Changeset 25097

Timestamp:

2012-02-20T15:14:01+13:00 (12 years ago)

Author:

sjm84

Message:

More changes to metadataaction as well as sorting the keys when printing

Location:

main/trunk/greenstone2/perllib/cgiactions

Files:

• baseaction.pm (modified) (1 diff)
• metadataaction.pm (modified) (20 diffs)

main/trunk/greenstone2/perllib/cgiactions/baseaction.pm

@@ -67 +67 @@
     $err_mess .= "  check-installation\n\n";
 
-    foreach my $a (keys %$action_table) {
+    foreach my $a (sort keys %$action_table) {
         $err_mess .= "  $a:\n";
         $err_mess .= "    Compulsory args: ";

main/trunk/greenstone2/perllib/cgiactions/metadataaction.pm

@@ -41 +41 @@
 }
 
-
 @metadataaction::ISA = ('baseaction');
 
-
 my $action_table =
-{ 
-    "get-live-metadata"     => { 'compulsory-args' => [ "d", "metaname" ],
-                     'optional-args'   => [] }, 
-
-    "get-metadata"          => { 'compulsory-args' => [ "d", "metaname" ],
-                     'optional-args'   => [ "metapos" ] }, 
-
-    "set-live-metadata"     => { 'compulsory-args' => [ "d", "metaname", "metavalue" ],
-                     'optional-args'   => [ ] },
-
-    "set-metadata"          => { 'compulsory-args' => [ "d", "metaname", "metavalue" ],
-                     'optional-args'   => [ "metapos" ] },
-
-    "set-metadata-array"     => { 'compulsory-args' => [ "json" ],
-                     'optional-args'   => [ ] },
+{
+    #GET METHODS
+    "get-import-metadata" => { 
+        'compulsory-args' => [ "d", "metaname" ],
+        'optional-args'   => [ ] },
+
+    "get-archives-metadata" => { 
+        'compulsory-args' => [ "d", "metaname" ],
+        'optional-args'   => [ "metapos" ] },
+    
+    "get-metadata" => { 
+        'compulsory-args' => [ "d", "metaname" ],
+        'optional-args'   => [ "metapos" ] }, 
+
+    "get-live-metadata" => { 
+        'compulsory-args' => [ "d", "metaname" ],
+        'optional-args'   => [ ] }, 
+
+    #SET METHODS
+    "set-live-metadata" => { 
+        'compulsory-args' => [ "d", "metaname", "metavalue" ],
+        'optional-args'   => [ ] },
+
+    "set-metadata" => { 
+        'compulsory-args' => [ "d", "metaname", "metavalue" ],
+        'optional-args'   => [ "metapos" ] },
+
+    "set-archives-metadata" => { 
+        'compulsory-args' => [ "d", "metaname", "metavalue" ],
+        'optional-args'   => [ "metapos", "metamode", "prevmetavalue" ] }, # metamode can be "accumulate", "override",
+    
+    "set-import-metadata" => { 
+        'compulsory-args' => [ "metaname", "metavalue" ],
+        'optional-args'   => [ "d", "f", "metamode" ] }, # metamode can be "accumulate", "override", or "unique-id". Also need to add the ability to specify a previous metadata value to overwrite (because we can't use metapos)
+                 
+    #SET METHODS (ARRAY)
+    "set-metadata-array" => { 
+        'compulsory-args' => [ "json" ],
+        'optional-args'   => [ ] },
                      
