;
568
569                 my $pid = open2($in, $out, 'git-mktag');
570                 print $out ("object $cid\n".
571                     "type commit\n".
572                     "tag $dest\n".
573                     "tagger $author_name <$author_email>\n") and
574                 close($out)
575                     or die "Cannot create tag object $dest: $!\n";
576
577                 my $tagobj = <$in>;
578                 chomp $tagobj;
579
580                 if ( !close($in) or waitpid($pid, 0) != $pid or
581                                 $? != 0 or $tagobj !~ /^[0123456789abcdef]{40}$/ ) {
582                         die "Cannot create tag object $dest: $!\n";
583                 }
584
585                 open(C,">$git_dir/refs/tags/$dest") and
586                 print C ("$tagobj\n") and
587                 close(C)
588                         or die "Cannot create tag $branch: $!\n";
589
590                 print "Created tag '$dest' on '$branch'\n" if $opt_v;
591         }
592         $branches{$branch}{"LAST"} = $cid;
593         $branches{$branch}{$revision} = $cid;
594         $last_rev = $cid;
595         print BRANCHES "$revision $branch $cid\n";
596         print "DONE: $revision $dest $cid\n" if $opt_v;
597 }
598
599 my ($changed_paths, $revision, $author, $date, $message, $pool) = @_;
600 sub _commit_all {
601         ($changed_paths, $revision, $author, $date, $message, $pool) = @_;
602         my %p;
603         while(my($path,$action) = each %$changed_paths) {
604                 $p{$path} = [ $action->action,$action->copyfrom_path, $action->copyfrom_rev ];
605         }
606         $changed_paths = \%p;
607 }
608
609 sub commit_all {
610         my %done;
611         my @col;
612         my $pref;
613         my $branch;
614
615         while(my($path,$action) = each %$changed_paths) {
616                 ($branch,$path) = split_path($revision,$path);
617                 next if not defined $branch;
618                 $done{$branch}{$path} = $action;
619         }
620         while(($branch,$changed_paths) = each %done) {
621                 commit($branch, $changed_paths, $revision, $author, $date, $message);
622         }
623 }
624
625 while(++$current_rev < $svn->{'maxrev'}) {
626         $svn->{'svn'}->get_log("/",$current_rev,$current_rev,$current_rev,1,1,\&_commit_all,"");
627         commit_all();
628         if($opt_l and not --$opt_l) {
629                 print STDERR "Exiting due to a memory leak. Repeat, please.\n";
630                 last;
631         }
632 }
633
634
635 unlink($git_index);
636
637 if (defined $orig_git_index) {
638         $ENV{GIT_INDEX_FILE} = $orig_git_index;
639 } else {
640         delete $ENV{GIT_INDEX_FILE};
641 }
642
643 # Now switch back to the branch we were in before all of this happened
644 if($orig_branch) {
645         print "DONE\n" if $opt_v and (not defined $opt_l or $opt_l > 0);
646         system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
647                 if $forward_master;
648         unless ($opt_i) {
649                 system('git-read-tree', '-m', '-u', 'SVN2GIT_HEAD', 'HEAD');
650                 die "read-tree failed: $?\n" if $?;
651         }
652 } else {
653         $orig_branch = "master";
654         print "DONE; creating $orig_branch branch\n" if $opt_v and (not defined $opt_l or $opt_l > 0);
655         system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
656                 unless -f "$git_dir/refs/heads/master";
657         unlink("$git_dir/HEAD");
658         symlink("refs/heads/$orig_branch","$git_dir/HEAD");
659         unless ($opt_i) {
660                 system('git checkout');
661                 die "checkout failed: $?\n" if $?;
662         }
663 }
664 unlink("$git_dir/SVN2GIT_HEAD");
665 close(BRANCHES);