git-archimport.perl 36.1 KB
Newer Older
1
#!/usr/bin/perl
2 3 4 5
#
# This tool is copyright (c) 2005, Martin Langhoff.
# It is released under the Gnu Public License, version 2.
#
Junio C Hamano's avatar
Junio C Hamano committed
6 7
# The basic idea is to walk the output of tla abrowse,
# fetch the changesets and apply them.
8
#
9

10 11
=head1 Invocation

12
    git archimport [ -h ] [ -v ] [ -o ] [ -a ] [ -f ] [ -T ]
Junio C Hamano's avatar
Junio C Hamano committed
13
	[ -D depth] [ -t tempdir ] <archive>/<branch> [ <archive>/<branch> ]
14

15 16
Imports a project from one or more Arch repositories. It will follow branches
and repositories within the namespaces defined by the <archive/branch>
17
parameters supplied. If it cannot find the remote branch a merge comes from
Junio C Hamano's avatar
Junio C Hamano committed
18
it will just import it as a regular commit. If it can find it, it will mark it
19
as a merge whenever possible.
20

21
See man (1) git-archimport for more details.
22

23
=head1 TODO
24

25
 - create tag objects instead of ref tags
26
 - audit shell-escaping of filenames
27
 - hide our private tags somewhere smarter
Junio C Hamano's avatar
Junio C Hamano committed
28
 - find a way to make "cat *patches | patch" safe even when patchfiles are missing newlines
29 30 31
 - sort and apply patches by graphing ancestry relations instead of just
   relying in dates supplied in the changeset itself.
   tla ancestry-graph -m could be helpful here...
32 33 34

=head1 Devel tricks

Junio C Hamano's avatar
Junio C Hamano committed
35
Add print in front of the shell commands invoked via backticks.
36

37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54
=head1 Devel Notes

There are several places where Arch and git terminology are intermixed
and potentially confused.

The notion of a "branch" in git is approximately equivalent to
a "archive/category--branch--version" in Arch.  Also, it should be noted
that the "--branch" portion of "archive/category--branch--version" is really
optional in Arch although not many people (nor tools!) seem to know this.
This means that "archive/category--version" is also a valid "branch"
in git terms.

We always refer to Arch names by their fully qualified variant (which
means the "archive" name is prefixed.

For people unfamiliar with Arch, an "archive" is the term for "repository",
and can contain multiple, unrelated branches.

55 56
=cut

57
use 5.008;
58 59 60
use strict;
use warnings;
use Getopt::Std;
61
use File::Temp qw(tempdir);
62
use File::Path qw(mkpath rmtree);
63 64 65 66 67 68 69
use File::Basename qw(basename dirname);
use Data::Dumper qw/ Dumper /;
use IPC::Open2;

$SIG{'PIPE'}="IGNORE";
$ENV{'TZ'}="UTC";

70 71
my $git_dir = $ENV{"GIT_DIR"} || ".git";
$ENV{"GIT_DIR"} = $git_dir;
72
my $ptag_dir = "$git_dir/archimport/tags";
73

74
our($opt_h,$opt_f,$opt_v,$opt_T,$opt_t,$opt_D,$opt_a,$opt_o);
75 76 77

sub usage() {
    print STDERR <<END;
78
usage: git archimport     # fetch/update GIT from Arch
79
       [ -h ] [ -v ] [ -o ] [ -a ] [ -f ] [ -T ] [ -D depth ] [ -t tempdir ]
80 81 82 83 84
       repository/arch-branch [ repository/arch-branch] ...
END
    exit(1);
}

85
getopts("fThvat:D:") or usage();
86 87 88
usage if $opt_h;

@ARGV >= 1 or usage();
89 90 91
# $arch_branches:
# values associated with keys:
#   =1 - Arch version / git 'branch' detected via abrowse on a limit
92
#   >1 - Arch version / git 'branch' of an auxiliary branch we've merged
93 94 95 96 97
my %arch_branches = map { my $branch = $_; $branch =~ s/:[^:]*$//; $branch => 1 } @ARGV;

# $branch_name_map:
# maps arch branches to git branch names
my %branch_name_map = map { m/^(.*):([^:]*)$/; $1 => $2 } grep { m/:/ } @ARGV;
98

99 100
$ENV{'TMPDIR'} = $opt_t if $opt_t; # $ENV{TMPDIR} will affect tempdir() calls:
my $tmp = tempdir('git-archimport-XXXXXX', TMPDIR => 1, CLEANUP => 1);
101
$opt_v && print "+ Using $tmp as temporary directory\n";
102

103 104 105 106 107 108 109 110 111
unless (-d $git_dir) { # initial import needs empty directory
    opendir DIR, '.' or die "Unable to open current directory: $!\n";
    while (my $entry = readdir DIR) {
        $entry =~ /^\.\.?$/ or
            die "Initial import needs an empty current working directory.\n"
    }
    closedir DIR
}

112
my $default_archive;		# default Arch archive
113 114
my %reachable = ();             # Arch repositories we can access
my %unreachable = ();           # Arch repositories we can't access :<
115
my @psets  = ();                # the collection
116
my %psets  = ();                # the collection, by name
117 118 119 120
my %stats  = (			# Track which strategy we used to import:
	get_tag => 0, replay => 0, get_new => 0, get_delta => 0,
        simple_changeset => 0, import_or_tag => 0
);
121 122 123

my %rptags = ();                # my reverse private tags
                                # to map a SHA1 to a commitid
124
my $TLA = $ENV{'ARCH_CLIENT'} || 'tla';
125

126 127 128 129
sub do_abrowse {
    my $stage = shift;
    while (my ($limit, $level) = each %arch_branches) {
        next unless $level == $stage;
Junio C Hamano's avatar
Junio C Hamano committed
130 131

	open ABROWSE, "$TLA abrowse -fkD --merges $limit |"
132
                                or die "Problems with tla abrowse: $!";
Junio C Hamano's avatar
Junio C Hamano committed
133

