root/main/trunk/greenstone2/perllib/util.pm @ 31694

Revision 31694, 61.8 KB (checked in by kjdon, 3 years ago)

testing on commdev I had the situation where the apache user couldn't write to gsdlhome, so it needed to use the tmp folder. But there is no TMP environment variable set (which I think was added for windows). so now try /tmp folder if all else fails

  • 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        if (defined $ENV{'TMP'}) {
265            $tmp_dirname = $ENV{'TMP'};
266        } else {
267            $tmp_dirname = "/tmp";
268        }
269        $tmp_dirname = &FileUtils::filenameConcatenate($tmp_dirname, "greenstone");
270            $ENV{'GS_TMP_OUTPUT_DIR'} = $tmp_dirname; # store for next time
271        }
272    }
273   
274    $tmp_dirname = &FileUtils::filenameConcatenate($tmp_dirname, "tmp");
275    &FileUtils::makeAllDirectories ($tmp_dirname) unless -e $tmp_dirname;
276
277    return $tmp_dirname;
278}
279
280sub get_tmp_filename
281{
282    my $file_ext = shift(@_) || undef;
283
284    my $opt_dot_file_ext = "";
285    if (defined $file_ext) {
286    if ($file_ext !~ m/\./) {
287        # no dot, so needs one added in at start
288        $opt_dot_file_ext = ".$file_ext"
289    }
290    else {
291        # allow for "extensions" such as _metadata.txt to be handled
292        # gracefully
293        $opt_dot_file_ext = $file_ext;
294    }
295    }
296
297    my $tmpdir = &util::determine_tmp_dir(0);
298
299    my $count = 1000;
300    my $rand = int(rand $count);
301    my $full_tmp_filename = &FileUtils::filenameConcatenate($tmpdir, "F$rand$opt_dot_file_ext");
302
303    while (-e $full_tmp_filename) {
304    $rand = int(rand $count);
305    $full_tmp_filename = &FileUtils::filenameConcatenate($tmpdir, "F$rand$opt_dot_file_ext");
306    $count++;
307    }
308   
309    return $full_tmp_filename;
310}
311
312# These 2 are "static" variables used by the get_timestamped_tmp_folder() subroutine below and
313# belong with that function. They help ensure the timestamped tmp folders generated are unique.
314my $previous_timestamp = undef;
315my $previous_timestamp_f = 0; # frequency
316
317sub get_timestamped_tmp_folder
318{
319    my $tmp_dirname = &util::determine_tmp_dir(1);
320   
321    # add the timestamp into the path otherwise we can run into problems
322    # if documents have the same name
323    my $timestamp = time;   
324   
325    if (!defined $previous_timestamp || ($timestamp > $previous_timestamp)) {
326        $previous_timestamp_f = 0;
327        $previous_timestamp = $timestamp;
328    } else {
329        $previous_timestamp_f++;
330    }
331   
332    my $time_tmp_dirname = &FileUtils::filenameConcatenate($tmp_dirname, $timestamp);
333    $tmp_dirname = $time_tmp_dirname;   
334    my $i = $previous_timestamp_f;
335   
336    if($previous_timestamp_f > 0) {
337        $tmp_dirname = $time_tmp_dirname."_".$i;
338        $i++;
339    }
340    while (-e $tmp_dirname) {
341    $tmp_dirname = $time_tmp_dirname."_".$i;
342    $i++;
343    }
344    &FileUtils::makeDirectory($tmp_dirname);
345
346    return $tmp_dirname;
347}
348
349sub get_timestamped_tmp_filename_in_collection
350{
351
352    my ($input_filename, $output_ext) = @_;
353    # derive tmp filename from input filename
354    my ($tailname, $dirname, $suffix)
355    = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
356
357    # softlink to collection tmp dir
358    my $tmp_dirname = &util::get_timestamped_tmp_folder();
359    $tmp_dirname = $dirname unless defined $tmp_dirname;
360
361    # following two steps copied from ConvertBinaryFile
362    # do we need them?? can't use them as is, as they use plugin methods.
363
364    #$tailname = $self->SUPER::filepath_to_utf8($tailname) unless &unicode::check_is_utf8($tailname);
365
366    # URLEncode this since htmls with images where the html filename is utf8 don't seem
367    # to work on Windows (IE or Firefox), as browsers are looking for filesystem-encoded
368    # files on the filesystem.
369    #$tailname = &util::rename_file($tailname, $self->{'file_rename_method'}, "without_suffix");
370    if (defined $output_ext) {
371    $output_ext = ".$output_ext"; # add the dot
372    } else {
373    $output_ext = $suffix;
374    }
375    $output_ext= lc($output_ext);
376    my $tmp_filename = &FileUtils::filenameConcatenate($tmp_dirname, "$tailname$output_ext");
377   
378    return $tmp_filename;
379}
380
381sub get_toplevel_tmp_dir
382{
383    return &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "tmp");
384}
385
386
387sub get_collectlevel_tmp_dir
388{
389    my $tmp_dirname = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "tmp");
390    &FileUtils::makeDirectory($tmp_dirname) if (!-e $tmp_dirname);
391
392    return $tmp_dirname;
393}
394
395sub get_parent_folder
396{
397    my ($path) = @_;
398    my ($tailname, $dirname, $suffix)
399    = &File::Basename::fileparse($path, "\\.[^\\.]+\$");
400
401    return &FileUtils::sanitizePath($dirname);
402}
403
404sub filename_to_regex {
405    my $filename = shift (@_);
406
407    # need to make single backslashes double so that regex works
408    $filename =~ s/\\/\\\\/g; # if ($ENV{'GSDLOS'} =~ /^windows$/i);   
409   
410    # note that the first part of a substitution is a regex, so RE chars need to be escaped,
411    # the second part of a substitution is not a regex, so for e.g. full-stop can be specified literally
412    $filename =~ s/\./\\./g; # in case there are extensions/other full stops, escape them
413    $filename =~ s@\(@\\(@g; # escape brackets
414    $filename =~ s@\)@\\)@g; # escape brackets
415    $filename =~ s@\[@\\[@g; # escape brackets
416    $filename =~ s@\]@\\]@g; # escape brackets
417   
418    return $filename;
419}
420
421sub unregex_filename {
422    my $filename = shift (@_);
423
424    # need to put doubled backslashes for regex back to single
425    $filename =~ s/\\\./\./g; # remove RE syntax for .
