Changeset 19498 for gsdl/trunk

Timestamp:

2009-05-18T11:00:51+12:00 (15 years ago)

Author:

davidb

Message:

Supporting routines that exploit the new 'metafiles' structures, introduction to track which metadata.xml file a piece of metadata came from

File:

• gsdl/trunk/perllib/inexport.pm (modified) (9 diffs)

gsdl/trunk/perllib/inexport.pm

@@ -125 +125 @@
     }
 
-    # figure of if new file or not
+    # figure out if new file or not
     if (defined $full_prev_all_files->{$full_curr_file}) {
         
@@ -152 +152 @@
     # In building the final list of files to delete, we test to see if
     # it exists on the filesystem and if it does (unusual for a file
-    # that's allegedly deleted!) , supress it from going into the final
+    # that's allegedly deleted!), supress it from going into the final
     # list
 
@@ -175 +175 @@
 
 
+sub is_assoc_file
+{
+    my ($file,$doc_rec) = @_;
+
+    foreach my $af (@{$doc_rec->{'assoc-file'}}) {
+    return 1 if ($af eq $file);
+    }
+
+    return 0;
+}
+
+
 sub _mark_docs_for_deletion
 {
-    my ($archive_info,$deleted_files_ref,$archivedir,$verbosity,$mode_text) = @_;
+    my ($archive_info,$block_hash,$deleted_files,$archivedir,$verbosity,$mode_text) = @_;
 
     my $doc_db = "archiveinf-doc.gdb";
@@ -186 +198 @@
 
     # record files marked for deletion in arcinfo
-    foreach my $file (@$deleted_files_ref) {
+    foreach my $file (@$deleted_files) {
     # use 'archiveinf-src' GDBM file to look up all the OIDs
     # this file is used in (note in most cases, it's just one OID)
@@ -194 +206 @@
     foreach my $oid (@$oids) {
 
-        # Find out if it's an assoc file or main doc
+        # Find out if it's a main doc, assoc file, or metadata
 
         my $doc_rec = GDBMUtils::gdbmRecordToHash($arcinfo_doc_filename,$oid);
-        if ($doc_rec->{'src-file'}->[0] eq $file) {
-        # It's the main doc
-        # => mark it for deletion
-    
-        if ($verbosity>1) {
-            print STDERR "$oid ($file) marked to be $mode_text on next buildcol.pl\n";
-        }
-        $archive_info->set_status_info($oid,"D");
-
-        my $val = &GDBMUtils::gdbmDatabaseGet($arcinfo_doc_filename,$oid);
-        my ($index_status) = ($val =~ m/^<index-status>(.*)$/m);
-        if ($index_status ne "D") {
-            $val =~ s/^<index-status>(.*)$/<index-status>D/m;
-            &GDBMUtils::gdbmDatabaseSet($arcinfo_doc_filename,$oid,$val);
-        }
-        }
-        else {
+
+        if (is_assoc_file($file,$doc_rec)) {
         # assoc file => mark it for re-indexing (safest thing to do)
         my $curr_status = $archive_info->get_status_info($oid);
@@ -229 +226 @@
             &GDBMUtils::gdbmDatabaseSet($arcinfo_doc_filename,$oid,$val);
         }
+        GDBMUtils::gdbmDatabaseRemove($arcinfo_src_filename,$file);
         }
-
-        GDBMUtils::gdbmDatabaseRemove($arcinfo_src_filename,$file);
+        else {
+        # either src-file or metadata.xml file linking to src-file
+
+        my $src_file;
+
+        if ($doc_rec->{'src-file'}->[0] ne $file) {
+            # it's a metadata file attached to this OID
+            # => workout the src-file it matches to
+
+            $src_file = $doc_rec->{'src-file'}->[0];
+            
+            my $src_filename = $src_file;
+            if (!&util::filename_is_absolute($src_file)) {
+            $src_filename = &util::filename_cat($ENV{'GSDLCOLLECTDIR'},$src_file);
+            }
+            
+            $block_hash->{'reindex_files'}->{$src_filename} = 1;
+        }
+        else {
+            # It's the main doc
+            # => make it the target and mark it for deletion
+            $src_file = $file;
+        }
+
+        # Whether the main file directly or indirectly, mark for deletion/reindex
+
+        if ($verbosity>1) {
+            print STDERR "$oid ($src_file) marked to be $mode_text on next buildcol.pl\n";
+        }
+        $archive_info->set_status_info($oid,"D");
+
+        my $val = &GDBMUtils::gdbmDatabaseGet($arcinfo_doc_filename,$oid);
+        my ($index_status) = ($val =~ m/^<index-status>(.*)$/m);
+        if ($index_status ne "D") {
+            $val =~ s/^<index-status>(.*)$/<index-status>D/m;
+            &GDBMUtils::gdbmDatabaseSet($arcinfo_doc_filename,$oid,$val);
+        }
+
+        GDBMUtils::gdbmDatabaseRemove($arcinfo_src_filename,$src_file);
+        }
+
     }
     }
@@ -245 +282 @@
 sub mark_docs_for_reindex
 {
-    my ($archive_info,$existing_files_ref,$archivedir,$verbosity) = @_;
+    my ($archive_info,$block_hash,$archivedir,$verbosity) = @_;
 
     # Reindexing is accomplished by deleting the previously indexed
@@ -260 +297 @@
     # deals with it appropriately.
 
+    my @existing_files = sort keys %{$block_hash->{'existing_files'}};
+
     my $doc_db = "archiveinf-doc.gdb";
     my $arcinfo_doc_filename = &util::filename_cat ($archivedir, $doc_db);
 
-
     my $archiveinf_timestamp = -M $arcinfo_doc_filename;
 
-    my $reindex_files_ref = [];
-
-    foreach my $existing_filename (@$existing_files_ref) {
+    my $reindex_files = [];
+
+    foreach my $existing_filename (@existing_files) {
     
     if (-M $existing_filename < $archiveinf_timestamp) {
@@ -281 +319 @@
 ###     print STDERR "**** Deleting existing file: $existing_file\n";
 
-        push(@$reindex_files_ref,$existing_file);
+        push(@$reindex_files,$existing_file);
+        $block_hash->{'reindex_files'}->{$existing_filename} = 1;
     }
 
     }
     
-    _mark_docs_for_deletion($archive_info,$reindex_files_ref,$archivedir,$verbosity, "reindex");
-
-    return @$reindex_files_ref;
+    _mark_docs_for_deletion($archive_info,$block_hash,$reindex_files,$archivedir,$verbosity, "reindex");
+
 }