Changeset 7588

Timestamp:

2004-06-11T11:41:29+12:00 (20 years ago)

Author:

kjdon

Message:

made a lot of changes to this but apparently it still doesn't work

File:

• trunk/gsdl/bin/script/unbuildv2.pl (modified) (20 diffs)

trunk/gsdl/bin/script/unbuildv2.pl

@@ -3 +3 @@
 ###########################################################################
 #
-# unindex.pl --
+# unbuildv2.pl --
 # A component of the Greenstone digital library software
 # from the New Zealand Digital Library Project at the 
@@ -33 +33 @@
 # updated. It probably needs some work done before using.
 
+# Katherine updated this but apparently it still doesn't work.
+
 BEGIN {
     die "GSDLHOME not set\n" unless defined $ENV{'GSDLHOME'};
@@ -40 +42 @@
 }
 
+use unbuildutil;
 use doc;
 use docsave;
@@ -127 +130 @@
     &get_toplevel_OID ($gdbmfile); # puts a list of the top level document OIDs into $toplevelinfo
 
-    tie (%infodb, "GDBM_File", $gdbmfile, 1, 0);
+    #tie (%infodb, "GDBM_File", $gdbmfile, 1, 0);
+    
+    #read ldb file into %infodb
+    &unbuildutil::read_gdbm($gdbmfile, \%infodb);
 
     #this makes the files specifying the hierarchy of subjects, titles etc
     foreach $classify (@$classifyinfo) {
     
-        &make_info_file($classify);
-
-    }
-
-   
+    &make_info_file($classify);
+
+    }
+
     
     #write out the collect.cfg
@@ -237 +242 @@
     next if ($key =~ /^CL/); #ignore classification entries
     next if ($value =~ /<section>/); #ignore docnum->OID entries
+    next if ($value =~ /<docoid>/); #ignore strange s133->OID entries
     next if ($key !~ /\d/); #ignore collection, browse entries
 
@@ -245 +251 @@
     $/ = "\n";
     #print STDERR "toplevel sections are: ", join ("\n", @$toplevelinfo);
+    #print STDERR "\n";
 }
     
@@ -251 +258 @@
     
     my ($gdb_str_ref, $hashref) = @_;
-    
     my @entries = split(/\n/, $gdb_str_ref);
     foreach $entry (@entries) {
@@ -257 +263 @@
     $$hashref{$key} .= '@' if defined $$hashref{$key};
     $$hashref{$key} .= $value;
-    #print STDERR "key: $key, value: $value\n";
+    
     }
    
@@ -295 +301 @@
     if (defined ($childs)) {
     $childs =~ s/\@$//;  #remove trailing @
-    if ($childs =~ s/^\"//) { #remove initial " if there
-        @items = split /\;\"/, $childs; #split into .1 .2 .3 etc
-        foreach $item (@items) {
-        push (@$children, "$parentoid$item");
-        }
-    }
-    else { # whole id is already present. split on ; and dont put parentoid in front
-        @items = split /\;/, $childs;
-        foreach $item (@items) {
-        push (@$children, "$item");
-        }
-    }
-    #print STDERR "children are: ", join ("\n", @$children);
-
+    @items = split /\;/, $childs; #split on ;
+    foreach $item (@items) {
+        $item =~ s/^\"/$parentoid/; # replace " with parentoid
+        push (@$children, "$item");
+    }
     }
     else {
       $children = [];
       }
-
-
-}    
+}
+
+    
 sub recurse_sections {
     my ($doc_obj, $children, $parentoid, $parentsection, $olddir) = @_;
@@ -482 +479 @@
     &mgcommand('.set query docnums');
     &mgcommand('.set term_freq off');
+    &mgcommand('.set briefstats off');
+    &mgcommand('.set memstats off');
+    &mgcommand('.set sizestats off');
+    &mgcommand('.set timestats off');
 }
 
@@ -567 +568 @@
 
     $/ = "\n";
-    print STDERR "classifications are: ", join(", ", @$classifyinfo);
+    #print STDERR "classifications are: ", join(", ", @$classifyinfo);
+    #print STDERR "\n";
+    
 }
 
@@ -577 +580 @@
     my ($classifier) = @_;
     my $info_file = "";
-
     my $entry = $infodb{$classifier};
-    #print STDERR "entry = $entry\n";
+
     my $hashref = {};
     &get_metadata($entry, $hashref);
@@ -586 +588 @@
     $classifier_name =~ s/\@$//; #remove trailing @
    
