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 require v5.8.0; # for shell-safe open("-|",LIST)
  14 use strict;
  15 use warnings;
  16 use Getopt::Std;
  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 CVN: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,$opt_T,$opt_b,$opt_s,$opt_l);
  34
  35 sub usage() {
  36         print STDERR <<END;
  37 Usage: ${\basename $0}     # fetch/update GIT from CVS
  38        [-o branch-for-HEAD] [-h] [-v] [-l max_num_changes]
  39        [-C GIT_repository] [-t tagname] [-T trunkname] [-b branchname]
  40        [-i] [-u] [-s start_chg] [-m] [-M regex] [SVN_URL]
  41 END
  42         exit(1);
  43 }
  44
  45 getopts("b:C:hil:mM:o:s:t:T:uv") or usage();
  46 usage if $opt_h;
  47
  48 my $tag_name = $opt_t || "tags";
  49 my $trunk_name = $opt_T || "trunk";
  50 my $branch_name = $opt_b || "branches";
  51
  52 @ARGV == 1 or usage();
  53
  54 $opt_o ||= "origin";
  55 $opt_s ||= 1;
  56 $opt_l = 100 unless defined $opt_l;
  57 my $git_tree = $opt_C;
  58 $git_tree ||= ".";
  59
  60 my $svn_url = $ARGV[0];
  61
  62 our @mergerx = ();
  63 if ($opt_m) {
  64         @mergerx = ( qr/\W(?:from|of|merge|merging|merged) (\w+)/i );
  65 }
  66 if ($opt_M) {
  67         push (@mergerx, qr/$opt_M/);
  68 }
  69
  70 select(STDERR); $|=1; select(STDOUT);
  71
  72
  73 package SVNconn;
  74 # Basic SVN connection.
  75 # We're only interested in connecting and downloading, so ...
  76
  77 use File::Spec;
  78 use File::Temp qw(tempfile);
  79 use POSIX qw(strftime dup2);
  80
  81 sub new {
  82         my($what,$repo) = @_;
  83         $what=ref($what) if ref($what);
  84
  85         my $self = {};
  86         $self->{'buffer'} = "";
  87         bless($self,$what);
  88
  89         $repo =~ s#/+$##;
  90         $self->{'fullrep'} = $repo;
  91         $self->conn();
  92
  93         return $self;
  94 }
  95
  96 sub conn {
  97         my $self = shift;
  98         my $repo = $self->{'fullrep'};
  99         my $s = SVN::Ra->new($repo);
 100
 101         die "SVN connection to $repo: $!\n" unless defined $s;
 102         $self->{'svn'} = $s;
 103         $self->{'repo'} = $repo;
 104         $self->{'maxrev'} = $s->get_latest_revnum();
 105 }
 106
 107 sub file {
 108         my($self,$path,$rev) = @_;
 109         my $res;
 110
 111         my ($fh, $name) = tempfile('gitsvn.XXXXXX',
 112                     DIR => File::Spec->tmpdir(), UNLINK => 1);
 113
 114         print "... $rev $path ...\n" if $opt_v;
 115         eval { $self->{'svn'}->get_file($path,$rev,$fh); };
 116         if ($@ and $@ !~ /Attempted to get checksum/) {
 117             # retry
 118             $self->conn();
 119                 eval { $self->{'svn'}->get_file($path,$rev,$fh); };
 120         };
 121         return () if $@ and $@ !~ /Attempted to get checksum/;
 122         die $@ if $@;
 123         close ($fh);
 124
 125         return ($name, $res);
 126 }
 127
 128
 129 package main;
 130
 131 my $svn = SVNconn->new($svn_url);
 132
 133
 134 sub pdate($) {
 135         my($d) = @_;
 136         $d =~ m#(\d\d\d\d)-(\d\d)-(\d\d)T(\d\d):(\d\d):(\d\d)#
 137                 or die "Unparseable date: $d\n";
 138         my $y=$1; $y-=1900 if $y>1900;
 139         return timegm($6||0,$5,$4,$3,$2-1,$y);
 140 }
 141
 142 sub getwd() {
 143         my $pwd = `pwd`;
 144         chomp $pwd;
 145         return $pwd;
 146 }
 147
 148
 149 sub get_headref($$) {
 150     my $name    = shift;
 151     my $git_dir = shift;
 152     my $sha;
 153
 154     if (open(C,"$git_dir/refs/heads/$name")) {
