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 use strict;
  14 use warnings;
  15 use Getopt::Std;
  16 use File::Copy;
  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 SVN: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,
  34     $opt_b,$opt_r,$opt_I,$opt_A,$opt_s,$opt_l,$opt_d,$opt_D);
  35
  36 sub usage() {
  37         print STDERR <<END;
  38 Usage: ${\basename $0}     # fetch/update GIT from SVN
  39        [-o branch-for-HEAD] [-h] [-v] [-l max_rev]
  40        [-C GIT_repository] [-t tagname] [-T trunkname] [-b branchname]
  41        [-d|-D] [-i] [-u] [-r] [-I ignorefilename] [-s start_chg]
  42        [-m] [-M regex] [-A author_file] [SVN_URL]
  43 END
  44         exit(1);
  45 }
  46
  47 getopts("A:b:C:dDhiI:l:mM:o:rs:t:T:uv") or usage();
  48 usage if $opt_h;
  49
  50 my $tag_name = $opt_t || "tags";
  51 my $trunk_name = $opt_T || "trunk";
  52 my $branch_name = $opt_b || "branches";
  53
  54 @ARGV == 1 or @ARGV == 2 or usage();
  55
  56 $opt_o ||= "origin";
  57 $opt_s ||= 1;
  58 my $git_tree = $opt_C;
  59 $git_tree ||= ".";
  60
  61 my $svn_url = $ARGV[0];
  62 my $svn_dir = $ARGV[1];
  63
  64 our @mergerx = ();
  65 if ($opt_m) {
  66         @mergerx = ( qr/\W(?:from|of|merge|merging|merged) (\w+)/i );
  67 }
  68 if ($opt_M) {
  69         push (@mergerx, qr/$opt_M/);
  70 }
  71
  72 # Absolutize filename now, since we will have chdir'ed by the time we
  73 # get around to opening it.
  74 $opt_A = File::Spec->rel2abs($opt_A) if $opt_A;
  75
  76 our %users = ();
  77 our $users_file = undef;
  78 sub read_users($) {
  79         $users_file = File::Spec->rel2abs(@_);
  80         die "Cannot open $users_file\n" unless -f $users_file;
  81         open(my $authors,$users_file);
  82         while(<$authors>) {
  83                 chomp;
  84                 next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
  85                 (my $user,my $name,my $email) = ($1,$2,$3);
  86                 $users{$user} = [$name,$email];
  87         }
  88         close($authors);
  89 }
  90
  91 select(STDERR); $|=1; select(STDOUT);
  92
  93
  94 package SVNconn;
  95 # Basic SVN connection.
  96 # We're only interested in connecting and downloading, so ...
  97
  98 use File::Spec;
  99 use File::Temp qw(tempfile);
 100 use POSIX qw(strftime dup2);
 101 use Fcntl qw(SEEK_SET);
 102
 103 sub new {
 104         my($what,$repo) = @_;
 105         $what=ref($what) if ref($what);
 106
 107         my $self = {};
 108         $self->{'buffer'} = "";
 109         bless($self,$what);
 110
 111         $repo =~ s#/+$##;
 112         $self->{'fullrep'} = $repo;
 113         $self->conn();
 114
 115         return $self;
 116 }
 117
 118 sub conn {
 119         my $self = shift;
 120         my $repo = $self->{'fullrep'};
 121         my $auth = SVN::Core::auth_open ([SVN::Client::get_simple_provider,
 122                           SVN::Client::get_ssl_server_trust_file_provider,
 123                           SVN::Client::get_username_provider]);
 124         my $s = SVN::Ra->new(url => $repo, auth => $auth);
 125         die "SVN connection to $repo: $!\n" unless defined $s;
 126         $self->{'svn'} = $s;
 127         $self->{'repo'} = $repo;
 128         $self->{'maxrev'} = $s->get_latest_revnum();
 129 }
 130
 131 sub file {
 132         my($self,$path,$rev) = @_;
 133
 134         my ($fh, $name) = tempfile('gitsvn.XXXXXX',
 135                     DIR => File::Spec->tmpdir(), UNLINK => 1);
 136
 137         print "... $rev $path ...\n" if $opt_v;
 138         my (undef, $properties);
 139         eval { (undef, $properties)
 140                    = $self->{'svn'}->get_file($path,$rev,$fh); };
 141         if($@) {
 142                 return undef if $@ =~ /Attempted to get checksum/;
 143                 die $@;
 144         }
 145         my $mode;
 146         if (exists $properties->{'svn:executable'}) {
 147                 $mode = '100755';
 148         } elsif (exists $properties->{'svn:special'}) {
 149                 my ($special_content, $filesize);
