Changeset 15696

Timestamp:

2008-05-26T13:05:16+12:00 (16 years ago)

Author:

mdewsnip

Message:

(Adding new DB support) Moved the GDBM-specific stuff out of infodb() into a new write_infodb_entry_gdbm() function, in preparation for adding sqlite support.

File:

• gsdl/trunk/perllib/basebuildproc.pm (modified) (14 diffs)

gsdl/trunk/perllib/basebuildproc.pm

@@ -34 +34 @@
 use doc;
 use docproc;
+use strict; no strict 'subs';
 use util;
 
@@ -330 +331 @@
     my $self = shift (@_);
     my ($doc_obj, $filename) = @_;
-    my $handle = $self->{'output_handle'};
-
+
+    # only output this document if it is a "indexed_doc" or "info_doc" (database only) document
     my $doctype = $doc_obj->get_doc_type();
-
-    # only output this document if it is a "indexed_doc" or "info_doc" (database only) document
     return if ($doctype ne "indexed_doc" && $doctype ne "info_doc");
 
     my $archivedir = "";
-
     if (defined $filename)
     {
     # doc_obj derived directly from file
-
     my ($dir) = $filename =~ /^(.*?)(?:\/|\\)[^\/\\]*$/;
     $dir = "" unless defined $dir;
@@ -361 +358 @@
     }
 
-
-    #GRB: moved 1/06/2004 from GRB01062004
     #add this document to the browse structure
     push(@{$self->{'doclist'}},$doc_obj->get_OID()) 
@@ -369 +364 @@
     # classify this document
     &classify::classify_doc ($self->{'classifiers'}, $doc_obj);
-    #GRB: end of moved block
 
     # this is another document
@@ -380 +374 @@
     my $doc_OID = $doc_obj->get_OID();
     my $first = 1;
-    my $url = "";
 
     $self->{'doc_mdprefix_fields'} = {};
@@ -391 +384 @@
         $section_OID = $doc_OID . "." . $section;
     }
+    my %section_infodb = ();
 
     # update a few statistics
     $self->{'num_bytes'} += $doc_obj->get_text_length ($section);
     $self->{'num_sections'} += 1 unless ($doctype eq "classification");
-
-    # output the section name
-    print $handle "[$section_OID]\n";
 
     # output the fact that this document is a document (unless doctype
@@ -403 +394 @@
     my $dtype = $doc_obj->get_metadata_element ($section, "doctype");
     if (!defined $dtype || $dtype !~ /\w/) {
-        print $handle "<doctype>doc\n";
+        $section_infodb{"doctype"} = "doc";
     }
 
@@ -411 +402 @@
     # explicitly add <hastxt> as this is preserved as metadata when
     # the database file is loaded in
-
     if (defined $filename)
     {
         # doc_obj derived directly from file
         if ($doc_obj->get_text_length($section) > 0) {
-        print $handle "<hastxt>1\n";
+        $section_infodb{"hastxt"} = "1";
         } else {
-        print $handle "<hastxt>0\n";
+        $section_infodb{"hastxt"} = "0";
         }
     }
@@ -443 +433 @@
         # special case for URL metadata
         if ($field =~ /^URL$/i) {
-                    $url .= "[$value]\n";
-            $url .= "<section>$section_OID\n";
-                    $url .= '-' x 70 . "\n";
+            $self->write_infodb_entry($value, { 'section'  => $section_OID });
         }
 
         if (!defined $self->{'dontdb'}->{$field}) {
-            print $handle "<$field>$value\n";
+            $section_infodb{$field} = $value;
 
             if ($section eq "")
@@ -465 +453 @@
         foreach my $prefix (keys %$doc_mdprefix_fields)
         {
-        print $handle "<metadataset>$prefix\n";
+        $section_infodb{"metadataset"} = $prefix;
 
         foreach my $field (keys %{$doc_mdprefix_fields->{$prefix}})
@@ -471 +459 @@
             my $val = $doc_mdprefix_fields->{$prefix}->{$field};
 
