#!/usr/bin/perl -w
-#
+use IPC::Run qw(run);
-$commits_list = "pld-cvs-commit\@pld-linux.org";
+use Sys::Syslog;
+
+openlog("distfiles-ff", "pid", "user");
+
+$commits_list = "pld-cvs-commit\@lists.pld-linux.org";
$spool_dir = "./spool";
$copy_dir = "src"; # relative to ftp root
$no_url_dir = "./upload";
$df_server = "distfiles.pld-linux.org";
-$df_scp = "plddist\@$df_server:ftp";
+$df_scp = "plddist\@distfiles.pld-linux.org:ftp";
+$user_agent = "PLD/distfiles";
@md5 = ();
-%url = ();
+@url = ();
$problems = "";
$normal_out = "";
$requester = "";
$fetched_count = 0;
$force_reply = 0;
$req_login = "nobody";
+$spec = "";
@files = ();
+sub fatal($)
+{
+ my $msg = shift;
+
+ syslog("err","FATAL: $msg");
+ die "$msg";
+}
+
# try lookup some file in spool, exit if it cannot be done
sub find_file_in_spool()
{
- opendir(DIR, $spool_dir) || die "can't opendir $spool_dir: $!";
+ opendir(DIR, $spool_dir) || fatal("can't opendir $spool_dir: $!");
while (1) {
my $f = readdir(DIR);
defined $f or last;
}
# read file from spool, and try unlink it. if cannot unlink -- exit
-# sets $requester (email), $problems, @md5 (arrays of md5's)
-# and %url (map from md5 to urls)
+# sets $requester (email), $problems, @md5 (arrays of md5's)
+# and @url (map from md5 to urls)
sub read_spool_file()
{
+ syslog("info","reading spool file $file");
open(F, "< $file") || exit 0;
$requester = <F>;
chomp $requester;
- $requester =~ /^[a-zA-Z_\-0-9\@\.]+$/
- or die "$file: evil requester: $requester";
+ $requester =~ /^[a-zA-Z_0-9@.-]+$/
+ or fatal("$file: evil requester: $requester");
$req_login = "";
$requester =~ /^([^@]+)\@/ and $req_login = $1;
- $req_login =~ /^[a-z0-9A-Z_]+$/ or die "$file: evil requester $requester";
+ $req_login =~ /^[a-z0-9A-Z_.]+$/ or fatal("$file: evil requester $requester");
+ $spec = <F>;
+ chomp $spec;
+ $spec =~ s/\.spec$//;
+
my $flags = <F>;
$force_reply++ if ($flags =~ /force-reply/);
-
+
while (<F>) {
if (/^ERROR/) {
s/^ERROR: //;
$problems .= $_;
next;
}
- /^([a-f0-9]{32})\s+((ftp|http|no-url|no-url-copy):\/\/([a-z0-9A-Z:\+\~\.\-\/_]|\%[0-9])+)\s*$/
- or die "$file: corrupted";
- push @md5, $1;
- $url{$1} = $2;
- /\/$/ and die "$file: cannot fetch dir";
+ if (/^([a-f0-9]{32})\s+((ftp|http|https|no-url|no-url-copy):\/\/([a-z0-9A-Z;:\=\?&\@\+\~\.,\-\/_]|\%[0-9])+(#\/[a-zA-Z0-9\._-]+)?)\s*$/) {
+ if (/\/$/) {
+ $problems .= "$file: cannot fetch dir";
+ } else {
+ push @md5, $1;
+ push @url, $2;
+ }
+ } else {
+ $problems .= "FILE: $file: corrupted";
+ }
}
close(F);
sub copy_to_df($$)
{
my ($from, $to) = @_;
- my $cmd = "scp -r -B -q $from $df_scp/$to";
- open(E, "$cmd 2>&1 |") or die;
+ my $cmd = "scp -pr -B -q $from $df_scp/$to";
+ open(E, "$cmd 2>&1 |") or fatal("$cmd failed");
my $oops = "";
while (<E>) {
$oops .= $_;
}
- close(E);
+ $oops .= "\nThe command has exited with a non-zero status."
