Allow \n literals in strings

example/bench_all.primi

@@ -111,7 +111,7 @@ xb_2 = b - i_2;
 xc = c - j;
 xc_k = c - k;
 l = "xoxoxomilanobergamo,anno:domini. Anno, neštvi ma.".string_replace(r"ann?o", "FAIL");
-m = "\ahoj\n\vo\le" - r"\\ahoj\s";
+m = "\\ahoj\n\\vo\\le" - r"\\ahoj\s";
 n = "a/b/c" - r"\/b"; // Test proper handling of escaped regex delimiters.
 a = 4;
 b = 5;

src/handlers/StringLiteral.php

@@ -1,9 +1,11 @@
 <?php
 
+declare(strict_types = 1);
+
 namespace Smuuf\Primi\Handlers;
 
 use \Smuuf\Primi\Structures\StringValue;
-use \Smuuf\Primi\Structures\NumberValue;
+use \Smuuf\Primi\Helpers\StringEscaping;
 use \Smuuf\Primi\Context;
 
 class StringLiteral extends \Smuuf\Primi\StrictObject implements IHandler {
@@ -16,6 +18,7 @@ public static function handle(array $node, Context $context) {
 		// Using trim("\"'", ...) would make "abc'" into abc instead of abc',
 		// so do this a little more directly.
 		$value = \mb_substr($content, 1, \mb_strlen($content) - 2);
+		$value = StringEscaping::unescapeString($value);
 
 		return new StringValue($value);
 

src/helpers/StringEscaping.php

@@ -0,0 +1,102 @@
+<?php
+
+declare(strict_types = 1);
+
+namespace Smuuf\Primi\Helpers;
+
+use \Smuuf\Primi\ErrorException;
+
+class StringEscaping extends \Smuuf\Primi\StrictObject {
+
+	/**
+	 * Escape sequences that are supported in string literals.
+	 *
+	 * Left side represents character after a backslash which together form a
+	 * supported escape sequence. Right side represents a result of such
+	 * sequence.
+	 *
+	 * @const
+	 */
+	private const ESCAPE_PAIRS = [
+		'\\\\' => "\\",
+		'\\n' => "\n",
+		'\\t' => "\t",
+		'\\"' => '"',
+		"\\'" => "'",
+	];
+
+	private const QUOTE_CHARS = ['"', "'"];
+
+	/**
+	 * Return the provided string but with known escape sequences being expanded
+	 * to their literal meaning. For example `\n` will be expanded to literal
+	 * new-line.
+	 */
+	public static function unescapeString(string $str): string {
+
+		return \preg_replace_callback('#(\\\\.)#', function($m) {
+
+			$char = $m[1];
+			foreach (self::ESCAPE_PAIRS as $in => $out) {
+				if ($char === $in) {
+					return $out;
+				}
+			}
+
+			// The backslashed character doesn't represent any known escape
+			// sequence, therefore error.
+			throw new ErrorException(
+				"Unrecognized string escape sequence '{$m[0]}'."
+			);
+
+		}, $str);
+
+	}
+
+	/**
+	 * The string provided as an argument will be returned, but with added
+	 * escaping for characters that represent a known escape sequence. For
+	 * example a literal new-line will be replaced by `\n`.
+	 *
+	 * This is useful for converting a internal string value of StringValue to
+	 * a source code representaton of it - that is how a string literal would
+	 * have to be written by hand for it to be - as a result - interpreted as
+	 * that original string).
+	 *
+	 * If a third optional $quoteChar argument is passed, all other known
+	 * quote characters will NOT be escaped - only the one specified by the
+	 * third argument. The only known quote characters are `'` and `"`. For
+	 * example the string `hello'there"` without $quoteChar specified will
+	 * be escaped as `hello\'there\"`. But with $quoteChar being `"` it would
+	 * result in `hello'there\"`, since the caller used the third argument to
+	 * specify that the single-quote does NOT have to be escaped.
+	 */
+	public static function escapeString(
+		string $str,
+		string $quoteChar = null
+	): string {
+
+		foreach (self::ESCAPE_PAIRS as $out => $in) {
+
+			// $in = <new line>
+			// $out = '\n'
+
+			if (
+				$quoteChar !== null
+				&& in_array($in, self::QUOTE_CHARS, true)
+				&& $in !== $quoteChar
+			) {
+				// Do not escape quote characters that aren't necessary to be
+				// escaped.
+				continue;
+			}
+
+			$str = str_replace($in, $out, $str);
+
+		}
+
+		return $str;
+
+	}
+
+}

src/parser/CompiledParser.php

@@ -14,11 +14,11 @@ class CompiledParser extends Parser\Packrat {
 	/** @var string **/
 	public $string;
 
