Add lib (with the modules config and structure) and adapt the main script (create_graph) to use it

config.xml

+<?xml version="1.0" encoding="UTF-8"?>
+<config>
+    
+    <directory
+        config="config/"
+        frame="frame/"
+        corpus="corpus/"
+        cts="cts/"
+        output="output/"
+        />
+    
+    <file
+        schema = "unl-rdf-schema.ttl"
+        semantic_net = "semantic-net.ttl"
+        dash = "dash-data-shapes.ttl"
+        config_param = "config-parameters.ttl"
+        cts = "transduction-schemes.ttl"        
+        />
+    
+    <reference
+        default_base_uri = "https://unsel.tetras-libre.fr/tenet/working"
+        default_ontology_suffix = "-ontology.ttl"
+        />
+
+</config>

create_graph.py

+#!/usr/bin/python3.5
+# -*-coding:Utf-8 -*
+
+#==============================================================================
+# TENET: crea
+#------------------------------------------------------------------------------
+# Script to prepare the work structure for extraction process
+#==============================================================================
+
+#==============================================================================
+# Importing required modules
+#==============================================================================
+
+from lib import structure
+
+
+#==============================================================================
+# Parameters
+#==============================================================================
+
+
+# Dev Tests
+base_uri = "https://unsel.tetras-libre.fr/tenet/working"
+corpus_40 = "CCTP-SRSA-IP-20210831/"
+req_100 = "CCTP-SRSA-IP-20210831-R100/"
+req_200 = "CCTP-SRSA-IP-20210831-R200/"
+req_300 = "CCTP-SRSA-IP-20210831-R300/"
+req_1100 = "CCTP-SRSA-IP-20210831-R1100/"
+req_13900 = "CCTP-SRSA-IP-20210831-R13900/"
+corpus_comp = "COMP/"
+req_f1 = "COMP/F1/"
+req_34 = "COMP/34/"
+req_50 = "COMP/50/"
+corpus_ERTMS = "ERTMS/"
+corpus_PEV = "PEV-RSE-Approach/"
+   
+   
+#==============================================================================
+# Process
+#==============================================================================
+ 
+target_ref = "system"
+#target_ref = "environment"
+    
+#structure.create_extraction_graph(corpus_comp, 'COMP-01', target_ref) 
+#structure.create_extraction_graph(req_f1, 'C-F1', target_ref) 
+#structure.create_extraction_graph(req_50, 'C-50', target_ref) 
+#structure.create_extraction_graph(req_34, 'C-34', target_ref) 
+    
+structure.create_extraction_graph(req_13900, 'R13900', target_ref)
+structure.create_extraction_graph(req_100, 'R100f', target_ref) 
+#structure.create_extraction_graph(req_200, 'R200f', target_ref)
+#structure.create_extraction_graph(req_300, 'R300f', target_ref) 
+#structure.create_extraction_graph(req_1100, 'R1100f', target_ref) 
+    
+#structure.create_extraction_graph(corpus_40, 'Corpus-CCTP-40f', target_ref)
+    
+#structure.create_extraction_graph(corpus_ERTMS, 'Corpus-ERTMS', target_ref)
+#structure.create_extraction_graph(corpus_PEV, 'Corpus-PEV', target_ref)
+
+
+    
+    
+    

