source: main/trunk/greenstone2/perllib/util.pm@ 31187

Last change on this file since 31187 was 31187, checked in by ak19, 7 years ago

Useful changes not specifically related to upcoming oaiinfo db related commit.
  • Property svn:keywords set to Author Date Id Revision
File size: 61.9 KB
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
841 # lower case the entire thing, eg for cover.jpg when its actually cover.JPG
842 my $lower_filename = lc($filename);
843 $block_hash->{'file_blocks'}->{$lower_filename} = 1;
844# my $lower_drive = $filename;
845# $lower_drive =~ s/^([A-Z]):/\l$1:/i;
846
847# my $upper_drive = $filename;
848# $upper_drive =~ s/^([A-Z]):/\u$1:/i;
849#
850# $block_hash->{'file_blocks'}->{$lower_drive} = 1;
851# $block_hash->{'file_blocks'}->{$upper_drive} = 1;
852 }
853 else {
854 $block_hash->{'file_blocks'}->{$filename} = 1;
855 }
856}
857
858
859sub filename_is_absolute
860{
861 warnings::warnif("deprecated", "util::filename_is_absolute() is deprecated, using FileUtils::isFilenameAbsolute() instead");
862 return &FileUtils::isFilenameAbsolute(@_);
863}
864
865
866## @method make_absolute()
867#
868# Ensure the given file path is absolute in respect to the given base path.
869#
870# @param $base_dir A string denoting the base path the given dir must be
871# absolute to.
872# @param $dir The directory to be made absolute as a string. Note that the
873# dir may already be absolute, in which case it will remain
874# unchanged.
875# @return The now absolute form of the directory as a string.
876#
877# @author John Thompson, DL Consulting Ltd.
878# @copy 2006 DL Consulting Ltd.
879#
880#used in buildcol.pl, doesn't work for all cases --kjdon
881sub make_absolute {
882
883 my ($base_dir, $dir) = @_;
884### print STDERR "dir = $dir\n";
885 $dir =~ s/[\\\/]+/\//g;
886 $dir = $base_dir . "/$dir" unless ($dir =~ m|^(\w:)?/|);
887 $dir =~ s|^/tmp_mnt||;
888 1 while($dir =~ s|/[^/]*/\.\./|/|g);
889 $dir =~ s|/[.][.]?/|/|g;
890 $dir =~ tr|/|/|s;
891### print STDERR "dir = $dir\n";
892
893 return $dir;
894}
895## make_absolute() ##
896
897sub get_dirsep {
898
899 if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
900 return "\\";
901 } else {
902 return "\/";
903 }
904}
905
906sub get_os_dirsep {
907
908 if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
909 return "\\\\";
910 } else {
911 return "\\\/";
912 }
913}
914
915sub get_re_dirsep {
916
917 return "\\\\|\\\/";
918}
919
920
921sub get_dirsep_tail {
922 my ($filename) = @_;
923
924 # returns last part of directory or filename
925 # On unix e.g. a/b.d => b.d
926 # a/b/c => c
927
928 my $dirsep = get_re_dirsep();
929 my @dirs = split (/$dirsep/, $filename);
930 my $tail = pop @dirs;
931
932 # - caused problems under windows
933 #my ($tail) = ($filename =~ m/^(?:.*?$dirsep)?(.*?)$/);
934
935 return $tail;
936}
937
938
939# if this is running on windows we want binaries to end in
940# .exe, otherwise they don't have to end in any extension
941sub get_os_exe {
942 return ".exe" if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin"));
943 return "";
944}
945
946
947# test to see whether this is a big or little endian machine
948sub is_little_endian
949{
950 # To determine the name of the operating system, the variable $^O is a cheap alternative to pulling it out of the Config module;
951 # 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
952 # Otherwise, it's little endian
953
954 #return 0 if $^O =~ /^darwin$/i;
955 #return 0 if $ENV{'GSDLOS'} =~ /^darwin$/i;
956
957 # Going back to stating exactly whether the machine is little endian
958 # or big endian, without any special case for Macs. Since for rata it comes
959 # back with little endian and for shuttle with bigendian.
960 return (ord(substr(pack("s",1), 0, 1)) == 1);
961}
962
963
964# will return the collection name if successful, "" otherwise
965sub use_collection {
966 my ($collection, $collectdir) = @_;
967
968 if (!defined $collectdir || $collectdir eq "") {
969 $collectdir = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "collect");
970 }
971
972 if (!defined $ENV{'GREENSTONEHOME'}) { # for GS3, would have been defined in use_site_collection, to GSDL3HOME
973 $ENV{'GREENSTONEHOME'} = $ENV{'GSDLHOME'};
974 }
975
976 # get and check the collection
977 if (!defined($collection) || $collection eq "") {
978 if (defined $ENV{'GSDLCOLLECTION'}) {
979 $collection = $ENV{'GSDLCOLLECTION'};
980 } else {
981 print STDOUT "No collection specified\n";
982 return "";
983 }
984 }
985
986 if ($collection eq "modelcol") {
987 print STDOUT "You can't use modelcol.\n";
988 return "";
989 }
990
991 # make sure the environment variables GSDLCOLLECTION and GSDLCOLLECTDIR
992 # are defined
993 $ENV{'GSDLCOLLECTION'} = $collection;
994 $ENV{'GSDLCOLLECTHOME'} = $collectdir;
995 $ENV{'GSDLCOLLECTDIR'} = &FileUtils::filenameConcatenate($collectdir, $collection);
996
997 # make sure this collection exists
998 if (!-e $ENV{'GSDLCOLLECTDIR'}) {
999 print STDOUT "Invalid collection ($collection).\n";
1000 return "";
1001 }
1002
1003 # everything is ready to go
1004 return $collection;
1005}
1006
1007sub get_current_collection_name {
1008 return $ENV{'GSDLCOLLECTION'};
1009}
1010
1011
1012# will return the collection name if successful, "" otherwise.
