source: main/trunk/greenstone2/perllib/util.pm@ 27326

###########################################################################
#
# 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;
# Config for getting the perlpath in the recommended way, though it uses paths that are
# hard-coded into the Config file that's generated upon configuring and compiling perl.
# $^X works better in some cases to return the path to perl used to launch the script,
# but if launched with plain "perl" (no full-path), that will be just what it returns.
use Config;
# New module for file related utility functions - intended as a
# placeholder for an extension that allows a variety of different
# filesystems (FTP, HTTP, SAMBA, WEBDav, HDFS etc)
use FileUtils;

# removes files (but not directories)
sub rm {
  warnings::warnif("deprecated", "util::rm() is deprecated, using FileUtils::removeFiles() instead");
  return &FileUtils::removeFiles(@_);
}

# recursive removal
sub filtered_rm_r {
  warnings::warnif("deprecated", "util::filtered_rm_r() is deprecated, using FileUtils::removeFilesFiltered() instead");
  return &FileUtils::removeFilesFiltered(@_);
}

# recursive removal
sub rm_r {
  warnings::warnif("deprecated", "util::rm_r() is deprecated, using FileUtils::removeFilesRecursive() instead");
  return &FileUtils::removeFilesRecursive(@_);
}

# moves a file or a group of files
sub mv {
  warnings::warnif("deprecated", "util::mv() is deprecated, using FileUtils::moveFiles() instead");
  return &FileUtils::moveFiles(@_);
}

# Move the contents of source directory into target directory
# (as opposed to merely replacing target dir with the src dir)
# This can overwrite any files with duplicate names in the target
# but other files and folders in the target will continue to exist
sub mv_dir_contents {
  warnings::warnif("deprecated", "util::mv_dir_contents() is deprecated, using FileUtils::moveDirectoryContents() instead");
  return &FileUtils::moveDirectoryContents(@_);
}

# copies a file or a group of files
sub cp {
  warnings::warnif("deprecated", "util::cp() is deprecated, using FileUtils::copyFiles() instead");
  return &FileUtils::copyFiles(@_);
}

# 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 {
  warnings::warnif("deprecated", "util::cp_r() is deprecated, using FileUtils::copyFilesrecursive() instead");
  return &FileUtils::copyFilesRecursive(@_);
}

# 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 {
  warnings::warnif("deprecated", "util::cp_r_nosvn() is deprecated, using FileUtils::copyFilesRecursiveNoSVN() instead");
  return &FileUtils::copyFilesRecursiveNoSVN(@_);
}

# copies a directory and its contents, excluding subdirectories, into a new directory
sub cp_r_toplevel {
  warnings::warnif("deprecated", "util::cp_r_toplevel() is deprecated, using FileUtils::recursiveCopyTopLevel() instead");
  return &FileUtils::recursiveCopyTopLevel(@_);
}

sub mk_dir {
  warnings::warnif("deprecated", "util::mk_dir() is deprecated, using FileUtils::makeDirectory() instead");
  return &FileUtils::makeDirectory(@_);
}

# 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 {
  warnings::warnif("deprecated", "util::mk_all_dir() is deprecated, using FileUtils::makeAllDirectories() instead");
  return &FileUtils::makeAllDirectories(@_);
}

# make hard link to file if supported by OS, otherwise copy the file
sub hard_link {
  warnings::warnif("deprecated", "util::hard_link() is deprecated, using FileUtils::hardLink() instead");
  return &FileUtils::hardLink(@_);
}

# make soft link to file if supported by OS, otherwise copy file
sub soft_link {
  warnings::warnif("deprecated", "util::soft_link() is deprecated, using FileUtils::softLink() instead");
  return &FileUtils::softLink(@_);
}

# 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 {
  warnings::warnif("deprecated", "util::fd_exists() is deprecated, using FileUtils::fileTest() instead");
  return &FileUtils::fileTest(@_);
}

sub file_exists {
  warnings::warnif("deprecated", "util::file_exists() is deprecated, using FileUtils::fileExists() instead");
  return &FileUtils::fileExists(@_);
}

sub dir_exists {
  warnings::warnif("deprecated", "util::dir_exists() is deprecated, using FileUtils::directoryExists() instead");
  return &FileUtils::directoryExists(@_);
}