426    $filename =~ s@\\\(@(@g; # remove RE syntax for ( => "\(" turns into "("
427    $filename =~ s@\\\)@)@g; # remove RE syntax for ) => "\)" turns into ")"
428    $filename =~ s@\\\[@[@g; # remove RE syntax for [ => "\[" turns into "["
429    $filename =~ s@\\\]@]@g; # remove RE syntax for ] => "\]" turns into "]"
430   
431    # \\ goes to \
432    # This is the last step in reverse mirroring the order of steps in filename_to_regex()
433    $filename =~ s/\\\\/\\/g; # remove RE syntax for \   
434    return $filename;
435}
436
437sub filename_cat {
438  # I've disabled this warning for now, as every Greenstone perl
439  # script seems to make use of this function and so you drown in a
440  # sea of deprecated warnings [jmt12]
441#  warnings::warnif("deprecated", "util::filename_cat() is deprecated, using FileUtils::filenameConcatenate() instead");
442  return &FileUtils::filenameConcatenate(@_);
443}
444
445
446sub _pathname_cat {
447    my $join_char  = shift(@_);
448    my $first_path = shift(@_);
449    my (@pathnames) = @_;
450
451    # If first_path is not null or empty, then add it back into the list
452    if (defined $first_path && $first_path =~ /\S/) {
453    unshift(@pathnames, $first_path);
454    }
455
456    my $pathname = join($join_char, @pathnames);
457
458    # remove duplicate slashes
459    if ($join_char eq ";") {
460    $pathname =~ s/[\\\/]+/\\/g;
461    if ($^O eq "cygwin") {
462        # Once we've collapsed muliple (potentialy > 2) slashes
463        # For cygwin, actually want things double-backslahed
464        $pathname =~ s/\\/\\\\/g;
465    }
466
467    } else {
468    $pathname =~ s/[\/]+/\//g;
469    # DB: want a pathname abc\de.html to remain like this
470    }
471
472    return $pathname;
473}
474
475
476sub pathname_cat {
477    my (@pathnames) = @_;
478
479    my $join_char;
480    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
481    $join_char = ";";
482    } else {
483    $join_char = ":";
484    }
485    return _pathname_cat($join_char,@pathnames);
486}
487
488
489sub javapathname_cat {
490    my (@pathnames) = @_;
491
492    my $join_char;
493
494    # Unlike pathname_cat() above, not interested if running in a Cygwin environment
495    # This is because the java we run is actually a native Windows executable
496
497    if (($ENV{'GSDLOS'} =~ /^windows$/i)) {
498    $join_char = ";";
499    } else {
500    $join_char = ":";
501    }
502    return _pathname_cat($join_char,@pathnames);
503}
504
505
506sub makeFilenameJavaCygwinCompatible
507{
508    my ($java_filename) = @_;
509
510    if ($^O eq "cygwin") {
511    # To be used with a Java program, but under Cygwin
512    # Because the java binary that is native to Windows, need to
513    # convert the Cygwin paths (i.e. Unix style) to be Windows
514    # compatible
515   
516    $java_filename = `cygpath -wp "$java_filename"`;
517    chomp($java_filename);
518    $java_filename =~ s%\\%\\\\%g;
519    }
520
521    return $java_filename;
522}
523
524sub tidy_up_oid {
525    my ($OID) = @_;
526    if ($OID =~ /[\.\/\\]/) {
527    print STDERR "Warning, identifier $OID contains periods or slashes(.\\\/), replacing them with _\n";
528    $OID =~ s/[\.\\\/]/_/g; #remove any periods
529    }
530    if ($OID =~ /^\s.*\s$/) {
531    print STDERR "Warning, identifier $OID starts or ends with whitespace. Removing it\n";
532    # remove starting and trailing whitespace
533    $OID =~ s/^\s+//;
534    $OID =~ s/\s+$//;
535    }
536    if ($OID =~ /^[\d]*$/) {
537    print STDERR "Warning, identifier $OID contains only digits. Prepending 'D'.\n";
538    $OID = "D" . $OID;
539    }       
540   
541    return $OID;
542}
543
544sub envvar_prepend {
545    my ($var,$val) = @_;
546
547    # 64 bit linux can't handle ";" as path separator, so make sure to set this to the right one for the OS
548##    my $pathsep = (defined $ENV{'GSDLOS'} && $ENV{'GSDLOS'} !~ m/windows/) ? ":" : ";";
549
550    # Rewritten above to make ":" the default (Windows is the special
551    # case, anything else 'unusual' such as Solaris etc is Unix)
552    my $pathsep = (defined $ENV{'GSDLOS'} && (($ENV{'GSDLOS'} =~ m/windows/) && ($^O ne "cygwin"))) ? ";" : ":";
553
554    # do not prepend any value/path that's already in the environment variable
555   
556    my $escaped_val = &filename_to_regex($val); # escape any backslashes and brackets for upcoming regex
557    if (!defined($ENV{$var})) {
558    $ENV{$var} = "$val";
559    }
560    elsif($ENV{$var} !~ m/$escaped_val/) {
561    $ENV{$var} = "$val".$pathsep.$ENV{$var};
562    }
563}
564
565sub envvar_append {
566    my ($var,$val) = @_;
567
568    # 64 bit linux can't handle ";" as path separator, so make sure to set this to the right one for the OS
569    my $pathsep = (defined $ENV{'GSDLOS'} && $ENV{'GSDLOS'} !~ m/windows/) ? ":" : ";";
570   
571    # do not append any value/path that's already in the environment variable
572
573    my $escaped_val = &filename_to_regex($val); # escape any backslashes and brackets for upcoming regex
574    if (!defined($ENV{$var})) {
575    $ENV{$var} = "$val";
576    }
577    elsif($ENV{$var} !~ m/$escaped_val/) {
578    $ENV{$var} = $ENV{$var}.$pathsep."$val";
579    }
580}
581
582
583
584# splits a filename into a prefix and a tail extension using the tail_re, or
585# if that fails, splits on the file_extension . (dot)
586sub get_prefix_and_tail_by_regex {
587
588    my ($filename,$tail_re) = @_;
589   
590    my ($file_prefix,$file_ext) = ($filename =~ m/^(.*?)($tail_re)$/);
591    if ((!defined $file_prefix) || (!defined $file_ext)) {
592    ($file_prefix,$file_ext) = ($filename =~ m/^(.*)(\..*?)$/);
593    }
594
595    return ($file_prefix,$file_ext);
596}
597
598# get full path and file only path from a base_dir (which may be empty) and
599# file (which may contain directories)
600sub get_full_filenames {
601    my ($base_dir, $file) = @_;
602   
603#    my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(0);
604#    my ($lcfilename) = ($cfilename =~ m/([^\\\/]*)$/);
605#    print STDERR "** Calling method: $lcfilename:$cline $cpackage->$csubr\n";
606
607
608    my $filename_full_path = $file;
609    # add on directory if present
610    $filename_full_path = &FileUtils::filenameConcatenate($base_dir, $file) if $base_dir =~ /\S/;
611   
612    my $filename_no_path = $file;
613
614    # remove directory if present
615    $filename_no_path =~ s/^.*[\/\\]//;
616    return ($filename_full_path, $filename_no_path);
617}
618
619# returns the path of a file without the filename -- ie. the directory the file is in
620sub filename_head {
621    my $filename = shift(@_);
622
623    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
624    $filename =~ s/[^\\\\]*$//;
625    }
626    else {
627    $filename =~ s/[^\\\/]*$//;
628    }
629
630    return $filename;
631}
632
633
634
635# returns 1 if filename1 and filename2 point to the same
636# file or directory
637sub filenames_equal {
638    my ($filename1, $filename2) = @_;
639
640    # use filename_cat to clean up trailing slashes and
641    # multiple slashes
642    $filename1 = &FileUtils::filenameConcatenate($filename1);
643    $filename2 = &FileUtils::filenameConcatenate($filename2);
644
645    # filenames not case sensitive on windows
646    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
647    $filename1 =~ tr/[A-Z]/[a-z]/;
648    $filename2 =~ tr/[A-Z]/[a-z]/;
649    }
650    return 1 if $filename1 eq $filename2;
651    return 0;
652}
653
654# If filename is relative to within_dir, returns the relative path of filename to that directory
655# with slashes in the filename returned as they were in the original (absolute) filename.
656sub filename_within_directory
657{
658    my ($filename,$within_dir) = @_;
659   
660    if ($within_dir !~ m/[\/\\]$/) {
661    my $dirsep = &util::get_dirsep();
662    $within_dir .= $dirsep;
663    }
664   
665    $within_dir = &filename_to_regex($within_dir); # escape DOS style file separator and brackets   
666    if ($filename =~ m/^$within_dir(.*)$/) {
667    $filename = $1;
668    }
669   
670    return $filename;
671}
672
673# If filename is relative to within_dir, returns the relative path of filename to that directory in URL format.
674# Filename and within_dir can be any type of slashes, but will be compared as URLs (i.e. unix-style slashes).
675# The subpath returned will also be a URL type filename.
676sub filename_within_directory_url_format
677{
678    my ($filename,$within_dir) = @_;
679   
680    # convert parameters only to / slashes if Windows
681   
682    my $filename_urlformat = &filepath_to_url_format($filename);
683    my $within_dir_urlformat = &filepath_to_url_format($within_dir);
684
685    #if ($within_dir_urlformat !~ m/\/$/) {
686        # make sure directory ends with a slash
687        #$within_dir_urlformat .= "/";
688    #}
689   
690    my $within_dir_urlformat_re = &filename_to_regex($within_dir_urlformat); # escape any special RE characters, such as brackets
691   
692    #print STDERR "@@@@@ $filename_urlformat =~ $within_dir_urlformat_re\n";
693   
694    # dir prefix may or may not end with a slash (this is discarded when extracting the sub-filepath)
695    if ($filename_urlformat =~ m/^$within_dir_urlformat_re(?:\/)*(.*)$/) {
696        $filename_urlformat = $1;
697    }
698   
699    return $filename_urlformat;
700}
701
702# Convert parameter to use / slashes if Windows (if on Linux leave any \ as is,
703# since on Linux it doesn't represent a file separator but an escape char).
