aboutsummaryrefslogtreecommitdiff
path: root/sbosrcarch
blob: ce269f8d6d78e9b92a057b03afc30998a3d456d8 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
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
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
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
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
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
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
#!/usr/bin/perl -w

=pod

=head1 NAME

sbosrcarch - Create and maintain an archive of source code for SBo

=head1 SYNOPSIS

sbosrcarch <create|update|trim|purge|check>

sbosrcarch add [-f] <category/prgnam> [<file> ...]

sbosrcarch rm <category/prgnam>

=head1 DESCRIPTION

sbosrcarch creates and maintains an archive of source code files linked
to by DOWNLOAD= and DOWNLOAD_x86_64= URLs in SlackBuilds.org .info files.

Since a full archive would be pretty large (45GB or so), sbosrcarch
allows limiting the size of the archive (but only indirectly, by
limiting the max file size it will download). This means we won't have
a full archive of every source tarball, but even a partial mirror is
still useful.

Rough guideline for choosing filesize:

 Max filesize | Approx. total archive size | Coverage
        1.0M  |                    803.1M  |  68%
        2.0M  |                      1.4G  |  77%
        5.0M  |                      2.7G  |  85%
       10.0M  |                      4.3G  |  90%
       20.0M  |                      6.6G  |  93%
       35.0M  |                      8.9G  |  95%
       50.0M  |                     11.6G  |  96%
      100.0M  |                     16.6G  |  98%
   unlimited  |                     43.0G  | 100%

"Coverage" is the percentage of all the URLs in all the .info files
that will be kept in this archive. Notice that about 75% of the storage
space is eaten up by 2% of the files, in the unlimited case. These
large files are mostly games, if that influences your decision any.

=head1 OPTIONS

=over

=item create

Create archive. Used for initial archive creation, and for downloading new
files to an existing archive when the size limit ($maxfilemegs) is increased.

Should be run interactively, from a login shell. Takes a long time to
run and uses a lot of bandwidth. Log output goes to stdout.

If the archive already exists, existing files will be kept instead of
being re-downloaded (provided of course their md5sums are correct).

=item update

Update archive, by checking the SBo git log and parsing any .info files that
have changed since the last create or update.

Should be run daily or weekly as a cron job.

If there are are few or no changed download URLs, update should run
quickly and not eat many resources. For each new URL, the file is
downloaded and added to the archive, but the old file is *not* deleted
(use 'sbosrcarch purge' to do that).

=item purge

Purge files from the archive that are no longer referenced by any
.info file.  Should be run monthly or quarterly as a cron job. This is
more resource-intensive than an update, as it must read and parse every
.info file in the SBo repository.

=item trim

Gets rid of files that are in the archive, but are larger than the size
limit. Should be run manually after lowering $maxfilemegs; there's no
reason to run it any other time.

=item check

TODO: this is not yet implemented.

Checks the integrity and coverage of the archive: Will report at least these conditions:

 - dangling symlinks
 - invalid md5sums
 - files present in only one of by-name or by-md5 but not the other
 - extraneous files in the tree
 - generates a detailed status report, giving the total size,
   coverage, and a list of slackbuilds not covered.

Will not modify the archive in any way, but might recommend fixes.

=item add [-f] <category/prgnam> [<file> ...]

Manually add a single (possibly already downloaded) file to the archive.

Use -f to skip the size limit checking, so your archive can include a
few large files (perhaps because they're for builds you maintain).

Files added this way will still be deleted by 'sbosrcarch trim', if
they're larger than the limit.

This is intended to let the mirror operator keep a few large files, over
the maxfilemegs limit, or save bandwidth by using already-downloaded
copies (e.g. of stuff that was built recently).

If files are given after the category/prgnam argument, they will be
used instead of downloading the URLs in the .info files (provided their
md5sums match the .info file). Size limits are not checked for files
added this way.

