ASD of UNL Graph

asd/doc.py

@@ -2,21 +2,16 @@
 # -*-coding:Utf-8 -*
 
 #==============================================================================
-# TENET: inference
+# unlAnt: ASD of UNL document
 #------------------------------------------------------------------------------
-# Module to execute the extraction process, 
-# by applying the transduction schemes with the SHACL inference engine
+# --
 #==============================================================================
 
 #==============================================================================
 # Importing required modules
 #==============================================================================
 
-import sys
-from subprocess import Popen, PIPE
-from antlr4 import FileStream, CommonTokenStream
-# from antlr.unlLexer import unlLexer
-# from antlr.unlParser import unlParser
+# None
 
 
 #==============================================================================

asd/unl.py

+#!/usr/bin/python3.10
+# -*-coding:Utf-8 -*
+
+#==============================================================================
+# unlAnt: ASD of UNL Graph
+#------------------------------------------------------------------------------
+# --
+#==============================================================================
+
+#==============================================================================
+# Importing required modules
+#==============================================================================
+
+# None
+
+
+#==============================================================================
+# Parameters
+#==============================================================================
+
+# None
+
+
+#==============================================================================
+# Document Class
+#==============================================================================
+
+  
+class Graph:
+    """ unl := Graph(link*) """
+    
+    def __init__(self, link_list):
+        self.link_list = link_list
+    
+    def to_string(self):
+        res = "Graph("
+        res_init_length = len(res) 
+        for link in self.link_list:
+            res += link.to_string() + ", "
+        if len(res) > res_init_length:
+            res = res[:-2]
+        res += ")"
+        return res
+    
+    
+class Link:
+    """ link := Link(relation, uw, uw) """
+    
+    def __init__(self, relation, uw1, uw2):
+        self.relation = relation
+        self.uw1 = uw1  
+        self.uw2 = uw2
+    
+    def to_string(self):
+        res = self.relation.to_string() + "("
+        res += self.uw1.to_string() + ", "
+        res += self.uw2.to_string() + ")"
+        return res
+    
+    
+class Relation:
+    """ relation := Relation(id) """
+    
+    def __init__(self, id):
+        self.id = id    
+    
+    def to_string(self):
+        return self.id
+    
+    
+class UniversalWord:
+    """ uw := UniversalWord(headword, restriction*, attr*) """
+    
+    def __init__(self, headword, restriction_list, attr_list):
+        self.headword = headword    
+        self.restriction_list = restriction_list   
+        self.attr_list = attr_list   
+    
+    def to_string(self):
+        res = self.headword.to_string() + "("
+        res_init_length = len(res) 
+        for restrictio in self.restriction_list:
+            res += restrictio.to_string() + ", "
+        res = res[:-1]
+        if len(res) > res_init_length:
+            res = res[:-1]
+            res += ")"
+        for attr in self.attr_list:
+            res += ".@" + attr.to_string()
+        return res
+    
+    
+class Headword:
+    """ headword := Headword(id) """
+    
+    def __init__(self, id):
+        self.id = id    
+    
+    def to_string(self):
+        return self.id.to_string()
+    
+    
+class Restriction:
+    """ restriction := Restriction(id) """
+    
+    def __init__(self, relation, id):
+        self.relation = relation
+        self.id = id
+    
+    def to_string(self):
+        return self.relation.to_string() + ">" + self.id.to_string()
+    
+    
+class Attribute:
+    """ attr := Attribute(id) """
+    
+    def __init__(self, id):
+        self.id = id    
+    
+    def to_string(self):
+        return self.id.to_string()
+    
+    
+class Ident:
+    """ id := Ident(string) """
+    
+    def __init__(self, id):
+        self.id = id    
+    
+    def to_string(self):
+        return self.id
+    

grammar/unl/unl.g4

@@ -4,6 +4,10 @@
 
 grammar unl;
 
+@header {
+from asd import unl
+}
+
 
 //=============================================================================
 // Parser Grammar
@@ -13,40 +17,50 @@ grammar unl;
 // UNL representation
 //---------------------------------------------------------
 