704sub filepath_to_url_format
705{
706    my ($filepath) = @_;
707    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
708        # Only need to worry about Windows, as Unix style directories already in url-format
709        # Convert Windows style \ => /
710        $filepath =~ s@\\@/@g;     
711    }
712    return $filepath;
713}
714
715# regex filepaths on windows may include \\ as path separator. Convert \\ to /
716sub filepath_regex_to_url_format
717{
718    my ($filepath) = @_;
719    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
720    # Only need to worry about Windows, as Unix style directories already in url-format
721    # Convert Windows style \\ => /
722    $filepath =~ s@\\\\@/@g;       
723    }
724    return $filepath;
725   
726}
727
728# Like File::Basename::fileparse, but expects filepath in url format (ie only / slash for dirsep)
729# and ignores trailing /
730# returns (file, dirs) dirs will be empty if no subdirs
731sub url_fileparse
732{
733    my ($filepath) = @_;
734    # remove trailing /
735    $filepath =~ s@/$@@;
736    if ($filepath !~ m@/@) {
737    return ($filepath, "");
738    }
739    my ($dirs, $file) = $filepath =~ m@(.+/)([^/]+)@;
740    return ($file, $dirs);
741   
742}
743
744
745sub filename_within_collection
746{
747    my ($filename) = @_;
748
749    my $collect_dir = $ENV{'GSDLCOLLECTDIR'};
750   
751    if (defined $collect_dir) {
752
753    # if from within GSDLCOLLECTDIR, then remove directory prefix
754    # so source_filename is realative to it.  This is done to aid
755    # portability, i.e. the collection can be moved to somewhere
756    # else on the file system and the archives directory will still
757    # work.  This is needed, for example in the applet version of
758    # GLI where GSDLHOME/collect on the server will be different to
759    # the collect directory of the remove user.  Of course,
760    # GSDLCOLLECTDIR subsequently needs to be put back on to turn
761    # it back into a full pathname.
762
763    $filename = filename_within_directory($filename,$collect_dir);
764    }
765   
766    return $filename;
767}
768
769sub prettyprint_file
770{
771    my ($base_dir,$file,$gli) = @_;
772
773    my $filename_full_path = &FileUtils::filenameConcatenate($base_dir,$file);
774
775    if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
776    require Win32;
777
778    # For some reason base_dir in the form c:/a/b/c
779    # This leads to confusion later on, so turn it back into
780    # the more usual Windows form
781    $base_dir =~ s/\//\\/g;
782    my $long_base_dir = Win32::GetLongPathName($base_dir);
783    my $long_full_path = Win32::GetLongPathName($filename_full_path);
784
785    $file = filename_within_directory($long_full_path,$long_base_dir);
786    $file = encode("utf8",$file) if ($gli);
787    }
788
789    return $file;
790}
791
792
793sub upgrade_if_dos_filename
794{
795    my ($filename_full_path,$and_encode) = @_;
796
797    if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
798    # Ensure any DOS-like filename, such as test~1.txt, has been upgraded
799    # to its long (Windows) version
800    my $long_filename = Win32::GetLongPathName($filename_full_path);
801    if (defined $long_filename) {
802       
803        $filename_full_path = $long_filename;
804    }
805    # Make sure initial drive letter is lower-case (to fit in with rest of Greenstone)
806    $filename_full_path =~ s/^(.):/\u$1:/;
807   
808    if ((defined $and_encode) && ($and_encode)) {
809        $filename_full_path = encode("utf8",$filename_full_path);
810    }
811    }
812
813    return $filename_full_path;
814}
815
816
817sub downgrade_if_dos_filename
818{
819    my ($filename_full_path) = @_;
820
821    if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
822    require Win32;
823
824    # Ensure the given long Windows filename is in a form that can
825    # be opened by Perl => convert it to a short DOS-like filename
826
827    my $short_filename = Win32::GetShortPathName($filename_full_path);
828    if (defined $short_filename) {
829        $filename_full_path = $short_filename;
830    }
831    # Make sure initial drive letter is lower-case (to fit in
832    # with rest of Greenstone)
833    $filename_full_path =~ s/^(.):/\u$1:/;
834    }
835
836    return $filename_full_path;
837}
838
839
840sub filename_is_absolute
841{
842  warnings::warnif("deprecated", "util::filename_is_absolute() is deprecated, using FileUtils::isFilenameAbsolute() instead");
843  return &FileUtils::isFilenameAbsolute(@_);
844}
845
846
847## @method make_absolute()
848#
849#  Ensure the given file path is absolute in respect to the given base path.
850#
851#  @param  $base_dir A string denoting the base path the given dir must be
852#                    absolute to.
853#  @param  $dir The directory to be made absolute as a string. Note that the
854#               dir may already be absolute, in which case it will remain
855#               unchanged.
856#  @return The now absolute form of the directory as a string.
857#
858#  @author John Thompson, DL Consulting Ltd.
859#  @copy 2006 DL Consulting Ltd.
860#
861#used in buildcol.pl, doesn't work for all cases --kjdon
862sub make_absolute {
863   
864    my ($base_dir, $dir) = @_;
865###    print STDERR "dir = $dir\n";
866    $dir =~ s/[\\\/]+/\//g;
867    $dir = $base_dir . "/$dir" unless ($dir =~ m|^(\w:)?/|);
868    $dir =~ s|^/tmp_mnt||;
869    1 while($dir =~ s|/[^/]*/\.\./|/|g);
870    $dir =~ s|/[.][.]?/|/|g;
871    $dir =~ tr|/|/|s;
872###    print STDERR "dir = $dir\n";
873   
874    return $dir;
875}
876## make_absolute() ##
877
878sub get_dirsep {
879
880    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
881    return "\\";
882    } else {
883    return "\/";
884    }
885}
886
887sub get_os_dirsep {
888
889    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
890    return "\\\\";
891    } else {
892    return "\\\/";
893    }
894}
895
896sub get_re_dirsep {
897
898    return "\\\\|\\\/";
899}
900
901
902sub get_dirsep_tail {
903    my ($filename) = @_;
904   
905    # returns last part of directory or filename
906    # On unix e.g. a/b.d => b.d
907    #              a/b/c => c
908
909    my $dirsep = get_re_dirsep();
910    my @dirs = split (/$dirsep/, $filename);
911    my $tail = pop @dirs;
912
913    # - caused problems under windows
914    #my ($tail) = ($filename =~ m/^(?:.*?$dirsep)?(.*?)$/);
915
916    return $tail;
917}
918
919
920# if this is running on windows we want binaries to end in
921# .exe, otherwise they don't have to end in any extension
922sub get_os_exe {
923    return ".exe" if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin"));
924    return "";
925}
926
927
928# test to see whether this is a big or little endian machine
929sub is_little_endian
930{
931    # To determine the name of the operating system, the variable $^O is a cheap alternative to pulling it out of the Config module;
932    # 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
933    # Otherwise, it's little endian
934
935    #return 0 if $^O =~ /^darwin$/i;
936    #return 0 if $ENV{'GSDLOS'} =~ /^darwin$/i;
937   
938    # Going back to stating exactly whether the machine is little endian
939    # or big endian, without any special case for Macs. Since for rata it comes
940    # back with little endian and for shuttle with bigendian.
