[537] | 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 | ###########################################################################
|
---|
[4] | 25 |
|
---|
| 26 | package util;
|
---|
| 27 |
|
---|
[23362] | 28 | use strict;
|
---|
[31863] | 29 | no strict 'refs'; # make an exception so we can use variables as filehandles
|
---|
[27509] | 30 | use FileUtils;
|
---|
[23362] | 31 |
|
---|
| 32 | use Encode;
|
---|
[29807] | 33 | use Unicode::Normalize 'normalize';
|
---|
| 34 |
|
---|
[4] | 35 | use File::Copy;
|
---|
[619] | 36 | use File::Basename;
|
---|
[24362] | 37 | # Config for getting the perlpath in the recommended way, though it uses paths that are
|
---|
| 38 | # hard-coded into the Config file that's generated upon configuring and compiling perl.
|
---|
| 39 | # $^X works better in some cases to return the path to perl used to launch the script,
|
---|
| 40 | # but if launched with plain "perl" (no full-path), that will be just what it returns.
|
---|
[27303] | 41 | use Config;
|
---|
| 42 | # New module for file related utility functions - intended as a
|
---|
| 43 | # placeholder for an extension that allows a variety of different
|
---|
| 44 | # filesystems (FTP, HTTP, SAMBA, WEBDav, HDFS etc)
|
---|
| 45 | use FileUtils;
|
---|
[4] | 46 |
|
---|
[28226] | 47 | if ($ENV{'GSDLOS'} =~ /^windows$/i) {
|
---|
[28225] | 48 | require Win32; # for working out Windows Long Filenames from Win 8.3 short filenames
|
---|
| 49 | }
|
---|
| 50 |
|
---|
[4] | 51 | # removes files (but not directories)
|
---|
| 52 | sub rm {
|
---|
[27303] | 53 | warnings::warnif("deprecated", "util::rm() is deprecated, using FileUtils::removeFiles() instead");
|
---|
| 54 | return &FileUtils::removeFiles(@_);
|
---|
[4] | 55 | }
|
---|
| 56 |
|
---|
| 57 | # recursive removal
|
---|
[10211] | 58 | sub filtered_rm_r {
|
---|
[27303] | 59 | warnings::warnif("deprecated", "util::filtered_rm_r() is deprecated, using FileUtils::removeFilesFiltered() instead");
|
---|
| 60 | return &FileUtils::removeFilesFiltered(@_);
|
---|
[4] | 61 | }
|
---|
| 62 |
|
---|
[10211] | 63 | # recursive removal
|
---|
| 64 | sub rm_r {
|
---|
[27326] | 65 | warnings::warnif("deprecated", "util::rm_r() is deprecated, using FileUtils::removeFilesRecursive() instead");
|
---|
[27303] | 66 | return &FileUtils::removeFilesRecursive(@_);
|
---|
[10211] | 67 | }
|
---|
| 68 |
|
---|
[721] | 69 | # moves a file or a group of files
|
---|
| 70 | sub mv {
|
---|
[27303] | 71 | warnings::warnif("deprecated", "util::mv() is deprecated, using FileUtils::moveFiles() instead");
|
---|
| 72 | return &FileUtils::moveFiles(@_);
|
---|
[721] | 73 | }
|
---|
| 74 |
|
---|
[25554] | 75 | # Move the contents of source directory into target directory
|
---|
| 76 | # (as opposed to merely replacing target dir with the src dir)
|
---|
| 77 | # This can overwrite any files with duplicate names in the target
|
---|
| 78 | # but other files and folders in the target will continue to exist
|
---|
| 79 | sub mv_dir_contents {
|
---|
[27303] | 80 | warnings::warnif("deprecated", "util::mv_dir_contents() is deprecated, using FileUtils::moveDirectoryContents() instead");
|
---|
| 81 | return &FileUtils::moveDirectoryContents(@_);
|
---|
[25554] | 82 | }
|
---|
[721] | 83 |
|
---|
[4] | 84 | # copies a file or a group of files
|
---|
| 85 | sub cp {
|
---|
[27303] | 86 | warnings::warnif("deprecated", "util::cp() is deprecated, using FileUtils::copyFiles() instead");
|
---|
| 87 | return &FileUtils::copyFiles(@_);
|
---|
[4] | 88 | }
|
---|
| 89 |
|
---|
| 90 | # recursively copies a file or group of files
|
---|
[1454] | 91 | # syntax: cp_r (sourcefiles, destination directory)
|
---|
| 92 | # destination must be a directory - to copy one file to
|
---|
| 93 | # another use cp instead
|
---|
[4] | 94 | sub cp_r {
|
---|
[27303] | 95 | warnings::warnif("deprecated", "util::cp_r() is deprecated, using FileUtils::copyFilesrecursive() instead");
|
---|
| 96 | return &FileUtils::copyFilesRecursive(@_);
|
---|
| 97 | }
|
---|
[4] | 98 |
|
---|
[21762] | 99 | # recursively copies a file or group of files
|
---|
| 100 | # syntax: cp_r (sourcefiles, destination directory)
|
---|
| 101 | # destination must be a directory - to copy one file to
|
---|
| 102 | # another use cp instead
|
---|
| 103 | sub cp_r_nosvn {
|
---|
[27303] | 104 | warnings::warnif("deprecated", "util::cp_r_nosvn() is deprecated, using FileUtils::copyFilesRecursiveNoSVN() instead");
|
---|
| 105 | return &FileUtils::copyFilesRecursiveNoSVN(@_);
|
---|
[21762] | 106 | }
|
---|
| 107 |
|
---|
[11179] | 108 | # copies a directory and its contents, excluding subdirectories, into a new directory
|
---|
| 109 | sub cp_r_toplevel {
|
---|
[27303] | 110 | warnings::warnif("deprecated", "util::cp_r_toplevel() is deprecated, using FileUtils::recursiveCopyTopLevel() instead");
|
---|
| 111 | return &FileUtils::recursiveCopyTopLevel(@_);
|
---|
[11179] | 112 | }
|
---|
| 113 |
|
---|
[721] | 114 | sub mk_dir {
|
---|
[27303] | 115 | warnings::warnif("deprecated", "util::mk_dir() is deprecated, using FileUtils::makeDirectory() instead");
|
---|
| 116 | return &FileUtils::makeDirectory(@_);
|
---|
[721] | 117 | }
|
---|
| 118 |
|
---|
[1046] | 119 | # in case anyone cares - I did some testing (using perls Benchmark module)
|
---|
| 120 | # on this subroutine against File::Path::mkpath (). mk_all_dir() is apparently
|
---|
| 121 | # slightly faster (surprisingly) - Stefan.
|
---|
[4] | 122 | sub mk_all_dir {
|
---|
[27303] | 123 | warnings::warnif("deprecated", "util::mk_all_dir() is deprecated, using FileUtils::makeAllDirectories() instead");
|
---|
| 124 | return &FileUtils::makeAllDirectories(@_);
|
---|
[4] | 125 | }
|
---|
| 126 |
|
---|
[619] | 127 | # make hard link to file if supported by OS, otherwise copy the file
|
---|
| 128 | sub hard_link {
|
---|
[27303] | 129 | warnings::warnif("deprecated", "util::hard_link() is deprecated, using FileUtils::hardLink() instead");
|
---|
| 130 | return &FileUtils::hardLink(@_);
|
---|
[619] | 131 | }
|
---|
| 132 |
|
---|
[2193] | 133 | # make soft link to file if supported by OS, otherwise copy file
|
---|
[721] | 134 | sub soft_link {
|
---|
[27303] | 135 | warnings::warnif("deprecated", "util::soft_link() is deprecated, using FileUtils::softLink() instead");
|
---|
| 136 | return &FileUtils::softLink(@_);
|
---|
[721] | 137 | }
|
---|
| 138 |
|
---|
[23362] | 139 | # Primarily for filenames generated by processing
|
---|
| 140 | # content of HTML files (which are mapped to UTF-8 internally)
|
---|
| 141 | #
|
---|
| 142 | # To turn this into an octet string that really exists on the file
|
---|
| 143 | # system:
|
---|
| 144 | # 1. don't need to do anything special for Unix-based systems
|
---|
| 145 | # (as underlying file system is byte-code)
|
---|
| 146 | # 2. need to map to short DOS filenames for Windows
|
---|
[721] | 147 |
|
---|
[23362] | 148 | sub utf8_to_real_filename
|
---|
| 149 | {
|
---|
| 150 | my ($utf8_filename) = @_;
|
---|
[721] | 151 |
|
---|
[23362] | 152 | my $real_filename;
|
---|
[721] | 153 |
|
---|
[28375] | 154 | if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
|
---|
[23362] | 155 | require Win32;
|
---|
[23388] | 156 |
|
---|
[23362] | 157 | my $unicode_filename = decode("utf8",$utf8_filename);
|
---|
| 158 | $real_filename = Win32::GetShortPathName($unicode_filename);
|
---|
| 159 | }
|
---|
| 160 | else {
|
---|
| 161 | $real_filename = $utf8_filename;
|
---|
| 162 | }
|
---|
| 163 |
|
---|
| 164 | return $real_filename;
|
---|
| 165 | }
|
---|
| 166 |
|
---|
[29794] | 167 | sub raw_filename_to_unicode
|
---|
| 168 | {
|
---|
[29810] | 169 | my ($directory, $raw_file, $filename_encoding ) = @_;
|
---|
[29794] | 170 |
|
---|
[29810] | 171 | my $unicode_filename = $raw_file;
|
---|
| 172 | if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
|
---|
[29816] | 173 | # Try turning a short version to the long version
|
---|
| 174 | # 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
|
---|
| 175 | $unicode_filename = &util::get_dirsep_tail(&util::upgrade_if_dos_filename(&FileUtils::filenameConcatenate($directory, $raw_file), 0));
|
---|
| 176 |
|
---|
| 177 |
|
---|
| 178 | if ($unicode_filename eq $raw_file) {
|
---|
| 179 | # 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
|
---|
[29810] | 180 |
|
---|
[29816] | 181 | $unicode_filename = &Encode::decode(locale_fs => $unicode_filename);
|
---|
| 182 | }
|
---|
| 183 | # 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.
|
---|
| 184 |
|
---|
[29810] | 185 |
|
---|
| 186 | } else {
|
---|
[29816] | 187 | # we had a utf-8 string, turn it into perl internal unicode
|
---|
| 188 | $unicode_filename = &Encode::decode("utf-8", $unicode_filename);
|
---|
[29810] | 189 |
|
---|
| 190 |
|
---|
[29816] | 191 | }
|
---|
| 192 | #Does the filename have url encoded chars in it?
|
---|
| 193 | if (&unicode::is_url_encoded($unicode_filename)) {
|
---|
| 194 | $unicode_filename = &unicode::url_decode($unicode_filename);
|
---|
| 195 | }
|
---|
| 196 |
|
---|
[29794] | 197 | # Normalise the filename to canonical composition - on mac, filenames use decopmposed form for accented chars
|
---|
| 198 | if ($ENV{'GSDLOS'} =~ m/^darwin$/i) {
|
---|
[29816] | 199 | $unicode_filename = normalize('C', $unicode_filename); # Composed form 'C'
|
---|
| 200 | }
|
---|
[29794] | 201 | return $unicode_filename;
|
---|
| 202 |
|
---|
| 203 | }
|
---|
[27303] | 204 | sub fd_exists {
|
---|
| 205 | warnings::warnif("deprecated", "util::fd_exists() is deprecated, using FileUtils::fileTest() instead");
|
---|
| 206 | return &FileUtils::fileTest(@_);
|
---|
[23362] | 207 | }
|
---|
| 208 |
|
---|
[27303] | 209 | sub file_exists {
|
---|
| 210 | warnings::warnif("deprecated", "util::file_exists() is deprecated, using FileUtils::fileExists() instead");
|
---|
| 211 | return &FileUtils::fileExists(@_);
|
---|
[23362] | 212 | }
|
---|
| 213 |
|
---|
[27303] | 214 | sub dir_exists {
|
---|
| 215 | warnings::warnif("deprecated", "util::dir_exists() is deprecated, using FileUtils::directoryExists() instead");
|
---|
| 216 | return &FileUtils::directoryExists(@_);
|
---|
[23362] | 217 | }
|
---|
| 218 |
|
---|
[4] | 219 | # updates a copy of a directory in some other part of the filesystem
|
---|
| 220 | # verbosity settings are: 0=low, 1=normal, 2=high
|
---|
| 221 | # both $fromdir and $todir should be absolute paths
|
---|
| 222 | sub cachedir {
|
---|
[27303] | 223 | warnings::warnif("deprecated", "util::cachedir() is deprecated, using FileUtils::synchronizeDirectories() instead");
|
---|
| 224 | return &FileUtils::synchronizeDirectories(@_);
|
---|
[4] | 225 | }
|
---|
| 226 |
|
---|
| 227 | # this function returns -1 if either file is not found
|
---|
| 228 | # assumes that $file1 and $file2 are absolute file names or
|
---|
| 229 | # in the current directory
|
---|
| 230 | # $file2 is allowed to be newer than $file1
|
---|
| 231 | sub differentfiles {
|
---|
[27303] | 232 | warnings::warnif("deprecated", "util::differentfiles() is deprecated, using FileUtils::differentFiles() instead");
|
---|
| 233 | return &FileUtils::differentFiles(@_);
|
---|
[4] | 234 | }
|
---|
| 235 |
|
---|
| 236 |
|
---|
[30574] | 237 | # works out the temporary directory, including in the case where Greenstone is not writable
|
---|
| 238 | # In that case, gs3-setup.bat would already have set the GS_TMP_OUTPUT_DIR temp variable
|
---|
| 239 | sub determine_tmp_dir
|
---|
| 240 | {
|
---|
| 241 | my $try_collect_dir = shift(@_) || 0;
|
---|
| 242 |
|
---|
| 243 | my $tmp_dirname;
|
---|
| 244 | if(defined $ENV{'GS_TMP_OUTPUT_DIR'}) {
|
---|
| 245 | $tmp_dirname = $ENV{'GS_TMP_OUTPUT_DIR'};
|
---|
| 246 | } elsif($try_collect_dir && defined $ENV{'GSDLCOLLECTDIR'}) {
|
---|
| 247 | $tmp_dirname = $ENV{'GSDLCOLLECTDIR'};
|
---|
| 248 | } elsif(defined $ENV{'GSDLHOME'}) {
|
---|
| 249 | $tmp_dirname = $ENV{'GSDLHOME'};
|
---|
| 250 | } else {
|
---|
| 251 | return undef;
|
---|
| 252 | }
|
---|
| 253 |
|
---|
| 254 | if(!defined $ENV{'GS_TMP_OUTPUT_DIR'}) {
|
---|
| 255 | # test the tmp_dirname folder is writable, by trying to write out a file
|
---|
| 256 | # Unfortunately, cound not get if(-w $dirname) to work on directories on Windows
|
---|
| 257 | ## http://alvinalexander.com/blog/post/perl/perl-file-test-operators-reference-cheat-sheet (test file/dir writable)
|
---|
| 258 | ## http://www.makelinux.net/alp/083 (real and effective user IDs)
|
---|
| 259 |
|
---|
| 260 | my $tmp_test_file = &FileUtils::filenameConcatenate($tmp_dirname, "writability_test.tmp");
|
---|
[30575] | 261 | if (open (FOUT, ">$tmp_test_file")) {
|
---|
[30574] | 262 | close(FOUT);
|
---|
| 263 | &FileUtils::removeFiles($tmp_test_file);
|
---|
| 264 | } else { # location not writable, use TMP location
|
---|
[31694] | 265 | if (defined $ENV{'TMP'}) {
|
---|
| 266 | $tmp_dirname = $ENV{'TMP'};
|
---|
| 267 | } else {
|
---|
| 268 | $tmp_dirname = "/tmp";
|
---|
| 269 | }
|
---|
| 270 | $tmp_dirname = &FileUtils::filenameConcatenate($tmp_dirname, "greenstone");
|
---|
[30574] | 271 | $ENV{'GS_TMP_OUTPUT_DIR'} = $tmp_dirname; # store for next time
|
---|
| 272 | }
|
---|
| 273 | }
|
---|
| 274 |
|
---|
| 275 | $tmp_dirname = &FileUtils::filenameConcatenate($tmp_dirname, "tmp");
|
---|
| 276 | &FileUtils::makeAllDirectories ($tmp_dirname) unless -e $tmp_dirname;
|
---|
| 277 |
|
---|
| 278 | return $tmp_dirname;
|
---|
| 279 | }
|
---|
| 280 |
|
---|
[16266] | 281 | sub get_tmp_filename
|
---|
| 282 | {
|
---|
| 283 | my $file_ext = shift(@_) || undef;
|
---|
| 284 |
|
---|
[22438] | 285 | my $opt_dot_file_ext = "";
|
---|
| 286 | if (defined $file_ext) {
|
---|
| 287 | if ($file_ext !~ m/\./) {
|
---|
| 288 | # no dot, so needs one added in at start
|
---|
| 289 | $opt_dot_file_ext = ".$file_ext"
|
---|
| 290 | }
|
---|
| 291 | else {
|
---|
| 292 | # allow for "extensions" such as _metadata.txt to be handled
|
---|
| 293 | # gracefully
|
---|
| 294 | $opt_dot_file_ext = $file_ext;
|
---|
| 295 | }
|
---|
| 296 | }
|
---|
[16266] | 297 |
|
---|
[31514] | 298 | my $tmpdir = &util::determine_tmp_dir(0);
|
---|
[4] | 299 |
|
---|
| 300 | my $count = 1000;
|
---|
| 301 | my $rand = int(rand $count);
|
---|
[27303] | 302 | my $full_tmp_filename = &FileUtils::filenameConcatenate($tmpdir, "F$rand$opt_dot_file_ext");
|
---|
[16266] | 303 |
|
---|
| 304 | while (-e $full_tmp_filename) {
|
---|
[4] | 305 | $rand = int(rand $count);
|
---|
[27303] | 306 | $full_tmp_filename = &FileUtils::filenameConcatenate($tmpdir, "F$rand$opt_dot_file_ext");
|
---|
[4] | 307 | $count++;
|
---|
| 308 | }
|
---|
[16266] | 309 |
|
---|
| 310 | return $full_tmp_filename;
|
---|
[4] | 311 | }
|
---|
| 312 |
|
---|
[28066] | 313 | # These 2 are "static" variables used by the get_timestamped_tmp_folder() subroutine below and
|
---|
| 314 | # belong with that function. They help ensure the timestamped tmp folders generated are unique.
