Merge pull request #1 from ACRIOS-Systems/feature/support-comments-parsing

Add support for variable comment headers parsing

Author: Hadatko

.gitignore

@@ -5,5 +5,6 @@
 .DS_Store
 .tox
 .pytest_cache
+build
 **/build/*
 **/dist/**

dissect/cstruct/cstruct.py

@@ -81,7 +81,7 @@ def _read(self, stream):
 
         return Instance(self, r, sizes)
 
-    def add_fields(self, name, type_, offset=None):
+    def add_fields(self, name, type_, offset=None, commentAttributes={}):
         raise NotImplementedError("Can't add fields to a compiled structure")
 
     def __repr__(self):
@@ -401,8 +401,9 @@ def _structs(self, data):
                     self.cstruct.addtype(td, st)
 
     def _parse_fields(self, s):
+        commentAttributes = {}
         fields = re.finditer(
-            r'(?P<type>[^\s]+)\s+(?P<name>[^\s\[:]+)(\s*:\s*(?P<bits>\d+))?(\[(?P<count>[^;\n]*)\])?;',
+            r'(?P<commentBlock>\/\*(\*(?!\/)|[^*])*\*\/)?[ \t\r\n]*(?P<type>[^\s]+)\s+(?P<name>[^\s\[:]+)(\s*:\s*(?P<bits>\d+))?(\[(?P<count>[^;\n]*)\])?;',
             s,
         )
         r = []
@@ -411,6 +412,21 @@ def _parse_fields(self, s):
             if d['type'].startswith('//'):
                 continue
 
+            commentAttributes={}
+
+            #parse the comment header
+            if d['commentBlock'] is not None and d['commentBlock'].startswith('/*'):
+                commentfields = re.finditer(
+                    r'@(?P<commentType>[a-zA-Z0-9_\-\/,.]+):[ \t]*(?P<commentVal>[a-zA-Z0-9_\-,.]+)',
+                    d['commentBlock'],
+                )
+                for cf in commentfields:
+                    cd=cf.groupdict()
+                    try:
+                        commentAttributes[cd['commentType']]=cd['commentVal']
+                    except Exception:
+                        pass
+
             type_ = self.cstruct.resolve(d['type'])
 
             d['name'] = d['name'].replace('(', '').replace(')', '')
@@ -433,7 +449,7 @@ def _parse_fields(self, s):
                 d['name'] = d['name'][1:]
                 type_ = Pointer(self.cstruct, type_)
 
-            field = Field(d['name'], type_, int(d['bits']) if d['bits'] else None)
+            field = Field(d['name'], type_, int(d['bits']) if d['bits'] else None, commentAttributes=commentAttributes)
             r.append(field)
 
         return r
@@ -747,6 +763,7 @@ def __init__(self, cstruct, name, fields=None):
         self.lookup = OrderedDict()
         self.fields = fields if fields else []
 
+
         for f in self.fields:
             self.lookup[f.name] = f
 
@@ -860,15 +877,15 @@ def _write(self, stream, data):
 
         return num
 
-    def add_field(self, name, type_, offset=None):
+    def add_fields(self, name, type_, offset=None, commentAttributes={}):
         """Add a field to this structure.
 
         Args:
             name: The field name.
             type_: The field type.
             offset: The field offset.
         """
-        field = Field(name, type_, offset=offset)
+        field = Field(name, type_, offset=offset, commentAttributes=commentAttributes)
         self.fields.append(field)
         self.lookup[name] = field
         self.size = None
@@ -969,14 +986,15 @@ def reset(self):
 class Field(object):
     """Holds a structure field."""
 
-    def __init__(self, name, type_, bits=None, offset=None):
+    def __init__(self, name, type_, bits=None, offset=None, commentAttributes={}):
         self.name = name
         self.type = type_
         self.bits = bits
         self.offset = offset
+        self.commentAttributes = commentAttributes
 
     def __repr__(self):
-        return '<Field {} {}>'.format(self.name, self.type)
+        return '<Field {} {} {}>'.format(self.name, self.type, self.commentAttributes)
 
 
 class Array(BaseType):

tests/test_basic.py

@@ -731,9 +731,9 @@ def test_struct_sizes():
     """, compiled=False)
 
     assert len(c.static) == 4
-    c.static.add_field("another", c.uint32)
+    c.static.add_fields("another", c.uint32)
     assert len(c.static) == 8
-    c.static.add_field("atoffset", c.uint32, 12)
+    c.static.add_fields("atoffset", c.uint32, 12)
     assert len(c.static) == 16
 
     a = c.static(b'\x01\x00\x00\x00\x02\x00\x00\x00\x00\x00\x00\x00\x03\x00\x00\x00')
@@ -770,3 +770,37 @@ def test_dumpstruct(capsys):
     captured_2 = capsys.readouterr()
 
     assert captured_1.out == captured_2.out
+
+
+def test_commentfieldparse(capsys):
+    c = cstruct.cstruct()
+    c.load("""
+    /*discardedCom1*/
+    struct test{
+        /*
+        * @scale: 0.001
+        * @unit: testUnit1
+        */
+        int testVar1;
+        int testVar2;
+        /* dicardedCom2
+        * @scale: 5
+        * @unit: testUnit2
+        */
+        int testVar3;
+    };
+    """, compiled=False)
+
+    assert c.test.name == 'test'
+
+    assert 'testVar1' in c.test.lookup
+    assert 'testVar2' in c.test.lookup
+    assert 'testVar2' in c.test.lookup
+
+    assert c.test.lookup['testVar1'].commentAttributes['scale'] == '0.001'
+    assert c.test.lookup['testVar1'].commentAttributes['unit'] == 'testUnit1'
+
+    assert c.test.lookup['testVar2'].commentAttributes == {}
+
+    assert c.test.lookup['testVar3'].commentAttributes['scale'] == '5'
+    assert c.test.lookup['testVar3'].commentAttributes['unit'] == 'testUnit2'