perf: optimize all_zeros using fast bytes comparison (#3078)

* perf: optimize all_zeros using fast bytes comparison

CHANGELOG.md

@@ -115,6 +115,7 @@
 - fix: replace assert with isinstance guard in get_callee for invalid MethodSpec tokens @williballenthin (SURF-41)
 - fix: incorrect bytes() constructor usage in buf_filled_with @mike-hunhoff #3077
 - fix: remove redundant code related to cli loading @mike-hunhoff #3076
+- fix: optimize all_zeros using fast bytes comparison @mike-hunhoff #3078
 
 ### capa Explorer Web
 

capa/features/extractors/helpers.py

@@ -14,7 +14,6 @@
 
 
 import struct
-import builtins
 from typing import Iterator
 
 MIN_STACKSTRING_LEN = 8
@@ -108,7 +107,12 @@ def reformat_forwarded_export_name(forwarded_name: str) -> str:
 
 
 def all_zeros(bytez: bytes) -> bool:
-    return all(b == 0 for b in builtins.bytes(bytez))
+    # Using `bytez == b'\x00' * len(bytez)` is much faster than `all(b == 0 for b in bytez)`
+    # because it relies on the optimized C implementation of bytes comparison.
+    # While it creates a temporary bytes object, the buffers passed here are small
+    # (typically capped at MAX_BYTES_FEATURE_SIZE = 256 bytes), so the memory overhead is negligible.
+    bytez = bytes(bytez)
+    return bytez == b"\x00" * len(bytez)
 
 
 def twos_complement(val: int, bits: int) -> int: