source: main/trunk/greenstone2/perllib/util.pm@ 23484

###########################################################################
#
# util.pm -- various useful utilities
# A component of the Greenstone digital library software
# from the New Zealand Digital Library Project at the 
# University of Waikato, New Zealand.
#
# Copyright (C) 1999 New Zealand Digital Library Project
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
#
###########################################################################

package util;

use strict;

use Encode;
use File::Copy;
use File::Basename;

# removes files (but not directories)
sub rm {
    my (@files) = @_;

    my @filefiles = ();

    # make sure the files we want to delete exist 
    # and are regular files
    foreach my $file (@files) {
    if (!-e $file) {
        print STDERR "util::rm $file does not exist\n";
    } elsif ((!-f $file) && (!-l $file)) {
        print STDERR "util::rm $file is not a regular (or symbolic) file\n";
    } else {
        push (@filefiles, $file);
    }
    }
    
    # remove the files
    my $numremoved = unlink @filefiles;

    # check to make sure all of them were removed
    if ($numremoved != scalar(@filefiles)) {
    print STDERR "util::rm Not all files were removed\n";
    }
}

# removes files (but not directories) - can rename this to the default
# "rm" subroutine  when debugging the deletion of individual files.
sub rm_debug {
    my (@files) = @_;
    my @filefiles = ();

    # make sure the files we want to delete exist 
    # and are regular files
    foreach my $file (@files) {
        if (!-e $file) {
            print STDERR "util::rm $file does not exist\n";
        } elsif ((!-f $file) && (!-l $file)) {
            print STDERR "util::rm $file is not a regular (or symbolic) file\n";
        } else { # debug message
            unlink($file) or warn "Could not delete file $file: $!\n";
        }
    }
}


# recursive removal
sub filtered_rm_r {
    my ($files,$file_accept_re,$file_reject_re) = @_;

    my @files_array = (ref $files eq "ARRAY") ? @$files : ($files);

    # recursively remove the files
    foreach my $file (@files_array) {
    $file =~ s/[\/\\]+$//; # remove trailing slashes
    
    if (!-e $file) {
        print STDERR "util::filtered_rm_r $file does not exist\n";

    } elsif ((-d $file) && (!-l $file)) { # don't recurse down symbolic link
        # get the contents of this directory
        if (!opendir (INDIR, $file)) {
        print STDERR "util::filtered_rm_r could not open directory $file\n";
        } else {
        my @filedir = grep (!/^\.\.?$/, readdir (INDIR));
        closedir (INDIR);
                
        # remove all the files in this directory
        map {$_="$file/$_";} @filedir;
        &filtered_rm_r (\@filedir,$file_accept_re,$file_reject_re);

        if (!defined $file_accept_re && !defined $file_reject_re) {
            # remove this directory
            if (!rmdir $file) {
            print STDERR "util::filtered_rm_r couldn't remove directory $file\n";
            }
        }
        }
    } else {
        next if (defined $file_reject_re && ($file =~ m/$file_reject_re/));

        if ((!defined $file_accept_re) || ($file =~ m/$file_accept_re/)) {
        # remove this file  
        &rm ($file);
        }
    }
    }
}


# recursive removal
sub rm_r {
    my (@files) = @_;
    
    # use the more general (but reterospectively written function
    # filtered_rm_r function()

    filtered_rm_r(\@files,undef,undef); # no accept or reject expressions
}




# moves a file or a group of files
sub mv {
    my $dest = pop (@_);
    my (@srcfiles) = @_;

    # remove trailing slashes from source and destination files
    $dest =~ s/[\\\/]+$//;
    map {$_ =~ s/[\\\/]+$//;} @srcfiles;

    # a few sanity checks
    if (scalar (@srcfiles) == 0) {
    print STDERR "util::mv no destination directory given\n";
    return;
    } elsif ((scalar (@srcfiles) > 1) && (!-d $dest)) {
    print STDERR "util::mv if multiple source files are given the ".
        "destination must be a directory\n";
    return;
    }

    # move the files
    foreach my $file (@srcfiles) {
    my $tempdest = $dest;
    if (-d $tempdest) {
        my ($filename) = $file =~ /([^\\\/]+)$/;
        $tempdest .= "/$filename";
    }
    if (!-e $file) {
        print STDERR "util::mv $file does not exist\n";
    } else {
        rename ($file, $tempdest);
    }
    }
}


# copies a file or a group of files
sub cp {
    my $dest = pop (@_);
    my (@srcfiles) = @_;

    # remove trailing slashes from source and destination files
    $dest =~ s/[\\\/]+$//;
    map {$_ =~ s/[\\\/]+$//;} @srcfiles;

    # a few sanity checks
    if (scalar (@srcfiles) == 0) {
    print STDERR "util::cp no destination directory given\n";
    return;
    } elsif ((scalar (@srcfiles) > 1) && (!-d $dest)) {
    print STDERR "util::cp if multiple source files are given the ".
        "destination must be a directory\n";
    return;
    }

    # copy the files
    foreach my $file (@srcfiles) {
    my $tempdest = $dest;
    if (-d $tempdest) {
        my ($filename) = $file =~ /([^\\\/]+)$/;
        $tempdest .= "/$filename";
    }
    if (!-e $file) {
        print STDERR "util::cp $file does not exist\n";
    } elsif (!-f $file) {
        print STDERR "util::cp $file is not a plain file\n";
    } else {
        &File::Copy::copy ($file, $tempdest);
    }
    }
}



# recursively copies a file or group of files
# syntax: cp_r (sourcefiles, destination directory)
# destination must be a directory - to copy one file to
# another use cp instead
sub cp_r {
    my $dest = pop (@_);
    my (@srcfiles) = @_;

    # a few sanity checks
    if (scalar (@srcfiles) == 0) {
    print STDERR "util::cp_r no destination directory given\n";
    return;
    } elsif (-f $dest) {
    print STDERR "util::cp_r destination must be a directory\n";
    return;
    }
    