134 135
        my %ps        = ();         # the current one
        my $lastseen  = '';
Junio C Hamano's avatar
Junio C Hamano committed
136

137 138
        while (<ABROWSE>) {
            chomp;
Junio C Hamano's avatar
Junio C Hamano committed
139

140 141 142 143 144 145 146 147 148 149 150
            # first record padded w 8 spaces
            if (s/^\s{8}\b//) {
                my ($id, $type) = split(m/\s+/, $_, 2);

                my %last_ps;
                # store the record we just captured
                if (%ps && !exists $psets{ $ps{id} }) {
                    %last_ps = %ps; # break references
                    push (@psets, \%last_ps);
                    $psets{ $last_ps{id} } = \%last_ps;
                }
Junio C Hamano's avatar
Junio C Hamano committed
151

152 153 154 155 156
                my $branch = extract_versionname($id);
                %ps = ( id => $id, branch => $branch );
                if (%last_ps && ($last_ps{branch} eq $branch)) {
                    $ps{parent_id} = $last_ps{id};
                }
Junio C Hamano's avatar
Junio C Hamano committed
157

158 159 160 161 162 163 164 165
                $arch_branches{$branch} = 1;
                $lastseen = 'id';

                # deal with types (should work with baz or tla):
                if ($type =~ m/\(.*changeset\)/) {
                    $ps{type} = 's';
                } elsif ($type =~ /\(.*import\)/) {
                    $ps{type} = 'i';
166
                } elsif ($type =~ m/\(tag.*?(\S+\@\S+).*?\)/) {
167 168
                    $ps{type} = 't';
                    # read which revision we've tagged when we parse the log
169
                    $ps{tag}  = $1;
Junio C Hamano's avatar
Junio C Hamano committed
170
                } else {
171 172 173 174 175
                    warn "Unknown type $type";
                }

                $arch_branches{$branch} = 1;
                $lastseen = 'id';
Junio C Hamano's avatar
Junio C Hamano committed
176 177
            } elsif (s/^\s{10}//) {
                # 10 leading spaces or more
178
                # indicate commit metadata
Junio C Hamano's avatar
Junio C Hamano committed
179

180 181 182 183 184 185 186 187 188 189
                # date
                if ($lastseen eq 'id' && m/^(\d{4}-\d\d-\d\d \d\d:\d\d:\d\d)/){
                    $ps{date}   = $1;
                    $lastseen = 'date';
                } elsif ($_ eq 'merges in:') {
                    $ps{merges} = [];
                    $lastseen = 'merges';
                } elsif ($lastseen eq 'merges' && s/^\s{2}//) {
                    my $id = $_;
                    push (@{$ps{merges}}, $id);
Junio C Hamano's avatar
Junio C Hamano committed
190

191 192 193 194
                    # aggressive branch finding:
                    if ($opt_D) {
                        my $branch = extract_versionname($id);
                        my $repo = extract_reponame($branch);
Junio C Hamano's avatar
Junio C Hamano committed
195

196 197 198 199 200 201 202
                        if (archive_reachable($repo) &&
                                !defined $arch_branches{$branch}) {
                            $arch_branches{$branch} = $stage + 1;
                        }
                    }
                } else {
                    warn "more metadata after merges!?: $_\n" unless /^\s*$/;
203 204 205 206
                }
            }
        }

207 208 209 210 211
        if (%ps && !exists $psets{ $ps{id} }) {
            my %temp = %ps;         # break references
            if (@psets && $psets[$#psets]{branch} eq $ps{branch}) {
                $temp{parent_id} = $psets[$#psets]{id};
            }
Junio C Hamano's avatar
Junio C Hamano committed
212
            push (@psets, \%temp);
213
            $psets{ $temp{id} } = \%temp;
Junio C Hamano's avatar
Junio C Hamano committed
214 215
        }

216 217
        close ABROWSE or die "$TLA abrowse failed on $limit\n";
    }
218 219
}                               # end foreach $root

220 221 222 223 224 225 226 227
do_abrowse(1);
my $depth = 2;
$opt_D ||= 0;
while ($depth <= $opt_D) {
    do_abrowse($depth);
    $depth++;
}

228
## Order patches by time
229 230 231
# FIXME see if we can find a more optimal way to do this by graphing
# the ancestry data and walking it, that way we won't have to rely on
# client-supplied dates
232 233 234 235 236 237 238 239
@psets = sort {$a->{date}.$b->{id} cmp $b->{date}.$b->{id}} @psets;

#print Dumper \@psets;

##
## TODO cleanup irrelevant patches
##      and put an initial import
##      or a full tag
240
my $import = 0;
241
unless (-d $git_dir) { # initial import
242 243
    if ($psets[0]{type} eq 'i' || $psets[0]{type} eq 't') {
        print "Starting import from $psets[0]{id}\n";
244
	`git-init`;
245 246
	die $! if $?;
	$import = 1;
247 248 249
    } else {
        die "Need to start from an import or a tag -- cannot use $psets[0]{id}";
    }
250 251
} else {    # progressing an import
    # load the rptags
252
    opendir(DIR, $ptag_dir)
253 254
	|| die "can't opendir: $!";
    while (my $file = readdir(DIR)) {
255 256
        # skip non-interesting-files
        next unless -f "$ptag_dir/$file";
Junio C Hamano's avatar
Junio C Hamano committed
257

258 259 260 261 262 263 264 265
        # convert first '--' to '/' from old git-archimport to use
        # as an archivename/c--b--v private tag
        if ($file !~ m!,!) {
            my $oldfile = $file;
            $file =~ s!--!,!;
            print STDERR "converting old tag $oldfile to $file\n";
            rename("$ptag_dir/$oldfile", "$ptag_dir/$file") or die $!;
        }
266 267 268 269 270
	my $sha = ptag($file);
	chomp $sha;
	$rptags{$sha} = $file;
    }
    closedir DIR;
271 272
}

