bench-encode-string.php

<?php

class Dumper
{
	/**
	 * @internal
	 * @return string UTF-8
	 */
	public static function encodeString($s, $maxLength = NULL)
	{
		static $table;
		if ($table === NULL) {
			foreach (array_merge(range("\x00", "\x1F"), range("\x7F", "\xFF")) as $ch) {
				$table[$ch] = '\x' . str_pad(dechex(ord($ch)), 2, '0', STR_PAD_LEFT);
			}
			$table['\\'] = '\\\\';
			$table["\r"] = '\r';
			$table["\n"] = '\n';
			$table["\t"] = '\t';
		}

		if (preg_match('#[^\x09\x0A\x0D\x20-\x7E\xA0-\x{10FFFF}]#u', $s) || preg_last_error()) {
			if ($shortened = ($maxLength && strlen($s) > $maxLength)) {
				$s = substr($s, 0, $maxLength);
			}
			$s = strtr($s, $table);

		} elseif ($maxLength && $s !== '') {
			if (function_exists('mb_substr')) {
				$s = mb_substr($tmp = $s, 0, $maxLength, 'UTF-8');
				$shortened = $s !== $tmp;
			} else {
				$i = $len = 0;
				do {
					if (($s[$i] < "\x80" || $s[$i] >= "\xC0") && (++$len > $maxLength)) {
						$s = substr($s, 0, $i);
						$shortened = TRUE;
						break;
					}
				} while (isset($s[++$i]));
			}
		}

		return $s . (empty($shortened) ? '' : ' ... ');
	}


	/**
	 * @internal
	 * @return string UTF-8
	 */
	public static function encodeString2($s, $maxLength = NULL)
	{
		static $table;
		if ($table === NULL) {
			foreach (array_merge(range("\x00", "\x1F"), range("\x7F", "\xFF")) as $ch) {
				$table[$ch] = '\x' . str_pad(dechex(ord($ch)), 2, '0', STR_PAD_LEFT);
			}
			$table['\\'] = '\\\\';
			$table["\r"] = '\r';
			$table["\n"] = '\n';
			$table["\t"] = '\t';
		}

		$shortened = FALSE;
		if ($maxLength && function_exists('mb_substr')) {
			$s = mb_substr($tmp = $s, 0, $maxLength, 'UTF-8');
			$shortened = $s !== $tmp;
		}

		if (!preg_match('#^[\x09\x0A\x0D\x20-\x7E\xA0-\x{10FFFF}]*+\z#u', $s) || preg_last_error()) {
			if ($maxLength && strlen($s) > $maxLength) {
				$s = substr($s, 0, $maxLength);
				$shortened = TRUE;
			}
			$s = strtr($s, $table);

		} elseif ($maxLength && strlen($s) > $maxLength && !function_exists('mb_substr')) {
			$i = $len = 0;
			do {
				if (($s[$i] < "\x80" || $s[$i] >= "\xC0") && (++$len > $maxLength)) {
					$s = substr($s, 0, $i);
					$shortened = TRUE;
					break;
				}
			} while (isset($s[++$i]));
		}

		return $s . (empty($shortened) ? '' : ' ... ');
	}


	/**
	 * @internal
	 * @return string UTF-8
	 */
	public static function encodeString3($s, $maxLength = NULL)
	{
		static $table;
		if ($table === NULL) {
			foreach (array_merge(range("\x00", "\x1F"), range("\x7F", "\xFF")) as $ch) {
				$table[$ch] = '\x' . str_pad(dechex(ord($ch)), 2, '0', STR_PAD_LEFT);
			}
			$table['\\'] = '\\\\';
			$table["\r"] = '\r';
			$table["\n"] = '\n';
			$table["\t"] = '\t';
		}

		if ($maxLength && $s !== '') {
			$tmp = $s;
			if (!preg_match('##u', $s)) {
				$s = substr($s, 0, $maxLength);
			} elseif (function_exists('mb_substr')) {
				$s = mb_substr($s, 0, $maxLength, 'UTF-8');
			} else {
				$i = $len = 0;
				do {
					if (($s[$i] < "\x80" || $s[$i] >= "\xC0") && (++$len > $maxLength)) {
						$s = substr($s, 0, $i);
						break;
					}
				} while (isset($s[++$i]));
			}
			if ($s !== $tmp) {
				$s .= ' ... ';
			}
		}