    # create destination directory if it doesn't exist already
    if (! -d $dest) {
    my $store_umask = umask(0002);
    mkdir ($dest, 0777);
    umask($store_umask);
    } 

    # copy the files
    foreach my $file (@srcfiles) {

    if (!-e $file) {
        print STDERR "util::cp_r $file does not exist\n";

    } elsif (-d $file) {
        # make the new directory
        my ($filename) = $file =~ /([^\\\/]*)$/;
        $dest = &util::filename_cat ($dest, $filename);
        my $store_umask = umask(0002);
        mkdir ($dest, 0777);
        umask($store_umask);

        # get the contents of this directory
        if (!opendir (INDIR, $file)) {
        print STDERR "util::cp_r could not open directory $file\n";
        } else {
        my @filedir = readdir (INDIR);
        closedir (INDIR);
        foreach my $f (@filedir) {
            next if $f =~ /^\.\.?$/;
            # copy all the files in this directory
            my $ff = &util::filename_cat ($file, $f); 
            &cp_r ($ff, $dest);
        }
        }

    } else {
        &cp($file, $dest);
    }
    }
}
# recursively copies a file or group of files
# syntax: cp_r (sourcefiles, destination directory)
# destination must be a directory - to copy one file to
# another use cp instead
sub cp_r_nosvn {
    my $dest = pop (@_);
    my (@srcfiles) = @_;

    # a few sanity checks
    if (scalar (@srcfiles) == 0) {
    print STDERR "util::cp_r no destination directory given\n";
    return;
    } elsif (-f $dest) {
    print STDERR "util::cp_r destination must be a directory\n";
    return;
    }
    
    # create destination directory if it doesn't exist already
    if (! -d $dest) {
    my $store_umask = umask(0002);
    mkdir ($dest, 0777);
    umask($store_umask);
    } 

    # copy the files
    foreach my $file (@srcfiles) {

    if (!-e $file) {
        print STDERR "util::cp_r $file does not exist\n";

    } elsif (-d $file) {
        # make the new directory
        my ($filename) = $file =~ /([^\\\/]*)$/;
        $dest = &util::filename_cat ($dest, $filename);
        my $store_umask = umask(0002);
        mkdir ($dest, 0777);
        umask($store_umask);

        # get the contents of this directory
        if (!opendir (INDIR, $file)) {
        print STDERR "util::cp_r could not open directory $file\n";
        } else {
        my @filedir = readdir (INDIR);
        closedir (INDIR);
        foreach my $f (@filedir) {
            next if $f =~ /^\.\.?$/;
            next if $f =~ /^\.svn$/;
            # copy all the files in this directory
            my $ff = &util::filename_cat ($file, $f); 
            &cp_r ($ff, $dest);
        }
        }

    } else {
        &cp($file, $dest);
    }
    }
}

# copies a directory and its contents, excluding subdirectories, into a new directory
sub cp_r_toplevel {
    my $dest = pop (@_);
    my (@srcfiles) = @_;

    # a few sanity checks
    if (scalar (@srcfiles) == 0) {
    print STDERR "util::cp_r no destination directory given\n";
    return;
    } elsif (-f $dest) {
    print STDERR "util::cp_r destination must be a directory\n";
    return;
    }
    
    # create destination directory if it doesn't exist already
    if (! -d $dest) {
    my $store_umask = umask(0002);
    mkdir ($dest, 0777);
    umask($store_umask);
    } 

    # copy the files
    foreach my $file (@srcfiles) {

    if (!-e $file) {
        print STDERR "util::cp_r $file does not exist\n";

    } elsif (-d $file) {
        # make the new directory
        my ($filename) = $file =~ /([^\\\/]*)$/;
        $dest = &util::filename_cat ($dest, $filename);
        my $store_umask = umask(0002);
        mkdir ($dest, 0777);
        umask($store_umask);

        # get the contents of this directory
        if (!opendir (INDIR, $file)) {
        print STDERR "util::cp_r could not open directory $file\n";
        } else {
        my @filedir = readdir (INDIR);
        closedir (INDIR);
        foreach my $f (@filedir) {
            next if $f =~ /^\.\.?$/;
            
            # copy all the files in this directory, but not directories
            my $ff = &util::filename_cat ($file, $f); 
            if (-f $ff) {
            &cp($ff, $dest);
            #&cp_r ($ff, $dest);
            }
        }
        }

    } else {
        &cp($file, $dest);
    }
    }
}

sub mk_dir {
    my ($dir) = @_;

    my $store_umask = umask(0002);
    my $mkdir_ok = mkdir ($dir, 0777);
    umask($store_umask);
    
    if (!$mkdir_ok) 
    {
    print STDERR "util::mk_dir could not create directory $dir\n";
    return;
    }
}

# in case anyone cares - I did some testing (using perls Benchmark module)
# on this subroutine against File::Path::mkpath (). mk_all_dir() is apparently
# slightly faster (surprisingly) - Stefan.
sub mk_all_dir {
    my ($dir) = @_;

    # use / for the directory separator, remove duplicate and
    # trailing slashes
    $dir=~s/[\\\/]+/\//g; 
    $dir=~s/[\\\/]+$//;

    # make sure the cache directory exists
    my $dirsofar = "";
    my $first = 1;
    foreach my $dirname (split ("/", $dir)) {
    $dirsofar .= "/" unless $first;
    $first = 0;

    $dirsofar .= $dirname;

    next if $dirname =~ /^(|[a-z]:)$/i;
    if (!-e $dirsofar)
        {
        my $store_umask = umask(0002);
        my $mkdir_ok = mkdir ($dirsofar, 0777);
        umask($store_umask);
        if (!$mkdir_ok)
        {
            print STDERR "util::mk_all_dir could not create directory $dirsofar\n";
            return;
        }
        }
    }
}

# make hard link to file if supported by OS, otherwise copy the file
sub hard_link {
    my ($src, $dest, $verbosity) = @_;

