1
0
Fork 0
mirror of https://github.com/denoland/deno.git synced 2024-12-27 01:29:14 -05:00
denoland-deno/cli/tests/node_compat/test/parallel/test-buffer-indexof.js

646 lines
24 KiB
JavaScript

// deno-fmt-ignore-file
// deno-lint-ignore-file
// Copyright Joyent and Node contributors. All rights reserved. MIT license.
// Taken from Node 18.12.1
// This file is automatically generated by `tools/node_compat/setup.ts`. Do not modify this file manually.
'use strict';
const common = require('../common');
const assert = require('assert');
const b = Buffer.from('abcdef');
const buf_a = Buffer.from('a');
const buf_bc = Buffer.from('bc');
const buf_f = Buffer.from('f');
const buf_z = Buffer.from('z');
const buf_empty = Buffer.from('');
const s = 'abcdef';
assert.strictEqual(b.indexOf('a'), 0);
assert.strictEqual(b.indexOf('a', 1), -1);
assert.strictEqual(b.indexOf('a', -1), -1);
assert.strictEqual(b.indexOf('a', -4), -1);
assert.strictEqual(b.indexOf('a', -b.length), 0);
assert.strictEqual(b.indexOf('a', NaN), 0);
assert.strictEqual(b.indexOf('a', -Infinity), 0);
assert.strictEqual(b.indexOf('a', Infinity), -1);
assert.strictEqual(b.indexOf('bc'), 1);
assert.strictEqual(b.indexOf('bc', 2), -1);
assert.strictEqual(b.indexOf('bc', -1), -1);
assert.strictEqual(b.indexOf('bc', -3), -1);
assert.strictEqual(b.indexOf('bc', -5), 1);
assert.strictEqual(b.indexOf('bc', NaN), 1);
assert.strictEqual(b.indexOf('bc', -Infinity), 1);
assert.strictEqual(b.indexOf('bc', Infinity), -1);
assert.strictEqual(b.indexOf('f'), b.length - 1);
assert.strictEqual(b.indexOf('z'), -1);
assert.strictEqual(b.indexOf(''), 0);
assert.strictEqual(b.indexOf('', 1), 1);
assert.strictEqual(b.indexOf('', b.length + 1), b.length);
assert.strictEqual(b.indexOf('', Infinity), b.length);
assert.strictEqual(b.indexOf(buf_a), 0);
assert.strictEqual(b.indexOf(buf_a, 1), -1);
assert.strictEqual(b.indexOf(buf_a, -1), -1);
assert.strictEqual(b.indexOf(buf_a, -4), -1);
assert.strictEqual(b.indexOf(buf_a, -b.length), 0);
assert.strictEqual(b.indexOf(buf_a, NaN), 0);
assert.strictEqual(b.indexOf(buf_a, -Infinity), 0);
assert.strictEqual(b.indexOf(buf_a, Infinity), -1);
assert.strictEqual(b.indexOf(buf_bc), 1);
assert.strictEqual(b.indexOf(buf_bc, 2), -1);
assert.strictEqual(b.indexOf(buf_bc, -1), -1);
assert.strictEqual(b.indexOf(buf_bc, -3), -1);
assert.strictEqual(b.indexOf(buf_bc, -5), 1);
assert.strictEqual(b.indexOf(buf_bc, NaN), 1);
assert.strictEqual(b.indexOf(buf_bc, -Infinity), 1);
assert.strictEqual(b.indexOf(buf_bc, Infinity), -1);
assert.strictEqual(b.indexOf(buf_f), b.length - 1);
assert.strictEqual(b.indexOf(buf_z), -1);
assert.strictEqual(b.indexOf(buf_empty), 0);
assert.strictEqual(b.indexOf(buf_empty, 1), 1);
assert.strictEqual(b.indexOf(buf_empty, b.length + 1), b.length);
assert.strictEqual(b.indexOf(buf_empty, Infinity), b.length);
assert.strictEqual(b.indexOf(0x61), 0);
assert.strictEqual(b.indexOf(0x61, 1), -1);
assert.strictEqual(b.indexOf(0x61, -1), -1);
assert.strictEqual(b.indexOf(0x61, -4), -1);