# 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 {
  warnings::warnif("deprecated", "util::cachedir() is deprecated, using FileUtils::synchronizeDirectories() instead");
  return &FileUtils::synchronizeDirectories(@_);
}

# 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 {
  warnings::warnif("deprecated", "util::differentfiles() is deprecated, using FileUtils::differentFiles() instead");
  return &FileUtils::differentFiles(@_);
}


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 $tmpdir = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "tmp");
    &mk_all_dir ($tmpdir) unless -e $tmpdir;

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

    while (-e $full_tmp_filename) {
    $rand = int(rand $count);
    $full_tmp_filename = &FileUtils::filenameConcatenate($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;
    }

    $tmp_dirname = &FileUtils::filenameConcatenate($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 $time_tmp_dirname = &FileUtils::filenameConcatenate($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 $tmp_filename = &FileUtils::filenameConcatenate($tmp_dirname, "$tailname$output_ext");
    
    return $tmp_filename;
}

sub get_toplevel_tmp_dir
{
    return &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "tmp");
}


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

    # need to make single backslashes double so that regex works
    $filename =~ s/\\/\\\\/g; # if ($ENV{'GSDLOS'} =~ /^windows$/i);    
    
    # note that the first part of a substitution is a regex, so RE chars need to be escaped,
    # the second part of a substitution is not a regex, so for e.g. full-stop can be specified literally
    $filename =~ s/\./\\./g; # in case there are extensions/other full stops, escape them
    $filename =~ s@\(@\\(@g; # escape brackets
    $filename =~ s@\)@\\)@g; # escape brackets
    $filename =~ s@\[@\\[@g; # escape brackets
    $filename =~ s@\]@\\]@g; # escape brackets
    
    return $filename;
}

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

    # need to put doubled backslashes for regex back to single
    $filename =~ s/\\\./\./g; # remove RE syntax for .
    $filename =~ s@\\\(@(@g; # remove RE syntax for ( => "\(" turns into "("
    $filename =~ s@\\\)@)@g; # remove RE syntax for ) => "\)" turns into ")"
    $filename =~ s@\\\[@[@g; # remove RE syntax for [ => "\[" turns into "["
    $filename =~ s@\\\]@]@g; # remove RE syntax for ] => "\]" turns into "]"
    
    # \\ goes to \
    # This is the last step in reverse mirroring the order of steps in filename_to_regex()
    $filename =~ s/\\\\/\\/g; # remove RE syntax for \    
    return $filename;
}

sub filename_cat {
  # I've disabled this warning for now, as every Greenstone perl
  # script seems to make use of this function and so you drown in a
  # sea of deprecated warnings [jmt12]
#  warnings::warnif("deprecated", "util::filename_cat() is deprecated, using FileUtils::filenameConcatenate() instead");
  return &FileUtils::filenameConcatenate(@_);
}


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) = @_;

    # 64 bit linux can't handle ";" as path separator, so make sure to set this to the right one for the OS
    my $pathsep = (defined $ENV{'GSDLOS'} && $ENV{'GSDLOS'} !~ m/windows/) ? ":" : ";";

    # do not prepend any value/path that's already in the environment variable
    
    my $escaped_val = &filename_to_regex($val); # escape any backslashes and brackets for upcoming regex
    if (!defined($ENV{$var})) {
    $ENV{$var} = "$val";
    }
    elsif($ENV{$var} !~ m/$escaped_val/) { 
    $ENV{$var} = "$val".$pathsep.$ENV{$var};
    }
}

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

    # 64 bit linux can't handle ";" as path separator, so make sure to set this to the right one for the OS
    my $pathsep = (defined $ENV{'GSDLOS'} && $ENV{'GSDLOS'} !~ m/windows/) ? ":" : ";";
    