 150                 $filesize = tell $fh;
 151                 seek $fh, 0, SEEK_SET;
 152                 read $fh, $special_content, $filesize;
 153                 if ($special_content =~ s/^link //) {
 154                         $mode = '120000';
 155                         seek $fh, 0, SEEK_SET;
 156                         truncate $fh, 0;
 157                         print $fh $special_content;
 158                 } else {
 159                         die "unexpected svn:special file encountered";
 160                 }
 161         } else {
 162                 $mode = '100644';
 163         }
 164         close ($fh);
 165
 166         return ($name, $mode);
 167 }
 168
 169 sub ignore {
 170         my($self,$path,$rev) = @_;
 171
 172         print "... $rev $path ...\n" if $opt_v;
 173         my (undef,undef,$properties)
 174             = $self->{'svn'}->get_dir($path,$rev,undef);
 175         if (exists $properties->{'svn:ignore'}) {
 176                 my ($fh, $name) = tempfile('gitsvn.XXXXXX',
 177                                            DIR => File::Spec->tmpdir(),
 178                                            UNLINK => 1);
 179                 print $fh $properties->{'svn:ignore'};
 180                 close($fh);
 181                 return $name;
 182         } else {
 183                 return undef;
 184         }
 185 }
 186
 187 package main;
 188 use URI;
 189
 190 our $svn = $svn_url;
 191 $svn .= "/$svn_dir" if defined $svn_dir;
 192 my $svn2 = SVNconn->new($svn);
 193 $svn = SVNconn->new($svn);
 194
 195 my $lwp_ua;
 196 if($opt_d or $opt_D) {
 197         $svn_url = URI->new($svn_url)->canonical;
 198         if($opt_D) {
 199                 $svn_dir =~ s#/*$#/#;
 200         } else {
 201                 $svn_dir = "";
 202         }
 203         if ($svn_url->scheme eq "http") {
 204                 use LWP::UserAgent;
 205                 $lwp_ua = LWP::UserAgent->new(keep_alive => 1, requests_redirectable => []);
 206         } else {
 207                 print STDERR "Warning: not HTTP; turning off direct file access\n";
 208                 $opt_d=0;
 209         }
 210 }
 211
 212 sub pdate($) {
 213         my($d) = @_;
 214         $d =~ m#(\d\d\d\d)-(\d\d)-(\d\d)T(\d\d):(\d\d):(\d\d)#
 215                 or die "Unparseable date: $d\n";
 216         my $y=$1; $y-=1900 if $y>1900;
 217         return timegm($6||0,$5,$4,$3,$2-1,$y);
 218 }
 219
 220 sub getwd() {
 221         my $pwd = `pwd`;
 222         chomp $pwd;
 223         return $pwd;
 224 }
 225
 226
 227 sub get_headref($$) {
 228     my $name    = shift;
 229     my $git_dir = shift;
 230     my $sha;
 231
 232     if (open(C,"$git_dir/refs/heads/$name")) {
 233         chomp($sha = <C>);
 234         close(C);
 235         length($sha) == 40
 236             or die "Cannot get head id for $name ($sha): $!\n";
 237     }
 238     return $sha;
 239 }
 240
 241
 242 -d $git_tree
 243         or mkdir($git_tree,0777)
 244         or die "Could not create $git_tree: $!";
 245 chdir($git_tree);
 246
 247 my $orig_branch = "";
 248 my $forward_master = 0;
 249 my %branches;
 250
 251 my $git_dir = $ENV{"GIT_DIR"} || ".git";
 252 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
 253 $ENV{"GIT_DIR"} = $git_dir;
 254 my $orig_git_index;
 255 $orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
 256 my ($git_ih, $git_index) = tempfile('gitXXXXXX', SUFFIX => '.idx',
 257                                     DIR => File::Spec->tmpdir());
 258 close ($git_ih);
 259 $ENV{GIT_INDEX_FILE} = $git_index;
 260 my $maxnum = 0;
 261 my $last_rev = "";
 262 my $last_branch;
 263 my $current_rev = $opt_s || 1;
 264 unless(-d $git_dir) {
 265         system("git-init-db");
 266         die "Cannot init the GIT db at $git_tree: $?\n" if $?;
 267         system("git-read-tree");
 268         die "Cannot init an empty tree: $?\n" if $?;
 269
 270         $last_branch = $opt_o;
 271         $orig_branch = "";
 272 } else {
 273         -f "$git_dir/refs/heads/$opt_o"
 274                 or die "Branch '$opt_o' does not exist.\n".