assert.strictEqual(b.indexOf(0x61, -b.length), 0);
assert.strictEqual(b.indexOf(0x61, NaN), 0);
assert.strictEqual(b.indexOf(0x61, -Infinity), 0);
assert.strictEqual(b.indexOf(0x61, Infinity), -1);
assert.strictEqual(b.indexOf(0x0), -1);
// test offsets
assert.strictEqual(b.indexOf('d', 2), 3);
assert.strictEqual(b.indexOf('f', 5), 5);
assert.strictEqual(b.indexOf('f', -1), 5);
assert.strictEqual(b.indexOf('f', 6), -1);
assert.strictEqual(b.indexOf(Buffer.from('d'), 2), 3);
assert.strictEqual(b.indexOf(Buffer.from('f'), 5), 5);
assert.strictEqual(b.indexOf(Buffer.from('f'), -1), 5);
assert.strictEqual(b.indexOf(Buffer.from('f'), 6), -1);
// TODO(Soremwar)
// Enable again once encoding is taking into account when evaluating indexOf
// assert.strictEqual(Buffer.from('ff').indexOf(Buffer.from('f'), 1, 'ucs2'), -1);
// Test invalid and uppercase encoding
assert.strictEqual(b.indexOf('b', 'utf8'), 1);
assert.strictEqual(b.indexOf('b', 'UTF8'), 1);
assert.strictEqual(b.indexOf('62', 'HEX'), 1);
assert.throws(() => b.indexOf('bad', 'enc'), /Unknown encoding: enc/);
// test hex encoding
assert.strictEqual(
Buffer.from(b.toString('hex'), 'hex')
.indexOf('64', 0, 'hex'),
3
);
assert.strictEqual(
Buffer.from(b.toString('hex'), 'hex')
.indexOf(Buffer.from('64', 'hex'), 0, 'hex'),
3
);
// Test base64 encoding
assert.strictEqual(
Buffer.from(b.toString('base64'), 'base64')
.indexOf('ZA==', 0, 'base64'),
3
);
assert.strictEqual(
Buffer.from(b.toString('base64'), 'base64')
.indexOf(Buffer.from('ZA==', 'base64'), 0, 'base64'),
3
);
// Test base64url encoding
assert.strictEqual(
Buffer.from(b.toString('base64url'), 'base64url')
.indexOf('ZA==', 0, 'base64url'),
3
);
// test ascii encoding
assert.strictEqual(
Buffer.from(b.toString('ascii'), 'ascii')
.indexOf('d', 0, 'ascii'),
3
);
assert.strictEqual(
Buffer.from(b.toString('ascii'), 'ascii')
.indexOf(Buffer.from('d', 'ascii'), 0, 'ascii'),
3
);
// Test latin1 encoding
assert.strictEqual(
Buffer.from(b.toString('latin1'), 'latin1')
.indexOf('d', 0, 'latin1'),
3
);
assert.strictEqual(
Buffer.from(b.toString('latin1'), 'latin1')
.indexOf(Buffer.from('d', 'latin1'), 0, 'latin1'),
3
);
assert.strictEqual(
Buffer.from('aa\u00e8aa', 'latin1')
.indexOf('\u00e8', 'latin1'),
2
);
assert.strictEqual(
Buffer.from('\u00e8', 'latin1')
.indexOf('\u00e8', 'latin1'),
0
);
assert.strictEqual(
Buffer.from('\u00e8', 'latin1')
.indexOf(Buffer.from('\u00e8', 'latin1'), 'latin1'),
0
);
// Test binary encoding
assert.strictEqual(
Buffer.from(b.toString('binary'), 'binary')
.indexOf('d', 0, 'binary'),
3
);
assert.strictEqual(
Buffer.from(b.toString('binary'), 'binary')
.indexOf(Buffer.from('d', 'binary'), 0, 'binary'),
3
);
assert.strictEqual(
Buffer.from('aa\u00e8aa', 'binary')
.indexOf('\u00e8', 'binary'),
2
);
assert.strictEqual(
Buffer.from('\u00e8', 'binary')
.indexOf('\u00e8', 'binary'),
0
);
assert.strictEqual(
Buffer.from('\u00e8', 'binary')
.indexOf(Buffer.from('\u00e8', 'binary'), 'binary'),
0
);
// Test optional offset with passed encoding
