| 1 | ###########################################################################
|
|---|
| 2 | #
|
|---|
| 3 | # RecPlug.pm --
|
|---|
| 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 |
|
|---|
| 26 | # RecPlug is a plugin which recurses through directories processing
|
|---|
| 27 | # each file it finds.
|
|---|
| 28 |
|
|---|
| 29 | # RecPlug has one option: use_metadata_files. When this is set, it will
|
|---|
| 30 | # check each directory for an XML file called "metadata.xml" that specifies
|
|---|
| 31 | # metadata for the files (and subdirectories) in the directory.
|
|---|
| 32 | #
|
|---|
| 33 | # Here's an example of a metadata file that uses three FileSet structures
|
|---|
| 34 | # (ignore the # characters):
|
|---|
| 35 |
|
|---|
| 36 | #<?xml version="1.0" encoding="UTF-8" standalone="no"?>
|
|---|
| 37 | #<!DOCTYPE DirectoryMetadata SYSTEM "http://greenstone.org/dtd/DirectoryMetadata/1.0/DirectoryMetadata.dtd">
|
|---|
| 38 | #<DirectoryMetadata>
|
|---|
| 39 | # <FileSet>
|
|---|
| 40 | # <FileName>nugget.*</FileName>
|
|---|
| 41 | # <Description>
|
|---|
| 42 | # <Metadata name="Title">Nugget Point, The Catlins</Metadata>
|
|---|
| 43 | # <Metadata name="Place" mode="accumulate">Nugget Point</Metadata>
|
|---|
| 44 | # </Description>
|
|---|
| 45 | # </FileSet>
|
|---|
| 46 | # <FileSet>
|
|---|
| 47 | # <FileName>nugget-point-1.jpg</FileName>
|
|---|
| 48 | # <Description>
|
|---|
| 49 | # <Metadata name="Title">Nugget Point Lighthouse, The Catlins</Metadata>
|
|---|
| 50 | # <Metadata name="Subject">Lighthouse</Metadata>
|
|---|
| 51 | # </Description>
|
|---|
| 52 | # </FileSet>
|
|---|
| 53 | # <FileSet>
|
|---|
| 54 | # <FileName>kaka-point-dir</FileName>
|
|---|
| 55 | # <Description>
|
|---|
| 56 | # <Metadata name="Title">Kaka Point, The Catlins</Metadata>
|
|---|
| 57 | # </Description>
|
|---|
| 58 | # </FileSet>
|
|---|
| 59 | #</DirectoryMetadata>
|
|---|
| 60 |
|
|---|
| 61 | # Metadata elements are read and applied to files in the order they appear
|
|---|
| 62 | # in the file.
|
|---|
| 63 | #
|
|---|
| 64 | # The FileName element describes the subfiles in the directory that the
|
|---|
| 65 | # metadata applies to as a perl regular expression (a FileSet group may
|
|---|
| 66 | # contain multiple FileName elements). So, <FileName>nugget.*</FileName>
|
|---|
| 67 | # indicates that the metadata records in the following Description block
|
|---|
| 68 | # apply to every subfile that starts with "nugget". For these files, a
|
|---|
| 69 | # Title metadata element is set, overriding any old value that the Title
|
|---|
| 70 | # might have had.
|
|---|
| 71 | #
|
|---|
| 72 | # Occasionally, we want to have multiple metadata values applied to a
|
|---|
| 73 | # document; in this case we use the "mode=accumulate" attribute of the
|
|---|
| 74 | # particular Metadata element. In the second metadata element of the first
|
|---|
| 75 | # FileSet above, the "Place" metadata is accumulating, and may therefore be
|
|---|
| 76 | # given several values. If we wanted to override these values and use a
|
|---|
| 77 | # single metadata element again, we could set the mode attribute to
|
|---|
| 78 | # "override" instead. Remember: every element is assumed to be in override
|
|---|
| 79 | # mode unless you specify otherwise, so if you want to accumulate metadata
|
|---|
| 80 | # for some field, every occurance must have "mode=accumulate" specified.
|
|---|
| 81 | #
|
|---|
| 82 | # The second FileSet element above applies to a specific file, called
|
|---|
| 83 | # nugget-point-1.jpg. This element overrides the Title metadata set in the
|
|---|
| 84 | # first FileSet, and adds a "Subject" metadata field.
|
|---|
| 85 | #
|
|---|
| 86 | # The third and final FileSet sets metadata for a subdirectory rather than
|
|---|
| 87 | # a file. The metadata specified (a Title) will be passed into the
|
|---|
| 88 | # subdirectory and applied to every file that occurs in the subdirectory
|
|---|
| 89 | # (and to every subsubdirectory and its contents, and so on) unless the
|
|---|
| 90 | # metadata is explictly overridden later in the import.
