Rewrite of base85.py algorithm (#9069)

* rewrite of base85.py

* changed maps to list comprehension

* Apply suggestions from code review

Co-authored-by: Tianyi Zheng <tianyizheng02@gmail.com>

---------

Co-authored-by: Tianyi Zheng <tianyizheng02@gmail.com>
This commit is contained in:
Chris O 2023-09-24 20:36:06 +13:00 committed by GitHub
parent 708d906141
commit 882fb2f3c9
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

View File

@ -1,30 +1,55 @@
import base64
"""
Base85 (Ascii85) encoding and decoding
https://en.wikipedia.org/wiki/Ascii85
"""
def base85_encode(string: str) -> bytes:
def _base10_to_85(d: int) -> str:
return "".join(chr(d % 85 + 33)) + _base10_to_85(d // 85) if d > 0 else ""
def _base85_to_10(digits: list) -> int:
return sum(char * 85**i for i, char in enumerate(reversed(digits)))
def ascii85_encode(data: bytes) -> bytes:
"""
>>> base85_encode("")
>>> ascii85_encode(b"")
b''
>>> base85_encode("12345")
>>> ascii85_encode(b"12345")
b'0etOA2#'
>>> base85_encode("base 85")
>>> ascii85_encode(b"base 85")
b'@UX=h+?24'
"""
# encoded the input to a bytes-like object and then a85encode that
return base64.a85encode(string.encode("utf-8"))
binary_data = "".join(bin(ord(d))[2:].zfill(8) for d in data.decode("utf-8"))
null_values = (32 * ((len(binary_data) // 32) + 1) - len(binary_data)) // 8
binary_data = binary_data.ljust(32 * ((len(binary_data) // 32) + 1), "0")
b85_chunks = [int(_s, 2) for _s in map("".join, zip(*[iter(binary_data)] * 32))]
result = "".join(_base10_to_85(chunk)[::-1] for chunk in b85_chunks)
return bytes(result[:-null_values] if null_values % 4 != 0 else result, "utf-8")
def base85_decode(a85encoded: bytes) -> str:
def ascii85_decode(data: bytes) -> bytes:
"""
>>> base85_decode(b"")
''
>>> base85_decode(b"0etOA2#")
'12345'
>>> base85_decode(b"@UX=h+?24")
'base 85'
>>> ascii85_decode(b"")
b''
>>> ascii85_decode(b"0etOA2#")
b'12345'
>>> ascii85_decode(b"@UX=h+?24")
b'base 85'
"""
# a85decode the input into bytes and decode that into a human readable string
return base64.a85decode(a85encoded).decode("utf-8")
null_values = 5 * ((len(data) // 5) + 1) - len(data)
binary_data = data.decode("utf-8") + "u" * null_values
b85_chunks = map("".join, zip(*[iter(binary_data)] * 5))
b85_segments = [[ord(_s) - 33 for _s in chunk] for chunk in b85_chunks]
results = [bin(_base85_to_10(chunk))[2::].zfill(32) for chunk in b85_segments]
char_chunks = [
[chr(int(_s, 2)) for _s in map("".join, zip(*[iter(r)] * 8))] for r in results
]
result = "".join("".join(char) for char in char_chunks)
offset = int(null_values % 5 == 0)
return bytes(result[: offset - null_values], "utf-8")
if __name__ == "__main__":