941    return (ord(substr(pack("s",1), 0, 1)) == 1);
942}
943
944
945# will return the collection name if successful, "" otherwise
946sub use_collection {
947    my ($collection, $collectdir) = @_;
948
949    if (!defined $collectdir || $collectdir eq "") {
950    $collectdir = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "collect");
951    }
952
953    if (!defined $ENV{'GREENSTONEHOME'}) { # for GS3, would have been defined in use_site_collection, to GSDL3HOME
954     $ENV{'GREENSTONEHOME'} = $ENV{'GSDLHOME'};
955    }
956
957    # get and check the collection
958    if (!defined($collection) || $collection eq "") {
959    if (defined $ENV{'GSDLCOLLECTION'}) {
960        $collection = $ENV{'GSDLCOLLECTION'};
961    } else {
962        print STDOUT "No collection specified\n";
963        return "";
964    }
965    }
966   
967    if ($collection eq "modelcol") {
968    print STDOUT "You can't use modelcol.\n";
969    return "";
970    }
971
972    # make sure the environment variables GSDLCOLLECTION and GSDLCOLLECTDIR
973    # are defined
974    $ENV{'GSDLCOLLECTION'} = $collection;
975    $ENV{'GSDLCOLLECTHOME'} = $collectdir;
976    $ENV{'GSDLCOLLECTDIR'} = &FileUtils::filenameConcatenate($collectdir, $collection);
977
978    # make sure this collection exists
979    if (!-e $ENV{'GSDLCOLLECTDIR'}) {
980    print STDOUT "Invalid collection ($collection).\n";
981    return "";
982    }
983
984    # everything is ready to go
985    return $collection;
986}
987
988sub get_current_collection_name {
989    return $ENV{'GSDLCOLLECTION'};
990}
991
992
993# will return the collection name if successful, "" otherwise. 
994# Like use_collection (above) but for greenstone 3 (taking account of site level)
995
996sub use_site_collection {
997    my ($site, $collection, $collectdir) = @_;
998
999    if (!defined $collectdir || $collectdir eq "") {
1000    die "GSDL3HOME not set.\n" unless defined $ENV{'GSDL3HOME'};
1001    $collectdir = &FileUtils::filenameConcatenate($ENV{'GSDL3HOME'}, "sites", $site, "collect");
1002    }
1003
1004    if (defined $ENV{'GSDL3HOME'}) {
1005    $ENV{'GREENSTONEHOME'} = $ENV{'GSDL3HOME'};     
1006    $ENV{'SITEHOME'} = &FileUtils::filenameConcatenate($ENV{'GREENSTONEHOME'}, "sites", $site);
1007    } elsif (defined $ENV{'GSDL3SRCHOME'}) {
1008    $ENV{'GREENSTONEHOME'} = &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'}, "web"); 
1009    $ENV{'SITEHOME'} = &FileUtils::filenameConcatenate($ENV{'GREENSTONEHOME'}, "sites", $site);
1010    } else {
1011    print STDERR "*** util::use_site_collection(). Warning: Neither GSDL3HOME nor GSDL3SRCHOME set.\n";
1012    }
1013
1014    # collectdir explicitly set by this point (using $site variable if required).
1015    # Can call "old" gsdl2 use_collection now.
1016
1017    return use_collection($collection,$collectdir);
1018}
1019
1020
1021
1022sub locate_config_file
1023{
1024    my ($file) = @_;
1025
1026    my $locations = locate_config_files($file);
1027
1028    return shift @$locations; # returns undef if 'locations' is empty
1029}
1030
1031
1032sub locate_config_files
1033{
1034    my ($file) = @_;
1035
1036    my @locations = ();
1037
1038    if (-e $file) {
1039    # Clearly specified (most likely full filename)
1040    # No need to hunt in 'etc' directories, return value unchanged
1041    push(@locations,$file);
1042    }
1043    else {
1044    # Check for collection specific one before looking in global GSDL 'etc'
1045    if (defined $ENV{'GSDLCOLLECTDIR'} && $ENV{'GSDLCOLLECTDIR'} ne "") {
1046        my $test_collect_etc_filename
1047        = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'},"etc", $file);
1048       
1049        if (-e $test_collect_etc_filename) {
1050        push(@locations,$test_collect_etc_filename);
1051        }
1052    }
1053    my $test_main_etc_filename
1054        = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"etc", $file);
1055    if (-e $test_main_etc_filename) {
1056        push(@locations,$test_main_etc_filename);
1057    }
1058    }
1059
1060    return \@locations;
1061}
1062
1063
1064sub hyperlink_text
1065{
1066    my ($text) = @_;
1067   
1068    $text =~ s/(http:\/\/[^\s]+)/<a href=\"$1\">$1<\/a>/mg;
1069    $text =~ s/(^|\s+)(www\.(\w|\.)+)/<a href=\"http:\/\/$2\">$2<\/a>/mg;
1070
1071    return $text;
1072}
1073
1074
1075# A method to check if a directory is empty (note that an empty directory still has non-zero size!!!)
1076# Code is from http://episteme.arstechnica.com/eve/forums/a/tpc/f/6330927813/m/436007700831
1077sub is_dir_empty {
1078  warnings::warnif("deprecated", "util::is_dir_empty() is deprecated, using FileUtils::isDirectoryEmpty() instead");
1079  return &FileUtils::isDirectoryEmpty(@_);
1080}
1081
1082# Returns the given filename converted using either URL encoding or base64
1083# encoding, as specified by $rename_method. If the given filename has no suffix
1084# (if it is just the tailname), then $no_suffix should be some defined value.
1085# rename_method can be url, none, base64
1086sub rename_file {
1087    my ($filename, $rename_method, $no_suffix)  = @_;
1088
1089    if(!$filename) { # undefined or empty string
1090    return $filename;
1091    }
1092
1093    if (!$rename_method) {
1094    print STDERR "WARNING: no file renaming method specified. Defaulting to using URL encoding...\n";
1095    # Debugging information
1096    # my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(1);
1097    # print STDERR "Called from method: $cfilename:$cline $cpackage->$csubr\n";
1098    $rename_method = "url";
1099    } elsif($rename_method eq "none") {
1100    return $filename; # would have already been renamed
1101    }
1102
1103    # No longer replace spaces with underscores, since underscores mess with incremental rebuild
1104    ### Replace spaces with underscore. Do this first else it can go wrong below when getting tailname
1105    ###$filename =~ s/ /_/g;
1106
1107    my ($tailname,$dirname,$suffix);
1108    if($no_suffix) { # given a tailname, no suffix
1109    ($tailname,$dirname) = File::Basename::fileparse($filename);
1110    }
1111    else {
1112    ($tailname,$dirname,$suffix) = File::Basename::fileparse($filename, "\\.(?:[^\\.]+?)\$");
1113    }
1114    if (!$suffix) {
1115    $suffix = "";
1116    }
1117    # This breaks GLI matching extracted metadata to files in Enrich panel, as
1118    # original is eg .JPG while gsdlsourcefilename ends up .jpg
1119    # Not sure why it was done in first place...
1120    #else {
1121    #$suffix = lc($suffix);
1122    #}
1123
1124    if ($rename_method eq "url") {
1125    $tailname = &unicode::url_encode($tailname);
1126    }
1127    elsif ($rename_method eq "base64") {
1128    $tailname = &unicode::base64_encode($tailname);
1129    $tailname =~ s/\s*//sg;      # for some reason it adds spaces not just at end but also in middle
1130    }
1131
1132    $filename = "$tailname$suffix";
1133    $filename = "$dirname$filename" if ($dirname ne "./" && $dirname ne ".\\");
1134
1135    return $filename;
1136}
1137
1138
1139# BACKWARDS COMPATIBILITY: Just in case there are old .ldb/.bdb files
1140sub rename_ldb_or_bdb_file {
1141    my ($filename_no_ext) = @_;
1142
1143    my $new_filename = "$filename_no_ext.gdb";
1144    return if (-f $new_filename); # if the file has the right extension, don't need to do anything
1145    # try ldb
1146    my $old_filename = "$filename_no_ext.ldb";
1147   
1148    if (-f $old_filename) {
1149    print STDERR "Renaming $old_filename to $new_filename\n";
1150    rename ($old_filename, $new_filename)
1151        || print STDERR "Rename failed: $!\n";
1152    return;
1153    }
1154    # try bdb
1155    $old_filename = "$filename_no_ext.bdb";
1156    if (-f $old_filename) {
1157    print STDERR "Renaming $old_filename to $new_filename\n";   
1158    rename ($old_filename, $new_filename)
1159        || print STDERR "Rename failed: $!\n";
1160    return;
1161    }
1162}
1163
1164sub os_dir() {
1165   
1166    my $gsdlarch = "";
1167    if(defined $ENV{'GSDLARCH'}) {
1168    $gsdlarch = $ENV{'GSDLARCH'};
1169    }
1170    return $ENV{'GSDLOS'}.$gsdlarch;
1171}
1172
1173# returns 1 if this (GS server) is a GS3 installation, returns 0 if it's GS2.