    # remove trailing slashes from source and destination files
    $src =~ s/[\\\/]+$//;
    $dest =~ s/[\\\/]+$//;

##    print STDERR "**** src = ", unicode::debug_unicode_string($src),"\n";
    # a few sanity checks
    if (-e $dest) {
    # destination file already exists
    return;
    }
    elsif (!-e $src) {
    print STDERR "util::hard_link source file \"$src\" does not exist\n";
    return 1;
    }
    elsif (-d $src) {
    print STDERR "util::hard_link source \"$src\" is a directory\n";
    return 1;
    }

    my $dest_dir = &File::Basename::dirname($dest);
    mk_all_dir($dest_dir) if (!-e $dest_dir);


    if (!link($src, $dest)) {
    if ((!defined $verbosity) || ($verbosity>2)) {
        print STDERR "util::hard_link: unable to create hard link. ";
        print STDERR " Copying file: $src -> $dest\n";
    }
    &File::Copy::copy ($src, $dest);
    }
    return 0;
}

# make soft link to file if supported by OS, otherwise copy file
sub soft_link {
    my ($src, $dest, $ensure_paths_absolute) = @_;

    # remove trailing slashes from source and destination files
    $src =~ s/[\\\/]+$//;
    $dest =~ s/[\\\/]+$//;

    # Ensure file paths are absolute IF requested to do so 
    # Soft_linking didn't work for relative paths
    if(defined $ensure_paths_absolute && $ensure_paths_absolute) {
    # We need to ensure that the src file is the absolute path 
    # See http://perldoc.perl.org/File/Spec.html
    if(!File::Spec->file_name_is_absolute( $src ))  { # it's relative
        $src = File::Spec->rel2abs($src); # make absolute
    }
    # Might as well ensure that the destination file's absolute path is used
    if(!File::Spec->file_name_is_absolute( $dest )) {
        $dest = File::Spec->rel2abs($dest); # make absolute
    }
    }

    # a few sanity checks
    if (!-e $src) {
    print STDERR "util::soft_link source file $src does not exist\n";
    return 0;
    }

    my $dest_dir = &File::Basename::dirname($dest);
    mk_all_dir($dest_dir) if (!-e $dest_dir);

    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
        
    # symlink not supported on windows
    &File::Copy::copy ($src, $dest);

    } elsif (!eval {symlink($src, $dest)}) {
    print STDERR "util::soft_link: unable to create soft link.\n";
    return 0;
    }

    return 1;
}

# Primarily for filenames generated by processing
# content of HTML files (which are mapped to UTF-8 internally)
#
# To turn this into an octet string that really exists on the file
# system:
# 1. don't need to do anything special for Unix-based systems
#   (as underlying file system is byte-code)
# 2. need to map to short DOS filenames for Windows

sub utf8_to_real_filename
{
    my ($utf8_filename) = @_;

    my $real_filename;

    if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
    require Win32;

    print STDERR "***** utf8 filename = $utf8_filename\n\n\n";

    my $unicode_filename = decode("utf8",$utf8_filename);
    $real_filename = Win32::GetShortPathName($unicode_filename);
    }
    else {
    $real_filename = $utf8_filename;
    }

    return $real_filename;
}


sub fd_exists
{
    my $filename_full_path = shift @_;
    my $test_op = shift @_ || "-e";

    # By default tests for existance of file or directory (-e)
    # Can be made more specific by providing second parameter (e.g. -f or -d)

    my $exists = 0;

    if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
    require Win32;
    my $filename_short_path = Win32::GetShortPathName($filename_full_path); 
    if (!defined $filename_short_path) {
        # Was probably still in UTF8 form (not what is needed on Windows)
        my $unicode_filename_full_path = eval "decode(\"utf8\",\$filename_full_path)";
        if (defined $unicode_filename_full_path) {
        $filename_short_path = Win32::GetShortPathName($unicode_filename_full_path);
        }
    }
    $filename_full_path = $filename_short_path;
    }

    if (defined $filename_full_path) {
    $exists = eval "($test_op \$filename_full_path)";
    }

    return $exists;
}

sub file_exists
{
    my ($filename_full_path) = @_;

    return fd_exists($filename_full_path,"-f");
}

sub dir_exists
{
    my ($filename_full_path) = @_;

    return fd_exists($filename_full_path,"-d");
}



# updates a copy of a directory in some other part of the filesystem
# verbosity settings are: 0=low, 1=normal, 2=high
# both $fromdir and $todir should be absolute paths
sub cachedir {
    my ($fromdir, $todir, $verbosity) = @_;
    $verbosity = 1 unless defined $verbosity;

    # use / for the directory separator, remove duplicate and
    # trailing slashes
    $fromdir=~s/[\\\/]+/\//g; 
    $fromdir=~s/[\\\/]+$//;
    $todir=~s/[\\\/]+/\//g; 
    $todir=~s/[\\\/]+$//;

    &mk_all_dir ($todir);

    # get the directories in ascending order
    if (!opendir (FROMDIR, $fromdir)) {
    print STDERR "util::cachedir could not read directory $fromdir\n";
    return;
    }
    my @fromdir = grep (!/^\.\.?$/, sort(readdir (FROMDIR)));
    closedir (FROMDIR);

    if (!opendir (TODIR, $todir)) {
    print STDERR "util::cacedir could not read directory $todir\n";
    return;
    }
    my @todir = grep (!/^\.\.?$/, sort(readdir (TODIR)));
    closedir (TODIR);

    my $fromi = 0;
    my $toi = 0;
            
    while ($fromi < scalar(@fromdir) || $toi < scalar(@todir)) {
#   print "fromi: $fromi toi: $toi\n";

    # see if we should delete a file/directory
    # this should happen if the file/directory
    # is not in the from list or if its a different
    # size, or has an older timestamp
    if ($toi < scalar(@todir)) {
        if (($fromi >= scalar(@fromdir)) ||
        ($todir[$toi] lt $fromdir[$fromi] || 
         ($todir[$toi] eq $fromdir[$fromi] && 
          &differentfiles("$fromdir/$fromdir[$fromi]","$todir/$todir[$toi]",
                  $verbosity)))) {

