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         $path =~ s#^/*##;
 152         eval { (undef, $properties)
 153                    = $self->{'svn'}->get_file($path,$rev,$fh,$pool); };
 154         $pool->clear;
 155         if($@) {
 156                 return undef if $@ =~ /Attempted to get checksum/;
 157                 die $@;
 158         }
 159         my $mode;
 160         if (exists $properties->{'svn:executable'}) {
 161                 $mode = '100755';
 162         } elsif (exists $properties->{'svn:special'}) {
 163                 my ($special_content, $filesize);
 164                 $filesize = tell $fh;
 165                 seek $fh, 0, SEEK_SET;
 166                 read $fh, $special_content, $filesize;
 167                 if ($special_content =~ s/^link //) {
 168                         $mode = '120000';
 169                         seek $fh, 0, SEEK_SET;
 170                         truncate $fh, 0;
 171                         print $fh $special_content;
 172                 } else {
 173                         die "unexpected svn:special file encountered";
 174                 }
 175         } else {
 176                 $mode = '100644';
 177         }
 178         close ($fh);
 179
 180         return ($name, $mode);
 181 }
 182
 183 sub ignore {
 184         my($self,$path,$rev) = @_;
 185
 186         print "... $rev $path ...\n" if $opt_v;
 187         $path =~ s#^/*##;
 188         my (undef,undef,$properties)
 189             = $self->{'svn'}->get_dir($path,$rev,undef);
 190         if (exists $properties->{'svn:ignore'}) {
 191                 my ($fh, $name) = tempfile('gitsvn.XXXXXX',
 192                                            DIR => File::Spec->tmpdir(),
 193                                            UNLINK => 1);
 194                 print $fh $properties->{'svn:ignore'};
 195                 close($fh);
 196                 return $name;
 197         } else {
 198                 return undef;
 199         }
 200 }
 201
 202 sub dir_list {
 203         my($self,$path,$rev) = @_;
 204         $path =~ s#^/*##;
 205         my ($dirents,undef,$properties)
 206             = $self->{'svn'}->get_dir($path,$rev,undef);
 207         return $dirents;
 208 }
 209
 210 package main;
 211 use URI;
 212
 213 our $svn = $svn_url;
 214 $svn .= "/$svn_dir" if defined $svn_dir;
 215 my $svn2 = SVNconn->new($svn);
 216 $svn = SVNconn->new($svn);
 217
 218 my $lwp_ua;
 219 if($opt_d or $opt_D) {
 220         $svn_url = URI->new($svn_url)->canonical;
 221         if($opt_D) {
 222                 $svn_dir =~ s#/*$#/#;
 223         } else {
 224                 $svn_dir = "";
 225         }
 226         if ($svn_url->scheme eq "http") {
 227                 use LWP::UserAgent;
 228                 $lwp_ua = LWP::UserAgent->new(keep_alive => 1, requests_redirectable => []);
 229         } else {
 230                 print STDERR "Warning: not HTTP; turning off direct file access\n";
 231                 $opt_d=0;
 232         }
 233 }
 234
 235 sub pdate($) {
 236         my($d) = @_;
 237         $d =~ m#(\d\d\d\d)-(\d\d)-(\d\d)T(\d\d):(\d\d):(\d\d)#
 238                 or die "Unparseable date: $d\n";
 239         my $y=$1; $y-=1900 if $y>1900;
 240         return timegm($6||0,$5,$4,$3,$2-1,$y);
 241 }
 242
 243 sub getwd() {
 244         my $pwd = `pwd`;
 245         chomp $pwd;
 246         return $pwd;
 247 }
 248
 249
 250 sub get_headref($$) {
 251     my $name    = shift;
 252     my $git_dir = shift;
 253     my $sha;
 254
 255     if (open(C,"$git_dir/refs/heads/$name")) {
 256         chomp($sha = <C>);
 257         close(C);
 258         length($sha) == 40
 259             or die "Cannot get head id for $name ($sha): $!\n";
 260     }
 261     return $sha;
 262 }
 263
 264
 265 -d $git_tree
 266         or mkdir($git_tree,0777)
 267         or die "Could not create $git_tree: $!";
 268 chdir($git_tree);
 269
 270 my $orig_branch = "";
 271 my $forward_master = 0;
 272 my %branches;
 273
 274 my $git_dir = $ENV{"GIT_DIR"} || ".git";
 275 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
 276 $ENV{"GIT_DIR"} = $git_dir;
 277 my $orig_git_index;
 278 $orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
 279 my ($git_ih, $git_index) = tempfile('gitXXXXXX', SUFFIX => '.idx',
 280                                     DIR => File::Spec->tmpdir());
 281 close ($git_ih);
 282 $ENV{GIT_INDEX_FILE} = $git_index;
 283 my $maxnum = 0;
 284 my $last_rev = "";
 285 my $last_branch;
 286 my $current_rev = $opt_s || 1;
 287 unless(-d $git_dir) {
 288         system("git-init-db");
 289         die "Cannot init the GIT db at $git_tree: $?\n" if $?;
 290         system("git-read-tree");
 291         die "Cannot init an empty tree: $?\n" if $?;
 292
 293         $last_branch = $opt_o;
 294         $orig_branch = "";
 295 } else {
 296         -f "$git_dir/refs/heads/$opt_o"
 297                 or die "Branch '$opt_o' does not exist.\n".
 298                        "Either use the correct '-o branch' option,\n".
 299                        "or import to a new repository.\n";
 300
 301         -f "$git_dir/svn2git"
 302                 or die "'$git_dir/svn2git' does not exist.\n".
 303                        "You need that file for incremental imports.\n";
 304         open(F, "git-symbolic-ref HEAD |") or
 305                 die "Cannot run git-symbolic-ref: $!\n";
 306         chomp ($last_branch = <F>);
 307         $last_branch = basename($last_branch);
 308         close(F);
 309         unless($last_branch) {
 310                 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
 311                 $last_branch = "master";
 312         }
 313         $orig_branch = $last_branch;
 314         $last_rev = get_headref($orig_branch, $git_dir);
 315         if (-f "$git_dir/SVN2GIT_HEAD") {
 316                 die <<EOM;
 317 SVN2GIT_HEAD exists.
 318 Make sure your working directory corresponds to HEAD and remove SVN2GIT_HEAD.
 319 You may need to run
 320
 321     git-read-tree -m -u SVN2GIT_HEAD HEAD
 322 EOM
 323         }
 324         system('cp', "$git_dir/HEAD", "$git_dir/SVN2GIT_HEAD");
 325
 326         $forward_master =
 327             $opt_o ne 'master' && -f "$git_dir/refs/heads/master" &&
 328             system('cmp', '-s', "$git_dir/refs/heads/master",
 329                                 "$git_dir/refs/heads/$opt_o") == 0;
 330
 331         # populate index
 332         system('git-read-tree', $last_rev);
 333         die "read-tree failed: $?\n" if $?;
 334
 335         # Get the last import timestamps
 336         open my $B,"<", "$git_dir/svn2git";
 337         while(<$B>) {
 338                 chomp;
 339                 my($num,$branch,$ref) = split;
 340                 $branches{$branch}{$num} = $ref;
 341                 $branches{$branch}{"LAST"} = $ref;
 342                 $current_rev = $num+1 if $current_rev <= $num;
 343         }
 344         close($B);
 345 }
 346 -d $git_dir
 347         or die "Could not create git subdir ($git_dir).\n";
 348
 349 my $default_authors = "$git_dir/svn-authors";
 350 if ($opt_A) {
 351         read_users($opt_A);
 352         copy($opt_A,$default_authors) or die "Copy failed: $!";
 353 } else {
 354         read_users($default_authors) if -f $default_authors;
 355 }
 356
 357 open BRANCHES,">>", "$git_dir/svn2git";
 358
 359 sub node_kind($$) {
 360         my ($svnpath, $revision) = @_;
 361         my $pool=SVN::Pool->new;
 362         $svnpath =~ s#^/*##;
 363         my $kind = $svn->{'svn'}->check_path($svnpath,$revision,$pool);
 364         $pool->clear;
 365         return $kind;
 366 }
 367
 368 sub get_file($$$) {
 369         my($svnpath,$rev,$path) = @_;
 370
 371         # now get it
 372         my ($name,$mode);
 373         if($opt_d) {
 374                 my($req,$res);
 375
 376                 # /svn/!svn/bc/2/django/trunk/django-docs/build.py
 377                 my $url=$svn_url->clone();
 378                 $url->path($url->path."/!svn/bc/$rev/$svn_dir$svnpath");
 379                 print "... $path...\n" if $opt_v;
 380                 $req = HTTP::Request->new(GET => $url);
 381                 $res = $lwp_ua->request($req);
 382                 if ($res->is_success) {
 383                         my $fh;
 384                         ($fh, $name) = tempfile('gitsvn.XXXXXX',
 385                         DIR => File::Spec->tmpdir(), UNLINK => 1);
 386                         print $fh $res->content;
 387                         close($fh) or die "Could not write $name: $!\n";
 388                 } else {
 389                         return undef if $res->code == 301; # directory?
