path: root/urpmi-proxy.cgi

#!/usr/bin/perl -w
## written by Maarten Vanraes (c) 2009-2013
## urpmi-proxy is GPLv2+

use strict;
use warnings;

use Time::HiRes qw( gettimeofday );

my $debug = 0;
my $proxy = 0;
my $config_file = '/etc/urpmi-proxy.conf';

# config defaults
my $cache_tmp_path = '/var/tmp/urpmi-proxy';
my $cache_path = '/var/cache/urpmi-proxy';
my $logfile = '/var/log/urpmi-proxy.log';
my $check_updates_only_files = undef;
my $check_no_updates_files = '\.rpm$';
my $merge_files = undef;
my $sources = [
	'urpmi'
];
my $connect_timeout = 120;
my $ftp_response_timeout = 30;
my $max_stall_speed = 8192;
my $max_stall_time = 60;
my $start_buffer = 4096;

# load config file
if (-R $config_file) {
	my $r = open(FILE, '<', $config_file);
	if ($r) {
		my $l = '';
		while (my $i = <FILE>) {
			$l .= $i;
		}
		eval $l;
		close FILE;
	}
}
print STDERR "logfile: $logfile\n" if $debug;

print STDERR "orig sources: " . scalar(@$sources) . "\n" if $debug;
if ($debug) {
	foreach my $sou (@$sources) {
		print STDERR " - " . $sou . "\n";
	}
}

# prepare cache path
system("mkdir -p $cache_tmp_path");

# check for valid request
my $file = $ENV{PATH_INFO};
return_error(500, 'Server error') if !$file;

# split up request
return_error(500, 'Server error') if $file !~ m!^(.*)/([^/]*)$!;
my $dest_path = $1;
my $filename = $2;
my $file_type = "";
my $merge = 0;

print STDERR "file: $file\n" if $debug;
print STDERR "dest_path: $dest_path\n" if $debug;
print STDERR "filename: $filename\n" if $debug;

# check if there's a time condition
my $modified_since = 0;
if (defined $ENV{HTTP_IF_MODIFIED_SINCE}) {
	$modified_since = `date --date="$ENV{HTTP_IF_MODIFIED_SINCE}" "+%s"` if defined $ENV{HTTP_IF_MODIFIED_SINCE};
	$modified_since =~ s/[\s\r\n]*$//;
	print STDERR "is modified since: $ENV{HTTP_IF_MODIFIED_SINCE} ($modified_since) ?\n" if $debug && $modified_since;
}

# set request ENV vars
my $ip = $ENV{REMOTE_ADDR};
my $user_agent = '';
$user_agent = $ENV{HTTP_USER_AGENT} if defined $ENV{HTTP_USER_AGENT};

# check if request needs update checking
my $check_file = 1;
$check_file = 0 if defined $check_no_updates_files && $filename =~ m/$check_no_updates_files/;
if (defined $check_updates_only_files) {
	$check_file = 0;
	$check_file = 1 if $filename =~ m/$check_updates_only_files/;
}

# check if request needs merging
$merge = 1 if defined $merge_files && $filename =~ m/$merge_files/;
$check_file = 1 if $merge;

print STDERR "check_file: $check_file\n" if $debug;
print STDERR "merge: $merge\n" if $debug;

# get datetime from local file if it exists
my @stat = lstat($cache_path . $file);
if ($filename && scalar(@stat) > 0 && $stat[7] > 0) {
	print STDERR "timestamp: $stat[9]\n" if $debug;
	# if the file needs no update checks, check in cache
	return_file($cache_path, $file, $logfile, 'HIT_NO_CHECK', \@stat, $ip, $user_agent) if !$check_file;
}

# set up curl with timecheck
my $curl;
my $r = 0;
my $file_sent = 0;
my $file_unmodified = 0;
my $file_time = -1;
my $err = 200;
my $buffer;
my $pos;

# prepare curl transfer
my $tmp_file = $cache_tmp_path . "/" . rand() . $$;
open(FILEHANDLE, ">", $tmp_file) or do {
	_log($logfile, $file, 500, 'MISS', -1, $ip, $user_agent);
	return_error(500, 'Server error');
};
binmode(FILEHANDLE);
my %curldata = (fh => \*FILEHANDLE, file_sent => \$file_sent, content_type => $file_type, size => -1, merge => \$merge, buffer => \$buffer, pos => \$pos, max => $start_buffer);

print STDERR "sources: " . scalar(@$sources) . "\n" if $debug;
if ($debug) {
	foreach my $sou (@$sources) {
		print STDERR " - " . $sou . "\n";
	}
}

