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); $|=1; select(STDOUT);
  92
  93
  94 package SVNconn;
  95 # Basic SVN connection.
  96 # We're only interested in connecting and downloading, so ...
  97
  98 use File::Spec;
  99 use File::Temp qw(tempfile);
 100 use POSIX qw(strftime dup2);
 101
 102 sub new {
 103         my($what,$repo) = @_;
 104         $what=ref($what) if ref($what);
 105
 106         my $self = {};
 107         $self->{'buffer'} = "";
 108         bless($self,$what);
 109
 110         $repo =~ s#/+$##;
 111         $self->{'fullrep'} = $repo;
 112         $self->conn();
 113
 114         return $self;
 115 }
 116
 117 sub conn {
 118         my $self = shift;
 119         my $repo = $self->{'fullrep'};
 120         my $auth = SVN::Core::auth_open ([SVN::Client::get_simple_provider,
 121                           SVN::Client::get_ssl_server_trust_file_provider,
 122                           SVN::Client::get_username_provider]);
 123         my $s = SVN::Ra->new(url => $repo, auth => $auth);
 124         die "SVN connection to $repo: $!\n" unless defined $s;
 125         $self->{'svn'} = $s;
 126         $self->{'repo'} = $repo;
 127         $self->{'maxrev'} = $s->get_latest_revnum();
 128 }
 129
 130 sub file {
 131         my($self,$path,$rev) = @_;
 132
 133         my ($fh, $name) = tempfile('gitsvn.XXXXXX',
 134                     DIR => File::Spec->tmpdir(), UNLINK => 1);
 135
 136         print "... $rev $path ...\n" if $opt_v;
 137         my (undef, $properties);
 138         eval { (undef, $properties)
 139                    = $self->{'svn'}->get_file($path,$rev,$fh); };
 140         if($@) {
 141                 return undef if $@ =~ /Attempted to get checksum/;
 142                 die $@;
 143         }
 144         my $mode;
 145         if (exists $properties->{'svn:executable'}) {
 146                 $mode = '0755';
 147         } else {
 148                 $mode = '0644';
 149         }
 150         close ($fh);
 151
 152         return ($name, $mode);
 153 }
 154
 155 sub ignore {
 156         my($self,$path,$rev) = @_;
 157
 158         print "... $rev $path ...\n" if $opt_v;
 159         my (undef,undef,$properties)
 160             = $self->{'svn'}->get_dir($path,$rev,undef);
 161         if (exists $properties->{'svn:ignore'}) {
 162                 my ($fh, $name) = tempfile('gitsvn.XXXXXX',
 163                                            DIR => File::Spec->tmpdir(),
 164                                            UNLINK => 1);
 165                 print $fh $properties->{'svn:ignore'};
 166                 close($fh);
 167                 return $name;
 168         } else {
 169                 return undef;
 170         }
 171 }
 172
 173 package main;
 174 use URI;
 175
 176 our $svn = $svn_url;
 177 $svn .= "/$svn_dir" if defined $svn_dir;
 178 my $svn2 = SVNconn->new($svn);
 179 $svn = SVNconn->new($svn);
 180
 181 my $lwp_ua;
 182 if($opt_d or $opt_D) {
 183         $svn_url = URI->new($svn_url)->canonical;
 184         if($opt_D) {
 185                 $svn_dir =~ s#/*$#/#;
 186         } else {
 187                 $svn_dir = "";
 188         }
 189         if ($svn_url->scheme eq "http") {
 190                 use LWP::UserAgent;
 191                 $lwp_ua = LWP::UserAgent->new(keep_alive => 1, requests_redirectable => []);
 192         } else {
 193                 print STDERR "Warning: not HTTP; turning off direct file access\n";
 194                 $opt_d=0;
 195         }
 196 }
 197
 198 sub pdate($) {
 199         my($d) = @_;
 200         $d =~ m#(\d\d\d\d)-(\d\d)-(\d\d)T(\d\d):(\d\d):(\d\d)#
 201                 or die "Unparseable date: $d\n";
 202         my $y=$1; $y-=1900 if $y>1900;
 203         return timegm($6||0,$5,$4,$3,$2-1,$y);
 204 }
 205
 206 sub getwd() {
 207         my $pwd = `pwd`;
 208         chomp $pwd;
 209         return $pwd;
 210 }
 211
 212
 213 sub get_headref($$) {
 214     my $name    = shift;
 215     my $git_dir = shift;
 216     my $sha;
 217
 218     if (open(C,"$git_dir/refs/heads/$name")) {
 219         chomp($sha = <C>);
 220         close(C);
 221         length($sha) == 40
 222             or die "Cannot get head id for $name ($sha): $!\n";
 223     }
 224     return $sha;
 225 }
 226
 227
 228 -d $git_tree
 229         or mkdir($git_tree,0777)
 230         or die "Could not create $git_tree: $!";
 231 chdir($git_tree);
 232
 233 my $orig_branch = "";
 234 my $forward_master = 0;
 235 my %branches;
 236
 237 my $git_dir = $ENV{"GIT_DIR"} || ".git";
 238 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
 239 $ENV{"GIT_DIR"} = $git_dir;
 240 my $orig_git_index;
 241 $orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
 242 my ($git_ih, $git_index) = tempfile('gitXXXXXX', SUFFIX => '.idx',
 243                                     DIR => File::Spec->tmpdir());
 244 close ($git_ih);
 245 $ENV{GIT_INDEX_FILE} = $git_index;
 246 my $maxnum = 0;
 247 my $last_rev = "";
 248 my $last_branch;
 249 my $current_rev = $opt_s || 1;
 250 unless(-d $git_dir) {
 251         system("git-init-db");
 252         die "Cannot init the GIT db at $git_tree: $?\n" if $?;
 253         system("git-read-tree");
 254         die "Cannot init an empty tree: $?\n" if $?;
 255
 256         $last_branch = $opt_o;
 257         $orig_branch = "";
 258 } else {
 259         -f "$git_dir/refs/heads/$opt_o"
 260                 or die "Branch '$opt_o' does not exist.\n".
 261                        "Either use the correct '-o branch' option,\n".
 262                        "or import to a new repository.\n";
 263
 264         -f "$git_dir/svn2git"
 265                 or die "'$git_dir/svn2git' does not exist.\n".
 266                        "You need that file for incremental imports.\n";
 267         open(F, "git-symbolic-ref HEAD |") or
 268                 die "Cannot run git-symbolic-ref: $!\n";
 269         chomp ($last_branch = <F>);
 270         $last_branch = basename($last_branch);
 271         close(F);
 272         unless($last_branch) {
 273                 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
 274                 $last_branch = "master";
 275         }
 276         $orig_branch = $last_branch;
 277         $last_rev = get_headref($orig_branch, $git_dir);
 278         if (-f "$git_dir/SVN2GIT_HEAD") {
 279                 die <<EOM;