 275                        "Either use the correct '-o branch' option,\n".
 276                        "or import to a new repository.\n";
 277
 278         -f "$git_dir/svn2git"
 279                 or die "'$git_dir/svn2git' does not exist.\n".
 280                        "You need that file for incremental imports.\n";
 281         open(F, "git-symbolic-ref HEAD |") or
 282                 die "Cannot run git-symbolic-ref: $!\n";
 283         chomp ($last_branch = <F>);
 284         $last_branch = basename($last_branch);
 285         close(F);
 286         unless($last_branch) {
 287                 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
 288                 $last_branch = "master";
 289         }
 290         $orig_branch = $last_branch;
 291         $last_rev = get_headref($orig_branch, $git_dir);
 292         if (-f "$git_dir/SVN2GIT_HEAD") {
 293                 die <<EOM;
 294 SVN2GIT_HEAD exists.
 295 Make sure your working directory corresponds to HEAD and remove SVN2GIT_HEAD.
 296 You may need to run
 297
 298     git-read-tree -m -u SVN2GIT_HEAD HEAD
 299 EOM
 300         }
 301         system('cp', "$git_dir/HEAD", "$git_dir/SVN2GIT_HEAD");
 302
 303         $forward_master =
 304             $opt_o ne 'master' && -f "$git_dir/refs/heads/master" &&
 305             system('cmp', '-s', "$git_dir/refs/heads/master",
 306                                 "$git_dir/refs/heads/$opt_o") == 0;
 307
 308         # populate index
 309         system('git-read-tree', $last_rev);
 310         die "read-tree failed: $?\n" if $?;
 311
 312         # Get the last import timestamps
 313         open my $B,"<", "$git_dir/svn2git";
 314         while(<$B>) {
 315                 chomp;
 316                 my($num,$branch,$ref) = split;
 317                 $branches{$branch}{$num} = $ref;
 318                 $branches{$branch}{"LAST"} = $ref;
 319                 $current_rev = $num+1 if $current_rev <= $num;
 320         }
 321         close($B);
 322 }
 323 -d $git_dir
 324         or die "Could not create git subdir ($git_dir).\n";
 325
 326 my $default_authors = "$git_dir/svn-authors";
 327 if ($opt_A) {
 328         read_users($opt_A);
 329         copy($opt_A,$default_authors) or die "Copy failed: $!";
 330 } else {
 331         read_users($default_authors) if -f $default_authors;
 332 }
 333
 334 open BRANCHES,">>", "$git_dir/svn2git";
 335
 336 sub node_kind($$$) {
 337         my ($branch, $path, $revision) = @_;
 338         my $pool=SVN::Pool->new;
 339         my $kind = $svn->{'svn'}->check_path(revert_split_path($branch,$path),$revision,$pool);
 340         $pool->clear;
 341         return $kind;
 342 }
 343
 344 sub revert_split_path($$) {
 345         my($branch,$path) = @_;
 346
 347         my $svnpath;
 348         $path = "" if $path eq "/"; # this should not happen, but ...
 349         if($branch eq "/") {
 350                 $svnpath = "$trunk_name/$path";
 351         } elsif($branch =~ m#^/#) {
 352                 $svnpath = "$tag_name$branch/$path";
 353         } else {
 354                 $svnpath = "$branch_name/$branch/$path";
 355         }
 356
 357         $svnpath =~ s#/+$##;
 358         return $svnpath;
 359 }
 360
 361 sub get_file($$$) {
 362         my($rev,$branch,$path) = @_;
 363
 364         my $svnpath = revert_split_path($branch,$path);
 365
 366         # now get it
 367         my ($name,$mode);
 368         if($opt_d) {
 369                 my($req,$res);
 370
 371                 # /svn/!svn/bc/2/django/trunk/django-docs/build.py
 372                 my $url=$svn_url->clone();
 373                 $url->path($url->path."/!svn/bc/$rev/$svn_dir$svnpath");
 374                 print "... $path...\n" if $opt_v;
 375                 $req = HTTP::Request->new(GET => $url);
 376                 $res = $lwp_ua->request($req);
 377                 if ($res->is_success) {
 378                         my $fh;
 379                         ($fh, $name) = tempfile('gitsvn.XXXXXX',
 380                         DIR => File::Spec->tmpdir(), UNLINK => 1);
 381                         print $fh $res->content;
 382                         close($fh) or die "Could not write $name: $!\n";
 383                 } else {
 384                         return undef if $res->code == 301; # directory?