assert.strictEqual(Buffer.from('aaaa0').indexOf('30', 'hex'), 4);
assert.strictEqual(Buffer.from('aaaa00a').indexOf('3030', 'hex'), 4);
{
// Test usc2 and utf16le encoding
['ucs2', 'utf16le'].forEach((encoding) => {
const twoByteString = Buffer.from(
'\u039a\u0391\u03a3\u03a3\u0395', encoding);
assert.strictEqual(twoByteString.indexOf('\u0395', 4, encoding), 8);
assert.strictEqual(twoByteString.indexOf('\u03a3', -4, encoding), 6);
assert.strictEqual(twoByteString.indexOf('\u03a3', -6, encoding), 4);
assert.strictEqual(twoByteString.indexOf(
Buffer.from('\u03a3', encoding), -6, encoding), 4);
assert.strictEqual(-1, twoByteString.indexOf('\u03a3', -2, encoding));
});
}
const mixedByteStringUcs2 =
Buffer.from('\u039a\u0391abc\u03a3\u03a3\u0395', 'ucs2');
assert.strictEqual(mixedByteStringUcs2.indexOf('bc', 0, 'ucs2'), 6);
assert.strictEqual(mixedByteStringUcs2.indexOf('\u03a3', 0, 'ucs2'), 10);
assert.strictEqual(-1, mixedByteStringUcs2.indexOf('\u0396', 0, 'ucs2'));
assert.strictEqual(
mixedByteStringUcs2.indexOf(Buffer.from('bc', 'ucs2'), 0, 'ucs2'), 6);
assert.strictEqual(
mixedByteStringUcs2.indexOf(Buffer.from('\u03a3', 'ucs2'), 0, 'ucs2'), 10);
assert.strictEqual(
-1, mixedByteStringUcs2.indexOf(Buffer.from('\u0396', 'ucs2'), 0, 'ucs2'));
{
const twoByteString = Buffer.from('\u039a\u0391\u03a3\u03a3\u0395', 'ucs2');
// Test single char pattern
assert.strictEqual(twoByteString.indexOf('\u039a', 0, 'ucs2'), 0);
let index = twoByteString.indexOf('\u0391', 0, 'ucs2');
assert.strictEqual(index, 2, `Alpha - at index ${index}`);
index = twoByteString.indexOf('\u03a3', 0, 'ucs2');
assert.strictEqual(index, 4, `First Sigma - at index ${index}`);
index = twoByteString.indexOf('\u03a3', 6, 'ucs2');
assert.strictEqual(index, 6, `Second Sigma - at index ${index}`);
index = twoByteString.indexOf('\u0395', 0, 'ucs2');
assert.strictEqual(index, 8, `Epsilon - at index ${index}`);
index = twoByteString.indexOf('\u0392', 0, 'ucs2');
assert.strictEqual(-1, index, `Not beta - at index ${index}`);
// Test multi-char pattern
index = twoByteString.indexOf('\u039a\u0391', 0, 'ucs2');
assert.strictEqual(index, 0, `Lambda Alpha - at index ${index}`);
index = twoByteString.indexOf('\u0391\u03a3', 0, 'ucs2');
assert.strictEqual(index, 2, `Alpha Sigma - at index ${index}`);
index = twoByteString.indexOf('\u03a3\u03a3', 0, 'ucs2');
assert.strictEqual(index, 4, `Sigma Sigma - at index ${index}`);
index = twoByteString.indexOf('\u03a3\u0395', 0, 'ucs2');
assert.strictEqual(index, 6, `Sigma Epsilon - at index ${index}`);
}
const mixedByteStringUtf8 = Buffer.from('\u039a\u0391abc\u03a3\u03a3\u0395');
assert.strictEqual(mixedByteStringUtf8.indexOf('bc'), 5);
assert.strictEqual(mixedByteStringUtf8.indexOf('bc', 5), 5);
assert.strictEqual(mixedByteStringUtf8.indexOf('bc', -8), 5);
assert.strictEqual(mixedByteStringUtf8.indexOf('\u03a3'), 7);
assert.strictEqual(mixedByteStringUtf8.indexOf('\u0396'), -1);
// Test complex string indexOf algorithms. Only trigger for long strings.
