diff --git a/lib/Archive/BagIt/Base.pm b/lib/Archive/BagIt/Base.pm
index 8c748523f7ba7d633bb1408a83f07927154dedbf..83a76adf9d70f31a33be92c3438e9583a10b0db5 100644
--- a/lib/Archive/BagIt/Base.pm
+++ b/lib/Archive/BagIt/Base.pm
@@ -15,6 +15,7 @@ use File::stat;
use Digest::MD5;
use Class::Load qw(load_class);
use Carp;
+use List::Util qw( uniq);
# VERSION
@@ -109,9 +110,8 @@ sub bag_info_by_key {
my $info = $self->bag_info();
if (defined $searchkey) {
foreach my $entry (@{$info}) {
- my ($key, $value) = each %{$entry};
- if (defined $key && $key eq $searchkey) {
- return $value;
+ if (exists $entry->{$searchkey}) {
+ return $entry->{$searchkey};
}
}
}
@@ -307,55 +307,42 @@ sub _build_tagmanifest_files {
}
-sub _build_tagmanifest_entries {
- my ($self) = @_;
-
- my @tagmanifests = @{$self->tagmanifest_files};
- my $tagmanifest_entries = {};
- foreach my $tagmanifest_file (@tagmanifests) {
- die("Cannot open $tagmanifest_file: $!") unless (open(my $TAGMANIFEST,"<:encoding(utf8)", $tagmanifest_file));
- while (my $line = <$TAGMANIFEST>) {
+sub __build_xxxmanifest_entries {
+ my ($self, $xxmanifestfiles) = @_;
+ my @xxmanifests = @{$xxmanifestfiles};
+ my $xxmanifest_entries = {};
+ my $bag_path=$self->bag_path();
+ foreach my $xxmanifest_file (@xxmanifests) {
+ die("Cannot open $xxmanifest_file: $!") unless (open(my $XXMANIFEST,"<:encoding(utf8)", $xxmanifest_file));
+ my $algo = $xxmanifest_file;
+ $algo =~ s#^($bag_path/).bagit/#$1#; # FIXME: only for dotbagit-variant, if dotbagit will be outdated, this should be removed
+ $algo =~ s#^$bag_path/##;
+ $algo =~ s#^tag##;
+ $algo =~ s#^manifest-([a-z0-9]+)\.txt$#$1#;
+ while (my $line = <$XXMANIFEST>) {
chomp($line);
my($digest,$file) = split(/\s+/, $line, 2);
- $tagmanifest_entries->{$file} = $digest;
+ $xxmanifest_entries->{$algo}->{$file} = $digest;
}
- close($TAGMANIFEST);
-
+ close($XXMANIFEST);
}
- return $tagmanifest_entries;
+ return $xxmanifest_entries;
}
-sub _build_manifest_entries {
+sub _build_tagmanifest_entries {
my ($self) = @_;
+ return $self->__build_xxxmanifest_entries($self->tagmanifest_files);
+}
- my @manifests = @{$self->manifest_files};
- my $manifest_entries = {};
- foreach my $manifest_file (@manifests) {
- die("Cannot open $manifest_file: $!") unless (open (my $MANIFEST, "<:encoding(utf8)", $manifest_file));
- while (my $line = <$MANIFEST>) {
- chomp($line);
- my ($digest,$file);
- ($digest, $file) = $line =~ /^([a-f0-9]+)\s+(.+)/;
- if(!$file) {
- die ("This is not a valid manifest file");
- } else {
- print "file: $file \n" if $DEBUG;
- $manifest_entries->{$file} = $digest;
- }
- }
- close($MANIFEST);
- }
-
- return $manifest_entries;
-
+sub _build_manifest_entries {
+ my ($self) = @_;
+ return $self->__build_xxxmanifest_entries($self->manifest_files);
}
sub _build_payload_files{
my($self) = @_;
-
my $payload_dir = $self->payload_path;
my $payload_reldir = $self->rel_payload_path;
-
my @payload=();
File::Find::find( sub{
$File::Find::name = decode ('utf8', $File::Find::name);
@@ -532,6 +519,96 @@ sub load_plugins {
return 1;
}
+
+sub _verify_XXX_manifests {
+ my ($self, $xxprefix, $xxmanifest_entries, $files, $return_all_errors) =@_;
+ # Read the manifest file
+ #use Data::Printer;
+ #p( $self);
+ #print Dumper($self->{entries});
+ my %manifest = %{$xxmanifest_entries};
+ my @payload = @{ $files };
+ my %invalids;
+ my $bagit = $self->bag_path;
+ my $version = $self->bag_version();
+ # Evaluate each file against the manifest
+ foreach my $alg (keys %{$xxmanifest_entries}) {
+ my $manifest_alg = $self->manifests->{$alg};
+ if (! defined $manifest_alg) {
+ next;
+ # TODO: return Errormessage?
