root/main/trunk/greenstone2/perllib/util.pm @ 31443

Revision 31443, 62.1 KB (checked in by ak19, 3 years ago)

when we check a filename in the block hash we are converting to windows long filenames, so lets make sure things going into hash are long filenames

  • Property svn:keywords set to Author Date Id Revision
Line 
1###########################################################################
2#
3# util.pm -- various useful utilities
4# A component of the Greenstone digital library software
5# from the New Zealand Digital Library Project at the
6# University of Waikato, New Zealand.
7#
8# Copyright (C) 1999 New Zealand Digital Library Project
9#
10# This program is free software; you can redistribute it and/or modify
11# it under the terms of the GNU General Public License as published by
12# the Free Software Foundation; either version 2 of the License, or
13# (at your option) any later version.
14#
15# This program is distributed in the hope that it will be useful,
16# but WITHOUT ANY WARRANTY; without even the implied warranty of
17# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18# GNU General Public License for more details.
19#
20# You should have received a copy of the GNU General Public License
21# along with this program; if not, write to the Free Software
22# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
23#
24###########################################################################
25
26package util;
27
28use strict;
29use FileUtils;
30
31use Encode;
32use Unicode::Normalize 'normalize';
33
34use File::Copy;
35use File::Basename;
36# Config for getting the perlpath in the recommended way, though it uses paths that are
37# hard-coded into the Config file that's generated upon configuring and compiling perl.
38# $^X works better in some cases to return the path to perl used to launch the script,
39# but if launched with plain "perl" (no full-path), that will be just what it returns.
40use Config;
41# New module for file related utility functions - intended as a
42# placeholder for an extension that allows a variety of different
43# filesystems (FTP, HTTP, SAMBA, WEBDav, HDFS etc)
44use FileUtils;
45
46if ($ENV{'GSDLOS'} =~ /^windows$/i) {
47    require Win32; # for working out Windows Long Filenames from Win 8.3 short filenames
48}
49
50# removes files (but not directories)
51sub rm {
52  warnings::warnif("deprecated", "util::rm() is deprecated, using FileUtils::removeFiles() instead");
53  return &FileUtils::removeFiles(@_);
54}
55
56# recursive removal
57sub filtered_rm_r {
58  warnings::warnif("deprecated", "util::filtered_rm_r() is deprecated, using FileUtils::removeFilesFiltered() instead");
59  return &FileUtils::removeFilesFiltered(@_);
60}
61
62# recursive removal
63sub rm_r {
64  warnings::warnif("deprecated", "util::rm_r() is deprecated, using FileUtils::removeFilesRecursive() instead");
65  return &FileUtils::removeFilesRecursive(@_);
66}
67
68# moves a file or a group of files
69sub mv {
70  warnings::warnif("deprecated", "util::mv() is deprecated, using FileUtils::moveFiles() instead");
71  return &FileUtils::moveFiles(@_);
72}
73
74# Move the contents of source directory into target directory
75# (as opposed to merely replacing target dir with the src dir)
76# This can overwrite any files with duplicate names in the target
77# but other files and folders in the target will continue to exist
78sub mv_dir_contents {
79  warnings::warnif("deprecated", "util::mv_dir_contents() is deprecated, using FileUtils::moveDirectoryContents() instead");
80  return &FileUtils::moveDirectoryContents(@_);
81}
82
83# copies a file or a group of files
84sub cp {
85  warnings::warnif("deprecated", "util::cp() is deprecated, using FileUtils::copyFiles() instead");
86  return &FileUtils::copyFiles(@_);
87}
88
89# recursively copies a file or group of files
90# syntax: cp_r (sourcefiles, destination directory)
91# destination must be a directory - to copy one file to
92# another use cp instead
93sub cp_r {
94  warnings::warnif("deprecated", "util::cp_r() is deprecated, using FileUtils::copyFilesrecursive() instead");
95  return &FileUtils::copyFilesRecursive(@_);
96}
97
98# recursively copies a file or group of files
99# syntax: cp_r (sourcefiles, destination directory)
100# destination must be a directory - to copy one file to
101# another use cp instead
102sub cp_r_nosvn {
103  warnings::warnif("deprecated", "util::cp_r_nosvn() is deprecated, using FileUtils::copyFilesRecursiveNoSVN() instead");
104  return &FileUtils::copyFilesRecursiveNoSVN(@_);
105}
106
107# copies a directory and its contents, excluding subdirectories, into a new directory
108sub cp_r_toplevel {
109  warnings::warnif("deprecated", "util::cp_r_toplevel() is deprecated, using FileUtils::recursiveCopyTopLevel() instead");
110  return &FileUtils::recursiveCopyTopLevel(@_);
111}
112
113sub mk_dir {
114  warnings::warnif("deprecated", "util::mk_dir() is deprecated, using FileUtils::makeDirectory() instead");
115  return &FileUtils::makeDirectory(@_);
116}
117
118# in case anyone cares - I did some testing (using perls Benchmark module)
119# on this subroutine against File::Path::mkpath (). mk_all_dir() is apparently
120# slightly faster (surprisingly) - Stefan.
121sub mk_all_dir {
122  warnings::warnif("deprecated", "util::mk_all_dir() is deprecated, using FileUtils::makeAllDirectories() instead");
123  return &FileUtils::makeAllDirectories(@_);
124}
125
126# make hard link to file if supported by OS, otherwise copy the file
127sub hard_link {
128  warnings::warnif("deprecated", "util::hard_link() is deprecated, using FileUtils::hardLink() instead");
129  return &FileUtils::hardLink(@_);
130}
131
132# make soft link to file if supported by OS, otherwise copy file
133sub soft_link {
134  warnings::warnif("deprecated", "util::soft_link() is deprecated, using FileUtils::softLink() instead");
135  return &FileUtils::softLink(@_);
136}
137
138# Primarily for filenames generated by processing
139# content of HTML files (which are mapped to UTF-8 internally)
140#
141# To turn this into an octet string that really exists on the file
142# system:
143# 1. don't need to do anything special for Unix-based systems
144#   (as underlying file system is byte-code)
145# 2. need to map to short DOS filenames for Windows
146
147sub utf8_to_real_filename
148{
149    my ($utf8_filename) = @_;
150
151    my $real_filename;
152
153    if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
154    require Win32;
155
156    my $unicode_filename = decode("utf8",$utf8_filename);
157    $real_filename = Win32::GetShortPathName($unicode_filename);
158    }
159    else {
160    $real_filename = $utf8_filename;
161    }
162
163    return $real_filename;
164}
165
166sub raw_filename_to_unicode
167{
168    my ($directory, $raw_file, $filename_encoding ) = @_;
169       
170    my $unicode_filename = $raw_file;
171    if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
172        # Try turning a short version to the long version
173        # If there are "funny" characters in the file name, that can't be represented in the ANSI code, then we will have a short weird version, eg E74~1.txt
174        $unicode_filename = &util::get_dirsep_tail(&util::upgrade_if_dos_filename(&FileUtils::filenameConcatenate($directory, $raw_file), 0));
175       
176       
177        if ($unicode_filename eq $raw_file) {
178        # This means the original filename *was* able to be encoded in the local ANSI file encoding (eg windows_1252), so now we turn it back to perl's unicode
179       
180        $unicode_filename = &Encode::decode(locale_fs => $unicode_filename);
181        }
182        # else This means we did have one of the funny filenames. the getLongPathName (used in upgrade_if_dos_filename) will return unicode, so we don't need to do anything more.
183       
184                   
185    } else {
186        # we had a utf-8 string, turn it into perl internal unicode
187        $unicode_filename = &Encode::decode("utf-8", $unicode_filename);
188   
189       
190    }
191    #Does the filename have url encoded chars in it?
