svn import: Add direct HTTP access
[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 SVN connection.
9 #
10 # The head revision is on branch "origin" by default.
11 # You can change that with the '-o' option.
12
13 require v5.8.0; # for shell-safe open("-|",LIST)
14 use strict;
15 use warnings;
16 use Getopt::Std;
17 use File::Spec;
18 use File::Temp qw(tempfile);
19 use File::Path qw(mkpath);
20 use File::Basename qw(basename dirname);
21 use Time::Local;
22 use IO::Pipe;
23 use POSIX qw(strftime dup2);
24 use IPC::Open2;
25 use SVN::Core;
26 use SVN::Ra;
27
28 die "Need CVN:Core 1.2.1 or better" if $SVN::Core::VERSION lt "1.2.1";
29
30 $SIG{'PIPE'}="IGNORE";
31 $ENV{'TZ'}="UTC";
32
33 our($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,$opt_d,$opt_D);
34
35 sub usage() {
36         print STDERR <<END;
37 Usage: ${\basename $0}     # fetch/update GIT from CVS
38        [-o branch-for-HEAD] [-h] [-v] [-l max_num_changes]
39        [-C GIT_repository] [-t tagname] [-T trunkname] [-b branchname]
40        [-d|-D] [-i] [-u] [-s start_chg] [-m] [-M regex] [SVN_URL]
41 END
42         exit(1);
43 }
44
45 getopts("b:C:dDhil:mM:o:s:t:T:uv") or usage();
46 usage if $opt_h;
47
48 my $tag_name = $opt_t || "tags";
49 my $trunk_name = $opt_T || "trunk";
50 my $branch_name = $opt_b || "branches";
51
52 @ARGV == 1 or @ARGV == 2 or usage();
53
54 $opt_o ||= "origin";
55 $opt_s ||= 1;
56 $opt_l = 100 unless defined $opt_l;
57 my $git_tree = $opt_C;
58 $git_tree ||= ".";
59
60 my $svn_url = $ARGV[0];
61 my $svn_dir = $ARGV[1];
62
63 our @mergerx = ();
64 if ($opt_m) {
65         @mergerx = ( qr/\W(?:from|of|merge|merging|merged) (\w+)/i );
66 }
67 if ($opt_M) {
68         push (@mergerx, qr/$opt_M/);
69 }
70
71 select(STDERR); $|=1; select(STDOUT);
72
73
74 package SVNconn;
75 # Basic SVN connection.
76 # We're only interested in connecting and downloading, so ...
77
78 use File::Spec;
79 use File::Temp qw(tempfile);
80 use POSIX qw(strftime dup2);
81
82 sub new {
83         my($what,$repo) = @_;
84         $what=ref($what) if ref($what);
85
86         my $self = {};
87         $self->{'buffer'} = "";
88         bless($self,$what);
89
90         $repo =~ s#/+$##;
91         $self->{'fullrep'} = $repo;
92         $self->conn();
93
94         return $self;
95 }
96
97 sub conn {
98         my $self = shift;
99         my $repo = $self->{'fullrep'};
100         my $s = SVN::Ra->new($repo);
101
102         die "SVN connection to $repo: $!\n" unless defined $s;
103         $self->{'svn'} = $s;
104         $self->{'repo'} = $repo;
105         $self->{'maxrev'} = $s->get_latest_revnum();
106 }
107
108 sub file {
109         my($self,$path,$rev) = @_;
110
111         my ($fh, $name) = tempfile('gitsvn.XXXXXX', 
112                     DIR => File::Spec->tmpdir(), UNLINK => 1);
113
114         print "... $rev $path ...\n" if $opt_v;
115         eval { $self->{'svn'}->get_file($path,$rev,$fh); };
116         if($@) {
117                 return undef if $@ =~ /Attempted to get checksum/;
118                 die $@;
119         }
120         close ($fh);
121
122         return $name;
123 }
124
125 package main;
126 use URI;
127
128 my $svn = $svn_url;
129 $svn .= "/$svn_dir" if defined $svn_dir;