+ unless (close (E));
$problems .= "scp problems: $cmd:\n$oops\n"
if ($oops ne "");
+ syslog("err","copy_to_df oops for '$cmd': $oops") if ($oops ne "");
return ($oops ne "");
}
my ($md5, $url, $local_copy) = @_;
my $bn = basename($url);
-
+
if ($local_copy ne "$tmp_dir/$md5/$bn") {
if (system("mv -f \"$local_copy\" \"$tmp_dir/$md5/$bn\"")) {
$problems .= "FATAL: cannot move $local_copy to $tmp_dir\n";
}
$local_copy = "$tmp_dir/$md5/$bn";
}
-
+ # no-url local copy may have wrong permissions (it's created by ftp upload)
+ # always fix them
+ chmod(0644, $local_copy);
+
if (open(D, "> $tmp_dir/$md5/$bn.desc")) {
print D "URL: $url\n";
print D "Login: $req_login\n";
my $dir = by_md5($md5, $url);
$dir =~ s|/[^/]+/[^/]+$||;
if (copy_to_df("$tmp_dir/$md5/", $dir) == 0) {
- $normal_out .=
+ $normal_out .=
"STORED: $url\n" .
"\t$md5 " . basename($url) . "\n" .
"\tSize: " . (-s $local_copy) . " bytes\n";
sub make_src_symlink($$)
{
my ($md5, $url) = @_;
-
+
return unless ($url =~ /^no-url/);
-
+
my $b = basename($url);
if (open(S, "> $tmp_dir/$b.link")) {
if ($url =~ /^no-url-copy/) {
sub md5($)
{
my $file = shift;
- my $md5 = `md5sum "$file" 2>/dev/null`;
+ my $in = "";
+ my $md5 = "";
+ my $err = "";
+ my @cmd = ("md5sum", $file);
+
+ run \@cmd, \$in, \$md5, \$err;
+ if ($err ne "") {
+ chomp($err);
+ $problems .= "FATAL: " . $err . "\n";
+ return "error";
+ }
+ chomp $md5;
$md5 =~ /^([a-f0-9]{32})/ and $md5 = $1;
return $md5;
}
sub handle_no_url($$)
{
my ($md5, $url) = @_;
-
- $url =~ m|://([^/]+)| or die "corrupted! (no-url)";
+
+ unless ($url =~ m#://([^/]+)#) {
+ $problems .= "$url: corrupted! (no-url)";
+ return;
+ }
my $basename = $1;
- my $file = "$no_url_dir/$req_login/$basename";
+ my $file = "$no_url_dir/$basename";
+
+ $file = "$no_url_dir/$req_login/$basename" unless (-f $file);
if (-f $file) {
my $computed_md5 = md5($file);
if ($computed_md5 ne $md5) {
- $problems .= "FATAL: $file md5 mismatch, needed $md5, got $computed_md5\n";
+ $problems .= "FATAL: $basename md5 mismatch, needed $md5, got $computed_md5\n";
} else {
move_file($md5, $url, $file);
make_src_symlink($md5, $url);
}
} else {
- $problems .= "FATAL: $file was not uploaded\n";
+ $problems .= "FATAL: $basename was not uploaded\n";
}
}
my $all_out = "";
my $bn = basename($url);
my $local = "$tmp_dir/$md5/$bn";
- my $cmd = "wget -nv -O $local \"$url\"";
+ my @cmd = ("wget", "-nv", "--no-check-certificate", "--user-agent=$user_agent", "-O", $local, $url);
+ my $cmd_joined = join(' ', @cmd);
+ my @cmd2 = ("wget", "-nv", "--no-check-certificate", "--user-agent=$user_agent", "--passive-ftp", "-O", $local, $url);
+ my $cmd2_joined = join(' ', @cmd2);
+
+ syslog("info","fetch_file($md5,$url)");
push @files, $bn;
+ if ( $bn =~ m/(%[0-9a-f]{2})/i ) {
+ $problems .= "$bn: refusing to download file with uri escape codes ($1) in the name\n";
+ $bn =~ s/%[0-9a-f]{2}/_/g;
+ $problems .= "HINT: use $url?/$bn as source to rename the file\n\n";
+ return;
+ }
+
if (got_on_distfiles($md5, $url)) {
- $normal_out .=
- "ALREADY GOT: $url\n" .
- "\t$md5 " . basename($url) . "\n";
+ $normal_out .=
+ "ALREADY GOT: $url\n" .