192    if (&unicode::is_url_encoded($unicode_filename)) {
193        $unicode_filename = &unicode::url_decode($unicode_filename);
194    }
195   
196    # Normalise the filename to canonical composition - on mac, filenames use decopmposed form for accented chars
197    if ($ENV{'GSDLOS'} =~ m/^darwin$/i) {
198        $unicode_filename = normalize('C', $unicode_filename); # Composed form 'C'
199    }
200    return $unicode_filename;
201
202}
203sub fd_exists {
204  warnings::warnif("deprecated", "util::fd_exists() is deprecated, using FileUtils::fileTest() instead");
205  return &FileUtils::fileTest(@_);
206}
207
208sub file_exists {
209  warnings::warnif("deprecated", "util::file_exists() is deprecated, using FileUtils::fileExists() instead");
210  return &FileUtils::fileExists(@_);
211}
212
213sub dir_exists {
214  warnings::warnif("deprecated", "util::dir_exists() is deprecated, using FileUtils::directoryExists() instead");
215  return &FileUtils::directoryExists(@_);
216}
217
218# updates a copy of a directory in some other part of the filesystem
219# verbosity settings are: 0=low, 1=normal, 2=high
220# both $fromdir and $todir should be absolute paths
221sub cachedir {
222  warnings::warnif("deprecated", "util::cachedir() is deprecated, using FileUtils::synchronizeDirectories() instead");
223  return &FileUtils::synchronizeDirectories(@_);
224}
225
226# this function returns -1 if either file is not found
227# assumes that $file1 and $file2 are absolute file names or
228# in the current directory
229# $file2 is allowed to be newer than $file1
230sub differentfiles {
231  warnings::warnif("deprecated", "util::differentfiles() is deprecated, using FileUtils::differentFiles() instead");
232  return &FileUtils::differentFiles(@_);
233}
234
235
236# works out the temporary directory, including in the case where Greenstone is not writable
237# In that case, gs3-setup.bat would already have set the GS_TMP_OUTPUT_DIR temp variable
238sub determine_tmp_dir
239{
240    my $try_collect_dir = shift(@_) || 0;
241
242    my $tmp_dirname;
243    if(defined $ENV{'GS_TMP_OUTPUT_DIR'}) {
244        $tmp_dirname = $ENV{'GS_TMP_OUTPUT_DIR'};
245    } elsif($try_collect_dir && defined $ENV{'GSDLCOLLECTDIR'}) {
246        $tmp_dirname = $ENV{'GSDLCOLLECTDIR'};
247    } elsif(defined $ENV{'GSDLHOME'}) {
248        $tmp_dirname = $ENV{'GSDLHOME'};
249    } else {
250        return undef;
251    }
252   
253    if(!defined $ENV{'GS_TMP_OUTPUT_DIR'}) {
254        # test the tmp_dirname folder is writable, by trying to write out a file
255        # Unfortunately, cound not get if(-w $dirname) to work on directories on Windows
256            ## http://alvinalexander.com/blog/post/perl/perl-file-test-operators-reference-cheat-sheet (test file/dir writable)
257            ## http://www.makelinux.net/alp/083 (real and effective user IDs)
258       
259        my $tmp_test_file = &FileUtils::filenameConcatenate($tmp_dirname, "writability_test.tmp");
260        if (open (FOUT, ">$tmp_test_file")) {
261            close(FOUT);
262            &FileUtils::removeFiles($tmp_test_file);
263        } else { # location not writable, use TMP location
264            $tmp_dirname = &FileUtils::filenameConcatenate($ENV{'TMP'}, "greenstone");
265            $ENV{'GS_TMP_OUTPUT_DIR'} = $tmp_dirname; # store for next time
266        }
267    }
268   
269    $tmp_dirname = &FileUtils::filenameConcatenate($tmp_dirname, "tmp");
270    &FileUtils::makeAllDirectories ($tmp_dirname) unless -e $tmp_dirname;
271
272    return $tmp_dirname;
273}
274
275sub get_tmp_filename
276{
277    my $file_ext = shift(@_) || undef;
278
279    my $opt_dot_file_ext = "";
280    if (defined $file_ext) {
281    if ($file_ext !~ m/\./) {
282        # no dot, so needs one added in at start
283        $opt_dot_file_ext = ".$file_ext"
284    }
285    else {
286        # allow for "extensions" such as _metadata.txt to be handled
287        # gracefully
288        $opt_dot_file_ext = $file_ext;
289    }
290    }
291
292    my $tmpdir = &util::determine_tmp_dir(0);
293
294    my $count = 1000;
295    my $rand = int(rand $count);
296    my $full_tmp_filename = &FileUtils::filenameConcatenate($tmpdir, "F$rand$opt_dot_file_ext");
297
298    while (-e $full_tmp_filename) {
299    $rand = int(rand $count);
300    $full_tmp_filename = &FileUtils::filenameConcatenate($tmpdir, "F$rand$opt_dot_file_ext");
301    $count++;
302    }
303   
304    return $full_tmp_filename;
305}
306
307# These 2 are "static" variables used by the get_timestamped_tmp_folder() subroutine below and
308# belong with that function. They help ensure the timestamped tmp folders generated are unique.
309my $previous_timestamp = undef;
310my $previous_timestamp_f = 0; # frequency
311
312sub get_timestamped_tmp_folder
313{
314    my $tmp_dirname = &util::determine_tmp_dir(1);
315   
316    # add the timestamp into the path otherwise we can run into problems
317    # if documents have the same name
318    my $timestamp = time;   
319   
320    if (!defined $previous_timestamp || ($timestamp > $previous_timestamp)) {
321        $previous_timestamp_f = 0;
322        $previous_timestamp = $timestamp;
323    } else {
324        $previous_timestamp_f++;
325    }
326   
327    my $time_tmp_dirname = &FileUtils::filenameConcatenate($tmp_dirname, $timestamp);
328    $tmp_dirname = $time_tmp_dirname;   
329    my $i = $previous_timestamp_f;
330   
331    if($previous_timestamp_f > 0) {
332        $tmp_dirname = $time_tmp_dirname."_".$i;
333        $i++;
334    }
335    while (-e $tmp_dirname) {
336    $tmp_dirname = $time_tmp_dirname."_".$i;
337    $i++;
338    }
339    &FileUtils::makeDirectory($tmp_dirname);
340
341    return $tmp_dirname;
342}
343
344sub get_parent_folder
345{
346    my ($path) = @_;
347    my ($tailname, $dirname, $suffix)
348    = &File::Basename::fileparse($path, "\\.[^\\.]+\$");
349
350    return &FileUtils::sanitizePath($dirname);
351}
352
353sub get_timestamped_tmp_filename_in_collection
354{
355
356    my ($input_filename, $output_ext) = @_;
357    # derive tmp filename from input filename
358    my ($tailname, $dirname, $suffix)
359    = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
360
361    # softlink to collection tmp dir
362    my $tmp_dirname = &util::get_timestamped_tmp_folder();
363    $tmp_dirname = $dirname unless defined $tmp_dirname;
364
365    # following two steps copied from ConvertBinaryFile
366    # do we need them?? can't use them as is, as they use plugin methods.
367
368    #$tailname = $self->SUPER::filepath_to_utf8($tailname) unless &unicode::check_is_utf8($tailname);
369
370    # URLEncode this since htmls with images where the html filename is utf8 don't seem
371    # to work on Windows (IE or Firefox), as browsers are looking for filesystem-encoded
372    # files on the filesystem.
373    #$tailname = &util::rename_file($tailname, $self->{'file_rename_method'}, "without_suffix");
374    if (defined $output_ext) {
375    $output_ext = ".$output_ext"; # add the dot
376    } else {
377    $output_ext = $suffix;
378    }
379    $output_ext= lc($output_ext);
380    my $tmp_filename = &FileUtils::filenameConcatenate($tmp_dirname, "$tailname$output_ext");
381   
382    return $tmp_filename;
383}
384
385sub get_toplevel_tmp_dir
386{
387    return &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "tmp");
388}
389
390
391sub get_collectlevel_tmp_dir
392{
393    my $tmp_dirname = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "tmp");
394    &FileUtils::makeDirectory($tmp_dirname) if (!-e $tmp_dirname);
395
396    return $tmp_dirname;
397}
398
399
400sub filename_to_regex {
401    my $filename = shift (@_);
402
403    # need to make single backslashes double so that regex works
404    $filename =~ s/\\/\\\\/g; # if ($ENV{'GSDLOS'} =~ /^windows$/i);   
405   
406    # note that the first part of a substitution is a regex, so RE chars need to be escaped,
407    # the second part of a substitution is not a regex, so for e.g. full-stop can be specified literally
408    $filename =~ s/\./\\./g; # in case there are extensions/other full stops, escape them
409    $filename =~ s@\(@\\(@g; # escape brackets
410    $filename =~ s@\)@\\)@g; # escape brackets
411    $filename =~ s@\[@\\[@g; # escape brackets
412    $filename =~ s@\]@\\]@g; # escape brackets
413   
414    return $filename;
415}
416
417sub unregex_filename {
418    my $filename = shift (@_);
419
420    # need to put doubled backslashes for regex back to single
421    $filename =~ s/\\\./\./g; # remove RE syntax for .
