root/main/trunk/greenstone2/perllib/plugins/DirectoryPlugin.pm @ 21308

Revision 21308, 20.5 KB (checked in by kjdon, 10 years ago)

removeold renamed to remove_all, added remove_one for deleting a single doc

  • Property svn:executable set to *
  • Property svn:keywords set to Author Date Id Revision
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    my $proc_package_name = ref $processor;
127
128    if ($proc_package_name !~ /buildproc$/ && $self->{'incremental'} == 1) {
129
130    # Only lookup timestamp info for import.pl, and only if incremental is set
131
132    my $output_dir = $processor->getoutputdir();
133##  my $archives_inf = &util::filename_cat($output_dir,"archives.inf");
134    my $doc_db = "archiveinf-doc.gdb";
135        my $archives_inf = &util::filename_cat($output_dir,$doc_db);
136
137    if ( -e $archives_inf ) {
138        $self->{'inf_timestamp'} = -M $archives_inf;
139    }
140    }
141}
142
143sub remove_all {
144    my $self = shift (@_);
145    my ($pluginfo, $base_dir, $processor, $maxdocs) = @_;
146
147}
148
149sub remove_one {
150    my $self = shift (@_);
151    my ($file, $oids) = @_;
152   
153}
154
155
156# called at the end of each plugin pass
157sub end {
158    my ($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        $block_hash->{'file_blocks'}->{"$prefix$e"} = 1;
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    if (defined $block_hash->{'file_blocks'}->{$filename_full_path}) {
260    $self->{'num_blocked'} ++;
261    return 1;
262    }
263    # check Directory plugin's own block_exp
264    if ($self->{'block_exp'} ne "" && $filename_full_path =~ /$self->{'block_exp'}/) {
265    $self->{'num_blocked'} ++;
266    return 1; # blocked
267    }
268    return 0;
269}
270
271
272
273sub file_block_read {
274    my $self = shift (@_);
275    my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $gli) = @_;
276
277    my $outhandle = $self->{'outhandle'};
278    my $verbosity = $self->{'verbosity'};
279   
280    # Calculate the directory name and ensure it is a directory and
281    # that it is not explicitly blocked.
282    my $dirname = $file;
283    $dirname = &util::filename_cat ($base_dir, $file) if $base_dir =~ /\w/;
284
285    my $directory_ok = $self->check_directory_path($dirname);
286    return $directory_ok unless (defined $directory_ok && $directory_ok == 1);
287
288    print $outhandle "Global file scan checking directory: $dirname\n";
289
290    $block_hash->{'all_files'} = {} unless defined $block_hash->{'all_files'};
291    $block_hash->{'metadata_files'} = {} unless defined $block_hash->{'metadata_files'};
292
293    $block_hash->{'file_blocks'} = {} unless defined $block_hash->{'file_blocks'};
294    $block_hash->{'shared_fileroot'} = {} unless defined $block_hash->{'shared_fileroot'};
295
296     # Recur over directory contents.
297    my (@dir, $subfile);
298    #my $count = 0;
299   
300    print $outhandle "DirectoryPlugin block: getting directory $dirname\n" if ($verbosity > 2);
301   
302    # find all the files in the directory
303    if (!opendir (DIR, $dirname)) {
304    if ($gli) {
305        print STDERR "<ProcessingError n='$file' r='Could not read directory $dirname'>\n";
306    }
307    print $outhandle "DirectoryPlugin: WARNING - couldn't read directory $dirname\n";
308    return -1; # error in processing
309    }
310    @dir = readdir (DIR);
311    closedir (DIR);
312   
313    for (my $i = 0; $i < scalar(@dir); $i++) {
314    my $subfile = $dir[$i];
315    my $this_file_base_dir = $base_dir;
316    next if ($subfile =~ m/^\.\.?$/);
317
318    # Recursively read each $subfile
319    print $outhandle "DirectoryPlugin block recurring: $subfile\n" if ($verbosity > 2);
320   
321    #$count += &plugin::file_block_read ($pluginfo, $this_file_base_dir,
322    &plugin::file_block_read ($pluginfo, $this_file_base_dir,
323                  &util::filename_cat($file, $subfile),
324                  $block_hash, $metadata, $gli);
325   
326    }
327    $self->sort_out_associated_files($block_hash);
328    #return $count;
329   
330}
331
332# We don't do metadata_read
333sub metadata_read {
334    my $self = shift (@_);
335    my ($pluginfo, $base_dir, $file, $block_hash,
336    $extrametakeys, $extrametadata, $extrametafile,
337    $processor, $maxdocs, $gli) = @_;
338
339    return undef;
340}
341
342
343# return number of files processed, undef if can't process
344# Note that $base_dir might be "" and that $file might
345# include directories
346
347# This function passes around metadata hash structures.  Metadata hash
348# structures are hashes that map from a (scalar) key (the metadata element
349# name) to either a scalar metadata value or a reference to an array of
350# such values.