 280 SVN2GIT_HEAD exists.
 281 Make sure your working directory corresponds to HEAD and remove SVN2GIT_HEAD.
 282 You may need to run
 283
 284     git-read-tree -m -u SVN2GIT_HEAD HEAD
 285 EOM
 286         }
 287         system('cp', "$git_dir/HEAD", "$git_dir/SVN2GIT_HEAD");
 288
 289         $forward_master =
 290             $opt_o ne 'master' && -f "$git_dir/refs/heads/master" &&
 291             system('cmp', '-s', "$git_dir/refs/heads/master",
 292                                 "$git_dir/refs/heads/$opt_o") == 0;
 293
 294         # populate index
 295         system('git-read-tree', $last_rev);
 296         die "read-tree failed: $?\n" if $?;
 297
 298         # Get the last import timestamps
 299         open my $B,"<", "$git_dir/svn2git";
 300         while(<$B>) {
 301                 chomp;
 302                 my($num,$branch,$ref) = split;
 303                 $branches{$branch}{$num} = $ref;
 304                 $branches{$branch}{"LAST"} = $ref;
 305                 $current_rev = $num+1 if $current_rev <= $num;
 306         }
 307         close($B);
 308 }
 309 -d $git_dir
 310         or die "Could not create git subdir ($git_dir).\n";
 311
 312 my $default_authors = "$git_dir/svn-authors";
 313 if ($opt_A) {
 314         read_users($opt_A);
 315         copy($opt_A,$default_authors) or die "Copy failed: $!";
 316 } else {
 317         read_users($default_authors) if -f $default_authors;
 318 }
 319
 320 open BRANCHES,">>", "$git_dir/svn2git";
 321
 322 sub node_kind($$$) {
 323         my ($branch, $path, $revision) = @_;
 324         my $pool=SVN::Pool->new;
 325         my $kind = $svn->{'svn'}->check_path(revert_split_path($branch,$path),$revision,$pool);
 326         $pool->clear;
 327         return $kind;
 328 }
 329
 330 sub revert_split_path($$) {
 331         my($branch,$path) = @_;
 332
 333         my $svnpath;
 334         $path = "" if $path eq "/"; # this should not happen, but ...
 335         if($branch eq "/") {
 336                 $svnpath = "$trunk_name/$path";
 337         } elsif($branch =~ m#^/#) {
 338                 $svnpath = "$tag_name$branch/$path";
 339         } else {
 340                 $svnpath = "$branch_name/$branch/$path";
 341         }
 342
 343         $svnpath =~ s#/+$##;
 344         return $svnpath;
 345 }
 346
 347 sub get_file($$$) {
 348         my($rev,$branch,$path) = @_;
 349
 350         my $svnpath = revert_split_path($branch,$path);
 351
 352         # now get it
 353         my ($name,$mode);
 354         if($opt_d) {
 355                 my($req,$res);
 356
 357                 # /svn/!svn/bc/2/django/trunk/django-docs/build.py
 358                 my $url=$svn_url->clone();
 359                 $url->path($url->path."/!svn/bc/$rev/$svn_dir$svnpath");
 360                 print "... $path...\n" if $opt_v;
 361                 $req = HTTP::Request->new(GET => $url);
 362                 $res = $lwp_ua->request($req);
 363                 if ($res->is_success) {
 364                         my $fh;
 365                         ($fh, $name) = tempfile('gitsvn.XXXXXX',
 366                         DIR => File::Spec->tmpdir(), UNLINK => 1);
 367                         print $fh $res->content;
 368                         close($fh) or die "Could not write $name: $!\n";
 369                 } else {
 370                         return undef if $res->code == 301; # directory?
 371                         die $res->status_line." at $url\n";
 372                 }
 373                 $mode = '0644'; # can't obtain mode via direct http request?
