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,$opt_S,$opt_F,
  35     $opt_P,$opt_R);
  36
  37 sub usage() {
  38         print STDERR <<END;
  39 Usage: ${\basename $0}     # fetch/update GIT from SVN
  40        [-o branch-for-HEAD] [-h] [-v] [-l max_rev] [-R repack_each_revs]
  41        [-C GIT_repository] [-t tagname] [-T trunkname] [-b branchname]
  42        [-d|-D] [-i] [-u] [-r] [-I ignorefilename] [-s start_chg]
  43        [-m] [-M regex] [-A author_file] [-S] [-F] [-P project_name] [SVN_URL]
  44 END
  45         exit(1);
  46 }
  47
  48 getopts("A:b:C:dDFhiI:l:mM:o:rs:t:T:SP:R:uv") or usage();
  49 usage if $opt_h;
  50
  51 my $tag_name = $opt_t || "tags";
  52 my $trunk_name = $opt_T || "trunk";
  53 my $branch_name = $opt_b || "branches";
  54 my $project_name = $opt_P || "";
  55 $project_name = "/" . $project_name if ($project_name);
  56 my $repack_after = $opt_R || 1000;
  57
  58 @ARGV == 1 or @ARGV == 2 or usage();
  59
  60 $opt_o ||= "origin";
  61 $opt_s ||= 1;
  62 my $git_tree = $opt_C;
  63 $git_tree ||= ".";
  64
  65 my $svn_url = $ARGV[0];
  66 my $svn_dir = $ARGV[1];
  67
  68 our @mergerx = ();
  69 if ($opt_m) {
  70         my $branch_esc = quotemeta ($branch_name);
  71         my $trunk_esc  = quotemeta ($trunk_name);
  72         @mergerx =
  73         (
  74                 qr!\b(?:merg(?:ed?|ing))\b.*?\b((?:(?<=$branch_esc/)[\w\.\-]+)|(?:$trunk_esc))\b!i,
  75                 qr!\b(?:from|of)\W+((?:(?<=$branch_esc/)[\w\.\-]+)|(?:$trunk_esc))\b!i,
  76                 qr!\b(?:from|of)\W+(?:the )?([\w\.\-]+)[-\s]branch\b!i
  77         );
  78 }
  79 if ($opt_M) {
  80         unshift (@mergerx, qr/$opt_M/);
  81 }
  82
  83 # Absolutize filename now, since we will have chdir'ed by the time we
  84 # get around to opening it.
  85 $opt_A = File::Spec->rel2abs($opt_A) if $opt_A;
  86
  87 our %users = ();
  88 our $users_file = undef;
  89 sub read_users($) {
  90         $users_file = File::Spec->rel2abs(@_);
  91         die "Cannot open $users_file\n" unless -f $users_file;
  92         open(my $authors,$users_file);
  93         while(<$authors>) {
  94                 chomp;
  95                 next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
  96                 (my $user,my $name,my $email) = ($1,$2,$3);
  97                 $users{$user} = [$name,$email];
  98         }
  99         close($authors);
 100 }
 101
 102 select(STDERR); $|=1; select(STDOUT);
 103
 104
 105 package SVNconn;
 106 # Basic SVN connection.
 107 # We're only interested in connecting and downloading, so ...
 108
 109 use File::Spec;
 110 use File::Temp qw(tempfile);
 111 use POSIX qw(strftime dup2);
 112 use Fcntl qw(SEEK_SET);
 113
 114 sub new {
 115         my($what,$repo) = @_;
 116         $what=ref($what) if ref($what);
 117
 118         my $self = {};
 119         $self->{'buffer'} = "";
 120         bless($self,$what);
 121
 122         $repo =~ s#/+$##;
 123         $self->{'fullrep'} = $repo;
 124         $self->conn();
 125
 126         return $self;
 127 }
 128
 129 sub conn {
 130         my $self = shift;
 131         my $repo = $self->{'fullrep'};
 132         my $auth = SVN::Core::auth_open ([SVN::Client::get_simple_provider,
 133                           SVN::Client::get_ssl_server_trust_file_provider,
 134                           SVN::Client::get_username_provider]);
 135         my $s = SVN::Ra->new(url => $repo, auth => $auth);
 136         die "SVN connection to $repo: $!\n" unless defined $s;
 137         $self->{'svn'} = $s;
 138         $self->{'repo'} = $repo;
 139         $self->{'maxrev'} = $s->get_latest_revnum();
 140 }
 141
 142 sub file {
 143         my($self,$path,$rev) = @_;
 144
 145         my ($fh, $name) = tempfile('gitsvn.XXXXXX',
 146                     DIR => File::Spec->tmpdir(), UNLINK => 1);
 147
 148         print "... $rev $path ...\n" if $opt_v;
 149         my (undef, $properties);
 150         my $pool = SVN::Pool->new();
 151         eval { (undef, $properties)
 152                    = $self->{'svn'}->get_file($path,$rev,$fh,$pool); };
 153         $pool->clear;
 154         if($@) {
 155                 return undef if $@ =~ /Attempted to get checksum/;
 156                 die $@;
 157         }
 158         my $mode;
 159         if (exists $properties->{'svn:executable'}) {
 160                 $mode = '100755';