|
|---|
| 91 |
|
|---|
| 92 |
|
|---|
| 93 |
|
|---|
| 94 | package RecPlug;
|
|---|
| 95 |
|
|---|
| 96 | use BasPlug;
|
|---|
| 97 | use plugin;
|
|---|
| 98 | use util;
|
|---|
| 99 |
|
|---|
| 100 | use File::Basename;
|
|---|
| 101 |
|
|---|
| 102 |
|
|---|
| 103 | BEGIN {
|
|---|
| 104 | @RecPlug::ISA = ('BasPlug');
|
|---|
| 105 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib/cpan");
|
|---|
| 106 | }
|
|---|
| 107 |
|
|---|
| 108 | use XMLParser;
|
|---|
| 109 |
|
|---|
| 110 | my $arguments =
|
|---|
| 111 | [ { 'name' => "block_exp",
|
|---|
| 112 | 'desc' => "{BasPlug.block_exp}",
|
|---|
| 113 | 'type' => "regexp",
|
|---|
| 114 | 'deft' => &get_default_block_exp(),
|
|---|
| 115 | 'reqd' => "no" },
|
|---|
| 116 | { 'name' => "use_metadata_files",
|
|---|
| 117 | 'desc' => "{RecPlug.use_metadata_files}",
|
|---|
| 118 | 'type' => "flag",
|
|---|
| 119 | 'reqd' => "no" },
|
|---|
| 120 | { 'name' => "recheck_directories",
|
|---|
| 121 | 'desc' => "{RecPlug.recheck_directories}",
|
|---|
| 122 | 'type' => "flag",
|
|---|
| 123 | 'reqd' => "no" } ];
|
|---|
| 124 |
|
|---|
| 125 | my $options = { 'name' => "RecPlug",
|
|---|
| 126 | 'desc' => "{RecPlug.desc}",
|
|---|
| 127 | 'abstract' => "no",
|
|---|
| 128 | 'inherits' => "yes",
|
|---|
| 129 | 'args' => $arguments };
|
|---|
| 130 |
|
|---|
| 131 |
|
|---|
| 132 | my ($self);
|
|---|
| 133 | sub new {
|
|---|
| 134 | my $class = shift (@_);
|
|---|
| 135 |
|
|---|
| 136 | # $self is global for use within subroutines called by XML::Parser
|
|---|
| 137 | $self = new BasPlug ($class, @_);
|
|---|
| 138 |
|
|---|
| 139 | # 14-05-02 To allow for proper inheritance of arguments - John Thompson
|
|---|
| 140 | my $option_list = $self->{'option_list'};
|
|---|
| 141 | push( @{$option_list}, $options );
|
|---|
| 142 |
|
|---|
| 143 | if (!parsargv::parse(\@_,
|
|---|
| 144 | q^use_metadata_files^, \$self->{'use_metadata_files'},
|
|---|
| 145 | q^recheck_directories^, \$self->{'recheck_directories'},
|
|---|
| 146 | "allow_extra_options")) {
|
|---|
| 147 | print STDERR "\nRecPlug uses an incorrect option.\n";
|
|---|
| 148 | print STDERR "Check your collect.cfg configuration file.\n\n";
|
|---|
| 149 | $self->print_txt_usage(""); # Use default resource bundle
|
|---|
| 150 | die "\n";
|
|---|
| 151 | }
|
|---|
| 152 |
|
|---|
| 153 | if ($self->{'use_metadata_files'}) {
|
|---|
| 154 | # create XML::Parser object for parsing metadata.xml files
|
|---|
| 155 | my $parser = new XML::Parser('Style' => 'Stream',
|
|---|
| 156 | 'Handlers' => {'Char' => \&Char,
|
|---|
| 157 | 'Doctype' => \&Doctype
|
|---|
| 158 | });
|
|---|
| 159 | $self->{'parser'} = $parser;
|
|---|
| 160 | $self->{'in_filename'} = 0;
|
|---|
| 161 | }
|
|---|
| 162 |
|
|---|
| 163 | $self->{'subdir_extrametakeys'} = {};
|
|---|
| 164 |
|
|---|
| 165 | return bless $self, $class;
|
|---|
| 166 | }
|
|---|
| 167 |
|
|---|
| 168 | sub begin {
|
|---|
| 169 | my $self = shift (@_);
|
|---|
| 170 | my ($pluginfo, $base_dir, $processor, $maxdocs) = @_;
|
|---|
| 171 |
|
|---|
| 172 | my $proc_package_name = ref $processor;
|
|---|
| 173 |
|
|---|
| 174 | if ($proc_package_name !~ /buildproc$/) {
|
|---|
| 175 |
|
|---|
| 176 | # Only lookup timestamp info for import.pl
|
|---|
| 177 |
|
|---|
| 178 | my $output_dir = $processor->getoutputdir();
|
|---|
| 179 | my $archives_inf = &util::filename_cat($output_dir,"archives.inf");
|
|---|
| 180 |
|
|---|
| 181 | if ( -e $archives_inf ) {
|
|---|
| 182 | $self->{'inf_timestamp'} = -M $archives_inf;
|
|---|
| 183 | }
|
|---|
| 184 | }
|
|---|
| 185 |
|
|---|
| 186 | $self->SUPER::begin($pluginfo, $base_dir, $processor, $maxdocs);
|
|---|
| 187 | }
|
|---|
| 188 |
|
|---|
| 189 |
|
|---|
| 190 | # return 1 if this class might recurse using $pluginfo
|
|---|
| 191 | sub is_recursive {
|
|---|
| 192 | my $self = shift (@_);
|
|---|
| 193 |
|
|---|
| 194 | return 1;
|
|---|
| 195 | }
|
|---|
| 196 |
|
|---|
| 197 | sub get_default_block_exp {
|
|---|
| 198 | my $self = shift (@_);
|
|---|
| 199 |
|
|---|
| 200 | return 'CVS';
|
|---|
| 201 | }
|
|---|
| 202 |
|
|---|
| 203 | # return number of files processed, undef if can't process
|
|---|
| 204 | # Note that $base_dir might be "" and that $file might
|
|---|
| 205 | # include directories
|
|---|
| 206 |
|
|---|
| 207 | # This function passes around metadata hash structures. Metadata hash
|
|---|
| 208 | # structures are hashes that map from a (scalar) key (the metadata element
|
|---|
| 209 | # name) to either a scalar metadata value or a reference to an array of
|
|---|
| 210 | # such values.
