root/main/trunk/greenstone2/perllib/util.pm @ 28394

Revision 28394, 50.5 KB (checked in by davidb, 7 years ago)

Further support for operation under Cygwin added

  • 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 File::Copy;
33use File::Basename;
34# Config for getting the perlpath in the recommended way, though it uses paths that are
35# hard-coded into the Config file that's generated upon configuring and compiling perl.
36# $^X works better in some cases to return the path to perl used to launch the script,
37# but if launched with plain "perl" (no full-path), that will be just what it returns.
38use Config;
39# New module for file related utility functions - intended as a
40# placeholder for an extension that allows a variety of different
41# filesystems (FTP, HTTP, SAMBA, WEBDav, HDFS etc)
42use FileUtils;
43
44if ($ENV{'GSDLOS'} =~ /^windows$/i) {
45    require Win32; # for working out Windows Long Filenames from Win 8.3 short filenames
46}
47
48# removes files (but not directories)
49sub rm {
50  warnings::warnif("deprecated", "util::rm() is deprecated, using FileUtils::removeFiles() instead");
51  return &FileUtils::removeFiles(@_);
52}
53
54# recursive removal
55sub filtered_rm_r {
56  warnings::warnif("deprecated", "util::filtered_rm_r() is deprecated, using FileUtils::removeFilesFiltered() instead");
57  return &FileUtils::removeFilesFiltered(@_);
58}
59
60# recursive removal
61sub rm_r {
62  warnings::warnif("deprecated", "util::rm_r() is deprecated, using FileUtils::removeFilesRecursive() instead");
63  return &FileUtils::removeFilesRecursive(@_);
64}
65
66# moves a file or a group of files
67sub mv {
68  warnings::warnif("deprecated", "util::mv() is deprecated, using FileUtils::moveFiles() instead");
69  return &FileUtils::moveFiles(@_);
70}
71
72# Move the contents of source directory into target directory
73# (as opposed to merely replacing target dir with the src dir)
74# This can overwrite any files with duplicate names in the target
75# but other files and folders in the target will continue to exist
76sub mv_dir_contents {
77  warnings::warnif("deprecated", "util::mv_dir_contents() is deprecated, using FileUtils::moveDirectoryContents() instead");
78  return &FileUtils::moveDirectoryContents(@_);
79}
80
81# copies a file or a group of files
82sub cp {
83  warnings::warnif("deprecated", "util::cp() is deprecated, using FileUtils::copyFiles() instead");
84  return &FileUtils::copyFiles(@_);
85}
86
87# recursively copies a file or group of files
88# syntax: cp_r (sourcefiles, destination directory)
89# destination must be a directory - to copy one file to
90# another use cp instead
91sub cp_r {
92  warnings::warnif("deprecated", "util::cp_r() is deprecated, using FileUtils::copyFilesrecursive() instead");
93  return &FileUtils::copyFilesRecursive(@_);
94}
95
96# recursively copies a file or group of files
97# syntax: cp_r (sourcefiles, destination directory)
98# destination must be a directory - to copy one file to
99# another use cp instead
100sub cp_r_nosvn {
101  warnings::warnif("deprecated", "util::cp_r_nosvn() is deprecated, using FileUtils::copyFilesRecursiveNoSVN() instead");
102  return &FileUtils::copyFilesRecursiveNoSVN(@_);
103}
104
105# copies a directory and its contents, excluding subdirectories, into a new directory
106sub cp_r_toplevel {
107  warnings::warnif("deprecated", "util::cp_r_toplevel() is deprecated, using FileUtils::recursiveCopyTopLevel() instead");
108  return &FileUtils::recursiveCopyTopLevel(@_);
109}
110
111sub mk_dir {
112  warnings::warnif("deprecated", "util::mk_dir() is deprecated, using FileUtils::makeDirectory() instead");
113  return &FileUtils::makeDirectory(@_);
114}
115
116# in case anyone cares - I did some testing (using perls Benchmark module)
117# on this subroutine against File::Path::mkpath (). mk_all_dir() is apparently
118# slightly faster (surprisingly) - Stefan.
119sub mk_all_dir {
120  warnings::warnif("deprecated", "util::mk_all_dir() is deprecated, using FileUtils::makeAllDirectories() instead");
121  return &FileUtils::makeAllDirectories(@_);
122}
123
124# make hard link to file if supported by OS, otherwise copy the file
125sub hard_link {
126  warnings::warnif("deprecated", "util::hard_link() is deprecated, using FileUtils::hardLink() instead");
127  return &FileUtils::hardLink(@_);
128}
129
130# make soft link to file if supported by OS, otherwise copy file
131sub soft_link {
132  warnings::warnif("deprecated", "util::soft_link() is deprecated, using FileUtils::softLink() instead");
133  return &FileUtils::softLink(@_);
134}
135
136# Primarily for filenames generated by processing
137# content of HTML files (which are mapped to UTF-8 internally)
138#
139# To turn this into an octet string that really exists on the file
140# system:
141# 1. don't need to do anything special for Unix-based systems
142#   (as underlying file system is byte-code)
143# 2. need to map to short DOS filenames for Windows
144
145sub utf8_to_real_filename
146{
147    my ($utf8_filename) = @_;
148
149    my $real_filename;
150
151    if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
152    require Win32;
153
154    print STDERR "***** utf8 filename = $utf8_filename\n\n\n";
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 fd_exists {
167  warnings::warnif("deprecated", "util::fd_exists() is deprecated, using FileUtils::fileTest() instead");
168  return &FileUtils::fileTest(@_);
169}
170
171sub file_exists {
172  warnings::warnif("deprecated", "util::file_exists() is deprecated, using FileUtils::fileExists() instead");
173  return &FileUtils::fileExists(@_);
174}
175
176sub dir_exists {
177  warnings::warnif("deprecated", "util::dir_exists() is deprecated, using FileUtils::directoryExists() instead");
178  return &FileUtils::directoryExists(@_);
179}
180
181# updates a copy of a directory in some other part of the filesystem
182# verbosity settings are: 0=low, 1=normal, 2=high
183# both $fromdir and $todir should be absolute paths
184sub cachedir {
185  warnings::warnif("deprecated", "util::cachedir() is deprecated, using FileUtils::synchronizeDirectories() instead");
186  return &FileUtils::synchronizeDirectories(@_);
187}
188
189# this function returns -1 if either file is not found
190# assumes that $file1 and $file2 are absolute file names or
191# in the current directory
192# $file2 is allowed to be newer than $file1
193sub differentfiles {
194  warnings::warnif("deprecated", "util::differentfiles() is deprecated, using FileUtils::differentFiles() instead");
195  return &FileUtils::differentFiles(@_);
196}
197
198
199sub get_tmp_filename
200{
201    my $file_ext = shift(@_) || undef;
202
203    my $opt_dot_file_ext = "";
204    if (defined $file_ext) {
205    if ($file_ext !~ m/\./) {
206        # no dot, so needs one added in at start
207        $opt_dot_file_ext = ".$file_ext"
208    }
209    else {
210        # allow for "extensions" such as _metadata.txt to be handled
211        # gracefully
212        $opt_dot_file_ext = $file_ext;
213    }
214    }
215
216    my $tmpdir = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "tmp");
217    &FileUtils::makeAllDirectories ($tmpdir) unless -e $tmpdir;
218
219    my $count = 1000;
220    my $rand = int(rand $count);
221    my $full_tmp_filename = &FileUtils::filenameConcatenate($tmpdir, "F$rand$opt_dot_file_ext");
222
223    while (-e $full_tmp_filename) {
224    $rand = int(rand $count);
225    $full_tmp_filename = &FileUtils::filenameConcatenate($tmpdir, "F$rand$opt_dot_file_ext");
226    $count++;
227    }
228   
229    return $full_tmp_filename;
230}
231
232# These 2 are "static" variables used by the get_timestamped_tmp_folder() subroutine below and
233# belong with that function. They help ensure the timestamped tmp folders generated are unique.