 374         } else {
 375                 ($name,$mode) = $svn->file("$svnpath",$rev);
 376                 return undef unless defined $name;
 377         }
 378
 379         my $pid = open(my $F, '-|');
 380         die $! unless defined $pid;
 381         if (!$pid) {
 382             exec("git-hash-object", "-w", $name)
 383                 or die "Cannot create object: $!\n";
 384         }
 385         my $sha = <$F>;
 386         chomp $sha;
 387         close $F;
 388         unlink $name;
 389         return [$mode, $sha, $path];
 390 }
 391
 392 sub get_ignore($$$$$) {
 393         my($new,$old,$rev,$branch,$path) = @_;
 394
 395         return unless $opt_I;
 396         my $svnpath = revert_split_path($branch,$path);
 397         my $name = $svn->ignore("$svnpath",$rev);
 398         if ($path eq '/') {
 399                 $path = $opt_I;
 400         } else {
 401                 $path = File::Spec->catfile($path,$opt_I);
 402         }
 403         if (defined $name) {
 404                 my $pid = open(my $F, '-|');
 405                 die $! unless defined $pid;
 406                 if (!$pid) {
 407                         exec("git-hash-object", "-w", $name)
 408                             or die "Cannot create object: $!\n";
 409                 }
 410                 my $sha = <$F>;
 411                 chomp $sha;
 412                 close $F;
 413                 unlink $name;
 414                 push(@$new,['0644',$sha,$path]);
 415         } else {
 416                 push(@$old,$path);
 417         }
 418 }
 419
 420 sub split_path($$) {
 421         my($rev,$path) = @_;
 422         my $branch;
 423
 424         if($path =~ s#^/\Q$tag_name\E/([^/]+)/?##) {
 425                 $branch = "/$1";
 426         } elsif($path =~ s#^/\Q$trunk_name\E/?##) {
 427                 $branch = "/";
 428         } elsif($path =~ s#^/\Q$branch_name\E/([^/]+)/?##) {
 429                 $branch = $1;
 430         } else {
 431                 my %no_error = (
 432                         "/" => 1,
 433                         "/$tag_name" => 1,
 434                         "/$branch_name" => 1
 435                 );
 436                 print STDERR "$rev: Unrecognized path: $path\n" unless (defined $no_error{$path});
 437                 return ()
 438         }
 439         $path = "/" if $path eq "";
 440         return ($branch,$path);
 441 }
 442
 443 sub branch_rev($$) {
 444
 445         my ($srcbranch,$uptorev) = @_;
 446
 447         my $bbranches = $branches{$srcbranch};
 448         my @revs = reverse sort { ($a eq 'LAST' ? 0 : $a) <=> ($b eq 'LAST' ? 0 : $b) } keys %$bbranches;
 449         my $therev;
 450         foreach my $arev(@revs) {
 451                 next if  ($arev eq 'LAST');
 452                 if ($arev <= $uptorev) {
 453                         $therev = $arev;
 454                         last;
 455                 }
 456         }
 457         return $therev;
 458 }
 459
 460 sub copy_path($$$$$$$$) {
 461         # Somebody copied a whole subdirectory.
 462         # We need to find the index entries from the old version which the
 463         # SVN log entry points to, and add them to the new place.
 464
 465         my($newrev,$newbranch,$path,$oldpath,$rev,$node_kind,$new,$parents) = @_;
 466
 467         my($srcbranch,$srcpath) = split_path($rev,$oldpath);
 468         unless(defined $srcbranch) {
 469                 print "Path not found when copying from $oldpath @ $rev\n";
 470                 return;
 471         }
 472         my $therev = branch_rev($srcbranch, $rev);
 473         my $gitrev = $branches{$srcbranch}{$therev};
 474         unless($gitrev) {
 475                 print STDERR "$newrev:$newbranch: could not find $oldpath \@ $rev\n";
 476                 return;
 477         }
 478         if ($srcbranch ne $newbranch) {
 479                 push(@$parents, $branches{$srcbranch}{'LAST'});
 480         }
 481         print "$newrev:$newbranch:$path: copying from $srcbranch:$srcpath @ $rev\n" if $opt_v;
 482         if ($node_kind eq $SVN::Node::dir) {
 483                         $srcpath =~ s#/*$#/#;
 484         }
 485
 486         my $pid = open my $f,'-|';
 487         die $! unless defined $pid;
 488         if (!$pid) {
 489                 exec("git-ls-tree","-r","-z",$gitrev,$srcpath)
 490                         or die $!;
 491         }
 492         local $/ = "\0";
 493         while(<$f>) {
 494                 chomp;
 495                 my($m,$p) = split(/\t/,$_,2);
 496                 my($mode,$type,$sha1) = split(/ /,$m);
 497                 next if $type ne "blob";
 498                 if ($node_kind eq $SVN::Node::dir) {
 499                         $p = $path . substr($p,length($srcpath)-1);
 500                 } else {