    # do not append any value/path that's already in the environment variable

    my $escaped_val = &filename_to_regex($val); # escape any backslashes and brackets for upcoming regex
    if (!defined($ENV{$var})) {
    $ENV{$var} = "$val";
    }
    elsif($ENV{$var} !~ m/$escaped_val/) { 
    $ENV{$var} = $ENV{$var}.$pathsep."$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
    $filename_full_path = &FileUtils::filenameConcatenate($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
    $filename1 = &FileUtils::filenameConcatenate($filename1);
    $filename2 = &FileUtils::filenameConcatenate($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;
}

# If filename is relative to within_dir, returns the relative path of filename to that directory
# with slashes in the filename returned as they were in the original (absolute) filename.
sub filename_within_directory
{
    my ($filename,$within_dir) = @_;
    
    if ($within_dir !~ m/[\/\\]$/) {
    my $dirsep = &util::get_dirsep();
    $within_dir .= $dirsep;
    }
    
    $within_dir = &filename_to_regex($within_dir); # escape DOS style file separator and brackets   
    if ($filename =~ m/^$within_dir(.*)$/) {
    $filename = $1;
    }
    
    return $filename;
}

# If filename is relative to within_dir, returns the relative path of filename to that directory in URL format.
# Filename and within_dir can be any type of slashes, but will be compared as URLs (i.e. unix-style slashes).
# The subpath returned will also be a URL type filename.
sub filename_within_directory_url_format
{
    my ($filename,$within_dir) = @_;
    
    # convert parameters only to / slashes if Windows
    
    my $filename_urlformat = &filepath_to_url_format($filename);
    my $within_dir_urlformat = &filepath_to_url_format($within_dir);

    #if ($within_dir_urlformat !~ m/\/$/) {
        # make sure directory ends with a slash
        #$within_dir_urlformat .= "/";
    #}
    
    my $within_dir_urlformat_re = &filename_to_regex($within_dir_urlformat); # escape any special RE characters, such as brackets
    
    #print STDERR "@@@@@ $filename_urlformat =~ $within_dir_urlformat_re\n";
    
    # dir prefix may or may not end with a slash (this is discarded when extracting the sub-filepath)
    if ($filename_urlformat =~ m/^$within_dir_urlformat_re(?:\/)*(.*)$/) {
        $filename_urlformat = $1;
    }
    
    return $filename_urlformat;
}

# Convert parameter to use / slashes if Windows (if on Linux leave any \ as is,
# since on Linux it doesn't represent a file separator but an escape char).
sub filepath_to_url_format
{
    my ($filepath) = @_;
    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
        # Only need to worry about Windows, as Unix style directories already in url-format
        # Convert Windows style \ => /
        $filepath =~ s@\\@/@g;      
    }
    return $filepath;
}

# regex filepaths on windows may include \\ as path separator. Convert \\ to /
sub filepath_regex_to_url_format
{
    my ($filepath) = @_;
    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
    # Only need to worry about Windows, as Unix style directories already in url-format
    # Convert Windows style \\ => /
    $filepath =~ s@\\\\@/@g;        
    }
    return $filepath;
    
}

# Like File::Basename::fileparse, but expects filepath in url format (ie only / slash for dirsep)
# and ignores trailing /
# returns (file, dirs) dirs will be empty if no subdirs
sub url_fileparse
{
    my ($filepath) = @_;
    # remove trailing /
    $filepath =~ s@/$@@;
    if ($filepath !~ m@/@) {
    return ($filepath, "");
    } 
    my ($dirs, $file) = $filepath =~ m@(.+/)([^/]+)@;
    return ($file, $dirs);
    
}


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 $filename_full_path = &FileUtils::filenameConcatenate($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 block_filename
{
    my ($block_hash,$filename) = @_;
    
    if ($ENV{'GSDLOS'} =~ m/^windows$/) {
    
    # lower case the entire thing, eg for cover.jpg when its actually cover.JPG
    my $lower_filename = lc($filename);
    $block_hash->{'file_blocks'}->{$lower_filename} = 1;
#   my $lower_drive = $filename;
#   $lower_drive =~ s/^([A-Z]):/\l$1:/i;
    