-unlPart
-  : '{unl}' (relationOccurrence)+ '{/unl}'
+unlGraph returns [out]
+  : '{unl}'                               {link_list = []}
+    (l=link                               {link_list.append($l.out)} )+
+    '{/unl}'                              {$out = unl.Graph(link_list)}
   ;
 
-relationOccurrence
-  : universalRelation LP universalWord COMMA universalWord RP
+link returns [out]
+  : r=universalRelation
+    LP uw1=universalWord COMMA uw2=universalWord RP
+                                          {$out = unl.Link($r.out, $uw1.out, $uw2.out)}
   ;
 
-universalWord
-  : headword
-    (LP restriction (COMMA restriction)* RP)?
-    (attribute)*
-  | value
+universalWord returns [out]
+  : h=headword                            {headword =  $h.out}
+                                          {restriction_list = []}
+    (LP
+    r1=restriction                        {restriction_list.append($r1.out)}
+    (COMMA r2=restriction                 {restriction_list.append($r2.out)} )*
+    RP)?
+                                          {attr_list = []}
+    (a=attribute                          {attr_list.append($a.out)} )*
+                                          {$out = unl.UniversalWord(headword, restriction_list, attr_list)}
+  | v=value                               {$out = unl.UniversalWord($v.out, [], [])}
   ;
 
-headword
-  : ident
+headword returns [out]
+  : i=ident                              {$out = unl.Headword($i.out)}
   ;
 
-restriction
-  : universalRelation GREATER ident
+restriction returns [out]
+  : r=universalRelation GREATER i=ident  {$out = unl.Restriction($r.out, $i.out)}
   ;
 
-attribute
-  : DOT AT ident
+attribute returns [out]
+  : DOT AT i=ident                       {$out = unl.Attribute($i.out)}
   ;
 
-value
-  : VALUE
+value returns [out]
+  : v=VALUE                               {$out = unl.Ident($v.text)}
   ;
 
-universalRelation
-  : ( AND | AOJ | BEN | CNT |
-      EQU | ICL | OBJ | QUA )
+universalRelation returns [out]
+  : r=( AND | AOJ | BEN | CNT
+      | EQU | ICL | OBJ | QUA )           {$out = unl.Relation($r.text)}
   ;
 
 
@@ -56,9 +70,16 @@ universalRelation
 
 sentence : (word | punctuation | bracket)* ;
 
-ident : word (UNDERSCORE word)* ;
+ident returns [out]
+  : w1=word                               {ident = $w1.out}
+    (UNDERSCORE w2=word                   {ident += "_" + $w2.out} )*
+                                          {$out = unl.Ident(ident)}
+  ;
 
-word : LETTER | WORD ;
+word returns [$out]
+  : l=LETTER                              {$out = $l.text}
+  | w=WORD                                {$out = $w.text}
+  ;
 
 punctuation : DOT | COMMA | SEMCOL | COLON | DASH ;
 

grammar/unl/unl.interp

@@ -59,8 +59,8 @@ WORD
 VALUE
 
 rule names:
-unlPart
-relationOccurrence
+unlGraph
+link
 universalWord
 headword
 restriction
@@ -75,4 +75,4 @@ bracket
 
