Merge pull request #360 from PJCampi/bugfix/348-import-rules-dependencies-only-once

load_grammar now collects all imports to make before loading them to …

lark/load_grammar.py

@@ -725,7 +725,7 @@ class GrammarLoader:
         rule_defs = [options_from_rule(*x) for x in rule_defs]
 
         # Execute statements
-        ignore = []
+        ignore, imports = [], {}
         for (stmt,) in statements:
             if stmt.data == 'ignore':
                 t ,= stmt.children
@@ -734,22 +734,20 @@ class GrammarLoader:
                 if len(stmt.children) > 1:
                     path_node, arg1 = stmt.children
                 else:
-                    path_node ,= stmt.children
+                    path_node, = stmt.children
                     arg1 = None
 
                 if isinstance(arg1, Tree):  # Multi import
-                    dotted_path = path_node.children
+                    dotted_path = tuple(path_node.children)
                     names = arg1.children
-                    aliases = names  # Can't have aliased multi import, so all aliases will be the same as names
+                    aliases = dict(zip(names, names))  # Can't have aliased multi import, so all aliases will be the same as names
                 else:  # Single import
-                    dotted_path = path_node.children[:-1]
-                    names = [path_node.children[-1]]  # Get name from dotted path
-                    aliases = [arg1] if arg1 else names  # Aliases if exist
-
-                grammar_path = os.path.join(*dotted_path) + EXT
+                    dotted_path = tuple(path_node.children[:-1])
+                    name = path_node.children[-1]  # Get name from dotted path
+                    aliases = {name: arg1 or name}  # Aliases if exist
 
                 if path_node.data == 'import_lib':  # Import from library
-                    g = import_grammar(grammar_path)
+                    base_paths = []
                 else:  # Relative import
                     if grammar_name == '<string>':  # Import relative to script file path if grammar is coded in script
                         try:
@@ -759,16 +757,16 @@ class GrammarLoader:
                     else:
                         base_file = grammar_name  # Import relative to grammar file path if external grammar file
                     if base_file:
-                        base_path = os.path.split(base_file)[0]
+                        base_paths = [os.path.split(base_file)[0]]
                     else:
-                        base_path = os.path.abspath(os.path.curdir)
-                    g = import_grammar(grammar_path, base_paths=[base_path])
-
-                aliases_dict = dict(zip(names, aliases))
-                new_td, new_rd = import_from_grammar_into_namespace(g, '__'.join(dotted_path), aliases_dict)
+                        base_paths = [os.path.abspath(os.path.curdir)]
 
-                term_defs += new_td
-                rule_defs += new_rd
+                try:
+                    import_base_paths, import_aliases = imports[dotted_path]
+                    assert base_paths == import_base_paths, 'Inconsistent base_paths for %s.' % '.'.join(dotted_path)
+                    import_aliases.update(aliases)
+                except KeyError:
+                    imports[dotted_path] = base_paths, aliases
 
             elif stmt.data == 'declare':
                 for t in stmt.children:
@@ -776,6 +774,14 @@ class GrammarLoader:
             else:
                 assert False, stmt
 
+        # import grammars
+        for dotted_path, (base_paths, aliases) in imports.items():
+            grammar_path = os.path.join(*dotted_path) + EXT
+            g = import_grammar(grammar_path, base_paths=base_paths)
+            new_td, new_rd = import_from_grammar_into_namespace(g, '__'.join(dotted_path), aliases)
+
+            term_defs += new_td
+            rule_defs += new_rd
 
         # Verify correctness 1
         for name, _ in term_defs:

tests/grammars/three_rules_using_same_token.lark

@@ -0,0 +1,7 @@
+%import common.INT
+
+a: A
+b: A
+c: A
+
+A: "A"

tests/test_parser.py

@@ -1111,6 +1111,13 @@ def _make_parser_test(LEXER, PARSER):
             x = l.parse('N')
             self.assertEqual(next(x.find_data('rule_to_import')).children, ['N'])
 
+        def test_relative_import_rules_dependencies_imported_only_once(self):
+            l = _Lark_open("test_relative_import_rules_dependencies_imported_only_once.lark", rel_to=__file__)
+            x = l.parse('AAA')
+            self.assertEqual(next(x.find_data('a')).children, ['A'])
+            self.assertEqual(next(x.find_data('b')).children, ['A'])
+            self.assertEqual(next(x.find_data('d')).children, ['A'])
+
         def test_import_errors(self):
             grammar = """
             start: NUMBER WORD

tests/test_relative_import_rules_dependencies_imported_only_once.lark

@@ -0,0 +1,5 @@
+%import .grammars.three_rules_using_same_token.a
+%import .grammars.three_rules_using_same_token.b
+%import .grammars.three_rules_using_same_token.c -> d
+
+start: a b d