 390                         die $res->status_line." at $url\n";
 391                 }
 392                 $mode = '0644'; # can't obtain mode via direct http request?
 393         } else {
 394                 ($name,$mode) = $svn->file("$svnpath",$rev);
 395                 return undef unless defined $name;
 396         }
 397
 398         my $pid = open(my $F, '-|');
 399         die $! unless defined $pid;
 400         if (!$pid) {
 401             exec("git-hash-object", "-w", $name)
 402                 or die "Cannot create object: $!\n";
 403         }
 404         my $sha = <$F>;
 405         chomp $sha;
 406         close $F;
 407         unlink $name;
 408         return [$mode, $sha, $path];
 409 }
 410
 411 sub get_ignore($$$$$) {
 412         my($new,$old,$rev,$path,$svnpath) = @_;
 413
 414         return unless $opt_I;
 415         my $name = $svn->ignore("$svnpath",$rev);
 416         if ($path eq '/') {
 417                 $path = $opt_I;
 418         } else {
 419                 $path = File::Spec->catfile($path,$opt_I);
 420         }
 421         if (defined $name) {
 422                 my $pid = open(my $F, '-|');
 423                 die $! unless defined $pid;
 424                 if (!$pid) {
 425                         exec("git-hash-object", "-w", $name)
 426                             or die "Cannot create object: $!\n";
 427                 }
 428                 my $sha = <$F>;
 429                 chomp $sha;
 430                 close $F;
 431                 unlink $name;
 432                 push(@$new,['0644',$sha,$path]);
 433         } elsif (defined $old) {
 434                 push(@$old,$path);
 435         }
 436 }
 437
 438 sub project_path($$)
 439 {
 440         my ($path, $project) = @_;
 441
 442         $path = "/".$path unless ($path =~ m#^\/#) ;
 443         return $1 if ($path =~ m#^$project\/(.*)$#);
 444
 445         $path =~ s#\.#\\\.#g;
 446         $path =~ s#\+#\\\+#g;
 447         return "/" if ($project =~ m#^$path.*$#);
 448
 449         return undef;
 450 }
 451
 452 sub split_path($$) {
 453         my($rev,$path) = @_;
 454         my $branch;
 455
 456         if($path =~ s#^/\Q$tag_name\E/([^/]+)/?##) {
 457                 $branch = "/$1";
 458         } elsif($path =~ s#^/\Q$trunk_name\E/?##) {
 459                 $branch = "/";
 460         } elsif($path =~ s#^/\Q$branch_name\E/([^/]+)/?##) {
 461                 $branch = $1;
 462         } else {
 463                 my %no_error = (
 464                         "/" => 1,
 465                         "/$tag_name" => 1,
 466                         "/$branch_name" => 1
 467                 );
 468                 print STDERR "$rev: Unrecognized path: $path\n" unless (defined $no_error{$path});
 469                 return ()
 470         }
 471         if ($path eq "") {
 472                 $path = "/";
 473         } elsif ($project_name) {
 474                 $path = project_path($path, $project_name);
 475         }
 476         return ($branch,$path);
 477 }
 478
 479 sub branch_rev($$) {
 480
 481         my ($srcbranch,$uptorev) = @_;
 482
 483         my $bbranches = $branches{$srcbranch};
 484         my @revs = reverse sort { ($a eq 'LAST' ? 0 : $a) <=> ($b eq 'LAST' ? 0 : $b) } keys %$bbranches;
 485         my $therev;
 486         foreach my $arev(@revs) {
 487                 next if  ($arev eq 'LAST');
 488                 if ($arev <= $uptorev) {
 489                         $therev = $arev;
 490                         last;
 491                 }
 492         }
 493         return $therev;
 494 }
 495
 496 sub expand_svndir($$$);
 497
 498 sub expand_svndir($$$)
 499 {
 500         my ($svnpath, $rev, $path) = @_;
 501         my @list;
 502         get_ignore(\@list, undef, $rev, $path, $svnpath);
 503         my $dirents = $svn->dir_list($svnpath, $rev);
 504         foreach my $p(keys %$dirents) {
 505                 my $kind = node_kind($svnpath.'/'.$p, $rev);
 506                 if ($kind eq $SVN::Node::file) {
 507                         my $f = get_file($svnpath.'/'.$p, $rev, $path.'/'.$p);
 508                         push(@list, $f) if $f;
 509                 } elsif ($kind eq $SVN::Node::dir) {
 510                         push(@list,
 511                              expand_svndir($svnpath.'/'.$p, $rev, $path.'/'.$p));
 512                 }
 513         }
 514         return @list;
