root/gs2-extensions/parallel-building/trunk/src/perllib/plugins/DirectoryPlugin.pm @ 25814

Revision 25814, 23.5 KB (checked in by jmt12, 8 years ago)

Added extra test to prevent warning when block_hash's manifest varaible not defined

  • Property svn:executable set to *
Line 
1###########################################################################
2#
3# DirectoryPlugin.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# DirectoryPlugin is a plugin which recurses through directories processing
27# each file it finds - which basically means passing it down the plugin
28# pipeline
29
30package DirectoryPlugin;
31
32use PrintInfo;
33use plugin;
34use util;
35use metadatautil;
36
37use File::Basename;
38use strict;
39no strict 'refs';
40no strict 'subs';
41
42use Encode;
43
44BEGIN {
45    @DirectoryPlugin::ISA = ('PrintInfo');
46}
47
48my $arguments =
49    [ { 'name' => "block_exp",
50    'desc' => "{BasePlugin.block_exp}",
51    'type' => "regexp",
52    'deft' => &get_default_block_exp(),
53    'reqd' => "no" },
54      # this option has been deprecated. leave it here for now so we can warn people not to use it
55      { 'name' => "use_metadata_files",
56    'desc' => "{DirectoryPlugin.use_metadata_files}",
57    'type' => "flag",
58    'reqd' => "no",
59    'hiddengli' => "yes" },
60      { 'name' => "recheck_directories",
61    'desc' => "{DirectoryPlugin.recheck_directories}",
62    'type' => "flag",
63    'reqd' => "no" } ];
64   
65my $options = { 'name'     => "DirectoryPlugin",
66        'desc'     => "{DirectoryPlugin.desc}",
67        'abstract' => "no",
68        'inherits' => "yes",
69        'args'     => $arguments };
70
71sub new {
72    my ($class) = shift (@_);
73    my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
74    push(@$pluginlist, $class);
75
76    push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
77    push(@{$hashArgOptLists->{"OptList"}},$options);
78
79    my $self = new PrintInfo($pluginlist, $inputargs, $hashArgOptLists);
80   
81    if ($self->{'info_only'}) {
82    # don't worry about any options or initialisations etc
83    return bless $self, $class;
84    }
85
86    # we have left this option in so we can warn people who are still using it
87    if ($self->{'use_metadata_files'}) {
88    die "ERROR: DirectoryPlugin -use_metadata_files option has been deprecated. Please remove the option and add MetadataXMLPlug to your plugin list instead!\n";
89    }
90   
91    $self->{'num_processed'} = 0;
92    $self->{'num_not_processed'} = 0;
93    $self->{'num_blocked'} = 0;
94    $self->{'num_archives'} = 0;
95
96    $self->{'subdir_extrametakeys'} = {};
97
98    return bless $self, $class;
99}
100
101# called once, at the start of processing
102sub init {
103    my $self = shift (@_);
104    my ($verbosity, $outhandle, $failhandle) = @_;
105
106    # verbosity is passed through from the processor
107    $self->{'verbosity'} = $verbosity;
108
109    # as are the outhandle and failhandle
110    $self->{'outhandle'} = $outhandle if defined $outhandle;
111    $self->{'failhandle'} = $failhandle;
112
113}
114
115# called once, after all passes have finished
116sub deinit {
117    my ($self) = @_;
118
119}
120
121# called at the beginning of each plugin pass (import has one, building has many)
122sub begin {
123    my $self = shift (@_);
124    my ($pluginfo, $base_dir, $processor, $maxdocs) = @_;
125
126    # Only lookup timestamp info for import.pl, and only if incremental is set
127    my $proc_package_name = ref $processor;
128    if ($proc_package_name !~ /buildproc$/ && $self->{'incremental'} == 1) {
129        # Get the infodbtype value for this collection from the arcinfo object
130        my $infodbtype = $processor->getoutputinfo()->{'infodbtype'};
131    $infodbtype = "gdbm" if $infodbtype eq "gdbm-txtgz"; # in archives, cannot use txtgz version
132    my $output_dir = $processor->getoutputdir();
133        my $archives_inf = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $output_dir);
134
135    if ( -e $archives_inf ) {
136        $self->{'inf_timestamp'} = -M $archives_inf;
137    }
138    }
139}
140
141sub remove_all {
142    my $self = shift (@_);
143    my ($pluginfo, $base_dir, $processor, $maxdocs) = @_;
144
145}
146
147
148sub remove_one {
149    my $self = shift (@_);
150    my ($file, $oids, $archivedir) = @_;
151    return undef; # this will never be called for directories (will it??)