 
 atn:
-[3, 24715, 42794, 33075, 47597, 16764, 15335, 30598, 22884, 3, 29, 103, 4, 2, 9, 2, 4, 3, 9, 3, 4, 4, 9, 4, 4, 5, 9, 5, 4, 6, 9, 6, 4, 7, 9, 7, 4, 8, 9, 8, 4, 9, 9, 9, 4, 10, 9, 10, 4, 11, 9, 11, 4, 12, 9, 12, 4, 13, 9, 13, 4, 14, 9, 14, 3, 2, 3, 2, 6, 2, 31, 10, 2, 13, 2, 14, 2, 32, 3, 2, 3, 2, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 4, 3, 4, 3, 4, 3, 4, 3, 4, 7, 4, 49, 10, 4, 12, 4, 14, 4, 52, 11, 4, 3, 4, 3, 4, 5, 4, 56, 10, 4, 3, 4, 7, 4, 59, 10, 4, 12, 4, 14, 4, 62, 11, 4, 3, 4, 5, 4, 65, 10, 4, 3, 5, 3, 5, 3, 6, 3, 6, 3, 6, 3, 6, 3, 7, 3, 7, 3, 7, 3, 7, 3, 8, 3, 8, 3, 9, 3, 9, 3, 10, 3, 10, 3, 10, 7, 10, 84, 10, 10, 12, 10, 14, 10, 87, 11, 10, 3, 11, 3, 11, 3, 11, 7, 11, 92, 10, 11, 12, 11, 14, 11, 95, 11, 11, 3, 12, 3, 12, 3, 13, 3, 13, 3, 14, 3, 14, 3, 14, 2, 2, 15, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 2, 6, 3, 2, 19, 26, 3, 2, 27, 28, 3, 2, 6, 10, 3, 2, 11, 14, 2, 98, 2, 28, 3, 2, 2, 2, 4, 36, 3, 2, 2, 2, 6, 64, 3, 2, 2, 2, 8, 66, 3, 2, 2, 2, 10, 68, 3, 2, 2, 2, 12, 72, 3, 2, 2, 2, 14, 76, 3, 2, 2, 2, 16, 78, 3, 2, 2, 2, 18, 85, 3, 2, 2, 2, 20, 88, 3, 2, 2, 2, 22, 96, 3, 2, 2, 2, 24, 98, 3, 2, 2, 2, 26, 100, 3, 2, 2, 2, 28, 30, 7, 3, 2, 2, 29, 31, 5, 4, 3, 2, 30, 29, 3, 2, 2, 2, 31, 32, 3, 2, 2, 2, 32, 30, 3, 2, 2, 2, 32, 33, 3, 2, 2, 2, 33, 34, 3, 2, 2, 2, 34, 35, 7, 4, 2, 2, 35, 3, 3, 2, 2, 2, 36, 37, 5, 16, 9, 2, 37, 38, 7, 11, 2, 2, 38, 39, 5, 6, 4, 2, 39, 40, 7, 7, 2, 2, 40, 41, 5, 6, 4, 2, 41, 42, 7, 12, 2, 2, 42, 5, 3, 2, 2, 2, 43, 55, 5, 8, 5, 2, 44, 45, 7, 11, 2, 2, 45, 50, 5, 10, 6, 2, 46, 47, 7, 7, 2, 2, 47, 49, 5, 10, 6, 2, 48, 46, 3, 2, 2, 2, 49, 52, 3, 2, 2, 2, 50, 48, 3, 2, 2, 2, 50, 51, 3, 2, 2, 2, 51, 53, 3, 2, 2, 2, 52, 50, 3, 2, 2, 2, 53, 54, 7, 12, 2, 2, 54, 56, 3, 2, 2, 2, 55, 44, 3, 2, 2, 2, 55, 56, 3, 2, 2, 2, 56, 60, 3, 2, 2, 2, 57, 59, 5, 12, 7, 2, 58, 57, 3, 2, 2, 2, 59, 62, 3, 2, 2, 2, 60, 58, 3, 2, 2, 2, 60, 61, 3, 2, 2, 2, 61, 65, 3, 2, 2, 2, 62, 60, 3, 2, 2, 2, 63, 65, 5, 14, 8, 2, 64, 43, 3, 2, 2, 2, 64, 63, 3, 2, 2, 2, 65, 7, 3, 2, 2, 2, 66, 67, 5, 20, 11, 2, 67, 9, 3, 2, 2, 2, 68, 69, 5, 16, 9, 2, 69, 70, 7, 16, 2, 2, 70, 71, 5, 20, 11, 2, 71, 11, 3, 2, 2, 2, 72, 73, 7, 6, 2, 2, 73, 74, 7, 17, 2, 2, 74, 75, 5, 20, 11, 2, 75, 13, 3, 2, 2, 2, 