Update script extract to generate a full graph as turtle

extract.py

@@ -12,6 +12,7 @@
 # Importing required modules
 #==============================================================================
 
+import glob
 from rdflib import Graph
 
 
@@ -26,16 +27,19 @@ CORPUS_DIR = "corpus/"
 CTS_DIR = "cts/"
 
 # Config Definition
+dash_file = "http://datashapes.org/dash.ttl"
 schema_file = "unl-rdf-schema.ttl"
 semantic_net_file = "semantic-net.ttl"
+cts_file = "transduction-schemes.ttl"
 c_param_file = "config-parameters.ttl"
 req_onto_file = "requirement-ontology.ttl"
 sys_onto_file = "system-ontology.ttl"
 f_param_file = "ontology-parameters.ttl"
 
 # Dev Tests
-corpus = "CCTP-SRSA-IP-20210625/"
-req_300_file = "SRSA-IP_STB_PHON_00300.ttl"
+base_uri = "https://unsel.tetras-libre.fr/tenet/working"
+corpus = "CCTP-SRSA-IP-20210831/"
+req_file = "SRSA-IP_STB_PHON_00100.ttl"
 
 
 #==============================================================================
@@ -63,6 +67,10 @@ def load_config(g):
     g.parse(file_ref)
     print("----- Semantic Net Definition (" + str(len(g)) + ")")
     
+    file_ref = dash_file
+    g.parse(file_ref)
+    print("----- Data Shapes Dash (" + str(len(g)) + ")")
+    
     file_ref = CONFIG_DIR + c_param_file
     g.parse(file_ref)
     print("----- Config Parameter Definition (" + str(len(g)) + ")")
@@ -84,119 +92,39 @@ def load_frame(g):
     print("----- Ontology Parameters (" + str(len(g)) + ")")
     
     
-def load_sentence(g, sentence_file):
+def load_sentences(g):
     print("-- Sentence Loading:")
     
-    file_ref = CORPUS_DIR + corpus + sentence_file
+    #file_ref = CORPUS_DIR + corpus + sentence_file
+    target_ref = CORPUS_DIR + corpus + '**/*.ttl'
+    for file_ref in glob.glob(target_ref, recursive = True):
         g.parse(file_ref)
-    print("----- " + sentence_file + " (" + str(len(g)) + ")")
-     
-
-#==============================================================================
-# Graph Preprocessing
-#==============================================================================    
-    
-def bypass_reification(g):
-    print("-- Bypass Reification:")
-    cts_group = "preprocessing/reification/"
-    
-    queryString = read_query(cts_group, 201)
-    g.update(queryString) 
-    print("----- update to extend UNL relation (" + str(len(g)) + ")")  
-    
-    queryString = read_query(cts_group, 202)
-    g.update(queryString) 
-    print("----- update to bypass scopes (" + str(len(g)) + ")")  
+        print("----- " + file_ref + " (" + str(len(g)) + ")") 
             
        
 
 #==============================================================================
-# Net Extension (CTS Application)
+# CT Schemes for Transduction Process
 #==============================================================================
     
-def extract_entity(g):
-    print("-- Net Extension to Extract Entity:")
-    cts_group = "net_extension/entity_extraction/"
+def load_cts(g):
+    print("-- CTS Loading:")
     
-    queryString = read_query(cts_group, 301)
-    g.update(queryString) 
-    print("----- update to create Entity net (" + str(len(g)) + ")")    
-    print("-------- net instances: ")   
-    for r in g.query(read_query("selection/", 101)):
-        print("--------   " + f"{r.net} ")
-    
-    queryString = read_query(cts_group, 302)
-    g.update(queryString) 
-    print("----- update to create Entity net (" + str(len(g)) + ")")    
-    print("-------- net instances: ")   
-    for r in g.query(read_query("selection/", 101)):
-        print("--------   " + f"{r.net} ")
-    
-    queryString = read_query(cts_group, 303)
-    g.update(queryString) 
-    print("----- update to create Entity net (" + str(len(g)) + ")")    
-    print("-------- net instances: ")   
-    for r in g.query(read_query("selection/", 101)):
-        print("--------   " + f"{r.net} ")
-    
-    queryString = read_query(cts_group, 304)
-    g.update(queryString) 
-    print("----- update to create Entity net (" + str(len(g)) + ")")    
-    print("-------- net instances: ")   
-    for r in g.query(read_query("selection/", 101)):
-        print("--------   " + f"{r.net} ")
-    
-    queryString = read_query(cts_group, 305)
-    g.update(queryString) 
-    print("----- update to create Entity net (" + str(len(g)) + ")")    
-    print("-------- net instances: ")   
-    for r in g.query(read_query("selection/", 101)):
-        print("--------   " + f"{r.net} ")
-    
-    queryString = read_query(cts_group, 306)
-    g.update(queryString) 
-    print("----- update to create Entity net (" + str(len(g)) + ")")    
-    print("-------- net instances: ")   
-    for r in g.query(read_query("selection/", 101)):
-        print("--------   " + f"{r.net} ")
-    
-    queryString = read_query(cts_group, 307)
-    g.update(queryString) 
-    print("----- update to create Entity net (" + str(len(g)) + ")")    
-    print("-------- net instances: ")   
-    for r in g.query(read_query("selection/", 101)):
-        print("--------   " + f"{r.net} ")
-    
-    queryString = read_query(cts_group, 308)
-    g.update(queryString) 
-    print("----- update to create Entity net (" + str(len(g)) + ")")    
-    print("-------- net instances: ")   
-    for r in g.query(read_query("selection/", 101)):
-        print("--------   " + f"{r.net} ")
+    file_ref = CONFIG_DIR + cts_file
+    g.parse(file_ref)
+    print("----- All Schemes (" + str(len(g)) + ")")
         
     
 
 #==============================================================================
-# Ontology Generation
+# Result (export)
 #==============================================================================
      
-def classify_entity(g):
-    print("-- Enity Classification (From Net Ontology to System Ontology):")
-    cts_group = "generation/classification/"
-        
-    queryString = read_query(cts_group, 401)
-    g.update(queryString) 
-    print("----- update to create Entity net (" + str(len(g)) + ")")    
-    print("-------- entity classes: ")   
-    for r in g.query(read_query("selection/", 102)):
-        print("--------   " + f"{r.entity}")
+def export_result(g):
+    export_file = 'output.ttl'
+    print("-- Export result as turtle: " + export_file)
+    g.serialize(destination=export_file, base=base_uri, format='turtle')
     
-    queryString = read_query(cts_group, 402)
-    g.update(queryString) 
-    print("----- update to create Entity net (" + str(len(g)) + ")")    
-    print("-------- entity classes: ")   
-    for r in g.query(read_query("selection/", 102)):
-        print("--------   " + f"{r.entity}")
     
         
 #==============================================================================
@@ -211,16 +139,15 @@ def run():
         g = Graph()
         load_config(g)   
         load_frame(g) 
-        load_sentence(g, req_300_file)
     
-        print("\n" + "- Graph Preprocessing")
-        bypass_reification(g)
+        print("\n" + "- Preparation of Transduction Process")
+        load_cts(g)
     
-        print("\n" + "- Net Extension (CTS Application)")
-        extract_entity(g)
+        print("\n" + "- Data Source Imports")
+        load_sentences(g)
     
-        print("\n" + "- Ontology Generation ")
-        classify_entity(g)
+        print("\n" + "- Result")
+        export_result(g)
     
     except:
         print("!!! An exception occurred !!!")