parse mbc fields

capa/render/default.py

@@ -179,18 +179,15 @@ def render_mbc(doc, ostream):
         if not rule["meta"].get("mbc"):
             continue
 
-        mbcs = rule["meta"]["mbc"]
-        if not isinstance(mbcs, list):
-            raise ValueError("invalid rule: MBC mapping is not a list")
+        for mbc in rule["meta"]["mbc"]:
+            id = mbc.get("id")
+            objective = mbc.get("objective")
+            behavior = mbc.get("behavior")
+            method = mbc.get("method")
 
-        for mbc in mbcs:
-            objective, _, rest = mbc.partition("::")
-            if "::" in rest:
-                behavior, _, rest = rest.partition("::")
-                method, _, id = rest.rpartition(" ")
+            if method:
                 objectives[objective].add((behavior, method, id))
             else:
-                behavior, _, id = rest.rpartition(" ")
                 objectives[objective].add((behavior, id))
 
     rows = []
@@ -199,10 +196,10 @@ def render_mbc(doc, ostream):
         for spec in sorted(behaviors):
             if len(spec) == 2:
                 behavior, id = spec
-                inner_rows.append("%s %s" % (rutils.bold(behavior), id))
+                inner_rows.append("%s [%s]" % (rutils.bold(behavior), id))
             elif len(spec) == 3:
                 behavior, method, id = spec
-                inner_rows.append("%s::%s %s" % (rutils.bold(behavior), method, id))
+                inner_rows.append("%s::%s [%s]" % (rutils.bold(behavior), method, id))
             else:
                 raise RuntimeError("unexpected MBC spec format")
         rows.append(

capa/rules.py

@@ -281,6 +281,42 @@ def parse_description(s, value_type, description=None):
     return value, description
 
 
+def parse_meta(meta):
+    """
+    further process meta items such as MBC
+    """
+    mbcs = meta.get("mbc", [])
+    if not isinstance(mbcs, list):
+        raise InvalidRule("MBC mapping must be a list")
+
+    if mbcs:
+        meta["mbc"] = [parse_canonical_mbc(mbc) for mbc in mbcs]
+
+    # TODO att&ck
+
+    return meta
+
+
+def parse_canonical_mbc(mbc):
+    """
+    parse capa's canonical MBC representation: `OBJECTIVE::Behavior::Method [Identifier]`
+    """
+    objective, _, rest = mbc.partition("::")
+    if "::" in rest:
+        behavior, _, rest = rest.partition("::")
+        method, _, id = rest.rpartition(" ")
+    else:
+        behavior, _, id = rest.rpartition(" ")
+        method = ""
+    return {
+        "id": id.lstrip("[").rstrip("]"),
+        "objective": objective,
+        "behavior": behavior,
+        "method": method,
+        # TODO "micro-behavior": "",
+    }
+
+
 def pop_statement_description_entry(d):
     """
     extracts the description for statements and removes the description entry from the document
@@ -577,7 +613,9 @@ class Rule(object):
         if scope not in SUPPORTED_FEATURES.keys():
             raise InvalidRule("{:s} is not a supported scope".format(scope))
 
-        return cls(name, scope, build_statements(statements[0], scope), d["rule"]["meta"], definition)
+        meta = parse_meta(d["rule"]["meta"])
+
+        return cls(name, scope, build_statements(statements[0], scope), meta, definition)
 
     @staticmethod
     @lru_cache()