-    "set-archives-metadata" => { 'compulsory-args' => [ "d", "metaname", "metavalue" ],
-                     'optional-args'   => [ "metapos", "metamode" ]
-                      # metamode can be "accumulate", "override",
-                },
-
-    "set-archives-metadata-array" => { 'compulsory-args' => [ "json" ],
-                     'optional-args'   => [ "metamode" ]
-                },
-                
-    "set-import-metadata"   => { 'compulsory-args' => [ "metaname", "metavalue" ],
-                     'optional-args'   => [ "d", "f", "metamode" ] # Need to add the ability to specify a previous metadata value to overwrite (because we can't use metapos)
-                    # metamode can be "accumulate", "override", or "unique-id"
-                 },
-
-    "remove-import-metadata" => { 'compulsory-args' => [ "d", "metaname", "metavalue" ], #TODO: add f argument
-                     'optional-args'   => [ ] },
+    "set-archives-metadata-array" => { 
+        'compulsory-args' => [ "json" ],
+        'optional-args'   => [ ] },
+        
+    "set-import-metadata-array" => {
+        'compulsory-args' => [ "json" ],
+        'optional-args'   => [ ] },
+        
+    #REMOVE METHODS
+    "remove-import-metadata" => { 
+        'compulsory-args' => [ "d", "metaname", "metavalue" ], #TODO: add f argument
+        'optional-args'   => [ ] },
                      
-    "remove-archives-metadata" => { 'compulsory-args' => [ "d", "metaname" ], #TODO: add f argument
-                     'optional-args'   => [ "metapos", "metavalue" ] },
-
-    "remove-live-metadata"  => { 'compulsory-args' => [ "d", "metaname" ],
-                     'optional-args'   => [ ] },
-
-    "remove-metadata"       => { 'compulsory-args' => [ "d", "metaname" ],
-                     'optional-args'   => [ "metapos", "metavalue" ] },
-
-    "insert-metadata"       => { 'compulsory-args' => [ "d", "metaname", "metavalue" ],
-                     'optional-args'   => [ ]
-                   }
+    "remove-archives-metadata" => { 
+        'compulsory-args' => [ "d", "metaname" ], #TODO: add f argument
+        'optional-args'   => [ "metapos", "metavalue" ] },
+
+    "remove-live-metadata" => {
+        'compulsory-args' => [ "d", "metaname" ],
+        'optional-args'   => [ ] },
+
+    "remove-metadata" => {
+        'compulsory-args' => [ "d", "metaname" ],
+        'optional-args'   => [ "metapos", "metavalue" ] },
+
+    #INSERT METHODS
+    "insert-metadata" => { 
+        'compulsory-args' => [ "d", "metaname", "metavalue" ],
+        'optional-args'   => [ ] }
 };
 