|
|---|
| 211 |
|
|---|
| 212 | sub read {
|
|---|
| 213 | my $self = shift (@_);
|
|---|
| 214 | my ($pluginfo, $base_dir, $file, $in_metadata, $processor, $maxdocs, $total_count, $gli) = @_;
|
|---|
| 215 |
|
|---|
| 216 | my $outhandle = $self->{'outhandle'};
|
|---|
| 217 | my $verbosity = $self->{'verbosity'};
|
|---|
| 218 | my $read_metadata_files = $self->{'use_metadata_files'};
|
|---|
| 219 |
|
|---|
| 220 | # Calculate the directory name and ensure it is a directory and
|
|---|
| 221 | # that it is not explicitly blocked.
|
|---|
| 222 | my $dirname = $file;
|
|---|
| 223 | $dirname = &util::filename_cat ($base_dir, $file) if $base_dir =~ /\w/;
|
|---|
| 224 | return undef unless (-d $dirname);
|
|---|
| 225 | return 0 if ($self->{'block_exp'} ne "" && $dirname =~ /$self->{'block_exp'}/);
|
|---|
| 226 |
|
|---|
| 227 | # check to make sure we're not reading the archives or index directory
|
|---|
| 228 | my $gsdlhome = quotemeta($ENV{'GSDLHOME'});
|
|---|
| 229 | if ($dirname =~ m/^$gsdlhome\/.*?\/import.*?\/(archives|index)$/) {
|
|---|
| 230 | print $outhandle "RecPlug: $dirname appears to be a reference to a Greenstone collection, skipping.\n";
|
|---|
| 231 | return 0;
|
|---|
| 232 | }
|
|---|
| 233 |
|
|---|
| 234 | # check to see we haven't got a cyclic path...
|
|---|
| 235 | if ($dirname =~ m%(/.*){,41}%) {
|
|---|
| 236 | print $outhandle "RecPlug: $dirname is 40 directories deep, is this a recursive path? if not increase constant in RecPlug.pm.\n";
|
|---|
| 237 | return 0;
|
|---|
| 238 | }
|
|---|
| 239 |
|
|---|
| 240 | # check to see we haven't got a cyclic path...
|
|---|
| 241 | if ($dirname =~ m%.*?import/(.+?)/import/\1.*%) {
|
|---|
| 242 | print $outhandle "RecPlug: $dirname appears to be in a recursive loop...\n";
|
|---|
| 243 | return 0;
|
|---|
| 244 | }
|
|---|
| 245 |
|
|---|
| 246 | if (($verbosity > 2) && ((scalar keys %$in_metadata) > 0)) {
|
|---|
| 247 | print $outhandle "RecPlug: metadata passed in: ",
|
|---|
| 248 | join(", ", keys %$in_metadata), "\n";
|
|---|
| 249 | }
|
|---|
| 250 |
|
|---|
| 251 | # Recur over directory contents.
|
|---|
| 252 | my (@dir, $subfile);
|
|---|
| 253 | my $count = 0;
|
|---|
| 254 |
|
|---|
| 255 | print $outhandle "RecPlug: getting directory $dirname\n" if ($verbosity);
|
|---|
| 256 |
|
|---|
| 257 | # find all the files in the directory
|
|---|
| 258 | if (!opendir (DIR, $dirname)) {
|
|---|
| 259 | if ($gli) {
|
|---|
| 260 | print STDERR "<ProcessingError n='$file' r='Could not read directory $dirname'>\n";
|
|---|
| 261 | }
|
|---|
| 262 | print $outhandle "RecPlug: WARNING - couldn't read directory $dirname\n";
|
|---|
| 263 | return -1; # error in processing
|
|---|
| 264 | }
|
|---|
| 265 | @dir = readdir (DIR);
|
|---|
| 266 | closedir (DIR);
|
|---|
| 267 |
|
|---|
| 268 | # Re-order the files in the list so any directories ending with .all are moved to the end
|
|---|
| 269 | for (my $i = scalar(@dir) - 1; $i >= 0; $i--) {
|
|---|
| 270 | if (-d &util::filename_cat($dirname, $dir[$i]) && $dir[$i] =~ /\.all$/) {
|
|---|
| 271 | push(@dir, splice(@dir, $i, 1));
|
|---|
| 272 | }
|
|---|
| 273 | }
|
|---|
| 274 |
|
|---|
| 275 | # read XML metadata files (if supplied)
|
|---|
| 276 | my $additionalmetadata = 0; # is there extra metadata available?