// Long string that isn't a simple repeat of a shorter string.
let longString = 'A';
for (let i = 66; i < 76; i++) { // from 'B' to 'K'
longString = longString + String.fromCharCode(i) + longString;
}
const longBufferString = Buffer.from(longString);
// Pattern of 15 chars, repeated every 16 chars in long
let pattern = 'ABACABADABACABA';
for (let i = 0; i < longBufferString.length - pattern.length; i += 7) {
const index = longBufferString.indexOf(pattern, i);
assert.strictEqual((i + 15) & ~0xf, index,
`Long ABACABA...-string at index ${i}`);
}
let index = longBufferString.indexOf('AJABACA');
assert.strictEqual(index, 510, `Long AJABACA, First J - at index ${index}`);
index = longBufferString.indexOf('AJABACA', 511);
assert.strictEqual(index, 1534, `Long AJABACA, Second J - at index ${index}`);
pattern = 'JABACABADABACABA';
index = longBufferString.indexOf(pattern);
assert.strictEqual(index, 511, `Long JABACABA..., First J - at index ${index}`);
index = longBufferString.indexOf(pattern, 512);
assert.strictEqual(
index, 1535, `Long JABACABA..., Second J - at index ${index}`);
// Search for a non-ASCII string in a pure ASCII string.
const asciiString = Buffer.from(
'arglebargleglopglyfarglebargleglopglyfarglebargleglopglyf');
assert.strictEqual(-1, asciiString.indexOf('\x2061'));
assert.strictEqual(asciiString.indexOf('leb', 0), 3);
// Search in string containing many non-ASCII chars.
const allCodePoints = [];
for (let i = 0; i < 65534; i++) allCodePoints[i] = i;
const allCharsString = String.fromCharCode.apply(String, allCodePoints) +
String.fromCharCode(65534, 65535);
const allCharsBufferUtf8 = Buffer.from(allCharsString);
const allCharsBufferUcs2 = Buffer.from(allCharsString, 'ucs2');
// Search for string long enough to trigger complex search with ASCII pattern
// and UC16 subject.
assert.strictEqual(-1, allCharsBufferUtf8.indexOf('notfound'));
assert.strictEqual(-1, allCharsBufferUcs2.indexOf('notfound'));
// Needle is longer than haystack, but only because it's encoded as UTF-16
assert.strictEqual(Buffer.from('aaaa').indexOf('a'.repeat(4), 'ucs2'), -1);
assert.strictEqual(Buffer.from('aaaa').indexOf('a'.repeat(4), 'utf8'), 0);
assert.strictEqual(Buffer.from('aaaa').indexOf('你好', 'ucs2'), -1);
// Haystack has odd length, but the needle is UCS2.
assert.strictEqual(Buffer.from('aaaaa').indexOf('b', 'ucs2'), -1);
{
// Find substrings in Utf8.
const lengths = [1, 3, 15]; // Single char, simple and complex.
const indices = [0x5, 0x60, 0x400, 0x680, 0x7ee, 0xFF02, 0x16610, 0x2f77b];
for (let lengthIndex = 0; lengthIndex < lengths.length; lengthIndex++) {
for (let i = 0; i < indices.length; i++) {
const index = indices[i];
let length = lengths[lengthIndex];
if (index + length > 0x7F) {
length = 2 * length;
}
if (index + length > 0x7FF) {
length = 3 * length;
}
if (index + length > 0xFFFF) {
length = 4 * length;
}
const patternBufferUtf8 = allCharsBufferUtf8.slice(index, index + length);
assert.strictEqual(index, allCharsBufferUtf8.indexOf(patternBufferUtf8));
const patternStringUtf8 = patternBufferUtf8.toString();
assert.strictEqual(index, allCharsBufferUtf8.indexOf(patternStringUtf8));
}
}
}
// TODO(Soremwar)
// Enable again once encoding is taking into account when evaluating indexOf
// {
// // Find substrings in Usc2.