234my $previous_timestamp = undef;
235my $previous_timestamp_f = 0; # frequency
236
237sub get_timestamped_tmp_folder
238{
239   
240    my $tmp_dirname;
241    if(defined $ENV{'GSDLCOLLECTDIR'}) {
242    $tmp_dirname = $ENV{'GSDLCOLLECTDIR'};
243    } elsif(defined $ENV{'GSDLHOME'}) {
244    $tmp_dirname = $ENV{'GSDLHOME'};
245    } else {
246    return undef;
247    }
248
249    $tmp_dirname = &FileUtils::filenameConcatenate($tmp_dirname, "tmp");
250    &FileUtils::makeDirectory($tmp_dirname) if (!-e $tmp_dirname);
251
252    # add the timestamp into the path otherwise we can run into problems
253    # if documents have the same name
254    my $timestamp = time;   
255   
256    if (!defined $previous_timestamp || ($timestamp > $previous_timestamp)) {
257        $previous_timestamp_f = 0;
258        $previous_timestamp = $timestamp;
259    } else {
260        $previous_timestamp_f++;
261    }
262   
263    my $time_tmp_dirname = &FileUtils::filenameConcatenate($tmp_dirname, $timestamp);
264    $tmp_dirname = $time_tmp_dirname;   
265    my $i = $previous_timestamp_f;
266   
267    if($previous_timestamp_f > 0) {
268        $tmp_dirname = $time_tmp_dirname."_".$i;
269        $i++;
270    }
271    while (-e $tmp_dirname) {
272    $tmp_dirname = $time_tmp_dirname."_".$i;
273    $i++;
274    }
275    &FileUtils::makeDirectory($tmp_dirname);
276
277    return $tmp_dirname;
278}
279
280sub get_timestamped_tmp_filename_in_collection
281{
282
283    my ($input_filename, $output_ext) = @_;
284    # derive tmp filename from input filename
285    my ($tailname, $dirname, $suffix)
286    = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
287
288    # softlink to collection tmp dir
289    my $tmp_dirname = &util::get_timestamped_tmp_folder();
290    $tmp_dirname = $dirname unless defined $tmp_dirname;
291
292    # following two steps copied from ConvertBinaryFile
293    # do we need them?? can't use them as is, as they use plugin methods.
294
295    #$tailname = $self->SUPER::filepath_to_utf8($tailname) unless &unicode::check_is_utf8($tailname);
296
297    # URLEncode this since htmls with images where the html filename is utf8 don't seem
298    # to work on Windows (IE or Firefox), as browsers are looking for filesystem-encoded
299    # files on the filesystem.
300    #$tailname = &util::rename_file($tailname, $self->{'file_rename_method'}, "without_suffix");
301    if (defined $output_ext) {
302    $output_ext = ".$output_ext"; # add the dot
303    } else {
304    $output_ext = $suffix;
305    }
306    $output_ext= lc($output_ext);
307    my $tmp_filename = &FileUtils::filenameConcatenate($tmp_dirname, "$tailname$output_ext");
308   
309    return $tmp_filename;
310}
311
312sub get_toplevel_tmp_dir
313{
314    return &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "tmp");
315}
316
317
318sub filename_to_regex {
319    my $filename = shift (@_);
320
321    # need to make single backslashes double so that regex works
322    $filename =~ s/\\/\\\\/g; # if ($ENV{'GSDLOS'} =~ /^windows$/i);   
323   
324    # note that the first part of a substitution is a regex, so RE chars need to be escaped,
325    # the second part of a substitution is not a regex, so for e.g. full-stop can be specified literally
326    $filename =~ s/\./\\./g; # in case there are extensions/other full stops, escape them
327    $filename =~ s@\(@\\(@g; # escape brackets
328    $filename =~ s@\)@\\)@g; # escape brackets
329    $filename =~ s@\[@\\[@g; # escape brackets
330    $filename =~ s@\]@\\]@g; # escape brackets
331   
332    return $filename;
333}
334
335sub unregex_filename {
336    my $filename = shift (@_);
337
338    # need to put doubled backslashes for regex back to single
339    $filename =~ s/\\\./\./g; # remove RE syntax for .
340    $filename =~ s@\\\(@(@g; # remove RE syntax for ( => "\(" turns into "("
341    $filename =~ s@\\\)@)@g; # remove RE syntax for ) => "\)" turns into ")"
342    $filename =~ s@\\\[@[@g; # remove RE syntax for [ => "\[" turns into "["
343    $filename =~ s@\\\]@]@g; # remove RE syntax for ] => "\]" turns into "]"
344   
345    # \\ goes to \
346    # This is the last step in reverse mirroring the order of steps in filename_to_regex()
347    $filename =~ s/\\\\/\\/g; # remove RE syntax for \   
348    return $filename;
349}
350
351sub filename_cat {
352  # I've disabled this warning for now, as every Greenstone perl
353  # script seems to make use of this function and so you drown in a
354  # sea of deprecated warnings [jmt12]
355#  warnings::warnif("deprecated", "util::filename_cat() is deprecated, using FileUtils::filenameConcatenate() instead");
356  return &FileUtils::filenameConcatenate(@_);
357}
358
359
360sub _pathname_cat {
361    my $join_char  = shift(@_);
362    my $first_path = shift(@_);
363    my (@pathnames) = @_;
364
365    # If first_path is not null or empty, then add it back into the list
366    if (defined $first_path && $first_path =~ /\S/) {
367    unshift(@pathnames, $first_path);
368    }
369
370    my $pathname = join($join_char, @pathnames);
371
372    # remove duplicate slashes
373    if ($join_char eq ";") {
374    $pathname =~ s/[\\\/]+/\\/g;
375    } else {
376    $pathname =~ s/[\/]+/\//g;
377    # DB: want a pathname abc\de.html to remain like this
378    }
379
380    return $pathname;
381}
382
383
384sub pathname_cat {
385    my (@pathnames) = @_;
386
387    my $join_char;
388    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
389    $join_char = ";";
390    } else {
391    $join_char = ":";
392    }
393    return _pathname_cat($join_char,@pathnames);
394}
395
396
397sub javapathname_cat {
398    my (@pathnames) = @_;
399
400    my $join_char;
401
402    # Unlike pathname_cat() above, not interested if running in a Cygwin environment
403    # This is because the java we run is actually a native Windows executable
404
405    if (($ENV{'GSDLOS'} =~ /^windows$/i)) {
406    $join_char = ";";
407    } else {
408    $join_char = ":";
409    }
410    return _pathname_cat($join_char,@pathnames);
411}
412
413
414
415sub tidy_up_oid {
416    my ($OID) = @_;
417    if ($OID =~ /\./) {
418    print STDERR "Warning, identifier $OID contains periods (.), removing them\n";
419    $OID =~ s/\.//g; #remove any periods
420    }