|
|---|
| 277 | my %extrametadata; # maps from filespec to extra metadata keys
|
|---|
| 278 | my @extrametakeys; # keys of %extrametadata in order read
|
|---|
| 279 |
|
|---|
| 280 | my $dirsepre = &util::get_re_dirsep();
|
|---|
| 281 | my $dirsep = &util::get_dirsep();
|
|---|
| 282 | my $local_dirname = $dirname;
|
|---|
| 283 | $local_dirname =~ s/^$base_dir($dirsepre)//;
|
|---|
| 284 | $local_dirname .= $dirsep;
|
|---|
| 285 |
|
|---|
| 286 | if (defined $self->{'subdir_extrametakeys'}->{$local_dirname}) {
|
|---|
| 287 | my $extrakeys = $self->{'subdir_extrametakeys'}->{$local_dirname};
|
|---|
| 288 | foreach my $ek (@$extrakeys) {
|
|---|
| 289 | my $extrakeys_re = $ek->{'re'};
|
|---|
| 290 | my $extrakeys_md = $ek->{'md'};
|
|---|
| 291 | push(@extrametakeys,$extrakeys_re);
|
|---|
| 292 | $extrametadata{$extrakeys_re} = $extrakeys_md;
|
|---|
| 293 | }
|
|---|
| 294 | delete($self->{'subdir_extrametakeys'}->{$local_dirname});
|
|---|
| 295 | }
|
|---|
| 296 |
|
|---|
| 297 | if ($read_metadata_files) {
|
|---|
| 298 | #read the directory "metadata.xml" file
|
|---|
| 299 | my $metadatafile = &util::filename_cat ($dirname, 'metadata.xml');
|
|---|
| 300 | if (-e $metadatafile) {
|
|---|
| 301 | print $outhandle "RecPlug: found metadata in $metadatafile\n"
|
|---|
| 302 | if ($verbosity);
|
|---|
| 303 | $self->read_metadata_xml_file($metadatafile, \%extrametadata, \@extrametakeys);
|
|---|
| 304 | $additionalmetadata = 1;
|
|---|
| 305 | }
|
|---|
| 306 | }
|
|---|
| 307 |
|
|---|
| 308 | # apply metadata pass for each of the files in the directory
|
|---|
| 309 | my $out_metadata;
|
|---|
| 310 | my $num_files = scalar(@dir);
|
|---|
| 311 | for (my $i = 0; $i < scalar(@dir); $i++) {
|
|---|
| 312 | my $subfile = $dir[$i];
|
|---|
| 313 | my $this_file_base_dir = $base_dir;
|
|---|
| 314 | last if ($maxdocs != -1 && $count >= $maxdocs);
|
|---|
| 315 | next if ($subfile =~ m/^\.\.?$/);
|
|---|
| 316 | #next if ($read_metadata_files && $subfile =~ /metadata\.xml$/);
|
|---|
| 317 |
|
|---|
| 318 | # Recursively read each $subfile
|
|---|
| 319 | print $outhandle "RecPlug metadata recurring: $subfile\n" if ($verbosity > 2);
|
|---|
| 320 |
|
|---|
| 321 | $count += &plugin::metadata_read ($pluginfo, $this_file_base_dir,
|
|---|
| 322 | &util::filename_cat($file, $subfile),
|
|---|
| 323 | $out_metadata, \@extrametakeys, \%extrametadata,
|
|---|
| 324 | $processor, $maxdocs, $gli);
|
|---|
| 325 | $additionalmetadata = 1;
|
|---|
| 326 | }
|
|---|
| 327 |
|
|---|
| 328 | # filter out any extrametakeys that mention subdirectories and store
|
|---|
| 329 | # for later use (i.e. when that sub-directory is being processed)
|
|---|
| 330 |
|
|---|
| 331 | foreach my $ek (@extrametakeys) {
|
|---|
| 332 | my ($subdir_re,$extrakey_dir) = &File::Basename::fileparse($ek);
|
|---|
| 333 | $extrakey_dir =~ s/\\\./\./g; # remove RE syntax
|
|---|
| 334 |
|
|---|
| 335 | my $dirsep_re = &util::get_re_dirsep();
|
|---|
| 336 |
|
|---|
| 337 | if ($ek =~ m/$dirsep_re/) { # specifies at least one directory
|
|---|
| 338 | my $md = $extrametadata{$ek};
|
|---|
| 339 |
|
|---|
| 340 | my $subdir_extrametakeys = $self->{'subdir_extrametakeys'};
|
|---|
| 341 |
|
|---|
| 342 | my $subdir_rec = { 're' => $subdir_re, 'md' => $md };
|
|---|
| 343 | push(@{$subdir_extrametakeys->{$extrakey_dir}},$subdir_rec);
|
|---|
| 344 | }
|
|---|
| 345 | }
|
|---|
| 346 |
|
|---|
| 347 | # import each of the files in the directory
|
|---|
| 348 | $count=0;
|
|---|
| 349 | for (my $i = 0; $i <= scalar(@dir); $i++) {
|
|---|
| 350 | # When every file in the directory has been done, pause for a moment (figuratively!)