152
153}
154
155
156# called at the end of each plugin pass
157sub end {
158    my ($self) = shift (@_);
159
160}
161
162
163
164# return 1 if this class might recurse using $pluginfo
165sub is_recursive {
166    my $self = shift (@_);
167   
168    return 1;
169}
170
171sub get_default_block_exp {
172    my $self = shift (@_);
173   
174    return '(?i)(CVS|\.svn|Thumbs\.db|OIDcount|~)$';
175}
176
177sub check_directory_path {
178
179    my $self = shift(@_);
180    my ($dirname) = @_;
181   
182    return undef unless (-d $dirname);
183
184    return 0 if ($self->{'block_exp'} ne "" && $dirname =~ /$self->{'block_exp'}/);
185
186    my $outhandle = $self->{'outhandle'};
187   
188    # check to make sure we're not reading the archives or index directory
189    my $gsdlhome = quotemeta($ENV{'GSDLHOME'});
190    if ($dirname =~ m/^$gsdlhome\/.*?\/import.*?\/(archives|index)$/) {
191    print $outhandle "DirectoryPlugin: $dirname appears to be a reference to a Greenstone collection, skipping.\n";
192        return 0;
193    }
194   
195    # check to see we haven't got a cyclic path...
196    if ($dirname =~ m%(/.*){,41}%) {
197    print $outhandle "DirectoryPlugin: $dirname is 40 directories deep, is this a recursive path? if not increase constant in DirectoryPlugin.pm.\n";
198    return 0;
199    }
200   
201    # check to see we haven't got a cyclic path...
202    if ($dirname =~ m%.*?import/(.+?)/import/\1.*%) {
203    print $outhandle "DirectoryPlugin: $dirname appears to be in a recursive loop...\n";
204    return 0;
205    }
206
207    return 1;
208}
209
210# this may be called more than once
211sub sort_out_associated_files {
212
213    my $self = shift (@_);
214    my ($block_hash) = @_;
215    if (!scalar (keys %{$block_hash->{'shared_fileroot'}})) {
216    return;
217    }
218
219    $self->{'assocfile_info'} = {} unless defined $self->{'assocfile_info'};
220    my $metadata = $self->{'assocfile_info'};
221    foreach my $prefix (keys %{$block_hash->{'shared_fileroot'}}) {
222    my $record = $block_hash->{'shared_fileroot'}->{$prefix};
223
224    my $tie_to = $record->{'tie_to'};
225    my $exts = $record->{'exts'};
226   
227    if ((defined $tie_to) && (scalar (keys %$exts) > 0)) {
228        # set up fileblocks and assocfile_tobe
229        my $base_file = "$prefix$tie_to";
230        $metadata->{$base_file} = {} unless defined $metadata->{$base_file};
231        my $base_file_metadata = $metadata->{$base_file};
232       
233        $base_file_metadata->{'gsdlassocfile_tobe'} = [] unless defined $base_file_metadata->{'gsdlassocfile_tobe'};
234        my $assoc_tobe = $base_file_metadata->{'gsdlassocfile_tobe'};
235        foreach my $e (keys %$exts) {
236        # block the file
237        &util::block_filename($block_hash,"$prefix$e");
238        # set up as an associatd file
239        print STDERR "  $self->{'plugin_type'}: Associating $prefix$e with $tie_to version\n";
240        my $mime_type = ""; # let system auto detect this
241        push(@$assoc_tobe,"$prefix$e:$mime_type:");
242
243        }
244    }
245    } # foreach record
246
247    $block_hash->{'shared_fileroot'} = undef;
248    $block_hash->{'shared_fileroot'} = {};
249
250}
251
252
253# do block exp OR special blocking ???