421    if ($OID =~ /^\s.*\s$/) {
422    print STDERR "Warning, identifier $OID starts or ends with whitespace. Removing it\n";
423    # remove starting and trailing whitespace
424    $OID =~ s/^\s+//;
425    $OID =~ s/\s+$//;
426    }
427    if ($OID =~ /^[\d]*$/) {
428    print STDERR "Warning, identifier $OID contains only digits. Prepending 'D'.\n";
429    $OID = "D" . $OID;
430    }       
431   
432    return $OID;
433}
434
435sub envvar_prepend {
436    my ($var,$val) = @_;
437
438    # 64 bit linux can't handle ";" as path separator, so make sure to set this to the right one for the OS
439##    my $pathsep = (defined $ENV{'GSDLOS'} && $ENV{'GSDLOS'} !~ m/windows/) ? ":" : ";";
440
441    # Rewritten above to make ":" the default (Windows is the special
442    # case, anything else 'unusual' such as Solaris etc is Unix)
443    my $pathsep = (defined $ENV{'GSDLOS'} && (($ENV{'GSDLOS'} =~ m/windows/) && ($^O ne "cygwin"))) ? ";" : ":";
444
445    # do not prepend any value/path that's already in the environment variable
446   
447    my $escaped_val = &filename_to_regex($val); # escape any backslashes and brackets for upcoming regex
448    if (!defined($ENV{$var})) {
449    $ENV{$var} = "$val";
450    }
451    elsif($ENV{$var} !~ m/$escaped_val/) {
452    $ENV{$var} = "$val".$pathsep.$ENV{$var};
453    }
454}
455
456sub envvar_append {
457    my ($var,$val) = @_;
458
459    # 64 bit linux can't handle ";" as path separator, so make sure to set this to the right one for the OS
460    my $pathsep = (defined $ENV{'GSDLOS'} && $ENV{'GSDLOS'} !~ m/windows/) ? ":" : ";";
461   
462    # do not append any value/path that's already in the environment variable
463
464    my $escaped_val = &filename_to_regex($val); # escape any backslashes and brackets for upcoming regex
465    if (!defined($ENV{$var})) {
466    $ENV{$var} = "$val";
467    }
468    elsif($ENV{$var} !~ m/$escaped_val/) {
469    $ENV{$var} = $ENV{$var}.$pathsep."$val";
470    }
471}
472
473
474
475# splits a filename into a prefix and a tail extension using the tail_re, or
476# if that fails, splits on the file_extension . (dot)
477sub get_prefix_and_tail_by_regex {
478
479    my ($filename,$tail_re) = @_;
480   
481    my ($file_prefix,$file_ext) = ($filename =~ m/^(.*?)($tail_re)$/);
482    if ((!defined $file_prefix) || (!defined $file_ext)) {
483    ($file_prefix,$file_ext) = ($filename =~ m/^(.*)(\..*?)$/);
484    }
485
486    return ($file_prefix,$file_ext);
487}
488
489# get full path and file only path from a base_dir (which may be empty) and
490# file (which may contain directories)
491sub get_full_filenames {
492    my ($base_dir, $file) = @_;
493   
494#    my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(0);
495#    my ($lcfilename) = ($cfilename =~ m/([^\\\/]*)$/);
496#    print STDERR "** Calling method: $lcfilename:$cline $cpackage->$csubr\n";
497
498
499    my $filename_full_path = $file;
500    # add on directory if present
501    $filename_full_path = &FileUtils::filenameConcatenate($base_dir, $file) if $base_dir =~ /\S/;
502   
503    my $filename_no_path = $file;
504
505    # remove directory if present
506    $filename_no_path =~ s/^.*[\/\\]//;
507    return ($filename_full_path, $filename_no_path);
508}
509
510# returns the path of a file without the filename -- ie. the directory the file is in
511sub filename_head {
512    my $filename = shift(@_);
513
514    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
515    $filename =~ s/[^\\\\]*$//;
516    }
517    else {
518    $filename =~ s/[^\\\/]*$//;
519    }
520
521    return $filename;
522}
523
524
525
526# returns 1 if filename1 and filename2 point to the same
527# file or directory
528sub filenames_equal {
529    my ($filename1, $filename2) = @_;
530
531    # use filename_cat to clean up trailing slashes and
532    # multiple slashes
533    $filename1 = &FileUtils::filenameConcatenate($filename1);
534    $filename2 = &FileUtils::filenameConcatenate($filename2);
535
536    # filenames not case sensitive on windows
537    if ($ENV{'GSDLOS'} =~ /^windows$/i) {
538    $filename1 =~ tr/[A-Z]/[a-z]/;
539    $filename2 =~ tr/[A-Z]/[a-z]/;
540    }
541    return 1 if $filename1 eq $filename2;
542    return 0;
543}
544
545# If filename is relative to within_dir, returns the relative path of filename to that directory
546# with slashes in the filename returned as they were in the original (absolute) filename.
547sub filename_within_directory
548{
549    my ($filename,$within_dir) = @_;
550   
551    if ($within_dir !~ m/[\/\\]$/) {
552    my $dirsep = &util::get_dirsep();
553    $within_dir .= $dirsep;
554    }
555   
556    $within_dir = &filename_to_regex($within_dir); # escape DOS style file separator and brackets   
557    if ($filename =~ m/^$within_dir(.*)$/) {
558    $filename = $1;
559    }
560   
561    return $filename;
562}
563
564# If filename is relative to within_dir, returns the relative path of filename to that directory in URL format.
565# Filename and within_dir can be any type of slashes, but will be compared as URLs (i.e. unix-style slashes).
566# The subpath returned will also be a URL type filename.
567sub filename_within_directory_url_format
568{
569    my ($filename,$within_dir) = @_;
570   
571    # convert parameters only to / slashes if Windows
572   
573    my $filename_urlformat = &filepath_to_url_format($filename);
574    my $within_dir_urlformat = &filepath_to_url_format($within_dir);
575
576    #if ($within_dir_urlformat !~ m/\/$/) {
577        # make sure directory ends with a slash
578        #$within_dir_urlformat .= "/";
579    #}
580   
581    my $within_dir_urlformat_re = &filename_to_regex($within_dir_urlformat); # escape any special RE characters, such as brackets
582   
583    #print STDERR "@@@@@ $filename_urlformat =~ $within_dir_urlformat_re\n";
584   
585    # dir prefix may or may not end with a slash (this is discarded when extracting the sub-filepath)
586    if ($filename_urlformat =~ m/^$within_dir_urlformat_re(?:\/)*(.*)$/) {
587        $filename_urlformat = $1;
588    }
589   
590    return $filename_urlformat;
591}
592
593# Convert parameter to use / slashes if Windows (if on Linux leave any \ as is,
594# since on Linux it doesn't represent a file separator but an escape char).
