Specific Main Method to ODRL extraction

tenet/__init__.py

@@ -8,4 +8,6 @@ sys.path.insert(0, os.path.abspath(LIB_PATH))
 # -- Main Methods
 from tenet.main import create_ontology_from_amrld_file
 from tenet.main import create_ontology_from_amrld_dir 
+from tenet.main import generate_odrl_from_amrld_file
+from tenet.main import generate_odrl_from_amrld_dir 
 #from main import create_ontology_from_unlrdf_file
\ No newline at end of file

tenet/main.py

@@ -19,7 +19,8 @@ from utility.timer import timed
 # -- Config File Path
 LIB_PATH = os.path.dirname(os.path.abspath(__file__)) + '/'
 LOGGING_CONF_FILE_PATH = f'{LIB_PATH}logging.conf'
-CONFIG_FILE_PATH = f'{LIB_PATH}config.xml'
+OWL_CONFIG_FILE_PATH = f'{LIB_PATH}owl_amr_config.xml'
+ODRL_CONFIG_FILE_PATH = f'{LIB_PATH}odrl_amr_config.xml'
 
 # -- Logging
 logging.config.fileConfig(LOGGING_CONF_FILE_PATH, disable_existing_loggers=False)
@@ -36,7 +37,9 @@ def __set_context():
     os.chdir(LIB_PATH)
 
 
-def __set_config(source_type, source_corpus, onto_prefix, 
+def __set_config(
+        config_file_path,
+        source_type, source_corpus, onto_prefix, 
         base_output_dir, technical_dir_path):
     
     logger.info("-- Process Setting ")
@@ -45,9 +48,9 @@ def __set_config(source_type, source_corpus, onto_prefix,
     logger.info(f'----- technical dir path: {technical_dir_path}')
     logger.info(f'----- Ontology target (id): {onto_prefix}')
     logger.info(f'----- Current path: {os.getcwd()}')
-    logger.debug(f'----- Config file: {CONFIG_FILE_PATH}')
+    logger.debug(f'----- Config file: {config_file_path}')
     
-    process_config = config.Config(CONFIG_FILE_PATH, 
+    process_config = config.Config(config_file_path, 
                                    onto_prefix, 
                                    source_corpus, 
                                    base_output_dir = base_output_dir,
@@ -104,7 +107,7 @@ def __serialize_factoid_graph(config, factoid_graph, out_file_path=None):
     
 
 #==============================================================================
-# AMR Main Methods 
+# AMR Main Methods (to create an ontology) 
 #==============================================================================
 
 @timed
@@ -137,7 +140,8 @@ def create_ontology_from_amrld_file(amrld_file_path,
     logger.info('\n === Process Initialization === ')
     __set_context()
     if onto_prefix is None: onto_prefix = 'DefaultId'
-    config = __set_config('amr', amrld_file_path, onto_prefix, 
+    config = __set_config(OWL_CONFIG_FILE_PATH,
+                          'amr', amrld_file_path, onto_prefix, 
                           out_file_path, technical_dir_path)
     assert os.path.exists(amrld_file_path), f'input file does not exists ({amrld_file_path})'
     