		if (preg_match('#[^\x09\x0A\x0D\x20-\x7E\xA0-\x{10FFFF}]#u', $s) || preg_last_error()) {
			$s = strtr($s, $table);
		}

		return $s;
	}
}

$testCount = 1000;
$dataSets = [
	'100 -> 100' => [$testCount, FALSE, str_repeat('a', 100), 100],
	'500 -> 100' => [$testCount, FALSE, str_repeat('a', 500), 100],
	'1e5 -> 100' => [$testCount, FALSE, str_repeat('a', 1e5), 100],
	'1e5 -> 100 (UTF-8)' => [$testCount, FALSE, str_repeat("\xC4\x9B", 1e5), 100],
	'1e5 -> 100 (UTF-8 mis-aligned 1)' => [$testCount, FALSE, 'a' . str_repeat("\xC4\x9B", 1e5), 100],
	'1e5 -> 100 (UTF-8 mis-aligned 2)' => [$testCount, FALSE, 'aa' . str_repeat("\xC4\x9B", 1e5), 100],
	'1e5 -> 100 (UTF-8 mis-aligned 3)' => [$testCount, FALSE, 'aaa' . str_repeat("\xC4\x9B", 1e5), 100],
	'1e5 -> 100 (binary 1)' => [$testCount, TRUE, 'a' . str_repeat("\xf0\x90\x28\xbc", 1e5), 100],
	'1e5 -> 100 (binary 2)' => [$testCount, TRUE, 'a' . str_repeat("\xc3\x28", 1e5), 100],
	'1e5 -> 100 (binary 3)' => [$testCount, TRUE, 'a' . str_repeat("\xa0\xa1", 1e5), 100],
	'1e5 -> 100 (binary 4)' => [$testCount, TRUE, 'a' . str_repeat("\xe2\x28\xa1", 1e5), 100],
	'1e5 -> 100 (binary 5)' => [$testCount, TRUE, 'a' . str_repeat("\xe2\x82\x28", 1e5), 100],
	'1e5 -> 100 (binary 6)' => [$testCount, TRUE, 'a' . str_repeat("\xf0\x90\x28\xbc", 1e5), 100],
	'1e5 -> 100 (binary 7)' => [$testCount, TRUE, 'a' . str_repeat("\xf8\xa1\xa1\xa1\xa1", 1e5), 100],
	'1e5 -> 100 (binary 8)' => [$testCount, TRUE, 'a' . str_repeat("\xfc\xa1\xa1\xa1\xa1\xa1", 1e5), 100],
	'1e5 -> 100 (binary 9)' => [$testCount, TRUE, 'a' . str_repeat("\xed\xa0\x80", 1e5), 100],
	'1e5 -> 100 (binary X)' => [$testCount, TRUE, 'a' . str_repeat("\xf0\x82\x82\xac", 1e5), 100],
];

$tests = [
	'consistency' => function ($count, $isBinary, $s, $maxLength) {
		$expected = Dumper::encodeString($s);
		$actualA = Dumper::encodeString2($s);
		$actualB = Dumper::encodeString3($s);

		assert($actualA === $expected);
		assert($actualB === $expected);
	},

	'original' => function ($count, $isBinary, $s, $maxLength) {
		while ($count--) {
			Dumper::encodeString($s, $maxLength);
		}
	},

	'pr' => function ($count, $isBinary, $s, $maxLength) {
		while ($count--) {
			Dumper::encodeString2($s, $maxLength);
		}
	},

	'dg' => function ($count, $isBinary, $s, $maxLength) {
		while ($count--) {
			Dumper::encodeString3($s, $maxLength);
		}
	},
];

$padLength = max(array_map('strlen', array_keys($dataSets))) + 2;
foreach ($tests as $testName => $test) {
	printf("Test %s:\n", $testName);
	foreach ($dataSets as $dataSetName => $dataSet) {
		$time = -microtime(TRUE);
		$test(...$dataSet);
		$time += microtime(TRUE);
		printf("  %s%5.0f ms\n", str_pad($dataSetName, $padLength), $time * 1e3);
	}
	printf("\n");
}