@@ -242 +264 @@
 
 
+sub get_import_metadata
+{
+    my $self = shift @_;
+
+    my $username  = $self->{'username'};
+    my $collect   = $self->{'collect'};
+    my $gsdl_cgi  = $self->{'gsdl_cgi'};
+    my $gsdlhome  = $self->{'gsdlhome'};
+
+    # Authenticate user if it is enabled
+    if ($baseaction::authentication_enabled) {
+        # Ensure the user is allowed to edit this collection
+        &authenticate_user($gsdl_cgi, $username, $collect);
+    }
+
+    # Obtain the collect dir
+    my $site = $self->{'site'};
+    my $collect_dir = $gsdl_cgi->get_collection_dir($site);
+    ##my $collect_dir = &util::filename_cat($gsdlhome, "collect");
+
+    # Make sure the collection isn't locked by someone else
+    $self->lock_collection($username, $collect);
+
+    # look up additional args
+    my $docid     = $self->{'d'};
+    my $metaname  = $self->{'metaname'};
+    my $infodbtype = $self->{'infodbtype'};
+    if (!defined $docid) 
+    {
+        $gsdl_cgi->generate_error_message("No docid (d=...) specified.\n");
+    } 
+
+    # Obtain where the metadata.xml is from the archiveinfo-doc.gdb file
+    # If the doc oid is not specified, we assume the metadata.xml is next to the specified "f"
+    my $metadata_xml_file;
+    my $import_filename = undef;
+    
+
+    my $archive_dir = &util::filename_cat($collect_dir, $collect, "archives");
+    my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
+    my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
+
+    # This now stores the full pathname
+    $import_filename = $doc_rec->{'src-file'}->[0];
+
+    # figure out correct metadata.xml file [?]
+    # Assuming the metadata.xml file is next to the source file
+    # Note: This will not work if it is using the inherited metadata from the parent folder
+    my ($import_tailname, $import_dirname) = File::Basename::fileparse($import_filename);
+    my $metadata_xml_filename = &util::filename_cat($import_dirname, "metadata.xml");
+
+    $gsdl_cgi->generate_ok_message($self->get_metadata_from_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname, $import_tailname));
+
+    # Release the lock once it is done
+    $self->unlock_collection($username, $collect);
+}
+
+sub get_metadata_from_metadata_xml
+{
+    my $self = shift @_;
+    my ($gsdl_cgi, $metadata_xml_filename, $metaname, $src_file) = @_;
+    
+    my @rules = 
+    ( 
+        _default => 'raw',
+        'Metadata' => \&gfmxml_metadata,
+        'FileName' => \&mxml_filename
+    );
+        
+    my $parser = XML::Rules->new
+    (
+        rules => \@rules,
+        output_encoding => 'utf8'
+    );
+    
+    my $xml_in = "";
+    if (!open(MIN,"<$metadata_xml_filename")) 
+    {
+        $gsdl_cgi->generate_error("Unable to read in $metadata_xml_filename: $!");
+    }
+    else 
+    {
+        # Read them in
+        my $line;
+        while (defined ($line=<MIN>)) {
+            $xml_in .= $line;
+        }
+        close(MIN); 
+
+        $parser->parse($xml_in, {metaname => $metaname, src_file => $src_file});
+        
+        if(defined $parser->{'pad'}->{'metavalue'})
+        {
+            return $parser->{'pad'}->{'metavalue'};
+        }
+        else
+        {
+            return "";
+        }
+    }
+}
+
+sub gfmxml_metadata
+{
+    my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
+
+    if (($parser->{'parameters'}->{'src_file'} eq $parser->{'parameters'}->{'current_file'}) && ($attrHash->{'name'} eq $parser->{'parameters'}->{'metaname'}))
+    {
+        $parser->{'pad'}->{'metavalue'} = $attrHash->{'_content'};
+    }
+}
+
+sub get_archives_metadata
+{
+    my $self = shift @_;
+
+    my $username  = $self->{'username'};
+    my $collect   = $self->{'collect'};
+    my $gsdl_cgi  = $self->{'gsdl_cgi'};
+    my $gsdlhome  = $self->{'gsdlhome'};
+    my $infodbtype = $self->{'infodbtype'};
+
+    # Authenticate user if it is enabled
+    if ($baseaction::authentication_enabled) {
+        # Ensure the user is allowed to edit this collection
+        &authenticate_user($gsdl_cgi, $username, $collect);
+    }
+
+    # Obtain the collect dir
+    my $site = $self->{'site'};
+    my $collect_dir = $gsdl_cgi->get_collection_dir($site);
+    
+    my $archive_dir = &util::filename_cat($collect_dir, $collect, "archives");
+
+    # Make sure the collection isn't locked by someone else
+    $self->lock_collection($username, $collect);
+    
+    # look up additional args
+    my ($docid, $docid_secnum) = ($self->{'d'} =~ m/^(.*?)(\..*)?$/);
+    $docid_secnum = "" if (!defined $docid_secnum);
+    
+    my $metaname = $self->{'metaname'};
+    my $metapos = $self->{'metapos'};
+    $metapos = 0 if (!defined $metapos);
+    
+    my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
+    my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
+
+    # This now stores the full pathname
+    my $doc_filename = $doc_rec->{'doc-file'}->[0]; 
+
+    $gsdl_cgi->generate_ok_message($self->get_metadata_from_archive_xml($gsdl_cgi, &util::filename_cat($archive_dir, $doc_filename), $metaname, $metapos, $docid_secnum));
+
+    # Release the lock once it is done
+    $self->unlock_collection($username, $collect);
+}
+
+sub get_metadata_from_archive_xml
+{
+    my $self = shift @_;
+    my ($gsdl_cgi, $doc_xml_filename, $metaname, $metapos, $secid) = @_;
+    
+    my @start_rules = ('Section' => \&dxml_start_section);
+    
+    my @rules = 
+    ( 
+        _default => 'raw',
+        'Metadata' => \&gfdxml_metadata
+    );
+        
+    my $parser = XML::Rules->new
+    (
+        start_rules => \@start_rules,
+        rules => \@rules,
+        output_encoding => 'utf8'
+    );
+    
+    my $xml_in = "";
+    if (!open(MIN,"<$doc_xml_filename")) 
+    {
+        $gsdl_cgi->generate_error("Unable to read in $doc_xml_filename: $!");
+    }
+    else 
+    {
+        # Read them in
+        my $line;
+        while (defined ($line=<MIN>)) {
+            $xml_in .= $line;
+        }
+        close(MIN); 
+
+        $parser->parse($xml_in, {metaname => $metaname, metapos => $metapos, secid => $secid});
+        
+        if(defined $parser->{'pad'}->{'metavalue'})
+        {
+            return $parser->{'pad'}->{'metavalue'};
+        }
+        else
+        {
+            return "";
+        }
+    }
+}
+
+sub gfdxml_metadata
+{
+    my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
+    
+    if(!($parser->{'parameters'}->{'secid'} eq $parser->{'parameters'}->{'curr_section_num'}))
+    {
+        return;
+    }
+
+    if ($parser->{'parameters'}->{'metaname'} eq $attrHash->{'name'})
+    {
+        if (!defined $parser->{'parameters'}->{'poscount'})
+        {
+            $parser->{'parameters'}->{'poscount'} = 0;
+        }
+        else
+        {
+            $parser->{'parameters'}->{'poscount'}++;
+        }
+    }
+
+    if (($parser->{'parameters'}->{'metaname'} eq $attrHash->{'name'}) && ($parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'}))
+    {   
+        $parser->{'pad'}->{'metavalue'} = $attrHash->{'_content'};
+    }
+}
+
 sub set_live_metadata
 {
@@ -394 +647 @@
     $self->unlock_collection($username, $collect);
 }
-
-
 
