Add symbol and structure support

[fur] / parsing.py

diff --git a/parsing.py b/parsing.py
index f16af14..20ac376 100644 (file)
--- a/parsing.py
+++ b/parsing.py
@@ -43,6 +43,14 @@ FurIntegerLiteralExpression = collections.namedtuple(
     ],
 )
 
+FurLambdaExpression = collections.namedtuple(
+    'FurLambdaExpression',
+    (
+        'argument_name_list',
+        'statement_list',
+    ),
+)
+
 FurStringLiteralExpression = collections.namedtuple(
     'FurStringLiteralExpression',
     [
@@ -53,6 +61,7 @@ FurStringLiteralExpression = collections.namedtuple(
 FurSymbolExpression = collections.namedtuple(
     'FurSymbolExpression',
     [
+        'metadata',
         'symbol',
     ],
 )
@@ -60,6 +69,7 @@ FurSymbolExpression = collections.namedtuple(
 FurNegationExpression = collections.namedtuple(
     'FurNegationExpression',
     [
+        'metadata',
         'value',
     ],
 )
@@ -67,6 +77,7 @@ FurNegationExpression = collections.namedtuple(
 FurInfixExpression = collections.namedtuple(
     'FurInfixExpression',
     [
+        'metadata',
         'order',
         'operator',
         'left',
@@ -90,6 +101,21 @@ FurIfExpression = collections.namedtuple(
     ],
 )
 
+FurSymbolExpressionPair = collections.namedtuple(
+    'FurSymbolExpressionPair',
+    [
+        'symbol',
+        'expression',
+    ],
+)
+
+FurStructureLiteralExpression = collections.namedtuple(
+    'FurStructureLiteralExpression',
+    [
+        'fields',
+    ],
+)
+
 def _integer_literal_expression_parser(index, tokens):
     failure = (False, index, None)
 
@@ -111,7 +137,14 @@ def _string_literal_expression_parser(index, tokens):
 
 def _symbol_expression_parser(index, tokens):
     if tokens[index].type == 'symbol':
-        return (True, index + 1, FurSymbolExpression(symbol=tokens[index].match))
+        return (
+            True,
+            index + 1,
+            FurSymbolExpression(
+                metadata=tokens[index].metadata,
+                symbol=tokens[index].match,
+            ),
+        )
 
     return (False, index, None)
 
@@ -150,6 +183,94 @@ def _parenthese_wrapped_parser(internal_parser):
 def _parenthesized_expression_parser(index, tokens):
     return _parenthese_wrapped_parser(_expression_parser)(index, tokens)
 
+def symbol_expression_pair_parser(index, tokens):
+    failure = (False, index, None)
+
+    if tokens[index].type == 'symbol':
+        symbol = tokens[index].match
+        index += 1
+    else:
+        return failure
+
+    if tokens[index].type == 'colon':
+        index += 1
+    else:
+        return failure
+
+    success, index, expression = _expression_parser(index, tokens)
+
+    if not success:
+        raise Exception()
+
+    return (
+        True,
+        index,
+        FurSymbolExpressionPair(
+            symbol=symbol,
+            expression=expression,
+        ),
+    )
+
+def _structure_literal_parser(index, tokens):
+    success, index, result = _parenthese_wrapped_parser(_comma_separated_list_parser(symbol_expression_pair_parser))(index, tokens)
+    return (
+        success,
+        index,
+        FurStructureLiteralExpression(
+            fields=result,
+        ),
+    )
+
+def _lambda_expression_parser(index, tokens):
+    failure = (False, index, None)
+
+    if tokens[index].type == 'keyword' and tokens[index].match == 'lambda':
+        index += 1
+    else:
+        return failure
+
+    if tokens[index].type == 'open_parenthese':
+        index += 1
+    else:
+        raise Exception('Expected "(", found "{}" on line {}'.format(
+            tokens[index].match,
+            tokens[index].metadata.line,
+        ))
+
+    success, index, argument_name_list = _comma_separated_list_parser(_symbol_expression_parser)(
+        index,
+        tokens,
+    )
+
+    if tokens[index].type == 'close_parenthese':
+        index += 1
+    else:
+        raise Exception('Expected ")", found "{}" on line {}'.format(
+            tokens[index].match,
+            tokens[index].line,
+        ))
+
+    if tokens[index].match == 'do':
+        index += 1
+    else:
+        return failure
+
+    success, index, statement_list = _zero_or_more_parser(tuple, _statement_parser)(index, tokens)
+
+    _, index, _ = consume_newlines(index, tokens)
+
+    if tokens[index].type == 'keyword' and tokens[index].match == 'end':
+        index += 1
+    else:
+        return failure
+
+    return True, index, FurLambdaExpression(
+        argument_name_list=tuple(an.symbol for an in argument_name_list),
+        statement_list=statement_list,
+    )
+
+
+
 def _list_literal_expression_parser(index, tokens):
     failure = (False, index, None)
 
@@ -162,29 +283,45 @@ def _list_literal_expression_parser(index, tokens):
     else:
         return failure
 
+def _literal_level_expression_parser(index, tokens):
+    return _or_parser(
+        _list_item_expression_parser,
+        _function_call_expression_parser,
+        _parenthesized_expression_parser,
+        _integer_literal_expression_parser,
+        _string_literal_expression_parser,
+        _list_literal_expression_parser,
+        _lambda_expression_parser,
+        _symbol_expression_parser,
+        _structure_literal_parser,
+    )(index, tokens)
+
+def _dot_expression_parser(index, tokens):
+    return _left_recursive_infix_operator_parser(
+        lambda token: token.type == 'period',
+        _literal_level_expression_parser,
+        'dot_level',
+    )(index, tokens)
+
 def _negation_expression_parser(index, tokens):
     failure = (False, index, None)
 
     if tokens[index].match != '-':
         return failure
 
-    success, index, value = _literal_level_expression_parser(index + 1, tokens)
+    metadata = tokens[index].metadata
+
+    success, index, value = _dot_expression_parser(index + 1, tokens)
 
     if not success:
         return failure
 
-    return (True, index, FurNegationExpression(value=value))
+    return (True, index, FurNegationExpression(metadata=metadata, value=value))
 
-def _literal_level_expression_parser(index, tokens):
+def _negation_level_expression_parser(index, tokens):
     return _or_parser(
+        _dot_expression_parser,
         _negation_expression_parser,
-        _list_item_expression_parser,
-        _function_call_expression_parser,
-        _parenthesized_expression_parser,
-        _integer_literal_expression_parser,
-        _string_literal_expression_parser,
-        _list_literal_expression_parser,
-        _symbol_expression_parser,
     )(index, tokens)
 
 def _left_recursive_infix_operator_parser(operator_token_matcher, operand_parser, order):
@@ -204,6 +341,7 @@ def _left_recursive_infix_operator_parser(operator_token_matcher, operand_parser
 
             if success:
                 result = FurInfixExpression(
+                    metadata=tokens[index].metadata,
                     order=order,
                     operator=tokens[index].match,
                     left=result,
@@ -218,7 +356,7 @@ def _left_recursive_infix_operator_parser(operator_token_matcher, operand_parser
 def _multiplication_level_expression_parser(index, tokens):
     return _left_recursive_infix_operator_parser(
         lambda token: token.type == 'multiplication_level_operator',
-        _literal_level_expression_parser,
+        _negation_level_expression_parser,
         'multiplication_level',
     )(index, tokens)
 
@@ -289,6 +427,7 @@ FurListItemExpression = collections.namedtuple(
     'FurListItemExpression',
     [
         'list_expression',
+        'metadata',
         'index_expression',
     ],
 )
@@ -296,6 +435,7 @@ FurListItemExpression = collections.namedtuple(
 FurFunctionCallExpression = collections.namedtuple(
     'FurFunctionCallExpression',
     [
+        'metadata',
         'function',
         'arguments',
     ],
@@ -345,6 +485,8 @@ def _list_item_expression_parser(index, tokens):
     if not success:
         return failure
 
+    metadata = tokens[index].metadata
+
     success, index, index_expression = _bracket_wrapped_parser(_expression_parser)(
         index,
         tokens,
@@ -358,9 +500,12 @@ def _list_item_expression_parser(index, tokens):
         # We can't give this a better name without a bunch of checks, however.
         list_expression = FurListItemExpression(
             list_expression=list_expression,
+            metadata=metadata,
             index_expression=index_expression,
         )
 
+        metadata = tokens[index].metadata
+
         success, index, index_expression = _bracket_wrapped_parser(_expression_parser)(
             index,
             tokens,
@@ -381,6 +526,8 @@ def _function_call_expression_parser(index, tokens):
     if not success:
         return failure
 
+    metadata = tokens[index].metadata
+
     success, index, arguments = _parenthese_wrapped_parser(_comma_separated_expression_list_parser)(
         index,
         tokens,
@@ -393,10 +540,13 @@ def _function_call_expression_parser(index, tokens):
         # "function" is actually the full function call if the next parse attempt doesn't succeed
         # We can't give this a better name without a bunch of checks, however.
         function = FurFunctionCallExpression(
+            metadata=metadata,
             function=function,
             arguments=arguments,
         )
 
+        metadata = tokens[index].metadata
+
         success, index, arguments = _parenthese_wrapped_parser(_comma_separated_expression_list_parser)(
             index,
             tokens,
@@ -448,9 +598,6 @@ def _if_expression_parser(index, tokens):
         ),
     )
 
-
-
-
 _expression_parser = _or_parser(
     _or_level_expression_parser,
     _if_expression_parser, # This should always be at the top level
@@ -473,7 +620,7 @@ def _assignment_statement_parser(index, tokens):
 
     if tokens[index].type == 'symbol':
         target = tokens[index].match
-        target_assignment_line = tokens[index].line
+        target_assignment_line = tokens[index].metadata.line
 
         index += 1
     else:
@@ -554,7 +701,7 @@ def _function_definition_statement_parser(index, tokens):
 
     return True, index, FurFunctionDefinitionStatement(
         name=name,
-    argument_name_list=tuple(an.symbol for an in argument_name_list),
+        argument_name_list=tuple(an.symbol for an in argument_name_list),
         statement_list=statement_list,
     )