Switch back to base64 from standard lib

We can assume this is always present in the Python world.
Also use the standard alphabet, because relay-js does so.
And, just like relay-js, we return '' in case of an error.

Author: Cito

src/graphql_relay/utils/base64.py

@@ -1,4 +1,5 @@
-from typing import Iterator, List
+from base64 import b64encode, b64decode
+import binascii
 
 __all__ = ["base64", "unbase64"]
 
@@ -7,131 +8,17 @@
 
 def base64(s: str) -> Base64String:
     """Encode the string s using Base64."""
-    if isinstance(s, (bytearray, bytes)):
-        s = s.decode("unicode")  # handle encoded string gracefully
+    b: bytes = s.encode("utf-8") if isinstance(s, str) else s
+    return b64encode(b).decode("ascii")
 
-    unicode_list = list(str_to_unicode_seq(s))
-    length = len(unicode_list)
-    rest = length % 3
-    result: List[str] = []
-    extend = result.extend
 
-    for i in range(0, length - rest, 3):
-        a, b, c = unicode_list[i : i + 3]
-        result.extend(
-            (
-                first_6_bits(a),
-                last_2_bits_and_first_4_bits(a, b),
-                last_4_bits_and_first_2_bits(b, c),
-                last_6_bits(c),
-            )
-        )
-
-    if rest == 1:
-        a = unicode_list[-1]
-        extend((first_6_bits(a), last_2_bits_and_first_4_bits(a, 0), "=="))
-    elif rest == 2:
-        a, b = unicode_list[-2:]
-        extend(
-            (
-                first_6_bits(a),
-                last_2_bits_and_first_4_bits(a, b),
-                last_4_bits_and_first_2_bits(b, 0),
-                "=",
-            )
-        )
-
-    return "".join(result)
-
-
-def first_6_bits(a: int) -> str:
-    return to_base_64_char(a >> 2 & 0x3F)
-
-
-def last_2_bits_and_first_4_bits(a: int, b: int) -> str:
-    return to_base_64_char((a << 4 | b >> 4) & 0x3F)
-
-
-def last_4_bits_and_first_2_bits(b: int, c: int) -> str:
-    return to_base_64_char((b << 2 | c >> 6) & 0x3F)
-
-
-def last_6_bits(c: int) -> str:
-    return to_base_64_char(c & 0x3F)
-
-
-def unbase64(s: str) -> str:
+def unbase64(s: Base64String) -> str:
     """Decode the string s using Base64."""
-    if isinstance(s, (bytearray, bytes)):
-        s = s.decode("ascii")  # handle encoded string gracefully
-
-    unicode_list: List[int] = []
-    extend = unicode_list.extend
-    length = len(s)
-
-    for i in range(0, length, 4):
-        try:
-            a, b, c, d = [from_base_64_char(char) for char in s[i : i + 4]]
-        except (KeyError, ValueError):
-            return ""  # for compatibility
-        bitmap_24 = a << 18 | b << 12 | c << 6 | d
-        extend((bitmap_24 >> 16 & 0xFF, bitmap_24 >> 8 & 0xFF, bitmap_24 & 0xFF))
-
-    i = length - 1
-    while i > 0 and s[i] == "=":
-        i -= 1
-        unicode_list.pop()
-
-    return "".join(unicode_list_to_str(unicode_list))
-
-
-b64_characters = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"
-
-b64_character_map = {c: i for i, c in enumerate(b64_characters)}
-
-
-def to_base_64_char(bit_map_6: int) -> str:
-    return b64_characters[bit_map_6]
-
-
-def from_base_64_char(base_64_char: str) -> int:
-    return 0 if base_64_char == "=" else b64_character_map[base_64_char]
-
-
-def str_to_unicode_seq(s: str) -> Iterator[int]:
-    for utf_char in s:
-        code = ord(utf_char)
-        if code < 0x80:
-            yield code
-        elif code < 0x800:
-            yield 0xC0 | code >> 6
-            yield 0x80 | code & 0x3F
-        elif code < 0x10000:
-            yield 0xE0 | code >> 12
-            yield 0x80 | code >> 6 & 0x3F
-            yield 0x80 | code & 0x3F
-        else:
-            yield 0xF0 | code >> 18
-            yield 0x80 | code >> 12 & 0x3F
-            yield 0x80 | code >> 6 & 0x3F
-            yield 0x80 | code & 0x3F
-
-
-def unicode_list_to_str(s: List[int]) -> Iterator[str]:
-    s.reverse()
-    next_code = s.pop
-    while s:
-        a = next_code()
-        if a & 0x80 == 0:
-            yield chr(a)
-            continue
-        b = next_code()
-        if a & 0xE0 == 0xC0:
-            yield chr((a & 0x1F) << 6 | b & 0x3F)
-            continue
-        c = next_code()
-        if a & 0xF0 == 0xE0:
-            yield chr((a & 0x0F) << 12 | (b & 0x3F) << 6 | c & 0x3F)
-            continue
-        d = next_code()
-        yield chr((a & 0x07) << 18 | (b & 0x3F) << 12 | (c & 0x3F) << 6 | d & 0x3F)
+    try:
+        b: bytes = s.encode("ascii") if isinstance(s, str) else s
+    except UnicodeEncodeError:
+        return ""
+    try:
+        return b64decode(b).decode("utf-8")
+    except (binascii.Error, UnicodeDecodeError):
+        return ""

tests/utils/test_base64.py

@@ -15,6 +15,9 @@ def converts_from_base64_to_unicode():
     def converts_invalid_base64_to_empty_string():
         assert unbase64("") == ""
         assert unbase64("invalid") == ""
+        assert unbase64(example_base64[-1:]) == ""
+        assert unbase64(example_base64[1:]) == ""
+        assert unbase64("!" + example_base64[1:]) == ""
         assert unbase64("Ü" + example_base64[1:]) == ""
 
     def converts_from_unicode_as_bytes_to_base64():