Repository navigation
Expand file tree
/
Copy path__main__.py
More file actions
248 lines (203 loc) · 8.29 KB
/
Copy path__main__.py
File metadata and controls
248 lines (203 loc) · 8.29 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
"""JSON benchmark."""
from __future__ import annotations
__all__: list[str] = []
import json
import sys
from functools import partial
from math import inf
from timeit import timeit
from typing import TYPE_CHECKING, Any
import fastsimdjson # type: ignore[import-not-found]
import msgspec
import orjson
from tabulate import tabulate # type: ignore[import-untyped]
if sys.version_info >= (3, 10):
from test.support.import_helper import import_fresh_module # type: ignore
else:
from test.support import import_fresh_module
if not (
(jsonyx := import_fresh_module("jsonyx", fresh=["_jsonyx"]))
and (pyjsonyx := import_fresh_module("jsonyx", blocked=["_jsonyx"]))
):
raise ImportError
if TYPE_CHECKING:
from collections.abc import Callable
_Func = Callable[[Any], Any]
_Factory = Callable[[int], Any]
def _raw(obj: bytes | list[bytes], encoding: str = "utf-8") -> bytes:
if isinstance(obj, bytes):
return '"'.encode(encoding) + obj + '"'.encode(encoding)
return (
"[".encode(encoding)
+ ",".encode(encoding).join(obj)
+ "]".encode(encoding)
)
_ENCODE_CASES: dict[str, _Factory] = {
# characters
"7-bit character": lambda n: "\u0020" * n,
"8-bit character": lambda n: "\u0080" * n,
"11-bit character": lambda n: "\u0100" * n,
"16-bit character": lambda n: "\u0800" * n,
"21-bit character": lambda n: "\U00010000" * n,
# variable length characters
"7-in-16-bit character": lambda n: "\u0020" * n + "\u0100",
"7-in-32-bit character": lambda n: "\u0020" * n + "\U00010000",
"11-in-32-bit character": lambda n: "\u0100" * n + "\U00010000",
"16-in-32-bit character": lambda n: "\u0800" * n + "\U00010000",
# escapes
"double quote escape": lambda n: '"' * n,
"backslash escape": lambda n: "\\" * n,
"character escape": lambda n: "\b" * n,
"7-bit escape": lambda n: "\u0000" * n,
# constants
"null": lambda n: [None] * n,
"boolean": lambda n: [False] * n,
# strings
"empty string": lambda n: [""] * n,
# floats
"fixed-point float": lambda n: [0.0] * n,
"scientific float": lambda n: [1e-05] * n,
"minimum float": lambda n: [5e-324] * n,
"maximum float": lambda n: [1.7976931348623157e+308] * n,
# integers
"32-bit (+) integer": lambda n: [1] * n,
"32-bit (-) integer": lambda n: [-1] * n,
"64-bit (+) integer": lambda n: [4294967296] * n,
"64-bit (-) integer": lambda n: [-4294967296] * n,
">64-bit (+) integer": lambda n: [18446744073709551616] * n,
">64-bit (-) integer": lambda n: [-18446744073709551616] * n,
# lists
"empty list": lambda n: [[]] * n, # type: ignore
"non-empty list": lambda n: [[None]] * n,
# dictionaries
"empty dictionary": lambda n: [{}] * n, # type: ignore
"non-empty dictionary": lambda n: [{"": None}] * n,
}
_DECODE_CASES: dict[str, _Factory] = {
# whitespace
"outer space": lambda n: b" " * n + b"null",
"inner space": lambda n: b"[" + b" " * n + b"]",
# characters
"7-bit character": lambda n: _raw(b"\x20" * n),
"8-bit character": lambda n: _raw(b"\xc2\x80" * n),
"11-bit character": lambda n: _raw(b"\xc4\x80" * n),
"16-bit character": lambda n: _raw(b"\xe0\xa0\x80" * n),
"21-bit character": lambda n: _raw(b"\xf0\x90\x80\x80" * n),
# variable length characters
"7-in-16-bit character": lambda n: _raw(b"\x20" * n + b"\xc4\x80"),
"7-in-32-bit character": lambda n: _raw(b"\x20" * n + b"\xf0\x90\x80\x80"),
"11-in-32-bit character": lambda n: _raw(
b"\xc4\x80" * n + b"\xf0\x90\x80\x80",
),
"16-in-32-bit character": lambda n: _raw(
b"\xe0\xa0\x80" * n + b"\xf0\x90\x80\x80",
),
# encodings
"utf-8-sig character": lambda n: b"\xef\xbb\xbf" + _raw(b"\x20" * n),
"utf-16le character": lambda n: _raw(b"\x00\x01" * n, "utf-16le"),
"utf-16be character": lambda n: _raw(b"\x01\x00" * n, "utf-16be"),
