Merge branch 'jc/verify-loose-object-header' into maint
[git] / contrib / fast-import / import-tars.perl
1 #!/usr/bin/perl
2
3 ## tar archive frontend for git-fast-import
4 ##
5 ## For example:
6 ##
7 ##  mkdir project; cd project; git init
8 ##  perl import-tars.perl *.tar.bz2
9 ##  git whatchanged import-tars
10 ##
11 ## Use --metainfo to specify the extension for a meta data file, where
12 ## import-tars can read the commit message and optionally author and
13 ## committer information.
14 ##
15 ##  echo 'This is the commit message' > myfile.tar.bz2.msg
16 ##  perl import-tars.perl --metainfo=msg myfile.tar.bz2
17
18 use strict;
19 use Getopt::Long;
20
21 my $metaext = '';
22
23 die "usage: import-tars [--metainfo=extension] *.tar.{gz,bz2,lzma,xz,Z}\n"
24         unless GetOptions('metainfo=s' => \$metaext) && @ARGV;
25
26 my $branch_name = 'import-tars';
27 my $branch_ref = "refs/heads/$branch_name";
28 my $author_name = $ENV{'GIT_AUTHOR_NAME'} || 'T Ar Creator';
29 my $author_email = $ENV{'GIT_AUTHOR_EMAIL'} || 'tar@example.com';
30 my $committer_name = $ENV{'GIT_COMMITTER_NAME'} || `git config --get user.name`;
31 my $committer_email = $ENV{'GIT_COMMITTER_EMAIL'} || `git config --get user.email`;
32
33 chomp($committer_name, $committer_email);
34
35 open(FI, '|-', 'git', 'fast-import', '--quiet')
36         or die "Unable to start git fast-import: $!\n";
37 foreach my $tar_file (@ARGV)
38 {
39         my $commit_time = time;
40         $tar_file =~ m,([^/]+)$,;
41         my $tar_name = $1;
42
43         if ($tar_name =~ s/\.(tar\.gz|tgz)$//) {
44                 open(I, '-|', 'gunzip', '-c', $tar_file)
45                         or die "Unable to gunzip -c $tar_file: $!\n";
46         } elsif ($tar_name =~ s/\.(tar\.bz2|tbz2)$//) {
47                 open(I, '-|', 'bunzip2', '-c', $tar_file)
48                         or die "Unable to bunzip2 -c $tar_file: $!\n";
49         } elsif ($tar_name =~ s/\.tar\.Z$//) {
50                 open(I, '-|', 'uncompress', '-c', $tar_file)
51                         or die "Unable to uncompress -c $tar_file: $!\n";
52         } elsif ($tar_name =~ s/\.(tar\.(lzma|xz)|(tlz|txz))$//) {
53                 open(I, '-|', 'xz', '-dc', $tar_file)
54                         or die "Unable to xz -dc $tar_file: $!\n";
55         } elsif ($tar_name =~ s/\.tar$//) {
56                 open(I, $tar_file) or die "Unable to open $tar_file: $!\n";
57         } else {
58                 die "Unrecognized compression format: $tar_file\n";
59         }
60
61         my $author_time = 0;
62         my $next_mark = 1;
63         my $have_top_dir = 1;
64         my ($top_dir, %files);
65
66         while (read(I, $_, 512) == 512) {
67                 my ($name, $mode, $uid, $gid, $size, $mtime,
68                         $chksum, $typeflag, $linkname, $magic,
69                         $version, $uname, $gname, $devmajor, $devminor,
70                         $prefix) = unpack 'Z100 Z8 Z8 Z8 Z12 Z12
71                         Z8 Z1 Z100 Z6
72                         Z2 Z32 Z32 Z8 Z8 Z*', $_;
73                 last unless length($name);
74                 if ($name eq '././@LongLink') {
75                         # GNU tar extension
76                         if (read(I, $_, 512) != 512) {
77                                 die ('Short archive');
78                         }
79                         $name = unpack 'Z257', $_;
80                         next unless $name;
81
82                         my $dummy;
83                         if (read(I, $_, 512) != 512) {
84                                 die ('Short archive');
85                         }
86                         ($dummy, $mode, $uid, $gid, $size, $mtime,
87                         $chksum, $typeflag, $linkname, $magic,
88                         $version, $uname, $gname, $devmajor, $devminor,
89                         $prefix) = unpack 'Z100 Z8 Z8 Z8 Z12 Z12
90                         Z8 Z1 Z100 Z6
91                         Z2 Z32 Z32 Z8 Z8 Z*', $_;
92                 }
93                 next if $name =~ m{/\z};
94                 $mode = oct $mode;
95                 $size = oct $size;
96                 $mtime = oct $mtime;
97                 next if $typeflag == 5; # directory
98