 385                         die $res->status_line." at $url\n";
 386                 }
 387                 $mode = '0644'; # can't obtain mode via direct http request?
 388         } else {
 389                 ($name,$mode) = $svn->file("$svnpath",$rev);
 390                 return undef unless defined $name;
 391         }
 392
 393         my $pid = open(my $F, '-|');
 394         die $! unless defined $pid;
 395         if (!$pid) {
 396             exec("git-hash-object", "-w", $name)
 397                 or die "Cannot create object: $!\n";
 398         }
 399         my $sha = <$F>;
 400         chomp $sha;
 401         close $F;
 402         unlink $name;
 403         return [$mode, $sha, $path];
 404 }
 405
 406 sub get_ignore($$$$$) {
 407         my($new,$old,$rev,$branch,$path) = @_;
 408
 409         return unless $opt_I;
 410         my $svnpath = revert_split_path($branch,$path);
 411         my $name = $svn->ignore("$svnpath",$rev);
 412         if ($path eq '/') {
 413                 $path = $opt_I;
 414         } else {
 415                 $path = File::Spec->catfile($path,$opt_I);
 416         }
 417         if (defined $name) {
 418                 my $pid = open(my $F, '-|');
 419                 die $! unless defined $pid;
 420                 if (!$pid) {
 421                         exec("git-hash-object", "-w", $name)
 422                             or die "Cannot create object: $!\n";
 423                 }
 424                 my $sha = <$F>;
 425                 chomp $sha;
 426                 close $F;
 427                 unlink $name;
 428                 push(@$new,['0644',$sha,$path]);
 429         } else {
 430                 push(@$old,$path);
 431         }
 432 }
 433
 434 sub split_path($$) {
 435         my($rev,$path) = @_;
 436         my $branch;
 437
 438         if($path =~ s#^/\Q$tag_name\E/([^/]+)/?##) {
 439                 $branch = "/$1";
 440         } elsif($path =~ s#^/\Q$trunk_name\E/?##) {
 441                 $branch = "/";
 442         } elsif($path =~ s#^/\Q$branch_name\E/([^/]+)/?##) {
 443                 $branch = $1;
 444         } else {
 445                 my %no_error = (
 446                         "/" => 1,
 447                         "/$tag_name" => 1,
 448                         "/$branch_name" => 1
 449                 );
 450                 print STDERR "$rev: Unrecognized path: $path\n" unless (defined $no_error{$path});
 451                 return ()
 452         }
 453         $path = "/" if $path eq "";
 454         return ($branch,$path);
 455 }
 456
 457 sub branch_rev($$) {
 458
 459         my ($srcbranch,$uptorev) = @_;
 460
 461         my $bbranches = $branches{$srcbranch};
 462         my @revs = reverse sort { ($a eq 'LAST' ? 0 : $a) <=> ($b eq 'LAST' ? 0 : $b) } keys %$bbranches;
 463         my $therev;
 464         foreach my $arev(@revs) {
 465                 next if  ($arev eq 'LAST');
 466                 if ($arev <= $uptorev) {
 467                         $therev = $arev;
 468                         last;
 469                 }
 470         }
 471         return $therev;
 472 }
 473
 474 sub copy_path($$$$$$$$) {
 475         # Somebody copied a whole subdirectory.
 476         # We need to find the index entries from the old version which the
 477         # SVN log entry points to, and add them to the new place.