lib/config.py

+#!/usr/bin/python3.5
+# -*-coding:Utf-8 -*
+
+#==============================================================================
+# TENET: config
+#------------------------------------------------------------------------------
+# Class grouping the configuration data 
+#==============================================================================
+
+#==============================================================================
+# Importing required modules
+#==============================================================================
+
+from lxml import etree
+
+
+#==============================================================================
+# Default Values
+#==============================================================================
+
+# Dev Tests
+req_100 = "CCTP-SRSA-IP-20210831-R100/"
+req_200 = "CCTP-SRSA-IP-20210831-R200/"
+req_300 = "CCTP-SRSA-IP-20210831-R300/"
+req_1100 = "CCTP-SRSA-IP-20210831-R1100/"
+corpus_40 = "CCTP-SRSA-IP-20210831/"
+corpus_ERTMS = "ERTMS/"
+corpus_PEV = "PEV-RSE-Approach/"
+
+
+#==============================================================================
+# Class
+#==============================================================================
+
+class Config:
+    """Configuration data for Tenet process """
+
+    
+    def __init__(self, config_file, source_corpus, source_ref, target_ref):
+        
+        # -- Config XML Tree
+        config_tree = etree.parse(config_file)
+        
+        # -- Source and Target
+        self.source_corpus = source_corpus
+        self.source_ref = source_ref
+        self.target_ref = target_ref
+
+        # -- Working directories
+        c_dir = config_tree.xpath("directory")[0]
+        self.config_dir = c_dir.get("config")
+        self.frame_dir = c_dir.get("frame")
+        self.corpus_dir = c_dir.get("corpus")
+        self.cts_dir = c_dir.get("cts")
+        self.output_dir = c_dir.get("output")
+        
+        # -- Config File Definition
+        c_file = config_tree.xpath("file")[0]
+        self.schema_file = self.config_dir + c_file.get("schema")
+        self.semantic_net_file = self.config_dir + c_file.get("semantic_net")
+        self.dash_file = self.config_dir + c_file.get("dash")
+        self.config_param_file = self.config_dir + c_file.get("config_param")
+        self.cts_file = self.config_dir + c_file.get("cts")
+        
+        # -- Reference
+        c_ref = config_tree.xpath("reference")[0]
+        self.base_uri = c_ref.get("default_base_uri")
+        self.onto_suffix = c_ref.get("default_ontology_suffix")
+        
+        # -- Source File Definition
+        self.source_sentence_files = self.corpus_dir + self.source_corpus
+        self.source_sentence_files += '**/*.ttl'
+        
+        # -- Target File Definition
+        self.frame_ontology_file = self.frame_dir + self.target_ref
+        self.frame_ontology_file += self.onto_suffix
+        self.output_file = self.output_dir + self.source_ref + ".ttl"
+            
+            
+    # TODO: fonction to print
+            
+    
+#==============================================================================
+# Test
+#==============================================================================
+
+if __name__ == '__main__':
+    default_config_file = "../config.xml"
+    config = Config(default_config_file, req_100, "R100", "system")
+    print(config.__dict__)
\ No newline at end of file

prepare_work_data.py → lib/structure.py

@@ -2,9 +2,9 @@
 # -*-coding:Utf-8 -*
 
 #==============================================================================
-# TENET: prepare work data
+# TENET: structure
 #------------------------------------------------------------------------------
-# Prepare work data for extraction processing. 
+# Work structure for extraction processing. 
 #==============================================================================
 
 #==============================================================================
@@ -12,139 +12,70 @@
 #==============================================================================
 
 import glob
-from rdflib import Graph, Namespace, URIRef
+from rdflib import Graph
+from lib.config import Config
 
 
 #==============================================================================
 # Parameters
 #==============================================================================
 
-# Working directories
-CONFIG_DIR = "config/"
-FRAME_DIR = "frame/"
-CORPUS_DIR = "corpus/"
-CTS_DIR = "cts/"
-OUTPUT_DIR = "output/"
-
-# Config Definition
-TURTLE_SUFFIX = ".ttl"
-ONTO_FILE = "-ontology" + TURTLE_SUFFIX
-dash_file = "dash-data-shapes.ttl" # data from "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"
-
-# Dev Tests
-corpus_40 = "CCTP-SRSA-IP-20210831/"
-base_uri = "https://unsel.tetras-libre.fr/tenet/working"
-req_100 = "CCTP-SRSA-IP-20210831-R100/"
-req_200 = "CCTP-SRSA-IP-20210831-R200/"
-req_300 = "CCTP-SRSA-IP-20210831-R300/"
-req_1100 = "CCTP-SRSA-IP-20210831-R1100/"
-corpus_comp = "COMP/"
-req_f1 = "COMP/F1/"
-req_34 = "COMP/34/"
-req_50 = "COMP/50/"
-corpus_ERTMS = "ERTMS/"
-corpus_PEV = "PEV-RSE-Approach/"
+# Config File
+config_file = "config.xml"
 