99                 if ($typeflag != 1) { # handle hard links later
100                         print FI "blob\n", "mark :$next_mark\n";
101                         if ($typeflag == 2) { # symbolic link
102                                 print FI "data ", length($linkname), "\n",
103                                         $linkname;
104                                 $mode = 0120000;
105                         } else {
106                                 print FI "data $size\n";
107                                 while ($size > 0 && read(I, $_, 512) == 512) {
108                                         print FI substr($_, 0, $size);
109                                         $size -= 512;
110                                 }
111                         }
112                         print FI "\n";
113                 }
114
115                 my $path;
116                 if ($prefix) {
117                         $path = "$prefix/$name";
118                 } else {
119                         $path = "$name";
120                 }
121
122                 if ($typeflag == 1) { # hard link
123                         $linkname = "$prefix/$linkname" if $prefix;
124                         $files{$path} = [ $files{$linkname}->[0], $mode ];
125                 } else {
126                         $files{$path} = [$next_mark++, $mode];
127                 }
128
129                 $author_time = $mtime if $mtime > $author_time;
130                 $path =~ m,^([^/]+)/,;
131                 $top_dir = $1 unless $top_dir;
132                 $have_top_dir = 0 if $top_dir ne $1;
133         }
134
135         my $commit_msg = "Imported from $tar_file.";
136         my $this_committer_name = $committer_name;
137         my $this_committer_email = $committer_email;
138         my $this_author_name = $author_name;
139         my $this_author_email = $author_email;
140         if ($metaext ne '') {
141                 # Optionally read a commit message from <filename.tar>.msg
142                 # Add a line on the form "Committer: name <e-mail>" to override
143                 # the committer and "Author: name <e-mail>" to override the
144                 # author for this tar ball.
145                 if (open MSG, '<', "${tar_file}.${metaext}") {
146                         my $header_done = 0;
147                         $commit_msg = '';
148                         while (<MSG>) {
149                                 if (!$header_done && /^Committer:\s+([^<>]*)\s+<(.*)>\s*$/i) {
150                                         $this_committer_name = $1;
151                                         $this_committer_email = $2;
152                                 } elsif (!$header_done && /^Author:\s+([^<>]*)\s+<(.*)>\s*$/i) {
153                                         $this_author_name = $1;
154                                         $this_author_email = $2;
155                                 } elsif (!$header_done && /^$/) { # empty line ends header.
156                                         $header_done = 1;
157                                 } else {
158                                         $commit_msg .= $_;
159                                         $header_done = 1;
160                                 }
161                         }
162                         close MSG;
163                 }
164         }
165
166         print FI <<EOF;
167 commit $branch_ref
168 author $this_author_name <$this_author_email> $author_time +0000
169 committer $this_committer_name <$this_committer_email> $commit_time +0000
170 data <<END_OF_COMMIT_MESSAGE
171 $commit_msg
172 END_OF_COMMIT_MESSAGE
173
174 deleteall
175 EOF
176
177         foreach my $path (keys %files)
178         {
179                 my ($mark, $mode) = @{$files{$path}};
180                 $path =~ s,^([^/]+)/,, if $have_top_dir;
181                 $mode = $mode & 0111 ? 0755 : 0644 unless $mode == 0120000;
182                 printf FI "M %o :%i %s\n", $mode, $mark, $path;
183         }
184         print FI "\n";
185
186         print FI <<EOF;
187 tag $tar_name
188 from $branch_ref
189 tagger $author_name <$author_email> $author_time +0000
190 data <<END_OF_TAG_MESSAGE
191 Package $tar_name
192 END_OF_TAG_MESSAGE
193
194 EOF
195
196         close I;
197 }
198 close FI;