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