273
# process patchsets
274 275 276 277 278
# extract the Arch repository name (Arch "archive" in Arch-speak)
sub extract_reponame {
    my $fq_cvbr = shift; # archivename/[[[[category]branch]version]revision]
    return (split(/\//, $fq_cvbr))[0];
}
Junio C Hamano's avatar
Junio C Hamano committed
279

280 281 282 283 284
sub extract_versionname {
    my $name = shift;
    $name =~ s/--(?:patch|version(?:fix)?|base)-\d+$//;
    return $name;
}
285

286
# convert a fully-qualified revision or version to a unique dirname:
Junio C Hamano's avatar
Junio C Hamano committed
287
#   normalperson@yhbt.net-05/mpd--uclinux--1--patch-2
288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303
# becomes: normalperson@yhbt.net-05,mpd--uclinux--1
#
# the git notion of a branch is closer to
# archive/category--branch--version than archive/category--branch, so we
# use this to convert to git branch names.
# Also, keep archive names but replace '/' with ',' since it won't require
# subdirectories, and is safer than swapping '--' which could confuse
# reverse-mapping when dealing with bastard branches that
# are just archive/category--version  (no --branch)
sub tree_dirname {
    my $revision = shift;
    my $name = extract_versionname($revision);
    $name =~ s#/#,#;
    return $name;
}

304 305 306 307 308 309 310 311
# old versions of git-archimport just use the <category--branch> part:
sub old_style_branchname {
    my $id = shift;
    my $ret = safe_pipe_capture($TLA,'parse-package-name','-p',$id);
    chomp $ret;
    return $ret;
}

312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339
*git_default_branchname = $opt_o ? *old_style_branchname : *tree_dirname;

# retrieve default archive, since $branch_name_map keys might not include it
sub get_default_archive {
    if (!defined $default_archive) {
        $default_archive = safe_pipe_capture($TLA,'my-default-archive');
        chomp $default_archive;
    }
    return $default_archive;
}

sub git_branchname {
    my $revision = shift;
    my $name = extract_versionname($revision);

    if (exists $branch_name_map{$name}) {
	return $branch_name_map{$name};

    } elsif ($name =~ m#^([^/]*)/(.*)$#
	     && $1 eq get_default_archive()
	     && exists $branch_name_map{$2}) {
	# the names given in the command-line lacked the archive.
	return $branch_name_map{$2};

    } else {
	return git_default_branchname($revision);
    }
}
340

341 342
sub process_patchset_accurate {
    my $ps = shift;
Junio C Hamano's avatar
Junio C Hamano committed
343

344 345 346
    # switch to that branch if we're not already in that branch:
    if (-e "$git_dir/refs/heads/$ps->{branch}") {
       system('git-checkout','-f',$ps->{branch}) == 0 or die "$! $?\n";
347

348 349 350
       # remove any old stuff that got leftover:
       my $rm = safe_pipe_capture('git-ls-files','--others','-z');
       rmtree(split(/\0/,$rm)) if $rm;
351
    }
Junio C Hamano's avatar
Junio C Hamano committed
352

353 354 355 356 357 358 359 360 361 362 363 364
    # Apply the import/changeset/merge into the working tree
    my $dir = sync_to_ps($ps);
    # read the new log entry:
    my @commitlog = safe_pipe_capture($TLA,'cat-log','-d',$dir,$ps->{id});
    die "Error in cat-log: $!" if $?;
    chomp @commitlog;

    # grab variables we want from the log, new fields get added to $ps:
    # (author, date, email, summary, message body ...)
    parselog($ps, \@commitlog);

    if ($ps->{id} =~ /--base-0$/ && $ps->{id} ne $psets[0]{id}) {
Junio C Hamano's avatar
Junio C Hamano committed
365
        # this should work when importing continuations
366
        if ($ps->{tag} && (my $branchpoint = eval { ptag($ps->{tag}) })) {
Junio C Hamano's avatar
Junio C Hamano committed
367

368
            # find where we are supposed to branch from
369 370 371 372 373 374 375 376 377 378 379 380 381 382
	    if (! -e "$git_dir/refs/heads/$ps->{branch}") {
		system('git-branch',$ps->{branch},$branchpoint) == 0 or die "$! $?\n";

		# We trust Arch with the fact that this is just a tag,
		# and it does not affect the state of the tree, so
		# we just tag and move on.  If the user really wants us
		# to consolidate more branches into one, don't tag because
		# the tag name would be already taken.
		tag($ps->{id}, $branchpoint);
		ptag($ps->{id}, $branchpoint);
		print " * Tagged $ps->{id} at $branchpoint\n";
	    }
	    system('git-checkout','-f',$ps->{branch}) == 0 or die "$! $?\n";

383 384 385 386 387 388 389 390 391
            # remove any old stuff that got leftover:
            my $rm = safe_pipe_capture('git-ls-files','--others','-z');
            rmtree(split(/\0/,$rm)) if $rm;
            return 0;
        } else {
            warn "Tagging from unknown id unsupported\n" if $ps->{tag};
        }
        # allow multiple bases/imports here since Arch supports cherry-picks
        # from unrelated trees
Junio C Hamano's avatar
Junio C Hamano committed
392 393
    }

394
    # update the index with all the changes we got
395 396
    system('git-diff-files --name-only -z | '.
            'git-update-index --remove -z --stdin') == 0 or die "$! $?\n";
397 398 399 400
    system('git-ls-files --others -z | '.
            'git-update-index --add -z --stdin') == 0 or die "$! $?\n";
    return 1;
}
401

402 403 404 405
# the native changeset processing strategy.  This is very fast, but
# does not handle permissions or any renames involving directories
sub process_patchset_fast {
    my $ps = shift;
Junio C Hamano's avatar
Junio C Hamano committed
406
    #
407 408
    # create the branch if needed
    #
409 410
    if ($ps->{type} eq 'i' && !$import) {
        die "Should not have more than one 'Initial import' per GIT import: $ps->{id}";
411 412
    }