130 $svn = SVNconn->new($svn);
131
132 my $lwp_ua;
133 if($opt_d or $opt_D) {
134         $svn_url = URI->new($svn_url)->canonical;
135         if($opt_D) {
136                 $svn_dir =~ s#/*$#/#;
137         } else {
138                 $svn_dir = "";
139         }
140         if ($svn_url->scheme eq "http") {
141                 use LWP::UserAgent;
142                 $lwp_ua = LWP::UserAgent->new(keep_alive => 1, requests_redirectable => []);
143         } else {
144                 print STDERR "Warning: not HTTP; turning off direct file access\n";
145                 $opt_d=0;
146         }
147 }
148
149 sub pdate($) {
150         my($d) = @_;
151         $d =~ m#(\d\d\d\d)-(\d\d)-(\d\d)T(\d\d):(\d\d):(\d\d)#
152                 or die "Unparseable date: $d\n";
153         my $y=$1; $y-=1900 if $y>1900;
154         return timegm($6||0,$5,$4,$3,$2-1,$y);
155 }
156
157 sub getwd() {
158         my $pwd = `pwd`;
159         chomp $pwd;
160         return $pwd;
161 }
162
163
164 sub get_headref($$) {
165     my $name    = shift;
166     my $git_dir = shift; 
167     my $sha;
168     
169     if (open(C,"$git_dir/refs/heads/$name")) {
170         chomp($sha = <C>);
171         close(C);
172         length($sha) == 40
173             or die "Cannot get head id for $name ($sha): $!\n";
174     }
175     return $sha;
176 }
177
178
179 -d $git_tree
180         or mkdir($git_tree,0777)
181         or die "Could not create $git_tree: $!";
182 chdir($git_tree);
183
184 my $orig_branch = "";
185 my $forward_master = 0;
186 my %branches;
187
188 my $git_dir = $ENV{"GIT_DIR"} || ".git";
189 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
190 $ENV{"GIT_DIR"} = $git_dir;
191 my $orig_git_index;
192 $orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
193 my ($git_ih, $git_index) = tempfile('gitXXXXXX', SUFFIX => '.idx',
194                                     DIR => File::Spec->tmpdir());
195 close ($git_ih);
196 $ENV{GIT_INDEX_FILE} = $git_index;
197 my $maxnum = 0;
198 my $last_rev = "";
199 my $last_branch;
200 my $current_rev = $opt_s-1;
201 unless(-d $git_dir) {
202         system("git-init-db");
203         die "Cannot init the GIT db at $git_tree: $?\n" if $?;
204         system("git-read-tree");
205         die "Cannot init an empty tree: $?\n" if $?;
206
207         $last_branch = $opt_o;
208         $orig_branch = "";
209 } else {
210         -f "$git_dir/refs/heads/$opt_o"
211                 or die "Branch '$opt_o' does not exist.\n".
212                        "Either use the correct '-o branch' option,\n".
213                        "or import to a new repository.\n";
214
215         -f "$git_dir/svn2git"
216                 or die "'$git_dir/svn2git' does not exist.\n".
217                        "You need that file for incremental imports.\n";
218         $last_branch = basename(readlink("$git_dir/HEAD"));
219         unless($last_branch) {
220                 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
221                 $last_branch = "master";
222         }
223         $orig_branch = $last_branch;
224         $last_rev = get_headref($orig_branch, $git_dir);
225         if (-f "$git_dir/SVN2GIT_HEAD") {
226                 die <<EOM;
