Add ConcreteBytecode.{instr_at_code_offset, index_at_code_offset}

Author: eric-wieser

bytecode/concrete.py

@@ -377,6 +377,45 @@ def to_bytecode(self):
         bytecode.extend(instructions)
         return bytecode
 
+    def index_at_code_offset(self, offset):
+        """
+        Returns the index `i`, for use in slicing, such that:
+            `self[i:]` is the decoded version of `self.to_code()[offset:]`
+            `self[:i]` is the decoded version of `self.to_code()[:offset]`
+
+        For getting a single instruction an offset, use instr_at_code_offset
+
+        Raises IndexError if `offset` > len(self.to_code()), or `offset`
+        lies midway through an instruction.
+        """
+        if offset < 0:
+            raise IndexError('Offset {} is out of range'.format(offset))
+
+        at = 0
+        for i, instr in enumerate(self):
+            if offset == at:
+                return i
+            elif offset < at:
+                raise IndexError('Offset {} lies within instruction #{}, {}'.format(offset, i, instr))
+            if isinstance(instr, ConcreteInstr):
+                at += instr.size
+
+        # returning the length of the array is ok
+        if offset == at:
+            return i + 1
+
+        raise IndexError('Offset {} is out of range for code of length {}'.format(offset, at))
+
+    def instr_at_code_offset(self, offset):
+        """
+        Return the instruction starting at `offset` within `self.to_code()`
+        """
+        i = self.index_at_code_offset(offset)
+        while i < len(self):
+            if isinstance(self[i], ConcreteInstr):
+                return self[i]
+            i += 1
+        raise IndexError('Instruction at {} is out of range'.format(i))
 
 class _ConvertBytecodeToConcrete:
 

bytecode/tests/test_concrete.py

@@ -335,6 +335,42 @@ def test_load_classderef(self):
         self.assertEqual(
             code.co_code, b'\x94\x01\x89\x01' if WORDCODE else b'\x94\x01\x00\x89\x01\x00')
 
+    def test_offset_index(self):
+        concrete = ConcreteBytecode()
+        concrete[:] = [
+            ConcreteInstr('LOAD_FAST', 0),
+            ConcreteInstr('LOAD_FAST', 1),
+            SetLineno(2),
+            ConcreteInstr('BINARY_ADD'),
+            ConcreteInstr('RETURN_VALUE')
+        ]
+        # simple cases
+        self.assertEqual(concrete.index_at_code_offset(0), 0)
+        self.assertEqual(concrete.instr_at_code_offset(0), concrete[0])
+        self.assertEqual(concrete.index_at_code_offset(3), 1)
+        self.assertEqual(concrete.instr_at_code_offset(3), concrete[1])
+        self.assertEqual(concrete.index_at_code_offset(7), 4)
+        self.assertEqual(concrete.instr_at_code_offset(7), concrete[4])
+
+        # these indices are deliberately different
+        # the index returns the lower bound, the SetLineno
+        # the instruction returns the actual instruction
+        self.assertEqual(concrete.index_at_code_offset(6), 2)
+        self.assertEqual(concrete.instr_at_code_offset(6), concrete[3])
+
+        # asking for the index at the end is OK, but not the instruction
+        self.assertEqual(concrete.index_at_code_offset(8), 5)
+        self.assertRaisesRegex(IndexError, 'out of range', concrete.instr_at_code_offset, 8)
+
+        # other disallowed things
+        self.assertRaisesRegex(IndexError, 'within', concrete.instr_at_code_offset, 1)
+        self.assertRaisesRegex(IndexError, 'within', concrete.instr_at_code_offset, 1)
+        self.assertRaisesRegex(IndexError, 'within', concrete.index_at_code_offset, 5)
+        self.assertRaisesRegex(IndexError, 'within', concrete.instr_at_code_offset, 5)
+        self.assertRaisesRegex(IndexError, 'out of range', concrete.index_at_code_offset, -1)
+        self.assertRaisesRegex(IndexError, 'out of range', concrete.instr_at_code_offset, -1)
+        self.assertRaisesRegex(IndexError, 'out of range', concrete.index_at_code_offset,  9)
+        self.assertRaisesRegex(IndexError, 'out of range', concrete.instr_at_code_offset,  9)
 
 class ConcreteFromCodeTests(TestCase):