 501                         $p = $path;
 502                 }
 503                 push(@$new,[$mode,$sha1,$p]);
 504         }
 505         close($f) or
 506                 print STDERR "$newrev:$newbranch: could not list files in $oldpath \@ $rev\n";
 507 }
 508
 509 sub commit {
 510         my($branch, $changed_paths, $revision, $author, $date, $message) = @_;
 511         my($author_name,$author_email,$dest);
 512         my(@old,@new,@parents);
 513
 514         if (not defined $author) {
 515                 $author_name = $author_email = "unknown";
 516         } elsif (defined $users_file) {
 517                 die "User $author is not listed in $users_file\n"
 518                     unless exists $users{$author};
 519                 ($author_name,$author_email) = @{$users{$author}};
 520         } elsif ($author =~ /^(.*?)\s+<(.*)>$/) {
 521                 ($author_name, $author_email) = ($1, $2);
 522         } else {
 523                 $author =~ s/^<(.*)>$/$1/;
 524                 $author_name = $author_email = $author;
 525         }
 526         $date = pdate($date);
 527
 528         my $tag;
 529         my $parent;
 530         if($branch eq "/") { # trunk
 531                 $parent = $opt_o;
 532         } elsif($branch =~ m#^/(.+)#) { # tag
 533                 $tag = 1;
 534                 $parent = $1;
 535         } else { # "normal" branch
 536                 # nothing to do
 537                 $parent = $branch;
 538         }
 539         $dest = $parent;
 540
 541         my $prev = $changed_paths->{"/"};
 542         if($prev and $prev->[0] eq "A") {
 543                 delete $changed_paths->{"/"};
 544                 my $oldpath = $prev->[1];
 545                 my $rev;
 546                 if(defined $oldpath) {
 547                         my $p;
 548                         ($parent,$p) = split_path($revision,$oldpath);
 549                         if($parent eq "/") {
 550                                 $parent = $opt_o;
 551                         } else {
 552                                 $parent =~ s#^/##; # if it's a tag
 553                         }
 554                 } else {
 555                         $parent = undef;
 556                 }
 557         }
 558
 559         my $rev;
 560         if($revision > $opt_s and defined $parent) {
 561                 open(H,"git-rev-parse --verify $parent |");
 562                 $rev = <H>;
 563                 close(H) or do {
 564                         print STDERR "$revision: cannot find commit '$parent'!\n";
 565                         return;
 566                 };
 567                 chop $rev;
 568                 if(length($rev) != 40) {
 569                         print STDERR "$revision: cannot find commit '$parent'!\n";
 570                         return;
 571                 }
 572                 $rev = $branches{($parent eq $opt_o) ? "/" : $parent}{"LAST"};
 573                 if($revision != $opt_s and not $rev) {
 574                         print STDERR "$revision: do not know ancestor for '$parent'!\n";
 575                         return;
 576                 }
 577         } else {
 578                 $rev = undef;
 579         }
 580
 581 #       if($prev and $prev->[0] eq "A") {
 582 #               if(not $tag) {
 583 #                       unless(open(H,"> $git_dir/refs/heads/$branch")) {
 584 #                               print STDERR "$revision: Could not create branch $branch: $!\n";
 585 #                               $state=11;
 586 #                               next;
 587 #                       }
 588 #                       print H "$rev\n"
 589 #                               or die "Could not write branch $branch: $!";
 590 #                       close(H)
 591 #                               or die "Could not write branch $branch: $!";
 592 #               }
 593 #       }
 594         if(not defined $rev) {
 595                 unlink($git_index);
 596         } elsif ($rev ne $last_rev) {
 597                 print "Switching from $last_rev to $rev ($branch)\n" if $opt_v;
 598                 system("git-read-tree", $rev);
 599                 die "read-tree failed for $rev: $?\n" if $?;
 600                 $last_rev = $rev;
 601         }
 602
 603         push (@parents, $rev) if defined $rev;
 604
 605         my $cid;
 606         if($tag and not %$changed_paths) {
 607                 $cid = $rev;
 608         } else {
 609                 my @paths = sort keys %$changed_paths;
 610                 foreach my $path(@paths) {
 611                         my $action = $changed_paths->{$path};
 612
 613                         if ($action->[0] eq "R") {
 614                                 # refer to a file/tree in an earlier commit