595sub filepath_to_url_format
596{
597    my ($filepath) = @_;
598    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
599        # Only need to worry about Windows, as Unix style directories already in url-format
600        # Convert Windows style \ => /
601        $filepath =~ s@\\@/@g;     
602    }
603    return $filepath;
604}
605
606# regex filepaths on windows may include \\ as path separator. Convert \\ to /
607sub filepath_regex_to_url_format
608{
609    my ($filepath) = @_;
610    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
611    # Only need to worry about Windows, as Unix style directories already in url-format
612    # Convert Windows style \\ => /
613    $filepath =~ s@\\\\@/@g;       
614    }
615    return $filepath;
616   
617}
618
619# Like File::Basename::fileparse, but expects filepath in url format (ie only / slash for dirsep)
620# and ignores trailing /
621# returns (file, dirs) dirs will be empty if no subdirs
622sub url_fileparse
623{
624    my ($filepath) = @_;
625    # remove trailing /
626    $filepath =~ s@/$@@;
627    if ($filepath !~ m@/@) {
628    return ($filepath, "");
629    }
630    my ($dirs, $file) = $filepath =~ m@(.+/)([^/]+)@;
631    return ($file, $dirs);
632   
633}
634
635
636sub filename_within_collection
637{
638    my ($filename) = @_;
639
640    my $collect_dir = $ENV{'GSDLCOLLECTDIR'};
641   
642    if (defined $collect_dir) {
643
644    # if from within GSDLCOLLECTDIR, then remove directory prefix
645    # so source_filename is realative to it.  This is done to aid
646    # portability, i.e. the collection can be moved to somewhere
647    # else on the file system and the archives directory will still
648    # work.  This is needed, for example in the applet version of
649    # GLI where GSDLHOME/collect on the server will be different to
650    # the collect directory of the remove user.  Of course,
651    # GSDLCOLLECTDIR subsequently needs to be put back on to turn
652    # it back into a full pathname.
653
654    $filename = filename_within_directory($filename,$collect_dir);
655    }
656   
657    return $filename;
658}
659
660sub prettyprint_file
661{
662    my ($base_dir,$file,$gli) = @_;
663
664    my $filename_full_path = &FileUtils::filenameConcatenate($base_dir,$file);
665
666    if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
667    require Win32;
668
669    # For some reason base_dir in the form c:/a/b/c
670    # This leads to confusion later on, so turn it back into
671    # the more usual Windows form
672    $base_dir =~ s/\//\\/g;
673    my $long_base_dir = Win32::GetLongPathName($base_dir);
674    my $long_full_path = Win32::GetLongPathName($filename_full_path);
675
676    $file = filename_within_directory($long_full_path,$long_base_dir);
677    $file = encode("utf8",$file) if ($gli);
678    }
679
680    return $file;
681}
682
683
684sub upgrade_if_dos_filename
685{
686    my ($filename_full_path,$and_encode) = @_;
687
688    if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
689    # Ensure any DOS-like filename, such as test~1.txt, has been upgraded
690    # to its long (Windows) version
691    my $long_filename = Win32::GetLongPathName($filename_full_path);
692    if (defined $long_filename) {
693        $filename_full_path = $long_filename;
694    }
695    # Make sure initial drive letter is lower-case (to fit in with rest of Greenstone)
696    $filename_full_path =~ s/^(.):/\u$1:/;
697    if ((defined $and_encode) && ($and_encode)) {
698        $filename_full_path = encode("utf8",$filename_full_path);
699    }
700    }
701
702    return $filename_full_path;
703}
704
705
706sub downgrade_if_dos_filename
707{
708    my ($filename_full_path) = @_;
709
710    if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
711    require Win32;
712
713    # Ensure the given long Windows filename is in a form that can
714    # be opened by Perl => convert it to a short DOS-like filename
715
716    my $short_filename = Win32::GetShortPathName($filename_full_path);
717    if (defined $short_filename) {
718        $filename_full_path = $short_filename;
719    }
720    # Make sure initial drive letter is lower-case (to fit in
721    # with rest of Greenstone)
722    $filename_full_path =~ s/^(.):/\u$1:/;
723    }
724
725    return $filename_full_path;
726}
727
728sub block_filename
729{
730    my ($block_hash,$filename) = @_;
731   
732    if (($ENV{'GSDLOS'} =~ m/^windows$/) && ($^O ne "cygwin")) {
733   
734    # lower case the entire thing, eg for cover.jpg when its actually cover.JPG
735    my $lower_filename = lc($filename);
736    $block_hash->{'file_blocks'}->{$lower_filename} = 1;
737#   my $lower_drive = $filename;
738#   $lower_drive =~ s/^([A-Z]):/\l$1:/i;
739   
740#   my $upper_drive = $filename;
741#   $upper_drive =~ s/^([A-Z]):/\u$1:/i;
742#   
743#   $block_hash->{'file_blocks'}->{$lower_drive} = 1;
744#   $block_hash->{'file_blocks'}->{$upper_drive} = 1;       
745    }
746    else {
747    $block_hash->{'file_blocks'}->{$filename} = 1;
748    }
749}
750
751
752sub filename_is_absolute
753{
754  warnings::warnif("deprecated", "util::filename_is_absolute() is deprecated, using FileUtils::isFilenameAbsolute() instead");
755  return &FileUtils::isFilenameAbsolute(@_);
756}
757
758
759## @method make_absolute()
760#
761#  Ensure the given file path is absolute in respect to the given base path.
762#
763#  @param  $base_dir A string denoting the base path the given dir must be
764#                    absolute to.
765#  @param  $dir The directory to be made absolute as a string. Note that the
766#               dir may already be absolute, in which case it will remain
767#               unchanged.
768#  @return The now absolute form of the directory as a string.
769#
770#  @author John Thompson, DL Consulting Ltd.
771#  @copy 2006 DL Consulting Ltd.
772#
773#used in buildcol.pl, doesn't work for all cases --kjdon
774sub make_absolute {
775   
776    my ($base_dir, $dir) = @_;
777###    print STDERR "dir = $dir\n";
778    $dir =~ s/[\\\/]+/\//g;
779    $dir = $base_dir . "/$dir" unless ($dir =~ m|^(\w:)?/|);
780    $dir =~ s|^/tmp_mnt||;
781    1 while($dir =~ s|/[^/]*/\.\./|/|g);
782    $dir =~ s|/[.][.]?/|/|g;
783    $dir =~ tr|/|/|s;
784###    print STDERR "dir = $dir\n";
785   
786    return $dir;
787}
788## make_absolute() ##
789
790sub get_dirsep {
791
792    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
793    return "\\";
794    } else {
795    return "\/";
796    }
797}
798
799sub get_os_dirsep {
800
801    if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
802    return "\\\\";
803    } else {
804    return "\\\/";
805    }
806}
807
808sub get_re_dirsep {
809
810    return "\\\\|\\\/";
811}
812
813
814sub get_dirsep_tail {
815    my ($filename) = @_;
816   
817    # returns last part of directory or filename
818    # On unix e.g. a/b.d => b.d
819    #              a/b/c => c
820
821    my $dirsep = get_re_dirsep();
822    my @dirs = split (/$dirsep/, $filename);
823    my $tail = pop @dirs;
824
825    # - caused problems under windows
826    #my ($tail) = ($filename =~ m/^(?:.*?$dirsep)?(.*?)$/);
827
828    return $tail;
829}
830
831
832# if this is running on windows we want binaries to end in
833# .exe, otherwise they don't have to end in any extension
834sub get_os_exe {
835    return ".exe" if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin"));
836    return "";
837}
838
839
840# test to see whether this is a big or little endian machine
841sub is_little_endian
842{
843    # To determine the name of the operating system, the variable $^O is a cheap alternative to pulling it out of the Config module;
844    # 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
845    # Otherwise, it's little endian
846
847    #return 0 if $^O =~ /^darwin$/i;
848    #return 0 if $ENV{'GSDLOS'} =~ /^darwin$/i;
849   
850    # Going back to stating exactly whether the machine is little endian
851    # or big endian, without any special case for Macs. Since for rata it comes
852    # back with little endian and for shuttle with bigendian.