// const lengths = [2, 4, 16]; // Single char, simple and complex.
// const indices = [0x5, 0x65, 0x105, 0x205, 0x285, 0x2005, 0x2085, 0xfff0];
// for (let lengthIndex = 0; lengthIndex < lengths.length; lengthIndex++) {
// for (let i = 0; i < indices.length; i++) {
// const index = indices[i] * 2;
// const length = lengths[lengthIndex];
// const patternBufferUcs2 =
// allCharsBufferUcs2.slice(index, index + length);
// assert.strictEqual(
// index, allCharsBufferUcs2.indexOf(patternBufferUcs2, 0, 'ucs2'));
// const patternStringUcs2 = patternBufferUcs2.toString('ucs2');
// assert.strictEqual(
// index, allCharsBufferUcs2.indexOf(patternStringUcs2, 0, 'ucs2'));
// }
// }
// }
[
() => {},
{},
[],
].forEach((val) => {
assert.throws(
() => b.indexOf(val),
{
code: 'ERR_INVALID_ARG_TYPE',
name: 'TypeError',
message: 'The "value" argument must be one of type number or string ' +
'or an instance of Buffer or Uint8Array.' +
common.invalidArgTypeHelper(val)
}
);
});
// Test weird offset arguments.
// The following offsets coerce to NaN or 0, searching the whole Buffer
assert.strictEqual(b.indexOf('b', undefined), 1);
assert.strictEqual(b.indexOf('b', {}), 1);
assert.strictEqual(b.indexOf('b', 0), 1);
assert.strictEqual(b.indexOf('b', null), 1);
assert.strictEqual(b.indexOf('b', []), 1);
// The following offset coerces to 2, in other words +[2] === 2
assert.strictEqual(b.indexOf('b', [2]), -1);
// Behavior should match String.indexOf()
assert.strictEqual(
b.indexOf('b', undefined),
s.indexOf('b', undefined));
assert.strictEqual(
b.indexOf('b', {}),
s.indexOf('b', {}));
assert.strictEqual(
b.indexOf('b', 0),
s.indexOf('b', 0));
assert.strictEqual(
b.indexOf('b', null),
s.indexOf('b', null));
assert.strictEqual(
b.indexOf('b', []),
s.indexOf('b', []));
assert.strictEqual(
b.indexOf('b', [2]),
s.indexOf('b', [2]));
// All code for handling encodings is shared between Buffer.indexOf and
// Buffer.lastIndexOf, so only testing the separate lastIndexOf semantics.
// Test lastIndexOf basic functionality; Buffer b contains 'abcdef'.
// lastIndexOf string:
assert.strictEqual(b.lastIndexOf('a'), 0);
assert.strictEqual(b.lastIndexOf('a', 1), 0);
assert.strictEqual(b.lastIndexOf('b', 1), 1);
assert.strictEqual(b.lastIndexOf('c', 1), -1);
assert.strictEqual(b.lastIndexOf('a', -1), 0);
assert.strictEqual(b.lastIndexOf('a', -4), 0);
assert.strictEqual(b.lastIndexOf('a', -b.length), 0);
assert.strictEqual(b.lastIndexOf('a', -b.length - 1), -1);
assert.strictEqual(b.lastIndexOf('a', NaN), 0);
assert.strictEqual(b.lastIndexOf('a', -Infinity), -1);
assert.strictEqual(b.lastIndexOf('a', Infinity), 0);
// lastIndexOf Buffer:
assert.strictEqual(b.lastIndexOf(buf_a), 0);
assert.strictEqual(b.lastIndexOf(buf_a, 1), 0);