-            print $handle "<metadatalist-$prefix>$field\n";
-            print $handle "<metadatafreq-$prefix-$field>$val\n";
+            $section_infodb{"metadatalist-$prefix"} = $field;
+            $section_infodb{"metadatafreq-$prefix-$field"} = $val;
         }
-
         }
     }
@@ -481 +468 @@
     # explicitly add <archivedir> as this is preserved as metadata when
     # the database file is loaded in
-
     if (defined $filename)
     {
         # output archivedir if at top level
         if ($section eq $doc_obj->get_top_section()) {
-        print $handle "<archivedir>$archivedir\n";
+        $section_infodb{"archivedir"} = $archivedir;
         }
     }
@@ -492 +478 @@
     # output document display type
     if ($first) {
-        print $handle "<thistype>$thistype\n";
-    }
-
+        $section_infodb{"thistype"} = $thistype;
+    }
 
     if ($self->{'db_level'} eq "document") {
         # doc num is num_docs not num_sections
         # output the matching document number
-        print $handle "<docnum>$self->{'num_docs'}\n";
-            
-    } else {
+        $section_infodb{"docnum"} = $self->{'num_docs'};
+    }
+    else {
         # output a list of children
         my $children = $doc_obj->get_children ($section);
         if (scalar(@$children) > 0) {
-        print $handle "<childtype>$childtype\n";
-        print $handle "<contains>";
-        my $firstchild = 1;
-        foreach my $child (@$children) {
-            print $handle ";" unless $firstchild;
-            $firstchild = 0;
-            if ($child =~ /^.*?\.(\d+)$/) {
-            print $handle "\".$1";
-            } else {
-            print $handle "\".$child";
+        $section_infodb{"childtype"} = $childtype;
+        my $contains = "";
+        foreach my $child (@$children)
+        {
+            $contains .= ";" unless ($contains eq "");
+            if ($child =~ /^.*?\.(\d+)$/)
+            {
+            $contains .= "\".$1";
+            }
+            else {
+            $contains .= "\".$child";
             }
         }
-        print $handle "\n";
-        }
-        #output the matching doc number
-        print $handle "<docnum>$self->{'num_sections'}\n";
-        
+        $section_infodb{"contains"} = $contains;
+        }
+        # output the matching doc number
+        $section_infodb{"docnum"} = $self->{'num_sections'};
     } 
     
-    print $handle '-' x 70, "\n";
-
+    $self->write_infodb_entry($section_OID, \%section_infodb);
     
     # output a database entry for the document number
     if ($self->{'db_level'} eq "document") {
-        print $handle "[$self->{'num_docs'}]\n";
-        print $handle "<section>$doc_OID\n";
+        $self->write_infodb_entry($self->{'num_docs'}, { 'section'  => $doc_OID });
     }
     else {
-        print $handle "[$self->{'num_sections'}]\n";
-        print $handle "<section>$section_OID\n";
-    }
-    print $handle '-' x 70, "\n";
-    
-        # output entry for url
-        if ($url ne "") {
-            print $handle $url;
-        }
+        $self->write_infodb_entry($self->{'num_sections'}, { 'section'  => $section_OID });
+    }
 
     $first = 0;
@@ -547 +523 @@
     last if ($self->{'db_level'} eq "document"); # if no sections wanted, only add the docs
     }
-
-    #GRB01062004: see code above moved from here
+}
+
+
+sub write_infodb_entry
+{
+    my $self = shift(@_);
+
+    $self->write_infodb_entry_gdbm(@_);
+}
+
+    
+sub write_infodb_entry_gdbm
+{
+    my $self = shift(@_);
+    my $infodb_key = shift(@_);
+    my $infodb_map = shift(@_);
+
+    my $handle = $self->{'output_handle'};
+
+    print $handle "[$infodb_key]\n";
+    foreach my $infodb_value_key (keys(%$infodb_map))
+    {
+    print $handle "<$infodb_value_key>" . $infodb_map->{$infodb_value_key} . "\n";
+    }
+    print $handle '-' x 70, "\n";
 }