853    return (ord(substr(pack("s",1), 0, 1)) == 1);
854}
855
856
857# will return the collection name if successful, "" otherwise
858sub use_collection {
859    my ($collection, $collectdir) = @_;
860
861    if (!defined $collectdir || $collectdir eq "") {
862    $collectdir = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "collect");
863    }
864
865    if (!defined $ENV{'GREENSTONEHOME'}) { # for GS3, would have been defined in use_site_collection, to GSDL3HOME
866     $ENV{'GREENSTONEHOME'} = $ENV{'GSDLHOME'};
867    }
868
869    # get and check the collection
870    if (!defined($collection) || $collection eq "") {
871    if (defined $ENV{'GSDLCOLLECTION'}) {
872        $collection = $ENV{'GSDLCOLLECTION'};
873    } else {
874        print STDOUT "No collection specified\n";
875        return "";
876    }
877    }
878   
879    if ($collection eq "modelcol") {
880    print STDOUT "You can't use modelcol.\n";
881    return "";
882    }
883
884    # make sure the environment variables GSDLCOLLECTION and GSDLCOLLECTDIR
885    # are defined
886    $ENV{'GSDLCOLLECTION'} = $collection;
887    $ENV{'GSDLCOLLECTHOME'} = $collectdir;
888    $ENV{'GSDLCOLLECTDIR'} = &FileUtils::filenameConcatenate($collectdir, $collection);
889
890    # make sure this collection exists
891    if (!-e $ENV{'GSDLCOLLECTDIR'}) {
892    print STDOUT "Invalid collection ($collection).\n";
893    return "";
894    }
895
896    # everything is ready to go
897    return $collection;
898}
899
900sub get_current_collection_name {
901    return $ENV{'GSDLCOLLECTION'};
902}
903
904
905# will return the collection name if successful, "" otherwise. 
906# Like use_collection (above) but for greenstone 3 (taking account of site level)
907
908sub use_site_collection {
909    my ($site, $collection, $collectdir) = @_;
910
911    if (!defined $collectdir || $collectdir eq "") {
912    die "GSDL3HOME not set.\n" unless defined $ENV{'GSDL3HOME'};
913    $collectdir = &FileUtils::filenameConcatenate($ENV{'GSDL3HOME'}, "sites", $site, "collect");
914    }
915
916    if (defined $ENV{'GSDL3HOME'}) {
917    $ENV{'GREENSTONEHOME'} = $ENV{'GSDL3HOME'};     
918    $ENV{'SITEHOME'} = &FileUtils::filenameConcatenate($ENV{'GREENSTONEHOME'}, "sites", $site);
919    } elsif (defined $ENV{'GSDL3SRCHOME'}) {
920    $ENV{'GREENSTONEHOME'} = &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'}, "web"); 
921    $ENV{'SITEHOME'} = &FileUtils::filenameConcatenate($ENV{'GREENSTONEHOME'}, "sites", $site);
922    } else {
923    print STDERR "*** util::use_site_collection(). Warning: Neither GSDL3HOME nor GSDL3SRCHOME set.\n";
924    }
925
926    # collectdir explicitly set by this point (using $site variable if required).
927    # Can call "old" gsdl2 use_collection now.
928
929    return use_collection($collection,$collectdir);
930}
931
932
933
934sub locate_config_file
935{
936    my ($file) = @_;
937
938    my $locations = locate_config_files($file);
939
940    return shift @$locations; # returns undef if 'locations' is empty
941}
942
943
944sub locate_config_files
945{
946    my ($file) = @_;
947
948    my @locations = ();
949
950    if (-e $file) {
951    # Clearly specified (most likely full filename)
952    # No need to hunt in 'etc' directories, return value unchanged
953    push(@locations,$file);
954    }
955    else {
956    # Check for collection specific one before looking in global GSDL 'etc'
957    if (defined $ENV{'GSDLCOLLECTDIR'} && $ENV{'GSDLCOLLECTDIR'} ne "") {
958        my $test_collect_etc_filename
959        = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'},"etc", $file);
960       
961        if (-e $test_collect_etc_filename) {
962        push(@locations,$test_collect_etc_filename);
963        }
964    }
965    my $test_main_etc_filename
966        = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"etc", $file);
967    if (-e $test_main_etc_filename) {
968        push(@locations,$test_main_etc_filename);
969    }
970    }
971
972    return \@locations;
973}
974
975
976sub hyperlink_text
977{
978    my ($text) = @_;
979   
980    $text =~ s/(http:\/\/[^\s]+)/<a href=\"$1\">$1<\/a>/mg;
981    $text =~ s/(^|\s+)(www\.(\w|\.)+)/<a href=\"http:\/\/$2\">$2<\/a>/mg;
982
983    return $text;
984}
985
986
987# A method to check if a directory is empty (note that an empty directory still has non-zero size!!!)
988# Code is from http://episteme.arstechnica.com/eve/forums/a/tpc/f/6330927813/m/436007700831
989sub is_dir_empty {
990  warnings::warnif("deprecated", "util::is_dir_empty() is deprecated, using FileUtils::isDirectoryEmpty() instead");
991  return &FileUtils::isDirectoryEmpty(@_);
992}
993
994# Returns the given filename converted using either URL encoding or base64
995# encoding, as specified by $rename_method. If the given filename has no suffix
996# (if it is just the tailname), then $no_suffix should be some defined value.
997# rename_method can be url, none, base64
998sub rename_file {
999    my ($filename, $rename_method, $no_suffix)  = @_;
1000
1001    if(!$filename) { # undefined or empty string
1002    return $filename;
1003    }
1004
1005    if (!$rename_method) {
1006    print STDERR "WARNING: no file renaming method specified. Defaulting to using URL encoding...\n";
1007    # Debugging information
1008    # my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(1);
1009    # print STDERR "Called from method: $cfilename:$cline $cpackage->$csubr\n";
1010    $rename_method = "url";
1011    } elsif($rename_method eq "none") {
1012    return $filename; # would have already been renamed
1013    }
1014
1015    # No longer replace spaces with underscores, since underscores mess with incremental rebuild
1016    ### Replace spaces with underscore. Do this first else it can go wrong below when getting tailname
1017    ###$filename =~ s/ /_/g;
1018
1019    my ($tailname,$dirname,$suffix);
1020    if($no_suffix) { # given a tailname, no suffix
1021    ($tailname,$dirname) = File::Basename::fileparse($filename);
1022    }
1023    else {
1024    ($tailname,$dirname,$suffix) = File::Basename::fileparse($filename, "\\.(?:[^\\.]+?)\$");
1025    }
1026    if (!$suffix) {
1027    $suffix = "";
1028    }
1029    # This breaks GLI matching extracted metadata to files in Enrich panel, as
1030    # original is eg .JPG while gsdlsourcefilename ends up .jpg
1031    # Not sure why it was done in first place...
