Changeset 22900

Timestamp:

2010-09-14T14:12:16+12:00 (13 years ago)

Author:

kjdon

Message:

getting this to work properly

File:

• main/trunk/greenstone2/perllib/plugins/LOMPlugin.pm (modified) (8 diffs)

main/trunk/greenstone2/perllib/plugins/LOMPlugin.pm

@@ -26 +26 @@
 
 ### Note this plugin currently can't download source documents from outside if you are behind a firewall.
+# Unless, you set the http_proxy environment variable to be your proxy server, 
+# and set proxy_user and proxy_password in .wgetrc file in home directory. 
+# (does that work on windows??)
 
 package LOMPlugin;
@@ -155 +158 @@
     $lom_srcdoc =~ s/^$base_dir($dirsep)//;
     $self->{'extra_blocks'}->{$file}++;
-
     $file_re = $lom_srcdoc;
     }
@@ -161 +163 @@
     $file_re = $tail;
     }
-    $file_re =~ s/\./\\\./g;
-
+    $file_re = &util::filename_to_regex($file_re);
     $self->{'lom_srcdoc'} = undef; # reset for next file to be processed
 
     push(@$extrametakeys,$file_re);
     $extrametadata->{$file_re} = $self->{'saved_metadata'};
-
+    if (defined $lom_srcdoc) {
+    # copied from oaiplugin
+    if (!defined $extrametafile->{$file_re}) {
+        $extrametafile->{$file_re} = {};
+    }
+     #maps the file to full path
+    $extrametafile->{$file_re}->{$file} = $filename_full_path;
+    }
+    
     return 1;
 }
@@ -268 +277 @@
     my $outhandle = $self->{'outhandle'};
 
-    print STDERR "<Processing n='$file' p='LOMPlugin'>\n" if ($gli);
-
-    print $outhandle "LOMPLug: processing $file\n";
-
     my $cursection = $doc_obj->get_top_section();
     $doc_obj->add_utf8_text($cursection, $$textref);
@@ -583 +588 @@
 
     my $full_file_url = &util::filename_cat($output_dir,$file_url);
-
+    # the path to srcdoc will be used later in extrametadata to associate
+    # the lom metadata with the document. Needs to be relative to current
+    # directory.
+    my $srcdoc_path = &util::filename_cat("_gsdldown.all", $file_url);
     my $check_timestamp = $self->{'check_timestamp'};
     my $status;
@@ -610 +618 @@
         $wget_cmd = "wget -nv -P \"$output_dir\" --timestamping -k -p \"$doc_url\""; 
     }
-    #my $wget_cmd = "wget -nv -P$output_dir --timestamping -k -p \"$doc_url\""; 
-    ### my $full_dir_url = &File::Basename::dirname($full_file_url);
-
-    ## my $wget_cmd = "wget -O $full_file_url --timestamping -k \"$doc_url\"";
-    #my $wget_cmd = "wget -nv -P$output_dir --timestamping -x -k \"$doc_url\"";
-    print STDERR "**** wget = $wget_cmd\n";
+    ##print STDERR "**** wget = $wget_cmd\n";
 
     
@@ -623 +626 @@
     }
     if ($status==0) {
-        $self->{'lom_srcdoc'} = $full_file_url; 
+        $self->{'lom_srcdoc'} = $srcdoc_path;   
     }
     else {
@@ -633 +636 @@
     # not time-stamping and file already exists
     $status=0;
-    $self->{'lom_srcdoc'} = $full_file_url; 
+    $self->{'lom_srcdoc'} = $srcdoc_path;   
     }