 478
 479         my($newrev,$newbranch,$path,$oldpath,$rev,$node_kind,$new,$parents) = @_;
 480
 481         my($srcbranch,$srcpath) = split_path($rev,$oldpath);
 482         unless(defined $srcbranch) {
 483                 print "Path not found when copying from $oldpath @ $rev\n";
 484                 return;
 485         }
 486         my $therev = branch_rev($srcbranch, $rev);
 487         my $gitrev = $branches{$srcbranch}{$therev};
 488         unless($gitrev) {
 489                 print STDERR "$newrev:$newbranch: could not find $oldpath \@ $rev\n";
 490                 return;
 491         }
 492         if ($srcbranch ne $newbranch) {
 493                 push(@$parents, $branches{$srcbranch}{'LAST'});
 494         }
 495         print "$newrev:$newbranch:$path: copying from $srcbranch:$srcpath @ $rev\n" if $opt_v;
 496         if ($node_kind eq $SVN::Node::dir) {
 497                         $srcpath =~ s#/*$#/#;
 498         }
 499
 500         my $pid = open my $f,'-|';
 501         die $! unless defined $pid;
 502         if (!$pid) {
 503                 exec("git-ls-tree","-r","-z",$gitrev,$srcpath)
 504                         or die $!;
 505         }
 506         local $/ = "\0";
 507         while(<$f>) {
 508                 chomp;
 509                 my($m,$p) = split(/\t/,$_,2);
 510                 my($mode,$type,$sha1) = split(/ /,$m);
 511                 next if $type ne "blob";
 512                 if ($node_kind eq $SVN::Node::dir) {
 513                         $p = $path . substr($p,length($srcpath)-1);
 514                 } else {
 515                         $p = $path;
 516                 }
 517                 push(@$new,[$mode,$sha1,$p]);
 518         }
 519         close($f) or
 520                 print STDERR "$newrev:$newbranch: could not list files in $oldpath \@ $rev\n";
 521 }
 522
 523 sub commit {
 524         my($branch, $changed_paths, $revision, $author, $date, $message) = @_;
 525         my($author_name,$author_email,$dest);
 526         my(@old,@new,@parents);
 527
 528         if (not defined $author) {
 529                 $author_name = $author_email = "unknown";
 530         } elsif (defined $users_file) {
 531                 die "User $author is not listed in $users_file\n"
 532                     unless exists $users{$author};
 533                 ($author_name,$author_email) = @{$users{$author}};
 534         } elsif ($author =~ /^(.*?)\s+<(.*)>$/) {
 535                 ($author_name, $author_email) = ($1, $2);
 536         } else {
 537                 $author =~ s/^<(.*)>$/$1/;
 538                 $author_name = $author_email = $author;
 539         }
 540         $date = pdate($date);
 541
 542         my $tag;
 543         my $parent;
 544         if($branch eq "/") { # trunk
 545                 $parent = $opt_o;
 546         } elsif($branch =~ m#^/(.+)#) { # tag
 547                 $tag = 1;
 548                 $parent = $1;
 549         } else { # "normal" branch
 550                 # nothing to do
 551                 $parent = $branch;
 552         }
 553         $dest = $parent;
 554
 555         my $prev = $changed_paths->{"/"};
 556         if($prev and $prev->[0] eq "A") {
 557                 delete $changed_paths->{"/"};
 558                 my $oldpath = $prev->[1];
 559                 my $rev;
 560                 if(defined $oldpath) {
 561                         my $p;
 562                         ($parent,$p) = split_path($revision,$oldpath);
 563                         if($parent eq "/") {
 564                                 $parent = $opt_o;
 565                         } else {
 566                                 $parent =~ s#^/##; # if it's a tag
 567                         }
 568                 } else {
 569                         $parent = undef;
 570                 }
 571         }
 572
 573         my $rev;
 574         if($revision > $opt_s and defined $parent) {
 575                 open(H,"git-rev-parse --verify $parent |");
 576                 $rev = <H>;
 577                 close(H) or do {
 578                         print STDERR "$revision: cannot find commit '$parent'!\n";
 579                         return;
 580                 };
 581                 chop $rev;
 582                 if(length($rev) != 40) {
 583                         print STDERR "$revision: cannot find commit '$parent'!\n";
 584                         return;
 585                 }
 586                 $rev = $branches{($parent eq $opt_o) ? "/" : $parent}{"LAST"};
 587                 if($revision != $opt_s and not $rev) {
 588                         print STDERR "$revision: do not know ancestor for '$parent'!\n";
 589                         return;
 590                 }
 591         } else {
 592                 $rev = undef;
 593         }
 594
 595 #       if($prev and $prev->[0] eq "A") {
 596 #               if(not $tag) {
 597 #                       unless(open(H,"> $git_dir/refs/heads/$branch")) {
 598 #                               print STDERR "$revision: Could not create branch $branch: $!\n";
 599 #                               $state=11;
 600 #                               next;
 601 #                       }
 602 #                       print H "$rev\n"
 603 #                               or die "Could not write branch $branch: $!";
 604 #                       close(H)