 161         } elsif (exists $properties->{'svn:special'}) {
 162                 my ($special_content, $filesize);
 163                 $filesize = tell $fh;
 164                 seek $fh, 0, SEEK_SET;
 165                 read $fh, $special_content, $filesize;
 166                 if ($special_content =~ s/^link //) {
 167                         $mode = '120000';
 168                         seek $fh, 0, SEEK_SET;
 169                         truncate $fh, 0;
 170                         print $fh $special_content;
 171                 } else {
 172                         die "unexpected svn:special file encountered";
 173                 }
 174         } else {
 175                 $mode = '100644';
 176         }
 177         close ($fh);
 178
 179         return ($name, $mode);
 180 }
 181
 182 sub ignore {
 183         my($self,$path,$rev) = @_;
 184
 185         print "... $rev $path ...\n" if $opt_v;
 186         my (undef,undef,$properties)
 187             = $self->{'svn'}->get_dir($path,$rev,undef);
 188         if (exists $properties->{'svn:ignore'}) {
 189                 my ($fh, $name) = tempfile('gitsvn.XXXXXX',
 190                                            DIR => File::Spec->tmpdir(),
 191                                            UNLINK => 1);
 192                 print $fh $properties->{'svn:ignore'};
 193                 close($fh);
 194                 return $name;
 195         } else {
 196                 return undef;
 197         }
 198 }
 199
 200 sub dir_list {
 201         my($self,$path,$rev) = @_;
 202         my ($dirents,undef,$properties)
 203             = $self->{'svn'}->get_dir($path,$rev,undef);
 204         return $dirents;
 205 }
 206
 207 package main;
 208 use URI;
 209
 210 our $svn = $svn_url;
 211 $svn .= "/$svn_dir" if defined $svn_dir;
 212 my $svn2 = SVNconn->new($svn);
 213 $svn = SVNconn->new($svn);
 214
 215 my $lwp_ua;
 216 if($opt_d or $opt_D) {
 217         $svn_url = URI->new($svn_url)->canonical;
 218         if($opt_D) {
 219                 $svn_dir =~ s#/*$#/#;
 220         } else {
 221                 $svn_dir = "";
 222         }
 223         if ($svn_url->scheme eq "http") {
 224                 use LWP::UserAgent;
 225                 $lwp_ua = LWP::UserAgent->new(keep_alive => 1, requests_redirectable => []);
 226         } else {
 227                 print STDERR "Warning: not HTTP; turning off direct file access\n";
 228                 $opt_d=0;
 229         }
 230 }
 231
 232 sub pdate($) {
 233         my($d) = @_;
 234         $d =~ m#(\d\d\d\d)-(\d\d)-(\d\d)T(\d\d):(\d\d):(\d\d)#
 235                 or die "Unparseable date: $d\n";
 236         my $y=$1; $y-=1900 if $y>1900;
 237         return timegm($6||0,$5,$4,$3,$2-1,$y);
 238 }
 239
 240 sub getwd() {
 241         my $pwd = `pwd`;
 242         chomp $pwd;
 243         return $pwd;
 244 }
 245
 246
 247 sub get_headref($$) {
 248     my $name    = shift;
 249     my $git_dir = shift;
 250     my $sha;
 251
 252     if (open(C,"$git_dir/refs/heads/$name")) {
 253         chomp($sha = <C>);
 254         close(C);
 255         length($sha) == 40
 256             or die "Cannot get head id for $name ($sha): $!\n";
 257     }
 258     return $sha;
 259 }
 260
 261
 262 -d $git_tree
 263         or mkdir($git_tree,0777)
 264         or die "Could not create $git_tree: $!";
 265 chdir($git_tree);
 266
 267 my $orig_branch = "";
 268 my $forward_master = 0;
 269 my %branches;
 270
 271 my $git_dir = $ENV{"GIT_DIR"} || ".git";
 272 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
 273 $ENV{"GIT_DIR"} = $git_dir;
 274 my $orig_git_index;
 275 $orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
 276 my ($git_ih, $git_index) = tempfile('gitXXXXXX', SUFFIX => '.idx',
 277                                     DIR => File::Spec->tmpdir());
 278 close ($git_ih);
 279 $ENV{GIT_INDEX_FILE} = $git_index;
 280 my $maxnum = 0;
 281 my $last_rev = "";
 282 my $last_branch;
 283 my $current_rev = $opt_s || 1;
 284 unless(-d $git_dir) {
 285         system("git-init-db");
 286         die "Cannot init the GIT db at $git_tree: $?\n" if $?;
 287         system("git-read-tree");
 288         die "Cannot init an empty tree: $?\n" if $?;
 289
 290         $last_branch = $opt_o;
 291         $orig_branch = "";
 292 } else {
 293         -f "$git_dir/refs/heads/$opt_o"
 294                 or die "Branch '$opt_o' does not exist.\n".