422    $filename =~ s@\\\(@(@g; # remove RE syntax for ( => "\(" turns into "("
423    $filename =~ s@\\\)@)@g; # remove RE syntax for ) => "\)" turns into ")"
424    $filename =~ s@\\\[@[@g; # remove RE syntax for [ => "\[" turns into "["
425    $filename =~ s@\\\]@]@g; # remove RE syntax for ] => "\]" turns into "]"
426   
427    # \\ goes to \
428    # This is the last step in reverse mirroring the order of steps in filename_to_regex()
429    $filename =~ s/\\\\/\\/g; # remove RE syntax for \   
430    return $filename;
431}
432
433sub filename_cat {
434  # I've disabled this warning for now, as every Greenstone perl
435  # script seems to make use of this function and so you drown in a
436  # sea of deprecated warnings [jmt12]
437#  warnings::warnif("deprecated", "util::filename_cat() is deprecated, using FileUtils::filenameConcatenate() instead");
438  return &FileUtils::filenameConcatenate(@_);
439}
440
441
442sub _pathname_cat {
443    my $join_char  = shift(@_);
444    my $first_path = shift(@_);
445    my (@pathnames) = @_;
446
447    # If first_path is not null or empty, then add it back into the list
448    if (defined $first_path && $first_path =~ /\S/) {
449    unshift(@pathnames, $first_path);
450    }
451
452    my $pathname = join($join_char, @pathnames);
453
454    # remove duplicate slashes
455    if ($join_char eq ";") {
456    $pathname =~ s/[\\\/]+/\\/g;
457    if ($^O eq "cygwin") {
458        # Once we've collapsed muliple (potentialy > 2) slashes
459        # For cygwin, actually want things double-backslahed
460        $pathname =~ s/\\/\\\\/g;
461    }
462
463    } else {
464    $pathname =~ s/[\/]+/\//g;
465    # DB: want a pathname abc\de.html to remain like this
466    }
467
468    return $pathname;
469}
470
471
472sub pathname_cat {
473    my (@pathnames) = @_;
474
475    my $join_char;
476    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
477    $join_char = ";";
478    } else {
479    $join_char = ":";
480    }
481    return _pathname_cat($join_char,@pathnames);
482}
483
484
485sub javapathname_cat {
486    my (@pathnames) = @_;
487
488    my $join_char;
489
490    # Unlike pathname_cat() above, not interested if running in a Cygwin environment
491    # This is because the java we run is actually a native Windows executable
492
493    if (($ENV{'GSDLOS'} =~ /^windows$/i)) {
494    $join_char = ";";
495    } else {
496    $join_char = ":";
497    }
498    return _pathname_cat($join_char,@pathnames);
499}
500
501
502sub makeFilenameJavaCygwinCompatible
503{
504    my ($java_filename) = @_;
505
506    if ($^O eq "cygwin") {
507    # To be used with a Java program, but under Cygwin
508    # Because the java binary that is native to Windows, need to
509    # convert the Cygwin paths (i.e. Unix style) to be Windows
510    # compatible
511   
512    $java_filename = `cygpath -wp "$java_filename"`;
513    chomp($java_filename);
514    $java_filename =~ s%\\%\\\\%g;
515    }
516
517    return $java_filename;
518}
519
520sub tidy_up_oid {
521    my ($OID) = @_;
522    if ($OID =~ /[\.\/\\]/) {
523    print STDERR "Warning, identifier $OID contains periods or slashes(.\\\/), replacing them with _\n";
524    $OID =~ s/[\.\\\/]/_/g; #remove any periods
525    }
526    if ($OID =~ /^\s.*\s$/) {
527    print STDERR "Warning, identifier $OID starts or ends with whitespace. Removing it\n";
528    # remove starting and trailing whitespace
529    $OID =~ s/^\s+//;
530    $OID =~ s/\s+$//;
531    }
532    if ($OID =~ /^[\d]*$/) {
533    print STDERR "Warning, identifier $OID contains only digits. Prepending 'D'.\n";
534    $OID = "D" . $OID;
535    }       
536   
537    return $OID;
538}
539
540sub envvar_prepend {
541    my ($var,$val) = @_;
542
543    # 64 bit linux can't handle ";" as path separator, so make sure to set this to the right one for the OS
544##    my $pathsep = (defined $ENV{'GSDLOS'} && $ENV{'GSDLOS'} !~ m/windows/) ? ":" : ";";
545
546    # Rewritten above to make ":" the default (Windows is the special
547    # case, anything else 'unusual' such as Solaris etc is Unix)
548    my $pathsep = (defined $ENV{'GSDLOS'} && (($ENV{'GSDLOS'} =~ m/windows/) && ($^O ne "cygwin"))) ? ";" : ":";
549
550    # do not prepend any value/path that's already in the environment variable
551   
552    my $escaped_val = &filename_to_regex($val); # escape any backslashes and brackets for upcoming regex
553    if (!defined($ENV{$var})) {
554    $ENV{$var} = "$val";
555    }
556    elsif($ENV{$var} !~ m/$escaped_val/) {
557    $ENV{$var} = "$val".$pathsep.$ENV{$var};
558    }
559}
560
561sub envvar_append {
562    my ($var,$val) = @_;
563
564    # 64 bit linux can't handle ";" as path separator, so make sure to set this to the right one for the OS
565    my $pathsep = (defined $ENV{'GSDLOS'} && $ENV{'GSDLOS'} !~ m/windows/) ? ":" : ";";
566   
567    # do not append any value/path that's already in the environment variable
568
569    my $escaped_val = &filename_to_regex($val); # escape any backslashes and brackets for upcoming regex
570    if (!defined($ENV{$var})) {
571    $ENV{$var} = "$val";
572    }
573    elsif($ENV{$var} !~ m/$escaped_val/) {
574    $ENV{$var} = $ENV{$var}.$pathsep."$val";
575    }
576}
577
578
579
580# splits a filename into a prefix and a tail extension using the tail_re, or
581# if that fails, splits on the file_extension . (dot)
582sub get_prefix_and_tail_by_regex {
583
584    my ($filename,$tail_re) = @_;
585   
586    my ($file_prefix,$file_ext) = ($filename =~ m/^(.*?)($tail_re)$/);
587    if ((!defined $file_prefix) || (!defined $file_ext)) {
588    ($file_prefix,$file_ext) = ($filename =~ m/^(.*)(\..*?)$/);
589    }
590
591    return ($file_prefix,$file_ext);
592}
593
594# get full path and file only path from a base_dir (which may be empty) and
595# file (which may contain directories)
596sub get_full_filenames {
597    my ($base_dir, $file) = @_;
598   
599#    my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(0);
600#    my ($lcfilename) = ($cfilename =~ m/([^\\\/]*)$/);
601#    print STDERR "** Calling method: $lcfilename:$cline $cpackage->$csubr\n";
602
603
604    my $filename_full_path = $file;
605    # add on directory if present
606    $filename_full_path = &FileUtils::filenameConcatenate($base_dir, $file) if $base_dir =~ /\S/;
607   
608    my $filename_no_path = $file;
609
610    # remove directory if present
611    $filename_no_path =~ s/^.*[\/\\]//;
612    return ($filename_full_path, $filename_no_path);
613}
614
615# returns the path of a file without the filename -- ie. the directory the file is in
616sub filename_head {
617    my $filename = shift(@_);
618
619    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
620    $filename =~ s/[^\\\\]*$//;
621    }
622    else {
623    $filename =~ s/[^\\\/]*$//;
624    }
625
626    return $filename;
627}
628
629
630
631# returns 1 if filename1 and filename2 point to the same
632# file or directory
633sub filenames_equal {
634    my ($filename1, $filename2) = @_;
635
636    # use filename_cat to clean up trailing slashes and
637    # multiple slashes
638    $filename1 = &FileUtils::filenameConcatenate($filename1);
639    $filename2 = &FileUtils::filenameConcatenate($filename2);
640
641    # filenames not case sensitive on windows
642    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
643    $filename1 =~ tr/[A-Z]/[a-z]/;
644    $filename2 =~ tr/[A-Z]/[a-z]/;
645    }
646    return 1 if $filename1 eq $filename2;
647    return 0;
648}
649
650# If filename is relative to within_dir, returns the relative path of filename to that directory
651# with slashes in the filename returned as they were in the original (absolute) filename.
652sub filename_within_directory
653{
654    my ($filename,$within_dir) = @_;
655   
656    if ($within_dir !~ m/[\/\\]$/) {
657    my $dirsep = &util::get_dirsep();
658    $within_dir .= $dirsep;
659    }
660   
661    $within_dir = &filename_to_regex($within_dir); # escape DOS style file separator and brackets   
662    if ($filename =~ m/^$within_dir(.*)$/) {
663    $filename = $1;
664    }
665   
666    return $filename;
667}
668
669# If filename is relative to within_dir, returns the relative path of filename to that directory in URL format.
670# Filename and within_dir can be any type of slashes, but will be compared as URLs (i.e. unix-style slashes).
671# The subpath returned will also be a URL type filename.
672sub filename_within_directory_url_format
673{
674    my ($filename,$within_dir) = @_;
675   
676    # convert parameters only to / slashes if Windows
677   
678    my $filename_urlformat = &filepath_to_url_format($filename);
679    my $within_dir_urlformat = &filepath_to_url_format($within_dir);
680
681    #if ($within_dir_urlformat !~ m/\/$/) {
682        # make sure directory ends with a slash
683        #$within_dir_urlformat .= "/";
684    #}
685   
686    my $within_dir_urlformat_re = &filename_to_regex($within_dir_urlformat); # escape any special RE characters, such as brackets
687   
688    #print STDERR "@@@@@ $filename_urlformat =~ $within_dir_urlformat_re\n";
689   
690    # dir prefix may or may not end with a slash (this is discarded when extracting the sub-filepath)
691    if ($filename_urlformat =~ m/^$within_dir_urlformat_re(?:\/)*(.*)$/) {
692        $filename_urlformat = $1;
693    }
694   
695    return $filename_urlformat;
696}
697
698# Convert parameter to use / slashes if Windows (if on Linux leave any \ as is,
699# since on Linux it doesn't represent a file separator but an escape char).
700sub filepath_to_url_format
701{
702    my ($filepath) = @_;
703    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
704        # Only need to worry about Windows, as Unix style directories already in url-format
705        # Convert Windows style \ => /
706        $filepath =~ s@\\@/@g;     
707    }
708    return $filepath;
709}
710
711# regex filepaths on windows may include \\ as path separator. Convert \\ to /
712sub filepath_regex_to_url_format
713{
714    my ($filepath) = @_;
715    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
716    # Only need to worry about Windows, as Unix style directories already in url-format
717    # Convert Windows style \\ => /
718    $filepath =~ s@\\\\@/@g;       
719    }
720    return $filepath;
721   
722}
723
724# Like File::Basename::fileparse, but expects filepath in url format (ie only / slash for dirsep)
725# and ignores trailing /
726# returns (file, dirs) dirs will be empty if no subdirs
727sub url_fileparse
728{
729    my ($filepath) = @_;
730    # remove trailing /
731    $filepath =~ s@/$@@;
732    if ($filepath !~ m@/@) {
733    return ($filepath, "");
734    }
735    my ($dirs, $file) = $filepath =~ m@(.+/)([^/]+)@;
736    return ($file, $dirs);
737   
738}
739
740
741sub filename_within_collection
742{
743    my ($filename) = @_;
744
745    my $collect_dir = $ENV{'GSDLCOLLECTDIR'};
746   
747    if (defined $collect_dir) {
748
749    # if from within GSDLCOLLECTDIR, then remove directory prefix
750    # so source_filename is realative to it.  This is done to aid
751    # portability, i.e. the collection can be moved to somewhere
752    # else on the file system and the archives directory will still
753    # work.  This is needed, for example in the applet version of
754    # GLI where GSDLHOME/collect on the server will be different to
755    # the collect directory of the remove user.  Of course,
756    # GSDLCOLLECTDIR subsequently needs to be put back on to turn
757    # it back into a full pathname.