=item rm <category/prgnam>

Manually remove files from the archive. All the files referenced by the
.info file for <category>/<prgnam> will be removed.

...but the next update will re-add anything you remove, if it's less than
the size limit. Mostly this is useful for manually-added files that are
over the limit.

=back

=head1 FILES

B<.sbosrcarch.conf> (or B<sbosrcarch.conf>) is the config file for
sbosrcarch.  It's searched for in the current directory, the user's
home directory, /etc/sbosrcarch, and /etc (in order). See the section
B<CONFIG FILE> for details.

The archive created by sbosrcarch consists of two top-level directories
called B<by-name> and B<by-md5>. All files are present in both hierarchies
(as hardlinked copies, to save space).

B<by-name> is organized by the familiar category and PRGNAM, like SBo
itself. Example:

  by-name/network/ifstatus/ifstatus-v1.1.0.tar.gz

This makes it easy for humans to browse the archive and find the source
file they're looking for.

B<by-md5> contains the same files, but organized in a hierarchy based on
the md5sum of the file, for automated systems to easily find the exact
file needed. The same file as the example above would be found at:

by-md5/f/4/f4d413f880754fd6677290160f8bc5d7/ifstatus-v1.1.0.tar.gz

Notice there are two layers of subdirectory, named after the first two
hex digits in the md5sum.

There is one other directory of files used/maintained by sbosrcarch:
a git clone of SBo's master git branch. This is cloned and updated
automatically as needed, and shouldn't need to be messed with. If you
need a git clone of SBo for some other purpose, create a separate one
to avoid confusing sbosrcarch with your changes and pulls.

=head1 CONFIG FILE

TODO: document the config options here.

For now, see the sample config file sbosrcarch.conf

=head1 SERVER CONFIGURATION

If you're planning to host a public archive, you'll need to make the
$archivedir available via whatever protocols you support (HTTP, FTP,
rsync, etc). This is the directory containing B<by-name> and B<by-md5>.
The git clone directory doesn't need to be served to the public.

TODO: example Apache, proftpd, etc configs for serving up the archive.

=head1 EXAMPLE

TODO: shell script that parses an .info file and tries to d/l the source
from the archive.

=head1 NOTES

sbosrcarch is written in perl, and is intended to work on at least
Slackware 13.0 through 14.1, using only perl modules that ship with the
OS (so no CPAN dependencies). If you want to run it on some other OS,
it might need some extra stuff installed and/or some slight porting
work. If you want to keep a SBo source archive on your non-Slackware
server, it might be best to just rsync someone else's (that they build
using this script).

Note that there's no need to run sbosrcarch as root. In fact, it's
recommended not to. Good choices for a user to run it as:
 - your everyday user you log in as
 - apache
 - nobody

=head1 BUGS/LIMITATIONS

Plenty of these, see FIXME TODO XXX comments in the code. Here are some
that I'm not planning to address any time soon:

No threading. Not likely to change. It would be possible to spawn wget
processes in the background, but I'm not going to complicate it that way.

Anything that checks referer header or otherwise tries to stop automated
downloads, will stop us. This isn't really a bug (sbopkg can't handle them
either).

Length: unspecified isn't handled (we just don't download these). Specifically,
dropbox URLs do this.

$sbogitdir and $archivedir must be located on the same filesystem, as files
are moved around by linking them.

=head1 AUTHOR

B. Watson <yalhcru@gmail.com>

=cut

# use only modules that ship with Slackware
use File::Temp qw/tempfile tempdir/;
use File::Find;
use Digest::MD5;
use Net::FTP;
use POSIX 'getcwd';

sub read_config {
	@configdirs = (
			".",
			$ENV{HOME},
			"/etc/sbosrcarch",
			"/etc",
			);

	for $dir (@configdirs) {
		for $file (qw/.sbosrcarch.conf sbosrcarch.conf/) {
			$_ = "$dir/$file";
			next unless -e $_;
			do $_;
			next if $!;
			die "reading config file $_: $@" if $@;
			$conf_used = $_;
			last;
		}
	}

	if($conf_used) {
		print "read config file: $conf_used\n";
	} else {
		die "can't find .sbosrcarch.conf or sbosrcarch.conf in any of the\n" .
			"following directories, giving up:\n" .
			join ("\n", @configdirs) . "\n";
	}