1174sub is_gs3() {
1175    if($ENV{'GSDL3SRCHOME'}) {
1176    return 1;
1177    } else {
1178    return 0;
1179    }
1180}
1181
1182# Returns the greenstone URL prefix extracted from the appropriate GS2/GS3 config file.
1183# By default, /greenstone3 for GS3 or /greenstone for GS2.
1184sub get_greenstone_url_prefix() {
1185    # if already set on a previous occasion, just return that
1186    # (Don't want to keep repeating this: cost of re-opening and scanning files.)
1187    return $ENV{'GREENSTONE_URL_PREFIX'} if($ENV{'GREENSTONE_URL_PREFIX'});
1188
1189    my ($configfile, $urlprefix, $defaultUrlprefix);
1190    my @propertynames = ();
1191
1192    if($ENV{'GSDL3SRCHOME'}) {
1193    $defaultUrlprefix = "/greenstone3";
1194    $configfile = &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'}, "packages", "tomcat", "conf", "Catalina", "localhost", "greenstone3.xml");
1195    push(@propertynames, qw/path\s*\=/);
1196    } else {
1197    $defaultUrlprefix = "/greenstone";
1198    $configfile = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "cgi-bin", &os_dir(), "gsdlsite.cfg");
1199    push(@propertynames, (qw/\nhttpprefix/, qw/\ngwcgi/)); # inspect one property then the other
1200    }
1201
1202    $urlprefix = &extract_propvalue_from_file($configfile, \@propertynames);
1203
1204    if(!$urlprefix) { # no values found for URL prefix, use default values
1205    $urlprefix = $defaultUrlprefix;
1206    } else {
1207    #gwcgi can contain more than the wanted prefix, we split on / to get the first "directory" level
1208    $urlprefix =~ s/^\///; # remove the starting slash
1209    my @dirs = split(/(\\|\/)/, $urlprefix);
1210    $urlprefix = shift(@dirs);
1211
1212    if($urlprefix !~ m/^\//) { # in all cases: ensure the required forward slash is at the front
1213        $urlprefix = "/$urlprefix";
1214    }
1215    }
1216
1217    # set for the future
1218    $ENV{'GREENSTONE_URL_PREFIX'} = $urlprefix;
1219#    print STDERR "*** in get_greenstone_url_prefix(): $urlprefix\n\n";
1220    return $urlprefix;
1221}
1222
1223
1224
1225#
1226# The following comes from activate.pl
1227#
1228# Designed to work with a server included with GS.
1229#  - For GS3, we ask ant for the library URL.
1230#  - For GS2, we derive the URL from the llssite.cfg file.
1231
1232sub get_full_greenstone_url_prefix
1233{   
1234    my ($gs_mode, $lib_name) = @_;
1235   
1236    # if already set on a previous occasion, just return that
1237    # (Don't want to keep repeating this: cost of re-opening and scanning files.)
1238    return $ENV{'FULL_GREENSTONE_URL_PREFIX'} if($ENV{'FULL_GREENSTONE_URL_PREFIX'});
1239
1240    # set gs_mode if it was not passed in (servercontrol.pm would pass it in, any other callers won't)
1241    $gs_mode = ($ENV{'GSDL3SRCHOME'}) ? "gs3" : "gs2" unless defined $gs_mode;
1242   
1243    my $url = undef;   
1244   
1245    if($gs_mode eq "gs2") {     
1246    my $llssite_cfg = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "llssite.cfg");
1247   
1248    if(-f $llssite_cfg) {
1249        # check llssite.cfg for line with url property
1250        # for server.exe also need to use portnumber and enterlib properties           
1251       
1252        # Read in the entire contents of the file in one hit
1253        if (!open (FIN, $llssite_cfg)) {
1254        print STDERR "util::get_full_greenstone_url_prefix() failed to open $llssite_cfg ($!)\n";
1255        return undef;
1256        }
1257       
1258        my $contents;
1259        sysread(FIN, $contents, -s FIN);           
1260        close(FIN);
1261       
1262        my @lines = split(/[\n\r]+/, $contents); # split on carriage-returns and/or linefeeds
1263        my $enterlib = "";
1264        my $portnumber = "8282"; # will remain empty (implicit port 80) unless it's specifically been assigned
1265       
1266        foreach my $line (@lines) {             
1267        if($line =~ m/^url=(.*)$/) {
1268            $url = $1;                 
1269        } elsif($line =~ m/^enterlib=(.*)$/) {
1270            $enterlib = $1;                 
1271        } elsif($line =~ m/^portnumber=(.*)$/) {
1272            $portnumber = $1;                   
1273        }   
1274        }
1275       
1276        if(!$url) {
1277        return undef;
1278        }
1279        elsif($url eq "URL_pending") { # library is not running
1280        # do not process url=URL_pending in the file, since for server.exe
1281        # this just means the Enter Library button hasn't been pressed yet             
1282        $url = undef;
1283        }
1284        else {
1285        # In the case of server.exe, need to do extra work to get the proper URL
1286        # But first, need to know whether we're indeed dealing with server.exe:
1287       
1288        # compare the URL's domain to the full URL
1289        # E.g. for http://localhost:8383/greenstone3/cgi-bin, the domain is localhost:8383
1290        my $uri = URI->new( $url );
1291        my $host = $uri->host;
1292        #print STDERR "@@@@@ host: $host\n";
1293        if($url =~ m/https?:\/\/$host(\/)?$/) {
1294            #if($url !~ m/https?:\/\/$host:$portnumber(\/)?/ || $url =~ m/https?:\/\/$host(\/)?$/) {
1295            # (if the URL does not contain the portnumber, OR if the port is implicitly 80 and)                 
1296            # If the domain with http:// prefix is completely the same as the URL, assume server.exe
1297            # then the actual URL is the result of suffixing the port and enterlib properties in llssite.cfg
1298            $url = $url.":".$portnumber.$enterlib;         
1299        } # else, apache web server         
1300       
1301        }           
1302    }
1303    } elsif($gs_mode eq "gs3") {
1304    # Either check build.properties for tomcat.server, tomcat.port and app.name (and default servlet name).
1305    # app.name is stored in app.path by build.xml. Need to move app.name in build.properties from build.xml
1306   
1307    # Or, run the new target get-default-servlet-url
1308    # the output can look like:
1309    #
1310    # Buildfile: build.xml
1311    #   [echo] os.name: Windows Vista
1312    #
1313    # get-default-servlet-url:
1314    #   [echo] http://localhost:8383/greenstone3/library
1315    # BUILD SUCCESSFUL
1316    # Total time: 0 seconds
1317   
1318    #my $output = qx/ant get-default-servlet-url/; # backtick operator, to get STDOUT (else 2>&1)
1319    # - see http://stackoverflow.com/questions/799968/whats-the-difference-between-perls-backticks-system-and-exec
1320   
1321    # The get-default-servlet-url ant target can be run from anywhere by specifying the
1322    # location of GS3's ant build.xml buildfile. Activate.pl can be run from anywhere for GS3
1323    # GSDL3SRCHOME will be set for GS3 by gs3-setup.sh, a step that would have been necessary
1324    # to run the activate.pl script in the first place
1325   
1326    my $full_build_xml = &FileUtils::javaFilenameConcatenate($ENV{'GSDL3SRCHOME'},"build.xml");
1327
1328    my $perl_command = "ant -buildfile \"$full_build_xml\" get-default-servlet-url";
1329   
1330    if (open(PIN, "$perl_command |")) {
1331        while (defined (my $perl_output_line = <PIN>)) {
1332
1333        if($perl_output_line =~ m@(https?):\/\/(\S*)@) { # grab all the non-whitespace chars
1334            $url="$1://".$2; # preserve the http protocol #$url="http://".$1;
1335        }
1336        }
1337        close(PIN);
1338       
1339        if (defined $lib_name) { # url won't be undef now
1340            # replace the servlet_name portion of the url found, with the given library_name
1341            $url =~ s@/[^/]*$@/$lib_name@;
1342        }
1343    } else {
1344        print STDERR "util::get_full_greenstone_url_prefix() failed to run $perl_command to work out library URL for $gs_mode\n";
1345    }
1346    }
1347   
1348    # either the url is still undef or it is now set
1349    #print STDERR "\n@@@@@ final URL:|$url|\n" if $url;     
1350    #print STDERR "\n@@@@@ URL still undef\n" if !$url;
1351
1352    $ENV{'FULL_GREENSTONE_URL_PREFIX'} = $url;
1353
1354    return $url;
1355}
1356
1357
1358# Given a config file (xml or java properties file) and a list/array of regular expressions
1359# that represent property names to match on, this function will return the value for the 1st
1360# matching property name. If the return value is undefined, no matching property was found.