|
|---|
| 351 | # If the -recheck_directories argument hasn't been provided, stop now (default)
|
|---|
| 352 | # Otherwise, re-read the contents of the directory to check for new files
|
|---|
| 353 | # Any new files are added to the @dir list and are processed as normal
|
|---|
| 354 | # This is necessary when documents to be indexed are specified in bibliographic DBs
|
|---|
| 355 | # These files are copied/downloaded and stored in a new folder at import time
|
|---|
| 356 | if ($i == $num_files) {
|
|---|
| 357 | last unless $self->{'recheck_directories'};
|
|---|
| 358 |
|
|---|
| 359 | # Re-read the files in the directory to see if there are any new files
|
|---|
| 360 | last if (!opendir (DIR, $dirname));
|
|---|
| 361 | my @dirnow = readdir (DIR);
|
|---|
| 362 | closedir (DIR);
|
|---|
| 363 |
|
|---|
| 364 | # We're only interested if there are more files than there were before
|
|---|
| 365 | last if (scalar(@dirnow) <= scalar(@dir));
|
|---|
| 366 |
|
|---|
| 367 | # Any new files are added to the end of @dir to get processed by the loop
|
|---|
| 368 | my $j;
|
|---|
| 369 | foreach my $subfilenow (@dirnow) {
|
|---|
| 370 | for ($j = 0; $j < $num_files; $j++) {
|
|---|
| 371 | last if ($subfilenow eq $dir[$j]);
|
|---|
| 372 | }
|
|---|
| 373 | if ($j == $num_files) {
|
|---|
| 374 | # New file
|
|---|
| 375 | push(@dir, $subfilenow);
|
|---|
| 376 | }
|
|---|
| 377 | }
|
|---|
| 378 | # When the new files have been processed, check again
|
|---|
| 379 | $num_files = scalar(@dir);
|
|---|
| 380 | }
|
|---|
| 381 |
|
|---|
| 382 | my $subfile = $dir[$i];
|
|---|
| 383 | my $this_file_base_dir = $base_dir;
|
|---|
| 384 | last if ($maxdocs != -1 && ($count + $total_count) >= $maxdocs);
|
|---|
| 385 | next if ($subfile =~ /^\.\.?$/);
|
|---|
| 386 | next if ($read_metadata_files && $subfile =~ /metadata\.xml$/);
|
|---|
| 387 |
|
|---|
| 388 | # Follow Windows shortcuts
|
|---|
| 389 | if ($subfile =~ /(?i)\.lnk$/ && $ENV{'GSDLOS'} =~ /^windows$/i) {
|
|---|
| 390 | require Win32::Shortcut;
|
|---|
| 391 | my $shortcut = new Win32::Shortcut(&util::filename_cat($dirname, $subfile));
|
|---|
| 392 | if ($shortcut) {
|
|---|
| 393 | # The file to be processed is now the target of the shortcut
|
|---|
| 394 | $this_file_base_dir = "";
|
|---|
| 395 | $file = "";
|
|---|
| 396 | $subfile = $shortcut->Path;
|
|---|
| 397 | }
|
|---|
| 398 | }
|
|---|
| 399 |
|
|---|
| 400 | # check for a symlink pointing back to a leading directory
|
|---|
| 401 | if (-d "$dirname/$subfile" && -l "$dirname/$subfile") {
|
|---|
| 402 | # readlink gives a "fatal error" on systems that don't implement
|
|---|
| 403 | # symlinks. This assumes the the -l test above would fail on those.
|
|---|
| 404 | my $linkdest=readlink "$dirname/$subfile";
|
|---|
| 405 | if (!defined ($linkdest)) {
|
|---|
| 406 | # system error - file not found?