 
@@ -471 +722 @@
 
 
-
 sub dxml_metadata
 {
@@ -491 +741 @@
 ##    print STDERR "**** metamode = $metamode\n";
     
-    if ((!defined $opt_doc_secnum) || ($opt_doc_secnum eq $curr_secnum)) {
-    my $name_attr = $attrHash->{'name'};
-    if (($name_attr eq $metaname) && ($metamode eq "override")) {
-##      print STDERR "**** got match!!\n";
-        # Get the value and override the current value
-        my $metavalue = $parser->{'parameters'}->{'metavalue'};
-        $attrHash->{'_content'} = $metavalue;
-        
-        # Don't want it to wipe out any other pieces of metadata
-        $parser->{'parameters'}->{'metamode'} = "done";
-    }
+    if ((!defined $opt_doc_secnum) || ($opt_doc_secnum eq $curr_secnum)) 
+    {
+        my $name_attr = $attrHash->{'name'};
+        if (($name_attr eq $metaname) && ($metamode eq "override")) 
+        {
+            if (!defined $parser->{'parameters'}->{'poscount'}) 
+            { 
+                $parser->{'parameters'}->{'poscount'} = 0; 
+            } 
+            else 
+            { 
+                $parser->{'parameters'}->{'poscount'}++; 
+            } 
+            
+            if(defined $parser->{'parameters'}->{'metapos'} && $parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'})
+            {
+                ##print STDERR "**** got match!!\n";
+                # Get the value and override the current value
+                my $metavalue = $parser->{'parameters'}->{'metavalue'};
+                $attrHash->{'_content'} = $metavalue;
+                
+                # Don't want it to wipe out any other pieces of metadata
+                $parser->{'parameters'}->{'metamode'} = "done";
+            }
+            else if(defined $parser->{'parameters'}->{'prevmetavalue'} && $parser->{'parameters'}->{'prevmetavalue'} == $attrHash->{'_content'})
+            {
+                $attrHash->{'_content'} = $metavalue;
+                $parser->{'parameters'}->{'metamode'} = "done";
+            }
+        }
     }
 