413
    unless ($import) { # skip for import
414
        if ( -e "$git_dir/refs/heads/$ps->{branch}") {
415
            # we know about this branch
416
            system('git-checkout',$ps->{branch});
417 418 419 420
        } else {
            # new branch! we need to verify a few things
            die "Branch on a non-tag!" unless $ps->{type} eq 't';
            my $branchpoint = ptag($ps->{tag});
Junio C Hamano's avatar
Junio C Hamano committed
421
            die "Tagging from unknown id unsupported: $ps->{tag}"
422
                unless $branchpoint;
Junio C Hamano's avatar
Junio C Hamano committed
423

424
            # find where we are supposed to branch from
425 426 427 428 429 430 431 432 433 434 435 436 437
	    if (! -e "$git_dir/refs/heads/$ps->{branch}") {
		system('git-branch',$ps->{branch},$branchpoint) == 0 or die "$! $?\n";

		# We trust Arch with the fact that this is just a tag,
		# and it does not affect the state of the tree, so
		# we just tag and move on.  If the user really wants us
		# to consolidate more branches into one, don't tag because
		# the tag name would be already taken.
		tag($ps->{id}, $branchpoint);
		ptag($ps->{id}, $branchpoint);
		print " * Tagged $ps->{id} at $branchpoint\n";
            }
            system('git-checkout',$ps->{branch}) == 0 or die "$! $?\n";
438
            return 0;
Junio C Hamano's avatar
Junio C Hamano committed
439
        }
440
        die $! if $?;
Junio C Hamano's avatar
Junio C Hamano committed
441
    }
442 443 444

    #
    # Apply the import/changeset/merge into the working tree
Junio C Hamano's avatar
Junio C Hamano committed
445
    #
446 447
    if ($ps->{type} eq 'i' || $ps->{type} eq 't') {
        apply_import($ps) or die $!;
448
        $stats{import_or_tag}++;
449
        $import=0;
450 451
    } elsif ($ps->{type} eq 's') {
        apply_cset($ps);
452
        $stats{simple_changeset}++;
453 454 455 456 457 458
    }

    #
    # prepare update git's index, based on what arch knows
    # about the pset, resolve parents, etc
    #
Junio C Hamano's avatar
Junio C Hamano committed
459 460

    my @commitlog = safe_pipe_capture($TLA,'cat-archive-log',$ps->{id});
461
    die "Error in cat-archive-log: $!" if $?;
Junio C Hamano's avatar
Junio C Hamano committed
462

463
    parselog($ps,\@commitlog);
464 465 466

    # imports don't give us good info
    # on added files. Shame on them
467 468 469
    if ($ps->{type} eq 'i' || $ps->{type} eq 't') {
        system('git-ls-files --deleted -z | '.
                'git-update-index --remove -z --stdin') == 0 or die "$! $?\n";
470 471
        system('git-ls-files --others -z | '.
                'git-update-index --add -z --stdin') == 0 or die "$! $?\n";
472 473
    }

474
    # TODO: handle removed_directories and renamed_directories:
475

476 477
    if (my $del = $ps->{removed_files}) {
        unlink @$del;
478 479
        while (@$del) {
            my @slice = splice(@$del, 0, 100);
480 481
            system('git-update-index','--remove','--',@slice) == 0 or
                            die "Error in git-update-index --remove: $! $?\n";
482 483
        }
    }
484 485

    if (my $ren = $ps->{renamed_files}) {                # renamed
486 487 488
        if (@$ren % 2) {
            die "Odd number of entries in rename!?";
        }
Junio C Hamano's avatar
Junio C Hamano committed
489

490
        while (@$ren) {
491
            my $from = shift @$ren;
Junio C Hamano's avatar
Junio C Hamano committed
492
            my $to   = shift @$ren;
493 494 495 496

            unless (-d dirname($to)) {
                mkpath(dirname($to)); # will die on err
            }
497
            # print "moving $from $to";
498 499 500 501 502
            rename($from, $to) or die "Error renaming '$from' '$to': $!\n";
            system('git-update-index','--remove','--',$from) == 0 or
                            die "Error in git-update-index --remove: $! $?\n";
            system('git-update-index','--add','--',$to) == 0 or
                            die "Error in git-update-index --add: $! $?\n";
503 504
        }
    }
505

506 507 508 509 510 511 512 513
    if (my $add = $ps->{new_files}) {
        while (@$add) {
            my @slice = splice(@$add, 0, 100);
            system('git-update-index','--add','--',@slice) == 0 or
                            die "Error in git-update-index --add: $! $?\n";
        }
    }

514
    if (my $mod = $ps->{modified_files}) {
515 516
        while (@$mod) {
            my @slice = splice(@$mod, 0, 100);
517 518
            system('git-update-index','--',@slice) == 0 or
                            die "Error in git-update-index: $! $?\n";
519 520
        }
    }
521 522 523 524 525 526 527 528 529 530 531 532
    return 1; # we successfully applied the changeset
}

if ($opt_f) {
    print "Will import patchsets using the fast strategy\n",
            "Renamed directories and permission changes will be missed\n";
    *process_patchset = *process_patchset_fast;
} else {
    print "Using the default (accurate) import strategy.\n",
            "Things may be a bit slow\n";
    *process_patchset = *process_patchset_accurate;
}
Junio C Hamano's avatar
Junio C Hamano committed
533

534 535 536 537 538
foreach my $ps (@psets) {
    # process patchsets
    $ps->{branch} = git_branchname($ps->{id});

    #
Junio C Hamano's avatar
Junio C Hamano committed
539 540
    # ensure we have a clean state
    #
541 542 543 544 545
    if (my $dirty = `git-diff-files`) {
        die "Unclean tree when about to process $ps->{id} " .
            " - did we fail to commit cleanly before?\n$dirty";
    }
    die $! if $?;
Junio C Hamano's avatar
Junio C Hamano committed
546

547 548 549 550 551
    #
    # skip commits already in repo
    #
    if (ptag($ps->{id})) {
      $opt_v && print " * Skipping already imported: $ps->{id}\n";
552
      next;
553 554 555 556 557 558
    }

    print " * Starting to work on $ps->{id}\n";

    process_patchset($ps) or next;