|
---|
| 315 | my $previous_timestamp = undef;
|
---|
| 316 | my $previous_timestamp_f = 0; # frequency
|
---|
| 317 |
|
---|
[22886] | 318 | sub get_timestamped_tmp_folder
|
---|
[22873] | 319 | {
|
---|
[30574] | 320 | my $tmp_dirname = &util::determine_tmp_dir(1);
|
---|
[28066] | 321 |
|
---|
[22873] | 322 | # add the timestamp into the path otherwise we can run into problems
|
---|
| 323 | # if documents have the same name
|
---|
[28066] | 324 | my $timestamp = time;
|
---|
| 325 |
|
---|
| 326 | if (!defined $previous_timestamp || ($timestamp > $previous_timestamp)) {
|
---|
| 327 | $previous_timestamp_f = 0;
|
---|
| 328 | $previous_timestamp = $timestamp;
|
---|
| 329 | } else {
|
---|
| 330 | $previous_timestamp_f++;
|
---|
| 331 | }
|
---|
| 332 |
|
---|
[27303] | 333 | my $time_tmp_dirname = &FileUtils::filenameConcatenate($tmp_dirname, $timestamp);
|
---|
[28066] | 334 | $tmp_dirname = $time_tmp_dirname;
|
---|
| 335 | my $i = $previous_timestamp_f;
|
---|
| 336 |
|
---|
| 337 | if($previous_timestamp_f > 0) {
|
---|
| 338 | $tmp_dirname = $time_tmp_dirname."_".$i;
|
---|
| 339 | $i++;
|
---|
| 340 | }
|
---|
[22873] | 341 | while (-e $tmp_dirname) {
|
---|
[28066] | 342 | $tmp_dirname = $time_tmp_dirname."_".$i;
|
---|
[22873] | 343 | $i++;
|
---|
| 344 | }
|
---|
[28066] | 345 | &FileUtils::makeDirectory($tmp_dirname);
|
---|
| 346 |
|
---|
[22886] | 347 | return $tmp_dirname;
|
---|
| 348 | }
|
---|
[22873] | 349 |
|
---|
[22886] | 350 | sub get_timestamped_tmp_filename_in_collection
|
---|
| 351 | {
|
---|
| 352 |
|
---|
| 353 | my ($input_filename, $output_ext) = @_;
|
---|
| 354 | # derive tmp filename from input filename
|
---|
| 355 | my ($tailname, $dirname, $suffix)
|
---|
| 356 | = &File::Basename::fileparse($input_filename, "\\.[^\\.]+\$");
|
---|
| 357 |
|
---|
| 358 | # softlink to collection tmp dir
|
---|
| 359 | my $tmp_dirname = &util::get_timestamped_tmp_folder();
|
---|
| 360 | $tmp_dirname = $dirname unless defined $tmp_dirname;
|
---|
| 361 |
|
---|
[22873] | 362 | # following two steps copied from ConvertBinaryFile
|
---|
[22886] | 363 | # do we need them?? can't use them as is, as they use plugin methods.
|
---|
| 364 |
|
---|
[22873] | 365 | #$tailname = $self->SUPER::filepath_to_utf8($tailname) unless &unicode::check_is_utf8($tailname);
|
---|
| 366 |
|
---|
| 367 | # URLEncode this since htmls with images where the html filename is utf8 don't seem
|
---|
| 368 | # to work on Windows (IE or Firefox), as browsers are looking for filesystem-encoded
|
---|
| 369 | # files on the filesystem.
|
---|
| 370 | #$tailname = &util::rename_file($tailname, $self->{'file_rename_method'}, "without_suffix");
|
---|
| 371 | if (defined $output_ext) {
|
---|
| 372 | $output_ext = ".$output_ext"; # add the dot
|
---|
| 373 | } else {
|
---|
| 374 | $output_ext = $suffix;
|
---|
| 375 | }
|
---|
| 376 | $output_ext= lc($output_ext);
|
---|
[27303] | 377 | my $tmp_filename = &FileUtils::filenameConcatenate($tmp_dirname, "$tailname$output_ext");
|
---|
[22873] | 378 |
|
---|
| 379 | return $tmp_filename;
|
---|
| 380 | }
|
---|
| 381 |
|
---|
[21218] | 382 | sub get_toplevel_tmp_dir
|
---|
| 383 | {
|
---|
[27303] | 384 | return &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "tmp");
|
---|
[21218] | 385 | }
|
---|
| 386 |
|
---|
| 387 |
|
---|
[28460] | 388 | sub get_collectlevel_tmp_dir
|
---|
| 389 | {
|
---|
| 390 | my $tmp_dirname = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'}, "tmp");
|
---|
| 391 | &FileUtils::makeDirectory($tmp_dirname) if (!-e $tmp_dirname);
|
---|
| 392 |
|
---|
| 393 | return $tmp_dirname;
|
---|
| 394 | }
|
---|
| 395 |
|
---|
[31513] | 396 | sub get_parent_folder
|
---|
| 397 | {
|
---|
| 398 | my ($path) = @_;
|
---|
| 399 | my ($tailname, $dirname, $suffix)
|
---|
| 400 | = &File::Basename::fileparse($path, "\\.[^\\.]+\$");
|
---|
| 401 |
|
---|
| 402 | return &FileUtils::sanitizePath($dirname);
|
---|
| 403 | }
|
---|
| 404 |
|
---|
[17512] | 405 | sub filename_to_regex {
|
---|
| 406 | my $filename = shift (@_);
|
---|
[4] | 407 |
|
---|
[24971] | 408 | # need to make single backslashes double so that regex works
|
---|
[24832] | 409 | $filename =~ s/\\/\\\\/g; # if ($ENV{'GSDLOS'} =~ /^windows$/i);
|
---|
[24829] | 410 |
|
---|
[24832] | 411 | # note that the first part of a substitution is a regex, so RE chars need to be escaped,
|
---|
| 412 | # the second part of a substitution is not a regex, so for e.g. full-stop can be specified literally
|
---|
[24829] | 413 | $filename =~ s/\./\\./g; # in case there are extensions/other full stops, escape them
|
---|
| 414 | $filename =~ s@\(@\\(@g; # escape brackets
|
---|
| 415 | $filename =~ s@\)@\\)@g; # escape brackets
|
---|
[24932] | 416 | $filename =~ s@\[@\\[@g; # escape brackets
|
---|
| 417 | $filename =~ s@\]@\\]@g; # escape brackets
|
---|
[24829] | 418 |
|
---|
[17512] | 419 | return $filename;
|
---|
| 420 | }
|
---|
| 421 |
|
---|
[24829] | 422 | sub unregex_filename {
|
---|
| 423 | my $filename = shift (@_);
|
---|
| 424 |
|
---|
| 425 | # need to put doubled backslashes for regex back to single
|
---|
| 426 | $filename =~ s/\\\./\./g; # remove RE syntax for .
