[537] | 1 | ###########################################################################
|
---|
| 2 | #
|
---|
[15870] | 3 | # DirectoryPlugin.pm --
|
---|
[537] | 4 | # A component of the Greenstone digital library software
|
---|
| 5 | # from the New Zealand Digital Library Project at the
|
---|
| 6 | # University of Waikato, New Zealand.
|
---|
| 7 | #
|
---|
| 8 | # Copyright (C) 1999 New Zealand Digital Library Project
|
---|
| 9 | #
|
---|
| 10 | # This program is free software; you can redistribute it and/or modify
|
---|
| 11 | # it under the terms of the GNU General Public License as published by
|
---|
| 12 | # the Free Software Foundation; either version 2 of the License, or
|
---|
| 13 | # (at your option) any later version.
|
---|
| 14 | #
|
---|
| 15 | # This program is distributed in the hope that it will be useful,
|
---|
| 16 | # but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
| 17 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
| 18 | # GNU General Public License for more details.
|
---|
| 19 | #
|
---|
| 20 | # You should have received a copy of the GNU General Public License
|
---|
| 21 | # along with this program; if not, write to the Free Software
|
---|
| 22 | # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
| 23 | #
|
---|
| 24 | ###########################################################################
|
---|
| 25 |
|
---|
[15870] | 26 | # DirectoryPlugin is a plugin which recurses through directories processing
|
---|
| 27 | # each file it finds - which basically means passing it down the plugin
|
---|
| 28 | # pipeline
|
---|
[4] | 29 |
|
---|
[15870] | 30 | package DirectoryPlugin;
|
---|
[2228] | 31 |
|
---|
[24951] | 32 | use extrametautil;
|
---|
[31492] | 33 | use CommonUtil;
|
---|
[4] | 34 | use plugin;
|
---|
[136] | 35 | use util;
|
---|
[27306] | 36 | use FileUtils;
|
---|
[13188] | 37 | use metadatautil;
|
---|
[4] | 38 |
|
---|
[8737] | 39 | use File::Basename;
|
---|
[10254] | 40 | use strict;
|
---|
| 41 | no strict 'refs';
|
---|
[15870] | 42 | no strict 'subs';
|
---|
| 43 |
|
---|
[29760] | 44 | use Encode::Locale;
|
---|
[13545] | 45 | use Encode;
|
---|
[29763] | 46 | use Unicode::Normalize;
|
---|
[4] | 47 |
|
---|
| 48 | BEGIN {
|
---|
[31492] | 49 | @DirectoryPlugin::ISA = ('CommonUtil');
|
---|
[4] | 50 | }
|
---|
| 51 |
|
---|
[4744] | 52 | my $arguments =
|
---|
| 53 | [ { 'name' => "block_exp",
|
---|
[31494] | 54 | 'desc' => "{CommonUtil.block_exp}",
|
---|
[6408] | 55 | 'type' => "regexp",
|
---|
[4744] | 56 | 'deft' => &get_default_block_exp(),
|
---|
| 57 | 'reqd' => "no" },
|
---|
[13417] | 58 | # this option has been deprecated. leave it here for now so we can warn people not to use it
|
---|
[4744] | 59 | { 'name' => "use_metadata_files",
|
---|
[15870] | 60 | 'desc' => "{DirectoryPlugin.use_metadata_files}",
|
---|
[4744] | 61 | 'type' => "flag",
|
---|
[13188] | 62 | 'reqd' => "no",
|
---|
| 63 | 'hiddengli' => "yes" },
|
---|
[7686] | 64 | { 'name' => "recheck_directories",
|
---|
[15870] | 65 | 'desc' => "{DirectoryPlugin.recheck_directories}",
|
---|
[7686] | 66 | 'type' => "flag",
|
---|
[4744] | 67 | 'reqd' => "no" } ];
|
---|
[13188] | 68 |
|
---|
[15870] | 69 | my $options = { 'name' => "DirectoryPlugin",
|
---|
| 70 | 'desc' => "{DirectoryPlugin.desc}",
|
---|
[6408] | 71 | 'abstract' => "no",
|
---|
[4744] | 72 | 'inherits' => "yes",
|
---|
| 73 | 'args' => $arguments };
|
---|
[3540] | 74 |
|
---|
[4] | 75 | sub new {
|
---|
[10218] | 76 | my ($class) = shift (@_);
|
---|
| 77 | my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
|
---|
| 78 | push(@$pluginlist, $class);
|
---|
[3540] | 79 |
|
---|
[15870] | 80 | push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
|
---|
| 81 | push(@{$hashArgOptLists->{"OptList"}},$options);
|
---|
[10218] | 82 |
|
---|
[31492] | 83 | my $self = new CommonUtil($pluginlist, $inputargs, $hashArgOptLists);
|
---|
[13188] | 84 |
|
---|
| 85 | if ($self->{'info_only'}) {
|
---|
| 86 | # don't worry about any options or initialisations etc
|
---|
| 87 | return bless $self, $class;
|
---|
| 88 | }
|
---|
| 89 |
|
---|
| 90 | # we have left this option in so we can warn people who are still using it
|
---|
[2813] | 91 | if ($self->{'use_metadata_files'}) {
|
---|
[15870] | 92 | die "ERROR: DirectoryPlugin -use_metadata_files option has been deprecated. Please remove the option and add MetadataXMLPlug to your plugin list instead!\n";
|
---|
[2813] | 93 | }
|