 295                        "Either use the correct '-o branch' option,\n".
 296                        "or import to a new repository.\n";
 297
 298         -f "$git_dir/svn2git"
 299                 or die "'$git_dir/svn2git' does not exist.\n".
 300                        "You need that file for incremental imports.\n";
 301         open(F, "git-symbolic-ref HEAD |") or
 302                 die "Cannot run git-symbolic-ref: $!\n";
 303         chomp ($last_branch = <F>);
 304         $last_branch = basename($last_branch);
 305         close(F);
 306         unless($last_branch) {
 307                 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
 308                 $last_branch = "master";
 309         }
 310         $orig_branch = $last_branch;
 311         $last_rev = get_headref($orig_branch, $git_dir);
 312         if (-f "$git_dir/SVN2GIT_HEAD") {
 313                 die <<EOM;
 314 SVN2GIT_HEAD exists.
 315 Make sure your working directory corresponds to HEAD and remove SVN2GIT_HEAD.
 316 You may need to run
 317
 318     git-read-tree -m -u SVN2GIT_HEAD HEAD
 319 EOM
 320         }
 321         system('cp', "$git_dir/HEAD", "$git_dir/SVN2GIT_HEAD");
 322
 323         $forward_master =
 324             $opt_o ne 'master' && -f "$git_dir/refs/heads/master" &&
 325             system('cmp', '-s', "$git_dir/refs/heads/master",
 326                                 "$git_dir/refs/heads/$opt_o") == 0;
 327
 328         # populate index
 329         system('git-read-tree', $last_rev);
 330         die "read-tree failed: $?\n" if $?;
 331
 332         # Get the last import timestamps
 333         open my $B,"<", "$git_dir/svn2git";
 334         while(<$B>) {
 335                 chomp;
 336                 my($num,$branch,$ref) = split;
 337                 $branches{$branch}{$num} = $ref;
 338                 $branches{$branch}{"LAST"} = $ref;
 339                 $current_rev = $num+1 if $current_rev <= $num;
 340         }
 341         close($B);
 342 }
 343 -d $git_dir
 344         or die "Could not create git subdir ($git_dir).\n";
 345
 346 my $default_authors = "$git_dir/svn-authors";
 347 if ($opt_A) {
 348         read_users($opt_A);
 349         copy($opt_A,$default_authors) or die "Copy failed: $!";
 350 } else {
 351         read_users($default_authors) if -f $default_authors;
 352 }
 353
 354 open BRANCHES,">>", "$git_dir/svn2git";
 355
 356 sub node_kind($$) {
 357         my ($svnpath, $revision) = @_;
 358         my $pool=SVN::Pool->new;
 359         my $kind = $svn->{'svn'}->check_path($svnpath,$revision,$pool);
 360         $pool->clear;
 361         return $kind;
 362 }
 363
 364 sub get_file($$$) {
 365         my($svnpath,$rev,$path) = @_;
 366
 367         # now get it
 368         my ($name,$mode);
 369         if($opt_d) {
 370                 my($req,$res);
 371
 372                 # /svn/!svn/bc/2/django/trunk/django-docs/build.py
 373                 my $url=$svn_url->clone();
 374                 $url->path($url->path."/!svn/bc/$rev/$svn_dir$svnpath");
 375                 print "... $path...\n" if $opt_v;
 376                 $req = HTTP::Request->new(GET => $url);
 377                 $res = $lwp_ua->request($req);
 378                 if ($res->is_success) {
 379                         my $fh;
 380                         ($fh, $name) = tempfile('gitsvn.XXXXXX',
 381                         DIR => File::Spec->tmpdir(), UNLINK => 1);
 382                         print $fh $res->content;
 383                         close($fh) or die "Could not write $name: $!\n";
 384                 } else {
 385                         return undef if $res->code == 301; # directory?
 386                         die $res->status_line." at $url\n";
 387                 }
 388                 $mode = '0644'; # can't obtain mode via direct http request?
