Minimal UNL-RDF generation

asd/unl.py

@@ -11,14 +11,17 @@
 # Importing required modules
 #==============================================================================
 
-import pydot
+import pydot as dot
+
+import asd.urdf as urdf
 
 
 #==============================================================================
 # Parameters
 #==============================================================================
 
-# None
+UNL_RDF_SCHEMA = "https://unl.tetras-libre.fr/rdf/schema/"
+UNL_RDF_BASE_URI = "https://unl.tetras-libre.fr/rdf/dev/"
 
 
 #==============================================================================
@@ -43,13 +46,19 @@ class Graph:
         return res
     
     def to_dot_graph(self):
-        graph = pydot.Dot("My_Graph", graph_type="graph", bgcolor="yellow")
+        graph = dot.Dot("My_Graph", graph_type="graph", bgcolor="yellow")
         for link in self.link_list:
             graph.add_node(link.uw1.to_dot_node())
             graph.add_node(link.uw2.to_dot_node())
             graph.add_edge(link.to_dot_edge())
         return graph
     
+    def to_rdf_graph(self):
+        graph = urdf.init_unl_rdf_graph()
+        for link in self.link_list:
+            graph = link.to_rdf_graph(graph) 
+        return graph
+    
     
 class Link:
     """ link := Link(relation, uw, uw) """
@@ -68,9 +77,15 @@ class Link:
     def to_dot_edge(self):
         n1_name = self.uw1.to_dot_node().get_name()
         n2_name = self.uw2.to_dot_node().get_name()
-        edge = pydot.Edge(n1_name, n2_name, color="blue")
+        edge = dot.Edge(n1_name, n2_name, color="blue")
         return edge
     
+    def to_rdf_graph(self, graph):
+        graph, rel_node = self.relation.to_rdf_graph(graph)
+        graph, uw1_node = self.uw1.to_rdf_graph(graph)
+        graph, uw2_node = self.uw2.to_rdf_graph(graph)
+        graph = urdf.add_relation_triple(graph, rel_node, uw1_node, uw2_node)
+        return graph
     
 class Relation:
     """ relation := Relation(id) """
@@ -81,6 +96,9 @@ class Relation:
     def to_string(self):
         return self.id
     
+    def to_rdf_graph(self, graph):
+        return urdf.add_relation_node(graph, self.id)
+    
     
 class UniversalWord:
     """ uw := UniversalWord(headword, restriction*, attr*) """
@@ -93,8 +111,8 @@ class UniversalWord:
     def to_string(self):
         res = self.headword.to_string() + "("
         res_init_length = len(res) 
-        for restrictio in self.restriction_list:
-            res += restrictio.to_string() + ", "
+        for restriction in self.restriction_list:
+            res += restriction.to_string() + ", "
         res = res[:-1]
         if len(res) > res_init_length:
             res = res[:-1]
@@ -103,12 +121,21 @@ class UniversalWord:
             res += ".@" + attr.to_string()
         return res
     
+    def to_id(self):
+        res = self.headword.to_string()
+        for restriction in self.restriction_list:
+            res += "-" + restriction.to_id()
+        return res        
+    
     def to_dot_node(self):
         node_name = self.headword.to_string() # TODO: à revoir pour id unique
         node_label = self.headword.to_string()
-        node = pydot.Node(node_name, label=node_label)
+        node = dot.Node(node_name, label=node_label)
         return node
     
+    def to_rdf_graph(self, graph):
+        return urdf.add_uw_node(graph, self.to_id(), self.to_string())
+    
     
 class Headword:
     """ headword := Headword(id) """
@@ -130,6 +157,9 @@ class Restriction:
     def to_string(self):
         return self.relation.to_string() + ">" + self.id.to_string()
     