758
759    $filename = filename_within_directory($filename,$collect_dir);
760    }
761   
762    return $filename;
763}
764
765sub prettyprint_file
766{
767    my ($base_dir,$file,$gli) = @_;
768
769    my $filename_full_path = &FileUtils::filenameConcatenate($base_dir,$file);
770
771    if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
772    require Win32;
773
774    # For some reason base_dir in the form c:/a/b/c
775    # This leads to confusion later on, so turn it back into
776    # the more usual Windows form
777    $base_dir =~ s/\//\\/g;
778    my $long_base_dir = Win32::GetLongPathName($base_dir);
779    my $long_full_path = Win32::GetLongPathName($filename_full_path);
780
781    $file = filename_within_directory($long_full_path,$long_base_dir);
782    $file = encode("utf8",$file) if ($gli);
783    }
784
785    return $file;
786}
787
788
789sub upgrade_if_dos_filename
790{
791    my ($filename_full_path,$and_encode) = @_;
792
793    if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
794    # Ensure any DOS-like filename, such as test~1.txt, has been upgraded
795    # to its long (Windows) version
796    my $long_filename = Win32::GetLongPathName($filename_full_path);
797    if (defined $long_filename) {
798       
799        $filename_full_path = $long_filename;
800    }
801    # Make sure initial drive letter is lower-case (to fit in with rest of Greenstone)
802    $filename_full_path =~ s/^(.):/\u$1:/;
803   
804    if ((defined $and_encode) && ($and_encode)) {
805        $filename_full_path = encode("utf8",$filename_full_path);
806    }
807    }
808
809    return $filename_full_path;
810}
811
812
813sub downgrade_if_dos_filename
814{
815    my ($filename_full_path) = @_;
816
817    if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
818    require Win32;
819
820    # Ensure the given long Windows filename is in a form that can
821    # be opened by Perl => convert it to a short DOS-like filename
822
823    my $short_filename = Win32::GetShortPathName($filename_full_path);
824    if (defined $short_filename) {
825        $filename_full_path = $short_filename;
826    }
827    # Make sure initial drive letter is lower-case (to fit in
828    # with rest of Greenstone)
829    $filename_full_path =~ s/^(.):/\u$1:/;
830    }
831
832    return $filename_full_path;
833}
834
835sub block_filename
836{
837    my ($block_hash,$filename) = @_;
838   
839    if (($ENV{'GSDLOS'} =~ m/^windows$/) && ($^O ne "cygwin")) {
840       # block hash contains long names, lets make sure that we were passed a long name
841       $filename = &util::upgrade_if_dos_filename($filename);
842       # lower case the entire thing, eg for cover.jpg when its actually cover.JPG
843       my $lower_filename = lc($filename);
844       $block_hash->{'file_blocks'}->{$lower_filename} = 1;
845   
846#   my $lower_drive = $filename;
847#   $lower_drive =~ s/^([A-Z]):/\l$1:/i;
848   
849#   my $upper_drive = $filename;
850#   $upper_drive =~ s/^([A-Z]):/\u$1:/i;
851#   
852#   $block_hash->{'file_blocks'}->{$lower_drive} = 1;
853#   $block_hash->{'file_blocks'}->{$upper_drive} = 1;       
854    }
855    else {
856    $block_hash->{'file_blocks'}->{$filename} = 1;
857    }
858}
859
860
861sub filename_is_absolute
862{
863  warnings::warnif("deprecated", "util::filename_is_absolute() is deprecated, using FileUtils::isFilenameAbsolute() instead");
864  return &FileUtils::isFilenameAbsolute(@_);
865}
866
867
868## @method make_absolute()
869#
870#  Ensure the given file path is absolute in respect to the given base path.
871#
872#  @param  $base_dir A string denoting the base path the given dir must be
873#                    absolute to.
874#  @param  $dir The directory to be made absolute as a string. Note that the
875#               dir may already be absolute, in which case it will remain
876#               unchanged.
877#  @return The now absolute form of the directory as a string.
878#
879#  @author John Thompson, DL Consulting Ltd.
880#  @copy 2006 DL Consulting Ltd.
881#
882#used in buildcol.pl, doesn't work for all cases --kjdon
883sub make_absolute {
884   
885    my ($base_dir, $dir) = @_;
886###    print STDERR "dir = $dir\n";
887    $dir =~ s/[\\\/]+/\//g;
888    $dir = $base_dir . "/$dir" unless ($dir =~ m|^(\w:)?/|);
889    $dir =~ s|^/tmp_mnt||;
890    1 while($dir =~ s|/[^/]*/\.\./|/|g);
891    $dir =~ s|/[.][.]?/|/|g;
892    $dir =~ tr|/|/|s;
893###    print STDERR "dir = $dir\n";
894   
895    return $dir;
896}
897## make_absolute() ##
898
899sub get_dirsep {
900
901    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
902    return "\\";
903    } else {
904    return "\/";
905    }
906}
907
908sub get_os_dirsep {
909
910    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
911    return "\\\\";
912    } else {
913    return "\\\/";
914    }
915}
916
917sub get_re_dirsep {
918
919    return "\\\\|\\\/";
920}
921
922
923sub get_dirsep_tail {
924    my ($filename) = @_;
925   
926    # returns last part of directory or filename
927    # On unix e.g. a/b.d => b.d
928    #              a/b/c => c
929
930    my $dirsep = get_re_dirsep();
931    my @dirs = split (/$dirsep/, $filename);
932    my $tail = pop @dirs;
933
934    # - caused problems under windows
935    #my ($tail) = ($filename =~ m/^(?:.*?$dirsep)?(.*?)$/);
936
937    return $tail;
938}
939
940
941# if this is running on windows we want binaries to end in
942# .exe, otherwise they don't have to end in any extension
943sub get_os_exe {
944    return ".exe" if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin"));
945    return "";
946}
947
948
949# test to see whether this is a big or little endian machine
950sub is_little_endian
951{
952    # To determine the name of the operating system, the variable $^O is a cheap alternative to pulling it out of the Config module;
953    # 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
954    # Otherwise, it's little endian
955
956    #return 0 if $^O =~ /^darwin$/i;
957    #return 0 if $ENV{'GSDLOS'} =~ /^darwin$/i;
958   
959    # Going back to stating exactly whether the machine is little endian
960    # or big endian, without any special case for Macs. Since for rata it comes
961    # back with little endian and for shuttle with bigendian.
962    return (ord(substr(pack("s",1), 0, 1)) == 1);
963}
964
965
966# will return the collection name if successful, "" otherwise
967sub use_collection {
968    my ($collection, $collectdir) = @_;
969
970    if (!defined $collectdir || $collectdir eq "") {
971    $collectdir = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "collect");
972    }
973
974    if (!defined $ENV{'GREENSTONEHOME'}) { # for GS3, would have been defined in use_site_collection, to GSDL3HOME
975     $ENV{'GREENSTONEHOME'} = $ENV{'GSDLHOME'};
976    }
977
978    # get and check the collection
979    if (!defined($collection) || $collection eq "") {
980    if (defined $ENV{'GSDLCOLLECTION'}) {
981        $collection = $ENV{'GSDLCOLLECTION'};
982    } else {
983        print STDOUT "No collection specified\n";
984        return "";
985    }
986    }
987   
988    if ($collection eq "modelcol") {
989    print STDOUT "You can't use modelcol.\n";
990    return "";
991    }
992
993    # make sure the environment variables GSDLCOLLECTION and GSDLCOLLECTDIR
994    # are defined
995    $ENV{'GSDLCOLLECTION'} = $collection;
996    $ENV{'GSDLCOLLECTHOME'} = $collectdir;
997    $ENV{'GSDLCOLLECTDIR'} = &FileUtils::filenameConcatenate($collectdir, $collection);
998
999    # make sure this collection exists
1000    if (!-e $ENV{'GSDLCOLLECTDIR'}) {
1001    print STDOUT "Invalid collection ($collection).\n";
1002    return "";
1003    }
1004
1005    # everything is ready to go
1006    return $collection;
1007}
1008
1009sub get_current_collection_name {
1010    return $ENV{'GSDLCOLLECTION'};
1011}
1012
1013
1014# will return the collection name if successful, "" otherwise. 
1015# Like use_collection (above) but for greenstone 3 (taking account of site level)
1016
1017sub use_site_collection {
1018    my ($site, $collection, $collectdir) = @_;
1019
1020    if (!defined $collectdir || $collectdir eq "") {
1021    die "GSDL3HOME not set.\n" unless defined $ENV{'GSDL3HOME'};
1022    $collectdir = &FileUtils::filenameConcatenate($ENV{'GSDL3HOME'}, "sites", $site, "collect");
1023    }
1024
1025    if (defined $ENV{'GSDL3HOME'}) {
1026    $ENV{'GREENSTONEHOME'} = $ENV{'GSDL3HOME'};     
1027    $ENV{'SITEHOME'} = &FileUtils::filenameConcatenate($ENV{'GREENSTONEHOME'}, "sites", $site);
1028    } elsif (defined $ENV{'GSDL3SRCHOME'}) {
1029    $ENV{'GREENSTONEHOME'} = &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'}, "web"); 
1030    $ENV{'SITEHOME'} = &FileUtils::filenameConcatenate($ENV{'GREENSTONEHOME'}, "sites", $site);
1031    } else {
1032    print STDERR "*** util::use_site_collection(). Warning: Neither GSDL3HOME nor GSDL3SRCHOME set.\n";
1033    }
1034
1035    # collectdir explicitly set by this point (using $site variable if required).