# filter out duplicate sources (and expand urpmi)
my @sources;
my %seen;
foreach my $s (@$sources) {
	# heh
	next if $seen{$s}++;
	if ($s eq "urpmi") {
		# urpmi support is required
		use urpm;
		use urpm::cfg;
		my $urpm = new urpm();
		urpm::get_global_options($urpm);
		my $config = urpm::cfg::load_config($urpm->{config});
		my %s;
		foreach my $media (@{$config->{media}}) {
			if (!$media->{ignore}) {
				if ($media->{mirrorlist} && !($seen{'mirrorlist:' . $media->{mirrorlist}}++)) {
					# push mirrorlists now so they'll be first
					push @sources, 'mirrorlist://' . $media->{mirrorlist};
				}
				elsif ($media->{url} && !($seen{$media->{url}}++)) {
					$s{$media->{url}} = 1;
				}
			}
		}
		# push the urls
		push @sources, keys %s if scalar(keys %s);
	}
	else {
		push @sources, $s;
	}
}

# check for source
print STDERR "interpolated sources: " . scalar(@sources) . "\n" if $debug;
if ($debug) {
	foreach my $sou (@sources) {
		print STDERR " - " . $sou . "\n";
	}
}
foreach my $source (@sources) {
	my ($type, @loc) = split('://', $source);
	# clear buffer
	$buffer = '';
	$pos = 0;
	my $loc = join('://', @loc);
	print STDERR "source of type $type: '" . $source . "'\n" if $debug;
	if ($type eq 'mirrorlist') {
		# get exact url from cache and parse
		my $res = open(FILE, '<', '/var/cache/urpmi/mirrors.cache');
		if ($res) {
			my $mirrorcache = '';
			while (my $i = <FILE>) {
				$mirrorcache .= $i;
			}
			close FILE;
			my $host_loc = $loc;
			$host_loc =~ s/\$/\\\$/g;
			if ($mirrorcache =~ m/'$host_loc'\s+=>\s+{[\r\n]+\s+'chosen'\s+=>\s+'([^']+)'/m) {
				$source = $1;
				# rectify source to remove '/distrib/version/arch'
				$source =~ s!/[^/]+/[^/]+/[^/]+$!!;
				print STDERR "mirrorlist returns source '$source'\n" if $debug;
				($type, @loc) = split('://', $source);
				$loc = join('://', @loc);
				if (defined $type) {
					print STDERR "mirrorlist returns type $type: '" . $source . "'\n" if $debug;
				}
				else {
					print STDERR "transfer error: mirrorlist is no url '" . $source . "'.\n" if $debug;
					$type = '';
				}
			}
			else {
				print STDERR "transfer error: mirrorlist has no chosen url '" . $source . "'.\n" if $debug;
				$type = '';
			}
		}
		else {
			print STDERR "transfer error: couldn't open mirrorlist cache.\n" if $debug;
		}
	}
	if ($type eq 'rsync') {
		# find the equivalent ftp mirror location by hostname
		my $res = open(FILE, '<', '/var/cache/urpmi/mirrors.cache');
		if ($res) {
			my $mirrorcache = '';
			while (my $i = <FILE>) {
				$mirrorcache .= $i;
			}
			close FILE;
			my $loc_host = $loc;
			$loc_host =~ s!/.+!!;
			if ($mirrorcache =~ m!'url'\s+=>\s+'((ftp|http)://$loc_host/[^']+)'!) {
				$source = $1;
				# rectify source to remove '/distrib/version/arch'
				$source =~ s!/[^/]+/[^/]+/[^/]+$!!;
				print STDERR "rsync switch returns source '$source'\n" if $debug;
				($type, @loc) = split('://', $source);
				$loc = join('://', @loc);
				if (defined $type) {
					print STDERR "rsync switch returns type $type: '" . $source . "'\n" if $debug;
				}
				else {
					print STDERR "transfer error: rsync switch is no url '" . $source . "'.\n" if $debug;
					$type = '';
				}
			}
			else {
				print STDERR "transfer error: rsync switch has no suitable url '" . $source . "'.\n" if $debug;
				$type = '';
			}
		}
		else {
			print STDERR "transfer error: couldn't open mirrorlist cache.\n" if $debug;
		}
	}
	if ($type eq 'file') {
		my @statl = lstat($loc . $file);
		if ($filename && scalar(@statl) > 0) {
			my $ft = `file -b --mime-type $loc$file`;