 515 }
 516
 517 sub copy_path($$$$$$$$) {
 518         # Somebody copied a whole subdirectory.
 519         # We need to find the index entries from the old version which the
 520         # SVN log entry points to, and add them to the new place.
 521
 522         my($newrev,$newbranch,$path,$oldpath,$rev,$node_kind,$new,$parents) = @_;
 523
 524         my($srcbranch,$srcpath) = split_path($rev,$oldpath);
 525         unless(defined $srcbranch && defined $srcpath) {
 526                 print "Path not found when copying from $oldpath @ $rev.\n".
 527                         "Will try to copy from original SVN location...\n"
 528                         if $opt_v;
 529                 push (@$new, expand_svndir($oldpath, $rev, $path));
 530                 return;
 531         }
 532         my $therev = branch_rev($srcbranch, $rev);
 533         my $gitrev = $branches{$srcbranch}{$therev};
 534         unless($gitrev) {
 535                 print STDERR "$newrev:$newbranch: could not find $oldpath \@ $rev\n";
 536                 return;
 537         }
 538         if ($srcbranch ne $newbranch) {
 539                 push(@$parents, $branches{$srcbranch}{'LAST'});
 540         }
 541         print "$newrev:$newbranch:$path: copying from $srcbranch:$srcpath @ $rev\n" if $opt_v;
 542         if ($node_kind eq $SVN::Node::dir) {
 543                 $srcpath =~ s#/*$#/#;
 544         }
 545
 546         my $pid = open my $f,'-|';
 547         die $! unless defined $pid;
 548         if (!$pid) {
 549                 exec("git-ls-tree","-r","-z",$gitrev,$srcpath)
 550                         or die $!;
 551         }
 552         local $/ = "\0";
 553         while(<$f>) {
 554                 chomp;
 555                 my($m,$p) = split(/\t/,$_,2);
 556                 my($mode,$type,$sha1) = split(/ /,$m);
 557                 next if $type ne "blob";
 558                 if ($node_kind eq $SVN::Node::dir) {
 559                         $p = $path . substr($p,length($srcpath)-1);
 560                 } else {
 561                         $p = $path;
 562                 }
 563                 push(@$new,[$mode,$sha1,$p]);
 564         }
 565         close($f) or
 566                 print STDERR "$newrev:$newbranch: could not list files in $oldpath \@ $rev\n";
 567 }
 568
 569 sub commit {
 570         my($branch, $changed_paths, $revision, $author, $date, $message) = @_;
 571         my($committer_name,$committer_email,$dest);
 572         my($author_name,$author_email);
 573         my(@old,@new,@parents);
 574
 575         if (not defined $author or $author eq "") {
 576                 $committer_name = $committer_email = "unknown";
 577         } elsif (defined $users_file) {
 578                 die "User $author is not listed in $users_file\n"
 579                     unless exists $users{$author};
 580                 ($committer_name,$committer_email) = @{$users{$author}};
 581         } elsif ($author =~ /^(.*?)\s+<(.*)>$/) {
 582                 ($committer_name, $committer_email) = ($1, $2);
 583         } else {
 584                 $author =~ s/^<(.*)>$/$1/;
 585                 $committer_name = $committer_email = $author;
 586         }
 587
 588         if ($opt_F && $message =~ /From:\s+(.*?)\s+<(.*)>\s*\n/) {
 589                 ($author_name, $author_email) = ($1, $2);
 590                 print "Author from From: $1 <$2>\n" if ($opt_v);;
 591         } elsif ($opt_S && $message =~ /Signed-off-by:\s+(.*?)\s+<(.*)>\s*\n/) {
 592                 ($author_name, $author_email) = ($1, $2);
 593                 print "Author from Signed-off-by: $1 <$2>\n" if ($opt_v);;
 594         } else {
 595                 $author_name = $committer_name;
 596                 $author_email = $committer_email;
 597         }
 598
 599         $date = pdate($date);
 600
 601         my $tag;
 602         my $parent;
 603         if($branch eq "/") { # trunk
 604                 $parent = $opt_o;
 605         } elsif($branch =~ m#^/(.+)#) { # tag
 606                 $tag = 1;
 607                 $parent = $1;
 608         } else { # "normal" branch
 609                 # nothing to do
 610                 $parent = $branch;
 611         }
 612         $dest = $parent;
 613
 614         my $prev = $changed_paths->{"/"};
 615         if($prev and $prev->[0] eq "A") {
 616                 delete $changed_paths->{"/"};
 617                 my $oldpath = $prev->[1];