# required stuff in the conf file:
	die "config file missing \$sbogiturl\n" unless defined $sbogiturl;
	die "config file missing \$sbogitdir\n" unless defined $sbogitdir;
	die "config file missing \$archivedir\n" unless defined $archivedir;

# not required, but warn if it's missing:
	if((not defined $maxfilemegs) || ($maxfilemegs < 0)) {
		warn "config file missing/invalid \$maxfilemegs, defaulting to 10\n";
		$maxfilemegs = 10;
	}

# quietly use defaults if missing:
	$wgetargs = "" unless defined $wgetargs;
	$symlinks = "" unless defined $symlinks;

	if(not defined $wgetrc_contents) {
		$wgetrc_contents = <<EOF;
robots = off
user_agent = Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)
check_certificate = off
content_disposition = off
EOF
	}

	if(not %user_agent_overrides) {
		%user_agent_overrides = (
			qr/(?:sourceforge|sf)\.net/ => 'wget',
		);
	}
}

# url_to_filename, gets the filename part of a URL (after the last slash)
# and un-escapes any %XX sequences.
sub url_to_filename {
	my $u = shift;
	$u =~ s,.*/,,;
	$u =~ s,%([0-9A-F]{2}),chr(hex($1)),ge;
	return $u;
}

# parse a single .info file, return a hashref where keys = URL(s)
# and values are their md5sums.
sub parse_info {
	local $/ = "";
	my $file = shift;
	open my $fh, "<", $file;
	my $got = <$fh>;

	$got =~ s/\\\s*\n//gs; # join \ continuation lines
	$got =~ s/[ \t]+/ /g;  # condense whitespace

	$got =~ /DOWNLOAD(?:_x86_64)?="([^"]+)"/;
	my @urls = split " ", $1;

	$got =~ /MD5SUM(?:_x86_64)?="([^"]+)"/;
	my @md5s = split " ", $1;

	my %ret;
	for(@urls) {
		next if /^un(test|support)ed$/;
		$ret{$_} = shift @md5s;
	}

	return \%ret;
}

# the download_* subs return:
# 0 - file too big (so skip it)
# positive integer - file size
# undef - download error (404, failed DNS, etc).
# FIXME: the above isn't really true, and the calling code doesn't
# check the return values as it should.

sub download_http {
	my $url = shift;
	my $size = wget($url, 1); # HEAD request first

	# $size will be 0 for 'too big' or undef if the HEAD failed.

	if($size) {
		$size = wget($url, 0);
	}
	return $size;
}

sub download_file {
	my $url = shift;
	my $filename = url_to_filename($url);

	my $dlresult;
	if($url =~ /^ftp:/) {
		$dlresult = download_ftp($url);
	} else {
		$dlresult = download_http($url);
	}

	return $dlresult;
}

# see %user_agent_overrides
sub user_agent {
	my $url = shift;

	my $ua = "";
	$url =~ m,^\w+://([^/]*)/,;
	my $site = $1;
	for (keys %user_agent_overrides) {
		$site =~ /$_/ && do {
			$ua = $user_agent_overrides{$_};
		};
	}
	$ua = "--user-agent='$ua'" if $ua;
	return $ua;
}

# return true if limit set and file size > limit.
# return false if no limit set, or file size <= limit.
sub toobig {
	return 0 if $maxfilemegs <= 0; # no limit
	return $_[0] > ($maxfilemegs * 1024 * 1024);
}

# wget_fake_head: What is a fake HEAD request?

# Stoopid github "cloud" bullshit (actually, amazonaws.com) doesn't
# allow HEAD requests, so we fake them by doing a GET, then closing the
# connection as soon as we've got the headers.

