prepare_work_data.py

#!/usr/bin/python3.5
# -*-coding:Utf-8 -*

#==============================================================================
# TENET: prepare work data
#------------------------------------------------------------------------------
# Prepare work data for extraction processing. 
#==============================================================================

#==============================================================================
# Importing required modules
#==============================================================================

import glob
from rdflib import Graph, Namespace, URIRef


#==============================================================================
# Parameters
#==============================================================================

# Working directories
CONFIG_DIR = "config/"
FRAME_DIR = "frame/"
CORPUS_DIR = "corpus/"
CTS_DIR = "cts/"

# Config Definition
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"
req_onto_file = "requirement-ontology.ttl"
sys_onto_file = "system-ontology.ttl"
f_param_file = "ontology-parameters.ttl"

# Dev Tests
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/"
corpus_40 = "CCTP-SRSA-IP-20210831/"
corpus_ERTMS = "ERTMS/"


#==============================================================================
# Utility
#==============================================================================

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):
    print("-- Configuration Loading:")
    
    file_ref = CONFIG_DIR + schema_file
    work_graph.parse(file_ref)
    print("----- RDF Schema (" + str(len(work_graph)) + ")")
    
    file_ref = CONFIG_DIR + semantic_net_file
    work_graph.parse(file_ref)
    print("----- Semantic Net Definition (" + str(len(work_graph)) + ")")
    
    file_ref = CONFIG_DIR + dash_file
    work_graph.parse(file_ref)
    print("----- Data Shapes Dash (" + str(len(work_graph)) + ")")
    
    file_ref = CONFIG_DIR + c_param_file
    work_graph.parse(file_ref)
    print("----- Config Parameter Definition (" + str(len(work_graph)) + ")")


def load_frame(work_graph):
    print("-- Frame Ontology Loading:")
    
    file_ref = FRAME_DIR + req_onto_file
    work_graph.parse(file_ref)
    print("----- Requirement Frame Ontology (" + str(len(work_graph)) + ")")
    
    file_ref = FRAME_DIR + sys_onto_file
    work_graph.parse(file_ref)
    print("----- System Frame Ontology (" + str(len(work_graph)) + ")")
    
    file_ref = FRAME_DIR + f_param_file
    work_graph.parse(file_ref)
    print("----- Ontology Parameters (" + str(len(work_graph)) + ")")


#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()))
    
    
def load_sentences(work_graph, corpus):
    print("-- Sentence Loading:")
    
    target_ref = CORPUS_DIR + corpus + '**/*.ttl'
    for file_ref in glob.glob(target_ref, 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):
    print("-- Export result as turtle: " + export_file)
    
    work_graph.serialize(destination=export_file, 
                         base=base_uri + '/' + export_ref, 
                         format='turtle')
    
    
def finalize_export_file(export_file):
    """ finalize the export file by adding some useful prefixes """
    
    with open(export_file, "rt") as file:
        x = file.read()
    
    with open(export_file, "wt") as file:
        x = x.replace(
            "@prefix sys: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/> .",
            """@prefix sys: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/> .
@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#> .
@prefix sys-attributive_verb: <https://unsel.tetras-libre.fr/tenet/frame/system-ontology/attributive_verb#> .
@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#> .""")
        file.write(x)
        
#==============================================================================
# Main Function
#==============================================================================
  
def run(corpus, output_ref, output_file):
    try:
        print("[Tenet] Prepare work data from corpus " + corpus)
    
        print("\n" + "- Graph Initialization")
        work_graph = Graph()
        load_config(work_graph)   
        load_frame(work_graph) 
        #define_namespace(work_graph)
    
        print("\n" + "- Preparation of Transduction Process")
        load_cts(work_graph)
    
        print("\n" + "- Data Source Imports")
        load_sentences(work_graph, corpus)
    
        print("\n" + "- Result")
        export_result(work_graph, output_ref, output_file)
        finalize_export_file(output_file)
        
        print()
    
    except:
        print("!!! An exception occurred !!!") 
    

#==============================================================================
# Execution
#==============================================================================

if __name__ == '__main__':
    #run(req_100, 'R100', 'output100.ttl') 
    #run(req_200, 'R200', 'output200.ttl')
    run(req_300, 'R300b', 'output300b.ttl') 
    #run(corpus_40, 'Corpus-CCTP-40', 'outputCCTP40.ttl')
    #run(corpus_ERTMS, 'Corpus-ERTMS', 'outputERTMS.ttl')