1032    #else {
1033    #$suffix = lc($suffix);
1034    #}
1035
1036    if ($rename_method eq "url") {
1037    $tailname = &unicode::url_encode($tailname);
1038    }
1039    elsif ($rename_method eq "base64") {
1040    $tailname = &unicode::base64_encode($tailname);
1041    $tailname =~ s/\s*//sg;      # for some reason it adds spaces not just at end but also in middle
1042    }
1043
1044    $filename = "$tailname$suffix";
1045    $filename = "$dirname$filename" if ($dirname ne "./" && $dirname ne ".\\");
1046
1047    return $filename;
1048}
1049
1050
1051# BACKWARDS COMPATIBILITY: Just in case there are old .ldb/.bdb files
1052sub rename_ldb_or_bdb_file {
1053    my ($filename_no_ext) = @_;
1054
1055    my $new_filename = "$filename_no_ext.gdb";
1056    return if (-f $new_filename); # if the file has the right extension, don't need to do anything
1057    # try ldb
1058    my $old_filename = "$filename_no_ext.ldb";
1059   
1060    if (-f $old_filename) {
1061    print STDERR "Renaming $old_filename to $new_filename\n";
1062    rename ($old_filename, $new_filename)
1063        || print STDERR "Rename failed: $!\n";
1064    return;
1065    }
1066    # try bdb
1067    $old_filename = "$filename_no_ext.bdb";
1068    if (-f $old_filename) {
1069    print STDERR "Renaming $old_filename to $new_filename\n";   
1070    rename ($old_filename, $new_filename)
1071        || print STDERR "Rename failed: $!\n";
1072    return;
1073    }
1074}
1075
1076sub os_dir() {
1077   
1078    my $gsdlarch = "";
1079    if(defined $ENV{'GSDLARCH'}) {
1080    $gsdlarch = $ENV{'GSDLARCH'};
1081    }
1082    return $ENV{'GSDLOS'}.$gsdlarch;
1083}
1084
1085# Returns the greenstone URL prefix extracted from the appropriate GS2/GS3 config file.
1086# By default, /greenstone3 for GS3 or /greenstone for GS2.
1087sub get_greenstone_url_prefix() {
1088    # if already set on a previous occasion, just return that
1089    # (Don't want to keep repeating this: cost of re-opening and scanning files.)
1090    return $ENV{'GREENSTONE_URL_PREFIX'} if($ENV{'GREENSTONE_URL_PREFIX'});
1091
1092    my ($configfile, $urlprefix, $defaultUrlprefix);
1093    my @propertynames = ();
1094
1095    if($ENV{'GSDL3SRCHOME'}) {
1096    $defaultUrlprefix = "/greenstone3";
1097    $configfile = &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'}, "packages", "tomcat", "conf", "Catalina", "localhost", "greenstone3.xml");
1098    push(@propertynames, qw/path\s*\=/);
1099    } else {
1100    $defaultUrlprefix = "/greenstone";
1101    $configfile = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "cgi-bin", &os_dir(), "gsdlsite.cfg");
1102    push(@propertynames, (qw/\nhttpprefix/, qw/\ngwcgi/)); # inspect one property then the other
1103    }
1104
1105    $urlprefix = &extract_propvalue_from_file($configfile, \@propertynames);
1106
1107    if(!$urlprefix) { # no values found for URL prefix, use default values
1108    $urlprefix = $defaultUrlprefix;
1109    } else {
1110    #gwcgi can contain more than the wanted prefix, we split on / to get the first "directory" level
1111    $urlprefix =~ s/^\///; # remove the starting slash
1112    my @dirs = split(/(\\|\/)/, $urlprefix);
1113    $urlprefix = shift(@dirs);
1114
1115    if($urlprefix !~ m/^\//) { # in all cases: ensure the required forward slash is at the front
1116        $urlprefix = "/$urlprefix";
1117    }
1118    }
1119
1120    # set for the future
1121    $ENV{'GREENSTONE_URL_PREFIX'} = $urlprefix;
1122#    print STDERR "*** in get_greenstone_url_prefix(): $urlprefix\n\n";
1123    return $urlprefix;
1124}
1125
1126
1127# Given a config file (xml or java properties file) and a list/array of regular expressions
1128# that represent property names to match on, this function will return the value for the 1st
1129# matching property name. If the return value is undefined, no matching property was found.
1130sub extract_propvalue_from_file() {
1131    my ($configfile, $propertynames) = @_;
1132
1133    my $value;
1134    unless(open(FIN, "<$configfile")) {
1135    print STDERR "extract_propvalue_from_file(): Unable to open $configfile. $!\n";
1136    return $value; # not initialised
1137    }
1138
1139    # Read the entire file at once, as one single line, then close it
1140    my $filecontents;
1141    {
1142    local $/ = undef;       
1143    $filecontents = <FIN>;
1144    }
1145    close(FIN);
1146
1147    foreach my $regex (@$propertynames) {
1148        ($value) = $filecontents=~ m/$regex\s*(\S*)/s; # read value of the property given by regex up to the 1st space
1149    if($value) {
1150            $value =~ s/^\"//;     # remove any startquotes
1151        $value =~ s/\".*$//;   # remove the 1st endquotes (if any) followed by any xml
1152        last;              # found value for a matching property, break from loop
1153    }
1154    }
1155
1156    return $value;
1157}
1158
1159# Subroutine that sources setup.bash, given GSDLHOME and GSDLOS and
1160# given that perllib is in @INC in order to invoke this subroutine.
1161# Call as follows -- after setting up INC to include perllib and
1162# after setting up GSDLHOME and GSDLOS:
1163#
1164# require util;
1165# &util::setup_greenstone_env($ENV{'GSDLHOME'}, $ENV{'GSDLOS'});
1166#
1167sub setup_greenstone_env() {
1168    my ($GSDLHOME, $GSDLOS) = @_;
1169
1170    #my %env_map = ();
1171    # Get the localised ENV settings of running a localised source setup.bash
1172    # and put it into the ENV here. Need to clear GSDLHOME before running setup
1173    #my $perl_command = "(cd $GSDLHOME; export GSDLHOME=; . ./setup.bash > /dev/null; env)";
1174    my $perl_command = "(cd $GSDLHOME; /bin/bash -c \"export GSDLHOME=; source setup.bash > /dev/null; env\")";     
1175    if (($GSDLOS =~ m/windows/i) && ($^O ne "cygwin"))  {
1176        #$perl_command = "cmd /C \"cd $GSDLHOME&& set GSDLHOME=&& setup.bat > nul&& set\"";
1177        $perl_command = "(cd $GSDLHOME&& set GSDLHOME=&& setup.bat > nul&& set)";
1178    }
1179    if (!open(PIN, "$perl_command |")) {
1180        print STDERR ("Unable to execute command: $perl_command. $!\n");
1181    }
1182
1183    while (defined (my $perl_output_line = <PIN>)) {
1184        my($key,$value) = ($perl_output_line =~ m/^([^=]*)[=](.*)$/);
1185        #$env_map{$key}=$value;     
1186        $ENV{$key}=$value;
1187    }
1188    close (PIN);
1189
1190    # If any keys in $ENV don't occur in Greenstone's localised env
1191    # (stored in $env_map), delete those entries from $ENV
1192    #foreach $key (keys %ENV) {
1193    #   if(!defined $env_map{$key}) {
1194    #       print STDOUT "**** DELETING ENV KEY: $key\tVALUE: $ENV{'$key'}\n";
1195    #       delete $ENV{$key}; # del $ENV(key, value) pair
1196    #   }
1197    #}
1198    #undef %env_map;
1199}
1200
1201sub get_perl_exec() {   
1202    my $perl_exec = $^X; # may return just "perl"
1203   
1204    if($ENV{'PERLPATH'}) {
1205        # OR: # $perl_exec = &FileUtils::filenameConcatenate($ENV{'PERLPATH'},"perl");
1206        if (($ENV{'GSDLOS'} =~ m/windows/) && ($^O ne "cygwin")) {
1207            $perl_exec = "$ENV{'PERLPATH'}\\Perl.exe";
1208        } else {
1209            $perl_exec = "$ENV{'PERLPATH'}/perl";
1210        }
1211    } else { # no PERLPATH, use Config{perlpath} else $^X: special variables
1212        # containing the full path to the current perl executable we're using
1213        $perl_exec = $Config{perlpath}; # configured path for perl
1214        if (!-e $perl_exec) { # may not point to location on this machine
1215            $perl_exec = $^X; # may return just "perl"
1216            if($perl_exec =~ m/^perl/i) { # warn if just perl or Perl.exe
1217                print STDERR "**** WARNING: Perl exec found contains no path: $perl_exec";             
1218            }
1219        }
1220    }
1221   
1222    return $perl_exec;
1223}
1224
1225# returns the path to the java command in the JRE included with GS (if any),
1226# quoted to safeguard any spaces in this path, otherwise a simple java
1227# command is returned which assumes and will try for a system java.