 389         } else {
 390                 ($name,$mode) = $svn->file("$svnpath",$rev);
 391                 return undef unless defined $name;
 392         }
 393
 394         my $pid = open(my $F, '-|');
 395         die $! unless defined $pid;
 396         if (!$pid) {
 397             exec("git-hash-object", "-w", $name)
 398                 or die "Cannot create object: $!\n";
 399         }
 400         my $sha = <$F>;
 401         chomp $sha;
 402         close $F;
 403         unlink $name;
 404         return [$mode, $sha, $path];
 405 }
 406
 407 sub get_ignore($$$$$) {
 408         my($new,$old,$rev,$path,$svnpath) = @_;
 409
 410         return unless $opt_I;
 411         my $name = $svn->ignore("$svnpath",$rev);
 412         if ($path eq '/') {
 413                 $path = $opt_I;
 414         } else {
 415                 $path = File::Spec->catfile($path,$opt_I);
 416         }
 417         if (defined $name) {
 418                 my $pid = open(my $F, '-|');
 419                 die $! unless defined $pid;
 420                 if (!$pid) {
 421                         exec("git-hash-object", "-w", $name)
 422                             or die "Cannot create object: $!\n";
 423                 }
 424                 my $sha = <$F>;
 425                 chomp $sha;
 426                 close $F;
 427                 unlink $name;
 428                 push(@$new,['0644',$sha,$path]);
 429         } elsif (defined $old) {
 430                 push(@$old,$path);
 431         }
 432 }
 433
 434 sub project_path($$)
 435 {
 436         my ($path, $project) = @_;
 437
 438         $path = "/".$path unless ($path =~ m#^\/#) ;
 439         return $1 if ($path =~ m#^$project\/(.*)$#);
 440
 441         $path =~ s#\.#\\\.#g;
 442         $path =~ s#\+#\\\+#g;
 443         return "/" if ($project =~ m#^$path.*$#);
 444
 445         return undef;
 446 }
 447
 448 sub split_path($$) {
 449         my($rev,$path) = @_;
 450         my $branch;
 451
 452         if($path =~ s#^/\Q$tag_name\E/([^/]+)/?##) {
 453                 $branch = "/$1";
 454         } elsif($path =~ s#^/\Q$trunk_name\E/?##) {
 455                 $branch = "/";
 456         } elsif($path =~ s#^/\Q$branch_name\E/([^/]+)/?##) {
 457                 $branch = $1;
 458         } else {
 459                 my %no_error = (
 460                         "/" => 1,
 461                         "/$tag_name" => 1,
 462                         "/$branch_name" => 1
 463                 );
 464                 print STDERR "$rev: Unrecognized path: $path\n" unless (defined $no_error{$path});
 465                 return ()
 466         }
 467         if ($path eq "") {
 468                 $path = "/";
 469         } elsif ($project_name) {
 470                 $path = project_path($path, $project_name);
 471         }
 472         return ($branch,$path);
 473 }
 474
 475 sub branch_rev($$) {
 476
 477         my ($srcbranch,$uptorev) = @_;
 478
 479         my $bbranches = $branches{$srcbranch};
 480         my @revs = reverse sort { ($a eq 'LAST' ? 0 : $a) <=> ($b eq 'LAST' ? 0 : $b) } keys %$bbranches;
 481         my $therev;
 482         foreach my $arev(@revs) {
 483                 next if  ($arev eq 'LAST');
 484                 if ($arev <= $uptorev) {
 485                         $therev = $arev;
 486                         last;
 487                 }
 488         }
 489         return $therev;
 490 }
 491
 492 sub expand_svndir($$$);
 493
 494 sub expand_svndir($$$)
 495 {
 496         my ($svnpath, $rev, $path) = @_;
 497         my @list;
 498         get_ignore(\@list, undef, $rev, $path, $svnpath);
 499         my $dirents = $svn->dir_list($svnpath, $rev);
 500         foreach my $p(keys %$dirents) {
 501                 my $kind = node_kind($svnpath.'/'.$p, $rev);
 502                 if ($kind eq $SVN::Node::file) {
 503                         my $f = get_file($svnpath.'/'.$p, $rev, $path.'/'.$p);
 504                         push(@list, $f) if $f;
 505                 } elsif ($kind eq $SVN::Node::dir) {
 506                         push(@list,
 507                              expand_svndir($svnpath.'/'.$p, $rev, $path.'/'.$p));
 508                 }
 509         }
 510         return @list;
 511 }
 512
 513 sub copy_path($$$$$$$$) {
 514         # Somebody copied a whole subdirectory.
 515         # We need to find the index entries from the old version which the
 516         # SVN log entry points to, and add them to the new place.
 517
 518         my($newrev,$newbranch,$path,$oldpath,$rev,$node_kind,$new,$parents) = @_;
 519
 520         my($srcbranch,$srcpath) = split_path($rev,$oldpath);
 521         unless(defined $srcbranch && defined $srcpath) {
 522                 print "Path not found when copying from $oldpath @ $rev.\n".