# Due to buffering, wget still downloads the first 16K or so of the file,
# which gets discarded when we close its filehandle. We could do better
# than this by implementing the HTTP protocol in terms of IO::Socket::INET
# or such, but I'm not writing & debugging the mess that would turn into.

# This gets called for any URL that doesn't return a Content-Length header
# in its HEAD request (for whatever reason, including because of a 404
# not found). Of course, a GET might not return a length header either :(

sub wget_fake_head {
	my $url = shift;
	my $cmd = "wget --config=$wgetrc " .
		"--quiet -O- --save-headers " .
		user_agent($url) . " " .
		" $wgetargs " .
		"'$url'";

	print "real HEAD failed, trying fake HEAD request: $cmd\n";

	my $size;
	open my $fh, "$cmd|" or return undef;
	while(<$fh>) {
		s/\r//;
		chomp;
		last if /^$/;
		$size = $1 if /^Content-Length:\s+(\d+)/;
	}
	close $fh;

	if($size && toobig($size)) {
		printf "file too large: %0.2fMB\n", $size / (1024 * 1024);
		$skipcount++;
		$size = 0;
	}

	return $size;
}

sub wget {
	my $url = shift;
	my $head = shift; # boolean, 0 = download (GET), 1 = HEAD request only

	my $size;

	# XXX: respect environment's $TMP or such?
	if(not defined $wgetrc) {
		($fh, $wgetrc) = tempfile("wgetrc.XXXXXXXX", DIR => "/tmp", UNLINK => 1);
		print $fh $wgetrc_contents;
		close $fh;
	}

	my $outfile;
	($fh, $outfile) = tempfile("wget.out.XXXXXXXX", DIR => "/tmp", UNLINK => 1);
	close $fh;

	my $cmd = "wget --config=$wgetrc " .
		user_agent($url) . " " .
		($head ? "--spider --tries 1" : "") .
		" $wgetargs " .
		"'$url' " .
		">$outfile 2>&1";

		#" --referer='$url' " . # don't use, it breaks sourceforge

	print "$cmd\n";
	my $retval = system($cmd);

	open $fh, "<$outfile";
	while(<$fh>) {
		print " ! $_" if $retval != 0;

		/^Length:\s*(\d+).*\[(.*?)\]/ && do {
			$size = $1; # $content_type = $2;
			if(toobig($size)) {
				printf "file too large: %0.2fMB\n", $size / (1024 * 1024);
				$skipcount++;
				$size = 0;
			}
		};
	}
	close $fh;
	unlink $outfile;

	# Grr. Some sites refuse HEAD requests, and some allow them but
	# don't return a Content-Length header. So we must resort to more
	# drastic measures.
	# FIXME: don't bother doing this if we got 404 (not found) from the HEAD,
	# or stuff like DNS errors.
	if($head && not(defined($size))) {
		return wget_fake_head($url);
	}

	return $size; # which might be undef!
}

# we could use wget for FTP links too, but doing it this way
# lets us check the filesize and do the download with only one
# FTP session.
sub download_ftp {
	my ($server, $dir, $filename) = ($_[0] =~ m,
		^ftp://   # proto
		([^/]+)   # server (no slashes)
		(/.*?)?   # optional path (always at least the initial slash)
		([^/]+)$  # filename (everything after last slash)
		,x);

	print "using Net::FTP to get $_[0]\n";
	my $size = undef;
	eval {
		my $ftp = Net::FTP->new($server, Debug => 0)
			or die "Can't connect to $server: $@";
		$ftp->login("anonymous",'-anonymous@')
			or die "Can't log in to $server: ", $ftp->message;
		$ftp->cwd($dir)
			or die "Can't chdir($dir) on $server: ", $ftp->message;
		$ftp->binary;
		$size = $ftp->size($filename)
			or die "Can't get $filename size from $server: ", $ftp->message;

		if(toobig($size)) {
			printf "file too large: %0.2fMB\n", $size / (1024 * 1024);
			$skipcount++;
			$size = 0;
		} else {
			$ftp->get($filename)
				or die "Can't download $filename from server: ", $ftp->message;
		}