assert.strictEqual(b.lastIndexOf(buf_a, -1), 0);
assert.strictEqual(b.lastIndexOf(buf_a, -4), 0);
assert.strictEqual(b.lastIndexOf(buf_a, -b.length), 0);
assert.strictEqual(b.lastIndexOf(buf_a, -b.length - 1), -1);
assert.strictEqual(b.lastIndexOf(buf_a, NaN), 0);
assert.strictEqual(b.lastIndexOf(buf_a, -Infinity), -1);
assert.strictEqual(b.lastIndexOf(buf_a, Infinity), 0);
assert.strictEqual(b.lastIndexOf(buf_bc), 1);
assert.strictEqual(b.lastIndexOf(buf_bc, 2), 1);
assert.strictEqual(b.lastIndexOf(buf_bc, -1), 1);
assert.strictEqual(b.lastIndexOf(buf_bc, -3), 1);
assert.strictEqual(b.lastIndexOf(buf_bc, -5), 1);
assert.strictEqual(b.lastIndexOf(buf_bc, -6), -1);
assert.strictEqual(b.lastIndexOf(buf_bc, NaN), 1);
assert.strictEqual(b.lastIndexOf(buf_bc, -Infinity), -1);
assert.strictEqual(b.lastIndexOf(buf_bc, Infinity), 1);
assert.strictEqual(b.lastIndexOf(buf_f), b.length - 1);
assert.strictEqual(b.lastIndexOf(buf_z), -1);
assert.strictEqual(b.lastIndexOf(buf_empty), b.length);
assert.strictEqual(b.lastIndexOf(buf_empty, 1), 1);
assert.strictEqual(b.lastIndexOf(buf_empty, b.length + 1), b.length);
assert.strictEqual(b.lastIndexOf(buf_empty, Infinity), b.length);
// lastIndexOf number:
assert.strictEqual(b.lastIndexOf(0x61), 0);
assert.strictEqual(b.lastIndexOf(0x61, 1), 0);
assert.strictEqual(b.lastIndexOf(0x61, -1), 0);
assert.strictEqual(b.lastIndexOf(0x61, -4), 0);
assert.strictEqual(b.lastIndexOf(0x61, -b.length), 0);
assert.strictEqual(b.lastIndexOf(0x61, -b.length - 1), -1);
assert.strictEqual(b.lastIndexOf(0x61, NaN), 0);
assert.strictEqual(b.lastIndexOf(0x61, -Infinity), -1);
assert.strictEqual(b.lastIndexOf(0x61, Infinity), 0);
assert.strictEqual(b.lastIndexOf(0x0), -1);
// Test weird offset arguments.
// The following offsets coerce to NaN, searching the whole Buffer
assert.strictEqual(b.lastIndexOf('b', undefined), 1);
assert.strictEqual(b.lastIndexOf('b', {}), 1);
// The following offsets coerce to 0
assert.strictEqual(b.lastIndexOf('b', 0), -1);
assert.strictEqual(b.lastIndexOf('b', null), -1);
assert.strictEqual(b.lastIndexOf('b', []), -1);
// The following offset coerces to 2, in other words +[2] === 2
assert.strictEqual(b.lastIndexOf('b', [2]), 1);
// Behavior should match String.lastIndexOf()
assert.strictEqual(
b.lastIndexOf('b', undefined),
s.lastIndexOf('b', undefined));
assert.strictEqual(
b.lastIndexOf('b', {}),
s.lastIndexOf('b', {}));
assert.strictEqual(
b.lastIndexOf('b', 0),
s.lastIndexOf('b', 0));
assert.strictEqual(
b.lastIndexOf('b', null),
s.lastIndexOf('b', null));
assert.strictEqual(
b.lastIndexOf('b', []),
s.lastIndexOf('b', []));
assert.strictEqual(
b.lastIndexOf('b', [2]),
s.lastIndexOf('b', [2]));
// Test needles longer than the haystack.