|
|---|
| 407 | warn "RecPlug: symlink problem - $!";
|
|---|
| 408 | } else {
|
|---|
| 409 | # see if link points to current or a parent directory
|
|---|
| 410 | if ($linkdest =~ m@^[\./\\]+$@ ||
|
|---|
| 411 | index($dirname, $linkdest) != -1) {
|
|---|
| 412 | warn "RecPlug: Ignoring recursive symlink ($dirname/$subfile -> $linkdest)\n";
|
|---|
| 413 | next;
|
|---|
| 414 | ;
|
|---|
| 415 | }
|
|---|
| 416 | }
|
|---|
| 417 | }
|
|---|
| 418 |
|
|---|
| 419 | print $outhandle "RecPlug: preparing metadata for $subfile\n" if ($verbosity > 2);
|
|---|
| 420 |
|
|---|
| 421 | # Make a copy of $in_metadata to pass to $subfile
|
|---|
| 422 | $out_metadata = {};
|
|---|
| 423 | &combine_metadata_structures($out_metadata, $in_metadata);
|
|---|
| 424 |
|
|---|
| 425 | # Next add metadata read in XML files (if it is supplied)
|
|---|
| 426 | if ($additionalmetadata == 1) {
|
|---|
| 427 | my ($filespec, $mdref);
|
|---|
| 428 | foreach $filespec (@extrametakeys) {
|
|---|
| 429 | if ($subfile =~ /^$filespec$/) {
|
|---|
| 430 | print $outhandle "File \"$subfile\" matches filespec \"$filespec\"\n"
|
|---|
| 431 | if ($verbosity > 2);
|
|---|
| 432 | $mdref = $extrametadata{$filespec};
|
|---|
| 433 | &combine_metadata_structures($out_metadata, $mdref);
|
|---|
| 434 | }
|
|---|
| 435 | }
|
|---|
| 436 | }
|
|---|
| 437 |
|
|---|
| 438 |
|
|---|
| 439 | my $file_subfile = &util::filename_cat($file, $subfile);
|
|---|
| 440 | my $filename_subfile
|
|---|
| 441 | = &util::filename_cat($this_file_base_dir,$file_subfile);
|
|---|
| 442 | if (defined $self->{'inf_timestamp'}) {
|
|---|
| 443 | my $inf_timestamp = $self->{'inf_timestamp'};
|
|---|
| 444 |
|
|---|
| 445 | if (! -d $filename_subfile) {
|
|---|
| 446 | my $filename_timestamp = -M $filename_subfile;
|
|---|
| 447 | if ($filename_timestamp > $inf_timestamp) {
|
|---|
| 448 | # filename has been around for longer than inf
|
|---|
| 449 | ##### print $outhandle "**** Skipping $subfile\n";
|
|---|
| 450 | next;
|
|---|
| 451 | }
|
|---|
| 452 | }
|
|---|
| 453 | }
|
|---|
| 454 |
|
|---|
| 455 | # Recursively read each $subfile
|
|---|
| 456 | print $outhandle "RecPlug recurring: $subfile\n" if ($verbosity > 2);
|
|---|
| 457 |
|
|---|
| 458 | $count += &plugin::read ($pluginfo, $this_file_base_dir,
|
|---|
| 459 | $file_subfile,
|
|---|
| 460 | $out_metadata, $processor, $maxdocs, ($total_count + $count), $gli);
|
|---|
| 461 | }
|
|---|
| 462 |
|
|---|
| 463 | return $count;
|
|---|
| 464 | }
|
|---|
| 465 |
|
|---|
| 466 |
|
|---|
| 467 |
|
|---|
| 468 | # Read a manually-constructed metadata file and store the data
|
|---|
| 469 | # it contains in the $metadataref structure.
|
|---|
| 470 | #
|
|---|
| 471 | # (metadataref is a reference to a hash whose keys are filenames
|
|---|
| 472 | # and whose values are metadata hash structures.)
|
|---|
| 473 |
|
|---|
| 474 | sub read_metadata_xml_file {
|
|---|
| 475 | my $self = shift(@_);
|
|---|
| 476 | my ($filename, $metadataref, $metakeysref) = @_;
|
|---|
| 477 | $self->{'metadataref'} = $metadataref;
|
|---|
| 478 | $self->{'metakeysref'} = $metakeysref;
|
|---|
| 479 |
|
|---|
| 480 | eval {
|
|---|
| 481 | $self->{'parser'}->parsefile($filename);
|
|---|
| 482 | };
|
|---|
| 483 |
|
|---|
| 484 | if ($@) {
|
|---|
| 485 | die "RecPlug: ERROR $filename is not a well formed metadata.xml file ($@)\n";
|
|---|
| 486 | }
|
|---|
| 487 | }
|
|---|
| 488 |
|
|---|
| 489 | sub Doctype {
|
|---|
| 490 | my ($expat, $name, $sysid, $pubid, $internal) = @_;
|
|---|
| 491 |
|
|---|
| 492 | # allow the short-lived and badly named "GreenstoneDirectoryMetadata" files
|
|---|
| 493 | # to be processed as well as the "DirectoryMetadata" files which should now
|
|---|
| 494 | # be created by import.pl
|
|---|
| 495 | die if ($name !~ /^(Greenstone)?DirectoryMetadata$/);