|
---|
| 427 | $filename =~ s@\\\(@(@g; # remove RE syntax for ( => "\(" turns into "("
|
---|
| 428 | $filename =~ s@\\\)@)@g; # remove RE syntax for ) => "\)" turns into ")"
|
---|
[24932] | 429 | $filename =~ s@\\\[@[@g; # remove RE syntax for [ => "\[" turns into "["
|
---|
| 430 | $filename =~ s@\\\]@]@g; # remove RE syntax for ] => "\]" turns into "]"
|
---|
[24940] | 431 |
|
---|
| 432 | # \\ goes to \
|
---|
| 433 | # This is the last step in reverse mirroring the order of steps in filename_to_regex()
|
---|
| 434 | $filename =~ s/\\\\/\\/g; # remove RE syntax for \
|
---|
[24829] | 435 | return $filename;
|
---|
| 436 | }
|
---|
| 437 |
|
---|
[4] | 438 | sub filename_cat {
|
---|
[27303] | 439 | # I've disabled this warning for now, as every Greenstone perl
|
---|
| 440 | # script seems to make use of this function and so you drown in a
|
---|
| 441 | # sea of deprecated warnings [jmt12]
|
---|
| 442 | # warnings::warnif("deprecated", "util::filename_cat() is deprecated, using FileUtils::filenameConcatenate() instead");
|
---|
| 443 | return &FileUtils::filenameConcatenate(@_);
|
---|
[4] | 444 | }
|
---|
| 445 |
|
---|
[21413] | 446 |
|
---|
[28394] | 447 | sub _pathname_cat {
|
---|
| 448 | my $join_char = shift(@_);
|
---|
| 449 | my $first_path = shift(@_);
|
---|
[21413] | 450 | my (@pathnames) = @_;
|
---|
| 451 |
|
---|
| 452 | # If first_path is not null or empty, then add it back into the list
|
---|
| 453 | if (defined $first_path && $first_path =~ /\S/) {
|
---|
| 454 | unshift(@pathnames, $first_path);
|
---|
| 455 | }
|
---|
| 456 |
|
---|
| 457 | my $pathname = join($join_char, @pathnames);
|
---|
| 458 |
|
---|
| 459 | # remove duplicate slashes
|
---|
[28394] | 460 | if ($join_char eq ";") {
|
---|
[21413] | 461 | $pathname =~ s/[\\\/]+/\\/g;
|
---|
[28460] | 462 | if ($^O eq "cygwin") {
|
---|
| 463 | # Once we've collapsed muliple (potentialy > 2) slashes
|
---|
| 464 | # For cygwin, actually want things double-backslahed
|
---|
| 465 | $pathname =~ s/\\/\\\\/g;
|
---|
| 466 | }
|
---|
| 467 |
|
---|
[21413] | 468 | } else {
|
---|
| 469 | $pathname =~ s/[\/]+/\//g;
|
---|
| 470 | # DB: want a pathname abc\de.html to remain like this
|
---|
| 471 | }
|
---|
| 472 |
|
---|
| 473 | return $pathname;
|
---|
| 474 | }
|
---|
| 475 |
|
---|
| 476 |
|
---|
[28394] | 477 | sub pathname_cat {
|
---|
| 478 | my (@pathnames) = @_;
|
---|
| 479 |
|
---|
| 480 | my $join_char;
|
---|
| 481 | if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
|
---|
| 482 | $join_char = ";";
|
---|
| 483 | } else {
|
---|
| 484 | $join_char = ":";
|
---|
| 485 | }
|
---|
| 486 | return _pathname_cat($join_char,@pathnames);
|
---|
| 487 | }
|
---|
| 488 |
|
---|
| 489 |
|
---|
| 490 | sub javapathname_cat {
|
---|
| 491 | my (@pathnames) = @_;
|
---|
| 492 |
|
---|
| 493 | my $join_char;
|
---|
| 494 |
|
---|
| 495 | # Unlike pathname_cat() above, not interested if running in a Cygwin environment
|
---|
| 496 | # This is because the java we run is actually a native Windows executable
|
---|
| 497 |
|
---|
| 498 | if (($ENV{'GSDLOS'} =~ /^windows$/i)) {
|
---|
| 499 | $join_char = ";";
|
---|
| 500 | } else {
|
---|
| 501 | $join_char = ":";
|
---|
| 502 | }
|
---|
| 503 | return _pathname_cat($join_char,@pathnames);
|
---|
| 504 | }
|
---|
| 505 |
|
---|
| 506 |
|
---|
[28460] | 507 | sub makeFilenameJavaCygwinCompatible
|
---|
| 508 | {
|
---|
| 509 | my ($java_filename) = @_;
|
---|
[28394] | 510 |
|
---|
[28460] | 511 | if ($^O eq "cygwin") {
|
---|
| 512 | # To be used with a Java program, but under Cygwin
|
---|
| 513 | # Because the java binary that is native to Windows, need to
|
---|
| 514 | # convert the Cygwin paths (i.e. Unix style) to be Windows
|
---|
| 515 | # compatible
|
---|
| 516 |
|
---|
| 517 | $java_filename = `cygpath -wp "$java_filename"`;
|
---|
| 518 | chomp($java_filename);
|
---|
| 519 | $java_filename =~ s%\\%\\\\%g;
|
---|
| 520 | }
|
---|
| 521 |
|
---|
| 522 | return $java_filename;
|
---|
| 523 | }
|
---|
| 524 |
|
---|
[19616] | 525 | sub tidy_up_oid {
|
---|
| 526 | my ($OID) = @_;
|
---|
[29112] | 527 | if ($OID =~ /[\.\/\\]/) {
|
---|
| 528 | print STDERR "Warning, identifier $OID contains periods or slashes(.\\\/), replacing them with _\n";
|
---|
| 529 | $OID =~ s/[\.\\\/]/_/g; #remove any periods
|
---|
[19616] | 530 | }
|
---|
| 531 | if ($OID =~ /^\s.*\s$/) {
|
---|
| 532 | print STDERR "Warning, identifier $OID starts or ends with whitespace. Removing it\n";
|
---|
| 533 | # remove starting and trailing whitespace
|
---|
| 534 | $OID =~ s/^\s+//;
|
---|
| 535 | $OID =~ s/\s+$//;
|
---|
| 536 | }
|
---|
| 537 | if ($OID =~ /^[\d]*$/) {
|
---|
| 538 | print STDERR "Warning, identifier $OID contains only digits. Prepending 'D'.\n";
|
---|
| 539 | $OID = "D" . $OID;
|
---|
| 540 | }
|
---|
| 541 |
|
---|
| 542 | return $OID;
|
---|
| 543 | }
|
---|
[26206] | 544 |
|
---|
[10212] | 545 | sub envvar_prepend {
|
---|
| 546 | my ($var,$val) = @_;
|
---|
| 547 |
|
---|
[26206] | 548 | # 64 bit linux can't handle ";" as path separator, so make sure to set this to the right one for the OS
|
---|
[28375] | 549 | ## my $pathsep = (defined $ENV{'GSDLOS'} && $ENV{'GSDLOS'} !~ m/windows/) ? ":" : ";";
|
---|
[26206] | 550 |
|
---|
[28375] | 551 | # Rewritten above to make ":" the default (Windows is the special
|
---|
| 552 | # case, anything else 'unusual' such as Solaris etc is Unix)
|
---|
| 553 | my $pathsep = (defined $ENV{'GSDLOS'} && (($ENV{'GSDLOS'} =~ m/windows/) && ($^O ne "cygwin"))) ? ";" : ":";
|
---|
| 554 |
|
---|
[16404] | 555 | # do not prepend any value/path that's already in the environment variable
|
---|
[24832] | 556 |
|
---|
| 557 | my $escaped_val = &filename_to_regex($val); # escape any backslashes and brackets for upcoming regex
|
---|
| 558 | if (!defined($ENV{$var})) {
|
---|
| 559 | $ENV{$var} = "$val";
|
---|
[16442] | 560 | }
|
---|
[24832] | 561 | elsif($ENV{$var} !~ m/$escaped_val/) {
|
---|
[26206] | 562 | $ENV{$var} = "$val".$pathsep.$ENV{$var};
|
---|
[10212] | 563 | }
|
---|
| 564 | }
|
---|
| 565 |
|
---|
| 566 | sub envvar_append {
|
---|
| 567 | my ($var,$val) = @_;
|
---|
[26206] | 568 |
|
---|
| 569 | # 64 bit linux can't handle ";" as path separator, so make sure to set this to the right one for the OS
|
---|
| 570 | my $pathsep = (defined $ENV{'GSDLOS'} && $ENV{'GSDLOS'} !~ m/windows/) ? ":" : ";";
|
---|
[24832] | 571 |
|
---|
[16404] | 572 | # do not append any value/path that's already in the environment variable
|
---|
[26206] | 573 |
|
---|
[24832] | 574 | my $escaped_val = &filename_to_regex($val); # escape any backslashes and brackets for upcoming regex
|
---|
| 575 | if (!defined($ENV{$var})) {
|
---|
| 576 | $ENV{$var} = "$val";
|
---|
[16442] | 577 | }
|
---|
[24832] | 578 | elsif($ENV{$var} !~ m/$escaped_val/) {
|
---|
[26206] | 579 | $ENV{$var} = $ENV{$var}.$pathsep."$val";
|
---|
[24832] | 580 | }
|
---|
[10212] | 581 | }
|
---|
| 582 |
|
---|
[31862] | 583 | # debug aid
|
---|
| 584 | sub print_env {
|
---|
[31863] | 585 | my ($handle, @envvars) = @_; # print to $handle, which can be STDERR/STDOUT/file, etc.
|
---|
| 586 |
|
---|
[31862] | 587 | if (scalar(@envvars) == 0) {
|
---|
[31876] | 588 | #print $handle "@@@ All env vars requested\n";
|
---|
| 589 |
|
---|
| 590 | my $output = "";
|
---|
| 591 |
|
---|
| 592 | print $handle "@@@ Environment was:\n********\n";
|
---|
| 593 | foreach my $envvar (sort keys(%ENV)) {
|
---|
| 594 | if(defined $ENV{$envvar}) {
|
---|
| 595 | print $handle "\t$envvar = $ENV{$envvar}\n";
|
---|
| 596 | } else {
|
---|
| 597 | print $handle "\t$envvar = \n";
|
---|
| 598 | }
|
---|
| 599 | }
|
---|
| 600 | print $handle "********\n";
|
---|
[31862] | 601 | } else {
|
---|
[31863] | 602 | print $handle "@@@ Environment was:\n********\n";
|
---|
[31862] | 603 | foreach my $envvar (@envvars) {
|
---|
| 604 | if(defined $ENV{$envvar}) {
|
---|
[31863] | 605 | print $handle "\t$envvar = ".$ENV{$envvar}."\n";
|
---|
[31876] | 606 | } else {
|
---|
| 607 | print $handle "Env var '$envvar' was not set\n";
|
---|
[31862] | 608 | }
|
---|
| 609 | }
|
---|
[31863] | 610 | print $handle "********\n";
|
---|
[31862] | 611 | }
|
---|
| 612 | }
|
---|
[28394] | 613 |
|
---|
[31862] | 614 |
|
---|
[16380] | 615 | # splits a filename into a prefix and a tail extension using the tail_re, or
|
---|
| 616 | # if that fails, splits on the file_extension . (dot)
|
---|
| 617 | sub get_prefix_and_tail_by_regex {
|
---|
[10212] | 618 |
|
---|
[16380] | 619 | my ($filename,$tail_re) = @_;
|
---|
| 620 |
|
---|
| 621 | my ($file_prefix,$file_ext) = ($filename =~ m/^(.*?)($tail_re)$/);
|
---|
| 622 | if ((!defined $file_prefix) || (!defined $file_ext)) {
|
---|
| 623 | ($file_prefix,$file_ext) = ($filename =~ m/^(.*)(\..*?)$/);
|
---|
| 624 | }
|
---|
| 625 |
|
---|
| 626 | return ($file_prefix,$file_ext);
|
---|
| 627 | }
|
---|
| 628 |
|
---|
| 629 | # get full path and file only path from a base_dir (which may be empty) and
|
---|
| 630 | # file (which may contain directories)
|
---|
| 631 | sub get_full_filenames {
|
---|
| 632 | my ($base_dir, $file) = @_;
|
---|
| 633 |
|
---|
[28375] | 634 | # my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(0);
|
---|
| 635 | # my ($lcfilename) = ($cfilename =~ m/([^\\\/]*)$/);
|
---|
| 636 | # print STDERR "** Calling method: $lcfilename:$cline $cpackage->$csubr\n";
|
---|
| 637 |
|
---|
| 638 |
|
---|
[16380] | 639 | my $filename_full_path = $file;
|
---|
| 640 | # add on directory if present
|
---|
[27303] | 641 | $filename_full_path = &FileUtils::filenameConcatenate($base_dir, $file) if $base_dir =~ /\S/;
|
---|
[16380] | 642 |
|
---|
| 643 | my $filename_no_path = $file;
|
---|
| 644 |
|
---|
| 645 | # remove directory if present
|
---|
| 646 | $filename_no_path =~ s/^.*[\/\\]//;
|
---|
| 647 | return ($filename_full_path, $filename_no_path);
|
---|
| 648 | }
|
---|
| 649 |
|
---|
[8682] | 650 | # returns the path of a file without the filename -- ie. the directory the file is in
|
---|
| 651 | sub filename_head {
|
---|
| 652 | my $filename = shift(@_);
|
---|
| 653 |
|
---|
[28375] | 654 | if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
|
---|
[8682] | 655 | $filename =~ s/[^\\\\]*$//;
|
---|
| 656 | }
|
---|
| 657 | else {
|
---|
| 658 | $filename =~ s/[^\\\/]*$//;
|
---|
| 659 | }
|
---|
| 660 |
|
---|
| 661 | return $filename;
|
---|
| 662 | }
|
---|
| 663 |
|
---|
| 664 |
|
---|
[23362] | 665 |
|
---|
[1454] | 666 | # returns 1 if filename1 and filename2 point to the same
|
---|
| 667 | # file or directory
|
---|
| 668 | sub filenames_equal {
|
---|
| 669 | my ($filename1, $filename2) = @_;
|
---|
| 670 |
|
---|
| 671 | # use filename_cat to clean up trailing slashes and
|
---|
| 672 | # multiple slashes
|
---|
[27303] | 673 | $filename1 = &FileUtils::filenameConcatenate($filename1);
|
---|
| 674 | $filename2 = &FileUtils::filenameConcatenate($filename2);
|
---|
[1454] | 675 |
|
---|
| 676 | # filenames not case sensitive on windows
|
---|
| 677 | if ($ENV{'GSDLOS'} =~ /^windows$/i) {
|
---|
| 678 | $filename1 =~ tr/[A-Z]/[a-z]/;
|
---|
| 679 | $filename2 =~ tr/[A-Z]/[a-z]/;
|
---|
| 680 | }
|
---|
| 681 | return 1 if $filename1 eq $filename2;
|
---|
| 682 | return 0;
|
---|
| 683 | }
|
---|
| 684 |
|
---|
[24932] | 685 | # If filename is relative to within_dir, returns the relative path of filename to that directory
|
---|
| 686 | # with slashes in the filename returned as they were in the original (absolute) filename.
|
---|
[23362] | 687 | sub filename_within_directory
|
---|
| 688 | {
|
---|
| 689 | my ($filename,$within_dir) = @_;
|
---|
| 690 |
|
---|
[23371] | 691 | if ($within_dir !~ m/[\/\\]$/) {
|
---|
| 692 | my $dirsep = &util::get_dirsep();
|
---|
[23362] | 693 | $within_dir .= $dirsep;
|
---|
| 694 | }
|
---|
| 695 |
|
---|
[24829] | 696 | $within_dir = &filename_to_regex($within_dir); # escape DOS style file separator and brackets
|
---|
[23362] | 697 | if ($filename =~ m/^$within_dir(.*)$/) {
|
---|
| 698 | $filename = $1;
|
---|
| 699 | }
|
---|
| 700 |
|
---|
| 701 | return $filename;
|
---|
| 702 | }
|
---|
| 703 |
|
---|
[24932] | 704 | # If filename is relative to within_dir, returns the relative path of filename to that directory in URL format.
|
---|
| 705 | # Filename and within_dir can be any type of slashes, but will be compared as URLs (i.e. unix-style slashes).
|
---|
| 706 | # The subpath returned will also be a URL type filename.
|
---|
| 707 | sub filename_within_directory_url_format
|
---|
| 708 | {
|
---|
| 709 | my ($filename,$within_dir) = @_;
|
---|
| 710 |
|
---|
| 711 | # convert parameters only to / slashes if Windows
|
---|
| 712 |
|
---|
[24971] | 713 | my $filename_urlformat = &filepath_to_url_format($filename);
|
---|
| 714 | my $within_dir_urlformat = &filepath_to_url_format($within_dir);
|
---|
| 715 |
|
---|
[24932] | 716 | #if ($within_dir_urlformat !~ m/\/$/) {
|
---|
| 717 | # make sure directory ends with a slash
|
---|
| 718 | #$within_dir_urlformat .= "/";
|
---|
| 719 | #}
|
---|
| 720 |
|
---|
| 721 | my $within_dir_urlformat_re = &filename_to_regex($within_dir_urlformat); # escape any special RE characters, such as brackets
|
---|
| 722 |
|
---|
| 723 | #print STDERR "@@@@@ $filename_urlformat =~ $within_dir_urlformat_re\n";
|
---|
| 724 |
|
---|
| 725 | # dir prefix may or may not end with a slash (this is discarded when extracting the sub-filepath)
|
---|
| 726 | if ($filename_urlformat =~ m/^$within_dir_urlformat_re(?:\/)*(.*)$/) {
|
---|
| 727 | $filename_urlformat = $1;
|
---|
| 728 | }
|
---|
| 729 |
|
---|
| 730 | return $filename_urlformat;
|
---|
| 731 | }
|
---|
| 732 |
|
---|
[24971] | 733 | # Convert parameter to use / slashes if Windows (if on Linux leave any \ as is,
|
---|
| 734 | # since on Linux it doesn't represent a file separator but an escape char).