			$ft =~ s/[\s\r\n]*$//;
			my $t = localtime($statl[9]);
			print STDERR "HTTP Header: 200 OK\n" if $debug;
			print STDERR "Content-Type: $ft\n" if $debug;
			print STDERR "Content-Length: " . $statl[7] . "\n" if $debug;
			print STDERR "Last-Modified: " . $t . "\n" if $debug;
			$r = open(FILE, "<", $loc . $file);
			if ($r) {
				print STDERR "file fetch url '" . $loc . $file . "'\n" if $debug;
				if (!$file_sent) {
					$file_sent = $statl[7];
					print "Status: 200 OK\r\n";
					print "Content-Type: " . $ft . "\r\n";
					print "Content-Length: " . $statl[7] . "\r\n" if !$merge;
					print "Last-Modified: " . $t . "\r\n";
					print "\r\n";
				}
				else {
					$file_sent += $statl[7];
				}
				binmode(FILE);
				my $buf;
				while (read(FILE, $buf, 1024)) {
					print FILEHANDLE $buf;
					print $buf;
				}
				close FILE;
				$r = 0;
			}
			else {
				print STDERR "transfer error: couldn't open file '" . $loc . $file . "'.\n" if $debug;
				$r = 1;
				$err = 404;
			}
		}
		else {
			print STDERR "transfer error: couldn't read file '" . $loc . $file . "'.\n" if $debug;
		}
	}
	elsif ($type) {
		if (!defined $curl) {
			use WWW::Curl::Easy;
			# set up curl stuff
			$curl = new WWW::Curl::Easy;
			if (scalar(@stat) > 0 && $stat[9] > $modified_since) {
				$curl->setopt(CURLOPT_TIMECONDITION, 1); # CURL_TIMECOND_IFMODSINCE
				$curl->setopt(CURLOPT_TIMEVALUE, $stat[9]);
			}
			elsif ($modified_since > 0) {
				$curl->setopt(CURLOPT_TIMECONDITION, 1); # CURL_TIMECOND_IFMODSINCE
				$curl->setopt(CURLOPT_TIMEVALUE, $modified_since);
			}
			$curl->setopt(CURLOPT_USERAGENT, $user_agent) if $user_agent;
			$curl->setopt(CURLOPT_PROXY, $proxy) if $proxy;
			$curl->setopt(CURLOPT_CONNECTTIMEOUT, $connect_timeout);
			$curl->setopt(CURLOPT_FTP_RESPONSE_TIMEOUT, $ftp_response_timeout);
			$curl->setopt(CURLOPT_LOW_SPEED_LIMIT, $max_stall_speed);
			$curl->setopt(CURLOPT_LOW_SPEED_TIME, $max_stall_time);
			$curl->setopt(CURLOPT_FOLLOWLOCATION, 1);
			$curl->setopt(CURLOPT_FILETIME, 1);
			# hook curl transfer functions for local caching
			$curl->setopt(CURLOPT_WRITEDATA, \%curldata);
			$curl->setopt(CURLOPT_WRITEFUNCTION, \&write_function);
			$curl->setopt(CURLOPT_WRITEHEADER, \%curldata);
			$curl->setopt(CURLOPT_HEADERFUNCTION, \&header_function);
		}
		# depending on type check if remote file is newer
		print STDERR "curl fetch url '" . $source . $file . "'\n" if $debug;
		$curl->setopt(CURLOPT_URL, $source . $file);
		$r = $curl->perform;
		print STDERR "curl return value: " . $err . "\n" if $debug;
		# use curl to get it and output it directly
		if ($r == 0) {
			$err = $curl->getinfo(CURLINFO_HTTP_CODE);
			if ($err =~ m/^2/ || $err == 304) {
				if ($curl->getinfo(CURLINFO_CONDITION_UNMET)) {
					$file_unmodified = 1;
					print STDERR "condition unmet\n" if $debug;
				}
				$file_time = $curl->getinfo(CURLINFO_FILETIME);
				flush_buffer(\%curldata);
			}
			else {
				# error stuff ?
				print STDERR "transfer error: http code " . $err . "\n" if $debug;
			}
		}
		else {
			# error stuff ?
			print STDERR "transfer error: " . $curl->strerror($r) . " ($r)\n" if $debug;
		}
	}
	else {
		print STDERR "transfer error: this source does not have a type\n" if $debug;
	}
	print STDERR "file_sent: $file_sent\n" if $debug;
	last if $file_sent && !$merge;
}
# if file_sent, flush the buffer
if ($file_sent && $pos > 0) {
	print STDERR "still in buffer: $pos\n" if $debug;
	flush_buffer(\%curldata);
	print STDERR "file_sent(after flush): $file_sent\n" if $debug;
}