 155         chomp($sha = <C>);
 156         close(C);
 157         length($sha) == 40
 158             or die "Cannot get head id for $name ($sha): $!\n";
 159     }
 160     return $sha;
 161 }
 162
 163
 164 -d $git_tree
 165         or mkdir($git_tree,0777)
 166         or die "Could not create $git_tree: $!";
 167 chdir($git_tree);
 168
 169 my $orig_branch = "";
 170 my $forward_master = 0;
 171 my %branches;
 172
 173 my $git_dir = $ENV{"GIT_DIR"} || ".git";
 174 $git_dir = getwd()."/".$git_dir unless $git_dir =~ m#^/#;
 175 $ENV{"GIT_DIR"} = $git_dir;
 176 my $orig_git_index;
 177 $orig_git_index = $ENV{GIT_INDEX_FILE} if exists $ENV{GIT_INDEX_FILE};
 178 my ($git_ih, $git_index) = tempfile('gitXXXXXX', SUFFIX => '.idx',
 179                                     DIR => File::Spec->tmpdir());
 180 close ($git_ih);
 181 $ENV{GIT_INDEX_FILE} = $git_index;
 182 my $maxnum = 0;
 183 my $last_rev = "";
 184 my $last_branch;
 185 my $current_rev = $opt_s-1;
 186 unless(-d $git_dir) {
 187         system("git-init-db");
 188         die "Cannot init the GIT db at $git_tree: $?\n" if $?;
 189         system("git-read-tree");
 190         die "Cannot init an empty tree: $?\n" if $?;
 191
 192         $last_branch = $opt_o;
 193         $orig_branch = "";
 194 } else {
 195         -f "$git_dir/refs/heads/$opt_o"
 196                 or die "Branch '$opt_o' does not exist.\n".
 197                        "Either use the correct '-o branch' option,\n".
 198                        "or import to a new repository.\n";
 199
 200         -f "$git_dir/svn2git"
 201                 or die "'$git_dir/svn2git' does not exist.\n".
 202                        "You need that file for incremental imports.\n";
 203         $last_branch = basename(readlink("$git_dir/HEAD"));
 204         unless($last_branch) {
 205                 warn "Cannot read the last branch name: $! -- assuming 'master'\n";
 206                 $last_branch = "master";
 207         }
 208         $orig_branch = $last_branch;
 209         $last_rev = get_headref($orig_branch, $git_dir);
 210         if (-f "$git_dir/SVN2GIT_HEAD") {
 211                 die <<EOM;
 212 SVN2GIT_HEAD exists.
 213 Make sure your working directory corresponds to HEAD and remove SVN2GIT_HEAD.
 214 You may need to run
 215
 216     git-read-tree -m -u SVN2GIT_HEAD HEAD
 217 EOM
 218         }
 219         system('cp', "$git_dir/HEAD", "$git_dir/SVN2GIT_HEAD");
 220
 221         $forward_master =
 222             $opt_o ne 'master' && -f "$git_dir/refs/heads/master" &&
 223             system('cmp', '-s', "$git_dir/refs/heads/master",
 224                                 "$git_dir/refs/heads/$opt_o") == 0;
 225
 226         # populate index
 227         system('git-read-tree', $last_rev);
 228         die "read-tree failed: $?\n" if $?;
 229
 230         # Get the last import timestamps
 231         open my $B,"<", "$git_dir/svn2git";
 232         while(<$B>) {
 233                 chomp;
 234                 my($num,$branch,$ref) = split;
 235                 $branches{$branch}{$num} = $ref;
 236                 $branches{$branch}{"LAST"} = $ref;
 237                 $current_rev = $num if $current_rev < $num;
 238         }
 239         close($B);
 240 }
 241 -d $git_dir
 242         or die "Could not create git subdir ($git_dir).\n";
 243
 244 open BRANCHES,">>", "$git_dir/svn2git";
 245
 246 sub get_file($$$) {
 247         my($rev,$branch,$path) = @_;
 248
 249         # revert split_path(), below
 250         my $svnpath;
 251         $path = "" if $path eq "/"; # this should not happen, but ...
 252         if($branch eq "/") {
 253                 $svnpath = "/$trunk_name/$path";
 254         } elsif($branch =~ m#^/#) {
 255                 $svnpath = "/$tag_name$branch/$path";
 256         } else {
 257                 $svnpath = "/$branch_name/$branch/$path";
 258         }
 259
 260         # now get it
 261         my ($name, $res) = eval { $svn->file($svnpath,$rev); };
 262         return () unless defined $name;
 263
 264         open my $F, '-|', "git-hash-object", "-w", $name
 265                 or die "Cannot create object: $!\n";
 266         my $sha = <$F>;
 267         chomp $sha;
 268         close $F;
 269         unlink $name;
 270         my $mode = "0644"; # SV does not seem to store any file modes
 271         return [$mode, $sha, $path];
 272 }
 273
 274 sub split_path($$) {
 275         my($rev,$path) = @_;
 276         my $branch;
 277
 278         if($path =~ s#^/\Q$tag_name\E/([^/]+)/?##) {
 279                 $branch = "/$1";
 280         } elsif($path =~ s#^/\Q$trunk_name\E/?##) {
 281                 $branch = "/";
 282         } elsif($path =~ s#^/\Q$branch_name\E/([^/]+)/?##) {
 283                 $branch = $1;
 284         } else {
 285                 print STDERR "$rev: Unrecognized path: $path\n";
 286                 return ()
 287         }
 288         $path = "/" if $path eq "";
 289         return ($branch,$path);
 290 }
 291
 292 sub commit {
 293         my($branch, $changed_paths, $revision, $author, $date, $message) = @_;
 294         my($author_name,$author_email,$dest);
 295         my(@old,@new);
 296
 297         if (not defined $author) {
 298                 $author_name = $author_email = "unknown";
 299         } elsif ($author =~ /^(.*?)\s+<(.*)>$/) {
 300                 ($author_name, $author_email) = ($1, $2);
 301         } else {
 302                 $author =~ s/^<(.*)>$/$1/;
 303                 $author_name = $author_email = $author;
 304         }
 305         $date = pdate($date);
 306
 307         my $tag;
 308         my $parent;
 309         if($branch eq "/") { # trunk
 310                 $parent = $opt_o;
 311         } elsif($branch =~ m#^/(.+)#) { # tag
 312                 $tag = 1;
 313                 $parent = $1;
 314         } else { # "normal" branch
 315                 # nothing to do
 316                 $parent = $branch;
 317         }
 318         $dest = $parent;
 319
 320         my $prev = $changed_paths->{"/"};
 321         if($prev and $prev->[0] eq "A") {
 322                 delete $changed_paths->{"/"};
 323                 my $oldpath = $prev->[1];
 324                 my $rev;
 325                 if(defined $oldpath) {
 326                         my $p;
 327                         ($parent,$p) = split_path($revision,$oldpath);
 328                         if($parent eq "/") {
 329                                 $parent = $opt_o;
 330                         } else {
 331                                 $parent =~ s#^/##; # if it's a tag
 332                         }
 333                 } else {
 334                         $parent = undef;
 335                 }
 336         }
 337
 338         my $rev;
 339         if($revision > $opt_s and defined $parent) {
 340                 open(H,"git-rev-parse --verify $parent |");
 341                 $rev = <H>;
 342                 close(H) or do {
 343                         print STDERR "$revision: cannot find commit '$parent'!\n";
 344                         return;
 345                 };
 346                 chop $rev;
 347                 if(length($rev) != 40) {
 348                         print STDERR "$revision: cannot find commit '$parent'!\n";
 349                         return;
 350                 }
 351                 $rev = $branches{($parent eq $opt_o) ? "/" : $parent}{"LAST"};
 352                 if($revision != $opt_s and not $rev) {
 353                         print STDERR "$revision: do not know ancestor for '$parent'!\n";
 354                         return;
 355                 }
 356         } else {
 357                 $rev = undef;
 358         }
 359
 360 #       if($prev and $prev->[0] eq "A") {
 361 #               if(not $tag) {
 362 #                       unless(open(H,"> $git_dir/refs/heads/$branch")) {
 363 #                               print STDERR "$revision: Could not create branch $branch: $!\n";
 364 #                               $state=11;