1361sub extract_propvalue_from_file() {
1362    my ($configfile, $propertynames) = @_;
1363
1364    my $value;
1365    unless(open(FIN, "<$configfile")) {
1366    print STDERR "extract_propvalue_from_file(): Unable to open $configfile. $!\n";
1367    return $value; # not initialised
1368    }
1369
1370    # Read the entire file at once, as one single line, then close it
1371    my $filecontents;
1372    {
1373    local $/ = undef;       
1374    $filecontents = <FIN>;
1375    }
1376    close(FIN);
1377
1378    foreach my $regex (@$propertynames) {
1379        ($value) = $filecontents=~ m/$regex\s*(\S*)/s; # read value of the property given by regex up to the 1st space
1380    if($value) {
1381            $value =~ s/^\"//;     # remove any startquotes
1382        $value =~ s/\".*$//;   # remove the 1st endquotes (if any) followed by any xml
1383        last;              # found value for a matching property, break from loop
1384    }
1385    }
1386
1387    return $value;
1388}
1389
1390# Subroutine that sources setup.bash, given GSDLHOME and GSDLOS and
1391# given that perllib is in @INC in order to invoke this subroutine.
1392# Call as follows -- after setting up INC to include perllib and
1393# after setting up GSDLHOME and GSDLOS:
1394#
1395# require util;
1396# &util::setup_greenstone_env($ENV{'GSDLHOME'}, $ENV{'GSDLOS'});
1397#
1398sub setup_greenstone_env() {
1399    my ($GSDLHOME, $GSDLOS) = @_;
1400
1401    #my %env_map = ();
1402    # Get the localised ENV settings of running a localised source setup.bash
1403    # and put it into the ENV here. Need to clear GSDLHOME before running setup
1404    #my $perl_command = "(cd $GSDLHOME; export GSDLHOME=; . ./setup.bash > /dev/null; env)";
1405    my $perl_command = "(cd $GSDLHOME; /bin/bash -c \"export GSDLHOME=; source setup.bash > /dev/null; env\")";     
1406    if (($GSDLOS =~ m/windows/i) && ($^O ne "cygwin"))  {
1407        #$perl_command = "cmd /C \"cd $GSDLHOME&& set GSDLHOME=&& setup.bat > nul&& set\"";
1408        $perl_command = "(cd $GSDLHOME&& set GSDLHOME=&& setup.bat > nul&& set)";
1409    }
1410    if (!open(PIN, "$perl_command |")) {
1411        print STDERR ("Unable to execute command: $perl_command. $!\n");
1412    }
1413
1414    my $lastkey;
1415    while (defined (my $perl_output_line = <PIN>)) {
1416        my($key,$value) = ($perl_output_line =~ m/^([^=]*)[=](.*)$/);
1417        if(defined $key) {
1418            #$env_map{$key}=$value;     
1419            $ENV{$key}=$value;
1420            $lastkey = $key;
1421        } elsif($lastkey && $perl_output_line !~ m/^\s*$/) {
1422            # there was no equals sign in $perl_output_line, so this
1423            # $perl_output_line may be a spillover from the previous
1424            $ENV{$lastkey} = $ENV{$lastkey}."\n".$perl_output_line;
1425        }
1426    }
1427    close (PIN);
1428
1429    # If any keys in $ENV don't occur in Greenstone's localised env
1430    # (stored in $env_map), delete those entries from $ENV
1431    #foreach $key (keys %ENV) {
1432    #   if(!defined $env_map{$key}) {
1433    #       print STDOUT "**** DELETING ENV KEY: $key\tVALUE: $ENV{$key}\n";
1434    #       delete $ENV{$key}; # del $ENV(key, value) pair
1435    #   }
1436    #}
1437    #undef %env_map;
1438}
1439
1440sub get_perl_exec() {   
1441    my $perl_exec = $^X; # may return just "perl"
1442   
1443    if($ENV{'PERLPATH'}) {
1444        # OR: # $perl_exec = &FileUtils::filenameConcatenate($ENV{'PERLPATH'},"perl");
1445        if (($ENV{'GSDLOS'} =~ m/windows/) && ($^O ne "cygwin")) {
1446            $perl_exec = "$ENV{'PERLPATH'}\\Perl.exe";
1447        } else {
1448            $perl_exec = "$ENV{'PERLPATH'}/perl";
1449        }
1450    } else { # no PERLPATH, use Config{perlpath} else $^X: special variables
1451        # containing the full path to the current perl executable we're using
1452        $perl_exec = $Config{perlpath}; # configured path for perl
1453        if (!-e $perl_exec) { # may not point to location on this machine
1454            $perl_exec = $^X; # may return just "perl"
1455            if($perl_exec =~ m/^perl/i) { # warn if just perl or Perl.exe
1456                print STDERR "**** WARNING: Perl exec found contains no path: $perl_exec";             
1457            }
1458        }
1459    }
1460   
1461    return $perl_exec;
1462}
1463
1464# returns the path to the java command in the JRE included with GS (if any),
1465# quoted to safeguard any spaces in this path, otherwise a simple java
1466# command is returned which assumes and will try for a system java.
1467sub get_java_command {
1468    my $java = "java";
1469    if(defined $ENV{'GSDLHOME'}) { # should be, as this script would be launched from the cmd line
1470                               # after running setup.bat or from GLI which also runs setup.bat
1471    my $java_bin = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"packages","jre","bin");
1472    if(-d $java_bin) {
1473        $java = &FileUtils::filenameConcatenate($java_bin,"java");
1474        $java = "\"".$java."\""; # quoted to preserve spaces in path
1475    }
1476    }
1477    return $java;
1478}
1479
1480
1481# Given the qualified collection name (colgroup/collection),
1482# returns the collection and colgroup parts
1483sub get_collection_parts {
1484    # http://perldoc.perl.org/File/Basename.html
1485    # my($filename, $directories, $suffix) = fileparse($path);
1486    # "$directories contains everything up to and including the last directory separator in the $path
1487    # including the volume (if applicable). The remainder of the $path is the $filename."
