Update config module (extraction scheme)

tenet/extraction/config.py

@@ -24,18 +24,6 @@ import os
 BASE_ONTOLOGY_REF = ['base', 'base-ontology']
 
 
-#==============================================================================
-# Utilities
-#==============================================================================
-
-# def get_according_source_type(source_type, c_xpath, amr_ref, unl_ref):
-#     get_result = 'None'
-#     if source_type == 'amr':
-#         get_result = c_xpath.get(amr_ref)
-#     else: # else source_type == 'unl'
-#         get_result = c_xpath.get(unl_ref)
-#     return get_result
-
 
 #==============================================================================
 # Class
@@ -66,10 +54,11 @@ class Config:
         self.source_corpus = source_corpus
         self.target_ref = target_ref
         self._process_level = c_base.get("process_level")
-        self._source_type = c_base.get("source_type")
+        self.source_type = c_base.get("source_type")
+        self.extraction_scheme = c_base.get("extraction_scheme")
         
-        # -- CTS Reference
-        self.cts_ref = ""
+        # # -- CTS Reference
+        # self.cts_ref = ""
 
         # -- Directories
         c_dir = self.config_tree.xpath("directory")[0]
@@ -98,7 +87,8 @@ class Config:
         self.config_param_file += c_file.get("config_param") + ".ttl"
         self.base_ontology_file = self.structure_dir
         self.base_ontology_file += 'base-ontology.ttl'
-        self.cts_file = ""
+        self.cts_file = f"{self.cts_dir}{self.extraction_scheme}.py"
+        # self.cts_file = ""
         
         # -- Ontology References
         c_ref = self.config_tree.xpath("reference")[0]
@@ -155,40 +145,6 @@ class Config:
     process_level = property(_get_process_level, _set_process_level)
         
     
-    # -- Property: source_type
-    
-    def _get_source_type(self):
-        return self._source_type
-    
-    def _set_source_type(self, source_type):
-        self._source_type = source_type
-        self.cts_ref = ""
-        self.input_doc_dir = ""
-        self.schema_file = ""
-        self.cts_file = ""
-        self.source_sentence_file = '**/*'
-        
-    source_type = property(_get_source_type, _set_source_type)
-    
-    
-    # -- Property: cts_ref
-    
-    def _get_cts_ref(self):
-        return self._cts_ref
-    
-    def _set_cts_ref(self, new_cts_ref):
-        c_cts = self.config_tree.xpath("cts")[0]
-        self._cts_ref = ""
-        if self.source_type == 'amr':
-            self._cts_ref += c_cts.get("amr_cts_ref")
-        if self.source_type == 'unl':
-            self._cts_ref += c_cts.get("unl_cts_ref")
-        if new_cts_ref != "":
-            self._cts_ref = new_cts_ref
-    
-    cts_ref = property(_get_cts_ref, _set_cts_ref)  
-    
-    
     # -- Property: target_dir
     
     def _get_target_frame_dir(self):
@@ -257,14 +213,6 @@ class Config:
     
     def _set_input_doc_dir(self, output_dir_complement):
         self._input_doc_dir = output_dir_complement
-        # -- old ---
-        # c_dir = self.config_tree.xpath("directory")[0]
-        # self._input_doc_dir = self.base_dir
-        # if self.source_type == 'amr':
-        #     self._input_doc_dir += c_dir.get("amr_input_documents")
-        # if self.source_type == 'unl':
-        #     self._input_doc_dir += c_dir.get("unl_input_documents")
-        # self._input_doc_dir += output_dir_complement
     
     input_doc_dir = property(_get_input_doc_dir, _set_input_doc_dir)  
     
@@ -286,19 +234,6 @@ class Config:
     schema_file = property(_get_schema_file, _set_schema_file) 
             
