1
// Copyright (c) 2025 Cloudflare, Inc.
2
// Licensed under the Apache 2.0 license found in the LICENSE file or at:
3
//     https://opensource.org/licenses/Apache-2.0
4
 
5
#include "encoding.h"
6
 
7
#include <kj/test.h>
8
 
9
namespace workerd::api {
10
namespace test {
11
 
12
// These tests verify the findBestFit() function used by TextEncoder.encodeInto().
13
//
14
// bestFit(input, bufferSize) returns the number of input code units that can be
15
// fully converted to UTF-8 and fit within the given output buffer size in bytes.
16
//
17
// Different characters expand to different UTF-8 byte lengths:
18
//   - ASCII (U+0000-U+007F): 1 byte per code unit
19
//   - Latin-1 extended (U+0080-U+00FF): 2 bytes per code unit
20
//   - BMP characters (U+0100-U+FFFF): 2-3 bytes per code unit
21
//   - Supplementary characters (U+10000+): 4 bytes, encoded as surrogate pairs in UTF-16
22
//
23
// The function must never split a surrogate pair, so if there's only room for part of
24
// a multi-byte character, it stops before that character.
25
KJ_TEST("BestFitASCII") {
26
  // If there's zero input or output space, the answer is zero.
27
  KJ_ASSERT(bestFit("", 0) == 0);
28
  KJ_ASSERT(bestFit("a", 0) == 0);
29
  KJ_ASSERT(bestFit("aa", 0) == 0);
30
  KJ_ASSERT(bestFit("aaa", 0) == 0);
31
  KJ_ASSERT(bestFit("aaaa", 0) == 0);
32
  KJ_ASSERT(bestFit("aaaaa", 0) == 0);
33
  KJ_ASSERT(bestFit("", 0) == 0);
34
  KJ_ASSERT(bestFit("", 1) == 0);
35
  KJ_ASSERT(bestFit("", 2) == 0);
36
  KJ_ASSERT(bestFit("", 3) == 0);
37
  KJ_ASSERT(bestFit("", 4) == 0);
38
  KJ_ASSERT(bestFit("", 5) == 0);
39
  // Zero cases with two-byte strings.
40
  KJ_ASSERT(bestFit(u"", 0) == 0);
41
  KJ_ASSERT(bestFit(u"€", 0) == 0);
42
  KJ_ASSERT(bestFit(u"€€", 0) == 0);
43
  KJ_ASSERT(bestFit(u"€€€", 0) == 0);
44
  KJ_ASSERT(bestFit(u"€€€€", 0) == 0);
45
  KJ_ASSERT(bestFit(u"€€€€€", 0) == 0);
46
  KJ_ASSERT(bestFit(u"", 0) == 0);
47
  KJ_ASSERT(bestFit(u"", 1) == 0);
48
  KJ_ASSERT(bestFit(u"", 2) == 0);
49
  KJ_ASSERT(bestFit(u"", 3) == 0);
50
  KJ_ASSERT(bestFit(u"", 4) == 0);
51
  KJ_ASSERT(bestFit(u"", 5) == 0);
52
  // Small buffers that only just fit.
53
  KJ_ASSERT(bestFit(u"a", 1) == 1);
54
  KJ_ASSERT(bestFit(u"å", 2) == 1);
55
  KJ_ASSERT(bestFit(u"€", 3) == 1);
56
  KJ_ASSERT(bestFit(u"😹", 4) == 2);
57
  // Small buffers that don't fit.
58
  KJ_ASSERT(bestFit(u"å", 1) == 0);
59
  KJ_ASSERT(bestFit(u"€", 2) == 0);
60
  KJ_ASSERT(bestFit(u"😹", 3) == 0);
61
  // Don't chop a surrogate pair.
62
  KJ_ASSERT(bestFit(u"1😹", 4) == 1);
63
  KJ_ASSERT(bestFit(u"12😹", 5) == 2);
64
  KJ_ASSERT(bestFit(u"123😹", 6) == 3);
65
  KJ_ASSERT(bestFit(u"1234😹", 7) == 4);
66
  KJ_ASSERT(bestFit(u"12345😹", 8) == 5);
67
  // Some bigger ones just for fun.
68
  KJ_ASSERT(bestFit(u"😹😹😹😹😹😹", 0) == 0);
69
  KJ_ASSERT(bestFit(u"😹😹😹😹😹😹", 1) == 0);
70
  KJ_ASSERT(bestFit(u"😹😹😹😹😹😹", 2) == 0);
71
  KJ_ASSERT(bestFit(u"😹😹😹😹😹😹", 3) == 0);
72
  KJ_ASSERT(bestFit(u"😹😹😹😹😹😹", 4) == 2);
73
  KJ_ASSERT(bestFit(u"😹😹😹😹😹😹", 5) == 2);
74
  KJ_ASSERT(bestFit(u"😹😹😹😹😹😹", 6) == 2);
75
  KJ_ASSERT(bestFit(u"😹😹😹😹😹😹", 7) == 2);
76
  KJ_ASSERT(bestFit(u"😹😹😹😹😹😹", 8) == 4);
77
  KJ_ASSERT(bestFit(u"😹😹😹😹😹😹", 9) == 4);
78
  KJ_ASSERT(bestFit(u"0😹😹😹😹😹😹", 9) == 5);          // 0😹😹 is 5 and takes 9.
79
  KJ_ASSERT(bestFit(u"01😹😹😹😹😹😹", 9) == 4);         // 01😹 is 4 and takes 6.
80
  KJ_ASSERT(bestFit(u"012😹😹😹😹😹😹", 9) == 5);        // 012😹 is 5 and takes 7.
81
  KJ_ASSERT(bestFit(u"0123😹😹😹😹😹😹", 9) == 6);       // 0123😹 is 6 and takes 8.
82
  KJ_ASSERT(bestFit(u"01234😹😹😹😹😹😹", 9) == 7);      // 01234😹 is 7 and takes 9.
83
  KJ_ASSERT(bestFit(u"012345😹😹😹😹😹😹", 9) == 6);     // 012345 is 6 and takes 6.
84
  KJ_ASSERT(bestFit(u"0123456😹😹😹😹😹😹", 9) == 7);    // 0123456 is 7 and takes 7.
85
  KJ_ASSERT(bestFit(u"01234567😹😹😹😹😹😹", 9) == 8);   // 0123456 is 8 and takes 8.
86
  KJ_ASSERT(bestFit(u"012345678😹😹😹😹😹😹", 9) == 9);  // 0123456 is 9 and takes 9.
87
}
88
 
89
}  // namespace test
90
}  // namespace workerd::api