angr.utils.constants

angr.utils.constants.MAGIC_CONSTANTS = frozenset({195948557, 2343432205, 2880289470, 3131961357, 3135097598, 3405691582, 3405697037, 3735927486, 3735928559, 3735929054, 3735936685, 4207869677, 4276215469, 4277009102})

Well-known “magic” constants that are universally recognized in hexadecimal.

angr.utils.constants.NUMERIC_BOUNDARY_VALUES = frozenset({2147483647, 4294967295, 9223372036854775807, 18446744073709551615})

Values sitting on a standard integer-type boundary (INT_MAX/UINT_MAX for 32- and 64-bit). These read better as signed decimal (e.g. -1, 2147483647) than as a long run of set bits, regardless of the width inferred for the surrounding expression.

angr.utils.constants.is_alignment_mask(n)
angr.utils.constants.should_use_hex(value, bits=None)

Heuristically decide whether an integer constant reads better in hexadecimal than in decimal in decompiled output.

The rules are checked in priority order:

Hexadecimal (returns True):

  1. The value is a well-known “magic” constant (e.g. 0xdeadbeef); see MAGIC_CONSTANTS.

  2. The value is a known alignment mask; see is_alignment_mask().

  3. The binary representation contains a run of >= 8 consecutive 1 bits – typical of sub-word bitmasks (e.g. 0xff, 0xfff). Type-saturating values (e.g. -1/0xffffffff, INT_MAX/0x7fffffff) are intentionally excluded: those read better as signed decimal, which the codegen renders separately.

  4. The binary representation is a short bit pattern (period 2, 3, 4, or 8) repeated to fill a byte-aligned width – typical of mask/flag constants (e.g. 0x55/0xaaaa for 0101..., 0x1111, 0xabab).

  5. The value is a single set bit (power of two) that is >= 256 – a bit flag (e.g. 0x100, 0x4000).

Decimal (returns False):

  1. The decimal representation of abs(value) contains a run of > 3 (i.e. >= 4) identical digits (e.g. 10000, 11111, 1000000).

  2. The value is a nonzero multiple of 1000 – a “round” human-authored decimal number (e.g. 5000, 20000).

  3. The magnitude is small (abs(value) <= 9) – e.g. small loop counters and offsets.

Hexadecimal (weak signal, returns True):

  1. The low byte is zero and the value is > 0xff – a “round” hex number (e.g. 0x1200).

Otherwise the value is shown in decimal (returns False).

Parameters:

  • value (int) – The integer value to format.

  • bits (int | None) – The bit-width of the value’s type. Used to normalize negative values to their unsigned form and to detect type-saturating values (rule 3). If None, the value’s own bit length is used and only the absolute boundary values in NUMERIC_BOUNDARY_VALUES are treated as saturating.

Return type:

bool

Returns:

True if the value should be displayed in hexadecimal, False for decimal.