+ }
+ my $digestobj = $manifest_alg->algorithm();
+
+ my $xxfilename = "$bagit/$xxprefix-$alg.txt";
+ foreach my $local_name (@payload) {
+ # local_name is relative to bagit base
+ my ($digest);
+ unless (exists $manifest{$alg}{"$local_name"}) {
+ #system ("tree $bagit");
+ #use File::Slurp;
+ #my $vontent = read_file($xxfilename);
+ #print "Content: '$vontent'\n";
+ #print "Alg=$alg\n";
+ #use Data::Printer;
+ #p( %manifest);
+ die("file found which is not in $xxfilename: [$local_name] (bag-path:$bagit)");
+ }
+ if (!-r "$bagit/$local_name") {die("Cannot open $bagit/$local_name");}
+ $digest = $digestobj->verify_file("$bagit/$local_name");
+ print "digest " . $digestobj->name() . " of $bagit/$local_name: $digest\n" if $DEBUG;
+ unless ($digest eq $manifest{$alg}{$local_name}) {
+ if ($return_all_errors) {
+ $invalids{$local_name} = $digest;
+ }
+ else {
+ die("file: $bagit/$local_name invalid, digest ($alg) calculated=$digest, but expected=$manifest{$alg}{$local_name} in file '$xxfilename'");
+ }
+ }
+ delete($manifest{$alg}{$local_name});
+ }
+ }
+ if($return_all_errors && keys(%invalids) ) {
+ foreach my $invalid (keys(%invalids)) {
+ print "invalid: $invalid hash: ".$invalids{$invalid}."\n";
+ }
+ die ("bag verify for bagit $version failed with invalid files");
+ }
+ # Make sure there are no missing files
+ foreach my $alg (keys %manifest){
+ my @localfiles = keys(%{ $manifest{$alg} });
+ if (@localfiles) {
+ use Data::Printer;
+ p( $self);
+ die("Missing files in bag $bagit for algorithm=$alg", join( " file=", @localfiles));
+ }
+ }
+
+}
+
+sub _verify_manifests {
+ my ($self, $alg, $return_all_errors) = @_;
+ $self->_verify_XXX_manifests(
+ "manifest",
+ $self->manifest_entries(),
+ $self->payload_files(),
+ $return_all_errors
+ );
+}
+
+sub _verify_tagmanifests {
+ my ($self, $alg, $return_all_errors) = @_;
+ # filter tagmanifest-files
+ my @non_payload_files = grep { $_ !~ m/tagmanifest-[a-z0-9]+\.txt/} @{ $self->non_payload_files };
+ $self->_verify_XXX_manifests(
+ "tagmanifest",
+ $self->tagmanifest_entries,
+ \@non_payload_files,
+ $return_all_errors
+ );
+}
+
+
=head2 verify_bag
An interface to verify a bag.
@@ -547,11 +624,12 @@ sub verify_bag {
#like $return all errors rather than dying on first one
my $bagit = $self->bag_path;
my $version = $self->bag_version(); # to call trigger
- my $manifest_file = $self->metadata_path."/manifest-".$self->forced_fixity_algorithm()->name().".txt"; # FIXME: use plugin instead
+ my $forced_fixity_alg = $self->forced_fixity_algorithm()->name();
+ my $manifest_file = $self->metadata_path."/manifest-$forced_fixity_alg.txt"; # FIXME: use plugin instead
my $payload_dir = $self->payload_path;
my $return_all_errors = $opts->{return_all_errors};
- my %invalids;
- my @payload = @{$self->payload_files};
+
+
die("$manifest_file is not a regular file for bagit $version") unless -f ($manifest_file);
die("$payload_dir is not a directory") unless -d ($payload_dir);
@@ -560,38 +638,14 @@ sub verify_bag {
die ("Bag Version $version is unsupported");
}
- # Read the manifest file
- #print Dumper($self->{entries});
- my %manifest = %{$self->manifest_entries};
+ # check forced fixity
+ $self->_verify_manifests($forced_fixity_alg, $return_all_errors);
+ $self->_verify_tagmanifests($forced_fixity_alg, $return_all_errors);
+
+
