from typing import Dict, Optional, Set, Tuple, Iterator, Union
import archinfo
import claripy
from angr.storage.memory_object import bv_slice
[docs]class MultiValues:
"""
Represents a byte vector where each byte can have one or multiple values.
As an implementation optimization (so that we do not create excessive sets and dicts), self._single_value stores a
claripy AST when this MultiValues object represents only one value at offset 0.
"""
__slots__ = (
"_values",
"_single_value",
)
_single_value: Optional[claripy.ast.Bits]
_values: Optional[Dict[int, Set[claripy.ast.Bits]]]
[docs] def __init__(
self,
v: Union[claripy.ast.Bits, "MultiValues", None, Dict[int, Set[claripy.ast.Bits]]] = None,
offset_to_values=None,
):
if v is not None and offset_to_values is not None:
raise TypeError("You cannot specify v and offset_to_values at the same time!")
self._single_value = (
None
if v is None
else v if isinstance(v, claripy.ast.Bits) else v._single_value if isinstance(v, MultiValues) else None
)
self._values = (
offset_to_values
if offset_to_values is not None
else v if isinstance(v, dict) else v._values if isinstance(v, MultiValues) else None
)
if self._single_value is None and self._values is None:
self._values = {}
# if only one value is passed in, assign it to self._single_value
if self._values:
if len(self._values) == 1 and 0 in self._values and len(self._values[0]) == 1:
self._single_value = next(iter(self._values[0]))
self._values = None
if self._values:
# sanity check
for vs in self._values.values():
if not isinstance(vs, set):
raise TypeError("Each value in offset_to_values must be a set!")
[docs] def add_value(self, offset: int, value: claripy.ast.Bits) -> None:
if len(value) == 0:
return
if self._single_value is not None:
if len(self._single_value) == 0:
if offset == 0:
self._single_value = value
else:
self._single_value = None
self._values = {offset: {value}}
return
self._values = {0: {self._single_value}}
self._single_value = None
if self._values is None:
self._values = {}
if offset not in self._values:
self._values[offset] = set()
# if value overlaps with existing values, we need to break them down
succ_offset = self._adjacent_offset(offset, before=False)
if succ_offset is not None:
value_end = offset + value.length // 8
if value_end > succ_offset:
# value is too long. we need to break value into two
mid_value_size = succ_offset - offset
remaining_value = value[value.length - mid_value_size * 8 - 1 : 0]
# update value
value = value[value.length - 1 : value.length - mid_value_size * 8]
self.add_value(succ_offset, remaining_value)
if self._values[offset]:
curr_value_size = next(iter(self._values[offset])).length // 8
if curr_value_size > value.length // 8:
# we need to break existing values
new_curr_values = set()
remaining_values = set()
for v in self._values[offset]:
new_curr_values.add(v[v.length - 1 : v.length - value.length])
remaining_values.add(v[v.length - value.length - 1 : 0])
self._values[offset] = new_curr_values
for v in remaining_values:
self.add_value(offset + value.length // 8, v)
elif curr_value_size < value.length // 8:
# value is too long. we need to break value into two
remaining_value = value[value.length - curr_value_size * 8 - 1 : 0]
# update value
value = value[value.length - 1 : value.length - curr_value_size * 8]
self.add_value(offset + curr_value_size, remaining_value)
self._values[offset].add(value)
pre_offset = self._adjacent_offset(offset, before=True)
if pre_offset is not None:
pre_values = self._values[pre_offset]
pre_values_size = next(iter(pre_values)).length // 8
if pre_offset + pre_values_size > offset:
# we need to break the preceding values
new_pre_value_size = offset - pre_offset
new_pre_values = set()
remaining_values = set()
for v in pre_values:
new_pre_values.add(v[v.length - 1 : v.length - new_pre_value_size * 8])
remaining_values.add(v[v.length - new_pre_value_size * 8 - 1 : 0])
self._values[pre_offset] = new_pre_values
for v in remaining_values:
self.add_value(offset, v)
[docs] def one_value(self, strip_annotations: bool = False) -> Optional[claripy.ast.Bits]:
if self._single_value is not None:
return self._single_value
assert self._values is not None
if len(self._values) == 1 and len(self._values[0]) == 1:
return next(iter(self._values[0]))
if strip_annotations:
all_values_wo_annotations = {
bv.remove_annotations(bv.annotations) if bv.annotations else bv for bv in self._values[0]
}
if len(all_values_wo_annotations) == 1:
return next(iter(all_values_wo_annotations))
return None
def __len__(self) -> int:
if self._single_value is not None:
return self._single_value.length
assert self._values is not None
if len(self._values) == 0:
return 0
max_offset = max(self._values.keys())
max_len = max(x.size() for x in self._values[max_offset])
return max_offset * 8 + max_len # FIXME: we are assuming byte_width of 8
[docs] def merge(self, mv: "MultiValues") -> "MultiValues":
new_values = {k: set(v) for k, v in self.items()}
for off, vs in mv.items():
if off not in new_values:
new_values[off] = set(vs)
else:
new_values[off] |= vs
return MultiValues(offset_to_values=new_values)
def __eq__(self, other) -> bool:
if not isinstance(other, MultiValues):
return False
if self._single_value is not None and other._single_value is not None:
return self._single_value is other._single_value
if self._single_value is not None and other._single_value is None:
return False
if self._single_value is None and other._single_value is not None:
return False
assert self._values is not None
assert other._values is not None
if set(self._values.keys()) != set(other._values.keys()):
return False
for k in self._values.keys():
if self._values[k] != other._values[k]:
return False
return True
def __repr__(self):
if self._single_value is not None:
return f"<{self.__class__.__name__}({self._single_value})>"
return f"<{self.__class__.__name__}({self._values})>"
def __contains__(self, offset: int) -> bool:
if self._single_value is not None:
return offset == 0
return False if not self._values else offset in self._values
def __getitem__(self, offset: int) -> Set[claripy.ast.Bits]:
if self._single_value is not None:
if offset == 0:
return {self._single_value}
raise KeyError()
elif not self._values:
raise KeyError()
return self._values[offset]
[docs] def keys(self) -> Set[int]:
if self._single_value is not None:
return {0}
return set() if not self._values else set(self._values.keys())
[docs] def values(self) -> Iterator[Set[claripy.ast.Bits]]:
if self._single_value is not None:
yield {self._single_value}
else:
if self._values is None:
return
yield from self._values.values()
[docs] def items(self) -> Iterator[Tuple[int, Set[claripy.ast.Bits]]]:
if self._single_value is not None:
yield 0, {self._single_value}
else:
if self._values is None:
yield 0, set()
else:
yield from self._values.items()
[docs] def count(self) -> int:
if self._single_value is not None:
return 1
if self._values is None:
return 0
return len(self._values)
[docs] def concat(self, other: Union["MultiValues", claripy.ast.Bits, bytes]) -> "MultiValues":
if isinstance(other, bytes):
other = claripy.BVV(other)
if isinstance(other, claripy.ast.Bits):
other = MultiValues(other)
offset = len(self) // 8
result = MultiValues(self)
for k, v in other.items():
for v2 in v:
result.add_value(k + offset, v2)
return result
#
# Private methods
#
def _adjacent_offset(self, offset: int, before: bool = True) -> Optional[int]:
"""
Find the offset that is right before or after the given offset.
:param offset: The specified offset.
:param before: True if we want to find the offset right before the specified offset, False if we want to find
the offset right after the specified offset.
:return: The adjacent offset as requested. If the requested adjacent offset does not exist, return None.
"""
sorted_offsets = list(sorted(self._values.keys())) if self._values is not None else [0]
for i, off in enumerate(sorted_offsets):
if off == offset:
if before:
return sorted_offsets[i - 1] if i > 0 else None
else:
return sorted_offsets[i + 1] if i + 1 < len(sorted_offsets) else None
if off > offset:
# we missed it...
return None
return None