1488    #my ($collection, $colgroup) = &File::Basename::fileparse($qualified_collection);   
1489
1490    my $qualified_collection = shift(@_);
1491
1492    # Since activate.pl can be launched from the command-line, including by a user,
1493    # best not to assume colgroup uses URL-style slashes as would be the case with GLI
1494    # Also allow for the accidental inclusion of multiple slashes
1495    my ($colgroup, $collection) = split(/[\/\\]+/, $qualified_collection); #split('/', $qualified_collection);
1496   
1497    if(!defined $collection) {
1498        $collection = $colgroup;
1499        $colgroup = "";
1500    }
1501    return ($collection, $colgroup);
1502}
1503
1504# work out the "collectdir/collection" location
1505sub resolve_collection_dir {
1506    my ($collect_dir, $qualified_collection, $site) = @_; #, $gs_mode
1507   
1508    if (defined $ENV{'GSDLCOLLECTDIR'}) { # a predefined collection dir exists
1509        return $ENV{'GSDLCOLLECTDIR'};
1510    }
1511
1512    my ($colgroup, $collection) = &util::get_collection_parts($qualified_collection);   
1513   
1514    if (!defined $collect_dir || !$collect_dir) { # if undefined or empty string
1515        $collect_dir = &util::get_working_collect_dir($site);
1516    }
1517
1518    return &FileUtils::filenameConcatenate($collect_dir,$colgroup,$collection);
1519}
1520
1521# work out the full path to "collect" of this greenstone 2/3 installation
1522sub get_working_collect_dir {
1523    my ($site) = @_;   
1524
1525    if (defined $ENV{'GSDLCOLLECTHOME'}) { # a predefined collect dir exists
1526    return $ENV{'GSDLCOLLECTHOME'};
1527    }
1528
1529    if (defined $site && $site) { # site non-empty, so get default collect dir for GS3
1530   
1531    if (defined $ENV{'GSDL3HOME'}) {
1532        return &FileUtils::filenameConcatenate($ENV{'GSDL3HOME'},"sites",$site,"collect"); # web folder
1533    }
1534    elsif (defined $ENV{'GSDL3SRCHOME'}) {
1535        return &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'},"web","sites",$site,"collect");
1536    }
1537    }
1538
1539    elsif (defined $ENV{'SITEHOME'}) {
1540    return &FileUtils::filenameConcatenate($ENV{'SITEHOME'},"collect");
1541    }
1542   
1543    else { # get default collect dir for GS2
1544    return &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"collect");
1545    }
1546}
1547
1548sub is_abs_path_any_os {
1549    my ($path) = @_;
1550
1551    # We can have filenames in our DBs that were produced on other OS, so this method exists
1552    # to help identify absolute paths in such cases.
1553
1554    return 1 if($path =~ m@^/@); # full paths begin with forward slash on linux/mac
1555    return 1 if($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
1556
1557    return 0;
1558}
1559
1560
1561# This subroutine is for improving portability of Greenstone collections from one OS to another,
1562# to be used to convert absolute paths going into db files into paths with placeholders instead.
1563# This sub works with util::get_common_gs_paths and takes a path to a greenstone file and, if it's
1564# an absolute path, then it will replace the longest matching greenstone-path prefix of the given
1565# path with a placeholder to match.
1566# The Greenstone-path prefixes that can be matched are the following common Greenstone paths:
1567# the path to the current (specific) collection, the path to the general GS collect directory,
1568# the path to the site directory if GS3, else the path to the GSDLHOME/GSDL3HOME folder.
1569# The longest matching prefix will be replaced with the equivalent placeholder:
1570# @THISCOLLECTPATH@, else @COLLECTHOME@, else @SITEHOME@, else @GSDLHOME@.
1571sub abspath_to_placeholders {
1572    my $path = shift(@_); # path to convert from absolute to one with placeholders
1573    my $opt_long_or_short_winfilenames = shift(@_) || "short"; # whether we want to force use of long file names even on windows, default uses short
1574
1575    return $path unless is_abs_path_any_os($path); # path is relative
1576
1577    if ($opt_long_or_short_winfilenames eq "long") {
1578    $path = &util::upgrade_if_dos_filename($path); # will only do something on windows
1579    }
1580   
1581    # now we know we're dealing with absolute paths and have to replace gs prefixes with placeholders
1582    my @gs_paths = ($ENV{'GSDLCOLLECTDIR'}, $ENV{'GSDLCOLLECTHOME'}, $ENV{'SITEHOME'}, $ENV{'GREENSTONEHOME'}); # list in this order: from longest to shortest path
1583
1584    my %placeholder_map = ($ENV{'GREENSTONEHOME'} => '@GSDLHOME@', # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
1585               $ENV{'GSDLCOLLECTHOME'} => '@COLLECTHOME@',
1586               $ENV{'GSDLCOLLECTDIR'} => '@THISCOLLECTPATH@'
1587    );
1588    $placeholder_map{$ENV{'SITEHOME'}} = '@SITEHOME@' if defined $ENV{'SITEHOME'};
1589
1590    $path = &util::_abspath_to_placeholders($path, \@gs_paths, \%placeholder_map);
1591
1592    if ($ENV{'GSDLOS'} =~ /^windows$/i && $opt_long_or_short_winfilenames eq "short") {
1593    # for windows need to look for matches on short file names too
1594    # matched paths are again to be replaced with the usual placeholders
1595
1596    my $gsdlcollectdir = &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTDIR'});
1597    my $gsdlcollecthome = &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTHOME'});
1598    my $sitehome = (defined $ENV{'SITEHOME'}) ? &util::downgrade_if_dos_filename($ENV{'SITEHOME'}) : undef;
1599    my $greenstonehome =  &util::downgrade_if_dos_filename($ENV{'GREENSTONEHOME'});
1600
1601    @gs_paths = ($gsdlcollectdir, $gsdlcollecthome, $sitehome, $greenstonehome); # order matters
1602
1603    %placeholder_map = ($greenstonehome => '@GSDLHOME@', # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
1604                $gsdlcollecthome => '@COLLECTHOME@',
1605                $gsdlcollectdir => '@THISCOLLECTPATH@'
1606        );
1607    $placeholder_map{$sitehome} = '@SITEHOME@' if defined $sitehome;
1608
1609    $path = &util::_abspath_to_placeholders($path, \@gs_paths, \%placeholder_map);
1610    }
1611
1612    return $path;
1613}
1614
1615sub _abspath_to_placeholders {
1616    my ($path, $gs_paths_ref, $placeholder_map_ref) = @_;
1617
1618    # The sequence of elements in @gs_paths matters   
1619    # Need to loop starting from the *longest* matching path (the path to the specific collection)
1620    # to the shortest matching path (the path to gsdlhome/gsdl3home folder):
1621
1622    foreach my $gs_path (@$gs_paths_ref) {
1623    next if(!defined $gs_path); # site undefined for GS2
1624
1625    my $re_path =  &util::filename_to_regex($gs_path); # escape for regex
1626
1627    if($path =~ m/^$re_path/i) { # case sensitive or not for OS?
1628
1629        my $placeholder = $placeholder_map_ref->{$gs_path}; # get the placeholder to replace the matched path with
1630
1631        $path =~ s/^$re_path/$placeholder/; #case sensitive or not?
1632        #$path =~ s/^[\\\/]//; # remove gs_path's trailing separator left behind at the start of the path
1633        # lowercase file extension, This is needed when shortfilenames are used, as case affects alphetical ordering, which affects diffcol     
1634        $path =~ s/\.([A-Z]+)$/".".lc($1)/e;
1635        last; # done
1636    }
1637    }
1638   
1639    return $path;
1640}
1641
1642# Function that does the reverse of the util::abspath_to_placeholders subroutine
1643# Once again, call this with the values returned from util::get_common_gs_paths
1644sub placeholders_to_abspath {
1645    my $path = shift(@_); # path that can contain placeholders to convert to resolved absolute path
1646    my $opt_long_or_short_winfilenames = shift(@_) || "short"; # whether we want to force use of long file names even on windows, default uses short
1647
1648    return $path if($path !~ m/@/); # path contains no placeholders
1649   
1650    # replace placeholders with gs prefixes
1651    my @placeholders = ('@THISCOLLECTPATH@', '@COLLECTHOME@', '@SITEHOME@', '@GSDLHOME@'); # order of paths not crucial in this case,
1652                       # but listed here from longest to shortest once placeholders are have been resolved
1653
1654    # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
1655    my %placeholder_to_gspath_map;
1656    if ($ENV{'GSDLOS'} =~ /^windows$/i && $opt_long_or_short_winfilenames eq "short") {
1657    # always replace placeholders with short file names of the absolute paths on windows?