1228sub get_java_command {
1229    my $java = "java";
1230    if(defined $ENV{'GSDLHOME'}) { # should be, as this script would be launched from the cmd line
1231                               # after running setup.bat or from GLI which also runs setup.bat
1232    my $java_bin = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"packages","jre","bin");
1233    if(-d $java_bin) {
1234        $java = &FileUtils::filenameConcatenate($java_bin,"java");
1235        $java = "\"".$java."\""; # quoted to preserve spaces in path
1236    }
1237    }
1238    return $java;
1239}
1240
1241
1242# Given the qualified collection name (colgroup/collection),
1243# returns the collection and colgroup parts
1244sub get_collection_parts {
1245    # http://perldoc.perl.org/File/Basename.html
1246    # my($filename, $directories, $suffix) = fileparse($path);
1247    # "$directories contains everything up to and including the last directory separator in the $path
1248    # including the volume (if applicable). The remainder of the $path is the $filename."
1249    #my ($collection, $colgroup) = &File::Basename::fileparse($qualified_collection);   
1250
1251    my $qualified_collection = shift(@_);
1252
1253    # Since activate.pl can be launched from the command-line, including by a user,
1254    # best not to assume colgroup uses URL-style slashes as would be the case with GLI
1255    # Also allow for the accidental inclusion of multiple slashes
1256    my ($colgroup, $collection) = split(/[\/\\]+/, $qualified_collection); #split('/', $qualified_collection);
1257   
1258    if(!defined $collection) {
1259        $collection = $colgroup;
1260        $colgroup = "";
1261    }
1262    return ($collection, $colgroup);
1263}
1264
1265# work out the "collectdir/collection" location
1266sub resolve_collection_dir {
1267    my ($collect_dir, $qualified_collection, $site) = @_; #, $gs_mode
1268   
1269    if (defined $ENV{'GSDLCOLLECTDIR'}) { # a predefined collection dir exists
1270        return $ENV{'GSDLCOLLECTDIR'};
1271    }
1272
1273    my ($colgroup, $collection) = &util::get_collection_parts($qualified_collection);   
1274   
1275    if (!defined $collect_dir || !$collect_dir) { # if undefined or empty string
1276        $collect_dir = &util::get_working_collect_dir($site);
1277    }
1278
1279    return &FileUtils::filenameConcatenate($collect_dir,$colgroup,$collection);
1280}
1281
1282# work out the full path to "collect" of this greenstone 2/3 installation
1283sub get_working_collect_dir {
1284    my ($site) = @_;   
1285
1286    if (defined $ENV{'GSDLCOLLECTHOME'}) { # a predefined collect dir exists
1287    return $ENV{'GSDLCOLLECTHOME'};
1288    }
1289
1290    if (defined $site && $site) { # site non-empty, so get default collect dir for GS3
1291   
1292    if (defined $ENV{'GSDL3HOME'}) {
1293        return &FileUtils::filenameConcatenate($ENV{'GSDL3HOME'},"sites",$site,"collect"); # web folder
1294    }
1295    elsif (defined $ENV{'GSDL3SRCHOME'}) {
1296        return &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'},"web","sites",$site,"collect");
1297    }
1298    }
1299
1300    elsif (defined $ENV{'SITEHOME'}) {
1301    return &FileUtils::filenameConcatenate($ENV{'SITEHOME'},"collect");
1302    }
1303   
1304    else { # get default collect dir for GS2
1305    return &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"collect");
1306    }
1307}
1308
1309sub is_abs_path_any_os {
1310    my ($path) = @_;
1311
1312    # We can have filenames in our DBs that were produced on other OS, so this method exists
1313    # to help identify absolute paths in such cases.
1314
1315    return 1 if($path =~ m@^/@); # full paths begin with forward slash on linux/mac
1316    return 1 if($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
1317
1318    return 0;
1319}
1320
1321
1322# This subroutine is for improving portability of Greenstone collections from one OS to another,
1323# to be used to convert absolute paths going into db files into paths with placeholders instead.
1324# This sub works with util::get_common_gs_paths and takes a path to a greenstone file and, if it's
1325# an absolute path, then it will replace the longest matching greenstone-path prefix of the given
1326# path with a placeholder to match.
1327# The Greenstone-path prefixes that can be matched are the following common Greenstone paths:
1328# the path to the current (specific) collection, the path to the general GS collect directory,
1329# the path to the site directory if GS3, else the path to the GSDLHOME/GSDL3HOME folder.
1330# The longest matching prefix will be replaced with the equivalent placeholder:
1331# @THISCOLLECTPATH@, else @COLLECTHOME@, else @SITEHOME@, else @GSDLHOME@.
1332sub abspath_to_placeholders {
1333    my $path = shift(@_); # path to convert from absolute to one with placeholders
1334    my $opt_long_or_short_winfilenames = shift(@_) || "short"; # whether we want to force use of long file names even on windows, default uses short
1335
1336    return $path unless is_abs_path_any_os($path); # path is relative
1337
1338    if ($opt_long_or_short_winfilenames eq "long") {
1339    $path = &util::upgrade_if_dos_filename($path); # will only do something on windows
1340    }
1341   
1342    # now we know we're dealing with absolute paths and have to replace gs prefixes with placeholders
1343    my @gs_paths = ($ENV{'GSDLCOLLECTDIR'}, $ENV{'GSDLCOLLECTHOME'}, $ENV{'SITEHOME'}, $ENV{'GREENSTONEHOME'}); # list in this order: from longest to shortest path
1344
1345    my %placeholder_map = ($ENV{'GREENSTONEHOME'} => '@GSDLHOME@', # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
1346               $ENV{'GSDLCOLLECTHOME'} => '@COLLECTHOME@',
1347               $ENV{'GSDLCOLLECTDIR'} => '@THISCOLLECTPATH@'
1348    );
1349    $placeholder_map{$ENV{'SITEHOME'}} = '@SITEHOME@' if defined $ENV{'SITEHOME'};
1350
1351    $path = &util::_abspath_to_placeholders($path, \@gs_paths, \%placeholder_map);
1352
1353    if ($ENV{'GSDLOS'} =~ /^windows$/i && $opt_long_or_short_winfilenames eq "short") {
1354    # for windows need to look for matches on short file names too
1355    # matched paths are again to be replaced with the usual placeholders
1356
1357    my $gsdlcollectdir = &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTDIR'});
1358    my $gsdlcollecthome = &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTHOME'});
1359    my $sitehome = (defined $ENV{'SITEHOME'}) ? &util::downgrade_if_dos_filename($ENV{'SITEHOME'}) : undef;
1360    my $greenstonehome =  &util::downgrade_if_dos_filename($ENV{'GREENSTONEHOME'});
1361
1362    @gs_paths = ($gsdlcollectdir, $gsdlcollecthome, $sitehome, $greenstonehome); # order matters
1363
1364    %placeholder_map = ($greenstonehome => '@GSDLHOME@', # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
1365                $gsdlcollecthome => '@COLLECTHOME@',
1366                $gsdlcollectdir => '@THISCOLLECTPATH@'
1367        );
1368    $placeholder_map{$sitehome} = '@SITEHOME@' if defined $sitehome;
1369
1370    $path = &util::_abspath_to_placeholders($path, \@gs_paths, \%placeholder_map);
1371    }
1372
1373    return $path;
1374}
1375
1376sub _abspath_to_placeholders {
1377    my ($path, $gs_paths_ref, $placeholder_map_ref) = @_;
1378
1379    # The sequence of elements in @gs_paths matters   
1380    # Need to loop starting from the *longest* matching path (the path to the specific collection)
1381    # to the shortest matching path (the path to gsdlhome/gsdl3home folder):
1382
1383    foreach my $gs_path (@$gs_paths_ref) {
1384    next if(!defined $gs_path); # site undefined for GS2
1385
1386    my $re_path =  &util::filename_to_regex($gs_path); # escape for regex
1387
1388    if($path =~ m/^$re_path/i) { # case sensitive or not for OS?