---|
[16391] | 94 |
|
---|
[17738] | 95 | $self->{'num_processed'} = 0;
|
---|
| 96 | $self->{'num_not_processed'} = 0;
|
---|
| 97 | $self->{'num_blocked'} = 0;
|
---|
| 98 | $self->{'num_archives'} = 0;
|
---|
| 99 |
|
---|
[8737] | 100 | $self->{'subdir_extrametakeys'} = {};
|
---|
| 101 |
|
---|
[4] | 102 | return bless $self, $class;
|
---|
| 103 | }
|
---|
| 104 |
|
---|
[17738] | 105 | # called once, at the start of processing
|
---|
| 106 | sub init {
|
---|
| 107 | my $self = shift (@_);
|
---|
| 108 | my ($verbosity, $outhandle, $failhandle) = @_;
|
---|
| 109 |
|
---|
| 110 | # verbosity is passed through from the processor
|
---|
| 111 | $self->{'verbosity'} = $verbosity;
|
---|
| 112 |
|
---|
| 113 | # as are the outhandle and failhandle
|
---|
| 114 | $self->{'outhandle'} = $outhandle if defined $outhandle;
|
---|
| 115 | $self->{'failhandle'} = $failhandle;
|
---|
| 116 |
|
---|
| 117 | }
|
---|
| 118 |
|
---|
| 119 | # called once, after all passes have finished
|
---|
| 120 | sub deinit {
|
---|
| 121 | my ($self) = @_;
|
---|
| 122 |
|
---|
| 123 | }
|
---|
| 124 |
|
---|
| 125 | # called at the beginning of each plugin pass (import has one, building has many)
|
---|
[10156] | 126 | sub begin {
|
---|
| 127 | my $self = shift (@_);
|
---|
| 128 | my ($pluginfo, $base_dir, $processor, $maxdocs) = @_;
|
---|
| 129 |
|
---|
[21586] | 130 | # Only lookup timestamp info for import.pl, and only if incremental is set
|
---|
[10156] | 131 | my $proc_package_name = ref $processor;
|
---|
[12969] | 132 | if ($proc_package_name !~ /buildproc$/ && $self->{'incremental'} == 1) {
|
---|
[21586] | 133 | # Get the infodbtype value for this collection from the arcinfo object
|
---|
| 134 | my $infodbtype = $processor->getoutputinfo()->{'infodbtype'};
|
---|
[23171] | 135 | $infodbtype = "gdbm" if $infodbtype eq "gdbm-txtgz"; # in archives, cannot use txtgz version
|
---|
[10156] | 136 | my $output_dir = $processor->getoutputdir();
|
---|
[21586] | 137 | my $archives_inf = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $output_dir);
|
---|
[18441] | 138 |
|
---|
[10156] | 139 | if ( -e $archives_inf ) {
|
---|
| 140 | $self->{'inf_timestamp'} = -M $archives_inf;
|
---|
| 141 | }
|
---|
| 142 | }
|
---|
| 143 | }
|
---|
| 144 |
|
---|
[21308] | 145 | sub remove_all {
|
---|
[21285] | 146 | my $self = shift (@_);
|
---|
| 147 | my ($pluginfo, $base_dir, $processor, $maxdocs) = @_;
|
---|
| 148 | }
|
---|
[21308] | 149 |
|
---|
[21315] | 150 |
|
---|
[21308] | 151 | sub remove_one {
|
---|
| 152 | my $self = shift (@_);
|
---|
[21315] | 153 | my ($file, $oids, $archivedir) = @_;
|
---|
| 154 | return undef; # this will never be called for directories (will it??)
|
---|
| 155 |
|
---|
[21308] | 156 | }
|
---|
| 157 |
|
---|
| 158 |
|
---|
[17738] | 159 | # called at the end of each plugin pass
|
---|
[15870] | 160 | sub end {
|
---|
[17738] | 161 | my ($self) = shift (@_);
|
---|
[10156] | 162 |
|
---|
[15870] | 163 | }
|
---|
| 164 |
|
---|
| 165 |
|
---|
[17738] | 166 |
|
---|
[4] | 167 | # return 1 if this class might recurse using $pluginfo
|
---|
| 168 | sub is_recursive {
|
---|
| 169 | my $self = shift (@_);
|
---|
[2813] | 170 |
|
---|
[4] | 171 | return 1;
|
---|
| 172 | }
|
---|
| 173 |
|
---|
[2228] | 174 | sub get_default_block_exp {
|
---|
| 175 | my $self = shift (@_);
|
---|
[2813] | 176 |
|
---|
[26867] | 177 | return '(?i)(CVS|\.svn|Thumbs\.db|OIDcount|\.DS_Store|~)$';
|
---|
[2228] | 178 | }
|
---|
| 179 |
|
---|
[16391] | 180 | sub check_directory_path {
|
---|
[2228] | 181 |
|
---|
[16391] | 182 | my $self = shift(@_);
|
---|
| 183 | my ($dirname) = @_;
|
---|
[2813] | 184 |
|
---|
[2228] | 185 | return undef unless (-d $dirname);
|
---|
[16391] | 186 |
|
---|
[2228] | 187 | return 0 if ($self->{'block_exp'} ne "" && $dirname =~ /$self->{'block_exp'}/);
|
---|
[7932] | 188 |
|
---|
[16391] | 189 | my $outhandle = $self->{'outhandle'};
|
---|
| 190 |
|
---|
[2228] | 191 | # check to make sure we're not reading the archives or index directory
|
---|
| 192 | my $gsdlhome = quotemeta($ENV{'GSDLHOME'});
|
---|
[2813] | 193 | if ($dirname =~ m/^$gsdlhome\/.*?\/import.*?\/(archives|index)$/) {
|
---|
[15870] | 194 | print $outhandle "DirectoryPlugin: $dirname appears to be a reference to a Greenstone collection, skipping.\n";
|
---|
[2228] | 195 | return 0;
|
---|
[1755] | 196 | }
|
---|
[2813] | 197 |
|
---|
[1755] | 198 | # check to see we haven't got a cyclic path...