Junio C Hamano's avatar
Junio C Hamano committed
559
    # warn "errors when running git-update-index! $!";
560
    my $tree = `git-write-tree`;
561 562
    die "cannot write tree $!" if $?;
    chomp $tree;
Junio C Hamano's avatar
Junio C Hamano committed
563

564 565 566 567
    #
    # Who's your daddy?
    #
    my @par;
568
    if ( -e "$git_dir/refs/heads/$ps->{branch}") {
569
        if (open HEAD, "<","$git_dir/refs/heads/$ps->{branch}") {
570 571 572 573
            my $p = <HEAD>;
            close HEAD;
            chomp $p;
            push @par, '-p', $p;
Junio C Hamano's avatar
Junio C Hamano committed
574
        } else {
575 576 577 578 579
            if ($ps->{type} eq 's') {
                warn "Could not find the right head for the branch $ps->{branch}";
            }
        }
    }
Junio C Hamano's avatar
Junio C Hamano committed
580

581 582 583
    if ($ps->{merges}) {
        push @par, find_parents($ps);
    }
584

Junio C Hamano's avatar
Junio C Hamano committed
585
    #
586 587 588 589 590 591 592 593 594 595
    # Commit, tag and clean state
    #
    $ENV{TZ}                  = 'GMT';
    $ENV{GIT_AUTHOR_NAME}     = $ps->{author};
    $ENV{GIT_AUTHOR_EMAIL}    = $ps->{email};
    $ENV{GIT_AUTHOR_DATE}     = $ps->{date};
    $ENV{GIT_COMMITTER_NAME}  = $ps->{author};
    $ENV{GIT_COMMITTER_EMAIL} = $ps->{email};
    $ENV{GIT_COMMITTER_DATE}  = $ps->{date};

Junio C Hamano's avatar
Junio C Hamano committed
596
    my $pid = open2(*READER, *WRITER,'git-commit-tree',$tree,@par)
597
        or die $!;
598
    print WRITER $ps->{summary},"\n\n";
599 600 601 602 603

    # only print message if it's not empty, to avoid a spurious blank line;
    # also append an extra newline, so there's a blank line before the
    # following "git-archimport-id:" line.
    print WRITER $ps->{message},"\n\n" if ($ps->{message} ne "");
Junio C Hamano's avatar
Junio C Hamano committed
604

605 606
    # make it easy to backtrack and figure out which Arch revision this was:
    print WRITER 'git-archimport-id: ',$ps->{id},"\n";
Junio C Hamano's avatar
Junio C Hamano committed
607

608 609 610 611 612 613 614 615 616 617 618
    close WRITER;
    my $commitid = <READER>;    # read
    chomp $commitid;
    close READER;
    waitpid $pid,0;             # close;

    if (length $commitid != 40) {
        die "Something went wrong with the commit! $! $commitid";
    }
    #
    # Update the branch
Junio C Hamano's avatar
Junio C Hamano committed
619
    #
620
    open  HEAD, ">","$git_dir/refs/heads/$ps->{branch}";
621 622
    print HEAD $commitid;
    close HEAD;
623
    system('git-update-ref', 'HEAD', "$ps->{branch}");
624 625 626 627 628 629 630 631 632

    # tag accordingly
    ptag($ps->{id}, $commitid); # private tag
    if ($opt_T || $ps->{type} eq 't' || $ps->{type} eq 'i') {
        tag($ps->{id}, $commitid);
    }
    print " * Committed $ps->{id}\n";
    print "   + tree   $tree\n";
    print "   + commit $commitid\n";
633
    $opt_v && print "   + commit date is  $ps->{date} \n";
634
    $opt_v && print "   + parents:  ",join(' ',@par),"\n";
635 636 637 638 639 640 641 642 643 644 645 646 647
}

if ($opt_v) {
    foreach (sort keys %stats) {
        print" $_: $stats{$_}\n";
    }
}
exit 0;

# used by the accurate strategy:
sub sync_to_ps {
    my $ps = shift;
    my $tree_dir = $tmp.'/'.tree_dirname($ps->{id});
Junio C Hamano's avatar
Junio C Hamano committed
648

649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681
    $opt_v && print "sync_to_ps($ps->{id}) method: ";

    if (-d $tree_dir) {
        if ($ps->{type} eq 't') {
	    $opt_v && print "get (tag)\n";
            # looks like a tag-only or (worse,) a mixed tags/changeset branch,
            # can't rely on replay to work correctly on these
            rmtree($tree_dir);
            safe_pipe_capture($TLA,'get','--no-pristine',$ps->{id},$tree_dir);
            $stats{get_tag}++;
        } else {
                my $tree_id = arch_tree_id($tree_dir);
                if ($ps->{parent_id} && ($ps->{parent_id} eq $tree_id)) {
                    # the common case (hopefully)
		    $opt_v && print "replay\n";
                    safe_pipe_capture($TLA,'replay','-d',$tree_dir,$ps->{id});
                    $stats{replay}++;
                } else {
                    # getting one tree is usually faster than getting two trees
                    # and applying the delta ...
                    rmtree($tree_dir);
		    $opt_v && print "apply-delta\n";
                    safe_pipe_capture($TLA,'get','--no-pristine',
                                        $ps->{id},$tree_dir);
                    $stats{get_delta}++;
                }
        }
    } else {
        # new branch work
        $opt_v && print "get (new tree)\n";
        safe_pipe_capture($TLA,'get','--no-pristine',$ps->{id},$tree_dir);
        $stats{get_new}++;
    }
Junio C Hamano's avatar
Junio C Hamano committed
682

683 684 685 686 687 688 689
    # added -I flag to rsync since we're going to fast! AIEEEEE!!!!
    system('rsync','-aI','--delete','--exclude',$git_dir,
#               '--exclude','.arch-inventory',
                '--exclude','.arch-ids','--exclude','{arch}',
                '--exclude','+*','--exclude',',*',
                "$tree_dir/",'./') == 0 or die "Cannot rsync $tree_dir: $! $?";
    return $tree_dir;
690 691 692 693
}

sub apply_import {
    my $ps = shift;
694
    my $bname = git_branchname($ps->{id});
695