351
352sub read {
353    my $self = shift (@_);
354    my ($pluginfo, $base_dir, $file, $block_hash, $in_metadata, $processor, $maxdocs, $total_count, $gli) = @_;
355
356    my $outhandle = $self->{'outhandle'};
357    my $verbosity = $self->{'verbosity'};
358   
359    # Calculate the directory name and ensure it is a directory and
360    # that it is not explicitly blocked.
361    my $dirname;
362    if ($file eq "") {
363    $dirname = $base_dir;
364    } else {
365    $dirname = $file;
366    $dirname = &util::filename_cat ($base_dir, $file) if $base_dir =~ /\w/;
367    }
368
369    my $directory_ok = $self->check_directory_path($dirname);
370    return $directory_ok unless (defined $directory_ok && $directory_ok == 1);
371       
372    if (($verbosity > 2) && ((scalar keys %$in_metadata) > 0)) {
373        print $outhandle "DirectoryPlugin: metadata passed in: ",
374    join(", ", keys %$in_metadata), "\n";
375    }
376   
377
378    # Recur over directory contents.
379    my (@dir, $subfile);
380    my $count = 0;
381   
382    print $outhandle "DirectoryPlugin read: getting directory $dirname\n" if ($verbosity > 2);
383   
384    # find all the files in the directory
385    if (!opendir (DIR, $dirname)) {
386    if ($gli) {
387        print STDERR "<ProcessingError n='$file' r='Could not read directory $dirname'>\n";
388    }
389    print $outhandle "DirectoryPlugin: WARNING - couldn't read directory $dirname\n";
390    return -1; # error in processing
391    }
392    @dir = readdir (DIR);
393    closedir (DIR);
394
395    # Re-order the files in the list so any directories ending with .all are moved to the end
396    for (my $i = scalar(@dir) - 1; $i >= 0; $i--) {
397    if (-d &util::filename_cat($dirname, $dir[$i]) && $dir[$i] =~ /\.all$/) {
398        push(@dir, splice(@dir, $i, 1));
399    }
400    }
401
402    # setup the metadata structures. we do a metadata_read pass to see if there is any additional metadata, then pass it to read
403   
404    my $additionalmetadata = 0;      # is there extra metadata available?
405    my %extrametadata;               # maps from filespec to extra metadata keys
406    my %extrametafile;               # maps from filespec to the metadata.xml (or similar) file it came from
407    my @extrametakeys;               # keys of %extrametadata in order read
408
409
410    my $os_dirsep = &util::get_os_dirsep();
411    my $dirsep    = &util::get_dirsep();
412    my $base_dir_regexp = $base_dir;
413    $base_dir_regexp =~ s/\//$os_dirsep/g;
414    my $local_dirname = $dirname;
415    $local_dirname =~ s/^$base_dir_regexp($os_dirsep)//;
416    $local_dirname .= $dirsep;
417   
418    if (defined $self->{'subdir_extrametakeys'}->{$local_dirname}) {
419    my $extrakeys = $self->{'subdir_extrametakeys'}->{$local_dirname};
420    foreach my $ek (@$extrakeys) {
421        my $extrakeys_re  = $ek->{'re'};
422        my $extrakeys_md  = $ek->{'md'};
423        my $extrakeys_mf  = $ek->{'mf'};
424        push(@extrametakeys,$extrakeys_re);
425        $extrametadata{$extrakeys_re} = $extrakeys_md;
426        $extrametafile{$extrakeys_re} = $extrakeys_mf;
427    }
428    delete($self->{'subdir_extrametakeys'}->{$local_dirname});
429    }
430   
431    # apply metadata pass for each of the files in the directory
432    my $num_files = scalar(@dir);
433    for (my $i = 0; $i < scalar(@dir); $i++) {
434    my $subfile = $dir[$i];
435    my $this_file_base_dir = $base_dir;
436    last if ($maxdocs != -1 && $count >= $maxdocs);
437    next if ($subfile =~ m/^\.\.?$/);
438    my $file_subfile = &util::filename_cat($file, $subfile);
439    my $full_filename = &util::filename_cat($this_file_base_dir, $file_subfile);
440    if ($self->file_is_blocked($block_hash,$full_filename)) {
441        print STDERR "DirectoryPlugin: file $full_filename was blocked for metadata_read\n" if ($verbosity > 2);
442        next;
443    }
444   
445    # Recursively read each $subfile
446    print $outhandle "DirectoryPlugin metadata recurring: $subfile\n" if ($verbosity > 2);
447   
448    $count += &plugin::metadata_read ($pluginfo, $this_file_base_dir,
449                      $file_subfile,$block_hash,
450                      \@extrametakeys, \%extrametadata,
451                      \%extrametafile,
452                      $processor, $maxdocs, $gli);
453    $additionalmetadata = 1;
454    }
455
456    # filter out any extrametakeys that mention subdirectories and store
457    # for later use (i.e. when that sub-directory is being processed)
458
459    foreach my $ek (@extrametakeys) {
460    my ($subdir_re,$extrakey_dir) = &File::Basename::fileparse($ek);
461   
462    $extrakey_dir =~ s/\\\./\./g; # remove RE syntax for .