1013# Like use_collection (above) but for greenstone 3 (taking account of site level)
1014
1015sub use_site_collection {
1016 my ($site, $collection, $collectdir) = @_;
1017
1018 if (!defined $collectdir || $collectdir eq "") {
1019 die "GSDL3HOME not set.\n" unless defined $ENV{'GSDL3HOME'};
1020 $collectdir = &FileUtils::filenameConcatenate($ENV{'GSDL3HOME'}, "sites", $site, "collect");
1021 }
1022
1023 if (defined $ENV{'GSDL3HOME'}) {
1024 $ENV{'GREENSTONEHOME'} = $ENV{'GSDL3HOME'};
1025 $ENV{'SITEHOME'} = &FileUtils::filenameConcatenate($ENV{'GREENSTONEHOME'}, "sites", $site);
1026 } elsif (defined $ENV{'GSDL3SRCHOME'}) {
1027 $ENV{'GREENSTONEHOME'} = &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'}, "web");
1028 $ENV{'SITEHOME'} = &FileUtils::filenameConcatenate($ENV{'GREENSTONEHOME'}, "sites", $site);
1029 } else {
1030 print STDERR "*** util::use_site_collection(). Warning: Neither GSDL3HOME nor GSDL3SRCHOME set.\n";
1031 }
1032
1033 # collectdir explicitly set by this point (using $site variable if required).
1034 # Can call "old" gsdl2 use_collection now.
1035
1036 return use_collection($collection,$collectdir);
1037}
1038
1039
1040
1041sub locate_config_file
1042{
1043 my ($file) = @_;
1044
1045 my $locations = locate_config_files($file);
1046
1047 return shift @$locations; # returns undef if 'locations' is empty
1048}
1049
1050
1051sub locate_config_files
1052{
1053 my ($file) = @_;
1054
1055 my @locations = ();
1056
1057 if (-e $file) {
1058 # Clearly specified (most likely full filename)
1059 # No need to hunt in 'etc' directories, return value unchanged
1060 push(@locations,$file);
1061 }
1062 else {
1063 # Check for collection specific one before looking in global GSDL 'etc'
1064 if (defined $ENV{'GSDLCOLLECTDIR'} && $ENV{'GSDLCOLLECTDIR'} ne "") {
1065 my $test_collect_etc_filename
1066 = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'},"etc", $file);
1067
1068 if (-e $test_collect_etc_filename) {
1069 push(@locations,$test_collect_etc_filename);
1070 }
1071 }
1072 my $test_main_etc_filename
1073 = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"etc", $file);
1074 if (-e $test_main_etc_filename) {
1075 push(@locations,$test_main_etc_filename);
1076 }
1077 }
1078
1079 return \@locations;
1080}
1081
1082
1083sub hyperlink_text
1084{
1085 my ($text) = @_;
1086
1087 $text =~ s/(http:\/\/[^\s]+)/<a href=\"$1\">$1<\/a>/mg;
1088 $text =~ s/(^|\s+)(www\.(\w|\.)+)/<a href=\"http:\/\/$2\">$2<\/a>/mg;
1089
1090 return $text;
1091}
1092
1093
1094# A method to check if a directory is empty (note that an empty directory still has non-zero size!!!)
1095# Code is from http://episteme.arstechnica.com/eve/forums/a/tpc/f/6330927813/m/436007700831
1096sub is_dir_empty {
1097 warnings::warnif("deprecated", "util::is_dir_empty() is deprecated, using FileUtils::isDirectoryEmpty() instead");
1098 return &FileUtils::isDirectoryEmpty(@_);
1099}
1100
1101# Returns the given filename converted using either URL encoding or base64
1102# encoding, as specified by $rename_method. If the given filename has no suffix
1103# (if it is just the tailname), then $no_suffix should be some defined value.
1104# rename_method can be url, none, base64
1105sub rename_file {
1106 my ($filename, $rename_method, $no_suffix) = @_;
1107
1108 if(!$filename) { # undefined or empty string
1109 return $filename;
1110 }
1111
1112 if (!$rename_method) {
1113 print STDERR "WARNING: no file renaming method specified. Defaulting to using URL encoding...\n";
1114 # Debugging information
1115 # my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(1);
1116 # print STDERR "Called from method: $cfilename:$cline $cpackage->$csubr\n";
1117 $rename_method = "url";
1118 } elsif($rename_method eq "none") {
1119 return $filename; # would have already been renamed
1120 }
1121
1122 # No longer replace spaces with underscores, since underscores mess with incremental rebuild
1123 ### Replace spaces with underscore. Do this first else it can go wrong below when getting tailname
1124 ###$filename =~ s/ /_/g;
1125
1126 my ($tailname,$dirname,$suffix);
1127 if($no_suffix) { # given a tailname, no suffix
1128 ($tailname,$dirname) = File::Basename::fileparse($filename);
1129 }
1130 else {
1131 ($tailname,$dirname,$suffix) = File::Basename::fileparse($filename, "\\.(?:[^\\.]+?)\$");
1132 }
1133 if (!$suffix) {
1134 $suffix = "";
1135 }
1136 # This breaks GLI matching extracted metadata to files in Enrich panel, as
1137 # original is eg .JPG while gsdlsourcefilename ends up .jpg
1138 # Not sure why it was done in first place...