696
    mkpath($tmp);
697

698
    safe_pipe_capture($TLA,'get','-s','--no-pristine',$ps->{id},"$tmp/import");
Junio C Hamano's avatar
Junio C Hamano committed
699
    die "Cannot get import: $!" if $?;
700 701 702
    system('rsync','-aI','--delete', '--exclude',$git_dir,
		'--exclude','.arch-ids','--exclude','{arch}',
		"$tmp/import/", './');
703
    die "Cannot rsync import:$!" if $?;
Junio C Hamano's avatar
Junio C Hamano committed
704

705
    rmtree("$tmp/import");
706
    die "Cannot remove tempdir: $!" if $?;
Junio C Hamano's avatar
Junio C Hamano committed
707

708 709 710 711 712 713 714

    return 1;
}

sub apply_cset {
    my $ps = shift;

715
    mkpath($tmp);
716 717

    # get the changeset
718
    safe_pipe_capture($TLA,'get-changeset',$ps->{id},"$tmp/changeset");
719
    die "Cannot get changeset: $!" if $?;
Junio C Hamano's avatar
Junio C Hamano committed
720

721 722 723 724
    # apply patches
    if (`find $tmp/changeset/patches -type f -name '*.patch'`) {
        # this can be sped up considerably by doing
        #    (find | xargs cat) | patch
725
        # but that can get mucked up by patches
Junio C Hamano's avatar
Junio C Hamano committed
726
        # with missing trailing newlines or the standard
727 728 729 730 731 732 733 734 735 736 737 738 739 740 741
        # 'missing newline' flag in the patch - possibly
        # produced with an old/buggy diff.
        # slow and safe, we invoke patch once per patchfile
        `find $tmp/changeset/patches -type f -name '*.patch' -print0 | grep -zv '{arch}' | xargs -iFILE -0 --no-run-if-empty patch -p1 --forward -iFILE`;
        die "Problem applying patches! $!" if $?;
    }

    # apply changed binary files
    if (my @modified = `find $tmp/changeset/patches -type f -name '*.modified'`) {
        foreach my $mod (@modified) {
            chomp $mod;
            my $orig = $mod;
            $orig =~ s/\.modified$//; # lazy
            $orig =~ s!^\Q$tmp\E/changeset/patches/!!;
            #print "rsync -p '$mod' '$orig'";
742
            system('rsync','-p',$mod,"./$orig");
743 744 745 746 747
            die "Problem applying binary changes! $!" if $?;
        }
    }

    # bring in new files
748
    system('rsync','-aI','--exclude',$git_dir,
Junio C Hamano's avatar
Junio C Hamano committed
749
		'--exclude','.arch-ids',
750 751
		'--exclude', '{arch}',
		"$tmp/changeset/new-files-archive/",'./');
752 753 754

    # deleted files are hinted from the commitlog processing

755
    rmtree("$tmp/changeset");
756 757 758 759
}


# =for reference
760 761 762
# notes: *-files/-directories keys cannot have spaces, they're always
# pika-escaped.  Everything after the first newline
# A log entry looks like:
763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779
# Revision: moodle-org--moodle--1.3.3--patch-15
# Archive: arch-eduforge@catalyst.net.nz--2004
# Creator: Penny Leach <penny@catalyst.net.nz>
# Date: Wed May 25 14:15:34 NZST 2005
# Standard-date: 2005-05-25 02:15:34 GMT
# New-files: lang/de/.arch-ids/block_glossary_random.php.id
#     lang/de/.arch-ids/block_html.php.id
# New-directories: lang/de/help/questionnaire
#     lang/de/help/questionnaire/.arch-ids
# Renamed-files: .arch-ids/db_sears.sql.id db/.arch-ids/db_sears.sql.id
#    db_sears.sql db/db_sears.sql
# Removed-files: lang/be/docs/.arch-ids/release.html.id
#     lang/be/docs/.arch-ids/releaseold.html.id
# Modified-files: admin/cron.php admin/delete.php
#     admin/editor.html backup/lib.php backup/restore.php
# New-patches: arch-eduforge@catalyst.net.nz--2004/moodle-org--moodle--1.3.3--patch-15
# Summary: Updating to latest from MOODLE_14_STABLE (1.4.5+)
780
#   summary can be multiline with a leading space just like the above fields
781 782 783 784
# Keywords:
#
# Updating yadda tadda tadda madda
sub parselog {
785 786 787 788 789 790 791 792 793 794 795 796
    my ($ps, $log) = @_;
    my $key = undef;

    # headers we want that contain filenames:
    my %want_headers = (
        new_files => 1,
        modified_files => 1,
        renamed_files => 1,
        renamed_directories => 1,
        removed_files => 1,
        removed_directories => 1,
    );
Junio C Hamano's avatar
Junio C Hamano committed
797

798 799 800 801 802 803
    chomp (@$log);
    while ($_ = shift @$log) {
        if (/^Continuation-of:\s*(.*)/) {
            $ps->{tag} = $1;
            $key = undef;
        } elsif (/^Summary:\s*(.*)$/ ) {
804 805
            # summary can be multiline as long as it has a leading space.
	    # we squeeze it onto a single line, though.
806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834
            $ps->{summary} = [ $1 ];
            $key = 'summary';
        } elsif (/^Creator: (.*)\s*<([^\>]+)>/) {
            $ps->{author} = $1;
            $ps->{email} = $2;
            $key = undef;
        # any *-files or *-directories can be read here:
        } elsif (/^([A-Z][a-z\-]+):\s*(.*)$/) {
            my $val = $2;
            $key = lc $1;
            $key =~ tr/-/_/; # too lazy to quote :P
            if ($want_headers{$key}) {
                push @{$ps->{$key}}, split(/\s+/, $val);
            } else {
                $key = undef;
            }
        } elsif (/^$/) {
            last; # remainder of @$log that didn't get shifted off is message
        } elsif ($key) {
            if (/^\s+(.*)$/) {
                if ($key eq 'summary') {
                    push @{$ps->{$key}}, $1;
                } else { # files/directories:
                    push @{$ps->{$key}}, split(/\s+/, $1);
                }
            } else {
                $key = undef;
            }
        }
835
    }
Junio C Hamano's avatar
Junio C Hamano committed
836