 605 #                               or die "Could not write branch $branch: $!";
 606 #               }
 607 #       }
 608         if(not defined $rev) {
 609                 unlink($git_index);
 610         } elsif ($rev ne $last_rev) {
 611                 print "Switching from $last_rev to $rev ($branch)\n" if $opt_v;
 612                 system("git-read-tree", $rev);
 613                 die "read-tree failed for $rev: $?\n" if $?;
 614                 $last_rev = $rev;
 615         }
 616
 617         push (@parents, $rev) if defined $rev;
 618
 619         my $cid;
 620         if($tag and not %$changed_paths) {
 621                 $cid = $rev;
 622         } else {
 623                 my @paths = sort keys %$changed_paths;
 624                 foreach my $path(@paths) {
 625                         my $action = $changed_paths->{$path};
 626
 627                         if ($action->[0] eq "R") {
 628                                 # refer to a file/tree in an earlier commit
 629                                 push(@old,$path); # remove any old stuff
 630                         }
 631                         if(($action->[0] eq "A") || ($action->[0] eq "R")) {
 632                                 my $node_kind = node_kind($branch,$path,$revision);
 633                                 if ($node_kind eq $SVN::Node::file) {
 634                                         my $f = get_file($revision,$branch,$path);
 635                                         if ($f) {
 636                                                 push(@new,$f) if $f;
 637                                         } else {
 638                                                 my $opath = $action->[3];
 639                                                 print STDERR "$revision: $branch: could not fetch '$opath'\n";
 640                                         }
 641                                 } elsif ($node_kind eq $SVN::Node::dir) {
 642                                         if($action->[1]) {
 643                                                 copy_path($revision, $branch,
 644                                                           $path, $action->[1],
 645                                                           $action->[2], $node_kind,
 646                                                           \@new, \@parents);
 647                                         } else {
 648                                                 get_ignore(\@new, \@old, $revision,
 649                                                            $branch, $path);
 650                                         }
 651                                 }
 652                         } elsif ($action->[0] eq "D") {
 653                                 push(@old,$path);
 654                         } elsif ($action->[0] eq "M") {
 655                                 my $node_kind = node_kind($branch,$path,$revision);
 656                                 if ($node_kind eq $SVN::Node::file) {
 657                                         my $f = get_file($revision,$branch,$path);
 658                                         push(@new,$f) if $f;
 659                                 } elsif ($node_kind eq $SVN::Node::dir) {
 660                                         get_ignore(\@new, \@old, $revision,
 661                                                    $branch,$path);
 662                                 }
 663                         } else {
 664                                 die "$revision: unknown action '".$action->[0]."' for $path\n";
 665                         }
 666                 }
 667
 668                 while(@old) {
 669                         my @o1;
 670                         if(@old > 55) {
 671                                 @o1 = splice(@old,0,50);
 672                         } else {
 673                                 @o1 = @old;
 674                                 @old = ();
 675                         }
 676                         my $pid = open my $F, "-|";
 677                         die "$!" unless defined $pid;
 678                         if (!$pid) {
 679                                 exec("git-ls-files", "-z", @o1) or die $!;
 680                         }
 681                         @o1 = ();
 682                         local $/ = "\0";
 683                         while(<$F>) {
 684                                 chomp;
 685                                 push(@o1,$_);
 686                         }
 687                         close($F);
 688
 689                         while(@o1) {
 690                                 my @o2;
 691                                 if(@o1 > 55) {
 692                                         @o2 = splice(@o1,0,50);
 693                                 } else {
 694                                         @o2 = @o1;
 695                                         @o1 = ();
 696                                 }
 697                                 system("git-update-index","--force-remove","--",@o2);
 698                                 die "Cannot remove files: $?\n" if $?;
 699                         }
 700                 }
 701                 while(@new) {
 702                         my @n2;
 703                         if(@new > 12) {
 704                                 @n2 = splice(@new,0,10);
 705                         } else {
 706                                 @n2 = @new;