1139 #else {
1140 #$suffix = lc($suffix);
1141 #}
1142
1143 if ($rename_method eq "url") {
1144 $tailname = &unicode::url_encode($tailname);
1145 }
1146 elsif ($rename_method eq "base64") {
1147 $tailname = &unicode::base64_encode($tailname);
1148 $tailname =~ s/\s*//sg; # for some reason it adds spaces not just at end but also in middle
1149 }
1150
1151 $filename = "$tailname$suffix";
1152 $filename = "$dirname$filename" if ($dirname ne "./" && $dirname ne ".\\");
1153
1154 return $filename;
1155}
1156
1157
1158# BACKWARDS COMPATIBILITY: Just in case there are old .ldb/.bdb files
1159sub rename_ldb_or_bdb_file {
1160 my ($filename_no_ext) = @_;
1161
1162 my $new_filename = "$filename_no_ext.gdb";
1163 return if (-f $new_filename); # if the file has the right extension, don't need to do anything
1164 # try ldb
1165 my $old_filename = "$filename_no_ext.ldb";
1166
1167 if (-f $old_filename) {
1168 print STDERR "Renaming $old_filename to $new_filename\n";
1169 rename ($old_filename, $new_filename)
1170 || print STDERR "Rename failed: $!\n";
1171 return;
1172 }
1173 # try bdb
1174 $old_filename = "$filename_no_ext.bdb";
1175 if (-f $old_filename) {
1176 print STDERR "Renaming $old_filename to $new_filename\n";
1177 rename ($old_filename, $new_filename)
1178 || print STDERR "Rename failed: $!\n";
1179 return;
1180 }
1181}
1182
1183sub os_dir() {
1184
1185 my $gsdlarch = "";
1186 if(defined $ENV{'GSDLARCH'}) {
1187 $gsdlarch = $ENV{'GSDLARCH'};
1188 }
1189 return $ENV{'GSDLOS'}.$gsdlarch;
1190}
1191
1192# returns 1 if this (GS server) is a GS3 installation, returns 0 if it's GS2.
1193sub is_gs3() {
1194 if($ENV{'GSDL3SRCHOME'}) {
1195 return 1;
1196 } else {
1197 return 0;
1198 }
1199}
1200
1201# Returns the greenstone URL prefix extracted from the appropriate GS2/GS3 config file.
1202# By default, /greenstone3 for GS3 or /greenstone for GS2.
1203sub get_greenstone_url_prefix() {
1204 # if already set on a previous occasion, just return that
1205 # (Don't want to keep repeating this: cost of re-opening and scanning files.)
1206 return $ENV{'GREENSTONE_URL_PREFIX'} if($ENV{'GREENSTONE_URL_PREFIX'});
1207
1208 my ($configfile, $urlprefix, $defaultUrlprefix);
1209 my @propertynames = ();
1210
1211 if($ENV{'GSDL3SRCHOME'}) {
1212 $defaultUrlprefix = "/greenstone3";
1213 $configfile = &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'}, "packages", "tomcat", "conf", "Catalina", "localhost", "greenstone3.xml");
1214 push(@propertynames, qw/path\s*\=/);
1215 } else {
1216 $defaultUrlprefix = "/greenstone";
1217 $configfile = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "cgi-bin", &os_dir(), "gsdlsite.cfg");
1218 push(@propertynames, (qw/\nhttpprefix/, qw/\ngwcgi/)); # inspect one property then the other
1219 }
1220
1221 $urlprefix = &extract_propvalue_from_file($configfile, \@propertynames);
1222
1223 if(!$urlprefix) { # no values found for URL prefix, use default values
1224 $urlprefix = $defaultUrlprefix;
1225 } else {
1226 #gwcgi can contain more than the wanted prefix, we split on / to get the first "directory" level
1227 $urlprefix =~ s/^\///; # remove the starting slash
1228 my @dirs = split(/(\\|\/)/, $urlprefix);
1229 $urlprefix = shift(@dirs);
1230
1231 if($urlprefix !~ m/^\//) { # in all cases: ensure the required forward slash is at the front
1232 $urlprefix = "/$urlprefix";
1233 }
1234 }
1235
1236 # set for the future
1237 $ENV{'GREENSTONE_URL_PREFIX'} = $urlprefix;
1238# print STDERR "*** in get_greenstone_url_prefix(): $urlprefix\n\n";
1239 return $urlprefix;
1240}
1241
1242
1243
1244#
1245# The following comes from activate.pl
1246#
1247# Designed to work with a server included with GS.
1248# - For GS3, we ask ant for the library URL.
1249# - For GS2, we derive the URL from the llssite.cfg file.
1250
1251sub get_full_greenstone_url_prefix
1252{
1253 # if already set on a previous occasion, just return that
1254 # (Don't want to keep repeating this: cost of re-opening and scanning files.)