 523                         "Will try to copy from original SVN location...\n"
 524                         if $opt_v;
 525                 push (@$new, expand_svndir($oldpath, $rev, $path));
 526                 return;
 527         }
 528         my $therev = branch_rev($srcbranch, $rev);
 529         my $gitrev = $branches{$srcbranch}{$therev};
 530         unless($gitrev) {
 531                 print STDERR "$newrev:$newbranch: could not find $oldpath \@ $rev\n";
 532                 return;
 533         }
 534         if ($srcbranch ne $newbranch) {
 535                 push(@$parents, $branches{$srcbranch}{'LAST'});
 536         }
 537         print "$newrev:$newbranch:$path: copying from $srcbranch:$srcpath @ $rev\n" if $opt_v;
 538         if ($node_kind eq $SVN::Node::dir) {
 539                 $srcpath =~ s#/*$#/#;
 540         }
 541
 542         my $pid = open my $f,'-|';
 543         die $! unless defined $pid;
 544         if (!$pid) {
 545                 exec("git-ls-tree","-r","-z",$gitrev,$srcpath)
 546                         or die $!;
 547         }
 548         local $/ = "\0";
 549         while(<$f>) {
 550                 chomp;
 551                 my($m,$p) = split(/\t/,$_,2);
 552                 my($mode,$type,$sha1) = split(/ /,$m);
 553                 next if $type ne "blob";
 554                 if ($node_kind eq $SVN::Node::dir) {
 555                         $p = $path . substr($p,length($srcpath)-1);
 556                 } else {
 557                         $p = $path;
 558                 }
 559                 push(@$new,[$mode,$sha1,$p]);
 560         }
 561         close($f) or
 562                 print STDERR "$newrev:$newbranch: could not list files in $oldpath \@ $rev\n";
 563 }
 564
 565 sub commit {
 566         my($branch, $changed_paths, $revision, $author, $date, $message) = @_;
 567         my($committer_name,$committer_email,$dest);
 568         my($author_name,$author_email);
 569         my(@old,@new,@parents);
 570
 571         if (not defined $author or $author eq "") {
 572                 $committer_name = $committer_email = "unknown";
 573         } elsif (defined $users_file) {
 574                 die "User $author is not listed in $users_file\n"
 575                     unless exists $users{$author};
 576                 ($committer_name,$committer_email) = @{$users{$author}};
 577         } elsif ($author =~ /^(.*?)\s+<(.*)>$/) {
 578                 ($committer_name, $committer_email) = ($1, $2);
 579         } else {
 580                 $author =~ s/^<(.*)>$/$1/;
 581                 $committer_name = $committer_email = $author;
 582         }
 583
 584         if ($opt_F && $message =~ /From:\s+(.*?)\s+<(.*)>\s*\n/) {
 585                 ($author_name, $author_email) = ($1, $2);
 586                 print "Author from From: $1 <$2>\n" if ($opt_v);;
 587         } elsif ($opt_S && $message =~ /Signed-off-by:\s+(.*?)\s+<(.*)>\s*\n/) {
 588                 ($author_name, $author_email) = ($1, $2);
 589                 print "Author from Signed-off-by: $1 <$2>\n" if ($opt_v);;
 590         } else {
 591                 $author_name = $committer_name;
 592                 $author_email = $committer_email;
 593         }
 594
 595         $date = pdate($date);
 596
 597         my $tag;
 598         my $parent;
 599         if($branch eq "/") { # trunk
 600                 $parent = $opt_o;
 601         } elsif($branch =~ m#^/(.+)#) { # tag
 602                 $tag = 1;
 603                 $parent = $1;
 604         } else { # "normal" branch
 605                 # nothing to do
 606                 $parent = $branch;
 607         }
 608         $dest = $parent;
 609
 610         my $prev = $changed_paths->{"/"};
 611         if($prev and $prev->[0] eq "A") {
 612                 delete $changed_paths->{"/"};
 613                 my $oldpath = $prev->[1];
 614                 my $rev;
 615                 if(defined $oldpath) {
 616                         my $p;
 617                         ($parent,$p) = split_path($revision,$oldpath);
 618                         if(defined $parent) {
 619                                 if($parent eq "/") {
 620                                         $parent = $opt_o;
 621                                 } else {
 622                                         $parent =~ s#^/##; # if it's a tag
 623                                 }
 624                         }
 625                 } else {
 626                         $parent = undef;
 627                 }
 628         }
 629
 630         my $rev;
 631         if($revision > $opt_s and defined $parent) {
 632                 open(H,"git-rev-parse --verify $parent |");
 633                 $rev = <H>;
 634                 close(H) or do {
 635                         print STDERR "$revision: cannot find commit '$parent'!\n";
 636                         return;
 637                 };
 638                 chop $rev;
 639                 if(length($rev) != 40) {
 640                         print STDERR "$revision: cannot find commit '$parent'!\n";
 641                         return;
 642                 }
 643                 $rev = $branches{($parent eq $opt_o) ? "/" : $parent}{"LAST"};
 644                 if($revision != $opt_s and not $rev) {
 645                         print STDERR "$revision: do not know ancestor for '$parent'!\n";
 646                         return;
 647                 }
 648         } else {
 649                 $rev = undef;
 650         }
 651
 652 #       if($prev and $prev->[0] eq "A") {
 653 #               if(not $tag) {
 654 #                       unless(open(H,"> $git_dir/refs/heads/$branch")) {