		$ftp->quit;
	};

	if($@) {
		print "$_[0]: $@";
		$size = 0;
	}

	return $size;
}

sub git_clone {
	system("git clone $sbogiturl $sbogitdir");
}

sub git_pull {
	return !system("git pull");
}

sub md5_dir {
	my $md5 = shift;
	return "$archivedir/by-md5/" .
		substr($md5, 0, 1) .
		"/" .
		substr($md5, 1, 1) .
		"/" .
		$md5 .
		"/";
}

sub name_dir {
	my ($cat, $prg) = @_;
	return "$archivedir/by-name/$cat/$prg/";
}

sub md5sum_file {
	my $filename = shift;
	open my $fh, "<", $filename; # XXX: error check (don't use die)
	binmode($fh);
	my $ret = Digest::MD5->new->addfile($fh)->hexdigest;
	close $fh;
	return $ret;
}

sub already_exists {
	my ($filename, $category, $prgnam, $md5) = @_;

	my $n = name_dir($category, $prgnam) . "/" . $filename;
	my $m = md5_dir($md5) . "/" . $filename;

	return
		-e $n &&
		-e $m &&
		($md5 eq md5sum_file($n)) &&
		($md5 eq md5sum_file($n));
}

# TODO: handle %20 => space (and other URL encodings)
# ...needs to be done elsewhere too, not just here.
sub store_file {
	my ($filename, $category, $prgnam, $md5) = @_;

	#warn "store_file($filename, $category, $prgnam, $md5);\n";

	system("mkdir -p " . md5_dir($md5));
	system("mkdir -p " . name_dir($category, $prgnam));
	link($filename, name_dir($category, $prgnam) . "/" . $filename);
	warn "symlinks not yet supported, using hardlink instead\n" if $symlinks;
	link($filename, md5_dir($md5) . "/" . $filename); # TODO: symlink option
}

# handle_info_file() is used as the 'wanted' sub for File::Find, but
# it's also called from add and update modes, so it doesn't use any of
# the File::Find stuff. Call while cd'ed to $sbogitdir, with $_ set to
# the relative path to the .info file.
sub handle_info_file {
	return unless /\.info$/;

	my $dls = parse_info("$_");

	s,^\./,,; # strip leading ./, if present
	my ($category, $prgnam) = split /\//, $_;
	print "=== $category/$prgnam: ";

	for(keys %$dls) {
		$urlcount++;
		my $url = $_;
		my $md5 = $dls->{$_};
		my $filename = url_to_filename($url);

		if(already_exists($filename, $category, $prgnam, $md5)) {
			print "already in archive, OK\n";
			$archivecount++;
		} else {
			$attemptcount++;
			download_file($url); # TODO: check result!
			if(! -f $filename) {
				$failcount++;
				print "$filename not downloaded\n";
				next;
			}

			if(md5sum_file($filename) ne $md5) {
				$failcount++;
				print "md5sum failed for $url";
				unlink($filename);
				next;
			}

			print "downloaded, OK\n";
			$archivecount++;
			$dlcount++;
			store_file($filename, $category, $prgnam, $md5);
			unlink($filename);
		}
	}
}

sub init_git {
	chdir($sbogitdir) && -d ".git" ||
		die "SBo git dir $sbogitdir not a git checkout, " .
			"do you need to run 'sbosrcarch create?'\n";
}

sub create_mode {
	chdir($sbogitdir) or git_clone;
	chdir($sbogitdir) or die "can't find or create SBo git dir $sbogitdir\n";
	git_clone unless -d ".git";
	git_pull or die "git pull failed, check $sbogitdir\n";