 618                 my $rev;
 619                 if(defined $oldpath) {
 620                         my $p;
 621                         ($parent,$p) = split_path($revision,$oldpath);
 622                         if(defined $parent) {
 623                                 if($parent eq "/") {
 624                                         $parent = $opt_o;
 625                                 } else {
 626                                         $parent =~ s#^/##; # if it's a tag
 627                                 }
 628                         }
 629                 } else {
 630                         $parent = undef;
 631                 }
 632         }
 633
 634         my $rev;
 635         if($revision > $opt_s and defined $parent) {
 636                 open(H,"git-rev-parse --verify $parent |");
 637                 $rev = <H>;
 638                 close(H) or do {
 639                         print STDERR "$revision: cannot find commit '$parent'!\n";
 640                         return;
 641                 };
 642                 chop $rev;
 643                 if(length($rev) != 40) {
 644                         print STDERR "$revision: cannot find commit '$parent'!\n";
 645                         return;
 646                 }
 647                 $rev = $branches{($parent eq $opt_o) ? "/" : $parent}{"LAST"};
 648                 if($revision != $opt_s and not $rev) {
 649                         print STDERR "$revision: do not know ancestor for '$parent'!\n";
 650                         return;
 651                 }
 652         } else {
 653                 $rev = undef;
 654         }
 655
 656 #       if($prev and $prev->[0] eq "A") {
 657 #               if(not $tag) {
 658 #                       unless(open(H,"> $git_dir/refs/heads/$branch")) {
 659 #                               print STDERR "$revision: Could not create branch $branch: $!\n";
 660 #                               $state=11;
 661 #                               next;
 662 #                       }
 663 #                       print H "$rev\n"
 664 #                               or die "Could not write branch $branch: $!";
 665 #                       close(H)
 666 #                               or die "Could not write branch $branch: $!";
 667 #               }
 668 #       }
 669         if(not defined $rev) {
 670                 unlink($git_index);
 671         } elsif ($rev ne $last_rev) {
 672                 print "Switching from $last_rev to $rev ($branch)\n" if $opt_v;
 673                 system("git-read-tree", $rev);
 674                 die "read-tree failed for $rev: $?\n" if $?;
 675                 $last_rev = $rev;
 676         }
 677
 678         push (@parents, $rev) if defined $rev;
 679
 680         my $cid;
 681         if($tag and not %$changed_paths) {
 682                 $cid = $rev;
 683         } else {
 684                 my @paths = sort keys %$changed_paths;
 685                 foreach my $path(@paths) {
 686                         my $action = $changed_paths->{$path};
 687
 688                         if ($action->[0] eq "R") {
 689                                 # refer to a file/tree in an earlier commit
 690                                 push(@old,$path); # remove any old stuff
 691                         }
 692                         if(($action->[0] eq "A") || ($action->[0] eq "R")) {
 693                                 my $node_kind = node_kind($action->[3], $revision);
 694                                 if ($node_kind eq $SVN::Node::file) {
 695                                         my $f = get_file($action->[3],
 696                                                          $revision, $path);
 697                                         if ($f) {
 698                                                 push(@new,$f) if $f;
 699                                         } else {
 700                                                 my $opath = $action->[3];
 701                                                 print STDERR "$revision: $branch: could not fetch '$opath'\n";
 702                                         }
 703                                 } elsif ($node_kind eq $SVN::Node::dir) {
 704                                         if($action->[1]) {
 705                                                 copy_path($revision, $branch,
 706                                                           $path, $action->[1],