 655 #                               print STDERR "$revision: Could not create branch $branch: $!\n";
 656 #                               $state=11;
 657 #                               next;
 658 #                       }
 659 #                       print H "$rev\n"
 660 #                               or die "Could not write branch $branch: $!";
 661 #                       close(H)
 662 #                               or die "Could not write branch $branch: $!";
 663 #               }
 664 #       }
 665         if(not defined $rev) {
 666                 unlink($git_index);
 667         } elsif ($rev ne $last_rev) {
 668                 print "Switching from $last_rev to $rev ($branch)\n" if $opt_v;
 669                 system("git-read-tree", $rev);
 670                 die "read-tree failed for $rev: $?\n" if $?;
 671                 $last_rev = $rev;
 672         }
 673
 674         push (@parents, $rev) if defined $rev;
 675
 676         my $cid;
 677         if($tag and not %$changed_paths) {
 678                 $cid = $rev;
 679         } else {
 680                 my @paths = sort keys %$changed_paths;
 681                 foreach my $path(@paths) {
 682                         my $action = $changed_paths->{$path};
 683
 684                         if ($action->[0] eq "R") {
 685                                 # refer to a file/tree in an earlier commit
 686                                 push(@old,$path); # remove any old stuff
 687                         }
 688                         if(($action->[0] eq "A") || ($action->[0] eq "R")) {
 689                                 my $node_kind = node_kind($action->[3], $revision);
 690                                 if ($node_kind eq $SVN::Node::file) {
 691                                         my $f = get_file($action->[3],
 692                                                          $revision, $path);
 693                                         if ($f) {
 694                                                 push(@new,$f) if $f;
 695                                         } else {
 696                                                 my $opath = $action->[3];
 697                                                 print STDERR "$revision: $branch: could not fetch '$opath'\n";
 698                                         }
 699                                 } elsif ($node_kind eq $SVN::Node::dir) {
 700                                         if($action->[1]) {
 701                                                 copy_path($revision, $branch,
 702                                                           $path, $action->[1],
 703                                                           $action->[2], $node_kind,
 704                                                           \@new, \@parents);
 705                                         } else {
 706                                                 get_ignore(\@new, \@old, $revision,
 707                                                            $path, $action->[3]);
 708                                         }
 709                                 }
 710                         } elsif ($action->[0] eq "D") {
 711                                 push(@old,$path);
 712                         } elsif ($action->[0] eq "M") {
 713                                 my $node_kind = node_kind($action->[3], $revision);
 714                                 if ($node_kind eq $SVN::Node::file) {
 715                                         my $f = get_file($action->[3],
 716                                                          $revision, $path);
 717                                         push(@new,$f) if $f;
 718                                 } elsif ($node_kind eq $SVN::Node::dir) {
 719                                         get_ignore(\@new, \@old, $revision,
 720                                                    $path, $action->[3]);
 721                                 }
 722                         } else {
 723                                 die "$revision: unknown action '".$action->[0]."' for $path\n";
 724                         }
 725                 }
 726
 727                 while(@old) {
 728                         my @o1;
 729                         if(@old > 55) {
 730                                 @o1 = splice(@old,0,50);
 731                         } else {
 732                                 @o1 = @old;
 733                                 @old = ();
 734                         }
 735                         my $pid = open my $F, "-|";
 736                         die "$!" unless defined $pid;
 737                         if (!$pid) {
 738                                 exec("git-ls-files", "-z", @o1) or die $!;
 739                         }
 740                         @o1 = ();
 741                         local $/ = "\0";
 742                         while(<$F>) {
 743                                 chomp;
 744                                 push(@o1,$_);
 745                         }
 746                         close($F);
 747
 748                         while(@o1) {
 749                                 my @o2;
 750                                 if(@o1 > 55) {
 751                                         @o2 = splice(@o1,0,50);
 752                                 } else {
 753                                         @o2 = @o1;
 754                                         @o1 = ();
 755                                 }
 756                                 system("git-update-index","--force-remove","--",@o2);
 757                                 die "Cannot remove files: $?\n" if $?;
 758                         }
