Moving grammar out of the python file.

Additionally, the grammar now uses custom semantics and creates flat
structures.

.gitignore

+app/openccg_parser.py
 
 # Created by https://www.gitignore.io/api/python
 

Makefile

+app/openccg_parser.py: OpenCCG.ebnf
+	tatsu --generate-parser $< --outfile $@

OpenCCG.ebnf

+@@grammar :: OpenCCG
+
+start
+    = semspec $
+    ;
+
+semspec
+    = nominal
+    | term
+    | conjunction
+    | expression
+    ;
+
+conjunction
+    = @:expression '^' ~ @:conjunction
+    | @:expression
+    ;
+
+expression
+    = role_expression
+    | nominal_expression
+    | variable_expression
+    | atom_expression
+    ;
+
+role_expression
+    = role
+    ;
+
+variable_expression
+    = variable:variable '^' ~ roles:conjunction
+    | variable:variable
+    ;
+
+nominal_expression
+    = nominal:nominal
+    ;
+
+atom_expression
+    = entity:atom
+    ;
+
+term
+    = '(' ~ @:conjunction ')'
+    ;
+
+nominal
+    = '@' ~ nominal:variable roles:term
+    ;
+
+role
+    = '<' type:atom '>' target:variable
+    | '<' type:atom '>' target:term
+    | '<' type:atom '>' target:atom
+    ;
+
+variable
+    = /[a-z]\d+(:[a-zA-Z\-]+)*/
+    ;
+
+atom
+    = /[a-zA-Z\-\.]+/
+    ;

app/ccgparser.py

-from tatsu import parse
+from tatsu.ast import AST
 from tatsu.util import asjson
 
-
-GRAMMAR = r"""
-
-start
-    = semspec $
-    ;
-
-semspec
-    = nominal
-    | term
-    | conjunction
-    | expression
-    ;
-
-conjunction
-    = @:expression '^' ~ @:conjunction
-    | @:expression
-    ;
-
-expression
-    = role_expression
-    | nominal_expression
-    | variable_expression
-    | atom_expression
-    ;
-
-role_expression
-    = role:role
-    ;
-
-variable_expression
-    = variable:variable
-    ;
-
-nominal_expression
-    = nominal:nominal
-    ;
-
-atom_expression
-    = atom:atom
-    ;
-
-term
-    = '(' ~ @:conjunction ')'
-    ;
-
-nominal
-    = '@' ~ nominal:variable roles:term
-    ;
-
-role
-    = '<' type:atom '>' target:variable
-    | '<' type:atom '>' target:term
-    | '<' type:atom '>' target:atom
-    ;
-
-variable
-    = /[a-z]\d+(:[a-zA-Z\-]+)*/
-    ;
-
-atom
-    = /[a-zA-Z\-\.]+/
-    ;
-"""
+from openccg_parser import OpenCCGParser, OpenCCGSemantics
 
 
 def ccg_to_json(to_parse):
@@ -75,4 +12,43 @@ def ccg_to_json(to_parse):
     Returns:
         A JSON representation of the OpenCCG string.
     """
-    return asjson(parse(GRAMMAR, to_parse))
+    return asjson(OpenCCGParser().parse(to_parse,
+                                        semantics=OpenCCGCustomSemantics(),
+                                        parseinfo=False))
+
+
+class OpenCCGCustomSemantics(OpenCCGSemantics):
+    """The custom semantics are used to flatten the role structure
+    caused by right-deep trees of the conjunction parses."""
+
+    def variable_expression(self, ast):
+        if ast['roles'] and len(ast['roles']) >= 2:
+            if isinstance(ast['roles'][-1], list):
+                role = ast['roles'][-1][0]
+                del ast['roles'][-1][0]
+                if len(ast['roles'][-1]) == 0:
+                    del ast['roles'][-1]
+                ast['roles'].insert(-1, role)
+        return ast
+
+    def role(self, ast):
+        if isinstance(ast['target'], str):
+            return ast
+        if ast['target']['roles'] and len(ast['target']['roles']) >= 2:
+            for i, maybe_role in enumerate(ast['target']['roles']):
+                if isinstance(maybe_role, list):
+                    role = ast['target']['roles'][i][0]
+                    del ast['target']['roles'][i][0]
+                    if len(ast['target']['roles'][i]) == 0:
+                        del ast['target']['roles'][i]
+                    ast['target']['roles'].insert(i, role)
+
+            if len(ast['target']['roles']) >= 3:
+                for i, role in enumerate(ast['target']['roles']):
+                    if isinstance(role, list):
+                        r0, r1 = role
+                        del ast['target']['roles'][i]
+                        ast['target']['roles'].insert(i, r1)
+                        ast['target']['roles'].insert(i, r0)
+
+        return ast