Dumper: improved encoding of strings

Author: dg

src/Framework/Dumper.php

@@ -33,17 +33,6 @@ class Dumper
 	 */
 	public static function toLine($var): string
 	{
-		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';
-			$table["\t"] = '\t';
-		}
-
 		if (is_bool($var)) {
 			return $var ? 'TRUE' : 'FALSE';
 
@@ -62,9 +51,7 @@ public static function toLine($var): string
 			} elseif (strlen($var) > self::$maxLength) {
 				$var = substr($var, 0, self::$maxLength) . '...';
 			}
-			return preg_match('#[^\x09\x0A\x0D\x20-\x7E\xA0-\x{10FFFF}]#u', $var) || preg_last_error()
-				? '"' . strtr($var, $table) . '"'
-				: "'$var'";
+			return self::encodeString($var, false);
 
 		} elseif (is_array($var)) {
 			$out = '';
@@ -146,20 +133,10 @@ private static function _toPhp(&$var, array &$list = [], int $level = 0, int &$l
 		} elseif ($var === null) {
 			return 'null';
 
-		} elseif (is_string($var) && (preg_match('#[^\x09\x20-\x7E\xA0-\x{10FFFF}]#u', $var) || preg_last_error())) {
-			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';
-				$table['$'] = '\$';
-				$table['"'] = '\"';
-			}
-			return '"' . strtr($var, $table) . '"';
+		} elseif (is_string($var)) {
+			$res = self::encodeString($var, true);
+			$line += substr_count($res, "\n");
+			return $res;
 
 		} elseif (is_array($var)) {
 			$space = str_repeat("\t", $level);
@@ -242,9 +219,56 @@ private static function _toPhp(&$var, array &$list = [], int $level = 0, int &$l
 			return '/* resource ' . get_resource_type($var) . ' */';
 
 		} else {
-			$res = var_export($var, true);
-			$line += substr_count($res, "\n");
-			return $res;
+			return var_export($var, true);
+		}
+	}
+
+
+	private static function encodeString(string $s, bool $asPhp): string
+	{
+		static $specials = [
+			true => [
+				"\r" => '\r',
+				"\n" => '\n',
+				"\t" => "\t",
+				"\e" => '\e',
+				'\\' => '\\\\',
+			],
+			false => [
+				"\r" => "\r",
+				"\n" => "\n",
+				"\t" => "\t",
+				'\\' => '\\\\',
+			],
+		];
+		$special = $specials[$asPhp];
+		$utf8 = preg_match('##u', $s);
+		$escaped = preg_replace_callback(
+			$utf8 ? '#[\p{C}\\\\]#u' : '#[\x00-\x1F\x7F-\xFF\\\\]#',
+			function ($m) use ($special) {
+				return $special[$m[0]] ?? (strlen($m[0]) === 1
+						? '\x' . str_pad(strtoupper(dechex(ord($m[0]))), 2, '0', STR_PAD_LEFT) . ''
+						: '\u{' . strtoupper(ltrim(dechex(self::utf8Ord($m[0])), '0')) . '}');
+			},
+			$s
+		);
+		return $s === str_replace('\\\\', '\\', $escaped)
+			? "'" . preg_replace('#\'|\\\\(?=[\'\\\\]|$)#D', '\\\\$0', $s) . "'"
+			: '"' . addcslashes($escaped, '"$') . '"';
+	}
+
+
+	private static function utf8Ord(string $c): int
+	{
+		$ord0 = ord($c[0]);
+		if ($ord0 < 0x80) {
+			return $ord0;
+		} elseif ($ord0 < 0xE0) {
+			return ($ord0 << 6) + ord($c[1]) - 0x3080;
+		} elseif ($ord0 < 0xF0) {
+			return ($ord0 << 12) + (ord($c[1]) << 6) + ord($c[2]) - 0xE2080;
+		} else {
+			return ($ord0 << 18) + (ord($c[1]) << 12) + (ord($c[2]) << 6) + ord($c[3]) - 0x3C82080;
 		}
 	}
 

tests/Framework/Dumper.toLine.phpt

@@ -22,8 +22,9 @@ Assert::match("''", Dumper::toLine(''));
 Assert::match("' '", Dumper::toLine(' '));
 Assert::match("'0'", Dumper::toLine('0'));
 Assert::match('"\\x00"', Dumper::toLine("\x00"));
+Assert::match('"\u{FEFF}"', Dumper::toLine("\xEF\xBB\xBF")); // BOM
 Assert::match("'	'", Dumper::toLine("\t"));
-Assert::match('"\\xff"', Dumper::toLine("\xFF"));
+Assert::match('"\\xFF"', Dumper::toLine("\xFF"));
 Assert::match("'multi\nline'", Dumper::toLine("multi\nline"));
 Assert::match("'Iñtërnâtiônàlizætiøn'", Dumper::toLine("I\xc3\xb1t\xc3\xabrn\xc3\xa2ti\xc3\xb4n\xc3\xa0liz\xc3\xa6ti\xc3\xb8n"));
 Assert::match('resource(stream)', Dumper::toLine(fopen(__FILE__, 'r')));

tests/Framework/Dumper.toPhp.phpt

@@ -29,8 +29,9 @@ Assert::match("''", Dumper::toPhp(''));
 Assert::match("' '", Dumper::toPhp(' '));
 Assert::match("'0'", Dumper::toPhp('0'));
 Assert::match('"\\x00"', Dumper::toPhp("\x00"));
+Assert::match('"\u{FEFF}"', Dumper::toPhp("\xEF\xBB\xBF")); // BOM
 Assert::match("'	'", Dumper::toPhp("\t"));
-Assert::match('"\\xff"', Dumper::toPhp("\xFF"));
+Assert::match('"\\xFF"', Dumper::toPhp("\xFF"));
 Assert::match('"multi\nline"', Dumper::toPhp("multi\nline"));
 Assert::match("'Iñtërnâtiônàlizætiøn'", Dumper::toPhp("I\xc3\xb1t\xc3\xabrn\xc3\xa2ti\xc3\xb4n\xc3\xa0liz\xc3\xa6ti\xc3\xb8n"));
 Assert::match('[
@@ -41,6 +42,10 @@ Assert::match('[
 	[1 => 1, 2, 3, 4, 5, 6, 7, \'abcdefgh\'],
 ]', Dumper::toPhp([1, 'hello', "\r" => [], [1, 2], [1 => 1, 2, 3, 4, 5, 6, 7, 'abcdefgh']]));
 
+Assert::match('\'$"\\\\\'', Dumper::toPhp('$"\\'));
+Assert::match('\'$"\\ \x00\'', Dumper::toPhp('$"\\ \x00'));
+Assert::match('"\\$\\"\\\\ \x00"', Dumper::toPhp("$\"\\ \x00"));
+
 Assert::match('/* resource stream */', Dumper::toPhp(fopen(__FILE__, 'r')));
 Assert::match('(object) /* #%a% */ []', Dumper::toPhp((object) null));
 Assert::match("(object) /* #%a% */ [