1036    # Can call "old" gsdl2 use_collection now.
1037
1038    return use_collection($collection,$collectdir);
1039}
1040
1041
1042
1043sub locate_config_file
1044{
1045    my ($file) = @_;
1046
1047    my $locations = locate_config_files($file);
1048
1049    return shift @$locations; # returns undef if 'locations' is empty
1050}
1051
1052
1053sub locate_config_files
1054{
1055    my ($file) = @_;
1056
1057    my @locations = ();
1058
1059    if (-e $file) {
1060    # Clearly specified (most likely full filename)
1061    # No need to hunt in 'etc' directories, return value unchanged
1062    push(@locations,$file);
1063    }
1064    else {
1065    # Check for collection specific one before looking in global GSDL 'etc'
1066    if (defined $ENV{'GSDLCOLLECTDIR'} && $ENV{'GSDLCOLLECTDIR'} ne "") {
1067        my $test_collect_etc_filename
1068        = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'},"etc", $file);
1069       
1070        if (-e $test_collect_etc_filename) {
1071        push(@locations,$test_collect_etc_filename);
1072        }
1073    }
1074    my $test_main_etc_filename
1075        = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"etc", $file);
1076    if (-e $test_main_etc_filename) {
1077        push(@locations,$test_main_etc_filename);
1078    }
1079    }
1080
1081    return \@locations;
1082}
1083
1084
1085sub hyperlink_text
1086{
1087    my ($text) = @_;
1088   
1089    $text =~ s/(http:\/\/[^\s]+)/<a href=\"$1\">$1<\/a>/mg;
1090    $text =~ s/(^|\s+)(www\.(\w|\.)+)/<a href=\"http:\/\/$2\">$2<\/a>/mg;
1091
1092    return $text;
1093}
1094
1095
1096# A method to check if a directory is empty (note that an empty directory still has non-zero size!!!)
1097# Code is from http://episteme.arstechnica.com/eve/forums/a/tpc/f/6330927813/m/436007700831
1098sub is_dir_empty {
1099  warnings::warnif("deprecated", "util::is_dir_empty() is deprecated, using FileUtils::isDirectoryEmpty() instead");
1100  return &FileUtils::isDirectoryEmpty(@_);
1101}
1102
1103# Returns the given filename converted using either URL encoding or base64
1104# encoding, as specified by $rename_method. If the given filename has no suffix
1105# (if it is just the tailname), then $no_suffix should be some defined value.
1106# rename_method can be url, none, base64
1107sub rename_file {
1108    my ($filename, $rename_method, $no_suffix)  = @_;
1109
1110    if(!$filename) { # undefined or empty string
1111    return $filename;
1112    }
1113
1114    if (!$rename_method) {
1115    print STDERR "WARNING: no file renaming method specified. Defaulting to using URL encoding...\n";
1116    # Debugging information
1117    # my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(1);
1118    # print STDERR "Called from method: $cfilename:$cline $cpackage->$csubr\n";
1119    $rename_method = "url";
1120    } elsif($rename_method eq "none") {
1121    return $filename; # would have already been renamed
1122    }
1123
1124    # No longer replace spaces with underscores, since underscores mess with incremental rebuild
1125    ### Replace spaces with underscore. Do this first else it can go wrong below when getting tailname
1126    ###$filename =~ s/ /_/g;
1127
1128    my ($tailname,$dirname,$suffix);
1129    if($no_suffix) { # given a tailname, no suffix
1130    ($tailname,$dirname) = File::Basename::fileparse($filename);
1131    }
1132    else {
1133    ($tailname,$dirname,$suffix) = File::Basename::fileparse($filename, "\\.(?:[^\\.]+?)\$");
1134    }
1135    if (!$suffix) {
1136    $suffix = "";
1137    }
1138    # This breaks GLI matching extracted metadata to files in Enrich panel, as
1139    # original is eg .JPG while gsdlsourcefilename ends up .jpg
1140    # Not sure why it was done in first place...
1141    #else {
1142    #$suffix = lc($suffix);
1143    #}
1144
1145    if ($rename_method eq "url") {
1146    $tailname = &unicode::url_encode($tailname);
1147    }
1148    elsif ($rename_method eq "base64") {
1149    $tailname = &unicode::base64_encode($tailname);
1150    $tailname =~ s/\s*//sg;      # for some reason it adds spaces not just at end but also in middle
1151    }
1152
1153    $filename = "$tailname$suffix";
1154    $filename = "$dirname$filename" if ($dirname ne "./" && $dirname ne ".\\");
1155
1156    return $filename;
1157}
1158
1159
1160# BACKWARDS COMPATIBILITY: Just in case there are old .ldb/.bdb files
1161sub rename_ldb_or_bdb_file {
1162    my ($filename_no_ext) = @_;
1163
1164    my $new_filename = "$filename_no_ext.gdb";
1165    return if (-f $new_filename); # if the file has the right extension, don't need to do anything
1166    # try ldb
1167    my $old_filename = "$filename_no_ext.ldb";
1168   
1169    if (-f $old_filename) {
1170    print STDERR "Renaming $old_filename to $new_filename\n";
1171    rename ($old_filename, $new_filename)
1172        || print STDERR "Rename failed: $!\n";
1173    return;
1174    }
1175    # try bdb
1176    $old_filename = "$filename_no_ext.bdb";
1177    if (-f $old_filename) {
1178    print STDERR "Renaming $old_filename to $new_filename\n";   
1179    rename ($old_filename, $new_filename)
1180        || print STDERR "Rename failed: $!\n";
1181    return;
1182    }
1183}
1184
1185sub os_dir() {
1186   
1187    my $gsdlarch = "";
1188    if(defined $ENV{'GSDLARCH'}) {
1189    $gsdlarch = $ENV{'GSDLARCH'};
1190    }
1191    return $ENV{'GSDLOS'}.$gsdlarch;
1192}
1193
1194# returns 1 if this (GS server) is a GS3 installation, returns 0 if it's GS2.
1195sub is_gs3() {
1196    if($ENV{'GSDL3SRCHOME'}) {
1197    return 1;
1198    } else {
1199    return 0;
1200    }
1201}
1202
1203# Returns the greenstone URL prefix extracted from the appropriate GS2/GS3 config file.
1204# By default, /greenstone3 for GS3 or /greenstone for GS2.
1205sub get_greenstone_url_prefix() {
1206    # if already set on a previous occasion, just return that
1207    # (Don't want to keep repeating this: cost of re-opening and scanning files.)
1208    return $ENV{'GREENSTONE_URL_PREFIX'} if($ENV{'GREENSTONE_URL_PREFIX'});
1209
1210    my ($configfile, $urlprefix, $defaultUrlprefix);
1211    my @propertynames = ();
1212
1213    if($ENV{'GSDL3SRCHOME'}) {
1214    $defaultUrlprefix = "/greenstone3";
1215    $configfile = &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'}, "packages", "tomcat", "conf", "Catalina", "localhost", "greenstone3.xml");
1216    push(@propertynames, qw/path\s*\=/);
1217    } else {
1218    $defaultUrlprefix = "/greenstone";
1219    $configfile = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "cgi-bin", &os_dir(), "gsdlsite.cfg");
1220    push(@propertynames, (qw/\nhttpprefix/, qw/\ngwcgi/)); # inspect one property then the other
1221    }
1222
1223    $urlprefix = &extract_propvalue_from_file($configfile, \@propertynames);
1224
1225    if(!$urlprefix) { # no values found for URL prefix, use default values
1226    $urlprefix = $defaultUrlprefix;
1227    } else {
1228    #gwcgi can contain more than the wanted prefix, we split on / to get the first "directory" level
1229    $urlprefix =~ s/^\///; # remove the starting slash
1230    my @dirs = split(/(\\|\/)/, $urlprefix);
1231    $urlprefix = shift(@dirs);
1232
1233    if($urlprefix !~ m/^\//) { # in all cases: ensure the required forward slash is at the front
1234        $urlprefix = "/$urlprefix";
1235    }
1236    }
1237
1238    # set for the future
1239    $ENV{'GREENSTONE_URL_PREFIX'} = $urlprefix;
1240#    print STDERR "*** in get_greenstone_url_prefix(): $urlprefix\n\n";
1241    return $urlprefix;
1242}
1243
1244
1245
1246#
1247# The following comes from activate.pl
1248#
1249# Designed to work with a server included with GS.
1250#  - For GS3, we ask ant for the library URL.
1251#  - For GS2, we derive the URL from the llssite.cfg file.
1252
1253sub get_full_greenstone_url_prefix
1254{   
1255    # if already set on a previous occasion, just return that
1256    # (Don't want to keep repeating this: cost of re-opening and scanning files.)