        # the files are different
        &rm_r("$todir/$todir[$toi]");
        splice(@todir, $toi, 1); # $toi stays the same

        } elsif ($todir[$toi] eq $fromdir[$fromi]) {
        # the files are the same
        # if it is a directory, check its contents
        if (-d "$todir/$todir[$toi]") {
            &cachedir ("$fromdir/$fromdir[$fromi]",
                   "$todir/$todir[$toi]", $verbosity);
        }

        $toi++;
        $fromi++;
        next;
        }
    }
  
    # see if we should insert a file/directory
    # we should insert a file/directory if there
    # is no tofiles left or if the tofile does not exist
    if ($fromi < scalar(@fromdir) && ($toi >= scalar(@todir) || 
                      $todir[$toi] gt $fromdir[$fromi])) {
        &cp_r ("$fromdir/$fromdir[$fromi]", "$todir/$fromdir[$fromi]");
        splice (@todir, $toi, 0, $fromdir[$fromi]);

        $toi++;
        $fromi++;
    }
    }
}

# this function returns -1 if either file is not found
# assumes that $file1 and $file2 are absolute file names or
# in the current directory
# $file2 is allowed to be newer than $file1
sub differentfiles {
    my ($file1, $file2, $verbosity) = @_;
    $verbosity = 1 unless defined $verbosity;

    $file1 =~ s/\/+$//;
    $file2 =~ s/\/+$//;
    
    my ($file1name) = $file1 =~ /\/([^\/]*)$/;
    my ($file2name) = $file2 =~ /\/([^\/]*)$/;

    return -1 unless (-e $file1 && -e $file2);
    if ($file1name ne $file2name) {
    print STDERR "filenames are not the same\n" if ($verbosity >= 2);
    return 1;
    }

    my @file1stat = stat ($file1);
    my @file2stat = stat ($file2);

    if (-d $file1) {
    if (! -d $file2) {
        print STDERR "one file is a directory\n" if ($verbosity >= 2);
        return 1;
    }
    return 0;
    }

    # both must be regular files
    unless (-f $file1 && -f $file2) {
    print STDERR "one file is not a regular file\n" if ($verbosity >= 2);
    return 1;
    }

    # the size of the files must be the same
    if ($file1stat[7] != $file2stat[7]) {
    print STDERR "different sized files\n" if ($verbosity >= 2);
    return 1;
    }

    # the second file cannot be older than the first
    if ($file1stat[9] > $file2stat[9]) {
    print STDERR "file is older\n" if ($verbosity >= 2);
    return 1;
    }

    return 0;
}


sub get_tmp_filename 
{
    my $file_ext = shift(@_) || undef;

    my $opt_dot_file_ext = "";
    if (defined $file_ext) {
    if ($file_ext !~ m/\./) {
        # no dot, so needs one added in at start
        $opt_dot_file_ext = ".$file_ext"
    }
    else {
        # allow for "extensions" such as _metadata.txt to be handled
        # gracefully
        $opt_dot_file_ext = $file_ext;
    }
    }

    my $tmpdir = filename_cat($ENV{'GSDLHOME'}, "tmp");
    &mk_all_dir ($tmpdir) unless -e $tmpdir;

    my $count = 1000;
    my $rand = int(rand $count);
    my $full_tmp_filename = &filename_cat($tmpdir, "F$rand$opt_dot_file_ext");

    while (-e $full_tmp_filename) {
    $rand = int(rand $count);
    $full_tmp_filename = &filename_cat($tmpdir, "F$rand$opt_dot_file_ext");
    $count++;
    }
    
    return $full_tmp_filename;
}

sub get_timestamped_tmp_folder
{

    my $tmp_dirname;
    if(defined $ENV{'GSDLCOLLECTDIR'}) {
    $tmp_dirname = $ENV{'GSDLCOLLECTDIR'};
    } elsif(defined $ENV{'GSDLHOME'}) {
    $tmp_dirname = $ENV{'GSDLHOME'};
    } else {
    return undef;
    }

    $tmp_dirname = &util::filename_cat($tmp_dirname, "tmp");
    &util::mk_dir($tmp_dirname) if (!-e $tmp_dirname);

    # add the timestamp into the path otherwise we can run into problems 
    # if documents have the same name
    my $timestamp = time;
    my $time_tmp_dirname = &util::filename_cat($tmp_dirname, $timestamp);
    $tmp_dirname = $time_tmp_dirname;
    my $i = 1;
    while (-e $tmp_dirname) {
    $tmp_dirname = "$time_tmp_dirname$i";
    $i++;
    }
    &util::mk_dir($tmp_dirname);
 
    return $tmp_dirname;
}

sub get_timestamped_tmp_filename_in_collection
{

    my ($input_filename, $output_ext) = @_;
    # derive tmp filename from input filename
    my ($tailname, $dirname, $suffix)
    = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");

    # softlink to collection tmp dir
    my $tmp_dirname = &util::get_timestamped_tmp_folder();
    $tmp_dirname = $dirname unless defined $tmp_dirname;

    # following two steps copied from ConvertBinaryFile
    # do we need them?? can't use them as is, as they use plugin methods.

