Changeset 33302 for main/trunk/greenstone2/perllib/basebuilder.pm

Timestamp:

2019-07-05T22:23:48+12:00 (5 years ago)

Author:

ak19

Message:

1. Adding GPSMapOverlayLabel extracted from GPS.mapOverlay meta to text indexes for searching, as with Coordinte and CoordShort. 2. Added a shortname for this index, ML for MapLabel. 3. On testing the indexing of the GPSMapOverlayLabel text, the old problem of increasingly duplicated Coordinate/CoordShort and now also GPSMapOverlayLabel meta in the infodb reappeared. Dr Bainbridge explained why this was (documented as comments in this commit) and fixed the problem by not processing GPS.mapOverlay meta into Coordinate and Label meta during the infodb pass (and dummy pass, so specifically specifically non-text passes) of buildcol. A natural consequence is that to check whether Coord and Label meta have been indexed, can no longer check the index/text/col.jdb but need to use Luke (if a lucene collection ) to check contents of index/sidx and index/didx. 4. An important change needed for the bugfix in 3 is reordering call to &classify::reconstruct_doc_objs_metadata() in basebuilder.pm to take place AFTER build_proc->set_mode(infodb) has taken place. 5. Changed cross-files global variables declared in doc.pm from our to my variables and tested this works.

File:

• main/trunk/greenstone2/perllib/basebuilder.pm (modified) (2 diffs)

main/trunk/greenstone2/perllib/basebuilder.pm

@@ -480 +480 @@
     }
 
-    if ($self->{'incremental'}) {
-    # reconstruct doc_obj metadata from database for all docs
-    $reconstructed_docs 
-        = &classify::reconstruct_doc_objs_metadata($infodb_type, 
-                               $infodb_file_path,
-                               $database_recs);
-    }
-
     # set up the document processor
 
@@ -498 +490 @@
     $self->{'buildproc'}->set_store_text(1);
 
+    if ($self->{'incremental'}) {
+    # reconstruct doc_obj metadata from database for all docs
+    $reconstructed_docs 
+        = &classify::reconstruct_doc_objs_metadata($infodb_type, 
+                               $infodb_file_path,
+                               $database_recs);
+    }
+    
     # make_infodatabase needs full reset even for incremental build
     # as incremental works by reconstructing all docs from the database and