     
-    # -- Property: cts_file
-    
-    def _get_cts_file(self):
-        return self._cts_file
-    
-    def _set_cts_file(self, cts_file_cmpl):
-        self._cts_file = self.cts_dir
-        self._cts_file += self.cts_ref
-        self._cts_file += cts_file_cmpl + '.py'
-    
-    cts_file = property(_get_cts_file, _set_cts_file) 
-        
-    
     # -- Property: source_sentence_files
     
     def _get_source_sentence_file(self):
@@ -336,8 +271,9 @@ class Config:
         config_str += '\n  ----- source corpus: ' + self.source_corpus
         config_str += '\n  ----- target reference: ' + self.target_ref
         config_str += '\n  ----- process level: ' + self.process_level
-        config_str += '\n  ----- source type: ' + self.source_type
-        config_str += '\n  ----- CTS reference: ' + self.cts_ref
+        config_str += f'\n  ----- source type: {self.source_type}'
+        config_str += f'\n  ----- extraction scheme: {self.extraction_scheme}'
+        # config_str += '\n  ----- CTS reference (old): ' + self.cts_ref
         config_str += '\n  ----- output file: ' + self.output_file
         config_str += f'\n  ----- technical dir path: {self.technical_dir_path}'
         
@@ -358,10 +294,11 @@ class Config:
         config_str += '\n  ----- source corpus: ' + self.source_corpus
         config_str += '\n  ----- target reference: ' + self.target_ref
         config_str += '\n  ----- process level: ' + self.process_level
-        config_str += '\n  ----- source type: ' + self.source_type
+        config_str += f'\n  ----- source type: {self.source_type}'
+        config_str += f'\n  ----- extraction scheme: {self.extraction_scheme}'
         
-        config_str += '\n' + '  -- Compositional Transduction Scheme (CTS)'
-        config_str += '\n  ----- CTS reference: ' + self.cts_ref
+        # config_str += '\n' + '  -- Compositional Transduction Scheme (CTS)'
+        # config_str += '\n  ----- CTS reference (old): ' + self.cts_ref
         
         config_str += '\n' + '  -- Directories'
         config_str += '\n  ----- base directory: ' + self.base_dir

tenet/extraction/process.py

@@ -31,10 +31,10 @@ logger = logging.getLogger(__name__)
 #==============================================================================
 
 def load_cts(config):
-    """ Load extraction scheme (CTS) from <cts_ref> file """
+    """ Load extraction scheme (CTS) from <extraction_scheme> file """
     
     try:
-        cts_module = SourceFileLoader(config.cts_ref, 
+        cts_module = SourceFileLoader(config.extraction_scheme, 
                                       config.cts_file).load_module()
     
         return cts_module.rule_dir, cts_module.prefix_list, cts_module.scheme
@@ -47,7 +47,7 @@ def load_cts(config):
     except Exception:
         logger.error(' *** Error while loading scheme (load_cts) ***') 
         logger.debug(f' ----- current work directory: {os.getcwd()}')
-        logger.debug(f' ----- cts_ref: {config.cts_ref}') 
+        logger.debug(f' ----- extraction_scheme: {config.extraction_scheme}') 
         logger.debug(f' ----- cts_file: {config.cts_file}') 
     
     
@@ -396,7 +396,7 @@ def apply(config, graph):
     try:
         
         # -- Loading Extraction Scheme
-        logger.info("-- Loading Extraction Scheme ({0})".format(config.cts_ref))
+        logger.info(f"-- Loading Extraction Scheme ({config.extraction_scheme})")
         rule_dir, prefix_list, scheme = load_cts(config)
         logger.debug("----- Step number: {0}".format(len(scheme)))
             
@@ -430,7 +430,7 @@ def apply(config, graph):
     
     except:
         logger.error(' *** Error while processing extraction (apply) ***') 
-        logger.debug(f' ----- config.cts_ref = {config.cts_ref}')  
+        logger.debug(f' ----- config.extraction_scheme = {config.extraction_scheme}')  
         logger.debug(f' ----- rule_dir = {rule_dir}')  
         logger.debug(f' ----- scheme = {scheme}')      
         logger.debug(f' ----- step = {step_name}, {step_sequence_def}')  
\ No newline at end of file

tenet/odrl_amr_config.xml

@@ -4,11 +4,7 @@
     <base
         process_level = "sentence"
         source_type = "amr"
-        />
-
-    <cts
-        amr_cts_ref = "odrl_amr_scheme_1"
-        unl_cts_ref = "None"
+        extraction_scheme = "odrl_amr_scheme_1"
         />
 
     <directory

tenet/owl_amr_config.xml

@@ -4,11 +4,7 @@
     <base
         process_level = "sentence"
         source_type = "amr"
-        />
-
-    <cts
-        amr_cts_ref = "owl_amr_scheme_1"
-        unl_cts_ref = "owl_unl_scheme_1"
+        extraction_scheme = "owl_amr_scheme_1"
         />
 
     <directory