 759                 }
 760                 while(@new) {
 761                         my @n2;
 762                         if(@new > 12) {
 763                                 @n2 = splice(@new,0,10);
 764                         } else {
 765                                 @n2 = @new;
 766                                 @new = ();
 767                         }
 768                         system("git-update-index","--add",
 769                                 (map { ('--cacheinfo', @$_) } @n2));
 770                         die "Cannot add files: $?\n" if $?;
 771                 }
 772
 773                 my $pid = open(C,"-|");
 774                 die "Cannot fork: $!" unless defined $pid;
 775                 unless($pid) {
 776                         exec("git-write-tree");
 777                         die "Cannot exec git-write-tree: $!\n";
 778                 }
 779                 chomp(my $tree = <C>);
 780                 length($tree) == 40
 781                         or die "Cannot get tree id ($tree): $!\n";
 782                 close(C)
 783                         or die "Error running git-write-tree: $?\n";
 784                 print "Tree ID $tree\n" if $opt_v;
 785
 786                 my $pr = IO::Pipe->new() or die "Cannot open pipe: $!\n";
 787                 my $pw = IO::Pipe->new() or die "Cannot open pipe: $!\n";
 788                 $pid = fork();
 789                 die "Fork: $!\n" unless defined $pid;
 790                 unless($pid) {
 791                         $pr->writer();
 792                         $pw->reader();
 793                         open(OUT,">&STDOUT");
 794                         dup2($pw->fileno(),0);
 795                         dup2($pr->fileno(),1);
 796                         $pr->close();
 797                         $pw->close();
 798
 799                         my @par = ();
 800
 801                         # loose detection of merges
 802                         # based on the commit msg
 803                         foreach my $rx (@mergerx) {
 804                                 if ($message =~ $rx) {
 805                                         my $mparent = $1;
 806                                         if ($mparent eq 'HEAD') { $mparent = $opt_o };
 807                                         if ( -e "$git_dir/refs/heads/$mparent") {
 808                                                 $mparent = get_headref($mparent, $git_dir);
 809                                                 push (@parents, $mparent);
 810                                                 print OUT "Merge parent branch: $mparent\n" if $opt_v;
 811                                         }
 812                                 }
 813                         }
 814                         my %seen_parents = ();
 815                         my @unique_parents = grep { ! $seen_parents{$_} ++ } @parents;
 816                         foreach my $bparent (@unique_parents) {
 817                                 push @par, '-p', $bparent;
 818                                 print OUT "Merge parent branch: $bparent\n" if $opt_v;
 819                         }
 820
 821                         exec("env",
 822                                 "GIT_AUTHOR_NAME=$author_name",
 823                                 "GIT_AUTHOR_EMAIL=$author_email",
 824                                 "GIT_AUTHOR_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
 825                                 "GIT_COMMITTER_NAME=$committer_name",
 826                                 "GIT_COMMITTER_EMAIL=$committer_email",
 827                                 "GIT_COMMITTER_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
 828                                 "git-commit-tree", $tree,@par);
 829                         die "Cannot exec git-commit-tree: $!\n";
 830                 }
 831                 $pw->writer();
 832                 $pr->reader();
 833
 834                 $message =~ s/[\s\n]+\z//;
 835                 $message = "r$revision: $message" if $opt_r;
 836
 837                 print $pw "$message\n"
 838                         or die "Error writing to git-commit-tree: $!\n";
 839                 $pw->close();
 840
 841                 print "Committed change $revision:$branch ".strftime("%Y-%m-%d %H:%M:%S",gmtime($date)).")\n" if $opt_v;
 842                 chomp($cid = <$pr>);
 843                 length($cid) == 40
 844                         or die "Cannot get commit id ($cid): $!\n";
 845                 print "Commit ID $cid\n" if $opt_v;
 846                 $pr->close();
 847
 848                 waitpid($pid,0);
 849                 die "Error running git-commit-tree: $?\n" if $?;
 850         }
 851
 852         if (not defined $cid) {
 853                 $cid = $branches{"/"}{"LAST"};
 854         }
 855
 856         if(not defined $dest) {
 857                 print "... no known parent\n" if $opt_v;
 858         } elsif(not $tag) {
 859                 print "Writing to refs/heads/$dest\n" if $opt_v;
 860                 open(C,">$git_dir/refs/heads/$dest") and
 861                 print C ("$cid\n") and
 862                 close(C)
 863                         or die "Cannot write branch $dest for update: $!\n";
 864         }
 865
 866         if($tag) {
 867                 my($in, $out) = ('','');
 868                 $last_rev = "-" if %$changed_paths;
 869                 # the tag was 'complex', i.e. did not refer to a "real" revision
 870
 871                 $dest =~ tr/_/\./ if $opt_u;
 872                 $branch = $dest;
 873
 874                 my $pid = open2($in, $out, 'git-mktag');
 875                 print $out ("object $cid\n".