 
 #==============================================================================
-# Utility
+# Graph Loading
 #==============================================================================
     
-def read_query(cts_group, query_ref):
-    query_file = CTS_DIR + cts_group + str(query_ref) + ".cts"
-    with open(query_file, "r") as file:
-        return file.read()
-
-
-#==============================================================================
-# Graph Initialization
-#==============================================================================
-    
-def load_config(work_graph):
+def load_config(config, work_graph):
     print("-- Configuration Loading:")
     
-    file_ref = CONFIG_DIR + schema_file
-    work_graph.parse(file_ref)
+    work_graph.parse(config.schema_file)
     print("----- RDF Schema (" + str(len(work_graph)) + ")")
     
-    file_ref = CONFIG_DIR + semantic_net_file
-    work_graph.parse(file_ref)
+    work_graph.parse(config.semantic_net_file)
     print("----- Semantic Net Definition (" + str(len(work_graph)) + ")")
     
-    file_ref = CONFIG_DIR + dash_file
-    work_graph.parse(file_ref)
+    work_graph.parse(config.dash_file)
     print("----- Data Shapes Dash (" + str(len(work_graph)) + ")")
     
-    file_ref = CONFIG_DIR + c_param_file
-    work_graph.parse(file_ref)
+    work_graph.parse(config.config_param_file)
     print("----- Config Parameter Definition (" + str(len(work_graph)) + ")")
 
     
-def load_frame(work_graph, target_system):
-    print("-- Frame Ontology Loading:")
-    
-    # -- old --- file_ref = FRAME_DIR + req_onto_file
-    # -- old --- work_graph.parse(file_ref)
-    # -- old --- print("----- Requirement Frame Ontology (" + str(len(work_graph)) + ")")
+def load_cts(config, work_graph):
+    print("-- CTS Loading:")
     
-    file_ref = FRAME_DIR + target_system + ONTO_FILE
-    work_graph.parse(file_ref)
-    print("----- System Frame Ontology (" + str(len(work_graph)) + ")")
+    work_graph.parse(config.cts_file)
+    print("----- All Schemes (" + str(len(work_graph)) + ")")
 
-    # -- old --- file_ref = FRAME_DIR + f_param_file
-    # -- old --- work_graph.parse(file_ref)
-    # -- old --- print("----- Ontology Parameters (" + str(len(work_graph)) + ")")
 
+def load_frame(config, work_graph):
+    print("-- Frame Ontology Loading:")
         
-#def define_namespace(work_graph):
-#    print("-- Namespace Definition:")
-#    
-#    sys_uri = "https://unsel.tetras-libre.fr/tenet/frame/system-ontology/"
-#    concept_classes = ["agent"]
-#    for concept in concept_classes:
-#        new_prefix = "sys-" + concept
-#        new_uri = URIRef(sys_uri + concept + '#') 
-#        work_graph.namespace_manager.bind(new_prefix, new_uri)
-#        print("----- " + new_prefix + ": " + new_uri)
-#    print(list(work_graph.namespace_manager.namespaces()))
+    work_graph.parse(config.frame_ontology_file)
+    print("----- System Frame Ontology (" + str(len(work_graph)) + ")")
     
     
-def load_sentences(work_graph, corpus):
+def load_sentences(config, work_graph):
     print("-- Sentence Loading:")
     
-    target_ref = CORPUS_DIR + corpus + '**/*.ttl'
-    for file_ref in glob.glob(target_ref, recursive = True):
+    for file_ref in glob.glob(config.source_sentence_files, recursive = True):
         work_graph.parse(file_ref)
         print("----- " + file_ref + " (" + str(len(work_graph)) + ")") 
         
     
 