254
255sub file_is_blocked {
256    my $self = shift (@_);
257    my ($block_hash, $filename_full_path) = @_;
258
259    $filename_full_path = &util::upgrade_if_dos_filename($filename_full_path);
260###    print STDERR "*** DirectoryPlugin::file_is_blocked $filename_full_path\n";
261
262    if ($ENV{'GSDLOS'} =~ m/^windows$/) {
263    # on windows, all block paths are lowercased.
264    my $lower_filename = lc ($filename_full_path);
265    if (defined $block_hash->{'file_blocks'}->{$lower_filename}) {
266        $self->{'num_blocked'} ++;
267        return 1;
268    }
269    }
270    else {
271    if (defined $block_hash->{'file_blocks'}->{$filename_full_path}) {
272        $self->{'num_blocked'} ++;
273        return 1;
274    }
275    }
276    # check Directory plugin's own block_exp
277    if ($self->{'block_exp'} ne "" && $filename_full_path =~ /$self->{'block_exp'}/) {
278    $self->{'num_blocked'} ++;
279    return 1; # blocked
280    }
281    return 0;
282}
283
284
285
286sub file_block_read {
287    my $self = shift (@_);
288    my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $gli) = @_;
289
290    my $outhandle = $self->{'outhandle'};
291    my $verbosity = $self->{'verbosity'};
292   
293    # Calculate the directory name and ensure it is a directory and
294    # that it is not explicitly blocked.
295    my $dirname = $file;
296    $dirname = &util::filename_cat ($base_dir, $file) if $base_dir =~ /\w/;
297
298    my $directory_ok = $self->check_directory_path($dirname);
299    return $directory_ok unless (defined $directory_ok && $directory_ok == 1);
300
301    print $outhandle "Global file scan checking directory: $dirname\n" if ($verbosity > 0);
302
303    $block_hash->{'all_files'} = {} unless defined $block_hash->{'all_files'};
304    $block_hash->{'metadata_files'} = {} unless defined $block_hash->{'metadata_files'};
305
306    $block_hash->{'file_blocks'} = {} unless defined $block_hash->{'file_blocks'};
307    $block_hash->{'shared_fileroot'} = {} unless defined $block_hash->{'shared_fileroot'};
308
309     # Recur over directory contents.
310    my (@dir, $subfile);
311    #my $count = 0;
312   
313    print $outhandle "DirectoryPlugin block: getting directory $dirname\n" if ($verbosity > 2);
314   
315    # find all the files in the directory
316    if (!opendir (DIR, $dirname)) {
317    if ($gli) {
318        print STDERR "<ProcessingError n='$file' r='Could not read directory $dirname'>\n";
319    }
320    print $outhandle "DirectoryPlugin: WARNING - couldn't read directory $dirname\n";
321    return -1; # error in processing
322    }
323    @dir = readdir (DIR);
324    closedir (DIR);
325   
326    for (my $i = 0; $i < scalar(@dir); $i++) {
327    my $raw_subfile = $dir[$i];
328    next if ($raw_subfile =~ m/^\.\.?$/);
329
330    my $this_file_base_dir = $base_dir;
331    my $raw_file_subfile = &util::filename_cat($file, $raw_subfile);
332
333    # Recursively read each $raw_subfile
334    print $outhandle "DirectoryPlugin block recurring: $raw_file_subfile\n" if ($verbosity > 2);
335   
336    #$count += &plugin::file_block_read ($pluginfo, $this_file_base_dir,
337
338    &plugin::file_block_read ($pluginfo, $this_file_base_dir,
339                  $raw_file_subfile,
340                  $block_hash, $metadata, $gli);
341   
342    }
343    $self->sort_out_associated_files($block_hash);
344    #return $count;
345    return 1;
346   
347}
348
349# We don't do metadata_read
350sub metadata_read {
351    my $self = shift (@_);
352    my ($pluginfo, $base_dir, $file, $block_hash,
353    $extrametakeys, $extrametadata, $extrametafile,
354    $processor, $gli, $aux) = @_;
355
356    return undef;
357}
358
359
360# return number of files processed, undef if can't process
361# Note that $base_dir might be "" and that $file might
362# include directories
363
364# This function passes around metadata hash structures.  Metadata hash
365# structures are hashes that map from a (scalar) key (the metadata element
366# name) to either a scalar metadata value or a reference to an array of
367# such values.