1658    %placeholder_to_gspath_map = ('@GSDLHOME@' => &util::downgrade_if_dos_filename($ENV{'GREENSTONEHOME'}),
1659                     '@COLLECTHOME@' => &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTHOME'}),
1660                     '@THISCOLLECTPATH@' => &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTDIR'})
1661    );
1662    $placeholder_to_gspath_map{'@SITEHOME@'} =  &util::downgrade_if_dos_filename($ENV{'SITEHOME'}) if defined $ENV{'SITEHOME'};
1663    } else {
1664    %placeholder_to_gspath_map = ('@GSDLHOME@' => $ENV{'GREENSTONEHOME'},
1665                      '@SITEHOME@' => $ENV{'SITEHOME'}, # can be undef
1666                      '@COLLECTHOME@' => $ENV{'GSDLCOLLECTHOME'},
1667                      '@THISCOLLECTPATH@' => $ENV{'GSDLCOLLECTDIR'}
1668        ); # $placeholder_to_gspath_map{'@SITEHOME@'} = $ENV{'SITEHOME'} if defined $ENV{'SITEHOME'};
1669    }
1670
1671    foreach my $placeholder (@placeholders) {
1672    my $gs_path = $placeholder_to_gspath_map{$placeholder};
1673
1674    next if(!defined $gs_path); # sitehome for GS2 is undefined
1675
1676    if($path =~ m/^$placeholder/) {
1677        $path =~ s/^$placeholder/$gs_path/;
1678        last; # done
1679    }
1680    }
1681   
1682    return $path;
1683}
1684
1685# Used by pdfpstoimg.pl and PDFBoxConverter to create a .item file from
1686# a directory containing sequentially numbered images.
1687sub create_itemfile
1688{
1689    my ($output_dir, $convert_basename, $convert_to) = @_;
1690    my $page_num = "";
1691
1692    opendir(DIR, $output_dir) || die "can't opendir $output_dir: $!";
1693    my @dir_files = grep {-f "$output_dir/$_"} readdir(DIR);
1694    closedir DIR;
1695
1696    # Sort files in the directory by page_num
1697    sub page_number {
1698    my ($dir) = @_;
1699    my ($pagenum) =($dir =~ m/^.*?[-\.]?(\d+)(\.(jpg|gif|png))?$/i);
1700#   my ($pagenum) =($dir =~ m/(\d+)(\.(jpg|gif|png))?$/i); # this works but is not as safe/strict about input filepatterns as the above
1701
1702    $pagenum = 1 unless defined $pagenum;
1703    return $pagenum;
1704    }
1705
1706    # sort the files in the directory in the order of page_num rather than lexically.
1707    @dir_files = sort { page_number($a) <=> page_number($b) } @dir_files;
1708
1709    # work out if the numbering of the now sorted image files starts at 0 or not
1710    # by checking the number of the first _image_ file (skipping item files)
1711    my $starts_at_0 = 0;
1712    my $firstfile = ($dir_files[0] !~ /\.item$/i) ? $dir_files[0] : $dir_files[1];
1713    if(page_number($firstfile) == 0) { # 00 will evaluate to 0 too in this condition
1714    $starts_at_0 = 1;
1715    }
1716
1717    my $item_file = &FileUtils::filenameConcatenate($output_dir, $convert_basename.".item");
1718    my $item_fh;
1719    &FileUtils::openFileHandle($item_file, 'w', \$item_fh);
1720    print $item_fh "<PagedDocument>\n";
1721
1722    foreach my $file (@dir_files){
1723    if ($file !~ /\.item/i){
1724        $page_num = page_number($file);
1725        $page_num++ if $starts_at_0; # image numbers start at 0, so add 1
1726        print $item_fh "   <Page pagenum=\"$page_num\" imgfile=\"$file\" txtfile=\"\"/>\n";
1727    }
1728    }
1729
1730    print $item_fh "</PagedDocument>\n";
1731    &FileUtils::closeFileHandle($item_file, \$item_fh);
1732    return $item_file;
1733}
1734
1735# Sets the gnomelib_env. Based on the logic in wvware.pl which can perhaps be replaced with a call to this function in future
1736sub set_gnomelib_env
1737{
1738    ## SET THE ENVIRONMENT AS DONE IN SETUP.BASH/BAT OF GNOME-LIB
1739    # Though this is only needed for darwin Lion at this point (and android, though that is untested)
1740
1741    my $libext = "so";
1742    if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
1743    return;
1744    } elsif ($ENV{'GSDLOS'} =~ m/^darwin$/i) {
1745    $libext = "dylib";
1746    }
1747
1748    if (!defined $ENV{'GEXTGNOME'}) {
1749        ##print STDERR "@@@ Setting GEXTGNOME env\n";
1750
1751    my $gnome_dir = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"ext","gnome-lib-minimal");
1752
1753    if(! -d $gnome_dir) {
1754        $gnome_dir = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"ext","gnome-lib");
1755
1756        if(! -d $gnome_dir) {
1757        $gnome_dir = "";
1758        }
1759    }
1760   
1761    # now set other the related env vars,
1762    # IF we've found the gnome-lib dir installed in the ext folder 
1763
1764    if ($gnome_dir ne "" && -f &FileUtils::filenameConcatenate($gnome_dir, $ENV{'GSDLOS'}, "lib", "libiconv.$libext")) {
1765        $ENV{'GEXTGNOME'} = $gnome_dir;
1766        $ENV{'GEXTGNOME_INSTALLED'}=&FileUtils::filenameConcatenate($ENV{'GEXTGNOME'}, $ENV{'GSDLOS'});
1767       
1768        my $gnomelib_bin = &FileUtils::filenameConcatenate($ENV{'GEXTGNOME_INSTALLED'}, "bin");
1769        if(-d $gnomelib_bin) { # no bin subfolder in GS binary's cutdown gnome-lib-minimal folder
1770        &util::envvar_prepend("PATH", $gnomelib_bin);
1771        }
1772
1773        # util's prepend will create LD/DYLD_LIB_PATH if it doesn't exist yet
1774        my $gextlib = &FileUtils::filenameConcatenate($ENV{'GEXTGNOME_INSTALLED'}, "lib");
1775
1776        if($ENV{'GSDLOS'} eq "linux") {
1777        &util::envvar_prepend("LD_LIBRARY_PATH", $gextlib);
1778        }
1779        elsif ($ENV{'GSDLOS'} eq "darwin") {
1780        #&util::envvar_prepend("DYLD_LIBRARY_PATH", $gextlib);
1781        &util::envvar_prepend("DYLD_FALLBACK_LIBRARY_PATH", $gextlib);
1782        }
1783    }
1784   
1785    # Above largely mimics the setup.bash of the gnome-lib-minimal.
1786    # Not doing the devel-srcpack that gnome-lib-minimal's setup.bash used to set
1787    # Not exporting GSDLEXTS variable either
1788    }
1789
1790#    print STDERR "@@@@@ GEXTGNOME: ".$ENV{'GEXTGNOME'}."\n\tINSTALL".$ENV{'GEXTGNOME_INSTALLED'}."\n";
1791#    print STDERR "\tPATH".$ENV{'PATH'}."\n";
1792#    print STDERR "\tLD_LIB_PATH".$ENV{'LD_LIBRARY_PATH'}."\n" if $ENV{'LD_LIBRARY_PATH};
1793#    print STDERR "\tDYLD_FALLBACK_LIB_PATH".$ENV{'DYLD_FALLBACK_LIBRARY_PATH'}."\n" if $ENV{'DYLD_FALLBACK_LIBRARY_PATH};
1794
1795    # if no GEXTGNOME, maybe users didn't need gnome-lib to run gnomelib/libiconv dependent binaries like hashfile, suffix, wget
1796    # (wvware is launched in a gnomelib env from its own script, but could possibly go through this script in future)
1797}
1798
1799
1800
1801## @function augmentINC()
1802#
1803#  Prepend a path (if it exists) onto INC but only if it isn't already in INC
1804#  @param $new_path The path to add
1805#  @author jmt12
1806#
1807sub augmentINC
1808{
1809  my ($new_path) = @_;
1810  my $did_add_path = 0;
1811  # might need to be replaced with FileUtils::directoryExists() call eventually
1812  if (-d $new_path)
1813  {
1814    my $did_find_path = 0;
1815    foreach my $existing_path (@INC)
1816    {
1817      if ($existing_path eq $new_path)
1818      {
1819        $did_find_path = 1;
1820        last;
1821      }
1822    }
1823    if (!$did_find_path)
1824    {
1825      unshift(@INC, $new_path);
1826      $did_add_path = 1;
1827    }
1828  }
1829  return $did_add_path;
1830}
1831## augmentINC()
1832
1833
18341;
Note: See TracBrowser for help on using the browser.