227 SVN2GIT_HEAD exists.
228 Make sure your working directory corresponds to HEAD and remove SVN2GIT_HEAD.
229 You may need to run
230
231     git-read-tree -m -u SVN2GIT_HEAD HEAD
232 EOM
233         }
234         system('cp', "$git_dir/HEAD", "$git_dir/SVN2GIT_HEAD");
235
236         $forward_master =
237             $opt_o ne 'master' && -f "$git_dir/refs/heads/master" &&
238             system('cmp', '-s', "$git_dir/refs/heads/master", 
239                                 "$git_dir/refs/heads/$opt_o") == 0;
240
241         # populate index
242         system('git-read-tree', $last_rev);
243         die "read-tree failed: $?\n" if $?;
244
245         # Get the last import timestamps
246         open my $B,"<", "$git_dir/svn2git";
247         while(<$B>) {
248                 chomp;
249                 my($num,$branch,$ref) = split;
250                 $branches{$branch}{$num} = $ref;
251                 $branches{$branch}{"LAST"} = $ref;
252                 $current_rev = $num if $current_rev < $num;
253         }
254         close($B);
255 }
256 -d $git_dir
257         or die "Could not create git subdir ($git_dir).\n";
258
259 open BRANCHES,">>", "$git_dir/svn2git";
260
261 sub get_file($$$) {
262         my($rev,$branch,$path) = @_;
263
264         # revert split_path(), below
265         my $svnpath;
266         $path = "" if $path eq "/"; # this should not happen, but ...
267         if($branch eq "/") {
268                 $svnpath = "/$trunk_name/$path";
269         } elsif($branch =~ m#^/#) {
270                 $svnpath = "/$tag_name$branch/$path";
271         } else {
272                 $svnpath = "/$branch_name/$branch/$path";
273         }
274
275         # now get it
276         my $name;
277         if($opt_d) {
278                 my($req,$res);
279
280                 # /svn/!svn/bc/2/django/trunk/django-docs/build.py
281                 my $url=$svn_url->clone();
282                 $url->path($url->path."/!svn/bc/$rev/$svn_dir$svnpath");
283                 print "Fetching $url...\n" if $opt_v;
284                 $req = HTTP::Request->new(GET => $url);
285                 $res = $lwp_ua->request($req);
286                 if ($res->is_success) {
287                         my $fh;
288                         ($fh, $name) = tempfile('gitsvn.XXXXXX', 
289                         DIR => File::Spec->tmpdir(), UNLINK => 1);
290                         print $fh $res->content;
291                         close($fh) or die "Could not write $name: $!\n";
292                 } else {
293                         return undef if $res->code == 301; # directory?
294                         die $res->status_line." at $url\n";
295                 }
296         } else {
297                 $name = $svn->file($svnpath,$rev);
298                 return undef unless defined $name;
299         }
300
301         open my $F, '-|', "git-hash-object", "-w", $name
302                 or die "Cannot create object: $!\n";
303         my $sha = <$F>;
304         chomp $sha;
305         close $F;
306         unlink $name;
307         my $mode = "0644"; # SV does not seem to store any file modes
308         return [$mode, $sha, $path];
309 }
310
311 sub split_path($$) {
312         my($rev,$path) = @_;
313         my $branch;
314
315         if($path =~ s#^/\Q$tag_name\E/([^/]+)/?##) {
316                 $branch = "/$1";
317         } elsif($path =~ s#^/\Q$trunk_name\E/?##) {
318                 $branch = "/";
319         } elsif($path =~ s#^/\Q$branch_name\E/([^/]+)/?##) {
320                 $branch = $1;
321         } else {
322                 print STDERR "$rev: Unrecognized path: $path\n";
323                 return ()
324         }
325         $path = "/" if $path eq "";
326         return ($branch,$path);
327 }
328
329 sub commit {
330         my($branch, $changed_paths, $revision, $author, $date, $message) = @_;
331         my($author_name,$author_email,$dest);
332         my(@old,@new);
333
334         if (not defined $author) {
335                 $author_name = $author_email = "unknown";
336         } elsif ($author =~ /^(.*?)\s+<(.*)>$/) {
337                 ($author_name, $author_email) = ($1, $2);
338         } else {
339                 $author =~ s/^<(.*)>$/$1/;
340                 $author_name = $author_email = $author;
341         }
342         $date = pdate($date);
343
344         my $tag;
345         my $parent;
346         if($branch eq "/") { # trunk
347                 $parent = $opt_o;
348         } elsif($branch =~ m#^/(.+)#) { # tag
349                 $tag = 1;
350                 $parent = $1;
351         } else { # "normal" branch
352                 # nothing to do
353                 $parent = $branch;
354         }
355         $dest = $parent;
356
357         my $prev = $changed_paths->{"/"};
358         if($prev and $prev->[0] eq "A") {
359                 delete $changed_paths->{"/"};
360                 my $oldpath = $prev->[1];
361                 my $rev;
362                 if(defined $oldpath) {
363                         my $p;
364                         ($parent,$p) = split_path($revision,$oldpath);