|
---|
| 199 | if ($dirname =~ m%(/.*){,41}%) {
|
---|
[15870] | 200 | print $outhandle "DirectoryPlugin: $dirname is 40 directories deep, is this a recursive path? if not increase constant in DirectoryPlugin.pm.\n";
|
---|
[2228] | 201 | return 0;
|
---|
[1755] | 202 | }
|
---|
[2813] | 203 |
|
---|
[1755] | 204 | # check to see we haven't got a cyclic path...
|
---|
| 205 | if ($dirname =~ m%.*?import/(.+?)/import/\1.*%) {
|
---|
[15870] | 206 | print $outhandle "DirectoryPlugin: $dirname appears to be in a recursive loop...\n";
|
---|
[2228] | 207 | return 0;
|
---|
[1755] | 208 | }
|
---|
[16391] | 209 |
|
---|
| 210 | return 1;
|
---|
| 211 | }
|
---|
| 212 |
|
---|
| 213 | # this may be called more than once
|
---|
| 214 | sub sort_out_associated_files {
|
---|
| 215 |
|
---|
| 216 | my $self = shift (@_);
|
---|
| 217 | my ($block_hash) = @_;
|
---|
| 218 | if (!scalar (keys %{$block_hash->{'shared_fileroot'}})) {
|
---|
| 219 | return;
|
---|
| 220 | }
|
---|
| 221 |
|
---|
| 222 | $self->{'assocfile_info'} = {} unless defined $self->{'assocfile_info'};
|
---|
| 223 | my $metadata = $self->{'assocfile_info'};
|
---|
| 224 | foreach my $prefix (keys %{$block_hash->{'shared_fileroot'}}) {
|
---|
| 225 | my $record = $block_hash->{'shared_fileroot'}->{$prefix};
|
---|
| 226 |
|
---|
| 227 | my $tie_to = $record->{'tie_to'};
|
---|
| 228 | my $exts = $record->{'exts'};
|
---|
| 229 |
|
---|
| 230 | if ((defined $tie_to) && (scalar (keys %$exts) > 0)) {
|
---|
| 231 | # set up fileblocks and assocfile_tobe
|
---|
| 232 | my $base_file = "$prefix$tie_to";
|
---|
| 233 | $metadata->{$base_file} = {} unless defined $metadata->{$base_file};
|
---|
| 234 | my $base_file_metadata = $metadata->{$base_file};
|
---|
| 235 |
|
---|
| 236 | $base_file_metadata->{'gsdlassocfile_tobe'} = [] unless defined $base_file_metadata->{'gsdlassocfile_tobe'};
|
---|
| 237 | my $assoc_tobe = $base_file_metadata->{'gsdlassocfile_tobe'};
|
---|
| 238 | foreach my $e (keys %$exts) {
|
---|
| 239 | # block the file
|
---|
[31477] | 240 | $self->block_filename($block_hash,"$prefix$e");
|
---|
[16391] | 241 | # set up as an associatd file
|
---|
| 242 | print STDERR " $self->{'plugin_type'}: Associating $prefix$e with $tie_to version\n";
|
---|
| 243 | my $mime_type = ""; # let system auto detect this
|
---|
| 244 | push(@$assoc_tobe,"$prefix$e:$mime_type:");
|
---|
| 245 |
|
---|
| 246 | }
|
---|
| 247 | }
|
---|
| 248 | } # foreach record
|
---|
| 249 |
|
---|
| 250 | $block_hash->{'shared_fileroot'} = undef;
|
---|
| 251 | $block_hash->{'shared_fileroot'} = {};
|
---|
| 252 |
|
---|
| 253 | }
|
---|
| 254 |
|
---|
| 255 |
|
---|
| 256 |
|
---|
| 257 |
|
---|
| 258 | sub file_block_read {
|
---|
| 259 | my $self = shift (@_);
|
---|
| 260 | my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $gli) = @_;
|
---|
| 261 |
|
---|
| 262 | my $outhandle = $self->{'outhandle'};
|
---|
| 263 | my $verbosity = $self->{'verbosity'};
|
---|
[2813] | 264 |
|
---|
[16391] | 265 | # Calculate the directory name and ensure it is a directory and
|
---|
| 266 | # that it is not explicitly blocked.