 615                                 push(@old,$path); # remove any old stuff
 616                         }
 617                         if(($action->[0] eq "A") || ($action->[0] eq "R")) {
 618                                 my $node_kind = node_kind($branch,$path,$revision);
 619                                 if($action->[1]) {
 620                                         copy_path($revision,$branch,$path,$action->[1],$action->[2],$node_kind,\@new,\@parents);
 621                                 } elsif ($node_kind eq $SVN::Node::file) {
 622                                         my $f = get_file($revision,$branch,$path);
 623                                         if ($f) {
 624                                                 push(@new,$f) if $f;
 625                                         } else {
 626                                                 my $opath = $action->[3];
 627                                                 print STDERR "$revision: $branch: could not fetch '$opath'\n";
 628                                         }
 629                                 } elsif ($node_kind eq $SVN::Node::dir) {
 630                                         get_ignore(\@new, \@old, $revision,
 631                                                    $branch,$path);
 632                                 }
 633                         } elsif ($action->[0] eq "D") {
 634                                 push(@old,$path);
 635                         } elsif ($action->[0] eq "M") {
 636                                 my $node_kind = node_kind($branch,$path,$revision);
 637                                 if ($node_kind eq $SVN::Node::file) {
 638                                         my $f = get_file($revision,$branch,$path);
 639                                         push(@new,$f) if $f;
 640                                 } elsif ($node_kind eq $SVN::Node::dir) {
 641                                         get_ignore(\@new, \@old, $revision,
 642                                                    $branch,$path);
 643                                 }
 644                         } else {
 645                                 die "$revision: unknown action '".$action->[0]."' for $path\n";
 646                         }
 647                 }
 648
 649                 while(@old) {
 650                         my @o1;
 651                         if(@old > 55) {
 652                                 @o1 = splice(@old,0,50);
 653                         } else {
 654                                 @o1 = @old;
 655                                 @old = ();
 656                         }
 657                         my $pid = open my $F, "-|";
 658                         die "$!" unless defined $pid;
 659                         if (!$pid) {
 660                                 exec("git-ls-files", "-z", @o1) or die $!;
 661                         }
 662                         @o1 = ();
 663                         local $/ = "\0";
 664                         while(<$F>) {
 665                                 chomp;
 666                                 push(@o1,$_);
 667                         }
 668                         close($F);
 669
 670                         while(@o1) {
 671                                 my @o2;
 672                                 if(@o1 > 55) {
 673                                         @o2 = splice(@o1,0,50);
 674                                 } else {
 675                                         @o2 = @o1;
 676                                         @o1 = ();
 677                                 }
 678                                 system("git-update-index","--force-remove","--",@o2);
 679                                 die "Cannot remove files: $?\n" if $?;
 680                         }
 681                 }
 682                 while(@new) {
 683                         my @n2;
 684                         if(@new > 12) {
 685                                 @n2 = splice(@new,0,10);
 686                         } else {
 687                                 @n2 = @new;
 688                                 @new = ();
 689                         }
 690                         system("git-update-index","--add",
 691                                 (map { ('--cacheinfo', @$_) } @n2));
 692                         die "Cannot add files: $?\n" if $?;
 693                 }
 694
 695                 my $pid = open(C,"-|");
 696                 die "Cannot fork: $!" unless defined $pid;
 697                 unless($pid) {
 698                         exec("git-write-tree");
 699                         die "Cannot exec git-write-tree: $!\n";
 700                 }
 701                 chomp(my $tree = <C>);
 702                 length($tree) == 40
