notes

4.9 KB

1. // Test-vector corpus generator for the hl:mpackdb Zig MessagePack codec.
2. //
3. // Generates corpus.json from the REAL npm msgpackr (the codec mpackdb v1.0.7
4. // is built on) so the Zig codec is validated against actual bytes, not specs.
5. //
6. // Usage (msgpackr is NOT vendored in this repo — install it in a scratch dir):
7. // mkdir -p /tmp/mpackdb-corpus && cd /tmp/mpackdb-corpus
8. // npm install [email protected]
9. // node /path/to/repo/plugins/mpackdb/testdata/gen_corpus.mjs corpus.json
10. // cp corpus.json /path/to/repo/plugins/mpackdb/testdata/
11. //
12. // The bare specifier is resolved from the CURRENT WORKING DIRECTORY (via
13. // createRequire), so the script itself can live inside the repo while
14. // msgpackr lives in the scratch dir.
15. //
16. // Vector format: { name, hex, framed? }
17. // hex — msgpackr.pack(value) output as hex
18. // framed — value additionally wrapped in mpackdb's mpack.serialize framing:
19. // [4-byte LE total size incl. prefix][msgpack bytes]
20. //
21. // The Zig side (msgpack.zig corpus test) decodes every vector and re-encodes
22. // it, asserting byte identity — msgpackr output is canonical for the subset
23. // mpackdb uses (no records/shared structures, no float32, map16 for objects).
25. import { createRequire } from 'module';
26. import { writeFileSync } from 'fs';
28. const require = createRequire(process.cwd() + '/');
29. const { pack } = require('msgpackr');
31. const out = process.argv[2] || 'corpus.json';
32. const vectors = [];
33. const add = (name, value) => vectors.push({ name, hex: Buffer.from(pack(value)).toString('hex') });
35. // --- integers at every width boundary ---
36. for (const n of [0, 1, 127, 128, 255, 256, 65535, 65536,
37. 2147483647, 2147483648, 4294967295, 4294967296,
38. Number.MAX_SAFE_INTEGER,
39. -1, -32, -33, -128, -129, -32768, -32769,
40. -2147483648, -2147483649, -Number.MAX_SAFE_INTEGER]) {
41. add(`int ${n}`, n);
42. }
43. // --- floats (msgpackr default: always float64, never float32) ---
44. for (const f of [0.5, -0.25, 3.14159, 1e15, 1.5e300, 5e-324, NaN, Infinity, -Infinity, -0]) {
45. add(`float ${f}`, f);
46. }
47. // --- strings at width boundaries (byte lengths incl. multi-byte UTF-8) ---
48. add('str empty', '');
49. add('str 1', 'a');
50. add('str 31', 'x'.repeat(31));
51. add('str 32', 'x'.repeat(32));
52. add('str 255', 'y'.repeat(255));
53. add('str 256', 'y'.repeat(256));
54. add('str 65535', 'z'.repeat(65535));
55. add('str 65536', 'z'.repeat(65536));
56. add('str utf8', 'héllo wörld ✓ 日本語');
57. add('str utf8 32bytes', 'é'.repeat(16)); // 32 UTF-8 bytes → str8, not fixstr
58. // --- booleans / nil / undefined (msgpackr: fixext1 type 0) ---
59. add('true', true);
60. add('false', false);
61. add('null', null);
62. add('undefined', undefined);
63. // --- arrays at width boundaries ---
64. add('arr empty', []);
65. add('arr 15', Array.from({ length: 15 }, (_, i) => i));
66. add('arr 16', Array.from({ length: 16 }, (_, i) => i));
67. add('arr 65535', Array.from({ length: 65535 }, () => 0));
68. add('arr 65536', Array.from({ length: 65536 }, () => 0));
69. add('arr mixed', [1, 'two', 3.5, null, true, [1, 2], { a: 1 }]);
70. // --- maps (msgpackr default: ALWAYS map16 0xde, even empty) ---
71. add('map empty', {});
72. add('map small', { name: 'Alice', age: 30 });
73. add('map numeric-string keys', { 1: 'a', 2: 'b' });
74. const map1000 = {};
75. for (let i = 0; i < 1000; i++) map1000['k' + i] = i;
76. add('map 1000', map1000);
77. add('map nested', { a: { b: { c: [1, 2, { d: 'deep' }] } } });
78. add('map undefined prop', { a: undefined, b: 1 });
79. // --- binary (decode-side coverage; Hybriel has no binary type) ---
80. add('bin 3', Buffer.from([1, 2, 3]));
81. add('bin 255', Buffer.alloc(255, 7));
82. add('bin 256', Buffer.alloc(256, 7));
83. add('bin 65536', Buffer.alloc(65536, 7));
84. // --- Date → msgpack timestamp ext -1 (decode fidelity; kept raw ext in Zig) ---
85. add('date ms', new Date(1721996422123)); // d7 ff (ns+sec 64-bit form)
86. add('date whole sec', new Date(1721996422000)); // d6 ff (32-bit sec form)
87. add('date pre-1970', new Date(-1000)); // c7 0c ff (96-bit form)
88. add('date far future', new Date(4294967296000));// c7 0c ff
90. // --- mpackdb record shapes (as its consumers actually store them) ---
91. add('record user', { id: 0, name: 'Alice Smith', email: '[email protected]', age: 30, active: true, score: 91.25, tags: ['admin', 'staff'], meta: { createdAt: 1721996422123, notes: null } });
92. add('record uuid pk', { uuid: '0kx2mrejh4a7', label: 'thing', x: -5, y: 12 });
93. add('record utxo-ish', { txid: 'f4184fc596403b9d638783cf57adfe4c75c605f6356fbc91338530e9831e9e16', vout: 0, value: 5000000000, height: 170 });
95. // --- framed records: mpack.serialize() = [4-byte LE total size][msgpack] ---
96. const frame = (name, value) => {
97. const m = pack(value);
98. const b = Buffer.allocUnsafe(4 + m.length);
99. b.writeInt32LE(4 + m.length, 0);
100. m.copy(b, 4);
101. vectors.push({ name, hex: b.toString('hex'), framed: true });
102. };
103. frame('framed record', { id: 1, name: 'Bob', balance: 12.5 });
104. frame('framed empty map', {});
106. writeFileSync(out, JSON.stringify(vectors, null, '\t') + '\n');
107. console.log(`wrote ${vectors.length} vectors to ${out}`);