|
---|
| 267 | my $dirname = $file;
|
---|
[27306] | 268 | $dirname = &FileUtils::filenameConcatenate($base_dir, $file) if $base_dir =~ /\w/;
|
---|
[16391] | 269 |
|
---|
| 270 | my $directory_ok = $self->check_directory_path($dirname);
|
---|
| 271 | return $directory_ok unless (defined $directory_ok && $directory_ok == 1);
|
---|
| 272 |
|
---|
[18523] | 273 | print $outhandle "Global file scan checking directory: $dirname\n";
|
---|
| 274 |
|
---|
[18441] | 275 | $block_hash->{'all_files'} = {} unless defined $block_hash->{'all_files'};
|
---|
[20577] | 276 | $block_hash->{'metadata_files'} = {} unless defined $block_hash->{'metadata_files'};
|
---|
[18441] | 277 |
|
---|
[16391] | 278 | $block_hash->{'file_blocks'} = {} unless defined $block_hash->{'file_blocks'};
|
---|
| 279 | $block_hash->{'shared_fileroot'} = {} unless defined $block_hash->{'shared_fileroot'};
|
---|
| 280 |
|
---|
| 281 | # Recur over directory contents.
|
---|
| 282 | my (@dir, $subfile);
|
---|
| 283 | #my $count = 0;
|
---|
| 284 |
|
---|
| 285 | print $outhandle "DirectoryPlugin block: getting directory $dirname\n" if ($verbosity > 2);
|
---|
| 286 |
|
---|
| 287 | # find all the files in the directory
|
---|
| 288 | if (!opendir (DIR, $dirname)) {
|
---|
| 289 | if ($gli) {
|
---|
| 290 | print STDERR "<ProcessingError n='$file' r='Could not read directory $dirname'>\n";
|
---|
| 291 | }
|
---|
| 292 | print $outhandle "DirectoryPlugin: WARNING - couldn't read directory $dirname\n";
|
---|
| 293 | return -1; # error in processing
|
---|
| 294 | }
|
---|
[27578] | 295 | @dir = sort readdir (DIR);
|
---|
[16391] | 296 | closedir (DIR);
|
---|
| 297 |
|
---|
| 298 | for (my $i = 0; $i < scalar(@dir); $i++) {
|
---|
[23335] | 299 | my $raw_subfile = $dir[$i];
|
---|
| 300 | next if ($raw_subfile =~ m/^\.\.?$/);
|
---|
| 301 |
|
---|
[16391] | 302 | my $this_file_base_dir = $base_dir;
|
---|
[27306] | 303 | my $raw_file_subfile = &FileUtils::filenameConcatenate($file, $raw_subfile);
|
---|
[16391] | 304 |
|
---|
[23335] | 305 | # Recursively read each $raw_subfile
|
---|
[31477] | 306 | print $outhandle "DirectoryPlugin block recurring: $raw_file_subfile\n" if ($verbosity > 2);
|
---|
[16391] | 307 | #$count += &plugin::file_block_read ($pluginfo, $this_file_base_dir,
|
---|
[23335] | 308 |
|
---|
[16391] | 309 | &plugin::file_block_read ($pluginfo, $this_file_base_dir,
|
---|
[23335] | 310 | $raw_file_subfile,
|
---|
[16391] | 311 | $block_hash, $metadata, $gli);
|
---|
| 312 |
|
---|
| 313 | }
|
---|
| 314 | $self->sort_out_associated_files($block_hash);
|
---|
| 315 | #return $count;
|
---|
[23419] | 316 | return 1;
|
---|
[16391] | 317 |
|
---|
| 318 | }
|
---|
[17738] | 319 |
|
---|
| 320 | # We don't do metadata_read
|
---|
| 321 | sub metadata_read {
|
---|
| 322 | my $self = shift (@_);
|
---|
[19493] | 323 | my ($pluginfo, $base_dir, $file, $block_hash,
|
---|
| 324 | $extrametakeys, $extrametadata, $extrametafile,
|
---|
[23212] | 325 | $processor, $gli, $aux) = @_;
|
---|
[17738] | 326 |
|
---|
| 327 | return undef;
|
---|
| 328 | }
|
---|
| 329 |
|
---|
| 330 |
|
---|
[16391] | 331 | # return number of files processed, undef if can't process
|
---|
| 332 | # Note that $base_dir might be "" and that $file might
|
---|
| 333 | # include directories
|
---|
| 334 |
|
---|
| 335 | # This function passes around metadata hash structures. Metadata hash
|
---|
| 336 | # structures are hashes that map from a (scalar) key (the metadata element
|
---|
| 337 | # name) to either a scalar metadata value or a reference to an array of
|
---|
| 338 | # such values.
|
---|
| 339 |
|
---|
| 340 | sub read {
|
---|
| 341 | my $self = shift (@_);
|
---|
| 342 | my ($pluginfo, $base_dir, $file, $block_hash, $in_metadata, $processor, $maxdocs, $total_count, $gli) = @_;
|
---|
[17320] | 343 |
|
---|
[16391] | 344 | my $outhandle = $self->{'outhandle'};
|
---|
| 345 | my $verbosity = $self->{'verbosity'};
|
---|
[24349] | 346 |
|
---|
[16391] | 347 | # Calculate the directory name and ensure it is a directory and
|
---|
| 348 | # that it is not explicitly blocked.