463    $extrakey_dir =~ s/\\\\/\\/g; # remove RE syntax for \
464
465    my $dirsep_re = &util::get_re_dirsep();
466
467    my $ek_non_re = $ek;
468    $ek_non_re =~ s/\\\./\./g; # remove RE syntax for .
469    $ek_non_re =~ s/\\\\/\\/g; # remove RE syntax for \
470    if ($ek_non_re =~ m/$dirsep_re/) { # specifies at least one directory
471        my $md = $extrametadata{$ek};
472        my $mf = $extrametafile{$ek};
473
474        my $subdir_extrametakeys = $self->{'subdir_extrametakeys'};
475
476        my $subdir_rec = { 're' => $subdir_re, 'md' => $md, 'mf' => $mf };
477
478        # when its looked up, it must be relative to the base dir
479        push(@{$subdir_extrametakeys->{"$local_dirname$extrakey_dir"}},$subdir_rec);
480        #push(@{$subdir_extrametakeys->{"$extrakey_dir"}},$subdir_rec);
481    }
482    }
483   
484    # import each of the files in the directory
485    $count=0;
486    for (my $i = 0; $i <= scalar(@dir); $i++) {
487    # When every file in the directory has been done, pause for a moment (figuratively!)
488    # If the -recheck_directories argument hasn't been provided, stop now (default)
489    # Otherwise, re-read the contents of the directory to check for new files
490    #   Any new files are added to the @dir list and are processed as normal
491    #   This is necessary when documents to be indexed are specified in bibliographic DBs
492    #   These files are copied/downloaded and stored in a new folder at import time
493    if ($i == $num_files) {
494        last unless $self->{'recheck_directories'};
495
496        # Re-read the files in the directory to see if there are any new files
497        last if (!opendir (DIR, $dirname));
498        my @dirnow = readdir (DIR);
499        closedir (DIR);
500
501        # We're only interested if there are more files than there were before
502        last if (scalar(@dirnow) <= scalar(@dir));
503
504        # Any new files are added to the end of @dir to get processed by the loop
505        my $j;
506        foreach my $subfilenow (@dirnow) {
507        for ($j = 0; $j < $num_files; $j++) {
508            last if ($subfilenow eq $dir[$j]);
509        }
510        if ($j == $num_files) {
511            # New file
512            push(@dir, $subfilenow);
513        }
514        }
515        # When the new files have been processed, check again
516        $num_files = scalar(@dir);
517    }
518
519    my $subfile = $dir[$i];
520    my $this_file_base_dir = $base_dir;
521    last if ($maxdocs != -1 && ($count + $total_count) >= $maxdocs);
522    next if ($subfile =~ /^\.\.?$/);
523
524    my $file_subfile = &util::filename_cat($file, $subfile);
525    my $full_filename
526        = &util::filename_cat($this_file_base_dir,$file_subfile);
527
528    if ($self->file_is_blocked($block_hash,$full_filename)) {
529        print STDERR "DirectoryPlugin: file $full_filename was blocked for read\n" if ($verbosity > 2);
530        next;
531    }
532    #print STDERR "processing $full_filename\n";
533    # Follow Windows shortcuts
534    if ($subfile =~ /(?i)\.lnk$/ && $ENV{'GSDLOS'} =~ /^windows$/i) {
535        require Win32::Shortcut;
536        my $shortcut = new Win32::Shortcut(&util::filename_cat($dirname, $subfile));
537        if ($shortcut) {
538        # The file to be processed is now the target of the shortcut
539        $this_file_base_dir = "";
540        $file = "";
541        $subfile = $shortcut->Path;
542        }
543    }
544
545    # check for a symlink pointing back to a leading directory
546    if (-d "$dirname/$subfile" && -l "$dirname/$subfile") {
547        # readlink gives a "fatal error" on systems that don't implement
548        # symlinks. This assumes the the -l test above would fail on those.