-#==============================================================================
-# CT Schemes for Transduction Process
-#==============================================================================
-    
-def load_cts(work_graph):
-    print("-- CTS Loading:")
-    
-    file_ref = CONFIG_DIR + cts_file
-    work_graph.parse(file_ref)
-    print("----- All Schemes (" + str(len(work_graph)) + ")")
-        
-    
-
 #==============================================================================
 # Result (export)
 #==============================================================================
      
-def export_result(work_graph, export_ref, export_file):
+def export_result(config, work_graph, export_file):
     print("-- Export result as turtle: " + export_file)
     
     work_graph.serialize(destination=export_file, 
-                         base=base_uri + '/' + export_ref, 
+                         base=config.base_uri + '/' + config.source_ref, 
                          format='turtle')
     
     
@@ -159,13 +90,8 @@ def finalize_export_file(export_file):
             "@prefix sys: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/> .",
             """
             @prefix sys: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/> .
-            @prefix sys-class: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/class/> .
-            @prefix sys-property: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/property/> .
-            @prefix sys-relation: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/relation/> .
             @prefix sys-Event: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/Event#> .
-            @prefix sys-event: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/eventObjectProperty#> .
             @prefix sys-State_Property: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/State_Property#> .
-            @prefix sys-stateProperty: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/statePropertyObjectProperty#> .
             @prefix sys-abstract_thing: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/abstract_thing#> .
             @prefix sys-action_verb: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/action_verb#> .
             @prefix sys-agent: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/agent#> .
@@ -173,65 +99,45 @@ def finalize_export_file(export_file):
             @prefix sys-component: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/component#> .
             @prefix sys-message: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/message#> .
             @prefix sys-place: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/place#> .
+            @prefix owl: <http://www.w3.org/2002/07/owl#> .
+            @prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
+            @prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
+            @prefix sys-relation: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/relation/> .
+            @prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
             """)
         file.write(x)
    
+     
 #==============================================================================
 # Main Function
 #==============================================================================
   
-def run(corpus, source_ref, target_ref):
+def create_extraction_graph(corpus, source_ref, target_ref):
     try:
         print("[Tenet] Prepare work data from corpus " + corpus)
     
+        print("\n" + "- Config Loading")
+        config = Config(config_file, corpus, source_ref, target_ref)
+        print(config.__dict__)
+    
         print("\n" + "- Graph Initialization")
         work_graph = Graph()
-        load_config(work_graph)   
-        load_frame(work_graph, target_ref) 
-        #define_namespace(work_graph)
+        load_config(config, work_graph)   
+        load_frame(config, work_graph) 
     
         print("\n" + "- Preparation of Transduction Process")
-        load_cts(work_graph)
+        load_cts(config, work_graph)
     
         print("\n" + "- Data Source Imports")
-        load_sentences(work_graph, corpus)
+        load_sentences(config, work_graph)
     
         print("\n" + "- Result")
-        output_file = OUTPUT_DIR + source_ref + TURTLE_SUFFIX
-        export_result(work_graph, source_ref, output_file)
-        finalize_export_file(output_file)
+        export_file = config.output_file
+        export_result(config, work_graph, export_file)
+        finalize_export_file(export_file)
         
         print()
     
     except:
         print("!!! An exception occurred !!!") 
     
-
-#==============================================================================
-# Execution
-#==============================================================================
-
-if __name__ == '__main__':
-    
-    target_ref = "system"
-    #target_ref = "environment"
-    
-    run(corpus_comp, 'COMP-01', target_ref) 
-    run(req_f1, 'C-F1', target_ref) 
-    run(req_50, 'C-50', target_ref) 
-    run(req_34, 'C-34', target_ref) 
-    
-    #run(req_100, 'R100f', target_ref) 
-    #run(req_200, 'R200f', target_ref)
-    #run(req_300, 'R300f', target_ref) 
-    #run(req_1100, 'R1100f', target_ref) 
-    
-    #run(corpus_40, 'Corpus-CCTP-40f', target_ref)
-    
-    #run(corpus_ERTMS, 'Corpus-ERTMS', target_ref)
-    #run(corpus_PEV, 'Corpus-PEV', target_ref)
-
-
-    
-    
-