76, 77, 7, 29, 2, 2, 77, 15, 3, 2, 2, 2, 78, 79, 9, 2, 2, 2, 79, 17, 3, 2, 2, 2, 80, 84, 5, 22, 12, 2, 81, 84, 5, 24, 13, 2, 82, 84, 5, 26, 14, 2, 83, 80, 3, 2, 2, 2, 83, 81, 3, 2, 2, 2, 83, 82, 3, 2, 2, 2, 84, 87, 3, 2, 2, 2, 85, 83, 3, 2, 2, 2, 85, 86, 3, 2, 2, 2, 86, 19, 3, 2, 2, 2, 87, 85, 3, 2, 2, 2, 88, 93, 5, 22, 12, 2, 89, 90, 7, 18, 2, 2, 90, 92, 5, 22, 12, 2, 91, 89, 3, 2, 2, 2, 92, 95, 3, 2, 2, 2, 93, 91, 3, 2, 2, 2, 93, 94, 3, 2, 2, 2, 94, 21, 3, 2, 2, 2, 95, 93, 3, 2, 2, 2, 96, 97, 9, 3, 2, 2, 97, 23, 3, 2, 2, 2, 98, 99, 9, 4, 2, 2, 99, 25, 3, 2, 2, 2, 100, 101, 9, 5, 2, 2, 101, 27, 3, 2, 2, 2, 10, 32, 50, 55, 60, 64, 83, 85, 93]
+[3, 24715, 42794, 33075, 47597, 16764, 15335, 30598, 22884, 3, 29, 134, 4, 2, 9, 2, 4, 3, 9, 3, 4, 4, 9, 4, 4, 5, 9, 5, 4, 6, 9, 6, 4, 7, 9, 7, 4, 8, 9, 8, 4, 9, 9, 9, 4, 10, 9, 10, 4, 11, 9, 11, 4, 12, 9, 12, 4, 13, 9, 13, 4, 14, 9, 14, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 6, 2, 34, 10, 2, 13, 2, 14, 2, 35, 3, 2, 3, 2, 3, 2, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 4, 3, 4, 3, 4, 3, 4, 3, 4, 3, 4, 3, 4, 3, 4, 3, 4, 3, 4, 7, 4, 59, 10, 4, 12, 4, 14, 4, 62, 11, 4, 3, 4, 3, 4, 5, 4, 66, 10, 4, 3, 4, 3, 4, 3, 4, 3, 4, 7, 4, 72, 10, 4, 12, 4, 14, 4, 75, 11, 4, 3, 4, 3, 4, 3, 4, 3, 4, 3, 4, 5, 4, 82, 10, 4, 3, 5, 3, 5, 3, 5, 3, 6, 3, 6, 3, 6, 3, 6, 3, 6, 3, 7, 3, 7, 3, 7, 3, 7, 3, 7, 3, 8, 3, 8, 3, 8, 3, 9, 3, 9, 3, 9, 3, 10, 3, 10, 3, 10, 7, 10, 106, 10, 10, 12, 10, 14, 10, 109, 11, 10, 3, 11, 3, 11, 3, 11, 3, 11, 3, 11, 3, 11, 7, 11, 117, 10, 11, 12, 11, 14, 11, 120, 11, 11, 3, 11, 3, 11, 3, 12, 3, 12, 3, 12, 3, 12, 5, 12, 128, 10, 12, 3, 13, 3, 13, 3, 14, 3, 14, 3, 14, 2, 2, 15, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26, 2, 5, 3, 2, 19, 26, 3, 2, 6, 10, 3, 2, 11, 14, 2, 130, 2, 28, 3, 2, 2, 2, 4, 40, 3, 2, 2, 2, 6, 81, 3, 2, 2, 2, 8, 83, 3, 2, 2, 2, 10, 86, 3, 2, 2, 2, 12, 91, 3, 2, 2, 2, 14, 96, 3, 2, 2, 2, 16, 99, 3, 2, 2, 2, 18, 107, 3, 2, 2, 2, 20, 110, 3, 2, 2, 2, 22, 127, 3, 2, 2, 2, 24, 129, 3, 2, 2, 2, 26, 131, 3, 2, 2, 2, 28, 29, 7, 3, 2, 2, 29, 33, 8, 2, 1, 2, 30, 31, 5, 4, 3, 2, 31, 32, 8, 2, 1, 2, 32, 34, 3, 2, 2, 2, 33, 30, 3, 2, 2, 2, 34, 35, 3, 2, 2, 2, 35, 33, 3, 2, 2, 2, 35, 36, 3, 2, 2, 2, 36, 37, 3, 2, 2, 2, 37, 38, 7, 4, 2, 