#   my $upper_drive = $filename;
#   $upper_drive =~ s/^([A-Z]):/\u$1:/i;
#   
#   $block_hash->{'file_blocks'}->{$lower_drive} = 1;
#   $block_hash->{'file_blocks'}->{$upper_drive} = 1;       
    }
    else {
    $block_hash->{'file_blocks'}->{$filename} = 1;
    }
}


sub filename_is_absolute
{
  warnings::warnif("deprecated", "util::filename_is_absolute() is deprecated, using FileUtils::isFilenameAbsolute() instead");
  return &FileUtils::isFilenameAbsolute(@_);
}


## @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 "") {
    $collectdir = &FileUtils::filenameConcatenate($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;
    $ENV{'GSDLCOLLECTDIR'} = &FileUtils::filenameConcatenate($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'};
    $collectdir = &FileUtils::filenameConcatenate($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 $test_collect_etc_filename 
        = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'},"etc", $file);
        
        if (-e $test_collect_etc_filename) {
        push(@locations,$test_collect_etc_filename);
        }
    }
    my $test_main_etc_filename 
        = &FileUtils::filenameConcatenate($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 {
  warnings::warnif("deprecated", "util::is_dir_empty() is deprecated, using FileUtils::isDirectoryEmpty() instead");
  return &FileUtils::isDirectoryEmpty(@_);
}

# 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) {
    $suffix = "";
    }
    # This breaks GLI matching extracted metadata to files in Enrich panel, as
    # original is eg .JPG while gsdlsourcefilename ends up .jpg
    # Not sure why it was done in first place...
    #else {
    #$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;
    }
}

sub os_dir() {
    
    my $gsdlarch = "";
    if(defined $ENV{'GSDLARCH'}) {
    $gsdlarch = $ENV{'GSDLARCH'};
    }
    return $ENV{'GSDLOS'}.$gsdlarch;
}

# 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";
    $configfile = &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'}, "packages", "tomcat", "conf", "Catalina", "localhost", "greenstone3.xml");
    push(@propertynames, qw/path\s*\=/);
    } else {
    $defaultUrlprefix = "/greenstone";
    $configfile = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "cgi-bin", &os_dir(), "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;
    }
    close (PIN);

    # 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;
}

sub get_perl_exec() {   
    my $perl_exec = $^X; # may return just "perl"
    
    if($ENV{'PERLPATH'}) {
        # OR: # $perl_exec = &FileUtils::filenameConcatenate($ENV{'PERLPATH'},"perl");
        if($ENV{'GSDLOS'} =~ m/windows/) {
            $perl_exec = "$ENV{'PERLPATH'}\\Perl.exe";
        } else {
            $perl_exec = "$ENV{'PERLPATH'}/perl";
        }
    } else { # no PERLPATH, use Config{perlpath} else $^X: special variables
        # containing the full path to the current perl executable we're using
        $perl_exec = $Config{perlpath}; # configured path for perl
        if (!-e $perl_exec) { # may not point to location on this machine
            $perl_exec = $^X; # may return just "perl"
            if($perl_exec =~ m/^perl/i) { # warn if just perl or Perl.exe
                print STDERR "**** WARNING: Perl exec found contains no path: $perl_exec";              
            }
        }
    }
    
    return $perl_exec;
}

# returns the path to the java command in the JRE included with GS (if any),
# quoted to safeguard any spaces in this path, otherwise a simple java
# command is returned which assumes and will try for a system java.
sub get_java_command {
    my $java = "java";
    if(defined $ENV{'GSDLHOME'}) { # should be, as this script would be launched from the cmd line 
                               # after running setup.bat or from GLI which also runs setup.bat
    my $java_bin = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"packages","jre","bin");
    if(-d $java_bin) {
        $java = &FileUtils::filenameConcatenate($java_bin,"java");
        $java = "\"".$java."\""; # quoted to preserve spaces in path
    }
    }
    return $java;
}


# Given the qualified collection name (colgroup/collection), 
# returns the collection and colgroup parts
sub get_collection_parts {
    # http://perldoc.perl.org/File/Basename.html
    # my($filename, $directories, $suffix) = fileparse($path);
    # "$directories contains everything up to and including the last directory separator in the $path 
    # including the volume (if applicable). The remainder of the $path is the $filename."
    #my ($collection, $colgroup) = &File::Basename::fileparse($qualified_collection);   

    my $qualified_collection = shift(@_); 