837 838 839 840 841 842 843 844 845 846 847 848
    # drop leading empty lines from the log message
    while (@$log && $log->[0] eq '') {
	shift @$log;
    }
    if (exists $ps->{summary} && @{$ps->{summary}}) {
	$ps->{summary} = join(' ', @{$ps->{summary}});
    }
    elsif (@$log == 0) {
	$ps->{summary} = 'empty commit message';
    } else {
	$ps->{summary} = $log->[0] . '...';
    }
849
    $ps->{message} = join("\n",@$log);
Junio C Hamano's avatar
Junio C Hamano committed
850

851 852 853
    # skip Arch control files, unescape pika-escaped files
    foreach my $k (keys %want_headers) {
        next unless (defined $ps->{$k});
854
        my @tmp = ();
855 856 857 858 859 860
        foreach my $t (@{$ps->{$k}}) {
           next unless length ($t);
           next if $t =~ m!\{arch\}/!;
           next if $t =~ m!\.arch-ids/!;
           # should we skip this?
           next if $t =~ m!\.arch-inventory$!;
861 862
           # tla cat-archive-log will give us filenames with spaces as file\(sp)name - why?
           # we can assume that any filename with \ indicates some pika escaping that we want to get rid of.
863
           if ($t =~ /\\/ ){
864
               $t = (safe_pipe_capture($TLA,'escape','--unescaped',$t))[0];
865
           }
866
           push @tmp, $t;
867
        }
868
        $ps->{$k} = \@tmp;
869 870 871 872 873 874
    }
}

# write/read a tag
sub tag {
    my ($tag, $commit) = @_;
Junio C Hamano's avatar
Junio C Hamano committed
875

876 877 878
    if ($opt_o) {
        $tag =~ s|/|--|g;
    } else {
879 880 881
	my $patchname = $tag;
	$patchname =~ s/.*--//;
        $tag = git_branchname ($tag) . '--' . $patchname;
882
    }
Junio C Hamano's avatar
Junio C Hamano committed
883

884
    if ($commit) {
885
        open(C,">","$git_dir/refs/tags/$tag")
886 887 888 889 890
            or die "Cannot create tag $tag: $!\n";
        print C "$commit\n"
            or die "Cannot write tag $tag: $!\n";
        close(C)
            or die "Cannot write tag $tag: $!\n";
891
        print " * Created tag '$tag' on '$commit'\n" if $opt_v;
892
    } else {                    # read
893
        open(C,"<","$git_dir/refs/tags/$tag")
894 895 896 897 898 899 900 901 902 903 904 905 906 907
            or die "Cannot read tag $tag: $!\n";
        $commit = <C>;
        chomp $commit;
        die "Error reading tag $tag: $!\n" unless length $commit == 40;
        close(C)
            or die "Cannot read tag $tag: $!\n";
        return $commit;
    }
}

# write/read a private tag
# reads fail softly if the tag isn't there
sub ptag {
    my ($tag, $commit) = @_;
908 909

    # don't use subdirs for tags yet, it could screw up other porcelains
Junio C Hamano's avatar
Junio C Hamano committed
910 911
    $tag =~ s|/|,|g;

912 913 914
    my $tag_file = "$ptag_dir/$tag";
    my $tag_branch_dir = dirname($tag_file);
    mkpath($tag_branch_dir) unless (-d $tag_branch_dir);
915 916

    if ($commit) {              # write
917
        open(C,">",$tag_file)
918 919 920 921 922
            or die "Cannot create tag $tag: $!\n";
        print C "$commit\n"
            or die "Cannot write tag $tag: $!\n";
        close(C)
            or die "Cannot write tag $tag: $!\n";
Junio C Hamano's avatar
Junio C Hamano committed
923
	$rptags{$commit} = $tag
924
	    unless $tag =~ m/--base-0$/;
925 926
    } else {                    # read
        # if the tag isn't there, return 0
927
        unless ( -s $tag_file) {
928 929
            return 0;
        }
930
        open(C,"<",$tag_file)
931 932 933 934 935 936
            or die "Cannot read tag $tag: $!\n";
        $commit = <C>;
        chomp $commit;
        die "Error reading tag $tag: $!\n" unless length $commit == 40;
        close(C)
            or die "Cannot read tag $tag: $!\n";
937 938 939
	unless (defined $rptags{$commit}) {
	    $rptags{$commit} = $tag;
	}
940 941 942
        return $commit;
    }
}
943 944 945 946 947 948

sub find_parents {
    #
    # Identify what branches are merging into me
    # and whether we are fully merged
    # git-merge-base <headsha> <headsha> should tell
Junio C Hamano's avatar
Junio C Hamano committed
949
    # me what the base of the merge should be
950 951 952 953 954 955 956 957 958 959 960 961 962
    #
    my $ps = shift;

    my %branches; # holds an arrayref per branch
                  # the arrayref contains a list of
                  # merged patches between the base
                  # of the merge and the current head

    my @parents;  # parents found for this commit

    # simple loop to split the merges
    # per branch
    foreach my $merge (@{$ps->{merges}}) {
963
	my $branch = git_branchname($merge);
964 965 966 967 968 969 970
	unless (defined $branches{$branch} ){
	    $branches{$branch} = [];
	}
	push @{$branches{$branch}}, $merge;
    }