"utf-32le character": lambda n: _raw(b"\x00\x00\x01\x00" * n, "utf-32le"),
"utf-32be character": lambda n: _raw(b"\x00\x01\x00\x00" * n, "utf-32be"),
# escapes
"double quote escape": lambda n: _raw(rb"\"" * n),
"backslash escape": lambda n: _raw(rb"\\" * n),
"character escape": lambda n: _raw(rb"\b" * n),
"7-bit escape": lambda n: _raw(rb"\u0000" * n),
"8-bit escape": lambda n: _raw(rb"\u0080" * n),
"11-bit escape": lambda n: _raw(rb"\u0100" * n),
"16-bit escape": lambda n: _raw(rb"\u0800" * n),
"21-bit escape": lambda n: _raw(rb"\ud800\udc00" * n),
# constants
"null": lambda n: _raw([b"null"] * n),
"boolean": lambda n: _raw([b"false"] * n),
# strings
"empty string": lambda n: _raw([b'""'] * n),
# floats
"fixed-point float": lambda n: _raw([b"0.0"] * n),
"scientific float": lambda n: _raw([b"1e-05"] * n),
"minimum float": lambda n: _raw([b"5e-324"] * n),
"<minimum float": lambda n: _raw([b"2e-324"] * n),
"maximum float": lambda n: _raw([b"1.7976931348623157e+308"] * n),
">maximum float": lambda n: _raw([b"1.7976931348623159e+308"] * n),
# integers
"32-bit (+) integer": lambda n: _raw([b"1"] * n),
"32-bit (-) integer": lambda n: _raw([b"-1"] * n),
"64-bit (+) integer": lambda n: _raw([b"4294967296"] * n),
"64-bit (-) integer": lambda n: _raw([b"-4294967296"] * n),
">64-bit (+) integer": lambda n: _raw([b"18446744073709551616"] * n),
">64-bit (-) integer": lambda n: _raw([b"-18446744073709551616"] * n),
# lists and dictionaries
"empty list": lambda n: _raw([b"[]"] * n),
"empty dictionary": lambda n: _raw([b"{}"] * n),
}
def _make_dumpb(func: _Func) -> _Func:
return lambda obj: func(obj).encode()
def _make_loadb(func: _Func, encodings: set[str]) -> _Func:
def loadb(s: Any) -> Any:
encoding: str = jsonyx.detect_encoding(s) # type: ignore
if encoding not in encodings:
s = s.decode(encoding)
return func(s)
return loadb
_ENCODE_FUNCS: dict[str, _Func] = {
"json": _make_dumpb(json.JSONEncoder().encode),
"jsonyx": _make_dumpb(jsonyx.Encoder().dumps),
"pyjsonyx": _make_dumpb(pyjsonyx.Encoder().dumps),
# pylint: disable-next=I1101
"fastsimdjson": fastsimdjson.dumpb, # type: ignore
"msgspec": msgspec.json.Encoder().encode,
# pylint: disable-next=E1101
"orjson": orjson.dumps,
}
_DECODE_FUNCS: dict[str, _Func] = {
"json": json.loads,
"jsonyx": jsonyx.Decoder().loads,
"pyjsonyx": pyjsonyx.Decoder().loads,
"fastsimdjson": _make_loadb(
# pylint: disable-next=I1101
fastsimdjson.loads, {"utf-8", "utf-8-sig"}, # type: ignore
),
"msgspec": _make_loadb(msgspec.json.Decoder().decode, {"utf-8"}),
# pylint: disable-next=E1101
"orjson": _make_loadb(orjson.loads, {"utf-8"}),
}
def _run_benchmark(
cases: dict[str, _Factory], funcs: dict[str, _Func],
) -> None:
rows: list[list[str]] = []
speedups: list[float] = []
for case, make_obj in cases.items():
times: dict[str, float] = {}
for lib, func in funcs.items():
print(end=".", flush=True)
try:
n: int = 1
while True:
if (timing := timeit(
partial(func, make_obj(n)), number=1,
)) >= 0.04 or n >= 2 ** 26:
times[lib] = min(timing, *[
timeit(partial(func, make_obj(n)), number=1)
for _ in range(4)
]) / n
break
n *= 2
except (ValueError, TypeError):
times[lib] = inf
speedups.append(times["pyjsonyx"] / times["jsonyx"])
del times["pyjsonyx"]
fastest_time: float = min(times.values())
row: list[str] = [case]
row.extend(f"{time / fastest_time:.02f}x" for time in times.values())
row.append(f"{1e9 * fastest_time:.02f} ns")
rows.append(row)
del funcs["pyjsonyx"]
headers: list[str] = ["case", *funcs.keys(), "fastest\xa0time"]
colalign: list[str] = ["left"] + ["right"] * (len(funcs) + 1)
print()
print(tabulate(rows, headers, "rst", colalign=colalign))
print(f"max speedup: {max(speedups):.02f}x")
if __name__ == "__main__":
print(f"Python {sys.version}")
_run_benchmark(_ENCODE_CASES, _ENCODE_FUNCS)
_run_benchmark(_DECODE_CASES, _DECODE_FUNCS)