|
---|
| 349 | my $dirname;
|
---|
| 350 | if ($file eq "") {
|
---|
| 351 | $dirname = $base_dir;
|
---|
| 352 | } else {
|
---|
| 353 | $dirname = $file;
|
---|
[27306] | 354 | $dirname = &FileUtils::filenameConcatenate($base_dir, $file) if $base_dir =~ /\w/;
|
---|
[16391] | 355 | }
|
---|
[24932] | 356 |
|
---|
[16391] | 357 | my $directory_ok = $self->check_directory_path($dirname);
|
---|
| 358 | return $directory_ok unless (defined $directory_ok && $directory_ok == 1);
|
---|
| 359 |
|
---|
[2228] | 360 | if (($verbosity > 2) && ((scalar keys %$in_metadata) > 0)) {
|
---|
[15870] | 361 | print $outhandle "DirectoryPlugin: metadata passed in: ",
|
---|
[2813] | 362 | join(", ", keys %$in_metadata), "\n";
|
---|
[2228] | 363 | }
|
---|
[2813] | 364 |
|
---|
[2228] | 365 | # Recur over directory contents.
|
---|
| 366 | my (@dir, $subfile);
|
---|
[6332] | 367 |
|
---|
[16391] | 368 | print $outhandle "DirectoryPlugin read: getting directory $dirname\n" if ($verbosity > 2);
|
---|
[2813] | 369 |
|
---|
[2228] | 370 | # find all the files in the directory
|
---|
| 371 | if (!opendir (DIR, $dirname)) {
|
---|
[9584] | 372 | if ($gli) {
|
---|
| 373 | print STDERR "<ProcessingError n='$file' r='Could not read directory $dirname'>\n";
|
---|
| 374 | }
|
---|
[15870] | 375 | print $outhandle "DirectoryPlugin: WARNING - couldn't read directory $dirname\n";
|
---|
[7362] | 376 | return -1; # error in processing
|
---|
[2228] | 377 | }
|
---|
[27578] | 378 | @dir = sort readdir (DIR);
|
---|
[29818] | 379 | map { $_ = &unicode::raw_filename_to_url_encoded($_); } @dir;
|
---|
[2228] | 380 | closedir (DIR);
|
---|
[7686] | 381 | # Re-order the files in the list so any directories ending with .all are moved to the end
|
---|
[8716] | 382 | for (my $i = scalar(@dir) - 1; $i >= 0; $i--) {
|
---|
[27306] | 383 | if (-d &FileUtils::filenameConcatenate($dirname, $dir[$i]) && $dir[$i] =~ /\.all$/) {
|
---|
[7686] | 384 | push(@dir, splice(@dir, $i, 1));
|
---|
| 385 | }
|
---|
| 386 | }
|
---|
| 387 |
|
---|
[13188] | 388 | # setup the metadata structures. we do a metadata_read pass to see if there is any additional metadata, then pass it to read
|
---|
| 389 |
|
---|
[2228] | 390 | my $additionalmetadata = 0; # is there extra metadata available?
|
---|
| 391 | my %extrametadata; # maps from filespec to extra metadata keys
|
---|
[19493] | 392 | my %extrametafile; # maps from filespec to the metadata.xml (or similar) file it came from
|
---|
[2228] | 393 | my @extrametakeys; # keys of %extrametadata in order read
|
---|
[8512] | 394 |
|
---|
[16391] | 395 |
|
---|
[29818] | 396 | my $os_dirsep = &util::get_os_dirsep();
|
---|
[8737] | 397 | my $dirsep = &util::get_dirsep();
|
---|
[11919] | 398 | my $base_dir_regexp = $base_dir;
|
---|
| 399 | $base_dir_regexp =~ s/\//$os_dirsep/g;
|
---|
[24932] | 400 |
|
---|
[29818] | 401 | # Want to get relative path of local_dirname within the base_directory
|
---|
| 402 | # but with URL style slashes.
|
---|
| 403 | my $local_dirname = &util::filename_within_directory_url_format($dirname, $base_dir);
|
---|
[24932] | 404 |
|
---|
[22896] | 405 | # if we are in import folder, then local_dirname will be empty
|
---|
| 406 | if ($local_dirname ne "") {
|
---|
[31477] | 407 | # convert to perl unicode
|
---|
| 408 | $local_dirname = $self->raw_filename_to_unicode($local_dirname);
|
---|
| 409 |
|
---|
[24932] | 410 | # look for extra metadata passed down from higher folders
|
---|
| 411 | $local_dirname .= "/"; # closing slash must be URL type slash also and not $dirsep;
|
---|
[22896] | 412 | if (defined $self->{'subdir_extrametakeys'}->{$local_dirname}) {
|
---|
| 413 | my $extrakeys = $self->{'subdir_extrametakeys'}->{$local_dirname};
|
---|
| 414 | foreach my $ek (@$extrakeys) {
|
---|
| 415 | my $extrakeys_re = $ek->{'re'};
|
---|
| 416 | my $extrakeys_md = $ek->{'md'};
|
---|
| 417 | my $extrakeys_mf = $ek->{'mf'};
|
---|
[24951] | 418 | &extrametautil::addmetakey(\@extrametakeys, $extrakeys_re);
|
---|
| 419 | &extrametautil::setmetadata(\%extrametadata, $extrakeys_re, $extrakeys_md);
|
---|
| 420 | &extrametautil::setmetafile(\%extrametafile, $extrakeys_re, $extrakeys_mf);
|
---|
[22896] | 421 | }
|
---|
| 422 | delete($self->{'subdir_extrametakeys'}->{$local_dirname});
|
---|
[31459] | 423 | }
|
---|
[8737] | 424 | }
|
---|
[23212] | 425 | # apply metadata pass for each of the files in the directory -- ignore
|
---|
| 426 | # maxdocs here
|
---|
[7686] | 427 | my $num_files = scalar(@dir);
|
---|
[8512] | 428 | for (my $i = 0; $i < scalar(@dir); $i++) {
|
---|
| 429 | my $subfile = $dir[$i];
|
---|
[23335] | 430 | next if ($subfile =~ m/^\.\.?$/);
|
---|
| 431 |
|
---|
[8512] | 432 | my $this_file_base_dir = $base_dir;
|
---|
[23335] | 433 | my $raw_subfile = &unicode::url_encoded_to_raw_filename($subfile);
|
---|
| 434 |
|
---|
[27306] | 435 | my $raw_file_subfile = &FileUtils::filenameConcatenate($file, $raw_subfile);
|
---|
| 436 | my $raw_full_filename = &FileUtils::filenameConcatenate($this_file_base_dir, $raw_file_subfile);
|
---|
[31477] | 437 | if ($self->raw_file_is_blocked($block_hash, $raw_full_filename)) {
|
---|
| 438 | print STDERR "DirectoryPlugin: file $raw_full_filename was blocked for metadata_read\n";# if ($verbosity > 2);
|
---|
| 439 | print STDERR "raw file was blocked for metadata read\n";
|
---|
| 440 | print STDERR &unicode::debug_unicode_string($raw_full_filename)."\n";
|
---|
[16391] | 441 | next;
|
---|
| 442 | }
|
---|
[31477] | 443 |
|
---|
[23335] | 444 | # Recursively read each $raw_subfile
|
---|
[31477] | 445 | print $outhandle "DirectoryPlugin metadata recurring: $raw_subfile\n"; # if ($verbosity > 2);
|
---|
[8512] | 446 |
|
---|
[23212] | 447 | &plugin::metadata_read ($pluginfo, $this_file_base_dir,
|
---|
[23335] | 448 | $raw_file_subfile,$block_hash,
|
---|
[23212] | 449 | \@extrametakeys, \%extrametadata,
|
---|
| 450 | \%extrametafile,
|
---|
| 451 | $processor, $gli);
|
---|
[8512] | 452 | $additionalmetadata = 1;
|
---|
| 453 | }
|
---|
[16391] | 454 |
|
---|
[8737] | 455 | # filter out any extrametakeys that mention subdirectories and store
|
---|
| 456 | # for later use (i.e. when that sub-directory is being processed)
|
---|
[25094] | 457 | foreach my $ek (@extrametakeys) { # where each Extrametakey (which is a filename) is stored as a url-style regex
|
---|
| 458 |
|
---|
| 459 | my ($subdir_re,$extrakey_dir) = &util::url_fileparse($ek);
|
---|
| 460 | if ($extrakey_dir ne "") {
|
---|
| 461 | # a subdir was specified
|
---|
[24951] | 462 | my $md = &extrametautil::getmetadata(\%extrametadata, $ek);
|
---|
| 463 | my $mf = &extrametautil::getmetafile(\%extrametafile, $ek);
|
---|
[8737] | 464 |
|
---|
| 465 | my $subdir_extrametakeys = $self->{'subdir_extrametakeys'};
|
---|
[19493] | 466 | my $subdir_rec = { 're' => $subdir_re, 'md' => $md, 'mf' => $mf };
|
---|
[15005] | 467 |
|
---|
[24951] | 468 | # when it's looked up, it must be relative to the base dir
|
---|
[17320] | 469 | push(@{$subdir_extrametakeys->{"$local_dirname$extrakey_dir"}},$subdir_rec);
|
---|
[8737] | 470 | }
|
---|
| 471 | }
|
---|
[8512] | 472 |
|
---|
| 473 | # import each of the files in the directory
|
---|
[23212] | 474 | my $count=0;
|
---|
[7686] | 475 | for (my $i = 0; $i <= scalar(@dir); $i++) {
|
---|
| 476 | # When every file in the directory has been done, pause for a moment (figuratively!)