    #$tailname = $self->SUPER::filepath_to_utf8($tailname) unless &unicode::check_is_utf8($tailname);

    # URLEncode this since htmls with images where the html filename is utf8 don't seem
    # to work on Windows (IE or Firefox), as browsers are looking for filesystem-encoded
    # files on the filesystem.
    #$tailname = &util::rename_file($tailname, $self->{'file_rename_method'}, "without_suffix");
    if (defined $output_ext) {
    $output_ext = ".$output_ext"; # add the dot
    } else {
    $output_ext = $suffix;
    }
    $output_ext= lc($output_ext);
    my $tmp_filename = &util::filename_cat($tmp_dirname, "$tailname$output_ext");
    
    return $tmp_filename;
}

sub get_toplevel_tmp_dir
{
    return filename_cat($ENV{'GSDLHOME'}, "tmp");
}


sub filename_to_regex {
    my $filename = shift (@_);

    # need to put single backslash back to double so that regex works
    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
    $filename =~ s/\\/\\\\/g;
    }
    return $filename;
}

sub filename_cat {
    my $first_file = shift(@_);
    my (@filenames) = @_;

#   Useful for debugging 
#     -- might make sense to call caller(0) rather than (1)??
#   my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(1);
#   print STDERR "Calling method: $cfilename:$cline $cpackage->$csubr\n";
   
    # If first_file is not null or empty, then add it back into the list
    if (defined $first_file && $first_file =~ /\S/) {
    unshift(@filenames, $first_file);
    }

    my $filename = join("/", @filenames);

    # remove duplicate slashes and remove the last slash
    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
    $filename =~ s/[\\\/]+/\\/g;
    } else {
    $filename =~ s/[\/]+/\//g; 
    # DB: want a filename abc\de.html to remain like this
    }
    $filename =~ s/[\\\/]$//;

    return $filename;
}


sub pathname_cat {
    my $first_path = shift(@_);
    my (@pathnames) = @_;

    # If first_path is not null or empty, then add it back into the list
    if (defined $first_path && $first_path =~ /\S/) {
    unshift(@pathnames, $first_path);
    }

    my $join_char;
    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
    $join_char = ";";
    } else {
    $join_char = ":";
    }

    my $pathname = join($join_char, @pathnames);

    # remove duplicate slashes
    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
    $pathname =~ s/[\\\/]+/\\/g;
    } else {
    $pathname =~ s/[\/]+/\//g; 
    # DB: want a pathname abc\de.html to remain like this
    }

    return $pathname;
}


sub tidy_up_oid {
    my ($OID) = @_;
    if ($OID =~ /\./) {
    print STDERR "Warning, identifier $OID contains periods (.), removing them\n";
    $OID =~ s/\.//g; #remove any periods
    }
    if ($OID =~ /^\s.*\s$/) {
    print STDERR "Warning, identifier $OID starts or ends with whitespace. Removing it\n";
    # remove starting and trailing whitespace
    $OID =~ s/^\s+//;
    $OID =~ s/\s+$//;
    }
    if ($OID =~ /^[\d]*$/) {
    print STDERR "Warning, identifier $OID contains only digits. Prepending 'D'.\n";
    $OID = "D" . $OID;
    }       
    
    return $OID;
}
sub envvar_prepend {
    my ($var,$val) = @_;

    # do not prepend any value/path that's already in the environment variable
    if ($ENV{'GSDLOS'} =~ /^windows$/i) 
    {
    my $escaped_val = $val;
    $escaped_val =~ s/\\/\\\\/g; # escape any Windows backslashes for upcoming regex
    if (!defined($ENV{$var})) {
        $ENV{$var} = "$val";
    }
    elsif($ENV{$var} !~ m/$escaped_val/) { 
        $ENV{$var} = "$val;".$ENV{$var};
    }
    }
    else {
    if (!defined($ENV{$var})) {
        $ENV{$var} = "$val";
    }
    elsif($ENV{$var} !~ m/$val/) { 
        $ENV{$var} = "$val:".$ENV{$var};
    }
    }
}

sub envvar_append {
    my ($var,$val) = @_;

    # do not append any value/path that's already in the environment variable
    if ($ENV{'GSDLOS'} =~ /^windows$/i) 
    {
    my $escaped_val = $val;
    $escaped_val =~ s/\\/\\\\/g; # escape any Windows backslashes for upcoming regex
    if (!defined($ENV{$var})) {
        $ENV{$var} = "$val";
    }
    elsif($ENV{$var} !~ m/$escaped_val/) { 
        $ENV{$var} .= ";$val";
    }
    }
    else {
    if (!defined($ENV{$var})) {
        $ENV{$var} = "$val";
    }
    elsif($ENV{$var} !~ m/$val/) { 
        $ENV{$var} .= ":$val";
    }
    }    
}


# splits a filename into a prefix and a tail extension using the tail_re, or 
# if that fails, splits on the file_extension . (dot) 
sub get_prefix_and_tail_by_regex {

    my ($filename,$tail_re) = @_;
    
    my ($file_prefix,$file_ext) = ($filename =~ m/^(.*?)($tail_re)$/);
    if ((!defined $file_prefix) || (!defined $file_ext)) {
    ($file_prefix,$file_ext) = ($filename =~ m/^(.*)(\..*?)$/);
    }

    return ($file_prefix,$file_ext);
}

# get full path and file only path from a base_dir (which may be empty) and 
# file (which may contain directories)
sub get_full_filenames {
    my ($base_dir, $file) = @_;
    
    my $filename_full_path = $file;
    # add on directory if present
    $filename_full_path = &util::filename_cat ($base_dir, $file) if $base_dir =~ /\S/;
    
    my $filename_no_path = $file;

    # remove directory if present
    $filename_no_path =~ s/^.*[\/\\]//;
    return ($filename_full_path, $filename_no_path);
}

# returns the path of a file without the filename -- ie. the directory the file is in
sub filename_head {
    my $filename = shift(@_);

    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
    $filename =~ s/[^\\\\]*$//;
    }
    else {
    $filename =~ s/[^\\\/]*$//;
    }

    return $filename;
}



# returns 1 if filename1 and filename2 point to the same
# file or directory
sub filenames_equal {
    my ($filename1, $filename2) = @_;

    # use filename_cat to clean up trailing slashes and 
    # multiple slashes
    $filename1 = filename_cat ($filename1);
    $filename2 = filename_cat ($filename2);