|
|---|
| 496 | }
|
|---|
| 497 |
|
|---|
| 498 | sub StartTag {
|
|---|
| 499 | my ($expat, $element) = @_;
|
|---|
| 500 |
|
|---|
| 501 | if ($element eq "FileSet") {
|
|---|
| 502 | $self->{'saved_targets'} = [];
|
|---|
| 503 | $self->{'saved_metadata'} = {};
|
|---|
| 504 | }
|
|---|
| 505 | elsif ($element eq "FileName") {
|
|---|
| 506 | $self->{'in_filename'} = 1;
|
|---|
| 507 | }
|
|---|
| 508 | elsif ($element eq "Metadata") {
|
|---|
| 509 | $self->{'metadata_name'} = $_{'name'};
|
|---|
| 510 | if ((defined $_{'mode'}) && ($_{'mode'} eq "accumulate")) {
|
|---|
| 511 | $self->{'metadata_accumulate'} = 1;
|
|---|
| 512 | } else {
|
|---|
| 513 | $self->{'metadata_accumulate'} = 0;
|
|---|
| 514 | }
|
|---|
| 515 | }
|
|---|
| 516 | }
|
|---|
| 517 |
|
|---|
| 518 | sub EndTag {
|
|---|
| 519 | my ($expat, $element) = @_;
|
|---|
| 520 |
|
|---|
| 521 | if ($element eq "FileSet") {
|
|---|
| 522 | push (@{$self->{'metakeysref'}}, @{$self->{'saved_targets'}});
|
|---|
| 523 | foreach my $target (@{$self->{'saved_targets'}}) {
|
|---|
| 524 | my $file_metadata = $self->{'metadataref'}->{$target};
|
|---|
| 525 | my $saved_metadata = $self->{'saved_metadata'};
|
|---|
| 526 | if (!defined $file_metadata) {
|
|---|
| 527 | $self->{'metadataref'}->{$target} = $saved_metadata;
|
|---|
| 528 | }
|
|---|
| 529 | else {
|
|---|
| 530 | $self->combine_metadata_structures($file_metadata,$saved_metadata);
|
|---|
| 531 | }
|
|---|
| 532 | }
|
|---|
| 533 | }
|
|---|
| 534 | elsif ($element eq "FileName") {
|
|---|
| 535 | $self->{'in_filename'} = 0;
|
|---|
| 536 | }
|
|---|
| 537 | elsif ($element eq "Metadata") {
|
|---|
| 538 | $self->{'metadata_name'} = "";
|
|---|
| 539 | }
|
|---|
| 540 |
|
|---|
| 541 | }
|
|---|
| 542 |
|
|---|
| 543 | sub store_saved_metadata
|
|---|
| 544 | {
|
|---|
| 545 | my $self = shift(@_);
|
|---|
| 546 | my ($mname,$mvalue,$md_accumulate) = @_;
|
|---|
| 547 |
|
|---|
| 548 | if (defined $self->{'saved_metadata'}->{$mname}) {
|
|---|
| 549 | if ($md_accumulate) {
|
|---|
| 550 | # accumulate mode - add value to existing value(s)
|
|---|
| 551 | if (ref ($self->{'saved_metadata'}->{$mname}) eq "ARRAY") {
|
|---|
| 552 | push (@{$self->{'saved_metadata'}->{$mname}}, $mvalue);
|
|---|
| 553 | } else {
|
|---|
| 554 | $self->{'saved_metadata'}->{$mname} =
|
|---|
| 555 | [$self->{'saved_metadata'}->{$mname}, $mvalue];
|
|---|
| 556 | }
|
|---|
| 557 | } else {
|
|---|
| 558 | # override mode
|
|---|
| 559 | $self->{'saved_metadata'}->{$mname} = $mvalue;
|
|---|
| 560 | }
|
|---|
| 561 | } else {
|
|---|
| 562 | if ($md_accumulate) {
|
|---|
| 563 | # accumulate mode - add value into (currently empty) array
|
|---|
| 564 | $self->{'saved_metadata'}->{$mname} = [$mvalue];
|
|---|
| 565 | } else {
|
|---|
| 566 | # override mode
|
|---|
| 567 | $self->{'saved_metadata'}->{$mname} = $mvalue;
|
|---|
| 568 | }
|
|---|
| 569 | }
|
|---|
| 570 | }
|
|---|
| 571 |
|
|---|
| 572 |
|
|---|
| 573 | sub Text {
|
|---|
| 574 |
|
|---|
| 575 | if ($self->{'in_filename'}) {
|
|---|
| 576 | # $_ == FileName content
|
|---|
| 577 | push (@{$self->{'saved_targets'}}, $_);
|
|---|
| 578 | }
|
|---|
| 579 | elsif (defined ($self->{'metadata_name'}) && $self->{'metadata_name'} ne "") {
|
|---|
| 580 | # $_ == Metadata content
|
|---|
| 581 | my $mname = $self->{'metadata_name'};
|
|---|
| 582 | my $mvalue = $_;
|
|---|
| 583 | my $md_accumulate = $self->{'metadata_accumulate'};
|
|---|
| 584 | $self->store_saved_metadata($mname,$mvalue,$md_accumulate);
|
|---|
| 585 | }
|
|---|
| 586 | }
|
|---|
| 587 |
|
|---|
| 588 | # This Char function overrides the one in XML::Parser::Stream to overcome a
|
|---|
| 589 | # problem where $expat->{Text} is treated as the return value, slowing
|
|---|
| 590 | # things down significantly in some cases.