|
---|
| 735 | sub filepath_to_url_format
|
---|
| 736 | {
|
---|
| 737 | my ($filepath) = @_;
|
---|
[28375] | 738 | if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
|
---|
[24971] | 739 | # Only need to worry about Windows, as Unix style directories already in url-format
|
---|
| 740 | # Convert Windows style \ => /
|
---|
| 741 | $filepath =~ s@\\@/@g;
|
---|
| 742 | }
|
---|
| 743 | return $filepath;
|
---|
| 744 | }
|
---|
[24932] | 745 |
|
---|
[25093] | 746 | # regex filepaths on windows may include \\ as path separator. Convert \\ to /
|
---|
| 747 | sub filepath_regex_to_url_format
|
---|
| 748 | {
|
---|
| 749 | my ($filepath) = @_;
|
---|
[28375] | 750 | if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
|
---|
[25093] | 751 | # Only need to worry about Windows, as Unix style directories already in url-format
|
---|
| 752 | # Convert Windows style \\ => /
|
---|
| 753 | $filepath =~ s@\\\\@/@g;
|
---|
| 754 | }
|
---|
| 755 | return $filepath;
|
---|
| 756 |
|
---|
| 757 | }
|
---|
[24971] | 758 |
|
---|
[25093] | 759 | # Like File::Basename::fileparse, but expects filepath in url format (ie only / slash for dirsep)
|
---|
| 760 | # and ignores trailing /
|
---|
| 761 | # returns (file, dirs) dirs will be empty if no subdirs
|
---|
| 762 | sub url_fileparse
|
---|
| 763 | {
|
---|
| 764 | my ($filepath) = @_;
|
---|
| 765 | # remove trailing /
|
---|
| 766 | $filepath =~ s@/$@@;
|
---|
| 767 | if ($filepath !~ m@/@) {
|
---|
| 768 | return ($filepath, "");
|
---|
| 769 | }
|
---|
| 770 | my ($dirs, $file) = $filepath =~ m@(.+/)([^/]+)@;
|
---|
| 771 | return ($file, $dirs);
|
---|
| 772 |
|
---|
| 773 | }
|
---|
| 774 |
|
---|
| 775 |
|
---|
[10281] | 776 | sub filename_within_collection
|
---|
| 777 | {
|
---|
| 778 | my ($filename) = @_;
|
---|
| 779 |
|
---|
| 780 | my $collect_dir = $ENV{'GSDLCOLLECTDIR'};
|
---|
| 781 |
|
---|
| 782 | if (defined $collect_dir) {
|
---|
[23362] | 783 |
|
---|
[15875] | 784 | # if from within GSDLCOLLECTDIR, then remove directory prefix
|
---|
| 785 | # so source_filename is realative to it. This is done to aid
|
---|
| 786 | # portability, i.e. the collection can be moved to somewhere
|
---|
| 787 | # else on the file system and the archives directory will still
|
---|
| 788 | # work. This is needed, for example in the applet version of
|
---|
| 789 | # GLI where GSDLHOME/collect on the server will be different to
|
---|
| 790 | # the collect directory of the remove user. Of course,
|
---|
| 791 | # GSDLCOLLECTDIR subsequently needs to be put back on to turn
|
---|
| 792 | # it back into a full pathname.
|
---|
[23362] | 793 |
|
---|
| 794 | $filename = filename_within_directory($filename,$collect_dir);
|
---|
[10281] | 795 | }
|
---|
| 796 |
|
---|
| 797 | return $filename;
|
---|
| 798 | }
|
---|
| 799 |
|
---|
[23362] | 800 | sub prettyprint_file
|
---|
| 801 | {
|
---|
[23484] | 802 | my ($base_dir,$file,$gli) = @_;
|
---|
[23362] | 803 |
|
---|
[27303] | 804 | my $filename_full_path = &FileUtils::filenameConcatenate($base_dir,$file);
|
---|
[23362] | 805 |
|
---|
[28375] | 806 | if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
|
---|
[23362] | 807 | require Win32;
|
---|
| 808 |
|
---|
| 809 | # For some reason base_dir in the form c:/a/b/c
|
---|
| 810 | # This leads to confusion later on, so turn it back into
|
---|
| 811 | # the more usual Windows form
|
---|
| 812 | $base_dir =~ s/\//\\/g;
|
---|
| 813 | my $long_base_dir = Win32::GetLongPathName($base_dir);
|
---|
| 814 | my $long_full_path = Win32::GetLongPathName($filename_full_path);
|
---|
| 815 |
|
---|
| 816 | $file = filename_within_directory($long_full_path,$long_base_dir);
|
---|
[23484] | 817 | $file = encode("utf8",$file) if ($gli);
|
---|
[23362] | 818 | }
|
---|
| 819 |
|
---|
| 820 | return $file;
|
---|
| 821 | }
|
---|
| 822 |
|
---|
| 823 |
|
---|
| 824 | sub upgrade_if_dos_filename
|
---|
| 825 | {
|
---|
[23371] | 826 | my ($filename_full_path,$and_encode) = @_;
|
---|
[23362] | 827 |
|
---|
[28375] | 828 | if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
|
---|
[23362] | 829 | # Ensure any DOS-like filename, such as test~1.txt, has been upgraded
|
---|
| 830 | # to its long (Windows) version
|
---|
[23416] | 831 | my $long_filename = Win32::GetLongPathName($filename_full_path);
|
---|
| 832 | if (defined $long_filename) {
|
---|
[29810] | 833 |
|
---|
[23416] | 834 | $filename_full_path = $long_filename;
|
---|
| 835 | }
|
---|
[23362] | 836 | # Make sure initial drive letter is lower-case (to fit in with rest of Greenstone)
|
---|
[23483] | 837 | $filename_full_path =~ s/^(.):/\u$1:/;
|
---|
[29810] | 838 |
|
---|
[23371] | 839 | if ((defined $and_encode) && ($and_encode)) {
|
---|
| 840 | $filename_full_path = encode("utf8",$filename_full_path);
|
---|
| 841 | }
|
---|
[23362] | 842 | }
|
---|
| 843 |
|
---|
| 844 | return $filename_full_path;
|
---|
| 845 | }
|
---|
| 846 |
|
---|
| 847 |
|
---|
[23388] | 848 | sub downgrade_if_dos_filename
|
---|
| 849 | {
|
---|
| 850 | my ($filename_full_path) = @_;
|
---|
| 851 |
|
---|
[28375] | 852 | if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
|
---|
[23388] | 853 | require Win32;
|
---|
| 854 |
|
---|
| 855 | # Ensure the given long Windows filename is in a form that can
|
---|
| 856 | # be opened by Perl => convert it to a short DOS-like filename
|
---|
| 857 |
|
---|
[23414] | 858 | my $short_filename = Win32::GetShortPathName($filename_full_path);
|
---|
| 859 | if (defined $short_filename) {
|
---|
| 860 | $filename_full_path = $short_filename;
|
---|
| 861 | }
|
---|
[23416] | 862 | # Make sure initial drive letter is lower-case (to fit in
|
---|
| 863 | # with rest of Greenstone)
|
---|
[23483] | 864 | $filename_full_path =~ s/^(.):/\u$1:/;
|
---|
[23388] | 865 | }
|
---|
| 866 |
|
---|
| 867 | return $filename_full_path;
|
---|
| 868 | }
|
---|
| 869 |
|
---|
| 870 |
|
---|
[18441] | 871 | sub filename_is_absolute
|
---|
| 872 | {
|
---|
[27303] | 873 | warnings::warnif("deprecated", "util::filename_is_absolute() is deprecated, using FileUtils::isFilenameAbsolute() instead");
|
---|
| 874 | return &FileUtils::isFilenameAbsolute(@_);
|
---|
[18441] | 875 | }
|
---|
| 876 |
|
---|
| 877 |
|
---|
[17572] | 878 | ## @method make_absolute()
|
---|
| 879 | #
|
---|
| 880 | # Ensure the given file path is absolute in respect to the given base path.
|
---|
| 881 | #
|
---|
| 882 | # @param $base_dir A string denoting the base path the given dir must be
|
---|
| 883 | # absolute to.
|
---|
| 884 | # @param $dir The directory to be made absolute as a string. Note that the
|
---|
| 885 | # dir may already be absolute, in which case it will remain
|
---|
| 886 | # unchanged.
|
---|
| 887 | # @return The now absolute form of the directory as a string.
|
---|
| 888 | #
|
---|
| 889 | # @author John Thompson, DL Consulting Ltd.
|
---|
| 890 | # @copy 2006 DL Consulting Ltd.
|
---|
| 891 | #
|
---|
| 892 | #used in buildcol.pl, doesn't work for all cases --kjdon
|
---|
| 893 | sub make_absolute {
|
---|
| 894 |
|
---|
| 895 | my ($base_dir, $dir) = @_;
|
---|
[18441] | 896 | ### print STDERR "dir = $dir\n";
|
---|
[17572] | 897 | $dir =~ s/[\\\/]+/\//g;
|
---|
| 898 | $dir = $base_dir . "/$dir" unless ($dir =~ m|^(\w:)?/|);
|
---|
| 899 | $dir =~ s|^/tmp_mnt||;
|
---|
| 900 | 1 while($dir =~ s|/[^/]*/\.\./|/|g);
|
---|
| 901 | $dir =~ s|/[.][.]?/|/|g;
|
---|
| 902 | $dir =~ tr|/|/|s;
|
---|
[18441] | 903 | ### print STDERR "dir = $dir\n";
|
---|
[17572] | 904 |
|
---|
| 905 | return $dir;
|
---|
| 906 | }
|
---|
| 907 | ## make_absolute() ##
|
---|
[10281] | 908 |
|
---|
[7929] | 909 | sub get_dirsep {
|
---|
| 910 |
|
---|
[28375] | 911 | if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
|
---|
[7929] | 912 | return "\\";
|
---|
| 913 | } else {
|
---|
| 914 | return "\/";
|
---|
| 915 | }
|
---|
| 916 | }
|
---|
| 917 |
|
---|
[619] | 918 | sub get_os_dirsep {
|
---|
[4] | 919 |
|
---|
[28375] | 920 | if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin")) {
|
---|
[619] | 921 | return "\\\\";
|
---|
| 922 | } else {
|
---|
| 923 | return "\\\/";
|
---|
| 924 | }
|
---|
| 925 | }
|
---|
| 926 |
|
---|
| 927 | sub get_re_dirsep {
|
---|
| 928 |
|
---|
| 929 | return "\\\\|\\\/";
|
---|
| 930 | }
|
---|
| 931 |
|
---|
| 932 |
|
---|
[15003] | 933 | sub get_dirsep_tail {
|
---|
| 934 | my ($filename) = @_;
|
---|
| 935 |
|
---|
| 936 | # returns last part of directory or filename
|
---|
| 937 | # On unix e.g. a/b.d => b.d
|
---|
| 938 | # a/b/c => c
|
---|
| 939 |
|
---|
[15088] | 940 | my $dirsep = get_re_dirsep();
|
---|
| 941 | my @dirs = split (/$dirsep/, $filename);
|
---|
| 942 | my $tail = pop @dirs;
|
---|
[15003] | 943 |
|
---|
[15088] | 944 | # - caused problems under windows
|
---|
| 945 | #my ($tail) = ($filename =~ m/^(?:.*?$dirsep)?(.*?)$/);
|
---|
| 946 |
|
---|
[15003] | 947 | return $tail;
|
---|
| 948 | }
|
---|
| 949 |
|
---|
| 950 |
|
---|
[4] | 951 | # if this is running on windows we want binaries to end in
|
---|
| 952 | # .exe, otherwise they don't have to end in any extension
|
---|
| 953 | sub get_os_exe {
|
---|
[28375] | 954 | return ".exe" if (($ENV{'GSDLOS'} =~ /^windows$/i) && ($^O ne "cygwin"));
|
---|
[4] | 955 | return "";
|
---|
| 956 | }
|
---|
| 957 |
|
---|
| 958 |
|
---|
[86] | 959 | # test to see whether this is a big or little endian machine
|
---|
[15713] | 960 | sub is_little_endian
|
---|
| 961 | {
|
---|
| 962 | # To determine the name of the operating system, the variable $^O is a cheap alternative to pulling it out of the Config module;
|
---|
| 963 | # 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
|
---|
| 964 | # Otherwise, it's little endian
|
---|
| 965 |
|
---|
| 966 | #return 0 if $^O =~ /^darwin$/i;
|
---|
[17714] | 967 | #return 0 if $ENV{'GSDLOS'} =~ /^darwin$/i;
|
---|
| 968 |
|
---|
| 969 | # Going back to stating exactly whether the machine is little endian
|
---|
| 970 | # or big endian, without any special case for Macs. Since for rata it comes
|
---|
| 971 | # back with little endian and for shuttle with bigendian.
|
---|
[15713] | 972 | return (ord(substr(pack("s",1), 0, 1)) == 1);
|
---|
[86] | 973 | }
|
---|
[4] | 974 |
|
---|
[86] | 975 |
|
---|
[135] | 976 | # will return the collection name if successful, "" otherwise
|
---|
| 977 | sub use_collection {
|
---|
[1454] | 978 | my ($collection, $collectdir) = @_;
|
---|
[135] | 979 |
|
---|
[1454] | 980 | if (!defined $collectdir || $collectdir eq "") {
|
---|
[27303] | 981 | $collectdir = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "collect");
|
---|
[1454] | 982 | }
|
---|
| 983 |
|
---|
[28211] | 984 | if (!defined $ENV{'GREENSTONEHOME'}) { # for GS3, would have been defined in use_site_collection, to GSDL3HOME
|
---|
| 985 | $ENV{'GREENSTONEHOME'} = $ENV{'GSDLHOME'};
|
---|
| 986 | }
|
---|
| 987 |
|
---|
[135] | 988 | # get and check the collection
|
---|
| 989 | if (!defined($collection) || $collection eq "") {
|
---|
| 990 | if (defined $ENV{'GSDLCOLLECTION'}) {
|
---|
| 991 | $collection = $ENV{'GSDLCOLLECTION'};
|
---|
| 992 | } else {
|
---|
[2359] | 993 | print STDOUT "No collection specified\n";
|
---|
[135] | 994 | return "";
|
---|
| 995 | }
|
---|
| 996 | }
|
---|
| 997 |
|
---|
| 998 | if ($collection eq "modelcol") {
|
---|
[2359] | 999 | print STDOUT "You can't use modelcol.\n";
|
---|
[135] | 1000 | return "";
|
---|
| 1001 | }
|
---|
| 1002 |
|
---|
| 1003 | # make sure the environment variables GSDLCOLLECTION and GSDLCOLLECTDIR
|
---|
| 1004 | # are defined
|
---|
[17204] | 1005 | $ENV{'GSDLCOLLECTION'} = $collection;
|
---|
[28211] | 1006 | $ENV{'GSDLCOLLECTHOME'} = $collectdir;
|
---|
[27303] | 1007 | $ENV{'GSDLCOLLECTDIR'} = &FileUtils::filenameConcatenate($collectdir, $collection);
|
---|
[135] | 1008 |
|
---|
| 1009 | # make sure this collection exists
|
---|
| 1010 | if (!-e $ENV{'GSDLCOLLECTDIR'}) {
|
---|
[2359] | 1011 | print STDOUT "Invalid collection ($collection).\n";
|
---|
[135] | 1012 | return "";
|
---|
| 1013 | }
|
---|
| 1014 |
|
---|
| 1015 | # everything is ready to go
|
---|
| 1016 | return $collection;
|
---|
| 1017 | }
|
---|
| 1018 |
|
---|
[21207] | 1019 | sub get_current_collection_name {
|
---|
| 1020 | return $ENV{'GSDLCOLLECTION'};
|
---|
| 1021 | }
|
---|
[14926] | 1022 |
|
---|
| 1023 |
|
---|
| 1024 | # will return the collection name if successful, "" otherwise.