 365 #                               next;
 366 #                       }
 367 #                       print H "$rev\n"
 368 #                               or die "Could not write branch $branch: $!";
 369 #                       close(H)
 370 #                               or die "Could not write branch $branch: $!";
 371 #               }
 372 #       }
 373         if(not defined $rev) {
 374                 unlink($git_index);
 375         } elsif ($rev ne $last_rev) {
 376                 print "Switching from $last_rev to $rev ($branch)\n" if $opt_v;
 377                 system("git-read-tree", $rev);
 378                 die "read-tree failed for $rev: $?\n" if $?;
 379                 $last_rev = $rev;
 380         }
 381
 382         my $cid;
 383         if($tag and not %$changed_paths) {
 384                 $cid = $rev;
 385         } else {
 386                 while(my($path,$action) = each %$changed_paths) {
 387                         if ($action->[0] eq "A") {
 388                                 my $f = get_file($revision,$branch,$path);
 389                                 push(@new,$f) if $f;
 390                         } elsif ($action->[0] eq "D") {
 391                                 push(@old,$path);
 392                         } elsif ($action->[0] eq "M") {
 393                                 my $f = get_file($revision,$branch,$path);
 394                                 push(@new,$f) if $f;
 395                         } elsif ($action->[0] eq "R") {
 396                                 # refer to a file/tree in an earlier commit
 397                                 push(@old,$path); # remove any old stuff
 398
 399                                 # ... and add any new stuff
 400                                 my($b,$p) = split_path($revision,$action->[1]);
 401                                 open my $F,"-|","git-ls-tree","-r","-z", $branches{$b}{$action->[2]}, $p;
 402                                 local $/ = '\0';
 403                                 while(<$F>) {
 404                                         chomp;
 405                                         my($m,$p) = split(/\t/,$_,2);
 406                                         my($mode,$type,$sha1) = split(/ /,$m);
 407                                         next if $type ne "blob";
 408                                         push(@new,[$mode,$sha1,$p]);
 409                                 }
 410                         } else {
 411                                 die "$revision: unknown action '".$action->[0]."' for $path\n";
 412                         }
 413                 }
 414
 415                 if(@old) {
 416                         open my $F, "-|", "git-ls-files", "-z", @old or die $!;
 417                         @old = ();
 418                         local $/ = '\0';
 419                         while(<$F>) {
 420                                 chomp;
 421                                 push(@old,$_);
 422                         }
 423                         close($F);
 424
 425                         while(@old) {
 426                                 my @o2;
 427                                 if(@old > 55) {
 428                                         @o2 = splice(@old,0,50);
 429                                 } else {
 430                                         @o2 = @old;
 431                                         @old = ();
 432                                 }
 433                                 system("git-update-index","--force-remove","--",@o2);
 434                                 die "Cannot remove files: $?\n" if $?;
 435                         }
 436                 }
 437                 while(@new) {
 438                         my @n2;
 439                         if(@new > 12) {
 440                                 @n2 = splice(@new,0,10);
 441                         } else {
 442                                 @n2 = @new;
 443                                 @new = ();
 444                         }
 445                         system("git-update-index","--add",
 446                                 (map { ('--cacheinfo', @$_) } @n2));
 447                         die "Cannot add files: $?\n" if $?;
 448                 }
 449
 450                 my $pid = open(C,"-|");
 451                 die "Cannot fork: $!" unless defined $pid;
 452                 unless($pid) {