|
|---|
| 591 | sub Char {
|
|---|
| 592 | use bytes; # Necessary to prevent encoding issues with XML::Parser 2.31+
|
|---|
| 593 | $_[0]->{'Text'} .= $_[1];
|
|---|
| 594 | return undef;
|
|---|
| 595 | }
|
|---|
| 596 |
|
|---|
| 597 | # Combine two metadata structures. Given two references to metadata
|
|---|
| 598 | # element structures, add every field of the second ($mdref2) to the first
|
|---|
| 599 | # ($mdref1).
|
|---|
| 600 | #
|
|---|
| 601 | # Afterwards $mdref1 will be updated, and $mdref2 will be unchanged.
|
|---|
| 602 | #
|
|---|
| 603 | # We have to be careful about the way we merge metadata when one metadata
|
|---|
| 604 | # structure is in "override" mode and one is in "merge" mode. In fact, we
|
|---|
| 605 | # use the mode from the second structure, $mdref2, because it is generally
|
|---|
| 606 | # defined later (lower in the directory structure) and is therefore more
|
|---|
| 607 | # "local" to the document concerned.
|
|---|
| 608 | #
|
|---|
| 609 | # Another issue is the use of references to pass metadata around. If we
|
|---|
| 610 | # simply copy one metadata structure reference to another, then we're
|
|---|
| 611 | # effectively just copyinga pointer, and changes to the new referene
|
|---|
| 612 | # will affect the old (copied) one also. This also applies to ARRAY
|
|---|
| 613 | # references used as metadata element values (hence the "clonedata"
|
|---|
| 614 | # function below).
|
|---|
| 615 |
|
|---|
| 616 | sub combine_metadata_structures {
|
|---|
| 617 | my ($mdref1, $mdref2) = @_;
|
|---|
| 618 | my ($key, $value1, $value2);
|
|---|
| 619 |
|
|---|
| 620 | foreach $key (keys %$mdref2) {
|
|---|
| 621 |
|
|---|
| 622 | $value1 = $mdref1->{$key};
|
|---|
| 623 | $value2 = $mdref2->{$key};
|
|---|
| 624 |
|
|---|
| 625 | # If there is no existing value for this metadata field in
|
|---|
| 626 | # $mdref1, so we simply copy the value from $mdref2 over.
|
|---|
| 627 | if (!defined $value1) {
|
|---|
| 628 | $mdref1->{$key} = &clonedata($value2);
|
|---|
| 629 | }
|
|---|
| 630 | # Otherwise we have to add the new values to the existing ones.
|
|---|
| 631 | # If the second structure is accumulated, then acculate all the
|
|---|
| 632 | # values into the first structure
|
|---|
| 633 | elsif ((ref $value2) eq "ARRAY") {
|
|---|
| 634 | # If the first metadata element is a scalar we have to
|
|---|
| 635 | # convert it into an array before we add anything more.
|
|---|
| 636 | if ((ref $value1) ne 'ARRAY') {
|
|---|
| 637 | $mdref1->{$key} = [$value1];
|
|---|
| 638 | $value1 = $mdref1->{$key};
|
|---|
| 639 | }
|
|---|
| 640 | # Now add the value(s) from the second array to the first
|
|---|
| 641 | $value2 = &clonedata($value2);
|
|---|
| 642 | push @$value1, @$value2;
|
|---|
| 643 | }
|
|---|
| 644 | # Finally, If the second structure is not an array erference, we
|
|---|
| 645 | # know it is in override mode, so override the first structure.
|
|---|
| 646 | else {
|
|---|
| 647 | $mdref1->{$key} = &clonedata($value2);
|
|---|
| 648 | }
|
|---|
| 649 | }
|
|---|
| 650 | }
|
|---|
| 651 |
|
|---|
| 652 |
|
|---|
| 653 | # Make a "cloned" copy of a metadata value.
|
|---|
| 654 | # This is trivial for a simple scalar value,
|
|---|
| 655 | # but not for an array reference.
|
|---|
| 656 |
|
|---|
| 657 | sub clonedata {
|
|---|
| 658 | my ($value) = @_;
|
|---|
| 659 | my $result;
|
|---|
| 660 |
|
|---|
| 661 | if ((ref $value) eq 'ARRAY') {
|
|---|
| 662 | $result = [];
|
|---|
| 663 | foreach my $item (@$value) {
|
|---|
| 664 | push @$result, $item;
|
|---|
| 665 | }
|
|---|
| 666 | } else {
|
|---|
| 667 | $result = $value;
|
|---|
| 668 | }
|
|---|
| 669 | return $result;
|
|---|
| 670 | }
|
|---|
| 671 |
|
|---|
| 672 |
|
|---|
| 673 | 1;
|
|---|