|
---|
| 1025 | # Like use_collection (above) but for greenstone 3 (taking account of site level)
|
---|
| 1026 |
|
---|
| 1027 | sub use_site_collection {
|
---|
| 1028 | my ($site, $collection, $collectdir) = @_;
|
---|
| 1029 |
|
---|
| 1030 | if (!defined $collectdir || $collectdir eq "") {
|
---|
| 1031 | die "GSDL3HOME not set.\n" unless defined $ENV{'GSDL3HOME'};
|
---|
[27303] | 1032 | $collectdir = &FileUtils::filenameConcatenate($ENV{'GSDL3HOME'}, "sites", $site, "collect");
|
---|
[14926] | 1033 | }
|
---|
| 1034 |
|
---|
[28211] | 1035 | if (defined $ENV{'GSDL3HOME'}) {
|
---|
| 1036 | $ENV{'GREENSTONEHOME'} = $ENV{'GSDL3HOME'};
|
---|
| 1037 | $ENV{'SITEHOME'} = &FileUtils::filenameConcatenate($ENV{'GREENSTONEHOME'}, "sites", $site);
|
---|
| 1038 | } elsif (defined $ENV{'GSDL3SRCHOME'}) {
|
---|
| 1039 | $ENV{'GREENSTONEHOME'} = &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'}, "web");
|
---|
| 1040 | $ENV{'SITEHOME'} = &FileUtils::filenameConcatenate($ENV{'GREENSTONEHOME'}, "sites", $site);
|
---|
| 1041 | } else {
|
---|
| 1042 | print STDERR "*** util::use_site_collection(). Warning: Neither GSDL3HOME nor GSDL3SRCHOME set.\n";
|
---|
| 1043 | }
|
---|
| 1044 |
|
---|
[14926] | 1045 | # collectdir explicitly set by this point (using $site variable if required).
|
---|
| 1046 | # Can call "old" gsdl2 use_collection now.
|
---|
| 1047 |
|
---|
| 1048 | return use_collection($collection,$collectdir);
|
---|
| 1049 | }
|
---|
| 1050 |
|
---|
| 1051 |
|
---|
| 1052 |
|
---|
[15018] | 1053 | sub locate_config_file
|
---|
| 1054 | {
|
---|
| 1055 | my ($file) = @_;
|
---|
| 1056 |
|
---|
| 1057 | my $locations = locate_config_files($file);
|
---|
| 1058 |
|
---|
| 1059 | return shift @$locations; # returns undef if 'locations' is empty
|
---|
| 1060 | }
|
---|
| 1061 |
|
---|
| 1062 |
|
---|
| 1063 | sub locate_config_files
|
---|
| 1064 | {
|
---|
| 1065 | my ($file) = @_;
|
---|
| 1066 |
|
---|
| 1067 | my @locations = ();
|
---|
| 1068 |
|
---|
| 1069 | if (-e $file) {
|
---|
| 1070 | # Clearly specified (most likely full filename)
|
---|
| 1071 | # No need to hunt in 'etc' directories, return value unchanged
|
---|
| 1072 | push(@locations,$file);
|
---|
| 1073 | }
|
---|
| 1074 | else {
|
---|
| 1075 | # Check for collection specific one before looking in global GSDL 'etc'
|
---|
[16969] | 1076 | if (defined $ENV{'GSDLCOLLECTDIR'} && $ENV{'GSDLCOLLECTDIR'} ne "") {
|
---|
| 1077 | my $test_collect_etc_filename
|
---|
[27303] | 1078 | = &FileUtils::filenameConcatenate($ENV{'GSDLCOLLECTDIR'},"etc", $file);
|
---|
[16969] | 1079 |
|
---|
| 1080 | if (-e $test_collect_etc_filename) {
|
---|
| 1081 | push(@locations,$test_collect_etc_filename);
|
---|
| 1082 | }
|
---|
[15018] | 1083 | }
|
---|
| 1084 | my $test_main_etc_filename
|
---|
[27303] | 1085 | = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"etc", $file);
|
---|
[15018] | 1086 | if (-e $test_main_etc_filename) {
|
---|
| 1087 | push(@locations,$test_main_etc_filename);
|
---|
| 1088 | }
|
---|
| 1089 | }
|
---|
| 1090 |
|
---|
| 1091 | return \@locations;
|
---|
| 1092 | }
|
---|
| 1093 |
|
---|
| 1094 |
|
---|
[9955] | 1095 | sub hyperlink_text
|
---|
| 1096 | {
|
---|
| 1097 | my ($text) = @_;
|
---|
| 1098 |
|
---|
| 1099 | $text =~ s/(http:\/\/[^\s]+)/<a href=\"$1\">$1<\/a>/mg;
|
---|
| 1100 | $text =~ s/(^|\s+)(www\.(\w|\.)+)/<a href=\"http:\/\/$2\">$2<\/a>/mg;
|
---|
| 1101 |
|
---|
| 1102 | return $text;
|
---|
| 1103 | }
|
---|
| 1104 |
|
---|
| 1105 |
|
---|
[16436] | 1106 | # A method to check if a directory is empty (note that an empty directory still has non-zero size!!!)
|
---|
| 1107 | # Code is from http://episteme.arstechnica.com/eve/forums/a/tpc/f/6330927813/m/436007700831
|
---|
[27303] | 1108 | sub is_dir_empty {
|
---|
| 1109 | warnings::warnif("deprecated", "util::is_dir_empty() is deprecated, using FileUtils::isDirectoryEmpty() instead");
|
---|
| 1110 | return &FileUtils::isDirectoryEmpty(@_);
|
---|
[16436] | 1111 | }
|
---|
| 1112 |
|
---|
[18337] | 1113 | # Returns the given filename converted using either URL encoding or base64
|
---|
| 1114 | # encoding, as specified by $rename_method. If the given filename has no suffix
|
---|
[20413] | 1115 | # (if it is just the tailname), then $no_suffix should be some defined value.
|
---|
| 1116 | # rename_method can be url, none, base64
|
---|
[18319] | 1117 | sub rename_file {
|
---|
[18337] | 1118 | my ($filename, $rename_method, $no_suffix) = @_;
|
---|
[18329] | 1119 |
|
---|
[18337] | 1120 | if(!$filename) { # undefined or empty string
|
---|
[18329] | 1121 | return $filename;
|
---|
| 1122 | }
|
---|
[18319] | 1123 |
|
---|
[20413] | 1124 | if (!$rename_method) {
|
---|
| 1125 | print STDERR "WARNING: no file renaming method specified. Defaulting to using URL encoding...\n";
|
---|
| 1126 | # Debugging information
|
---|
[22856] | 1127 | # my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(1);
|
---|
| 1128 | # print STDERR "Called from method: $cfilename:$cline $cpackage->$csubr\n";
|
---|
[20413] | 1129 | $rename_method = "url";
|
---|
| 1130 | } elsif($rename_method eq "none") {
|
---|
| 1131 | return $filename; # would have already been renamed
|
---|
| 1132 | }
|
---|
| 1133 |
|
---|
[19762] | 1134 | # No longer replace spaces with underscores, since underscores mess with incremental rebuild
|
---|
| 1135 | ### Replace spaces with underscore. Do this first else it can go wrong below when getting tailname
|
---|
| 1136 | ###$filename =~ s/ /_/g;
|
---|
[18337] | 1137 |
|
---|
| 1138 | my ($tailname,$dirname,$suffix);
|
---|
| 1139 | if($no_suffix) { # given a tailname, no suffix
|
---|
| 1140 | ($tailname,$dirname) = File::Basename::fileparse($filename);
|
---|
| 1141 | }
|
---|
| 1142 | else {
|
---|
| 1143 | ($tailname,$dirname,$suffix) = File::Basename::fileparse($filename, "\\.(?:[^\\.]+?)\$");
|
---|
| 1144 | }
|
---|
[23388] | 1145 | if (!$suffix) {
|
---|
| 1146 | $suffix = "";
|
---|
| 1147 | }
|
---|
[26973] | 1148 | # This breaks GLI matching extracted metadata to files in Enrich panel, as
|
---|
| 1149 | # original is eg .JPG while gsdlsourcefilename ends up .jpg
|
---|
| 1150 | # Not sure why it was done in first place...
|
---|
| 1151 | #else {
|
---|
| 1152 | #$suffix = lc($suffix);
|
---|
| 1153 | #}
|
---|
[18337] | 1154 |
|
---|
[20413] | 1155 | if ($rename_method eq "url") {
|
---|
[18319] | 1156 | $tailname = &unicode::url_encode($tailname);
|
---|
| 1157 | }
|
---|
| 1158 | elsif ($rename_method eq "base64") {
|
---|
[18341] | 1159 | $tailname = &unicode::base64_encode($tailname);
|
---|
[18319] | 1160 | $tailname =~ s/\s*//sg; # for some reason it adds spaces not just at end but also in middle
|
---|
| 1161 | }
|
---|
[18326] | 1162 |
|
---|
[18319] | 1163 | $filename = "$tailname$suffix";
|
---|
[18326] | 1164 | $filename = "$dirname$filename" if ($dirname ne "./" && $dirname ne ".\\");
|
---|
[18319] | 1165 |
|
---|
| 1166 | return $filename;
|
---|
| 1167 | }
|
---|
| 1168 |
|
---|
[21616] | 1169 |
|
---|
| 1170 | # BACKWARDS COMPATIBILITY: Just in case there are old .ldb/.bdb files
|
---|
[21664] | 1171 | sub rename_ldb_or_bdb_file {
|
---|
[18657] | 1172 | my ($filename_no_ext) = @_;
|
---|
| 1173 |
|
---|
| 1174 | my $new_filename = "$filename_no_ext.gdb";
|
---|
[21615] | 1175 | return if (-f $new_filename); # if the file has the right extension, don't need to do anything
|
---|
[18657] | 1176 | # try ldb
|
---|
| 1177 | my $old_filename = "$filename_no_ext.ldb";
|
---|
| 1178 |
|
---|
| 1179 | if (-f $old_filename) {
|
---|
[19056] | 1180 | print STDERR "Renaming $old_filename to $new_filename\n";
|
---|
| 1181 | rename ($old_filename, $new_filename)
|
---|
| 1182 | || print STDERR "Rename failed: $!\n";
|
---|
[18657] | 1183 | return;
|
---|
| 1184 | }
|
---|
| 1185 | # try bdb
|
---|
| 1186 | $old_filename = "$filename_no_ext.bdb";
|
---|
| 1187 | if (-f $old_filename) {
|
---|
[19056] | 1188 | print STDERR "Renaming $old_filename to $new_filename\n";
|
---|
| 1189 | rename ($old_filename, $new_filename)
|
---|
| 1190 | || print STDERR "Rename failed: $!\n";
|
---|
[18657] | 1191 | return;
|
---|
| 1192 | }
|
---|
| 1193 | }
|
---|
| 1194 |
|
---|
[24874] | 1195 | sub os_dir() {
|
---|
| 1196 |
|
---|
| 1197 | my $gsdlarch = "";
|
---|
| 1198 | if(defined $ENV{'GSDLARCH'}) {
|
---|
| 1199 | $gsdlarch = $ENV{'GSDLARCH'};
|
---|
| 1200 | }
|
---|
| 1201 | return $ENV{'GSDLOS'}.$gsdlarch;
|
---|
| 1202 | }
|
---|
[18657] | 1203 |
|
---|
[28997] | 1204 | # returns 1 if this (GS server) is a GS3 installation, returns 0 if it's GS2.
|
---|
| 1205 | sub is_gs3() {
|
---|
| 1206 | if($ENV{'GSDL3SRCHOME'}) {
|
---|
| 1207 | return 1;
|
---|
| 1208 | } else {
|
---|
| 1209 | return 0;
|
---|
| 1210 | }
|
---|
| 1211 | }
|
---|
| 1212 |
|
---|
[21719] | 1213 | # Returns the greenstone URL prefix extracted from the appropriate GS2/GS3 config file.
|
---|
| 1214 | # By default, /greenstone3 for GS3 or /greenstone for GS2.
|
---|
| 1215 | sub get_greenstone_url_prefix() {
|
---|
| 1216 | # if already set on a previous occasion, just return that
|
---|
| 1217 | # (Don't want to keep repeating this: cost of re-opening and scanning files.)
|
---|
| 1218 | return $ENV{'GREENSTONE_URL_PREFIX'} if($ENV{'GREENSTONE_URL_PREFIX'});
|
---|
[18657] | 1219 |
|
---|
[21719] | 1220 | my ($configfile, $urlprefix, $defaultUrlprefix);
|
---|
| 1221 | my @propertynames = ();
|
---|
| 1222 |
|
---|
| 1223 | if($ENV{'GSDL3SRCHOME'}) {
|
---|
| 1224 | $defaultUrlprefix = "/greenstone3";
|
---|
[27303] | 1225 | $configfile = &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'}, "packages", "tomcat", "conf", "Catalina", "localhost", "greenstone3.xml");
|
---|
[21719] | 1226 | push(@propertynames, qw/path\s*\=/);
|
---|
| 1227 | } else {
|
---|
| 1228 | $defaultUrlprefix = "/greenstone";
|
---|
[27303] | 1229 | $configfile = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "cgi-bin", &os_dir(), "gsdlsite.cfg");
|
---|
[21719] | 1230 | push(@propertynames, (qw/\nhttpprefix/, qw/\ngwcgi/)); # inspect one property then the other
|
---|
| 1231 | }
|
---|
| 1232 |
|
---|
| 1233 | $urlprefix = &extract_propvalue_from_file($configfile, \@propertynames);
|
---|
| 1234 |
|
---|
| 1235 | if(!$urlprefix) { # no values found for URL prefix, use default values
|
---|
| 1236 | $urlprefix = $defaultUrlprefix;
|
---|
| 1237 | } else {
|
---|
| 1238 | #gwcgi can contain more than the wanted prefix, we split on / to get the first "directory" level
|
---|
| 1239 | $urlprefix =~ s/^\///; # remove the starting slash
|
---|
| 1240 | my @dirs = split(/(\\|\/)/, $urlprefix);
|
---|
| 1241 | $urlprefix = shift(@dirs);
|
---|
| 1242 |
|
---|
| 1243 | if($urlprefix !~ m/^\//) { # in all cases: ensure the required forward slash is at the front
|
---|
| 1244 | $urlprefix = "/$urlprefix";
|
---|
| 1245 | }
|
---|
| 1246 | }
|
---|
| 1247 |
|
---|
| 1248 | # set for the future
|
---|
| 1249 | $ENV{'GREENSTONE_URL_PREFIX'} = $urlprefix;
|
---|
| 1250 | # print STDERR "*** in get_greenstone_url_prefix(): $urlprefix\n\n";
|
---|
| 1251 | return $urlprefix;
|
---|
| 1252 | }
|
---|
| 1253 |
|
---|
| 1254 |
|
---|
[28460] | 1255 |
|
---|
| 1256 | #
|
---|
[30561] | 1257 | # The following comes from activate.pl
|
---|
[28460] | 1258 | #
|
---|
| 1259 | # Designed to work with a server included with GS.