|
---|
| 477 | # If the -recheck_directories argument hasn't been provided, stop now (default)
|
---|
| 478 | # Otherwise, re-read the contents of the directory to check for new files
|
---|
| 479 | # Any new files are added to the @dir list and are processed as normal
|
---|
| 480 | # This is necessary when documents to be indexed are specified in bibliographic DBs
|
---|
| 481 | # These files are copied/downloaded and stored in a new folder at import time
|
---|
| 482 | if ($i == $num_files) {
|
---|
| 483 | last unless $self->{'recheck_directories'};
|
---|
| 484 |
|
---|
| 485 | # Re-read the files in the directory to see if there are any new files
|
---|
| 486 | last if (!opendir (DIR, $dirname));
|
---|
[27578] | 487 | my @dirnow = sort readdir (DIR);
|
---|
[23335] | 488 | map { $_ = &unicode::raw_filename_to_url_encoded($_) } @dirnow;
|
---|
[7686] | 489 | closedir (DIR);
|
---|
| 490 |
|
---|
| 491 | # We're only interested if there are more files than there were before
|
---|
| 492 | last if (scalar(@dirnow) <= scalar(@dir));
|
---|
| 493 |
|
---|
| 494 | # Any new files are added to the end of @dir to get processed by the loop
|
---|
[8716] | 495 | my $j;
|
---|
[7686] | 496 | foreach my $subfilenow (@dirnow) {
|
---|
| 497 | for ($j = 0; $j < $num_files; $j++) {
|
---|
| 498 | last if ($subfilenow eq $dir[$j]);
|
---|
| 499 | }
|
---|
| 500 | if ($j == $num_files) {
|
---|
| 501 | # New file
|
---|
| 502 | push(@dir, $subfilenow);
|
---|
| 503 | }
|
---|
| 504 | }
|
---|
| 505 | # When the new files have been processed, check again
|
---|
| 506 | $num_files = scalar(@dir);
|
---|
| 507 | }
|
---|
| 508 |
|
---|
| 509 | my $subfile = $dir[$i];
|
---|
[9853] | 510 | last if ($maxdocs != -1 && ($count + $total_count) >= $maxdocs);
|
---|
[2228] | 511 | next if ($subfile =~ /^\.\.?$/);
|
---|
[29818] | 512 |
|
---|
[23335] | 513 | my $this_file_base_dir = $base_dir;
|
---|
| 514 | my $raw_subfile = &unicode::url_encoded_to_raw_filename($subfile);
|
---|
[29818] | 515 | # get the canonical unicode version of the filename. This may not match
|
---|
| 516 | # the filename on the file system. We will use it to compare to regex
|
---|
| 517 | # in the metadata table.
|
---|
[29795] | 518 | my $unicode_subfile = &util::raw_filename_to_unicode($dirname, $raw_subfile);
|
---|
[27306] | 519 | my $raw_file_subfile = &FileUtils::filenameConcatenate($file, $raw_subfile);
|
---|
[23335] | 520 | my $raw_full_filename
|
---|
[27306] | 521 | = &FileUtils::filenameConcatenate($this_file_base_dir,$raw_file_subfile);
|
---|
[31477] | 522 | my $full_unicode_file = $self->raw_filename_to_unicode($raw_full_filename);
|
---|
| 523 | print STDERR "full unicode filename $full_unicode_file\n";
|
---|
| 524 | print STDERR &unicode::debug_unicode_string($full_unicode_file)."\n";
|
---|
| 525 | if ($self->file_is_blocked($block_hash,$full_unicode_file)) {
|
---|
| 526 | print STDERR "Actually, we have blocked the unicode version\n";
|
---|
| 527 | next;
|
---|
| 528 | }
|
---|
[23335] | 529 | if ($self->file_is_blocked($block_hash,$raw_full_filename)) {
|
---|
[31477] | 530 | print STDERR "DirectoryPlugin: file $raw_full_filename was blocked for read\n"; # if ($verbosity > 2);
|
---|
[16391] | 531 | next;
|
---|
| 532 | }
|
---|
[7932] | 533 | # Follow Windows shortcuts
|
---|
[28489] | 534 | if ($raw_subfile =~ m/(?i)\.lnk$/ && (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin"))) {
|
---|
[7932] | 535 | require Win32::Shortcut;
|
---|
[27306] | 536 | my $shortcut = new Win32::Shortcut(&FileUtils::filenameConcatenate($dirname, $raw_subfile));
|
---|
[7932] | 537 | if ($shortcut) {
|
---|
| 538 | # The file to be processed is now the target of the shortcut
|
---|
| 539 | $this_file_base_dir = "";
|
---|
| 540 | $file = "";
|
---|
[23335] | 541 | $raw_subfile = $shortcut->Path;
|
---|
[7932] | 542 | }
|
---|
[30681] | 543 | $shortcut->Close(); # see http://cpansearch.perl.org/src/JDB/Win32-Shortcut-0.08/docs/reference.html
|
---|
[7932] | 544 | }
|
---|
| 545 |
|
---|
[3108] | 546 | # check for a symlink pointing back to a leading directory
|
---|
[23335] | 547 | if (-d "$dirname/$raw_subfile" && -l "$dirname/$raw_subfile") {
|
---|
[3108] | 548 | # readlink gives a "fatal error" on systems that don't implement
|
---|
| 549 | # symlinks. This assumes the the -l test above would fail on those.
|
---|
[23335] | 550 | my $linkdest=readlink "$dirname/$raw_subfile";
|
---|
[3108] | 551 | if (!defined ($linkdest)) {
|
---|
| 552 | # system error - file not found?