1257    return $ENV{'GREENSTONE_URL_PREFIX'} if($ENV{'GREENSTONE_URL_PREFIX'});
1258
1259    my $gs_mode = ($ENV{'GSDL3SRCHOME'}) ? "gs3" : "gs2";
1260   
1261    my $url = undef;   
1262   
1263    if($gs_mode eq "gs2") {     
1264    my $llssite_cfg = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "llssite.cfg");
1265   
1266    if(-f $llssite_cfg) {
1267        # check llssite.cfg for line with url property
1268        # for server.exe also need to use portnumber and enterlib properties           
1269       
1270        # Read in the entire contents of the file in one hit
1271        if (!open (FIN, $llssite_cfg)) {
1272        print STDERR "util::get_library_URL() failed to open $llssite_cfg ($!)\n";
1273        return undef;
1274        }
1275       
1276        my $contents;
1277        sysread(FIN, $contents, -s FIN);           
1278        close(FIN);
1279       
1280        my @lines = split(/[\n\r]+/, $contents); # split on carriage-returns and/or linefeeds
1281        my $enterlib = "";
1282        my $portnumber = "8282"; # will remain empty (implicit port 80) unless it's specifically been assigned
1283       
1284        foreach my $line (@lines) {             
1285        if($line =~ m/^url=(.*)$/) {
1286            $url = $1;                 
1287        } elsif($line =~ m/^enterlib=(.*)$/) {
1288            $enterlib = $1;                 
1289        } elsif($line =~ m/^portnumber=(.*)$/) {
1290            $portnumber = $1;                   
1291        }   
1292        }
1293       
1294        if(!$url) {
1295        return undef;
1296        }
1297        elsif($url eq "URL_pending") { # library is not running
1298        # do not process url=URL_pending in the file, since for server.exe
1299        # this just means the Enter Library button hasn't been pressed yet             
1300        $url = undef;
1301        }
1302        else {
1303        # In the case of server.exe, need to do extra work to get the proper URL
1304        # But first, need to know whether we're indeed dealing with server.exe:
1305       
1306        # compare the URL's domain to the full URL
1307        # E.g. for http://localhost:8383/greenstone3/cgi-bin, the domain is localhost:8383
1308        my $uri = URI->new( $url );
1309        my $host = $uri->host;
1310        #print STDERR "@@@@@ host: $host\n";
1311        if($url =~ m/http:\/\/$host(\/)?$/) {
1312            #if($url !~ m/http:\/\/$host:$portnumber(\/)?/ || $url =~ m/http:\/\/$host(\/)?$/) {
1313            # (if the URL does not contain the portnumber, OR if the port is implicitly 80 and)                 
1314            # If the domain with http:// prefix is completely the same as the URL, assume server.exe
1315            # then the actual URL is the result of suffixing the port and enterlib properties in llssite.cfg
1316            $url = $url.":".$portnumber.$enterlib;         
1317        } # else, apache web server         
1318       
1319        }           
1320    }
1321    } elsif($gs_mode eq "gs3") {
1322    # Either check build.properties for tomcat.server, tomcat.port and app.name (and default servlet name).
1323    # app.name is stored in app.path by build.xml. Need to move app.name in build.properties from build.xml
1324   
1325    # Or, run the new target get-default-servlet-url
1326    # the output can look like:
1327    #
1328    # Buildfile: build.xml
1329    #   [echo] os.name: Windows Vista
1330    #
1331    # get-default-servlet-url:
1332    #   [echo] http://localhost:8383/greenstone3/library
1333    # BUILD SUCCESSFUL
1334    # Total time: 0 seconds
1335   
1336    #my $output = qx/ant get-default-servlet-url/; # backtick operator, to get STDOUT (else 2>&1)
1337    # - see http://stackoverflow.com/questions/799968/whats-the-difference-between-perls-backticks-system-and-exec
1338   
1339    # The get-default-servlet-url ant target can be run from anywhere by specifying the
1340    # location of GS3's ant build.xml buildfile. Activate.pl can be run from anywhere for GS3
1341    # GSDL3SRCHOME will be set for GS3 by gs3-setup.sh, a step that would have been necessary
1342    # to run the activate.pl script in the first place
1343   
1344    my $full_build_xml = &FileUtils::javaFilenameConcatenate($ENV{'GSDL3SRCHOME'},"build.xml");
1345
1346    my $perl_command = "ant -buildfile \"$full_build_xml\" get-default-servlet-url";
1347   
1348    if (open(PIN, "$perl_command |")) {
1349        while (defined (my $perl_output_line = <PIN>)) {
1350
1351        if($perl_output_line =~ m@http:\/\/(\S*)@) { # grab all the non-whitespace chars
1352            $url="http://".$1;
1353        }
1354        }
1355        close(PIN);
1356    } else {
1357        print STDERR "util::get_library_URL() failed to run $perl_command to work out library URL for $gs_mode\n";
1358    }       
1359    }
1360   
1361    # either the url is still undef or it is now set
1362    #print STDERR "\n@@@@@ final URL:|$url|\n" if $url;     
1363    #print STDERR "\n@@@@@ URL still undef\n" if !$url;
1364
1365    $ENV{'FULL_GREENSTONE_URL_PREFIX'} = $url;
1366
1367    return $url;
1368}
1369
1370
1371# Given a config file (xml or java properties file) and a list/array of regular expressions
1372# that represent property names to match on, this function will return the value for the 1st
1373# matching property name. If the return value is undefined, no matching property was found.
1374sub extract_propvalue_from_file() {
1375    my ($configfile, $propertynames) = @_;
1376
1377    my $value;
1378    unless(open(FIN, "<$configfile")) {
1379    print STDERR "extract_propvalue_from_file(): Unable to open $configfile. $!\n";
1380    return $value; # not initialised
1381    }
1382
1383    # Read the entire file at once, as one single line, then close it
1384    my $filecontents;
1385    {
1386    local $/ = undef;       
1387    $filecontents = <FIN>;
1388    }
1389    close(FIN);
1390
1391    foreach my $regex (@$propertynames) {
1392        ($value) = $filecontents=~ m/$regex\s*(\S*)/s; # read value of the property given by regex up to the 1st space
1393    if($value) {
1394            $value =~ s/^\"//;     # remove any startquotes
1395        $value =~ s/\".*$//;   # remove the 1st endquotes (if any) followed by any xml
1396        last;              # found value for a matching property, break from loop
1397    }
1398    }
1399
1400    return $value;
1401}
1402
1403# Subroutine that sources setup.bash, given GSDLHOME and GSDLOS and
1404# given that perllib is in @INC in order to invoke this subroutine.
1405# Call as follows -- after setting up INC to include perllib and
1406# after setting up GSDLHOME and GSDLOS:
1407#
1408# require util;
1409# &util::setup_greenstone_env($ENV{'GSDLHOME'}, $ENV{'GSDLOS'});
1410#
1411sub setup_greenstone_env() {
1412    my ($GSDLHOME, $GSDLOS) = @_;
1413
1414    #my %env_map = ();
1415    # Get the localised ENV settings of running a localised source setup.bash
1416    # and put it into the ENV here. Need to clear GSDLHOME before running setup
1417    #my $perl_command = "(cd $GSDLHOME; export GSDLHOME=; . ./setup.bash > /dev/null; env)";
1418    my $perl_command = "(cd $GSDLHOME; /bin/bash -c \"export GSDLHOME=; source setup.bash > /dev/null; env\")";     
1419    if (($GSDLOS =~ m/windows/i) && ($^O ne "cygwin"))  {
1420        #$perl_command = "cmd /C \"cd $GSDLHOME&& set GSDLHOME=&& setup.bat > nul&& set\"";
1421        $perl_command = "(cd $GSDLHOME&& set GSDLHOME=&& setup.bat > nul&& set)";
1422    }
1423    if (!open(PIN, "$perl_command |")) {
1424        print STDERR ("Unable to execute command: $perl_command. $!\n");
1425    }
1426
1427    my $lastkey;
1428    while (defined (my $perl_output_line = <PIN>)) {
1429        my($key,$value) = ($perl_output_line =~ m/^([^=]*)[=](.*)$/);
1430        if(defined $key) {
1431            #$env_map{$key}=$value;     
1432            $ENV{$key}=$value;
1433            $lastkey = $key;
1434        } elsif($lastkey && $perl_output_line !~ m/^\s*$/) {
1435            # there was no equals sign in $perl_output_line, so this
1436            # $perl_output_line may be a spillover from the previous
1437            $ENV{$lastkey} = $ENV{$lastkey}."\n".$perl_output_line;
1438        }
1439    }
1440    close (PIN);
1441
1442    # If any keys in $ENV don't occur in Greenstone's localised env
1443    # (stored in $env_map), delete those entries from $ENV
1444    #foreach $key (keys %ENV) {
1445    #   if(!defined $env_map{$key}) {
1446    #       print STDOUT "**** DELETING ENV KEY: $key\tVALUE: $ENV{$key}\n";
1447    #       delete $ENV{$key}; # del $ENV(key, value) pair
1448    #   }
1449    #}
1450    #undef %env_map;
1451}
1452
1453sub get_perl_exec() {   
1454    my $perl_exec = $^X; # may return just "perl"
1455   
1456    if($ENV{'PERLPATH'}) {
1457        # OR: # $perl_exec = &FileUtils::filenameConcatenate($ENV{'PERLPATH'},"perl");
1458        if (($ENV{'GSDLOS'} =~ m/windows/) && ($^O ne "cygwin")) {
1459            $perl_exec = "$ENV{'PERLPATH'}\\Perl.exe";
1460        } else {
1461            $perl_exec = "$ENV{'PERLPATH'}/perl";
1462        }
1463    } else { # no PERLPATH, use Config{perlpath} else $^X: special variables
1464        # containing the full path to the current perl executable we're using
1465        $perl_exec = $Config{perlpath}; # configured path for perl
1466        if (!-e $perl_exec) { # may not point to location on this machine
1467            $perl_exec = $^X; # may return just "perl"
1468            if($perl_exec =~ m/^perl/i) { # warn if just perl or Perl.exe
1469                print STDERR "**** WARNING: Perl exec found contains no path: $perl_exec";             
1470            }
1471        }
1472    }
1473   
1474    return $perl_exec;
1475}
1476
1477# returns the path to the java command in the JRE included with GS (if any),
1478# quoted to safeguard any spaces in this path, otherwise a simple java
1479# command is returned which assumes and will try for a system java.