 707                                 @new = ();
 708                         }
 709                         system("git-update-index","--add",
 710                                 (map { ('--cacheinfo', @$_) } @n2));
 711                         die "Cannot add files: $?\n" if $?;
 712                 }
 713
 714                 my $pid = open(C,"-|");
 715                 die "Cannot fork: $!" unless defined $pid;
 716                 unless($pid) {
 717                         exec("git-write-tree");
 718                         die "Cannot exec git-write-tree: $!\n";
 719                 }
 720                 chomp(my $tree = <C>);
 721                 length($tree) == 40
 722                         or die "Cannot get tree id ($tree): $!\n";
 723                 close(C)
 724                         or die "Error running git-write-tree: $?\n";
 725                 print "Tree ID $tree\n" if $opt_v;
 726
 727                 my $pr = IO::Pipe->new() or die "Cannot open pipe: $!\n";
 728                 my $pw = IO::Pipe->new() or die "Cannot open pipe: $!\n";
 729                 $pid = fork();
 730                 die "Fork: $!\n" unless defined $pid;
 731                 unless($pid) {
 732                         $pr->writer();
 733                         $pw->reader();
 734                         open(OUT,">&STDOUT");
 735                         dup2($pw->fileno(),0);
 736                         dup2($pr->fileno(),1);
 737                         $pr->close();
 738                         $pw->close();
 739
 740                         my @par = ();
 741
 742                         # loose detection of merges
 743                         # based on the commit msg
 744                         foreach my $rx (@mergerx) {
 745                                 if ($message =~ $rx) {
 746                                         my $mparent = $1;
 747                                         if ($mparent eq 'HEAD') { $mparent = $opt_o };
 748                                         if ( -e "$git_dir/refs/heads/$mparent") {
 749                                                 $mparent = get_headref($mparent, $git_dir);
 750                                                 push (@parents, $mparent);
 751                                                 print OUT "Merge parent branch: $mparent\n" if $opt_v;
 752                                         }
 753                                 }
 754                         }
 755                         my %seen_parents = ();
 756                         my @unique_parents = grep { ! $seen_parents{$_} ++ } @parents;
 757                         foreach my $bparent (@unique_parents) {
 758                                 push @par, '-p', $bparent;
 759                                 print OUT "Merge parent branch: $bparent\n" if $opt_v;
 760                         }
 761
 762                         exec("env",
 763                                 "GIT_AUTHOR_NAME=$author_name",
 764                                 "GIT_AUTHOR_EMAIL=$author_email",
 765                                 "GIT_AUTHOR_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
 766                                 "GIT_COMMITTER_NAME=$author_name",
 767                                 "GIT_COMMITTER_EMAIL=$author_email",
 768                                 "GIT_COMMITTER_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
 769                                 "git-commit-tree", $tree,@par);
 770                         die "Cannot exec git-commit-tree: $!\n";
 771                 }
 772                 $pw->writer();
 773                 $pr->reader();
 774
 775                 $message =~ s/[\s\n]+\z//;
 776                 $message = "r$revision: $message" if $opt_r;
 777
 778                 print $pw "$message\n"
 779                         or die "Error writing to git-commit-tree: $!\n";
 780                 $pw->close();
 781
 782                 print "Committed change $revision:$branch ".strftime("%Y-%m-%d %H:%M:%S",gmtime($date)).")\n" if $opt_v;
 783                 chomp($cid = <$pr>);
 784                 length($cid) == 40
 785                         or die "Cannot get commit id ($cid): $!\n";
 786                 print "Commit ID $cid\n" if $opt_v;
 787                 $pr->close();
 788
 789                 waitpid($pid,0);
 790                 die "Error running git-commit-tree: $?\n" if $?;
 791         }
 792
 793         if (not defined $cid) {
 794                 $cid = $branches{"/"}{"LAST"};
 795         }
 796
 797         if(not defined $dest) {
 798                 print "... no known parent\n" if $opt_v;
 799         } elsif(not $tag) {
 800                 print "Writing to refs/heads/$dest\n" if $opt_v;
 801                 open(C,">$git_dir/refs/heads/$dest") and
 802                 print C ("$cid\n") and
 803                 close(C)
 804                         or die "Cannot write branch $dest for update: $!\n";
 805         }
 806
 807         if($tag) {
 808                 my($in, $out) = ('','');
 809                 $last_rev = "-" if %$changed_paths;
 810                 # the tag was 'complex', i.e. did not refer to a "real" revision
 811
 812                 $dest =~ tr/_/\./ if $opt_u;
 813                 $branch = $dest;
 814
 815                 my $pid = open2($in, $out, 'git-mktag');
 816                 print $out ("object $cid\n".