-/* StringLiteral: / (?:".*?(?<!\\)")|(?:'.*?(?<!\\)') /s */
+/* StringLiteral: / ("[^"\\]*(\\.[^"\\]*)*")|('[^'\\]*(\\.[^'\\]*)*') /s */
 protected $match_StringLiteral_typestack = array('StringLiteral');
 function match_StringLiteral ($stack = []) {
 	$matchrule = "StringLiteral"; $result = $this->construct($matchrule, $matchrule, \null);
-	if (( $subres = $this->rx( '/ (?:".*?(?<!\\\\)")|(?:\'.*?(?<!\\\\)\') /s' ) ) !== \false) {
+	if (( $subres = $this->rx( '/ ("[^"\\\\]*(\\\\.[^"\\\\]*)*")|(\'[^\'\\\\]*(\\\\.[^\'\\\\]*)*\') /s' ) ) !== \false) {
 		$result["text"] .= $subres;
 		return $this->finalise($result);
 	}

src/parser/Grammar.peg

@@ -21,7 +21,7 @@ class CompiledParser extends Parser\Packrat {
 
 # Carefully crafted (or stolen somewhere?) string matching regex which supports
 # both " and ' quotes, doesn't cause JIT stack overflow and also supports escaped quotes.
-StringLiteral: / (?:".*?(?<!\\)")|(?:'.*?(?<!\\)') /s
+StringLiteral: / ("[^"\\]*(\\.[^"\\]*)*")|('[^'\\]*(\\.[^'\\]*)*') /s
 NumberLiteral: /-?\d+(\.\d+)?/
 BoolLiteral: "true" | "false"
 NullLiteral: "null"

src/structures/StringValue.php

@@ -3,6 +3,7 @@
 namespace Smuuf\Primi\Structures;
 
 use \Smuuf\Primi\Helpers\Common;
+use \Smuuf\Primi\Helpers\StringEscaping;
 use \Smuuf\Primi\ISupportsComparison;
 use \Smuuf\Primi\ISupportsAddition;
 use \Smuuf\Primi\ISupportsSubtraction;
@@ -22,14 +23,14 @@ class StringValue extends Value implements
 	const TYPE = "string";
 
 	public function __construct(string $value) {
-		$this->value = self::expandSequences($value);
+		$this->value = $value;
 	}
 
 	public function getStringValue(): string {
 
 		// We are about to put double-quotes around the return value,
 		// so let's "escape" double-quotes present in the string value.
-		$escaped = \str_replace('"', '\"', $this->value);
+		$escaped = StringEscaping::escapeString($this->value, '"');
 		return "\"$escaped\"";
 
 	}
@@ -159,7 +160,10 @@ public function getIterator(): \Iterator {
 	protected static function expandSequences(string $string) {
 
 		// Primi strings support some escape sequences.
-		return \str_replace('\n', "\n", $string);
+		//return \str_replace('\n', "\n", $string);
+		//return \preg_replace('#(?<!\\\\)\\\\n#', "\n", $string);
+		$string = \preg_replace('#(?<!\\\\)\\\\n#', "\n", $string);
+		return \str_replace('\\\\', "\\", $string);
 