1255 return $ENV{'GREENSTONE_URL_PREFIX'} if($ENV{'GREENSTONE_URL_PREFIX'});
1256
1257 my $gs_mode = ($ENV{'GSDL3SRCHOME'}) ? "gs3" : "gs2";
1258
1259 my $url = undef;
1260
1261 if($gs_mode eq "gs2") {
1262 my $llssite_cfg = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "llssite.cfg");
1263
1264 if(-f $llssite_cfg) {
1265 # check llssite.cfg for line with url property
1266 # for server.exe also need to use portnumber and enterlib properties
1267
1268 # Read in the entire contents of the file in one hit
1269 if (!open (FIN, $llssite_cfg)) {
1270 print STDERR "util::get_library_URL() failed to open $llssite_cfg ($!)\n";
1271 return undef;
1272 }
1273
1274 my $contents;
1275 sysread(FIN, $contents, -s FIN);
1276 close(FIN);
1277
1278 my @lines = split(/[\n\r]+/, $contents); # split on carriage-returns and/or linefeeds
1279 my $enterlib = "";
1280 my $portnumber = "8282"; # will remain empty (implicit port 80) unless it's specifically been assigned
1281
1282 foreach my $line (@lines) {
1283 if($line =~ m/^url=(.*)$/) {
1284 $url = $1;
1285 } elsif($line =~ m/^enterlib=(.*)$/) {
1286 $enterlib = $1;
1287 } elsif($line =~ m/^portnumber=(.*)$/) {
1288 $portnumber = $1;
1289 }
1290 }
1291
1292 if(!$url) {
1293 return undef;
1294 }
1295 elsif($url eq "URL_pending") { # library is not running
1296 # do not process url=URL_pending in the file, since for server.exe
1297 # this just means the Enter Library button hasn't been pressed yet
1298 $url = undef;
1299 }
1300 else {
1301 # In the case of server.exe, need to do extra work to get the proper URL
1302 # But first, need to know whether we're indeed dealing with server.exe:
1303
1304 # compare the URL's domain to the full URL
1305 # E.g. for http://localhost:8383/greenstone3/cgi-bin, the domain is localhost:8383
1306 my $uri = URI->new( $url );
1307 my $host = $uri->host;
1308 #print STDERR "@@@@@ host: $host\n";
1309 if($url =~ m/http:\/\/$host(\/)?$/) {
1310 #if($url !~ m/http:\/\/$host:$portnumber(\/)?/ || $url =~ m/http:\/\/$host(\/)?$/) {
1311 # (if the URL does not contain the portnumber, OR if the port is implicitly 80 and)
1312 # If the domain with http:// prefix is completely the same as the URL, assume server.exe
1313 # then the actual URL is the result of suffixing the port and enterlib properties in llssite.cfg
1314 $url = $url.":".$portnumber.$enterlib;
1315 } # else, apache web server
1316
1317 }
1318 }
1319 } elsif($gs_mode eq "gs3") {
1320 # Either check build.properties for tomcat.server, tomcat.port and app.name (and default servlet name).
1321 # app.name is stored in app.path by build.xml. Need to move app.name in build.properties from build.xml
1322
1323 # Or, run the new target get-default-servlet-url
1324 # the output can look like:
1325 #
1326 # Buildfile: build.xml
1327 # [echo] os.name: Windows Vista
1328 #
1329 # get-default-servlet-url:
1330 # [echo] http://localhost:8383/greenstone3/library
1331 # BUILD SUCCESSFUL
1332 # Total time: 0 seconds
1333
1334 #my $output = qx/ant get-default-servlet-url/; # backtick operator, to get STDOUT (else 2>&1)
1335 # - see http://stackoverflow.com/questions/799968/whats-the-difference-between-perls-backticks-system-and-exec
1336
1337 # The get-default-servlet-url ant target can be run from anywhere by specifying the
1338 # location of GS3's ant build.xml buildfile. Activate.pl can be run from anywhere for GS3
1339 # GSDL3SRCHOME will be set for GS3 by gs3-setup.sh, a step that would have been necessary
1340 # to run the activate.pl script in the first place
1341
1342 my $full_build_xml = &FileUtils::javaFilenameConcatenate($ENV{'GSDL3SRCHOME'},"build.xml");
1343
1344 my $perl_command = "ant -buildfile \"$full_build_xml\" get-default-servlet-url";
1345
1346 if (open(PIN, "$perl_command |")) {
1347 while (defined (my $perl_output_line = <PIN>)) {
1348
1349 if($perl_output_line =~ m@http:\/\/(\S*)@) { # grab all the non-whitespace chars
1350 $url="http://".$1;
1351 }
1352 }
1353 close(PIN);
1354 } else {
1355 print STDERR "util::get_library_URL() failed to run $perl_command to work out library URL for $gs_mode\n";
1356 }
1357 }
1358
1359 # either the url is still undef or it is now set
1360 #print STDERR "\n@@@@@ final URL:|$url|\n" if $url;
1361 #print STDERR "\n@@@@@ URL still undef\n" if !$url;
1362
1363 $ENV{'FULL_GREENSTONE_URL_PREFIX'} = $url;
1364
1365 return $url;
1366}
1367
1368
1369# Given a config file (xml or java properties file) and a list/array of regular expressions
1370# that represent property names to match on, this function will return the value for the 1st
1371# matching property name. If the return value is undefined, no matching property was found.
1372sub extract_propvalue_from_file() {
1373 my ($configfile, $propertynames) = @_;
1374
1375 my $value;
1376 unless(open(FIN, "<$configfile")) {
1377 print STDERR "extract_propvalue_from_file(): Unable to open $configfile. $!\n";
1378 return $value; # not initialised
1379 }
1380
1381 # Read the entire file at once, as one single line, then close it
1382 my $filecontents;
1383 {
1384 local $/ = undef;
1385 $filecontents = <FIN>;
1386 }
1387 close(FIN);
1388
1389 foreach my $regex (@$propertynames) {
1390 ($value) = $filecontents=~ m/$regex\s*(\S*)/s; # read value of the property given by regex up to the 1st space
1391 if($value) {
1392 $value =~ s/^\"//; # remove any startquotes
1393 $value =~ s/\".*$//; # remove the 1st endquotes (if any) followed by any xml
1394 last; # found value for a matching property, break from loop
1395 }
1396 }
1397
1398 return $value;
1399}
1400
1401# Subroutine that sources setup.bash, given GSDLHOME and GSDLOS and
1402# given that perllib is in @INC in order to invoke this subroutine.