my $extra = '';

close(FILEHANDLE);

if ($file_sent && $r == 0 && $err =~ m/^2/ && $filename) {
	# clean up file and move to correct location
	if (system("mkdir -p $cache_path$dest_path") == 0) {
		if (rename($tmp_file, $cache_path . $file)) {
			utime(time(), $file_time, $cache_path . $file) if $file_time > 0;
		}
		else {
			print STDERR "WARNING: file '$tmp_file' could not be moved to '$cache_path$file'\n";
		}
	}
	else {
		print STDERR "WARNING: containing path for '$cache_path$file' could not be created\n";
	}
	_log($logfile, $file, 200, 'MISS', $file_sent, $ip, $user_agent);
}
else {
	unlink($tmp_file);
	if ($file_sent) {
		if ($filename) {
			_log($logfile, $file, $err, 'MISS_FAIL_SENT', $file_sent, $ip, $user_agent);
		}
		else {
			# It was actually successful, but paths can't be saved...
			print STDERR "NOTICE: paths cant be saved: '$file'\n" if $debug;
			_log($logfile, $file, $err, 'MISS', $file_sent, $ip, $user_agent);
		}
		exit 0;
	}
	if ($file_unmodified) {
		$extra = '_UNMODIFIED';
		if ($modified_since > 0 && (scalar(@stat) == 0 || $stat[9] <= $modified_since)) {
			# it's been requested, so we can answer unmodified
			_log($logfile, $file, 304, 'MISS' . $extra, 0, $ip, $user_agent);
			return_error(304, 'Unmodified');
		}
	}
	else {
		$extra = 'AFTER_FAIL';
	}
	return_file($cache_path, $file, $logfile, 'HIT' . $extra, \@stat, $ip, $user_agent) if $filename && scalar(@stat) > 0 && $stat[7] > 0;
	_log($logfile, $file, 404, 'MISS_FAIL', -1, $ip, $user_agent);
	return_error(404, 'File not found');
}

print STDERR "finished." if $debug;

exit 0;

sub header_function {
	my ($ptr, $data) = @_;
	if (!${$data->{file_sent}}) {
		$data->{http_header} = $1 if $ptr =~ m!^HTTP/[0-9.]+\s+(.+?)[\s\r\n]*$!;
		$data->{content_type} = $1 if $ptr =~ m/^Content-[tT]ype:\s+(.+?)[\s\r\n]*$/;
		$data->{size} = $1 if $ptr =~ m/^Content-[lL]ength:\s+(.+?)[\s\r\n]*$/;
		$data->{date} = $1 if $ptr =~ m/^Last-[mM]odified:\s+(.+?)[\s\r\n]*$/;
		$data->{size} = $1 if $ptr =~ m/^213\s+(.+?)[\s\r\n]*$/;
	}
	return length($ptr);
}

sub print_headers {
	my ($data) = @_;
	print STDERR "HTTP header: " . $data->{http_header} . "\n" if $debug && defined $data->{http_header};
	print STDERR "Content-Type: " . $data->{content_type} . "\n" if $debug && defined $data->{content_type};
	print STDERR "Content-Length: " . $data->{size} . "\n" if $debug;
	print STDERR "Last-Modified: " . $data->{date} . "\n" if $debug && defined $data->{date};
	print "Status: " . $data->{http_header} . "\r\n" if $data->{http_header} && $data->{http_header} !~ m/^2/;
	print "Content-Type: " . $data->{content_type} . "\r\n" if $data->{content_type};
	print "Content-Length: " . $data->{size} . "\r\n" if $data->{size} > -1 && !${$data->{merge}};
	print "Last-Modified: " . $data->{date} . "\r\n" if $data->{date};
	print "\r\n";
}