    # Since activate.pl can be launched from the command-line, including by a user, 
    # best not to assume colgroup uses URL-style slashes as would be the case with GLI
    # Also allow for the accidental inclusion of multiple slashes
    my ($colgroup, $collection) = split(/[\/\\]+/, $qualified_collection); #split('/', $qualified_collection);
    
    if(!defined $collection) {
        $collection = $colgroup;
        $colgroup = "";
    }
    return ($collection, $colgroup);
}

# work out the "collectdir/collection" location
sub resolve_collection_dir {
    my ($collect_dir, $qualified_collection, $site) = @_; #, $gs_mode
    
    my ($colgroup, $collection) = &util::get_collection_parts($qualified_collection);   
    
    if (defined $collect_dir && $collect_dir) { # ensure not empty string either
        return &FileUtils::filenameConcatenate($collect_dir,$colgroup, $collection);
    }
    elsif (defined($ENV{'GSDLCOLLECTDIR'})) {
        return $ENV{'GSDLCOLLECTDIR'};
    }
    else {
        if (defined $site) {
            return &FileUtils::filenameConcatenate($ENV{'GSDL3HOME'},"sites",$site,"collect",$colgroup, $collection);
        }
        else {
            return &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"collect",$colgroup, $collection);
        }
    }
}

# Used by pdfpstoimg.pl and PDFBoxConverter to create a .item file from
# a directory containing sequentially numbered images.
sub create_itemfile
{
    my ($output_dir, $convert_basename, $convert_to) = @_;
    my $page_num = "";

    opendir(DIR, $output_dir) || die "can't opendir $output_dir: $!";
    my @dir_files = grep {-f "$output_dir/$_"} readdir(DIR);
    closedir DIR;

    # Sort files in the directory by page_num
    sub page_number {
    my ($dir) = @_;
    my ($pagenum) =($dir =~ m/^.*[-\.]?(\d+)(\.(jpg|gif|png))?$/i);

    $pagenum = 1 unless defined $pagenum;
    return $pagenum;
    }

    # sort the files in the directory in the order of page_num rather than lexically.
    @dir_files = sort { page_number($a) <=> page_number($b) } @dir_files;

    # work out if the numbering of the now sorted image files starts at 0 or not
    # by checking the number of the first _image_ file (skipping item files)
    my $starts_at_0 = 0;
    my $firstfile = ($dir_files[0] !~ /\.item$/i) ? $dir_files[0] : $dir_files[1];
    if(page_number($firstfile) == 0) { # 00 will evaluate to 0 too in this condition
    $starts_at_0 = 1;
    }

    my $item_file = &FileUtils::filenameConcatenate($output_dir, $convert_basename.".item");
    my $item_fh;
    &FileUtils::openFileHandle($item_file, 'w', \$item_fh);
    print $item_fh "<PagedDocument>\n";

    foreach my $file (@dir_files){
    if ($file !~ /\.item/i){
        $page_num = page_number($file);
        $page_num++ if $starts_at_0; # image numbers start at 0, so add 1
        print $item_fh "   <Page pagenum=\"$page_num\" imgfile=\"$file\" txtfile=\"\"/>\n";
    }
    }

    print $item_fh "</PagedDocument>\n";
    &FileUtils::closeFileHandle($item_file, \$item_fh);
    return $item_file;
}


# /** @function augmentINC()
#  * Prepend a path (if it exists) onto INC but only if it isn't already in
#  * INC
#  * @param $new_path The path to add as necessary
#  * @author jmt12
#  */
sub augmentINC
{
  my ($new_path) = @_;
  my $did_add_path = 0;
  # will need to be replaced with FileUtils::directoryExists() call eventually
  if (-d $new_path)
  {
    my $did_find_path = 0;
    foreach my $existing_path (@INC)
    {
      if ($existing_path eq $new_path)
      {
        $did_find_path = 1;
        last;
      }
    }
    if (!$did_find_path)
    {
      unshift(@INC, $new_path);
      $did_add_path = 1;
    }
  }
  return $did_add_path;
}
# /** augmentINC($new_path) **/

1;