@@ -192,7 +196,126 @@ def create_ontology_from_amrld_dir(amrld_dir_path,
     logger.info('\n === Process Initialization === ')
     __set_context()
     if onto_prefix is None: onto_prefix = 'DefaultId'
-    config = __set_config('amr', amrld_dir_path, onto_prefix, 
+    config = __set_config(OWL_CONFIG_FILE_PATH,
+                          'amr', amrld_dir_path, onto_prefix, 
+                          out_file_path, technical_dir_path)
+    assert os.path.exists(amrld_dir_path), f'input directory does not exists ({amrld_dir_path})'
+    __count_number_of_graph(config)
+    
+    # -- Extraction Processing    
+    logger.info('\n === Extraction Processing === ')
+    sentence_dir = config.source_sentence_file
+    sentence_count = 0
+    result_triple_list = []
+    for sentence_file in glob.glob(sentence_dir, recursive = True):
+        sentence_count += 1
+        logger.info(f'     *** sentence {sentence_count} *** ')
+        config.sentence_output_dir = f'-{sentence_count}'
+        new_triple_list = __apply_extraction(config, sentence_file)
+        result_triple_list.extend(new_triple_list)
+    
+    # -- Final Ontology Generation (factoid_graph)
+    logger.info('\n === Final Ontology Generation  === ') 
+    factoid_graph = __generate_final_ontology(result_triple_list)
+    ontology_turtle_string = __serialize_factoid_graph(config, factoid_graph, out_file_path)
+        
+    # -- Done
+    logger.info('\n === Done === ')
+    if config.technical_dir_path is not None:
+        log_file_name = 'tenet.log'
+        dest_file_path = f'{config.technical_dir_path}{log_file_name}'
+        shutil.copy(log_file_name, dest_file_path)
+    
+    return ontology_turtle_string
+
+
+
+#==============================================================================
+# AMR Main Methods (to generate ODRL statements) 
+#==============================================================================
+
+@timed
+def generate_odrl_from_amrld_file(
+        amrld_file_path, onto_prefix=None, out_file_path=None, 
+        technical_dir_path=None):
+    """
+    Method to generate an ODRL statement (as Turtle String) from a transduction 
+    analysis of an AMRLD file.
+
+    Parameters
+    ----------
+    amrld_file_path: a path to an AMR-LD Turtle File.
+    onto_prefix: the target ontology prefix if defined (if not defined a prefix based on the amrld filename is used).
+    out_file_path: a file path where the ODRL statement is written if defined (the function still outputs the string). 
+    technical_dir_path: a dir path where some technical and log files are written if defined.
+
+    Returns
+    -------
+    Ontology Turtle String.
+
+    """
+    
+    logger.info('[TENET] Extraction Processing')    
+    
+    # -- Process Initialization
+    logger.info('\n === Process Initialization === ')
+    __set_context()
+    if onto_prefix is None: onto_prefix = 'DefaultId'
+    config = __set_config(ODRL_CONFIG_FILE_PATH,
+                          'amr', amrld_file_path, onto_prefix, 
+                          out_file_path, technical_dir_path)
+    assert os.path.exists(amrld_file_path), f'input file does not exists ({amrld_file_path})'
+    
+    # -- Extraction Processing   
+    logger.info('\n === Extraction Processing === ')
+    config.sentence_output_dir = f'-0'
+    result_triple_list = __apply_extraction(config, amrld_file_path)
+    
+    # -- Final Ontology Generation (factoid_graph)
+    logger.info('\n === Final Ontology Generation  === ') 
+    factoid_graph = __generate_final_ontology(result_triple_list)
+    ontology_turtle_string = __serialize_factoid_graph(config, factoid_graph, out_file_path)
+        
+    # -- Done
+    logger.info('\n === Done === ')
+    if config.technical_dir_path is not None:
+        log_file_name = 'tenet.log'
+        dest_file_path = f'{config.technical_dir_path}{log_file_name}'
+        shutil.copy(log_file_name, dest_file_path)
+    
+    return ontology_turtle_string
+
+
+@timed
+def generate_odrl_from_amrld_dir(
+        amrld_dir_path, onto_prefix=None, out_file_path=None, 
+        technical_dir_path=None):
+    """
+    Method to create an ontology (as Turtle String) from a transduction 
+    analysis of an AMRLD file.
+
+    Parameters
+    ----------
+    amrld_file_path: a path to an AMR-LD Turtle File.
+    onto_prefix: the target ontology prefix if defined (if not defined a prefix based on the amrld filename is used).
+    out_file_path: a file path where the ODRL statement is written if defined (the function still outputs the string). 
+    technical_dir_path: a dir path where some technical and log files are written if defined.
+
+    Returns
+    -------
+    Dictionary [filename -> Ontology Turtle String].
+    Complete Ontology Turtle String (synthesis of all ontology)
+
+    """
+
+    logger.info('[TENET] Extraction Processing')    
+    
+    # -- Process Initialization
+    logger.info('\n === Process Initialization === ')
+    __set_context()
+    if onto_prefix is None: onto_prefix = 'DefaultId'
+    config = __set_config(ODRL_CONFIG_FILE_PATH,
+                          'amr', amrld_dir_path, onto_prefix, 
                           out_file_path, technical_dir_path)
     assert os.path.exists(amrld_dir_path), f'input directory does not exists ({amrld_dir_path})'
     __count_number_of_graph(config)

tenet/config.xml → tenet/odrl_amr_config.xml

@@ -7,8 +7,8 @@
         />
 
     <cts
-        amr_cts_ref = "amr_scheme_clara_1"
-        unl_cts_ref = "unl_scheme_1"
+        amr_cts_ref = "odrl_amr_scheme_1"
+        unl_cts_ref = "None"
         />
 
     <directory

tenet/owl_amr_config.xml

+<?xml version="1.0" encoding="UTF-8"?>
+<config>
+
+    <base
+        process_level = "sentence"
+        source_type = "amr"
+        />
+
+    <cts
+        amr_cts_ref = "owl_amr_scheme_1"
+        unl_cts_ref = "owl_unl_scheme_1"
+        />
+
+    <directory
+        base_dir = "./"
+        structure = "structure/"
+        cts = "scheme/"
+        target_frame = "../input/targetFrameStructure/"
+        amr_input_documents = "../input/amrDocuments/"
+        unl_input_documents = "../input/unlDocuments/"
+        output = "../output/"
+        />
+
+    <file
+        amr_input_data_schema = "amr-rdf-schema"
+        unl_input_data_schema = "unl-rdf-schema"
+        semantic_net_schema = "odrl-snet-schema"
+        config_param = "config-parameters"
+        />
+
+    <reference
+        default_base_uri = "https://tenet.tetras-libre.fr/working"
+        default_ontology_suffix = "-ontology.ttl"
+        default_ontology_seed_suffix = "-ontology-seed.ttl"
+        />
+
+    <output
+        ontology_namespace = "https://tenet.tetras-libre.fr/base-ontology/"
+        />
+
+</config>

tests/context.py → tests/main_tests/context.py

 import os
 import sys
-LIB_PATH = f'{os.path.dirname(os.path.abspath(__file__))}/..'
+LIB_PATH = f'{os.path.dirname(os.path.abspath(__file__))}/../..'
 print(f'Test Context: {LIB_PATH}')
 sys.path.insert(0, os.path.abspath(LIB_PATH))