1403# Call as follows -- after setting up INC to include perllib and
1404# after setting up GSDLHOME and GSDLOS:
1405#
1406# require util;
1407# &util::setup_greenstone_env($ENV{'GSDLHOME'}, $ENV{'GSDLOS'});
1408#
1409sub setup_greenstone_env() {
1410 my ($GSDLHOME, $GSDLOS) = @_;
1411
1412 #my %env_map = ();
1413 # Get the localised ENV settings of running a localised source setup.bash
1414 # and put it into the ENV here. Need to clear GSDLHOME before running setup
1415 #my $perl_command = "(cd $GSDLHOME; export GSDLHOME=; . ./setup.bash > /dev/null; env)";
1416 my $perl_command = "(cd $GSDLHOME; /bin/bash -c \"export GSDLHOME=; source setup.bash > /dev/null; env\")";
1417 if (($GSDLOS =~ m/windows/i) && ($^O ne "cygwin")) {
1418 #$perl_command = "cmd /C \"cd $GSDLHOME&& set GSDLHOME=&& setup.bat > nul&& set\"";
1419 $perl_command = "(cd $GSDLHOME&& set GSDLHOME=&& setup.bat > nul&& set)";
1420 }
1421 if (!open(PIN, "$perl_command |")) {
1422 print STDERR ("Unable to execute command: $perl_command. $!\n");
1423 }
1424
1425 my $lastkey;
1426 while (defined (my $perl_output_line = <PIN>)) {
1427 my($key,$value) = ($perl_output_line =~ m/^([^=]*)[=](.*)$/);
1428 if(defined $key) {
1429 #$env_map{$key}=$value;
1430 $ENV{$key}=$value;
1431 $lastkey = $key;
1432 } elsif($lastkey && $perl_output_line !~ m/^\s*$/) {
1433 # there was no equals sign in $perl_output_line, so this
1434 # $perl_output_line may be a spillover from the previous
1435 $ENV{$lastkey} = $ENV{$lastkey}."\n".$perl_output_line;
1436 }
1437 }
1438 close (PIN);
1439
1440 # If any keys in $ENV don't occur in Greenstone's localised env
1441 # (stored in $env_map), delete those entries from $ENV
1442 #foreach $key (keys %ENV) {
1443 # if(!defined $env_map{$key}) {
1444 # print STDOUT "**** DELETING ENV KEY: $key\tVALUE: $ENV{$key}\n";
1445 # delete $ENV{$key}; # del $ENV(key, value) pair
1446 # }
1447 #}
1448 #undef %env_map;
1449}
1450
1451sub get_perl_exec() {
1452 my $perl_exec = $^X; # may return just "perl"
1453
1454 if($ENV{'PERLPATH'}) {
1455 # OR: # $perl_exec = &FileUtils::filenameConcatenate($ENV{'PERLPATH'},"perl");
1456 if (($ENV{'GSDLOS'} =~ m/windows/) && ($^O ne "cygwin")) {
1457 $perl_exec = "$ENV{'PERLPATH'}\\Perl.exe";
1458 } else {
1459 $perl_exec = "$ENV{'PERLPATH'}/perl";
1460 }
1461 } else { # no PERLPATH, use Config{perlpath} else $^X: special variables
1462 # containing the full path to the current perl executable we're using
1463 $perl_exec = $Config{perlpath}; # configured path for perl
1464 if (!-e $perl_exec) { # may not point to location on this machine
1465 $perl_exec = $^X; # may return just "perl"
1466 if($perl_exec =~ m/^perl/i) { # warn if just perl or Perl.exe
1467 print STDERR "**** WARNING: Perl exec found contains no path: $perl_exec";
1468 }
1469 }
1470 }
1471
1472 return $perl_exec;
1473}
1474
1475# returns the path to the java command in the JRE included with GS (if any),
1476# quoted to safeguard any spaces in this path, otherwise a simple java
1477# command is returned which assumes and will try for a system java.
1478sub get_java_command {
1479 my $java = "java";
1480 if(defined $ENV{'GSDLHOME'}) { # should be, as this script would be launched from the cmd line
1481 # after running setup.bat or from GLI which also runs setup.bat
1482 my $java_bin = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"packages","jre","bin");
1483 if(-d $java_bin) {
1484 $java = &FileUtils::filenameConcatenate($java_bin,"java");
1485 $java = "\"".$java."\""; # quoted to preserve spaces in path
1486 }
1487 }
1488 return $java;
1489}
1490
1491
1492# Given the qualified collection name (colgroup/collection),
1493# returns the collection and colgroup parts
1494sub get_collection_parts {
1495 # http://perldoc.perl.org/File/Basename.html
1496 # my($filename, $directories, $suffix) = fileparse($path);
1497 # "$directories contains everything up to and including the last directory separator in the $path
1498 # including the volume (if applicable). The remainder of the $path is the $filename."
