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