assert.strictEqual(b.lastIndexOf('aaaaaaaaaaaaaaa', 'ucs2'), -1);
assert.strictEqual(b.lastIndexOf('aaaaaaaaaaaaaaa', 'utf8'), -1);
assert.strictEqual(b.lastIndexOf('aaaaaaaaaaaaaaa', 'latin1'), -1);
assert.strictEqual(b.lastIndexOf('aaaaaaaaaaaaaaa', 'binary'), -1);
assert.strictEqual(b.lastIndexOf(Buffer.from('aaaaaaaaaaaaaaa')), -1);
assert.strictEqual(b.lastIndexOf('aaaaaaaaaaaaaaa', 2, 'ucs2'), -1);
assert.strictEqual(b.lastIndexOf('aaaaaaaaaaaaaaa', 3, 'utf8'), -1);
assert.strictEqual(b.lastIndexOf('aaaaaaaaaaaaaaa', 5, 'latin1'), -1);
assert.strictEqual(b.lastIndexOf('aaaaaaaaaaaaaaa', 5, 'binary'), -1);
assert.strictEqual(b.lastIndexOf(Buffer.from('aaaaaaaaaaaaaaa'), 7), -1);
// 你好 expands to a total of 6 bytes using UTF-8 and 4 bytes using UTF-16
assert.strictEqual(buf_bc.lastIndexOf('你好', 'ucs2'), -1);
assert.strictEqual(buf_bc.lastIndexOf('你好', 'utf8'), -1);
assert.strictEqual(buf_bc.lastIndexOf('你好', 'latin1'), -1);
assert.strictEqual(buf_bc.lastIndexOf('你好', 'binary'), -1);
assert.strictEqual(buf_bc.lastIndexOf(Buffer.from('你好')), -1);
assert.strictEqual(buf_bc.lastIndexOf('你好', 2, 'ucs2'), -1);
assert.strictEqual(buf_bc.lastIndexOf('你好', 3, 'utf8'), -1);
assert.strictEqual(buf_bc.lastIndexOf('你好', 5, 'latin1'), -1);
assert.strictEqual(buf_bc.lastIndexOf('你好', 5, 'binary'), -1);
assert.strictEqual(buf_bc.lastIndexOf(Buffer.from('你好'), 7), -1);
// Test lastIndexOf on a longer buffer:
const bufferString = Buffer.from('a man a plan a canal panama');
assert.strictEqual(bufferString.lastIndexOf('canal'), 15);
assert.strictEqual(bufferString.lastIndexOf('panama'), 21);
assert.strictEqual(bufferString.lastIndexOf('a man a plan a canal panama'), 0);
assert.strictEqual(-1, bufferString.lastIndexOf('a man a plan a canal mexico'));
assert.strictEqual(-1, bufferString
.lastIndexOf('a man a plan a canal mexico city'));
assert.strictEqual(-1, bufferString.lastIndexOf(Buffer.from('a'.repeat(1000))));
assert.strictEqual(bufferString.lastIndexOf('a man a plan', 4), 0);
assert.strictEqual(bufferString.lastIndexOf('a '), 13);
assert.strictEqual(bufferString.lastIndexOf('a ', 13), 13);
assert.strictEqual(bufferString.lastIndexOf('a ', 12), 6);
assert.strictEqual(bufferString.lastIndexOf('a ', 5), 0);
assert.strictEqual(bufferString.lastIndexOf('a ', -1), 13);
assert.strictEqual(bufferString.lastIndexOf('a ', -27), 0);
assert.strictEqual(-1, bufferString.lastIndexOf('a ', -28));
// Test lastIndexOf for the case that the first character can be found,
// but in a part of the buffer that does not make search to search
// due do length constraints.
const abInUCS2 = Buffer.from('ab', 'ucs2');
assert.strictEqual(-1, Buffer.from('µaaaa¶bbbb', 'latin1').lastIndexOf('µ'));
assert.strictEqual(-1, Buffer.from('µaaaa¶bbbb', 'binary').lastIndexOf('µ'));
assert.strictEqual(-1, Buffer.from('bc').lastIndexOf('ab'));
assert.strictEqual(-1, Buffer.from('abc').lastIndexOf('qa'));
assert.strictEqual(-1, Buffer.from('abcdef').lastIndexOf('qabc'));
assert.strictEqual(-1, Buffer.from('bc').lastIndexOf(Buffer.from('ab')));
assert.strictEqual(-1, Buffer.from('bc', 'ucs2').lastIndexOf('ab', 'ucs2'));
assert.strictEqual(-1, Buffer.from('bc', 'ucs2').lastIndexOf(abInUCS2));
assert.strictEqual(Buffer.from('abc').lastIndexOf('ab'), 0);
assert.strictEqual(Buffer.from('abc').lastIndexOf('ab', 1), 0);
assert.strictEqual(Buffer.from('abc').lastIndexOf('ab', 2), 0);
assert.strictEqual(Buffer.from('abc').lastIndexOf('ab', 3), 0);
// The above tests test the LINEAR and SINGLE-CHAR strategies.