    # filenames not case sensitive on windows
    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
    $filename1 =~ tr/[A-Z]/[a-z]/;
    $filename2 =~ tr/[A-Z]/[a-z]/;
    }
    return 1 if $filename1 eq $filename2;
    return 0;
}


sub filename_within_directory
{
    my ($filename,$within_dir) = @_;
    
    if ($within_dir !~ m/[\/\\]$/) {
    my $dirsep = &util::get_dirsep();
    $within_dir .= $dirsep;
    }
    
    $within_dir =~ s/\\/\\\\/g; # escape DOS style file separator

    if ($filename =~ m/^$within_dir(.*)$/) {
    $filename = $1;
    }
    
    return $filename;
}

sub filename_within_collection
{
    my ($filename) = @_;

    my $collect_dir = $ENV{'GSDLCOLLECTDIR'};
    
    if (defined $collect_dir) {

    # if from within GSDLCOLLECTDIR, then remove directory prefix
    # so source_filename is realative to it.  This is done to aid
    # portability, i.e. the collection can be moved to somewhere
    # else on the file system and the archives directory will still
    # work.  This is needed, for example in the applet version of
    # GLI where GSDLHOME/collect on the server will be different to
    # the collect directory of the remove user.  Of course,
    # GSDLCOLLECTDIR subsequently needs to be put back on to turn
    # it back into a full pathname.

    $filename = filename_within_directory($filename,$collect_dir);
    }
    
    return $filename;
}

sub prettyprint_file
{
    my ($base_dir,$file,$gli) = @_;

    my $filename_full_path = &util::filename_cat($base_dir,$file);

    if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
    require Win32;

    # For some reason base_dir in the form c:/a/b/c
    # This leads to confusion later on, so turn it back into
    # the more usual Windows form
    $base_dir =~ s/\//\\/g; 
    my $long_base_dir = Win32::GetLongPathName($base_dir);
    my $long_full_path = Win32::GetLongPathName($filename_full_path);

    $file = filename_within_directory($long_full_path,$long_base_dir);
    $file = encode("utf8",$file) if ($gli);
    }

    return $file;
}


sub upgrade_if_dos_filename
{
    my ($filename_full_path,$and_encode) = @_;

    if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
    # Ensure any DOS-like filename, such as test~1.txt, has been upgraded
    # to its long (Windows) version
    my $long_filename = Win32::GetLongPathName($filename_full_path);
    if (defined $long_filename) {
        $filename_full_path = $long_filename;
    }
    # Make sure initial drive letter is lower-case (to fit in with rest of Greenstone)
    $filename_full_path =~ s/^(.):/\u$1:/;
    if ((defined $and_encode) && ($and_encode)) {
        $filename_full_path = encode("utf8",$filename_full_path);
    }
    }

    return $filename_full_path;
}


sub downgrade_if_dos_filename
{
    my ($filename_full_path) = @_;

    if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
    require Win32;

    # Ensure the given long Windows filename is in a form that can
    # be opened by Perl => convert it to a short DOS-like filename

    my $short_filename = Win32::GetShortPathName($filename_full_path);
    if (defined $short_filename) {
        $filename_full_path = $short_filename;
    }
    # Make sure initial drive letter is lower-case (to fit in
    # with rest of Greenstone)
    $filename_full_path =~ s/^(.):/\u$1:/;
    }

    return $filename_full_path;
}


sub filename_is_absolute
{
    my ($filename) = @_;

    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
    return ($filename =~ m/^(\w:)?\\/);
    }
    else {
    return ($filename =~ m/^\//);
    }
}


## @method make_absolute()
#
#  Ensure the given file path is absolute in respect to the given base path.
#
#  @param  $base_dir A string denoting the base path the given dir must be
#                    absolute to.
#  @param  $dir The directory to be made absolute as a string. Note that the
#               dir may already be absolute, in which case it will remain
#               unchanged.
#  @return The now absolute form of the directory as a string.
#
#  @author John Thompson, DL Consulting Ltd.
#  @copy 2006 DL Consulting Ltd.
#
#used in buildcol.pl, doesn't work for all cases --kjdon
sub make_absolute {
    
    my ($base_dir, $dir) = @_;
###    print STDERR "dir = $dir\n";
    $dir =~ s/[\\\/]+/\//g;
    $dir = $base_dir . "/$dir" unless ($dir =~ m|^(\w:)?/|); 
    $dir =~ s|^/tmp_mnt||;
    1 while($dir =~ s|/[^/]*/\.\./|/|g);
    $dir =~ s|/[.][.]?/|/|g;
    $dir =~ tr|/|/|s;
###    print STDERR "dir = $dir\n";
    
    return $dir;
}
## make_absolute() ##

sub get_dirsep {

    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
    return "\\";
    } else {
    return "\/";
    }
}

sub get_os_dirsep {

    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
    return "\\\\";
    } else {
    return "\\\/";
    }
}

sub get_re_dirsep {

    return "\\\\|\\\/";
}


sub get_dirsep_tail {
    my ($filename) = @_;
    
    # returns last part of directory or filename
    # On unix e.g. a/b.d => b.d
    #              a/b/c => c

    my $dirsep = get_re_dirsep();
    my @dirs = split (/$dirsep/, $filename);
    my $tail = pop @dirs;

    # - caused problems under windows
    #my ($tail) = ($filename =~ m/^(?:.*?$dirsep)?(.*?)$/); 

    return $tail;
}


# if this is running on windows we want binaries to end in
# .exe, otherwise they don't have to end in any extension
sub get_os_exe {
    return ".exe" if $ENV{'GSDLOS'} =~ /^windows$/i;
    return "";
}


# test to see whether this is a big or little endian machine
sub is_little_endian
{
    # To determine the name of the operating system, the variable $^O is a cheap alternative to pulling it out of the Config module;
    # If it is a Macintosh machine (i.e. the Darwin operating system), regardless if it's running on the IBM power-pc cpu or the x86 Intel-based chip with a power-pc emulator running on top of it, it's big-endian
    # Otherwise, it's little endian

    #return 0 if $^O =~ /^darwin$/i;
    #return 0 if $ENV{'GSDLOS'} =~ /^darwin$/i;
    
