From bee77791622171590fe18da817b2f245a2f31736 Mon Sep 17 00:00:00 2001
From: Alex Skrenchuk <alexskr@stanford.edu>
Date: Fri, 26 Jan 2024 20:17:19 -0800
Subject: [PATCH] use max_depth calculated by owlapi_wrapper for UMLS
 ontologies

Get max depth from the metrics.csv file which is already generated
by owlapi_wrapper when new submission of UMLS ontology is created.
Ruby code/sparql for calculating max_depth fails for large UMLS
ontologie with AllegroGraph backend

Addresses #181
---
 lib/ontologies_linked_data/metrics/metrics.rb |  5 ++--
 .../models/ontology_submission.rb             | 24 ++++++++++---------
 test/models/test_ontology_submission.rb       |  2 +-
 3 files changed, 17 insertions(+), 14 deletions(-)

diff --git a/lib/ontologies_linked_data/metrics/metrics.rb b/lib/ontologies_linked_data/metrics/metrics.rb
index 5909db69..6ec9a208 100644
--- a/lib/ontologies_linked_data/metrics/metrics.rb
+++ b/lib/ontologies_linked_data/metrics/metrics.rb
@@ -35,7 +35,7 @@ def self.metrics_for_submission(submission, logger)
         logger.flush
 
         # re-generate metrics file
-        submission.generate_metrics_file2(cls_metrics[:classes], indiv_count, prop_count, cls_metrics[:maxDepth])
+        submission.generate_metrics_file(cls_metrics[:classes], indiv_count, prop_count, cls_metrics[:maxDepth])
         logger.info("generation of metrics file finished")
         logger.flush
 
@@ -54,10 +54,11 @@ def self.max_depth_fn(submission, logger, is_flat, rdfsSC)
       if (mx_from_file && mx_from_file.length == 2 && mx_from_file[0].length >= 4)
       then
         max_depth = mx_from_file[1][3].to_i
+        logger.info("Metrics max_depth retrieved #{max_depth} from the metrics csv file.")
       else
         logger.info("Unable to find metrics providing max_depth in file for submission #{submission.id.to_s}.  Using ruby calculation of max_depth.")  
         roots = submission.roots
-      
+
         unless is_flat
           depths = []
           roots.each do |root|
diff --git a/lib/ontologies_linked_data/models/ontology_submission.rb b/lib/ontologies_linked_data/models/ontology_submission.rb
index a0146920..24f7bfb8 100644
--- a/lib/ontologies_linked_data/models/ontology_submission.rb
+++ b/lib/ontologies_linked_data/models/ontology_submission.rb
@@ -418,14 +418,7 @@ def metrics_from_file(logger=nil)
         metrics
       end
 
-      def generate_metrics_file(class_count, indiv_count, prop_count)
-        CSV.open(self.metrics_path, "wb") do |csv|
-          csv << ["Class Count", "Individual Count", "Property Count"]
-          csv << [class_count, indiv_count, prop_count]
-        end
-      end
-
-      def generate_metrics_file2(class_count, indiv_count, prop_count, max_depth)
+      def generate_metrics_file(class_count, indiv_count, prop_count, max_depth)
         CSV.open(self.metrics_path, "wb") do |csv|
           csv << ["Class Count", "Individual Count", "Property Count", "Max Depth"]
           csv << [class_count, indiv_count, prop_count, max_depth]
@@ -433,10 +426,11 @@ def generate_metrics_file2(class_count, indiv_count, prop_count, max_depth)
       end
 
       def generate_umls_metrics_file(tr_file_path=nil)
-        tr_file_path ||= self.triples_file_path
+        tr_file_path ||= triples_file_path
         class_count = 0
         indiv_count = 0
         prop_count = 0
+        max_depth = 0
 
         File.foreach(tr_file_path) do |line|
           class_count += 1 if line =~ /owl:Class/
@@ -444,7 +438,15 @@ def generate_umls_metrics_file(tr_file_path=nil)
           prop_count += 1 if line =~ /owl:ObjectProperty/
           prop_count += 1 if line =~ /owl:DatatypeProperty/
         end
-        self.generate_metrics_file(class_count, indiv_count, prop_count)
+
+        # Get max depth from the metrics.csv file which is already generated
+        # by owlapi_wrapper when new submission of UMLS ontology is created.
+        # Ruby code/sparql for calculating max_depth fails for large UMLS
+        # ontologie with AllegroGraph backend
+        metrics_from_owlapi = metrics_from_file
+        max_depth = metrics_from_owlapi[1][3] unless metrics_from_owlapi.empty?
+
+        generate_metrics_file(class_count, indiv_count, prop_count, max_depth)
       end
 
       def generate_rdf(logger, reasoning: true)
@@ -452,7 +454,7 @@ def generate_rdf(logger, reasoning: true)
 
         if self.hasOntologyLanguage.umls?
           triples_file_path = self.triples_file_path
-          logger.info("Using UMLS turtle file found, skipping OWLAPI parse")
+          logger.info("UMLS turtle file found; doing OWLAPI parse to extract metrics")
           logger.flush
           mime_type = LinkedData::MediaTypes.media_type_from_base(LinkedData::MediaTypes::TURTLE)
           generate_umls_metrics_file(triples_file_path)
diff --git a/test/models/test_ontology_submission.rb b/test/models/test_ontology_submission.rb
index dffab3c5..3049b6f1 100644
--- a/test/models/test_ontology_submission.rb
+++ b/test/models/test_ontology_submission.rb
@@ -603,7 +603,7 @@ def test_umls_metrics_file
     metrics = sub.metrics_from_file(Logger.new(sub.parsing_log_path))
     assert !metrics.nil?, "Metrics is nil: #{metrics}"
     assert !metrics.empty?, "Metrics is empty: #{metrics}"
-    metrics.each { |m| assert_equal 3, m.length }
+    metrics.each { |m| assert_equal 4, m.length }
     assert_equal "Individual Count", metrics[0][1]
     assert_equal 133, metrics[1][0].to_i
   end