|
---|
| 1260 | # - For GS3, we ask ant for the library URL.
|
---|
| 1261 | # - For GS2, we derive the URL from the llssite.cfg file.
|
---|
| 1262 |
|
---|
| 1263 | sub get_full_greenstone_url_prefix
|
---|
| 1264 | {
|
---|
[31518] | 1265 | my ($gs_mode, $lib_name) = @_;
|
---|
| 1266 |
|
---|
[28460] | 1267 | # if already set on a previous occasion, just return that
|
---|
| 1268 | # (Don't want to keep repeating this: cost of re-opening and scanning files.)
|
---|
[31517] | 1269 | return $ENV{'FULL_GREENSTONE_URL_PREFIX'} if($ENV{'FULL_GREENSTONE_URL_PREFIX'});
|
---|
[28460] | 1270 |
|
---|
[31518] | 1271 | # set gs_mode if it was not passed in (servercontrol.pm would pass it in, any other callers won't)
|
---|
| 1272 | $gs_mode = ($ENV{'GSDL3SRCHOME'}) ? "gs3" : "gs2" unless defined $gs_mode;
|
---|
[28460] | 1273 |
|
---|
| 1274 | my $url = undef;
|
---|
| 1275 |
|
---|
| 1276 | if($gs_mode eq "gs2") {
|
---|
| 1277 | my $llssite_cfg = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "llssite.cfg");
|
---|
| 1278 |
|
---|
| 1279 | if(-f $llssite_cfg) {
|
---|
| 1280 | # check llssite.cfg for line with url property
|
---|
| 1281 | # for server.exe also need to use portnumber and enterlib properties
|
---|
| 1282 |
|
---|
| 1283 | # Read in the entire contents of the file in one hit
|
---|
| 1284 | if (!open (FIN, $llssite_cfg)) {
|
---|
[31515] | 1285 | print STDERR "util::get_full_greenstone_url_prefix() failed to open $llssite_cfg ($!)\n";
|
---|
[28460] | 1286 | return undef;
|
---|
| 1287 | }
|
---|
| 1288 |
|
---|
| 1289 | my $contents;
|
---|
| 1290 | sysread(FIN, $contents, -s FIN);
|
---|
| 1291 | close(FIN);
|
---|
| 1292 |
|
---|
| 1293 | my @lines = split(/[\n\r]+/, $contents); # split on carriage-returns and/or linefeeds
|
---|
| 1294 | my $enterlib = "";
|
---|
[30561] | 1295 | my $portnumber = "8282"; # will remain empty (implicit port 80) unless it's specifically been assigned
|
---|
[28460] | 1296 |
|
---|
| 1297 | foreach my $line (@lines) {
|
---|
| 1298 | if($line =~ m/^url=(.*)$/) {
|
---|
| 1299 | $url = $1;
|
---|
| 1300 | } elsif($line =~ m/^enterlib=(.*)$/) {
|
---|
| 1301 | $enterlib = $1;
|
---|
| 1302 | } elsif($line =~ m/^portnumber=(.*)$/) {
|
---|
| 1303 | $portnumber = $1;
|
---|
| 1304 | }
|
---|
| 1305 | }
|
---|
| 1306 |
|
---|
| 1307 | if(!$url) {
|
---|
| 1308 | return undef;
|
---|
| 1309 | }
|
---|
| 1310 | elsif($url eq "URL_pending") { # library is not running
|
---|
| 1311 | # do not process url=URL_pending in the file, since for server.exe
|
---|
| 1312 | # this just means the Enter Library button hasn't been pressed yet
|
---|
| 1313 | $url = undef;
|
---|
| 1314 | }
|
---|
| 1315 | else {
|
---|
| 1316 | # In the case of server.exe, need to do extra work to get the proper URL
|
---|
| 1317 | # But first, need to know whether we're indeed dealing with server.exe:
|
---|
| 1318 |
|
---|
| 1319 | # compare the URL's domain to the full URL
|
---|
| 1320 | # E.g. for http://localhost:8383/greenstone3/cgi-bin, the domain is localhost:8383
|
---|
| 1321 | my $uri = URI->new( $url );
|
---|
| 1322 | my $host = $uri->host;
|
---|
| 1323 | #print STDERR "@@@@@ host: $host\n";
|
---|
[31514] | 1324 | if($url =~ m/https?:\/\/$host(\/)?$/) {
|
---|
[31515] | 1325 | #if($url !~ m/https?:\/\/$host:$portnumber(\/)?/ || $url =~ m/https?:\/\/$host(\/)?$/) {
|
---|
[28460] | 1326 | # (if the URL does not contain the portnumber, OR if the port is implicitly 80 and)
|
---|
| 1327 | # If the domain with http:// prefix is completely the same as the URL, assume server.exe
|
---|
| 1328 | # then the actual URL is the result of suffixing the port and enterlib properties in llssite.cfg
|
---|
| 1329 | $url = $url.":".$portnumber.$enterlib;
|
---|
| 1330 | } # else, apache web server
|
---|
| 1331 |
|
---|
| 1332 | }
|
---|
| 1333 | }
|
---|
| 1334 | } elsif($gs_mode eq "gs3") {
|
---|
| 1335 | # Either check build.properties for tomcat.server, tomcat.port and app.name (and default servlet name).
|
---|
| 1336 | # app.name is stored in app.path by build.xml. Need to move app.name in build.properties from build.xml
|
---|
| 1337 |
|
---|
| 1338 | # Or, run the new target get-default-servlet-url
|
---|
| 1339 | # the output can look like:
|
---|
| 1340 | #
|
---|
| 1341 | # Buildfile: build.xml
|
---|
| 1342 | # [echo] os.name: Windows Vista
|
---|
| 1343 | #
|
---|
| 1344 | # get-default-servlet-url:
|
---|
| 1345 | # [echo] http://localhost:8383/greenstone3/library
|
---|
| 1346 | # BUILD SUCCESSFUL
|
---|
| 1347 | # Total time: 0 seconds
|
---|
| 1348 |
|
---|
| 1349 | #my $output = qx/ant get-default-servlet-url/; # backtick operator, to get STDOUT (else 2>&1)
|
---|
| 1350 | # - see http://stackoverflow.com/questions/799968/whats-the-difference-between-perls-backticks-system-and-exec
|
---|
| 1351 |
|
---|
| 1352 | # The get-default-servlet-url ant target can be run from anywhere by specifying the
|
---|
| 1353 | # location of GS3's ant build.xml buildfile. Activate.pl can be run from anywhere for GS3
|
---|
| 1354 | # GSDL3SRCHOME will be set for GS3 by gs3-setup.sh, a step that would have been necessary
|
---|
| 1355 | # to run the activate.pl script in the first place
|
---|
| 1356 |
|
---|
| 1357 | my $full_build_xml = &FileUtils::javaFilenameConcatenate($ENV{'GSDL3SRCHOME'},"build.xml");
|
---|
| 1358 |
|
---|
| 1359 | my $perl_command = "ant -buildfile \"$full_build_xml\" get-default-servlet-url";
|
---|
| 1360 |
|
---|
| 1361 | if (open(PIN, "$perl_command |")) {
|
---|
| 1362 | while (defined (my $perl_output_line = <PIN>)) {
|
---|
| 1363 |
|
---|
[31514] | 1364 | if($perl_output_line =~ m@(https?):\/\/(\S*)@) { # grab all the non-whitespace chars
|
---|
| 1365 | $url="$1://".$2; # preserve the http protocol #$url="http://".$1;
|
---|
[28460] | 1366 | }
|
---|
| 1367 | }
|
---|
| 1368 | close(PIN);
|
---|
[31518] | 1369 |
|
---|
| 1370 | if (defined $lib_name) { # url won't be undef now
|
---|
| 1371 | # replace the servlet_name portion of the url found, with the given library_name
|
---|
| 1372 | $url =~ s@/[^/]*$@/$lib_name@;
|
---|
| 1373 | }
|
---|
[28460] | 1374 | } else {
|
---|
[31515] | 1375 | print STDERR "util::get_full_greenstone_url_prefix() failed to run $perl_command to work out library URL for $gs_mode\n";
|
---|
[31514] | 1376 | }
|
---|
[28460] | 1377 | }
|
---|
| 1378 |
|
---|
| 1379 | # either the url is still undef or it is now set
|
---|
| 1380 | #print STDERR "\n@@@@@ final URL:|$url|\n" if $url;
|
---|
| 1381 | #print STDERR "\n@@@@@ URL still undef\n" if !$url;
|
---|
| 1382 |
|
---|
| 1383 | $ENV{'FULL_GREENSTONE_URL_PREFIX'} = $url;
|
---|
| 1384 |
|
---|
| 1385 | return $url;
|
---|
| 1386 | }
|
---|
| 1387 |
|
---|
| 1388 |
|
---|
[21719] | 1389 | # Given a config file (xml or java properties file) and a list/array of regular expressions
|
---|
| 1390 | # that represent property names to match on, this function will return the value for the 1st
|
---|
| 1391 | # matching property name. If the return value is undefined, no matching property was found.
|
---|
| 1392 | sub extract_propvalue_from_file() {
|
---|
| 1393 | my ($configfile, $propertynames) = @_;
|
---|
| 1394 |
|
---|
| 1395 | my $value;
|
---|
| 1396 | unless(open(FIN, "<$configfile")) {
|
---|
| 1397 | print STDERR "extract_propvalue_from_file(): Unable to open $configfile. $!\n";
|
---|
| 1398 | return $value; # not initialised
|
---|
| 1399 | }
|
---|
| 1400 |
|
---|
| 1401 | # Read the entire file at once, as one single line, then close it
|
---|
| 1402 | my $filecontents;
|
---|
| 1403 | {
|
---|
| 1404 | local $/ = undef;
|
---|
| 1405 | $filecontents = <FIN>;
|
---|
| 1406 | }
|
---|
| 1407 | close(FIN);
|
---|
| 1408 |
|
---|
| 1409 | foreach my $regex (@$propertynames) {
|
---|
| 1410 | ($value) = $filecontents=~ m/$regex\s*(\S*)/s; # read value of the property given by regex up to the 1st space
|
---|
| 1411 | if($value) {
|
---|
| 1412 | $value =~ s/^\"//; # remove any startquotes
|
---|
| 1413 | $value =~ s/\".*$//; # remove the 1st endquotes (if any) followed by any xml
|
---|
| 1414 | last; # found value for a matching property, break from loop
|
---|
| 1415 | }
|
---|
| 1416 | }
|
---|
| 1417 |
|
---|
| 1418 | return $value;
|
---|
| 1419 | }
|
---|
| 1420 |
|
---|
[23306] | 1421 | # Subroutine that sources setup.bash, given GSDLHOME and GSDLOS and
|
---|
| 1422 | # given that perllib is in @INC in order to invoke this subroutine.
|
---|
| 1423 | # Call as follows -- after setting up INC to include perllib and
|
---|
| 1424 | # after setting up GSDLHOME and GSDLOS:
|
---|
| 1425 | #
|
---|
| 1426 | # require util;
|
---|
| 1427 | # &util::setup_greenstone_env($ENV{'GSDLHOME'}, $ENV{'GSDLOS'});
|
---|
| 1428 | #
|
---|
| 1429 | sub setup_greenstone_env() {
|
---|
| 1430 | my ($GSDLHOME, $GSDLOS) = @_;
|
---|
| 1431 |
|
---|
| 1432 | #my %env_map = ();
|
---|
| 1433 | # Get the localised ENV settings of running a localised source setup.bash
|
---|
[23314] | 1434 | # and put it into the ENV here. Need to clear GSDLHOME before running setup
|
---|
| 1435 | #my $perl_command = "(cd $GSDLHOME; export GSDLHOME=; . ./setup.bash > /dev/null; env)";
|
---|
| 1436 | my $perl_command = "(cd $GSDLHOME; /bin/bash -c \"export GSDLHOME=; source setup.bash > /dev/null; env\")";
|
---|
[28375] | 1437 | if (($GSDLOS =~ m/windows/i) && ($^O ne "cygwin")) {
|
---|
[23314] | 1438 | #$perl_command = "cmd /C \"cd $GSDLHOME&& set GSDLHOME=&& setup.bat > nul&& set\"";
|
---|
| 1439 | $perl_command = "(cd $GSDLHOME&& set GSDLHOME=&& setup.bat > nul&& set)";
|
---|
[23306] | 1440 | }
|
---|
| 1441 | if (!open(PIN, "$perl_command |")) {
|
---|
| 1442 | print STDERR ("Unable to execute command: $perl_command. $!\n");
|
---|
[24563] | 1443 | }
|
---|
[23306] | 1444 |
|
---|
[29582] | 1445 | my $lastkey;
|
---|
[23306] | 1446 | while (defined (my $perl_output_line = <PIN>)) {
|
---|
| 1447 | my($key,$value) = ($perl_output_line =~ m/^([^=]*)[=](.*)$/);
|
---|
[29582] | 1448 | if(defined $key) {
|
---|
| 1449 | #$env_map{$key}=$value;
|
---|
| 1450 | $ENV{$key}=$value;
|
---|
| 1451 | $lastkey = $key;
|
---|
| 1452 | } elsif($lastkey && $perl_output_line !~ m/^\s*$/) {
|
---|
| 1453 | # there was no equals sign in $perl_output_line, so this
|
---|
| 1454 | # $perl_output_line may be a spillover from the previous
|
---|
| 1455 | $ENV{$lastkey} = $ENV{$lastkey}."\n".$perl_output_line;
|
---|
| 1456 | }
|
---|
[23306] | 1457 | }
|
---|
[24563] | 1458 | close (PIN);
|
---|
| 1459 |
|
---|
[23306] | 1460 | # If any keys in $ENV don't occur in Greenstone's localised env
|
---|
| 1461 | # (stored in $env_map), delete those entries from $ENV
|
---|
| 1462 | #foreach $key (keys %ENV) {
|
---|
| 1463 | # if(!defined $env_map{$key}) {
|
---|
[29708] | 1464 | # print STDOUT "**** DELETING ENV KEY: $key\tVALUE: $ENV{$key}\n";
|
---|
[23306] | 1465 | # delete $ENV{$key}; # del $ENV(key, value) pair
|
---|
| 1466 | # }
|
---|
| 1467 | #}
|
---|
| 1468 | #undef %env_map;
|
---|
| 1469 | }
|
---|
| 1470 |
|
---|
[24362] | 1471 | sub get_perl_exec() {
|
---|
| 1472 | my $perl_exec = $^X; # may return just "perl"
|
---|
| 1473 |
|
---|
| 1474 | if($ENV{'PERLPATH'}) {
|
---|
[27303] | 1475 | # OR: # $perl_exec = &FileUtils::filenameConcatenate($ENV{'PERLPATH'},"perl");
|
---|
[28375] | 1476 | if (($ENV{'GSDLOS'} =~ m/windows/) && ($^O ne "cygwin")) {
|
---|
[24362] | 1477 | $perl_exec = "$ENV{'PERLPATH'}\\Perl.exe";
|
---|
| 1478 | } else {
|
---|
| 1479 | $perl_exec = "$ENV{'PERLPATH'}/perl";
|
---|
| 1480 | }
|
---|
| 1481 | } else { # no PERLPATH, use Config{perlpath} else $^X: special variables
|
---|
| 1482 | # containing the full path to the current perl executable we're using
|
---|
| 1483 | $perl_exec = $Config{perlpath}; # configured path for perl
|
---|
| 1484 | if (!-e $perl_exec) { # may not point to location on this machine
|
---|
| 1485 | $perl_exec = $^X; # may return just "perl"
|
---|
| 1486 | if($perl_exec =~ m/^perl/i) { # warn if just perl or Perl.exe
|
---|
| 1487 | print STDERR "**** WARNING: Perl exec found contains no path: $perl_exec";
|
---|
| 1488 | }
|
---|
| 1489 | }
|
---|
| 1490 | }
|
---|
| 1491 |
|
---|
| 1492 | return $perl_exec;
|
---|
| 1493 | }
|
---|
| 1494 |
|
---|
[25533] | 1495 | # returns the path to the java command in the JRE included with GS (if any),
|
---|
| 1496 | # quoted to safeguard any spaces in this path, otherwise a simple java
|
---|
| 1497 | # command is returned which assumes and will try for a system java.