1499 #my ($collection, $colgroup) = &File::Basename::fileparse($qualified_collection);
1500
1501 my $qualified_collection = shift(@_);
1502
1503 # Since activate.pl can be launched from the command-line, including by a user,
1504 # best not to assume colgroup uses URL-style slashes as would be the case with GLI
1505 # Also allow for the accidental inclusion of multiple slashes
1506 my ($colgroup, $collection) = split(/[\/\\]+/, $qualified_collection); #split('/', $qualified_collection);
1507
1508 if(!defined $collection) {
1509 $collection = $colgroup;
1510 $colgroup = "";
1511 }
1512 return ($collection, $colgroup);
1513}
1514
1515# work out the "collectdir/collection" location
1516sub resolve_collection_dir {
1517 my ($collect_dir, $qualified_collection, $site) = @_; #, $gs_mode
1518
1519 if (defined $ENV{'GSDLCOLLECTDIR'}) { # a predefined collection dir exists
1520 return $ENV{'GSDLCOLLECTDIR'};
1521 }
1522
1523 my ($colgroup, $collection) = &util::get_collection_parts($qualified_collection);
1524
1525 if (!defined $collect_dir || !$collect_dir) { # if undefined or empty string
1526 $collect_dir = &util::get_working_collect_dir($site);
1527 }
1528
1529 return &FileUtils::filenameConcatenate($collect_dir,$colgroup,$collection);
1530}
1531
1532# work out the full path to "collect" of this greenstone 2/3 installation
1533sub get_working_collect_dir {
1534 my ($site) = @_;
1535
1536 if (defined $ENV{'GSDLCOLLECTHOME'}) { # a predefined collect dir exists
1537 return $ENV{'GSDLCOLLECTHOME'};
1538 }
1539
1540 if (defined $site && $site) { # site non-empty, so get default collect dir for GS3
1541
1542 if (defined $ENV{'GSDL3HOME'}) {
1543 return &FileUtils::filenameConcatenate($ENV{'GSDL3HOME'},"sites",$site,"collect"); # web folder
1544 }
1545 elsif (defined $ENV{'GSDL3SRCHOME'}) {
1546 return &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'},"web","sites",$site,"collect");
1547 }
1548 }
1549
1550 elsif (defined $ENV{'SITEHOME'}) {
1551 return &FileUtils::filenameConcatenate($ENV{'SITEHOME'},"collect");
1552 }
1553
1554 else { # get default collect dir for GS2
1555 return &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"collect");
1556 }
1557}
1558
1559sub is_abs_path_any_os {
1560 my ($path) = @_;
1561
1562 # We can have filenames in our DBs that were produced on other OS, so this method exists
1563 # to help identify absolute paths in such cases.
1564
1565 return 1 if($path =~ m@^/@); # full paths begin with forward slash on linux/mac
1566 return 1 if($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
1567
1568 return 0;
1569}
1570
1571
1572# This subroutine is for improving portability of Greenstone collections from one OS to another,
1573# to be used to convert absolute paths going into db files into paths with placeholders instead.
1574# This sub works with util::get_common_gs_paths and takes a path to a greenstone file and, if it's
1575# an absolute path, then it will replace the longest matching greenstone-path prefix of the given
1576# path with a placeholder to match.
1577# The Greenstone-path prefixes that can be matched are the following common Greenstone paths:
1578# the path to the current (specific) collection, the path to the general GS collect directory,
1579# the path to the site directory if GS3, else the path to the GSDLHOME/GSDL3HOME folder.
1580# The longest matching prefix will be replaced with the equivalent placeholder:
1581# @THISCOLLECTPATH@, else @COLLECTHOME@, else @SITEHOME@, else @GSDLHOME@.
1582sub abspath_to_placeholders {
1583 my $path = shift(@_); # path to convert from absolute to one with placeholders
1584 my $opt_long_or_short_winfilenames = shift(@_) || "short"; # whether we want to force use of long file names even on windows, default uses short
1585
1586 return $path unless is_abs_path_any_os($path); # path is relative
1587
1588 if ($opt_long_or_short_winfilenames eq "long") {
1589 $path = &util::upgrade_if_dos_filename($path); # will only do something on windows
1590 }
1591
1592 # now we know we're dealing with absolute paths and have to replace gs prefixes with placeholders
1593 my @gs_paths = ($ENV{'GSDLCOLLECTDIR'}, $ENV{'GSDLCOLLECTHOME'}, $ENV{'SITEHOME'}, $ENV{'GREENSTONEHOME'}); # list in this order: from longest to shortest path
1594
1595 my %placeholder_map = ($ENV{'GREENSTONEHOME'} => '@GSDLHOME@', # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
1596 $ENV{'GSDLCOLLECTHOME'} => '@COLLECTHOME@',
1597 $ENV{'GSDLCOLLECTDIR'} => '@THISCOLLECTPATH@'
1598 );
1599 $placeholder_map{$ENV{'SITEHOME'}} = '@SITEHOME@' if defined $ENV{'SITEHOME'};
1600
1601 $path = &util::_abspath_to_placeholders($path, \@gs_paths, \%placeholder_map);
1602
1603 if ($ENV{'GSDLOS'} =~ /^windows$/i && $opt_long_or_short_winfilenames eq "short") {
1604 # for windows need to look for matches on short file names too
1605 # matched paths are again to be replaced with the usual placeholders
1606
1607 my $gsdlcollectdir = &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTDIR'});
1608 my $gsdlcollecthome = &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTHOME'});
1609 my $sitehome = (defined $ENV{'SITEHOME'}) ? &util::downgrade_if_dos_filename($ENV{'SITEHOME'}) : undef;
1610 my $greenstonehome = &util::downgrade_if_dos_filename($ENV{'GREENSTONEHOME'});
1611
1612 @gs_paths = ($gsdlcollectdir, $gsdlcollecthome, $sitehome, $greenstonehome); # order matters
1613
1614 %placeholder_map = ($greenstonehome => '@GSDLHOME@', # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
1615 $gsdlcollecthome => '@COLLECTHOME@',
1616 $gsdlcollectdir => '@THISCOLLECTPATH@'
1617 );
1618 $placeholder_map{$sitehome} = '@SITEHOME@' if defined $sitehome;
1619
1620 $path = &util::_abspath_to_placeholders($path, \@gs_paths, \%placeholder_map);
1621 }
1622
1623 return $path;
1624}
1625
1626sub _abspath_to_placeholders {
1627 my ($path, $gs_paths_ref, $placeholder_map_ref) = @_;
1628
1629 # The sequence of elements in @gs_paths matters
1630 # Need to loop starting from the *longest* matching path (the path to the specific collection)
1631 # to the shortest matching path (the path to gsdlhome/gsdl3home folder):
1632
1633 foreach my $gs_path (@$gs_paths_ref) {
1634 next if(!defined $gs_path); # site undefined for GS2
1635
1636 my $re_path = &util::filename_to_regex($gs_path); # escape for regex
1637
1638 if($path =~ m/^$re_path/i) { # case sensitive or not for OS?