-    # check children - if all are documents at this level, dont need a 
-    # hierarchy so just put a AZList into config file
+    # check children - if there is a classifier node at this level, 
+    # use a hierarchy, otherwise use an AZList.
     
     my $children=[];
-    &get_whole_children($classifier, $hashref, $children);
-    if (&is_document($$children[0])) {
+    my $hierarchy = 0;
+    &get_whole_children($classifier, $hashref, $children); #returns a list of the child ids
+    foreach $child(@$children) {
+    if(not &is_document($child)) {
+        $hierarchy = 1;
+        last;
+    }
+    }
+
+    if (!$hierarchy) {
     &add_classify_cfg_list($classifier_name);
-    
     }else { #there is a hierarchy so create a file
-
     $info_file = "./etc/$classifier_name.txt";
 
@@ -615 +623 @@
 sub process_entry {
     my ($handle, $classifier_name, $classify_id) = @_;
-
     my $value = $infodb{$classify_id};
     
     my $hashref={};
-    
     &get_metadata($value, $hashref);
     my $title = $hashref->{'Title'};
@@ -631 +637 @@
         &add_doc_metadata($child, $classifier_name, $title);
     }else {
-        process_entry($handle, $classifier_name, $child);
+        &process_entry($handle, $classifier_name, $child);
     }
     }
@@ -660 +666 @@
     my ($handle, $classify_id, $title) = @_;
     #print STDERR "classify id= $classify_id, title= $title\n";
-    $title = &unicode::ascii2utf8($title);
-
+    $title = &unicode::ascii2utf8(\$title);
     my ($num) = $classify_id =~ /^CL\d\.(.*)$/; #remove the CL1. from the front
     
@@ -687 +692 @@
     $collect_cfg->{'creator'}="$username\@cs.waikato.ac.nz";
     $collect_cfg->{'maintainer'}="$username\@cs.waikato.ac.nz";
-    $collect_cfg->{'public'}="false";
+    $collect_cfg->{'public'}="true";
     $collect_cfg->{'beta'}="true";
     
     $collect_cfg->{'plugin'}=[];
-    push (@{$collect_cfg->{'plugin'}}, ["GMLPlug"]);
+    push (@{$collect_cfg->{'plugin'}}, ["GAPlug"]);
     push (@{$collect_cfg->{'plugin'}}, ["ArcPlug"]);
     push (@{$collect_cfg->{'plugin'}}, ["RecPlug"]);
@@ -704 +709 @@
     $collect_cfg->{'collectionmeta'}={};
     $collect_cfg->{'collectionmeta'}->{'collectionname'}="\"$collection\"";
-    $collect_cfg->{'collectionmeta'}->{'iconcollection'}="\"http://www.nzdl.org/gsdl/collect/$collection/images/$collection.gif\"";
-    $collect_cfg->{'collectionmeta'}->{'iconcollectionsmall'}="\"http://www.nzdl.org/gsdl/collect/$collection/images/${collection}sm.gif\"";
+    $collect_cfg->{'collectionmeta'}->{'iconcollection'}="\"_httpprefix_/collect/$collection/images/$collection.gif\"";
+    $collect_cfg->{'collectionmeta'}->{'iconcollectionsmall'}="\"_httpprefix_/collect/$collection/images/${collection}sm.gif\"";
     $collect_cfg->{'collectionmeta'}->{'collectionextra'} = "\"This is a collection rebuilt from CDROM.\"";
-
-      
-
 
 }
@@ -751 +753 @@
     my ($title) = $classifier_name =~ /^CL(.*)$/;
     my ($filename) = $file =~ /\/([^\/]*)$/;
-    my $entry = "Hierarchy hfile=$filename metadata=$classifier_name title=$title sort=Title";
+    my $entry = "Hierarchy -hfile $filename -metadata $classifier_name -buttonname $title -sort Title";
     push (@{$collect_cfg->{'classify'}},[$entry]);
    
@@ -762 +764 @@
     $collect_cfg->{'classify'} = [] unless defined $collect_cfg->{'classify'};
     my ($title) = $classifier =~ /^CL(.*)$/;
-    my $entry = "AZList metadata=$classifier title=$title";
+    my $entry = "AZList -metadata $classifier -buttonname $title";
     push (@{$collect_cfg->{'classify'}},[$entry]);
 }