Changeset 16206

Timestamp:

2008-06-26T13:52:53+12:00 (16 years ago)

Author:

mdewsnip

Message:

Added support for a "-split_greenstone_dates" option that builds a hierarchy from Greenstone "YYYYMMDD" date metadata values.

Location:

gsdl/trunk/src/recpt

Files:

• dynamicclassifieraction.cpp (modified) (7 diffs)
• dynamicclassifieraction.h (modified) (1 diff)

gsdl/trunk/src/recpt/dynamicclassifieraction.cpp

@@ -198 +198 @@
   text_t classifier_node_OID = args["dcn"];
   text_t classifier_node_metadata_value = classifier_node_OID;
+  text_t classifier_node_metadata_value_filter = classifier_node_metadata_value + "|*";
 
   // Split the classifier node OID into its components, then remove any we've already dealt with
@@ -206 +207 @@
     classifier_node_OID_parts_remaining.pop_front();
 
-    // Update classifier_node_metadata_value to remove any grouping node components
+    // Update classifier_node_metadata_value (and consequently the filter) to remove the grouping node component
     joinchar (classifier_node_OID_parts_remaining, '|', classifier_node_metadata_value);
+    classifier_node_metadata_value_filter = classifier_node_metadata_value + "|*";
   }
   if (classifier_options["-use_hlist_at_top"] == "1")
   {
     classifier_node_OID_parts_remaining.pop_front();
+  }
+
+  // If "-split_greenstone_dates" is on the metadata value and filter doesn't quite match the OID -- fix this
+  if (classifier_options["-split_greenstone_dates"] == "1")
+  {
+    classifier_node_metadata_value.replace("|", "");
+    classifier_node_metadata_value_filter = classifier_node_metadata_value;
+    classifier_node_metadata_value_filter += ((classifier_node_metadata_value_filter.size() == 8) ? "|*" : "*");
   }
 
@@ -217 +227 @@
   if (classifier_node_metadata_value == "")
   {
-    output_child_classifier_nodes (classifier_node_OID, classifier_node_metadata_value, current_metadata_value_filter, classifier_node_indent, classifier_options, args, collectproto, browsers, disp, outconvert, textout, logout);
+    output_child_classifier_nodes (classifier_node_OID, "", current_metadata_value_filter, classifier_node_indent, classifier_options, args, collectproto, browsers, disp, outconvert, textout, logout);
   }
 
@@ -227 +237 @@
 
     // Output the child classifier nodes
-    text_t metadata_value_filter = classifier_node_metadata_value + "|*";
-    output_child_classifier_nodes (classifier_node_OID, classifier_node_metadata_value, metadata_value_filter, classifier_node_indent, classifier_options, args, collectproto, browsers, disp, outconvert, textout, logout);
+    output_child_classifier_nodes (classifier_node_OID, classifier_node_metadata_value, classifier_node_metadata_value_filter, classifier_node_indent, classifier_options, args, collectproto, browsers, disp, outconvert, textout, logout);
 
     // Get the document nodes at this level
@@ -316 +325 @@
   while (metadata_value_iterator != metadata_values_response.docInfo.end())
   {
-    text_t metadata_value = (*metadata_value_iterator).OID;
-
-    // Is this metadata value hierarchical?
-    text_t::iterator hierarchy_split_position = findchar(metadata_value.begin(), metadata_value.end(), '|');
-    if (hierarchy_split_position != metadata_value.end())
-    {
-      // Yes, so use the first part of the hierarchy only
-      metadata_value = substr(metadata_value.begin(), hierarchy_split_position);
-    }
-
-    // Create a node for this metadata value if we haven't seen it before
-    if (hlist_classifier_nodes.find(metadata_value) == hlist_classifier_nodes.end())
-    {
-      hlist_classifier_nodes[metadata_value] = 0;
+    text_t hierarchical_metadata_value = split_metadata_value ((*metadata_value_iterator).OID, classifier_options);
+
+    // Assume for now we're always at the top
+    text_t hlist_metadata_value = hierarchical_metadata_value;
+
+    // Determine the label for the hlist classifier node from the metadata value
+    text_tarray hlist_metadata_value_parts;
+    splitchar (hlist_metadata_value.begin(), hlist_metadata_value.end(), '|', hlist_metadata_value_parts);
+    text_t hlist_classifier_node_label = hlist_metadata_value_parts.front();
+
+    // Create a node for this value if we haven't seen it before
+    if (hlist_classifier_nodes.find(hlist_classifier_node_label) == hlist_classifier_nodes.end())
+    {
+      hlist_classifier_nodes[hlist_classifier_node_label] = 0;
     }
 
     // Increment the occurrence count
-    hlist_classifier_nodes[metadata_value] += (*metadata_value_iterator).result_num;
+    hlist_classifier_nodes[hlist_classifier_node_label] += (*metadata_value_iterator).result_num;
 
     metadata_value_iterator++;
@@ -446 +455 @@
   while (metadata_value_iterator != metadata_values_response.docInfo.end())
   {
-    text_t metadata_value = (*metadata_value_iterator).OID;
+    text_t hierarchical_metadata_value = split_metadata_value ((*metadata_value_iterator).OID, classifier_options);
+    text_t classifier_node_hierarchical_metadata_value = split_metadata_value (classifier_node_metadata_value, classifier_options);
 
     // We need to remove the current position from the metadata value to leave the child metadata value
-    text_t child_metadata_value = metadata_value;
-    if (starts_with(metadata_value, classifier_node_metadata_value + "|"))
-    {
-      child_metadata_value = substr(metadata_value.begin() + (classifier_node_metadata_value + "|").size(), metadata_value.end());
+    text_t child_metadata_value = hierarchical_metadata_value;
+    if (starts_with(hierarchical_metadata_value, classifier_node_hierarchical_metadata_value + "|"))
+    {
+      child_metadata_value = substr(hierarchical_metadata_value.begin() + (classifier_node_hierarchical_metadata_value + "|").size(), hierarchical_metadata_value.end());
     }
 
@@ -499 +509 @@
 
 
+text_t dynamicclassifieraction::split_metadata_value (text_t metadata_value, text_tmap classifier_options)
+{
+  text_t hierarchical_metadata_value = metadata_value;
+
+  // Add hierarchy separators into Greenstone date values (YYYYMMDD -> YYYY|MM|DD)
+  if (classifier_options["-split_greenstone_dates"] == "1")
+  {
+    if (metadata_value.size() == 4 || metadata_value.size() == 6 || metadata_value.size() == 8)
+    {
+      // Add year
+      hierarchical_metadata_value = substr(metadata_value.begin(), metadata_value.begin() + 4);
+      if (metadata_value.size() == 6 || metadata_value.size() == 8)
+      {
+    // Add month
+    hierarchical_metadata_value += "|" + substr(metadata_value.begin() + 4, metadata_value.begin() + 6);
+    if (metadata_value.size() == 8)
+    {
+      // Add day
+      hierarchical_metadata_value += "|" + substr(metadata_value.begin() + 6, metadata_value.begin() + 8);
+    }
+      }
+    }
+  }
+
+  return hierarchical_metadata_value;
+}
+
+
 void dynamicclassifieraction::display_classifier_nodes (FilterResponse_t classifier_nodes_response,
                                 text_t classifier_nodes_type,

gsdl/trunk/src/recpt/dynamicclassifieraction.h

@@ -90 +90 @@
                       ostream &textout, ostream &logout);
 
+  text_t split_metadata_value (text_t metadata_value, text_tmap classifier_options);
+
   void display_classifier_nodes (FilterResponse_t classifier_nodes_response,
                  text_t classifier_nodes_type,