Changeset 32615 for main/trunk/greenstone2/perllib/inexport.pm

Timestamp:

2018-11-20T10:54:07+13:00 (5 years ago)

Author:

kjdon

Message:

mods for incremental build. now we load in the reverse info db (archiveinf-src) and make changes to that, then write it out again at the end. did a bit of work to get html + supporting images working a bit better. for the cases where the main file is modified, or if the images are modified, or if links to those images are changed.

File:

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

main/trunk/greenstone2/perllib/inexport.pm

@@ -628 +628 @@
     my $archive_info = new arcinfo ($collectcfg->{'infodbtype'});
     $archive_info->load_info ($arcinfo_doc_filename);
+    # load in rev info so we don't overwrite existing info when we do incremental import
+    # from here on, make all changes to this object, then write out the file at the end.
+    $archive_info->load_rev_info($arcinfo_src_filename);
 
     if ($manifest eq "") {
@@ -713 +716 @@
     $block_hash->{'metadata_files'} = {};
     $block_hash->{'shared_fileroot'} = '';
-    # a new flag so we can tell we had a manifest way down in the plugins
-    # [jmt12]
     $block_hash->{'manifest'} = 'false';
     my $metadata = {};
@@ -855 +856 @@
         }
       }
-
+    
       # If we are not using complex inherited metadata (and thus have skipped
       # the global file scan) we need to at least check for a matching
@@ -1374 +1375 @@
 
 
-# this is used to delete "deleted" docs, and to remove old versions of "changed" docs
+# this is used to delete "deleted" docs and to remove old versions of "changed" docs
 # $mode is 'delete' or 'reindex'
 sub mark_docs_for_deletion
@@ -1396 +1397 @@
     # use 'archiveinf-src' info database file to look up all the OIDs
     # that this file is used in (note in most cases, it's just one OID)
-    
-    my $relfile = &util::abspath_to_placeholders($file);
-
-    my $src_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_src_filename, $relfile);
-    my $oids = $src_rec->{'oid'};
-    my $file_record_deleted = 0;
-
-    # delete the src record
-    my $src_infodb_file_handle = &dbutil::open_infodb_write_handle($infodbtype, $arcinfo_src_filename, "append");
-    &dbutil::delete_infodb_entry($infodbtype, $src_infodb_file_handle, $relfile);
-    &dbutil::close_infodb_write_handle($infodbtype, $src_infodb_file_handle);
-
-
-    foreach my $oid (@$oids) {
-
+);
+    my $oids = $archive_info->get_reverseinfo($file);
+    $archive_info->remove_reverseinfo($file);
+    
+    foreach my $oid (@$oids) {
+        # get the record for this OID from doc db
+        my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $oid);
         # find the source doc (the primary file that becomes this oid)
-        my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $oid);
         my $doc_source_file = $doc_rec->{'src-file'}->[0];
         $doc_source_file = &util::placeholders_to_abspath($doc_source_file);
@@ -1426 +1418 @@
         $block_hash->{'reindex_files'}->{$doc_source_file} = 1;
         
+        } else {
+
+        # the file to be deleted/reindexed is a primary file. We need to remove all references to this in the src db
+        my $assoc_files = $doc_rec->{'assoc-file'};
+        foreach my $assocfile (@$assoc_files) {
+            $assocfile = &util::placeholders_to_abspath($assocfile);
+            $archive_info->remove_reverseinfo($assocfile, $oid);
+            if (!defined $archive_info->get_reverseinfo($assocfile)) {
+            # nothing refers to it anymore, mark for reindex.
+            $block_hash->{'reindex_files'}->{$assocfile} = 1;
+            }
+        }
+
         }
         my $curr_status = $archive_info->get_status_info($oid);