 	}
 

tests/language/suites/extensions/ext.string.expect

@@ -14,3 +14,8 @@ m:BoolValue:true
 n:BoolValue:false
 o:StringValue:"1,2,3"
 p:StringValue:"1,nested,stuff-yeah,3"
+esc_1:StringValue:"\n"
+esc_2:StringValue:"\\n"
+esc_3:StringValue:"\n\n"
+esc_4:StringValue:"\\n\\n"
+esc_5:StringValue:"\n\\n\t\\"

tests/language/suites/extensions/ext.string.primi

@@ -23,3 +23,9 @@ n = "ahoj ahoj ahoj!".contains(r"(čau ){2}!?");
 
 o = ",".join([1, 2, 3]);
 p = ",".join([1, ['nested', "stuff-{}".format("yeah")], 3]);
+
+esc_1 = "\n"
+esc_2 = "\\n"
+esc_3 = "\n" + "\n"
+esc_4 = "\\n" + "\\n"
+esc_5 = "\n" + "\\n\t\\"

tests/language/suites/structures/regexes.expect

@@ -16,7 +16,7 @@ xb_2:StringValue:"ab123dfg"
 xc:StringValue:"xyzčaukomňauko"
 xc_k:StringValue:"xyz456čako"
 l:StringValue:"xoxoxomilFAILbergamo,FAIL:domini. Anno, neštvi ma."
-m:StringValue:"\vo\le"
+m:StringValue:"\\vo\\le"
 n:StringValue:"a/c"
 o:StringValue:"00011"
 p:StringValue:"00011234455"

tests/language/suites/structures/regexes.primi

@@ -20,7 +20,7 @@ xc = c - j;
 xc_k = c - k;
 
 l = "xoxoxomilanobergamo,anno:domini. Anno, neštvi ma.".string_replace(r"ann?o", "FAIL");
-m = "\ahoj\n\vo\le" - r"\\ahoj\s";
+m = "\\ahoj\n\\vo\\le" - r"\\ahoj\s";
 n = "a/b/c" - r"/b"; // Test proper handling of escaped regex delimiters.
 
 o = "[0-9]{5}".to_regex().match("abc00011234455");

tests/unit/method.helper.phpt

@@ -5,8 +5,6 @@ use \Tester\Assert;
 use \Smuuf\Primi\Helpers\Common;
 use \Smuuf\Primi\Structures\StringValue;
 use \Smuuf\Primi\Structures\Value;
-use \Smuuf\Primi\Structures\RegexValue;
-use \Smuuf\Primi\ErrorException;
 
 require __DIR__ . '/../bootstrap.php';
 

tests/unit/value.string.phpt

@@ -22,13 +22,33 @@ $fns = ExtensionHub::get();
 $string = new StringValue("this is a string.");
 $letterA = new StringValue("a");
 $unicode = new StringValue("ťhiš íš á ŠTřing.");
-$withNewline = new StringValue('a \n b');
 
 //
-// Test sequence expanding...
+// Escaping sequences are NOT handled inside StringValue, but instead during the
+// handling of source code string literal.
 //
 
-Assert::same(2, count(explode("\n", get_val($withNewline))));
+$stringsWithEscapeSequences = [
+	'a\nb',
+	'a \n b',
+	'a \\n b',
+	'a\\n\nb',
+	'a\\\n\t\\tb',
+];
+foreach ($stringsWithEscapeSequences as $s) {
+	Assert::same($s, get_val(new StringValue($s)));
+}
+
+// Get correct repr - things should be quoted and escaped properly.
+// REMEMBER: Escape characters are NOT HANDLED when creating StringValue
+// objects. Whatever is put into StringValue as argument will literally be
+// what's inside.
+Assert::same('"\""', (new StringValue('"'))->getStringValue());
+Assert::same('"\\\\\""', (new StringValue('\\"'))->getStringValue());
+Assert::same('"\\\\\'"', (new StringValue("\'"))->getStringValue());
+Assert::same('"\\\\\\\\\'"', (new StringValue("\\\\'"))->getStringValue());
+Assert::same('"\\\\n"', (new StringValue('\n'))->getStringValue());
+Assert::same('"\\n"', (new StringValue("\n"))->getStringValue());
 
 //
 // Test adding and subtracting...
@@ -267,8 +287,6 @@ Assert::same(17, get_val($fns['string_length']->invoke([$string])));
 Assert::same(1, get_val($fns['string_length']->invoke([$letterA])));
 // Multibyte strings should report length correctly.
 Assert::same(17, get_val($fns['string_length']->invoke([$unicode])));
-// "\n" is expanded as newline - that's one character.
-Assert::same(5, get_val($fns['string_length']->invoke([$withNewline])));
 
 //
 // Test replacing.