368
369sub read {
370    my $self = shift (@_);
371    my ($pluginfo, $base_dir, $file, $block_hash, $in_metadata, $processor, $maxdocs, $total_count, $gli) = @_;
372
373    my $outhandle = $self->{'outhandle'};
374    my $verbosity = $self->{'verbosity'};
375   
376    # Calculate the directory name and ensure it is a directory and
377    # that it is not explicitly blocked.
378    my $dirname;
379    if ($file eq "") {
380    $dirname = $base_dir;
381    } else {
382    $dirname = $file;
383    $dirname = &util::filename_cat ($base_dir, $file) if $base_dir =~ /\w/;
384    }
385
386    my $directory_ok = $self->check_directory_path($dirname);
387    return $directory_ok unless (defined $directory_ok && $directory_ok == 1);
388       
389    if (($verbosity > 2) && ((scalar keys %$in_metadata) > 0)) {
390        print $outhandle "DirectoryPlugin: metadata passed in: ",
391    join(", ", keys %$in_metadata), "\n";
392    }
393   
394
395    # Recur over directory contents.
396    my (@dir, $subfile);
397   
398    print $outhandle "DirectoryPlugin read: getting directory $dirname\n" if ($verbosity > 2);
399   
400    # find all the files in the directory
401    if (!opendir (DIR, $dirname)) {
402    if ($gli) {
403        print STDERR "<ProcessingError n='$file' r='Could not read directory $dirname'>\n";
404    }
405    print $outhandle "DirectoryPlugin: WARNING - couldn't read directory $dirname\n";
406    return -1; # error in processing
407    }
408    @dir = readdir (DIR);
409    map { $_ = &unicode::raw_filename_to_url_encoded($_) } @dir;
410    closedir (DIR);
411
412    # Re-order the files in the list so any directories ending with .all are moved to the end
413    for (my $i = scalar(@dir) - 1; $i >= 0; $i--) {
414    if (-d &util::filename_cat($dirname, $dir[$i]) && $dir[$i] =~ /\.all$/) {
415        push(@dir, splice(@dir, $i, 1));
416    }
417    }
418
419    # setup the metadata structures. we do a metadata_read pass to see if there is any additional metadata, then pass it to read
420   
421    my $additionalmetadata = 0;      # is there extra metadata available?