+ # TODO: check if additional algorithms are used
+
- # Evaluate each file against the manifest
- my $digestobj = $self->forced_fixity_algorithm();
- foreach my $local_name (@payload) { # local_name is relative to bagit base
- my ($digest);
- unless ($manifest{"$local_name"}) {
- die ("file found not in manifest: [$local_name] (bag-path:$bagit)");
- }
- if (! -r "$bagit/$local_name" ) {die ("Cannot open $bagit/$local_name");}
- $digest = $digestobj->verify_file( "$bagit/$local_name");
- print "digest of $bagit/$local_name: $digest\n" if $DEBUG;
- unless ($digest eq $manifest{$local_name}) {
- if($return_all_errors) {
- $invalids{$local_name} = $digest;
- }
- else {
- die ("file: $bagit/$local_name invalid");
- }
- }
- delete($manifest{$local_name});
- }
- if($return_all_errors && keys(%invalids) ) {
- foreach my $invalid (keys(%invalids)) {
- print "invalid: $invalid hash: ".$invalids{$invalid}."\n";
- }
- die ("bag verify for bagit $version failed with invalid files");
- }
- # Make sure there are no missing files
- if (keys(%manifest)) { die ("Missing files in bag".p(%manifest)); }
return 1;
}
diff --git a/tmp/bagit/test/bag-info.txt b/tmp/bagit/test/bag-info.txt
index 3964b4191b230fe5ce9e6d68a5b5b948b57c690f..746e31a9f3dbc86984924a072667bc50d57eb12c 100644
--- a/tmp/bagit/test/bag-info.txt
+++ b/tmp/bagit/test/bag-info.txt
@@ -1,6 +1,6 @@
SLUBArchiv-sipVersion: v2020.1
Author: Taras Schevchenko
-Bagging-Date: 2020-02-25
+Bagging-Date: 2020-03-09
Bag-Software-Agent: Archive::BagIt <https://metacpan.org/pod/Archive::BagIt>
Payload-Oxum: 23.1
Bag-Size: 23 B
diff --git a/tmp/bagit/test/tagmanifest-md5.txt b/tmp/bagit/test/tagmanifest-md5.txt
index 58ffb033203ceb4d9bc54c52fd7d0a33dde7fba8..04e2a1bbfddf84d466f8f5b6c391702bd4bd44cf 100644
--- a/tmp/bagit/test/tagmanifest-md5.txt
+++ b/tmp/bagit/test/tagmanifest-md5.txt
@@ -1,5 +1,5 @@
78bfb3f0189a50e8652f8fcfffebeb12 manifest-sha512.txt
-4f1e53d2b0c3a731c560078589ad3202 bag-info.txt
+1e2916f186cc750d5dc68bc27e7efe21 bag-info.txt
7cb0edfcba877a10170214f4f97d0319 bagit.txt
4aa497234e99e2f9302d10bed31db634 manifest-md5.txt
8895d8bf80620a0c2f79969cac46782a meta/Fallbeispiel-13.xml
diff --git a/tmp/bagit/test/tagmanifest-sha512.txt b/tmp/bagit/test/tagmanifest-sha512.txt
index baaeefb8b5059f00e2e96120fa145ca3ead0f335..785841bd1a7a89e4503b5ece5ee4174ebe9af1dc 100644
--- a/tmp/bagit/test/tagmanifest-sha512.txt
+++ b/tmp/bagit/test/tagmanifest-sha512.txt
@@ -1,5 +1,5 @@
b1c9a4bc23f472db7fff27884e69567fd62ee2dc17b36909a373b80fda9b0bcf84491ce0bc163e00ce43e4f02e085a3d5315556604cf23f40e81a25d3c0a57a8 manifest-sha512.txt
-95447138cd491c96d8f65baa9a95a974055e11a414f1ccf721472ed0174a12c7cdf3e5ff02eda2477bb8628279ce59e12382ce9b2ba430901c17c967d726b344 bag-info.txt
+f21212570b7b64e142c866cafcb0c891ad7beb288f439b1489b68445b93e22450f07c13a7b0ac9ec2c9d7e9cfc91bb9102af2c71954f0b7e2eb67511bd8ff6eb bag-info.txt
da429ec4f5ca60aebd25aa4c26ccf51ae87609129ae76f09683c810da4af4414277250ecc0253bf7871c7a3f9c5b3ed6918a5d5c092a7dc394051601eb8bb3ab bagit.txt
0e892c5219c5b1153f6cdbdcdc4c08d5fb7d0eeb63725d4ec19110ccc28ed59d0129ccdb29117d8094f5e8f04924db6dc55b8a0416216a60df46047d4ede4114 manifest-md5.txt
a9c97cc5c5f4677472965fb988cc0e4eff57beb534fcdadb838b29226d212bb00a3d3ae5ca15390f41b2d0f2f57bf02bbb779f7db41f128c32893fd0f85038d1 meta/Fallbeispiel-13.xml