+    def to_id(self):
+        return self.relation.to_string() + "-" + self.id.to_string()
+    
     
 class Attribute:
     """ attr := Attribute(id) """

asd/urdf.py

+#!/usr/bin/python3.10
+# -*-coding:Utf-8 -*
+
+#==============================================================================
+# unlAnt: uRDF
+#------------------------------------------------------------------------------
+# This module contains methods to generate an UNL-RDF graph
+#==============================================================================
+
+#==============================================================================
+# Importing required modules
+#==============================================================================
+
+import rdflib as rdf
+from rdflib import Namespace, URIRef, Literal  #, BNode
+from rdflib import XSD, RDFS, RDF, OWL
+
+
+#==============================================================================
+# Parameters
+#==============================================================================
+
+# -- Schémas, URI
+UNL_RDF_SCHEMA = "https://unl.tetras-libre.fr/rdf/schema/"
+UNL_RDF_BASE_URI = "https://unl.tetras-libre.fr/rdf/dev/"
+
+# -- Namespace(s)
+URDF = Namespace(UNL_RDF_SCHEMA)
+
+
+#==============================================================================
+# Main Functions
+#==============================================================================
+
+# ----------------------------------------------
+# Methods to initialize the UNL-RDF graph
+# ----------------------------------------------
+
+def init_unl_rdf_graph():
+    graph = rdf.Graph()
+    graph.bind("urdf", UNL_RDF_SCHEMA) # prefix for UNL-RDF schema
+    return graph
+
+
+def add_useful_prefix(graph):  
+    """ Add useful prefix, as RDFS, OWL... """
+    
+    graph.bind("xsd", XSD)
+    graph.bind("rdfs", RDFS)
+    graph.bind("owl", OWL)
+    graph.bind("rdf", RDF)
+    
+    return graph
+
+
+# ----------------------------------------------
+# Methods to add a node in the UNL-RDF graph
+# ----------------------------------------------
+          
+def add_relation_node(graph, relation_id):
+    """ Add a node corresponding to a relation in UNL. 
+    
+    Attrs:
+    - relation_id (str): relation identifier
+    
+    Returns:
+    - a RDF node corresponding to the UNL relation
+    """
+    
+    relation_node = URIRef(UNL_RDF_BASE_URI + relation_id)
+    relation_label = Literal(relation_id)
+    graph.add((relation_node, RDF.type, URDF.UW_Occurrence))
+    graph.add((relation_node, RDFS.label, relation_label))
+    
+    return graph, relation_node
+
+
+def add_uw_node(graph, uw_id, uw_label):
+    """ Add a node corresponding to an universal word (UW) in UNL. 
+    
+    Attrs:
+    - uw_id (str): UW identifier
+    - uw_label (str): label for the UW node
+    
+    Returns:
+    - a RDF node corresponding to the UW
+    """
+    
+    uw_node = URIRef(UNL_RDF_BASE_URI + uw_id)
+    uw_label = Literal(uw_label)
+    graph.add((uw_node, RDF.type, URDF.UW_Occurrence))
+    graph.add((uw_node, RDFS.label, uw_label))
+    
+    return graph, uw_node
+      
+    
+# ----------------------------------------------
+# Methods to add a triple in the UNL-RDF graph
+# ----------------------------------------------    
+    
+def add_relation_triple(graph, relation_node, uw1_node, uw2_node):
+    """ Add a triple corresponding to a relation between two nodes in UNL. 
+    
+    Attrs:
+    - relation_node (RDF node): relation node
+    - uw1_node (RDF node): UW1 node
+    - uw2_node (RDF node): UW2 node
+    """
+    
+    graph.add((uw1_node, relation_node, uw2_node))
+    
+    return graph
+
+    
+    

parse.py

@@ -21,11 +21,13 @@ from antlr4 import FileStream, CommonTokenStream, InputStream
 #==============================================================================
 
 # ANTLR Grammar
-
 doc_grammar = 'grammar/doc/doc.g4'
 org_grammar = 'grammar/org/org.g4'
 unl_grammar = 'grammar/unl/unl.g4'
 
+# RDF URIs
+UNL_RDF_BASE_URI = "https://unl.tetras-libre.fr/rdf/dev/"
+
 
 #==============================================================================
 # Utilities
@@ -152,6 +154,13 @@ def main(argv):
     unl_dot_graph.write_raw("output/output_raw.dot")
     unl_dot_graph.write_png("output/output.png")
     
+    # -- Generate rdf file
+    unl_rdf_graph = unl.to_rdf_graph()
+    print("----- RDF file generated: " + "output/output.ttl")
+    unl_rdf_graph.serialize(destination ="output/output.ttl", 
+                            base = UNL_RDF_BASE_URI + "output", 
+                            format = 'turtle')
+
 
 if __name__ == '__main__':
     main(sys.argv)

requirements.txt

 antlr4-python3-runtime==4.9.3
 pydot
+rdflib