1 |
#!/usr/bin/perl -w |
2 |
use strict; |
3 |
$|++; |
4 |
|
5 |
# Mirror CPAN latest archive. Based on article "Mirroring your own mini-CPAN" |
6 |
# by Randal L. Schwartz for Linux Magazine Column 42 (Nov 2002) available on |
7 |
# http://www.stonehenge.com/merlyn/LinuxMag/col42.html |
8 |
# http://www.stonehenge.com/merlyn/LinuxMag/col42.listing.txt |
9 |
|
10 |
# TODO: |
11 |
# - support for ZIP archives (and fix .tar.gz cludges) |
12 |
# - add version requirement for Archive::Tar (0.22 has a bug which |
13 |
# prevents it to extract some tars) |
14 |
|
15 |
### CONFIG |
16 |
|
17 |
#my $REMOTE = "http://ftp.linux.hr/CPAN/"; |
18 |
my $REMOTE = "http://www.cpan.org/"; |
19 |
# my $REMOTE = "http://fi.cpan.org/"; |
20 |
# my $REMOTE = "http://au.cpan.org/"; |
21 |
# my $REMOTE = "file://Users/merlyn/MIRROR/CPAN/"; |
22 |
|
23 |
## warning: unknown files below this dir are deleted! |
24 |
my $LOCAL = "/mirrors/cpan/CPAN/"; |
25 |
|
26 |
my $TRACE = 0; |
27 |
|
28 |
### END CONFIG |
29 |
|
30 |
## core - |
31 |
use File::Path qw(mkpath); |
32 |
use File::Basename qw(dirname); |
33 |
use File::Spec::Functions qw(catfile devnull); |
34 |
use File::Find qw(find); |
35 |
|
36 |
## LWP - |
37 |
use URI (); |
38 |
use LWP::Simple qw(mirror RC_OK RC_NOT_MODIFIED); |
39 |
|
40 |
## Compress::Zlib - |
41 |
use Compress::Zlib qw(gzopen $gzerrno); |
42 |
|
43 |
## Archive::Tar - |
44 |
use Archive::Tar qw(); |
45 |
|
46 |
## first, get index files |
47 |
my_mirror($_) for qw( |
48 |
authors/01mailrc.txt.gz |
49 |
modules/02packages.details.txt.gz |
50 |
modules/03modlist.data.gz |
51 |
); |
52 |
|
53 |
## now walk the packages list |
54 |
my $details = catfile($LOCAL, qw(modules 02packages.details.txt.gz)); |
55 |
my $gz = gzopen($details, "rb") or die "Cannot open details: $gzerrno"; |
56 |
my $inheader = 1; |
57 |
while ($gz->gzreadline($_) > 0) { |
58 |
if ($inheader) { |
59 |
$inheader = 0 unless /\S/; |
60 |
next; |
61 |
} |
62 |
|
63 |
my ($module, $version, $path) = split; |
64 |
next if $path =~ m{/perl-5}; # skip Perl distributions |
65 |
my_mirror("authors/id/$path", 1); |
66 |
} |
67 |
|
68 |
## finally, clean the files we didn't stick there |
69 |
clean_unmirrored(); |
70 |
|
71 |
exit 0; |
72 |
|
73 |
BEGIN { |
74 |
## %mirrored tracks the already done, keyed by filename |
75 |
## 1 = local-checked, 2 = remote-mirrored |
76 |
my %mirrored; |
77 |
|
78 |
sub my_mirror { |
79 |
my $path = shift; # partial URL |
80 |
my $skip_if_present = shift; # true/false |
81 |
|
82 |
my $remote_uri = URI->new_abs($path, $REMOTE)->as_string; # full URL |
83 |
my $local_file = catfile($LOCAL, split "/", $path); # native absolute file |
84 |
my $checksum_might_be_up_to_date = 1; |
85 |
|
86 |
if ($skip_if_present and -f $local_file) { |
87 |
## upgrade to checked if not already |
88 |
$mirrored{$local_file} = 1 unless $mirrored{$local_file}; |
89 |
} elsif (($mirrored{$local_file} || 0) < 2) { |
90 |
## upgrade to full mirror |
91 |
$mirrored{$local_file} = 2; |
92 |
|
93 |
mkpath(dirname($local_file), $TRACE, 0711); |
94 |
print $path if $TRACE; |
95 |
my $status = mirror($remote_uri, $local_file); |
96 |
|
97 |
if ($status == RC_OK) { |
98 |
$checksum_might_be_up_to_date = 0; |
99 |
print " ... updated\n" if $TRACE; |
100 |
} elsif ($status != RC_NOT_MODIFIED) { |
101 |
warn "\n$remote_uri: $status\n"; |
102 |
return; |
103 |
} else { |
104 |
print " ... up to date\n" if $TRACE; |
105 |
} |
106 |
} |
107 |
|
108 |
if ($path =~ m{^authors/id}) { # maybe fetch CHECKSUMS |
109 |
my $checksum_path = |
110 |
URI->new_abs("CHECKSUMS", $remote_uri)->rel($REMOTE); |
111 |
if ($path ne $checksum_path) { |
112 |
my_mirror($checksum_path, $checksum_might_be_up_to_date); |
113 |
} |
114 |
} |
115 |
} |
116 |
|
117 |
sub clean_unmirrored { |
118 |
find sub { |
119 |
return if /\.readme$/; # don't erase readme files |
120 |
check_readme($File::Find::name); |
121 |
return unless -f and not $mirrored{$File::Find::name}; |
122 |
print "$File::Find::name ... removed\n" if $TRACE; |
123 |
unlink $_ or warn "Cannot remove $File::Find::name: $!"; |
124 |
my $path = $File::Find::name; |
125 |
if ($path =~ s/\.tar\.gz/.readme/g && -f $path) { |
126 |
# only if we erase archive also! |
127 |
unlink $path or warn "Cannot remove $path: $!"; |
128 |
} |
129 |
}, $LOCAL; |
130 |
} |
131 |
|
132 |
sub check_readme { |
133 |
|
134 |
my $path = shift; |
135 |
|
136 |
# fixup some things |
137 |
my $readme_path = $path; |
138 |
$readme_path =~ s/\.tar\.gz/.readme/g || return; # just .tar.gz is supported! |
139 |
|
140 |
my $at = Archive::Tar->new($path) or die "Archive::Tar failed on $path\n"; |
141 |
|
142 |
if (! -f $readme_path) { |
143 |
# create readme file |
144 |
my @readmes = sort grep m{^[^/]+/README\z}, $at->list_files(); |
145 |
my $readme; |
146 |
|
147 |
if ($readme = shift @readmes) { |
148 |
open(R, "> $readme_path") || die "Cannot create $readme_path: $!"; |
149 |
print R $at->get_content($readme); |
150 |
close(R); |
151 |
|
152 |
print "$readme_path ... created\n" if $TRACE; |
153 |
|
154 |
} else { |
155 |
|
156 |
$readme_path =~ s/^.+\/(.+)$/$1/; |
157 |
print "can't find readme for $readme_path ...\n" if $TRACE; |
158 |
|
159 |
} |
160 |
|
161 |
} |
162 |
} |
163 |
|
164 |
|
165 |
} |