1480sub get_java_command {
1481    my $java = "java";
1482    if(defined $ENV{'GSDLHOME'}) { # should be, as this script would be launched from the cmd line
1483                               # after running setup.bat or from GLI which also runs setup.bat
1484    my $java_bin = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"packages","jre","bin");
1485    if(-d $java_bin) {
1486        $java = &FileUtils::filenameConcatenate($java_bin,"java");
1487        $java = "\"".$java."\""; # quoted to preserve spaces in path
1488    }
1489    }
1490    return $java;
1491}
1492
1493
1494# Given the qualified collection name (colgroup/collection),
1495# returns the collection and colgroup parts
1496sub get_collection_parts {
1497    # http://perldoc.perl.org/File/Basename.html
1498    # my($filename, $directories, $suffix) = fileparse($path);
1499    # "$directories contains everything up to and including the last directory separator in the $path
1500    # including the volume (if applicable). The remainder of the $path is the $filename."
1501    #my ($collection, $colgroup) = &File::Basename::fileparse($qualified_collection);   
1502
1503    my $qualified_collection = shift(@_);
1504
1505    # Since activate.pl can be launched from the command-line, including by a user,
1506    # best not to assume colgroup uses URL-style slashes as would be the case with GLI
1507    # Also allow for the accidental inclusion of multiple slashes
1508    my ($colgroup, $collection) = split(/[\/\\]+/, $qualified_collection); #split('/', $qualified_collection);
1509   
1510    if(!defined $collection) {
1511        $collection = $colgroup;
1512        $colgroup = "";
1513    }
1514    return ($collection, $colgroup);
1515}
1516
1517# work out the "collectdir/collection" location
1518sub resolve_collection_dir {
1519    my ($collect_dir, $qualified_collection, $site) = @_; #, $gs_mode
1520   
1521    if (defined $ENV{'GSDLCOLLECTDIR'}) { # a predefined collection dir exists
1522        return $ENV{'GSDLCOLLECTDIR'};
1523    }
1524
1525    my ($colgroup, $collection) = &util::get_collection_parts($qualified_collection);   
1526   
1527    if (!defined $collect_dir || !$collect_dir) { # if undefined or empty string
1528        $collect_dir = &util::get_working_collect_dir($site);
1529    }
1530
1531    return &FileUtils::filenameConcatenate($collect_dir,$colgroup,$collection);
1532}
1533
1534# work out the full path to "collect" of this greenstone 2/3 installation
1535sub get_working_collect_dir {
1536    my ($site) = @_;   
1537
1538    if (defined $ENV{'GSDLCOLLECTHOME'}) { # a predefined collect dir exists
1539    return $ENV{'GSDLCOLLECTHOME'};
1540    }
1541
1542    if (defined $site && $site) { # site non-empty, so get default collect dir for GS3
1543   
1544    if (defined $ENV{'GSDL3HOME'}) {
1545        return &FileUtils::filenameConcatenate($ENV{'GSDL3HOME'},"sites",$site,"collect"); # web folder
1546    }
1547    elsif (defined $ENV{'GSDL3SRCHOME'}) {
1548        return &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'},"web","sites",$site,"collect");
1549    }
1550    }
1551
1552    elsif (defined $ENV{'SITEHOME'}) {
1553    return &FileUtils::filenameConcatenate($ENV{'SITEHOME'},"collect");
1554    }
1555   
1556    else { # get default collect dir for GS2
1557    return &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"collect");
1558    }
1559}
1560
1561sub is_abs_path_any_os {
1562    my ($path) = @_;
1563
1564    # We can have filenames in our DBs that were produced on other OS, so this method exists
1565    # to help identify absolute paths in such cases.
1566
1567    return 1 if($path =~ m@^/@); # full paths begin with forward slash on linux/mac
1568    return 1 if($path =~ m@^([a-zA-Z]\:|\\)@); # full paths begin with drive letter colon for Win or \ for volume, http://stackoverflow.com/questions/13011013/get-only-volume-name-from-filepath
1569
1570    return 0;
1571}
1572
1573
1574# This subroutine is for improving portability of Greenstone collections from one OS to another,
1575# to be used to convert absolute paths going into db files into paths with placeholders instead.
1576# This sub works with util::get_common_gs_paths and takes a path to a greenstone file and, if it's
1577# an absolute path, then it will replace the longest matching greenstone-path prefix of the given
1578# path with a placeholder to match.
1579# The Greenstone-path prefixes that can be matched are the following common Greenstone paths:
1580# the path to the current (specific) collection, the path to the general GS collect directory,
1581# the path to the site directory if GS3, else the path to the GSDLHOME/GSDL3HOME folder.
1582# The longest matching prefix will be replaced with the equivalent placeholder:
1583# @THISCOLLECTPATH@, else @COLLECTHOME@, else @SITEHOME@, else @GSDLHOME@.
1584sub abspath_to_placeholders {
1585    my $path = shift(@_); # path to convert from absolute to one with placeholders
1586    my $opt_long_or_short_winfilenames = shift(@_) || "short"; # whether we want to force use of long file names even on windows, default uses short
1587
1588    return $path unless is_abs_path_any_os($path); # path is relative
1589
1590    if ($opt_long_or_short_winfilenames eq "long") {
1591    $path = &util::upgrade_if_dos_filename($path); # will only do something on windows
1592    }
1593   
1594    # now we know we're dealing with absolute paths and have to replace gs prefixes with placeholders
1595    my @gs_paths = ($ENV{'GSDLCOLLECTDIR'}, $ENV{'GSDLCOLLECTHOME'}, $ENV{'SITEHOME'}, $ENV{'GREENSTONEHOME'}); # list in this order: from longest to shortest path
1596
1597    my %placeholder_map = ($ENV{'GREENSTONEHOME'} => '@GSDLHOME@', # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
1598               $ENV{'GSDLCOLLECTHOME'} => '@COLLECTHOME@',
1599               $ENV{'GSDLCOLLECTDIR'} => '@THISCOLLECTPATH@'
1600    );
1601    $placeholder_map{$ENV{'SITEHOME'}} = '@SITEHOME@' if defined $ENV{'SITEHOME'};
1602
1603    $path = &util::_abspath_to_placeholders($path, \@gs_paths, \%placeholder_map);
1604
1605    if ($ENV{'GSDLOS'} =~ /^windows$/i && $opt_long_or_short_winfilenames eq "short") {
1606    # for windows need to look for matches on short file names too
1607    # matched paths are again to be replaced with the usual placeholders
1608
1609    my $gsdlcollectdir = &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTDIR'});
1610    my $gsdlcollecthome = &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTHOME'});
1611    my $sitehome = (defined $ENV{'SITEHOME'}) ? &util::downgrade_if_dos_filename($ENV{'SITEHOME'}) : undef;
1612    my $greenstonehome =  &util::downgrade_if_dos_filename($ENV{'GREENSTONEHOME'});
1613
1614    @gs_paths = ($gsdlcollectdir, $gsdlcollecthome, $sitehome, $greenstonehome); # order matters
1615
1616    %placeholder_map = ($greenstonehome => '@GSDLHOME@', # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
1617                $gsdlcollecthome => '@COLLECTHOME@',
1618                $gsdlcollectdir => '@THISCOLLECTPATH@'
1619        );
1620    $placeholder_map{$sitehome} = '@SITEHOME@' if defined $sitehome;
1621
1622    $path = &util::_abspath_to_placeholders($path, \@gs_paths, \%placeholder_map);
1623    }
1624
1625    return $path;
1626}
1627
1628sub _abspath_to_placeholders {
1629    my ($path, $gs_paths_ref, $placeholder_map_ref) = @_;
1630
1631    # The sequence of elements in @gs_paths matters   
1632    # Need to loop starting from the *longest* matching path (the path to the specific collection)
1633    # to the shortest matching path (the path to gsdlhome/gsdl3home folder):
1634
1635    foreach my $gs_path (@$gs_paths_ref) {
1636    next if(!defined $gs_path); # site undefined for GS2
1637
1638    my $re_path =  &util::filename_to_regex($gs_path); # escape for regex
1639
1640    if($path =~ m/^$re_path/i) { # case sensitive or not for OS?
1641
1642        my $placeholder = $placeholder_map_ref->{$gs_path}; # get the placeholder to replace the matched path with
1643
1644        $path =~ s/^$re_path/$placeholder/; #case sensitive or not?