1639
1640 my $placeholder = $placeholder_map_ref->{$gs_path}; # get the placeholder to replace the matched path with
1641
1642 $path =~ s/^$re_path/$placeholder/; #case sensitive or not?
1643 #$path =~ s/^[\\\/]//; # remove gs_path's trailing separator left behind at the start of the path
1644 # lowercase file extension, This is needed when shortfilenames are used, as case affects alphetical ordering, which affects diffcol
1645 $path =~ s/\.([A-Z]+)$/".".lc($1)/e;
1646 last; # done
1647 }
1648 }
1649
1650 return $path;
1651}
1652
1653# Function that does the reverse of the util::abspath_to_placeholders subroutine
1654# Once again, call this with the values returned from util::get_common_gs_paths
1655sub placeholders_to_abspath {
1656 my $path = shift(@_); # path that can contain placeholders to convert to resolved absolute path
1657 my $opt_long_or_short_winfilenames = shift(@_) || "short"; # whether we want to force use of long file names even on windows, default uses short
1658
1659 return $path if($path !~ m/@/); # path contains no placeholders
1660
1661 # replace placeholders with gs prefixes
1662 my @placeholders = ('@THISCOLLECTPATH@', '@COLLECTHOME@', '@SITEHOME@', '@GSDLHOME@'); # order of paths not crucial in this case,
1663 # but listed here from longest to shortest once placeholders are have been resolved
1664
1665 # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
1666 my %placeholder_to_gspath_map;
1667 if ($ENV{'GSDLOS'} =~ /^windows$/i && $opt_long_or_short_winfilenames eq "short") {
1668 # always replace placeholders with short file names of the absolute paths on windows?
1669 %placeholder_to_gspath_map = ('@GSDLHOME@' => &util::downgrade_if_dos_filename($ENV{'GREENSTONEHOME'}),
1670 '@COLLECTHOME@' => &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTHOME'}),
1671 '@THISCOLLECTPATH@' => &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTDIR'})
1672 );
1673 $placeholder_to_gspath_map{'@SITEHOME@'} = &util::downgrade_if_dos_filename($ENV{'SITEHOME'}) if defined $ENV{'SITEHOME'};
1674 } else {
1675 %placeholder_to_gspath_map = ('@GSDLHOME@' => $ENV{'GREENSTONEHOME'},
1676 '@SITEHOME@' => $ENV{'SITEHOME'}, # can be undef
1677 '@COLLECTHOME@' => $ENV{'GSDLCOLLECTHOME'},
1678 '@THISCOLLECTPATH@' => $ENV{'GSDLCOLLECTDIR'}
1679 ); # $placeholder_to_gspath_map{'@SITEHOME@'} = $ENV{'SITEHOME'} if defined $ENV{'SITEHOME'};
1680 }
1681
1682 foreach my $placeholder (@placeholders) {
1683 my $gs_path = $placeholder_to_gspath_map{$placeholder};
1684
1685 next if(!defined $gs_path); # sitehome for GS2 is undefined
1686
1687 if($path =~ m/^$placeholder/) {
1688 $path =~ s/^$placeholder/$gs_path/;
1689 last; # done
1690 }
1691 }
1692
1693 return $path;
1694}
1695
1696# Used by pdfpstoimg.pl and PDFBoxConverter to create a .item file from
1697# a directory containing sequentially numbered images.
1698sub create_itemfile
1699{
1700 my ($output_dir, $convert_basename, $convert_to) = @_;
1701 my $page_num = "";
1702
1703 opendir(DIR, $output_dir) || die "can't opendir $output_dir: $!";
1704 my @dir_files = grep {-f "$output_dir/$_"} readdir(DIR);
1705 closedir DIR;
1706
1707 # Sort files in the directory by page_num
1708 sub page_number {
1709 my ($dir) = @_;
1710 my ($pagenum) =($dir =~ m/^.*?[-\.]?(\d+)(\.(jpg|gif|png))?$/i);
1711# my ($pagenum) =($dir =~ m/(\d+)(\.(jpg|gif|png))?$/i); # this works but is not as safe/strict about input filepatterns as the above
1712
1713 $pagenum = 1 unless defined $pagenum;
1714 return $pagenum;
1715 }
1716
1717 # sort the files in the directory in the order of page_num rather than lexically.