 817                     "type commit\n".
 818                     "tag $dest\n".
 819                     "tagger $author_name <$author_email>\n") and
 820                 close($out)
 821                     or die "Cannot create tag object $dest: $!\n";
 822
 823                 my $tagobj = <$in>;
 824                 chomp $tagobj;
 825
 826                 if ( !close($in) or waitpid($pid, 0) != $pid or
 827                                 $? != 0 or $tagobj !~ /^[0123456789abcdef]{40}$/ ) {
 828                         die "Cannot create tag object $dest: $!\n";
 829                 }
 830
 831                 open(C,">$git_dir/refs/tags/$dest") and
 832                 print C ("$tagobj\n") and
 833                 close(C)
 834                         or die "Cannot create tag $branch: $!\n";
 835
 836                 print "Created tag '$dest' on '$branch'\n" if $opt_v;
 837         }
 838         $branches{$branch}{"LAST"} = $cid;
 839         $branches{$branch}{$revision} = $cid;
 840         $last_rev = $cid;
 841         print BRANCHES "$revision $branch $cid\n";
 842         print "DONE: $revision $dest $cid\n" if $opt_v;
 843 }
 844
 845 sub commit_all {
 846         # Recursive use of the SVN connection does not work
 847         local $svn = $svn2;
 848
 849         my ($changed_paths, $revision, $author, $date, $message, $pool) = @_;
 850         my %p;
 851         while(my($path,$action) = each %$changed_paths) {
 852                 $p{$path} = [ $action->action,$action->copyfrom_path, $action->copyfrom_rev, $path ];
 853         }
 854         $changed_paths = \%p;
 855
 856         my %done;
 857         my @col;
 858         my $pref;
 859         my $branch;
 860
 861         while(my($path,$action) = each %$changed_paths) {
 862                 ($branch,$path) = split_path($revision,$path);
 863                 next if not defined $branch;
 864                 $done{$branch}{$path} = $action;
 865         }
 866         while(($branch,$changed_paths) = each %done) {
 867                 commit($branch, $changed_paths, $revision, $author, $date, $message);
 868         }
 869 }
 870
 871 $opt_l = $svn->{'maxrev'} if not defined $opt_l or $opt_l > $svn->{'maxrev'};
 872
 873 if ($opt_l < $current_rev) {
 874     print "Up to date: no new revisions to fetch!\n" if $opt_v;
 875     unlink("$git_dir/SVN2GIT_HEAD");
 876     exit;
 877 }
 878
 879 print "Fetching from $current_rev to $opt_l ...\n" if $opt_v;
 880
 881 my $pool=SVN::Pool->new;
 882 $svn->{'svn'}->get_log("/",$current_rev,$opt_l,0,1,1,\&commit_all,$pool);
 883 $pool->clear;
 884
 885
 886 unlink($git_index);
 887
 888 if (defined $orig_git_index) {
 889         $ENV{GIT_INDEX_FILE} = $orig_git_index;
 890 } else {
 891         delete $ENV{GIT_INDEX_FILE};
 892 }
 893
 894 # Now switch back to the branch we were in before all of this happened
 895 if($orig_branch) {
 896         print "DONE\n" if $opt_v and (not defined $opt_l or $opt_l > 0);
 897         system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
 898                 if $forward_master;
 899         unless ($opt_i) {
 900                 system('git-read-tree', '-m', '-u', 'SVN2GIT_HEAD', 'HEAD');
 901                 die "read-tree failed: $?\n" if $?;
 902         }
 903 } else {
 904         $orig_branch = "master";
 905         print "DONE; creating $orig_branch branch\n" if $opt_v and (not defined $opt_l or $opt_l > 0);
 906         system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
 907                 unless -f "$git_dir/refs/heads/master";
 908         system('git-update-ref', 'HEAD', "$orig_branch");
 909         unless ($opt_i) {
 910                 system('git checkout');
 911                 die "checkout failed: $?\n" if $?;
 912         }
 913 }
 914 unlink("$git_dir/SVN2GIT_HEAD");
 915 close(BRANCHES);