|
---|
[25512] | 1498 | sub get_java_command {
|
---|
| 1499 | my $java = "java";
|
---|
| 1500 | if(defined $ENV{'GSDLHOME'}) { # should be, as this script would be launched from the cmd line
|
---|
| 1501 | # after running setup.bat or from GLI which also runs setup.bat
|
---|
[27303] | 1502 | my $java_bin = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"packages","jre","bin");
|
---|
[25512] | 1503 | if(-d $java_bin) {
|
---|
[27303] | 1504 | $java = &FileUtils::filenameConcatenate($java_bin,"java");
|
---|
[25533] | 1505 | $java = "\"".$java."\""; # quoted to preserve spaces in path
|
---|
[25512] | 1506 | }
|
---|
| 1507 | }
|
---|
| 1508 | return $java;
|
---|
| 1509 | }
|
---|
[24362] | 1510 |
|
---|
[25512] | 1511 |
|
---|
[25577] | 1512 | # Given the qualified collection name (colgroup/collection),
|
---|
| 1513 | # returns the collection and colgroup parts
|
---|
| 1514 | sub get_collection_parts {
|
---|
| 1515 | # http://perldoc.perl.org/File/Basename.html
|
---|
| 1516 | # my($filename, $directories, $suffix) = fileparse($path);
|
---|
| 1517 | # "$directories contains everything up to and including the last directory separator in the $path
|
---|
| 1518 | # including the volume (if applicable). The remainder of the $path is the $filename."
|
---|
| 1519 | #my ($collection, $colgroup) = &File::Basename::fileparse($qualified_collection);
|
---|
| 1520 |
|
---|
| 1521 | my $qualified_collection = shift(@_);
|
---|
| 1522 |
|
---|
| 1523 | # Since activate.pl can be launched from the command-line, including by a user,
|
---|
| 1524 | # best not to assume colgroup uses URL-style slashes as would be the case with GLI
|
---|
| 1525 | # Also allow for the accidental inclusion of multiple slashes
|
---|
| 1526 | my ($colgroup, $collection) = split(/[\/\\]+/, $qualified_collection); #split('/', $qualified_collection);
|
---|
| 1527 |
|
---|
| 1528 | if(!defined $collection) {
|
---|
| 1529 | $collection = $colgroup;
|
---|
| 1530 | $colgroup = "";
|
---|
| 1531 | }
|
---|
| 1532 | return ($collection, $colgroup);
|
---|
| 1533 | }
|
---|
| 1534 |
|
---|
| 1535 | # work out the "collectdir/collection" location
|
---|
| 1536 | sub resolve_collection_dir {
|
---|
| 1537 | my ($collect_dir, $qualified_collection, $site) = @_; #, $gs_mode
|
---|
| 1538 |
|
---|
[28211] | 1539 | if (defined $ENV{'GSDLCOLLECTDIR'}) { # a predefined collection dir exists
|
---|
| 1540 | return $ENV{'GSDLCOLLECTDIR'};
|
---|
| 1541 | }
|
---|
| 1542 |
|
---|
[25577] | 1543 | my ($colgroup, $collection) = &util::get_collection_parts($qualified_collection);
|
---|
| 1544 |
|
---|
[28175] | 1545 | if (!defined $collect_dir || !$collect_dir) { # if undefined or empty string
|
---|
| 1546 | $collect_dir = &util::get_working_collect_dir($site);
|
---|
[25577] | 1547 | }
|
---|
[28175] | 1548 |
|
---|
| 1549 | return &FileUtils::filenameConcatenate($collect_dir,$colgroup,$collection);
|
---|
| 1550 | }
|
---|
| 1551 |
|
---|
| 1552 | # work out the full path to "collect" of this greenstone 2/3 installation
|
---|
| 1553 | sub get_working_collect_dir {
|
---|
[28211] | 1554 | my ($site) = @_;
|
---|
| 1555 |
|
---|
| 1556 | if (defined $ENV{'GSDLCOLLECTHOME'}) { # a predefined collect dir exists
|
---|
| 1557 | return $ENV{'GSDLCOLLECTHOME'};
|
---|
[28175] | 1558 | }
|
---|
[28211] | 1559 |
|
---|
| 1560 | if (defined $site && $site) { # site non-empty, so get default collect dir for GS3
|
---|
| 1561 |
|
---|
| 1562 | if (defined $ENV{'GSDL3HOME'}) {
|
---|
| 1563 | return &FileUtils::filenameConcatenate($ENV{'GSDL3HOME'},"sites",$site,"collect"); # web folder
|
---|
| 1564 | }
|
---|
| 1565 | elsif (defined $ENV{'GSDL3SRCHOME'}) {
|
---|
| 1566 | return &FileUtils::filenameConcatenate($ENV{'GSDL3SRCHOME'},"web","sites",$site,"collect");
|
---|
| 1567 | }
|
---|
| 1568 | }
|
---|
| 1569 |
|
---|
[28213] | 1570 | elsif (defined $ENV{'SITEHOME'}) {
|
---|
[28211] | 1571 | return &FileUtils::filenameConcatenate($ENV{'SITEHOME'},"collect");
|
---|
[28177] | 1572 | }
|
---|
| 1573 |
|
---|
| 1574 | else { # get default collect dir for GS2
|
---|
| 1575 | return &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"collect");
|
---|
| 1576 | }
|
---|
| 1577 | }
|
---|
| 1578 |
|
---|
[28211] | 1579 | sub is_abs_path_any_os {
|
---|
[28177] | 1580 | my ($path) = @_;
|
---|
| 1581 |
|
---|
[28211] | 1582 | # We can have filenames in our DBs that were produced on other OS, so this method exists
|
---|
| 1583 | # to help identify absolute paths in such cases.
|
---|
| 1584 |
|
---|
[28177] | 1585 | return 1 if($path =~ m@^/@); # full paths begin with forward slash on linux/mac
|
---|
| 1586 | return 1 if($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
|
---|
| 1587 |
|
---|
| 1588 | return 0;
|
---|
| 1589 | }
|
---|
| 1590 |
|
---|
| 1591 |
|
---|
| 1592 | # This subroutine is for improving portability of Greenstone collections from one OS to another,
|
---|
| 1593 | # to be used to convert absolute paths going into db files into paths with placeholders instead.
|
---|
| 1594 | # This sub works with util::get_common_gs_paths and takes a path to a greenstone file and, if it's
|
---|
| 1595 | # an absolute path, then it will replace the longest matching greenstone-path prefix of the given
|
---|
| 1596 | # path with a placeholder to match.
|
---|
| 1597 | # The Greenstone-path prefixes that can be matched are the following common Greenstone paths:
|
---|
| 1598 | # the path to the current (specific) collection, the path to the general GS collect directory,
|
---|
| 1599 | # the path to the site directory if GS3, else the path to the GSDLHOME/GSDL3HOME folder.
|
---|
| 1600 | # The longest matching prefix will be replaced with the equivalent placeholder:
|
---|
| 1601 | # @THISCOLLECTPATH@, else @COLLECTHOME@, else @SITEHOME@, else @GSDLHOME@.
|
---|
| 1602 | sub abspath_to_placeholders {
|
---|
| 1603 | my $path = shift(@_); # path to convert from absolute to one with placeholders
|
---|
[28236] | 1604 | my $opt_long_or_short_winfilenames = shift(@_) || "short"; # whether we want to force use of long file names even on windows, default uses short
|
---|
[28177] | 1605 |
|
---|
[28211] | 1606 | return $path unless is_abs_path_any_os($path); # path is relative
|
---|
[28236] | 1607 |
|
---|
| 1608 | if ($opt_long_or_short_winfilenames eq "long") {
|
---|
| 1609 | $path = &util::upgrade_if_dos_filename($path); # will only do something on windows
|
---|
| 1610 | }
|
---|
[28225] | 1611 |
|
---|
[28177] | 1612 | # now we know we're dealing with absolute paths and have to replace gs prefixes with placeholders
|
---|
[28213] | 1613 | my @gs_paths = ($ENV{'GSDLCOLLECTDIR'}, $ENV{'GSDLCOLLECTHOME'}, $ENV{'SITEHOME'}, $ENV{'GREENSTONEHOME'}); # list in this order: from longest to shortest path
|
---|
[28177] | 1614 |
|
---|
[28211] | 1615 | my %placeholder_map = ($ENV{'GREENSTONEHOME'} => '@GSDLHOME@', # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
|
---|
| 1616 | $ENV{'GSDLCOLLECTHOME'} => '@COLLECTHOME@',
|
---|
| 1617 | $ENV{'GSDLCOLLECTDIR'} => '@THISCOLLECTPATH@'
|
---|
[28177] | 1618 | );
|
---|
[28213] | 1619 | $placeholder_map{$ENV{'SITEHOME'}} = '@SITEHOME@' if defined $ENV{'SITEHOME'};
|
---|
[28177] | 1620 |
|
---|
[28228] | 1621 | $path = &util::_abspath_to_placeholders($path, \@gs_paths, \%placeholder_map);
|
---|
[28177] | 1622 |
|
---|
[28236] | 1623 | if ($ENV{'GSDLOS'} =~ /^windows$/i && $opt_long_or_short_winfilenames eq "short") {
|
---|
[28228] | 1624 | # for windows need to look for matches on short file names too
|
---|
| 1625 | # matched paths are again to be replaced with the usual placeholders
|
---|
| 1626 |
|
---|
| 1627 | my $gsdlcollectdir = &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTDIR'});
|
---|
| 1628 | my $gsdlcollecthome = &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTHOME'});
|
---|
| 1629 | my $sitehome = (defined $ENV{'SITEHOME'}) ? &util::downgrade_if_dos_filename($ENV{'SITEHOME'}) : undef;
|
---|
| 1630 | my $greenstonehome = &util::downgrade_if_dos_filename($ENV{'GREENSTONEHOME'});
|
---|
| 1631 |
|
---|
| 1632 | @gs_paths = ($gsdlcollectdir, $gsdlcollecthome, $sitehome, $greenstonehome); # order matters
|
---|
| 1633 |
|
---|
| 1634 | %placeholder_map = ($greenstonehome => '@GSDLHOME@', # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
|
---|
| 1635 | $gsdlcollecthome => '@COLLECTHOME@',
|
---|
| 1636 | $gsdlcollectdir => '@THISCOLLECTPATH@'
|
---|
| 1637 | );
|
---|
| 1638 | $placeholder_map{$sitehome} = '@SITEHOME@' if defined $sitehome;
|
---|
| 1639 |
|
---|
| 1640 | $path = &util::_abspath_to_placeholders($path, \@gs_paths, \%placeholder_map);
|
---|
| 1641 | }
|
---|
| 1642 |
|
---|
| 1643 | return $path;
|
---|
| 1644 | }
|
---|
| 1645 |
|
---|
| 1646 | sub _abspath_to_placeholders {
|
---|
| 1647 | my ($path, $gs_paths_ref, $placeholder_map_ref) = @_;
|
---|
| 1648 |
|
---|
[28177] | 1649 | # The sequence of elements in @gs_paths matters
|
---|
| 1650 | # Need to loop starting from the *longest* matching path (the path to the specific collection)
|
---|
| 1651 | # to the shortest matching path (the path to gsdlhome/gsdl3home folder):
|
---|
| 1652 |
|
---|
[28228] | 1653 | foreach my $gs_path (@$gs_paths_ref) {
|
---|
[28213] | 1654 | next if(!defined $gs_path); # site undefined for GS2
|
---|
[28211] | 1655 |
|
---|
[28177] | 1656 | my $re_path = &util::filename_to_regex($gs_path); # escape for regex
|
---|
| 1657 |
|
---|
[28211] | 1658 | if($path =~ m/^$re_path/i) { # case sensitive or not for OS?
|
---|
[28177] | 1659 |
|
---|
[28228] | 1660 | my $placeholder = $placeholder_map_ref->{$gs_path}; # get the placeholder to replace the matched path with
|
---|
[28177] | 1661 |
|
---|
| 1662 | $path =~ s/^$re_path/$placeholder/; #case sensitive or not?