	$skipcount = $attemptcount = $urlcount =
	$archivecount = $dlcount = $failcount = $nowarchived = 0;

	find({wanted => \&handle_info_file, no_chdir => 1}, ".");

	$nowarchived = $dlcount + $archivecount;
	$coverage = sprintf("%.1d", ($nowarchived * 100 / $urlcount));
	print <<EOF;

---
Total URLs: $urlcount
Already archived: $archivecount
Skipped downloads due to size limit: $skipcount
Attempted downloads: $attemptcount
Successful downloads: $dlcount
Failed downloads: $failcount
Now archived: $nowarchived
Coverage: $coverage%
EOF
}

sub update_mode {
	my $oldcommit;

	init_git();

	open my $fh, "git log|" or die "$!";
	my $logline = <$fh>;
	(undef, $oldcommit) = split /\s+/, $logline;
	print "git repo was at commit $oldcommit\n";
	close $fh;

	git_pull();

	open $fh, "git diff --numstat $oldcommit|" or die "$!";
	while(<$fh>) {
		(undef, undef, $_) = split /\s+/;
		next unless /\.info$/;
		handle_info_file();
	}
	exit 0;
}

# purge_mode() does 3 passes.
# 1. get all the filenames from all the info files, build a hash of filenames.
# 2. walk the archive tree with File::Find and rm any file that's in a
#    category/name dir, but not mentioned in the filename hash (also, rm its
#    md5_dir() counterpart).
# 3. do a trim_post() pass to delete any empty dirs and/or dangling symlinks

# FIXME: files from different URLs but with the same filename will not be
# purged when they should, because the comparison is solely filename-based!
sub purge_mode {
	init_git();

	$purgebytes = $purgefiles = 0;

	# pass 1
	%keep_filenames = (); # populated by the find():
	find({wanted => \&purge_pass_1_wanted, no_chdir => 1}, ".");

#	for(keys %keep_filenames) {
#		warn "keep $_\n";
#	}

	# pass 2
	chdir($archivedir) or die "$archivedir: $!\n";
	find({wanted => \&purge_pass_2_wanted, no_chdir => 1}, "by-name");

	# pass 3
	trim_post();

	printf("Purged $purgefiles files, %.1fMB\n", ($purgebytes / (1024 * 1024)));
	exit 0;
}

# helper for purge_mode, populates %keep_filenames
sub purge_pass_1_wanted {
	return unless /\.info$/;
	my $dls = parse_info($_);
	for(keys %$dls) {
		$_ = url_to_filename($_);
		$keep_filenames{$_}++;
	}
}

# helper for purge_mode, removes all files in category/prgnam/
# dirs that aren't listed in %keep_filenames
sub purge_pass_2_wanted {
	s,^\./,,; # remove leading ./
	my (undef, $cat, $name, $file) = split /\//, $_;
	return unless defined $file;
	return if $keep_filenames{$file};

	print "purge $_\n";
	$purgebytes += -s $_;
	$purgefiles++;

	unlink md5_dir(md5sum_file($_)). "$file";
	unlink $_;
}

# helper for trim_mode
sub trim_wanted {
	return unless -f $_;
	my $size = -s _;
	if(toobig($size)) {
		unlink($_);
		$trimcount++;
		$trimbytes += $size;
	}
}

# helper for trim_post
sub trim_post_wanted {
	unlink $_ if -l $_ && ! -e _;
	return unless -d _;
	push @trim_empty_dirs, $_ if !<*>;
}

# pass 2 of trim_mode, also called by purge_mode. removes
# empty directories and dangling symlinks.
sub trim_post {
	chdir($archivedir) or die "$archivedir: $!\n";

	# can't rmdir from within find's wanted sub, or we get
	# lots of 'Can't opendir()' warnings. So collect all the
	# empty dirs in an array during the find, then rmdir them
	# all in one swell foop afterwards.
	@trim_empty_dirs = ();