@@ -544 +813 @@
 
 
-
 sub dxml_start_section
 {
@@ -581 +849 @@
     $parser->{'parameters'}->{'curr_section_num'}   = $new_secnum;
 
-    print STDERR "*** In Section: $new_secnum\n";
+    ##print STDERR "*** In Section: $new_secnum\n";
 }
 
@@ -625 +893 @@
 }
 
-
 sub edit_doc_xml
 {
     my $self = shift @_;
-    my ($gsdl_cgi, $doc_xml_filename, $metaname, $metavalue, $metapos, $metamode, $opt_secnum) = @_;
+    my ($gsdl_cgi, $doc_xml_filename, $metaname, $metavalue, $metapos, $metamode, $opt_secnum, $prevmetavalue) = @_;
 
     # To monitor which section/subsection number we are in
@@ -646 +913 @@
             'metapos'   => $metapos,
             'metavalue' => $metavalue,
-            'metamode'  => $metamode };
+            'metamode'  => $metamode,
+            'prevmetavalue' => $prevmetavalue };
             
     if (defined $opt_secnum) {
@@ -658 +926 @@
 {
     my $self = shift @_;
-    my ($gsdl_cgi,$archive_dir, $collect_dir,$collect, $infodbtype,$docid,$metaname,$metapos,$metavalue,$metamode) = @_;
+    my ($gsdl_cgi,$archive_dir, $collect_dir,$collect, $infodbtype,$docid,$metaname,$metapos,$metavalue,$metamode,$prevmetavalue) = @_;
     
     # Obtain the doc.xml path for the specified docID
@@ -676 +944 @@
     # Running import.pl -groupsize will cause this to have multiple sections in one doc.xml
     
-    print STDERR "** away to call edit_doc_xml\n";
-    
     $self->edit_doc_xml($gsdl_cgi,$doc_xml_filename,
-            $metaname,$metavalue,$metapos,$metamode,$docid_secnum);
-            
-    print STDERR "*** finished edit_doc_xml\n";
+            $metaname,$metavalue,$metapos,$metamode,$docid_secnum,$prevmetavalue);
     
     return 0; # return 0 for now to indicate no error
@@ -717 +981 @@
     my $metaname   = $self->{'metaname'};
     my $metavalue  = $self->{'metavalue'};
+    my $prevmetavalue = $self->{'prevmetavalue'}
     
     my $metapos    = $self->{'metapos'};
@@ -729 +994 @@
     
     my $status = $self->set_archives_metadata_entry($gsdl_cgi,$archive_dir, $collect_dir,$collect, $infodbtype,$docid,
-                $metaname,$metapos,$metavalue,$metamode);
+                $metaname,$metapos,$metavalue,$metamode,$prevmetavalue);
    
     # Release the lock once it is done
@@ -876 +1141 @@
 
     # This now stores the full pathname
-    my $import_filename = $doc_rec->{'doc-file'}->[0];  
-
-    my $status = $self->remove_from_doc_xml($gsdl_cgi, &util::filename_cat($archive_dir, $import_filename), $metaname, $metapos, undef, $docid_secnum);
+    my $doc_filename = $doc_rec->{'doc-file'}->[0]; 
+
+    my $status = $self->remove_from_doc_xml($gsdl_cgi, &util::filename_cat($archive_dir, $doc_filename), $metaname, $metapos, undef, $docid_secnum);
     
     # Release the lock once it is done
@@ -1000 +1265 @@
 
     # Report error if we don't see FileName tag before this
-    die "Fatel Error: Unexpected metadata.xml structure. Undefind current_file, possiblely encountered Description before FileName" if (!defined($parser->{'parameters'}->{'current_file'}));
+    die "Fatal Error: Unexpected metadata.xml structure. Undefind current_file, possiblely encountered Description before FileName" if (!defined($parser->{'parameters'}->{'current_file'}));
     