1389
1390        my $placeholder = $placeholder_map_ref->{$gs_path}; # get the placeholder to replace the matched path with
1391
1392        $path =~ s/^$re_path/$placeholder/; #case sensitive or not?
1393        #$path =~ s/^[\\\/]//; # remove gs_path's trailing separator left behind at the start of the path
1394        last; # done
1395    }
1396    }
1397   
1398    return $path;
1399}
1400
1401# Function that does the reverse of the util::abspath_to_placeholders subroutine
1402# Once again, call this with the values returned from util::get_common_gs_paths
1403sub placeholders_to_abspath {
1404    my $path = shift(@_); # path that can contain placeholders to convert to resolved absolute path
1405    my $opt_long_or_short_winfilenames = shift(@_) || "short"; # whether we want to force use of long file names even on windows, default uses short
1406
1407    return $path if($path !~ m/@/); # path contains no placeholders
1408   
1409    # replace placeholders with gs prefixes
1410    my @placeholders = ('@THISCOLLECTPATH@', '@COLLECTHOME@', '@SITEHOME@', '@GSDLHOME@'); # order of paths not crucial in this case,
1411                       # but listed here from longest to shortest once placeholders are have been resolved
1412
1413    # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
1414    my %placeholder_to_gspath_map;
1415    if ($ENV{'GSDLOS'} =~ /^windows$/i && $opt_long_or_short_winfilenames eq "short") {
1416    # always replace placeholders with short file names of the absolute paths on windows?
1417    %placeholder_to_gspath_map = ('@GSDLHOME@' => &util::downgrade_if_dos_filename($ENV{'GREENSTONEHOME'}),
1418                     '@COLLECTHOME@' => &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTHOME'}),
1419                     '@THISCOLLECTPATH@' => &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTDIR'})
1420    );
1421    $placeholder_to_gspath_map{'@SITEHOME@'} =  &util::downgrade_if_dos_filename($ENV{'SITEHOME'}) if defined $ENV{'SITEHOME'};
1422    } else {
1423    %placeholder_to_gspath_map = ('@GSDLHOME@' => $ENV{'GREENSTONEHOME'},
1424                      '@SITEHOME@' => $ENV{'SITEHOME'}, # can be undef
1425                      '@COLLECTHOME@' => $ENV{'GSDLCOLLECTHOME'},
1426                      '@THISCOLLECTPATH@' => $ENV{'GSDLCOLLECTDIR'}
1427        ); # $placeholder_to_gspath_map{'@SITEHOME@'} = $ENV{'SITEHOME'} if defined $ENV{'SITEHOME'};
1428    }
1429
1430    foreach my $placeholder (@placeholders) {
1431    my $gs_path = $placeholder_to_gspath_map{$placeholder};
1432
1433    next if(!defined $gs_path); # sitehome for GS2 is undefined
1434
1435    if($path =~ m/^$placeholder/) {
1436        $path =~ s/^$placeholder/$gs_path/;
1437        last; # done
1438    }
1439    }
1440   
1441    return $path;
1442}
1443
1444# Used by pdfpstoimg.pl and PDFBoxConverter to create a .item file from
1445# a directory containing sequentially numbered images.
1446sub create_itemfile
1447{
1448    my ($output_dir, $convert_basename, $convert_to) = @_;
1449    my $page_num = "";
1450
1451    opendir(DIR, $output_dir) || die "can't opendir $output_dir: $!";
1452    my @dir_files = grep {-f "$output_dir/$_"} readdir(DIR);
1453    closedir DIR;
1454
1455    # Sort files in the directory by page_num
1456    sub page_number {
1457    my ($dir) = @_;
1458    my ($pagenum) =($dir =~ m/^.*?[-\.]?(\d+)(\.(jpg|gif|png))?$/i);
1459#   my ($pagenum) =($dir =~ m/(\d+)(\.(jpg|gif|png))?$/i); # this works but is not as safe/strict about input filepatterns as the above
1460
1461    $pagenum = 1 unless defined $pagenum;
1462    return $pagenum;
1463    }
1464
1465    # sort the files in the directory in the order of page_num rather than lexically.
1466    @dir_files = sort { page_number($a) <=> page_number($b) } @dir_files;
1467
1468    # work out if the numbering of the now sorted image files starts at 0 or not
1469    # by checking the number of the first _image_ file (skipping item files)
1470    my $starts_at_0 = 0;
1471    my $firstfile = ($dir_files[0] !~ /\.item$/i) ? $dir_files[0] : $dir_files[1];
1472    if(page_number($firstfile) == 0) { # 00 will evaluate to 0 too in this condition
1473    $starts_at_0 = 1;
1474    }
1475
1476    my $item_file = &FileUtils::filenameConcatenate($output_dir, $convert_basename.".item");
1477    my $item_fh;
1478    &FileUtils::openFileHandle($item_file, 'w', \$item_fh);
1479    print $item_fh "<PagedDocument>\n";
1480
1481    foreach my $file (@dir_files){
1482    if ($file !~ /\.item/i){
1483        $page_num = page_number($file);
1484        $page_num++ if $starts_at_0; # image numbers start at 0, so add 1
1485        print $item_fh "   <Page pagenum=\"$page_num\" imgfile=\"$file\" txtfile=\"\"/>\n";
1486    }
1487    }
1488
1489    print $item_fh "</PagedDocument>\n";
1490    &FileUtils::closeFileHandle($item_file, \$item_fh);
1491    return $item_file;
1492}
1493
1494
1495## @function augmentINC()
1496#
1497#  Prepend a path (if it exists) onto INC but only if it isn't already in INC
1498#  @param $new_path The path to add
1499#  @author jmt12
1500#
1501sub augmentINC
1502{
1503  my ($new_path) = @_;
1504  my $did_add_path = 0;
1505  # might need to be replaced with FileUtils::directoryExists() call eventually
1506  if (-d $new_path)
1507  {
1508    my $did_find_path = 0;
1509    foreach my $existing_path (@INC)
1510    {
1511      if ($existing_path eq $new_path)
1512      {
1513        $did_find_path = 1;
1514        last;
1515      }
1516    }
1517    if (!$did_find_path)
1518    {
1519      unshift(@INC, $new_path);
1520      $did_add_path = 1;
1521    }
1522  }
1523  return $did_add_path;
1524}
1525## augmentINC()
1526
1527
15281;
Note: See TracBrowser for help on using the browser.