Merge branch 'jn/mime-type-with-params'
[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 $typeflag == 5; # directory
98
99                 print FI "blob\n", "mark :$next_mark\n";
100                 if ($typeflag == 2) { # symbolic link
101                         print FI "data ", length($linkname), "\n", $linkname;
102                         $mode = 0120000;
103                 } else {
104                         print FI "data $size\n";
105                         while ($size > 0 && read(I, $_, 512) == 512) {
106                                 print FI substr($_, 0, $size);
107                                 $size -= 512;
108                         }
109                 }
110                 print FI "\n";
111
112                 my $path;
113                 if ($prefix) {
114                         $path = "$prefix/$name";
115                 } else {
116                         $path = "$name";
117                 }
118                 $files{$path} = [$next_mark++, $mode];
119
120                 $author_time = $mtime if $mtime > $author_time;
121                 $path =~ m,^([^/]+)/,;
122                 $top_dir = $1 unless $top_dir;
123                 $have_top_dir = 0 if $top_dir ne $1;
124         }
125
126         my $commit_msg = "Imported from $tar_file.";
127         my $this_committer_name = $committer_name;
128         my $this_committer_email = $committer_email;
129         my $this_author_name = $author_name;
130         my $this_author_email = $author_email;
131         if ($metaext ne '') {
132                 # Optionally read a commit message from <filename.tar>.msg
133                 # Add a line on the form "Committer: name <e-mail>" to override
134                 # the committer and "Author: name <e-mail>" to override the
135                 # author for this tar ball.
136                 if (open MSG, '<', "${tar_file}.${metaext}") {
137                         my $header_done = 0;
138                         $commit_msg = '';
139                         while (<MSG>) {
140                                 if (!$header_done && /^Committer:\s+([^<>]*)\s+<(.*)>\s*$/i) {
141                                         $this_committer_name = $1;
142                                         $this_committer_email = $2;
143                                 } elsif (!$header_done && /^Author:\s+([^<>]*)\s+<(.*)>\s*$/i) {
144                                         $this_author_name = $1;
145                                         $this_author_email = $2;
146                                 } elsif (!$header_done && /^$/) { # empty line ends header.
147                                         $header_done = 1;
148                                 } else {
149                                         $commit_msg .= $_;
150                                         $header_done = 1;
151                                 }
152                         }
153                         close MSG;
154                 }
155         }
156
157         print FI <<EOF;
158 commit $branch_ref
159 author $this_author_name <$this_author_email> $author_time +0000
160 committer $this_committer_name <$this_committer_email> $commit_time +0000
161 data <<END_OF_COMMIT_MESSAGE
162 $commit_msg
163 END_OF_COMMIT_MESSAGE
164
165 deleteall
166 EOF
167
168         foreach my $path (keys %files)
169         {
170                 my ($mark, $mode) = @{$files{$path}};
171                 $path =~ s,^([^/]+)/,, if $have_top_dir;
172                 $mode = $mode & 0111 ? 0755 : 0644 unless $mode == 0120000;
173                 printf FI "M %o :%i %s\n", $mode, $mark, $path;
174         }
175         print FI "\n";
176
177         print FI <<EOF;
178 tag $tar_name
179 from $branch_ref
180 tagger $author_name <$author_email> $author_time +0000
181 data <<END_OF_TAG_MESSAGE
182 Package $tar_name
183 END_OF_TAG_MESSAGE
184
185 EOF
186
187         close I;
188 }
189 close FI;