 453                         exec("git-write-tree");
 454                         die "Cannot exec git-write-tree: $!\n";
 455                 }
 456                 chomp(my $tree = <C>);
 457                 length($tree) == 40
 458                         or die "Cannot get tree id ($tree): $!\n";
 459                 close(C)
 460                         or die "Error running git-write-tree: $?\n";
 461                 print "Tree ID $tree\n" if $opt_v;
 462
 463                 my $pr = IO::Pipe->new() or die "Cannot open pipe: $!\n";
 464                 my $pw = IO::Pipe->new() or die "Cannot open pipe: $!\n";
 465                 $pid = fork();
 466                 die "Fork: $!\n" unless defined $pid;
 467                 unless($pid) {
 468                         $pr->writer();
 469                         $pw->reader();
 470                         open(OUT,">&STDOUT");
 471                         dup2($pw->fileno(),0);
 472                         dup2($pr->fileno(),1);
 473                         $pr->close();
 474                         $pw->close();
 475
 476                         my @par = ();
 477                         @par = ("-p",$rev) if defined $rev;
 478
 479                         # loose detection of merges
 480                         # based on the commit msg
 481                         foreach my $rx (@mergerx) {
 482                                 if ($message =~ $rx) {
 483                                         my $mparent = $1;
 484                                         if ($mparent eq 'HEAD') { $mparent = $opt_o };
 485                                         if ( -e "$git_dir/refs/heads/$mparent") {
 486                                                 $mparent = get_headref($mparent, $git_dir);
 487                                                 push @par, '-p', $mparent;
 488                                                 print OUT "Merge parent branch: $mparent\n" if $opt_v;
 489                                         }
 490                                 }
 491                         }
 492
 493                         exec("env",
 494                                 "GIT_AUTHOR_NAME=$author_name",
 495                                 "GIT_AUTHOR_EMAIL=$author_email",
 496                                 "GIT_AUTHOR_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
 497                                 "GIT_COMMITTER_NAME=$author_name",
 498                                 "GIT_COMMITTER_EMAIL=$author_email",
 499                                 "GIT_COMMITTER_DATE=".strftime("+0000 %Y-%m-%d %H:%M:%S",gmtime($date)),
 500                                 "git-commit-tree", $tree,@par);
 501                         die "Cannot exec git-commit-tree: $!\n";
 502                 }
 503                 $pw->writer();
 504                 $pr->reader();
 505
 506                 $message =~ s/[\s\n]+\z//;
 507
 508                 print $pw "$message\n"
 509                         or die "Error writing to git-commit-tree: $!\n";
 510                 $pw->close();
 511
 512                 print "Committed change $revision:$branch ".strftime("%Y-%m-%d %H:%M:%S",gmtime($date)).")\n" if $opt_v;
 513                 chomp($cid = <$pr>);
 514                 length($cid) == 40
 515                         or die "Cannot get commit id ($cid): $!\n";
 516                 print "Commit ID $cid\n" if $opt_v;
 517                 $pr->close();
 518
 519                 waitpid($pid,0);
 520                 die "Error running git-commit-tree: $?\n" if $?;
 521         }
 522
 523         if(not defined $dest) {
 524                 print "... no known parent\n" if $opt_v;
 525         } elsif(not $tag) {
 526                 print "Writing to refs/heads/$dest\n" if $opt_v;
 527                 open(C,">$git_dir/refs/heads/$dest") and
 528                 print C ("$cid\n") and
 529                 close(C)
 530                         or die "Cannot write branch $dest for update: $!\n";
 531         }
 532
 533         if($tag) {
 534                 my($in, $out) = ('','');
 535                 $last_rev = "-" if %$changed_paths;
 536                 # the tag was 'complex', i.e. did not refer to a "real" revision
 537
 538                 $dest =~ tr/_/\./ if $opt_u;
 539
 540                 my $pid = open2($in, $out, 'git-mktag');
 541                 print $out ("object $cid\n".