422    my %extrametadata;               # maps from filespec to extra metadata keys
423    my %extrametafile;               # maps from filespec to the metadata.xml (or similar) file it came from
424    my @extrametakeys;               # keys of %extrametadata in order read
425
426
427    my $os_dirsep = &util::get_os_dirsep();
428    my $dirsep    = &util::get_dirsep();
429    my $base_dir_regexp = $base_dir;
430    $base_dir_regexp =~ s/\//$os_dirsep/g;
431    my $local_dirname = $dirname;
432   
433    $local_dirname =~ s/^$base_dir_regexp($os_dirsep)*//;
434    # if we are in import folder, then local_dirname will be empty
435    if ($local_dirname ne "") {
436    # look for extra metadata passed down from higher folders
437    $local_dirname .= $dirsep;
438    if (defined $self->{'subdir_extrametakeys'}->{$local_dirname}) {
439        my $extrakeys = $self->{'subdir_extrametakeys'}->{$local_dirname};
440        foreach my $ek (@$extrakeys) {
441        my $extrakeys_re  = $ek->{'re'};
442        my $extrakeys_md  = $ek->{'md'};
443        my $extrakeys_mf  = $ek->{'mf'};
444        push(@extrametakeys,$extrakeys_re);
445        $extrametadata{$extrakeys_re} = $extrakeys_md;
446        $extrametafile{$extrakeys_re} = $extrakeys_mf;
447        }
448        delete($self->{'subdir_extrametakeys'}->{$local_dirname});
449    }
450    }
451    # apply metadata pass for each of the files in the directory -- ignore
452    # maxdocs here
453    my $num_files = scalar(@dir);
454    for (my $i = 0; $i < scalar(@dir); $i++) {
455    my $subfile = $dir[$i];
456    next if ($subfile =~ m/^\.\.?$/);
457
458    my $this_file_base_dir = $base_dir;
459    my $raw_subfile = &unicode::url_encoded_to_raw_filename($subfile);
460
461    my $raw_file_subfile = &util::filename_cat($file, $raw_subfile);
462    my $raw_full_filename = &util::filename_cat($this_file_base_dir, $raw_file_subfile);
463
464    if ($self->file_is_blocked($block_hash,$raw_full_filename)) {
465        print STDERR "DirectoryPlugin: file $raw_full_filename was blocked for metadata_read\n" if ($verbosity > 2);
466        next;
467    }
468
469        # If we are processing a manifest file, then we don't have to guess at
470        # what metadata source files to process. Only process those that appear
471        # in the metadata_files list in the block_hash. [jmt12]
472        if (defined $block_hash && defined $block_hash->{'manifest'} && $block_hash->{'manifest'} eq 'true'
473            && !-d $raw_full_filename
474            && !$block_hash->{'metadata_files'}->{$raw_full_filename})
475        {
476          print $outhandle "**** Not in manifest. Ignoring metadata in $subfile\n" if ($verbosity > 3);
477          next;
478        }
479
480   
481    # Recursively read each $raw_subfile
482    print $outhandle "DirectoryPlugin metadata recurring: $raw_subfile\n" if ($verbosity > 2);
483   
484    &plugin::metadata_read ($pluginfo, $this_file_base_dir,
485                $raw_file_subfile,$block_hash,
486                \@extrametakeys, \%extrametadata,
487                \%extrametafile,
488                $processor, $gli);
489    $additionalmetadata = 1;
490    }
491
492    # filter out any extrametakeys that mention subdirectories and store
493    # for later use (i.e. when that sub-directory is being processed)
494    foreach my $ek (@extrametakeys) {
495    my ($subdir_re,$extrakey_dir) = &File::Basename::fileparse($ek);
496    $extrakey_dir =~ s/\\\./\./g; # remove RE syntax for .
497    $extrakey_dir =~ s/\\\\/\\/g; # remove RE syntax for \
498
499    my $dirsep_re = &util::get_re_dirsep();
500
501    my $ek_non_re = $ek;
502    $ek_non_re =~ s/\\\./\./g; # remove RE syntax for .
503    $ek_non_re =~ s/\\\\/\\/g; # remove RE syntax for \
504    if ($ek_non_re =~ m/$dirsep_re/) { # specifies at least one directory
505        my $md = $extrametadata{$ek};
506        my $mf = $extrametafile{$ek};
507
508        my $subdir_extrametakeys = $self->{'subdir_extrametakeys'};
509
510        my $subdir_rec = { 're' => $subdir_re, 'md' => $md, 'mf' => $mf };
511
512        # when its looked up, it must be relative to the base dir
513        push(@{$subdir_extrametakeys->{"$local_dirname$extrakey_dir"}},$subdir_rec);
514        #push(@{$subdir_extrametakeys->{"$extrakey_dir"}},$subdir_rec);
515    }
516    }
517   
518    # import each of the files in the directory
519    my $count=0;
520    for (my $i = 0; $i <= scalar(@dir); $i++) {
521    # When every file in the directory has been done, pause for a moment (figuratively!)
522    # If the -recheck_directories argument hasn't been provided, stop now (default)
523    # Otherwise, re-read the contents of the directory to check for new files
524    #   Any new files are added to the @dir list and are processed as normal
525    #   This is necessary when documents to be indexed are specified in bibliographic DBs
526    #   These files are copied/downloaded and stored in a new folder at import time
527    if ($i == $num_files) {
528        last unless $self->{'recheck_directories'};
529
530        # Re-read the files in the directory to see if there are any new files
531        last if (!opendir (DIR, $dirname));
532        my @dirnow = readdir (DIR);
533        map { $_ = &unicode::raw_filename_to_url_encoded($_) } @dirnow;
534        closedir (DIR);
535
536        # We're only interested if there are more files than there were before
537        last if (scalar(@dirnow) <= scalar(@dir));
538
539        # Any new files are added to the end of @dir to get processed by the loop
540        my $j;
541        foreach my $subfilenow (@dirnow) {
542        for ($j = 0; $j < $num_files; $j++) {
543            last if ($subfilenow eq $dir[$j]);
544        }
545        if ($j == $num_files) {
546            # New file
547            push(@dir, $subfilenow);
548        }
549        }
550        # When the new files have been processed, check again
551        $num_files = scalar(@dir);
552    }
553
554    my $subfile = $dir[$i];
555    last if ($maxdocs != -1 && ($count + $total_count) >= $maxdocs);
556    next if ($subfile =~ /^\.\.?$/);
557
558    my $this_file_base_dir = $base_dir;
559    my $raw_subfile = &unicode::url_encoded_to_raw_filename($subfile);
560
561    my $raw_file_subfile = &util::filename_cat($file, $raw_subfile);
562    my $raw_full_filename
563        = &util::filename_cat($this_file_base_dir,$raw_file_subfile);
564
565    if ($self->file_is_blocked($block_hash,$raw_full_filename)) {
566        print STDERR "DirectoryPlugin: file $raw_full_filename was blocked for read\n" if ($verbosity > 2);
567        next;
568    }
569    #print STDERR "processing $raw_full_filename\n";
570    # Follow Windows shortcuts
571    if ($raw_subfile =~ /(?i)\.lnk$/ && $ENV{'GSDLOS'} =~ /^windows$/i) {
572        require Win32::Shortcut;
573        my $shortcut = new Win32::Shortcut(&util::filename_cat($dirname, $raw_subfile));
574        if ($shortcut) {
575        # The file to be processed is now the target of the shortcut
576        $this_file_base_dir = "";
577        $file = "";
578        $raw_subfile = $shortcut->Path;
579        }
580    }
581
582    # check for a symlink pointing back to a leading directory
583    if (-d "$dirname/$raw_subfile" && -l "$dirname/$raw_subfile") {
584        # readlink gives a "fatal error" on systems that don't implement
585        # symlinks. This assumes the the -l test above would fail on those.
586        my $linkdest=readlink "$dirname/$raw_subfile";
587        if (!defined ($linkdest)) {
588        # system error - file not found?