365                         if($parent eq "/") {
366                                 $parent = $opt_o;
367                         } else {
368                                 $parent =~ s#^/##; # if it's a tag
369                         }
370                 } else {
371                         $parent = undef;
372                 }
373         }
374
375         my $rev;
376         if($revision > $opt_s and defined $parent) {
377                 open(H,"git-rev-parse --verify $parent |");
378                 $rev = <H>;
379                 close(H) or do {
380                         print STDERR "$revision: cannot find commit '$parent'!\n";
381                         return;
382                 };
383                 chop $rev;
384                 if(length($rev) != 40) {
385                         print STDERR "$revision: cannot find commit '$parent'!\n";
386                         return;
387                 }
388                 $rev = $branches{($parent eq $opt_o) ? "/" : $parent}{"LAST"};
389                 if($revision != $opt_s and not $rev) {
390                         print STDERR "$revision: do not know ancestor for '$parent'!\n";
391                         return;
392                 }
393         } else {
394                 $rev = undef;
395         }
396
397 #       if($prev and $prev->[0] eq "A") {
398 #               if(not $tag) {
399 #                       unless(open(H,"> $git_dir/refs/heads/$branch")) {
400 #                               print STDERR "$revision: Could not create branch $branch: $!\n";
401 #                               $state=11;
402 #                               next;
403 #                       }
404 #                       print H "$rev\n"
405 #                               or die "Could not write branch $branch: $!";
406 #                       close(H)
407 #                               or die "Could not write branch $branch: $!";
408 #               }
409 #       }
410         if(not defined $rev) {
411                 unlink($git_index);
412         } elsif ($rev ne $last_rev) {
413                 print "Switching from $last_rev to $rev ($branch)\n" if $opt_v;
414                 system("git-read-tree", $rev);
415                 die "read-tree failed for $rev: $?\n" if $?;
416                 $last_rev = $rev;
417         }
418
419         my $cid;
420         if($tag and not %$changed_paths) {
421                 $cid = $rev;
422         } else {
423                 while(my($path,$action) = each %$changed_paths) {
424                         if ($action->[0] eq "A") {
425                                 my $f = get_file($revision,$branch,$path);
426                                 push(@new,$f) if $f;
427                         } elsif ($action->[0] eq "D") {
428                                 push(@old,$path);
429                         } elsif ($action->[0] eq "M") {
430                                 my $f = get_file($revision,$branch,$path);
431                                 push(@new,$f) if $f;
432                         } elsif ($action->[0] eq "R") {
433                                 # refer to a file/tree in an earlier commit
434                                 push(@old,$path); # remove any old stuff
435
436                                 # ... and add any new stuff
437                                 my($b,$p) = split_path($revision,$action->[1]);
438                                 open my $F,"-|","git-ls-tree","-r","-z", $branches{$b}{$action->[2]}, $p;
439                                 local $/ = '\0';
440                                 while(<$F>) {
441                                         chomp;
442                                         my($m,$p) = split(/\t/,$_,2);
443                                         my($mode,$type,$sha1) = split(/ /,$m);
444                                         next if $type ne "blob";
445                                         push(@new,[$mode,$sha1,$p]);
446                                 }
447                         } else {
448                                 die "$revision: unknown action '".$action->[0]."' for $path\n";
449                         }
450                 }
451
452                 if(@old) {
453                         open my $F, "-|", "git-ls-files", "-z", @old or die $!;
454                         @old = ();
455                         local $/ = '\0';
456                         while(<$F>) {
457                                 chomp;
458                                 push(@old,$_);
459                         }
460                         close($F);
461
462                         while(@old) {
463                                 my @o2;
464                                 if(@old > 55) {
465                                         @o2 = splice(@old,0,50);
466                                 } else {
467                                         @o2 = @old;
468                                         @old = ();
469                                 }
470                                 system("git-update-index","--force-remove","--",@o2);