 542                     "type commit\n".
 543                     "tag $dest\n".
 544                     "tagger $author_name <$author_email>\n") and
 545                 close($out)
 546                     or die "Cannot create tag object $dest: $!\n";
 547
 548                 my $tagobj = <$in>;
 549                 chomp $tagobj;
 550
 551                 if ( !close($in) or waitpid($pid, 0) != $pid or
 552                                 $? != 0 or $tagobj !~ /^[0123456789abcdef]{40}$/ ) {
 553                         die "Cannot create tag object $dest: $!\n";
 554                 }
 555
 556                 open(C,">$git_dir/refs/tags/$dest") and
 557                 print C ("$tagobj\n") and
 558                 close(C)
 559                         or die "Cannot create tag $branch: $!\n";
 560
 561                 print "Created tag '$dest' on '$branch'\n" if $opt_v;
 562         }
 563         $branches{$branch}{"LAST"} = $cid;
 564         $branches{$branch}{$revision} = $cid;
 565         $last_rev = $cid;
 566         print BRANCHES "$revision $branch $cid\n";
 567         print "DONE: $revision $dest $cid\n" if $opt_v;
 568 }
 569
 570 my ($changed_paths, $revision, $author, $date, $message, $pool) = @_;
 571 sub _commit_all {
 572         ($changed_paths, $revision, $author, $date, $message, $pool) = @_;
 573         my %p;
 574         while(my($path,$action) = each %$changed_paths) {
 575                 $p{$path} = [ $action->action,$action->copyfrom_path, $action->copyfrom_rev ];
 576         }
 577         $changed_paths = \%p;
 578 }
 579
 580 sub commit_all {
 581         my %done;
 582         my @col;
 583         my $pref;
 584         my $branch;
 585
 586         while(my($path,$action) = each %$changed_paths) {
 587                 ($branch,$path) = split_path($revision,$path);
 588                 next if not defined $branch;
 589                 $done{$branch}{$path} = $action;
 590         }
 591         while(($branch,$changed_paths) = each %done) {
 592                 commit($branch, $changed_paths, $revision, $author, $date, $message);
 593         }
 594 }
 595
 596 while(++$current_rev <= $svn->{'maxrev'}) {
 597         $svn->{'svn'}->get_log("/",$current_rev,$current_rev,$current_rev,1,1,\&_commit_all,"");
 598         commit_all();
 599         if($opt_l and not --$opt_l) {
 600                 print STDERR "Exiting due to a memory leak. Repeat, please.\n";
 601                 last;
 602         }
 603 }
 604
 605
 606 unlink($git_index);
 607
 608 if (defined $orig_git_index) {
 609         $ENV{GIT_INDEX_FILE} = $orig_git_index;
 610 } else {
 611         delete $ENV{GIT_INDEX_FILE};
 612 }
 613
 614 # Now switch back to the branch we were in before all of this happened
 615 if($orig_branch) {
 616         print "DONE\n" if $opt_v and (not defined $opt_l or $opt_l > 0);
 617         system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
 618                 if $forward_master;
 619         unless ($opt_i) {
 620                 system('git-read-tree', '-m', '-u', 'SVN2GIT_HEAD', 'HEAD');
 621                 die "read-tree failed: $?\n" if $?;
 622         }
 623 } else {
 624         $orig_branch = "master";
 625         print "DONE; creating $orig_branch branch\n" if $opt_v and (not defined $opt_l or $opt_l > 0);
 626         system("cp","$git_dir/refs/heads/$opt_o","$git_dir/refs/heads/master")
 627                 unless -f "$git_dir/refs/heads/master";
 628         unlink("$git_dir/HEAD");
 629         symlink("refs/heads/$orig_branch","$git_dir/HEAD");
 630         unless ($opt_i) {
 631                 system('git checkout');
 632                 die "checkout failed: $?\n" if $?;
 633         }
 634 }
 635 unlink("$git_dir/SVN2GIT_HEAD");
 636 close(BRANCHES);