|
---|
[15870] | 553 | warn "DirectoryPlugin: symlink problem - $!";
|
---|
[3108] | 554 | } else {
|
---|
| 555 | # see if link points to current or a parent directory
|
---|
| 556 | if ($linkdest =~ m@^[\./\\]+$@ ||
|
---|
| 557 | index($dirname, $linkdest) != -1) {
|
---|
[23335] | 558 | warn "DirectoryPlugin: Ignoring recursive symlink ($dirname/$raw_subfile -> $linkdest)\n";
|
---|
[3108] | 559 | next;
|
---|
| 560 | ;
|
---|
| 561 | }
|
---|
| 562 | }
|
---|
| 563 | }
|
---|
| 564 |
|
---|
[23335] | 565 | print $outhandle "DirectoryPlugin: preparing metadata for $raw_subfile\n" if ($verbosity > 2);
|
---|
[317] | 566 |
|
---|
[23335] | 567 | # Make a copy of $in_metadata to pass to $raw_subfile
|
---|
[17313] | 568 | my $out_metadata = {};
|
---|
[13188] | 569 | &metadatautil::combine_metadata_structures($out_metadata, $in_metadata);
|
---|
[317] | 570 |
|
---|
[16391] | 571 | # check the assocfile_info
|
---|
[23335] | 572 | if (defined $self->{'assocfile_info'}->{$raw_full_filename}) {
|
---|
| 573 | &metadatautil::combine_metadata_structures($out_metadata, $self->{'assocfile_info'}->{$raw_full_filename});
|
---|
[16391] | 574 | }
|
---|
[23335] | 575 |
|
---|
[29818] | 576 | ### Now we need to look up the metadata table to see if there is any
|
---|
| 577 | # extra metadata for us. We need the canonical unicode version here.
|
---|
[2228] | 578 | if ($additionalmetadata == 1) {
|
---|
[19493] | 579 | foreach my $filespec (@extrametakeys) {
|
---|
[29745] | 580 | if ($unicode_subfile =~ /^$filespec$/) {
|
---|
[29818] | 581 | print $outhandle "File \"$unicode_subfile\" matches filespec \"$filespec\"\n"
|
---|
[2228] | 582 | if ($verbosity > 2);
|
---|
[24951] | 583 | my $mdref = &extrametautil::getmetadata(\%extrametadata, $filespec);
|
---|
| 584 | my $mfref = &extrametautil::getmetafile(\%extrametafile, $filespec);
|
---|
[19493] | 585 |
|
---|
| 586 | # Add the list files where the metadata came from
|
---|
| 587 | # into the metadata table so we can track this
|
---|
| 588 | # This mechanism is similar to how gsdlassocfile works
|
---|
| 589 |
|
---|
| 590 | my @metafile_pair = ();
|
---|
| 591 | foreach my $l (keys %$mfref) {
|
---|
| 592 | my $f = $mfref->{$l};
|
---|
[19516] | 593 | push (@metafile_pair, "$f : $l");
|
---|
[19493] | 594 | }
|
---|
| 595 |
|
---|
| 596 | $mdref->{'gsdlmetafile'} = \@metafile_pair;
|
---|
| 597 |
|
---|
[13188] | 598 | &metadatautil::combine_metadata_structures($out_metadata, $mdref);
|
---|
[2228] | 599 | }
|
---|
[4] | 600 | }
|
---|
| 601 | }
|
---|
[10156] | 602 |
|
---|
| 603 | if (defined $self->{'inf_timestamp'}) {
|
---|
[18469] | 604 | # Look to see if it's a completely new file
|
---|
[10156] | 605 |
|
---|
[23335] | 606 | if (!$block_hash->{'new_files'}->{$raw_full_filename}) {
|
---|
[18469] | 607 | # Not a new file, must be an existing file
|
---|
| 608 | # Let' see if it's newer than the last import.pl
|
---|
| 609 |
|
---|
| 610 |
|
---|
[23335] | 611 | if (! -d $raw_full_filename) {
|
---|
| 612 | if (!$block_hash->{'reindex_files'}->{$raw_full_filename}) {
|
---|
[20577] | 613 | # filename has been around for longer than inf_timestamp
|
---|
[29745] | 614 | print $outhandle "**** Skipping $unicode_subfile\n" if ($verbosity >3);
|
---|
[18469] | 615 | next;
|
---|
| 616 | }
|
---|
| 617 | else {
|
---|
| 618 | # Remove old folder in archives (might hash to something different)
|
---|
| 619 | # *** should be doing this on a Del one as well
|
---|
| 620 | # but leave folder name?? and ensure hashs to
|
---|
| 621 | # same again??
|
---|
| 622 |
|
---|
| 623 | # Then let through as new doc??
|
---|
| 624 |
|
---|
| 625 | # mark to doc-oids that rely on it for re-indexing
|
---|
| 626 | }
|
---|
[10156] | 627 | }
|
---|
| 628 | }
|
---|
| 629 | }
|
---|
| 630 |
|
---|
[2228] | 631 | # Recursively read each $subfile
|
---|
[29745] | 632 | print $outhandle "DirectoryPlugin recurring: $unicode_subfile\n" if ($verbosity > 2);
|
---|
[8512] | 633 |
|
---|
| 634 | $count += &plugin::read ($pluginfo, $this_file_base_dir,
|
---|
[23335] | 635 | $raw_file_subfile, $block_hash,
|
---|
[9853] | 636 | $out_metadata, $processor, $maxdocs, ($total_count + $count), $gli);
|
---|
[4] | 637 | }
|
---|
[7686] | 638 |
|
---|
[8512] | 639 | return $count;
|
---|
[2228] | 640 | }
|
---|
[4] | 641 |
|
---|
[17738] | 642 | sub compile_stats {
|
---|
| 643 | my $self = shift(@_);
|
---|
| 644 | my ($stats) = @_;
|
---|
| 645 | }
|
---|
| 646 |
|
---|
[4] | 647 | 1;
|
---|