source: gsdl/trunk/perllib/g2futil.pm@ 15656

package g2futil;


BEGIN 
{
    if (!defined $ENV{'FEDORA_HOME'}) {
    print STDERR "Error: Environment variable FEDORA_HOME not set.\n";
    exit 1;
    }

    my $fedora_home = $ENV{'FEDORA_HOME'};
    my $fedora_client_bin = &util::filename_cat($fedora_home,"client","bin");
    if($ENV{'GSDLOS'} =~ m/^windows/) {
    $ENV{'PATH'} .= ";$fedora_client_bin";
    } else {
    $ENV{'PATH'} .= ":$fedora_client_bin";
    }
}

use util;

sub run_cmd_old
{
    my ($cmd,$verbosity,$tolerate_error) = @_;

    if (($verbosity == 0) 
    || (defined $tolerate_error && ($tolerate_error eq "tolerate_error"))) {
    $cmd .= " > /dev/null"; # Too Unix specific?
    }

    if ($verbosity >= 2) {
    print "Runing command:\n";
    print "$cmd\n";
    }

    my $status = system($cmd);

    if ($verbosity >= 2) {
    print "Exit status = ", $status/256, "\n";
    }

    if ((!defined $tolerate_error) || ($tolerate_error ne "tolerate_error")) {
    if ($status>0) {
        print STDERR "Error executing:\n$cmd\n";
        print STDERR "$!\n";
    }
    }

    return $status;
}


sub run_cmd
{
    my ($prog,$arguments,$verbosity,$tolerate_error) = @_;

    my $script_ext = ($ENV{'GSDLOS'} =~ m/^windows/) ? ".bat" : ".sh";

    if ($prog =~ m/^fedora-/) {
    $prog .= $script_ext;
    }
    if (($prog =~ m/.pl$/i) && ($ENV{'GSDLOS'} =~ m/^windows/)) {
    $prog ="perl -S $prog";
    }
 
    my $cmd = "$prog $arguments";

###    print "*** cmd = $cmd\n";

    if (open(CMD,"$cmd 2>&1 |"))
    {
    my $result = "";
    my $line;
    while (defined ($line = <CMD>))
    {   
        $result .= $line;       

        if ((!defined $tolerate_error) || ($tolerate_error ne "tolerate_error"))
        {
        print $line;
        }


    }
    
    close(CMD);
    
    $cmd_status = $?;

    if ($cmd_status == 0) {
        # Check for any lines in result begining 'Error:'
        
        if ($result =~ m/^Error\s*:/m) {
        # Fedora script generated an error, but did not exit
        # with an error status => artificially raise one

        $cmd_status = -1;
        }
    }

    if ($cmd_status != 0) {

        if ((!defined $tolerate_error) || ($tolerate_error ne "tolerate_error"))
        {
        print STDERR "Error: processing command failed.  Exit status $cmd_status\n";
        
        if ($verbosity >= 2) {
            print STDERR "  Command was: $cmd\n";
        }
        if ($verbosity >= 3) {
            print STDERR "result: $result\n";
        }

        }
    }
    }
    else 
    {
    print STDERR "Error: failed to execute $cmd\n";
    }


    return $cmd_status;
}


sub run_datastore_info
{
    my ($pid,$options) = @_;

    my $verbosity = $options->{'verbosity'};

    my $hostname = $options->{'hostname'};
    my $port     = $options->{'port'};
    my $username = $options->{'username'};
    my $password = $options->{'password'};
    my $protocol = $options->{'protocol'};

    my $prog = "fedora-dsinfo";
    my $arguments = "$hostname $port $username $password $pid $protocol";
    my $status = run_cmd($prog,$arguments,$verbosity,"tolerate_error");

    return $status;
}

sub run_purge
{
    my ($pid,$options) = @_;

    my $verbosity = $options->{'verbosity'};

    my $hostname = $options->{'hostname'};
    my $port     = $options->{'port'};
    my $username = $options->{'username'};
    my $password = $options->{'password'};
    my $protocol = $options->{'protocol'};

    my $server = "$hostname:$port";