	# remove dangling symlinks and make a list of empty dirs
	find({wanted => \&trim_post_wanted, no_chdir => 1}, ".");

	rmdir $_ for @trim_empty_dirs; # the aforementioned swell foop
}

# this mode doesn't know/care about the git stuff, it operates purely
# on the archive file tree.
sub trim_mode {
	chdir($archivedir) or die "$archivedir: $!\n";

	$trimcount = $trimbytes = 0;

	# first pass: remove files that are too big
	find({wanted => \&trim_wanted, no_chdir => 1}, ".");

	# 2nd pass
	trim_post();

	printf("Trimmed $trimcount files, %.1fMB\n", ($trimbytes / (1024 * 1024)));
	exit 0;
}

# in: "category/name"
# out: "category/name/name.info"
sub find_info_file {
	my $info = shift;
	$info =~ s,/([^/]+)$,/$1/$1.info,;
	return $info;
}

# FIXME: this will fail if @localfiles are absolute paths!
sub local_add {
	my ($oldcwd, $catname, $info, @localfiles) = @_;
	$catname =~ s,^\./,,;
	my ($category, $prgnam) = split /\//, $catname;
	my %localmd5s;

	for(@localfiles) {
		$localmd5s{md5sum_file("$oldcwd/$_")} = "$oldcwd/$_";
	}

	my $dls = parse_info($info);

	chdir($archivedir) or die "$archivedir: $!";
	for(keys %$dls) {
		my $targetfile = url_to_filename($_);

		my $md5 = $dls->{$_};
		my $localfile = $localmd5s{$md5};
		next unless $localfile;

		delete $localmd5s{$md5};

		system("cp \"$localfile\" \"./$targetfile\"");
		store_file($targetfile, $category, $prgnam, $md5);
		unlink($targetfile);
	}

	for(keys %localmd5s) {
		print "$localmd5s{$_} ($_) ignored: doesn't match any md5sum in $info\n";
	}

	exit 0;
}

sub add_or_rm_mode {
	my $oldcwd = POSIX::getcwd();
	init_git();
	my $mode = shift @ARGV;
	my $catname = shift @ARGV or usage();

	if($catname eq '-f') {
		$maxfilemegs = 0;
		$catname = shift(@ARGV) or usage();
	}

	my $info = find_info_file($catname);
	if(! -f $info) {
		die "Can't find $info in repo\n";
	}

	if($mode eq "add") {
		if(!@ARGV) { # no args, use URL(s) in .info file
			$_ = $info;
			handle_info_file();
			exit 0;
		} else {
			local_add($oldcwd, $catname, $info, @ARGV);
		}
	} elsif($mode eq "rm") {
		my $dls = parse_info($info);
		for(keys %$dls) {
			my $md5 = $dls->{$_};
			my $filename = url_to_filename($_);
			my ($category, $prgname) = split /\//, $catname;
			unlink(name_dir($category, $prgname) . "/$filename");
			rmdir(name_dir($category, $prgname));
			unlink(md5_dir($md5) . "/$filename");
			rmdir(md5_dir($md5));
			exit 0;
		}
	} else {
		die "this never happens";
	}
}

sub usage {
	my $self = $0;
	$self =~ s,.*/,,;

	print <<EOF;
$self - create and maintain SBo source archive

Usage: $self <mode>

<mode> is one of:

  create
  update
  purge
  trim
  check
  add <category/prgname> [<file> ...]
  rm <category/prgname>

For full documentation try:
  perldoc $self
EOF

	exit 1
}

#main()

usage() unless defined $ARGV[0];
read_config();
for ($ARGV[0]) {
	/create/ && do { create_mode();    };
	/update/ && do { update_mode();    };
	/purge/  && do { purge_mode();     };
	/add/    && do { add_or_rm_mode(); };
	/rm/     && do { add_or_rm_mode(); };
	/trim/   && do { trim_mode();      };
	usage();
}

__END__