 703                         or die "Cannot get tree id ($tree): $!\n";
 704                 close(C)
 705                         or die "Error running git-write-tree: $?\n";
 706                 print "Tree ID $tree\n" if $opt_v;
 707
 708                 my $pr = IO::Pipe->new() or die "Cannot open pipe: $!\n";
 709                 my $pw = IO::Pipe->new() or die "Cannot open pipe: $!\n";
 710                 $pid = fork();
 711                 die "Fork: $!\n" unless defined $pid;
 712                 unless($pid) {
 713                         $pr->writer();
 714                         $pw->reader();
 715                         open(OUT,">&STDOUT");
 716                         dup2($pw->fileno(),0);
 717                         dup2($pr->fileno(),1);
 718                         $pr->close();
 719                         $pw->close();
 720
 721                         my @par = ();
 722
 723                         # loose detection of merges
 724                         # based on the commit msg
 725                         foreach my $rx (@mergerx) {
 726                                 if ($message =~ $rx) {
 727                                         my $mparent = $1;
 728                                         if ($mparent eq 'HEAD') { $mparent = $opt_o };
 729                                         if ( -e "$git_dir/refs/heads/$mparent") {
 730                                                 $mparent = get_headref($mparent, $git_dir);
 731                                                 push (@parents, $mparent);
 732                                                 print OUT "Merge parent branch: $mparent\n" if $opt_v;
 733                                         }
 734                                 }
 735                         }
 736                         my %seen_parents = ();
 737                         my @unique_parents = grep { ! $seen_parents{$_} ++ } @parents;
 738                         foreach my $bparent (@unique_parents) {
 739                                 push @par, '-p', $bparent;
 740                                 print OUT "Merge parent branch: $bparent\n" if $opt_v;
 741                         }
 742
 743                         exec("env",
 744                                 "GIT_AUTHOR_NAME=$author_name",
 745                                 "GIT_AUTHOR_EMAIL=$author_email",
 746                                 "GIT_AUTHOR_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
 747                                 "GIT_COMMITTER_NAME=$author_name",
 748                                 "GIT_COMMITTER_EMAIL=$author_email",
 749                                 "GIT_COMMITTER_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
 750                                 "git-commit-tree", $tree,@par);
 751                         die "Cannot exec git-commit-tree: $!\n";
 752                 }
 753                 $pw->writer();
 754                 $pr->reader();
 755
 756                 $message =~ s/[\s\n]+\z//;
 757                 $message = "r$revision: $message" if $opt_r;
 758
 759                 print $pw "$message\n"
 760                         or die "Error writing to git-commit-tree: $!\n";
 761                 $pw->close();
 762
 763                 print "Committed change $revision:$branch ".strftime("%Y-%m-%d %H:%M:%S",gmtime($date)).")\n" if $opt_v;
 764                 chomp($cid = <$pr>);
 765                 length($cid) == 40
 766                         or die "Cannot get commit id ($cid): $!\n";
 767                 print "Commit ID $cid\n" if $opt_v;
 768                 $pr->close();
 769
 770                 waitpid($pid,0);
 771                 die "Error running git-commit-tree: $?\n" if $?;
 772         }
 773
 774         if (not defined $cid) {
 775                 $cid = $branches{"/"}{"LAST"};
 776         }
 777
 778         if(not defined $dest) {
 779                 print "... no known parent\n" if $opt_v;
 780         } elsif(not $tag) {
 781                 print "Writing to refs/heads/$dest\n" if $opt_v;
 782                 open(C,">$git_dir/refs/heads/$dest") and
 783                 print C ("$cid\n") and
 784                 close(C)
 785                         or die "Cannot write branch $dest for update: $!\n";
 786         }
 787
 788         if($tag) {
 789                 my($in, $out) = ('','');
 790                 $last_rev = "-" if %$changed_paths;
 791                 # the tag was 'complex', i.e. did not refer to a "real" revision
 792
 793                 $dest =~ tr/_/\./ if $opt_u;
 794                 $branch = $dest;
 795
 796                 my $pid = open2($in, $out, 'git-mktag');
 797                 print $out ("object $cid\n".
 798                     "type commit\n".
 799                     "tag $dest\n".