+ "\t$md5 " . basename($url) . "\n";
make_src_symlink($md5, $url);
return;
}
- mkdir("$tmp_dir/$md5") or die;
-
+ mkdir("$tmp_dir/$md5");
+
if ($url =~ /^no-url/) {
handle_no_url($md5, $url);
return;
}
-
- open(W, "$cmd 2>&1 |");
+
+ my $pid = open(W, "-|");
+ fatal("Cannot fork $!") unless defined $pid;
+ unless ( $pid ) {
+ open STDERR, ">&", \*STDOUT or fatal("$0: open: $!");
+ exec { $cmd[0] } @cmd or fatal("$0: exec: $!");
+ }
while (<W>) {
$all_out .= $_;
/URL:.*\s+\-\>\s+.*/ and next;
}
close(W);
if ($out ne "") {
- $problems .= "$cmd:\n$out\n\n";
+ syslog("err","$cmd_joined: $out");
+ $problems .= "$cmd_joined:\n$out\n\n";
+ }
+ if ( $? ) {
+ $problems .= sprintf "%s:\nexited with code %d (0x%02x)\n\n",
+ $cmd_joined,
+ $? >> 8,
+ $? & 0xff;
}
- if (-f $local && -s $local > 0) {
+ if (-f $local && -s $local == 0 && $url =~ /^ftp:/) {
+ $out = "";
+ my $pid = open(W, "-|");
+ fatal("Cannot fork $!") unless defined $pid;
+ unless ( $pid ) {
+ open STDERR, ">&", \*STDOUT or fatal("$0: open: $!");
+ exec { $cmd2[0] } @cmd2 or fatal("$0: exec: $!");
+ }
+ while (<W>) {
+ $all_out .= "\n\t\t$_";
+ /URL:.*\s+\-\>\s+.*/ and next;
+ $out .= $_;
+ }
+ close(W);
+ if ($out ne "") {
+ $problems .= "$cmd2_joined:\n$out\n\n";
+ }
+ if ( $? ) {
+ $problems .= sprintf "%s:\nexited with code %d (0x%02x)\n\n",
+ $cmd2_joined,
+ $? >> 8,
+ $? & 0xff;
+ }
+ }
+ if (-r $local && -s $local > 0) {
my $computed_md5 = md5($local);
if ($computed_md5 ne $md5) {
$problems .= "FATAL: $url md5 mismatch, needed $md5, got $computed_md5\n";
} else {
my $testcmd = "file \"$local\" |";
my $testres = "";
- if ($url =~ /^http:/ && $local =~ /\.(tar\.(bz2|gz)|tgz|zip|jar|xpi)$/) {
- open(T, $testcmd) or die;
+ if ($url =~ /^(http|https):/ && $local =~ /\.(tar\.(bz2|gz)|tgz|zip|jar|xpi)$/) {
+ open(T, $testcmd) or fatal("$testcmd failed");
$testres = <T>;
close(T);
}
move_file($md5, $url, $local);
}
}
+ } elsif (-f $local && -s $local > 0) {
+ $problems .= "FATAL: $url ($md5) was not fetched correctly ($cmd_joined: $all_out): file is not readable\n";
+ } elsif (-f $local && not -s $local) {
+ $problems .= "FATAL: $url ($md5) was not fetched correctly ($cmd_joined: $all_out): file fetched but has 0 length\n";
} else {
- $problems .= "FATAL: $url ($md5) was not fetched ($cmd: $all_out)\n";
+ $problems .= "FATAL: $url ($md5) was not fetched correctly ($cmd_joined: $all_out)\n";
}
# save space
unlink($local);
sub fetch_files()
{
$problems .= "\n\n" if ($problems ne "");
- foreach $md5 (@md5) {
- fetch_file($md5, $url{$md5});
+ foreach $i (0..$#md5) {
+ fetch_file($md5[$i], $url[$i]);
}
}
sub send_email()
{
+ syslog("info","sending email to $requester");
open(EMAIL, "| /usr/sbin/sendmail -t");
#open(EMAIL, "| cat");
my $marker = "";
splice(@files, 10, @files - 10, "...")
if (@files > 10);
-
- print EMAIL
+
+ print EMAIL
"From: $req_login <$requester>
To: $commits_list
Cc: $requester
-Subject: DISTFILES: ${marker}@{files}
+Subject: DISTFILES: ${spec}: ${marker}@{files}
Message-ID: <$$." . time . "\@distfiles.pld-linux.org>
X-distfiles-program: file-fetcher.pl
X-distfiles-version: " . '$Id$' . "
--
Virtually Yours: distfiles.
";
- close(EMAIL) or die;
+ close(EMAIL) or fatal("close() failed");
}
sub make_tmp_dir()
chomp $id;
$id = rand if (!defined $id or $id eq "");
$tmp_dir = "./tmp/$id";
- mkdir($tmp_dir) or die;
+ mkdir($tmp_dir) or fatal("mkdir($tmp_dir) failed");
}
sub clean_tmp_dir()
fetch_files();
send_email() unless (!$force_reply and $problems eq "" and $fetched_count == 0);
clean_tmp_dir();
+
+# vim: ts=2:sw=2:et:fdm=marker