    # Going back to stating exactly whether the machine is little endian
    # or big endian, without any special case for Macs. Since for rata it comes
    # back with little endian and for shuttle with bigendian.
    return (ord(substr(pack("s",1), 0, 1)) == 1);
}


# will return the collection name if successful, "" otherwise
sub use_collection {
    my ($collection, $collectdir) = @_;

    if (!defined $collectdir || $collectdir eq "") {
    $collectdir = &filename_cat ($ENV{'GSDLHOME'}, "collect");
    }

    # get and check the collection
    if (!defined($collection) || $collection eq "") {
    if (defined $ENV{'GSDLCOLLECTION'}) {
        $collection = $ENV{'GSDLCOLLECTION'};
    } else {
        print STDOUT "No collection specified\n";
        return "";
    }
    }
    
    if ($collection eq "modelcol") {
    print STDOUT "You can't use modelcol.\n";
    return "";
    }

    # make sure the environment variables GSDLCOLLECTION and GSDLCOLLECTDIR
    # are defined
    $ENV{'GSDLCOLLECTION'} = $collection;
    $ENV{'GSDLCOLLECTDIR'} = &filename_cat ($collectdir, $collection);

    # make sure this collection exists
    if (!-e $ENV{'GSDLCOLLECTDIR'}) {
    print STDOUT "Invalid collection ($collection).\n";
    return "";
    }

    # everything is ready to go
    return $collection;
}

sub get_current_collection_name {
    return $ENV{'GSDLCOLLECTION'};
}


# will return the collection name if successful, "" otherwise.  
# Like use_collection (above) but for greenstone 3 (taking account of site level)

sub use_site_collection {
    my ($site, $collection, $collectdir) = @_;

    if (!defined $collectdir || $collectdir eq "") {
    die "GSDL3HOME not set.\n" unless defined $ENV{'GSDL3HOME'};
    $collectdir = &filename_cat ($ENV{'GSDL3HOME'}, "sites", $site, "collect");
    }

    # collectdir explicitly set by this point (using $site variable if required).
    # Can call "old" gsdl2 use_collection now.

    return use_collection($collection,$collectdir);
}



sub locate_config_file
{
    my ($file) = @_;

    my $locations = locate_config_files($file);

    return shift @$locations; # returns undef if 'locations' is empty
}


sub locate_config_files
{
    my ($file) = @_;

    my @locations = ();

    if (-e $file) {
    # Clearly specified (most likely full filename)
    # No need to hunt in 'etc' directories, return value unchanged
    push(@locations,$file);
    }
    else {
    # Check for collection specific one before looking in global GSDL 'etc'
    if (defined $ENV{'GSDLCOLLECTDIR'} && $ENV{'GSDLCOLLECTDIR'} ne "") {
        my $test_collect_etc_filename 
        = &util::filename_cat($ENV{'GSDLCOLLECTDIR'},"etc", $file);
        
        if (-e $test_collect_etc_filename) {
        push(@locations,$test_collect_etc_filename);
        }
    }
    my $test_main_etc_filename 
        = &util::filename_cat($ENV{'GSDLHOME'},"etc", $file);
    if (-e $test_main_etc_filename) {
        push(@locations,$test_main_etc_filename);
    }
    }

    return \@locations;
}


sub hyperlink_text
{
    my ($text) = @_;
    
    $text =~ s/(http:\/\/[^\s]+)/<a href=\"$1\">$1<\/a>/mg;
    $text =~ s/(^|\s+)(www\.(\w|\.)+)/<a href=\"http:\/\/$2\">$2<\/a>/mg;

    return $text;
}


# A method to check if a directory is empty (note that an empty directory still has non-zero size!!!) 
# Code is from http://episteme.arstechnica.com/eve/forums/a/tpc/f/6330927813/m/436007700831
sub is_dir_empty
{
    my ($path) = @_;
    opendir DIR, $path;
    while(my $entry = readdir DIR) {
        next if($entry =~ /^\.\.?$/);
        closedir DIR;
        return 0;
    }
    closedir DIR;
    return 1;
}

# Returns the given filename converted using either URL encoding or base64
# encoding, as specified by $rename_method. If the given filename has no suffix
# (if it is just the tailname), then $no_suffix should be some defined value.
# rename_method can be url, none, base64 
sub rename_file {
    my ($filename, $rename_method, $no_suffix)  = @_;

    if(!$filename) { # undefined or empty string
    return $filename;
    }

    if (!$rename_method) {
    print STDERR "WARNING: no file renaming method specified. Defaulting to using URL encoding...\n";
    # Debugging information
    # my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(1);
    # print STDERR "Called from method: $cfilename:$cline $cpackage->$csubr\n";
    $rename_method = "url";
    } elsif($rename_method eq "none") {
    return $filename; # would have already been renamed
    }

    # No longer replace spaces with underscores, since underscores mess with incremental rebuild
    ### Replace spaces with underscore. Do this first else it can go wrong below when getting tailname
    ###$filename =~ s/ /_/g;

    my ($tailname,$dirname,$suffix); 
    if($no_suffix) { # given a tailname, no suffix
    ($tailname,$dirname) = File::Basename::fileparse($filename);
    } 
    else {
    ($tailname,$dirname,$suffix) = File::Basename::fileparse($filename, "\\.(?:[^\\.]+?)\$");
    }
    if (!$suffix) {
    $suffix = "";
    }
    else {
    $suffix = lc($suffix);
    }

    if ($rename_method eq "url") {
    $tailname = &unicode::url_encode($tailname);
    }
    elsif ($rename_method eq "base64") {
    $tailname = &unicode::base64_encode($tailname);
    $tailname =~ s/\s*//sg;      # for some reason it adds spaces not just at end but also in middle
    }