 876                     "type commit\n".
 877                     "tag $dest\n".
 878                     "tagger $committer_name <$committer_email> 0 +0000\n") and
 879                 close($out)
 880                     or die "Cannot create tag object $dest: $!\n";
 881
 882                 my $tagobj = <$in>;
 883                 chomp $tagobj;
 884
 885                 if ( !close($in) or waitpid($pid, 0) != $pid or
 886                                 $? != 0 or $tagobj !~ /^[0123456789abcdef]{40}$/ ) {
 887                         die "Cannot create tag object $dest: $!\n";
 888                 }
 889
 890                 open(C,">$git_dir/refs/tags/$dest") and
 891                 print C ("$tagobj\n") and
 892                 close(C)
 893                         or die "Cannot create tag $branch: $!\n";
 894
 895                 print "Created tag '$dest' on '$branch'\n" if $opt_v;
 896         }
 897         $branches{$branch}{"LAST"} = $cid;
 898         $branches{$branch}{$revision} = $cid;
 899         $last_rev = $cid;
 900         print BRANCHES "$revision $branch $cid\n";
 901         print "DONE: $revision $dest $cid\n" if $opt_v;
 902 }
 903
 904 sub commit_all {
 905         # Recursive use of the SVN connection does not work
 906         local $svn = $svn2;
 907
 908         my ($changed_paths, $revision, $author, $date, $message, $pool) = @_;
 909         my %p;
 910         while(my($path,$action) = each %$changed_paths) {
 911                 $p{$path} = [ $action->action,$action->copyfrom_path, $action->copyfrom_rev, $path ];
 912         }
 913         $changed_paths = \%p;
 914
 915         my %done;
 916         my @col;
 917         my $pref;
 918         my $branch;
 919
 920         while(my($path,$action) = each %$changed_paths) {
 921                 ($branch,$path) = split_path($revision,$path);
 922                 next if not defined $branch;
 923                 next if not defined $path;
 924                 $done{$branch}{$path} = $action;
 925         }
 926         while(($branch,$changed_paths) = each %done) {
 927                 commit($branch, $changed_paths, $revision, $author, $date, $message);
 928         }
 929 }
 930
 931 $opt_l = $svn->{'maxrev'} if not defined $opt_l or $opt_l > $svn->{'maxrev'};
 932
 933 if ($opt_l < $current_rev) {
 934     print "Up to date: no new revisions to fetch!\n" if $opt_v;
 935     unlink("$git_dir/SVN2GIT_HEAD");
 936     exit;
 937 }
 938
 939 print "Processing from $current_rev to $opt_l ...\n" if $opt_v;
 940
 941 my $from_rev;
 942 my $to_rev = $current_rev;
 943
 944 while ($to_rev < $opt_l) {
 945         $from_rev = $to_rev;
 946         $to_rev = $from_rev + $repack_after;
 947         $to_rev = $opt_l if $opt_l < $to_rev;
 948         print "Fetching from $from_rev to $to_rev ...\n" if $opt_v;
 949         my $pool=SVN::Pool->new;
 950         $svn->{'svn'}->get_log("/",$from_rev,$to_rev,0,1,1,\&commit_all,$pool);
 951         $pool->clear;
 952         my $pid = fork();
 953         die "Fork: $!\n" unless defined $pid;
 954         unless($pid) {
 955                 exec("git-repack", "-d")
 956                         or die "Cannot repack: $!\n";
 957         }
 958         waitpid($pid, 0);
 959 }
 960
 961
 962 unlink($git_index);
 963
 964 if (defined $orig_git_index) {
 965         $ENV{GIT_INDEX_FILE} = $orig_git_index;
 966 } else {
 967         delete $ENV{GIT_INDEX_FILE};
 968 }
 969
 970 # Now switch back to the branch we were in before all of this happened
 971 if($orig_branch) {
 972         print "DONE\n" if $opt_v and (not defined $opt_l or $opt_l > 0);
 973         system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
 974                 if $forward_master;
 975         unless ($opt_i) {
 976                 system('git-read-tree', '-m', '-u', 'SVN2GIT_HEAD', 'HEAD');
 977                 die "read-tree failed: $?\n" if $?;
 978         }
 979 } else {
 980         $orig_branch = "master";
 981         print "DONE; creating $orig_branch branch\n" if $opt_v and (not defined $opt_l or $opt_l > 0);
 982         system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
 983                 unless -f "$git_dir/refs/heads/master";
 984         system('git-update-ref', 'HEAD', "$orig_branch");
 985         unless ($opt_i) {
 986                 system('git checkout');
 987                 die "checkout failed: $?\n" if $?;
 988         }
 989 }
 990 unlink("$git_dir/SVN2GIT_HEAD");
 991 close(BRANCHES);