471                                 die "Cannot remove files: $?\n" if $?;
472                         }
473                 }
474                 while(@new) {
475                         my @n2;
476                         if(@new > 12) {
477                                 @n2 = splice(@new,0,10);
478                         } else {
479                                 @n2 = @new;
480                                 @new = ();
481                         }
482                         system("git-update-index","--add",
483                                 (map { ('--cacheinfo', @$_) } @n2));
484                         die "Cannot add files: $?\n" if $?;
485                 }
486
487                 my $pid = open(C,"-|");
488                 die "Cannot fork: $!" unless defined $pid;
489                 unless($pid) {
490                         exec("git-write-tree");
491                         die "Cannot exec git-write-tree: $!\n";
492                 }
493                 chomp(my $tree = <C>);
494                 length($tree) == 40
495                         or die "Cannot get tree id ($tree): $!\n";
496                 close(C)
497                         or die "Error running git-write-tree: $?\n";
498                 print "Tree ID $tree\n" if $opt_v;
499
500                 my $pr = IO::Pipe->new() or die "Cannot open pipe: $!\n";
501                 my $pw = IO::Pipe->new() or die "Cannot open pipe: $!\n";
502                 $pid = fork();
503                 die "Fork: $!\n" unless defined $pid;
504                 unless($pid) {
505                         $pr->writer();
506                         $pw->reader();
507                         open(OUT,">&STDOUT");
508                         dup2($pw->fileno(),0);
509                         dup2($pr->fileno(),1);
510                         $pr->close();
511                         $pw->close();
512
513                         my @par = ();
514                         @par = ("-p",$rev) if defined $rev;
515
516                         # loose detection of merges
517                         # based on the commit msg
518                         foreach my $rx (@mergerx) {
519                                 if ($message =~ $rx) {
520                                         my $mparent = $1;
521                                         if ($mparent eq 'HEAD') { $mparent = $opt_o };
522                                         if ( -e "$git_dir/refs/heads/$mparent") {
523                                                 $mparent = get_headref($mparent, $git_dir);
524                                                 push @par, '-p', $mparent;
525                                                 print OUT "Merge parent branch: $mparent\n" if $opt_v;
526                                         }
527                                 } 
528                         }
529
530                         exec("env",
531                                 "GIT_AUTHOR_NAME=$author_name",
532                                 "GIT_AUTHOR_EMAIL=$author_email",
533                                 "GIT_AUTHOR_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
534                                 "GIT_COMMITTER_NAME=$author_name",
535                                 "GIT_COMMITTER_EMAIL=$author_email",
536                                 "GIT_COMMITTER_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
537                                 "git-commit-tree", $tree,@par);
538                         die "Cannot exec git-commit-tree: $!\n";
539                 }
540                 $pw->writer();
541                 $pr->reader();
542
543                 $message =~ s/[\s\n]+\z//;
544
545                 print $pw "$message\n"
546                         or die "Error writing to git-commit-tree: $!\n";
547                 $pw->close();
548
549                 print "Committed change $revision:$branch ".strftime("%Y-%m-%d %H:%M:%S",gmtime($date)).")\n" if $opt_v;
550                 chomp($cid = <$pr>);
551                 length($cid) == 40
552                         or die "Cannot get commit id ($cid): $!\n";
553                 print "Commit ID $cid\n" if $opt_v;
554                 $pr->close();
555
556                 waitpid($pid,0);
557                 die "Error running git-commit-tree: $?\n" if $?;
558         }
559
560         if(not defined $dest) {
561                 print "... no known parent\n" if $opt_v;
562         } elsif(not $tag) {
563                 print "Writing to refs/heads/$dest\n" if $opt_v;
564                 open(C,">$git_dir/refs/heads/$dest") and 
565                 print C ("$cid\n") and
566                 close(C)
567                         or die "Cannot write branch $dest for update: $!\n";
568         }
569
570         if($tag) {
571                 my($in, $out) = ('','');
572                 $last_rev = "-" if %$changed_paths;
573                 # the tag was 'complex', i.e. did not refer to a "real" revision
574                 
575                 $dest =~ tr/_/\./ if $opt_u;
576
577                 my $pid = open2($in, $out, 'git-mktag');
578                 print $out ("object $cid\n".