sub flush_buffer {
	my ($data) = @_;
	if (${$data->{pos}} > 0) {
		print STDERR "flushing buffer\n" if $debug;
		my $f = ${$data->{fh}};
		# output to file
		print $f (${$data->{buffer}});
		# set file_sent and print headers if required
		if (!${$data->{file_sent}}) {
			${$data->{file_sent}} = ${$data->{pos}};
			print_headers($data);
		}
		else {
			${$data->{file_sent}} += ${$data->{pos}};
		}
		# output
		binmode(STDOUT);
		print ${$data->{buffer}};
		# clear buffer
		${$data->{buffer}} = '';
		${$data->{pos}} = 0;
	}
}

sub write_function {
	my ($ptr, $data) = @_;
	my $l = length($ptr);
	my $f = ${$data->{fh}};
	my ($sec, $usec) = gettimeofday;
	print STDERR "[$sec,$usec]: write($l,${$data->{pos}},$data->{max},${$data->{file_sent}})\n" if $debug;
	# if not sent yet and we can fit in the data in the buffer, we'll use the buffer instead and do nothing
	if (!${$data->{file_sent}} && (${$data->{pos}} + $l < $data->{max})) {
		${$data->{buffer}} .= $ptr;
		${$data->{pos}} += $l;
		return $l;
	}
	($sec, $usec) = gettimeofday;
	print STDERR "[$sec,$usec]: write: flush\n" if $debug;
	# now that we are printing, flush the buffer
	flush_buffer($data);
	($sec, $usec) = gettimeofday;
	print STDERR "[$sec,$usec]: write: print file\n" if $debug;
	# output to file
	print $f ($ptr);
	# set file_sent and print headers if required
	if (!${$data->{file_sent}}) {
		($sec, $usec) = gettimeofday;
		print STDERR "[$sec,$usec]: write: headers\n" if $debug;
		${$data->{file_sent}} = $l;
		print_headers($data);
	}
	else {
		${$data->{file_sent}} += $l;
	}
	($sec, $usec) = gettimeofday;
	print STDERR "[$sec,$usec]: write: print STDOUT\n" if $debug;
	# output
	binmode(STDOUT);
	print $ptr;
	($sec, $usec) = gettimeofday;
	print STDERR "[$sec,$usec]: write: return\n" if $debug;
	return $l;
}

sub _log {
	my ($logfile, $file, $code, $cached, $size, $ip, $user_agent) = @_;
	my $date = `date`;
	$date =~ s/[\s\r\n]*$//;
	$size = '-' if $size < 0;
	open(FILE, ">>" . $logfile) or return;
	print FILE "[" . $date . "] $ip $code $size $cached '$file' '$user_agent'\n";
	close(FILE);
}

sub return_file {
	my ($cache_path, $file, $logfile, $cached, $stat, $ip, $user_agent)=@_;
	open(FILE, "<", $cache_path . $file) or do {
		_log($logfile, $file, 500, $cached, -1, $ip, $user_agent);
		return_error(500, 'Server error');
	};
	my $ft = `file -b --mime-type $cache_path$file`;
	$ft =~ s/[\s\r\n]*$//;
	my $t = localtime($stat->[9]);
	print STDERR "HTTP header: 200 OK\n" if $debug;
	print STDERR "Content-Type: '$ft'\n" if $debug;
	print STDERR "Content-Length: " . $stat->[7] . "\n" if $debug;
	print STDERR "Last-Modified: " . $t . "\n" if $debug;
	print "Status: 200 OK\r\n";
	print "Content-Type: " . $ft . "\r\n";
	print "Content-Length: " . $stat->[7] . "\r\n";
	print "Last-Modified: " . $t . "\r\n";
	print "\r\n";
	binmode(FILE);
	binmode(STDOUT);
	my $buf;
	while (read(FILE, $buf, 1024)) {
		print $buf;
	}
	close FILE;
	_log($logfile, $file, 200, $cached, $stat->[7], $ip, $user_agent);
	exit 0;
}

sub return_error {
	my ($code, $text) = @_;
	print "Status: $code $text\r\n\r\n";
	print STDERR "$code $text.\n" if $debug;
	exit 0;
}