    #
Junio C Hamano's avatar
Junio C Hamano committed
971
    # foreach branch find a merge base and walk it to the
972 973 974 975 976 977
    # head where we are, collecting the merged patchsets that
    # Arch has recorded. Keep that in @have
    # Compare that with the commits on the other branch
    # between merge-base and the tip of the branch (@need)
    # and see if we have a series of consecutive patches
    # starting from the merge base. The tip of the series
Junio C Hamano's avatar
Junio C Hamano committed
978
    # of consecutive patches merged is our new parent for
979 980 981
    # that branch.
    #
    foreach my $branch (keys %branches) {
982 983 984 985

	# check that we actually know about the branch
	next unless -e "$git_dir/refs/heads/$branch";

986
	my $mergebase = safe_pipe_capture(qw(git-merge-base), $branch, $ps->{branch});
Junio C Hamano's avatar
Junio C Hamano committed
987 988 989 990 991 992 993
	if ($?) {
	    # Don't die here, Arch supports one-way cherry-picking
	    # between branches with no common base (or any relationship
	    # at all beforehand)
	    warn "Cannot find merge base for $branch and $ps->{branch}";
	    next;
	}
994 995 996 997
	chomp $mergebase;

	# now walk up to the mergepoint collecting what patches we have
	my $branchtip = git_rev_parse($ps->{branch});
998
	my @ancestors = `git-rev-list --topo-order $branchtip ^$mergebase`;
999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017
	my %have; # collected merges this branch has
	foreach my $merge (@{$ps->{merges}}) {
	    $have{$merge} = 1;
	}
	my %ancestorshave;
	foreach my $par (@ancestors) {
	    $par = commitid2pset($par);
	    if (defined $par->{merges}) {
		foreach my $merge (@{$par->{merges}}) {
		    $ancestorshave{$merge}=1;
		}
	    }
	}
	# print "++++ Merges in $ps->{id} are....\n";
	# my @have = sort keys %have;	print Dumper(\@have);

	# merge what we have with what ancestors have
	%have = (%have, %ancestorshave);

Junio C Hamano's avatar
Junio C Hamano committed
1018
	# see what the remote branch has - these are the merges we
1019 1020
	# will want to have in a consecutive series from the mergebase
	my $otherbranchtip = git_rev_parse($branch);
1021
	my @needraw = `git-rev-list --topo-order $otherbranchtip ^$mergebase`;
1022 1023 1024 1025
	my @need;
	foreach my $needps (@needraw) { 	# get the psets
	    $needps = commitid2pset($needps);
	    # git-rev-list will also
Junio C Hamano's avatar
Junio C Hamano committed
1026
	    # list commits merged in via earlier
1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061
	    # merges. we are only interested in commits
	    # from the branch we're looking at
	    if ($branch eq $needps->{branch}) {
		push @need, $needps->{id};
	    }
	}

	# print "++++ Merges from $branch we want are....\n";
	# print Dumper(\@need);

	my $newparent;
	while (my $needed_commit = pop @need) {
	    if ($have{$needed_commit}) {
		$newparent = $needed_commit;
	    } else {
		last; # break out of the while
	    }
	}
	if ($newparent) {
	    push @parents, $newparent;
	}


    } # end foreach branch

    # prune redundant parents
    my %parents;
    foreach my $p (@parents) {
	$parents{$p} = 1;
    }
    foreach my $p (@parents) {
	next unless exists $psets{$p}{merges};
	next unless ref    $psets{$p}{merges};
	my @merges = @{$psets{$p}{merges}};
	foreach my $merge (@merges) {
Junio C Hamano's avatar
Junio C Hamano committed
1062
	    if ($parents{$merge}) {
1063 1064 1065 1066
		delete $parents{$merge};
	    }
	}
    }
1067

1068 1069 1070 1071
    @parents = ();
    foreach (keys %parents) {
        push @parents, '-p', ptag($_);
    }
1072 1073 1074 1075 1076
    return @parents;
}

sub git_rev_parse {
    my $name = shift;
1077
    my $val  = safe_pipe_capture(qw(git-rev-parse), $name);
1078 1079 1080 1081 1082 1083 1084 1085 1086
    die "Error: git-rev-parse $name" if $?;
    chomp $val;
    return $val;
}

# resolve a SHA1 to a known patchset
sub commitid2pset {
    my $commitid = shift;
    chomp $commitid;
Junio C Hamano's avatar
Junio C Hamano committed
1087
    my $name = $rptags{$commitid}
1088
	|| die "Cannot find reverse tag mapping for $commitid";
1089
    $name =~ s|,|/|;
Junio C Hamano's avatar
Junio C Hamano committed
1090
    my $ps   = $psets{$name}
1091 1092 1093
	|| (print Dumper(sort keys %psets)) && die "Cannot find patchset for $name";
    return $ps;
}
1094

1095

1096
# an alternative to `command` that allows input to be passed as an array
1097 1098 1099 1100 1101 1102 1103
# to work around shell problems with weird characters in arguments
sub safe_pipe_capture {
    my @output;
    if (my $pid = open my $child, '-|') {
        @output = (<$child>);
        close $child or die join(' ',@_).": $! $?";
    } else {
1104
	exec(@_) or die "$! $?"; # exec() can fail the executable can't be found
1105 1106 1107 1108
    }
    return wantarray ? @output : join('',@output);
}

1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119
# `tla logs -rf -d <dir> | head -n1` or `baz tree-id <dir>`
sub arch_tree_id {
    my $dir = shift;
    chomp( my $ret = (safe_pipe_capture($TLA,'logs','-rf','-d',$dir))[0] );
    return $ret;
}

sub archive_reachable {
    my $archive = shift;
    return 1 if $reachable{$archive};
    return 0 if $unreachable{$archive};
Junio C Hamano's avatar
Junio C Hamano committed
1120

1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134
    if (system "$TLA whereis-archive $archive >/dev/null") {
        if ($opt_a && (system($TLA,'register-archive',
                      "http://mirrors.sourcecontrol.net/$archive") == 0)) {
            $reachable{$archive} = 1;
            return 1;
        }
        print STDERR "Archive is unreachable: $archive\n";
        $unreachable{$archive} = 1;
        return 0;
    } else {
        $reachable{$archive} = 1;
        return 1;
    }
}