checking for unessisary disassembly

subdisassem/disassemble.py

@@ -30,20 +30,42 @@ import logging
 
 
 class _CapstoneBase:
-    def __init__(self, payload: bytes, offset: int = 0):
-        self.disassembly = list()
-
-        for opcode in self.capstone.disasm(payload, offset):
-            self.disassembly.append(opcode)
+    def __init__(self):
+        self.arch = self.__class__.__name__
+        pass
 
     def __repr__(self) -> str:
         return self.objdump
 
     def __len__(self) -> int:
+        if not self.disassembly:
+            logging.debug(
+                f"payload_missing: use {self.__class__}.load(payload=bytes) prior"
+            )
+            return 0
+
         return len(self.disassembly)
 
+    def load(self, payload: bytes, offset: int = 0):
+        disassembly = list()
+
+        for opcode in self.capstone.disasm(payload, offset):
+            disassembly.append(opcode)
+
+        if disassembly:
+            self.disassembly = disassembly
+        else:
+            logging.debug("disassembly_empty")
+            self.disassembly = list()
+
     @property
     def objdump(self) -> str:
+        if not self.disassembly:
+            logging.debug(
+                f"payload_missing: use {self.__class__}.load(payload=bytes) prior"
+            )
+            return ""
+
         opcodes = str()
 
         for opcode in self.disassembly:
@@ -53,6 +75,12 @@ class _CapstoneBase:
 
     @property
     def disasm(self) -> list:
+        if not self.disassembly:
+            logging.debug(
+                f"payload_missing: use {self.__class__}.load(payload=bytes) prior"
+            )
+            return []
+
         opcodes = list()
 
         for opcode in self.disassembly:
@@ -70,64 +98,51 @@ class _CapstoneBase:
 
 class X86_intel(_CapstoneBase):
     capstone = Cs(CS_ARCH_X86, CS_MODE_16)
-    arch = "x86-16"
 
 
 class X86(_CapstoneBase):
     capstone = Cs(CS_ARCH_X86, CS_MODE_32)
-    arch = "x86-32"
 
 
 class X86_64(_CapstoneBase):
     capstone = Cs(CS_ARCH_X86, CS_MODE_64)
-    arch = "x86-64"
 
 
 class ARM(_CapstoneBase):
     capstone = Cs(CS_ARCH_ARM, CS_MODE_ARM)
-    arch = "ARM"
 
 
 class Thumb(_CapstoneBase):
     capstone = Cs(CS_ARCH_ARM, CS_MODE_THUMB)
-    arch = "Thumb"
 
 
 class ARM_64(_CapstoneBase):
     capstone = Cs(CS_ARCH_ARM64, CS_MODE_ARM)
-    arch = "ARM 64"
 
 
 class MIPS_32_eb(_CapstoneBase):
     capstone = Cs(CS_ARCH_MIPS, CS_MODE_MIPS32 + CS_MODE_BIG_ENDIAN)
-    arch = "MIPS-32 (Big-endian)"
 
 
 class MIPS_64_el(_CapstoneBase):
     capstone = Cs(CS_ARCH_MIPS, CS_MODE_MIPS64 + CS_MODE_LITTLE_ENDIAN)
-    arch = "MIPS-64-EL (Little-endian)"
 
 
 class PPC_64(_CapstoneBase):
     capstone = Cs(CS_ARCH_PPC, CS_MODE_BIG_ENDIAN)
-    arch = "PPC-64"
 
 
 class Sparc(_CapstoneBase):
     capstone = Cs(CS_ARCH_SPARC, CS_MODE_BIG_ENDIAN)
-    arch = "Sparc"
 
 
 class SparcV9(_CapstoneBase):
     capstone = Cs(CS_ARCH_SPARC, CS_MODE_BIG_ENDIAN + CS_MODE_V9)
-    arch = "SparcV9"
 
 
 class SystemZ(_CapstoneBase):
     capstone = Cs(CS_ARCH_SYSZ, 0)
-    arch = "SystemZ"
 
 
 class XCore(_CapstoneBase):
     capstone = Cs(CS_ARCH_XCORE, 0)
-    arch = "XCore"

subdisassem/schema.py

@@ -33,12 +33,11 @@ class Disassembly(Base):
     @property
     def values(self) -> dict:
         values_dict = {
+            "id": self.id,
             "arch": self.arch,
-            "checksum": self.checksum,
             "count": self.count,
             "size": self.size,
             "offset": self.offset,
-            "path": self.path,
         }
 
         return values_dict

subdisassem/scripts.py

@@ -89,7 +89,22 @@ def subdisassem_script():
 
     for arch in archs:
         for offset in range(args.fuzz):
-            disasembler = arch(payload=raw_bytes, offset=offset)
+            disasembler = arch()
+            exists = (
+                session.query(Disassembly)
+                .filter(Disassembly.checksum == checksum)
+                .filter(Disassembly.offset == offset)
+                .filter(Disassembly.arch == disasembler.arch)
+                .first()
+            )
+
+            if exists:
+                logging.debug(
+                    f"subdiassembly_exists: {[disasembler.arch, checksum, offset]}"
+                )
+                continue
+
+            disasembler.load(payload=raw_bytes, offset=offset)
             row = Disassembly()
             row.arch = disasembler.arch
             row.checksum = checksum
@@ -98,16 +113,6 @@ def subdisassem_script():
             row.offset = offset
             row.opcodes = disasembler.objdump
             row.path = str(args.bin_path.absolute())
-
-            exists = (
-                session.query(Disassembly)
-                .filter(Disassembly.checksum == row.checksum)
-                .filter(Disassembly.offset == row.offset)
-                .filter(Disassembly.arch == row.arch)
-                .first()
-            )
-
-            if not exists:
             session.add(row)
 
     session.commit()