589        warn "DirectoryPlugin: symlink problem - $!";
590        } else {
591        # see if link points to current or a parent directory
592        if ($linkdest =~ m@^[\./\\]+$@ ||
593            index($dirname, $linkdest) != -1) {
594            warn "DirectoryPlugin: Ignoring recursive symlink ($dirname/$raw_subfile -> $linkdest)\n";
595            next;
596            ;
597        }
598        }
599    }
600
601    print $outhandle "DirectoryPlugin: preparing metadata for $raw_subfile\n" if ($verbosity > 2);
602
603    # Make a copy of $in_metadata to pass to $raw_subfile
604    my $out_metadata = {};
605    &metadatautil::combine_metadata_structures($out_metadata, $in_metadata);
606
607    # check the assocfile_info
608    if (defined $self->{'assocfile_info'}->{$raw_full_filename}) {
609        &metadatautil::combine_metadata_structures($out_metadata, $self->{'assocfile_info'}->{$raw_full_filename});
610    }
611
612    # $subfile by this point is url-encoded => all ASCII chars => no need to encode as UTF8
613
614    # Next add metadata read in XML files (if it is supplied)
615    if ($additionalmetadata == 1) {
616        foreach my $filespec (@extrametakeys) {
617        ## use the url-encoded filename to do the filename comparison
618
619        if ($subfile =~ /^$filespec$/) {
620            print $outhandle "File \"$subfile\" matches filespec \"$filespec\"\n"
621            if ($verbosity > 2);
622            my $mdref = $extrametadata{$filespec};
623            my $mfref = $extrametafile{$filespec};
624
625            # Add the list files where the metadata came from
626            # into the metadata table so we can track this
627            # This mechanism is similar to how gsdlassocfile works
628
629            my @metafile_pair = ();
630            foreach my $l (keys %$mfref) {
631            my $f = $mfref->{$l};
632            push (@metafile_pair, "$f : $l");
633            }
634
635            $mdref->{'gsdlmetafile'} = \@metafile_pair;
636
637            &metadatautil::combine_metadata_structures($out_metadata, $mdref);
638        }
639        }
640    }
641
642        # So, here's my problem. I have an empty archives directory, and a
643        # manifest file that says to add 1 out of 2 documents in the import
644        # directory. The directory scan finds both files, and doc 1 is in
645        # new_files, so that is all good. But doc 2 is also processed, and
646        # without any blocking or other rule, also gets imported.
647        # If we are processing a Manifest then files (not docs) that aren't in
648        # either the new files or reindex files are skipped.
649        # We'll also test for metadata files here, even though they should
650        # have been handled above - just in case we have the weird case where
651        # a file is both a metadata source and a file in its own right. [jmt12]
652        if (defined $block_hash
653            && defined $block_hash->{'manifest'}
654            && $block_hash->{'manifest'} eq 'true'
655            && !-d $raw_full_filename
656            && !$block_hash->{'new_files'}->{$raw_full_filename}
657            && !$block_hash->{'reindex_files'}->{$raw_full_filename}
658            && !$block_hash->{'metadata_files'}->{$raw_full_filename})
659        {
660          print $outhandle "**** Not in manifest. Skipping $subfile\n" if ($verbosity > 3);
661          next;
662        }
663
664    if (defined $self->{'inf_timestamp'}) {
665        # Look to see if it's a completely new file
666
667        if (!$block_hash->{'new_files'}->{$raw_full_filename}) {
668        # Not a new file, must be an existing file
669        # Let' see if it's newer than the last import.pl
670
671
672        if (! -d $raw_full_filename) {
673            if (!$block_hash->{'reindex_files'}->{$raw_full_filename}) {
674            # filename has been around for longer than inf_timestamp
675            print $outhandle "**** Skipping $subfile\n" if ($verbosity >3);
676            next;
677            }
678            else {
679            # Remove old folder in archives (might hash to something different)
680            # *** should be doing this on a Del one as well
681            # but leave folder name?? and ensure hashs to
682            # same again??
683
684            # Then let through as new doc??
685
686            # mark to doc-oids that rely on it for re-indexing
687            }
688        }
689        }
690        }
691
692    # Recursively read each $subfile
693    print $outhandle "DirectoryPlugin recurring: $subfile\n" if ($verbosity > 2);
694   
695    $count += &plugin::read ($pluginfo, $this_file_base_dir,
696                 $raw_file_subfile, $block_hash,
697                 $out_metadata, $processor, $maxdocs, ($total_count + $count), $gli);
698    }
699
700    return $count;
701}
702
703sub compile_stats {
704    my $self = shift(@_);
705    my ($stats) = @_;
706}
707
7081;
Note: See TracBrowser for help on using the browser.