Changeset 32062 for main/trunk/greenstone2/perllib

Timestamp:

2017-11-14T13:40:47+13:00 (6 years ago)

Author:

ak19

Message:

Committing unused and untested new subroutines, in case they may prove useful in future.

File:

• main/trunk/greenstone2/perllib/cgiactions/metadataaction.pm (modified) (6 diffs)

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

@@ -67 +67 @@
 my $getmeta_action_table =
 {
+    # unused and untested
+    # when DocEdit=1, need to retrieve a doc's full text (or doc section's full text) from archives
+    "get-archives-text" => {
+    'compulsory-args' => [ "d" ] },
+    #'compulsory-args' => [ "d" ],
+    #'optional-args' => [ "section" ] },
+    
     #GET METHODS
     "get-import-metadata" => { 
@@ -888 +895 @@
 }
 
+# unused and untested
+sub get_archives_text
+{
+    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'};
+
+    # 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, $collect, "archives");
+
+    # look up additional args
+    my ($docid, $docid_secnum) = ($self->{'d'} =~ m/^(.*?)(\..*)?$/);
+    $docid_secnum = "" if (!defined $docid_secnum); 
+    
+    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];
+    
+    my $metaname = undef;
+    my $metapos = -1;
+    
+    $gsdl_cgi->generate_ok_message($self->get_metadata_from_archive_xml($gsdl_cgi, &util::filename_cat($archive_dir, $doc_filename), $metaname, $metapos, $docid_secnum));
+
+}
+
 sub get_metadata_from_archive_xml
 {
@@ -935 +976 @@
 }
 
+# unused and untested
+sub get_text_from_archive_xml
+{
+    my $self = shift @_;
+    my ($gsdl_cgi, $doc_xml_filename, $secid) = @_;
+
+    # To monitor which section/subsection number we are in
+    my @start_rules = ('Section' => \&dxml_start_section);
+
+    # set the callback functions for the elements in doc.xml we're interested in, <Content>
+    my @rules = 
+    ( 
+        _default => 'raw',
+        'Content' => \&gfdxml_text # gfdxml = get from doc xml?
+    );
+        
+    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, {secid => $secid});
+        
+        if(defined $parser->{'pad'}->{'textcontent'})
+        {
+            return $parser->{'pad'}->{'textcontent'};
+        }
+        else
+        {
+            return "";
+        }
+    }
+}
+
 sub gfdxml_metadata
 {
@@ -962 +1053 @@
 }
 
+# unused and untested - for get_archives_text
+sub gfdxml_text
+{
+    my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
+    
+    if($parser->{'parameters'}->{'secid'} eq $parser->{'parameters'}->{'curr_section_num'})
+    {
+        $parser->{'pad'}->{'textcontent'} = $attrHash->{'_content'}; # the textnode content
+
+    }
+    else {
+        return;
+    }
+}
 
 sub dxml_start_section
@@ -993 +1098 @@
     }
     else {
-    # back up to parent section => lopp off tail
+    # back up to parent section => lop off tail
     $new_secnum = $old_secnum;
     $new_secnum =~ s/\.\d+$//;
@@ -1001 +1106 @@
     $parser->{'parameters'}->{'curr_section_num'}   = $new_secnum;
     
-    1;
+    1;
 }