2, 38, 39, 8, 2, 1, 2, 39, 3, 3, 2, 2, 2, 40, 41, 5, 16, 9, 2, 41, 42, 7, 11, 2, 2, 42, 43, 5, 6, 4, 2, 43, 44, 7, 7, 2, 2, 44, 45, 5, 6, 4, 2, 45, 46, 7, 12, 2, 2, 46, 47, 8, 3, 1, 2, 47, 5, 3, 2, 2, 2, 48, 49, 5, 8, 5, 2, 49, 50, 8, 4, 1, 2, 50, 65, 8, 4, 1, 2, 51, 52, 7, 11, 2, 2, 52, 53, 5, 10, 6, 2, 53, 60, 8, 4, 1, 2, 54, 55, 7, 7, 2, 2, 55, 56, 5, 10, 6, 2, 56, 57, 8, 4, 1, 2, 57, 59, 3, 2, 2, 2, 58, 54, 3, 2, 2, 2, 59, 62, 3, 2, 2, 2, 60, 58, 3, 2, 2, 2, 60, 61, 3, 2, 2, 2, 61, 63, 3, 2, 2, 2, 62, 60, 3, 2, 2, 2, 63, 64, 7, 12, 2, 2, 64, 66, 3, 2, 2, 2, 65, 51, 3, 2, 2, 2, 65, 66, 3, 2, 2, 2, 66, 67, 3, 2, 2, 2, 67, 73, 8, 4, 1, 2, 68, 69, 5, 12, 7, 2, 69, 70, 8, 4, 1, 2, 70, 72, 3, 2, 2, 2, 71, 68, 3, 2, 2, 2, 72, 75, 3, 2, 2, 2, 73, 71, 3, 2, 2, 2, 73, 74, 3, 2, 2, 2, 74, 76, 3, 2, 2, 2, 75, 73, 3, 2, 2, 2, 76, 77, 8, 4, 1, 2, 77, 82, 3, 2, 2, 2, 78, 79, 5, 14, 8, 2, 79, 80, 8, 4, 1, 2, 80, 82, 3, 2, 2, 2, 81, 48, 3, 2, 2, 2, 81, 78, 3, 2, 2, 2, 82, 7, 3, 2, 2, 2, 83, 84, 5, 20, 11, 2, 84, 85, 8, 5, 1, 2, 85, 9, 3, 2, 2, 2, 86, 87, 5, 16, 9, 2, 87, 88, 7, 16, 2, 2, 88, 89, 5, 20, 11, 2, 89, 90, 8, 6, 1, 2, 90, 11, 3, 2, 2, 2, 91, 92, 7, 6, 2, 2, 92, 93, 7, 17, 2, 2, 93, 94, 5, 20, 11, 2, 94, 95, 8, 7, 1, 2, 95, 13, 3, 2, 2, 2, 96, 97, 7, 29, 2, 2, 97, 98, 8, 8, 1, 2, 98, 15, 3, 2, 2, 2, 99, 100, 9, 2, 2, 2, 100, 101, 8, 9, 1, 2, 101, 17, 3, 2, 2, 2, 102, 106, 5, 22, 12, 2, 103, 106, 5, 24, 13, 2, 104, 106, 5, 26, 14, 2, 105, 102, 3, 2, 2, 2, 105, 103, 3, 2, 2, 2, 105, 104, 3, 2, 2, 2, 106, 109, 3, 2, 2, 2, 107, 105, 3, 2, 2, 2, 107, 108, 3, 2, 2, 2, 108, 19, 3, 2, 2, 2, 109, 107, 3, 2, 2, 2, 110, 111, 5, 22, 12, 2, 111, 118, 8, 11, 1, 2, 112, 113, 7, 18, 2, 2, 113, 114, 5, 22, 12, 2, 114, 115, 8, 11, 1, 2, 115, 117, 3, 2, 2, 2, 116, 112, 3, 2, 2, 2, 117, 120, 3, 2, 2, 2, 118, 116, 3, 2, 2, 2, 118, 119, 3, 2, 2, 2, 119, 121, 3, 2, 2, 2, 120, 118, 3, 2, 2, 2, 121, 122, 8, 11, 1, 2, 122, 21, 3, 2, 2, 2, 123, 124, 7, 27, 2, 2, 124, 128, 8, 12, 1, 2, 125, 126, 7, 28, 2, 2, 126, 128, 8, 12, 1, 2, 127, 123, 3, 2, 2, 2, 127, 125, 3, 2, 2, 2, 128, 23, 3, 2, 2, 2, 129, 130, 9, 3, 2, 2, 130, 25, 3, 2, 2, 2, 131, 132, 9, 4, 2, 2, 132, 27, 3, 2, 2, 2, 11, 35, 60, 65, 73, 81, 105, 107, 118, 127]
\ No newline at end of file