    my $prog = "fedora-purge";
    my $arguments = "$server $username $password $pid $protocol";
    $arguments .= " \\\n \"Automated_purge_by_g2f_script\"";

    my $status = run_cmd($prog,$arguments,$verbosity);

    return $status;
}

sub run_ingest
{
    my ($docmets_filename,$options) = @_;

    my $verbosity = $options->{'verbosity'};

    my $hostname = $options->{'hostname'};
    my $port     = $options->{'port'};
    my $username = $options->{'username'};
    my $password = $options->{'password'};
    my $protocol = $options->{'protocol'};

    my $server = "$hostname:$port";

    my $prog = "fedora-ingest";

    my $type = undef;
    
    if ($ENV{'FEDORA2_HOME'}) {
    $type = "metslikefedora1";
    }
    else {
    $type = "info:fedora/fedora-system:METSFedoraExt-1.1";
    }

    my $arguments = "file \"$docmets_filename\" $type $server $username $password $protocol";
    $arguments .= " \\\n \"Automated_purge_by_g2f_script\"";

    my $status = run_cmd($prog,$arguments,$verbosity);

    return $status;
}



sub get_hash_id
{
    my ($hash_dir) = @_;

    my $hash_id = undef;

    my $docmets_filename = &util::filename_cat($hash_dir,"docmets.xml");

    if (open(DIN,"<$docmets_filename"))
    {
    while (defined (my $line = <DIN>))
    {
        if ($line =~ m/<dc:identifier>(.*?)<\/dc:identifier>/)
        {
        $hash_id = $1;
        last;
        }
    }
    
    close(DIN);
    }
    else
    {
    print STDERR "Warning: Unable to open \"$docmets_filename\"\n";
    }

    return $hash_id;

}


# Subroutine to write the gsdl.xml file in FEDORA_HOME/tomcat/conf/Catalina/<host/localhost>/
# This xml file will tell Fedora where to find the parent folder of the GS collect dir
# so that it can obtain the FedoraMETS files for ingestion. 
# It depends on the Fedora server being on the same machine as the Greenstone server that 
# this code is part of.
sub write_gsdl_xml_file
{
    my ($fedora_host, $collect_dir, $options) = @_;
    my $verbosity = $options->{'verbosity'};
    my $hostname = $options->{'hostname'};
    my $port     = $options->{'port'};
    my $protocol = $options->{'protocol'};

    print STDERR "Ensuring that a correct gsdl.xml file exists on the Fedora server end\n";
    # The top of this file has already made sure that FEDORA_HOME is set

    # 1. Find out which folder to write to: fedora_host or localhost 
    # whichever contains fedora.xml is the one we want (if none, exit with error value?)
    my $fedora_home = $ENV{'FEDORA_HOME'};
    my $base_path = &util::filename_cat($fedora_home, "tomcat", "conf", "Catalina");

    my $host_path = &util::filename_cat($base_path, $fedora_host);
    my $xmlFile = &util::filename_cat($host_path, "fedora.xml");
    if (!-e $xmlFile) {
    # check if the folder localhost contains fedoraXML
    $host_path = &util::filename_cat($base_path, "localhost");
    $xmlFile = &util::filename_cat($host_path, "fedora.xml");
    if(!-e $xmlFile) {
        # try putting gsdl in this folder, but still print a warning
        print STDERR "**** $host_path does not contain file fedora.xml. Hoping gsdl.xml belongs there anyway\n";
    }
    }

    # 2. Construct the string we are going write to the gsdl.xml file
    # a. get the parent directory of collect_dir by removinbg the word 
    # "collect" from it and any optional OS-type slash at the end. 
    # (Path slash direction does not matter here.)
    my $collectParentDir = $collect_dir;
    $collectParentDir =~ s/collect(\/|\\)?//;
  
    # b. Use the collectParentDir to create the contents of gsdl.xml
    my $gsdlXMLcontents = "<?xml version='1.0' encoding='utf-8'?>\n<Context docBase=\"";
    $gsdlXMLcontents = $gsdlXMLcontents.$collectParentDir."\" path=\"/gsdl\"></Context>";
    
