fix(ciphers): operate on raw bytes in base32

ciphers/base32.py

@@ -9,35 +9,47 @@
 
 def base32_encode(data: bytes) -> bytes:
     """
+    >>> base32_encode(b"")
+    b''
     >>> base32_encode(b"Hello World!")
     b'JBSWY3DPEBLW64TMMQQQ===='
     >>> base32_encode(b"123456")
     b'GEZDGNBVGY======'
     >>> base32_encode(b"some long complex string")
     b'ONXW2ZJANRXW4ZZAMNXW24DMMV4CA43UOJUW4ZY='
+    >>> base32_encode("héllo".encode())
+    b'NDB2S3DMN4======'
     """
-    binary_data = "".join(bin(ord(d))[2:].zfill(8) for d in data.decode("utf-8"))
-    binary_data = binary_data.ljust(5 * ((len(binary_data) // 5) + 1), "0")
-    b32_chunks = map("".join, zip(*[iter(binary_data)] * 5))
+    if not data:
+        return b""
+    binary_data = "".join(bin(byte)[2:].zfill(8) for byte in data)
+    if (remainder := len(binary_data) % 5) != 0:
+        binary_data = binary_data + "0" * (5 - remainder)
+    b32_chunks = [binary_data[i : i + 5] for i in range(0, len(binary_data), 5)]
     b32_result = "".join(B32_CHARSET[int(chunk, 2)] for chunk in b32_chunks)
-    return bytes(b32_result.ljust(8 * ((len(b32_result) // 8) + 1), "="), "utf-8")
+    pad_len = (8 - (len(b32_result) % 8)) % 8
+    return (b32_result + "=" * pad_len).encode("ascii")
 
 
 def base32_decode(data: bytes) -> bytes:
     """
+    >>> base32_decode(b"")
+    b''
     >>> base32_decode(b'JBSWY3DPEBLW64TMMQQQ====')
     b'Hello World!'
     >>> base32_decode(b'GEZDGNBVGY======')
     b'123456'
     >>> base32_decode(b'ONXW2ZJANRXW4ZZAMNXW24DMMV4CA43UOJUW4ZY=')
     b'some long complex string'
+    >>> base32_decode(b'NDB2S3DMN4======').decode()
+    'héllo'
     """
-    binary_chunks = "".join(
-        bin(B32_CHARSET.index(_d))[2:].zfill(5)
-        for _d in data.decode("utf-8").strip("=")
-    )
-    binary_data = list(map("".join, zip(*[iter(binary_chunks)] * 8)))
-    return bytes("".join([chr(int(_d, 2)) for _d in binary_data]), "utf-8")
+    if not data:
+        return b""
+    s = data.decode("ascii").strip("=")
+    binary_chunks = "".join(bin(B32_CHARSET.index(c))[2:].zfill(5) for c in s)
+    num_bytes = len(binary_chunks) // 8
+    return bytes(int(binary_chunks[i : i + 8], 2) for i in range(0, num_bytes * 8, 8))
 
 
 if __name__ == "__main__":