enhanced x86 parser for directives

osaca/osaca.py

@@ -188,9 +188,9 @@ def insert_byte_marker(args):
         from kerncraft.incore_model import asm_instrumentation
     except ImportError:
         print(
-            "Module kerncraft not installed. Use 'pip install --user "
-            "kerncraft' for installation.\nFor more information see "
-            "https://github.com/RRZE-HPC/kerncraft",
+            'Module kerncraft not installed. Use \'pip install --user '
+            'kerncraft\' for installation.\nFor more information see '
+            'https://github.com/RRZE-HPC/kerncraft',
             file=sys.stderr,
         )
         sys.exit(1)

osaca/parser/parser_x86att.py

@@ -1,5 +1,6 @@
 #!/usr/bin/env python3
 
+import re
 import string
 
 import pyparsing as pp
@@ -108,23 +109,19 @@ class ParserX86ATT(BaseParser):
         ).setResultsName(self.MEMORY_ID)
 
         # Directive
-        directive_option = pp.Combine(
-            pp.Word('#@.', exact=1) + pp.Word(pp.printables, excludeChars=',')
-        )
         directive_parameter = (
             pp.quotedString
-            ^ directive_option
-            ^ identifier
-            ^ hex_number
-            ^ decimal_number
-            ^ self.register
-            ^ pp.Group(pp.Word(pp.alphanums + '_').setResultsName('name'))
+            ^ (
+                pp.Word(pp.printables, excludeChars=',#')
+                + pp.Optional(pp.Suppress(pp.Literal(',')))
+            )
+            ^ pp.Suppress(pp.Literal(','))
         )
-        commaSeparatedList = pp.delimitedList(pp.Optional(directive_parameter), delim=',')
         self.directive = pp.Group(
             pp.Literal('.')
             + pp.Word(pp.alphanums + '_').setResultsName('name')
-            + commaSeparatedList.setResultsName('parameters')
+            # + pp.ZeroOrMore(directive_parameter).setResultsName('parameters')
+            + pp.ZeroOrMore(directive_parameter).setResultsName('parameters')
             + pp.Optional(self.comment)
         ).setResultsName(self.DIRECTIVE_ID)
 
@@ -277,8 +274,18 @@ class ParserX86ATT(BaseParser):
             return self.process_immediate(operand[self.IMMEDIATE_ID])
         if self.LABEL_ID in operand:
             return self.process_label(operand[self.LABEL_ID])
+        if self.DIRECTIVE_ID in operand:
+            return self.process_directive(operand[self.DIRECTIVE_ID])
         return operand
 
+    def process_directive(self, directive):
+        directive_new = {'name': directive['name'], 'parameters': []}
+        if 'parameters' in directive:
+            directive_new['parameters'] = directive['parameters']
+        if 'comment' in directive:
+            directive_new['comment'] = directive['comment']
+        return AttrDict({self.DIRECTIVE_ID: directive_new})
+
     def process_memory_address(self, memory_address):
         """Post-process memory address operand"""
         # Remove unecessarily created dictionary entries during memory address parsing

tests/test_parser_x86att.py

@@ -47,18 +47,26 @@ class TestParserX86ATT(unittest.TestCase):
         self.assertEqual(len(self._get_directive(self.parser, '\t.align\t16,0x90').parameters), 2)
         self.assertEqual(
             self._get_directive(self.parser, '\t.set\tL$set$0,LECIE1-LSCIE1').parameters,
-            [{'name': 'L$set$0'}, {'name': 'LECIE1-LSCIE1'}])
+            ['L$set$0', 'LECIE1-LSCIE1'],
+        )
         self.assertEqual(
             self._get_directive(
                 self.parser,
-                '\t.section __TEXT,__eh_frame,coalesced,no_toc+strip_static_syms+live_support'
-                ).parameters,
-            [{'name': v} for v in 
-             ['__TEXT', '__eh_frame', 'coalesced', 'no_toc+strip_static_syms+live_support']])
+                '\t.section __TEXT,__eh_frame,coalesced,no_toc+strip_static_syms+live_support',
+            ).parameters,
+            [
+                '__TEXT',
+                '__eh_frame',
+                'coalesced',
+                'no_toc+strip_static_syms+live_support',
+            ],
+        )
         self.assertEqual(
             self._get_directive(
-                self.parser, '\t.section\t__TEXT,__literal16,16byte_literals').parameters,
-            [{'name': v} for v in ['__TEXT', '__literal16', '16byte_literals']])
+                self.parser, '\t.section\t__TEXT,__literal16,16byte_literals'
+            ).parameters,
+            ['__TEXT', '__literal16', '16byte_literals'],
+        )
         self.assertEqual(
             self._get_directive(self.parser, '\t.align\t16,0x90').parameters[1], '0x90'
         )
@@ -169,7 +177,7 @@ class TestParserX86ATT(unittest.TestCase):
         instruction_form_3 = {
             'instruction': None,
             'operands': [],
-            'directive': {'name': 'quad', 'parameters': [{'name': '.2.3_2__kmpc_loc_pack.2'}]},
+            'directive': {'name': 'quad', 'parameters': ['.2.3_2__kmpc_loc_pack.2']},
             'comment': 'qed',
             'label': None,
             'line': '.quad   .2.3_2__kmpc_loc_pack.2 #qed',
@@ -186,9 +194,7 @@ class TestParserX86ATT(unittest.TestCase):
                         'scale': 1,
                     }
                 },
-                {
-                    'register': {'name': 'ecx'}
-                }
+                {'register': {'name': 'ecx'}},
             ],
             'directive': None,
             'comment': '12.9',