579                     "type commit\n".
580                     "tag $dest\n".
581                     "tagger $author_name <$author_email>\n") and
582                 close($out)
583                     or die "Cannot create tag object $dest: $!\n";
584
585                 my $tagobj = <$in>;
586                 chomp $tagobj;
587
588                 if ( !close($in) or waitpid($pid, 0) != $pid or
589                                 $? != 0 or $tagobj !~ /^[0123456789abcdef]{40}$/ ) {
590                         die "Cannot create tag object $dest: $!\n";
591                 }
592
593                 open(C,">$git_dir/refs/tags/$dest") and
594                 print C ("$tagobj\n") and
595                 close(C)
596                         or die "Cannot create tag $branch: $!\n";
597
598                 print "Created tag '$dest' on '$branch'\n" if $opt_v;
599         }
600         $branches{$branch}{"LAST"} = $cid;
601         $branches{$branch}{$revision} = $cid;
602         $last_rev = $cid;
603         print BRANCHES "$revision $branch $cid\n";
604         print "DONE: $revision $dest $cid\n" if $opt_v;
605 }
606
607 my ($changed_paths, $revision, $author, $date, $message, $pool) = @_;
608 sub _commit_all {
609         ($changed_paths, $revision, $author, $date, $message, $pool) = @_;
610         my %p;
611         while(my($path,$action) = each %$changed_paths) {
612                 $p{$path} = [ $action->action,$action->copyfrom_path, $action->copyfrom_rev ];
613         }
614         $changed_paths = \%p;
615 }
616
617 sub commit_all {
618         my %done;
619         my @col;
620         my $pref;
621         my $branch;
622
623         while(my($path,$action) = each %$changed_paths) {
624                 ($branch,$path) = split_path($revision,$path);
625                 next if not defined $branch;
626                 $done{$branch}{$path} = $action;
627         }
628         while(($branch,$changed_paths) = each %done) {
629                 commit($branch, $changed_paths, $revision, $author, $date, $message);
630         }
631 }
632
633 while(++$current_rev <= $svn->{'maxrev'}) {
634         $svn->{'svn'}->get_log("/",$current_rev,$current_rev,$current_rev,1,1,\&_commit_all,"");
635         commit_all();
636         if($opt_l and not --$opt_l) {
637                 print STDERR "Exiting due to a memory leak. Repeat, please.\n";
638                 last;
639         }
640 }
641
642
643 unlink($git_index);
644
645 if (defined $orig_git_index) {
646         $ENV{GIT_INDEX_FILE} = $orig_git_index;
647 } else {
648         delete $ENV{GIT_INDEX_FILE};
649 }
650
651 # Now switch back to the branch we were in before all of this happened
652 if($orig_branch) {
653         print "DONE\n" if $opt_v and (not defined $opt_l or $opt_l > 0);
654         system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
655                 if $forward_master;
656         unless ($opt_i) {
657                 system('git-read-tree', '-m', '-u', 'SVN2GIT_HEAD', 'HEAD');
658                 die "read-tree failed: $?\n" if $?;
659         }
660 } else {
661         $orig_branch = "master";
662         print "DONE; creating $orig_branch branch\n" if $opt_v and (not defined $opt_l or $opt_l > 0);
663         system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
664                 unless -f "$git_dir/refs/heads/master";
665         unlink("$git_dir/HEAD");
666         symlink("refs/heads/$orig_branch","$git_dir/HEAD");
667         unless ($opt_i) {
668                 system('git checkout');
669                 die "checkout failed: $?\n" if $?;
670         }
671 }
672 unlink("$git_dir/SVN2GIT_HEAD");
673 close(BRANCHES);