1645        #$path =~ s/^[\\\/]//; # remove gs_path's trailing separator left behind at the start of the path
1646        # lowercase file extension, This is needed when shortfilenames are used, as case affects alphetical ordering, which affects diffcol     
1647        $path =~ s/\.([A-Z]+)$/".".lc($1)/e;
1648        last; # done
1649    }
1650    }
1651   
1652    return $path;
1653}
1654
1655# Function that does the reverse of the util::abspath_to_placeholders subroutine
1656# Once again, call this with the values returned from util::get_common_gs_paths
1657sub placeholders_to_abspath {
1658    my $path = shift(@_); # path that can contain placeholders to convert to resolved absolute path
1659    my $opt_long_or_short_winfilenames = shift(@_) || "short"; # whether we want to force use of long file names even on windows, default uses short
1660
1661    return $path if($path !~ m/@/); # path contains no placeholders
1662   
1663    # replace placeholders with gs prefixes
1664    my @placeholders = ('@THISCOLLECTPATH@', '@COLLECTHOME@', '@SITEHOME@', '@GSDLHOME@'); # order of paths not crucial in this case,
1665                       # but listed here from longest to shortest once placeholders are have been resolved
1666
1667    # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
1668    my %placeholder_to_gspath_map;
1669    if ($ENV{'GSDLOS'} =~ /^windows$/i && $opt_long_or_short_winfilenames eq "short") {
1670    # always replace placeholders with short file names of the absolute paths on windows?
1671    %placeholder_to_gspath_map = ('@GSDLHOME@' => &util::downgrade_if_dos_filename($ENV{'GREENSTONEHOME'}),
1672                     '@COLLECTHOME@' => &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTHOME'}),
1673                     '@THISCOLLECTPATH@' => &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTDIR'})
1674    );
1675    $placeholder_to_gspath_map{'@SITEHOME@'} =  &util::downgrade_if_dos_filename($ENV{'SITEHOME'}) if defined $ENV{'SITEHOME'};
1676    } else {
1677    %placeholder_to_gspath_map = ('@GSDLHOME@' => $ENV{'GREENSTONEHOME'},
1678                      '@SITEHOME@' => $ENV{'SITEHOME'}, # can be undef
1679                      '@COLLECTHOME@' => $ENV{'GSDLCOLLECTHOME'},
1680                      '@THISCOLLECTPATH@' => $ENV{'GSDLCOLLECTDIR'}
1681        ); # $placeholder_to_gspath_map{'@SITEHOME@'} = $ENV{'SITEHOME'} if defined $ENV{'SITEHOME'};
1682    }
1683
1684    foreach my $placeholder (@placeholders) {
1685    my $gs_path = $placeholder_to_gspath_map{$placeholder};
1686
1687    next if(!defined $gs_path); # sitehome for GS2 is undefined
1688
1689    if($path =~ m/^$placeholder/) {
1690        $path =~ s/^$placeholder/$gs_path/;
1691        last; # done
1692    }
1693    }
1694   
1695    return $path;
1696}
1697
1698# Used by pdfpstoimg.pl and PDFBoxConverter to create a .item file from
1699# a directory containing sequentially numbered images.
1700sub create_itemfile
1701{
1702    my ($output_dir, $convert_basename, $convert_to) = @_;
1703    my $page_num = "";
1704
1705    opendir(DIR, $output_dir) || die "can't opendir $output_dir: $!";
1706    my @dir_files = grep {-f "$output_dir/$_"} readdir(DIR);
1707    closedir DIR;
1708
1709    # Sort files in the directory by page_num
1710    sub page_number {
1711    my ($dir) = @_;
1712    my ($pagenum) =($dir =~ m/^.*?[-\.]?(\d+)(\.(jpg|gif|png))?$/i);
1713#   my ($pagenum) =($dir =~ m/(\d+)(\.(jpg|gif|png))?$/i); # this works but is not as safe/strict about input filepatterns as the above
1714
1715    $pagenum = 1 unless defined $pagenum;
1716    return $pagenum;
1717    }
1718
1719    # sort the files in the directory in the order of page_num rather than lexically.
1720    @dir_files = sort { page_number($a) <=> page_number($b) } @dir_files;
1721
1722    # work out if the numbering of the now sorted image files starts at 0 or not
1723    # by checking the number of the first _image_ file (skipping item files)
1724    my $starts_at_0 = 0;
1725    my $firstfile = ($dir_files[0] !~ /\.item$/i) ? $dir_files[0] : $dir_files[1];
1726    if(page_number($firstfile) == 0) { # 00 will evaluate to 0 too in this condition
1727    $starts_at_0 = 1;
1728    }
1729
1730    my $item_file = &FileUtils::filenameConcatenate($output_dir, $convert_basename.".item");
1731    my $item_fh;
1732    &FileUtils::openFileHandle($item_file, 'w', \$item_fh);
1733    print $item_fh "<PagedDocument>\n";
1734
1735    foreach my $file (@dir_files){
1736    if ($file !~ /\.item/i){
1737        $page_num = page_number($file);
1738        $page_num++ if $starts_at_0; # image numbers start at 0, so add 1
1739        print $item_fh "   <Page pagenum=\"$page_num\" imgfile=\"$file\" txtfile=\"\"/>\n";
1740    }
1741    }
1742
1743    print $item_fh "</PagedDocument>\n";
1744    &FileUtils::closeFileHandle($item_file, \$item_fh);
1745    return $item_file;
1746}
1747
1748# Sets the gnomelib_env. Based on the logic in wvware.pl which can perhaps be replaced with a call to this function in future
1749sub set_gnomelib_env
1750{
1751    ## SET THE ENVIRONMENT AS DONE IN SETUP.BASH/BAT OF GNOME-LIB
1752    # Though this is only needed for darwin Lion at this point (and android, though that is untested)
1753
1754    my $libext = "so";
1755    if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
1756    return;
1757    } elsif ($ENV{'GSDLOS'} =~ m/^darwin$/i) {
1758    $libext = "dylib";
1759    }
1760
1761    if (!defined $ENV{'GEXTGNOME'}) {
1762        ##print STDERR "@@@ Setting GEXTGNOME env\n";
1763
1764    my $gnome_dir = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"ext","gnome-lib-minimal");
1765
1766    if(! -d $gnome_dir) {
1767        $gnome_dir = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"ext","gnome-lib");
1768
1769        if(! -d $gnome_dir) {
1770        $gnome_dir = "";
1771        }
1772    }
1773   
1774    # now set other the related env vars,
1775    # IF we've found the gnome-lib dir installed in the ext folder 
1776
1777    if ($gnome_dir ne "" && -f &FileUtils::filenameConcatenate($gnome_dir, $ENV{'GSDLOS'}, "lib", "libiconv.$libext")) {
1778        $ENV{'GEXTGNOME'} = $gnome_dir;
1779        $ENV{'GEXTGNOME_INSTALLED'}=&FileUtils::filenameConcatenate($ENV{'GEXTGNOME'}, $ENV{'GSDLOS'});
1780       
1781        my $gnomelib_bin = &FileUtils::filenameConcatenate($ENV{'GEXTGNOME_INSTALLED'}, "bin");
1782        if(-d $gnomelib_bin) { # no bin subfolder in GS binary's cutdown gnome-lib-minimal folder
1783        &util::envvar_prepend("PATH", $gnomelib_bin);
1784        }
1785
1786        # util's prepend will create LD/DYLD_LIB_PATH if it doesn't exist yet
1787        my $gextlib = &FileUtils::filenameConcatenate($ENV{'GEXTGNOME_INSTALLED'}, "lib");
1788
1789        if($ENV{'GSDLOS'} eq "linux") {
1790        &util::envvar_prepend("LD_LIBRARY_PATH", $gextlib);
1791        }
1792        elsif ($ENV{'GSDLOS'} eq "darwin") {
1793        #&util::envvar_prepend("DYLD_LIBRARY_PATH", $gextlib);
1794        &util::envvar_prepend("DYLD_FALLBACK_LIBRARY_PATH", $gextlib);
1795        }
1796    }
1797   
1798    # Above largely mimics the setup.bash of the gnome-lib-minimal.
1799    # Not doing the devel-srcpack that gnome-lib-minimal's setup.bash used to set
1800    # Not exporting GSDLEXTS variable either
1801    }
1802
1803#    print STDERR "@@@@@ GEXTGNOME: ".$ENV{'GEXTGNOME'}."\n\tINSTALL".$ENV{'GEXTGNOME_INSTALLED'}."\n";
1804#    print STDERR "\tPATH".$ENV{'PATH'}."\n";
1805#    print STDERR "\tLD_LIB_PATH".$ENV{'LD_LIBRARY_PATH'}."\n" if $ENV{'LD_LIBRARY_PATH};
1806#    print STDERR "\tDYLD_FALLBACK_LIB_PATH".$ENV{'DYLD_FALLBACK_LIBRARY_PATH'}."\n" if $ENV{'DYLD_FALLBACK_LIBRARY_PATH};
1807
1808    # if no GEXTGNOME, maybe users didn't need gnome-lib to run gnomelib/libiconv dependent binaries like hashfile, suffix, wget
1809    # (wvware is launched in a gnomelib env from its own script, but could possibly go through this script in future)
1810}
1811
1812
1813
1814## @function augmentINC()
1815#
1816#  Prepend a path (if it exists) onto INC but only if it isn't already in INC
1817#  @param $new_path The path to add
1818#  @author jmt12
1819#
1820sub augmentINC
1821{
1822  my ($new_path) = @_;
1823  my $did_add_path = 0;
1824  # might need to be replaced with FileUtils::directoryExists() call eventually
1825  if (-d $new_path)
1826  {
1827    my $did_find_path = 0;
1828    foreach my $existing_path (@INC)
1829    {
1830      if ($existing_path eq $new_path)
1831      {
1832        $did_find_path = 1;
1833        last;
1834      }
1835    }
1836    if (!$did_find_path)
1837    {
1838      unshift(@INC, $new_path);
1839      $did_add_path = 1;
1840    }
1841  }
1842  return $did_add_path;
1843}
1844## augmentINC()
1845
1846
18471;
Note: See TracBrowser for help on using the browser.