     # Don't do anything if we are not in the right FileSet
@@ -1040 +1305 @@
 
     # Failed... Report error if we don't see FileName tag before this
-    die "Fatel Error: Unexpected metadata.xml structure. Undefind current_file, possiblely encountered Description before FileName" if (!defined($parser->{'parameters'}->{'current_file'}));
+    die "Fatal Error: Unexpected metadata.xml structure. Undefind current_file, possiblely encountered Description before FileName" if (!defined($parser->{'parameters'}->{'current_file'}));
 
     # Don't do anything if we are not in the right FileSet
@@ -1218 +1483 @@
     
     if ($baseaction::authentication_enabled) {
-    # Ensure the user is allowed to edit this collection
-    $self->authenticate_user($username, $collect);
+        # Ensure the user is allowed to edit this collection
+        $self->authenticate_user($username, $collect);
     }
 
@@ -1297 +1562 @@
 }
 
+sub set_import_metadata_array
+{
+    my $self = shift @_;
+
+    my $username  = $self->{'username'};
+    my $collect   = $self->{'collect'};
+    my $gsdl_cgi  = $self->{'gsdl_cgi'};
+    my $gsdlhome  = $self->{'gsdlhome'};
+
+    if ($baseaction::authentication_enabled) {
+        # Ensure the user is allowed to edit this collection
+        &authenticate_user($gsdl_cgi, $username, $collect);
+    }
+
+    my $site = $self->{'site'};
+    my $collect_dir = $gsdl_cgi->get_collection_dir($site);
+    
+    $gsdl_cgi->checked_chdir($collect_dir);
+
+    # Make sure the collection isn't locked by someone else
+    $self->lock_collection($username, $collect);
+
+    # look up additional args
+    
+    my $infodbtype = $self->{'infodbtype'};
+
+    my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
+    
+    my $json_str = $self->{'json'};
+    my $doc_array = decode_json $json_str;
+    
+    my $global_status = 0;
+    my $global_mess = "";
+    
+    my @all_docids = ();
+    
+    foreach my $doc_array_rec ( @$doc_array ) 
+    {   
+        my $docid = $doc_array_rec->{'docid'};
+        my $metaname = $doc_array_rec->{'metaname'};
+        my $metamode = $self->{'metamode'};
+        my $metavalue = $doc_array_rec->{'metavalue'};
+
+        if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
+            # make "accumulate" the default (less destructive, as won't actually 
+            # delete any existing values)
+            $metamode = "accumulate";
+        } 
+
+        push(@all_docids,$docid);
+
+        # Obtain where the metadata.xml is from the archiveinfo-doc.gdb file
+        # If the doc oid is not specified, we assume the metadata.xml is next to the specified "f"
+        my $metadata_xml_file;
+        my $import_filename = undef;
+
+        my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
+        my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
+
+        # This now stores the full pathname
+        $import_filename = $doc_rec->{'src-file'}->[0]; 
+
+        # figure out correct metadata.xml file [?]
+        # Assuming the metadata.xml file is next to the source file
+        # Note: This will not work if it is using the inherited metadata from the parent folder
+        my ($import_tailname, $import_dirname) = File::Basename::fileparse($import_filename);
+        my $metadata_xml_filename = &util::filename_cat($import_dirname,"metadata.xml");
+
+        $self->edit_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname, $metavalue, $metamode, $import_tailname);
+    }
+
+    my $mess = "set-archives-metadata-array successful: Keys[ ".join(", ",@all_docids)."]\n";
+    $gsdl_cgi->generate_ok_message($mess);
+    
+    # Release the lock once it is done
+    $self->unlock_collection($username, $collect);
+}
+
 sub remove_import_metadata
 {
@@ -1328 +1671 @@
     if ((!defined $docid) || ($docid =~ m/^\s*$/)) 
     {
-        die "No docid (d=...) specified.\n";
+        $gsdl_cgi->generate_error_message("No docid (d=...) specified.\n");
     }
     
@@ -1350 +1693 @@
     if((!defined $import_filename) || ($import_filename =~ m/^\s*$/))
     {
-        die "There is no metadata\n";
+        $gsdl_cgi->generate_error_message("There is no metadata\n");
     }