549        my $linkdest=readlink "$dirname/$subfile";
550        if (!defined ($linkdest)) {
551        # system error - file not found?
552        warn "DirectoryPlugin: symlink problem - $!";
553        } else {
554        # see if link points to current or a parent directory
555        if ($linkdest =~ m@^[\./\\]+$@ ||
556            index($dirname, $linkdest) != -1) {
557            warn "DirectoryPlugin: Ignoring recursive symlink ($dirname/$subfile -> $linkdest)\n";
558            next;
559            ;
560        }
561        }
562    }
563
564    print $outhandle "DirectoryPlugin: preparing metadata for $subfile\n" if ($verbosity > 2);
565
566    # Make a copy of $in_metadata to pass to $subfile
567    my $out_metadata = {};
568    &metadatautil::combine_metadata_structures($out_metadata, $in_metadata);
569
570    # check the assocfile_info
571    if (defined $self->{'assocfile_info'}->{$full_filename}) {
572        &metadatautil::combine_metadata_structures($out_metadata, $self->{'assocfile_info'}->{$full_filename});
573    }
574        ## encode the filename as perl5 doesn't handle unicode filenames   
575       
576        my $tmpfile = Encode::encode_utf8($subfile);
577    # Next add metadata read in XML files (if it is supplied)
578    if ($additionalmetadata == 1) {
579        foreach my $filespec (@extrametakeys) {
580       
581        ## use the utf8 encoded filename to do the filename comparison
582        if ($tmpfile =~ /^$filespec$/) {
583            print $outhandle "File \"$subfile\" matches filespec \"$filespec\"\n"
584            if ($verbosity > 2);
585            my $mdref = $extrametadata{$filespec};
586            my $mfref = $extrametafile{$filespec};
587
588            # Add the list files where the metadata came from
589            # into the metadata table so we can track this
590            # This mechanism is similar to how gsdlassocfile works
591
592            my @metafile_pair = ();
593            foreach my $l (keys %$mfref) {
594            my $f = $mfref->{$l};
595            push (@metafile_pair, "$f : $l");
596            }
597
598            $mdref->{'gsdlmetafile'} = \@metafile_pair;
599
600            &metadatautil::combine_metadata_structures($out_metadata, $mdref);
601        }
602        }
603    }
604
605    if (defined $self->{'inf_timestamp'}) {
606        # Look to see if it's a completely new file
607
608        if (!$block_hash->{'new_files'}->{$full_filename}) {
609        # Not a new file, must be an existing file
610        # Let' see if it's newer than the last import.pl
611
612
613        if (! -d $full_filename) {
614            if (!$block_hash->{'reindex_files'}->{$full_filename}) {
615            # filename has been around for longer than inf_timestamp
616            print $outhandle "**** Skipping $subfile\n" if ($verbosity >3);
617            next;
618            }
619            else {
620            # Remove old folder in archives (might hash to something different)
621            # *** should be doing this on a Del one as well
622            # but leave folder name?? and ensure hashs to
623            # same again??
624
625            # Then let through as new doc??
626
627            # mark to doc-oids that rely on it for re-indexing
628            ## &GDBMUtils::gdbmDatabase();
629            }
630        }
631        }
632    }
633
634    # Recursively read each $subfile
635    print $outhandle "DirectoryPlugin recurring: $subfile\n" if ($verbosity > 2);
636   
637    $count += &plugin::read ($pluginfo, $this_file_base_dir,
638                 $file_subfile, $block_hash,
639                 $out_metadata, $processor, $maxdocs, ($total_count + $count), $gli);
640    }
641
642    return $count;
643}
644
645sub compile_stats {
646    my $self = shift(@_);
647    my ($stats) = @_;
648}
649
6501;
Note: See TracBrowser for help on using the browser.