 800                     "tagger $author_name <$author_email>\n") and
 801                 close($out)
 802                     or die "Cannot create tag object $dest: $!\n";
 803
 804                 my $tagobj = <$in>;
 805                 chomp $tagobj;
 806
 807                 if ( !close($in) or waitpid($pid, 0) != $pid or
 808                                 $? != 0 or $tagobj !~ /^[0123456789abcdef]{40}$/ ) {
 809                         die "Cannot create tag object $dest: $!\n";
 810                 }
 811
 812                 open(C,">$git_dir/refs/tags/$dest") and
 813                 print C ("$tagobj\n") and
 814                 close(C)
 815                         or die "Cannot create tag $branch: $!\n";
 816
 817                 print "Created tag '$dest' on '$branch'\n" if $opt_v;
 818         }
 819         $branches{$branch}{"LAST"} = $cid;
 820         $branches{$branch}{$revision} = $cid;
 821         $last_rev = $cid;
 822         print BRANCHES "$revision $branch $cid\n";
 823         print "DONE: $revision $dest $cid\n" if $opt_v;
 824 }
 825
 826 sub commit_all {
 827         # Recursive use of the SVN connection does not work
 828         local $svn = $svn2;
 829
 830         my ($changed_paths, $revision, $author, $date, $message, $pool) = @_;
 831         my %p;
 832         while(my($path,$action) = each %$changed_paths) {
 833                 $p{$path} = [ $action->action,$action->copyfrom_path, $action->copyfrom_rev, $path ];
 834         }
 835         $changed_paths = \%p;
 836
 837         my %done;
 838         my @col;
 839         my $pref;
 840         my $branch;
 841
 842         while(my($path,$action) = each %$changed_paths) {
 843                 ($branch,$path) = split_path($revision,$path);
 844                 next if not defined $branch;
 845                 $done{$branch}{$path} = $action;
 846         }
 847         while(($branch,$changed_paths) = each %done) {
 848                 commit($branch, $changed_paths, $revision, $author, $date, $message);
 849         }
 850 }
 851
 852 $opt_l = $svn->{'maxrev'} if not defined $opt_l or $opt_l > $svn->{'maxrev'};
 853
 854 if ($svn->{'maxrev'} < $current_rev) {
 855     print "Up to date: no new revisions to fetch!\n" if $opt_v;
 856     unlink("$git_dir/SVN2GIT_HEAD");
 857     exit;
 858 }
 859
 860 print "Fetching from $current_rev to $opt_l ...\n" if $opt_v;
 861
 862 my $pool=SVN::Pool->new;
 863 $svn->{'svn'}->get_log("/",$current_rev,$opt_l,0,1,1,\&commit_all,$pool);
 864 $pool->clear;
 865
 866
 867 unlink($git_index);
 868
 869 if (defined $orig_git_index) {
 870         $ENV{GIT_INDEX_FILE} = $orig_git_index;
 871 } else {
 872         delete $ENV{GIT_INDEX_FILE};
 873 }
 874
 875 # Now switch back to the branch we were in before all of this happened
 876 if($orig_branch) {
 877         print "DONE\n" if $opt_v and (not defined $opt_l or $opt_l > 0);
 878         system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
 879                 if $forward_master;
 880         unless ($opt_i) {
 881                 system('git-read-tree', '-m', '-u', 'SVN2GIT_HEAD', 'HEAD');
 882                 die "read-tree failed: $?\n" if $?;
 883         }
 884 } else {
 885         $orig_branch = "master";
 886         print "DONE; creating $orig_branch branch\n" if $opt_v and (not defined $opt_l or $opt_l > 0);
 887         system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
 888                 unless -f "$git_dir/refs/heads/master";
 889         system('git-update-ref', 'HEAD', "$orig_branch");
 890         unless ($opt_i) {
 891                 system('git checkout');
 892                 die "checkout failed: $?\n" if $?;
 893         }
 894 }
 895 unlink("$git_dir/SVN2GIT_HEAD");
 896 close(BRANCHES);