    $filename = "$tailname$suffix";
    $filename = "$dirname$filename" if ($dirname ne "./" && $dirname ne ".\\");

    return $filename;
}


# BACKWARDS COMPATIBILITY: Just in case there are old .ldb/.bdb files
sub rename_ldb_or_bdb_file {
    my ($filename_no_ext) = @_;

    my $new_filename = "$filename_no_ext.gdb";
    return if (-f $new_filename); # if the file has the right extension, don't need to do anything
    # try ldb
    my $old_filename = "$filename_no_ext.ldb";
    
    if (-f $old_filename) {
    print STDERR "Renaming $old_filename to $new_filename\n";
    rename ($old_filename, $new_filename)
        || print STDERR "Rename failed: $!\n";
    return;
    }
    # try bdb
    $old_filename = "$filename_no_ext.bdb";
    if (-f $old_filename) {
    print STDERR "Renaming $old_filename to $new_filename\n";   
    rename ($old_filename, $new_filename)
        || print STDERR "Rename failed: $!\n";
    return;
    }
}


# Returns the greenstone URL prefix extracted from the appropriate GS2/GS3 config file. 
# By default, /greenstone3 for GS3 or /greenstone for GS2.
sub get_greenstone_url_prefix() {
    # if already set on a previous occasion, just return that
    # (Don't want to keep repeating this: cost of re-opening and scanning files.)
    return $ENV{'GREENSTONE_URL_PREFIX'} if($ENV{'GREENSTONE_URL_PREFIX'});

    my ($configfile, $urlprefix, $defaultUrlprefix); 
    my @propertynames = ();

    if($ENV{'GSDL3SRCHOME'}) {
    $defaultUrlprefix = "/greenstone3";
    $configfile = &util::filename_cat($ENV{'GSDL3SRCHOME'}, "packages", "tomcat", "conf", "Catalina", "localhost", "greenstone3.xml");
    push(@propertynames, qw/path\s*\=/);
    } else {
    $defaultUrlprefix = "/greenstone";
    $configfile = &util::filename_cat($ENV{'GSDLHOME'}, "cgi-bin", "gsdlsite.cfg");
    push(@propertynames, (qw/\nhttpprefix/, qw/\ngwcgi/)); # inspect one property then the other 
    }

    $urlprefix = &extract_propvalue_from_file($configfile, \@propertynames);

    if(!$urlprefix) { # no values found for URL prefix, use default values
    $urlprefix = $defaultUrlprefix;
    } else {
    #gwcgi can contain more than the wanted prefix, we split on / to get the first "directory" level
    $urlprefix =~ s/^\///; # remove the starting slash
    my @dirs = split(/(\\|\/)/, $urlprefix); 
    $urlprefix = shift(@dirs);

    if($urlprefix !~ m/^\//) { # in all cases: ensure the required forward slash is at the front
        $urlprefix = "/$urlprefix";
    }
    }

    # set for the future
    $ENV{'GREENSTONE_URL_PREFIX'} = $urlprefix;
#    print STDERR "*** in get_greenstone_url_prefix(): $urlprefix\n\n";
    return $urlprefix;
}


# Given a config file (xml or java properties file) and a list/array of regular expressions
# that represent property names to match on, this function will return the value for the 1st
# matching property name. If the return value is undefined, no matching property was found.
sub extract_propvalue_from_file() {
    my ($configfile, $propertynames) = @_;

    my $value;
    unless(open(FIN, "<$configfile")) { 
    print STDERR "extract_propvalue_from_file(): Unable to open $configfile. $!\n";
    return $value; # not initialised
    }

    # Read the entire file at once, as one single line, then close it
    my $filecontents;
    {
    local $/ = undef;        
    $filecontents = <FIN>;
    }
    close(FIN);

    foreach my $regex (@$propertynames) {
        ($value) = $filecontents=~ m/$regex\s*(\S*)/s; # read value of the property given by regex up to the 1st space
    if($value) { 
            $value =~ s/^\"//;     # remove any startquotes
        $value =~ s/\".*$//;   # remove the 1st endquotes (if any) followed by any xml
        last;              # found value for a matching property, break from loop
    }
    }

    return $value;
}

# Subroutine that sources setup.bash, given GSDLHOME and GSDLOS and
# given that perllib is in @INC in order to invoke this subroutine.
# Call as follows -- after setting up INC to include perllib and 
# after setting up GSDLHOME and GSDLOS:
#
# require util;
# &util::setup_greenstone_env($ENV{'GSDLHOME'}, $ENV{'GSDLOS'});
#
sub setup_greenstone_env() {
    my ($GSDLHOME, $GSDLOS) = @_;

    #my %env_map = ();
    # Get the localised ENV settings of running a localised source setup.bash 
    # and put it into the ENV here. Need to clear GSDLHOME before running setup
    #my $perl_command = "(cd $GSDLHOME; export GSDLHOME=; . ./setup.bash > /dev/null; env)";
    my $perl_command = "(cd $GSDLHOME; /bin/bash -c \"export GSDLHOME=; source setup.bash > /dev/null; env\")";     
    if($GSDLOS =~ m/windows/i) {
        #$perl_command = "cmd /C \"cd $GSDLHOME&& set GSDLHOME=&& setup.bat > nul&& set\"";
        $perl_command = "(cd $GSDLHOME&& set GSDLHOME=&& setup.bat > nul&& set)";
    }
    if (!open(PIN, "$perl_command |")) {
        print STDERR ("Unable to execute command: $perl_command. $!\n");
    }

    while (defined (my $perl_output_line = <PIN>)) {
        my($key,$value) = ($perl_output_line =~ m/^([^=]*)[=](.*)$/);
        #$env_map{$key}=$value;     
        $ENV{$key}=$value;
    }
    
    # If any keys in $ENV don't occur in Greenstone's localised env 
    # (stored in $env_map), delete those entries from $ENV
    #foreach $key (keys %ENV) { 
    #   if(!defined $env_map{$key}) { 
    #       print STDOUT "**** DELETING ENV KEY: $key\tVALUE: $ENV{'$key'}\n";
    #       delete $ENV{$key}; # del $ENV(key, value) pair
    #   }
    #}
    #undef %env_map;
}

1;