 707                                                           $action->[2], $node_kind,
 708                                                           \@new, \@parents);
 709                                         } else {
 710                                                 get_ignore(\@new, \@old, $revision,
 711                                                            $path, $action->[3]);
 712                                         }
 713                                 }
 714                         } elsif ($action->[0] eq "D") {
 715                                 push(@old,$path);
 716                         } elsif ($action->[0] eq "M") {
 717                                 my $node_kind = node_kind($action->[3], $revision);
 718                                 if ($node_kind eq $SVN::Node::file) {
 719                                         my $f = get_file($action->[3],
 720                                                          $revision, $path);
 721                                         push(@new,$f) if $f;
 722                                 } elsif ($node_kind eq $SVN::Node::dir) {
 723                                         get_ignore(\@new, \@old, $revision,
 724                                                    $path, $action->[3]);
 725                                 }
 726                         } else {
 727                                 die "$revision: unknown action '".$action->[0]."' for $path\n";
 728                         }
 729                 }
 730
 731                 while(@old) {
 732                         my @o1;
 733                         if(@old > 55) {
 734                                 @o1 = splice(@old,0,50);
 735                         } else {
 736                                 @o1 = @old;
 737                                 @old = ();
 738                         }
 739                         my $pid = open my $F, "-|";
 740                         die "$!" unless defined $pid;
 741                         if (!$pid) {
 742                                 exec("git-ls-files", "-z", @o1) or die $!;
 743                         }
 744                         @o1 = ();
 745                         local $/ = "\0";
 746                         while(<$F>) {
 747                                 chomp;
 748                                 push(@o1,$_);
 749                         }
 750                         close($F);
 751
 752                         while(@o1) {
 753                                 my @o2;
 754                                 if(@o1 > 55) {
 755                                         @o2 = splice(@o1,0,50);
 756                                 } else {
 757                                         @o2 = @o1;
 758                                         @o1 = ();
 759                                 }
 760                                 system("git-update-index","--force-remove","--",@o2);
 761                                 die "Cannot remove files: $?\n" if $?;
 762                         }
 763                 }
 764                 while(@new) {
 765                         my @n2;
 766                         if(@new > 12) {
 767                                 @n2 = splice(@new,0,10);
 768                         } else {
 769                                 @n2 = @new;
 770                                 @new = ();
 771                         }
 772                         system("git-update-index","--add",
 773                                 (map { ('--cacheinfo', @$_) } @n2));
 774                         die "Cannot add files: $?\n" if $?;
 775                 }
 776
 777                 my $pid = open(C,"-|");
 778                 die "Cannot fork: $!" unless defined $pid;
 779                 unless($pid) {
 780                         exec("git-write-tree");
 781                         die "Cannot exec git-write-tree: $!\n";
 782                 }
 783                 chomp(my $tree = <C>);
 784                 length($tree) == 40
 785                         or die "Cannot get tree id ($tree): $!\n";
 786                 close(C)
 787                         or die "Error running git-write-tree: $?\n";
 788                 print "Tree ID $tree\n" if $opt_v;
 789
 790                 my $pr = IO::Pipe->new() or die "Cannot open pipe: $!\n";
 791                 my $pw = IO::Pipe->new() or die "Cannot open pipe: $!\n";