grammar/unl/unlLexer.py

@@ -8,6 +8,9 @@ else:
     from typing.io import TextIO
 
 
+from asd import unl
+
+
 
 def serializedATN():
     with StringIO() as buf:

grammar/unl/unlListener.py

@@ -5,24 +5,27 @@ if __name__ is not None and "." in __name__:
 else:
     from unlParser import unlParser
 
+from asd import unl
+
+
 # This class defines a complete listener for a parse tree produced by unlParser.
 class unlListener(ParseTreeListener):
 
-    # Enter a parse tree produced by unlParser#unlPart.
-    def enterUnlPart(self, ctx:unlParser.UnlPartContext):
+    # Enter a parse tree produced by unlParser#unlGraph.
+    def enterUnlGraph(self, ctx:unlParser.UnlGraphContext):
         pass
 
-    # Exit a parse tree produced by unlParser#unlPart.
-    def exitUnlPart(self, ctx:unlParser.UnlPartContext):
+    # Exit a parse tree produced by unlParser#unlGraph.
+    def exitUnlGraph(self, ctx:unlParser.UnlGraphContext):
         pass
 
 
-    # Enter a parse tree produced by unlParser#relationOccurrence.
-    def enterRelationOccurrence(self, ctx:unlParser.RelationOccurrenceContext):
+    # Enter a parse tree produced by unlParser#link.
+    def enterLink(self, ctx:unlParser.LinkContext):
         pass
 
-    # Exit a parse tree produced by unlParser#relationOccurrence.
-    def exitRelationOccurrence(self, ctx:unlParser.RelationOccurrenceContext):
+    # Exit a parse tree produced by unlParser#link.
+    def exitLink(self, ctx:unlParser.LinkContext):
         pass
 
 

parse.py

@@ -112,8 +112,11 @@ def parse_unl(input):
     # -- Parse UNL part
     parser = instantiate_lexer_parser(input, unlLexer, unlParser)
     print("--- Parse UNL representation")
-    tree = parser.unlPart()
+    tree = parser.unlGraph()
     print("----- resulting tree:\n" + tree.toStringTree(recog=parser))  
+    unl = tree.out
+    
+    return unl
     
      
 #==============================================================================
@@ -140,7 +143,8 @@ def main(argv):
     
     # -- UNL Parsing (Sentence UNL Part)
     print("-- UNL Parsing (UNL Representation) ")
-    parse_unl(InputStream(unl_part)) 
+    unl = parse_unl(InputStream(unl_part))
+    print("----- UNL string:\n" + unl.to_string())
 
 
 if __name__ == '__main__':