1718 @dir_files = sort { page_number($a) <=> page_number($b) } @dir_files;
1719
1720 # work out if the numbering of the now sorted image files starts at 0 or not
1721 # by checking the number of the first _image_ file (skipping item files)
1722 my $starts_at_0 = 0;
1723 my $firstfile = ($dir_files[0] !~ /\.item$/i) ? $dir_files[0] : $dir_files[1];
1724 if(page_number($firstfile) == 0) { # 00 will evaluate to 0 too in this condition
1725 $starts_at_0 = 1;
1726 }
1727
1728 my $item_file = &FileUtils::filenameConcatenate($output_dir, $convert_basename.".item");
1729 my $item_fh;
1730 &FileUtils::openFileHandle($item_file, 'w', \$item_fh);
1731 print $item_fh "<PagedDocument>\n";
1732
1733 foreach my $file (@dir_files){
1734 if ($file !~ /\.item/i){
1735 $page_num = page_number($file);
1736 $page_num++ if $starts_at_0; # image numbers start at 0, so add 1
1737 print $item_fh " <Page pagenum=\"$page_num\" imgfile=\"$file\" txtfile=\"\"/>\n";
1738 }
1739 }
1740
1741 print $item_fh "</PagedDocument>\n";
1742 &FileUtils::closeFileHandle($item_file, \$item_fh);
1743 return $item_file;
1744}
1745
1746# Sets the gnomelib_env. Based on the logic in wvware.pl which can perhaps be replaced with a call to this function in future
1747sub set_gnomelib_env
1748{
1749 ## SET THE ENVIRONMENT AS DONE IN SETUP.BASH/BAT OF GNOME-LIB
1750 # Though this is only needed for darwin Lion at this point (and android, though that is untested)
1751
1752 my $libext = "so";
1753 if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
1754 return;
1755 } elsif ($ENV{'GSDLOS'} =~ m/^darwin$/i) {
1756 $libext = "dylib";
1757 }
1758
1759 if (!defined $ENV{'GEXTGNOME'}) {
1760 ##print STDERR "@@@ Setting GEXTGNOME env\n";
1761
1762 my $gnome_dir = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"ext","gnome-lib-minimal");
1763
1764 if(! -d $gnome_dir) {
1765 $gnome_dir = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"ext","gnome-lib");
1766
1767 if(! -d $gnome_dir) {
1768 $gnome_dir = "";
1769 }
1770 }
1771
1772 # now set other the related env vars,
1773 # IF we've found the gnome-lib dir installed in the ext folder
1774
1775 if ($gnome_dir ne "" && -f &FileUtils::filenameConcatenate($gnome_dir, $ENV{'GSDLOS'}, "lib", "libiconv.$libext")) {
1776 $ENV{'GEXTGNOME'} = $gnome_dir;
1777 $ENV{'GEXTGNOME_INSTALLED'}=&FileUtils::filenameConcatenate($ENV{'GEXTGNOME'}, $ENV{'GSDLOS'});
1778
1779 my $gnomelib_bin = &FileUtils::filenameConcatenate($ENV{'GEXTGNOME_INSTALLED'}, "bin");
1780 if(-d $gnomelib_bin) { # no bin subfolder in GS binary's cutdown gnome-lib-minimal folder
1781 &util::envvar_prepend("PATH", $gnomelib_bin);
1782 }
1783
1784 # util's prepend will create LD/DYLD_LIB_PATH if it doesn't exist yet
1785 my $gextlib = &FileUtils::filenameConcatenate($ENV{'GEXTGNOME_INSTALLED'}, "lib");
1786
1787 if($ENV{'GSDLOS'} eq "linux") {
1788 &util::envvar_prepend("LD_LIBRARY_PATH", $gextlib);
1789 }
1790 elsif ($ENV{'GSDLOS'} eq "darwin") {
1791 #&util::envvar_prepend("DYLD_LIBRARY_PATH", $gextlib);
1792 &util::envvar_prepend("DYLD_FALLBACK_LIBRARY_PATH", $gextlib);
1793 }
1794 }
1795
1796 # Above largely mimics the setup.bash of the gnome-lib-minimal.
1797 # Not doing the devel-srcpack that gnome-lib-minimal's setup.bash used to set
1798 # Not exporting GSDLEXTS variable either
1799 }
1800
1801# print STDERR "@@@@@ GEXTGNOME: ".$ENV{'GEXTGNOME'}."\n\tINSTALL".$ENV{'GEXTGNOME_INSTALLED'}."\n";
1802# print STDERR "\tPATH".$ENV{'PATH'}."\n";
1803# print STDERR "\tLD_LIB_PATH".$ENV{'LD_LIBRARY_PATH'}."\n" if $ENV{'LD_LIBRARY_PATH};
1804# print STDERR "\tDYLD_FALLBACK_LIB_PATH".$ENV{'DYLD_FALLBACK_LIBRARY_PATH'}."\n" if $ENV{'DYLD_FALLBACK_LIBRARY_PATH};
1805
1806 # if no GEXTGNOME, maybe users didn't need gnome-lib to run gnomelib/libiconv dependent binaries like hashfile, suffix, wget
1807 # (wvware is launched in a gnomelib env from its own script, but could possibly go through this script in future)
1808}
1809
1810
1811
1812## @function augmentINC()
1813#
1814# Prepend a path (if it exists) onto INC but only if it isn't already in INC
1815# @param $new_path The path to add
1816# @author jmt12
1817#
1818sub augmentINC
1819{
1820 my ($new_path) = @_;
1821 my $did_add_path = 0;
1822 # might need to be replaced with FileUtils::directoryExists() call eventually
1823 if (-d $new_path)
1824 {
1825 my $did_find_path = 0;
1826 foreach my $existing_path (@INC)
1827 {
1828 if ($existing_path eq $new_path)
1829 {
1830 $did_find_path = 1;
1831 last;
1832 }
1833 }
1834 if (!$did_find_path)
1835 {
1836 unshift(@INC, $new_path);
1837 $did_add_path = 1;
1838 }
1839 }
1840 return $did_add_path;
1841}
1842## augmentINC()
1843
1844
18451;
Note: See TracBrowser for help on using the repository browser.