// Now, we test the BOYER-MOORE-HORSPOOL strategy.
// Test lastIndexOf on a long buffer w multiple matches:
pattern = 'JABACABADABACABA';
assert.strictEqual(longBufferString.lastIndexOf(pattern), 1535);
assert.strictEqual(longBufferString.lastIndexOf(pattern, 1535), 1535);
assert.strictEqual(longBufferString.lastIndexOf(pattern, 1534), 511);
// Finally, give it a really long input to trigger fallback from BMH to
// regular BOYER-MOORE (which has better worst-case complexity).
// Generate a really long Thue-Morse sequence of 'yolo' and 'swag',
// "yolo swag swag yolo swag yolo yolo swag" ..., goes on for about 5MB.
// This is hard to search because it all looks similar, but never repeats.
// countBits returns the number of bits in the binary representation of n.
function countBits(n) {
let count;
for (count = 0; n > 0; count++) {
n = n & (n - 1); // remove top bit
}
return count;
}
const parts = [];
for (let i = 0; i < 1000000; i++) {
parts.push((countBits(i) % 2 === 0) ? 'yolo' : 'swag');
}
const reallyLong = Buffer.from(parts.join(' '));
assert.strictEqual(reallyLong.slice(0, 19).toString(), 'yolo swag swag yolo');
// Expensive reverse searches. Stress test lastIndexOf:
pattern = reallyLong.slice(0, 100000); // First 1/50th of the pattern.
assert.strictEqual(reallyLong.lastIndexOf(pattern), 4751360);
assert.strictEqual(reallyLong.lastIndexOf(pattern, 4000000), 3932160);
assert.strictEqual(reallyLong.lastIndexOf(pattern, 3000000), 2949120);
pattern = reallyLong.slice(100000, 200000); // Second 1/50th.
assert.strictEqual(reallyLong.lastIndexOf(pattern), 4728480);
pattern = reallyLong.slice(0, 1000000); // First 1/5th.
assert.strictEqual(reallyLong.lastIndexOf(pattern), 3932160);
pattern = reallyLong.slice(0, 2000000); // first 2/5ths.
assert.strictEqual(reallyLong.lastIndexOf(pattern), 0);
// Test truncation of Number arguments to uint8
// TODO(Soremwar)
// Enable once multi byte number search is available
// {
// const buf = Buffer.from('this is a test');
// assert.strictEqual(buf.indexOf(0x6973), 3);
// assert.strictEqual(buf.indexOf(0x697320), 4);
// assert.strictEqual(buf.indexOf(0x69732069), 2);
// assert.strictEqual(buf.indexOf(0x697374657374), 0);
// assert.strictEqual(buf.indexOf(0x69737374), 0);
// assert.strictEqual(buf.indexOf(0x69737465), 11);
// assert.strictEqual(buf.indexOf(0x69737465), 11);
// assert.strictEqual(buf.indexOf(-140), 0);
// assert.strictEqual(buf.indexOf(-152), 1);
// assert.strictEqual(buf.indexOf(0xff), -1);
// assert.strictEqual(buf.indexOf(0xffff), -1);
// }
// Test that Uint8Array arguments are okay.
{
const needle = new Uint8Array([ 0x66, 0x6f, 0x6f ]);
const haystack = Buffer.from('a foo b foo');
assert.strictEqual(haystack.indexOf(needle), 2);
assert.strictEqual(haystack.lastIndexOf(needle), haystack.length - 3);
}
// Avoid abort because of invalid usage
// see https://github.com/nodejs/node/issues/32753
{
assert.throws(() => {
const buffer = require('buffer');
new buffer.Buffer.prototype.lastIndexOf(1, 'str');
}, {
code: 'ERR_INVALID_ARG_TYPE',
name: 'TypeError',
message: 'The "buffer" argument must be an instance of Buffer, ' +
'TypedArray, or DataView. ' +
'Received an instance of lastIndexOf'
});
}