 792                 $pid = fork();
 793                 die "Fork: $!\n" unless defined $pid;
 794                 unless($pid) {
 795                         $pr->writer();
 796                         $pw->reader();
 797                         open(OUT,">&STDOUT");
 798                         dup2($pw->fileno(),0);
 799                         dup2($pr->fileno(),1);
 800                         $pr->close();
 801                         $pw->close();
 802
 803                         my @par = ();
 804
 805                         # loose detection of merges
 806                         # based on the commit msg
 807                         foreach my $rx (@mergerx) {
 808                                 if ($message =~ $rx) {
 809                                         my $mparent = $1;
 810                                         if ($mparent eq 'HEAD') { $mparent = $opt_o };
 811                                         if ( -e "$git_dir/refs/heads/$mparent") {
 812                                                 $mparent = get_headref($mparent, $git_dir);
 813                                                 push (@parents, $mparent);
 814                                                 print OUT "Merge parent branch: $mparent\n" if $opt_v;
 815                                         }
 816                                 }
 817                         }
 818                         my %seen_parents = ();
 819                         my @unique_parents = grep { ! $seen_parents{$_} ++ } @parents;
 820                         foreach my $bparent (@unique_parents) {
 821                                 push @par, '-p', $bparent;
 822                                 print OUT "Merge parent branch: $bparent\n" if $opt_v;
 823                         }
 824
 825                         exec("env",
 826                                 "GIT_AUTHOR_NAME=$author_name",
 827                                 "GIT_AUTHOR_EMAIL=$author_email",
 828                                 "GIT_AUTHOR_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
 829                                 "GIT_COMMITTER_NAME=$committer_name",
 830                                 "GIT_COMMITTER_EMAIL=$committer_email",
 831                                 "GIT_COMMITTER_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
 832                                 "git-commit-tree", $tree,@par);
 833                         die "Cannot exec git-commit-tree: $!\n";
 834                 }
 835                 $pw->writer();
 836                 $pr->reader();
 837
 838                 $message =~ s/[\s\n]+\z//;
 839                 $message = "r$revision: $message" if $opt_r;
 840
 841                 print $pw "$message\n"
 842                         or die "Error writing to git-commit-tree: $!\n";
 843                 $pw->close();
 844
 845                 print "Committed change $revision:$branch ".strftime("%Y-%m-%d %H:%M:%S",gmtime($date)).")\n" if $opt_v;
 846                 chomp($cid = <$pr>);
 847                 length($cid) == 40
 848                         or die "Cannot get commit id ($cid): $!\n";
 849                 print "Commit ID $cid\n" if $opt_v;
 850                 $pr->close();
 851
 852                 waitpid($pid,0);
 853                 die "Error running git-commit-tree: $?\n" if $?;
 854         }
 855
 856         if (not defined $cid) {
 857                 $cid = $branches{"/"}{"LAST"};
 858         }
 859
 860         if(not defined $dest) {
 861                 print "... no known parent\n" if $opt_v;
 862         } elsif(not $tag) {
 863                 print "Writing to refs/heads/$dest\n" if $opt_v;
 864                 open(C,">$git_dir/refs/heads/$dest") and
 865                 print C ("$cid\n") and
 866                 close(C)
 867                         or die "Cannot write branch $dest for update: $!\n";
 868         }
 869
 870         if($tag) {
 871                 my($in, $out) = ('','');
 872                 $last_rev = "-" if %$changed_paths;
 873                 # the tag was 'complex', i.e. did not refer to a "real" revision
 874
 875                 $dest =~ tr/_/\./ if $opt_u;
 876                 $branch = $dest;
 877
 878                 my $pid = open2($in, $out, 'git-mktag');
 879                 print $out ("object $cid\n".
 880                     "type commit\n".
 881                     "tag $dest\n".