    # 3. If there is already a gsdl.xml file in host_path, compare the string we
    # want to write with what is already in there. If they're the same, we can return
    $xmlFile = &util::filename_cat($host_path, "gsdl.xml");
    if(-e $xmlFile) {
    # such a file exists, so read the contents
    unless(open(FIN, "<$xmlFile")) { 
        print STDERR "***g2f-import.pl: Unable to open existing $xmlFile for comparing...Recoverable. $!\n";
        # doesn't matter, we'll just overwrite it then
    }    
    my $xml_contents;
    {
        local $/ = undef;        # Read entire file at once
        $xml_contents = <FIN>;   # Now file is read in as one single 'line'
    }
    close(FIN); # close the file
    if($xml_contents eq $gsdlXMLcontents) { 
        print STDERR "The old gsdl.xml file already contains the same.\n";
        # it already contains what we want, we're done
        return "gsdl.xml";
    }
    }

    # 4. If we're here, the contents of gsdl.xml need to be updated:
    # a. First stop the fedora server
    my $script_ext = ($ENV{'GSDLOS'} =~ m/^windows/) ? ".bat" : ".sh";
    my $stop_tomcat = &util::filename_cat($fedora_home, "tomcat", "bin", "shutdown".$script_ext);
    # execute the command 
    $!=0; # does this initialise the return value?
    my $status = system($stop_tomcat);
    if ($status!=0) { # to get the actual exit value, divide by 256, but not useful here
    # possible tomcat was already stopped - it's not the end of the world
    print STDERR "**** Failed to stop Fedora server. Perhaps it was not running. $!\n";
    print "Exit status = ", $status/256, "\n";
    }

    # b. overwrite the file that has outdated contents with the contents we just constructed
    unless(open(FOUT, ">$xmlFile")) {  # create or overwrite gsdl.xml file
    die "g2f-import.pl: Unable to open $xmlFile for telling Fedora where the collect dir is...ERROR: $!\n";
    }
    # write out the updated contents and close the file
    print FOUT $gsdlXMLcontents;
    close(FOUT);

    # c. Restart the fedora server
    my $start_tomcat = &util::filename_cat($fedora_home, "tomcat", "bin", "startup".$script_ext);
    $!=0;
    $status = system($start_tomcat);
    if ($status!=0) { 
    print STDERR "Failed to restart the Fedora server... ERROR: $!\n";
    print "Exit status = ", $status/256, "\n";
    }
   
    
    # Starting up the Fedora server takes a long time. We need to wait for the server to be
    # ready before import can continue, because g2f-import relies on an up-and-running Fedora
    # server to purge the collection from it while g2f-build.pl needs a ready Fedora server 
    # in order to make it ingest the FedoraMETS. Sleeping is not sufficient (#sleep 10;) since
    # the subsequent steps depend on a proper server restart.
    # Dr Bainbridge's suggestion: test the server is ready with a call to wget.
    
    # Wget tries to retrieve the fedora search page (protocol://host:port/fedora/search)
    # 20 times, waiting 3 seconds between each failed attempt. If it ultimately fails, we
    # print a message to the user.
    # The wget --spider option makes it check that the page is merely there rather than 
    # downloading it (see http://www.gnu.org/software/wget/manual/wget.html#Download-Options)
    # -q is for quiet, --tries for the number of retries, --waitretry is the number of seconds
    # between each attempt. Usually wget returns the contents of the page, but in our case it
    # will return 0 for success since we are not downloading.

    print STDERR "Fedora server restarted. Waiting for it to become ready...\n";
    
    # print STDERR "****$protocol://$hostname:$port/fedora/search\n";
    my $fedoraServerReady = system("wget -q --spider --waitretry=3 --tries=20 $protocol://$hostname:$port/fedora/search"); 
    if($fedoraServerReady != 0) {
    print STDERR "Fedora server is still not ready... ERROR: $!\n";
    print "Exit status = ", $status/256, "\n";
    die "Exiting....\n";
    }
    
    # return some indication that things went well
    return "gsdl.xml";
}


1;