|
---|
| 1663 | #$path =~ s/^[\\\/]//; # remove gs_path's trailing separator left behind at the start of the path
|
---|
[28604] | 1664 | # lowercase file extension, This is needed when shortfilenames are used, as case affects alphetical ordering, which affects diffcol
|
---|
| 1665 | $path =~ s/\.([A-Z]+)$/".".lc($1)/e;
|
---|
[28177] | 1666 | last; # done
|
---|
| 1667 | }
|
---|
| 1668 | }
|
---|
| 1669 |
|
---|
| 1670 | return $path;
|
---|
| 1671 | }
|
---|
| 1672 |
|
---|
| 1673 | # Function that does the reverse of the util::abspath_to_placeholders subroutine
|
---|
| 1674 | # Once again, call this with the values returned from util::get_common_gs_paths
|
---|
| 1675 | sub placeholders_to_abspath {
|
---|
| 1676 | my $path = shift(@_); # path that can contain placeholders to convert to resolved absolute path
|
---|
[28236] | 1677 | my $opt_long_or_short_winfilenames = shift(@_) || "short"; # whether we want to force use of long file names even on windows, default uses short
|
---|
[28177] | 1678 |
|
---|
| 1679 | return $path if($path !~ m/@/); # path contains no placeholders
|
---|
| 1680 |
|
---|
[28213] | 1681 | # replace placeholders with gs prefixes
|
---|
[28177] | 1682 | my @placeholders = ('@THISCOLLECTPATH@', '@COLLECTHOME@', '@SITEHOME@', '@GSDLHOME@'); # order of paths not crucial in this case,
|
---|
| 1683 | # but listed here from longest to shortest once placeholders are have been resolved
|
---|
| 1684 |
|
---|
[28213] | 1685 | # can't use double-quotes around at-sign, else perl tries to evaluate it as referring to an array
|
---|
[28228] | 1686 | my %placeholder_to_gspath_map;
|
---|
[28236] | 1687 | if ($ENV{'GSDLOS'} =~ /^windows$/i && $opt_long_or_short_winfilenames eq "short") {
|
---|
[28228] | 1688 | # always replace placeholders with short file names of the absolute paths on windows?
|
---|
| 1689 | %placeholder_to_gspath_map = ('@GSDLHOME@' => &util::downgrade_if_dos_filename($ENV{'GREENSTONEHOME'}),
|
---|
| 1690 | '@COLLECTHOME@' => &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTHOME'}),
|
---|
| 1691 | '@THISCOLLECTPATH@' => &util::downgrade_if_dos_filename($ENV{'GSDLCOLLECTDIR'})
|
---|
| 1692 | );
|
---|
| 1693 | $placeholder_to_gspath_map{'@SITEHOME@'} = &util::downgrade_if_dos_filename($ENV{'SITEHOME'}) if defined $ENV{'SITEHOME'};
|
---|
| 1694 | } else {
|
---|
| 1695 | %placeholder_to_gspath_map = ('@GSDLHOME@' => $ENV{'GREENSTONEHOME'},
|
---|
| 1696 | '@SITEHOME@' => $ENV{'SITEHOME'}, # can be undef
|
---|
| 1697 | '@COLLECTHOME@' => $ENV{'GSDLCOLLECTHOME'},
|
---|
| 1698 | '@THISCOLLECTPATH@' => $ENV{'GSDLCOLLECTDIR'}
|
---|
| 1699 | ); # $placeholder_to_gspath_map{'@SITEHOME@'} = $ENV{'SITEHOME'} if defined $ENV{'SITEHOME'};
|
---|
| 1700 | }
|
---|
[28177] | 1701 |
|
---|
| 1702 | foreach my $placeholder (@placeholders) {
|
---|
| 1703 | my $gs_path = $placeholder_to_gspath_map{$placeholder};
|
---|
| 1704 |
|
---|
[28213] | 1705 | next if(!defined $gs_path); # sitehome for GS2 is undefined
|
---|
[28177] | 1706 |
|
---|
| 1707 | if($path =~ m/^$placeholder/) {
|
---|
| 1708 | $path =~ s/^$placeholder/$gs_path/;
|
---|
| 1709 | last; # done
|
---|
| 1710 | }
|
---|
| 1711 | }
|
---|
| 1712 |
|
---|
| 1713 | return $path;
|
---|
| 1714 | }
|
---|
| 1715 |
|
---|
[25994] | 1716 | # Used by pdfpstoimg.pl and PDFBoxConverter to create a .item file from
|
---|
| 1717 | # a directory containing sequentially numbered images.
|
---|
| 1718 | sub create_itemfile
|
---|
| 1719 | {
|
---|
| 1720 | my ($output_dir, $convert_basename, $convert_to) = @_;
|
---|
[27303] | 1721 | my $page_num = "";
|
---|
[25994] | 1722 |
|
---|
[27303] | 1723 | opendir(DIR, $output_dir) || die "can't opendir $output_dir: $!";
|
---|
[25994] | 1724 | my @dir_files = grep {-f "$output_dir/$_"} readdir(DIR);
|
---|
[27303] | 1725 | closedir DIR;
|
---|
[25994] | 1726 |
|
---|
[27303] | 1727 | # Sort files in the directory by page_num
|
---|
[25994] | 1728 | sub page_number {
|
---|
| 1729 | my ($dir) = @_;
|
---|
[27970] | 1730 | my ($pagenum) =($dir =~ m/^.*?[-\.]?(\d+)(\.(jpg|gif|png))?$/i);
|
---|
| 1731 | # my ($pagenum) =($dir =~ m/(\d+)(\.(jpg|gif|png))?$/i); # this works but is not as safe/strict about input filepatterns as the above
|
---|
[25994] | 1732 |
|
---|
| 1733 | $pagenum = 1 unless defined $pagenum;
|
---|
| 1734 | return $pagenum;
|
---|
| 1735 | }
|
---|
| 1736 |
|
---|
[27303] | 1737 | # sort the files in the directory in the order of page_num rather than lexically.
|
---|
[25994] | 1738 | @dir_files = sort { page_number($a) <=> page_number($b) } @dir_files;
|
---|
| 1739 |
|
---|
| 1740 | # work out if the numbering of the now sorted image files starts at 0 or not
|
---|
| 1741 | # by checking the number of the first _image_ file (skipping item files)
|
---|
| 1742 | my $starts_at_0 = 0;
|
---|
| 1743 | my $firstfile = ($dir_files[0] !~ /\.item$/i) ? $dir_files[0] : $dir_files[1];
|
---|
| 1744 | if(page_number($firstfile) == 0) { # 00 will evaluate to 0 too in this condition
|
---|
| 1745 | $starts_at_0 = 1;
|
---|
| 1746 | }
|
---|
| 1747 |
|
---|
[27303] | 1748 | my $item_file = &FileUtils::filenameConcatenate($output_dir, $convert_basename.".item");
|
---|
| 1749 | my $item_fh;
|
---|
| 1750 | &FileUtils::openFileHandle($item_file, 'w', \$item_fh);
|
---|
| 1751 | print $item_fh "<PagedDocument>\n";
|
---|
[25994] | 1752 |
|
---|
| 1753 | foreach my $file (@dir_files){
|
---|
| 1754 | if ($file !~ /\.item/i){
|
---|
| 1755 | $page_num = page_number($file);
|
---|
| 1756 | $page_num++ if $starts_at_0; # image numbers start at 0, so add 1
|
---|
[27303] | 1757 | print $item_fh " <Page pagenum=\"$page_num\" imgfile=\"$file\" txtfile=\"\"/>\n";
|
---|
| 1758 | }
|
---|
[25994] | 1759 | }
|
---|
| 1760 |
|
---|
[27303] | 1761 | print $item_fh "</PagedDocument>\n";
|
---|
| 1762 | &FileUtils::closeFileHandle($item_file, \$item_fh);
|
---|
[25994] | 1763 | return $item_file;
|
---|
| 1764 | }
|
---|
| 1765 |
|
---|
[28560] | 1766 | # Sets the gnomelib_env. Based on the logic in wvware.pl which can perhaps be replaced with a call to this function in future
|
---|
| 1767 | sub set_gnomelib_env
|
---|
| 1768 | {
|
---|
| 1769 | ## SET THE ENVIRONMENT AS DONE IN SETUP.BASH/BAT OF GNOME-LIB
|
---|
| 1770 | # Though this is only needed for darwin Lion at this point (and android, though that is untested)
|
---|
[27303] | 1771 |
|
---|
[28560] | 1772 | my $libext = "so";
|
---|
| 1773 | if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
|
---|
| 1774 | return;
|
---|
| 1775 | } elsif ($ENV{'GSDLOS'} =~ m/^darwin$/i) {
|
---|
| 1776 | $libext = "dylib";
|
---|
| 1777 | }
|
---|
| 1778 |
|
---|
| 1779 | if (!defined $ENV{'GEXTGNOME'}) {
|
---|
| 1780 | ##print STDERR "@@@ Setting GEXTGNOME env\n";
|
---|
| 1781 |
|
---|
| 1782 | my $gnome_dir = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"ext","gnome-lib-minimal");
|
---|
| 1783 |
|
---|
| 1784 | if(! -d $gnome_dir) {
|
---|
| 1785 | $gnome_dir = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'},"ext","gnome-lib");
|
---|
| 1786 |
|
---|
| 1787 | if(! -d $gnome_dir) {
|
---|
| 1788 | $gnome_dir = "";
|
---|
| 1789 | }
|
---|
| 1790 | }
|
---|
| 1791 |
|
---|
| 1792 | # now set other the related env vars,
|
---|
| 1793 | # IF we've found the gnome-lib dir installed in the ext folder
|
---|
| 1794 |
|
---|
| 1795 | if ($gnome_dir ne "" && -f &FileUtils::filenameConcatenate($gnome_dir, $ENV{'GSDLOS'}, "lib", "libiconv.$libext")) {
|
---|
| 1796 | $ENV{'GEXTGNOME'} = $gnome_dir;
|
---|
| 1797 | $ENV{'GEXTGNOME_INSTALLED'}=&FileUtils::filenameConcatenate($ENV{'GEXTGNOME'}, $ENV{'GSDLOS'});
|
---|
| 1798 |
|
---|
| 1799 | my $gnomelib_bin = &FileUtils::filenameConcatenate($ENV{'GEXTGNOME_INSTALLED'}, "bin");
|
---|
| 1800 | if(-d $gnomelib_bin) { # no bin subfolder in GS binary's cutdown gnome-lib-minimal folder
|
---|
| 1801 | &util::envvar_prepend("PATH", $gnomelib_bin);
|
---|
| 1802 | }
|
---|
| 1803 |
|
---|
| 1804 | # util's prepend will create LD/DYLD_LIB_PATH if it doesn't exist yet
|
---|
| 1805 | my $gextlib = &FileUtils::filenameConcatenate($ENV{'GEXTGNOME_INSTALLED'}, "lib");
|
---|
| 1806 |
|
---|
| 1807 | if($ENV{'GSDLOS'} eq "linux") {
|
---|
| 1808 | &util::envvar_prepend("LD_LIBRARY_PATH", $gextlib);
|
---|
| 1809 | }
|
---|
| 1810 | elsif ($ENV{'GSDLOS'} eq "darwin") {
|
---|
| 1811 | #&util::envvar_prepend("DYLD_LIBRARY_PATH", $gextlib);
|
---|
| 1812 | &util::envvar_prepend("DYLD_FALLBACK_LIBRARY_PATH", $gextlib);
|
---|
| 1813 | }
|
---|
| 1814 | }
|
---|
| 1815 |
|
---|
| 1816 | # Above largely mimics the setup.bash of the gnome-lib-minimal.
|
---|
| 1817 | # Not doing the devel-srcpack that gnome-lib-minimal's setup.bash used to set
|
---|
| 1818 | # Not exporting GSDLEXTS variable either
|
---|
| 1819 | }
|
---|
| 1820 |
|
---|
| 1821 | # print STDERR "@@@@@ GEXTGNOME: ".$ENV{'GEXTGNOME'}."\n\tINSTALL".$ENV{'GEXTGNOME_INSTALLED'}."\n";
|
---|
| 1822 | # print STDERR "\tPATH".$ENV{'PATH'}."\n";
|
---|
| 1823 | # print STDERR "\tLD_LIB_PATH".$ENV{'LD_LIBRARY_PATH'}."\n" if $ENV{'LD_LIBRARY_PATH};
|
---|
| 1824 | # print STDERR "\tDYLD_FALLBACK_LIB_PATH".$ENV{'DYLD_FALLBACK_LIBRARY_PATH'}."\n" if $ENV{'DYLD_FALLBACK_LIBRARY_PATH};
|
---|
| 1825 |
|
---|
| 1826 | # if no GEXTGNOME, maybe users didn't need gnome-lib to run gnomelib/libiconv dependent binaries like hashfile, suffix, wget
|
---|
| 1827 | # (wvware is launched in a gnomelib env from its own script, but could possibly go through this script in future)
|
---|
| 1828 | }
|
---|
| 1829 |
|
---|
| 1830 |
|
---|
| 1831 |
|
---|
[27374] | 1832 | ## @function augmentINC()
|
---|
| 1833 | #
|
---|
| 1834 | # Prepend a path (if it exists) onto INC but only if it isn't already in INC
|
---|
| 1835 | # @param $new_path The path to add
|
---|
| 1836 | # @author jmt12
|
---|
| 1837 | #
|
---|
[27303] | 1838 | sub augmentINC
|
---|
| 1839 | {
|
---|
| 1840 | my ($new_path) = @_;
|
---|
| 1841 | my $did_add_path = 0;
|
---|
[27374] | 1842 | # might need to be replaced with FileUtils::directoryExists() call eventually
|
---|
[27303] | 1843 | if (-d $new_path)
|
---|
| 1844 | {
|
---|
| 1845 | my $did_find_path = 0;
|
---|
| 1846 | foreach my $existing_path (@INC)
|
---|
| 1847 | {
|
---|
| 1848 | if ($existing_path eq $new_path)
|
---|
| 1849 | {
|
---|
| 1850 | $did_find_path = 1;
|
---|
| 1851 | last;
|
---|
| 1852 | }
|
---|
| 1853 | }
|
---|
| 1854 | if (!$did_find_path)
|
---|
| 1855 | {
|
---|
| 1856 | unshift(@INC, $new_path);
|
---|
| 1857 | $did_add_path = 1;
|
---|
| 1858 | }
|
---|
| 1859 | }
|
---|
| 1860 | return $did_add_path;
|
---|
| 1861 | }
|
---|
[27374] | 1862 | ## augmentINC()
|
---|
[27303] | 1863 |
|
---|
[27374] | 1864 |
|
---|
[4] | 1865 | 1;
|
---|