 882                     "tagger $committer_name <$committer_email> 0 +0000\n") and
 883                 close($out)
 884                     or die "Cannot create tag object $dest: $!\n";
 885
 886                 my $tagobj = <$in>;
 887                 chomp $tagobj;
 888
 889                 if ( !close($in) or waitpid($pid, 0) != $pid or
 890                                 $? != 0 or $tagobj !~ /^[0123456789abcdef]{40}$/ ) {
 891                         die "Cannot create tag object $dest: $!\n";
 892                 }
 893
 894                 open(C,">$git_dir/refs/tags/$dest") and
 895                 print C ("$tagobj\n") and
 896                 close(C)
 897                         or die "Cannot create tag $branch: $!\n";
 898
 899                 print "Created tag '$dest' on '$branch'\n" if $opt_v;
 900         }
 901         $branches{$branch}{"LAST"} = $cid;
 902         $branches{$branch}{$revision} = $cid;
 903         $last_rev = $cid;
 904         print BRANCHES "$revision $branch $cid\n";
 905         print "DONE: $revision $dest $cid\n" if $opt_v;
 906 }
 907
 908 sub commit_all {
 909         # Recursive use of the SVN connection does not work
 910         local $svn = $svn2;
 911
 912         my ($changed_paths, $revision, $author, $date, $message, $pool) = @_;
 913         my %p;
 914         while(my($path,$action) = each %$changed_paths) {
 915                 $p{$path} = [ $action->action,$action->copyfrom_path, $action->copyfrom_rev, $path ];
 916         }
 917         $changed_paths = \%p;
 918
 919         my %done;
 920         my @col;
 921         my $pref;
 922         my $branch;
 923
 924         while(my($path,$action) = each %$changed_paths) {
 925                 ($branch,$path) = split_path($revision,$path);
 926                 next if not defined $branch;
 927                 next if not defined $path;
 928                 $done{$branch}{$path} = $action;
 929         }
 930         while(($branch,$changed_paths) = each %done) {
 931                 commit($branch, $changed_paths, $revision, $author, $date, $message);
 932         }
 933 }
 934
 935 $opt_l = $svn->{'maxrev'} if not defined $opt_l or $opt_l > $svn->{'maxrev'};
 936
 937 if ($opt_l < $current_rev) {
 938     print "Up to date: no new revisions to fetch!\n" if $opt_v;
 939     unlink("$git_dir/SVN2GIT_HEAD");
 940     exit;
 941 }
 942
 943 print "Processing from $current_rev to $opt_l ...\n" if $opt_v;
 944
 945 my $from_rev;
 946 my $to_rev = $current_rev;
 947
 948 while ($to_rev < $opt_l) {
 949         $from_rev = $to_rev;
 950         $to_rev = $from_rev + $repack_after;
 951         $to_rev = $opt_l if $opt_l < $to_rev;
 952         print "Fetching from $from_rev to $to_rev ...\n" if $opt_v;
 953         my $pool=SVN::Pool->new;
 954         $svn->{'svn'}->get_log("/",$from_rev,$to_rev,0,1,1,\&commit_all,$pool);
 955         $pool->clear;
 956         my $pid = fork();
 957         die "Fork: $!\n" unless defined $pid;
 958         unless($pid) {
 959                 exec("git-repack", "-d")
 960                         or die "Cannot repack: $!\n";
 961         }
 962         waitpid($pid, 0);
 963 }
 964
 965
 966 unlink($git_index);
 967
 968 if (defined $orig_git_index) {
 969         $ENV{GIT_INDEX_FILE} = $orig_git_index;
 970 } else {
 971         delete $ENV{GIT_INDEX_FILE};
 972 }
 973
 974 # Now switch back to the branch we were in before all of this happened
 975 if($orig_branch) {
 976         print "DONE\n" if $opt_v and (not defined $opt_l or $opt_l > 0);
 977         system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
 978                 if $forward_master;
 979         unless ($opt_i) {
 980                 system('git-read-tree', '-m', '-u', 'SVN2GIT_HEAD', 'HEAD');
 981                 die "read-tree failed: $?\n" if $?;
 982         }
 983 } else {
 984         $orig_branch = "master";
 985         print "DONE; creating $orig_branch branch\n" if $opt_v and (not defined $opt_l or $opt_l > 0);
 986         system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
 987                 unless -f "$git_dir/refs/heads/master";
 988         system('git-update-ref', 'HEAD', "$orig_branch");
 989         unless ($opt_i) {
 990                 system('git checkout');
 991                 die "checkout failed: $?\n" if $?;
 992         }
 993 }
 994 unlink("$git_dir/SVN2GIT_HEAD");
 995 close(BRANCHES);